Metabolismo Celular

Cell Metabolism
Previews
Treating Obesity Like a Tumor
Randy J. Seeley1,*
1University of Cincinnati, Cincinnati, OH 45237, USA
*Correspondence: randy.seeley@uc.edu
DOI 10.1016/j.cmet.2011.12.007
Expanding adipose tissue in obesity requires a great deal of angiogenesis to support increasing volumes of
tissue. A growing body of evidence indicates that inhibiting these blood vessels can result in substantial
weight loss, and now this has been demonstrated in nonhuman primates.
Extremely skeptical. That is the best
description of my response to publications indicating that either diet-induced
or genetic forms of obesity could be reversed by giving inhibitors to blood vessel
formation. The first of these reports came
from Maria Rupnick and Judah Folkman,
who used agents that inhibit tumor growth
and found profound weight loss in mice
(Rupnick et al., 2002). The next of these
reports used a more sophisticated approach in an attempt to direct the inhibition of the blood vessels specifically to
adipose tissue. In an effort led by Wadih
Arap and Renata Pasqulini, they used
a technology called phage display that
had been designed to identify the signatures of tumor-associated blood vessels
that might be different than other blood
vessels. However, they determined that
many tissues, including adipose tissue,
had unique properties. They were able to
identify short peptide sequences that
would selectively bind to the vasculature
of white adipose tissue, but not other
tissues they examined. As cancer researchers, they took the next logical step.
They used this peptide and attached a
poison pill that would produce apoptosis
in the targeted blood vessels (Kolonin
et al., 2004). This targeted approach also
led to rapid and substantial weight loss
in mice. In their most recent manuscript
published in Science Translational Medicine, this group has extended these
results to obese nonhuman primates,
showing substantial weight and body fat
loss after 28 days of treatment with this
peptide (Barnhart et al., 2011).
The reason for my skepticism toward
this approach centered on the assumption that, whether untargeted or targeted,
reducing adipose tissue blood vessels
would impair adipose tissue function.
While obesity is a scourge to be fought
and is the direct result of expanding
adipose tissue, the truth is that healthy

adipose tissue serves an important function to protect the rest of the body from
nutrients that, when stored in other tissues such as muscle and liver, cause
metabolic dysfunction. The most obvious
example of this comes from humans
or mice who fail to make sufficient
adipocytes. While leaner, such individuals
nevertheless have severe metabolic problems, including liver steatosis, hyperlipidemia, and severe insulin resistance
(Huang-Doran and Savage, 2011). Thus,
it seemed likely that such an approach
that compromised adipose tissue function could result in leaner individuals
who were more at risk for metabolic
disease.
Ultimately, the data in mice and
nonhuman primates simply do not support my assumption. In nonhuman primates, weight loss is accompanied by
reduced insulin resistance (Barnhart et al.,
2011). In mice, our own work has demonstrated that treatment with this peptide results in rapid weight loss that is
primarily due to reduced intake, and it
is accompanied by metabolic improvements (Kim et al., 2010). The important
point here is that the response to targeting the adipose tissue vasculature is
the exact opposite of what is observed
when adipocytes are not present or are
pushed into apoptosis. While removal of
adipocytes is associated with increased
intake and decreased insulin sensitivity
(Pajvani et al., 2005), removal of the supporting vasculature results in decreased
intake and increased insulin sensitivity
(Kim et al., 2010). The conclusion to be
drawn is that adipocytes are an important
source of signals to both the brain and
other tissues and that the removal of
those signals is deleterious. Targeting
the adipose tissue vasculature results in
changes in adipocyte communication
that promote weight loss and improved

metabolic regulation.
This work has opened up an entirely
unappreciated aspect of adipose tissue
biology that explores the intimate relationship between adipocytes and their supporting vasculature. More importantly,
understanding and manipulating this relationship has important therapeutic implications, given the potent effects of this
particular peptide. A crucial question is
whether this approach borrowed from
cancer treatment is going to be sufficiently safe to be used in the growing
number of individuals suffering under
the burden of obesity and its comorbid
conditions. After all, treating cancer is
considerably different from treating a
chronic condition such as obesity. Cancer
patients are often under a short-term
threat, while obesity is a much longerterm threat to an individuals health. As
a consequence, the risk-benefit analysis
is considerably different. For example,
the specificity of the targeting is less of a
concern in cancer as compared to obesity
treatment. Imagine a peptide with 90%
targeting selectivity to the tumor. Given
that the plan would be to treat the cancer
patient for weeks or months, as long as
the tumor is being undermined faster
than normal tissue, and that normal tissue
can recover once treatment is terminated,
it can be a viable therapy. For the obese
patient who is likely to be taking such a
treatment for the better part of his or
her life, would 90% targeting selectivity
be sufficient to avoid adverse effects
on other tissues? Ninety-five percent?
Ninety-nine percent? This is a complex
question that will need to be answered
before we know whether this approach
will become therapy.
The important new insights driven
by the work with these targeted peptides are an important advance in an
Cell Metabolism 15, January 4, 2012 2012 Elsevier Inc. 1
Cell Metabolism
Previews
environment where few effective treatment strategies short of bariatric surgery are available to help obese patients.
It is clear that more creative strategies
from a wider range of disciplines are
needed. To that end, further understanding of how adipose tissue signaling is altered by various aspects of its
milieu, including the supporting blood
vessels, macrophages, and its extracellular matrix, is necessary if we are to bring
more therapies to the large unmet medical
need presented by increasing rates of

obesity.
Kim, D.H., Woods, S.C., and Seeley, R.J. (2010).

Diabetes 59, 907915.
REFERENCES
Kolonin, M.G., Saha, P.K., Chan, L., Pasqualini, R.,

and Arap, W. (2004). Nat. Med. 10, 625632.
Barnhart, K.F., Christianson, D.R., Hanley, P.W.,

Driessen, W.H., Bernacky, B.J., Baze, W.B., Wen,
S., Tian, M., Ma, J., Kolonin, M.G., Saha, P.K.,
Do, K.A., Hulvat, J.F., Gelovani, J.G., Chan, L.,
Arap, W., and Pasqualini, R. (2011). Sci. Transl.
Med. 3, 108ra112.
Huang-Doran, I., and Savage, D.B. (2011). Pediatr.
Endocrinol. Rev. 8, 190199.
Pajvani, U.B., Trujillo, M.E., Combs, T.P., Iyengar,

P., Jelicks, L., Roth, K.A., Kitsis, R.N., and Scherer,
P.E. (2005). Nat. Med. 11, 797803.
Rupnick, M.A., Panigrahy, D., Zhang, C.Y., Dallabrida, S.M., Lowell, B.B., Langer, R., and Folkman,
M.J. (2002). Proc. Natl. Acad. Sci. USA 99,
1073010735.
Reactive Oxygen Species Resulting

from Mitochondrial Mutation Diminishes
Stem and Progenitor Cell Function
Brian S. Garrison1,2 and Derrick J. Rossi1,2,*
1Stem
Cell and Regenerative Biology Department, Harvard University, Cambridge, MA 02138, USA
Immune Disease Institute and Program in Cellular and Molecular Medicine, Childrens Hospital Boston, Boston, MA 02115, USA
*Correspondence: rossi@idi.harvard.edu
DOI 10.1016/j.cmet.2011.12.008
2The
While age-dependent stem cell decline is widely recognized as being a key component of organismal aging,
the underlying mechanisms remain elusive. In this issue of Cell Metabolism, Suomalainen and colleagues
provide evidence that mitochondrial mutation and associated reactive oxygen species can adversely impact
tissue-specific stem and progenitor cell function.
Physiological aging invariably leads to
a loss in normal tissue maintenance and
reduced regenerative potential. The fact
that these processes are normally under
the functional purview of adult tissuespecific stem cells implicates age-associated stem cell decline as a fundamental
contributor to the aging of tissues and
organisms. Indeed, the importance of
the stem cell compartment in contributing
to age-associated pathophysiology has
been demonstrated in a number of studies (Rossi et al., 2008). Consistent with
the inherent complexity of physiological
aging, the mechanistic basis for agerelated stem cell decline is similarly
complex, with evidence suggesting the
involvement of cellular, genetic, and epigenetic components (Rossi et al., 2008).
However, recent evidence from a number
of papers, including a paper by Suomalainen and colleagues in this issue of Cell
Metabolism, suggests that accumulating

mitochondrial DNA damage may also be
an important contributor to somatic stem
cell decline with age.
Mitochondria are frequently referred
to as the cells power plants since
they play a fundamental role in the
production of adenosine triphosphate
(ATP) through oxidative phosphorylation
(OXPHOS). Most aerobic organisms use
some form of OXPHOS because it is
a highly efficient method for producing ATP; however, the downside of this
energy-producing pathway is that it also
leads to the production of reactive oxygen
species (ROS) that have the potential to
damage cellular macromolecules and, in
such a way, contribute to aging. Mitochondrial DNA (mtDNA) is believed to be
highly susceptible to oxidative damage
in part because of its proximity to the electron transport chain, but also because
2 Cell Metabolism 15, January 4, 2012 2012 Elsevier Inc.
mtDNA lacks protective histones. Accumulation of damage in the mitochondrial

genome has been proposed to lead to
mitochondrial dysfunction and concomitant cellular decline and, in such a way,
contribute to physiological aging (Harman, 1972). This long-held theory was
supported experimentally with the generation of mtDNA mutator mice bearing a
proofreading-deficient mtDNA polymerase (POLG) that exhibit a spectrum of
degenerative phenotypes reminiscent of
aging (Kujoth et al., 2005; Trifunovic
et al., 2004). More recently, these mice
have also provided the necessary experimental tool to address how mtDNA
mutation accumulation impacts stem cell
function and to determine whether it
contributes to age-associated stem cell
decline.
Within mammalian tissues, aging has
been most comprehensively studied in
Cell Metabolism
Previews
environment where few effective treatment strategies short of bariatric surgery are available to help obese patients.
It is clear that more creative strategies
from a wider range of disciplines are
needed. To that end, further understanding of how adipose tissue signaling is altered by various aspects of its
milieu, including the supporting blood
vessels, macrophages, and its extracellular matrix, is necessary if we are to bring
more therapies to the large unmet medical
need presented by increasing rates of

obesity.
Kim, D.H., Woods, S.C., and Seeley, R.J. (2010).

REFERENCES
Kolonin, M.G., Saha, P.K., Chan, L., Pasqualini, R.,

and Arap, W. (2004). Nat. Med. 10, 625632.
Barnhart, K.F., Christianson, D.R., Hanley, P.W.,

Driessen, W.H., Bernacky, B.J., Baze, W.B., Wen,
S., Tian, M., Ma, J., Kolonin, M.G., Saha, P.K.,
Do, K.A., Hulvat, J.F., Gelovani, J.G., Chan, L.,
Arap, W., and Pasqualini, R. (2011). Sci. Transl.
Med. 3, 108ra112.
Huang-Doran, I., and Savage, D.B. (2011). Pediatr.
Endocrinol. Rev. 8, 190199.
Pajvani, U.B., Trujillo, M.E., Combs, T.P., Iyengar,

P., Jelicks, L., Roth, K.A., Kitsis, R.N., and Scherer,
P.E. (2005). Nat. Med. 11, 797803.
Rupnick, M.A., Panigrahy, D., Zhang, C.Y., Dallabrida, S.M., Lowell, B.B., Langer, R., and Folkman,
M.J. (2002). Proc. Natl. Acad. Sci. USA 99,
1073010735.
Reactive Oxygen Species Resulting

from Mitochondrial Mutation Diminishes
Stem and Progenitor Cell Function
Brian S. Garrison1,2 and Derrick J. Rossi1,2,*
1Stem
Cell and Regenerative Biology Department, Harvard University, Cambridge, MA 02138, USA
Immune Disease Institute and Program in Cellular and Molecular Medicine, Childrens Hospital Boston, Boston, MA 02115, USA
*Correspondence: rossi@idi.harvard.edu
DOI 10.1016/j.cmet.2011.12.008
2The
While age-dependent stem cell decline is widely recognized as being a key component of organismal aging,
the underlying mechanisms remain elusive. In this issue of Cell Metabolism, Suomalainen and colleagues
provide evidence that mitochondrial mutation and associated reactive oxygen species can adversely impact
tissue-specific stem and progenitor cell function.
Physiological aging invariably leads to
a loss in normal tissue maintenance and
reduced regenerative potential. The fact
that these processes are normally under
the functional purview of adult tissuespecific stem cells implicates age-associated stem cell decline as a fundamental
contributor to the aging of tissues and
organisms. Indeed, the importance of
the stem cell compartment in contributing
to age-associated pathophysiology has
been demonstrated in a number of studies (Rossi et al., 2008). Consistent with
the inherent complexity of physiological
aging, the mechanistic basis for agerelated stem cell decline is similarly
complex, with evidence suggesting the
involvement of cellular, genetic, and epigenetic components (Rossi et al., 2008).
However, recent evidence from a number
of papers, including a paper by Suomalainen and colleagues in this issue of Cell
Metabolism, suggests that accumulating

mitochondrial DNA damage may also be
an important contributor to somatic stem
cell decline with age.
Mitochondria are frequently referred
to as the cells power plants since
they play a fundamental role in the
production of adenosine triphosphate
(ATP) through oxidative phosphorylation
(OXPHOS). Most aerobic organisms use
some form of OXPHOS because it is
a highly efficient method for producing ATP; however, the downside of this
energy-producing pathway is that it also
leads to the production of reactive oxygen
species (ROS) that have the potential to
damage cellular macromolecules and, in
such a way, contribute to aging. Mitochondrial DNA (mtDNA) is believed to be
highly susceptible to oxidative damage
in part because of its proximity to the electron transport chain, but also because
mtDNA lacks protective histones. Accumulation of damage in the mitochondrial

genome has been proposed to lead to
mitochondrial dysfunction and concomitant cellular decline and, in such a way,
contribute to physiological aging (Harman, 1972). This long-held theory was
supported experimentally with the generation of mtDNA mutator mice bearing a
proofreading-deficient mtDNA polymerase (POLG) that exhibit a spectrum of
degenerative phenotypes reminiscent of
aging (Kujoth et al., 2005; Trifunovic
et al., 2004). More recently, these mice
have also provided the necessary experimental tool to address how mtDNA
mutation accumulation impacts stem cell
function and to determine whether it
contributes to age-associated stem cell
decline.
Within mammalian tissues, aging has
been most comprehensively studied in
Cell Metabolism
Previews
the hematopoietic system and, to a lesser
extent, the brain. In the blood, the hematopoietic stem cell (HSC) compartment
has been well documented to contribute
to pathophysiological conditions associated with aging, which include reduced
regenerative potential, diminished adaptive immune competence, and myelogenous disease predisposition (Rossi et al.,
2008). And while genomic DNA damage
accrual is thought to limit the regenerative
response of HSCs from old mice (Rossi
et al., 2007), the impact of mitochondrial
mutagenesis on hematopoietic aging has
until recently been unexplored. Two
groups have recently shown that the
mtDNA mutator mice exhibited a number
of hematopoietic phenotypes, including
abnormalities in erythroid and lymphocyte
development (Chen et al., 2009; Norddahl
et al., 2011). The fact that these hematologic deficiencies could be transplanted
into wild-type recipients indicated that
they were cell-autonomous defects transmitted by HSCs (Chen et al., 2009; Norddahl et al., 2011). However, although
anemia and lymphoid deficiencies are
commonly associated with aging in both
mice and people, Bryder and colleagues
went on to show that the aging of mutator stem cells was molecularly distinct
from normal physiological stem cell aging
(Norddahl et al., 2011). Using a transcriptional profiling strategy to compare
steady-state mutator HSCs against stem
cells purified from old wild-type mice, little
similarity was noted in their respective
profiles, suggesting that, at least at this
resolution, mitochondrial mutation-driven
stem cell decline could be uncoupled
from normal physiological stem cell aging
(Norddahl et al., 2011). Nonetheless, it remains possible that mitochondrial mutation contributes mechanistically to stem
cell aging despite the lack of evidence
at the transcriptional level, perhaps by
influencing the mobilization of energy
stores that these normally dormant stem
cells must harness when called into
action under conditions of stress or
regeneration.
An important outstanding question was
whether the hematopoietic phenotypes
exhibited by the mutator mice arose only
after a threshold level of mtDNA mutation

had occurred in the adult animals as they
aged. This is one of several issues addressed in a new article by Suomalainen
and colleagues featured in this issue
(Ahlqvist et al., 2012). To examine the
timing of the mutator hematopoietic defect manifestation, the investigators examined embryonic hematopoietic development at E13.515.5 in the fetal liver of
mutator fetuses and discovered diminished erythropoiesis and reduced frequencies of erythroid progenitors, likely
representing a precursor phenotype to
the later-observed adult anemia (Chen
et al., 2009; Norddahl et al., 2011). They
also observed a defect in fetal lymphopoiesis characterized by elevated B220positive B cells. Importantly, they discovered that administration of the antioxidant
N-acetylcysteine (NAC) throughout pregnancy could normalize fetal hematopoiesis to wild-type levels, implicating ROS
as an underlying mediator leading to the
fetal hematopoietic phenotypes observed
in the mutator mice. The researchers also
questioned how other tissue-specific
stem cell populations might be affected
in the mutator mice. In the brain, the
numbers of stem cells in the subventricular zone (SVZ) are the result of a tightly
controlled balance between the cell fates
of self-renewal, differentiation, and apoptosis. This means that controlling of the
fate of SVZ NSCs, which can be identified
by their nestin positivity, plays a critical
role in determining the number of neurons, astrocytes, and oligodendrocytes
in the brain. The researchers therefore
examined the number of nestin-positive
neural cells in the SVZ of wild-type and
mutator mice and discovered a significant
decrease in this important cell type in the
mutants. Despite the observed decrease
in NSCs, however, a multiparameter histological and biochemical examination
failed to uncover significant in vivo neurologic phenotypes in aged mutator mice.
However, when the authors examined
the NSCs using an in vitro neurosphere
self-renewal assay, they discovered that
mutator NSCs formed significantly fewer
neurospheres than wild-type cells, indicating that these cells suffer from a defect
in self-renewal that could explain the decreased NSC numbers observed in vivo.
Importantly, as the authors had observed
with the hematopoietic defects, the selfrenewal deficits of the mutator NSCs
could also be largely ameliorated through
treatment with NAC.
The findings of Suomalainen and colleagues at once support a growing body
of evidence showing how critical ROS
management is for proper stem and progenitor cell function (Ito et al., 2006; Tothova and Gilliland, 2007) and, at the
same time, identify that a consequence
of mutation accrual in the mitochondrial
genome is deregulation of ROS, which
can then work its mischief by impairing
tissue-specific stem and progenitor cells.
However, the connection of mitochondrial
genome maintenance to physiological
stem cell aging remains unclear.
REFERENCES
Ahlqvist, K., Hamalainen, R.H., Yatsuga, S., Uutela,
M., Terzioglu, M., Gotz, A., Forsstrom, S., Salven,
P., Angers-Loustau, A., Kopra, O.H., et al. (2012).
Cell Metab. 15, this issue, 100109.
Chen, M.L., Logan, T.D., Hochberg, M.L., Shelat,
S.G., Yu, X., Wilding, G.E., Tan, W., Kujoth, G.C.,
Prolla, T.A., Selak, M.A., et al. (2009). Blood 114,
40454053.
Harman, D. (1972). J. Am. Geriatr. Soc. 20,
145147.
Ito, K., Hirao, A., Arai, F., Takubo, K., Matsuoka, S.,
Miyamoto, K., Ohmura, M., Naka, K., Hosokawa,
K., Ikeda, Y., and Suda, T. (2006). Nat. Med. 12,
446451.
Kujoth, G.C., Hiona, A., Pugh, T.D., Someya, S.,
Panzer, K., Wohlgemuth, S.E., Hofer, T., Seo,
A.Y., Sullivan, R., Jobling, W.A., et al. (2005).
Science 309, 481484.
Norddahl, G.L., Pronk, C.J., Wahlestedt, M., Sten,
G., Nygren, J.M., Ugale, A., Sigvardsson, M., and
Bryder, D. (2011). Cell Stem Cell 8, 499510.
Rossi, D.J., Bryder, D., Seita, J., Nussenzweig, A.,
Hoeijmakers, J., and Weissman, I.L. (2007). Nature
447, 725729.
Rossi, D.J., Jamieson, C.H., and Weissman, I.L.
(2008). Cell 132, 681696.
Tothova, Z., and Gilliland, D.G. (2007). Cell Stem
Cell 1, 140152.
Trifunovic, A., Wredenberg, A., Falkenberg, M.,
Spelbrink, J.N., Rovio, A.T., Bruder, C.E., Bohlooly-Y, M., Gidlof, S., Oldfors, A., Wibom, R., et al.
(2004). Nature 429, 417423.
Cell Metabolism
Previews
Power Surge: Supporting Cells
Fuel Cancer Cell Mitochondria
Ubaldo E. Martinez-Outschoorn,1,2,3,4 Federica Sotgia,1,2,3,5,* and Michael P. Lisanti1,2,3,4,5,*
1The
Jefferson Stem Cell Biology and Regenerative Medicine Center

of Stem Cell Biology and Regenerative Medicine
3Department of Cancer Biology
4Department of Medical Oncology
Kimmel Cancer Center, Thomas Jefferson University, Philadelphia, PA 19107, USA
5Manchester Breast Centre and Breakthrough Breast Cancer Research Unit, Paterson Institute for Cancer Research, School of Cancer,
Enabling Sciences and Technology, Manchester Academic Health Science Centre, University of Manchester, Manchester M13 9PL, UK
*Correspondence: federica.sotgia@jefferson.edu (F.S.), michael.lisanti@kimmelcancercenter.org (M.P.L.)
DOI 10.1016/j.cmet.2011.12.011
2Department
An emerging paradigm in tumor metabolism is that catabolism in host cells fuels the anabolic growth of
cancer cells via energy transfer. A study in Nature Medicine (Nieman et al., 2011) supports this; they
show that triglyceride catabolism in adipocytes drives ovarian cancer metastasis by providing fatty acids
as mitochondrial fuels.
Our understanding of tumor metabolism
is evolving. A new central concept in
cancer metabolism is that tumor cells
function as metabolic parasites to extract
energy from supporting host cells, such
as fibroblasts and adipocytes. It has recently been demonstrated that metabolic
coupling exists in human tumors (Sotgia
et al., 2011). In two-compartment tumor
metabolism, the tumor stroma and adjacent host tissues are catabolic and the
cancer cells are anabolic (Figure 1). In
this model, energy is transferred from the
catabolic compartment to the anabolic
compartment via the sharing of nutrients
that promote tumor growth, behaving as
onco-metabolites. Although most studies
on two-compartment tumor metabolism
were first performed on fibroblasts and
breast cancer cells (Martinez-Outschoorn
et al., 2011; Sotgia et al., 2011, 2012;
Whitaker-Menezes et al., 2011), an elegant study in Nature Medicine now
broadens this emerging paradigm to adipocytes and ovarian cancer cells (Nieman
et al., 2011).
The tumor cellular microenvironment
contains supporting host cells, including
fibroblasts, adipocytes, smooth muscle
cells, endothelia, and immune cells, which
functionally promote tumor growth. In
two-compartment tumor metabolism, anabolic cancer cells extract energy from
the surrounding host cells by inducing
catabolic processes, such as autophagy,
mitophagy, and aerobic glycolysis. These
processes provide high-energy mitochondrial fuels (L-lactate, ketones, and gluta-
mine) for cancer cells to burn. In response,

cancer cells amplify or hyperactivate
their capacity for oxidative phosphorylation (OXPHOS) by increasing their mitochondrial mass (Sotgia et al., 2012). For
example, cancer-associated fibroblasts
show a shift toward aerobic glycolysis
and secrete L-lactate via MCT4 transporters. L-lactate is taken up by cancer
cells via MCT1 transporters, leading to
the generation of ATP via OXPHOS (Sotgia
et al., 2012). This process can be phenocopied by incubating cancer cells alone
with high-energy fuels, such as L-lactate.
Tumor cells can also exert metabolic
effects at a distance, which leads to increased fatty acid generation in adipose
tissue and catabolism in muscle (Das
et al., 2011). These key examples show
that energy transfer occurs in human
tumors and that cancer cells can exert
metabolic effects locally, in different tumor
compartments, and at distant sites.
Over 80% of ovarian cancers are
metastatic to the omental fat. It is not
known why ovarian cancer cells preferentially seed the omentum as compared to other sites. To address this issue,
the study by Nieman et al. (2011) uses
SKOV3ip1 human ovarian cancer cells
intraperitoneally (i.p.) injected into nude
mice or cocultured with adipocytes. They
describe how omental adipocytes are
metabolically reprogrammed to become
highly catabolic, generating free fatty
acids that are transferred to cancer cells.
Cancer cells then reutilize these fatty
acids to generate ATP via mitochondrial
b-oxidation. Utilization of adipocytederived fatty acids was related to the

production of fatty acid binding protein 4
(FABP4) by adipocytes. Importantly, this
study evaluates tumor metabolism in the
more physiological context of its proper
microenvironment.
Energy production and apoptosis are
important mitochondrial functions that
are biologically linked in normal cells.
For example, the mitochondrial proteins
Bcl-2 and Bcl-xL are antiapoptotic and
favor mitochondrial OXPHOS (Chen and
Pervaiz, 2010; Vander Heiden et al.,
2001). A mitochondrial paradox exists
in cancer research, since it is not understood why cancer cells, which are resistant to apoptosis, would use energetically
inefficient low mitochondrial metabolism
(aerobic glycolysis, also known as the
Warburg effect) (Le et al., 2010; Fogal
et al., 2010). Interestingly, Nieman et al.
(2011) demonstrate that ovarian cancer
cells have high mitochondrial metabolic
activity, specifically fatty acid b-oxidation,
when cocultured with adipocytes. This
type of mitochondrial metabolism was
not observed when ovarian cancer cells
were cultured alone, highlighting the importance of catabolite transfer to cancer
cells. As such, high-energy nutrients provided by host cells may bolster mitochondrial metabolism in cancer cells, protecting them against apoptosis. The answer
to the mitochondrial paradox may lie
in the metabolic reprogramming of cancer cells toward anabolic metabolism in
the presence of catabolic host cells,
Cell Metabolism
Previews
leading to mitochondrial biogenesis
and effectively treat advanced
Two-Compartment Tumor Metabolism
and OXPHOS in tumor cells, drivand metastatic cancers. New imaging chemoresistance and distant
ing techniques to visualize twoSupporting Host Cell
Cancer Cell
metastasis.
compartment tumor metabolism in
Catabolism
Anabolic Growth
Thus, the authors demonstrate
real time will allow us to measure
-Autophagy
that it is crucial to include the
the effectiveness of anticancer ther-Mitophagy
supporting microenvironment when
apies and facilitate more personal-Glycolysis
Energy
-Lipolysis
studying cancer cell metabolism
ized cancer treatments.
Mitochondrial
and that simply examining primary
Nutrients
Metabolism
cancer cells alone may not be
REFERENCES
-L-lactate
-OXPHOS
-Ketones
adequate. Unfortunately, most tradi-Beta-OX
-Glutamine
Chen, Z.X., and Pervaiz, S. (2010). Cell
Therapy
tional cancer metabolism studies
-Fatty Acids
Death Differ. 17, 408420.
have been carried out using tumor
Das, S.K., Eder, S., Schauer, S., Diwoky, C.,
cells alone, or using whole tumors,
Temmel, H., Guertl, B., Gorkiewicz, G.,
Figure 1. Two-Compartment Tumor Metabolism:
without modeling the host microenTamilarasan, K.P., Kumari, P., Trauner, M.,
Catabolic Host Cells Fuel Anabolic Cancer Cell
vironment. As such, one might gain
et al. (2011). Science 333, 233238.
Growth and Metastasis via Mitochondrial Metabolism
incomplete or inaccurate informaCatabolic host cells that make up the microenvironment (e.g.,
Fogal, V., Richardson, A.D., Karmali, P.P.,
fibroblasts and adipocytes) generate and transfer high-energy
tion by studying primary cancer cells
Scheffler, I.E., Smith, J.W., and Ruoslahti,
metabolites (L-lactate, ketones, glutamine, and free fatty
or cancer cell lines in the absence of
E. (2010). Mol. Cell. Biol. 30, 13031318.
acids) to epithelial cancer cells, energetically promoting tumor
supporting host cells.
growth and metastasis. Cancer cells increase their mitochonLe, A., Cooper, C.R., Gouw, A.M., Dinavahi,
drial mass and activity (OXPHOS and b-oxidation) to efficiently
Studies describing the compartR., Maitra, A., Deck, L.M., Royer, R.E., Vanburn
these
energy-rich
mitochondrial
fuels.
Targeted
therader Jagt, D.L., Semenza, G.L., and Dang,
mentalization of tumor metabolism
pies that metabolically uncouple parasitic cancer cells
C.V. (2010). Proc. Natl. Acad. Sci. USA
and energy transfer may pave the
from catabolic host cells (such as mitochondrial inhibitors
107, 20372042.
way toward the development of re[metformin and arsenic trioxide], as well as powerful antiMartinez-Outschoorn, U.E., Goldberg, A.,
oxidants) will starve tumor cells. Effective therapies would
lated predictive biomarkers and
Lin, Z., Ko, Y.H., Flomenberg, N., Wang,
block energy transfer, cutting off the fuel supply to cancer
targeted personalized therapies. It
C., Pavlides, S., Pestell, R.G., Howell, A.,
cells.
Sotgia, F., and Lisanti, M.P. (2011). Cancer
will be important to investigate
Biol. Ther. 12, 924938.
whether metabolically uncoupling
cancer cells from catabolic host cells IL-8 (Sotgia et al., 2012). Most importantly, Nieman, K.M., Kenny, H.A., Penicka, C.V.,
can be used as a new effective anticancer FDA-approved drugs that inhibit mito- Ladanyi, A., Buell-Gutbrod, R., Zillhardt, M.R.,
Romero, I.L., Carey, M.S., Mills, G.B., Hotamisligil,
strategy. Earlier studies have suggested chondrial metabolism (metformin, arsenic G.S., et al. (2011). Nat. Med. 17, 14981503.
that the detection of host-tumor meta- trioxide) or strong antioxidants (catalase)
bolic coupling may be useful for identi- can uncouple two-compartment tumor Sotgia, F., Martinez-Outschoorn, U.E., Pavlides,
S., Howell, A., Pestell, R.G., and Lisanti, M.P.
fying high-risk patients at diagnosis in metabolism and induce apoptosis in (2011). Breast Cancer Res. 13, 213.
human breast cancers. For example, cancer cells (Martinez-Outschoorn et al.,
Sotgia, F., Martinez-Outschoorn, U.E., Howell, A.,
loss of expression of the caveolin-1 pro- 2011; Sotgia et al., 2012) (Figure 1).
Pestell, R.G., Pavlides, S., and Lisanti, M.P.
tein in cancer-associated fibroblasts is
In conclusion, the importance of the (2012). Annu. Rev. Pathol. 7, 423467. Published
a marker for tumor-stroma metabolic host microenvironment and energy trans- online November 7, 2011. 10.1146/annurevpathol-011811-120856.
coupling (Sotgia et al., 2011) and is tightly fer in cancer metabolism is highlighted by
correlated with recurrence, metastasis, Nieman et al. (2011). More studies on two- Vander Heiden, M.G., Li, X.X., Gottleib, E., Hill,
and tamoxifen resistance as well as compartment tumor metabolism will be R.B., Thompson, C.B., and Colombini, M. (2001).
J. Biol. Chem. 276, 1941419419.
poor clinical outcome. Metabolic coupling necessary to understand and therabetween host cells and breast cancer peutically exploit the metabolic coupling Whitaker-Menezes, D., Martinez-Outschoorn,
cells also results in the generation of reac- between parasitic tumor cells and their U.E., Flomenberg, N., Birbe, R.C., Witkiewicz,
A.K., Howell, A., Pavlides, S., Tsirigos, A., Ertel,
tive oxygen species and inflammatory hosts. Uncoupling parasitic cancer A., Pestell, R.G., et al. (2011). Cell Cycle 10,
cytokine production, such as IL-6 and cells should allow us to starve cancer cells 40474064.
Cell Metabolism
Previews
Transgenerational Inheritance of Longevity:
Epigenetic Mysteries Abound
Shelley L. Berger1,2,3,*
1Department
of Cell and Developmental Biology

of Biology
3Department of Genetics
1057 BRB, 421 Curie Boulevard, University of Pennsylvania, Philadelphia, PA 19104, USA
*Correspondence: bergers@upenn.edu
DOI 10.1016/j.cmet.2011.12.012
2Department
Transgenerational inheritance of epigenetic characteristics in plants has been reported, whereas nongenetic
persistence of complex phenotypes in animals is controversial. A recent report by Anne Brunet and
colleagues describes a fascinating example of persistence across generations of extended life span in
worm and explores whether epigenetic mechanisms account for the longevity.
Major questions in the field of epigenetics

are whether chromatin states persist
through, first, mitotic cell division and,
second, meiotic germ cell generation, in
both cases to provide memory of the epigenome. The mitotic memory issue is
relevant to cell type differentiation during
development of multicellular organisms
and to loss of differentiation in human
disease states such as cancer, as well
as to tissue regenerative medicine and
the difficulties inherent in erasing or profoundly altering cell identity. The second
question, of meiotic memory through
germ cells to embryos, while similar in
principle, is more provocative, since this
could perpetuate nongenetic inheritance
across generations. A recent report in
Nature by Anne Brunet and colleagues
unveils a fascinating example of transgenerational inheritance, extending life
span in the worm C. elegans and potentially regulated by an epigenetic state
(Greer et al., 2011).
As background, it is important to understand certain theoretical and practical
considerations regarding epigenetic
memory. The key theoretical issue is the
mechanism(s) underlying cell recollection
of identity through cell division, as well
as through gametogenesis and early
embryogenesis, i.e., whether the molecules that transmit memory consist of
DNA-bound transcription factors, long
noncoding RNAs, bona fide chromatinregulatory enzymes, specialized histones,
or some combination thereof (Berger
et al., 2009). The practical consideration
is that many epigenetic regulators are
enzymes, enabling discovery of small
molecule modulators; such therapeutics

are already in the clinic or in aggressive
pharmaceutical development and thus
have huge potential impact on human
health (Rodrguez-Paredes and Esteller,
2011).
There is little doubt that epigenetic
memory exists through mitosis; this
phenomenon was initially recognized
through genetic analysis of development
in complex model organisms (Ringrose
and Paro, 2004). However, the question
of transgenerational memory via chromatin epigenetic states is far more controversial, since it challenges conventional
dogma about genetic inheritance. Indeed,
previous observations of transgenerational inheritance in animals have been
largely anecdotal or epidemiological (Daxinger and Whitelaw, 2010); the evolution of
this emerging field requires both molecular experimentation and manipulation.
A number of recent studies have found
a chromatin basis for setting life span,
including chromatin posttranslational
modifications such as reversible histone
acetylation. Sirtuins have long been
known to have a role in longevity, and although these enzymes have many cellular
substrates, histones indeed appear to be
key age-relevant acetylated substrates
in yeast S. cerevisiae and in mouse, functioning at telomeres (Dang et al., 2009;
Michishita et al., 2008). Further, histone
methylation also contributes to setting
life span, as shown in C. elegans, in that
deletion of the Set2 methylase enzyme
or other protein components of the H3
Lys4 (H3 K4) methylase complex extends
life span (Greer et al., 2010) (Figure 1A).
Thus, there appears to be chromatin

regulation of aging through organismal
life span involving relaxation of chromatin,
which may be deleterious to genomic
integrity and lead to aberrant gene expression. Hence, life span is extended
by reduction of certain histone modifications (Dang et al., 2009; Greer et al.,
2010) or via increased expression of
histones themselves (Feser et al., 2010).
Interestingly, Brunet and colleagues
now show that deletion of the same
worm histone H3 K4 methylase Set2
leads to transgenerational inheritance of
life span extension in wild-type offspring
(Greer et al., 2011). The authors designed
an experimental protocol to avoid maternal effects that might be transmitted
to offspringoften a question in transgenerational inheritance studies (Daxinger
and Whitelaw, 2010)by testing offspring
from the F3 to F5 generations (Figure 1B).
In brief, set-2-deficient mutants with
extended life span were mated to SET-2
wild-type worms to yield a heterozygous
F1 generation, which was then mated to
produce either genetically wild-type or
genetically mutant offspring. The authors
showed that, remarkably, wild-type offspring in generations F3 and F4 showed
the same extended life span as the
set2 / offspring (Figure 1B), thereby
demonstrating transgenerational inheritance of extended life span in worms.
Although this is a fascinating observation, whether the explanation is truly
due to altered chromatin remains to be
investigated, since the study does not
demonstrate a persistent chromatin
effect. That is, an obvious mechanism
Cell Metabolism
Previews
would be reduced H3 K4 methylagenerational longevity exist and
SET-2
A
tion in the F3 and F4 generations in
have not yet been tested.
H3K4me
spite of wild-type SET-2; however,
Thus, this report establishes a
the authors tested genome-wide
precedent that longevity can be
gene
methylation in the long-lived but
maintained
transgenerationally;
Aging
wild-type offspring, but found no
however, major challenges remain
lowering of H3 K4 methylation by
to show a direct epigenetic basis
global western analysis. This does
for the transgenerational inheritance.
not rule out possible localized reTaken together with other emerging
duction in methylation, for example,
examples in animals of transB
at specific genes that might regulate
generational inheritance of complex
set-2-/- X WT
Parents
longevity. To indirectly test this
phenotypes with possible underlying
Shorter LS
Longer LS
hypothesis, the authors performed
epigenetic mechanisms (Carone
genome-wide RNA expression miet al., 2010), we can anticipate
F1 (set-2+/-)
croarray analysis. Although the
many new studies exploring this
authors detected certain restricted
fascinating question in the near
F2
gene expression similarities of the
future.
F3 and F4 transgenerational offlonger
F3,F4 SET-2+/+
LS
spring to the actual mutant, the
REFERENCES
overall transcriptional picture unexH3K4me
shorter
pectedly showed expression clusF5
Berger, S.L., Kouzarides, T., Shiekhattar, R.,
LS
gene
tering more similar to the actual
and Shilatifard, A. (2009). Genes Dev. 23,
wild-type expression spectrum
781783.
Figure 1. Transgenerational Inheritance of Longevity
than to the actual mutant expression
(A) Aging may lead to decompaction of chromatin, resulting in
Carone, B.R., Fauquier, L., Habib, N., Shea,
spectrum. The authors also pointed
inappropriate gene expression. In the worm C. elegans, deleJ.M., Hart, C.E., Li, R., Bock, C., Li, C., Gu,
tion of the H3 K4 methylase, SET-2, extends life span.
out that certain specific GO cateH., Zamore, P.D., et al. (2010). Cell 143,
(B) Transgenerational inheritance of extended life span (LS) in
10841096.
gories that regulate metabolism are
the worm C. elegans results from parental deletion of SET-2
altered like the actual mutant; thus,
(set2 / ) and transmittal to wild-type F3 and F4 generations.
Dang, W., Steffen, K.K., Perry, R., Dorsey,
modest changes at a few genes
J.A., Johnson, F.B., Shilatifard, A., KaeberShorter life span is resumed in the wild-type F5 generation.
lein, M., Kennedy, B.K., and Berger, S.L.
may lead to persistent longevity in
(2009). Nature 459, 802807.
the F3 and F4 generations. To
address whether this is a direct or indirect mined. One possibility is that, since there Daxinger, L., and Whitelaw, E. (2010). Genome
effect, it will be important to explore is no transitional partial state, a threshold Res. 20, 16231628.
whether H3 K4 methylation is reduced at effect may exist, which would support Feser, J., Truong, D., Das, C., Carson, J.J., Kieft,
these genes in the wild-type offspring.
the idea that there are only a few genes J., Harkness, T., and Tyler, J.K. (2010). Mol. Cell
Another intriguing observation is that critical to life span extension, and, once 39, 724735.
the extended life span abruptly returns the methylation level increases to a certain Greer, E.L., Maures, T.J., Hauswirth, A.G., Green,
to normal length in the F5 generation level in the F5, the life span is no longer E.M., Leeman, D.S., Maro, G.S., Han, S., Banko,
M.R., Gozani, O., and Brunet, A. (2010). Nature
(Figure 1B) without passing through any extended.
466, 383387.
The study also reports that several
intermediate longevity state. The authors
E.L., Maures, T.J., Ucar, D., Hauswirth,
show an F5 RNA expression microarray other longevity genes, including deletion Greer,
A.G., Mancini, E., Lim, J.P., Benayoun, B.A., Shi,
with levels similar to the true SET-2 wild- of a few chromatin modulators of tran- Y., and Brunet, A. (2011). Nature 479, 365371.
type, but since the F3 and F4 transcription scription that extend life span (e.g., the
Michishita, E., McCord, R.A., Berber, E., Kioi, M.,
results are not clearly like set-2 mutant, H3 K9 methylase and H3 K27 demethy- Padilla-Nash, H., Damian, M., Cheung, P., Kusuthe interpretation of these findings re- lases) (Greer et al., 2010; Carone et al., moto, R., Kawahara, T.L., Barrett, J.C., et al.
mains ambiguous. Moreover, since no 2010), fail to show a transgenerational (2008). Nature 452, 492496.
altered chromatin state was detected in longevity effect. Thus, it is tempting to Ringrose, L., and Paro, R. (2004). Annu. Rev.
the F3 and F4 wild-type-but-extended speculate that K4 methylation plays a Genet. 38, 413443.
offspring, at present the chromatin state key role in transgenerational longevity or Rodrguez-Paredes, M., and Esteller, M. (2011).
in the F5 generation has yet to be deter- that other epigenetic mediators of trans- Nat. Med. 17, 330339.
Cell Metabolism
Previews
IL-6 Muscles In on the Gut and Pancreas
to Enhance Insulin Secretion
Tamara L. Allen,1 Martin Whitham,1 and Mark A. Febbraio1,*
1Cellular and Molecular Metabolism Laboratory, Baker IDI Heart and Diabetes Institute, Melbourne 3008, Victoria, Australia
*Correspondence: mark.febbraio@bakeridi.edu.au
DOI 10.1016/j.cmet.2011.12.004
The role of the cytokine interleukin-6 (IL-6) in metabolic homeostasis is the subject of conjecture. Recent
work in Nature Medicine (Ellingsgaard et al., 2011) demonstrates that IL-6 released from skeletal muscle
during exercise mediates crosstalk between insulin-sensitive tissues, intestinal L cells, and pancreatic islets
to adapt to changes in insulin demand.
The debate regarding the pathological
versus beneficial nature of IL-6 in metabolism remains unclear and the subject
of continuing debate (Mooney, 2007; Pedersen and Febbraio, 2007). On one hand,
increased IL-6 in obesity is associated
with the physiopathology of type 2 diabetes (Klover et al., 2003), while on the
other, muscle-derived IL-6 appears to
contribute to improved glycemia following
exercise (Pedersen and Febbraio, 2008).
A recent study published in Nature Medicine (Ellingsgaard et al., 2011) makes
a telling contribution to our understanding
of this apparent paradox. The authors
found that IL-6, released from either contracting skeletal muscle or white adipose
tissue, stimulated glucagon-like peptide-1
(GLP-1) from the gut and the pancreas.
As GLP-1 is a hormone that induces
insulin secretion, this complex organto-organ crosstalk provides strong evidence that IL-6 is a cytokine that has
positive effects on maintaining glucose
homeostasis.
The basic concept that IL-6 is an inflammatory cytokine that leads to insulin resistance was first challenged with the finding
that skeletal muscle releases IL-6 during
contraction (Steensberg et al., 2000) to
act in an endocrine-like manner (Febbraio
et al., 2004). Since this discovery, IL-6 has
been found to increase glucose uptake
and fat oxidation in skeletal muscle and
improve glucose tolerance and insulin
sensitivity (Carey et al., 2006; van Hall
et al., 2003), effects that oppose those
seen in the development of metabolic
syndrome. The recent paper by Donath
and colleagues (Ellingsgaard et al., 2011)
neatly demonstrates that IL-6 mediated
increases in GLP-1 secretion from L cells
in the intestine and a cells in the pancreas
that led to improved b cell function, insulin

secretion, and glycemic control. This
study is the first to provide evidence of
a previously unknown link between IL-6
secretion from insulin-sensitive tissues
and the beneficial effects of GLP-1 on
insulin action. By using IL-6 knockout
mice and coadministration of an IL-6 antibody, this recent paper (Ellingsgaard
et al., 2011) demonstrated that the exercise-induced increase in GLP-1 was IL-6
dependent. Furthermore, IL-6 was unable
to improve glucose tolerance in mice
lacking the GLP-1 receptor or mice
treated with the GLP-1 receptor antagonist, exendin (9-39). Interestingly, in addition to acute administration of IL-6, the
authors were also able to demonstrate
improved glycemia and glucose tolerance
following twice daily injections of IL-6 for
7 days. These effects were associated
with increases in synthesis and expression of circulating GLP-1. Significantly,
pancreatic GLP-1, insulin, and glucagon
content were also increased with IL-6
administration, along with insulin secretion from the islets of treated mice.
It is well known that exercise improves
insulin action in the immediate postexercise period (Wojtaszewski et al., 2000),
but until now the mechanism has been
unclear. Two important experiments in
the current study (Ellingsgaard et al.,
2011) provide new and important information on the phenomenon of enhanced
postexercise insulin action. First, IL-6deficient mice displayed no improvement
in glucose tolerance postexercise, and
second, the improvement was seen only
when the glucose tolerance test was
administered orally, not intraperitoneally,
indicative of the dependence on GLP-1.
This study supports the notion that
skeletal muscle is an endocrine organ,

capable of secreting metabolically active
proteins termed myokines (Pedersen
and Febbraio, 2008).
As the role of IL-6 in obesity and
metabolic disease is controversial, the
authors sought to investigate whether
additional administration of IL-6 could improve b cell function in overfed, insulinresistant, and diabetic animals. Exogenous IL-6 improved glucose tolerance
and insulin secretion in high-fat-fed
ob/ob and db/db mice. By way of support
for the proposed IL-6/GLP-1 b cell mechanism, high-fat-fed mice whose b cells
were destroyed with streptozotocin failed
to improve insulin secretion with administration of IL-6. Additionally, endogenous
IL-6 in db/db mice was blocked with
an IL-6 antibody treatment. Increased
fasting glucose levels and decreased glucose tolerance were reported in animals
treated with the IL-6 antibody, as well as
lowered glucagon levels and undetectable GLP-1 levels. GLP-1 content in the
pancreas of treated animals was also
lower than in control mice. Importantly,
analysis of high-fat-fed mice found that
IL-6 was required for the high-fat-dietinduced expansion of a cell mass and
increases in pancreatic GLP-1 content
and subsequent insulin secretion.
This paper highlights the importance of
organ crosstalk and may offer an additional therapeutic target for the modulation of glucose metabolism in obesity
and diabetes. However, these findings
still need to be interpreted with caution.
There are still many instances in the literature that suggest that IL-6 may be detrimental to maintaining metabolic homeostasis in obesity. While the study by
Ellingsgaard et al. (2011) adds weight to
Cell Metabolism
Previews
REFERENCES
the argument that IL-6 may be
elevated in obesity in order to
Carey, A.L., Steinberg, G.R.,
counteract other more delMacaulay, S.L., Thomas, W.G.,
eterious factors associated
Holmes, A.G., Ramm, G., Prelovsek,
O., Hohnen-Behrens, C., Watt, M.J.,
with obesity, the pleiotropic
James, D.E., et al. (2006). Diabetes
nature of IL-6 makes thera55, 26882697.
peutic strategies difficult.
Ellingsgaard, H., Hauselmann, I.,
In summary, the work by
Schuler, B., Habib, A.M., Baggio,
Donath and colleagues has
L.L., Meier, D.T., Eppler, E., Bouzakri,
K., Wueest, S., Muller, Y.D., et al.
provided crucial data that
(2011). Nat. Med. 17, 14811489.
adipose tissue-derived IL-6
in obesity and diabetes as
Febbraio, M.A., Hiscock, N., Sacchetti, M., Fischer, C.P., and Pederwell as skeletal muscle IL-6
sen, B.K. (2004). Diabetes 53, 1643
during exercise mediate the
1648.
production and secretion of
Klover, P.J., Zimmers, T.A., KoniaGLP-1 from the intestine and
ris, L.G., and Mooney, R.A. (2003).
pancreas, leading to an enDiabetes 52, 27842789.
hanced insulin response and
Mooney, R.A. (2007). J. Appl.
improved glycemia (see FigPhysiol. 102, 816818, discussion
ure 1). This study provides
818819.
further evidence of the imporPedersen, B.K., and Febbraio, M.A.
tance of IL-6 in glucose
(2007). J. Appl. Physiol. 102,
metabolism and uncovers its
814816.
previously unknown role in
Figure 1. The Pleiotropic Metabolic Effects of the Cytokine IL-6
Pedersen, B.K., and Febbraio, M.A.
IL-6 is released from contracting skeletal muscle and adipose tissue to trigger
linking adipose tissue, skel(2008). Physiol. Rev. 88, 13791406.
GLP-1 release from the gut and the pancreas to improve glucose tolerance.
etal muscle, intestines, and
Steensberg, A., van Hall, G., Osada,
pancreas through GLP-1.
T., Sacchetti, M., Saltin, B., and
Moreover, the study strengthens the secretory factors that may modulate Klarlund Pedersen, B. (2000). J. Physiol. 529,
237242.
concept that the skeletal muscle is an metabolic processes.
van Hall, G., Steensberg, A., Sacchetti, M., Fischer,
endocrine organ, capable of secreting
C., Keller, C., Schjerling, P., Hiscock, N., Mller, K.,
factors that can affect not only the
Saltin, B., Febbraio, M.A., and Pedersen, B.K.
adipose tissue and the liver, but also the ACKNOWLEDGMENTS
(2003). J. Clin. Endocrinol. Metab. 88, 30053010.
gut, pancreas, and perhaps many other
M.A.F. is a Senior Principal Research Fellow of the
Wojtaszewski, J.F., Hansen, B.F., Gade, Kiens, B.,
organs. The work opens the door to the National Health and Medical Research Council of Markuns, J.F., Goodyear, L.J., and Richter, E.A.
(2000). Diabetes 49, 325331.
identification of other skeletal muscle Australia.
Cell Metabolism
Review
The Inflammasome Puts Obesity in the Danger Zone
Rinke Stienstra,1,2,3 Cees J. Tack,1 Thirumala-Devi Kanneganti,4 Leo A.B. Joosten,1,2 and Mihai G. Netea1,2,*
1Department
of Medicine
Institute for Infection, Inflammation and Immunity (N4I)
Radboud University Nijmegen Medical Centre, Nijmegen 6525 GA, The Netherlands
3Nutrition, Metabolism and Genomics Group, Wageningen University, Wageningen 6703 HD, The Netherlands
4Department of Immunology, St. Jude Childrens Research Hospital, Memphis, TN 38105, USA
*Correspondence: m.netea@aig.umcn.nl
DOI 10.1016/j.cmet.2011.10.011
2Nijmegen
Obesity-induced inflammation is an important contributor to the induction of insulin resistance. Recently, the
cytokine interleukin-1b (IL-1b) has emerged as a prominent instigator of the proinflammatory response in
obesity. Several studies over the last year have subsequently deciphered the molecular mechanisms responsible for IL-1b activation in adipose tissue, liver, and macrophages and demonstrated a central role of the
processing enzyme caspase-1 and of the protein complex leading to its activation called the inflammasome.
These data suggest that activation of the inflammasome represents a crucial step in the road from obesity to
insulin resistance and type 2 diabetes.
Introduction
Accumulating evidence links inflammation to metabolic changes
associated with obesity and type 2 diabetes (Donath and Shoelson, 2011). While several studies suggest that expanding
adipose tissue is an important first step in driving the enhanced
state of inflammation, the underlying molecular mechanisms
modulating this process are less clear. A wide variety of immune
cells, including macrophages, monocytes, T cells, and b cells,
have been shown to infiltrate the adipose tissue and affect its
homeostasis by releasing inflammatory cytokines (Anderson
et al., 2010). Adipocytes themselves are also capable of
releasing inflammatory mediators and contribute to the inflammatory response (McGillicuddy et al., 2011; Stienstra et al.,
2010; Meijer et al., 2011). In addition to adipose tissue, inflammation in liver and pancreatic islets is also evident in obese individuals and participates in the pathogenesis of type 2 diabetes (Gregor and Hotamisligil, 2011). One of the proinflammatory
cytokines mediating obesity-induced inflammation is interleukin
(IL)-1b, which is processed by caspase-1, a cysteine protease
regulated by a protein complex called the inflammasome.
Although growing evidence points to a crucial role for IL-1b in
mediating the development of insulin resistance, it should be
stressed that the inflammatory response driving the development of insulin resistance probably is comprised of a combination of proinflammatory cytokines that jointly effectuate type 2
diabetes progression. For example, involvement of TNFa in
obesity-associated insulin resistance has been frequently reported. Since biological processes are often multifactorial,
involvement of other cytokines like TNFa is plausible.
Although detrimental effects of IL-1b on b cell function have
been well documented, the proinflammatory effects of IL-1b that
mediate the development of tissue dysfunction and peripheral
insulin resistance have only recently received more interest. While
several lines of evidence have shown involvement of IL-1b in the
development of obesity-associated insulin resistance peripherally,
the quantitative contribution remains to be defined in more detail.
In the present review we will discuss recently identified metabolic triggers that may function as potential danger signals,
promoting activation of inflammasome-dependent caspase-1,

and highlight new findings regarding the mechanisms involved
in the processing of IL-1b during the progression of obesity
and insulin resistance. In light of the growing interest to block
low-grade inflammation in obese and insulin-resistant subjects,
this review will particularly focus on the potential of the inflammasome as a therapeutic target in the treatment of obesity-induced
insulin resistance.
Pathogenic Role of IL-1b in the Development of Obesity,
Insulin Resistance, and Diabetes
IL-1b elicits potent proinflammatory actions through its binding
to the IL-1 receptor that in turn recruits the IL-1 receptor accessory protein. This receptor complex signals through the myeloid
differentiation factor 88 (MyD88) adaptor protein that spurs on
IL-1 receptor-activated kinases (IRAK1 to 4). This leads to activation of several protein kinases, including mitogen-activated
protein kinase 8 (JNK), mitogen-activated protein kinase 1
(ERK), p38 MAPK, and inhibitor of kappaB kinase (IKK), that
initiate the transcription factors nuclear factor kB (NF-kB) and
activator protein 1 (AP1) to stimulate inflammatory gene expression (Dinarello, 2011) (see Figure 1). The proinflammatory actions
of IL-1b are recognized as an important contributor to the development of both type 1 and type 2 diabetes (Donath and Shoelson, 2011; Mandrup-Poulsen et al., 2010). While IL-1b has toxic
effects on pancreatic b cells in the process of autoimmune diabetes (Mandrup-Poulsen et al., 1986), the cytokine also appears
to be involved in b cell deterioration related to glucotoxicity in
type 2 diabetes (Donath et al., 2003), with both processes
leading to defective insulin production. By activation of Fas-triggered apoptosis, which involves the transcription factor NF-kB,
IL-1b mediates b cell dysfunction. More peripherally, IL-1b
directly inhibits insulin signaling pathways by reducing tyrosine
phosphorylation of insulin receptor substrate 1 (IRS1) and negative regulation of IRS1 gene expression levels, thus inducing
a state of insulin resistance (Jager et al., 2007). In animal studies,
the lack of IL-1b or its receptor protects against the development
of adipose tissue inflammation and insulin resistance upon
Cell Metabolism
Review
Figure 1. IL-1b Signaling Pathway: Overview of the Intracellular Signaling Pathways Activated by Binding of IL-1b to the IL-1 Receptor
Upon activation of the IL-1 receptor complex, IL-1b induces recruitment of the myeloid differentiation primary response gene 88 (MyD88), which in turn promotes
activation of the interleukin-1 receptor kinase (IRAK) cascade. Via tumor necrosis factor-associated factor 6 (TRAF6) and the c-Jun N terminus (JNK), p38
mitogen-activated protein (p38 MAPK), and the inhibitor of nuclear factor b (IKK) kinases, the IkB cofactor is degraded, which subsequently promotes the nuclear
translocation of NF-kB and AP-1. Both transcription factors have the capacity to induce proinflammatory gene expression of various cytokines and chemokines
that modulate the inflammatory response.
high-fat diet feeding (Wen et al., 2011; McGillicuddy et al., 2011).

This protective effect may partially be mediated by the absence
of IL-1b-controlled chemokine synthesis (Dinarello, 2011), which
governs the influx of various immune cells into adipose tissue,
thus promoting inflammation.
In line with the proposed pathogenic role of IL-1b, increased
circulating levels of this cytokine accompanied by elevated
levels of IL-6 positively correlate with the development of type
2 diabetes in humans (Spranger et al., 2003). Conversely,
blockade of excessive IL-1 signaling in subjects with type 2 diabetes improves glycemic control and b cell function while
reducing markers of systemic inflammation (Larsen et al.,
2007). Although clinical studies have revealed that inhibition of
IL-1 signaling improves glucose tolerance, data pointing to an
improvement of insulin sensitivity upon anti-IL-1 treatment are
scarce. However, clinical studies using salsalate, an unspecific
anti-inflammatory agent, have demonstrated a significant
improvement in insulin sensitivity (Goldfine et al., 2008). It should
be noted that highly targeted anticytokine treatment approaches, such as treatment with the anti-TNF antibody (infliximab), have been found to be successful in improving insulin
sensitivity in patients diagnosed with rheumatoid arthritis (Gonzalez-Gay et al., 2010; Huvers et al., 2007).
Until recently, the triggers and molecular switches controlling
IL-1b production during the development of obesity and insulin
resistance have remained largely unknown. In as much as
IL-1b elicits a vigorous inflammatory response, activation must
be tightly controlled and requires processing from an inactive
procytokine into the biologically active form by proteolytic
enzymes. Processing of cytokines of the IL-1 family, such as
IL-1b and IL-18, is mainly mediated by the cysteine protease
caspase-1, which in turn is activated by a protein platform
termed the inflammasome.
The Inflammasome
The inflammasome is an important part of our innate immune
system that responds to danger signals that are sensed by
a number of different intracellular NOD-like receptors (NLRs).
Different inflammasomes have been identified, including NLR
family pyrin domain-containing 1 (NLRP1), NLR family pyrin
domain-containing 3 (NLRP3), NLR family pyrin domain-containing 6 (NLRP6), AIM2 (absent in melanoma 2), and IPAF
Cell Metabolism
Review
(IL-1b-converting enzyme protease-activating factor), which
each have the ability to respond to a wide variety of microbial
products or endogenous danger signals (Dunne, 2011). Hitherto,
NLRP3 is the most extensively studied inflammasome that, upon
its activation, forms a complex with its adaptor molecule PYD
and CARD domain containing protein (ASC) and thereby facilitates caspase-1-dependent processing of pro-IL-1b into its
active form. Normally, this proinflammatory response provides
protection for the host by inducing an acute phase response
(Dinarello, 2011).
Recent studies have identified the inflammasome as an important contributor to the development of insulin resistance by
mediation of processing and release of IL-1b in various tissues
and cell types during the development of obesity. In addition to
IL-1b, caspase-1 is also able to process and activate IL-18 and
IL-33 (Arend et al., 2008). In contrast to IL-1b, IL-18 ameliorates
development of obesity and insulin resistance (Netea et al.,
2006). Interestingly, caspase-1 deficient mice lacking both
IL-1b and IL-18 are characterized by an improvement in insulin
resistance (Stienstra et al., 2010), suggesting that the insulindesensitizing effects of IL-1b override IL-18 action. The fact
that expression and circulating levels of IL-18 are easily detectable in healthy subjects, as compared to IL-1b, suggests that this
cytokine executes different functions, among which are the
opposing effects on insulin resistance.
NLRP3-Mediated Caspase-1 Activity in the Drivers Seat
Several lines of evidence suggest that activation of inflammasome-mediated caspase-1 is one of the culprits behind the
enhanced inflammatory state characteristic of obesity and has
center stage in the pathogenesis of type 2 diabetes by acting
at two different levels.
First, caspase-1 appears to instigate defective insulin secretion by promoting pancreatic dysfunction. Pancreatic b cells itself are capable of producing IL-1b (Boni-Schnetzler et al.,
2008; Maedler et al., 2002) through mechanisms involving the
NLRP3 inflammasome (Zhou et al., 2010). Additionally,
enhanced macrophages infiltration of the pancreas observed
in human type 2 diabetic patients and high-fat diet (HFD)-fed
mice (Ehses et al., 2007) may further potentiate IL-1b production.
The IL-1b-driven inflammation of the islets is proposed as the
central mediator of glucose-, lipid-, and amyloid-induced b cell
failure leading to defective insulin secretion (Masters et al.,
2011; Mandrup-Poulsen, 2010) and ultimately b cell death, two
processes distinctive for the pathogenesis of type 2 diabetes.
Second, activation of caspase-1 can alter the function of
peripheral tissues, including adipose tissue and liver, both critically involved in maintaining glucose homeostasis. Adipose
tissue of animals fed a high-fat diet to induce obesity and insulin
resistance is characterized by enhanced gene expression and
increased protein levels of caspase-1 (Stienstra et al., 2010; Vandanmagsar et al., 2011). Similarly, feeding mice a methioninecholine-deficient diet, which promotes the development of
nonalcoholic steatohepatitis (NASH), has been shown to
promote NLRP3-dependent caspase-1 activation within hepatocytes (Csak et al., 2011). The elevated activity of caspase-1 leads
to increased processing of IL-1b and promotes a proinflammatory environment that drives tissue dysfunction. Indeed, the
absence of caspase-1 provides protection for the host against
the deleterious effects of high-fat diet feeding. For example,

caspase-1 / mice have a decreased influx of macrophages
into adipose tissue upon high-fat diet feeding (Stienstra
et al., 2011). Similarly, ASC / (Stienstra et al., 2011) and
NLRP3 / animals (Vandanmagsar et al., 2011) were protected
against the development of hepatosteatosis instigated by highfat diet feeding. Importantly, HFD-fed mice lacking caspase-1
were characterized by a robust improvement in insulin sensitivity
and were rescued from the development of obesity as compared
to wild-type mice (Stienstra et al., 2011). Notably, part of the
protective effects of the absence of caspase-1 may be accomplished by the reduction in weight gain as compared to the
wild-type animals fed the high-fat diet. A similar protection
against insulin resistance was observed in HFD-fed animals
lacking ASC or NLRP3 (Wen et al., 2011). However, different
phenotypical responses were observed between the various inflammasome knockout models upon high-fat diet feeding.
Whereas the absence of NLRP3 only provided protection against
high-fat diet-induced insulin resistance and fatty liver disease
after prolonged exposure to the diet for up to 9 months in one
model (Vandanmagsar et al., 2011), shorter periods of high-fat
diet feeding also appear to activate NLRP3 (Wen et al., 2011).
The absence of caspase-1 and ASC conferred protection
against the development of insulin resistance after 12 or
16 weeks of diet intervention. These results imply that NLRP3
activation may require secondary danger signals that are only
apparent after prolonged high-fat diet feeding and may include
lipotoxic metabolites like sphingolipids. Hypothetically, differences in the composition of the diets that were used to induce
obesity may have altered the supply of NLRP3 danger signals.
Alternatively, other inflammasomes next to NLRP3 may also be
involved in the development of obesity-induced insulin resistance. Furthermore, differences in expression levels between
distinct adipose tissue cell populations may explain any phenotypical variance. While caspase-1 is highly expressed in both
adipocytes and nonadipoycte cells, the inflammasome
members NLRP3 and ASC are predominantly expressed in nonadipocyte cells that are part of the adipose tissue. Although the
lower expression levels do not rule out any function of ASC and
NLRP3 in adipocytes, it appears that crosstalk between both cell
types determines the secretion levels of IL-1b by adipose tissue.
However, it should be emphasized that isolated adipocytes do
have the potential to produce IL-1b (Koenen et al., 2011b).
Hence, caspase-1 activation in adipocytes may rely on nonNLRP3 inflammasome members or is controlled independently
of the inflammasome.
Additionally, important tissue-specific effects of various inflammasomes have recently been described, such as the regulation of IL-18 production in intestinal epithelial cells by the NLRP6
inflammasome (Elinav et al., 2011). This opens up the intriguing
and yet unexplored possibility of specific inflammasome components present in adipocytes, hepatocytes, macrophages, or
pancreatic b cells that may differentially affect the development
of insulin resistance. Alternatively, differential regulation of
similar inflammasome subtypes may also occur and mediate
tissue-specific effects.
There is supportive evidence for the involvement of the inflammasome in obesity-induced inflammation in humans. For
example, caspase-1 activity levels are more pronounced in the
Cell Metabolism
Review
visceral fat depot, which is a strong determinant of insulin resistance, as compared to subcutaneous adipose tissue of mildly
obese subjects (Koenen et al., 2011b). In liver, expression levels
of NLRP3, caspase-1, and ASC are elevated in patients suffering
from NASH compared to healthy controls (Csak et al., 2011).
Conversely, substantial weight loss in obese subjects with type
2 diabetes was shown to reduce expression levels of IL-1b and
NLRP3 in adipose tissue and positively correlated with changes
in fasting plasma glucose levels (Vandanmagsar et al., 2011).
Overall, these results argue that activation of IL-1b by the inflammasome plays a central role in the pathogenesis of
obesity-induced insulin resistance. Since activation of caspase-1 leading to the cleavage and secretion of IL-1b is under
tight control of the inflammasome, specific danger signals
should exist that arise during the development of obesity and
that are sensed by intracellular NLRs. Indeed, several studies
have identified metabolic danger signals that can efficiently activate the inflammasome, leading to caspase-1-driven cleavage of
IL-1b in a variety of cell types, both from nonmyeloid and myeloid
origin (Vandanmagsar et al., 2011; Wen et al., 2011; Csak et al.,
2011; Zhou et al., 2010; Masters et al., 2010).
Potential Metabolic Danger Signals that Activate IL-1b
Recent studies have revealed that activation of IL-1b requires
two signals (Gong et al., 2010; Joosten et al., 2010; Hornung
et al., 2009). Whereas the first signal primes the cell and acts
as an inducer of IL-1b and NLR mRNA transcription, the second
signal induces conformational changes in the inflammasome
that instigates caspase-1 activation. Classically, the first signal
is provided by invading pathogens driving IL-1b mRNA transcription through pattern recognition receptors such as toll-like
receptors (TLRs) (Lamkanfi et al., 2008). The second signal is
of a more heterogeneous nature, as it can be represented by
microbial components such as microbial DNA (Muruve et al.,
2008), cell-wall components, and toxins (Ali et al., 2011), but
also by endogenous ligands such as adenosine triphosphate
(ATP) or uric acid (Mariathasan et al., 2006). The precise unifying
mechanism through which all these ligands induce caspase-1
activation is not known, although a rapid K+ efflux from cells
seems to be a common denominator that triggers activation of
the inflammasome (Lamkanfi et al., 2008). Notably, not all cell
types need sequential hits to induce IL-1b production. For
example, monocytes only require the first signal to induce inflammasome-dependent IL-1b release, due to the constitutive
activation of caspase-1 in this cell type (Netea et al., 2009).
Moreover, IL-1b can also be processed by caspase-1-independent cleavage through neutrophil-derived serine proteases
(Joosten et al., 2009), yet no information is available concerning
the mechanism that controls this process.
Very recently, fatty acids, glucose, uric acid, and islet amyloid
polypeptide (IAPP) have been put forward as metabolic danger
signals that possess the capacity to activate the inflammasome
and stimulate IL-1b production (Figure 2). Below we will discuss
where exactly these signals might act to promote IL-1b release.
Fatty Acids
Elevated circulating levels of free fatty acids are one of the hallmarks of type 2 diabetes (Krebs and Roden, 2005). Interestingly,
it has been suggested that saturated fatty acids bind and activate members of the TLR-family in vitro (Lee et al., 2001). For
example, palmitate acts as a ligand for TLR4 and induces IL-6

mRNA expression through activation of the transcription factor
NF-kB (Shi et al., 2006). Moreover, treatment of human THP-1
cells with palmitate has been shown to induce IL-1b mRNA
expression (Haversen et al., 2009) although some doubt has
been cast on whether saturated fatty acids can induce TLR
signaling (Erridge and Samani, 2009). However, elevated levels
of palmitate may provide the first signal needed for IL-1b production though induction of IL-1b gene transcription, although it
has to be kept in mind that levels of all fatty acids are elevated
in type 2 diabetes and may alter the effect of palmitate on
IL-1b production.
In addition, recent work demonstrates that palmitate also has
the potential to directly activate the NLRP3-inflammasome in
macrophages and thus provide the second signal needed for
IL-1b release (Wen et al., 2011). Via detailed mechanistic
in vitro studies, it was shown that palmitate negatively affects
the activation of AMP-activated protein kinase (AMPK), an
energy-sensing kinase that regulates multiple biochemical pathways in all eukaryotes, including lipid and glucose metabolism
and apoptosis (Steinberg and Kemp, 2009). The reduction in
AMPK phosphorylation leads to defective autophagy, a process
known to regulate the clearance of dysfunctional mitochondria.
The authors propose that the subsequent intracellular accumulation of mitochondrial-derived reactive oxygen species (ROS)
represents the trigger activating the inflammasome. Indeed,
ROS have long been proposed to activate the inflammasome,
although some controversy exists. For example, cells isolated
from patients diagnosed with chronic granulomatous disease
(CGD) have defective NADPH activity and thus cannot generate
NADPH-dependent ROS (van de Veerdonk et al., 2010; Meissner
et al., 2010). Upon stimulation, monocytes from CGD patients
produced similar or even increased amounts of IL-1b as
compared to cells from unaffected subjects, indicating
NADPH-independent activation of the inflammasome. In this
respect, enhanced production of ROS by mitochondria, a hallmark of insufficient mitochondrial function that is strongly associated with type 2 diabetes (Jin and Patti, 2009), may be an alternative source driving NLRP3 inflammasome activation
(Tschopp, 2011).
Other pathways that mediate palmitate-induced inflammasome activation can also be envisaged. Exposure of cells to
palmitate induces intracellular accumulation of ceramide, which
is produced by a ubiquitous biosynthetic pathway initiated by the
condensation of palmitoyl-coenzyme A (CoA) and serine.
Prevention of de novo ceramide synthesis limited the development of insulin resistance evoked by treatment of cells with saturated fatty acids (Chavez et al., 2003). Given the fact that ceramide has recently been identified as a potent activator of the
inflammasome in macrophages (Vandanmagsar et al., 2011),
palmitate may fuel ceramide biosynthesis, thereby providing
a continuous supply of danger signals for NLRP3 inflammasome
activation.
The palmitate-driven activation of the inflammasome is not
limited to macrophages, as nonmyeloid cell types including
hepatocytes have recently been shown to be a target of palmitate-driven activation of the NLRP3-inflammasome as well.
Though palmitic acid has been nominated as the metabolic activator that links obesity-induced insulin resistance to
Cell Metabolism
Review
Figure 2. Activation of the Inflammasome

An overview of the metabolic triggers that have been shown to provide the first signal (induction of transcription) or the second signal (inflammasome activation)
leading to the production and release of IL-1b. The first signal that is aimed at inducing IL-1b transcription and subsequent production of pro-IL-1b can be
provided by glucose, palmitate, uric acid, or LPS. In order to facilitate processing of pro-IL-1b into its active form, a second signal is essential to facilitate inflammasome activation and caspase-1-dependent cleavage of pro-IL-1b. NLRP3-dependent activation of caspase-1 can be triggered by palmitate, ceramide,
glucose, uric acid, reactive oxygen species (ROS), or amyloid. Promotion of this cascade occurs in a variety of tissues, including the pancreas, liver, and adipose
tissue and may subsequently contribute to tissue dysfunction and the development of insulin resistance.
activation of innate immunity, this may not imply that all fatty acids
cause activation of the inflammasome. Since unsaturated fatty
acids have anti-inflammatory effects (Browning, 2003) and lack
the ability to activate TLR4 or evoke ceramide synthesis (Chavez
et al., 2003), one would predict that highly unsaturated species
prevent inflammasome activation. Accordingly, it is important to
characterize the inflammasome-activating potential of various
fatty acids differing in length and degree of unsaturation.
A potential confounder in most of the experiments pinpointing
ceramide or palmitate as activators of the inflammasome (and
other metabolic danger signals) may be the fact that cells were
primed with LPS (Wen et al., 2011; Vandanmagsar et al.,
2011). Notably, both palmitate and ceramide did not induce inflammasome activation and subsequent cleavage and secretion
of IL-1b without prior exposure of macrophages to LPS. Since it
has been reported that serum endotoxin levels are elevated in
obese animals due to a disruption in intestinal integrity (Cani
et al., 2008), LPS may serve as a first signal in obese individuals.
However, other priming signals may exist in vivo that synergize
with ceramide or palmitate to induce both proIL-1b transcription
and inflammasome activation, and these include other fatty
acids, glucose, or adipocytokines.
Glucose
It has been known for almost a decade that high glucose levels
promote IL-1b mRNA transcription (Shanmugam et al., 2003).
Glucose has been shown to activate PKC-alpha and, via phosphorylation of p38 MAPK and ERK1/2, lead to NF-kB activation
and subsequent IL-1b transcription in monocytes (Dasu et al.,
2007), hence priming cells for inflammasome activation.
Additional pathways may also participate in high glucosemediated activation of IL-1b release by providing the second
signal that is required to activate the inflammasome. Indeed,
glucose has been identified as a direct stimulator of caspase-1
activity (Vincent and Mohr, 2007). Recent evidence has suggested that the NLRP3 inflammasome is activated in response
to high levels of glucose (Zhou et al., 2010). The mechanism of
action involves thioredoxin-interacting protein (TXNIP), a protein
that acts as an endogenous inhibitor of the ROS scavenging
protein thioredoxin (Minn et al., 2005; Parikh et al., 2007). Interestingly, TXNIP levels are elevated in subjects with type 2 diabetes mellitus (Parikh et al., 2007), and its expression is robustly
induced by glucose (Stoltzman et al., 2008). Zhou et al. (2010)
have shown that, upon activation, TXNIP is able to directly
interact with NLRP3 in a ROS-dependent manner, leading to
Cell Metabolism
Review
activation of caspase-1 and processing of IL-1b in pancreatic
islets. However, these observations could not be reproduced
in bone-marrow-derived macrophages lacking TXNIP and
exposed to high glucose levels (Masters et al., 2010). Nevertheless, treatment of the cells with 2-deoxy-D-glucose, which
blocks metabolic processing, prevents IL-1b processing and
release, implying that adequate glucose metabolism is indispensable for inflammasome function (Masters et al., 2010).
It is reasonable to expect that the effects of high levels of
glucose on caspase-1-mediated IL-1b production differ among
various cell types. Indeed, although adipose tissue ex vivo
responds to high levels of glucose by inducing TXNIP and
producing higher levels of IL-1b, TXNIP ablation in adipocytes
did not affect caspase-1 activity levels (Koenen et al., 2011a).
However, knockdown of TXNIP did reduce IL-1b mRNA levels
within adipocytes, indicating that TXNIP regulates IL-1b gene
expression levels during the presence of high levels of glucose.
As TXNIP itself has the capacity to induce chromatin modification (Perrone et al., 2009), it is conceivable that TXNIP may regulate IL-1b promoter availability.
In summary, high glucose levels may provide the priming
signal for transcription of IL-1b by activation of TXNIP, which
subsequently facilitates enhanced IL-1b expression levels. Additionally, high concentrations of glucose also promote the generation of ROS that is sufficient as second signal that promotes inflammasome activation and subsequently induces release of
bioactive IL-1b.
Uric Acid
Hyperuricemia often occurs in obese individuals and frequently
precedes the development of type 2 diabetes. It has been thought
that high levels of uric acid are a result of hyperinsulinemia, since
insulin normally reduces the renal secretion of uric acid. Interestingly, lowering of circulating levels of uric acid improved insulin
sensitivity in fructose-fed rats (Nakagawa et al., 2006).
It has been suggested that uric acid drives inflammatory and
oxidative changes in adipocytes. Uric acid directly affects the
inflammatory status of the adipose tissue by regulating MCP-1
levels in adipocytes (Baldwin et al., 2011), thereby promoting
infiltration of macrophages (Ito et al., 2008). Interestingly, crystals of monosodium urate, frequently observed in patients diagnosed with gout, have been uncovered as a robust inducer of the
NLRP3 inflammasome in macrophages (Martinon et al.,
2006).Can uric acid also drive obesity-induced inflammation by
activating the inflammasome?
Interestingly, uric acid is indispensable for normal adipogenesis, as animals that lack xanthine oxidoreductase (XOR), an
enzyme that generates uric acid from xanthine, exhibit a
decrease in adipose content and are characterized by a reduction in PPARg activity as compared to wild-type mice. More
importantly, uric acid levels within adipose tissue are enhanced
in obese animals as compared to lean mice, and this is accompanied by elevated transcription levels of XOR (Cheung et al.,
2007). Thus, it may be hypothesized that uric acid can directly
promote the inflammasome or modulate IL-1b mRNA transcription. In theory, adipocyte death, which has been suggested to be
the driving force of adipose tissue inflammation by attracting
macrophages (Cinti et al., 2005), may also activate the inflammasome, since dying cells are known to release uric acid to alert the
immune system (Shi et al., 2003).
Although more work is needed, uric acid is an important candidate that may modulate inflammasome activation during the
development of obesity and insulin resistance.
Islet Amyloid Polypeptide
Islet amyloid polypeptide (IAPP), also known as amylin, is a regulatory peptide of 37 amino acids that is produced by pancreatic
b cells and inhibits insulin and glucagon secretion. The ability of
amylin to form protein aggregates that cluster in pancreatic islets
as amyloid deposits is believed to be of critical importance for
the loss of b cell mass and type 2 diabetes progression (Westermark et al., 2011). Recent evidence has uncovered that IAPP has
the ability to energize the NLRP3 inflammasome, subsequently
promoting IL-1b production by the pancreas (Masters et al.,
2010). In contrast to glucose and palmitate, IAPP lacks the ability
to drive IL-1b mRNA expression, yet robustly promotes NLRP3mediated caspase-1 activation, partly by a disturbance in the
phagocytic machinery causing oxidative stress.
Interestingly, expression of IAPP by the pancreatic b cell line
MIN6 cells is activated by palmitate (Westermark et al., 2011),
suggesting that fatty acids may boost pancreatic IL-1b production in type 2 diabetes.
Additional lines of evidence suggest that IAPP may also have
extrapancreatic sites of action, since expression of the peptide
has been detected in the gastrointestinal tract and sensory
neurons (Westermark et al., 2011). Moreover, inasmuch as circulating levels of IAPP are elevated upon high-fat diet feeding of
animals (Westermark et al., 2011), it is plausible to expect that
IAAP-dependent activation of the inflammasome also occurs at
extrapancreatic sites.
Altogether, these data suggest that IAPP functions as a second
signal promoting NLRP3 activation and subsequent caspase-1dependent cleavage of IL-1b.
Therapeutic Interventions
Given the wealth of data that have implicated activation of the inflammasome in the pathogenesis of obesity-induced insulin
resistance, inhibition of IL-1b and caspase-1 may represent an
attractive therapeutic target. So far, only a few studies have
been conducted aimed at limiting IL-1b action, either by using
receptor antagonists or by neutralizing IL-1b. Blockade of
IL-1R-mediated signaling by the recombinant IL-1 receptor
antagonist (IL-1Ra) Anakinra was effective in improving glycemic
control in patients with type 2 diabetes (Larsen et al., 2007).
However, no improvement in insulin sensitivity levels as determined by euglycemic clamp was observed. In addition, no
changes in muscle gene expression or serum adipokine levels
were seen. Additionally, treatment with Anakinra had no direct
beneficial effect on insulin sensitivity in nondiabetic insulin-resistant subjects (van Asseldonk et al., 2011). Although these observations suggest that IL-1 blockade in human subjects does not
reverse insulin resistance, only a limited number of studies
have been conducted so far. Additionally, Anakinra has a short
half-life and requires daily injections that often cause side effects
at the site of the subcutaneous injection, limiting its therapeutic
potential. Inhibition of IL-1 may be of more benefit in reversing
hyperglycemia-associated insulin resistance. It should be also
underlined that one cannot exclude the possibility that effects
on hepatic insulin resistance explain the effects observed in
the study of Larsen et al. (2007).
Cell Metabolism
Review
More recently, the development of IL-1b-neutralizing antibodies that have successfully ameliorated insulin resistance
(Donath et al.,2008; Sloan-Lancaster et al.,2011; Rissanen
et al.,2011) has been reported. These preparations have a longer
half-life and require fewer injections and may therefore bear
superior therapeutic potential. Nevertheless, it must be emphasized that, to date, no clinical studies using anti-IL-1b therapies
have reported any improvement in insulin sensitivity levels.
A second potential approach would be direct inhibition of caspase-1. Specific caspase-1 inhibitors have potent beneficial
effects in animal models of insulin resistance and type 2 diabetes
(Stienstra et al., 2010), and such pharmacological agents may
prove useful in humans as well. However, at this moment, no
caspase-1 inhibitors are available for human use. Interestingly,
Glyburide, a frequently used sulfonylurea drug for the treatment
of type 2 diabetes, has been shown to inhibit the NLRP3 inflammasome (Lamkanfi et al., 2009; Masters et al., 2010), and one
may hypothesize that at least part of its effects may be due to
its anti-inflammatory action.
However, one has to keep in mind that caspase-1 cleaves
alternative substrates including caspase-7 (Lamkanfi et al.,
2008), which may mediate its deleterious effects during the
development of obesity and insulin resistance. Therefore, therapies aimed at specifically blocking caspase-1 may prove to be
more successful in attenuating type 2 diabetes, as compared
to strategies aimed solely at blockage of IL-1b.
REFERENCES
Summary
Inflammation is one of the key events that underlie the development of obesity-induced insulin resistance. Although different
roads may lead to its activation, the contribution of IL-1b to
the development of insulin resistance at the level of the b cell,
as well as peripherally, in obese individuals is now well established. Active IL-1b is produced by cleavage of pro-IL-1b by
caspase-1, which is part of the inflammasome protein complex.
Although most studies performed to date provide indirect and
associative evidence, growing evidence indicates that the inflammasome can be activated by fatty acids, high glucose
levels, uric acid, and IAPP, linking metabolic danger signals to
activation of IL-1b synthesis. The described correlations and
associations, however, do not necessarily prove causative relationships.
Inhibition of the IL-1R signaling or IL-1b production by the
NLRP3-mediated activation of caspase-1 may ward off loss of
pancreatic b cell function, yet may also prevent the development
of insulin resistance in liver, muscle, and adipose tissue.
Although clinical evidence is currently lacking, inhibition of the
IL-1R signaling or IL-1b production by the NLRP3-mediated activation of caspase-1 may avert the development of insulin resistance and represents an attractive therapeutic target to limit
pathological complications associated with obesity, insulin
resistance, and type 2 diabetes.
Cheung, K.J., Tzameli, I., Pissios, P., Rovira, I., Gavrilova, O., Ohtsubo, T.,
Chen, Z., Finkel, T., Flier, J.S., and Friedman, J.M. (2007). Xanthine oxidoreductase is a regulator of adipogenesis and PPARgamma activity. Cell Metab.
5, 115128.
ACKNOWLEDGMENTS
The authors would like to thank Sander Kersten for his valuable comments.
R.S. was supported by a Ruby Grant of the Dutch Diabetes Research Foundation. M.G.N. was supported by a Vici Grant from the Netherlands Organization
for Scientific Research (NWO).
Ali, S.R., Timmer, A.M., Bilgrami, S., Park, E.J., Eckmann, L., Nizet, V., and
Karin, M. (2011). Anthrax toxin induces macrophage death by p38 MAPK inhibition but leads to inflammasome activation via ATP leakage. Immunity 35,
3444.
Anderson, E.K., Gutierrez, D.A., and Hasty, A.H. (2010). Adipose tissue recruitment of leukocytes. Curr. Opin. Lipidol. 21, 172177.
Arend, W.P., Palmer, G., and Gabay, C. (2008). IL-1, IL-18, and IL-33 families of
cytokines. Immunol. Rev. 223, 2038.
Baldwin, W., McRae, S., Marek, G., Wymer, D., Pannu, V., Baylis, C., Johnson,
R.J., and Sautin, Y.Y. (2011). Hyperuricemia as a mediator of the proinflammatory endocrine imbalance in the adipose tissue in a murine model of the metabolic syndrome. Diabetes 60, 12581269.
Boni-Schnetzler, M., Thorne, J., Parnaud, G., Marselli, L., Ehses, J.A., KerrConte, J., Pattou, F., Halban, P.A., Weir, G.C., and Donath, M.Y. (2008).
Increased interleukin (IL)-1beta messenger ribonucleic acid expression in
beta -cells of individuals with type 2 diabetes and regulation of IL-1beta in
human islets by glucose and autostimulation. J. Clin. Endocrinol. Metab. 93,
40654074.
Browning, L.M. (2003). n-3 Polyunsaturated fatty acids, inflammation and
obesity-related disease. Proc. Nutr. Soc. 62, 447453.
Cani, P.D., Bibiloni, R., Knauf, C., Waget, A., Neyrinck, A.M., Delzenne, N.M.,
and Burcelin, R. (2008). Changes in gut microbiota control metabolic endotoxemia-induced inflammation in high-fat diet-induced obesity and diabetes in
mice. Diabetes 57, 14701481.
Chavez, J.A., Knotts, T.A., Wang, L.P., Li, G., Dobrowsky, R.T., Florant, G.L.,
and Summers, S.A. (2003). A role for ceramide, but not diacylglycerol, in the
antagonism of insulin signal transduction by saturated fatty acids. J. Biol.
Chem. 278, 1029710303.
Cinti, S., Mitchell, G., Barbatelli, G., Murano, I., Ceresi, E., Faloia, E., Wang, S.,
Fortier, M., Greenberg, A.S., and Obin, M.S. (2005). Adipocyte death defines
macrophage localization and function in adipose tissue of obese mice and humans. J. Lipid Res. 46, 23472355.
Csak, T., Ganz, M., Pespisa, J., Kodys, K., Dolganiuc, A., and Szabo, G. (2011).
Fatty acid and endotoxin activate inflammasomes in mouse hepatocytes that
release danger signals to stimulate immune cells. Hepatology 54, 133144.
Dasu, M.R., Devaraj, S., and Jialal, I. (2007). High glucose induces IL-1beta
expression in human monocytes: mechanistic insights. Am. J. Physiol. Endocrinol. Metab. 293, E337E346.
Dinarello, C.A. (2011). Interleukin-1 in the pathogenesis and treatment of
inflammatory diseases. Blood 117, 37203732.
Donath, M.Y., and Shoelson, S.E. (2011). Type 2 diabetes as an inflammatory
disease. Nat. Rev. Immunol. 11, 98107.
Donath, M.Y., Strling, J., Maedler, K., and Mandrup-Poulsen, T. (2003).
Inflammatory mediators and islet beta-cell failure: a link between type 1 and
type 2 diabetes. J. Mol. Med. (Berl) 81, 455470.
Donath, M.Y., Weder, C., Whitmore, J., Bauer, R.J., Der, K., Scannon, P.J.,
Dinarello, C.A., and Solinger, A.M. (2008). XOMA 052, an anti-IL-1b antibody,
in a double-blind, placebo-controlled, does escalation study of the safety and
pharmacokinetics in patients with type 2 diabetes mellitus- a new approach to
therapy. Diabetologia 51, s7.
Dunne, A. (2011). Inflammasome activation: from inflammatory disease to
infection. Biochem. Soc. Trans. 39, 669673.
Ehses, J.A., Perren, A., Eppler, E., Ribaux, P., Pospisilik, J.A., Maor-Cahn, R.,
Gueripel, X., Ellingsgaard, H., Schneider, M.K., Biollaz, G., et al. (2007).
Increased number of islet-associated macrophages in type 2 diabetes. Diabetes 56, 23562370.
Elinav, E., Strowig, T., Kau, A.L., Henao-Mejia, J., Thaiss, C.A., Booth, C.J.,
Peaper, D.R., Bertin, J., Eisenbarth, S.C., Gordon, J.I., and Flavell, R.A.
Cell Metabolism
Review
(2011). NLRP6 inflammasome regulates colonic microbial ecology and risk for
colitis. Cell 145, 745757.
Erridge, C., and Samani, N.J. (2009). Saturated fatty acids do not directly stimulate Toll-like receptor signaling. Arterioscler. Thromb. Vasc. Biol. 29, 1944
1949.
Goldfine, A.B., Silver, R., Aldhahi, W., Cai, D., Tatro, E., Lee, J., and Shoelson,
S.E. (2008). Use of salsalate to target inflammation in the treatment of insulin
resistance and type 2 diabetes. Clin Transl Sci 1, 3643.
Gong, Y.N., Wang, X., Wang, J., Yang, Z., Li, S., Yang, J., Liu, L., Lei, X., and
Shao, F. (2010). Chemical probing reveals insights into the signaling mechanism of inflammasome activation. Cell Res. 20, 12891305.
Gonzalez-Gay, M.A., Gonzalez-Juanatey, C., Vazquez-Rodriguez, T.R.,
Miranda-Filloy, J.A., and Llorca, J. (2010). Insulin resistance in rheumatoid
arthritis: the impact of the anti-TNF-alpha therapy. Ann. N Y Acad. Sci.
1193, 153159.
Lamkanfi, M., Mueller, J.L., Vitari, A.C., Misaghi, S., Fedorova, A., Deshayes,
K., Lee, W.P., Hoffman, H.M., and Dixit, V.M. (2009). Glyburide inhibits the Cryopyrin/Nalp3 inflammasome. J. Cell Biol. 187, 6170.
Larsen, C.M., Faulenbach, M., Vaag, A., Vlund, A., Ehses, J.A., Seifert, B.,
Mandrup-Poulsen, T., and Donath, M.Y. (2007). Interleukin-1-receptor antagonist in type 2 diabetes mellitus. N. Engl. J. Med. 356, 15171526.
Lee, J.Y., Sohn, K.H., Rhee, S.H., and Hwang, D. (2001). Saturated fatty acids,
but not unsaturated fatty acids, induce the expression of cyclooxygenase-2
mediated through Toll-like receptor 4. J. Biol. Chem. 276, 1668316689.
Maedler, K., Sergeev, P., Ris, F., Oberholzer, J., Joller-Jemelka, H.I., Spinas,
G.A., Kaiser, N., Halban, P.A., and Donath, M.Y. (2002). Glucose-induced
beta cell production of IL-1beta contributes to glucotoxicity in human pancreatic islets. J. Clin. Invest. 110, 851860.
Mandrup-Poulsen, T. (2010). IAPP boosts islet macrophage IL-1 in type 2 diabetes. Nat. Immunol. 11, 881883.
Gregor, M.F., and Hotamisligil, G.S. (2011). Inflammatory mechanisms in

obesity. Annu. Rev. Immunol. 29, 415445.
Mandrup-Poulsen, T., Bendtzen, K., Nerup, J., Dinarello, C.A., Svenson, M.,
and Nielsen, J.H. (1986). Affinity-purified human interleukin I is cytotoxic to isolated islets of Langerhans. Diabetologia 29, 6367.
Haversen, L., Danielsson, K.N., Fogelstrand, L., and Wiklund, O. (2009). Induction of proinflammatory cytokines by long-chain saturated fatty acids in human
macrophages. Atherosclerosis 202, 382393.
Mandrup-Poulsen, T., Pickersgill, L., and Donath, M.Y. (2010). Blockade of

interleukin 1 in type 1 diabetes mellitus. Nat Rev Endocrinol 6, 158166.
Hornung, V., Ablasser, A., Charrel-Dennis, M., Bauernfeind, F., Horvath, G.,
Caffrey, D.R., Latz, E., and Fitzgerald, K.A. (2009). AIM2 recognizes cytosolic
dsDNA and forms a caspase-1-activating inflammasome with ASC. Nature
458, 514518.
Huvers, F.C., Popa, C., Netea, M.G., van den Hoogen, F.H., and Tack, C.J.
(2007). Improved insulin sensitivity by anti-TNFalpha antibody treatment in
patients with rheumatic diseases. Ann. Rheum. Dis. 66, 558559.
Ito, A., Suganami, T., Yamauchi, A., Degawa-Yamauchi, M., Tanaka, M.,
Kouyama, R., Kobayashi, Y., Nitta, N., Yasuda, K., Hirata, Y., et al. (2008).
Role of CC chemokine receptor 2 in bone marrow cells in the recruitment of
macrophages into obese adipose tissue. J. Biol. Chem. 283, 3571535723.
Jager, J., Gremeaux, T., Cormont, M., Le Marchand-Brustel, Y., and Tanti, J.F.
(2007). Interleukin-1beta-induced insulin resistance in adipocytes through
down-regulation of insulin receptor substrate-1 expression. Endocrinology
148, 241251.
Jin, W., and Patti, M.E. (2009). Genetic determinants and molecular pathways
in the pathogenesis of Type 2 diabetes. Clin. Sci. 116, 99111.
Joosten, L.A., Netea, M.G., Fantuzzi, G., Koenders, M.I., Helsen, M.M., Sparrer, H., Pham, C.T., van der Meer, J.W., Dinarello, C.A., and van den Berg, W.B.
(2009). Inflammatory arthritis in caspase 1 gene-deficient mice: contribution of
proteinase 3 to caspase 1-independent production of bioactive interleukin1beta. Arthritis Rheum. 60, 36513662.
Joosten, L.A., Netea, M.G., Mylona, E., Koenders, M.I., Malireddi, R.K., Oosting, M., Stienstra, R., van de Veerdonk, F.L., Stalenhoef, A.F., GiamarellosBourboulis, E.J., et al. (2010). Engagement of fatty acids with Toll-like receptor
2 drives interleukin-1b production via the ASC/caspase 1 pathway in monosodium urate monohydrate crystal-induced gouty arthritis. Arthritis Rheum. 62,
32373248.
Koenen, T.B., Stienstra, R., van Tits, L.J., de Graaf, J., Stalenhoef, A.F., Joosten, L.A., Tack, C.J., and Netea, M.G. (2011a). Hyperglycemia activates caspase-1 and TXNIP-mediated IL-1beta transcription in human adipose tissue.
Koenen, T.B., Stienstra, R., van Tits, L.J., Joosten, L.A., van Velzen, J.F., Hijmans, A., Pol, J.A., van der Vliet, J.A., Netea, M.G., Tack, C.J., et al. (2011b).
The inflammasome and caspase-1 activation: a new mechanism underlying
increased inflammatory activity in human visceral adipose tissue. Endocrinology 152, 37693778.
Krebs, M., and Roden, M. (2005). Molecular mechanisms of lipid-induced
insulin resistance in muscle, liver and vasculature. Diabetes Obes. Metab. 7,
621632.
Lamkanfi, M., Kanneganti, T.D., Van Damme, P., Vanden Berghe, T., Vanoverberghe, I., Vandekerckhove, J., Vandenabeele, P., Gevaert, K., and Nunez, G.
(2008). Targeted peptidecentric proteomics reveals caspase-7 as a substrate
of the caspase-1 inflammasomes. Mol. Cell. Proteomics 7, 23502363.
Mariathasan, S., Weiss, D.S., Newton, K., McBride, J., ORourke, K., RooseGirma, M., Lee, W.P., Weinrauch, Y., Monack, D.M., and Dixit, V.M. (2006).
Cryopyrin activates the inflammasome in response to toxins and ATP. Nature
440, 228232.
Martinon, F., Petrilli, V., Mayor, A., Tardivel, A., and Tschopp, J. (2006). Goutassociated uric acid crystals activate the NALP3 inflammasome. Nature 440,
237241.
Masters, S.L., Dunne, A., Subramanian, S.L., Hull, R.L., Tannahill, G.M., Sharp,
F.A., Becker, C., Franchi, L., Yoshihara, E., Chen, Z., et al. (2010). Activation of
the NLRP3 inflammasome by islet amyloid polypeptide provides a mechanism
for enhanced IL-1b in type 2 diabetes. Nat. Immunol. 11, 897904.
Masters, S.L., Latz, E., and ONeill, L.A. (2011). The inflammasome in atherosclerosis and type 2 diabetes. Sci. Transl. Med. 3, ps17.
McGillicuddy, F.C., Harford, K.A., Reynolds, C.M., Oliver, E., Claessens, M.,
Mills, K.H., and Roche, H.M. (2011). Lack of interleukin-1 receptor I (IL-1RI)
protects mice from high-fat diet-induced adipose tissue inflammation coincident with improved glucose homeostasis. Diabetes 60, 16881698.
Meijer, K., de Vries, M., Al-Lahham, S., Bruinenberg, M., Weening, D., Dijkstra,
M., Kloosterhuis, N., van der Leij, R.J., van der Want, H., Kroesen, B.J., et al.
(2011). Human primary adipocytes exhibit immune cell function: adipocytes
prime inflammation independent of macrophages. PLoS ONE 6, e17154.
Meissner, F., Seger, R.A., Moshous, D., Fischer, A., Reichenbach, J., and Zychlinsky, A. (2010). Inflammasome activation in NADPH oxidase defective
mononuclear phagocytes from patients with chronic granulomatous disease.
Blood 116, 15701573.
Minn, A.H., Hafele, C., and Shalev, A. (2005). Thioredoxin-interacting protein is
stimulated by glucose through a carbohydrate response element and induces
beta-cell apoptosis. Endocrinology 146, 23972405.
Muruve, D.A., Petrilli, V., Zaiss, A.K., White, L.R., Clark, S.A., Ross, P.J., Parks,
R.J., and Tschopp, J. (2008). The inflammasome recognizes cytosolic microbial and host DNA and triggers an innate immune response. Nature 452,
103107.
Nakagawa, T., Hu, H., Zharikov, S., Tuttle, K.R., Short, R.A., Glushakova, O.,
Ouyang, X., Feig, D.I., Block, E.R., Herrera-Acosta, J., et al. (2006). A causal
role for uric acid in fructose-induced metabolic syndrome. Am. J. Physiol.
Renal Physiol. 290, F625F631.
Netea, M.G., Joosten, L.A., Lewis, E., Jensen, D.R., Voshol, P.J., Kullberg,
B.J., Tack, C.J., van Krieken, H., Kim, S.H., Stalenhoef, A.F., et al. (2006). Deficiency of interleukin-18 in mice leads to hyperphagia, obesity and insulin resistance. Nat. Med. 12, 650656.
Netea, M.G., Nold-Petry, C.A., Nold, M.F., Joosten, L.A., Opitz, B., van der
Meer, J.H., van de Veerdonk, F.L., Ferwerda, G., Heinhuis, B., Devesa, I.,
et al. (2009). Differential requirement for the activation of the inflammasome
for processing and release of IL-1beta in monocytes and macrophages. Blood
113, 23242335.
Cell Metabolism
Review
Parikh, H., Carlsson, E., Chutkow, W.A., Johansson, L.E., Storgaard, H., Poulsen, P., Saxena, R., Ladd, C., Schulze, P.C., Mazzini, M.J., et al. (2007). TXNIP
regulates peripheral glucose metabolism in humans. PLoS Med. 4, e158.
Perrone, L., Devi, T.S., Hosoya, K., Terasaki, T., and Singh, L.P. (2009). Thioredoxin interacting protein (TXNIP) induces inflammation through chromatin
modification in retinal capillary endothelial cells under diabetic conditions. J.
Cell. Physiol. 221, 262272.
Rissanen, A., Howard, C., Botha, J., and Thuren, T. (2011). IL-1b antibody
(Canakinumab) improves insulin secretion rates in subjects with impaired
glucose tolerance (IGT) and type 2 diabetes (T2DM). Diabetes 60 (Suppl 1A ),
LB11.
Shanmugam, N., Reddy, M.A., Guha, M., and Natarajan, R. (2003). High
glucose-induced expression of proinflammatory cytokine and chemokine
genes in monocytic cells. Diabetes 52, 12561264.
Shi, Y., Evans, J.E., and Rock, K.L. (2003). Molecular identification of a danger
signal that alerts the immune system to dying cells. Nature 425, 516521.
Shi, H., Kokoeva, M.V., Inouye, K., Tzameli, I., Yin, H., and Flier, J.S. (2006).
TLR4 links innate immunity and fatty acid-induced insulin resistance. J. Clin.
Invest. 116, 30153025.
Sloan-Lancaster, J., Abu-Raddad, E., Polzer, J., Miller, J.W., Scherer, J.C.,
Berg, J.K., and Landschulz, W.H. (2011). Safety, tolerability and efficacy of
subcutaneous (SC) LY2189102 (LY), a neutralizing IL-1b antibody, in patients
(Pts) with type 2 diabetes (T2DM). Diabetes 60 (Suppl 1A ), LB14.
Spranger, J., Kroke, A., Mohlig, M., Hoffmann, K., Bergmann, M.M., Ristow,
M., Boeing, H., and Pfeiffer, A.F. (2003). Inflammatory cytokines and the risk
to develop type 2 diabetes: results of the prospective population-based European Prospective Investigation into Cancer and Nutrition (EPIC)-Potsdam
Study. Diabetes 52, 812817.
Steinberg, G.R., and Kemp, B.E. (2009). AMPK in Health and Disease. Physiol.
Rev. 89, 10251078.
Stienstra, R., Joosten, L.A., Koenen, T., van Tits, B., van Diepen, J.A., van den
Berg, S.A., Rensen, P.C., Voshol, P.J., Fantuzzi, G., Hijmans, A., et al. (2010).
The inflammasome-mediated caspase-1 activation controls adipocyte differentiation and insulin sensitivity. Cell Metab. 12, 593605.
Stienstra, R., van Diepen, J.A., Tack, C.J., Zaki, M.H., van de Veerdonk, F.L.,
Perera, D., Neale, G.A., Hooiveld, G.J., Hijmans, A., Vroegrijk, I., et al. (2011).
Inflammasome is a central player in the induction of obesity and insulin resistance. Proc. Natl. Acad. Sci. USA 108, 1532415329.
Stoltzman, C.A., Peterson, C.W., Breen, K.T., Muoio, D.M., Billin, A.N., and
Ayer, D.E. (2008). Glucose sensing by MondoA:Mlx complexes: a role for
hexokinases and direct regulation of thioredoxin-interacting protein expression. Proc. Natl. Acad. Sci. USA 105, 69126917.
Tschopp, J. (2011). Mitochondria: Sovereign of inflammation? Eur. J. Immunol.
41, 11961202.
van Asseldonk, E.J., Stienstra, R., Koenen, T.B., Joosten, L.A., Netea, M.G.,
and Tack, C.J. (2011). Treatment with Anakinra improves disposition index
but not insulin sensitivity in nondiabetic subjects with the metabolic syndrome:
a randomized, double-blind, placebo-controlled study. J. Clin. Endocrinol.
Metab. 96, 21192126.
van de Veerdonk, F.L., Smeekens, S.P., Joosten, L.A., Kullberg, B.J., Dinarello, C.A., van der Meer, J.W., and Netea, M.G. (2010). Reactive oxygen
species-independent activation of the IL-1beta inflammasome in cells from
patients with chronic granulomatous disease. Proc. Natl. Acad. Sci. USA
107, 30303033.
Vandanmagsar, B., Youm, Y.H., Ravussin, A., Galgani, J.E., Stadler, K.,
Mynatt, R.L., Ravussin, E., Stephens, J.M., and Dixit, V.D. (2011). The
NLRP3 inflammasome instigates obesity-induced inflammation and insulin
resistance. Nat. Med. 17, 179188.
Vincent, J.A., and Mohr, S. (2007). Inhibition of caspase-1/interleukin-1beta
signaling prevents degeneration of retinal capillaries in diabetes and galactosemia. Diabetes 56, 224230.
Wen, H., Gris, D., Lei, Y., Jha, S., Zhang, L., Huang, M.T., Brickey, W.J., and
Ting, J.P. (2011). Fatty acid-induced NLRP3-ASC inflammasome activation
interferes with insulin signaling. Nat. Immunol. 12, 408415.
Westermark, P., Andersson, A., and Westermark, G.T. (2011). Islet amyloid
polypeptide, islet amyloid, and diabetes mellitus. Physiol. Rev. 91, 795826.
Zhou, R., Tardivel, A., Thorens, B., Choi, I., and Tschopp, J. (2010). Thioredoxin-interacting protein links oxidative stress to inflammasome activation.
Nat. Immunol. 11, 136140.
Cell Metabolism
Perspective
Metabolic Disease Drug Discovery
Hitting the Target Is Easier Said Than Done
David E. Moller1,*
1Division of Endocrinology and Cardiovascular Discovery Research and Clinical Investigation, Eli Lilly and Company, Lilly Research
Laboratories, Indianapolis, IN 46285, USA
*Correspondence: mollerda@lilly.com
DOI 10.1016/j.cmet.2011.10.012
Despite the advent of new drug classes, the global epidemic of cardiometabolic disease has not abated.
Continuing unmet medical needs remain a major driver for new research. Drug discovery approaches in
this field have mirrored industry trends, leading to a recent increase in the number of molecules entering
development. However, worrisome trends and newer hurdles are also apparent. The history of two newer
drug classesglucagon-like peptide-1 receptor agonists and dipeptidyl peptidase-4 inhibitorsillustrates
both progress and challenges. Future success requires that researchers learn from these experiences and
continue to explore and apply new technology platforms and research paradigms.
Introduction
The global epidemic of obesity and diabetes continues to progress relentlessly. The International Diabetes Federation predicts
an even greater diabetes burden (>430 million people afflicted)
by 2030, which will disproportionately affect developing nations
(International Diabetes Federation, 2011). Yet existing drug
classes for diabetes, obesity, and comorbid cardiovascular
(CV) conditions have substantial limitations.
Currently available prescription drugs for treatment of hyperglycemia in patients with type 2 diabetes (Table 1) have notable
shortcomings. In general, their efficacy profiles fall short of
achieving accepted treatment goals (American Diabetes Association, 2011). Therefore, clinicians must often use combination
therapy, adding additional agents over time. Ultimately many
patients will need to use insulina therapeutic class first introduced in 1922. Most existing agents also have issues around
safety and tolerability as well as dosing convenience (which
can impact patient compliance).
Future therapies must focus on unmet medical needs in the
following ways: (1) they must have mechanisms that can safely
achieve greater glycemic efficacy; (2) they must mitigate the
CV risk associated with diabetes, obesity, and dyslipidemia; (3)
they should help patients with or at-risk of diabetes to achieve
meaningful and safe weight loss; and (4) they should attenuate
the progression of microvascular complications. For diabetes
and obesity in particular, an overriding need exists for true
disease-modifying agents that deliver longer term benefitsto
prevent disease and/or disease progression. In examining the
extent to which existing therapies address these needs, consider
the fact that pharmacotherapy for obesity has been available in
the U.S. since the FDA approval of amphetamine sulfate (Benzedrine) in 1939. Yet, after many intervening episodes of transient
success and spectacular failure, today there are only two
approved (U.S.) weight-loss agents, orlistat and phentermine,
both with limited effectiveness. Phentermine is only indicated
for transient use, whereas treatment of metabolic disease generally requires lifelong intervention! Moreover, improvements in
glucose control over the past 20 years have been generally
modest. Even in a wealthy developed nation like the U.S.,
more than 40% of patients diagnosed with diabetes do not

achieve accepted glycemic goals (Hoerger et al., 2008); an
even greater proportion (80%) fail to also achieve critical goals
pertaining to blood pressure and lipids.
The confluence of an expanding epidemic, limitations of
current drugs, and compelling unmet medical needs provide
a strong rationale for continued emphasis on finding and developing novel therapeutic approaches.
The Current Paradigm
Tools from modern biology and an understanding of pathophysiology have lead to a common model for drug discovery and
development (Milne, 2008; Kola, 2008; Paul et al., 2010)
(Figure 1). R&D efforts aimed at metabolic disease have largely
mirrored industry trends in this regard.
Most projects focus on individual drug targets: discrete
gene products/proteins (Yang et al., 2009). However, strict
adherence to the single-target approach may be ill-advised
since all older and even many newer (since 1999) drug classes
were discovered using phenotypic screening, without initial
knowledge of a defined molecular target (Swinney and Anthony,
2011).
Screening technologies and chemistry trends have driven the
selection of targets toward drugable classes. Contemporary
examples include enzymes (e.g., glucokinase), circulating
proteins such as cholesterol ester transfer protein (CETP), and
cell-surface receptors (e.g., glucagon receptor). Peptides and
larger circulating proteins are also popular targets for biomolecules (e.g., peptide YY, ghrelin, fibroblast growth factor 21
[FGF21], and proprotein convertase subtilisin/kexin type 9
[PCSK9]).
Target validation is the buzzword for establishing links
between the target and human disease. This includes human
genetic association, regulation of expression in the context of
disease, phenotypes detected in transgenic or knockout mice,
consequences of siRNA-mediated gene knockdown, and pharmacology in animals or early clinical testing (Yang et al., 2009;
Milne, 2009; Libby et al., 2011). Validation is best seen as a
continuum. For example, early validation via a knockout mouse
Cell Metabolism
Perspective
Table 1. Major Existing Drug Classes Approved for Therapy of Hyperglycemia in Patients with Type 2 Diabetes (Excluding Insulin
Products)
Drug Class (Examples)
Molecular Mechanism
Efficacy Profile (D Hb A1c*)
Adverse Effects/Tolerability (other)
Sulfonylureas (glyburide, glipizide,

glimepiride) and meglitinides
(nateglinide, repaglinide)
K+-ATP channel modulator

(insulin secretion)
Moderate-good - but lacking

long term durability
( 0.7%1.3%)
Hypoglycemia; weight gain;

3 doses per day with meglitinides
Metformin
Unknown
Moderate ( 0.7%1.0%)
Gastrointestinal symptoms;
twice-daily dosing; lactic
acidosis (rare); contraindicated
with reduced renal function
Thiazolidinediones
(pioglitazone, rosiglitazone)
PPARg agonist
(insulin sensitization)
Moderate ( 0.6%1.1%)
Weight gain, edema, exacerbation

of heart failure, increased bone
fractures; possible increased
bladder cancer (pioglitazone);
potential increased risk of
myocardial infarction (rosiglitazone)
a-glucosidase inhibitors
(acarbose)
Inhibition of intestinal
glucose absorption
Fair ( 0.4%0.8%)
Gastrointestinal symptoms
(frequent)
Incretin-mimetics
(exenatide; liraglutide)
GLP-1 receptor agonist
Moderate-good ( 0.5%1.3%);
plus modest weight loss
Daily or twice-daily injection required;

nausea and vomiting plus other
gastrointestinal symptoms; possible
increased risk of pancreatitis
Dipeptidyl-peptidase-4
inhibitors (sitagliptin,
saxagliptin, linagliptin)
Increased endogenous
GLP-1 and GIP levels
Fair ( 0.5%0.8%)
Hypersensitivity reactions (rare);

nasopharyngitis; upper respiratory
infection (infrequent); possible
increased risk of pancreatitis
Table information derived from US prescribing and label information for representative products available at http://www.FDA.gov. PPARg = peroxisome proliferator-activated receptor g. * Hb A1c = hemoglobin A1c (glycosylated hemoglobin); D Hb A1c ranges noted represent effects seen as
monotherapy and/or incremental when added to metformin.
could be followed by confirmation via pharmacologic modulation

in animal models.
Once molecules that modulate target function are discovered
(screening and lead generation), lead optimization typically
follows. In this phase, many molecules are synthesized in an iterative fashion seeking improved potency, selectivity, pharmacokinetics, etc. Molecules selected for early development undergo
toxicology tests and initial (Phase I) human trials, but clear
signals of efficacy are usually not evident until the end of Phase
I or in Phase II.
The current process is long, arduous, and increasingly expensive. Since earlier phases of drug discovery/development have
greater risks of failure, the number of molecules entering
Phase I must exceed the number of desired new molecular entity
(NME) launches by an order of magnitude (Paul et al., 2010). The
daunting prospects for failure in preclinical phases demand
even greater numbers of basic research projects to supply new
Phase I opportunities.
Worrying Trends
The biopharmaceutical industry is facing an unprecedented
wave of patent expiries leading to a cumulative loss of U.S.
$78 billion in worldwide sales during 20102014 (Harrison,
2011). Drugs for diabetes and CV disease represented in this
cohort include atorvastatin, clopidogrel, and pioglitazone. These
and other macroeconomic forces, including rising healthcare
costs and pressure from third party payers, have created an
environment where new blockbusters are needed but funding
to support R&D is being curtailed.
A recent decline in total new drug approvals is worrisome.

Only 21 NMEs were approved by the FDA in 2010 and only 19
were approved in 2007, the lowest yearly total in 20 years
(Mullard, 2011). However, it is encouraging to note that new
approvals for the first half of 2011 have already matched the
2010 total.
Despite substantial industry investments, attrition rates for
pipeline molecules have not significantly improved. Thus, the
overall probability of success (POS) for molecules entering
Phase I remains low (%10%) and the aggregate average cost
to get a new drug approved is increasing (>$1 billion by current
estimates). Factors contributing to low productivity include (1)
increasing focus on disease areas characterized by lower POS,
oncology, and neuroscience in particular (Pammolli et al.,
2011; Kola, 2008); (2) research efforts oriented toward novel
poorly validated and difficult targets; (3) higher safety-regulatory hurdles; and (4) the requirement to achieve a clinical profile
that predicts commercial success. The poor track record in
translating efficacy in animal models to meaningful clinical efficacy remains challenging. This occurs frequently in metabolic
disease drug developmentespecially when weight loss is the
desired clinical effect. Therefore, the major inflection point in
POS during drug development is in Phase II where definitive
proof-of-concept is typically established (Kola, 2008; Pammolli
et al., 2011; Paul et al., 2010; Milne, 2009).
Stringent regulatory requirements and specific CV safety
concerns are substantial hurdles for new diabetes drugs. Questions were raised by the Action to Control CV Risk in Diabetes
(ACCORD) trial, which showed increased CV mortality with
Cell Metabolism
Perspective
Figure 1. Roadmap for Common Practices in Current Drug Discovery/Development

Industry-based efforts focusing on cardiometabolic disease have largely mirrored the trends and approaches noted in this generic R&D paradigm.
intensive glucose control in type 2 diabetes (Gerstein et al., 2011)

as well as by concerns pertaining to specific drugs such as
rosiglitazone (Hirshberg and Raz, 2011; Drucker and Goldfine,
2011). In response, recent FDA guidance mandates that new
antihyperglycemic drugs exclude significant CV risk prior to
initial approval (Hirshberg and Raz, 2011), thus requiring longer
(4 to 5 yr), larger (>5000 patient), and more costly (incremental
increase of $300 million) Phase III development programs.
An obvious consequence is that only companies with deep
pockets will have the resources to develop and launch diabetes
drugs. Hopefully, higher quality, safer drugs offering genuine CV
benefits may result despite fewer agents being selected for late
stage development.
Cases in Point
The history of glucagon-like peptide-1 (GLP-1) receptor agonists
and dipeptidyl peptidase-4 (DPP-4) inhibitors highlight important
aspects of the drug discovery process (Figure 2). The proglucagon gene, which encodes GLP-1, was cloned in 1982 (Lund
et al., 1982), several years after the discovery of another major
incretingastric inhibitory peptide (GIP); however, GLP-1s
ability to potentiate glucose-stimulated insulin secretion in humans was not reported until 1987 (Kreymann et al., 1987).
Normalization of glucose in patients with type 2 diabetes by
exogenous GLP-1 was described thereafter (Nauck et al.,
1993), leading to intense interest in this hormone and its
surrounding pathways as potential targets of therapy (Deacon
and Holst, 2002; Drucker, 2006; Nauck, 2011; Hansen et al.,
2009). The ability of GLP-1 to promote weight loss and enhance
b cell mass and function in rodents added to the enthusiasm.
A major challenge for developing native GLP-1 (or GIP) as

a drug was the finding that it is rapidly cleaved and inactivated
(via removal of two N-terminal residues) by DPP-4 (Mentlein
et al., 1993). This in vitro finding was ultimately confirmed in
animal and human studies (Deacon et al., 1995). Exendin-4,
a naturally occurring peptide, was isolated from the saliva of
Heloderma suspectum in the early 1990s (Eng et al., 1992). Its
characterization as an incretin-mimetic GLP-1 receptor agonist
resistant to degradation by DPP-4 was a key contribution to
the field (Drucker, 2006).
A decade of academic research lead to industry-based efforts
focused on the discovery of novel degradation-resistant incretinmimetic peptides and small molecule DPP-4 inhibitors, which
began in earnest in 19951996 (Figure 2). Since no modifications
to the peptide were required, development of exendin-4 as
a twice-daily injectable therapy (exenatide) proceeded fairly
rapidly. Its clinical efficacy profile was defined in 20012004
(Fineman et al., 2003), and FDA approval was achieved in
2005. A second once-daily injectable human GLP-1 analog, liraglutide (Degn et al., 2004), was subsequently developed and
launched several years later (Jeong and Yoo, 2011). Other longer
acting forms of exenatide or GLP-1 analogs are now in development (Nauck, 2011; Hansen et al., 2009).
The pursuit of small molecule DPP-4 inhibitors was aided by
the availability of tool compounds such as valine-pyrrolide,
which exert glycemic benefits in animals (Pauly et al., 1996;
Demuth et al., 2002). Initial concerns and issues surrounding
the DPP-4 mechanism, such as the potential for promiscuous
effects via regulation of substrates other than GLP-1 and
GIP and the importance of selectivity versus other prolyl
Cell Metabolism
Perspective
19 9 0
1995
2000
2005
2010
Figure 2. Abbreviated Timeline of the Discovery and Development of GLP-1 Receptor Agonists and DPP-4 Inhibitors
oligopeptidase family members, were successfully addressed

(Conarello et al., 2003; Lankas et al., 2005) in parallel with the
pursuit of novel and improved molecules. Early human proofof-concept was established in 20012002 (Ahren et al., 2002)
leading to even more intensive industry-based efforts and registration and approval of sitagliptin in 2006 (Subbarayan and
Kipnes, 2011). Several additional DPP-4 inhibitors have been
subsequently launched.
What are the take-home messages? Industry-based R&D
efforts were preceded by years of academic research establishing the importance of this pathway in animals and in human
disease. However, for both first-in-class drugs, progress from
initiation of formal drug discovery to approval was still
a decade-long adventure, even with important contributions
from continued academic research and close industry-academic
collaboration. Whats often ignored but can be gleaned
patents and public company recordsis that many additional
clinical stage molecules were discovered and partially developed, before being terminated for technical (e.g., safety) or strategic (not viewed as commercially viable) reasons. I estimate that
more than five additional GLP-1 related peptides and ten additional DPP-4 inhibitors were characterized in early human clinical
trials! Its also noteworthy that U.S. FDA approval of second or
third-in-class molecules for both mechanisms lagged behind
by 36 years. This was largely a consequence of the previously
described regulatory requirements imposed to insure greater
safety.
Along with the successes achieved in the GLP-1 and DPP-4
drug classes were high profile failures, such as torcetrapib and
rimonabant. The late Phase III failure of Torcetrapib, the first
developed CETP inhibitor, was a dramatic example of how offtarget vascular toxicity (mediated by increases in aldosterone)
led to unexpected adverse outcomes rather than the CV benefit
expected from increased HDL cholesterol (Food and Drug
Administration, 2006; Hu et al., 2009). An even more spectacular
failure involved withdrawal of rimonabant, a cannabanoid 1
(CB1) receptor inverse agonist for treatment of obesity, from
the market. Adverse psychiatric effects (no doubt mechanismbased) proved to be the Achilles heel for CB1. In fact, the demise
of rimonabant incited wholesale retreat from this approach by
companies across the globe (Lee et al., 2009).
Improving the Odds

Incretin-based therapies are contemporary examples of elegant
approaches harnessing a well-characterized pathway. It can be
argued that the POS for these mechanisms was relatively high
at the stage when industry efforts were ramping up. Yet the
field of cardiometabolic disease is also littered with failures.
Past commercial successes have been generally sufficient to
fund newer R&D efforts, but the aggregate expense and protracted timelines we face today (even for the most successful
ventures) implies that the current paradigm is not sustainable.
To improve the odds, a paradigm shift must focus on three
strategic goals.
1. Leverage New Technology Platforms
Several newer platforms can provide important target validation
information. Cell-based knockdown or overexpression as well as
in vivo genetic models provide valuable insights. A good
example is the broad use of mouse knockouts to screen for
obesity-related targets (Brommage et al., 2008). However,
compensation by other genes can mask or alter the phenotype
and background strain may also exert major effects (Lariviere
et al., 2001). Thus, results obtained using knockout mice must
be interpreted with caution. In my experience, a mouse phenotype of resistance to diet-induced obesity typically does not
translate into favorable pharmacology when the appropriate
tool compound(s) are tested in a similar context. But despite
shortcomings, these technologies can and should still be efficiently used to winnow longer lists of candidate targets and
pathways.
Advances in drug screening and medicinal chemistry technologies are also paving the way for pursuit of intractable targets,
such as small molecule GLP-1 receptor agonists (Sloop et al.,
2010).
Target identification and prioritization can also be augmented
by applying new computational data mining approaches to
preexisting information sourcesincluding genetic data, microarray data, proteomic data, etc.conjoined by integrated
pathway mapping (Yang et al., 2009). One newer application
involves integrating information from genetic mapping with
tissue gene-expression profiling to generate expanded
disease-associated networks (Schadt, 2009). This approach
might soon yield novel drug targets, but the volume of data
Cell Metabolism
Perspective
generated presents great challenges. Even major fans concede
that systems approaches are still in their infancy (Schadt,
2009).
Importantly, many of the more modern techniques noted
above were not routinely applied before 19952000. Therefore,
their impact on net industry output cannot yet be fully assessed.
2. Better Human Validation
Our understanding of human disease has not kept pace with the
abundance of identified target opportunities. Better understanding of disease pathogenesis remains essential. For type 2
diabetes, obesity, and CV disease, drug target selection and
prioritization should take human genetics into account (McCarthy, 2010; Libby et al., 2011). Moreover, increased POS can be
achieved with selected targets such as PCSK9 where monogenic traits affecting LDL cholesterol strongly predict future
clinical efficacy (reviewed by Libby et al., 2011). The application
of metabolite profiling (metabolomics) to larger clinical
samples also provides new insights. Rather than relying on
cross-sectional studies, it is important to incorporate prospective clinical endpoints. Findings related to diabetes pathogenesis
obtained using mass spectrometry in >2000 blood samples
from the Framingham Offspring cohort illustrate this point
(Wang et al., 2011). Similarly, tissue biobanks with associated
prospective clinical outcomes data can be used to detect critical
pathogenic nodes as exemplified by recent results of osteopontin levels in carotid atherosclerotic lesions (de Kleijn et al., 2010).
3. Apply Translational Medicine Principles
Efficient progression of drug discovery projects toward successful clinical readouts is at the heart of translational medicine
(Milne, 2009). Information from high-quality human experiments
elucidates next steps in research. Biomarkers that inform target
engagement and early efficacy signals are critical to making
succinct go/no-go decisions. This requires exploiting a full
repertoire of imaging, gene expression profiles, and circulating
markers (including metabolomics). Fortunately, many mechanisms targeting cardiometabolic disease lend themselves to
early efficacy signal detection. For example, 24 hr glucose
profiles can predict longer term glycemic efficacy (HbA1c
lowering). Remember, however, that even accepted surrogate
biomarkers for outcomes like HbA1c may have limitations as
illustrated by the unexpected results from ACCORD. Methods
for detecting small changes in human energy balance (intake
or expenditure) that predict future weight loss have also been
perfected (Gaich and Moller, 2011). With better translational
medicine tools and molecular probes in hand, a shift toward
earlier implementation of exploratory human testing (clinical
target assessment) is warranted. For example, a candidate
obesity pathway was examined in humans via continuous infusion of a melanocortin 4 selective agonist peptide over several
days (Greenfield et al., 2009).
Where Do We Go from Here?
Future success requires a closer relationship between industry
and academia as well as active knowledge sharing between
research groups through multiparty partnerships and consortia.
The Innovative Medicines Initiative for Diabetes is an excellent
example. This industry-academic consortium focused on relevant questions pertaining to the regulation of b cell mass and
function (IMIDIA, 2011). Other consortia have the potential for
great impact on the identification and use of biomarkers across

companies (Milne, 2009). Additional examples of open source
innovation that can facilitate academic-industry collaboration
in identifying novel drug-like molecules are also emerging (Lee
et al., 2011). The focused NIH Center for Advancing Translational
Sciences is being established to foster knowledge-sharing that
can favorably impact both academic- and industry-based
applied research (Bristol, 2011).
Finally, we need to carefully consider the concept of disease
stratification. Just as breast cancer is not a single disease, we
must acknowledge and embrace the notion that type 2 diabetes,
obesity, and related comorbid CV conditions are syndromes
with many distinct underlying molecular causes. This will
undoubtedly result in a shift toward the pursuit of tailored
therapies rather than perseverating with once-size-fits-all
approaches. However, the feasibility of developing drugs that
target only 10% of the population when CV risk must also be
discharged prior to drug approval remains to be seen.
Metabolic disease research has revealed many compelling
molecular pathways with possible connections to human
disease. Signs of improved productivity for industry-based
R&D efforts are also emerging. But rethinking the current paradigm, embracing new technologies, and adopting translational
medicine principles are crucial to progress in our field.
ACKNOWLEDGMENTS
I wish to thank Michael Coghlan, Bei B. Zhang, Robert Heine, David D.K.
Kendall, and Jennie Jacobson (all at Eli Lilly and Company) for helpful
suggestions. The author is an employee and shareholder of Eli Lilly and
Company. Information provided is derived from literature sources cited in
this Perspective and FDA approval dates (*) noted at http://www.FDA.gov.
Note that two additional DPP-4 inhibitors (vildagliptin and alogliptin) are
also approved for use outside the U.S.
REFERENCES
Ahren, B., Simonsson, E., Larsson, H., Landin-Olsson, M., Torgeirsson, H.,
Jansson, P.A., Sandqvist, M., Bavenholm, P., Efendic, S., Eriksson, J.W.,
et al. (2002). Inhibition of dipeptidyl peptidase IV improves metabolic control
over a 4-week study period in type 2 diabetes. Diabetes Care 25, 869875.
American Diabetes Association. (2011). Standards of medical care in diabetes2011. Diabetes Care 34 (Suppl 1 ), S11S61.
Bristol, N. (2011). NIH proposes new drug development centre. Lancet 377,
705706.
Brommage, R., Desai, U., Revelli, J.P., Donoviel, D.B., Fontenot, G.K., Dacosta, C.M., Smith, D.D., Kirkpatrick, L.L., Coker, K.J., Donoviel, M.S., et al.
(2008). High-throughput screening of mouse knockout lines identifies true
lean and obese phenotypes. Obesity (Silver Spring) 16, 23622367.
Conarello, S.L., Li, Z., Ronan, J., Roy, R.S., Zhu, L., Jiang, G., Liu, F., Woods,
J., Zycband, E., Moller, D.E., et al. (2003). Mice lacking dipeptidyl peptidase IV
are protected against obesity and insulin resistance. Proc. Natl. Acad. Sci.
USA 100, 68256830.
de Kleijn, D.P.V., Moll, F.L., Hellings, W.E., Ozsarlak-Sozer, G., de Bruin, P.,
Doevendans, P.A., Vink, A., Catanzariti, L.M., Schoneveld, A.H., Algra, A.,
et al. (2010). Local atherosclerotic plaques are a source of prognostic
biomarkers for adverse cardiovascular events. Arterioscler. Thromb. Vasc.
Biol. 30, 612619.
Deacon, C.F., Nauck, M.A., Toft-Nielsen, M., Pridal, L., Willms, B., and Holst,
J.J. (1995). Both subcutaneously and intravenously administered glucagonlike peptide I are rapidly degraded from the NH2-terminus in type II diabetic
patients and in healthy subjects. Diabetes 44, 11261131.
Cell Metabolism
Perspective
Deacon, C.F., and Holst, J.J. (2002). Dipeptidyl peptidase IV inhibition as an
approach to the treatment and prevention of type 2 diabetes: a historical
perspective. Biochem. Biophys. Res. Commun. 294, 14.
Lariviere, W.R., Chesler, E.J., and Mogil, J.S. (2001). Transgenic studies of
pain and analgesia: mutation or background genotype? J. Pharmacol. Exp.
Ther. 297, 467473.
Degn, K.B., Juhl, C.B., Sturis, J., Jakobsen, G., Brock, B., Chandramouli, V.,
Rungby, J., Landau, B.R., and Schmitz, O. (2004). One weeks treatment
with the long-acting glucagon-like peptide 1 derivative liraglutide (NN2211)
markedly improves 24-h glycemia and alpha- and beta-cell function and
reduces endogenous glucose release in patients with type 2 diabetes. Diabetes 53, 11871194.
Lee, H.K., Choi, E.B., and Pak, C.S. (2009). The current status and future
perspectives of studies of cannabinoid receptor 1 antagonists as anti-obesity
agents. Curr. Top. Med. Chem. 9, 482503.
Demuth, H.-U., Hinke, S.A., Pederson, R.A., and McIntosh, C.H. (2002).
Rebuttal to Deacon and Holst: Metformin effects on dipeptidyl peptidase
IV degradation of glucagon-like peptide-1 versus Dipeptidyl peptidase
inhibition as an approach to the treatment and prevention of type 2 diabetes:
a historical perspective. Biochem. Biophys. Res. Commun. 296, 229232.
Drucker, D.J. (2006). The biology of incretin hormones. Cell Metab. 3, 153165.
Drucker, D.J., and Goldfine, A.B. (2011). Cardiovascular safety and diabetes
drug development. Lancet 377, 977979.
Eng, J., Kleinman, W.A., Singh, L., Singh, G., and Raufman, J.P. (1992).
Isolation and characterization of exendin-4, an exendin-3 analogue, from
Heloderma suspectum venom. Further evidence for an exendin receptor on
dispersed acini from guinea pig pancreas. J. Biol. Chem. 267, 74027405.
Fineman, M.S., Bicsak, T.A., Shen, L.Z., Taylor, K., Gaines, E., Varns, A., Kim,
D., and Baron, A.D. (2003). Effect on glycemic control of exenatide (synthetic
exendin-4) additive to existing metformin and/or sulfonylurea treatment in
patients with type 2 diabetes. Diabetes Care 26, 23702377.
Food and Drug Administration (2006). Press Release: Pfizer Stops all
Torcetrapib Clinical Trials in Interest of Patient Safety, December 3, 2006.
http://www.FDA.gov.
Gaich, G., and Moller, D. (2011). New mechanisms and translational paradigms. In Translational Medicine and Drug Discovery, B.H. Littman and R.
Krishnan, eds. (Cambridge, UK: Cambridge University Press), pp. 89108.
Greenfield, J.R., Miller, J.W., Keogh, J.M., Henning, E., Satterwhite, J.H.,
Cameron, G.S., Astruc, B., Mayer, J.P., Brage, S., See, T.C., et al. (2009).
Modulation of blood pressure by central melanocortinergic pathways.
N. Engl. J. Med. 360, 4452.
Hansen, K.B., Knop, F.K., Holst, J.J., and Vilsbll, T. (2009). Treatment of type
2 diabetes with glucagon-like peptide-1 receptor agonists. Int. J. Clin. Pract.
63, 11541160.
Harrison, C. (2011). Patent watch: the patent cliff steepens. Nat. Rev. Drug
Discov. 10, 1213.
Hirshberg, B., and Raz, I. (2011). Impact of the U.S. Food and Drug Administration cardiovascular assessment requirements on the development of novel
antidiabetes drugs. Diabetes Care 34 (Suppl 2 ), S101S106.
Hoerger, T.J., Segel, J.E., Gregg, E.W., and Saaddine, J.B. (2008). Is glycemic
control improving in U.S. adults? Diabetes Care 31, 8186.
Hu, X., Dietz, J.D., Xia, C., Knight, D.R., Loging, W.T., Smith, A.H., Yuan, H.,
Perry, D.A., and Keiser, J. (2009). Torcetrapib induces aldosterone and cortisol
production by an intracellular calcium-mediated mechanism independently of
cholesteryl ester transfer protein inhibition. Endocrinology 150, 22112219.
IMIDIA. (2011). European compined excellence in diabetes research (http://
www.imidia.org).
International Diabetes Federation (2011). IDF Disease Atlas (http://www.
diabetesatlas.org).
Jeong, K.-H., and Yoo, B.K. (2011). The efficacy and safety of liraglutide. Int. J.
Clin. Pharmacol. 33, 740749.
Kola, I. (2008). The state of innovation in drug development. Clin. Pharmacol.
Ther. 83, 227230.
Kreymann, B., Williams, G., Ghatei, M.A., and Bloom, S.R. (1987). Glucagonlike peptide-1 7-36: a physiological incretin in man. Lancet 2, 13001304.
Lankas, G.R., Leiting, B., Roy, R.S., Eiermann, G.J., Beconi, M.G., Biftu, T.,
Chan, C.C., Edmondson, S., Feeney, W.P., He, H., et al. (2005). Dipeptidyl
peptidase IV inhibition for the treatment of type 2 diabetes: potential importance
of selectivity over dipeptidyl peptidases 8 and 9. Diabetes 54, 29882994.
Lee, J.A., Chu, S., Willard, F.S., Cox, K.L., Sells Galvin, R.J., Peery, R.B.,
Oliver, S.E., Oler, J., Meredith, T.D., Heidler, S.A., et al. (2011). Open innovation for phenotypic drug discovery: The PD2 assay panel. J. Biomol. Screen.
16, 588602.
Libby, P., Ridker, P.M., and Hansson, G.K. (2011). Progress and challenges in
translating the biology of atherosclerosis. Nature 473, 317325.
Lund, P.K., Goodman, R.H., Dee, P.C., and Habener, J.F. (1982). Pancreatic
preproglucagon cDNA contains two glucagon-related coding sequences
arranged in tandem. Proc. Natl. Acad. Sci. USA 79, 345349.
McCarthy, M.I. (2010). Genomics, type 2 diabetes, and obesity. N. Engl. J.
Med. 363, 23392350.
Mentlein, R., Gallwitz, B., and Schmidt, W.E. (1993). Dipeptidyl-peptidase IV
hydrolyses gastric inhibitory polypeptide, glucagon-like peptide-1(7-36)
amide, peptide histidine methionine and is responsible for their degradation
in human serum. Eur. J. Biochem. 214, 829835.
Milne, C.-P. (2008). Fixing the paradigm for biopharmaceutical R&D: Where to
start? Int. J. Biotechnol. 10, 404415.
Milne, C.-P. (2009). Can translational medicine bring us out of the R&D wilderness? Personalized Medicine 6, 543553.
Mullard, A. (2011). 2010 FDA drug approvals. Nat. Rev. Drug Discov. 10, 8285.
Nauck, M.A., Kleine, N., Orskov, C., Holst, J.J., Willms, B., and Creutzfeldt, W.
(1993). Normalization of fasting hyperglycaemia by exogenous glucagon-like
peptide 1 (7-36 amide) in type 2 (non-insulin-dependent) diabetic patients.
Diabetologia 36, 741744.
Nauck, M.A. (2011). Incretin-based therapies for type 2 diabetes mellitus:
properties, functions, and clinical implications. Am. J. Med. 124 (1, Suppl),
S3S18.
Pammolli, F., Magazzini, L., and Riccaboni, M. (2011). The productivity crisis in
pharmaceutical R&D. Nat. Rev. Drug Discov. 10, 428438.
Paul, S.M., Mytelka, D.S., Dunwiddie, C.T., Persinger, C.C., Munos, B.H.,
Lindborg, S.R., and Schacht, A.L. (2010). How to improve R&D productivity:
the pharmaceutical industrys grand challenge. Nat. Rev. Drug Discov. 9,
203214.
Pauly, R.P., Demuth, H.-U., Rosche, F., Schmidt, J., White, H.A., Lynn, F.,
McIntosh, C.H., and Pederson, R.A. (1996). Inhibition of dipeptidyl peptidase
IV (DP IV) in rat results in improved glucose tolerance. Regul. Pept. 64, 148.
Schadt, E.E. (2009). Molecular networks as sensors and drivers of common
human diseases. Nature 461, 218223.
Sloop, K.W., Willard, F.S., Brenner, M.B., Ficorilli, J., Valasek, K., Showalter,
A.D., Farb, T.B., Cao, J.X., Cox, A.L., Michael, M.D., et al. (2010). Novel small
molecule glucagon-like peptide-1 receptor agonist stimulates insulin secretion
in rodents and from human islets. Diabetes 59, 30993107.
Subbarayan, S., and Kipnes, M. (2011). Sitagliptin: a review. Expert Opin.
Pharmacother. 12, 16131622.
Swinney, D.C., and Anthony, J. (2011). How were new medicines discovered?
Nat. Rev. Drug Discov. 10, 507519.
Gerstein, H.C., Miller, M.E., Genuth, S., Ismail-Beigi, F., Buse, J.B., Goff, D.C.,
Jr., Probstfield, J.L., Cushman, W.C., Ginsberg, H.N., Bigger, J.T., et al;
ACCORD Study Group. (2011). Long-term effects of intensive glucose
lowering on cardiovascular outcomes. N. Engl. J. Med. 364, 818828.
Wang, T.J., Larson, M.G., Vasan, R.S., Cheng, S., Rhee, E.P., McCabe, E.,
Lewis, G.D., Fox, C.S., Jacques, P.F., Fernandez, C., et al. (2011). Metabolite
profiles and the risk of developing diabetes. Nat. Med. 17, 448453.
Yang, Y., Adelstein, S.J., and Kassis, A.I. (2009). Target discovery from data
mining approaches. Drug Discov. Today 14, 147154.
Cell Metabolism
Article
Srf-Dependent Paracrine Signals Produced
by Myofibers Control Satellite Cell-Mediated
Skeletal Muscle Hypertrophy
Aline Guerci,1,2,8 Charlotte Lahoute,1,2,4,8 Sophie Hebrard,1,2,4 Laura Collard,1,2,4 Dany Graindorge,1,2,4
Maryline Favier,1,2,4 Nicolas Cagnard,3 Sabrina Batonnet-Pichon,1,2,4 Guillaume Precigout,5,6,7 Luis Garcia,5,6,7
David Tuil,1,2,4 Dominique Daegelen,1,2,4 and Athanassia Sotiropoulos1,2,4,*
1Inserm
U1016, Institut Cochin, F-75014 Paris, France

UMR8104, F-75014 Paris, France
3Bioinformatic Platform, Universite
Paris Descartes, F-75006 Paris, France
4Universite
Paris Descartes, F-75006 Paris, France
5Inserm U974, Institut de Myologie, F-75013 Paris, France
6CNRS, UMR7215, F-75013 Paris, France
7Universite
Pierre et Marie Curie, F-75005 Paris, France
8These authors contributed equally to this work
*Correspondence: athanassia.sotiropoulos@inserm.fr
DOI 10.1016/j.cmet.2011.12.001
2CNRS
SUMMARY
Adult skeletal muscles adapt their fiber size to workload. We show that serum response factor (Srf) is
required for satellite cell-mediated hypertrophic
muscle growth. Deletion of Srf from myofibers and
not satellite cells blunts overload-induced hypertrophy, and impairs satellite cell proliferation and
recruitment to pre-existing fibers. We reveal a gene
network in which Srf within myofibers modulates
interleukin-6 and cyclooxygenase-2/interleukin-4 expressions and therefore exerts a paracrine control of
satellite cell functions. In Srf-deleted muscles, in vivo
overexpression of interleukin-6 is sufficient to restore
satellite cell proliferation but not satellite cell fusion
and overall growth. In contrast cyclooxygenase2/interleukin-4 overexpression rescue satellite cell
recruitment and muscle growth without affecting
satellite cell proliferation, identifying altered fusion
as the limiting cellular event. These findings unravel
a role for Srf in the translation of mechanical cues
applied to myofibers into paracrine signals, which in
turn will modulate satellite cell functions and support
muscle growth.
INTRODUCTION
Adult skeletal muscle is a highly plastic tissue, the mass of which
changes in response to environmental cues and physiological
stimuli. The basic cellular building blocks of adult muscle are
the multinucleated myofibers, which are able to grow during
postnatal growth, during regeneration after injury, and in
response to a functional demand, such as external loads.
Mature myofibers can grow following the addition of new contractile proteins to pre-existing sarcomere units, which involve
the phosphatidylinositol 3-kinase PI3K/Akt signaling pathway.

Anabolic hormones, such as insulin-like growth factor 1
(IGF1), induce myofiber hypertrophy by activating the Akt1
pathway and the translational machinery through mTOR,
S6K1, and 4E-BP1 (Glass, 2010). Postmitotic myofiber growth
can also result from the accretion of new nuclei provided by
muscle stem cells called satellite cells. Once activated in
response to the adaptive requirements of the muscle, satellite
cells go through an ordered series of events, including proliferation, migration, and fusion to growing myofibers (Le Grand and
Rudnicki, 2007).
A large set of growth factors and cytokines modulate the proliferation and differentiation of satellite cells (Kuang et al., 2008).
Among these, interleukin 6 (Il6), a myokine detected at high
concentrations in contracting muscle fibers and after increased
load (Carson et al., 2002; Penkowa et al., 2003), enhances satellite cell proliferation and migration during muscle hypertrophy
(Serrano et al., 2008). Muscle-secreted interleukin 4 (Il4)
promotes muscle regeneration and postnatal growth by facilitating the migration of myoblasts (Lafreniere et al., 2006) and
the fusion of myoblasts to nascent myotubes (Horsley et al.,
2003). Prostaglandins produced by cyclooxygenase (Cox)
enzymes, which catalyze the rate-limiting step in their synthesis,
are bioactive lipid mediators that can also regulate satellite cell
behavior. In particular, Cox2 isoform activity and Cox2-derived
prostaglandins modulate satellite cell proliferation, migration,
and fusion (Bondesen et al., 2007; Otis et al., 2005; Shen et al.,
2006).
Although significant progress has been made in understanding the signaling pathways that control muscle mass, the
molecules that sense muscle overload and translate it into
signals that support hypertrophy are unclear. Furthermore,
very little is known about the transcription factors and target
genes that are involved in promoting adult muscle growth.
Serum response factor (Srf) is a ubiquitously expressed
transcription factor that binds the CArG box sequence. One
major class of Srf targets is expressed specifically in muscles
and comprises several genes encoding sarcomeric proteins
Cell Metabolism 15, 2537, January 4, 2012 2012 Elsevier Inc. 25
Cell Metabolism
Control of Skeletal Muscle Hypertrophy by Srf
Figure 1. Srf Loss within Myofiber Impairs Overload-Induced Muscle Hypertrophy

Control and Srf-deleted (mutant) plantaris muscles were
obtained from nonoverloaded mice (c) and after 7 and
21 days of compensatory hypertrophy (CH).
(A) Srf protein was analyzed by western blotting. Gapdh
was used as a loading control.
(B) Srf mRNA expression was analyzed by qRT-PCR in
plantaris muscles from control and mutant mice before (c)
and after 7 days of CH. Data (mean SEM) were
normalized by 18S rRNA expression. *p < 0.05 versus
controls (c); xp < 0.05 versus controls (7).
(C) Muscle sections immunostained for dystrophin and
nuclear staining with DAPI from control and mutant mice
before (c) and after 21 days of CH.
(D) Ratio of plantaris mass (mg) to body weight (g) before
(c) and after 7 and 21 days of CH in control and mutant
mice. Data are mean SEM. *p < 0.01 versus controls (c).
(E) Mean CSA ( SEM) of muscle fibers before (c) and after
7 and 21 days of CH in control and mutant mice. *p < 0.05
versus c.
(F) Mean myofiber number ( SEM) before (c) and after
7 and 21 days of CH in control and mutant mice.
(Pipes et al., 2006). The Rho family of small GTPases and actin
dynamics has been shown to modulate Srf activity by controlling
the nuclear accumulation of its coactivator, myocardin-related
transcription factor A (Mrtf-A) (Miralles et al., 2003). Data obtained from mouse models with skeletal muscle-specific loss
of Srf or Mrtfs functions emphasize their crucial role in postnatal
muscle growth (Charvet et al., 2006; Li et al., 2005). In adults, Srf
activity could also be important for the control of skeletal muscle
growth, as suggested by increased Srf expression during overload-induced hypertrophy (Fluck et al., 1999). In cardiac muscle,
Srf and Mrtf-A are necessary for mediating cardiomyocyte
hypertrophy (Kuwahara et al., 2010; Parlakian et al., 2005). Taken
together, these data point toward Srf as a good candidate transcription factor in the control of skeletal muscle mass during
hypertrophy.
To examine the role of Srf during overload-induced muscle
hypertrophy, and the underlying cellular and molecular mechanisms, we used a mouse model of conditional and inducible
deletion of Srf within myofibers but not in satellite cells (Lahoute
et al., 2008). We show here that plantaris myofibers lacking Srf
are no longer able to display a hypertrophic response when
subjected to an experimental overload, thus revealing the
requirement of Srf for muscle growth response to increased
26 Cell Metabolism 15, 2537, January 4, 2012 2012 Elsevier Inc.
activity. Using a combination of in vivo and in vitro assays, we demonstrate that this hypertrophic growth defect is due to lack of satellite
cell proliferation and fusion to the pre-existing
myofibers, independent of the IGF1/Akt
pathway. We show that impaired production of
secreted factors by myofibers lacking Srf is
responsible for these defects. We also unraveled a gene network associating Srf, Il6, Il4,
and Cox2 genes, which drive overload-induced
muscle growth. By modulating Il6, Cox2/Il4
expression levels in the myofibers, Srf controls
both satellite cell proliferation and fusion. Altogether, these data identify Srf as a crucial transcription factor
required for the translation of mechanical cues applied to the
myofibers into paracrine signals that support the proliferation
and fusion of satellite cells, leading to muscle growth.
RESULTS
Srf Is Required for Overload-Induced Hypertrophy
To investigate the role of Srf in skeletal muscle hypertrophy,
compensatory hypertrophy (CH) of the plantaris muscle was
performed in HSA-Cre-ERT2:Srfflox/flox mutant mice previously
tamoxifen-injected to induce myofiber-specific Srf loss. In addition, control and mutant mice were injected with tamoxifen
during the CH procedure. Efficient Srf loss was achieved at the
protein (Figure 1A) and transcript levels (Figure 1B). As previously
described, no obvious differences in muscle mass, myofiber
cross-sectional area (CSA), and myofiber number were observed between control and mutant plantaris muscles before
overload (Figures 1C1F, and Lahoute et al., 2008). In this
muscle growth model, we observed a significant increase
in muscle mass (Figure 1D) and the myofibers CSA of control
muscles 21 days after overload (Figure 1E), which was not
accompanied by a modification of fiber number (Figure 1F). In
Cell Metabolism
Figure 2. Srf Is Not Required for IGF1/Akt-Dependent muscle growth

(A) Phosphorylated Akt and total Akt proteins were
analyzed by western blotting from muscle samples obtained from nonoverloaded mice (c) and after 7 and
21 days of compensatory hypertrophy (CH). Gapdh was
used as a loading control.
(B) Myotubes (Srfflox/flox) were transduced with AdGFP or
AdCreGFP (Srf deleted) adenoviruses; 48 hr later they
were treated or not with recombinant 250 ng/ml IGF1
(rIGF1) for an additional 48 hr. Srf protein expression was
analyzed by western blotting. Gapdh was used as
a loading control.
(C) Myotube diameter was measured after AdCreGFP or
AdGFP transduction and after treatment with rIGF1 or not
(). Data are mean SEM. *p < 0.05 versus not treated ().
(D) Tibialis muscles were electroporated with control (Ctl)
or Myr-Akt plasmids and isolated 10 days later. Immunostaining for P-Akt (green) and nuclear staining with DAPI
illustrated Myr-Akt electroporation. Numbers of DAPIstained nuclei within the dystrophin-positive sarcolemma
were counted in nonelectroporated (Nonelec) and MyrAkt-electroporated fibers and are expressed per fiber.
Ratio of fiber CSA in electroporated (with Ctl or Myr-Akt
plasmids) or versus nonelectroporated fibers was
measured in control and mutant tibialis muscles. Data are
mean SEM. *p < 0.05 versus Ctl plasmid.
contrast, growth was completely blunted in mutant muscles

(Figures 1C1E), showing that Srf is necessary for overloadinduced myofiber hypertrophy.
Srf Is Dispensable for IGF1/Akt-Dependent
Muscle Growth
We investigated the possibility of an alteration in Akt signaling
in mutant muscle. As illustrated in Figure 2A, we observed no
difference in activated Akt levels between control and mutant
muscles at the different times examined after overloading, indicating that Srf deletion in the myofiber does not affect Akt
signaling and that the lack of growth observed in mutant muscle
is not the consequence of altered Akt signaling.
We then determined whether Srf is needed for the recruitment
of new nuclei to the growing myofiber and/or an increase in the
cytoplasmic volume. Having previously shown that myotubes
treated with IGF1 increase in size with no associated increase
of their nuclei number (Ohanna et al., 2005), we assessed
whether myotubes lacking Srf were responsive to IGF1-induced
hypertrophy in culture. The loss of Srf was obtained by transducing myotubes (Srfflox/flox) with an adenovirus driving Cre
recombinase expression (AdCre) (Figure 2B). IGF1 treatment
led to a similar increase in myotube diameter in both Srf-
expressing and non-Srf-expressing myotubes

(Figure 2C). In addition, in vivo overexpression
of a constitutively active form of Akt1 (Myr-Akt)
can be used to induce muscle hypertrophy
involving myofiber volume increase without
satellite cell fusion (Blaauw et al., 2009). Accordingly, following in vivo gene transfer of a Myr-Akt
vector in control and mutant tibialis muscles, we
observed the same number of myonuclei in both
Myr-Akt-expressing and non-Myr-Akt-expressing myofibers
(Figure 2D, middle). Similar levels of hypertrophy were observed
in control and mutant Myr-Akt electroporated muscles, showing
that Srf is not required for in vivo Myr-Akt-induced hypertrophy
(Figure 2D, right). Taken together, these data demonstrate that
Srf is not required for IGF1/Akt-induced skeletal muscle hypertrophy, which depends exclusively on an increase in the cytoplasmic volume.
Srf Loss within Myofibers Impairs Satellite
Cell Functions
To analyze the consequences of Srf loss within myofibers on
satellite cell proliferation in vivo during CH, the number of satellite cells expressing both Pax7 (a marker of satellite cells) and
Ki67 (a marker of cycling cells) was quantified. Before CH, the
number of quiescent satellite cells (Pax7+) was similar in control
and mutant plantaris muscles (Figure 3B, left). In contrast, 3 and
7 days after overloading, the number of Pax7+Ki67+ cells was
significantly lower in mutant muscles compared to controls,
showing that in the absence of Srf within myofibers, satellite cells
present a decreased rate of proliferation (Figure 3B, right).
Accordingly, 7 days post-CH, the number of Pax7+ cells was
diminished in mutant muscles (Figures 3A and 3B, left). This
Cell Metabolism
Figure 3. Srf-Dependent Paracrine Control of Satellite Cell Proliferation and Fusion

(A) Immunostaining for Pax7 (green) and Ki67 (red) to illustrate cycling satellite cells (Pax7+Ki67+, white arrow) after 7 days of CH in control or mutant plantaris
muscles.
(B) Numbers of Pax7+ cells per mm2 and Pax7+Ki67+ cells per mm2 were quantified in control and mutant plantaris muscle sections before (c) and after 3 and
7 days of CH. Data are mean SEM. *p < 0.05 versus controls (0); xp < 0.05 versus controls.
(C) Number of DAPI-stained nuclei within the dystrophin-positive sarcolemma was counted before (c) and after 7 and 21 days of CH and are expressed per fiber.
Data are mean SEM. *p < 0.05 versus (c); xp < 0.05 versus controls.
(D) Myotubes transduced with AdGFP (Ctl) or AdCreGFP (Mut) were labeled in green (cell tracker) and mixed with Ctl myocytes labeled in red (expressing nlslacZ).
After 48 hr of coculture, myotubes were analyzed for dual labeling. Representative picture of the dual labeling: immunostaining for LacZ and GFP/green cell
tracker (yellow arrow). Normalized percentage of fusion events: % of dual labeled myotubes/LacZ cells of myotubes transduced with AdGFP or AdCreGFP
cultured without () or with (+) conditioned medium (CM) from control myotubes. Data are mean SEM. *p < 0.05 versus AdeGFP(-CM), xp < 0.05 versus
AdCreGFP(-CM).
(E) Normalized percentage of BrdU+ cells in control myoblasts cultured with or without CM from myotubes transduced with AdGFP or AdCreGFP. Data are
mean SEM. *p < 0.05 versus CM from AdGFP.
(F) Immunostaining for Pax7 (green) and Ki67 (red) to illustrate satellite cells (Pax7+) and cycling satellite cells (Pax7+Ki67+) on single fibers. Normalized
percentage of cycling satellite cells (Pax7+Ki67+ over Pax7+ cells) on single fibers isolated from control or mutant EDL muscles cultured with or without CM from
control myotubes. Data are mean SEM. *p < 0.05 versus control, xp < 0.05 versus mutant.
Cell Metabolism
decreased proliferation was not accompanied by increased cell

death as assessed by TUNEL (Figure S1, available online). These
data demonstrate that the presence of Srf within the myofibers is
critical for stimulating efficient satellite cell proliferation in
response to overload.
To investigate whether the lack of Srf in myofibers might also
affect satellite cell fusion to mutant myofibers, myonuclei
numbers were determined in control and mutant plantaris
muscles at various time points post-CH. Although significant
increases in myonuclei numbers were observed in control
muscles at 7 and 21 days post-CH, no such increases were
observed in mutant muscles (Figure 3C). Because decreased
satellite cell proliferation in mutant muscles may impede new
nuclei accretion, we set up an in vitro assay to specifically assess
whether the absence of Srf within myotubes (green) affects
fusion with Srf-expressing mononucleated myocytes (nls-lacZ,
red) (Figure 3D, top). The lack of Srf expression in myotubes
was accompanied by a 50% decrease in fusion events (duallabeling, Figure 3D) compared to the control myotubes. These
data show that Srf plays an important role within the myofibers
in controlling satellite cell recruitment.
Altogether, the lack of growth of Srf deficient myofibers in
response to increased load could result from decreased satellite
cell proliferation and decreased ability to fuse to the growing
myofibers.
Srf-Dependent Paracrine Control of Myoblast
Proliferation and Fusion to Myotubes
In this genetic model, Srf deletion occurs in myofibers of mutant
muscles but never in quiescent satellite cells or in satellite cells
engaged in differentiation/fusion (myocytes). Indeed, satellite
cells isolated from mutant single myofibers (from tamoxifen-injected HSA-Cre-ERT2: Srfflox/flox mice) did not display an excision
of floxed Srf allele as opposed to the whole muscle (Figure S2A).
In addition, Srf deletion did not occur in fusing HSA-Cre-ERT2:
Srfflox/flox myocytes (day 1 of differentiation) treated with tamoxifen but was observed in terminally differentiated HSA-Cre-ERT2:
Srfflox/flox myotubes (day 5 of differentiation) after tamoxifen
addition (Figure S2B).
We further checked control and mutant muscle-growth
responses in conditions excluding Cre-ERT2 activation in myofibers and in satellite cells during the overloading process.
Because Cre-ERT2 protein is only active in the presence of
tamoxifen, control and mutant mice were injected with tamoxifen
only prior to and not during CH. The complete lack of hypertrophy observed in mutant muscles as compared to control
muscles 21 days post-CH reinforced our idea that the loss of
Srf only within myofibers is responsible for the absence of growth
of mutant muscles (Figure S2C).
We therefore hypothesized that during CH myofibers exert
a paracrine control of satellite cell proliferation and fusion via
Srf. The potential involvement of factors secreted from the
myofibers on satellite cell proliferation was first assessed
in vitro by determining the BrdU incorporation rate of control
myoblasts in response to conditioned medium (CM) issued
from both Srf-expressing and non-Srf-expressing myotubes.
CM from myotubes lacking Srf (AdCreGFP-transducted) stimulated myoblast proliferation to a much lesser extent than CM
from control myotubes (AdGFP) (Figure 3E). We further moni-
tored the percentage of cycling satellite cells (Pax7+Ki67+

over Pax7+) on single-fiber cultures from control and mutant
muscles cultivated in differentiation media (2% HS) or in
presence of CM issued from Srf-expressing myotubes. The
normalized proportion of cycling satellite cells was significantly
decreased in mutant single fibers in comparison with control
single fibers and was restored to control level in presence of
CM (Figure 3F). Thus, the presence of Srf within myofiber/
myotube is necessary for controlling the expression of secreted
factors that promote adequate satellite cell/myoblast proliferation in vitro.
Whether an alteration of the set of factors secreted by mutant
myofibers could be responsible for inefficient satellite cell fusion
was also investigated in vitro. CM from control myotubes
almost totally rescued the fusion defect of control myocytes to
myotubes lacking Srf (Figure 3D). These data suggested that
CM from control myotubes contained secreted factors that are
no longer produced in Srf-depleted myotubes.
Taken together, these findings highlight a previously undiscovered contribution of Srf within myofibers in controlling the
expression of secreted factors that are necessary for promoting
the proliferation of muscle satellite cells and their fusion to preexisting myofibers during CH.
To identify molecular Srf target genes that control satellite cell
functions, we performed a microarray analysis of gene expression. Two conditions were used: Srfflox/flox myotubes were transduced with either AdGFP or AdSRFVP16 driving the expression
of a constitutively active Srf derivative. Among the genes whose
expression was induced by SRFVP16, a certain number of
known Srf targets serving as positive controls for the suitability
of the approach were found (Table S1, indicated in italics).
Following qRT-PCR validation, attention was focused on genes
encoding secreted factors that were differentially expressed
between overloaded mutant and control muscles (data not
shown). This was the case for Il6, whose role in Srf-dependent
CH was tested as detailed hereafter.
Il6 Rescues Satellite Cell proliferation
but not Muscle Growth
Il6 has previously been shown to control satellite cell proliferation in a paracrine manner (Serrano et al., 2008). First, we investigate a possible Srf-dependent regulation of Il6 expression.
Although a 2.7-fold increase in Il6 expression in gain-offunction myotubes was observed 2 days post-AdSRFVP16
transduction, Il6 expression decreased by more than half in
the absence of Srf 3 days post-AdCreGFP transduction (Figure 4A). In vivo, 7 days post-overload, the level of Il6 transcripts
had greatly increased in control plantaris muscles, whereas
only a faint increase was detected in mutant muscles (Figure 4B). Taken together, these data reveal an impaired Il6
expression in growing muscles lacking Srf, which is intrinsic
to muscle myofibers.
To evaluate the contribution of decreased Il6 expression to the
lack of growth of Srf-deleted myofibers, control and mutant
plantaris muscles were injected with an adeno-associated
virus (AAV) driving Il6 expression (AAV-IL6) 3weeks prior to
CH. Plantaris muscles were isolated 7 and 14 days post-CH
(Figure 4C). A significant increase in Il6 expression was
measured in plantaris muscle injected with AAV-IL6 as
Cell Metabolism
Figure 4. Il6 Overexpression Rescues Satellite Cell

Proliferation but Not Overload-Induced Muscle
Growth
(A) Il6 mRNA expression was analyzed by qRT-PCR in
myotubes (SRFflox/flox) transduced with AdGFP or AdCreGFP for 72 hr or with AdSRFVP16 for 48 hr. Data
(mean SEM) are normalized by cyclophilin expression
and presented as fold-induction relative to AdGFP.
*p < 0.05 versus AdGFP.
(B) Il6 mRNA expression was analyzed by qRT-PCR in
plantaris muscles from control and mutant mice before (c)
and after 7 days of CH. Data (mean SEM) are normalized
by 18S rRNA expression and as fold-induction relative to
controls (c). *p < 0.05 versus controls (c); xp < 0.05 versus
controls (7).
(C) Plantaris muscles from one leg of control and mutant
mice were injected with AAV-IL6 21 days prior to muscle
overloading procedure of both legs. Muscles were isolated after 7 days of CH for satellite cell proliferation and
fusion analyses, and after 14 days of CH for satellite cell
fusion and overall growth analyses.
(DF) Numbers of Pax7+ cells per mm2 (D), Pax7+Ki67+
cells per mm2 (E), and myonuclei per fiber (F) were quantified in control and mutant plantaris muscle sections injected (+) or not () with AAV-IL6 after 7 days of CH. Data
are mean SEM. *p < 0.05 versus controls (-); xp < 0.05
versus mutants ().
(G) Muscle sections immunostained for dystrophin and
mean CSA ( SEM) from control and mutant muscles
injected (+) or not () with AAV-IL6 after 14 days of CH.
*p < 0.05 versus controls ().
compared to non-injected muscles using qRT-PCR (Figure S3A).

Il6 overexpression did not affect the numbers of Pax7+ and
Pax7+Ki67+- proliferating satellite cells in control muscles
7 days post-CH, suggesting that the level of Il6 in control
muscles does not limit growth. Significantly, the numbers of
satellite cells (Pax7+, Figure 4D) and proliferating satellite cells
(Pax7+Ki67+, Figure 4E) increased to control levels in overloaded
mutant muscles overexpressing Il6, indicating a functional role
for decreased Il6 expression in the altered satellite cell proliferation of mutant CH muscles. In contrast, Il6 overexpression was
not able to rescue the altered fusion of satellite

cells to the growing Srf-deleted myofibers: at 7
and 14 days post-CH, the mean myonuclei
number was unchanged in mutant myofibers
and was still inferior to that of control myofibers
(Figure 4F and S3B). Finally, overall growth was
assessed by measuring the mean CSA of
myofibers in AAV-IL6-injected and non-AAVIL6-injected muscles 14 days post-CH. Il6
overexpression had no effect on the growth
of control and mutant myofibers, and the
mean CSA of myofibers lacking Srf remained
lower than that of controls (Figure 4G). Taken
together, these data demonstrate that Il6 overexpression was able to rescue in vivo satellite
cell proliferation in mutant muscles but not
satellite cell fusion and myofiber growth. Therefore, decreased satellite cell proliferation is not
the major limiting step for hypertrophy in response to overload
in Srf-deleted muscles.
Il4 Restores Satellite Cell Fusion and In Vivo
Muscle Growth
To discern whether altered satellite cell recruitment could determine the lack of hypertrophy of Srf-deleted myofibers, the genes
involved must be identified. Il4 is a good candidate because 1) it
has previously been shown to control fusion of myoblasts to
nascent myotubes during postnatal growth (Horsley et al.,
Cell Metabolism
Figure 5. Il4 Overexpression Restores Satellite Cell Fusion and Muscle Growth
(A) Il4 mRNA expression was analyzed by qRT-PCR in myotubes (SRFflox/flox) transduced with AdGFP or AdCreGFP for 72 hr or with AdSRFVP16 for 48 hr.
Data (mean SEM) are normalized by cyclophilin expression and presented as fold-induction relative to AdGFP. *p < 0.05 versus AdGFP.
(BD) Numbers of Pax7+ cells per mm2 (B), Pax7+Ki67+ cells per mm2 (C), and myonuclei per fiber (D) were quantified in control and mutant plantaris muscle
sections injected (+) or not () with AAV-IL4 after 7 days of CH. Data are mean SEM. *p < 0.05 versus controls (); xp < 0.05 versus mutants ().
(E) Muscle sections immunostained for dystrophin and mean CSA ( SEM) from control and mutant muscles injected (+) or not () with AAV-IL4 after 14 days
of CH. *p < 0.05 versus controls (); xp < 0.05 versus mutants ().
2003) and 2) the postnatal muscle-growth defect linked to skeletal muscle-specific loss of Srf has been correlated to decreased
Il4 expression (Charvet et al., 2006). In primary myotubes lacking
Srf (AdCreGFP-transduced for 3 days), Il4 expression is reduced
by 40% when compared to control myotubes (AdGFP) (Figure 5A). To determine whether decreased Il4 expression could
account for defective fusion of control myocytes to Srf-deleted
myotubes, in vitro fusion experiments were conducted using
CM issued from Il4/ myotubes and exogenous Il4. Importantly,
Il4 addition rescued fusion, whereas CM from Il4/ myotubes
did not ameliorate fusion (Figure S4). The implication of
decreased Il4 expression in the lack of growth in Srf-deleted
muscles was then addressed in vivo by injecting control and
mutant plantaris muscles with an AAV driving Il4 expression
(AAV-IL4) prior to CH. A robust Il4 overexpression was measured
in plantaris muscle injected with AAV-IL4 (Figure S5A). When Il4
was overexpressed in control muscles on day 7 post-CH, the
number of Pax7+ cells was unchanged and the number of
proliferating Pax7+Ki67+ satellite cells was lower, suggesting
that Il4 might drive satellite cells toward fusion and accelerate
the growth process. In mutant muscles, we observed that Il4
overexpression had no effect on the total number of Pax7+ or
Pax7+Ki67+ satellite cells (Figures 5B and 5C). Most interestingly, the impaired fusion of satellite cells to mutant myofibers
was fully rescued by Il4 overexpression, as assessed by count-
ing myonuclei numbers 7 and 14 days post-CH (Figures 5D

and S5B). Moreover, Il4 overexpression was able to rescue the
lack of response to overload in mutant muscle because mutant
plantaris muscle was now able to grow following an overloadinduced hypertrophy (Figure 5E). The 30% increase in size of
mutant myofibers expressing Il4 allowed mutant muscles to
reach the size of hypertrophied control muscles.
Taken together, these data show that Il4 is necessary for Srfcontrolled fusion in vitro and is sufficient to rescue both the
impaired satellite cell fusion of Srf mutant muscles and their
lack of hypertrophic growth in vivo without improving the
decreased proliferation of satellite cells. These results suggest
that satellite cell fusion to myofibers is the main limiting step
for growth occurring in myofibers lacking Srf. In addition, we
identify Il4 in vivo as a crucial secreted factor involved in the
paracrine control of satellite cell fusion during CH.
Cox2 Is a Direct Srf Target Gene and Controls
Il4 Expression
Whether Il4 could be a direct target gene of Srf, as found in the
muscle cell line C2C12 (Charvet et al., 2006), was further investigated through ChIP experiments. In contrast to the results in
C2C12 cells, none of the putative CArG boxes of Il4 promoter
bound Srf protein in cultured primary myotubes (Figure S6A).
However, Srf was bound to a-skeletal actin and myl9 promoters,
Cell Metabolism
indicating the efficacy of the assay. This discrepancy could be

attributed to the different cellular systems used in this study.
To further examine the potential control of Il4 transcription by
Srf, various Il4 promoter-reporter constructs were created and
the luciferase activity in the presence or absence of activated
Srf (SRFVP16) was analyzed. SRFVP16 only marginally activated
pIL4(1670) reporter construct, which contains two of the
putative Srf binding sites, and pIL4(838) reporter, which lacks
putative CArG boxes (Figure S6B). In line with these results,
endogenous Il4 expression was not modulated in gain-of-function myotube samples in a short time point post-AdSRFVP16
transduction (2 days) which favors the regulation of direct Srf
target genes rather than secondary targets (Figure 5A). Taken
together, these data argue against a direct regulation of Il4 by
Srf in primary myotubes.
We looked for potential candidate genes whose expression
could be directly regulated by Srf and potentially linked to Il4
expression in the list of genes upregulated upon Srf activation
(Table S1). Attention was focused on the Cox2 gene, whose
activity has been reported to be required for skeletal muscle
hypertrophy (Novak et al., 2009). A common signaling pathway,
the calcineurin/NFAT pathway, has previously been implicated in
the control of fusion by Cox-derived prostaglandins and in the
regulation of Il4 expression in muscle cells (Horsley et al.,
2003; Horsley and Pavlath, 2003). We first confirmed by qRTPCR that Cox2 expression is greatly increased in myotubes
transduced with AdSRFVP16, whereas a significant decrease
of Cox2 expression in myotubes lacking Srf was observed (Figure 6A). Interestingly, an increase of Cox2 expression in control
muscles, which was significantly blunted in Srf-deleted muscles,
was observed 7 days post-CH (Figure 6B), thus reinforcing the
potential role of Cox2 in Srf-dependent growth. To decipher
whether Cox2 could be a direct target of Srf, Cox2 reporter
plasmids were constructed and we showed that SRFVP16
increased the activation of pCox2(2570) reporter by 3.5-fold.
In contrast, the pCox2(2250) reporter was activated significantly less, suggesting the presence of a Srf-responding region
in the portion 2570 to 2250 of the Cox2 promoter (Figure 6C),
which contains a putative CArG motif at position 2320. In ChIP
experiments performed in primary myotubes, the Cox2 promoter
covering position 2320 was robustly amplified from Srf immunoprecipitates compared to IgG controls, demonstrating Srf
binding (Figure 6D). Together, these results suggest that Cox2
expression is directly regulated by Srf through binding to
a CArG motif at position 2320 of the Cox2 promoter.
To gain insight into the possible regulation of Il4 expression
by Cox2, primary myotubes were treated with a specific Cox2inhibitor (SC-791) and Il4 transcripts were quantified. Cox2
inhibition led to a 40% decrease of Il4 expression in a doseresponse manner (Figure 6E). This effect seemed to be specific
to Il4 because Il6 and Srf expression levels were unaffected
under the same conditions (data not shown). More importantly,
in transfection experiments, Cox2 overexpression was able
to increase the activation of the pIL4(1730) reporter more
than 5-fold (Figure 6F). In addition, both treatments with cyclosporin (CsA), a potent inhibitor of calcineurin, and coexpression
of VIVIT, a NFAT inhibitory peptide, significantly decreased
Cox2-induced activation of the pIL4(1730) reporter, showing
that at least part of this activation is dependent upon the
calcineurin/NFAT pathway (Figure 6F). To investigate whether

Cox2 could regulate Il4 expression in vivo, AAV driving Cox2
expression (AAV-Cox2) was injected in plantaris muscle. AAVCox2 injection strongly increased Cox2 expression (Figure S7A).
Post-CH, the overexpression of Cox2 increased by 4.5-fold the
amount of endogenous Il4 transcripts, whereas Il6 transcript
levels were unaffected (Figure 6G). Overall, these data suggest
that Cox2 positively regulates Il4 gene transcription in skeletal
muscle.
Cox2 Rescues Satellite Cell Fusion and Overall
Muscle Growth
If direct control of Cox2 expression by Srf mediates muscle
growth through Il4, it is expected that Cox2 overexpression
in vivo will rescue Srf-deleted myofiber growth in a similar
manner to Il4. Therefore, control and mutant plantaris muscles
were injected with AAV-Cox2 prior to CH. Cox2 overexpression
had no effect on satellite cell proliferation and fusion in control
muscles. Significantly, after overload-induced hypertrophy of
mutant muscles, Cox2 overexpression did not influence the
number of proliferating satellite cells (Figures 7A and 7B) but still
effectively rescued to control levels in both satellite cell fusion
(Figure 7C and S7B) and overall growth of Srf-deleted myofibers.
Indeed, mean CSA of mutant plantaris muscles injected with
AAV-Cox2 was comparable to control muscles (Figure 7D)
and, thus, display a rescue of growth similar to that obtained
by Il4 overexpression (Figure 5E).
These data show a previously unknown Srf-dependent paracrine control of muscle growth by Cox2/Il4 within myofibers.
DISCUSSION
Taking advantage of a genetic model that allows the invalidation
of Srf in myofibers and not in satellite cells, we demonstrate
that the recruitment of satellite cells to the myofibers is a limiting
step for physiological hypertrophy in adult Srf-deleted muscles.
We provide evidence for a gene network operating in myofibers
during overload-induced muscle growth in which Srf modulates
Il6 and Cox2/Il4 expression levels, which control satellite cell
proliferation and fusion, respectively. We propose that, within
overloaded myofibers, Srf acts as a sensor for mechanical
cues that translates them into paracrine signals, which in turn
regulate satellite cell functions and support muscle growth
(Figure 7E).
Two nonexclusive mechanisms are involved in muscle growth:
1) the increase of net protein content through the activation of
Akt/TOR pathway and 2) the addition of satellite cell-derived
myonuclei to the adult myofiber. Our data show that the lack of
recruitment of satellite cells is responsible for the absence of
compensatory hypertrophy in Srf-deleted muscles. Indeed, an
absence of Srf does not compromise Akt signaling and Srf is
dispensable for Akt-driven muscle growth that does not involve
satellite cells. Using in vivo and in vitro experiments, we further
deciphered the Srf-dependent molecular and cellular events
that could be implicated in overload-induced hypertrophy. Our
data show that, in response to overload, Srf within myofibers
controls both satellite cell proliferation and fusion in a paracrine
manner. Srf promotes the production of extracellular factors (Il6
and Il4) by the myofiber and, thus, controls the satellite cell local
Cell Metabolism
Figure 6. Cox2 Is a Direct Target of Srf and Vontrols Il4 Expression

(A) Cox2 mRNA expression was analyzed by qRT-PCR in myotubes (SRFflox/flox) transduced with AdGFP or AdCreGFP for 72 hr or with AdSRFVP16 for 48 hr. Data
(mean SEM) are normalized by cyclophilin expression and presented as fold-induction relative to AdGFP. *p < 0.05 versus AdGFP.
(B) Cox2 mRNA expression was analyzed by qRT-PCR in plantaris muscles from control and mutant mice before (c) and after 7 days of CH. Data (mean SEM) are
normalized by 18S rRNA expression and as fold-induction relative to controls (c). *p < 0.05 versus controls (c); xp < 0.05 versus controls (7).
(C) Deletion analysis of Cox2 promoter fragments ranging from 2570 to +1, cloned in front of a luciferase reporter and cotransfected with SRFVP16.
Putative CArG box is indicated. Shown is the mean of relative luciferase activity ( SEM) normalized to Renilla luciferase. *p < 0.05 versus pCox2(2570)
SRFVP16.
(D) ChIP performed using myotubes and antibodies specific to Srf or IgG. Bound Cox2 promoter was amplified by qRT-PCR and normalized to input and to an
additional negative control (primers spanning the first intron of Il4). Data are mean SEM. *p < 0.05 versus IgG.
(E) Cox2 mRNA expression was analyzed by qRT-PCR in myotubes (SRFflox/flox) treated with Cox2 inhibitor (SC-791) for 48 hr. Data (mean SEM) are normalized
by cyclophilin expression and as fold-induction relative to untreated. *p < 0.05 versus (0).
(F) Responsiveness of pIL4(1670) luciferase reporter construct to Cox2 treated with calcineurin inhibitor (CsA) or cotransfected with a NFAT inhibitor (VIVIT).
Data are mean SEM. *p < 0.05 versus pIL4(1670) alone, xp < 0.05 versus pIL4(1670) plus Cox2.
(G) Il4 and Il6 mRNA expressions were analyzed by qRT-PCR in control plantaris muscles injected (+) or not () with AAV-Cox2 after 7 days of CH. Data
(mean SEM) are normalized by cyclophilin expression and as fold-induction relative to ().*p < 0.05 versus ().
Cell Metabolism
Figure 7. Cox2 Recues Satellite Cell Fusion and Muscle Growth

(AC) Numbers of Pax7+ cells per mm2 (A), Pax7+Ki67+ cells per mm2 (B) and myonuclei per fiber (C) were quantified in control and mutant plantaris muscle
sections injected (+) or not () with AAV-Cox2 after 7 days of CH. Data are mean SEM. *p < 0.05 versus controls (); xp < 0.05 versus mutants ().
(D) Muscle sections immunostained for dystrophin and mean CSA ( SEM) from control and mutant muscles injected (+) or not () with AAV-Cox2 after 14 days of
CH. *p < 0.05 versus controls (); xp < 0.05 versus mutants ().
(E) Schematic model: in response to increased workload, Srf within myofibers modulates Il6 and Cox2/Il4 expressions and, therefore, exerts a paracrine control of
satellite cell proliferation and fusion, respectively, which in turn support skeletal muscle hypertrophy.
milieu in conditions of disturbed muscle homeostasis. Thus,

coordinated regulation by Srf of both satellite cell proliferation
and fusion ensures proper muscle growth in response to
increased load. We verified that Il6 and Il4 do not stimulate the
hypertrophy of fully differentiated myotubes in an autocrine
manner because myotubes treated with Il6 or Il4 did not display
an increase in size (data not shown).
It is notable that reduced proliferation of satellite cells is not the
limiting step responsible for the lack of hypertrophy of muscles
deleted for Srf because the normalization of satellite cell prolifer34 Cell Metabolism 15, 2537, January 4, 2012 2012 Elsevier Inc.
ation by Il6 overexpression did not improve satellite cell fusion

and was unable to drive muscle hypertrophy. In contrast, the
restoration of satellite cell fusion by either Il4 or Cox2 overexpression was sufficient to rescue overload-induced growth of
muscles lacking Srf. Interestingly, this rescue was obtained
without any increase in satellite cell proliferation, showing that
satellite cell fusion was the major limiting cellular event. The
number of cycling satellite cells remaining in Srf-deleted muscles
(50%) was probably sufficient to support growth when fusion is
restored. This situation differs from that reported for Il6/
Cell Metabolism
mice, in which reduced muscle compensatory hypertrophy was

attributed mainly to a defect in satellite cell proliferation (Serrano
et al., 2008). In that case, satellite cell proliferation was more
profoundly affected (with only 30% of cycling satellite cells left)
than muscles lacking Srf. Our data support an important role
for satellite cell recruitment in activity-induced hypertrophy of
Srf-deleted muscles and are in line with a recent study showing
that increased fiber size during compensatory hypertrophy is
preceded by the addition of nuclei and that this constitutes the
major cause of hypertrophy (Bruusgaard et al., 2010). Interestingly, overload-induced muscle hypertrophy can still take place
in the absence of satellite cells (McCarthy et al., 2011). The
compensatory mechanism allowing growth in satellite celldepleted skeletal muscle may be impaired in our model due to
the lack of Srf expression in myofibers.
While searching for mechanisms underlying such defects in
satellite cell recruitment, the Srf gain of function transcriptome
study led us to focus our interest on the Cox2 gene. Here, we
show that Cox2 expression increases strongly in response to
muscle overload in a Srf-dependent manner, and also show
that Cox2 is a direct target gene of Srf in differentiated muscle
cells. Significantly, the sole restoration of Cox2 expression in myofibers lacking Srf was able to rescue satellite cell recruitment to
the growing fiber and to subsequent muscle growth. This study
constitutes the first in vivo demonstration of the specific implication of Cox2 activity within the myofiber for the control of satellite
cell fusion during overload-induced hypertrophy. Interestingly,
Cox2 activity has been shown to be required for efficient muscle
regeneration (Bondesen et al., 2004; Shen et al., 2006), recovery
from atrophy (Bondesen et al., 2006), and overload-induced
hypertrophy (Novak et al., 2009). However, in this last case, the
blunting of muscle growth by a specific Cox2 inhibitor was attributed to reduction of macrophage accumulation, extracellular
protease activity, and overall cell proliferation. The effect on
proliferation was relatively modest and did not discriminate
satellite cells from other cell types. In our experimental procedure of hypertrophy, inflammation did not appear to be
differentially modulated between mutant and control muscles
as assessed by quantifying macrophage-specific genes expression levels and F4/80 positive cells on muscle sections
(data not shown). Moreover, our present data argue against
paracrine control of satellite cell proliferation by Cox2 during
hypertrophy because Cox2 overexpression in Srf-deleted myofibers failed to rescue satellite cell proliferation.
How could a Srf-dependent increase of Cox2 expression in
myofibers control satellite cell fusion during overload? We
propose that Cox2 and Il4 genes are linked. Here, we show for
the first time the implication of Il4 during overload-induced
muscle hypertrophy. As for Cox2, in vivo overexpression of Il4
in Srf-deleted myofibers was sufficient to restore satellite cell
fusion and muscle growth. However, in contrast to Cox2, Il4
does not appear to be a direct target gene of Srf in myotubes
derived from primary cultures. Significantly, our data clearly
demonstrate that Cox2 controls Il4 expression. Indeed, inhibition
of Cox2 activity downregulates endogenous Il4 expression in
cultured myotubes, whereas Cox2 overexpression activates
the Il4 promoter. Moreover, AAV-driven Cox2 overexpression
was sufficient to increase the amount of endogenous Il4 transcripts in vivo. These data constitute the first report showing
the upregulation of Il4 expression by Cox2 in muscle cells.

Thus, Il4 could mediate at least some of the action of Cox2 on
satellite cell recruitment during muscle overload. Furthermore,
we show that the calcineurin/NFAT pathway relays in part the
increase of Il4 expression by Cox2 activity. Because Cox2derived prostaglandin PGF2a has been shown to control myoblast fusion with nascent myotubes in a NFATc2-dependent
manner (Horsley and Pavlath, 2003) and Il4 expression has
been shown to be controlled by NFATc2 (Horsley et al., 2003),
NFATc2 could be the transcription factor that links Cox2 to Il4
activation.
The adaptation of skeletal muscle to external mechanical
stress, such as increased loading, requires initial sensing of the
stress by myofibers, followed by its transduction into signals
that will generate the appropriate physiological response. We
show that, within myofibers, Srf is required for muscle growth
in response to increased loading but is dispensable for MyrAkt-induced muscle hypertrophy, which occurs in the absence
of increased mechanical signals. Another difference between
these two growth models is that only overload-induced hypertrophy relies on the recruitment of satellite cells. Therefore, it
appears that Srf may be a central transcription factor required
for the translation of mechanical signals into a myofiber transcriptional program leading to local milieu modifications that
are interpreted by satellite cells and support muscle growth.
Several observations support the role of Srf as a sensor of
mechanical cues. In cardiac muscle cells, the nuclear localization of the Srf cofactor Mrtf-A is induced by mechanical stretch
and Mrtf-A/ mice show significantly attenuated mechanical
stress-induced cardiac hypertrophy (Kuwahara et al., 2010). In
human skeletal muscle, STARS (an activator of Srf by increasing
actin polymerization), MRTFs, and SRF expressions are upregulated following resistance training-induced hypertrophy (Lamon
et al., 2009), and Stars expression level is increased during
overload-induced muscle hypertrophy in mice (data not shown).
Collectively, these observations point to the Rho-actin-Mrtf-Srf
pathway as a signaling mechanism mediating mechanical
stress-induced gene expression and hypertrophic responses.
In this study, we show that, within myofibers, Srf is a key regulator of skeletal muscle hypertrophy in response to workload by
enhancing satellite cell proliferation and their subsequent fusion
to the growing fibers in a paracrine fashion through Il6 and
Cox2/Il4, respectively. Interestingly, hypertrophy induced by
overload is greatly attenuated in older animals (Alway et al.,
2002; Carson et al., 1995). We previously reported a decreased
expression of Srf in aged human and mouse muscles. Accordingly, loss of Srf within myofibers of young adult mice induced
premature aging in skeletal muscle (Lahoute et al., 2008). Therefore, during aging, there is a further link between Srf activity
and muscle hypertrophic capacities. Thus, the identification of
Srf as a master controller of physiological hypertrophy carries
potential significance in the search for muscle atrophy therapies
and treatments alleviating muscle aging.
EXPERIMENTAL PROCEDURES
Mouse Protocols
Mice homozygous for Srf floxed alleles (Srfflox/flox) and HSA-Cre-ERT2:
Srfflox/flox tamoxifen-dependent Cre recombinase premutant mice have been
Cell Metabolism
previously described (Lahoute et al., 2008). In all experiments, 2-month-old

pre-mutant and Srfflox/flox female mice were given intraperitoneal tamoxifen
(1 mg per day; Sigma) injections for 5 consecutive days and are referred as
mutants and controls, respectively.
Compensatory hypertrophy (CH) of plantaris muscles was induced as
described (Serrano et al., 2008). During the process of CH, control and mutant
mice were injected with tamoxifen every 3 days. A minimum of four animals of
each genotype were analyzed for every experiments. Control and mutant
plantaris muscles were injected with 7 3 1010 viral AAV genomes. Mice were
given a 3-week recovery period, allowing the expression of the transgene
before performing CH.
For electrogene transfer, tibialis muscles were injected with 8 U of hyaluronidase 1 hr prior injection of 15 mg pMyr-Akt or 5 mg pH2B-CFP+10 mg pCDNA3
plasmids. Six 65 V/cm pulses of 60 ms, with a 100-ms interval, were applied.
Muscles were collected 10 days after gene delivery.
All experiments were conducted in accordance with European guidelines
for the care and use of laboratory animals and were approved by the institutional animal care and use committee.
Plasmids and Viral Productions
For detailed plasmid and viral productions descriptions, see Supplemental
Information.
Primary Muscle Cell Culture and Infections
Primary cultures were derived from hindlimb muscles of Srfflox/flox mice and
cultured as described (Ohanna et al., 2005). Myoblasts were grown in growth
medium (DMEM/F12, 2% Ultroser G, 20% Fetal Calf Serum). For differentiation, myoblasts were seeded in Matrigel-coated dishes and cultured in differentiation medium (DM; DMEM/F-12, 2% horse serum).
Myotubes at day 2 of differentiation were transduced with 2 i.p./cell of
adenoviruses AdGFP, AdCreGFP, or AdSRFVP16.
Myotubes were treated with Cox2-inhibitor II (SC-791, Calbiochem; 5 and
15 mM) or cyclosporine (Csa, Sigma, 5 mM), added daily to DM.
To generate a stable cell line expressing nls-LacZ, myoblasts were infected
twice with the supernatant of a stable 293 Phoenix cell line producing nls-LacZ
retrovirus and 8 mg /ml of polybrene.
Single-Fiber Culture
Individual fibers were isolated from EDL muscles as described (Rosenblatt
et al., 1995). Myofibers were grown as nonadherent cultures for 48 hr in
DMEM medium supplemented with 2% horse serum or in conditioned media
from control myotubes.
To detect cycling satellite cells, single myofibers were immunostained using
anti-Pax7 (Santa Cruz) and anti-Ki67 (Abcam) and counterstained with DAPI.
The % of Pax7+Ki67+ cells is represented as relative to the total number of
Pax7+ cells. At least 60 cells were counted for each assay.
formed using a Light Cycler instrument with SYBR Green I kit (Roche). Primer
sequences are listed in Supplemental Information. The values were normalized
to housekeeping genes 18S rRNA or Cyclophilin.
ChIPs
Myotubes (5.106 cells) were fixed with 1% paraformaldehyde for 11 min. For
each immunoprecipitation, sonicated chromatin was incubated with 3 mg
anti-Srf (G20, Santa Cruz) or IgG and immunocomplexes were recovered
using a Magna ChIP G Kit (Millipore). Quantification of immunoprecipitated
DNA was done by real-time PCR using the appropriated primers (see Supplemental Information) and reported to input chromatin. Results shown are
normalized to the recovery of a non-Srf-regulated gene region (in the intronic
part of Il4).
Reporter Assays
Fifty thousand myoblasts were transfected using Lipofectamine 2000 reagent
(Invitrogen) with 80 ng luciferase plasmid reporter, 100 ng pRL-TK and 200
400 ng of indicated plasmid in a total of 800 ng DNA. Twenty-four hours
post-transfection, cells were cultured in DM and harvested 24 hr later. Firefly
and Renilla luciferase activities were determined using the Dual-Luciferase
Reporter Assay System (Promega). Each experiment was performed in triplicate and repeated three times.
Immunostaining
Plantaris muscles were frozen in cooled isopentane and cut in 7 mm-thick
sections. Fiber size, myonuclei number, and satellite cells were analyzed
by immunostaining using anti-dystrophin (Novocastra), anti-P-Akt (Cell
Signaling), anti-Pax7 (Santa Cruz), and/or anti-Ki67 (Abcam) and DAPI staining. Between 300 and 600 myofibers were analyzed per muscle. The distribution of fiber cross-sectional area was determined using Metamorph.
Western Blot Analysis
Western blotting was performed as described (Lahoute et al., 2008). Immunoblots were done using anti-phospho-Akt Ser473 and anti-Akt (Cell Signaling),
anti-Srf, and anti-Gapdh (Santa Cruz) antibodies.
Statistical Analysis
The significance of differences between means was assessed with a Students
t-test. P values of < 0.05 were considered statistically significant.
SUPPLEMENTAL INFORMATION
Supplemental Information includes Supplemental Experimental Procedures, seven figures, and one table and appears with this article online at
doi:10.1016/j.cmet.2011.12.001.
Fusion and Proliferation Assays

To analyze fusion, myotubes at day 2 of differentiation transduced with AdCreGFP or AdGFP were loaded with 6 mM of Green Cell Tracker (Molecular
Probes) for 15 min and then were cocultured with nls-LacZ myocytes
(myoblasts incubated overnight in DM). When indicated, conditioned media
(CM) from control myotubes was added. Two days later, fusion events were
scored after immunostaining with anti-LacZ antibody (Invitrogen) by counting
the dual-labeled cells (green cells GFP+/ nls-lacZ+). The number of fusion
events was normalized by the number of stained nls-LacZ nuclei. The presence of dual labels was analyzed in 70150 myotubes in each experiment.
To detect S phase entry, Srfflox/flox myoblasts were cultured for 12 hr in
CM from SRFflox/flox myotubes transduced with AdCreGFP or AdGFP
supplemented with 10% FCS, 0.1% Ultroser and pulsed with BrdU (Sigma;
5 mg/ml) for 2 hr prior to fixation and subsequent immunostaining using
anti-BrdU antibody (Dako). The percentage of BrdU+ cells is represented as
relative to total number of cells counted. At least 1,000 cells were counted
for each assay.
ACKNOWLEDGMENTS
Isolation of mRNA, RT-PCR, and qRT-PCR

Total RNA was extracted using TRIzol reagent and reverse-transcribed with
VILO reverse transcriptase (Invitrogen). Quantitative PCR analysis was per-
Alway, S.E., Degens, H., Krishnamurthy, G., and Smith, C.A. (2002). Potential
role for Id myogenic repressors in apoptosis and attenuation of hypertrophy in
muscles of aged rats. Am. J. Physiol. Cell Physiol. 283, C66C76.
We thank R. Treisman, M. Foretz, G. Pavlath and A.S. Armand for providing

materials. We acknowledge the Genomic Core Facility of Cochin Institute.
We are grateful to L. Dandolo, B. Chazaud, R. Mounier, C. Desdouets,
S. Alonso-Martin, J.P. Concordet, P. Maire, F. Legrand, and members of the
laboratory for assistance and critical reading of the manuscript. Figure 4 was
produced using Servier Medical Art. This work was supported by grants
from AFM (13523 to A.S.) and from ANR (JC08-327703 to A.G. and A.S.;
and ANR-08-GENO-023 to S.H., D.D., A.S.).
Received: March 3, 2011
Revised: July 20, 2011
Accepted: December 2, 2011
Published online: January 3, 2012
REFERENCES
Cell Metabolism
Blaauw, B., Canato, M., Agatea, L., Toniolo, L., Mammucari, C., Masiero, E.,
Abraham, R., Sandri, M., Schiaffino, S., and Reggiani, C. (2009). Inducible
activation of Akt increases skeletal muscle mass and force without satellite
cell activation. FASEB J. 23, 38963905.
Bondesen, B.A., Mills, S.T., Kegley, K.M., and Pavlath, G.K. (2004). The COX-2
pathway is essential during early stages of skeletal muscle regeneration.
Am. J. Physiol. Cell Physiol. 287, C475C483.
Bondesen, B.A., Mills, S.T., and Pavlath, G.K. (2006). The COX-2 pathway
regulates growth of atrophied muscle via multiple mechanisms. Am. J.
Physiol. Cell Physiol. 290, C1651C1659.
Bondesen, B.A., Jones, K.A., Glasgow, W.C., and Pavlath, G.K. (2007).
Inhibition of myoblast migration by prostacyclin is associated with enhanced
cell fusion. FASEB J. 21, 33383345.
Bruusgaard, J.C., Johansen, I.B., Egner, I.M., Rana, Z.A., and Gundersen, K.
(2010). Myonuclei acquired by overload exercise precede hypertrophy and
are not lost on detraining. Proc. Natl. Acad. Sci. USA 107, 1511115116.
Carson, J.A., Yamaguchi, M., and Alway, S.E. (1995). Hypertrophy and proliferation of skeletal muscle fibers from aged quail. J. Appl. Physiol. 78, 293299.
Carson, J.A., Nettleton, D., and Reecy, J.M. (2002). Differential gene expression in the rat soleus muscle during early work overload-induced hypertrophy.
FASEB J. 16, 207209.
Charvet, C., Houbron, C., Parlakian, A., Giordani, J., Lahoute, C., Bertrand, A.,
Sotiropoulos, A., Renou, L., Schmitt, A., Melki, J., et al. (2006). New role for
serum response factor in postnatal skeletal muscle growth and regeneration
via the interleukin 4 and insulin-like growth factor 1 pathways. Mol. Cell.
Biol. 26, 66646674.
Fluck, M., Carson, J.A., Schwartz, R.J., and Booth, F.W. (1999). SRF protein is
upregulated during stretch-induced hypertrophy of rooster ALD muscle.
J. Appl. Physiol. 86, 17931799.
Glass, D.J. (2010). Signaling pathways perturbing muscle mass. Curr. Opin.
Clin. Nutr. Metab. Care 13, 225229.
Horsley, V., and Pavlath, G.K. (2003). Prostaglandin F2(alpha) stimulates
growth of skeletal muscle cells via an NFATC2-dependent pathway. J. Cell
Biol. 161, 111118.
Horsley, V., Jansen, K.M., Mills, S.T., and Pavlath, G.K. (2003). IL-4 acts as
a myoblast recruitment factor during mammalian muscle growth. Cell 113,
483494.
Kuang, S., Gillespie, M.A., and Rudnicki, M.A. (2008). Niche regulation of
muscle satellite cell self-renewal and differentiation. Cell Stem Cell 2, 2231.
Lahoute, C., Sotiropoulos, A., Favier, M., Guillet-Deniau, I., Charvet, C., Ferry,
A., Butler-Browne, G., Metzger, D., Tuil, D., and Daegelen, D. (2008). Premature
aging in skeletal muscle lacking serum response factor. PLoS ONE 3, e3910.
Lamon, S., Wallace, M.A., Leger, B., and Russell, A.P. (2009). Regulation of
STARS and its downstream targets suggest a novel pathway involved in
human skeletal muscle hypertrophy and atrophy. J. Physiol. 587, 17951803.
Le Grand, F., and Rudnicki, M.A. (2007). Skeletal muscle satellite cells and
adult myogenesis. Curr. Opin. Cell Biol. 19, 628633.
Li, J., Zhu, X., Chen, M., Cheng, L., Zhou, D., Lu, M.M., Du, K., Epstein, J.A.,
and Parmacek, M.S. (2005). Myocardin-related transcription factor B is
required in cardiac neural crest for smooth muscle differentiation and cardiovascular development. Proc. Natl. Acad. Sci. USA 102, 89168921.
McCarthy, J.J., Mula, J., Miyazaki, M., Erfani, R., Garrison, K., Farooqui, A.B.,
Srikuea, R., Lawson, B.A., Grimes, B., Keller, C., et al. (2011). Effective fiber
hypertrophy in satellite cell-depleted skeletal muscle. Development 138,
36573666.
Miralles, F., Posern, G., Zaromytidou, A.I., and Treisman, R. (2003). Actin
dynamics control SRF activity by regulation of its coactivator MAL. Cell 113,
329342.
Novak, M.L., Billich, W., Smith, S.M., Sukhija, K.B., McLoughlin, T.J.,
Hornberger, T.A., and Koh, T.J. (2009). COX-2 inhibitor reduces skeletal
muscle hypertrophy in mice. Am. J. Physiol. Regul. Integr. Comp. Physiol.
296, R1132R1139.
Ohanna, M., Sobering, A.K., Lapointe, T., Lorenzo, L., Praud, C., Petroulakis,
E., Sonenberg, N., Kelly, P.A., Sotiropoulos, A., and Pende, M. (2005). Atrophy
of S6K1(-/-) skeletal muscle cells reveals distinct mTOR effectors for cell cycle
and size control. Nat. Cell Biol. 7, 286294.
Otis, J.S., Burkholder, T.J., and Pavlath, G.K. (2005). Stretch-induced
myoblast proliferation is dependent on the COX2 pathway. Exp. Cell Res.
310, 417425.
Parlakian, A., Charvet, C., Escoubet, B., Mericskay, M., Molkentin, J.D., GaryBobo, G., De Windt, L.J., Ludosky, M.A., Paulin, D., Daegelen, D., et al. (2005).
Temporally controlled onset of dilated cardiomyopathy through disruption of
the SRF gene in adult heart. Circulation 112, 29302939.
Penkowa, M., Keller, C., Keller, P., Jauffred, S., and Pedersen, B.K. (2003).
Immunohistochemical detection of interleukin-6 in human skeletal muscle
fibers following exercise. FASEB J. 17, 21662168.
Pipes, G.C., Creemers, E.E., and Olson, E.N. (2006). The myocardin family of
transcriptional coactivators: versatile regulators of cell growth, migration, and
myogenesis. Genes Dev. 20, 15451556.
Rosenblatt, J.D., Lunt, A.I., Parry, D.J., and Partridge, T.A. (1995). Culturing
satellite cells from living single muscle fiber explants. In Vitro Cell. Dev. Biol.
Anim. 31, 773779.
Kuwahara, K., Kinoshita, H., Kuwabara, Y., Nakagawa, Y., Usami, S., Minami,
T., Yamada, Y., Fujiwara, M., and Nakao, K. (2010). Myocardin-related
transcription factor A is a common mediator of mechanical stress- and neurohumoral stimulation-induced cardiac hypertrophic signaling leading to activation of brain natriuretic peptide gene expression. Mol. Cell. Biol. 30, 4134
4148.
Serrano, A.L., Baeza-Raja, B., Perdiguero, E., Jard, M., and Munoz-Canoves,
P. (2008). Interleukin-6 is an essential regulator of satellite cell-mediated
skeletal muscle hypertrophy. Cell Metab. 7, 3344.
Lafreniere, J.F., Mills, P., Bouchentouf, M., and Tremblay, J.P. (2006).
Interleukin-4 improves the migration of human myogenic precursor cells
in vitro and in vivo. Exp. Cell Res. 312, 11271141.
Shen, W., Prisk, V., Li, Y., Foster, W., and Huard, J. (2006). Inhibited skeletal
muscle healing in cyclooxygenase-2 gene-deficient mice: the role of PGE2
and PGF2alpha. J. Appl. Physiol. 101, 12151221.
Cell Metabolism
Article
PGRN is a Key Adipokine Mediating
High Fat Diet-Induced Insulin Resistance
and Obesity through IL-6 in Adipose Tissue
Toshiya Matsubara,1,3,6,10 Ayako Mita,1,2,10 Kohtaro Minami,1 Tetsuya Hosooka,2 Sohei Kitazawa,4,11 Kenichi Takahashi,9
Yoshikazu Tamori,2,7 Norihide Yokoi,1 Makoto Watanabe,6,8 Ei-ichi Matsuo,6,8 Osamu Nishimura,3,8,*
and Susumu Seino1,2,3,5,*
1Division
of Cellular and Molecular Medicine

of Diabetes and Endocrinology
3The Integrated Center for Mass Spectrometry
4Division of Diagnostic Molecular Pathology
Kobe University Graduate School of Medicine, 7-5-1 Kusunoki-cho, Chuo-ku, Kobe, Hyogo 650-0017, Japan
5Core Research for Evolutional Science and Technology (CREST), Japan Science and Technology Agency, 4-1-8 Hon-cho, Kawaguchi,
Saitama 332-0012, Japan
6Life Science Research Center, Technology Research Laboratory, Shimadzu Corporation, 3-9-4 Hikaridai, Seika-cho, Soraku-gun,
Kyoto 619-0237, Japan
7Department of Internal Medicine, Diabetes Center, Chibune Hospital, 2-2-45 Tsukuda, Nishiyodogawa-ku, Osaka 555-0001, Japan
8Division of Disease Proteomics, Institute for Protein Research, Osaka University, 3-2 Yamadaoka, Suita, Osaka 565-0871, Japan
9JCR Pharmaceuticals Co., Ltd., 2-2-10 Murotani, Nishi-ku, Kobe, Hyogo, 651-2241, Japan
11Present address: Division of Molecular Pathology, Ehime University School of Medicine, Shitsukawa, Toon, Ehime 791-0295, Japan
*Correspondence: seino@med.kobe-u.ac.jp (S.S.), osamu_nishimura@protein.osaka-u.ac.jp (O.N.)
DOI 10.1016/j.cmet.2011.12.002
2Division
SUMMARY
Adipose tissue secretes adipokines that mediate

insulin resistance, a characteristic feature of obesity
and type 2 diabetes. By differential proteome analysis of cellular models of insulin resistance, we identified progranulin (PGRN) as an adipokine induced by
TNF-a and dexamethasone. PGRN in blood and
adipose tissues was markedly increased in obese
mouse models and was normalized with treatment
of pioglitazone, an insulin-sensitizing agent. Ablation
of PGRN (Grn / ) prevented mice from high fat diet
(HFD)-induced insulin resistance, adipocyte hypertrophy, and obesity. Grn deficiency blocked elevation of IL-6, an inflammatory cytokine, induced by
HFD in blood and adipose tissues. Insulin resistance
induced by chronic administration of PGRN was suppressed by neutralizing IL-6 in vivo. Thus, PGRN is
a key adipokine that mediates HFD-induced insulin
resistance and obesity through production of IL-6
in adipose tissue, and may be a promising therapeutic target for obesity.
INTRODUCTION
Insulin resistance is a characteristic feature of obesity and type 2
diabetes. Adipose tissue is now recognized as not only an
energy-storage tissue but also an endocrine tissue that secretes
a variety of bioactive substances (adipokines) including adiponectin, resistin, tumor necrosis factor (TNF)-a, interleukin-6
(IL-6), and monocyte chemoattractant protein (MCP)-1 (Shoelson et al., 2007; Waki and Tontonoz, 2007). Defects in adipokine
secretion accompanying adipose tissue dysfunction contribute
to the pathophysiology of insulin resistance and obesity (Kahn
and Flier, 2000). Reduced expression and secretion of adiponectin in obesity promotes the development of systemic insulin
resistance by enhancing hepatic gluconeogenesis and suppressing glucose uptake in skeletal muscle (Guilherme et al.,
2008; Berg et al., 2001). In contrast, resistin, TNF-a, IL-6 and
MCP-1, the levels of which in adipose tissues and blood are
elevated in obesity, have been shown to be mediators in progression of insulin resistance (Waki and Tontonoz, 2007).
The relationship between inflammatory process and insulin
resistance has recently drawn considerable attention. For example, TNF-a, a proinflammatory cytokine, has been shown to
contribute to the development of insulin resistance by altering
insulin signaling mediated by activation of the IKK-NFkB and
JNK-AP1 signaling pathways (Hotamisligil et al., 1994, 1995;
Uysal et al., 1997). On the other hand, glucocorticoids, which
are known to have an anti-inflammatory action, also induce
insulin resistance in human and animals (Caro and Amatruda,
1982). Dexamethasone, a glucocorticoid, has been reported to
impair insulin signaling and insulin-stimulated glucose uptake
in adipose tissue, liver, and skeletal muscle (Qi and Rodrigues,
2007). Since TNF-a and dexamethasone both induce insulin
resistance despite their opposite inflammatory properties (Hotamisligil et al., 1994; Hotamisligil, 2006; Wellen and Hotamisligil,
2005; Qi and Rodrigues, 2007; van Putten et al., 1985; Turnbow
et al., 1994; Sakoda et al., 2000), we reasoned that there might
be a common mediator responsible for the cellular basis of
insulin resistance induced by TNF-a and dexamethasone. In
the present study, we searched for a novel adipokine(s) that
Cell Metabolism
Role of PGRN in Insulin Resistance and Obesity
play a key role in developing insulin resistance using 3T3-L1

adipocytes treated with TNF-a or dexamethasone. For this
purpose, we utilized a method of differential proteome analysis
based on stable isotope labeling of proteins with chemical
reagent 2-nitrobenzenesulfenyl chloride (NBSCl) incorporating
six 13C (13C6) or six 12C (13C0) in the tryptophan residues (Kuyama
et al., 2003; Matsuo et al., 2009). The NBS method has an advantage in reducing the complexity of the analysis because the
method targets only peptides that contain tryptophan, which is
the least abundant amino acid but is widespread in proteins
(Matsuo et al., 2009), and has been used successfully for differential expression analysis in clinical samples (Watanabe et al.,
2008; Okamura et al., 2008).
We applied differential proteome analysis using the NBSbased method to search for novel adipokines and identified
progranulin (PGRN) as a candidate. PGRN, also known as proepithelin, granulin/epithelin precursor (GEP) or PC cell-derived
growth factor (PCDGF), was originally discovered as an acrosomal glycoprotein, named acrogranin, which is synthesized
during guinea pig spermatogenesis (Anakwe and Gerton,
1990). PGRN is a 68-88 kDa secreted protein having seven
and one-half granulin (GRN) motifs connected by short linker
domains (He and Bateman, 2003). It is expressed widely in
tissues, especially at high levels in spleen and placenta (Bateman and Bennett, 1998). PGRN has been shown to be a pluripotent growth factor that mediates cell-cycle progression,
tumorigenesis, and wound healing (He and Bateman, 2003).
PGRN is also implicated in various disease states in humans,
including cancers of the breast and ovaries (He and Bateman,
2003), neurodegenerative diseases such as frontotemporal
dementia (Cruts and Van Broeckhoven, 2008), and rheumatoid
arthritis (Tang et al., 2011).
In the course of this study, it was reported that serum PGRN
concentrations in patients with type 2 diabetes are higher than
those in normal subjects (Youn et al., 2009). However, the role
of PGRN in insulin resistance and obesity remains unknown. In
the present study, we found that PGRN is a key adipokine that
mediates high fat diet (HFD)-induced insulin resistance, adipocyte hypertrophy, and obesity through production of IL-6 in
adipose tissue and that it is a potential target for treating HFDinduced obesity.
RESULTS
Identification of PGRN as an Adipokine Involved
in Insulin Resistance In Vitro
To identify proteins associated with insulin resistance in adipocytes in vitro, differential proteome analysis using the NBS
method (Matsuo et al., 2009) was performed in 3T3-L1 adipocytes in which insulin resistance was induced by TNF-a or dexamethasone, as outlined in Figure 1A. Expression levels of
proteins in these adipocytes were compared with those in
untreated adipocytes. The relative ratio of protein expressions
was estimated from the intensity of paired peaks with mass
difference of 6 Da derived from 13C0 and 13C6NBS-tagged
peptides in mass spectrum (Figure 1B). Since expression levels
of glyceraldehydes-3-phosphate dehydrogenase (GAPDH; aa
308-321; m/z 1933.1 [13C0], 1939.1 [13C6]) were not changed
by these treatments (Figure S1A available online), we used
GAPDH as an internal control. We found that 37 and 43 proteins

were upregulated by TNF-a treatment and dexamethasone
treatment, respectively, among which 21 proteins are common
in the two treatments (Figure S1D). We also found that 11 and
7 proteins were downregulated by TNF-a treatment and dexamethasone treatment, respectively, among which 2 proteins
are common in these treatments (Figure S1D). Identification of
haptoglobin, serum amyloid A-3 (SAA3) protein precursor, and
nicotinamide phosphoribosyltransferase, all of which are known
as adipokines to be upregulated by such treatment (Shoelson
et al., 2007; Lago et al., 2007; do Nascimento et al., 2004; Chiellini et al., 2002), confirmed the validity of the method (Table S1).
After excluding known adipokines among the 23 proteins identified, we confirmed the results of differential proteome analysis
on 8 proteins by immunoblot analysis using antibodies currently
available. We finally selected progranulin (PGRN) because it
is the only protein with both secretory and proinflammatory
properties.
PGRN, detected as an NBS-modified peptide pair (m/z 1634.70
[13C0], 1640.70 [13C6]), was upregulated in both TNF-a-treated
(1.66-fold versus control) and dexamethasone-treated (3.01fold versus control) adipocytes (Figures 1B and S1C). Using
tandem mass spectrometry (MS/MS) (Perkins et al., 1999), the
heavier peptide peak (m/z 1640.70 [13C6]) was found to match
the amino acid sequence (N-LNTGAWGCCPFAK-C) of trypsinized peptide derived from PGRN with a 13C6NBS modification
of the tryptophan residue (MASCOT score 49, expected P-value
0.0018) (Figure S1B). Using immunoblot analysis, we confirmed
that PGRN detected as an 80 kDa protein was significantly
increased under both conditions (Figure S1E).
Induction of PGRN expression by TNF-a or dexamethasone
was completely blocked by pioglitazone (Figure 1C), a peroxisome proliferators-activated receptor (PPAR) g agonist that
improves insulin resistance (Olefsky, 2000). In addition, we found
that PGRN expression decreased with differentiation of the cells
(Figure 1D), as assessed by Pref-1, PPARg, and FAPB4. These
results indicate that PGRN expression is associated strongly
with insulin resistance at cell level.
PGRN Has a Causative Role in Insulin Resistance In Vivo
We investigated the role of PGRN in insulin resistance in vivo.
Among liver, skeletal muscle, and adipose tissues, which are
the major target tissues of insulin, PGRN was expressed at
high levels in epididymal fat, at moderate levels in mesenteric
fat, and at low levels in liver in wild-type mice but was not expressed in skeletal muscle (Figure 2A). PGRN levels were significantly increased in adipose tissues and liver by HFD but not in
skeletal muscle (Figures 2A, S2A, and S2B). Grn (the gene
symbol of PGRN) was also expressed in leukocytes, spleen,
and lung, which abundantly contain immune cells, but was not
increased by HFD in these cells and tissues (Figure 2B). Immunohistochemistry of epididymal fat revealed PGRN to be
present in adipose tissue (Figure 2C) but the cellular distribution
was not clear. Therefore, we performed immunoblot analysis
of adipocytes and stromal vascular fraction (SVF) separated
from epididymal fat and found the presence of PGRN in both
adipocytes and stroma (Figure 2D). Expression levels of
PGRN were significantly increased by HFD in both fractions
(Figure 2D).
Cell Metabolism
A
TNF-
or
Dexamethasone
Control
100
3T3-L1 adipocytes
60
NBS tagging of tryptophan

13C
13C
0NBS
60
40
20
0
80
Protein extraction
40
6NBS
100
80
m/z
20
0
Tryptic digestion and enrichment

of NBS-tagged peptides
m/z
Fractionation and MS by LC-MALDI-TOF MS
100
80
100
80
m/z = 6
60
Relative quantification of NBS-tagged peptide pair
40
MS/MS
20
60
40
20
0
m/z
Protein identification
m/z
PGRN (aa297-309) peptide: R.LNTGAW*GCCPFAK.A
D
PGRN
Relative
protein expression
GAPDH
**
-1 0 1 2 3 4 5 6 7 8 9 10
**
2.5
PGRN
2.0
Pref-1
1.5
1.0
PPAR
0.5
FABP4
0
Pio (10 M)
ACTIN
tro
n
Co
F-
TN
m
xa
De
e
on
as
h
et
Figure 1. Proteome Analysis of 3T3-L1 Adipocytes Treated with TNF-a or Dexamethasone

(A) Outline of NBS method for differential proteome analysis in 3T3-L1 adipocytes. Proteins from untreated adipocytes were tagged with 13C0NBS reagents and
those from TNF-a-treated or dexamethasone-treated were tagged with 13C6NBS reagents. Relative quantification of the NBS-tagged proteins in the two samples
was performed from the MS spectra; proteins were then identified by a database search using queries based on data from the MS/MS spectra.
(B) MS spectrum of NBS proteome analysis. Inset shows 13C0NBS- and 13C6NBS-tagged peptide (LNTGAWGCCPFAK) from PGRN. The asterisk shown in the
peptide sequence indicates the tryptophan residue with a NBS modification.
(C) Effect of pioglitazone (Pio) on PGRN expression in 3T3-L1 adipocytes treated with TNF-a or dexamethasone. Quantitative data are presented as means SEM
from three independent experiments. *p < 0.05; **p < 0.01 (Students unpaired t-test).
(D) Changes in PGRN expression during adipocyte differentiation. The differentiation of 3T3-L1 preadipocytes were initiated by addition of differentiation medium
at Day 0. PGRN, Pref-1, PPARg, FABP4, and ACTIN expressions were determined by immunoblot analysis. Data are representative of three independent
experiments.
We then examined Grn expression in ob/ob mice, a well-characterized obese and insulin resistance model (Friedman and
Halaas, 1998). Grn expression in ob/ob mice was upregulated
in both peritoneal and subcutaneous white adipose tissues
(WAT) but not in brown adipose tissue (BAT), as compared
with that of ob/+ mice (Figure 2E, left). Serum PGRN levels of
ob/ob mice were also higher than those of ob/+ mice (Figure 2E,
right). Immunohistochemistry of epididymal fat revealed that
PGRN was detected predominantly in macrophages (as assessed by Mac-3 costaining) and also was present in the cytoplasm of adipocytes (Figure 2F).
ob/ob mice treated with pioglitazone exhibited a significant
improvement of glucose tolerance (Figure S2D) and increases
in expressions of TZD/PPARg-dependent genes in adipose
tissues (Table S2) but no significant change in body weight (Figure S2C). Under these conditions, Grn expressions in both
Cell Metabolism
peritoneal and subcutaneous WAT but not in BAT were decreased significantly (Figure 2G, left). Pioglitazone also normalized serum PGRN levels (Figure 2G, right). These results indicate
that PGRN levels in both adipose tissues and blood are associated with insulin resistance and obesity in vivo.
To determine whether PGRN causes insulin resistance in vivo
directly, recombinant mouse PGRN (rmPGRN) was administrated intraperitoneally to wild-type (WT) mice under standard
diet (SD) condition. We found that serum PGRN levels increased
about 2.4-fold within 1 hr after administration and were kept
constant for 24 hr onward (Figure S2E); we also found that
PGRN levels increased about 2.02.5-fold after treatment with
rmPGRN for 14 days (Figure S2F). This level of increased serum
PGRN is similar to that observed in obese (ob/ob) mice (Figure 2G, right). Under these conditions, the fasting insulin level
in the mice tended to increase (Figure S2I) despite no change
in either body weight or blood glucose level (Figures S2G,
and S2H). We also found that WT mice treated with rmPGRN
for 14 days exhibited insulin resistance, as assessed by insulin
tolerance test (ITT) (Figure 2H). Thus, PGRN has a causative
role in insulin resistance in vivo. These findings indicate that
PGRN is associated with insulin resistance and obesity and
that PGRN directly causes insulin resistance in vivo.
Ablation of Grn Prevents HFD-Induced Obesity
and Insulin Resistance In Vivo
To clarify the physiological and pathophysiological roles of
PGRN directly, we utilized Grn deficient (Grn / ) mice (Kayasuga
et al., 2007). The body weight of Grn / mice fed SD was similar
to that of WT mice (Figure 3A,top left), whereas the body weight
of Grn / mice fed HFD was significantly lower than that of WT
mice (Figure 3A, top right), despite similar food intake (Figure 3A,
bottom). In addition, Grn / mice fed HFD exhibited a marked
reduction in deposition of peritoneal fat compared to WT mice
(Figure 3C, left) as well as in fat mass of both visceral and subcutaneous fat pad (Figure 3B). Immunohistochemistry revealed that
the size of adipocytes in epididymal fat pads of Grn / mice was
significantly smaller than that of WT mice (Figures 3C, middle,
and 3D) and that infiltration of mac-3 positive inflammatory cells
was significantly less in Grn / mice than that in WT mice under
HFD condition (Figures 3C,right, and 3E). In addition, glucose
intolerance induced by HFD was improved in Grn / mice with
a decrease in serum insulin levels (Figures 3F and S3A), suggesting enhanced insulin sensitivity in Grn / mice. Indeed, insulin
tolerance test confirmed that insulin resistance induced by
HFD, which was seen in WT mice, was prevented in Grn /
mice (Figure 3G). These results indicate that ablation of Grn
prevents HFD-induced obesity and insulin resistance in vivo.
PGRN Impairs Insulin Signaling in Adipocytes
Since PGRN has been shown to be involved in the PI3K/Akt
signaling pathway (He and Bateman, 2003; Youn et al., 2009;
Zanocco-Marani et al., 1999; Lu and Serrero, 2001), we reasoned that PGRN might directly affect insulin signaling in
3T3-L1 adipocytes. Although PGRN treatment did not affect
phosphorylation of insulin receptor (IR) (Figure 4A), it decreased
insulin-stimulated phosphorylation of both insulin receptor
substrate (IRS)-1 (Figure 4A) and Akt in a dose-dependent
manner (Figure 4B). PGRN treatment also suppressed insulin-
stimulated glucose uptake (Figure 4C). To further confirm the

involvement of PGRN in insulin signaling, we utilized shRNA
against Grn. The phosphorylations of both IRS-1 and Akt were
increased at basal state and were further increased by insulin
treatment in Grn knockdown (KD) adipocytes compared to those
of respective controls, but the phosphorylation of IR was not
changed (Figures 4D and 4E). In addition, Grn KD also enhanced
insulin-stimulated glucose uptake (Figure 4F). Furthermore, the
TNF-a-induced decrement of insulin-stimulated Akt phosphorylation was reduced by Grn KD (52% in control versus 24% in KD)
(Figure 4G). These results indicate that PGRN in adipocytes
impairs insulin signaling downstream of IR and suppresses
insulin-stimulated glucose uptake and that PGRN mediates
TNF-a-induced insulin resistance at cell level.
PGRN Mediates TNF-a-Induced Insulin Resistance
through IL-6 Expression in 3T3-L1 Adipocytes
To clarify the mechanism by which PGRN mediates insulin resistance at cell level, we first examined the expressions of adipogenic genes (Pparg and Cebpa), adipose-specific genes (Fabp4
and Glut4), and inflammatory adipokines (Lep, Il6, Tnf, and Ccl2)
that are involved in the development of insulin resistance.
Expressions of Pparg and Cebpa in Grn KD adipocytes were
decreased significantly compared to control (Figure 5A). Among
Lep, Il6, Tnf, and Ccl2, which are known to be induced by TNF-a,
induction of Il6 was blocked completely in Grn KD 3T3-L1 adipocytes (Figure 5A). IL-6 has been reported to induce insulin resistance through JAK/STAT signaling and suppression of cytokine
signaling-3 (SOCS3) in both adipocytes and hepatocytes (He
et al., 2002; Rotter et al., 2003; Fasshauer et al., 2004; Ueki
et al., 2004; Emanuelli et al., 2001; Howard and Flier, 2006).
We found that PGRN induced both Il6 and Socs3 expression in
a dose-dependent manner (Figures 5B and 5C). In addition,
induction of STAT3 phosphorylation and Socs3 expression by
TNF-a were also abolished in Grn KD adipocytes (Figures 5D
and 5E). These results indicate that PGRN promotes IL-6 expression in adipocytes and suggest that the resultant increase in IL-6
enhances SOCS3 expression through activating JAK-STAT
signaling, leading to insulin resistance in adipocytes.
PGRN Mediates HFD-Induced Insulin Resistance
through IL-6 Expression in Adipose Tissue In Vivo
We then examined expressions of the genes involved in adipocyte hypertrophy and inflammation in adipose tissue. In epididymal fat of Grn / mice, expressions of Pparg and Cebpa,
Fabp4, Glut4, and Adipoq decreased significantly compared to
those in WT mice (Figure 6A). HFD-induced elevations of inflammatory markers Il6, Tnf, and Emr1 were found in WT mice but not
in Grn / mice (Figure 6A).
Among Il6, Tnf, and Ccl2 induced by HFD, induction of Il6 was
blocked almost completely in adipose tissue of Grn / mice (Figure 6A), as was found in Grn KD adipocytes (Figure 5A). In addition, HFD-induced elevation of serum IL-6 concentration was
also blocked markedly in Grn / mice (Figure 6B). Moreover,
HFD-induced Socs3 expressions were completely diminished
in both epididymal fat and liver in Grn / mice (Figures 6C and
6D). Because expression of Il6 in liver was not changed by
HFD (Figure 6E), HFD-induced SOCS3 expression is likely to
be mediated by IL-6 derived from adipose tissue.
Cell Metabolism
EF
(Grn/18S)
**
**
**
**
**
N.
D
N. .
D.
**
S
HF D
D
S
HF D
D
5.0
4.0
3.0
2.0
1.0
0
PGRN
GAPDH
MF
S
HF D
D
SD
HF
D
Protein expression
(PGRN/GAPDH)
Liver Skeletal
muscle
Adipocytes
SD
HFD
SVF
SD
HFD
**
**
D
SD HF
D
SD HF
SVF
Adipocytes
mRNA expression
(Grn/Rplp0)
Protein expression
(PGRN/ACTIN)
**
300
250
200
150
100
50
0
**
**
ob/+
ob/ob
1 2 3 4 51 2 3 4 5 1 2 3 4 5 1 2 3 4 5
PGRN
ACTIN
4.0
3.0
ob/+
ob/ob
** **
** ** * ** **
**
2.0
1.0
MF
EF
SF
BAT
**
N.S.
MF
SF
BAT
WAT
EF
WAT
Vehicle
Pio
N.S.
3.5
3.0
2.5
2.0
1.5
1.0
0.5
0
**
Pio
ob
/+
Ve
hic
le
1.6
1.4
1.2
1.0
0.8
0.6
0.4
0.2
0
Serum PGRN conc. (g/ml)
Merged
DAPI
Relative Grn mRNA expression
G
Mac-3
PGRN
ob/ob
**
**
Figure 2. Relationship between PGRN and Insulin Resistance In Vivo

(A) Immunoblot analysis of PGRN in epididymal fat (EF; 6 mg/lane, n = 5), mesenteric fat (MF; 6 mg/lane, n = 5), liver (12 mg/lane, n = 5), and skeletal muscle
(12 mg/lane, n = 5) in 26-week-old C57BL/6J mice fed standard diet (SD) or high fat diet (HFD).
(B) mRNA expression of Grn in EF, MF, subcutaneous fat (SF), brown adipose tissue (BAT), spleen, leukocytes, lung, and brain in 23-week-old C57BL/6J mice (n =
5) fed SD or HFD.
(C) Immunohistochemistry of PGRN in EF of 11-week-old C57BL/6J mice on SD. Immunoreactivities of PGRN (red arrow) are shown. Scale bar, 50 mm.
(D) Immunoblot analysis of PGRN in stromal vascular fraction (SVF) and adipocytes from EF of 33-week-old C57BL/6J mice (SD, n = 5; HFD, n = 5).
(E) PGRN levels in adipose tissues (left) and serum (right) of ob/+ (n = 5) and ob/ob (n = 5) mice. Grn expression in EF, MF, SF, and BAT at 10-week-old were
quantified by quantitative real-time RT-PCR analysis, and serum concentrations of PGRN were measured by ELISA assay.
(F) Immunofluorescent staining of PGRN and mac-3, and nuclear counterstaining with DAPI in EF of 14-week-old ob/ob mice. Arrows indicate mac-3 positive
cells. Scale bars, 25 mm.
(G) Effect of pioglitazone (Pio) on PGRN levels in adipose tissues (left) and serum (right) of ob/ob mice. Pio (n = 5) or vehicle (n = 4) was administered to ob/ob mice
for 7 days. Grn expressions in adipose tissues and serum concentrations of PGRN were quantified by quantitative real-time RT-PCR analysis and ELISA assay,
respectively.
Cell Metabolism
Neutralizing IL-6 Improves PGRN-Induced Insulin

Resistance In Vivo
To ascertain that IL-6 mediates PGRN-induced insulin resistance
in vivo, we utilized neutralizing antibody against IL-6. We found
that PGRN level was increased about 2.1-fold after chronic treatment of WT mice fed SD with rmPGRN (20 mg/day) once daily for
3 weeks (Figure S4A). Under this condition, the increased fasting
serum insulin level in the mice treated with rmPGRN tended to be
decreased by neutralizing antibody against IL-6 despite no
change in either body weight or blood glucose level (Figures
S4BS4D). Importantly, ITT reveals that insulin resistance
induced by rmPGRN was significantly improved by neutralizing
antibody against IL-6 (Figure 7) without change in body weight
(Figure S4B), indicating that IL-6 is a mediator of PGRN-induced
insulin resistance in vivo.
DISCUSSION
In the present study, we found that PGRN is a key adipokine
mediating HFD-induced insulin resistance and obesity through
IL-6 in adipose tissues. Despite an extensive search for novel
adipokines, PGRN has not been reported as an adipokine to
date. The NBS method used here has an advantage in reducing
the complexity of analysis by targeting peptides containing tryptophan, which is the least abundant amino acid but is widespread in proteins (Matsuo et al., 2009), and has permitted identification of PGRN as an adipokine.
Proinflammatory adipokines, which are secreted from adipocytes and/or macrophages in adipose tissue, induce a low-grade
chronic inflammatory state that plays a critical role in insulin
resistance associated with obesity (Hotamisligil, 2006; Wellen
and Hotamisligil, 2005; Weisberg et al., 2003; Xu et al., 2003).
However, the molecular basis for adipocyte hypertrophy and
the inflammation process underlying obesity is not fully understood. PGRN caught our attention because it is a secreted
protein associated with proinflammatory properties and, therefore, is a strong candidate for an adipokine involved in insulin
resistance. PGRN has been characterized as the precursor of
granulins (GRNs), some of which have been shown to modulate
inflammation and wound repair (He and Bateman, 2003). The
physiological and pathophysiological functions of PGRN are
complex: PGRN has both anti-inflammatory and proinflammatory properties (Zhu et al., 2002; Kessenbrock et al., 2008; He
and Bateman, 2003). It has been suggested that the full-length
form of the protein has trophic and anti-inflammatory activity,
whereas proteolytic cleavage generates GRNs that promote
inflammatory activity (Eriksen and Mackenzie, 2008). However,
we found that PGRN levels were increased in the insulin resistant
state both in vivo and in vitro and that PGRN induced the expression of IL-6, a proinflammatory adipokine. Accordingly, PGRN in
adipose tissues may well participate in chronic inflammation
associated with insulin resistance and obesity. A recent study
has shown that PGRN binds to TNF receptor and prevents
mice from inflammatory arthritis by blocking interaction with

TNF-a (Tang et al., 2011). Thus, it is possible that PGRN has
dual roles in inflammation and exerts proinflammatory or antiinflammatory function in different tissues.
In the present study, we also found that ablation of PGRN
protected against HFD-induced obesity and insulin resistance
in vivo. Although Grn / mice were originally reported to
exhibit offensive behavior against intruders, the mice did not
show hyperactivity under SD condition (Kayasuga et al.,
2007). In addition, neither energy metabolism nor locomotor
activity under HFD condition was investigated in their study.
To determine whether hyperactive behavior and/or energy
expenditure contributed to the development of obesity in
Grn / mice, we measured locomotor activity and performed
respiratory gas analysis before the development of obesity in
these mice. Grn / mice exhibited neither hyperactivity (Figure S3B) nor increased energy expenditure (Figures S3CS3I)
under SD or HFD condition. It is unlikely, therefore, that the
protective phenotype against HFD-induced obesity is due to
hyperactivity. Interestingly, the respiratory quotient in Grn /
mice fed HFD was significantly lower at dark phase, suggesting that ablation of PGRN suppressed HFD-induced obesity
by consuming lipids more preferentially than carbohydrate
(Figure S3H).
We have also found that PGRN induces insulin resistance
through IL-6 both in vivo and in vitro. It has been hypothesized
that chronic increase of IL-6 plays a role in causing insulin resistance associated with obesity. IL-6 alters insulin signaling differently in various tissues (Mooney, 2007). The IL-6 / STAT3
pathway is required for the action of insulin signaling in the brain
on hepatic gluconeogenesis (Wallenius et al., 2002; Inoue et al.,
2006). Therefore, IL-6 has both central and peripheral roles in
metabolism and its effects on systemic insulin resistance are
complex. However, IL-6 expression in adipose tissue is known
to contribute to developing chronic inflammatory states, such
as obesity (Senn et al., 2002; Fried et al., 1998; Shoelson et al.,
2007). In addition, it has been shown that circulating IL-6 is
elevated in obese, diabetic subjects (Pickup et al., 1997; Kern
et al., 2001) and that adipose tissue is a major site of IL-6 secretion, accounting for 15%35% of circulating levels (Fried et al.,
1998; Mohamed-Ali et al., 1997). Considered with our finding
that adipose tissue is a major source of increased PGRN in the
blood of mice fed HFD, PGRN may well induce IL-6 expression
in adipose tissues in obesity.
The mechanisms by which IL-6 inhibits insulin signaling have
been studied extensively in adipocytes and hepatocytes
(Mooney, 2007). IL-6 has been shown to attenuate insulin
signaling, which is mediated by increasing SOCS3 expression
through activation of JAK-STAT signaling in adipocytes and
hepatocytes (Ueki et al., 2004; Shi et al., 2004; Senn et al.,
2003; Emanuelli et al., 2001). SOCS3 impairs tyrosine phosphorylation of IRS-1 by direct interaction and promotes proteasomal
degradation of IRS-1 (Ueki et al., 2004; Emanuelli et al., 2001).
(H) Induction of insulin resistance by administration of recombinant PGRN in vivo. Recombinant mouse PGRN (rmPGRN, i.p. 20 mg/day) or PBS (vehicle) was
administered to C57BL/6J mice once daily for 14 days (n = 7) under SD condition. Insulin sensitivity was assessed by insulin tolerance test (ITT) (left). Blood
glucose levels were determined at the indicated times after intraperitoneal injection of insulin (0.5 IU/kg). Inverse area under curve (AUC) of ITT was shown (right).
All data are means SEM. *p < 0.05; **p < 0.01 in (A), (B), (D), (E), left of (G), (H) (Students unpaired t-test) and in right (G) (Dunnets method); N.S., not significant.
Cell Metabolism
B
Body weight (g)
A
45
40
35
30
25
20
15
10
5
0
SD
WT
Grn -/6
8 10 12 14 16
45
40
35
30
25
20
15
10
5
0
N.S.
HFD
WT
Grn -/-
*
*****
WT
Grn-/6
N.S.
8 10 12 14 16
N.S.
HE
Mac-3
WT
Grn/-
WT
(SD)
WT
(HFD)
250
200
150
Grn-/(SD)
N.S.
100
50
SD
Grn-/(HFD)
Grn-/-
WT
HF
SD
HF
Adipocyte diameter (m)
WT
Grn/-
F
N.S.
WT (HFD)
Grn-/- (HFD)
WT (HFD)
Grn-/- (HFD)
N.S.
N.S.
WT (SD)
WT (HFD)
Grn-/- (HFD)
Grn-/- (HFD)
WT
Grn-/-
**
**
Figure 3. Prevention of High Fat Diet-Induced Insulin Resistance, Adipocyte Hypertrophy, and Obesity by Ablation of Grn In Vivo
(A) Changes in body weight and food consumption on SD or HFD. Changes in body weights in WT (SD, n = 8, HFD, n = 8) and Grn / (SD, n = 5, HFD, n = 9) mice
were measured. Food consumptions in WT (n = 11) and Grn / (n = 9) mice fed SD or HFD for 1 week are shown as food intake (g) per day.
(B) Tissue weight of WATs in WT (SD, n = 8; HFD, n = 8) and Grn / mice (SD, n = 8; HFD, n = 7).
(C-E) Gross appearance (C left), histology (H&E staining: C middle, mac-3 immunostaining: C right), adipocyte diameter (n = 100) (D), and number of mac-3
positive cells (n = 10) (E) in epididymal fat of WT and Grn / mice. Red arrows indicate mac-3 positive cells. Scale bars, 50 mm.
(F) Oral glucose tolerance test. Blood glucose (left) and serum insulin (right) levels in WT mice (SD, n = 8; HFD, n = 10) and Grn / mice (SD, n = 5; HFD, n = 9) were
determined at the indicated times after oral administration of glucose.
(G) Insulin tolerance test (ITT). Blood glucose levels in WT (SD, n = 8; HFD, n = 9) and Grn / (SD, n = 5; HFD, n = 9) mice were determined at the indicated times
after intraperitoneal injection of insulin (0.3 IU/kg) (left). Insulin sensitivity was assessed by inverse AUC of ITT (right).
All data are means SEM. *p < 0.05; **p < 0.01 in (A), (B), (D), (E), (F) (Students unpaired t-test), and in (G), compared with WT mice fed SD (Dunnets method);
N.S., not significant.
Cell Metabolism
Figure 4. Inhibition of Insulin Signaling by PGRN in

3T3-L1 Adipocytes
*
**
** **
C
D
N.S.
N.S.
**
(A) Effects of exogenous PGRN on IR and IRS-1 phosphorylation. 3T3-L1 adipocytes were treated with various
concentrations of recombinant mouse PGRN protein
(rmPGRN) for 20 hr and subsequently stimulated with
100 nM of insulin for 10 min. Activation of insulin signaling
was then assessed by phosphorylation of IR (Tyr1146)
and tyrosine phosphorylation of immunoprecipitated (IP)
IRS-1.
(B) Effect of exogenous PGRN on Akt phosphorylation
(n = 4). 3T3-L1 adipocytes were treated with various
concentrations of rmPGRN for 4 hr and subsequently
stimulated with 10 nM of insulin for 10 min. Activation of
insulin signaling was then assessed by phosphorylation of
Akt (Ser743).
(C) Effects of exogenous PGRN on glucose uptake.
3T3-L1 adipocytes were treated with various concentrations of rmPGRN for 20 hr and subsequently stimulated
with 10 nM of insulin for 10 min (n = 3).
(D) Effects of Grn knockdown (KD) on IR and IRS-1
phosphorylation. 3T3-L1 adipocytes were infected with
adenovirus carrying shRNA for Grn (GrnshRNA) or adenovirus carrying nontarget shRNA (control) at MOI of 100.
Activation of insulin signaling was then assessed by
phosphorylation of IR (Tyr1146) and tyrosine phosphorylation of immunoprecipitated (IP) IRS-1.
(E) Effect of Grn KD on insulin-stimulated Akt phosphorylation. Insulin-stimulated Akt phosphorylation in KD or
control cells was analyzed by immunoblot analysis (n = 4).
(F) Effects of Grn KD on glucose uptake (n = 3).
(G) Effect of Grn KD on the suppression by TNF-a of
insulin-stimulated Akt phosphorylation. KD or control cells
were treated with 10 ng/ml TNF-a for 16 hr. Insulinstimulated Akt phosphorylation was then analyzed by
immunoblot analysis (n = 4).
All data are means SEM. *p < 0.05; **p < 0.01; ***p <
0.0005 in (B), (C) (Dunnets method) and in (E), (F), (G)
(Students unpaired t-test); N.S., not significant.
1.6
1.2
0.8
0.4
0
-
Thus, the finding that PGRN in vitro impairs insulin signaling

downstream of IR and meditates TNF-a-induced IL-6 expression
implicates IL-6 as a mediator of PGRN-induced insulin resistance in adipocytes.
In the present study, we also found that neutralizing antibody
against IL-6 improved PGRN-induced insulin resistance in vivo.
Treatment with IL-6 enhances hepatic insulin resistance (Klover
et al., 2003; Lagathu et al., 2003), while neutralization of IL-6
by administration of antibody specific for IL-6 reduces HFDinduced insulin resistance (Klover et al., 2005). It has been reported also that reduced expression of IL-6 in adipose tissues
by adipocyte-specific deficiency of JNK potentiates hepatic
insulin sensitivity and prevents mice from the development of
insulin resistance by HFD (Sabio et al., 2008). Taken together,
these findings suggest that PGRN in adipose tissues triggers
systemic insulin resistance by elevating levels of IL-6 in adipose
tissues and blood.
SOCS3 has been shown to be a physiological

regulator of insulin signaling in both hepatocytes and adipocytes (Rnn et al., 2007).
SOCS3 expression was found to be elevated
in adipose tissue of obese mice (Emanuelli
et al., 2001). In addition, ablation of Socs3 in liver improved
hepatic insulin sensitivity (Torisu et al., 2007; Sachithanandan
et al., 2010). Considered together with our present findings,
SOCS3 might, therefore, contribute to the development of
systemic insulin resistance by PGRN through elevated levels of
IL-6.
In conclusion, PGRN is a key adipokine that mediates HFDinduced insulin resistance and obesity through IL-6 and may
be a promising therapeutic target for preventing obesity.
Mice
We obtained male C57BL/6J from CLEA Japan (Tokyo, Japan), and ob/ob and
ob/+ mice from Charles River Japan (Yokohama, Japan). Grn+/ mice were
purchased from RIKEN BioResource Center (BRC) (Tsukuba, Japan). Genotyping of Grn / mice was performed as described previously (Kayasuga
et al., 2007). C57BL/6J mice were fed HFD from 7 through 33 weeks of age.
Cell Metabolism
Figure 5. Effects of Grn Knockdown or PGRN

Treatment on Il6 and Socs3 Expressions in 3T3-L1
Adipocytes
Il6
rn
G
2
C
cl
f
Tn
Il6
p
Le
oq
ip
Ad
lu
t4
4
bp
Fa
eb
Socs3
Pp
ar
pa
(A) Quantitative real-time RT-PCR analysis of various

genes in Grn KD adipocytes (n = 6). 3T3-L1 adipocytes
infected with adenovirus carrying GrnshRNA or adenovirus
carrying nontarget shRNA (control) at MOI of 100 were
treated with or without 10 ng/ml TNF-a for 16 hr.
(B) Effect of exogenous PGRN on Il6 mRNA expression. Il6
expression in 3T3-L1 adipocytes treated with indicated
concentrations of recombinant PGRN protein for 16 hr
were examined by quantitative real-time RT-PCR analysis
(n = 6).
(C) Effect of exogenous PGRN on Socs3 mRNA expression (n = 6). Socs3 expression in 3T3-L1 adipocytes
treated with indicated concentrations of recombinant
PGRN protein for 16 hr were examined by quantitative
real-time RT-PCR analysis.
(D) Quantitative real-time RT-PCR analysis of Socs3 in Grn
KD adipocytes (n = 6). Grn KD and control 3T3-L1 adipocytes were treated as in (A).
(E) Effect of Grn KD on TNF-a stimulated STAT3 phosphorylation (n = 6). STAT3 phosphorylation in KD and
control 3T3-L1 adipocytes treated as in (A) was analyzed
by immunoblot analysis.
All data are means SEM. *p < 0.05; **p < 0.01 in (A), (D),
and (E) (Tukey-Kramers method) and in (B) and (C)
(Dunnets method); N.S., not significant.
Socs3
(IBMX), 5 mg/ml insulin, and 1 mM dexamethasone was

added to preadipocytes 2 days after reaching confluence
(day 0). On day 3, the medium was replaced with DMEM
D
E
containing 25 mM glucose, 10% FBS and 5 mg/ml
insulin. From day 5 onward, cells were maintained in
DMEM containing 25 mM glucose and 10% FBS, with
a media change every other day until experimental treatments were initiated. Dexamethasone (20 nM) or TNF-a
(4 ng/ml) was added to mature adipocytes at any time
from day 8 to day 14 of differentiation. Media containing
TNF-a was changed daily for a total incubation period
of 4 days. Media containing dexamethasone was
changed every other day for a total of 8 days. Cells
were then collected by scraping and were lysed in buffer
- +
- - +
- +
containing 6 M guanidine-HCl, 50 mM Tris-HCl (pH 8.0),
+
TNF- 2 mM EDTA, 1 mM phenylmethylsulfonyl fluoride
Control
KD
(PMSF), 10 mg/ml leupeptin, and 10 mg/ml aprotinin)
for NBS reagent labeling, or in TNE buffer [1% (w/w)
Nonidet-P40, 150 mM NaCl, 20 mM Tris-HCl (pH 7.4),
Grn / mice were fed HFD from 5 through 17 weeks of age. The SD (CE-2,
2 mM EDTA, 10 mg/ml leupeptin, 10 mg/ml aprotinin, 5 mM mercaptoethanol,
CLEA Japan) supplied 4.8% of calories as fat with an energy density of
1 mM PMSF, 1 mM Na3VO4, 10 mM Na2MoO4, 50 mM NaF] for immunoblot
analysis.
3.43 kcal/g. The HFD (D12492, Research Diet Inc., NJ) supplied 60% of calories as fat with an energy density of 5.24 kcal/g. Animal care and experimental
NBS Tagging, Peptide Fractionation, and Mass Spectrometry
procedures were approved by the Institutional Animal Care and Use
Committee and carried out according to the Kobe University Experimentation
NBS tagging was performed according to the manufacturers protocol
Regulations.
(13CNBS stable isotope labeling kit-N; Shimadzu Biotech, Kyoto, Japan).
Briefly, each cell lysate (each containing 200 mg of protein) was labeled with
Cell Culture
isotopically 13C0NBS or 13C6NBS reagent. NBS-tagged proteins were then
3T3-L1 cells were purchased from the American Tissue Culture Collection
mixed, reduced, alkylated, and digested by trypsin. NBS-tagged peptides
(Manassas, VA) and maintained at 37 C in a humidified atmosphere of 5%
were enriched and separated by reversed-phase liquid choromatography
CO295% air in Dulbeccos modified Eagles medium (DMEM) containing
(LC-10ADvp mHPLC System; Shimadzu, Kyoto, Japan) as described previ5.6 mM glucose (Wako) supplemented with 10% (vol/vol) heat-inactivated
ously (Matsuo et al., 2009). Eluates were automatically deposited onto MALDI
newborn calf serum (Invitrogen, Carlsbad, CA). Differentiation medium, which
target plates by the LC spotting system (AccuSpot; Shimadzu). These spotted
consists of DMEM containing 25 mM glucose supplemented with 10% fetal
samples were automatically analyzed by MALDI-TOF MS (AXIMA-CFR Plus or
bovine serum (FBS) (Hyclone, Logan, UT), 0.5 mM isobutylmethylxanthine
AXIMA-TOF2; Shimadzu/Kratos, Manchester, UK).
Cell Metabolism
Figure 6. Effects of Grn Deficiency on HFDInduced Elevation of IL-6 and SOCS3 In Vivo
(A) Quantitative real-time RT-PCR analysis of EF of WT

(SD, n = 8; HFD, n = 8) and Grn / mice (SD, n = 7; HFD,
n = 5).
(B) Serum concentrations of IL-6 in WT (SD, n = 19; HFD,
n = 27) and Grn / mice (SD, n = 15; HFD, n = 22).
(C) Quantitative real-time RT-PCR analysis of Socs3 in
EF of WT (SD, n = 8; HFD, n = 8) and Grn / mice (SD, n = 7;
HFD, n = 5).
(D) Quantitative real-time RT-PCR analysis of Socs3 in liver
of WT (SD, n = 6; HFD, n = 8) and Grn / mice (SD, n = 7;
HFD, n = 7).
(E) Quantitative real-time RT-PCR analysis of Il6 in liver of
WT mice (SD, n = 6; HFD, n = 8) and Grn / mice (SD, n = 7;
HFD, n = 7).
All data are means SEM. *p < 0.05; **p < 0.01 (TukeyKramers method); N.S., not significant.
rn
G
r1
Em
2
C
cl
Il6
Tn
ip
Ad
Le
oq
t4
lu
N.S.
WT
MASCOT Distiller. The Mascot search parameters were

as follows: trypsin digestion allowing up to 2 missed cleavages, fixed modifications of 13C0NBS (or 13C6NBS) (W) and
carbamidomethyl (C), variable modifications of oxidation
(M), peptide tolerance of 0.3 Da, and MS/MS tolerance
of 0.5 Da. Search results with p values less than 0.05
were judged as positive identifications.
N.S.
Fractionation of Epididymal White Adipose Tissue

Epididymal white adipose tissues (WAT) dissected from
mice were minced and digested with 2 mg/ml collagenase
P (Roche, Mannheim, Germany) in DMEM containing 1%
BSA and antibiotics for 45 min at 37 C. The digested
tissues were passed through a nylon mesh filter (pore
size, 150 mm) to remove undigested material, and the
filtrates were centrifuged for 5 min at 250 x g. Floating cells
and the pellet was recovered as the mature adipocyte
fraction and the SVF, respectively, and washed twice
with phosphate buffered-saline (pH 7.4).
Relative Socs3 mRNA
Fa
bp
pa
eb
g
C
ar
Pp
Grn
Grn
Grn-/-
Grn-/-
WT
N.S.
Relative Il6 mRNA
Relative Socs3 mRNA
N.S.
N.S.
Treatment of ob/ob Mice with Pioglitazone

Pioglitazone (30 mg/kg) or vehicle (0.25% carboxymethyl
cellulose) was administered orally to 13-week-old ob/ob
mice once daily for 7 consecutive days.
Preparation of Recombinant PGRN

Mouse PGRN (mPGRN) cDNA clone, MGC Image clone,
was purchased from Invitrogen. pCAGIPuro-FLAGmPGRN was constructed by subcloning the insert encoding the mPGRN without signal peptide (amino acid 18 to
589), into pCAGIPuro-FLAG (Satoh-Horikawa et al.,
2000). To prepare CHO-K1 cells stably expressing the
WT
Grn-/FLAG-tagged mPGRN, CHO-K1 cells were transfected
WT
Grn-/with pCAGIPuro-FLAG-mPGRN construct by electroporation. Stably expressed cells were maintained in CD
Relative Quantification and Identification of Differentially Expressed
OptiCHO medium (Invitrogen) supplemented with 10 mg/ml puromycin
Proteins
(SIGMA), 4 mM GlutaMAX (Invitrogen) 1 3 HT supplement (Invitrogen), and
Relative quantification between 13C0NBS- and 13C6NBS-tagged peptides was
10 mg/ml insulin (SIGMA). Culture supernatants were collected and subjected
performed using the proteome analysis assistant software for relative quantito anti-FLAG M1 agarose affinity gel (SIGMA) column. The column was
fication, TWIP Version 1.0 (DYNACOM, Chiba, Japan), referring to a monoisowashed with 50 mM Tris-HCl (pH7.5), 150 mM NaCl and 1 mM CaCl2, and
then eluted with 50 mM Tris-HCl (pH7.5), 150 mM NaCl, and 2 mM EDTA.
topic mass list from MASCOT Distiller Ver. 1.1.2 (Matrix Science, London, UK)
The eluted proteins were dialyzed against PBS.
as described previously (Matsuo et al., 2009). Threshold values of 13C6/13C0
ratios in NBS-tagged peptide pairs were set to either larger than 1.25 or less
than 0.8. Candidate peptides having significant difference in peptide pair ratios
were selected and further subjected to MS/MS analysis (AXIMA-QIT-TOF MS;
The data are expressed as means SEM. Comparisons were made using
Shimadzu / Kratos). Proteins were identified by MASCOT MS/MS Ion Search
Students t-test, Dunnets method or Tukey-Kramers method as indicated
algorithm (Version 2.0; Matrix Science) using mass lists generated by
in the legends. A P value of < 0.05 was considered statistically significant.
Cell Metabolism
Figure 7. Effects of Neutralizing Antibody against

IL-6 on Insulin Resistance Induced by PGRN In Vivo
Recombinant mouse PGRN (rmPGRN, i.p. 20 mg/day) or
PBS (control) was administered to C57BL/6J mice once
daily for 21 days (n = 7). Neutralizing antibody against IL-6
(n = 7) or IgG (n = 6) was administered to the mice 72 hr
before ITT. Insulin sensitivity was assessed by inverse
AUC (right) of ITT (left) (insulin injection: 0.3 IU/kg).
All data are means SEM. *p < 0.05 (Tukey-Kramers
method); N.S., not significant.
Supplemental Information includes Supplemental Experimental Procedures,
four figures, and two tables and appears with this article online at doi:10.
1016/j.cmet.2011.12.002.
Emanuelli, B., Peraldi, P., Filloux, C., Chavey, C., Freidinger, K., Hilton, D.J.,
Hotamisligil, G.S., and Van Obberghen, E. (2001). SOCS-3 inhibits
insulin signaling and is up-regulated in response to tumor necrosis
factor-a in the adipose tissue of obese mice. J. Biol. Chem. 276, 47944
47949.
ACKNOWLEDGMENTS
Eriksen, J.L., and Mackenzie, I.R. (2008). Progranulin: normal function and role
in neurodegeneration. J. Neurochem. 104, 287297.
We thank for G.K. Honkawa for assistance with the manuscript. This study was
supported by a CREST grant from the Japan Science and Technology Agency
and Grant-in-Aid for Scientific Research and by grants for the Kobe Translational Research Cluster, the Knowledge Cluster Initiative, and the Global
Centers of Excellence Program Global Center of Excellence for Education
and Research on Signal Transduction Medicine in the Coming Generation
from the Ministry of Education, Culture, Sport, Science and Technology,
Japan.
Received: November 4, 2010
Revised: September 23, 2011
Published: January 3, 2012
Fasshauer, M., Kralisch, S., Klier, M., Lossner, U., Bluher, M., Klein, J., and
Paschke, R. (2004). Insulin resistance-inducing cytokines differentially
regulate SOCS mRNA expression via growth factor- and Jak/Stat-signaling
pathways in 3T3-L1 adipocytes. J. Endocrinol. 181, 129138.
Fried, S.K., Bunkin, D.A., and Greenberg, A.S.J. (1998). Omental and subcutaneous adipose tissues of obese subjects release interleukin-6: depot difference and regulation by glucocorticoid. J. Clin. Endocrinol. Metab. 83,
847850.
Friedman, J.M., and Halaas, J.L. (1998). Leptin and the regulation of body
weight in mammals. Nature 395, 763770.
Guilherme, A., Virbasius, J.V., Puri, V., and Czech, M.P. (2008). Adipocyte
dysfunctions linking obesity to insulin resistance and type 2 diabetes. Nat.
Rev. Mol. Cell Biol. 9, 367377.
REFERENCES
He, Z., and Bateman, A. (2003). Progranulin (granulin-epithelin precursor,

PC-cell-derived growth factor, acrogranin) mediates tissue repair and tumorigenesis. J. Mol. Med. 81, 600612.
Anakwe, O.O., and Gerton, G.L. (1990). Acrosome biogenesis begins during
meiosis: evidence from the synthesis and distribution of an acrosomal glycoprotein, acrogranin, during guinea pig spermatogenesis. Biol. Reprod. 42,
317328.
He, Z., Ismail, A., Kriazhev, L., Sadvakassova, G., and Bateman, A. (2002).
Progranulin (PC-cell-derived growth factor/acrogranin) regulates invasion
and cell survival. Cancer Res. 62, 55905596.
Bateman, A., and Bennett, H.P.J. (1998). Granulins: the structure and function
of an emerging family of growth factors. J. Endocrinol. 158, 145151.
Berg, A.H., Combs, T.P., Du, X., Brownlee, M., and Scherer, P.E. (2001). The
adipocyte-secreted protein Acrp30 enhances hepatic insulin action. Nat.
Med. 7, 947953.
Caro, J.F., and Amatruda, J.M. (1982). Glucocorticoid-induced insulin resistance: the importance of postbinding events in the regulation of insulin binding,
action, and degradation in freshly isolated and primary cultures of rat hepatocytes. J. Clin. Invest. 69, 866875.
Chiellini, C.A., Bertacca, A., Novelli, S.E., Gorgun, C.Z., Ciccarone, A.,
Giordano, A., Xu, H., Soukas, A., Costa, M., Gandini, D., et al. (2002).
Obesity modulates the expression of haptoglobin in the white adipose tissue
via TNFalpha. J. Cell. Physiol. 190, 251258.
Cruts, M., and Van Broeckhoven, C. (2008). Loss of progranulin function in
frontotemporal lobar degeneration. Trends Genet. 24, 186194.
do Nascimento, C.O., Hunter, L., and Trayhurn, P. (2004). Regulation of haptoglobin gene expression in 3T3-L1 adipocytes by cytokines, catecholamines,
and PPARgamma. Biochem. Biophys. Res. Commun. 313, 702708.
Hotamisligil, G.S. (2006). Inflammation and metabolic disorders. Nature 444,

860867.
Hotamisligil, G.S., Murray, D.L., Choy, L.N., and Spiegelman, B.M. (1994).
Tumor necrosis factor a inhibits signaling from the insulin receptor. Proc.
Natl. Acad. Sci. USA 91, 48544858.
Hotamisligil, G.S., Arner, P., Caro, J.F., Atkinson, R.L., and Spiegelman, B.M.
(1995). Increased adipose tissue expression of tumor necrosis factor-a in
human obesity and insulin resistance. J. Clin. Invest. 95, 24092415.
Howard, J.K., and Flier, J.S. (2006). Attenuation of leptin and insulin signaling
by SOCS proteins. Trends Endocrinol. Metab. 17, 365371.
Inoue, H., Ogawa, W., Asakawa, A., Okamoto, Y., Nishizawa, A., Matsumoto,
M., Teshigawara, K., Matsuki, Y., Watanabe, E., Hiramatsu, R., et al. (2006).
Role of hepatic STAT3 in brain-insulin action on hepatic glucose production.
Cell Metab. 3, 267275.
Kahn, B.B., and Flier, J.S. (2000). Obesity and insulin resistance. J. Clin. Invest.
106, 473481.
Kayasuga, Y., Chiba, S., Suzuki, M., Kikusui, T., Matsuwaki, T., Yamanouchi,
K., Kotaki, H., Horai, R., Iwakura, Y., and Nishihara, M. (2007). Alteration of
Cell Metabolism
behavioural phenotype in mice by targeted disruption of the progranulin gene.

Behav. Brain Res. 185, 110118.
Kern, P.A., Ranganathan, S., Li, C., Wood, L., and Ranganathan, G. (2001).
Adipose tissue tumor necrosis factor and interleukin-6 expression in human
obesity and insulin resistance. Am. J. Physiol. Endocrinol. Metab. 280,
E745E751.
Kessenbrock, K., Frohlich, L., Sixt, M., Lammermann, T., Pfister, H.,
Bateman, A., Belaaouaj, A., Ring, J., Ollert, M., Fassler, R., and Jenne,
D.E. (2008). Proteinase 3 and neutrophil elastase enhance inflammation in
mice by inactivating antiinflammatory progranulin. J. Clin. Invest. 118,
24382447.
Klover, P.J., Zimmers, T.A., Koniaris, L.G., and Mooney, R.A. (2003). Chronic
exposure to interleukin-6 causes hepatic insulin resistance in mice. Diabetes
52, 27842789.
Klover, P.J., Clementi, A.H., and Mooney, R.A. (2005). Interleukin-6 depletion
selectively improves hepatic insulin action in obesity. Endocrinology 146,
34173427.
Kuyama, H., Watanabe, M., Toda, C., Ando, E., Tanaka, K., and Nishimura, O.
(2003). An approach to quantitative proteome analysis by labeling tryptophan
residues. Rapid Commun. Mass Spectrom. 17, 16421650.
Lagathu, C., Bastard, J.P., Auclair, M., Maachi, M., Capeau, J., and Caron, M.
(2003). Chronic interleukin-6 (IL-6) treatment increased IL-6 secretion and
induced insulin resistance in adipocyte: prevention by rosiglitazone.
Biochem. Biophys. Res. Commun. 311, 372379.
Lago, F., Dieguez, C., Gomez-Reino, J., and Gualillo, O. (2007). Adipokines as
emerging mediators of immune response and inflammation. Nat. Clin. Pract.
Rheumatol. 3, 716724.
Lu, R., and Serrero, G. (2001). Mediation of estrogen mitogenic effect in human
breast cancer MCF-7 cells by PC-cell-derived growth factor (PCDGF/granulin
precursor). Proc. Natl. Acad. Sci. USA 98, 142147.
Sabio, G., Das, M., Mora, A., Zhang, Z., Jun, J.Y., Ko, H.J., Barrett, T., Kim,
J.K., and Davis, R.J. (2008). A stress signaling pathway in adipose tissue regulates hepatic insulin resistance. Science 322, 15391543.
Sachithanandan, N., Fam, B.C., Fynch, S., Dzamko, N., Watt, M.J., Wormald,
S., Honeyman, J., Galic, S., Proietto, J., Andrikopoulos, S., et al. (2010). Liverspecific suppressor of cytokine signaling-3 deletion in mice enhances hepatic
insulin sensitivity and lipogenesis resulting in fatty liver and obesity.
Hepatology 52, 16321642.
Sakoda, H., Ogihara, T., Anai, M., Funaki, M., Inukai, K., Katagiri, H.,
Fukushima, Y., Onishi, Y., Ono, H., Fujishiro, M., et al. (2000).
Dexamethasone-induced insulin resistance in 3T3-L1 adipocytes is due to
inhibition of glucose transport rather than insulin signal transduction.
Diabetes 49, 17001708.
Satoh-Horikawa, K., Nakanishi, H., Takahashi, K., Miyahara, M., Nishimura,
M., Tachibana, K., Mizoguchi, A., and Takai, Y. (2000). Nectin-3, a new
member of immunoglobulin-like cell adhesion molecules that shows homophilic and heterophilic cell-cell adhesion activities. J. Biol. Chem. 275,
1029110299.
Senn, J.J., Klover, P.J., Nowak, I.A., and Mooney, R.A. (2002).
Interleukin-6 induces cellular insulin resistance in hepatocytes. Diabetes
51, 33913399.
Senn, J.J., Klover, P.J., Nowak, I.A., Zimmers, T.A., Koniaris, L.G., Furlanetto,
R.W., and Mooney, R.A. (2003). Suppressor of cytokine signaling-3 (SOCS-3),
a potential mediator of interleukin-6-dependent insulin resistance in hepatocytes. J. Biol. Chem. 278, 1374013746.
Shi, H., Tzameli, I., Bjrbaek, C., and Flier. (2004). Suppressor of cytokine
signaling 3 is a physiological regulator of adipocyte insulin signaling. J. Biol.
Chem. 279, 3473334740.
Shoelson, S.E., Herrero, L., and Naaz, A. (2007). Obesity, inflammation, and
insulin resistance. Gastroenterology 132, 21692180.
Matsuo, E., Watanabe, M., Kuyama, H., and Nishimura, O. (2009). A new
strategy for protein biomarker discovery utilizing 2-nitrobenzenesulfenyl
(NBS) reagent and its applications to clinical samples. J. Chromatogr. B
Analyt. Technol. Biomed. Life Sci. 877, 26072614.
Tang, W., Lu, Y., Tian, Q.Y., Zhang, Y., Guo, F.J., Liu, G.Y., Syed, N.M., Lai, Y.,
Lin, E.A., Kong, L., et al. (2011). The growth factor progranulin binds to TNF
receptors and is therapeutic against inflammatory arthritis in mice. Science
332, 478484.
Mohamed-Ali, V., Goodrick, S., Rawesh, A., Katz, D.R., Miles, J.M., Yudkin,
J.S., Klein, S., and Coppack, S.W. (1997). Subcutaneous adipose tissue
releases interleukin-6, but not tumor necrosis factor-a, in vivo. J. Clin.
Endocrinol. Metab. 82, 41964200.
Torisu, T., Sato, N., Yoshiga, D., Kobayashi, T., Yoshioka, T., Mori, H., Iida, M.,
and Yoshimura, A. (2007). The dual function of hepatic SOCS3 in insulin resistance in vivo. Genes Cells 12, 143154.
Mooney, R.A. (2007). Counterpoint: Interleukin-6 does not have a beneficial

role in insulin sensitivity and glucose homeostasis. J. Appl. Physiol. 102,
816818, discussion 818819.
Okamura, N., Masuda, T., Gotoh, A., Shirakawa, T., Terao, S., Kaneko, N.,
Suganuma, K., Watanabe, M., Matsubara, T., Seto, R., et al. (2008).
Quantitative proteomic analysis to discover potential diagnostic markers
and therapeutic targets in human renal cell carcinoma. Proteomics 8, 3194
3203.
Turnbow, M.A., Keller, S.R., Rice, K.M., and Garner, C.W. (1994).
Dexamethasone down-regulation of insulin receptor substrate-1 in 3T3-L1
adipocytes. J. Biol. Chem. 269, 25162520.
Ueki, K., Kondo, T., and Kahn, C.R. (2004). Suppressor of cytokine signaling 1
(SOCS-1) and SOCS-3 cause insulin resistance through inhibition of tyrosine
phosphorylation of insulin receptor substrate proteins by discrete mechanisms. Mol. Cell. Biol. 24, 54345446.
Olefsky, J.M. (2000). Treatment of insulin resistance with peroxisome proliferator-activated receptor g agonists. J. Clin. Invest. 106, 467472.
Uysal, K.T., Wiesbrock, S.M., Marino, M.W., and Hotamisligil, G.S. (1997).
Protection from obesity-induced insulin resistance in mice lacking TNF-a function. Nature 389, 610614.
Perkins, D.N., Pappin, D.J., Creasy, D.M., and Cottrell, J.S. (1999). Probabilitybased protein identification by searching sequence databases using mass
spectrometry data. Electrophoresis 20, 35513567.
van Putten, J.P., Wieringa, T., and Krans, H.M. (1985). Corticosteroids as longterm regulators of the insulin effectiveness in mouse 3T3 adipocytes.
Pickup, J.C., Mattock, M.B., Chusney, G.D., and Burt, D. (1997). NIDDM
as a disease of the innate immune system: association of acute-phase reactants and interleukin-6 with metabolic syndrome X. Diabetologia 40, 1286
1292.
Waki, H., and Tontonoz, P. (2007). Endocrine functions of adipose tissue.

Annu. Rev. Pathol. 2, 3156.
Qi, D., and Rodrigues, B. (2007). Glucocorticoids produce whole body insulin
resistance with changes in cardiac metabolism. Am. J. Physiol. Endocrinol.
Metab. 292, E654E667.
Rnn, S.G., Billestrup, N., and Mandrup-Poulsen, T. (2007). Diabetes and
suppressors of cytokine signaling proteins. Diabetes 56, 541548.
Rotter, V., Nagaev, I., and Smith, U. (2003). Interleukin-6 (IL-6) induces insulin
resistance in 3T3-L1 adipocytes and is, like IL-8 and tumor necrosis factor-a,
overexpressed in human fat cells from insulin-resistant subjects. J. Biol.
Chem. 278, 4577745784.
Wallenius, V., Wallenius, K., Ahren, B., Rudling, M., Carlsten, H., Dickson, S.L.,
Ohlsson, C., and Jansson, J.O. (2002). Interleukin-6-deficient mice develop
mature-onset obesity. Nat. Med. 8, 7579.
Watanabe, M., Takemasa, I., Kawaguchi, N., Miyake, M., Nishimura, N.,
Matsubara, T., Matsuo, E., Sekimoto, M., Nagai, K., Matsuura, N., et al.
(2008). An application of the 2-nitrobenzenesulfenyl method to proteomic
profiling of human colorectal carcinoma: A novel approach for biomarker
discovery. Proteomics Clin Appl 2, 925935.
Weisberg, S.P., McCann, D., Desai, M., Rosenbaum, M., Leibel, R.L., and
Ferrante, A.W., Jr. (2003). Obesity is associated with macrophage accumulation in adipose tissue. J. Clin. Invest. 112, 17961808.
Cell Metabolism
Wellen, K.E., and Hotamisligil, G.S. (2005). Inflammation, stress, and diabetes.
J. Clin. Invest. 5, 11111119.
Xu, H., Barnes, G.T., Yang, Q., Tan, G., Yang, D., Chou, C.J., Sole, J., Nichols,
A., Ross, J.S., Tartaglia, L.A., and Chen, H. (2003). Chronic inflammation in fat
plays a crucial role in the development of obesity-related insulin resistance.
J. Clin. Invest. 112, 18211830.
Youn, B.S., Bang, S.I., Kloting, N., Park, J.W., Lee, N., Oh, J.E., Pi, K.B., Lee,
T.H., Ruschke, K., Fasshauer, M., et al. (2009). Serum progranulin concentra-
tions may be associated with macrophage infiltration into omental adipose

tissue. Diabetes 58, 627636.
Zanocco-Marani, T., Bateman, A., Romano, G., Valentinis, B., He, Z.H., and
Baserga, R. (1999). Biological activities and signaling pathways of the granulin/epithelin precursor. Cancer Res. 59, 53315340.
Zhu, J., Nathan, C., Jin, W., Sim, D., Ashcroft, G.S., Wahl, S.M., Lacomis, L.,
Erdjument-Bromage, H., Tempst, P., Wright, C.D., and Ding, A. (2002).
Conversion of proepithelin to epithelins: roles of SLPI and elastase in host
defense and wound repair. Cell 111, 867878.
Cell Metabolism
Article
p53-Induced Adipose Tissue Inflammation
Is Critically Involved in the Development
of Insulin Resistance in Heart Failure
Ippei Shimizu,1,5 Yohko Yoshida,1,5 Taro Katsuno,1 Kaoru Tateno,1 Sho Okada,1 Junji Moriya,1 Masataka Yokoyama,1
Aika Nojima,1 Takashi Ito,1 Rudolf Zechner,2 Issei Komuro,3 Yoshio Kobayashi,1 and Tohru Minamino1,4,*
1Department
of Cardiovascular Science and Medicine, Chiba University Graduate School of Medicine, Chiba 260-8670, Japan
of Molecular Biosciences, University of Graz, A-8010 Graz, Austria
3Department of Cardiovascular Medicine, Osaka University School of Medicine, Osaka 565-0871, Japan
4PRESTO, Japan Science and Technology Agency, Saitama 332-0012, Japan
*Correspondence: t_minamino@yahoo.co.jp
DOI 10.1016/j.cmet.2011.12.006
2Institute
SUMMARY
Several clinical studies have shown that insulin resistance is prevalent among patients with heart failure,
but the underlying mechanisms have not been fully
elucidated. Here, we report a mechanism of insulin
resistance associated with heart failure that involves
upregulation of p53 in adipose tissue. We found
that pressure overload markedly upregulated p53
expression in adipose tissue along with an increase
of adipose tissue inflammation. Chronic pressure
overload accelerated lipolysis in adipose tissue. In
the presence of pressure overload, inhibition of lipolysis by sympathetic denervation significantly downregulated adipose p53 expression and inflammation,
thereby improving insulin resistance. Likewise, disruption of p53 activation in adipose tissue attenuated
inflammation and improved insulin resistance but
also ameliorated cardiac dysfunction induced by
chronic pressure overload. These results indicate
that chronic pressure overload upregulates adipose
tissue p53 by promoting lipolysis via the sympathetic
nervous system, leading to an inflammatory response of adipose tissue and insulin resistance.
INTRODUCTION
The p53 tumor suppressor pathway coordinates DNA repair,
cell-cycle arrest, apoptosis, and senescence to preserve
genomic stability and prevent oncogenesis. Activation of p53 is
driven by a wide variety of stress signals that have the potential
to promote tumor formation, such as DNA damage, telomere
shortening, oxidative stress, and oncogene activation (Harris
and Levine, 2005; Meek, 2009; Vousden and Prives, 2009).
Recently, the contribution of p53 to many undesirable aspects
of aging and age-associated diseases, such as cardiovascular
and metabolic disorders, has been recognized (Royds and Iacopetta, 2006; Vousden and Lane, 2007). It has been reported that
aging is associated with an increase of the p53-mediated transcriptional activity (Edwards et al., 2007) and that slight constitutive overactivation of p53 is associated with premature aging in
mice (Maier et al., 2004; Tyner et al., 2002). Activation of p53
has also been observed in aged vessels and failing hearts and
has been implicated in atherosclerosis and heart failure (Minamino and Komuro, 2007, 2008; Sano et al., 2007). Recent findings have indicated a role of p53 in determining the response
of cells to nutrient stress and in regulating metabolism (Vousden
and Ryan, 2009). It has also been demonstrated that excessive
calorie intake induces p53-induced inflammation in adipose
tissue, leading to insulin resistance and diabetes in mice (Minamino et al., 2009).
A close link between heart failure and diabetes has long been
recognized in the clinical setting (Ashrafian et al., 2007; Lopaschuk et al., 2007; Witteles and Fowler, 2008). Many mechanisms have been suggested to explain the increased incidence
of heart failure in diabetic patients, including the hypertrophic
influence of insulin, the adverse effects of hyperglycemia,
increased oxidative stress, and hyperactivity of neurohumoral
systems, such as the renin-angiotensin-aldosterone system
and the adrenergic system. Recently, increasing attention has
been paid to insulin resistance as a distinct cause of cardiac
dysfunction and heart failure in diabetic patients. A study of
Swedish patients without prior cardiac dysfunction found that
insulin resistance predicted the subsequent onset of heart failure
independently of established risk factors (Ingelsson et al., 2005).
In another clinical study, the plasma level of proinsulin (a marker
of insulin resistance) was found to be higher in patients who
subsequently developed heart failure than in control patients
20 years before the actual diagnosis of heart failure (Arnlov
et al., 2001). These findings indicate that insulin resistance
precedes heart failure rather than being a consequence of it.
Evidence has emerged that myocardial insulin resistance is
central to altered metabolism in the failing heart and may play
a crucial role in the development of heart failure (Ashrafian
et al., 2007; Lopaschuk et al., 2007; Witteles and Fowler,
2008). The adaptive response of the failing heart involves
a complex series of enzymatic shifts and changes in the regulation of transcriptional factors, which result in an increase of
glucose metabolism and a decrease of fatty acid metabolism
Cell Metabolism
Adipose Inflammation in Heart Failure
to maximize the efficacy of energy production (Neubauer, 2007).

Insulin resistance of the myocardium inhibits these adaptive
responses, leading to increased reliance on fatty acid metabolism. This increases oxygen consumption and decreases cardiac
function, raising the potential for lipotoxicity in the heart (Sharma
et al., 2007; Young et al., 2002). Another line of evidence indicates that insulin signaling is upregulated in the failing heart
and that excessive cardiac insulin signaling exacerbates systolic
dysfunction (Shimizu et al., 2010).
Moreover, there is increasing evidence that heart failure reciprocally augments the risk of insulin resistance and clinical diabetes (Ashrafian et al., 2007). Insulin resistance and abnormal
glucose metabolism are very common in heart failure patients,
being identified in 43% of these patients, and such abnormalities
are associated with decreased cardiac function (Suskin et al.,
2000). Surprisingly, the link between heart failure and insulin
resistance grows stronger when patients with ischemic heart
disease are excluded (Witteles and Fowler, 2008). Heart failure
also predicts the development of type 2 diabetes in a graded
way (Tenenbaum et al., 2003). Although the above mentioned
clinical evidence supports a role of insulin resistance in the
occurrence of heart failure, evidence for the reciprocal statement
that heart failure promotes insulin resistance is largely associative. Moreover, the role of heart failure in the promotion of insulin
resistance has been demonstrated by only a few animal studies
(Nikolaidis et al., 2004; Shimizu et al., 2010) and the underlying
mechanisms are largely speculative.
Here, we studied the role of heart failure in the development of
insulin resistance and sought to elucidate the molecular mechanisms involved. We found that insulin resistance developed in
two murine models of heart failure, a chronic pressure overload
model and a myocardial infarction model. Heart failure markedly
upregulated p53 expression in adipose tissue in association with
increased inflammation of adipose tissue. Heart failure accelerated lipolysis in adipose tissue, whereas inhibition of lipolysis
by sympathetic denervation or treatment with a lipase inhibitor
significantly downregulated adipose tissue p53 expression and
inflammation, thereby improving insulin resistance. Likewise,
disruption of p53 activation in adipose tissue not only ameliorated inflammation in this tissue and improved insulin resistance
but also improved cardiac dysfunction associated with heart
failure. We conclude that heart failure upregulates p53 in adipose
tissue by promoting lipolysis via activation of the sympathetic
nervous system, leading to an inflammatory response of adipose
tissue and insulin resistance. Our results indicate that inhibition
of p53-induced adipose inflammation is a potential target for
treating metabolic abnormalities and systolic dysfunction in
patients with heart failure.
RESULTS
Pressure Overload Induces Adipose Tissue
Inflammation and Insulin Resistance
To examine the effect of cardiac pressure overload on glucose
homeostasis, we produced transverse aortic constriction (TAC)
in 11-week-old mice. In this mouse model, systolic cardiac
function deteriorated significantly along with left ventricular
(LV) dilatation 26 weeks after surgery (Figure S1A available online). The insulin tolerance test (ITT) and the glucose tolerance
test (GTT) showed that insulin sensitivity and glucose tolerance

were impaired at 46 weeks after TAC (Figure 1A) without any
change of food intake (Figure S1B). In patients with metabolic
disorders, the recruitment of inflammatory macrophages to
adipose tissue has been shown to increase the production of
proinflammatory cytokines, such as tumor necrosis factor
(TNF)-a and chemokine (CC motif) ligand 2 (CCL2), also known
as monocyte chemoattractant protein-1 (MCP-1), leading to the
development of systemic insulin resistance (Hotamisligil et al.,
1993; Kamei et al., 2006; Weisberg et al., 2003). Therefore, we
investigated whether pressure overload provokes adipose tissue
inflammation. Examination of hematoxylin- and eosin-stained
sections demonstrated the infiltration of mononuclear cells
into visceral fat, with most of these cells being identified as
macrophages by immunofluorescent staining for Mac3 (Figure 1B). Consistent with these results, expression of a marker
for macrophages (Egf-like module containing, mucin-like, hormone receptor-like 1; EMR1) and production of proinflammatory
cytokines were significantly upregulated in the adipose tissue of
TAC mice along with a decrease of adiponectin (Figure 1C)
compared with sham-operated mice. Treatment of TAC mice
with a neutralizing antibody for Tnf-a significantly improved
insulin resistance and glucose intolerance, suggesting a crucial
role in the upregulation of proinflammatory cytokines in the
development of metabolic abnormalities during heart failure
(Figure S1C).
Pressure Overload Increases Lipolysis and Induces
p53-Dependent Inflammation in Adipose Tissue
during Heart Failure
Computed tomography (CT) showed a significant decrease of
visceral fat after the creation of pressure overload (Figure 1D).
It is well accepted that sympathetic activity increases with heart
failure (Floras, 2009), and norepinephrine regulates lipolysis in
adipose tissue. We found that the norepinephrine levels of
plasma and adipose tissue increased significantly and plasma
fatty acid levels were markedly elevated in TAC mice compared
with sham-operated mice, suggesting acceleration of lipolysis
via the sympathetic nervous system in response to pressure
overload (Figure 1E). It has been reported that exposure to an
excess of fatty acids leads to p53 activation in various cells
(Zeng et al., 2008) and that p53 is crucially involved in the
regulation of adipose tissue inflammation in obese animals
(Minamino et al., 2009). Therefore, we hypothesized that chronic
pressure overload promotes lipolysis and the resultant increase
of fatty acids leads to p53-induced inflammation in adipose
tissue.
Consistent with this concept, we found that p53 expression
was upregulated in the adipose tissue of TAC mice at 24 weeks
after surgery and the change was sustained until 6 weeks
(Figures 2A and S2A). To further investigate the role of adipose
tissue p53 in the response to pressure overload, we performed
TAC in adipocyte-specific p53 knockout (adipo-p53 KO) mice.
The pressure overload-induced increase of p53 expression
was attenuated in adipo-p53 KO mice compared with littermate
controls (Figure S2B). Production of proinflammatory cytokines
as well as cyclin-dependent kinase inhibitor 1A (Cdkn1a) expression was also decreased in adipo-p53 KO mice, along with
a decline in the infiltration of macrophages into visceral fat
Cell Metabolism
Figure 1. Pressure Overload Induces Systemic Insulin Resistance and Adipose Tissue Lipolysis and Inflammation
(A) Insulin tolerance test (ITT) and glucose tolerance test (GTT) in mice at 6 weeks after sham operation (Sham) or TAC (n = 30).
(B) Hematoxylin and eosin staining of adipose tissues of mice at 6 weeks after sham operation (Sham) or TAC (upper panel). In the lower panel, the infiltration of
macrophages was evaluated by immunofluorescent staining for Mac3 (green). Nuclei were stained with Hoechst dye (blue). Scale bar, 50 mm. The right graph
indicates the quantitative data on the infiltration of macrophages (n = 5).
(C) Real-time PCR assessing the expression of Emr1, Tnf (Tnfa), Ccl2 (MCP1), and Adipoq (Adiponectin) levels in adipose tissues of mice at 6 weeks after sham
operation (Sham) or TAC (n = 10).
(D) CT analysis of mice at 6 weeks after sham operation (Sham) or TAC. The graph shows the ratio of visceral fat tissue weight estimated by CT to whole body
weight (n = 7).
(E) Norepinephrine level in adipose tissue (left) and plasma (middle), and plasma free fatty acid (FFA) level (right) of mice at 6 weeks after sham operation (Sham) or
TAC (n = 10). Data are shown as the means S.E.M. *p < 0.05, **p < 0.01.
Cell Metabolism
Cell Metabolism
(Figures 2B and 2C). Consequently, adipo-p53 KO mice showed

improved insulin sensitivity and glucose tolerance after induction
of pressure overload compared with littermate controls (Figure 2D) without any change of food intake (Figure S2C). These
results suggest that p53 has a critical role in the regulation of
adipose tissue inflammation and insulin resistance during
pressure overload. In contrast, a decrease of fat mass and an
increase of plasma free fatty acids were observed to a similar
extent in both adipo-p53 KO and control mice after TAC (Figures
S2DS2F), suggesting that pressure overload accelerates lipolysis in a p53-independent manner.
Pressure Overload Promotes Lipolysis via the
Sympathetic Nervous System
We inhibited sympathetic activity in epididymal fat tissue by
surgical denervation and then performed TAC. As a result,
surgical denervation effectively inhibited an increase of the
norepinephrine level of adipose tissue and attenuated lipolysis
after the onset of pressure overload (Figures S3A and S3B and
data not shown). Histological examination of adipose tissue
showed that infiltration of inflammatory cells after TAC was
attenuated by denervation (Figures S3C and S3D). Likewise,
disruption of the sympathetic efferent nerves significantly
reduced pressure overload-induced upregulation of Emr1, a
proinflammatory cytokine expression in adipose tissue (Figure 3A), and this reduction was associated with significant
improvement of insulin resistance and glucose tolerance in
TAC mice (Figure 3B). Surgical denervation attenuated pressure
overload-induced upregulation of p53 and Cdkn1a expression
in adipose tissue (Figures 3A and 3C). We also pharmacologically inhibited the sympathetic activity in adipose tissue by
injecting guanethidine directly into epididymal fat and then
performed TAC. As a result, pharmacological denervation also
significantly inhibited lipolysis (Figures S3A and S3B) and attenuated upregulation of p53 and Cdkn1a expression and inflammation in adipose tissues (Figures S3C, S3D, S4A and S4B).
Mice treated with guanethidine showed better insulin sensitivity
and glucose tolerance after creation of pressure overload (Figure S4C), indicating that pressure overload-induced activation
of the sympathetic nervous system accelerates lipolysis and,
thus, leads to adipose tissue inflammation and insulin resistance
in TAC mice.
Role of Lipolysis in the Regulation of Adipose p53
Expression and Inflammation
To examine the role of lipolysis in influencing adipose tissue
expression of p53 and inflammation after TAC, we inhibited lipolysis by administering acipimox, a selective inhibitor of lipolysis,
to mice with TAC. Treatment with acipimox markedly inhibited
lipolysis and also reduced infiltration of inflammatory cells into

adipose tissue during pressure overload (Figures S3AS3D). Inhibition of lipolysis also significantly reduced pressure overloadinduced upregulation of Emr1 and proinflammatory cytokine
production in adipose tissue (Figure 4A), along with significant
improvement of insulin resistance and glucose intolerance in
TAC mice (Figure 4B). Furthermore, treatment with acipimox
attenuated pressure overload-induced upregulation of p53 and
Cdkn1a expression in adipose tissue (Figures 4A and 4C), confirming a close relationship between lipolysis and p53 expression.
Next, we promoted lipolysis by administering isoproterenol
to mice via an infusion pump. Treatment with isoproterenol
significantly decreased the visceral fat mass and increased
plasma fatty acid levels (Figures S5AS5C) and increased p53
expression in adipose tissue (Figure 5A). Isoproterenol also
induced adipose tissue inflammation (Figures 5B and 5C). To
further investigate the role of lipolysis in the regulation of p53
expression and inflammation in adipose tissue, we tested the
influence of deleting adipose triglyceride lipase (patatin-like
phospholipase domain containing protein 2, encoded by Pnpla2;
hereafter referred to as Atgl) on adipose tissue expression of
p53. It has been reported that Atgl homozygous KO mice show
massive accumulation of lipids in the heart, causing cardiac
dysfunction and premature death (Haemmerle et al., 2006).
When we generated TAC mice, we also noted that cardiac function was worse and LV enlargement was more marked in Atgl
heterozygous KO mice compared with their littermates (Figure S5D). In fact, most of the KO mice died of heart failure within
4 weeks after TAC. Therefore, we utilized Atgl-deficient adipose
tissue for ex vivo experiments. We cultured epididymal fat pad
tissues from Atgl KO mice or wild-type littermates and examined
the effect of isoproterenol on p53 expression. Treatment of wildtype fat pads with isoproterenol significantly induced lipolysis
(Figure 5D) and upregulated the expression of both p53 and
Cdkn1a expression (Figures 5E and 5F). Disruption of Atgl inhibited isoproterenol-induced lipolysis (Figure 5D) and prevented the upregulation of adipose p53 and Cdkn1a expression
(Figures 5E and 5F), suggesting a crucial role of lipolysis in the
regulation of p53 expression and inflammation in adipose tissue.
Myocardial Infarction Induces Adipose Tissue
Inflammation and Insulin Resistance
To investigate whether myocardial infarction (MI) induced insulin
resistance, we created MI in 11-week-old mice and assessed
the animals 6 weeks after surgery. Insulin sensitivity and
glucose tolerance were significantly impaired in MI mice compared with sham-operated mice (Figure S5E). Significant loss
of fat tissue was also observed in MI mice (Figures S5F and
S5G) and this was associated with upregulation of adipose
Figure 2. p53-Dependent Adipose Tissue Inflammation Provokes Systemic Insulin Resistance during Heart Failure
(A) Expression of p53 was examined in adipose tissues of mice by western blot analysis at indicated time points after sham operation (Sham) or TAC. Actin was
used as an equal loading control. The graph indicates the quantitative data on p53 expression (n = 3).
(B) Real-time PCR assessing the expression of Emr1, Tnf (Tnfa), Ccl2 (MCP1), and Cdkn1a (p21) levels in adipose tissue of adipocyte-specific p53-deficient mice
(adipo-p53 KO) and littermate controls (Cont) at 6 weeks after sham operation or TAC procedure (n = 12).
(C) Hematoxylin and eosin staining of adipose tissues of adipocyte-specific p53-deficient mice (adipo-p53 KO) and littermate controls (Cont) at 6 weeks after
sham operation (Sham) or TAC procedure. Scale bar, 50 mm. The right graph indicates the quantitative data on the infiltration of macrophages (n = 4).
(D) Insulin tolerance test (ITT) and glucose tolerance test (GTT) in adipocyte-specific p53-deficient mice (KO) and littermate controls (Cont) at 6 weeks after sham
operation (Sham) or TAC procedure (n = 16). Data are shown as the means S.E.M. *p < 0.05, **p < 0.01.
Cell Metabolism
Figure 3. Surgical Transection of the Sympathetic Nerves Attenuates Adipose Tissue Inflammation and Systemic Insulin Resistance
(A) Real-time PCR assessing the expression of Emr1, Tnf (Tnfa), Ccl2 (MCP1), and Cdkn1a (p21) levels in adipose tissues of mice at 6 weeks after sham operation
(Sham) or TAC with or without surgical transection of the sympathetic nerves (Denervation) of epidydimal fat (n = 8).
(B) Insulin tolerance test (ITT) and glucose tolerance test (GTT) of mice at 6 weeks after sham operation (Sham) or TAC with or without surgical denervation (n = 20).
(C) Western blot analysis of p53 in adipose tissues of mice at 6 weeks after sham operation (Sham) or TAC with or without surgical denervation. The right graph
indicates the quantitative data on p53 expression (n = 3). Data are shown as the means S.E.M. *p < 0.05, **p < 0.01.
tissue p53 expression and inflammation (Figures S5HS5J).

Inhibition of p53 activation in adipose tissue by genetic disruption significantly attenuated inflammation of this tissue and
improved metabolic abnormalities (Figures S5K and S5L). These
results suggest that the same mechanism underlies insulin
resistance associated with heart failure due to both pressure
overload and MI.
Influence of Inhibiting p53-Induced Adipose Tissue

Inflammation on Cardiac Function
To investigate whether inhibition of p53-induced adipose tissue
inflammation could influence cardiac function in the development of heart failure, we performed TAC and monitored cardiac
function in adipo-p53 KO mice. We found that adipo-p53 KO
mice showed significantly better cardiac function and less LV
Cell Metabolism
Figure 4. Treatment with a Lipolysis Inhibitor Ameliorates Adipose Tissue Inflammation and Systemic Insulin Resistance
(A) Real-time PCR assessing the expression of Emr1, Tnf (Tnfa), Ccl2 (MCP1), and Cdkn1a (p21) levels in adipose tissue of mice at 6 weeks after sham operation
(Sham) or TAC with or without acipimox treatment (n = 8).
(B) Insulin tolerance test (ITT) and glucose tolerance test (GTT) of mice at 6 weeks after sham operation (Sham) or TAC with or without acipimox treatment (n = 32).
(C) Western blot analysis of p53 in adipose tissues of mice at 6 weeks after sham operation (Sham) or TAC with or without acipimox treatment. Actin was used as
an equal loading control. The right graph indicates the quantitative data on p53 expression (n = 3). Data are shown as the means S.E.M. *p < 0.05, **p < 0.01.
enlargement compared with their littermate controls (Figure 6A).

They also showed better survival during the chronic phase of
heart failure (Figure 6B). Similar results were observed in another
model of heart failure induced by MI (Figure S6A). Furthermore,
administration of a p53 inhibitor (pifithrin-a) into the adipose
tissue of the TAC or MI model mice after the onset of heart failure
improved cardiac dysfunction, as well as adipose tissue inflammation, and metabolic abnormalities (Figures 6C6E and S6B
S6D), indicating that inhibition of p53 may be useful for the
treatment of heart failure and its associated metabolic abnormalities. Moreover, we noted significant improvement of cardiac
function after sympathetic nerve blockade (Figures S6E and
S6F). However, treatment of TAC mice with acipimox was found
to exacerbate cardiac dysfunction (Figure S6G), presumably

because it impaired fatty acid metabolism and energy production in cardiomyocytes, as reported previously (Tuunanen
et al., 2006).
Mechanism of p53-Induced Adipose Tissue
Inflammation during Heart Failure
Because our results indicated that adrenergic activation induced
lipolysis that upregulated p53 and promoted adipose tissue
inflammation, we speculated that an excess of fatty acids might
be involved in the upregulation of p53 in adipose tissue. Therefore, we examined the effect of palmitic acid on cultured preadipocytes. Treatment with palmitic acid significantly increased the
Cell Metabolism
Figure 5. Role of Lipolysis in the Regulation of Adipose p53 Expression and Inflammation
(A) Western blot analysis of p53 in adipose tissues of wild-type mice treated with PBS or isoproterenol (ISO). Actin was used as an equal loading control. The right
graph indicates the quantitative data on p53 expression (n = 3).
(B) Real-time PCR assessing the expression of Emr1, Tnf (Tnfa), and Ccl2 (MCP1) levels in adipose tissues of wild-type mice treated with PBS or isoproterenol
(ISO) (n = 8).
(C) Hematoxylin and eosin staining of adipose tissues of wild-type mice treated with PBS or isoproterenol (ISO). Scale bar, 50 mm. The right graph indicates the
quantitative data on macrophage infiltration (n = 4).
(D) The changes in weight of adipose tissues isolated from Atgl-deficient mice (KO) and littermate controls (Cont) after treatment with PBS or isoproterenol (ISO)
(n = 6).
(E) Expression of p53 was examined in adipose tissues of Atgl-deficient mice (KO) and littermate controls (Cont) treated with PBS or isoproterenol (ISO) by
western blot analysis. The right graph indicates the quantitative data on p53 expression (n = 3).
(F) Real-time PCR assessing the expression of Cdkn1a (p21) level in adipose tissues isolated from Atgl-deficient mice (KO) and littermate controls (Cont) after
treatment with PBS or isoproterenol (ISO) (n = 6). Data are shown as the means S.E.M. *p < 0.05, **p < 0.01.
intracellular level of reactive oxygen species (ROS) and caused

DNA damage, as demonstrated by the increase of gH2AX, which
in turn upregulated p53 expression (Figures 7A7C, S7A, and
S7B). This upregulation of p53 was associated with an increase
of NF-kB activity and proinflammatory cytokine expression
(Figures 7D and 7E). Because it has been reported that p53
enhances the activity of NF-kB, which regulates various cytokines including TNF-a and CCL2 (Benoit et al., 2006; Ryan
et al., 2000), we examined the relationship between p53 expression and NF-kB activation. We demonstrated that the
disruption of p53 expression significantly attenuated palmitic
acid-induced activation of NF-kB and upregulation of Ccl2
Cell Metabolism
Figure 6. Influence of Inhibiting p53-Induced Adipose Tissue Inflammation on Cardiac Function

(A) Echocardiography to assess systolic function (FS) and ventricular size (LVDs) in adipocyte-specific p53-deficient mice (adipo-p53 KO) and littermate controls
(Cont) at 6 weeks after sham operation or TAC (n = 8). FS, fractional shortening; LVDs, left ventricular end-systolic diameter.
(B) Survival rate of adipocyte-specific p53-deficient mice (adipo-p53 KO) and littermate controls (Cont) after TAC procedure (n = 25).
(C) Pifithrin-a (PFT) was administered into the adipose tissue of mice at 24 weeks after TAC, and systolic function (FS) and ventricular size (LVDs) were estimated
before (2w, 2 weeks after TAC) and after (4w, 4 weeks after TAC) treatment by echocardiography (n = 5).
(D) Real-time PCR assessing the expression of Emr1, Tnf (Tnfa), and Ccl2 (MCP1) levels in adipose tissue of mice at 4 weeks after sham operation or TAC with or
without pifithrin-a (PFT) treatment (n = 4).
(E) Insulin tolerance test (ITT) and glucose tolerance test (GTT) of mice at 4 weeks after sham operation or TAC with or without pifithrin-a (PFT) treatment (n = 12).
Data are shown as the means S.E.M. *p < 0.05, **p < 0.01.
(Figures 7D and 7E), whereas knockdown of the NF-kB component p50 markedly inhibited palmitic acid-induced upregulation
of Ccl2 (Figure 7E). In addition, treatment with an antioxidant inhibited palmitic acid-induced DNA damage and upregulation of
p53 (Figures S7A and S7B). We also found that ROS and
gH2AX expression were increased in the adipose tissue of

mice with heart failure (Figures 7F and 7G). Furthermore, nuclear
localization of p50 was enhanced in adipose tissue during heart
failure (Figures 7H and S7C). This increase of nuclear p50
expression and the upregulation of proinflammatory cytokines
Cell Metabolism
Figure 7. Mechanism of p53-Induced Adipose Tissue Inflammation during Heart Failure

(A) Dihydroethidium (DHE) staining (red) of preadipocytes treated with vehicle (Cont) or palmitic acid (500 mM) for 10 min. Nuclei were stained with Hoechst dye
(blue). Scale bar indicates 50 mm. The right graph indicates the quantitative data on DHE-positive area (n = 4).
(B) Immunofluorescent staining for g-H2AX (red) in preadipocytes treated with vehicle (Cont) or palmitic acid (500 mM) for 1 hr. Nuclei and plasma membranes
were stained with Hoechst dye (blue) and Wheat Germ agglutinin lectin (green). Scale bar indicates 50 mm.
(C) Western blot analysis of phospho-p53 and p53 expression in preadipocytes treated with vehicle (Cont) or palmitic acid (500 mM).
(D) Small-interfering RNA targeting p53 (sip53) or negative control RNA (siNC) was introduced into preadipocytes treated with or without palmitic acid (500 mM) for
12 hr. The NF-kB activity was examined by luciferase assay (n = 5).
(E) Real-time PCR assessing the expression of Cdkn1a (p21) and Ccl2 (MCP1) levels in preadipocytes prepared in Figure 7D (n = 9). The effect of small-interfering
RNA targeting the NF-kB component p50 (sip50) on the expression of Ccl2 (MCP1) was also examined (n = 9).
(F) Dihydroethidium (DHE) staining (red) in adipose tissue from sham-operated (Sham) and TAC mice. Nuclei were stained with Hoechst dye (blue). Scale bar
indicates 20 mm. The right graph indicates the quantitative data on DHE-positive area (n = 5).
Cell Metabolism
were inhibited by disruption of p53 in adipose tissue (Figures 2B,

7H, and S7C). Moreover, treatment with a lipolysis inhibitor
significantly inhibited the heart failure-induced increase of ROS
and nuclear p50 expression (Figures S7D and S7E). Inhibition
of NF-kB activation in adipose tissue by BAY 11-7082 also significantly attenuated adipose tissue inflammation and improved
metabolic abnormalities and cardiac dysfunction in TAC mice
(Figures S7FS7H). These results indicate that adrenergic activation by heart failure induces lipolysis in adipose tissue, which
increases DNA damage due to ROS and thus upregulates p53.
Activation of p53 then induces adipose tissue inflammation
and metabolic abnormalities by upregulating the expression of
NF-kB-dependent proinflammatory cytokines.
DISCUSSION
Although treatments that achieve neurohumoral antagonism
have successfully reduced the morbidity and mortality of heart
failure, the death rate remains unacceptably high (Kannel,
2000). Various metabolic abnormalities are associated with heart
failure, and recent data have suggested that heart failure itself
promotes adverse changes of metabolism, such as systemic
insulin resistance (Ashrafian et al., 2007; Witteles and Fowler,
2008). Thus, a detrimental vicious cycle may be postulated, in
which heart failure induces insulin resistance that in turn accelerates cardiac dysfunction (Opie, 2004). However, studies on the
molecular mechanisms of such metabolic abnormalities in heart
failure are largely preliminary and the results have sometimes
been conflicting. In the present study, we demonstrated a causal
role for heart failure in the development of insulin resistance by
using two mouse models of heart failure, and we elucidated
the underlying mechanisms. We found that the hyperadrenergic
state of heart failure initiated a vicious metabolic cycle by
promoting lipolysis in adipose tissue that increased the release
of free fatty acids and upregulated p53 expression and proinflammatory cytokine production in adipose tissue, which then
promoted systemic insulin resistance. Cardiac insulin resistance
is considered to contribute to the development of heart failure.
Because excessive cardiac insulin signaling has been reported
to exacerbate systolic dysfunction in both TAC and MI models
(Shimizu et al., 2010), hyperinsulinemia associated with systemic
insulin resistance may also have a pathological role in heart
failure until insulin resistance becomes evident in the myocardium. Inhibition of lipolysis by sympathetic denervation or by
treatment with a lipolysis inhibitor improved insulin resistance
in our heart failure model. Plasma free fatty acid levels were
significantly elevated after the onset of heart failure, whereas
this increase was attenuated by inhibition of lipolysis with acipimox, denervation, or guanethidine. Disruption of p53 in adipose
tissue also markedly attenuated adipose inflammation and
metabolic abnormalities associated with heart failure, whereas
fatty acid levels were unaffected. Thus, adipose tissue inflamma-
tion rather than the increase of plasma free fatty acids per se is
involved in the impairment of insulin sensitivity and glucose tolerance associated with heart failure. We also noted that p53 was
modestly upregulated in the liver and skeletal muscle, presumably due to the increase of circulating free fatty acids. However,
we did not detect a strong inflammatory response in those
tissues under our experimental conditions (I. Shimizu and
T. Minamino, unpublished data), suggesting that upregulation
of adipose tissue p53 is more important for the development
of metabolic abnormalities during heart failure. This concept is
further supported by our finding that disruption of p53 activation
in adipose tissue nearly normalized insulin resistance and
glucose intolerance provoked by heart failure.
We observed that systolic cardiac function and survival with
chronic heart failure were significantly better for adipo-p53 KO
mice than their control littermates. Suppression of p53 activity
in adipose tissue by administration of a p53 inhibitor after the
onset of heart failure improved cardiac dysfunction and also
reduced adipose tissue inflammation and metabolic abnormalities in both the TAC and MI models. Inhibition of NF-kB activity
in adipose tissue also improved cardiac dysfunction, as well as
adipose tissue inflammation and insulin resistance. Improvement of cardiac dysfunction by disruption of p53 in adipose
tissue was not associated with a decrease of plasma free fatty
acid levels. Systemic inhibition of lipolysis (Atgl deficiency or acipimox treatment) and disturbance of lipolysis in adipose tissue
(denervation or guanethidine treatment) significantly reduced
plasma free fatty acid levels (Haemmerle et al., 2006). However,
the former intervention accelerated heart failure, whereas
cardiac dysfunction was improved by the latter. Thus, the beneficial effect of inhibiting p53-induced adipose tissue inflammation on cardiac function is independent of changes in circulating
free fatty acid levels, and lipolysis in cardiomyocytes appears to
have a crucial role in cardiac metabolism and energy production.
Although there is evidence suggesting that p53 has a protective
role against damage due to ROS and lipotoxicity (Bazuine et al.,
2009), our results indicate that chronic activation of p53 in
adipose tissue causes inflammation and that inhibition of p53induced adipose tissue inflammation is a potential target for
treating metabolic abnormalities and systolic dysfunction in
patients with heart failure.
Adipose tissue was traditionally considered to be a simple
energy storage organ, but it is now appreciated that it also has
endocrine functions and secretes a variety of factors referred
to as adipokines (Donath and Shoelson, 2011; Hotamisligil,
2006; Ouchi et al., 2011). With high calorie intake, the size and
number of adipocytes increase, and hypertrophic adipocytes
shift the balance toward production of proinflammatory adipokines. This shift in the adipokine profile causes the modification
of adipose tissue macrophages from the anti-inflammatory M2
type to the proinflammatory M1 type, and further increases
the production of proinflammatory molecules, which in turn
(G) The number of g-H2AX-positive nuclei (white arrows and inset) in adipose tissue of mice at 6 weeks after sham operation (Sham) or TAC procedure was
estimated by immunofluorescent staining for g-H2AX (red) (n = 5). Nuclei and plasma membranes were stained with Hoechst dye (blue) and Wheat Germ
agglutinin lectin (green). Scale bar indicates 50 mm.
(H) The number of p50-positive nuclei in adipose tissue of adipocyte-specific p53-deficient mice (adipo-p53 KO) and littermate controls (Cont) at 6 weeks
after sham operation (Sham) or TAC procedure was estimated by immunofluorescent staining for p50 (n = 6). Data are shown as the means S.E.M. *p < 0.05,
**p < 0.01.
Cell Metabolism
accelerates the recruitment of activated macrophages into inflamed fatty tissue. Adipokines produced by inflamed adipose
tissue have been suggested to play a crucial role in the regulation
of glucose and lipid metabolism and to contribute to the development of diabetes (Donath and Shoelson, 2011; Hotamisligil,
2006; Ouchi et al., 2011). It has been reported that excessive
calorie intake leads to accumulation of ROS in adipose tissue
and subsequently causes DNA damage that activates p53 (Minamino et al., 2009). In contrast to obesity, heart failure decreases
body fat tissue mass by inducing lipolysis. Accelerated lipolysis
and a subsequent increase of free fatty acids are likely to cause
p53 activation because we found that the promotion of lipolysis
by treatment with isoproterenol upregulated adipose tissue
expression of p53, whereas inhibition of lipolysis by acipimox
or disruption of lipase activity attenuated p53 expression. These
results are consistent with a recent report describing that fasting-induced lipolysis promotes an immune response in murine
adipose tissue (Kosteli et al., 2010). Various molecular mechanisms of p53 activation by heart failure may be postulated,
including hypoxia, increased oxidative stress, and induction of
endoplasmic reticulum stress (Harris and Levine, 2005; Schenk
et al., 2008). Our in vitro and in vivo studies have indicated that
an increase of free fatty acids causes ROS-induced DNA
damage that upregulates p53 in adipose tissue. Activation of
p53 then upregulates the expression of proinflammatory adipokines via the NF-kB signaling pathway and promotes systemic
insulin resistance.
The b-blockers are competitive antagonists of b-adrenergic
receptors. At one time, b-blockers were contraindicated in
patients with heart failure due to their negative inotropic effect.
However, several large-scale clinical trials demonstrated the
efficacy of b-blockers for reducing morbidity and mortality in
heart failure patients with impaired systolic function, so
b-blockers are now recommended as first-line agent for these
patients (Hjalmarson et al., 2000; Leizorovicz et al., 2002; Packer
et al., 2001, 2002). A reduction of heart rate due to inhibition of
cardiac b1-adrenergic receptors is believed to be responsible
for most of the therapeutic benefits associated with b-blocker
treatment, although this is not the only mechanism of action
that may be important in heart failure. It is interesting that treatment with a nonselective b-blocker (carvedilol) achieved a more
marked improvement of survival in patients with chronic heart
failure than treatment with a b1-selective blocker (metoprolol)
(Poole-Wilson et al., 2003), whereas new-onset diabetes was
frequent in heart failure patients during treatment with the
b1-selective blocker (Torp-Pedersen et al., 2007). It has been
reported that carvedilol antagonizes the b3-adrenergic receptor
as well as the b1/2-adrenergic receptors (Schnabel et al., 2000).
Taking our results together with these reports, it seems that
inhibition of b3-adrenergic activity in adipose tissue partially
accounts for the better clinical outcome in patients treated with
this nonselective b-blocker. Recent evidence has suggested
that treatment with insulin sensitizers improves systolic function
of the failing heart in animal models (Asakawa et al., 2002; Nemoto et al., 2005) but such treatment increases the incidence of
heart failure in diabetic patients, presumably because of sodium
retention (Home et al., 2009). Inhibition of p53-induced adipose
tissue inflammation could be an alternative therapeutic target to
block the metabolic vicious cycle in patients with heart failure.
Animal Models
All animal study protocols were approved by the Chiba University review
board. C57BL/6 mice were purchased from the SLC Japan (Shizuoka, Japan).
TAC and MI were performed in 11-week-old male mice as described previously (Harada et al., 2005; Sano et al., 2007). Sham-operated mice underwent
the same procedure except for aortic constriction. Mice that expressed Cre
recombinase in adipocytes (Fabp4-Cre) were purchased from Jackson Laboratories. We then crossed Fabp4-Cre mice (with a C57BL/6 background) with
mice that carried floxed Trp53 alleles with a C57BL/6 background (Marino
et al., 2000) to generate adipocyte-specific p53 knockout mice. The genotype
of littermate controls was Fabp4-Cre- Trp53 flox/flox. The generation and genotyping of Atgl-deficient mice has been described previously (Haemmerle et al.,
2006). Surgical or chemical denervation was performed before TAC operation
as described previously (Demas and Bartness, 2001; Foster and Bartness,
2006), with slight modification. In brief, the epidydimal fat pad was gently
separated from the skin and the abdominal wall by using a dissecting microscope. For surgical denervation, a drop of 1% toluidine blue was applied to
the fat pad to facilitate visualization of the nerves. The nerves were then freed
from the surrounding tissue and vasculature and cut in two or more locations,
and the segments were removed to prevent possible reconnection. Chemical
denervation was performed by the local injection of guanethidine sulfate
(400 mg, Santa Cruz) into bilateral epidydimal fat. Sham-operated mice for
surgical denervation underwent the same procedure except for transection
of the nerve. For the control group for chemical denervation, saline was injected into adipose tissue rather than guanethidine. Acipimox (Sigma) were
provided in drinking water (at a concentration of 0.05%) for 6 weeks after
TAC operation as described previously (Guo et al., 2009). Isoproterenol
(30 mg/kg/day, Sigma) were delivered by infusion pump (DURECT Corporation) for 2 weeks as described previously (Iaccarino et al., 1999). The local
injection of pifithrin-a (2.2 mg/kg/week, Carbiochem) or BAY 11-7082
(20 mg/kg/week, Carbiochem) into bilateral epidydimal fat was performed to
inhibit adipose p53 or NF-kB activity, respectively, from 2 weeks to 4 weeks
after operation.
Physiological and Histological Analyses
Echocardiography was performed with a Vevo 770 High Resolution Imaging
System (Visual Sonics Inc, Toronto, Ontario, Canada). To minimize variation
of the data, the heart rate was always approximately 550650 beats per minute
when cardiac function was assessed. Epidydimal fat samples were harvested
and fixed in 10% formalin overnight. The samples were embedded in paraffin
and sectioned (Narabyouri research Co., Ltd). The sections were subjected to
immunohistochemistry or HE staining. The antibodies used are Mac3-specific
primary antibody (PharMingen) for macrophages, p50-specific primary antibody (Cell signaling), and phospho-H2AX-specific antibody (Cell signaling).
Laboratory Tests
For the intraperitoneal glucose tolerance test (IGTT), mice were starved for
6 hr and were given glucose intraperitoneally at a dose of 2 g / kg (body weight)
in the early afternoon. For the insulin tolerance test, mice were given human
insulin intraperitoneally (1 U / kg body weight) at 1:00 pm without starvation.
Blood glucose levels were measured with a glucose analyzer (Roche Diagnostics). We analyzed free fatty acid (Biovision, Inc) and norepinephrine levels
(LDN) by using ELISA-based immunoassay kits according to the manufacturers instruction.
Western Blot Analysis
The lysates were resolved by SDS-polyacrylamide gel electrophoresis.
Proteins were transferred to a polyvinylidene difluoride membrane (Millipore,
Bedford, MA), which was incubated with the primary antibody followed by
anti-rabbit or anti-mouse immunoglobulin-G conjugated with horseradish
peroxidase (Jackson, West Grove, PA).
Cell Culture
Human preadipocytes were purchased from Sanko (Tokyo, Japan) and were
cultured according to the manufacturers instructions. NIH 3T3-L1 cells were
cultured in high-glucose DMEM plus 10% fetal bovine serum.
Cell Metabolism
Ex Vivo Culture
Epididymal fat was extracted from Atgl-deficient or littermate mice at 17 weeks
of age. Freshly isolated fat pads (100120 mg) were incubated in Dulbeccos
modified Eagles medium supplemented with 10% fetal bovine serum in the
presence of isoproterenol (10 mM) for 48 hr. Fat pads were treated with PBS
instead of isoproterenol in the control group.
Data are shown as the mean SEM. Differences between groups were
examined by Students t-test or ANOVA followed by Bonferronis correction
for comparison of means. For survival analysis, the Kaplan-Meier method
and log-rank test were used. For all analyses, p < 0.05 was considered statistically significant.
Donath, M.Y., and Shoelson, S.E. (2011). Type 2 diabetes as an inflammatory

disease. Nat. Rev. Immunol. 11, 98107.
Edwards, M.G., Anderson, R.M., Yuan, M., Kendziorski, C.M., Weindruch, R.,
and Prolla, T.A. (2007). Gene expression profiling of aging reveals activation
of a p53-mediated transcriptional program. BMC Genomics 8, 80.
Floras, J.S. (2009). Sympathetic nervous system activation in human heart
failure: clinical implications of an updated model. J. Am. Coll. Cardiol. 54,
375385.
Foster, M.T., and Bartness, T.J. (2006). Sympathetic but not sensory denervation stimulates white adipocyte proliferation. Am. J. Physiol. Regul. Integr.
Comp. Physiol. 291, R1630R1637.
Guo, W., Wong, S., Pudney, J., Jasuja, R., Hua, N., Jiang, L., Miller, A., Hruz,
P.W., Hamilton, J.A., and Bhasin, S. (2009). Acipimox, an inhibitor of lipolysis,
attenuates atherogenesis in LDLR-null mice treated with HIV protease inhibitor
ritonavir. Arterioscler. Thromb. Vasc. Biol. 29, 20282032.
Supplemental Information includes Supplemental Experimental Procedures

and seven figures and can be found with this article online at doi:10.1016/
j.cmet.2011.12.006.
Haemmerle, G., Lass, A., Zimmermann, R., Gorkiewicz, G., Meyer, C.,
Rozman, J., Heldmaier, G., Maier, R., Theussl, C., Eder, S., et al. (2006).
Defective lipolysis and altered energy metabolism in mice lacking adipose
triglyceride lipase. Science 312, 734737.
ACKNOWLEDGMENTS
Harada, M., Qin, Y., Takano, H., Minamino, T., Zou, Y., Toko, H., Ohtsuka, M.,
Matsuura, K., Sano, M., Nishi, J., et al. (2005). G-CSF prevents cardiac remodeling after myocardial infarction by activating the Jak-Stat pathway in cardiomyocytes. Nat. Med. 11, 305311.
We thank A. Berns (The Netherlands Cancer Institute) for floxed p53 mice,
T. Fujita (The Tokyo Metropolitan Institute of Medical Science) for reagents,
and E .Takahashi, M. Iijima, and I. Sakamoto for their excellent technical assistance. This work was supported by a Grant-in-Aid for Scientific Research from
the Ministry of Education, Culture, Sports, Science and Technology of Japan
and grants from the Ono Medical Research Foundation; the Uehara Memorial
Foundation; the Daiichi-Sankyo Foundation of Life Science; the NOVARTIS
Foundation for the Promotion Science; the Japan Diabetes Foundation; the
Mitsui Life Social Welfare Foundation; the Naito Foundation; the Japanese
Society of Anti-Aging Medicine; and the Mitsubishi Pharma Research Foundation (to T.M.); a Grant-in-Aid for Scientific Research from the Ministry of Education, Science, Sports, and Culture and Health and Labor Sciences Research
Grants (to I.K.); and a Grant-in-Aid for Scientific Research from the Ministry
of Education, Science, Sports, and Culture, and Health and a grant from the
Uehara Memorial Foundation, Takeda Science Foundation, and Kowa Life
Science Foundation (to I.S.).
Received: June 10, 2011
Revised: October 27, 2011
REFERENCES
Arnlov, J., Lind, L., Zethelius, B., Andren, B., Hales, C.N., Vessby, B., and
Lithell, H. (2001). Several factors associated with the insulin resistance
syndrome are predictors of left ventricular systolic dysfunction in a male population after 20 years of follow-up. Am. Heart J. 142, 720724.
Asakawa, M., Takano, H., Nagai, T., Uozumi, H., Hasegawa, H., Kubota, N.,
Saito, T., Masuda, Y., Kadowaki, T., and Komuro, I. (2002). Peroxisome proliferator-activated receptor gamma plays a critical role in inhibition of cardiac
hypertrophy in vitro and in vivo. Circulation 105, 12401246.
Ashrafian, H., Frenneaux, M.P., and Opie, L.H. (2007). Metabolic mechanisms
in heart failure. Circulation 116, 434448.
Bazuine, M., Stenkula, K.G., Cam, M., Arroyo, M., and Cushman, S.W. (2009).
Guardian of corpulence: a hypothesis on p53 signaling in the fat cell. Clin.
Lipidol. 4, 231243.
Benoit, V., de Moraes, E., Dar, N.A., Taranchon, E., Bours, V., Hautefeuille, A.,
Tanie`re, P., Chariot, A., Scoazec, J.Y., de Moura Gallo, C.V., et al. (2006).
Transcriptional activation of cyclooxygenase-2 by tumor suppressor p53
requires nuclear factor-kappaB. Oncogene 25, 57085718.
Demas, G.E., and Bartness, T.J. (2001). Novel method for localized, functional
sympathetic nervous system denervation of peripheral tissue using guanethidine. J. Neurosci. Methods 112, 2128.
Harris, S.L., and Levine, A.J. (2005). The p53 pathway: positive and negative
feedback loops. Oncogene 24, 28992908.
Hjalmarson, A., Goldstein, S., Fagerberg, B., Wedel, H., Waagstein, F.,
Kjekshus, J., Wikstrand, J., El Allaf, D., Vtovec, J., Aldershvile, J., et al;
MERIT-HF Study Group. (2000). Effects of controlled-release metoprolol on
total mortality, hospitalizations, and well-being in patients with heart failure:
the Metoprolol CR/XL Randomized Intervention Trial in congestive heart failure
(MERIT-HF). JAMA 283, 12951302.
Home, P.D., Pocock, S.J., Beck-Nielsen, H., Curtis, P.S., Gomis, R., Hanefeld,
M., Jones, N.P., Komajda, M., and McMurray, J.J.; RECORD Study Team.
(2009). Rosiglitazone evaluated for cardiovascular outcomes in oral agent
combination therapy for type 2 diabetes (RECORD): a multicentre, randomised, open-label trial. Lancet 373, 21252135.
Hotamisligil, G.S. (2006). Inflammation and metabolic disorders. Nature 444,
860867.
Hotamisligil, G.S., Shargill, N.S., and Spiegelman, B.M. (1993). Adipose
expression of tumor necrosis factor-alpha: direct role in obesity-linked insulin
resistance. Science 259, 8791.
Iaccarino, G., Dolber, P.C., Lefkowitz, R.J., and Koch, W.J. (1999). Bbetaadrenergic receptor kinase-1 levels in catecholamine-induced myocardial
hypertrophy: regulation by beta- but not alpha1-adrenergic stimulation.
Hypertension 33, 396401.
Ingelsson, E., Sundstrom, J., Arnlov, J., Zethelius, B., and Lind, L. (2005).
Insulin resistance and risk of congestive heart failure. JAMA 294, 334341.
Kamei, N., Tobe, K., Suzuki, R., Ohsugi, M., Watanabe, T., Kubota, N.,
Ohtsuka-Kowatari, N., Kumagai, K., Sakamoto, K., Kobayashi, M., et al.
(2006). Overexpression of monocyte chemoattractant protein-1 in adipose
tissues causes macrophage recruitment and insulin resistance. J. Biol.
Chem. 281, 2660226614.
Kannel, W.B. (2000). Incidence and epidemiology of heart failure. Heart Fail.
Rev. 5, 167173.
Kosteli, A., Sugaru, E., Haemmerle, G., Martin, J.F., Lei, J., Zechner, R., and
Ferrante, A.W., Jr. (2010). Weight loss and lipolysis promote a dynamic
immune response in murine adipose tissue. J. Clin. Invest. 120, 34663479.
Leizorovicz, A., Lechat, P., Cucherat, M., and Bugnard, F. (2002). Bisoprolol for
the treatment of chronic heart failure: a meta-analysis on individual data of two
placebo-controlled studiesCIBIS and CIBIS II. Cardiac Insufficiency
Bisoprolol Study. Am. Heart J. 143, 301307.
Lopaschuk, G.D., Folmes, C.D., and Stanley, W.C. (2007). Cardiac energy
metabolism in obesity. Circ. Res. 101, 335347.
Cell Metabolism
Maier, B., Gluba, W., Bernier, B., Turner, T., Mohammad, K., Guise, T.,
Sutherland, A., Thorner, M., and Scrable, H. (2004). Modulation of mammalian
life span by the short isoform of p53. Genes Dev. 18, 306319.
Marino, S., Vooijs, M., van Der Gulden, H., Jonkers, J., and Berns, A. (2000).
Induction of medulloblastomas in p53-null mutant mice by somatic inactivation
of Rb in the external granular layer cells of the cerebellum. Genes Dev. 14, 994
1004.
Meek, D.W. (2009). Tumour suppression by p53: a role for the DNA damage
response? Nat. Rev. Cancer 9, 714723.
Minamino, T., and Komuro, I. (2007). Vascular cell senescence: contribution to
atherosclerosis. Circ. Res. 100, 1526.
Hif-1 causes cardiac dysfunction during pressure overload. Nature 446,

444448.
Schenk, S., Saberi, M., and Olefsky, J.M. (2008). Insulin sensitivity: modulation
by nutrients and inflammation. J. Clin. Invest. 118, 29923002.
Schnabel, P., Maack, C., Mies, F., Tyroller, S., Scheer, A., and Bohm, M.
(2000). Binding properties of beta-blockers at recombinant beta1-, beta2-,
and beta3-adrenoceptors. J. Cardiovasc. Pharmacol. 36, 466471.
Sharma, N., Okere, I.C., Duda, M.K., Chess, D.J., OShea, K.M., and Stanley,
W.C. (2007). Potential impact of carbohydrate and fat intake on pathological
left ventricular hypertrophy. Cardiovasc. Res. 73, 257268.
Minamino, T., and Komuro, I. (2008). Vascular aging: insights from studies on
cellular senescence, stem cell aging, and progeroid syndromes. Nat. Clin.
Pract. Cardiovasc. Med. 5, 637648.
Shimizu, I., Minamino, T., Toko, H., Okada, S., Ikeda, H., Yasuda, N., Tateno,
K., Moriya, J., Yokoyama, M., Nojima, A., et al. (2010). Excessive cardiac
insulin signaling exacerbates systolic dysfunction induced by pressure overload in rodents. J. Clin. Invest. 120, 15061514.
Minamino, T., Orimo, M., Shimizu, I., Kunieda, T., Yokoyama, M., Ito, T.,
Nojima, A., Nabetani, A., Oike, Y., Matsubara, H., et al. (2009). A crucial role
for adipose tissue p53 in the regulation of insulin resistance. Nat. Med. 15,
10821087.
Suskin, N., McKelvie, R.S., Burns, R.J., Latini, R., Pericak, D., Probstfield, J.,
Rouleau, J.L., Sigouin, C., Solymoss, C.B., Tsuyuki, R., et al. (2000).
Glucose and insulin abnormalities relate to functional capacity in patients
with congestive heart failure. Eur. Heart J. 21, 13681375.
Nemoto, S., Razeghi, P., Ishiyama, M., De Freitas, G., Taegtmeyer, H., and
Carabello, B.A. (2005). PPAR-gamma agonist rosiglitazone ameliorates
ventricular dysfunction in experimental chronic mitral regurgitation. Am. J.
Physiol. Heart Circ. Physiol. 288, H77H82.
Tenenbaum, A., Motro, M., Fisman, E.Z., Leor, J., Freimark, D., Boyko, V.,
Mandelzweig, L., Adler, Y., Sherer, Y., and Behar, S. (2003). Functional class
in patients with heart failure is associated with the development of diabetes.
Am. J. Med. 114, 271275.
Neubauer, S. (2007). The failing heartan engine out of fuel. N. Engl. J. Med.
356, 11401151.
Nikolaidis, L.A., Sturzu, A., Stolarski, C., Elahi, D., Shen, Y.T., and Shannon,
R.P. (2004). The development of myocardial insulin resistance in conscious
dogs with advanced dilated cardiomyopathy. Cardiovasc. Res. 61, 297306.
Opie, L.H. (2004). The metabolic vicious cycle in heart failure. Lancet 364,
17331734.
Ouchi, N., Parker, J.L., Lugus, J.J., and Walsh, K. (2011). Adipokines in inflammation and metabolic disease. Nat. Rev. Immunol. 11, 8597.
Packer, M., Coats, A.J., Fowler, M.B., Katus, H.A., Krum, H., Mohacsi, P.,
Rouleau, J.L., Tendera, M., Castaigne, A., Roecker, E.B., et al; Carvedilol
Prospective Randomized Cumulative Survival Study Group. (2001). Effect of
carvedilol on survival in severe chronic heart failure. N. Engl. J. Med. 344,
16511658.
Packer, M., Fowler, M.B., Roecker, E.B., Coats, A.J., Katus, H.A., Krum, H.,
Mohacsi, P., Rouleau, J.L., Tendera, M., Staiger, C., et al; Carvedilol
Prospective Randomized Cumulative Survival (COPERNICUS) Study Group.
(2002). Effect of carvedilol on the morbidity of patients with severe chronic
heart failure: results of the carvedilol prospective randomized cumulative
survival (COPERNICUS) study. Circulation 106, 21942199.
Poole-Wilson, P.A., Swedberg, K., Cleland, J.G., Di Lenarda, A., Hanrath, P.,
Komajda, M., Lubsen, J., Lutiger, B., Metra, M., Remme, W.J., et al;
Carvedilol Or Metoprolol European Trial Investigators. (2003). Comparison of
carvedilol and metoprolol on clinical outcomes in patients with chronic heart
failure in the Carvedilol Or Metoprolol European Trial (COMET): randomised
controlled trial. Lancet 362, 713.
Royds, J.A., and Iacopetta, B. (2006). p53 and disease: when the guardian
angel fails. Cell Death Differ. 13, 10171026.
Ryan, K.M., Ernst, M.K., Rice, N.R., and Vousden, K.H. (2000). Role of NFkappaB in p53-mediated programmed cell death. Nature 404, 892897.
Sano, M., Minamino, T., Toko, H., Miyauchi, H., Orimo, M., Qin, Y., Akazawa,
H., Tateno, K., Kayama, Y., Harada, M., et al. (2007). p53-induced inhibition of
Torp-Pedersen, C., Metra, M., Charlesworth, A., Spark, P., Lukas, M.A., PooleWilson, P.A., Swedberg, K., Cleland, J.G., Di Lenarda, A., Remme, W.J., and
Scherhag, A.; COMET investigators. (2007). Effects of metoprolol and carvedilol on pre-existing and new onset diabetes in patients with chronic heart failure:
data from the Carvedilol Or Metoprolol European Trial (COMET). Heart 93,
968973.
Tuunanen, H., Engblom, E., Naum, A., Nagren, K., Hesse, B., Airaksinen, K.E.,
Nuutila, P., Iozzo, P., Ukkonen, H., Opie, L.H., and Knuuti, J. (2006). Free fatty
acid depletion acutely decreases cardiac work and efficiency in cardiomyopathic heart failure. Circulation 114, 21302137.
Tyner, S.D., Venkatachalam, S., Choi, J., Jones, S., Ghebranious, N.,
Igelmann, H., Lu, X., Soron, G., Cooper, B., Brayton, C., et al. (2002). p53
mutant mice that display early ageing-associated phenotypes. Nature 415,
4553.
Vousden, K.H., and Lane, D.P. (2007). p53 in health and disease. Nat. Rev.
Mol. Cell Biol. 8, 275283.
Vousden, K.H., and Prives, C. (2009). Blinded by the Light: The Growing
Complexity of p53. Cell 137, 413431.
Vousden, K.H., and Ryan, K.M. (2009). p53 and metabolism. Nat. Rev. Cancer
9, 691700.
Weisberg, S.P., McCann, D., Desai, M., Rosenbaum, M., Leibel, R.L., and
Ferrante, A.W., Jr. (2003). Obesity is associated with macrophage accumulation in adipose tissue. J. Clin. Invest. 112, 17961808.
Witteles, R.M., and Fowler, M.B. (2008). Insulin-resistant cardiomyopathy clinical evidence, mechanisms, and treatment options. J. Am. Coll. Cardiol. 51,
93102.
Young, M.E., McNulty, P., and Taegtmeyer, H. (2002). Adaptation and maladaptation of the heart in diabetes: Part II: potential mechanisms. Circulation 105,
18611870.
Zeng, L., Wu, G.Z., Goh, K.J., Lee, Y.M., Ng, C.C., You, A.B., Wang, J., Jia, D.,
Hao, A., Yu, Q., and Li, B. (2008). Saturated fatty acids modulate cell response
to DNA damage: implication for their role in tumorigenesis. PLoS ONE 3,
e2329.
Cell Metabolism
Article
Impaired Generation of 12-Hydroxylated Bile Acids
Links Hepatic Insulin Signaling with Dyslipidemia
Rebecca A. Haeusler,1 Matthew Pratt-Hyatt,2 Carrie L. Welch,1 Curtis D. Klaassen,2 and Domenico Accili1,*
1Department
of Medicine, Columbia University, New York, NY 10032

of Pharmacology, Toxicology and Therapeutics, University of Kansas Medical Center, Kansas City, KS 66160
*Correspondence: da230@columbia.edu
DOI 10.1016/j.cmet.2011.11.010
2Department
SUMMARY
The association of type 2 diabetes with elevated

plasma triglyceride (TG) and very low-density lipoproteins (VLDL), and intrahepatic lipid accumulation
represents a pathophysiological enigma and an unmet therapeutic challenge. Here, we uncover a link
between insulin action through FoxO1, bile acid (BA)
composition, and altered lipid homeostasis that
brings new insight to this longstanding conundrum.
FoxO1 ablation brings about two signature lipid
abnormalities of diabetes and the metabolic syndrome, elevated liver and plasma TG. These changes
are associated with deficiency of 12a-hydroxylated
BAs and their synthetic enzyme, Cyp8b1, that hinders
the TG-lowering effects of the BA receptor, Fxr.
Accordingly, pharmacological activation of Fxr with
GW4064 overcomes the BA imbalance, restoring
hepatic and plasma TG levels of FoxO1-deficient
mice to normal levels. We propose that generation
of 12a-hydroxylated products of BA metabolism
represents a signaling mechanism linking hepatic
lipid abnormalities with type 2 diabetes, and a treatment target for this condition.
INTRODUCTION
The metabolic syndrome and diabetes are linked to abnormalities in lipid metabolism that can lead to cardiovascular disease
and hepatic steatosis (DeFronzo, 2010). This dyslipidemia poses
a double therapeutic challenge: on one hand, its non-LDL based
mechanism renders it poorly responsive to statins and cholesterol absorption inhibitors; on the other, macrovascular complications arising from it are singularly unresponsive to tight
glycemia control and remain the leading cause of death of diabetic patients (National Institute of Diabetes and Digestive and
Kidney Diseases, 2005).
In addition to its role in causing hyperglycemia through excessive glucose production (HGP) (Lin and Accili, 2011), the liver is
also the source of increased triglyceride (TG) levels in diabetes
(Choi and Ginsberg, 2011). Interestingly, the latter antedate the
onset of hyperglycemia, seemingly indicating that alterations of
hepatic TG metabolism are an intrinsic component of the metabolic syndrome (Srensen et al., 2011).
FoxO1 promotes HGP (Haeusler et al., 2010b; Matsumoto

et al., 2007) and regulates different aspects of liver sensitivity
to insulin. Genetic ablation studies show that FoxO1 is dispensable for insulin regulation of hepatic lipid metabolism (Dong et al.,
2008; Wan et al., 2011). This finding led to a model in which
insulin signaling to glucose and lipid production diverges at
Akt, with FoxO1 regulating HGP, and mTOR or other serine
kinasessuch as atypical Pkcregulating lipid synthesis and
secretion (Biddinger et al., 2008b; Li et al., 2010). However, it
has proven difficult to obtain in vivo evidence for selective insulin
resistance (e.g., for a condition in which FoxO1 signaling is suppressed, but mTOR or Pkc signaling is preserved) (Brown and
Goldstein, 2008; Haeusler and Accili, 2008).
The hypothesis of this work was that FoxO1 mediates aspects
of insulin-dependent hepatic lipid metabolism that have not been
identified in previous studies because animals were not challenged appropriately. Using genetic (crosses with low-density
lipoprotein receptor knockout mice, Ldlr / ) or dietary approaches (cholesterol-rich, western-type diet [WTD]) in mice
lacking hepatic FoxO1 (L-FoxO1), we discovered that FoxO1 is
required to suppress hepatic and plasma TG levels through qualitative rather than quantitative regulation of BA synthesis.
As potent regulators of hepatic cholesterol and TG metabolism, BAs affect insulin sensitivity and glucose disposal. Bile
acid sequestrants reduce glucose and cholesterol levels in
type 2 diabetics (Staels and Kuipers, 2007). In addition to their
established role in cholesterol and fat absorption as well as
cholesterol detoxification, BAs activate the nuclear receptor
Fxr and the transmembrane receptor Tgr5 (Thomas et al.,
2008). In this work, we show that FoxO1 is required for expression of a subset of BA synthetic genes and loss of hepatic
FoxO1 results in an unusual BA profile that impairs the ability
of Fxr to reduce TG levels, and can be reversed by pharmacological Fxr activation. Our finding that insulin regulates hepatic lipid
metabolism through FoxO1-dependent modulation of BA profiles points to alterations of BA composition as an exploitable
opportunity in diabetes therapeutics.
RESULTS
Western Diet Exacerbates Lipid Abnormalities in Mice
Lacking Liver FoxO1
We examined liver and plasma lipids in L-FoxO1 and
L-FoxO1:Ldlr / mice consuming either a regular chow diet or
western diet (WTD). After WTD feeding, weight and adiposity
increased similarly in all groups (data not shown). Livers of
Cell Metabolism
Bile Acids and Insulin Action
also elevated in fasted L-FoxO1:Ldlr / mice on WTD (Figure 1F),

with trends toward higher LDL- and lower HDL-cholesterol
(Figure 1H).
We also assessed aortic atherosclerotic plaque area in
WTD-fed L-FoxO1:Ldlr / and controls but found no differences
between the two groups (Figures S1B and S1C). We reckoned
that this unexpected finding of normal plaque size despite
elevated TG and cholesterol might be due to the concurrent
increase in insulin sensitivity of the FoxO1 mutants, as seen
previously in chow-fed mice (Matsumoto et al., 2007). To test
this hypothesis, we measured glucose and insulin levels in
WTD-fed mice. We found that WTD tended to raise both glucose
and insulin, but L-FoxO1 and L-FoxO1: Ldlr / mice showed
lower fasting glucose and insulin (Figures S2AS2D) and better
glucose tolerance (Figures S2ES2H) than littermate controls.
There were no changes in the response to insulin injection
(Figures S2I and S2J). These observations reveal that hepatic
FoxO1 ablation preserves insulin sensitivity of glucose metabolism during WTD, possibly contributing to the lower than expected plaque size (Tabas et al., 2010).
Despite improvements in glucose metabolism and insulin
sensitivity due to FoxO1 ablation, the elevated serum TG in
L-FoxO1:Ldlr / mice, as well as the increased liver TG in
WTD-fed L-FoxO1 and L-FoxO1:Ldlr / mice, resemble key
aspects of the lipid profile of the metabolic syndrome (Isomaa
et al., 2001). These findings could be construed to support
a model in which lipid abnormalities in the metabolic syndrome
arise as a consequence of excessive (or preserved) sensitivity
to the lipogenic actions of insulin (Brown and Goldstein, 2008;
Haeusler and Accili, 2008).
Figure 1. Lipid Parameters after Feeding with Western Diet

(A and B) Livers (A) and hematoxylin and eosin staining (B) of liver sections after
WTD feeding.
(C) Liver weights as a percentage of body weight (n = 714).
(D) Liver TG (n = 78).
(E and F) Total serum TG (E) and cholesterol (F) in chow and WTD-fed L-FoxO1
and L-FoxO1:Ldlr / and respective controls after a 5-hr fast (n = 712).
(G and H) Ultracentrifuge-fractionated TG (G) and cholesterol (H) in WTD-fed
L-FoxO1:Ldlr / and controls (n = 5). *p < 0.05, **p < 0.01, ***p < 0.001, by
Students t-tests. Data are presented as mean SEM. See also Figures S1
and S2.
L-FoxO1 and L-FoxO1:Ldlr / mice were paler, larger, and laden

with larger lipid droplets compared to controls (Figure 1A1C),
due to a large increase in liver TG content (Figure 1D). In addition,
L-FoxO1:Ldlr / mice showed significantly higher serum TG
compared to Ldlr / on both chow and WTD (Figure 1E). The
increase in TG was associated with a large, though not statistically significant, increase of VLDL-TGs, as assessed by density
ultracentrifugation (Figure 1G). We did not detect a difference
in TG secretion rate (Figure S1A). Total serum cholesterol was
Altered Enterohepatic Bile Acid Composition

in L-FoxO1 Mice
To understand the mechanism underlying the alterations in lipid
levels, we performed comprehensive, unbiased liver metabolite
analyses. Because the liver abnormalities were identical in the
Ldlr-competent and Ldlr-deficient background, we limited subsequent analyses to the former. Notable metabolite changes in
L-FoxO1 mice that were consistent with the hypertriglyceridemia
included increases of glycerolipids (data not shown) and very
long-chain fatty acids (FAs) (R20 carbons), with decreases of
some shorter FAs (Figure S3A). However, the most striking
changes were among BA species (Figure 2A). Because BAs
are known to reduce TGs (Lefebvre et al., 2009; Thomas et al.,
2008), we hypothesized that alterations of bile acid metabolism
link FoxO1 deficiency to increased TG levels.
BAs are synthesized in liver from cholesterol in two multi-step
pathways, leading to the formation of primary BAs cholic acid
(CA), chenodeoxycholic acid (CDCA) and, in mice, muricholic
acids (MCAs) (Russell, 2009). After secretion into the duodenum,
these are converted to secondary BAs by microbiota and largely
recycled to the liver (Ridlon et al., 2006). We identified 10 hepatic
BA species with a distinct distribution: in WTD-fed L-FoxO1
mice, all products that had undergone hydroxylation at the
carbon 12 position (12a OH) had relative values less than one,
indicating a decrease compared to controls, a trend that showed
p < 0.05 for T-DCA (Figure 2A, left). Conversely, all products that
had never undergone 12a-hydroxylation had values greater than
one, indicating an increase compared to controls; for T-CDCA
Cell Metabolism
this trend achieved p < 0.05 (Figure 2A, left). In chow-fed

L-FoxO1 mice, we found that CA, the major 12a OH BA,
showed a trend toward decrease (p = NS), whereas two non12a-OH BAs, CDCA and T-CDCA, showed significant increases
(p = 0.001 and 0.03, respectively) (Figure 2A, right). Moreover,
WTD itself increased two non-12a-OH BAs (Figure S3B), suggesting that WTD exacerbates the BA composition defects of
L-FoxO1 mice.
Using LC-MS/MS, we quantitated BAs in gallbladder bile and
feces, as well as the total BA pool in WTD-fed mice (Alnouti et al.,
2008). In every case, L-FoxO1 mice showed relative deficiency of
12a OH BAs, leading to a marked increase in the ratios of non12a-OH:12a-OH BAs or MCAs:CA (Figures 2B2D, S3C, and
S3D). However, the total size of the BA pool was unchanged
because an increase in non-12a-OH BAs compensated for the
deficiency (Figure 2E). We also found no difference in conjugated
versus unconjugated BAs (Figure S3E). Quantitation of all
detected BAs is available in Table S1.
FoxO1 Ablation Affects Expression of Bile Acid, Fatty
Acid, and Cholesterol Synthesis Genes
We hypothesized that alterations of the BA pool in FoxO1deficient livers reflected altered FoxO1-dependent transcription.
Accordingly, we surveyed gene expression changes using an
unbiased exon microarray approach to identify possible effectors (Table S2). Gene ontology enrichment analyses demonstrated highly significant changes in lipid metabolism, transport,
storage, and biosynthesis in response to FoxO1 ablation after
WTD feeding (Figure 3A) (Eden et al., 2009).
We saw a striking dysregulation of BA synthesis genes in
L-FoxO1 mice. Consistent with the changes in BA composition,
Cyp8b1, encoding the 12a-hydroxylase enzyme, was sharply
downregulated in all tested conditions (Figures 3BD), as were
enzymes involved in other steps of BA metabolism, including
Cyp7a1, Cyp27a1, Cyp7b1 (Figures 3BD, Table S2). Canalicular
cholesterol transporters Abcg5 and Abcg8, two FoxO1 targets
(Biddinger et al., 2008a), were also downregulated, as were the
canalicular bile salt export pump (Bsep) and some sinusoidal
BA uptake transporters (Figures 3B and C, Table S2).
Consistent with their high liver TGs, very long chain FAs, and
glycerolipids, WTD-fed L-FoxO1 (and L-FoxO1:Ldlr / ) mice
showed increased expression of genes involved in FA synthesis,
elongation, glycerolipid formation, and lipid storage, including
Fasn, Acly, and Gpam (Figures 3B and 3C, Table S2). However,
these changes occurred in the absence of changes to known
regulators of lipid synthesis and oxidation, including sterol
response element-binding protein (Srebp)-1c and its posttranslational activator Insig-2a (Goldstein et al., 2006), liver X
receptora (Lxra, encoded by Nr1h3), and Ppara and its target
genes (Figures 3B and C). Carbohydrate response elementbinding protein (Chrebp) was reduced, but its target Pklr was
unchanged (Table S2). Genes involved in lipoprotein assembly
and turnoverLpl, Angptl3, Angptl4, Apoc3, and Mttpshowed
minor or no changes (Table S2).
Cholesterol biosynthesis genes, such as Hmgcs1, Hmgcr,
and Cyp51, were increased in WTD-fed L-FoxO1 and
L-FoxO1:Ldlr / mice (Figures 3B and 3C, Table S2). Srebp-2,
which promotes cholesterol biosynthesis genes as well as its
own expression, also tended to be elevated. These findings
might be related to the elevated serum cholesterol levels in

L-FoxO1:Ldlr / mice. Their significance in L-FoxO1 mice will
be discussed below.
Changes in Known Bile Acid Regulators Cannot Explain
the Defects of L-FoxO1 Mice
The observed changes in BAs could potentially explain the alterations of TG and cholesterol homeostasis seen in L-FoxO1 mice.
Such changes could arise either as a direct effect of impaired
FoxO1-dependent transcription of BA synthetic enzymes or as
an indirect effect of the FoxO1 knockout on known regulators
of BA synthetic genes. To examine the latter possibility, we
measured expression of nuclear receptors required to regulate
BA synthesis. Hepatic Fxr (encoded by Nr1h4) represses
Cyp7a1 and Cyp8b1 by inducing the nuclear corepressor Shp
(encoded by Nr0b2), which inhibits transcription factor Lrh-1
(encoded by Nr5a2) (Goodwin et al., 2000; Lu et al., 2000).
However, although Fxr and Lrh-1 were expressed normally,
Shp was reduced, as were other Fxr target genes, including
Bsep, Scarb1, and Fetub (Figures 3B and 3C, Table S2). Fxr activation in the small intestine also represses hepatic Cyp7a1
through Fgf15 and its receptor, Fgfr4 (Chiang, 2009). However,
hepatic Fgfr4 and its cofactor, b-Klotho (encoded by Klb), were
expressed normally. Fgf15 was not expected to be affected
directly by loss of liver FoxO1 because it is made in ileum.
Consistent with this, we found that Fgf15 mRNA tended to be
slightly reduced, although this reduction did not reach statistical
significance. Two intestinal Fxr targets, Ibabp and Shp, also
tended to be reduced (Figure 3E and Table S2). The apparent
reduction of Fxr activity would be expected to result in higher
expression of Cyp7a1 and Cyp8b1. Because we saw the opposite result (Figures 3B and 3D), we conclude that Fxr activity
cannot be directly responsible for the defects in BA synthesis
genes.
We examined other transcriptional regulators of Cyp7a1,
including Lxra, Hnf4a (hepatocyte nuclear factor 4a), Ppara,
Pgc1a (PPARg coactivator 1a), and Rev-erb (encoded by
Nr1d2) (Chiang, 2009), but neither they nor their target genes
were changed (Table S2). Thus, these data support the alternative hypothesis that reduction of BA synthetic genes is a primary
effect of FoxO1 ablation in L-FoxO1 mice.
Regulation of Bile Acid Synthesis Genes by FoxO1
We next interrogated whether the observed changes in BA
synthetic genes can be imputed to FoxO1. Based on the dearth
of 12a-OH BAs in L-FoxO1 mice, we expected a primary requirement for FoxO1 to promote Cyp8b1. FoxO1 is also reported
to promote Cyp7a1 in rodents (Li et al., 2006; Shin and
Osborne, 2009). We compared single hepatic FoxO1 knockouts
with compound hepatic knockouts of three FoxO isoforms
(L-FoxO1,O3,O4) (Haeusler et al., 2010b). Loss of FoxOs reduced the expression of Cyp8b1, Cyp7a1, and Cyp27a1 compared to control littermates, and the reduction in Cyp8b1 tended
to be exacerbated by triple knockout, compared to single FoxO1
knockout (Figure 4A). The changes in Cyp8b1 were identified as
early as post-natal day 2 in neonates (Figure 4B) and, thus, likely
represent primary effects of the FoxO ablation rather than
acquired or compensatory changes (Haeusler et al., 2010b; Matsumoto et al., 2007). In contrast, Cyp7a1, Cyp27a1, and Cyp7b1
Cell Metabolism
Figure 2. Metabolomics and Bile Acid Composition

(A) Hepatic BAs in L-FoxO1 mice, relative to controls, on WTD, and chow, separated by origination from 12a-hydroxylation (n = 8). Note that 12-oxo-CDCA does
not have a hydroxyl group at the 12 position, but it is derived by oxidation of CA by bacteria (Ridlon et al., 2006). A complete list of BA abbreviations is available in
Table S4.
(B) 12a-OH (blue) and non-12a-OH (red) BAs, as a percentage of total in gallbladder bile (pooled from 5 mice per genotype), feces (n = 34), and total BA pool
(n = 56), in WTD-fed L-FoxO1 mice and controls.
Cell Metabolism
were expressed normally at this age, suggesting that changes

occur later in life (Figure 4B). Although this developmental
snapshot might not adequately capture the complex developmental regulation of BA synthetic genes, the decrease of
Cyp8b1 emerges as a key feature of this model. In fact, when
we examined the nutritional regulation of these genes, we found
only Cyp8b1 to be induced by fasting, in a fashion consistent
with its regulation by FoxO1 (Gafvels et al., 1999). Moreover,
this regulation was lost in L-FoxO1 mice, indicating that FoxO1
is required for Cyp8b1 induction in vivo (Figure 4C).
Finally, we assessed the effect of WTD on these genes.
Cyp7a1 tended to be induced to the same extent in L-FoxO1
and control mice, consistent with an Lxr-mediated effect, in
response to cholesterol contained in the WTD (Kalaany and
Mangelsdorf, 2006) (Figure 4D). Cyp8b1, Cyp27a1 and Cyp7b1
were repressed by WTD (Figure 4D). Thus, it appears that the
physiologic response to WTD entails repression of Cyp8b1
with simultaneous induction of Cyp7a1, the rate-limiting enzyme
of BA synthesis. This combined effect explains the limited rise of
some non-12a-OH BAs in normal mice. In the absence of FoxO1,
the basal deficiency of Cyp8b1 is compounded by the failure to
fully induce Cyp7a1, resulting in the characteristic BA profile of
L-FoxO1 mice.
Lipid Defects in L-FoxO1 Mice Are Secondary
to Impaired Bile Acid Signaling, Not to Altered
Glucose Homeostasis
We hypothesized that lack of 12a-OH BAs renders the BA pool
of L-FoxO1 mice abnormally hydrophilic (Heuman, 1989) and
affects TG metabolism and cholesterol absorption (Lefebvre
et al., 2009). First, we examined TG metabolism. Hydrophobic
BAs are known to reduce TGs (Angelin et al., 1978; Bilz et al.,
2006; Miller and Nestel, 1974; Watanabe et al., 2004), possibly
through the actions of Fxr because Fxr / mice display hypertriglyceridemia, whereas Fxr activation reduces TGs (Evans et al.,
2009; Kong et al., 2009; Maloney et al., 2000; Sinal et al., 2000;
Watanabe et al., 2004; Zhang et al., 2006). However, hydrophilic
BAs are unable to activate Fxr (Makishima et al., 1999; Parks
et al., 1999). Thus, the hydrophilic BA pool of L-FoxO1 mice
might explain the reduced Fxr activity.
We hypothesized that the hypertriglyceridemia of L-FoxO1 mice
was due to low Fxr activity and tested the hypothesis by treating
WTD-fed control and L-FoxO1 mice with low doses of the Fxr
agonist GW4064 (Maloney et al., 2000). This treatment did not
affect body weight (Figure S4A) but was sufficient to activate Fxr,
as shown by the induction of Bsep and the repression of Cyp7a1
and Cyp8b1 (Figure 5A). Although this dose of GW4064 did not
affect liver TG in control mice, liver TG decreased to control levels
in L-FoxO1 mice (Figure 5B). Additionally, we treated mice with
cholic acid (CA), which is known to reduce TGs (Lefebvre et al.,
2009; Watanabe et al., 2004). Consistent with our hypothesis,
CA reduced liver TGs in both control and L-FoxO1 mice and eliminated differences between genotypes (Figure 5B). GW4064 and
CA had similar effects in L-FoxO1:Ldlr / mice (Figure S4B).
In L-FoxO1:Ldlr / and Ldlr / mice, GW4064 or CA treatment

also reduced serum TGs and eliminated differences between
genotypes (Figure 5C). The decrease occurred primarily, though
not exclusively, in VLDL-TGs (Figures 5C and S4C). Together,
these data support the hypothesis that after WTD feeding,
L-FoxO1 mice develop hypertriglyceridemia due to a dearth of
endogenous Fxr ligands (i.e,. 12a-OH BAs).
Second, we examined cholesterol homeostasis. Absorption of
dietary cholesterol is promoted by hydrophobic, micelle-forming
BAs (Murphy et al., 2005; Wang et al., 2003). Thus, L-FoxO1 mice
are expected to be ineffective at absorbing cholesterol, thereby
activating Srebp-2 (Heuman et al., 1989; Murphy et al., 2005).
Furthermore, the effect of WTD on the repression of cholesterol
synthetic genes was blunted in L-FoxO1 mice (Table S3), consistent with inefficient Srebp-2 inhibition, providing an explanation
for the inappropriate activation of this pathway (Figures 3B,
3C, and Table S2). Because these effects of BA are Fxr-independent, we expected that CA, but not GW4064, would increase
liver cholesterol and normalize cholesterol synthetic genes in
L-FoxO1 mice. Indeed CA, but not GW4064, increased liver
cholesterol and normalized Hmgcs1 in L-FoxO1 mice (Figures
S4DS4F), although the increase in liver cholesterol after CA
treatment may also cause cellular stress.
Finally, we examined glucose metabolism. We expected that
the improved glucose phenotype of L-FoxO1 mice was independent of the BA phenotype (Lin and Accili, 2011) because the BA
composition of L-FoxO1 mice predicts worse, not improved,
glycemia (Staels and Kuipers, 2007). To test this, we measured
blood glucose in L-FoxO1 neonates; nursing mothers (Forsyth
et al., 1983) were fed either chow alone or chow containing
0.5% CA. This dose was effective because treated neonates
showed induction of Shp and repression of Cyp8b1 and Cyp7a1
(Figure 5D).
As previously observed (Haeusler et al., 2010b; Matsumoto
et al., 2007), L-FoxO1 neonates from chow-fed mothers showed
hypoglycemia (Figure 5E). After supplementing mothers with CA,
L-FoxO1 neonates still showed hypoglycemia compared to
controls (Figure 5E). Intriguingly, CA treatment slightly elevated
the glycemia of L-FoxO1 neonates but not controls. We predicted that this may reflect a slight improvement in absorption
of milk, which contains a high percentage of lipid, including
cholesterol (Meier et al., 1965). Consistent with this, cholesterol
synthetic genes were repressed by CA treatment (Figure 5D).
Overall, these data suggest that the regulation of glucose metabolism by FoxO1 is independent of its effect on BA composition.
DISCUSSION
The constellation of metabolic dysfunctions known as the metabolic syndrome includes abnormalities of glucose and lipid
metabolism that are purportedly caused by defective insulin
signaling. The liver is a key site of development of mixed
abnormalities of insulin signaling that affect glucose and lipid
metabolism (Haeusler and Accili, 2008). However, although the
(C) Ratio of non-12a-OH BAs to 12a-OH BAs in bile, feces, and total pool.
(D) Composition of total BA pool in WTD-fed L-FoxO1 mice and controls (mean values of n = 56).
(E) Quantitation of total BAs, 12a-OH BAs, and non-12a-OH BAs in the total BA pool from WTD-fed mice (n = 56). *p < 0.05, **p < 0.01, by Welchs two-sample
t-test (A) or Students t-tests (B, C, and E). Data are presented as mean SEM. See also Figure S3.
Cell Metabolism
mechanism by which insulin regulates HGP has been well characterized (Lin and Accili, 2011), its control over TG synthesis,
secretion, and clearance is less defined (Choi and Ginsberg,
2011). In this work we show that insulin signaling through
FoxO1 controls production of 12a-OH BAs and we propose
that these molecules act as regulatory mediators, relaying a
signal from insulin to TG synthesis.
This discovery can be integrated into a new model for the role
of the insulin-Akt-FoxO1 pathway in regulating multiple aspects
of hepatic metabolism (Figure 5F). In normal physiology, FoxO1
maintains the production of 12a-OH BAs by regulating Cyp8b1
and, potentially, other BA synthetic genes. In turn, these BAs
maintain normal Fxr activity, curbing TG levels. In overt diabetes,
FoxO1 becomes trapped inside the nucleus, due to impaired
ability of insulin to cause its nuclear export, compounded by
hyperglycemia (Haeusler et al., 2010a; Kitamura et al., 2005).
In this condition, BAs would become skewed toward excess
12a-OH BAs. Indeed, diabetic models, such as alloxan-treated
rodents, NOD mice, and mice lacking hepatic insulin receptors,
show disproportionately high 12a-OH BAs (Akiyoshi et al.,
1986; Biddinger et al., 2008a; Uchida et al., 1985; Uchida
et al., 1996).
Compellingly, the pattern of increased 12a-OH BAs has also
been identified in human patients with diabetes (Brufau et al.,
2010). Contrary to rodents, however, CDCAthe most abundant
non-12a-OH BAis hydrophobic and also is the most potent
FXR ligand (Makishima et al., 1999; Parks et al., 1999); thus,
FOXO1 induction of CYP8B1 (which promotes CA formation) at
the expense of CDCA would cause a relative reduction in FXR
activation because CA and its conjugated form T-CA are less
potent FXR ligands (Makishima et al., 1999; Parks et al., 1999).
The constitutive activation of the FOXO1-CYP8B1-12a-OH BA
pathway during diabetes would be expected to result in reduced
FXR activity, raising TG.
Interestingly, L-FoxO1 mice express inadequate Cyp8b1 and
have a distinct functional defect in 12a-hydroxylation; however,
they do not phenocopy Cyp8b1 / mice (Li-Hawkins et al., 2002;
Murphy et al., 2005). The latter presumably have normal Fxr
activity because they express normal Shp and Bsep, and reportedly have no steatosis (Li-Hawkins et al., 2002; Murphy et al.,
2005). This is likely due to their compensatory upregulation of
Cyp7a1, which is associated with an expanded BA pool, including a substantial increase in CDCA (Li-Hawkins et al., 2002).
In contrast, L-FoxO1 mice show low Cyp7a1 and Cyp27a1 under
several different conditions and low Cyp7b1 after WTD feeding,
suggesting that FoxO1 is required for full expression of several
of these genes. Mice lacking Lrh-1 specifically in liver show
several similarities to L-FoxO1 mice, including reduced expression of Cyp8b1 and deficiency of 12-OH BAs (Lee et al., 2008;
Mataki et al., 2007). However, the two models are not identical:
Lrh-1 knockouts express normal levels of Cyp7a1 and they
have not been reported to develop steatosis. The latter may be
attributed to the fact that 1) they have not been challenged with
Figure 3. Gene Expression Changes Due to Hepatic FoxO1 Ablation

(A) Gene ontology enrichment analysis, based on microarrays of liver tissue
from WTD-fed L-FoxO1 and L-FoxO1:Ldlr / versus littermate controls. Only
nonredundant pathways, where p < 0.001 for the pathway, are shown.
(BD) Relative hepatic gene expression by qPCR in (B) WTD-fed

L-FoxO:Ldlr / ,WTD-fed L-FoxO1 (C), and chow-fed (D) L-FoxO1 mice.
(E) Relative gene expression by qPCR in ileum from WTD-fed L-FoxO1 mice.
(n = 67 for all) *p < 0.05, **p < 0.01, ***p < 0.001, by Students t-tests. Data are
presented as mean SEM.
Cell Metabolism
Prior observations that gain of FoxO1 function in liver results in

hyperlipidemia (Kamagate et al., 2008; Matsumoto et al., 2006)
seem to be at odds with our current findings, but there are key
differences. Transgenic mice overexpressing FoxO1 become
hyperglycemic, introducing a major confounder (Kamagate
et al., 2008; Nakae et al., 2002). Moreover, acute overexpression
of FoxO1 by adenovirus causes a paradoxical increase in Akt
activity that is seen only rarely in diabetes (Matsumoto et al.,
2006). Therefore, both models subsume independent downstream signaling events (e.g., hyperglycemia and Akt activation)
that independently contribute to excess lipid production. In contrast, L-FoxO1 mice have improved glycemia and insulin sensitivity, without ancillary Akt activation.
In addition to their role as FXR agonists, BAs also signal
through TGR5. We have not yet explored the consequences of
this model on TGR5 signaling, but there may be effects on other
aspects of the L-FoxO1 phenotype, such as energy expenditure
and metabolic efficiency (Thomas et al., 2008). The present data
raise the possibility that altered BA composition is part of the
underlying pathogenesis of diabetes and that redressing BA
composition may therapeutically benefit a wide population of
patients.
Figure 4. Regulation of Bile Acid Synthetic Genes

(A) Hepatic gene expression in chow-fed L-FoxO1 and L-FoxO1,O3,O4 mice.
Values are shown relative to same-strain littermate controls (n = 6-8).
(B) Hepatic gene expression from mice sacrificed on postnatal day 2 (P2)
(n = 6).
(C) Hepatic gene expression during a fasting-refeeding time course (n = 56).
Mice were fasted up to 24 hr, starting at 8:00 p.m., or fasted 24 hr then refed
chow for 0.25 or 4.25 hr. Black and white boxes at the bottom of each graph
indicate the light/dark cycle.
(D) Comparison of gene expression on chow and WTD (n = 6-7). *p < 0.05,
**p < 0.01, ***p < 0.001, by Students t-tests (AC) or two-way analysis of
variance (D). Data are presented as mean SEM.
WTD feeding, and 2) Lrh-1 is reportedly required for full induction

of Fasn, the rate limiting enzyme of FA synthesis (Matsukuma
et al., 2007). This pattern makes L-FoxO1 mice a unique model
of concerted impairment of BA synthesis, creating an environment that is permissive for the development of hyperlipidemia.
This pattern also suggests that the present phenotype is unlikely
to be due to a single FoxO1 target gene but rather results from the
multiple abnormalities of BA production seen in this model.
Mice and Diets

Only male mice were studied, with the exception of the experiment in P2 mice,
for which we did not separate male and female neonates. L-FoxO1 and
L-FoxO1,O3,O4 mice have been described (Haeusler et al., 2010b; Matsumoto
et al., 2007). L-FoxO1 mice were backcrossed more than 9 generations to
C57BL/6J and crossed with Ldlr / mice (both from The Jackson Laboratories),
except mice used in Figures 4A4C, which were on a mixed genetic background
(C57BL/6J 3 129 3 FVB). Mice were fed either standard chow diet (Purina), or
WTD containing 42% kcal from fat and 0.2% cholesterol (TD 88137, Harlan
Teklad), starting at 47 weeks of age. Mice were kept on the diet for 1013 weeks
before sacrifice. For GW4064 or CA feeding, mice were fed WTD for
910 weeks, then switched to the same WTD, crushed to powder and mixed
with either 0.035% GW4064 (Tocris), or sodium cholate (Sigma): 0.5% for
3 weeks or 1 day (Watanabe et al., 2004) (for liver lipids and gene expression,
respectively in L-FoxO1), or 1% for 1 week (L-FoxO1:Ldlr / ). GW4064 treatment in L-FoxO1:Ldlr / mice was delivered by oral gavage, dissolved in corn
oil, at a dose of 18 mg/kg/day for 3 days and a final dose of 9 mg/kg, 5 hr before
sacrifice. Mice are housed in a 12-hr light/dark cycle, with the dark cycle occurring between 7:00 p.m. and 7:00 a.m. For the time-course experiment shown in
Figure 4C, mice were first synchronized by removing food for 4 hr (12:00 p.m.
4:00 p.m.) and then replacing food for 4 hr (4:008:00 p.m.). The 24-hr fast began
at 8:00 p.m., and mice were sacrificed before the fast (0) and periodically
throughout the fast. Refed mice were refed chow at 8:00 p.m., following the
full 24-hr fast. The light/dark cycle was maintained during this experiment.
Glucose Metabolic Tests
Ad libitum fed measurements were taken between 8:00 a.m. and 10:00 a.m.
and 5 fasted measurements were taken between 1:00 p.m. and 3:00 p.m.,
using OneTouch glucose monitor and strips (Lifescan) and insulin ELISA
(Millipore). Intraperitoneal glucose and insulin tolerance tests were performed
as described (Haeusler et al., 2010b; Nakae et al., 2002). For phosphorylated
Akt analysis, mice were fasted 5 hr, anesthetized, and then injected with 5 mU
of insulin (NovoLog, Novo Nordisk) into the inferior vena cava. Livers were
collected 5 min after injection. Antibodies against total and phospho-Ser473
Akt were from Cell Signaling.
Lipid Metabolic Tests
To measure hepatic lipids, livers were extracted as described (Folch et al., 1957).
Liver and serum lipids were measured by colorimetric assay: TG (Infinity, Thermo
Cell Metabolism
Figure 5. Defects in Bile Acid Composition, Lipid, and Glucose Metabolism in L-FoxO1 Mice
(A) Hepatic gene expression showing FXR activation due to GW4064 treatment (n = 67).
(B) Liver TGs from L-FoxO1 and control mice fed WTD supplemented with either GW4064 (0.035%) or CA (0.5%) for 3 weeks.
(C) Serum total and VLDL-TGs in L-FoxO1:Ldlr / and Ldlr / controls after GW4064 (4 daily treatments, orally) or supplementation of WTD with CA (1% for
1 week). For comparison in B and C, we also show lipid levels from mice fed chow and WTD (without treatment); these are the same mice whose mean values are
shown in Figure 1.
(D and E) Hepatic gene expression (D) and blood glucose levels (E) from P2 neonates, where nursing mothers ate either chow or chow supplemented with CA.
Expression of Cyp8b1 and Cyp7a1 from untreated mice are the same measurements presented in Figure 4B.
(F) Model of metabolic functions of FoxO1 in liver. Data are presented as mean SEM. *p < 0.05, **p < 0.01, ***p < 0.001, by Students t-tests (B, C, and E) or
two-way analysis of variance (A and D). N.S., not significant; HGP, hepatic glucose production. See also Figure S4.
Scientific), cholesterol E, and free cholesterol (Wako). Lipoprotein fractions were
separated by density ultracentrifugation as described (Haeusler et al., 2010a).
Bile Acid Measurements and Metabolomics
A complete list of BA abbreviations is available in Table S4. The metabolomics
study was performed after 10 weeks on chow or WTD. Livers were collected
after a 5-hr fast and snap-frozen. Hepatic BAs (relative values) were measured
by LC-MS/MS (-ESI), with the exception of a-MCA, which was measured by
GC/MS (Metabolon). The relative increase of T-CDCA in L-FoxO1 mice is
underestimated in both WTD and chow-fed mice because it was detectable
in only 4 out of 16 control mice but 12 out of 16 L-FoxO1 mice; samples below
the limit of detection were considered to be at the lowest detectable level.
Cell Metabolism
Gallbladder bile was collected by syringe from mice fed the WTD for 10 weeks
and pooled (5 per genotype). Fecal samples were collected for two consecutive days following 8 weeks of WTD feeding. For total BA pool, gallbladder and
small intestine were collected together with whole liver, a small piece of which
was used to confirm genotype. For bile, feces, and total BA pool, BAs were
extracted and measured by LC-MS/MS as described (Alnouti et al., 2008).
Detailed methods are provided in Supplemental Information.
Gene Expression
RNA was isolated using TRIzol (Invitrogen). GeneChip Mouse Exon 1.0 ST
Arrays were from Affymetrix. Expression analysis was performed using Partek
Genomics Suite (Partek). For qPCR, cDNA was synthesized using qScript
(QantaBioSciences), and qPCR was performed using SyBr Green (New
England Biolabs). Genes were normalized to 18 S. Primer sequences are
available upon request.
Statistical Analyses
Data are presented as mean SEM. Data were analyzed by two-tailed
Students t-tests, with the following exceptions: metabolomics data were
analyzed by Welchs two-sample t-tests; microarray data was analyzed by
analysis of variance (ANOVA), using Partek Genomics Suite; and data comparing two variables (Figures 4D, 5A, 5D, S4E, and S4F) were analyzed by
two-way ANOVA; p < 0.05 was considered significant.
Supplemental Information includes Supplemental Experimental Procedures,
four figures, and four tables and can be found with this article online at
doi:10.1016/j.cmet.2011.11.010.
ACKNOWLEDGMENTS
This work was supported by NIH grants F32HL103103 to R.A.H, HL87123 and
DK58282 to D.A., and DK63608 (Columbia University Diabetes and Endocrinology Research Center). We thank A. Tall, I. Tabas, D. Mangelsdorf, J. Horton,
and U. Pajvani for valuable comments and discussions, Ana Flete for technical
assistance, and members of the Accili laboratory for discussion of the data.
R.A.H. and D.A. designed and analyzed the experiments and wrote the paper.
R.A.H. performed experiments. M.P.-H. and C.D.K. performed bile acid
measurements by LC-MS/MS. C.L.W. performed aortic root lesion analysis.
All authors edited the manuscript.
Received: July 18, 2011
Accepted: November 28, 2011
Published online: December 22, 2011
REFERENCES
combined hyperlipidemic hamsters. Am. J. Physiol. Endocrinol. Metab. 290,

E716E722.
Brown, M.S., and Goldstein, J.L. (2008). Selective versus total insulin resistance: a pathogenic paradox. Cell Metab. 7, 9596.
Brufau, G., Stellaard, F., Prado, K., Bloks, V.W., Jonkers, E., Boverhof, R.,
Kuipers, F., and Murphy, E.J. (2010). Improved glycemic control with colesevelam treatment in patients with type 2 diabetes is not directly associated
with changes in bile acid metabolism. Hepatology.
Chiang, J.Y. (2009). Bile acids: regulation of synthesis. J. Lipid Res. 50, 1955
1966.
Choi, S.H., and Ginsberg, H.N. (2011). Increased very low density lipoprotein
(VLDL) secretion, hepatic steatosis, and insulin resistance. Trends
Endocrinol. Metab. 22, 353363.
DeFronzo, R.A. (2010). Insulin resistance, lipotoxicity, type 2 diabetes and
atherosclerosis: the missing links. The Claude Bernard Lecture 2009.
Dong, X.C., Copps, K.D., Guo, S., Li, Y., Kollipara, R., DePinho, R.A., and
White, M.F. (2008). Inactivation of hepatic Foxo1 by insulin signaling is required
for adaptive nutrient homeostasis and endocrine growth regulation. Cell
Metab. 8, 6576.
Eden, E., Navon, R., Steinfeld, I., Lipson, D., and Yakhini, Z. (2009). GOrilla:
a tool for discovery and visualization of enriched GO terms in ranked gene lists.
BMC Bioinformatics 10, 48.
Evans, M.J., Mahaney, P.E., Borges-Marcucci, L., Lai, K., Wang, S., Krueger,
J.A., Gardell, S.J., Huard, C., Martinez, R., Vlasuk, G.P., and Harnish, D.C.
(2009). A synthetic farnesoid X receptor (FXR) agonist promotes cholesterol
lowering in models of dyslipidemia. Am. J. Physiol. Gastrointest. Liver
Physiol. 296, G543G552.
Folch, J., Lees, M., and Sloane Stanley, G.H. (1957). A simple method for the
isolation and purification of total lipides from animal tissues. J. Biol. Chem.
226, 497509.
Forsyth, J.S., Ross, P.E., and Bouchier, I.A. (1983). Bile salts in breast milk.
Eur. J. Pediatr. 140, 126127.
Gafvels, M., Olin, M., Chowdhary, B.P., Raudsepp, T., Andersson, U., Persson,
B., Jansson, M., Bjorkhem, I., and Eggertsen, G. (1999). Structure and chromosomal assignment of the sterol 12alpha-hydroxylase gene (CYP8B1) in
human and mouse: eukaryotic cytochrome P-450 gene devoid of introns.
Genomics 56, 184196.
Goldstein, J.L., DeBose-Boyd, R.A., and Brown, M.S. (2006). Protein sensors
for membrane sterols. Cell 124, 3546.
Goodwin, B., Jones, S.A., Price, R.R., Watson, M.A., McKee, D.D., Moore,
L.B., Galardi, C., Wilson, J.G., Lewis, M.C., Roth, M.E., et al. (2000). A regulatory cascade of the nuclear receptors FXR, SHP-1, and LRH-1 represses bile
acid biosynthesis. Mol. Cell 6, 517526.
Haeusler, R.A., and Accili, D. (2008). The double life of Irs. Cell Metab. 8, 79.
Akiyoshi, T., Uchida, K., Takase, H., Nomura, Y., and Takeuchi, N. (1986).
Cholesterol gallstones in alloxan-diabetic mice. J. Lipid Res. 27, 915924.
Alnouti, Y., Csanaky, I.L., and Klaassen, C.D. (2008). Quantitative-profiling of
bile acids and their conjugates in mouse liver, bile, plasma, and urine using
LC-MS/MS. J. Chromatogr. B Analyt. Technol. Biomed. Life Sci. 873, 209217.
Haeusler, R.A., Han, S., and Accili, D. (2010a). Hepatic FoxO1 ablation exacerbates lipid abnormalities during hyperglycemia. J. Biol. Chem. 285,
2686126868.
Haeusler, R.A., Kaestner, K.H., and Accili, D. (2010b). FoxOs function synergistically to promote glucose production. J. Biol. Chem. 285, 3524535248.
Angelin, B., Einarsson, K., Hellstrom, K., and Leijd, B. (1978). Effects of cholestyramine and chenodeoxycholic acid on the metabolism of endogenous
triglyceride in hyperlipoproteinemia. J. Lipid Res. 19, 10171024.
Heuman, D.M. (1989). Quantitative estimation of the hydrophilic-hydrophobic

balance of mixed bile salt solutions. J. Lipid Res. 30, 719730.
Biddinger, S.B., Haas, J.T., Yu, B.B., Bezy, O., Jing, E., Zhang, W., Unterman,
T.G., Carey, M.C., and Kahn, C.R. (2008a). Hepatic insulin resistance directly
promotes formation of cholesterol gallstones. Nat. Med. 14, 778782.
Heuman, D.M., Hylemon, P.B., and Vlahcevic, Z.R. (1989). Regulation of bile
acid synthesis. III. Correlation between biliary bile salt hydrophobicity index
and the activities of enzymes regulating cholesterol and bile acid synthesis
in the rat. J. Lipid Res. 30, 11611171.
Biddinger, S.B., Hernandez-Ono, A., Rask-Madsen, C., Haas, J.T., Aleman,

J.O., Suzuki, R., Scapa, E.F., Agarwal, C., Carey, M.C., Stephanopoulos, G.,
et al. (2008b). Hepatic insulin resistance is sufficient to produce dyslipidemia
and susceptibility to atherosclerosis. Cell Metab. 7, 125134.
Bilz, S., Samuel, V., Morino, K., Savage, D., Choi, C.S., and Shulman, G.I.
(2006). Activation of the farnesoid X receptor improves lipid metabolism in
Isomaa, B., Almgren, P., Tuomi, T., Forsen, B., Lahti, K., Nissen, M., Taskinen,
M.R., and Groop, L. (2001). Cardiovascular morbidity and mortality associated
with the metabolic syndrome. Diabetes Care 24, 683689.
Kalaany, N.Y., and Mangelsdorf, D.J. (2006). LXRS and FXR: the yin and yang
of cholesterol and fat metabolism. Annu. Rev. Physiol. 68, 159191.
Cell Metabolism
Kamagate, A., Qu, S., Perdomo, G., Su, D., Kim, D.H., Slusher, S., Meseck, M.,
and Dong, H.H. (2008). FoxO1 mediates insulin-dependent regulation of
hepatic VLDL production in mice. J. Clin. Invest. 118, 23472364.
Murphy, C., Parini, P., Wang, J., Bjorkhem, I., Eggertsen, G., and Gafvels, M.
(2005). Cholic acid as key regulator of cholesterol synthesis, intestinal absorption and hepatic storage in mice. Biochim. Biophys. Acta 1735, 167175.
Kitamura, Y.I., Kitamura, T., Kruse, J.P., Raum, J.C., Stein, R., Gu, W., and
Accili, D. (2005). FoxO1 protects against pancreatic beta cell failure through
NeuroD and MafA induction. Cell Metab. 2, 153163.
Nakae, J., Biggs, W.H., III, Kitamura, T., Cavenee, W.K., Wright, C.V., Arden,
K.C., and Accili, D. (2002). Regulation of insulin action and pancreatic betacell function by mutated alleles of the gene encoding forkhead transcription
factor Foxo1. Nat. Genet. 32, 245253.
Kong, B., Luyendyk, J.P., Tawfik, O., and Guo, G.L. (2009). Farnesoid X
receptor deficiency induces nonalcoholic steatohepatitis in low-density lipoprotein receptor-knockout mice fed a high-fat diet. J. Pharmacol. Exp. Ther.
328, 116122.
Lee, Y.K., Schmidt, D.R., Cummins, C.L., Choi, M., Peng, L., Zhang, Y.,
Goodwin, B., Hammer, R.E., Mangelsdorf, D.J., and Kliewer, S.A. (2008).
Liver receptor homolog-1 regulates bile acid homeostasis but is not essential
for feedback regulation of bile acid synthesis. Mol. Endocrinol. 22, 13451356.
Lefebvre, P., Cariou, B., Lien, F., Kuipers, F., and Staels, B. (2009). Role of bile
acids and bile acid receptors in metabolic regulation. Physiol. Rev. 89,
147191.
Li, T., Kong, X., Owsley, E., Ellis, E., Strom, S., and Chiang, J.Y. (2006). Insulin
regulation of cholesterol 7alpha-hydroxylase expression in human hepatocytes: roles of forkhead box O1 and sterol regulatory element-binding protein
1c. J. Biol. Chem. 281, 2874528754.
Li, S., Brown, M.S., and Goldstein, J.L. (2010). Bifurcation of insulin signaling
pathway in rat liver: mTORC1 required for stimulation of lipogenesis, but not
inhibition of gluconeogenesis. Proc. Natl. Acad. Sci. USA 107, 34413446.
Li-Hawkins, J., Gafvels, M., Olin, M., Lund, E.G., Andersson, U., Schuster, G.,
Bjorkhem, I., Russell, D.W., and Eggertsen, G. (2002). Cholic acid mediates
negative feedback regulation of bile acid synthesis in mice. J. Clin. Invest.
110, 11911200.
Lin, H.V., and Accili, D. (2011). Hormonal regulation of hepatic glucose
production in health and disease. Cell Metab. 14, 919.
Lu, T.T., Makishima, M., Repa, J.J., Schoonjans, K., Kerr, T.A., Auwerx, J., and
Mangelsdorf, D.J. (2000). Molecular basis for feedback regulation of bile acid
synthesis by nuclear receptors. Mol. Cell 6, 507515.
Makishima, M., Okamoto, A.Y., Repa, J.J., Tu, H., Learned, R.M., Luk, A., Hull,
M.V., Lustig, K.D., Mangelsdorf, D.J., and Shan, B. (1999). Identification of
a nuclear receptor for bile acids. Science 284, 13621365.
Maloney, P.R., Parks, D.J., Haffner, C.D., Fivush, A.M., Chandra, G., Plunket,
K.D., Creech, K.L., Moore, L.B., Wilson, J.G., Lewis, M.C., et al. (2000).
Identification of a chemical tool for the orphan nuclear receptor FXR.
J. Med. Chem. 43, 29712974.
Mataki, C., Magnier, B.C., Houten, S.M., Annicotte, J.S., Argmann, C.,
Thomas, C., Overmars, H., Kulik, W., Metzger, D., Auwerx, J., and
Schoonjans, K. (2007). Compromised intestinal lipid absorption in mice with
a liver-specific deficiency of liver receptor homolog 1. Mol. Cell. Biol. 27,
83308339.
Matsukuma, K.E., Wang, L., Bennett, M.K., and Osborne, T.F. (2007). A key
role for orphan nuclear receptor liver receptor homologue-1 in activation of
fatty acid synthase promoter by liver X receptor. J. Biol. Chem. 282, 20164
20171.
Matsumoto, M., Han, S., Kitamura, T., and Accili, D. (2006). Dual role of transcription factor FoxO1 in controlling hepatic insulin sensitivity and lipid metabolism. J. Clin. Invest. 116, 24642472.
Matsumoto, M., Pocai, A., Rossetti, L., Depinho, R.A., and Accili, D. (2007).
Impaired regulation of hepatic glucose production in mice lacking the forkhead
transcription factor Foxo1 in liver. Cell Metab. 6, 208216.
Meier, H., Hoag, W.G., and McBurney, J.J. (1965). Chemical characterization
of inbred-strain mouse milk. I. Gross composition and amino acid analysis.
J. Nutr. 85, 305308.
Miller, N.E., and Nestel, P.J. (1974). Triglyceride-lowering effect of chenodeoxycholic acid in patients with endogenous hypertriglyceridaemia. Lancet 2,
929931.
National Institute of Diabetes and Digestive and Kidney Diseases. (2005).

National Diabetes Statistics fact sheet: general information and national estimates on diabetes in the United States (Bethesda, MD: U.S. Department of
Health and Human Services, National Institute of Health).
Parks, D.J., Blanchard, S.G., Bledsoe, R.K., Chandra, G., Consler, T.G.,
Kliewer, S.A., Stimmel, J.B., Willson, T.M., Zavacki, A.M., Moore, D.D., and
Lehmann, J.M. (1999). Bile acids: natural ligands for an orphan nuclear
receptor. Science 284, 13651368.
Ridlon, J.M., Kang, D.J., and Hylemon, P.B. (2006). Bile salt biotransformations by human intestinal bacteria. J. Lipid Res. 47, 241259.
Russell, D.W. (2009). Fifty years of advances in bile acid synthesis and metabolism. J. Lipid Res. Suppl. 50, S120S125.
Shin, D.J., and Osborne, T.F. (2009). FGF15/FGFR4 integrates growth factor
signaling with hepatic bile acid metabolism and insulin action. J. Biol. Chem.
284, 1111011120.
Sinal, C.J., Tohkin, M., Miyata, M., Ward, J.M., Lambert, G., and Gonzalez, F.J.
(2000). Targeted disruption of the nuclear receptor FXR/BAR impairs bile acid
and lipid homeostasis. Cell 102, 731744.
Srensen, L.P., Sndergaard, E., Nellemann, B., Christiansen, J.S., Gormsen,
L.C., and Nielsen, S. (2011). Increased VLDL-triglyceride secretion precedes
impaired control of endogenous glucose production in obese, normoglycemic
men. Diabetes 60, 22572264.
Staels, B., and Kuipers, F. (2007). Bile acid sequestrants and the treatment of
type 2 diabetes mellitus. Drugs 67, 13831392.
Tabas, I., Tall, A., and Accili, D. (2010). The impact of macrophage insulin resistance on advanced atherosclerotic plaque progression. Circ. Res. 106, 5867.
Thomas, C., Pellicciari, R., Pruzanski, M., Auwerx, J., and Schoonjans, K.
(2008). Targeting bile-acid signalling for metabolic diseases. Nat. Rev. Drug
Discov. 7, 678693.
Uchida, K., Makino, S., and Akiyoshi, T. (1985). Altered bile acid metabolism in
nonobese, spontaneously diabetic (NOD) mice. Diabetes 34, 7983.
Uchida, K., Satoh, T., Takase, H., Nomura, Y., Takasu, N., Kurihara, H., and
Takeuchi, N. (1996). Altered bile acid metabolism related to atherosclerosis
in alloxan diabetic rats. J. Atheroscler. Thromb. 3, 5258.
Wan, M., Leavens, K.F., Saleh, D., Easton, R.M., Guertin, D.A., Peterson, T.R.,
Kaestner, K.H., Sabatini, D.M., and Birnbaum, M.J. (2011). Postprandial
hepatic lipid metabolism requires signaling through Akt2 independent of the
transcription factors FoxA2, FoxO1, and SREBP1c. Cell Metab. 14, 516527.
Wang, D.Q., Tazuma, S., Cohen, D.E., and Carey, M.C. (2003). Feeding natural
hydrophilic bile acids inhibits intestinal cholesterol absorption: studies in the
gallstone-susceptible mouse. Am. J. Physiol. Gastrointest. Liver Physiol.
285, G494G502.
Watanabe, M., Houten, S.M., Wang, L., Moschetta, A., Mangelsdorf, D.J.,
Heyman, R.A., Moore, D.D., and Auwerx, J. (2004). Bile acids lower triglyceride
levels via a pathway involving FXR, SHP, and SREBP-1c. J. Clin. Invest. 113,
14081418.
Zhang, Y., Lee, F.Y., Barrera, G., Lee, H., Vales, C., Gonzalez, F.J., Willson,
T.M., and Edwards, P.A. (2006). Activation of the nuclear receptor FXR
improves hyperglycemia and hyperlipidemia in diabetic mice. Proc. Natl.
Acad. Sci. USA 103, 10061011.
Cell Metabolism
Article
Acetylation Negatively Regulates Glycogen
Phosphorylase by Recruiting Protein Phosphatase 1
Tengfei Zhang,1,2 Shiwen Wang,1,2 Yan Lin,2 Wei Xu,1,2 Dan Ye,2 Yue Xiong,1,2,4,* Shimin Zhao,1,2,*
and Kun-Liang Guan1,2,3,5,6,*
1State
Key Lab of Genetic Engineering, School of Life Sciences

and Cell Biology Lab, Institutes of Biomedical Sciences
3Department of Biochemistry, College of Medicine
Fudan University, Shanghai 20032, China
4Department of Biochemistry and Biophysics, Lineberger Comprehensive Cancer Center, University of North Carolina at Chapel Hill,
Chapel Hill, NC 27599, USA
5Department of Pharmacology
6Moores Cancer Center
University of California, San Diego, La Jolla, CA 92093, USA
*Correspondence: yxiong@email.unc.edu (Y.X.), zhaosm@fudan.edu.cn (S.Z.), kuguan@ucsd.edu (K.-L.G.)
DOI 10.1016/j.cmet.2011.12.005
2Molecular
SUMMARY
Glycogen phosphorylase (GP) catalyzes the ratelimiting step in glycogen catabolism and plays a
key role in maintaining cellular and organismal
glucose homeostasis. GP is the first protein whose
function was discovered to be regulated by reversible protein phosphorylation, which is controlled by
phosphorylase kinase (PhK) and protein phosphatase 1 (PP1). Here we report that lysine acetylation
negatively regulates GP activity by both inhibiting
enzyme activity directly and promoting dephosphorylation. Acetylation of GP Lys470 enhances its interaction with the PP1 substrate-targeting subunit, GL,
and PP1, thereby promoting GP dephosphorylation
and inactivation. We show that GP acetylation is
stimulated by glucose and insulin and inhibited by
glucagon. Our results provide molecular insights
into the intricate regulation of the classical GP and
a functional crosstalk between protein acetylation
and phosphorylation.
INTRODUCTION
Glycogen phosphorylase (GP) catalyzes phosphorolytic cleavage of glycogen to produce glucose-1-phosphate for glucosedependent tissues when body glucose is scarce. GP activity
plays an important role in glucose homeostasis and glycogen
metabolism. Defects in glycogen synthesis and breakdown
in liver, muscle, and other glucose-dependent tissues often
cause glycogen storage diseases (Stegelmeier et al., 1995).
For example, McArdle disease is caused by mutations in muscle
GP, and patients with this disorder are intolerant of exercise
and show early fatigue (Andreu et al., 2007; Tang et al., 2003).
Due to their critical roles in glucose homeostasis, regulations
of glycogen synthase (GS) and GP activities have been extensively investigated in hopes of finding therapeutic strategy for
type II diabetes.
GP was the first allosteric enzyme to be discovered by Carl

and Gerty Cori, who demonstrated that GP existed in two interconvertible forms, referred to subsequently as an active a form or
inactive b form, that are regulated by adenosine monophosphate
(AMP) (Cori and Cori, 1936). GP is activated by the binding of
AMP and IMP, whereas it is inhibited by the binding of ATP
and glucose-6-phosphate (Barford et al., 1991; Barford and
Johnson, 1989). These allosteric controls provide excellent
means of GP activity regulation in response to energy and metabolic status. Moreover, GP activity is tightly regulated by reversible phosphorylation and dephosphorylation. In addition to
allosteric regulation, GP is also regulated by posttranslational
modifications (PTMs) (Johnson, 1992). In fact, GP was also the
first protein discovered to be regulated by reversible protein
phosphorylation (Fischer and Krebs, 1955; Sutherland and
Wosilait, 1955), which exemplifies a signal transduction pathway
by phosphorylation cascades. Under high serum glucose conditions, release of insulin indirectly activates protein phosphatase
1 (PP1), which dephosphorylates Ser-15 and converts the active
a form of GP to the unphosphorylated inactive b form, leading to
the inhibition of glycogen breakdown (Browner and Fletterick,
1992). Conversely, when glucose concentration is low, glucagon
triggers a cascade of signal transduction that activates protein
kinase A (PKA) which phosphorylates and activates phosphorylase kinase (PhK), which in turn activates GP by phosphorylating
Ser-15 and leads to increased glycogen breakdown and ultimately higher glucose levels.
PP1 regulates glycogen metabolism by inhibiting GP activity
and activating GS activity. The hepatic glycogen binding subunit
GL is a glycogen-metabolizing scaffold protein that binds to PP1,
glycogen, GS, and GP (Armstrong et al., 1998). GL targets PP1
to glycogen, where it dephosphorylates and inhibits GP, in addition to dephosphorylating and activating GS, thereby increasing
glycogen synthesis and reducing glucose output (Alemany and
Cohen, 1986). Therefore, the phosphorylation status of GP is
critically regulated by its interaction with GL.
Lysine acetylation has emerged as a common regulatory
mechanism of diverse cellular processes, including metabolism
(Choudhary et al., 2009; Kim et al., 2006; Wang et al., 2010; Zhao
et al., 2010). Acetylation modulates enzymes involved in fatty
Cell Metabolism
Acetylation Inhibits Glycogen Phosphorylase
acid metabolism, urea cycle, TCA cycle, and gluconeogenesis

via different mechanisms such as inhibition, activation, and
protein destabilization (Guan and Xiong, 2011; Kim and Yang,
2011). In the present study, we show that acetylation inhibits
GP activity by promoting its dephosphorylation. This is accomplished by an acetylation-induced interaction between GP and
the PP1-targeting subunit GL. Our study provides an example
of crosstalk between acetylation and phosphorylation in regulation of a key enzyme of the glycogen metabolism in response to
different physiologic conditions.
RESULTS
Acetylation Negatively Regulates GP Catalytic Activity
In an attempt to profile liver protein acetylation, we previously
enriched acetylated peptides of human liver proteins by affinity
purification (Zhao et al., 2010). Among the many liver acetylated proteins identified, two peptides were found to contain
acetylated lysine 470 (K470) and lysine 796 (K796) of GP (see
Figure S1A available online). To confirm GP acetylation, we expressed Flag-tagged GP in Changs liver cells and then treated
cells with nicotinamide (NAM) and trichostatin A (TSA), two
commonly used deacetylase inhibitors that inhibit all four classes
of deacetylases (Xu et al., 2007). GP was found to be acetylated,
and its acetylation was significantly elevated (approximately
2-fold) after NAM and TSA treatment (Figure 1A). Mutation of
both K470 and K796 (GPK2R) dramatically decreased GP acetylation (Figure 1B), indicating that K470 and K796 are the major,
if not the sole, acetylation sites in GP.
Next, the functional significance of acetylation in GP catalytic
activity was investigated. For GP assay, we adapted a published
protocol (Jones and Wright, 1970) and confirmed that under the
conditions used in our experiments, the assay was linear to GP
enzyme concentrations and displayed a typical MichaelisMenten response to glycogen concentrations (Figures S1B and
S1C). Treatment with deacetylase inhibitors decreased the
specific activity of GP by 75% (Figure 1C), suggesting that GP
activity is negatively regulated by acetylation. Moreover, when
both K470 and K796 of GP were changed to acetylation mimetic
glutamines (GPK2Q), the GPK2Q mutant displayed significantly
lower (about 55%) specific activity than the wild-type. Interestingly, the GPK2Q was no longer inhibited by deacetylase inhibitor
treatment (Figure 1C), indicating that the K470 and K796 in GP
are the primary acetylation sites for its enzymatic inhibition. To
clarify the individual contribution of K470 and K796 acetylation
to GP activity, we generated single lysine to glutamine substitution on either K470 (GPK470Q) or K796 (GPK796Q). GPK470Q and
GPK796Q had about 30% and 45% less specific activity than
wild-type GP, respectively (Figure 1D), suggesting that acetylation of either K470 or K796 contributed to GP inhibition. Notably,
unlike GPK2Q, which had negligible response to deacetylase
inhibitor treatment, both GPK470Q and GPK796Q activities were
still partially inhibited by NAM and TSA treatment (Figure 1D),
suggesting that both K470 and K796 acetylation contribute to
GP catalytic activity regulation and that these two acetylation
sites may function additively.
Because GP can form homodimers, we investigated whether
endogenous GP might interfere with our assays by forming heterodimers with the transfected mutant GP, and whether the
mutant GP could dominantly inhibit wild-type GP. In the transfected cells, the ectopically expressed GP was much higher
(about 10-fold) than the endogenous protein GP (Figure S1D),
indicating that endogenous GP should have a negligible effect
on our assay results. Moreover, we purified WT homodimer,
WT-K2R heterodimer, and K2R homodimer and assayed their
catalytic activity. The WT-K2R heterodimer exhibited activity
that was the average of wild-type and K2R homodimers (Figure S1E), showing that the K2R mutant has no dominant inhibition on wild-type GP. Together, our results suggest that K470
and K796 are two major acetylation sites in GP responsible for
its inhibition by acetylation. To further test the effect of acetylation on GP activity and glycogen metabolism, we determined
cellular glycogen content in response to NAM and TSA treatment. Consistent with an inhibitory effect of acetylation on GP
activity, we found that inhibition of deacetylases by NAM and
TSA increased cellular glycogen levels (Figure S1F). Next, we
determined glycogen hydrolysis in cells expressing different
GP mutants. Expression of wild-type GP promoted a faster
glycogen hydrolysis than the GPK2Q mutant (Figure 1E), further
supporting a lower enzymatic activity of the acetylation mimetic
mutant GPK2Q.
GP Acetylation and Activity Are Regulated by Glucose,
Insulin, and Glucagon
Given that glucose concentration is a key factor in GP regulation,
we determined GP acetylation under different glucose concentrations. The acetylation level of ectopically expressed GP was
low when cells were maintained in glucose-free medium, and
increased significantly with elevated glucose concentrations
in a dose-dependent manner by as much as 4-fold (Figure 2A),
suggesting that acetylation of GP was upregulated by glucose.
Moreover, acetylation of endogenous GP in L02 human hepatocytes was increased by glucose (Figure 2B, Figure S2A). NAM
and TSA treatment of cells cultured in high-glucose medium
could further increase the acetylation level of endogenous GP
(Figure 2B), showing that deacetylation of GP occurs even under
high glucose.
Next, the effect of glucose on GP activity was determined. As
shown in Figure 2C, the activity of wild-type GP was significantly inhibited by increasing concentrations of glucose, while
GPK2Q was largely refractory to inhibition by glucose (Figure 2C). These results suggest that glucose inhibits GP activity
through K470 and K796 acetylation. To further test the function
of acetylation in mediating glucose-induced GP inhibition, we
measured the effect of glucose on GP activity after NAM and
TSA treatment. We hypothesized that since the inhibition of deacetylases would increase GP acetylation, high glucose may
not have a significant effect on GP activity in the presence of
deacetylase inhibitors. As expected, glucose increased GP
acetylation and decreased GP activity in the absence of deacetylase inhibitors (Figure 2D, lanes 14). However, in the presence of deacetylase inhibitors, glucose had little effect on GP
activity and did not further increase the elevated GP acetylation
(Figure 2D, lanes 58). These results further support the notion
that acetylation plays a major role in GP inhibition in response
to glucose.
Both insulin and glucagon are important signals that regulate
GP activity and glycogen metabolism in opposite manners
Cell Metabolism
Figure 1. GP Activity Is Negatively Regulated by Acetylation

(A) GP acetylation is increased by deacetylase inhibitor. Flag-tagged GP was expressed in Changs cells with or without nicotinamide and trichostatin
A (NAM+TSA) treatment. Acetylation levels of Flag bead-purified GP were blotted with pan-anti-acetyllysine antibody (a-AcK). IB and IP denote immunoblotting
and immunoprecipitation, respectively.
(B) K470 and K796 are the major acetylation sites in GP. Ectopically expressed Flag-GP and GPK2R in Changs cells were immunopurified and immunoblotted with
pan-anti-acetyllysine antibody.
(C) GP catalytic activity is negatively regulated by acetylation. Both wild-type GP and GPK2Q were expressed in Changs liver cells with or without NAM+TSA
treatment. Catalytic activity of affinity-purified GP proteins was determined and normalized to protein levels. Activity of wild-type GP under no treatment condition
was set as 100%.
(D) The acetylation targets K470 and K796 are important for GP inhibition by the deacetylase inhibitor treatment. GP, GPK470Q, GPK796Q, and GPK2Q were each expressed in Changs liver cells with (hatched bars) or without (solid bars) NAM+TSA treatment as indicated. Specific activity of each purified enzyme was determined.
(E) Wild-type GP degrades glycogen faster than K2Q mutant. Wild-type GP and mutant GPK2Q were expressed at similar levels in Changs liver cells maintained in
regular DMEM medium. The medium was replaced by glucose-free DMEM at time 0, and the glycogen degradation rate was measured.
All error bars represent standard deviation (SD). n = 3 for each experimental group.
(Lok et al., 1994; Massague and Guinovart, 1977). We tested

whether these two hormones could regulate GP acetylation. In
L02 human hepatocytes, the acetylation of endogenous GP
was increased with insulin treatment in time- and dose-dependent manners (Figure 2E, Figure S2B). Moreover, acetylation of
ectopically expressed GP increased within 30 min of insulin
Cell Metabolism
Figure 2. Regulation of GP Acetylation by Glucose, Insulin, and Glucagon

(A) Glucose increases GP acetylation. Acetylation levels of GP ectopically expressed in Changs cells maintained in different glucose concentrations were probed
by anti-acetyllysine antibody.
(B) Glucose and deacetylase inhibitor increase endogenous GP acetylation. Human hepatic L02 cells were treated with various concentrations of glucose and
deacetylase inhibitors as indicated. Endogenous GP was immunoprecipitated with a GP antibody. Acetylation levels of GP were determined by western blot.
(C) Glucose decreases GP activity through acetylation. Relative specific activity of GP (solid bars) and GPK2Q (hatched bars) expressed in Changs cells
maintained in different glucose concentrations was determined. Specific activity of GP from glucose-free medium was arbitrarily set as 100.
(D) Deacetylase inhibitor treatment blocks the glucose inhibition on GP activity. Relative specific activities for GP expressed in Changs cells maintained in
different glucose concentrations with (hatched bars) and without NAM+TSA treatment (solid bars) were determined.
(E) Insulin and deacetylase inhibitor increase endogenous GP acetylation. Human hepatic L02 cells were treated with increasing concentrations of insulin and
treated with deacetylase inhibitors as indicated. Acetylation of immunoprecipitated endogenous GP was detected by western blot.
(F) Insulin decreases GP activity through acetylation. Flag-GP (solid bars) and Flag-GPK2Q (hatched bars) were expressed in Changs cells and treated with
increasing concentrations of insulin. GP acetylation and activity were determined.
(G) Glucagon decreases GP acetylation. Acetylation levels of Flag-GP expressed in Changs cells treated with 10 nM glucagon for different time periods
(as indicated) were determined.
Cell Metabolism
addition (Figure S2C). We then examined the combined effect of

both glucose and insulin. GP acetylation in cells cultured in
the presence of both glucose and insulin was higher than cells
cultured in either glucose or insulin (Figure S2D). Consistent
with the stimulatory effect of insulin on GP acetylation, we found
that insulin inhibited GP activity in a dose-dependent manner
(Figure 2F, lanes 15). These effects required K470 and K796,
as the GPK2Q mutant showed little acetylation and its catalytic
activity did not respond to insulin treatment (Figure 2F, lanes
610). Contrary to insulin treatment, glucagon decreased GP
acetylation in a time-dependent manner (Figure 2G). Together,
these results suggest that insulin and glucagon may regulate
GP activity by affecting K470 and K796 acetylation.
Acetylation Decreases GP Phosphorylation
It is well-characterized that GP is activated by PhK-mediated
serine phosphorylation at Ser-15 (Nolan et al., 1964). To investigate the mechanism of acetylation in regulating GP activity, we
looked into a possible crosstalk between acetylation and phosphorylation by determining GP serine phosphorylation under
different conditions that are known to affect GP acetylation
levels. We found that deacetylase inhibitor treatment increased
GP acetylation and at the same time decreased Ser-15 phosphorylation by 70% (Figure 3A). Similar results were observed
when we utilized a panphosphoserine antibody to detect the
GP serine phosphorylation level after NAM and TSA treatment
(Figure S3A). The inverse relationship between endogenous GP
acetylation and phosphorylation in responding to deacetylase
inhibitor was also examined in L02 human hepatocytes and
freshly isolated mouse primary hepatocytes. In both hepatocytes, endogenous GP showed a significant increase in acetylation and a concomitant decrease in Ser-15 phosphorylation
upon NAM and TSA treatment (Figure 3B). Similar results were
also observed in freshly isolated mouse skeletal muscle cells
(Figure S3B). These observations demonstrate that acetylation
negatively regulates GP phosphorylation.
To provide direct evidence that acetylation regulates GP
phosphorylation, we examined GP phosphorylation status
when GP was coexpressed with deacetylases. When GP was
coexpressed with SIRT1 and SIRT2, two cytosolic deacetylases,
an evident decrease in GP acetylation and concomitant increase in GP Ser-15 phosphorylation were observed, with
SIRT2 being more potent in deacetylating GP and increasing
Ser-15 phosphorylation (Figure 3C). Collaborating with these
findings, the catalytic activity of GP was also activated by either
SIRT1 or SIRT2 coexpression. This result indicates a causal role
of acetylation in modulating GP phosphorylation.
To further elucidate the relationship between GP acetylation
and phosphorylation, we determined the effect of deacetylase
inhibitors on the activity of GPS15D, which had Ser-15 replaced
by an aspartic acid and displayed a lower activity (Buchbinder
et al., 1997). Interestingly, the activity of GPS15D mutant was
not inhibited by deacetylase inhibitor treatment, whereas the
wild-type GP was potently inhibited by a similar treatment
(Figure 3D). Of note, glucose still increased the acetylation of
GPS15D mutant (Figure 3E). Therefore, Ser-15 phosphorylation
is not required for high-glucose-stimulated GP acetylation;
rather, acetylation may downregulate Ser-15 phosphorylation
and activity of GP.
Because of the inhibitory effect of acetylation on phosphorylation, the above data were unable to show whether acetylation
may directly affect GP activity. To address this question, we
prepared GP with or without acetylation and with or without
phosphorylation. Hyperacetylated GP was obtained by expressing Flag-GP in Changs liver cells maintained in high-glucose
(25 mM) medium supplemented with NAM and TSA, while hypoacetylated GP was obtained by glucose-free medium without
NAM and TSA. The immunopurified GP proteins were treated
with l-phosphatase or PhK in vitro. After verifying both the acetylation and the Ser-15 phosphorylation of GP by western blotting
(right panel, Figure 3F), GP activity was determined. We found
that phosphorylation by GP kinase (PhK) dramatically increased
GP activity regardless of whether GP was hyperacetylated (lane
1 versus lane 2, left panel, Figure 3F) or hypoacetylated (lane 3
versus lane 4). Acetylation did not affect GP activity when GP
was hypophosphorylated (lanes 1 versus lane 3). However,
when GP was hyperphosphorylated, acetylation exhibited an
inhibitory effect on GP in vitro (lane 2 versus lane 4). These results
indicate that acetylation inhibits GP only when it is hyperphosphorylated and active.
GP is also regulated by allosteric effect, including activation
of inactive b form by AMP. To explore the effect of acetylation
on allosteric regulation of GP, we determined the activity of
hypo- versus hyperacetylated GP in response to AMP. We found
that AMP induced a similar activation to both the hyperacetylated (lane 1 versus lane 1A, 20.5-fold, Figure 3F) and hypoacetylated GP (lane 3 versus lane 3A, 20.4-fold) when GP was
hypophosphorylated, indicating that acetylation does not
directly affect allosteric activation of GP by AMP. However,
when GP was fully activated (hypoacetylated/hyperphosphorylated), AMP could not activate GP further (lanes 4 versus 4A).
To provide more direct evidence to determine the effect of
acetylation on allosteric regulation, we treated immunopurified
GP with recombinant deacetylase CobB in vitro and measured
its allosteric activation by AMP after confirming the decrease
of acetylation (right panel, Figure 3G). We found that AMP activated GP equally regardless of the levels of GPs acetylation
(5.54-fold, 5.58-fold, and 5.50-fold, left panel, Figure 3G). We
therefore conclude that acetylation does not directly affect allosteric activation of GP by AMP.
Physiological Stimuli Regulate GP Acetylation-Induced
Dephosphorylation
The notion that acetylation negatively regulates GP-Ser-15
phosphorylation was further pursued by analyzing acetylation
and phosphorylation levels in response to physiological stimuli.
We found that glucose increased GP acetylation and decreased
phosphorylation in a dose-dependent manner (Figure 4A), indicating an inverse relationship between GP acetylation and
phosphorylation. Moreover, insulin treatment increased GP
acetylation and decreased GP phosphorylation (Figure 4B),
whereas glucagon treatment decreased GP acetylation and
increased GP phosphorylation (Figure 4C). A time-dependent
inverse correlation between acetylation and phosphorylation in
GP was observed in response to glucose, insulin, and glucagon
(Figure 4D, Figures S4A and S4B), supporting the notion that
acetylation may inhibit GP phosphorylation. It is worth noting
that glucagon can regulate GP activity through a direct signaling
Cell Metabolism
Figure 3. Acetylation Decreases GP Ser-15 Phosphorylation

(A) Deacetylase inhibitor treatment inhibits GP phosphorylation. Flag-GP was expressed in Changs liver cells and treated with or without NAM+TSA. Acetylation
and GP-Ser-15 phosphorylation levels of Flag-GP proteins were determined by western blot.
(B) Deacetylase inhibitor treatment increases acetylation and decreases phosphorylation of hepatic GP. Human hepatic L02 cells and mouse primary hepatocytes were treated with or without NAM+TSA as indicated. Acetylation and Ser-15 phosphorylation of GP proteins were determined by western blot (L02,
#
p = 0.0094, n = 3, and ##p = 0.0107, n = 3; primary hepatocytes, #p = 0.0090, n = 3, and ##p = 0.0270, n = 3).
(C) Coexpression of deacetylases reduces GP acetylation and increases GP phosphorylation. Flag-GP was expressed alone or coexpressed with Sirt1 and Sirt2
in Changs liver cells as indicated. Acetylation and Ser-15 phosphorylation of affinity-purified GP were measured. Relative specific activities of GP were
determined by normalizing GP activity against GP protein.
(D) GP Ser-15 is required for deacetylase inhibitor-induced GP inactivation. Wild-type GP and GPS15D mutant GP were expressed in Changs liver cells, with or
without NAM+TSA treatment. Activities of affinity-purified GP were determined.
Cell Metabolism
pathway. Of note, although Ser-15 phosphorylation occurred

rapidly and was evident as early as 30 min after glucagon
treatment, it increased continuously throughout the 4 hr of
experimental duration (Figures 2G and 4D). Thus, it is possible
that deacetylation may not be required for the acute GP phosphorylation by glucagon but rather may play a role to maintain
the phosphorylated and active GP in the active state (Figure S7).
We next examined the response of Ser-15 phosphorylation
in GPK2R, the nonacetylatable form of GP, to different glucose
concentrations. Contrary to wild-type GP, which showed
increased acetylation and decreased Ser-15 phosphorylation
by glucose, Ser-15 phosphorylation of GPK2R was largely unaffected by increasing glucose concentrations (Figure S4C),
indicating an essential role of acetylation in regulating Ser-15
phosphorylation. Notably, Ser-15 phosphorylation remained
constitutively high in GPK2R regardless of the glucose concentrations, suggesting that the deacetylated GP favors being
phosphorylated at Ser-15. This notion was supported by the
observation that GPK2Q, the acetylation mimetic mutant, not
only showed a very weak phosphorylation level but that its
phosphorylation was unresponsive to varying glucose concentrations (Figure S4D). Similarly, insulin did not influence the
phosphorylation levels of either GPK2Q or GPK2R (Figure S4E).
Collectively, the above data indicate that acetylation of K470
and K796 is required for GP phosphorylation regulation by
glucose and insulin.
To obtain in vivo data to support the inverse relationship
between GP acetylation and Ser-15 phosphorylation, mouse
experiments were performed. Upon feeding, a state that causes
high glucose or high insulin, mouse liver GP acetylation level was
high, whereas the level of Ser-15 phosphorylation was low (Figure 4E, Figure S4F). After overnight fasting, hepatic GP acetylation was decreased, while hepatic Ser-15 phosphorylation
was inversely increased. Furthermore, when mice were intraperitoneally injected with glucose (1 g/kg), insulin (5 U/kg), or
glucagon (0.2 mg/kg), in line with our observations in vitro,
glucose and insulin injection increased endogenous GP acetylation and decreased Ser-15 phosphorylation (Figures 4F and 4G,
Figures S4G and S4H). In contrast, glucagon injection decreased
endogenous GP acetylation and increased Ser-15 phosphorylation in mouse livers (Figure 4H, Figure S4I). Together, these data
indicate an inverse coregulation of GP acetylation and Ser-15
phosphorylation, and the regulation of these two PTMs under
physiological conditions.
Acetylation Increases the GP-PP1a Interaction
After establishing a causal relationship between acetylation and
phosphorylation, a key question is how acetylation modulates
GP phosphorylation, which is controlled by the PhK and PP1.

The interaction between GP and PP1 is important for GP
dephosphorylation. As expected, coexpression of PP1a, the
phosphatase that dephosphorylates P-Ser-15 of GP, largely
abolished the inhibitory effect of deacetylase inhibitor on GP
activity (Figure S5A). We thus investigated whether acetylation
affected the PP1a-GP interaction. When PP1a and GP were
coexpressed in Changs cells, the interaction between PP1a
and GP was readily detectable (Figure 5A). Interestingly, the
PP1a-GP interaction was increased (by more than 100%) after
deacetylase inhibitor treatment (Figure 5A), suggesting that
acetylation enhances the recruitment of PP1a to GP. Consistent
with a role of acetylation in promoting the interaction between
PP1a and GP, deacetylase inhibitor treatment did not further
increase the interaction between PP1a and GPK2Q (Figure 5B).
To further confirm the function of acetylation in enhancing the
PP1a-GP interaction, we compared the binding of PP1a to GP
and the acetylation mimic GPK2Q mutant. When PP1a and GP
proteins were coexpressed in Changs cells, the amount of
PP1a protein coimmunoprecipitated with GPK2Q was about 2fold more than that associated with the wild-type GP protein,
suggesting that the acetylation of GP may enhance its interaction
with PP1a. Consistently, the phosphorylation level of GPK2Q was
about 60% weaker than that of the wild-type GP (Figure 5C).
Moreover, deacetylase inhibitor treatment decreased Ser-15
phosphorylation of GP that was already reduced by PP1a coexpression (Figure 5A, Figure S5B). Furthermore, coexpression of
SIRT2 impaired the interaction between GP and PP1a and at
the same time stimulated GP serine phosphorylation (Figure 5D).
Taken together, the above data support a model in which acetylation of K470 and K796 in GP enhances its interaction with
PP1a, thereby resulting in GP dephosphorylation and inactivation (Figure 7).
K470 Acetylation Increases GP-GL Interaction
The PP1-GP interaction is mediated by GL, a substrate-targeting
subunit of PP1 (Armstrong et al., 1998). To determine whether
the observed PP1a-GP interaction resulted from increased
PP1a-GL, Flag-PP1a and HA-GL were coexpressed in Changs
cells, and the interaction between PP1a and GL was determined.
We found that the association between PP1a and GL was not
altered by deacetylase inhibitor treatment (Figure S6A), indicating that acetylation does not regulate interaction between
PP1 and GL. On the other hand, the interaction between GP
and GL was increased upon deacetylase inhibitor treatment (Figure 6A, lanes 3 and 4). GL binds to GP through its C-terminal 269
284 residues, and deletion of these five residues in GL diminishes
its interaction with GP (Kelsall et al., 2007; Pautsch et al., 2008).
(E) The effect of glucose on GP activity but not acetylation depends on Ser-15. GPS15D-transfected Changs liver cells were treated with different concentrations
of glucose. Acetylation, phosphorylation, and relative activity of GPS15D were determined. Activity of GP from glucose-free medium was set as 100% arbitrarily.
(F) Effect of acetylation and phosphorylation on GP activity and activation by AMP. Hyperacetylated GP was immunoprecipitated from Flag-GP-transfected
Changs liver cells maintained in high-glucose (25 mM) medium supplemented with NAM+TSA. Hypoacetylated GP was immunoprecipitated from FlagGP-transfected cells cultured in glucose-free medium without NAM+TSA. The immunopurified GP was incubated with l-phosphatase (l) or phosphorylase
kinase (PhK) in appropriate buffers for 2 hr. Both phosphatase and kinase were removed by washing with PBS, and the treated GP was measured for enzyme
activity with (hatched bars) or without AMP (5 mM, solid bars) as indicated. Relative enzyme activity was normalized against GP protein. Ser-15 phosphorylation
and lysine acetylation levels were determined by western blotting.
(G) GP allosteric activation by AMP is not affected by acetylation. Flag-GP was purified from Changs liver cells and then treated for 1 or 2 hr with buffer or CobB as
indicated. Relative activity of purified GP was measured in the absence (solid bars) or presence of 5 mM AMP (hatched bars) and normalized by protein quantity.
Cell Metabolism
F
E
Figure 4. GP Acetylation and Phosphorylation Are Inversely Regulated by Physiological Stimuli

(AC) The effect of glucose, insulin, and glucagon on GP acetylation and phosphorylation. Flag-GP-expressing Changs cells were treated with different
conditions as indicated. Acetylation and serine phosphorylation of Flag-GP were determined by western blotting.
(D) Time course of glucagon on GP acetylation and phosphorylation. Flag-GP-expressing Changs liver cells were treated with glucagon (10 nM) for different times
as indicated. Levels of GP acetylation and Ser-15 phosphorylation were determined by western blotting.
(E) Fasting decreases GP acetylation and increases GP phosphorylation in mouse liver. Mice were fasted overnight before sacrifice, and GP proteins were
immunoprecipitated from the liver. Acetylation and Ser-15 phosphorylation levels of GP were determined by western blot (#p = 0.0021, n = 3; ##p = 0.0064, n = 3).
(F and G) Both glucose and insulin increase acetylation and decrease phosphorylation of GP in mouse liver. Glucose (1 g/kg) or insulin (5 U/kg) was intraperitoneally injected into overnight-fasted mice. At 30 min postinjection, mice were sacrificed and liver samples were harvested. GP proteins were immunoprecipitated from the liver, and acetylation and Ser-15 phosphorylation levels of GP proteins were determined by western blot (glucose, #p = 0.0191, n = 3, and
##
p = 0.0013, n = 3; insulin, #p = 0.0021, n = 3, and ##p = 0.0036, n = 3).
(H) Glucagon decreases acetylation and increases phosphorylation of GP in mouse liver. Glucagon (0.2 mg/kg) was intraperitoneally injected into fed mice, and
the mice were sacrificed at 30 min postinjection. GP protein was immunoprecipitated from liver, and acetylation and serine phosphorylation were determined by
western blotting (#p = 0.0200, n = 3; ##p = 0.0100, n = 3).
Cell Metabolism
Figure 5. Acetylation Increases GP-PP1a

Interaction
(A) Deacetylase inhibitor treatment enhances GP

binding to PP1a and decreases GP phosphorylation. Flag-GP was coexpressed with HA-PP1a in
Changs liver cells and treated with or without
NAM+TSA. GP coprecipitation with PP1a and
serine15 phosphorylation was determined by
western blot.
(B) Deacetylase inhibitor treatment has no affect
on the interaction between GPK2Q and PP1a. FlagGPK2Q was coexpressed with HA-PP1a in Changs
liver cells, and NAM+TSA treatment was indicated. GPK2Q coprecipitation with PP1a was
determined.
(C) GPK2Q binds more strongly to PP1a than wildtype GP. HA-PP1a was coexpressed with Flag-GP
or Flag-GPK2Q in Changs liver cells. PP1a-GP
and PP1a-GPK2Q interactions were detected
by coimmunoprecipitation. Phosphorylation of
GP and GPK2Q were determined by panphosphoserine antibody.
(D) Ectopic expression of SIRT2 decreases
PP1a-GP binding and increases GP serine phosphorylation. Flag-GP was expressed alone or
coexpressed with HA-PP1a or Myc-SIRT2 in
Changs liver cells as indicated; PP1a-GP interaction and GP serine phosphorylation level were
determined.
All error bars represent standard deviation (SD).
n = 3 for each experimental group.
We thus generated a GLDC5 mutant by deleting the C-terminal

five residue deletion of GL and confirmed that the deletion weakened the binding of GL with GP. Notably, inhibition of deacetylases increased the interaction between GLDC5 and GP to a level
similar to that of wild-type GP (Figure 6A). This result suggests
that the acetylation enhances the GP-GL interaction through
a site in GL that is independent of the C-terminal five residues
in GL. Consistently, when GPK2R was coexpressed with GLDC5
in Changs cells, virtually no interaction was detected between
them in either the presence or the absence of deacetylase inhibitors (Figure S6B). Meanwhile, we characterized another protein,
PTG (protein targeting to glycogen), which also functions as
a molecular scaffold targeting GP to PP1 (Brady et al., 1997),
and investigated whether acetylation regulates GP and PTG
interaction. The result demonstrated that GP-PTG binding could
also be stimulated by NAM and TSA treatment (Figure S6C).
To obtain direct evidence to support a role of acetylation in
promoting GP-GL interaction, we performed in vitro deacetylation and binding experiments. Hyperacetylated GP was purified
from cells treated with deacetylase inhibitors, and the purified
GP was deacetylated in vitro by incubating with the NAD+dependent bacterial deacetylase CobB. We found that CobB
treatment decreased GP acetylation and also reduced the
interaction of GP-GL in a NAD+-dependent manner (Figure 6B),
providing direct biochemical evidence that acetylation of
GP enhances its interaction with GL. We extended these
experiments in mouse liver. Plasmids encoding epitope-tagged
GP and GL were intravenously injected into mouse tail veins
to express the two proteins in liver. The interaction of the
ectopically expressed GP and GL in mouse liver was determined

in response to glucose signal. When glucose (1 g/kg) was injected intraperitoneally into fasted mice, we found that the
hepatic GP-GL interaction was increased significantly (Figure 6C). As expected, glucose injection indeed increased blood
glucose concentration (Figure S6D). These results further
support the notion that acetylation of GP increases GP-GL interaction in vivo.
We next investigated the importance of K470 and K796
acetylation sites in regulating the GP-GL interaction. HA-GL
with Flag-GPK470Q or GPK796Q was coexpressed in Changs
cells. As a positive control, deacetylase inhibitors increased
the interaction between wild-type GP and HA-GL (Figure 6D,
lanes 3 and 4). In contrast, a similar treatment did not increase
the interaction between Flag-GPK470Q and GL, and the mutant
GPK470Q displayed a stronger interaction with GL than the wildtype GP (Figure 6D, lanes 6 and 7). On the other hand, the
binding between GPK796Q and GL was increased by NAM+TSA
treatment, and GPK796Q showed a basal GL interaction similar
to the wild-type protein (Figure S6E). These results suggest
that the acetylation-enhanced interaction between GP and GL
is predominantly mediated by K470, but not K796. We performed iTRAQ mass spectrometry analyses to quantify GP
K470 acetylation. Our results showed that as much as 50% of
GP was acetylated at K470 in this assay and that inhibition of
deacetylases resulted in an increase in the ratio of acetylated
K470 versus unacetylated K470 from roughly 1:1 to 2:1 (Figure 6E, Figures S6FS6I), suggesting that a substantial fraction
of K470 in GP is acetylated in the cell.
Cell Metabolism
Figure 6. Acetylation
Interaction
Finally, to investigate whether the GP-GL interaction is regulated by physiological stimuli, we determined the interaction
between GL with either wild-type or K470Q mutant GP in
response to glucose concentration and insulin stimulation. We
found that the GP-GL interaction was increased by approximately 100% after switching from glucose-free medium to
25 mM glucose medium (Figure 6F). However, a similar glucose
switch did not affect the interaction between GPK470Q and GL.
Furthermore, insulin stimulated the interaction between the
wild-type GP, but not GPK470Q, and GL (Figure 6G). These results
support the notion that GP-GL interaction is regulated by K470
acetylation in response to nutrient and hormonal signals.
GP-GL
(A) Acetylation enhances GP-GL interaction. FlagGP was transfected into Changs liver cells either
alone or with HA-tagged GL or GLDC5. NAM+TSA
treatment was indicated. GP-GL and GP-GLDC5
interactions were determined by coimmunoprecipitation.
(B) Deacetylation of GP decreases GP-GL interaction in vitro. Flag-GP was expressed in HEK293
cells in the presence of deacetylase inhibitor,
and HA-GL was separately expressed. Both
proteins were affinity purified. Flag-GP was deacetylated with purified bacterial deacetylase
CobB in vitro as indicated. In vitro binding
between Flag-GP and HA-GL was performed.
Acetylation level of Flag-GP and the amount of
HA-GL coprecipitated by Flag-GP were determined by western blot.
(C) Glucose increases GP-GL interaction in vivo.
Human Myc-GP and HA-GL plasmids were intravenously injected into mice. The injected mice
were fasted overnight and then intraperitoneally
injected with glucose (1 g/kg). The mice were
sacrificed at 30 min postinjection. GP and GL
proteins expressed in the injected mouse liver
were immunoprecipitated by Myc beads for MycGP. The amount of GP and coprecipitated GL
proteins was measured, and the GL/GP ratio was
calculated.
(D) K470 is required for deacetylase inhibitors to
stimulate the GP-GL interaction. Flag-GP and
Flag-GPK470Q were transfected in Changs liver
cells, either alone or with HA-tagged GL. Cells
were treated with or without NAM+TSA. GP-GL
interactions were determined by coimmunoprecipitation.
(E) Quantification of GP K470 acetylation. Acetylation levels of K470 of GP expressed in Changs
liver cells treated with or without NAM+TSA were
quantified by iTRAQ.
(F and G) K470 in GP is required for regulation of
GP-GL interaction by glucose and insulin. Flag-GP
and Flag-GPK470Q were transfected in Changs
liver cells, either alone or with HA-tagged GL. Cells
were treated with or without glucose (F) or insulin
(G). GP-GL interactions were determined by
coimmunoprecipitation.
All error bars represent standard deviation (SD).
n = 34 for each experimental group.
Increases
DISCUSSION
Following the initial discovery of histone acetylation (Allfrey et al.,
1964; Phillips, 1963), extensive studies over the last four
decades have identified not only the enzymes that catalyze
reversible acetylation, the protein lysine acetyltransferases
(KATs, formerly termed histone acetyltransferases, HATs), and
deacetylases (commonly known as histone deacetylases, or
HDACs) but also many nonhistone substrates. Until relatively
recently, nearly all well-characterized acetylation substrates
were nuclear proteins, including transcription factors and coregulators (Yang and Seto, 2008). Recent studies, in particular
Cell Metabolism
Figure 7. Acetylation Negatively Regulates

GP Activity by Promoting the Binding of
Phosphatase and the Dephosphorylation
of GP
The acetylation of GP is stimulated by glucose and
insulin and inhibited by glucagon. The acetylation
enhances the binding of GL subunit to GP and
dephosphorylation of GP by PP1.
genetic studies of various strains of SIRT mutants (Finkel et al.,

2009) and proteomic studies of acetylomes in different organisms (Guan and Xiong, 2011; Kim and Yang, 2011), have revealed important and broad roles of acetylation in the regulation
of nontranscriptional processes, including especially the regulation of metabolic enzyme activity. In this study we have uncovered an important function of acetylation in the regulation of
GP, the key enzyme in glycogen breakdown. Our study has
gained new molecular insights into the regulation of this extensively investigated and historically significant enzyme.
Crosstalk between different protein PTMs, such as phosphorylation-targeted protein ubiquitylation (Hunter, 2007), plays an
important role in coordinating and connecting different cellular
processes. As metabolism needs to respond to a variety of intraand extracellular conditions such as cell growth signals and
nutrient availability, metabolic enzymes, which have now been
found to be frequently modified by acetylation, are expected to
be subject to such crosstalk regulation between acetylation
and other type of PTMs. We have demonstrated recently that
acetylation of phosphoenolpyruvate carboxykinase (PEPCK1),
a rate-limiting enzyme in gluconeogenesis, promotes its association with UBR5/EDD1 HECT E3 ligase and thus its degradation
in the presence of high glucose (Jiang et al., 2011). The current
study adds another distinct examplepromoting protein
dephosphorylation by facilitating the recruitment of a phosphataseof crosstalk between acetylation and phosphorylation.
Our results support a model in which acetylation of GP modulates its phosphorylation status by enhancing GPs binding
with the GL subunit of PP1 phosphatase, thereby stimulating
GP dephosphorylation by PP1 and inhibiting GP activity (Figure 7). Although it remains to be elucidated how the acetylation
of GP, mostly on K470, facilitates the binding of GL to GP, we
speculate that controlling protein-protein interaction by acetylation could be a broad mechanism for acetylation to regulate
activities of other metabolic enzymes or proteins.
Beside reversible phosphorylation, another remarkable
feature of GP regulation is its interaction with several allosteric
effectors, including glucose and AMP, and regulation by
hormones, such as insulin and glucagon, thereby allowing cells
to integrate different cellular conditions and energy status to
the regulation of GP. The finding that GP acetylation and activity
are controlled by glucose, insulin, and glucagon places acetylation downstream of these physiological stimuli. Binding of
glucose shifts GP in liver from a more relaxed and thus active
state (R) to a tense and less active state (T). As a result, low
glucose would lead to decreased glucose binding and higher
GP activity, resulting in an increased production of glucose
from glycogen. Under such low glucose conditions, GP is hypoacetylated in the cell and its interaction with GL (hence PP1) is
weakened, which in turn would increase GP Ser-15 phosphorylation and activity (Figure 7). Conversely, in cells with high
glucose concentration, GP is hyperacetylated and the acetylation enhances GPs binding with GL and thus PP1, leading to
decreased Ser-15 phopshorylation and activity of GP and eventually reduced glycogen degradation. Therefore, it appears that
high glucose can downregulate GP activity by two different
mechanisms: a conformational change by the direct binding of
glucose and acetylation-mediated dephosphorylation. It will be
important to determine whether these two regulations operate
separately or sequentially, and whether GP conformational
change brought by the initial glucose binding facilitates the
acetylation.
The role of acetylation in recruiting PP1 is clearly consistent
with the inverse correlation between acetylation and phosphorylation of GP. Glucagon can induce a direct signaling pathway to
phosphorylate and activate GP in a manner independent of
acetylation. However, we speculate that the decrease of GP
acetylation induced by glucagon may contribute to the magnitude and duration of GP activation in response to glucagon.
Glucagon stimulates GP by activating the GP kinase via the
classical phosphorylation cascade and also by dissociating GP
phosphatase via acetylation (Figure S7). It should be noted that
acetylation appears to have no direct role in GP allosteric activation by AMP, although AMP cannot further activate GP when GP
is fully active. Although the precise mechanism by which glucose
regulates GP acetylation remains to be elucidated, using
acetylation machinery to control GP adds another layer of
regulation and thus lends cells further versatility in integrating
multifaceted signaling pathways and nutrient conditions to this
enzyme that is not only central to the glycogen metabolism,
but is also interlocked with multiple energy metabolic pathways.
Mouse Liver Collection and Primary Hepatocytes
and Muscle Cell Isolation
Male BALB/c mice (46 weeks old, 2025 g) were divided into two groups, fed,
and overnight fasted. Fasting started from late afternoon and lasted for 16 hr
before experiments. The fasted mice were further subdivided into glucoseand insulin-treated groups, and the fed mice were treated with glucagon.
Mice blood glucose levels were measured by Accu-Chek Active Blood
Glucose Meter (Roche). At 30 min postinjection, mice were sacrificed and liver
samples were harvested. In addition, mouse primary hepatocytes and muscle
cells were isolated. Please refer to the Supplemental Experimental Procedures
for detailed information. Animal experiments were performed at Fudan Animal
Center in accordance with the animal welfare guidelines.
Glycogen Content Measurement
Cells were washed twice and then lysed with supersonic. The cell lysate in 1 ml
PBS (pH 4.8) was heated at a boiling point for 10 min to liberate stored
Cell Metabolism
glycogen and to inactivate enzymes, which may produce extra glucose. After
centrifugation for 15 min at 13,000 rpm, 4U amyloglucosidase (Sigma) was
added into the supernatant. The resulting mixture was incubated for 1 hr at
50 C and followed by boiling for 10 min at 99 C. The cell lysate without
amyloglucosidase was included as a control. Subsequently, the glycogen
content was colorimetrically measured using a glucose assay kit (GAGO20,
Sigma). The mixtures were incubated for 30 min at 37 C, and absorbance at
540 nm was measured by using a UV/Visible spectrophotometer reader
(Ultrospec 3100 pro, Amersham Biosciences).
iTRAQ Quantification
Flag-GP proteins were expressed in Changs liver cells and quantified by
iTRAQ following the method modified according to Zhao (Zhao et al., 2010).
GP was immunopurified from cells untreated or treated with NAM and TSA,
resolved on 10% SDS-PAGE, and stained by Coomassie blue and sliced.
The dye of gel slice was removed by soaking with 50 mM NH4HCO3 and
50% acetonitrile, followed by water wash twice and removing water by
acetonitrile. The gel was dried and digested in 100 ml 50mM NH4HCO3 with
trypsin (trypsin:protein at 1:30) at 37 C overnight. The trypsin-treated peptides
were extracted by a volume containing 50% acetonitrile and 0.1% trifluoroacetic acid (TFA) and then followed by vacuum dry. Standard control peptides
and GP samples were separately labeled with different iTRAQ-labeling
reagents (ABI) as indicated in Table S1 and then subjected to LTQ-OrbiTrap
MS analysis. Quantification of peptides was calculated by comparing relative
intensity of the iTRAQ tags.
Cell Treatment
TSA (0.5 mM) and NAM (5 mM) were added to the culture medium 18 and 6 hr
before cell harvest, respectively. Glucose-free medium was prepared with
DMEM base (GIBCO, #11966) and supplemented with glucose (Sigma), insulin
(Sigma), and glucagon (Sigma) of different concentrations as indicated.
Glucose, insulin, and glucagon treatments were carried out by culturing cells
in DMEM medium for 24 hr before the desired medium was used to replace
DMEM medium.
Glycogen Phosphorylase Activity Assay and CobB Treatment
Flag-tagged proteins were expressed in Changs liver cells, eluted by Flag
peptides (Gilson Biochemical), and measured using the method of Jones
and Wright (Jones and Wright, 1970). The GP activity assay consists of
50 mM sodium glycerol-phosphate (pH 7.1), 10 mM potassium phosphate,
5 mM MgCl2, 0.5 mM NAD+, 1 mM DTT, 1.6 unit phosphoglucomutase, 1.6
unit glucose-6-phosphate dehydrogenase, and 0.2% glycogen in a total
volume of 0.3 ml. The reaction was started by adding GP into the volume
and assayed at 25 C. The reaction was monitored by measuring the increase
of fluorescence (excitation 350 nm, emission 470 nm, HITACHI F-4,600 fluorescence spectrophotometer) for NADH generation.
CobB deacetylation treatment was performed by using a method modified
elsewhere (Hallows et al., 2006). CobB was expressed in E. coli, purified
with nickel beads, and stored at 80 C in 10% glycerol. The CobB deacetylation
assay buffer consists of 40 mM HEPES (pH 7.0), 6 mM MgCl2, 1 mM NAD+, and
n1 mM DTT in a total volume of 0.1 ml. The reaction was started by adding
10 mg CobB and GP into the volume and was assayed at 37 C for 1 hr.
l-Phosphatase and Phosphorylase Kinase Treatment
Flag-GP was ectopically expressed in Changs liver cells and immunoprecipitated by Flag beads. Subsequently, the bead-linked GP was washed by
PBS (pH 7.4) three times before being treated with phosphatase and kinase
(PhK). The l-phosphatase assay consists of NEBuffer Pack for Protein
MetalloPhosphatases (50 mM HEPES, 100 mM NaCl, 2 mM DTT, 0.01% Brij
35 [pH 7.5]) and 1 mM MnCl2 in a total volume of 0.4 ml. The reaction was
started by adding 50 units lambda protein phosphatase (#P0753S, NEB) and
GP into the volume and assayed at 30 C for 2 hr by shaking. The PhK reaction
assay consists of 41 mM glycerophosphate, 41 mM Tris, 0.2 mM CaCl2, 3 mM
ATP, MgCl2 (pH 8.2). Reaction was started by adding 2U PhK (#P2014, Sigma)
and GP into the volume and assayed at 30 C for 2 hr by shaking. The reaction
was terminated by washing the beads three times by PBS and eluting by Flag
peptide, and the activity was determined by fluorescence spectrophotometer
(Cohen, 1973; Shenolikar et al., 1979).
In Vitro Binding
Flag-GP purified by Flag beads was subject to in vitro deacetylation by CobB
before it was mixed with purified HA-GL. The binding was allowed in 4 C for
4 hr before the beads were washed three times by PBS. Proteins on beads
were denatured by SDS loading buffer and detected by western blot.
Statistics were performed with a two-tailed unpaired Students t test. All data
shown represent the results obtained from triplicated independent experiments with standard deviations (mean SD). The values of p < 0.05 were
considered statistically significant.
Supplemental Information includes seven figures, one table, Supplemental
Experimental Procedures, and Supplemental References and can be found
with this article online at doi:10.1016/j.cmet.2011.12.005.
ACKNOWLEDGMENTS
We thank the members of the Fudan MCB laboratory for discussions
throughout this study. We appreciate Dr. P.T. Cohen for providing the GP
Ser-15 phospho-antibody. This work was supported by the 985 Program, 973
Program (grant numbers 2009CB918401, 2011CB910600, 2012CB910101,
and 2012CB910103), NSFC (grant numbers 30971485/C0706, 31030042,
31071192), Shanghai key project (grant numbers 09JC1402300 and
11JC1401100), the Shanghai Leading Academic Discipline Project (project
number B110), and National Institutes of Health (NIH) grants to Y.X. and K.-L.G.
Received: January 24, 2011
Revised: July 7, 2011
REFERENCES
Alemany, S., and Cohen, P. (1986). Phosphorylase a is an allosteric inhibitor of
the glycogen and microsomal forms of rat hepatic protein phosphatase-1.
FEBS Lett. 198, 194202.
Allfrey, V.G., Faulkner, R., and Mirsky, A.E. (1964). Acetylation and methylation
of histones and their possible role in the regulation of RNA synthesis. Proc.
Natl. Acad. Sci. USA 51, 786794.
Andreu, A.L., Nogales-Gadea, G., Cassandrini, D., Arenas, J., and Bruno, C.
(2007). McArdle disease: molecular genetic update. Acta Myol. 26, 5357.
Armstrong, C.G., Doherty, M.J., and Cohen, P.T. (1998). Identification of
the separate domains in the hepatic glycogen-targeting subunit of protein
phosphatase 1 that interact with phosphorylase a, glycogen and protein phosphatase 1. Biochem. J. 336, 699704.
Barford, D., and Johnson, L.N. (1989). The allosteric transition of glycogen
phosphorylase. Nature 340, 609616.
Barford, D., Hu, S.H., and Johnson, L.N. (1991). Structural mechanism for
glycogen phosphorylase control by phosphorylation and AMP. J. Mol. Biol.
218, 233260.
Brady, M.J., Printen, J.A., Mastick, C.C., and Saltiel, A.R. (1997). Role of
protein targeting to glycogen (PTG) in the regulation of protein phosphatase1 activity. J. Biol. Chem. 272, 2019820204.
Browner, M.F., and Fletterick, R.J. (1992). Phosphorylase: a biological transducer. Trends Biochem. Sci. 17, 6671.
Buchbinder, J.L., Luong, C.B., Browner, M.F., and Fletterick, R.J. (1997).
Partial activation of muscle phosphorylase by replacement of serine 14 with
acidic residues at the site of regulatory phosphorylation. Biochemistry 36,
80398044.
Choudhary, C., Kumar, C., Gnad, F., Nielsen, M.L., Rehman, M., Walther, T.C.,
Olsen, J.V., and Mann, M. (2009). Lysine acetylation targets protein complexes
and co-regulates major cellular functions. Science 325, 834840.
Cell Metabolism
Cohen, P. (1973). The subunit structure of rabbit-skeletal-muscle phosphorylase kinase, and the molecular basis of its activation reactions. Eur. J.
Biochem. 34, 114.
Cori, C.F., and Cori, G.T. (1936). Mechansim of formation of hexosemonophosphate in muscle and isolation of a new phosphate ester. Proc. Soc.
Exp. Biol. Med. 34, 702705.
Finkel, T., Deng, C.X., and Mostoslavsky, R. (2009). Recent progress in the
biology and physiology of sirtuins. Nature 460, 587591.
Fischer, E.H., and Krebs, E.G. (1955). Conversion of phosphorylase b to phosphorylase a in muscle extracts. J. Biol. Chem. 216, 121132.
Guan, K.L., and Xiong, Y. (2011). Regulation of intermediary metabolism by
protein acetylation. Trends Biochem. Sci. 36, 108116.
Hallows, W.C., Lee, S., and Denu, J.M. (2006). Sirtuins deacetylate and
activate mammalian acetyl-CoA synthetases. Proc. Natl. Acad. Sci. USA
103, 1023010235.
Hunter, T. (2007). The age of crosstalk: phosphorylation, ubiquitination, and
beyond. Mol. Cell 28, 730738.
Jiang, W., Wang, S., Xiao, M., Lin, Y., Zhou, L., Lei, Q., Xiong, Y., Guan, K.-L.,
and Zhao, S. (2011). Acetylation regulates gluconeogenesis by promoting
PEPCK1 degradation via recruiting the UBR5 ubiquitin ligase. Mol. Cell 43,
3344.
Johnson, L.N. (1992). Glycogen phosphorylase: control by phosphorylation
and allosteric effectors. FASEB J. 6, 22742282.
The human glucagon receptor encoding gene: structure, cDNA sequence

and chromosomal localization. Gene 140, 203209.
Massague, J., and Guinovart, J.J. (1977). Insulin control of rat hepatocyte
glycogen synthase and phosphorylase in the absence of glucose. FEBS
Lett. 82, 317320.
Nolan, C., Novoa, W.B., Krebs, E.G., and Fischer, E.H. (1964). Further studies
on the site phosphorylated in the Phosphorylase B to a reaction. Biochemistry
3, 542551.
Pautsch, A., Stadler, N., Wissdorf, O., Langkopf, E., Moreth, W., and Streicher,
R. (2008). Molecular recognition of the protein phosphatase 1 glycogen targeting subunit by glycogen phosphorylase. J. Biol. Chem. 283, 89138918.
Phillips, D.M. (1963). The presence of acetyl groups of histones. Biochem. J.
87, 258263.
Shenolikar, S., Cohen, P.T.W., Cohen, P., Nairn, A.C., and Perry, S.V. (1979).
The role of calmodulin in the structure and regulation of phosphorylase kinase
from rabbit skeletal muscle. Eur. J. Biochem. 100, 329337.
Stegelmeier, B.L., Molyneux, R.J., Elbein, A.D., and James, L.F. (1995). The
lesions of locoweed (Astragalus mollissimus), swainsonine, and castanospermine in rats. Vet. Pathol. 32, 289298.
Sutherland, E.W., Jr., and Wosilait, W.D. (1955). Inactivation and activation of
liver phosphorylase. Nature 175, 169170.
Jones, T.H., and Wright, B.E. (1970). Partial purification and characterization of
glycogen phosphorylase from Dictyostelium discoideum. J. Bacteriol. 104,
754761.
Tang, N.L., Hui, J., Young, E., Worthington, V., To, K.F., Cheung, K.L., Li, C.K.,
and Fok, T.F. (2003). A novel mutation (G233D) in the glycogen phosphorylase
gene in a patient with hepatic glycogen storage disease and residual enzyme
activity. Mol. Genet. Metab. 79, 142145.
Kelsall, I.R., Munro, S., Hallyburton, I., Treadway, J.L., and Cohen, P.T. (2007).
The hepatic PP1 glycogen-targeting subunit interaction with phosphorylase
a can be blocked by C-terminal tyrosine deletion or an indole drug. FEBS
Lett. 581, 47494753.
Wang, Q., Zhang, Y., Yang, C., Xiong, H., Lin, Y., Yao, J., Li, H., Xie, L., Zhao,
W., Yao, Y., et al. (2010). Acetylation of metabolic enzymes coordinates carbon
source utilization and metabolic flux. Science 327, 10041007.
Kim, G.W., and Yang, X.J. (2011). Comprehensive lysine acetylomes emerging
from bacteria to humans. Trends Biochem. Sci. 36, 211220.
Xu, W.S., Parmigiani, R.B., and Marks, P.A. (2007). Histone deacetylase inhibitors: molecular mechanisms of action. Oncogene 26, 55415552.
Kim, S.C., Sprung, R., Chen, Y., Xu, Y., Ball, H., Pei, J., Cheng, T., Kho, Y.,
Xiao, H., Xiao, L., et al. (2006). Substrate and functional diversity of lysine acetylation revealed by a proteomics survey. Mol. Cell 23, 607618.
Yang, X.J., and Seto, E. (2008). Lysine acetylation: codified crosstalk with
other posttranslational modifications. Mol. Cell 31, 449461.
Lok, S., Kuijper, J.L., Jelinek, L.J., Kramer, J.M., Whitmore, T.E., Sprecher,
C.A., Mathewes, S., Grant, F.J., Biggs, S.H., Rosenberg, G.B., et al. (1994).
Zhao, S., Xu, W., Jiang, W., Yu, W., Lin, Y., Zhang, T., Yao, J., Zhou, L., Zeng,
Y., Li, H., et al. (2010). Regulation of cellular metabolism by protein lysine
acetylation. Science 327, 10001004.
Cell Metabolism
Article
Lysosome-Related Organelles in Intestinal Cells
Are a Zinc Storage Site in C. elegans
Hyun Cheol Roh,1 Sara Collier,1 James Guthrie,2 J. David Robertson,2,3 and Kerry Kornfeld1,*
1Department
of Developmental Biology, Washington University School of Medicine, St. Louis, MO 63110, USA
Reactor Center
3Department of Chemistry
University of Missouri, Columbia, MO 65211, USA
*Correspondence: kornfeld@wustl.edu
DOI 10.1016/j.cmet.2011.12.003
2Research
SUMMARY
Zinc is an essential trace element involved in many

biological processes and human diseases. Because
zinc deficiency and excess are deleterious, animals
require homeostatic mechanisms to maintain zinc
levels in response to dietary fluctuations. Here, we
demonstrate that lysosome-related organelles in
intestinal cells of C. elegans, called gut granules,
function as the major site of zinc storage. Zinc
storage in gut granules promotes detoxification and
subsequent mobilization, linking cellular and organismal zinc metabolism. The cation diffusion facilitator
protein CDF-2 plays a critical role in this process by
transporting zinc into gut granules. In response to
high dietary zinc, gut granules displayed structural
changes characterized by a bilobed morphology
with asymmetric distributions of zinc and molecular
markers. We defined a genetic pathway that mediates the formation of bilobed morphology. These
findings elucidate mechanisms of zinc storage,
detoxification, and mobilization in C. elegans and
may be relevant to other animals.
INTRODUCTION
Zinc is a nutrient that is essential for all life. Zinc has roles in
many biological processes; protein-bound zinc contributes
to enzymatic activity and protein structure, and labile zinc
functions in signal transduction (Murakami and Hirano, 2008;
Vallee and Falchuk, 1993). Zinc is important for human health,
since zinc deficiency causes a broad range of defects in
multiple organ systems including skin, immune, skeletal, and
reproductive (Hambidge, 2000). Zinc deficiency is associated
with genetic diseases caused by mutations of zinc transporters, such as acrodermatitis enteropathica and inadequate
dietary intake, which is a major world-wide problem. Excess
zinc is also deleterious, since it may displace other trace metals
or bind low affinity sites, leading to protein dysfunction (Fosmire, 1990). Therefore, organisms require homeostatic mechanisms to control the levels and distribution of this essential
metal.
Zinc metabolism in animals is regulated at the organismal

and cellular levels. In vertebrates, the gastrointestinal tract
mediates zinc absorption, and absorbed zinc is distributed by
the circulatory system (King, 2011). The gastrointestinal tract
also plays a major role in zinc excretion, with smaller contributions from other tissues, including the kidney and pancreas. At
the cellular level, zinc is partitioned between the cytoplasm
and the lumen of intracellular organelles, and it can be labile or
protein bound (Eide, 2006). Two families of zinc transporters
play critical roles in these processes: cation diffusion facilitator
(CDF/ZnT/SLC30) and Zrt-like, Irt-like protein (ZIP/SLC39)
(Cragg et al., 2005; Feeney et al., 2005). CDF proteins decrease
cytoplasmic levels by transporting zinc across the plasma
membrane or into intracellular organelles, whereas ZIP proteins
increase cytoplasmic levels by transporting zinc in the opposite
direction. Mammals contain 10 CDF and 14 ZIP proteins that
have specific tissue distributions and intracellular localizations
(Lichten and Cousins, 2009). Thus, a network of zinc transporters
is important for zinc metabolism in animals.
Mechanisms of zinc storage and mobilization have been
characterized in the yeast Saccharomyces cerevisiae (Eide,
2009). When zinc is abundant, excess zinc is stored in the
vacuole by the CDF proteins Cot1 and Zrc1. In response to
zinc deficiency, the ZIP protein Zrt3 mobilizes stored zinc. Zinc
accumulation has been demonstrated in mammalian cells. In
response to high levels of zinc in the culture media, labile zinc
accumulates in endosomal or lysosomal organelles, termed
zincosomes (Haase and Beyersmann, 1999; Palmiter et al.,
1996). In higher animals such as birds, a high zinc diet causes
organismal zinc accumulation and promotes resistance to a
subsequent dietary zinc deficiency (Emmert and Baker, 1995).
Humans appear to have only a limited capacity for zinc storage
and mobilization, since symptoms of zinc deficiency develop
rapidly in response to dietary deficiency (King, 2011). An important question is how zinc storage at the cellular level contributes
to the response to zinc deficiency at the organismal level in
animals.
The nematode C. elegans is a useful model organism for the
study of metal biology, including iron and heme metabolism,
metal toxicity, and zinc signaling (Bruinsma et al., 2002; Gourley
et al., 2003; Liao and Freedman, 1998; Rajagopal et al., 2008;
Vatamaniuk et al., 2001; Yoder et al., 2004). To study zinc
metabolism, we developed methods to manipulate dietary zinc
and used forward and reverse genetic approaches to identify
genes important for zinc metabolism (Bruinsma et al., 2008;
Cell Metabolism
Lysosome-Related Organelles Store Zinc in Worms
Figure 1. Zinc Is Stored in Gut Granules

(A) Fluorescence images of live wild-type
hermaphrodites cultured with FluoZin-3 and the
indicated levels of supplemental zinc and TPEN.
Panels display the anterior half of the intestine of
a single animal with pharynx to the left and tail to
the right. Scale bar: 50 mm.
(B) Quantification of fluorescence images like
those shown in (A) using ImageJ software. The
fluorescence intensity (shown in arbitrary units,
A.U.) was normalized by setting the value at 0 mM
supplemental Zn equal to 1.0. Bars indicate mean
values SEM (n = 15) (**p < 0.001; ***p < 0.0001).
(C) Fluorescence images of live wild-type animals
costained with FluoZin-3 (green) and LysoTracker
(red). Boxed regions are magnified in the right
panels. Animals cultured with 100 mM supplemental zinc displayed bilobed gut granules with
asymmetric staining; one side was strongly positive for FluoZin-3 (arrow), and the other side was
strongly positive for LysoTracker (arrowhead).
Scale bars: 10 mm and 2 mm (boxed regions) (see
also Figures S1 and S2).
Davis et al., 2009; Murphy et al., 2011). C. elegans contains

highly conserved families of CDF, ZIP, and metallothionein
genes, suggesting that fundamental mechanisms of zinc
metabolism may be similar to other animals. In response to a
high zinc diet, C. elegans accumulates zinc (Davis et al., 2009).
To characterize zinc storage and mobilization, we developed
methods to visualize zinc in C. elegans using a zinc-specific
fluorescent dye. We determined that zinc is stored in lysosome-related organelles in intestinal cells, and these organelles
undergo a transition to a bilobed morphology in response to
high dietary zinc. We demonstrated that zinc accumulation in
these organelles, which is mediated by the activity of the
CDF-2 zinc transporter, plays a critical role in zinc detoxification
and mobilization in the physiological setting of an intact animal.
RESULTS
Gut Granules Contain Labile Zinc
By using inductively coupled plasma-mass spectrometry (ICPMS) to measure total zinc content in worm extracts, we demonstrated that C. elegans cultured with high dietary zinc display
elevated levels of total zinc, suggesting excess zinc is stored
(Davis et al., 2009). To identify the site

of zinc storage, we used zinc-specific
fluorescent dyes to visualize zinc. We
conducted pilot studies with several
dyes and selected FluoZin-3 based on
its high zinc sensitivity and specificity
(Gee et al., 2002). Hermaphrodites were
cultured on noble agar minimal medium
(NAMM) dishes (Bruinsma et al., 2008)
containing FluoZin-3, and live animals
were analyzed by fluorescence microscopy. Wild-type animals cultured without
supplemental zinc displayed green fluorescence in vesicles in intestinal cells
(Figure 1A). FluoZin-3 fluorescence intensity displayed significant, dose-dependent enhancement and diminishment in
worms cultured with supplemental zinc and the zinc chelator,
N,N,N0 ,N0 -tetrakis (2-pyridylmethyl) ethylenediamine (TPEN),
respectively (Figures 1A and 1B). These results indicate that
FluoZin-3 monitors labile zinc in live worms, and zinc is concentrated in vesicles of intestinal cells.
Gut granules in intestinal cells have been classified as lysosome-related organelles based on the presence of lysosomal
proteins and staining with lysosome-specific fluorescent dyes
such as LysoTracker (Clokey and Jacobson, 1986; Hermann
et al., 2005; Kostich et al., 2000). To investigate the relationship
between FluoZin-3 fluorescent vesicles and gut granules, we
performed costaining experiments using LysoTracker. With no
supplemental zinc, the patterns of FluoZin-3 and LysoTracker
fluorescence were highly overlapping in intestinal cells (Figures
1C and S1), indicating that zinc detected by FluoZin-3 is stored
in gut granules. Because gut granules contain birefringent and
autofluorescent materials, we determined how autofluorescence
compares to FluoZin-3 fluorescence by comparing control
animals cultured with no dye to animals cultured with FluoZin-3.
Animals cultured with FluoZin-3 displayed 2.6-fold, 3.5-fold,
Cell Metabolism
FluoZin-3
A
WT
pgp-2
glo-1
glo-3
0
100
Supplemental Zn (M)
Zn content (ppm)
300
250
WT
pgp-2
glo-1
cdf-2
200
150
100
50
0
0
200
Supplemental Zn (M)
Figure 2. Gut Granules Are the Major Site of Zinc Storage
(A) Fluorescence images of live wild-type, pgp-2(kx48), glo-1(zu391), and
glo-3(zu446) animals cultured with FluoZin-3 and the indicated levels of
supplemental zinc. Images show the intestine with pharynx to the left and tail
to the right. Scale bar: 50 mm.
(B) Total zinc content of wild-type, pgp-2(kx48), glo-1(zu391), and cdf-2(tm788)
animals. Populations of animals consisting of a mixture of developmental
stages were cultured on NAMM dishes with the indicated levels of supplemental zinc. Total zinc content was determined by ICP-MS and calculated in
parts per million (ppm). Bars indicate mean values SEM of two independent
experiments (see also Figure S3).
and 4.3-fold higher signal than control animals when cultured

with 0 mM, 100 mM, and 200 mM supplemental zinc, respectively
(Figure S2). These results indicate that the signal is primarily
due to FluoZin-3 binding zinc with a minor contribution from
autofluorescence.
Gut Granules Are the Major Site of Zinc Storage
To characterize the function of gut granules in zinc storage,
we analyzed Glo mutant animals that have reduced numbers
of gut granules due to defects in lysosome biogenesis. We
analyzed three genes, pgp-2, glo-1, and glo-3, because wellcharacterized loss-of-function mutations in these genes cause
Glo phenotypes of different severities; wild-type animals contain

hundreds of gut granules, whereas pgp-2(kx49) animals contain
10100 gut granules, and glo-1(zu391) and glo-3(zu446) animals
contain fewer than 10 gut granules (Hermann et al., 2005; Rabbitts et al., 2008; Schroeder et al., 2007). pgp-2 encodes an
ABC transporter that localizes to the membrane of gut granules,
glo-1 encodes a predicted Rab GTPase that localizes to gut
granules, and glo-3 has not been molecularly identified. All three
mutant strains displayed reduced FluoZin-3 fluorescence
compared to wild-type (Figure 2A). Consistent with the severity
of the Glo phenotype, pgp-2 mutant animals displayed an
intermediate number of FluoZin-3-positive granules, and glo-1
and glo-3 mutant animals displayed very few positive granules.
These results indicate that gut granules are the site of labile
zinc detected by FluoZin-3.
To quantify zinc storage defects in Glo animals, we measured
total zinc content using the independent method of ICP-MS.
pgp-2 mutant animals displayed a moderate reduction of total
zinc content, and glo-1 mutant animals displayed a severe
reduction of total zinc content compared to wild-type animals
(Figure 2B). The total zinc content of the Glo animals was well
correlated with the severity of the Glo phenotype. glo-1 mutant
animals cultured in high dietary zinc contained approximately
50% of the total zinc content of wild-type animals. These results
indicate that gut granules are the major site of zinc storage in
C. elegans that contains about half of the total zinc in the body.
By contrast, Glo animals did not display a consistent change
in total levels of magnesium, iron, manganese, or copper
compared to wild-type animals (Figure S3). These results indicate that gut granules are specifically involved in zinc storage.
CDF-2 Promotes Zinc Storage in Gut Granules
The cation diffusion facilitator protein CDF-2 is expressed
specifically in intestinal cells and localized to autofluorescent
vesicles (Davis et al., 2009). To elucidate the relationship
between CDF-2 and zinc storage, we cultured transgenic
animals expressing CDF-2::mCherry with FluoZin-3. In animals
cultured with no supplemental zinc, FluoZin-3 and CDF2::mCherry fluorescence overlapped almost completely (Figure S4, left), indicating that CDF-2 is localized to the gut granules
that concentrate zinc. cdf-2 mRNA levels are increased by high
dietary zinc (Davis et al., 2009). To analyze the regulation of
CDF-2 protein expression by dietary zinc, we used transgenic
animals expressing CDF-2::GFP. The level of CDF-2::GFP
was induced in a concentration-dependent manner by approximately 3-fold and 4-fold at 100 mM and 200 mM supplemental
zinc, respectively, compared to 0 mM supplemental zinc (Figures
3A and 3B). These results suggest that high levels of CDF-2
play an important role in the response to high dietary zinc.
To determine the function of CDF-2 in zinc storage in gut
granules, we analyzed animals with the cdf-2(tm788) deletion
mutation that causes a strong loss-of-function. cdf-2(tm788)
mutant animals displayed significantly lower FluoZin-3 fluorescence at both 0 mM and 100 mM supplemental zinc compared
to wild-type animals (Figures 3C and 3D). ICP-MS analysis
revealed that cdf-2(tm788) mutant animals had a large reduction
of total zinc content, similar to glo-1 mutant animals (Figure 2B),
consistent with our previous studies (Davis et al., 2009). These
results indicate that CDF-2 is necessary to concentrate zinc
Cell Metabolism
Figure 3. CDF-2 Functions Cell-Autonomously to Promote Zinc Storage in Gut

Granules
(A) Fluorescence microscope images of live
transgenic animals containing an integrated array,
amIs4, expressing CDF-2::GFP, and the cdf2(tm788) mutation. L4 stage hermaphrodites were
cultured with the indicated levels of supplemental
zinc. Each panel displays one representative
animal oriented with pharynx to the left and tail to
the right. Scale bar: 100 mm.
(B) Quantification of fluorescence images like
those shown in (A). The fluorescence intensity
(shown in arbitrary units, A.U.) was normalized by
setting the value at 0 mM supplemental zinc equal
to 1.0. Bars indicate mean values SEM (n = 15)
(***p < 0.0001).
(C) Fluorescence microscope images of wild-type,
cdf-2(tm788), and transgenic cdf-2(tm788) animals containing a multicopy extrachromosomal
array; amEx132, expressing CDF-2::mCherry.
Images show the intestine (pharynx to the left, tail
to the right) of live animals cultured with FluoZin-3
and the indicated levels of supplemental zinc.
Scale bar: 50 mm.
(D) Quantification of fluorescence images like
those shown in (C). The fluorescence intensity
(shown in arbitrary units, A.U.) was normalized by
setting the value of wild-type animals at 0 mM
supplemental Zn equal to 1.0. Bars indicate mean
values SEM (n = 15) (*p < 0.01; **p < 0.001;
***p < 0.0001).
(E) Images of a live cdf-2(tm788);amEx132 animal
that displayed mosaic expression of the CDF2::mCherry. The animal was cultured with FluoZin3 and no supplemental zinc, and the intestine was
imaged by fluorescence microscopy (pharynx to
the left, tail to the right). The intestinal cell lacking
CDF-2::mCherry expression is marked with a
star (*). Scale bar: 50 mm (see also Figure S4).
into gut granules. cdf-2(tm788) mutant animals displayed slightly

increased FluoZin-3 fluorescence at 100 mM compared to 0 mM
supplemental zinc, suggesting that a CDF-2 independent mechanism might also promote zinc accumulation in gut granules.
One possible mechanism is an alternative zinc transporter that
might be induced in cdf-2 mutant animals. ttm-1 encodes a predicted CDF protein that is highly related to CDF-2 (H.C.R. and
K.K., unpublished data). An analysis of ttm-1 transcripts using
qRT-PCR showed a small increase in ttm-1 transcript levels in
cdf-2 mutant animals compared to wild-type animals (data not
shown). Transgenic animals that contain multicopy, extrachromosomal arrays expressing CDF-2::mCherry displayed higher
FluoZin-3 fluorescence compared to wild-type animals (Figures
3C and 3D). Thus, overexpression of CDF-2 was sufficient to
concentrate zinc in gut granules.
Transgenic animals containing extrachromosomal arrays
display mosaic expression of transgenes spontaneously and at
a low frequency. To determine whether CDF-2 functions cellautonomously, we analyzed mosaic animals that lack transgene
expression in specific intestinal cells. Because these animals
contain the cdf-2(tm788) mutation, an intestinal cell that lacks
transgene expression lacks all CDF-2 function. The intestinal
cells lacking CDF-2::mCherry expression displayed lower Fluo-
Zin-3 fluorescence compared to the flanking cells expressing

CDF-2::mCherry (Figure 3E). These results indicate that CDF-2
functions cell-autonomously in intestinal cells to promote zinc
concentration in gut granules.
High Dietary Zinc Alters Gut Granule Morphology
To characterize how the intracellular localization of CDF-2
responds to high dietary zinc, we examined the colocalization
of CDF-2 and LysoTracker. With no supplemental zinc, CDF2::GFP colocalized completely with LysoTracker (Figure 4A,
left), demonstrating directly that CDF-2 localizes to the
membrane of gut granules. In the presence of 200 mM supplemental zinc, CDF-2::GFP expression remained restricted to gut
granules, but gut granules displayed altered morphology.
Many vesicles had a bilobed appearance, and the two lobes displayed distinct staining patterns; one lobe was positive for both
CDF-2 and LysoTracker, whereas the other lobe was positive for
CDF-2 and negative for LysoTracker (Figure 4A, right). A dose
response analysis showed that bilobed granules were induced
by 100 mM and 200 mM supplemental zinc in worms cultured
on NAMM dishes. The phenotype was highly penetrant, since
bilobed granules were observed in nearly every animal, but
only a subset of gut granules display a bilobed morphology.
Cell Metabolism
Figure 4. High Zinc Induces the Formation of

Asymmetric Bilobed Gut Granules
(A) Fluorescence images of live transgenic animals expressing CDF-2::GFP cultured with LysoTracker and the
indicated levels of supplemental zinc. The differential
interference contrast (DIC) images show the intestinal
lumen (triangle) and adjacent intestinal cells with pharynx
to the left and tail to the right. Boxed regions are magnified
in the right panels. With 200 mM supplemental zinc, many
gut granules appear to be bilobed and asymmetric; one
side is positive for CDF-2::GFP and LysoTracker (arrowhead), whereas the other side is positive for CDF-2::GFP
and negative for LysoTracker (arrow). Scale bars: 10 mm
and 2 mm (boxed regions) (see also Figure S5).
(B and C) Confocal microscope images of live transgenic
animals expressing LMP-1::GFP cultured with LysoTracker (B) or expressing CDF-2::mCherry (C) cultured
with the indicated levels of supplemental zinc. Images
show intestinal cells with pharynx to the left and tail to the
right. Insets are magnified images of the boxed regions.
Scale bar: 10 mm and 2 mm (insets) (see also Figure S6).
Since LysoTracker labels acidic organelles, these results indicate that one lobe may lack the determinants that concentrate
the LysoTracker or that the LysoTracker dye may be present
but not fluorescent due to high pH. To examine a possible pH
difference, we used LysoSensor Green DND-153, which is highly
fluorescent in neutral compartments. Colocalization analysis
with CDF-2::mCherry demonstrated that LysoSensor Green
DND-153 was fluorescent only in one lobe of the bilobed gut
granules (Figure S5), similar to LysoTracker staining. These
results suggest that asymmetric staining of bilobed vesicles
may reflect an asymmetric distribution of the molecules that
bind these fluorescent probes.
To characterize additional differences between the two sides
of bilobed gut granules, we first investigated the distribution of
zinc using FluoZin-3. With 100 mM supplemental zinc, FluoZin-3
displayed asymmetric staining in bilobed gut granules; the
LysoTracker-positive lobe displayed weak FluoZin-3 fluorescence, whereas the LysoTrackernegative lobe displayed strong FluoZin-3
fluorescence (Figure 1C, right). To examine the
relationship between zinc levels and CDF-2
protein in bilobed gut granules, we examined
the colocalization of FluoZin-3 and CDF2::mCherry. While FluoZin-3 fluorescence was
asymmetric and strong in only one lobe of
bilobed granules, CDF-2::mCherry was localized to both lobes (Figure S4, right).
To define the molecular properties of bilobed
gut granules, we examined the localization of
several well-characterized endosomal or lysosomal marker proteins. The GTPase RAB-5
localizes to early endosomes (Chen et al.,
2006; Hermann et al., 2005). RAB-5 did not colocalize with LysoTracker or CDF-2 (data not
shown), suggesting that gut granules are
distinct from early endosomes. Lysosome associated membrane proteins (LAMPs) are localized predominantly in lysosomes in vertebrates,
and C. elegans LMP-1 localizes to late endosomal or lysosomal
vesicles of intestinal cells (Chen et al., 2006; Kostich et al., 2000).
A subset of LMP-1 colocalized with LysoTracker in gut granules
(Figure 4B). In addition, a subset of LMP-1 localized to other
membrane compartments that were LysoTracker negative,
including the plasma membrane, indicating this marker is not
specific for lysosomes. To determine whether LMP-1 was
present on both lobes of bilobed gut granules, we exposed
worms to high zinc and used CDF-2::mCherry to define bilobed
morphology. LMP-1 was present only in one lobe of bilobed gut
granules (Figure 4C). These results indicate that bilobed gut
granules displayed asymmetric molecular properties; one lobe
has late endosomal or lysosomal characteristics whereas the
other lobe lacks at least one lysosomal protein.
C. elegans PGP-2 is an ABC transporter that localizes specifically to the gut granule membrane (Schroeder et al., 2007). With
Cell Metabolism
Figure 5. Glo Genes Are Necessary for the

Formation of Bilobed Gut Granules
(A) Fluorescence microscope images of live transgenic animals expressing CDF-2::GFP. L1 stage
animals were fed RNAi bacteria to reduce expression of the indicated genes, and L4 stage animals
were cultured with LysoTracker and 200 mM supplemental zinc for 16 hr and visualized. Images
show intestinal cells with pharynx to the left and
tail to the right. Boxed regions are magnified in
the bottom panels. Bilobed gut granules are
indicated by arrows and arrowheads. Scale bar:
10 mm and 2 mm (bottom) (see also Figure S7).
(B) A genetic pathway for the formation of bilobed
gut granules and a summary of molecular properties of gut granules in basal zinc (left) and
bilobed gut granules in high zinc (right). Plus (+)
and minus () signs indicate the presence and
absence of proteins/staining, respectively. Low
and high indicate relative levels of FluoZin-3
staining.
no supplemental zinc, CDF-2::mCherry, PGP-2::GFP, and autofluorescence completely colocalized in gut granules (Figure S6B,
left). With 100 mM supplemental zinc, CDF-2 and PGP-2 fully
colocalized on both sides of bilobed gut granules, while autofluorescence displayed an asymmetric pattern and was only
prominent on one lobe, similar to the LysoTracker staining
pattern (Figure S6B, right). These results indicate that bilobed
gut granules contain proteins that are distributed on both lobes,
such as PGP-2 and CDF-2, and at least one protein that is
asymmetrically localized to the LysoTracker positive lobe,
LMP-1 (Figure 5B).
Glo Genes Are Necessary for the Formation of Bilobed
Gut Granules
To elucidate the role of vesicular trafficking pathways in bilobed
gut granule formation, we reduced the activity of key genes
using the method of feeding RNAi. Animals were exposed to
RNAi starting at the first larval (L1) stage to allow normal embry-
onic development. RAB-7 and RME-8

are necessary for lysosome biogenesis
and endocytosis in multiple cell types
(Bucci et al., 2000; Zhang et al., 2001),
although a function in adult intestinal
cells has not been reported. When
rab-7 or rme-8 activities were reduced
by RNAi, bilobed morphology was still
detected (Figure 5A), indicating that
these genes may not be required for
the formation of bilobed gut granules or
that residual gene activity was sufficient
to mediate formation of bilobed granules. When pgp-2 or glo-3 activities
were reduced by RNAi, bilobed
morphology was not observed (Figure 5A). We did observe vesicles that
contained CDF-2::GFP and were LysoTracker negative, which we have not
observed in wild-type animals. These
results suggest that glo genes that function in gut granule
biogenesis are also required for the formation of bilobed morphology in response to high zinc.
To determine the role of zinc transport in the formation of
bilobed gut granules, we analyzed cdf-2. cdf-2 mutant animals
displayed bilobed gut granules that contained PGP-2::GFP on
both lobes and LysoTracker asymmetrically on one lobe (Figure S7A), similar to wild-type. In contrast to wild-type, bilobed
gut granules were not stained asymmetrically with FluoZin-3 in
cdf-2 mutant animals (Figure S7B). These results indicate that
CDF-2 activity is not necessary for the formation of bilobed gut
granules but is necessary for the asymmetric accumulation of
zinc. These results define a genetic pathway for the formation
of bilobed organelles (Figure 5B).
Gut Granules Are Necessary for Zinc Detoxification
To determine if zinc storage in gut granules is a protective mechanism that promotes detoxification, we analyzed the ability of
Cell Metabolism
B
WT
pgp-2
glo-1
glo-3
Length of worms (normalized)
1.4
1.2
1
0.8
0.6
0.4
0.2
0
1
0.8
0.6
0.4
0.2
0
50
100
150
200
Supplemental Zn (M)
C
1.4
WT
cdf-2
1.2
Length of worms (normalized)
50
100
150
200
Supplemental Zn (M)
WT
pgp-2
glo-1
l 1
glo-3
Figure 6. Zinc Storage in Gut Granules

Promotes Detoxification
(AD) Wild-type, pgp-2(kx48), glo-1(zu391), glo3(zu446), and cdf-2(tm788) hermaphrodites were
synchronized at the L1 stage and cultured on
NAMM dishes for 3 days with the indicated levels
of supplemental zinc in (A and B) or supplemental
cadmium (C and D). The length of individual
animals was measured using microscopy and
ImageJ software. To compare strains that have
different growth rates under optimal conditions,
we normalized the length of worms by setting
the value at 0 mM supplemental metal equal to
1.0 for each strain. Each point indicates mean
values SD (n = 10 for Zn, and n = 20 for Cd).
D
1.2
WT
cdf-2
cdf
2
Length of worms (normalized

d)
d)
Length of worms (normalized
animals to the toxic effects of additional

metals. Glo animals were similar to wildtype animals in sensitivity to dietary
1
0.8
cadmium (Figure 6C) and copper (data
0.8
not shown), indicating that gut granules
0.6
0.6
are not necessary to detoxify these
0.4
metals. The growth of cdf-2(tm788)
0.4
mutant animals was also similar to wild0.2
0.2
type animals in the presence of supple0
0
mental cadmium (Figure 6D) and copper
0
10
20
30
40
50
0
10
20
30
40
50
(data not shown). These results indicate
Supplemental Cd (M)
Supplemental Cd (M)
that CDF-2 function is specific to zinc
and support the conclusion that gut grananimals to tolerate high levels of dietary zinc. Dose-dependent ules are not a general site of metal storage and may be specific
zinc toxicity was previously determined using fully defined liquid for zinc storage.
CeMM medium (Davis et al., 2009). Here, we show that wild-type
animals cultured on NAMM dishes displayed a dose-dependent Gut Granules Provide a Source of Zinc during Deficiency
decrease in growth rate in response to supplemental zinc, indi- One possible function of storing zinc in gut granules is to provide
cating that high dietary zinc inhibits growth in a different culture a source of zinc that can be mobilized in response to zinc defimedium (Figure 6A). The concentration of zinc in these two ciency. To test this model, we monitored the levels of zinc in
experiments cannot be compared directly due to differences in gut granules using FluoZin-3 during a shift from high zinc to
culture conditions, particularly the undefined contribution of low zinc conditions. Wild-type animals cultured with 200 mM
bacteria to dietary zinc for worms cultured on NAMM dishes. supplemental zinc were transferred to NAMM dishes containing
pgp-2, glo-1, and glo-3 mutant animals displayed significantly 200 mM supplemental zinc, 0 mM supplemental zinc, or 100 mM
lower growth rates than wild-type animals at all concentrations TPEN and analyzed for FluoZin-3 fluorescence after 24 hr and
of supplemental zinc (Figure 6A). The severity of the zinc sensi- 48 hr. Animals continuously cultured with 200 mM supplemental
tivity phenotype correlated with the severity of Glo phenotype; zinc displayed a progressive increase of FluoZin-3 fluorescence
pgp-2 mutant animals were moderately zinc sensitive, whereas (Figures 7A and 7B). Animals shifted to 0 mM supplemental
glo-1 and glo-3 mutant animals were strongly zinc sensitive. zinc displayed FluoZin-3 fluorescence that increased by
These results indicate that gut granules play an important 1.7-fold after 24 hr but then slightly decreased after 48 hr.
protective role in response to zinc toxicity. To test the hypothesis Animals shifted to 100 mM TPEN displayed FluoZin-3 fluoresthat zinc storage in gut granules is important for zinc resistance cence that increased by 1.4-fold after 24 hr but then substanrather than another function of gut granules, we analyzed cdf-2. tially decreased after 48 hr to a level below the initial value. These
cdf-2(tm788) mutant animals appeared to have a normal number results suggest that zinc stored in gut granules is released during
and morphology of gut granules based on LysoTracker and zinc deficiency.
To investigate the physiologic significance of zinc mobilizaPGP-2::GFP staining, but these granules were defective in zinc
storage (Figure S7). Similar to Glo animals, cdf-2 mutant animals tion, we analyzed the growth of animals that had either low or
were hypersensitive to dietary zinc compared to wild-type high zinc storage and then were exposed to zinc-deficient
animals (Figure 6B). These findings indicate that CDF-2 medi- conditions. Wild-type animals were precultured with either
ated zinc transport into gut granules is a critical mechanism for 0 mM or 50 mM supplemental zinc and then cultured in the
presence of TPEN. Animals precultured with 50 mM supplezinc detoxification.
To investigate whether gut granules play a more general role mental zinc displayed a significantly increased growth rate in
in metal detoxification, we examined the sensitivity of Glo 100 mM TPEN compared to animals precultured with 0 mM
1.2
Cell Metabolism
Figure 7. Zinc in Gut Granules Can Be Mobilized in

Response to Zinc Deficiency
(A) Wild-type L4 stage hermaphrodites were cultured on
NAMM dishes containing 200 mM supplemental zinc to
promote zinc storage. Animals were then transferred to
NAMM dishes containing the indicated levels of supplemental zinc or TPEN, and FluoZin-3 fluorescence was
analyzed by fluorescence microscopy after 24 hr and
48 hr. Images show intestinal cells (pharynx to the left and
tail to the right). Scale bar: 50 mm.
(B) Quantification of fluorescence images like those shown
in (A). The fluorescence intensity (shown in arbitrary units,
A.U.) was normalized by setting the value at 200 mM
supplemental Zn (time 0) equal to 1.0. Each point indicates
mean values SEM (n = 10).
(C) Wild-type and cdf-2(tm788) L1 stage hermaphrodites
were precultured for 12 hr on NAMM dishes containing
0 or 50 mM supplemental zinc (High Zn), cultured for 3 days
on NGM dishes with the indicated levels of TPEN, and
analyzed individually for length. Each point indicates mean
value SD (n = 20).
(D) Wild-type, glo-1(zu391), glo-3(zu446) and pgp-2(kx48)
animals were analyzed as described in (C). The results
with 100 mM TPEN are presented, because this concentration was the most informative with wild-type animals.
The length of individual worms at 100 mM TPEN was
divided by the average length at 0 mM TPEN. Bars indicate
mean values SD (n = 20). This ratio compares the growth
of worms in deficient and normal zinc conditionslower
values indicate more severely reduced growth in response
to zinc deficiency. Animals were precultured with 0 (black)
or 50 mM (white) supplemental zinc. For each strain, the
white bar was compared to the black bar (***p < 0.0001).
All the mutant strain black and white bar values were
significantly different than the wild-type black and white
bar values, respectively.
DISCUSSION
supplemental zinc (Figure 7C), indicating that stored zinc can

be mobilized during dietary deficiency. cdf-2 mutant animals
precultured with 50 mM supplemental zinc did not display
a significant increase in growth rate (Figures 7C). These results
indicate that CDF-2 is necessary for zinc storage, which
promotes the resistance to zinc deficiency. To determine if
zinc is mobilized from gut granules during deficiency, we
examined Glo animals. glo-1, glo-3, and pgp-2 mutant animals
precultured with 0 mM supplemental zinc displayed slower
growth rates in the presence of TPEN compared to wild-type
animals (Figure 7D). In addition, the growth rate of the Glo
animals was not significantly affected by preculture with
50 mM supplemental zinc, whereas wild-type animals displayed
a significant increase (Figure 7D). These results indicate that
gut granules, which are the major site of zinc storage during
dietary excess, are the source of zinc mobilized during zinc
deficiency.
Lysosome-Related Organelles Are a Site

of Zinc Storage and Mobilization in an
Animal
Zinc is essential, but zinc availability can fluctuate. Thus, mechanisms to store and mobilize
zinc are important. We used C. elegans to characterize the site of zinc storage in an animal by developing
methods to visualize labile zinc using a zinc-specific fluorescent
dye, FluoZin-3. Labile zinc was detected primarily in lysosomerelated organelles in intestinal cells called gut granules. The
biogenesis of gut granules has been studied using Glo mutant
animals that have reduced numbers of gut granules (Hermann
et al., 2005; Rabbitts et al., 2008; Schroeder et al., 2007). Glo
mutant animals were used to demonstrate that reducing the
number of gut granules caused a corresponding reduction of
labile and total zinc. By contrast, Glo mutant animals had wildtype levels of other metals such as copper. These results indicate that gut granules function specifically in zinc storage and
provide direct evidence for the presence of a zinc-specific
storage site in an animal.
The cation diffusion facilitator protein CDF-2 plays a critical
role in zinc storage in these organelles. Colocalization experiments with gut granule markers demonstrated that CDF-2 is
Cell Metabolism
localized specifically to the membrane of gut granules. The level

of CDF-2 protein was increased by high dietary zinc, indicating
CDF-2 has an important function in responding to high levels
of zinc. This function was demonstrated using genetic analysis,
since reducing the activity of cdf-2 with a loss-of-function mutation and increasing the activity of cdf-2 by overexpression
showed that cdf-2 is both necessary and sufficient for zinc
storage in gut granules. Consistent with the model that CDF-2
directly transports zinc into gut granules, CDF-2 functioned
cell-autonomously in intestinal cells to promote zinc accumulation. The protein sequence of C. elegans CDF-2 is similar to
mammalian ZnT-2, which was discovered by Palmiter et al.
(1996) and shown to promote cellular resistance to zinc by
facilitating vesicular sequestration in cultured cells. Similar to
CDF-2, ZnT-2 is localized to intracellular organelles and regulated by dietary zinc (Falcon-Perez and DellAngelica, 2007;
Guo et al., 2010; Liuzzi et al., 2001). Mutations in human ZnT-2
are associated with low milk zinc concentrations, indicating
one function of ZnT-2 in humans is secretion of zinc into breast
milk (Chowanadisai et al., 2006). These similarities suggest that
C. elegans CDF-2 and mammalian ZnT-2 have important and
conserved functions in cellular and organismal zinc metabolism.
We characterized two functions of zinc storage in gut granules. One function is detoxification. Glo mutant animals that
had reduced numbers of gut granules were hypersensitive to
zinc toxicity, and the hypersensitivity correlated with the severity
of the Glo phenotype, demonstrating the importance of gut granules in zinc tolerance. cdf-2 loss-of-function mutant animals had
normal numbers of gut granules, but these organelles were deficient in zinc accumulation, and cdf-2 mutant animals were also
hypersensitive to dietary zinc. Thus, CDF-2-mediated transport
of zinc into gut granules is a key mechanism to detoxify excess
dietary zinc. In the unicellular yeast Saccharomyces cerevisiae,
the vacuole plays an important role in zinc tolerance, and zinc
transport into the vacuole is mediated by the CDF proteins
Cot1 and Zrc1 (Eide, 2006). Acidocalcisomes are specialized
organelles that accumulate zinc and other metals and have
been observed in a wide range of organisms (Docampo et al.,
2005). In vertebrates, several cell types have been shown to
accumulate zinc in specific organelles. For example, glutamatergic neurons, pancreatic b-cells and acinar cells, and intestinal
paneth cells contain secretary vesicles with high concentrations
of labile zinc (Kelly et al., 2004; Lichten and Cousins, 2009). Dietary zinc levels influence the number of paneth cells and the
morphology of the zinc-containing organelles, suggesting there
may be parallels with the gut granules of C. elegans intestinal
cells (Kelly et al., 2004). In several types of mammalian cells,
labile zinc accumulates in intracellular organelles that have lysosomal properties and have been called zincosomes (Haase and
Beyersmann, 1999; Palmiter et al., 1996). ZnT-2 mediates zinc
transport into zincosomes and promotes zinc tolerance
(Falcon-Perez and DellAngelica, 2007). Because yeast vacuoles, C. elegans gut granules, and mammalian zincosomes all
have lysosomal properties, lysosome-related organelles may
have an evolutionarily conserved function in cellular zinc storage.
High levels of dietary zinc induce changes in gene expression,
such as induction of metallothionein genes, and these alterations
in gene expression may play a role in detoxification (Andrews,
2001). An important direction for future research will be to
determine how zinc sequestration in lysosome-related organelles relates to other mechanisms of zinc tolerance such as
expression of zinc-binding proteins.
The second function of zinc storage in gut granules is to
provide a source of zinc that can be mobilized during dietary
deficiency. Animals shifted from high to low zinc conditions
displayed decreased zinc levels in gut granules after 48 hr, suggesting that stored zinc is released in response to deficiency.
Importantly, our data indicate that stored zinc is utilized for the
physiologic process of growth, since wild-type animals with
normal zinc storage grew more robustly than Glo and cdf-2
mutant animals with defective zinc storage. These results
demonstrate the existence of a specific site of zinc storage in
an animal that provides a physiologically important source of
zinc during dietary deficiency. Elegant studies reported by Eide
and colleagues show that zinc stored in the yeast vacuole is
mobilized by the ZIP protein Zrt3, and stored zinc can supply
the needs of as many as eight generations of progeny cells under
zinc starvation conditions (MacDiarmid et al., 2000; Simm et al.,
2007). In mammalian T cells, ZIP8 localizes to lysosomes and
mediates release of zinc that plays a regulatory role in T cell
activation (Aydemir et al., 2009; Begum et al., 2002). Zinc storage
and mobilization at the organismal level has been investigated
in vertebrates. Chickens fed a high zinc diet display accumulation of zinc in several tissues including the liver, bone, and small
intestine (Emmert and Baker, 1995). When these animals are
shifted to a zinc-deficient diet, the level of accumulated zinc
decreases, and the onset of zinc deficiency symptoms is
delayed compared to control chickens fed a low zinc diet, suggesting that accumulated zinc may be mobilized. In humans,
symptoms of zinc deficiency develop rapidly in response to dietary zinc deficiency, indicating there are only small pools of zinc
available for mobilization (King, 2011). Our studies indicate that
the capacity for zinc storage and mobilization of C. elegans is
intermediate between the large capacity of yeast and the small
capacity of mammals. The limited capacity of animals compared
to yeast may reflect the strategy of storing zinc in specialized cell
types, such as intestinal cells in C. elegans, compared to storage
in the vacuole of every yeast cell.
Lysosome-Related organelles Adopt a Bilobed
Morphology in Response to High Zinc
Gut granules displayed striking morphological changes in
response to high dietary zinc. In standard culture conditions,
gut granules were typically round in shape, autofluorescent,
positive for LysoTracker and FluoZin-3 staining, and positive
for membrane localization of CDF-2, LMP-1, and PGP-2. In
high dietary zinc, gut granules were frequently bilobed in shape,
and the two lobes displayed an asymmetric distribution of molecules. One lobe displayed molecular properties that were similar
to gut granules in standard culture conditions, whereas the other
lobe displayed strong FluoZin-3 staining, was positive for the gut
granule-specific proteins PGP-2 and CDF-2, and was negative
for the lysosomal markers Lysotracker and LMP-1. The high level
of zinc in one lobe raises the possibility that CDF-2 is more
abundant or active in the high zinc lobe. Using epifluorescence
microscopy, CDF-2 levels appeared to be similar on both lobes.
However, this analysis is complicated by autofluorescence from
the LysoTracker positive lobe. Using confocal microscopy that
Cell Metabolism
minimizes signal due to autofluorescence, CDF-2 levels appeared to be higher on the lobe with high zinc, suggesting that
there may be a positive correlation between CDF-2 levels and
zinc accumulation. While these studies are suggestive, strong
conclusions about the relative levels of CDF-2 on the two lobes
will require a more quantitative analysis than described here. We
speculate that the bilobed structure may facilitate zinc storage
by generating a compartment that is specialized to accommodate a large amount of zinc. Alternatively, it may facilitate zinc
excretion through budding of a secretory vesicle containing
excess zinc. Secretory granules of mammalian paneth cells
also display nonhomogenous staining of the contents, but the
arrangement is distinct from the bilobed granules. Based on
EM analysis, paneth cell granules have a central core and a
distinct surround, called a biphasic structure, which differ in
carbohydrate composition (Leis et al., 1997).
We used genetic analysis to elucidate a pathway for the formation of bilobed gut granules (Figure 5B). Reducing the activity of
pgp-2 and glo-3 by the method of RNAi inhibited the formation
of bilobed morphology. Because glo genes were inactivated
after gut granule biogenesis during embryonic development,
the absence of bilobed morphology is not likely to be caused
by defects in gut granule biogenesis. Rather, glo genes are likely
to act in both gut granule biogenesis and the transition to bilobed
morphology in response to high levels of zinc. Interestingly,
animals exposed to RNAi of glo genes displayed vesicles that
contained CDF-2::GFP and were LysoTracker negative. These
vesicles were spatially separate from gut granules that contained
CDF-2::GFP and were LysoTracker positive, suggesting that glo
genes may be involved in a vesicle fusion event that generates
bilobed morphology. cdf-2 mutant animals displayed organelles
with bilobed morphology, based on membrane staining with
PGP-2, but these organelles did not display FluoZin-3 fluorescence. Thus, CDF-2 is not necessary for the formation of bilobed
morphology but is necessary to accumulate zinc in bilobed
organelles. Therefore, the formation of bilobed morphology is
separable from zinc transport and accumulation. Lysosome
biogenesis and function can be regulated by signaling pathways
in response to cellular conditions (Karageorgos et al., 1997), and
the transcription factor EB was recently demonstrated to play
a critical role in this process (Sardiello et al., 2009). These results
indicate lysosomes are not static organelles, but rather respond
dynamically to cellular conditions. Our results showed that environmental zinc induced transcription of cdf-2, a resident protein
of lysosome-related organelles, and caused lysosome-related
organelles to adopt a bilobed morphology. These findings are
consistent with the model that the transcriptional response
to environmental zinc causes dynamic changes in lysosomerelated organelle structure and function. These studies establish
a genetically tractable model system to dissect the contribution
of transcriptional programs to the biogenesis of specialized
lysosome-related organelles that play a critical role in coping
with the environmental challenge of high zinc.
General Methods and Strains
C. elegans strains were cultured at 20 C on nematode growth medium (NGM)
seeded with E. coli OP50 unless otherwise noted (Brenner, 1974). The wild-
type C. elegans and parent of all mutant strains was Bristol N2. The following
mutations and transgenes were used: pgp-2(kx48) I (Schroeder et al., 2007),
unc-119(ed3) III (Praitis et al., 2001), cdf-2(tm788) X (Davis et al., 2009),
glo-1(zu391) X (Hermann et al., 2005), glo-3(zu446) X (Rabbitts et al., 2008),
amIs4(cdf-2::GFP::unc-119(+)) (Davis et al., 2009), pwIs50 (lmp-1:GFP)
(Treusch et al., 2004), pwIs72 (Pvha-6::GFP::rab-5) (Hermann et al., 2005),
and kxEx98(pgp-2::GFP;rol-6D) (Schroeder et al., 2007). amEx132(cdf2::mCherry;rol-6D) and amEx142(cdf-2::mCherry::unc-119(+)) are described
here. Double mutant animals were generated by standard methods, and genotypes were confirmed by PCR or DNA sequencing.
Metal Sensitivity Assays
Gravid adult hermaphrodites were treated with NaOH and bleach, and eggs
were incubated in M9 solution overnight to allow hatching and synchronized
arrest at the L1 stage. L1 animals were transferred to NAMM dishes (Bruinsma
et al., 2008) supplemented with zinc sulfate (ZnSO4), cadmium chloride
(CdCl2), or copper sulfate (CuSO4) and seeded with concentrated OP50. After
3 days, animals were washed twice in M9 containing 0.01% Tween-20, paralyzed with 10 mM sodium azide (NaN3) in M9, and mounted on a 2% agarose
pad on a microscope slide. Images were captured with a Zeiss Axioplan 2
microscope equipped with a Zeiss AxioCam MRm digital camera. Length of
individual animals was measured as an indicator of growth using ImageJ
software (NIH) by drawing a line from the nose to the tail tip.
Quantitative Analysis of CDF-2::GFP Expression by Fluorescence
Microscopy
cdf-2(tm788);amIs4 animals were synchronized at L1 stage and cultured on
NGM dishes. L4 stage hermaphrodites were then cultured for 24 hr on
NAMM dishes supplemented with ZnSO4 and seeded with concentrated
OP50. Animals were paralyzed with 0.1% tricaine and 0.01% tetramisole in
M9, mounted on 2% agarose pads on microscope slides, and imaged with
a Zeiss Axioplan 2 microscope equipped with a Zeiss AxioCam MRm digital
camera using identical settings and exposure times. GFP fluorescence intensity was quantified using ImageJ software (NIH). Briefly, the Spot Enhancing
Filter 2D plugin was used to amplify signals from gut granules, and then
threshold settings were used to specifically select the fluorescent regions of
gut granules. The selected regions were overlaid on the original images and
analyzed for mean fluorescence intensity of the area.
Staining with FluoZin-3, LysoTracker, and LysoSensor
FluoZin-3 acetoxymethyl (AM) ester (Molecular Probes F24195) was reconstituted in dimethylsulfoxide (DMSO) to generate a 1 mM stock solution, diluted in
M9 and dispensed on NAMM dishes to yield a final concentration of 3 mM.
L4 stage hermaphrodites were cultured on these dishes for 1224 hr in the
dark, transferred to NGM dishes without FluoZin-3 for 30 min to reduce
FluoZin-3 in the intestinal lumen, and analyzed by fluorescence microscopy
as described above. The intestine on the anterior half of each animal was
analyzed because this structure was typically observed in the same focal
plane. Residual fluorescence from the intestinal lumen was manually removed
and excluded from the analysis.
LysoTracker RED DND-99 (1 mM, Invitrogen L7528), or LysoSensor Green
DND-153 (1 mM, Invitrogen L7534) were diluted in M9 and dispensed on
NAMM dishes to yield a final concentration of 2 mM. L4 stage hermaphrodites
were cultured on these dishes for 1224 hr in the dark, transferred to NGM
dishes without dye for 30 min, and imaged as described above. Confocal
microscopy was performed using an Olympus FV500 confocal microscope
system equipped with multiline argon (458/488/515 nm) and krypton
(568 nm) lasers.
Zinc Shift Assays
To monitor zinc levels in gut granules, we cultured L4 stage animals for
1216 hr on NAMM dishes containing FluoZin-3 and 200 mM ZnSO4 and
then analyzed them by fluorescence microscopy as described above. Next,
animals were transferred to NAMM dishes with FluoZin-3 containing 0 or
200 mM ZnSO4 or 100 mM TPEN and analyzed by fluorescence microscopy
after 24 hr and 48 hr. To analyze growth, we cultured synchronized L1 stage
animals on NAMM dishes supplemented with 0 or 50 mM ZnSO4 for 12 hr.
We chose 50 mM supplemental zinc because it caused relatively mild toxicity
Cell Metabolism
as measured by subsequent growth, whereas 100 and 200 mM supplemental

zinc caused substantial toxicity (data not shown). Animals were washed three
times, incubated in M9 containing 0.01% Tween-20 for 30 min to minimize
residual bacteria, and washed one time in M9 containing 0.01% Tween-20.
Animals were cultured for 3 days on NGM dishes supplemented with TPEN
and seeded with concentrated OP50, and the length of each animal was
determined.
Statistics
All data were analyzed by two-tailed unpaired Students t test, and p < 0.05
was considered statistically significant.
Supplemental Information includes seven figures, Supplemental Experimental Procedures, and Supplemental References and can be found with
this article online at doi:10.1016/j.cmet.2011.12.003.
ACKNOWLEDGMENTS
We thank Greg Hermann, Barth Grant, the Caenorhabditis Genetics Center,
and the National Bioresource Project for providing strains; Andrew Fire,
Michael Nonet, and Judith Austin for providing plasmids; and Daniel Schneider
for technical assistance. We are grateful to Tim Schedl, Stuart Kornfeld,
Jeanne Nerbonne, Jason Mills, and Peter Chivers for helpful advice about
the manuscript. This research was supported by grants from the National
Institutes of Health to K.K. (GM068598, CA84271, and AG026561). K.K. was
a Senior Scholar of the Ellison Medical Foundation. H.C.R. was a scholar of
the McDonnell International Scholars Academy.
REFERENCES
Andrews, G.K. (2001). Cellular zinc sensors: MTF-1 regulation of gene expression. Biometals 14, 223237.
Aydemir, T.B., Liuzzi, J.P., McClellan, S., and Cousins, R.J. (2009). Zinc transporter ZIP8 (SLC39A8) and zinc influence IFN-gamma expression in activated
human T cells. J. Leukoc. Biol. 86, 337348.
Begum, N.A., Kobayashi, M., Moriwaki, Y., Matsumoto, M., Toyoshima, K.,
and Seya, T. (2002). Mycobacterium bovis BCG cell wall and lipopolysaccharide induce a novel gene, BIGM103, encoding a 7-TM protein: identification of
a new protein family having Zn-transporter and Zn-metalloprotease signatures. Genomics 80, 630645.
Brenner, S. (1974). The genetics of Caenorhabditis elegans. Genetics 77,
7194.
Bruinsma, J.J., Jirakulaporn, T., Muslin, A.J., and Kornfeld, K. (2002). Zinc ions
and cation diffusion facilitator proteins regulate Ras-mediated signaling. Dev.
Cell 2, 567578.
Bruinsma, J.J., Schneider, D.L., Davis, D.E., and Kornfeld, K. (2008).
Identification of mutations in Caenorhabditis elegans that cause resistance
to high levels of dietary zinc and analysis using a genomewide map of single
nucleotide polymorphisms scored by pyrosequencing. Genetics 179,
811828.
Bucci, C., Thomsen, P., Nicoziani, P., McCarthy, J., and van Deurs, B. (2000).
Rab7: a key to lysosome biogenesis. Mol. Biol. Cell 11, 467480.
Chen, C.C., Schweinsberg, P.J., Vashist, S., Mareiniss, D.P., Lambie, E.J., and
Grant, B.D. (2006). RAB-10 is required for endocytic recycling in the
Caenorhabditis elegans intestine. Mol. Biol. Cell 17, 12861297.
Chowanadisai, W., Lonnerdal, B., and Kelleher, S.L. (2006). Identification of
a mutation in SLC30A2 (ZnT-2) in women with low milk zinc concentration
that results in transient neonatal zinc deficiency. J. Biol. Chem. 281, 39699
39707.
Clokey, G.V., and Jacobson, L.A. (1986). The autofluorescent lipofuscin

granules in the intestinal cells of Caenorhabditis elegans are secondary
lysosomes. Mech. Ageing Dev. 35, 7994.
Cragg, R.A., Phillips, S.R., Piper, J.M., Varma, J.S., Campbell, F.C., Mathers,
J.C., and Ford, D. (2005). Homeostatic regulation of zinc transporters in the
human small intestine by dietary zinc supplementation. Gut 54, 469478.
Davis, D.E., Roh, H.C., Deshmukh, K., Bruinsma, J.J., Schneider, D.L.,
Guthrie, J., Robertson, J.D., and Kornfeld, K. (2009). The cation diffusion
facilitator gene cdf-2 mediates zinc metabolism in Caenorhabditis elegans.
Genetics 182, 10151033.
Docampo, R., de Souza, W., Miranda, K., Rohloff, P., and Moreno, S.N. (2005).
Acidocalcisomes - conserved from bacteria to man. Nat. Rev. Microbiol. 3,
251261.
Eide, D.J. (2006). Zinc transporters and the cellular trafficking of zinc. Biochim.
Biophys. Acta 1763, 711722.
Eide, D.J. (2009). Homeostatic and adaptive responses to zinc deficiency in
Saccharomyces cerevisiae. J. Biol. Chem. 284, 1856518569.
Emmert, J.L., and Baker, D.H. (1995). Zinc stores in chickens delay the onset of
zinc deficiency symptoms. Poult. Sci. 74, 10111021.
Falcon-Perez, J.M., and DellAngelica, E.C. (2007). Zinc transporter 2
(SLC30A2) can suppress the vesicular zinc defect of adaptor protein
3-depleted fibroblasts by promoting zinc accumulation in lysosomes. Exp.
Cell Res. 313, 14731483.
Feeney, G.P., Zheng, D., Kille, P., and Hogstrand, C. (2005). The phylogeny of
teleost ZIP and ZnT zinc transporters and their tissue specific expression and
response to zinc in zebrafish. Biochim. Biophys. Acta 1732, 8895.
Fosmire, G.J. (1990). Zinc toxicity. Am. J. Clin. Nutr. 51, 225227.
Gee, K.R., Zhou, Z.L., Ton-That, D., Sensi, S.L., and Weiss, J.H. (2002).
Measuring zinc in living cells. A new generation of sensitive and selective fluorescent probes. Cell Calcium 31, 245251.
Gourley, B.L., Parker, S.B., Jones, B.J., Zumbrennen, K.B., and Leibold, E.A.
(2003). Cytosolic aconitase and ferritin are regulated by iron in Caenorhabditis
elegans. J. Biol. Chem. 278, 32273234.
Guo, L., Lichten, L.A., Ryu, M.S., Liuzzi, J.P., Wang, F., and Cousins, R.J.
(2010). STAT5-glucocorticoid receptor interaction and MTF-1 regulate the
expression of ZnT2 (Slc30a2) in pancreatic acinar cells. Proc. Natl. Acad.
Sci. USA 107, 28182823.
Haase, H., and Beyersmann, D. (1999). Uptake and intracellular distribution of
labile and total Zn(II) in C6 rat glioma cells investigated with fluorescent probes
and atomic absorption. Biometals 12, 247254.
Hambidge, M. (2000). Human zinc deficiency. J. Nutr. 130 (5S, Suppl), 1344S
1349S.
Hermann, G.J., Schroeder, L.K., Hieb, C.A., Kershner, A.M., Rabbitts, B.M.,
Fonarev, P., Grant, B.D., and Priess, J.R. (2005). Genetic analysis of lysosomal
trafficking in Caenorhabditis elegans. Mol. Biol. Cell 16, 32733288.
Karageorgos, L.E., Isaac, E.L., Brooks, D.A., Ravenscroft, E.M., Davey, R.,
Hopwood, J.J., and Meikle, P.J. (1997). Lysosomal biogenesis in lysosomal
storage disorders. Exp. Cell Res. 234, 8597.
Kelly, P., Feakins, R., Domizio, P., Murphy, J., Bevins, C., Wilson, J., McPhail,
G., Poulsom, R., and Dhaliwal, W. (2004). Paneth cell granule depletion in the
human small intestine under infective and nutritional stress. Clin. Exp.
Immunol. 135, 303309.
King, J.C. (2011). Zinc: an essential but elusive nutrient. Am. J. Clin. Nutr. 94,
679S684S.
Kostich, M., Fire, A., and Fambrough, D.M. (2000). Identification and molecular-genetic characterization of a LAMP/CD68-like protein from
Caenorhabditis elegans. J. Cell Sci. 113, 25952606.
Leis, O., Madrid, J.F., Ballesta, J., and Hernandez, F. (1997). N- and O-linked
oligosaccharides in the secretory granules of rat Paneth cells: an ultrastructural cytochemical study. J. Histochem. Cytochem. 45, 285293.
Liao, V.H., and Freedman, J.H. (1998). Cadmium-regulated genes from
the nematode Caenorhabditis elegans. Identification and cloning of new
Cell Metabolism
cadmium-responsive genes by differential display. J. Biol. Chem. 273, 31962

31970.
is regulated by the conserved and concerted functions of HRG-1 proteins.

Nature 453, 11271131.
Lichten, L.A., and Cousins, R.J. (2009). Mammalian zinc transporters: nutritional and physiologic regulation. Annu. Rev. Nutr. 29, 153176.
Sardiello, M., Palmieri, M., di Ronza, A., Medina, D.L., Valenza, M., Gennarino,
V.A., Di Malta, C., Donaudy, F., Embrione, V., Polishchuk, R.S., et al. (2009). A
gene network regulating lysosomal biogenesis and function. Science 325,
473477.
Liuzzi, J.P., Blanchard, R.K., and Cousins, R.J. (2001). Differential regulation of
zinc transporter 1, 2, and 4 mRNA expression by dietary zinc in rats. J. Nutr.
131, 4652.
MacDiarmid, C.W., Gaither, L.A., and Eide, D. (2000). Zinc transporters that
regulate vacuolar zinc storage in Saccharomyces cerevisiae. EMBO J. 19,
28452855.
Schroeder, L.K., Kremer, S., Kramer, M.J., Currie, E., Kwan, E., Watts, J.L.,
Lawrenson, A.L., and Hermann, G.J. (2007). Function of the Caenorhabditis elegans ABC transporter PGP-2 in the biogenesis of a lysosome-related fat
storage organelle. Mol. Biol. Cell 18, 9951008.
Murakami, M., and Hirano, T. (2008). Intracellular zinc homeostasis and zinc
signaling. Cancer Sci. 99, 15151522.
Simm, C., Lahner, B., Salt, D., LeFurgey, A., Ingram, P., Yandell, B., and Eide,
D.J. (2007). Saccharomyces cerevisiae vacuole in zinc storage and intracellular zinc distribution. Eukaryot. Cell 6, 11661177.
Murphy, J.T., Bruinsma, J.J., Schneider, D.L., Collier, S., Guthrie, J.,
Chinwalla, A., Robertson, J.D., Mardis, E.R., and Kornfeld, K. (2011).
Histidine protects against zinc and nickel toxicity in Caenorhabditis elegans.
PLoS Genet. 7, e1002013.
Treusch, S., Knuth, S., Slaugenhaupt, S.A., Goldin, E., Grant, B.D., and Fares,
H. (2004). Caenorhabditis elegans functional orthologue of human protein hmucolipin-1 is required for lysosome biogenesis. Proc. Natl. Acad. Sci. USA
101, 44834488.
Palmiter, R.D., Cole, T.B., and Findley, S.D. (1996). ZnT-2, a mammalian
protein that confers resistance to zinc by facilitating vesicular sequestration.
EMBO J. 15, 17841791.
Vallee, B.L., and Falchuk, K.H. (1993). The biochemical basis of zinc physiology. Physiol. Rev. 73, 79118.
Praitis, V., Casey, E., Collar, D., and Austin, J. (2001). Creation of low-copy
integrated transgenic lines in Caenorhabditis elegans. Genetics 157, 1217
1226.
Rabbitts, B.M., Ciotti, M.K., Miller, N.E., Kramer, M., Lawrenson, A.L., Levitte,
S., Kremer, S., Kwan, E., Weis, A.M., and Hermann, G.J. (2008). glo-3, a novel
Caenorhabditis elegans gene, is required for lysosome-related organelle
biogenesis. Genetics 180, 857871.
Rajagopal, A., Rao, A.U., Amigo, J., Tian, M., Upadhyay, S.K., Hall, C., Uhm,
S., Mathew, M.K., Fleming, M.D., Paw, B.H., et al. (2008). Haem homeostasis
Vatamaniuk, O.K., Bucher, E.A., Ward, J.T., and Rea, P.A. (2001). A new
pathway for heavy metal detoxification in animals. Phytochelatin synthase is
required for cadmium tolerance in Caenorhabditis elegans. J. Biol. Chem.
276, 2081720820.
Yoder, J.H., Chong, H., Guan, K.L., and Han, M. (2004). Modulation of KSR
activity in Caenorhabditis elegans by Zn ions, PAR-1 kinase and PP2A phosphatase. EMBO J. 23, 111119.
Zhang, Y., Grant, B., and Hirsh, D. (2001). RME-8, a conserved J-domain
protein, is required for endocytosis in Caenorhabditis elegans. Mol. Biol. Cell
12, 20112021.
Cell Metabolism
Article
Somatic Progenitor Cell Vulnerability
to Mitochondrial DNA Mutagenesis
Underlies Progeroid Phenotypes in Polg Mutator Mice
Kati J. Ahlqvist,1 Riikka H. Hamalainen,1 Shuichi Yatsuga,1 Marko Uutela,1 Mugen Terzioglu,6,7 Alexandra Gotz,1
Saara Forsstrom,1 Petri Salven,2 Alexandre Angers-Loustau,3 Outi H. Kopra,4,8 Henna Tyynismaa,1 Nils-Goran Larsson,6,7
Kirmo Wartiovaara,3,5 Tomas Prolla,9 Aleksandra Trifunovic,6,10 and Anu Suomalainen1,11,*
1Research
Programs Unit, Molecular Neurology, Biomedicum-Helsinki

Programs Unit, Molecular Cancer Biology, Biomedicum-Helsinki
3Developmental Biology, Institute of Biomedicine
4Haartman Institute, Department of Medical Genetics and Research Programs Unit, Molecular Medicine, and Neuroscience Center
5Institute of Biotechnology
University of Helsinki, 00290 Helsinki, Finland
6Department of Laboratory Medicine, Karolinska Institutet, S-14186 Stockholm, Sweden
7Max Planck Institute for Biology of Aging, 50931 Cologne, Germany
8Folkha
lsan Institute of Genetics, 00290 Helsinki, Finland
9University of Wisconsin, Department of Genetics, Madison, WI 53706, USA
10Cologne Excellence Cluster on Cellular Stress Responses in Aging-Associated Diseases, Cologne University, D-50674 Cologne, Germany
11Helsinki University Central Hospital, Department of Neurology, 00290 Helsinki, Finland
*Correspondence: anu.wartiovaara@helsinki.fi
DOI 10.1016/j.cmet.2011.11.012
2Research
SUMMARY
Somatic stem cell (SSC) dysfunction is typical for

different progeroid phenotypes in mice with genomic
DNA repair defects. MtDNA mutagenesis in mice with
defective Polg exonuclease activity also leads to
progeroid symptoms, by an unknown mechanism.
We found that Polg-Mutator mice had neural (NSC)
and hematopoietic progenitor (HPC) dysfunction
already from embryogenesis. NSC self-renewal was
decreased in vitro, and quiescent NSC amounts
were reduced in vivo. HPCs showed abnormal lineage differentiation leading to anemia and lymphopenia. N-acetyl-L-cysteine treatment rescued both
NSC and HPC abnormalities, suggesting that subtle
ROS/redox changes, induced by mtDNA mutagenesis, modulate SSC function. Our results show that
mtDNA mutagenesis affected SSC function early but
manifested as respiratory chain deficiency in nondividing tissues in old age. Deletor mice, having mtDNA
deletions in postmitotic cells and no progeria, had
normal SSCs. We propose that SSC compartment is
sensitive to mtDNA mutagenesis, and that mitochondrial dysfunction in SSCs can underlie progeroid
manifestations.
INTRODUCTION
Somatic stem cell (SSC) dysfunction has been proposed to lead
to decreased ability for tissue regeneration and aging (Schlessinger and Van Zant, 2001; Sharpless and DePinho, 2007). This
hypothesis was supported by mouse models with genomic

DNA repair defects, which showed premature aging-like phenotypes and severe SSC dysfunction (Frappart et al., 2005; Ito
et al., 2004; Narasimhaiah et al., 2005; Nijnik et al., 2007; Rossi
et al., 2007). Mice with Ku70 and Ku80 inactivation, leading to
defective nonhomologous DNA end-joining, had disrupted lymphopoiesis (Gu et al., 1997) as well as decreased amounts
of neural and hematopoietic progenitors (Narasimhaiah et al.,
2005; Rossi et al., 2007). A mouse model for trichothiodystrophy
(TTD), with a nuclear DNA helicase (xpd) mutation and defective
nucleotide excision repair, showed a premature aging-like phenotype (de Boer et al., 2002), with severe decrease in hematopoietic progenitors (Rossi et al., 2007). Furthermore, mice with
ataxia-teleangiectasia (Atm) inactivation had reduced cell-cycle
checkpoint activity in response to DNA damage and telomeric
instability, resulting in infertility and self-renewal defect of hematopoietic stem cells (HSCs) (Barlow et al., 1996; Ito et al., 2004).
The progressive bone marrow failure of these mice was associated with elevated reactive oxygen species (ROS), and the HSC
defect could be reversed by treating the animals with n-acetyl-Lcysteine (NAC), a compound with antioxidant capacity and an
effect on redox balance (Ito et al., 2004). Double inactivation of
Atm and telomerase RNA component (Terc) resulted in a proliferation and self-renewal defect of adult neural stem cells (NSCs)
in vitro, as well as decreased amount of dopaminergic neurons
in substantia nigra (Wong et al., 2003). These reports indicate
that SSCs are sensitive to defects of genomic DNA repair, and
suggest a role for ROS and/or redox status in regulating SSC
quiescence and proliferation. They also strongly suggest that
SSC dysfunction is intimately connected with premature aginglike symptoms in mice.
Increased mitochondrial DNA (mtDNA) mutagenesis can result
in progeroid phenotype in mice. Mutator mice show extensive mtDNA mutagenesis, with point mutations and complex
Cell Metabolism
mtDNA Mutagenesis Affects Somatic Stem Cells
rearrangements, because of an inactivated exonuclease function of the mitochondrial replicative DNA polymerase gamma
(POLG) (Ameur et al., 2011; Kujoth et al., 2005; Trifunovic
et al., 2004; Williams et al., 2010). Their phenotype mimics
premature aging, starting from 68 months of age, with progressive hair graying, alopecia, osteoporosis, general wasting, and
reduced fertility (Kujoth et al., 2005; Trifunovic et al., 2004). Their
life span is limited to 1315 months because of severe anemia,
with age-dependent decline in erythro- and lymphopoiesis,
recently suggested to be due to HSC dysfunction (Chen et al.,
2009; Norddahl et al., 2011). POLG forms the minimal mtDNA
replisome together with the mitochondrial single-stranded DNA
binding protein and a replicative helicase, Twinkle (Korhonen
et al., 2004; Spelbrink et al., 2001). If dominant mutant Twinkle
is overexpressed in mice, large-scale mtDNA deletions accumulate in postmitotic tissues (hence the Deletor mice), leading to
progressive late-onset mitochondrial myopathy at 12 months of
age and respiratory chain (RC)-deficient neurons, but normal life
span without progeroid features (Tyynismaa et al., 2005).
We asked whether SSC dysfunction could contribute to the
mtDNA mutagenesis-linked premature aging, and utilized the
two mouse models with mtDNA maintenance defects, Mutator
and Deletor, of which only the former showed a premature aging
phenotype. We report here NSC and HSC dysfunction in murine
mitochondrial progeria.
RESULTS
Old Mutator Neurons and Skeletal Muscle Show Mild
Respiratory Chain Deficiency
To characterize the consequences of high mtDNA mutation
load in postmitotic cells, we analyzed the activity and presence
of RC enzymes of Mutator mice at the time when they already
manifest progeroid phenotype, with hair loss, kyphosis, and
severe anemia (1314 months, Figures 1A and 1B). Histochemical enzyme assay of cytochrome c oxidase (COX, partially encoded by mtDNA) and succinate dehydrogenase (SDH, nuclear
encoded) showed no COX-negative, SDH-positive neurons in
the hippocampus or dentate nucleus (Figures 1C and 1D), or
any area with large neurons, including the cortex and cerebellum. However, as previously described (Tyynismaa et al.,
2005), Deletor mice at the age of 18 months showed occasional COX-negative neurons in all areas harboring large neurons, including hippocampus and dentate nucleus (Figures 1C
and 1D). Immunohistochemistry for RC complexes I (Figure 1F
and see Figure S1C available online) and II and IV (Figure S1C)
showed high content of these complexes, with only occasional
CI-low Purkinje cells in cerebellum, and were similar to WT in
the cortex even at 46 weeks of age (Figures S1A and S1B). By
western blot, Mutator brains showed a tendency to reduced
amounts of complex I (80% of WT) and COX (72% of WT) (Figure 1H). No signs of gliosis were seen in the Mutator brain
even at the age of 46 weeks, in SVZ, hippocampus, cortex,
and striatum (Figures S1F), and the number of neurons was
similar in Mutators and WT littermates, when counted from
hippocampal regions CA1 and dentate gyrus (Figures S1G
and S1H). Skeletal muscle of Mutator showed less than 1% of
COX-negative fibers, whereas Deletors showed 5% of total
muscle fibers to be COX negative (Figure 1G), and the heart
showed patches of COX-negative cardiomyocytes (data not

shown). The late onset of the Mutator mitochondrial myopathy
was further supported by the increase of FGF21 cytokine in
serum of the Mutators (Figure 1I), which we recently have shown
to be secreted from COX-negative muscle fibers in Deletor mice
and to correlate with the degree of RC deficiency (Tyynismaa
et al., 2010). Contrasting the otherwise good COX activity
in CNS, the neurogenic region of subventricular zone (SVZ)
(Figures 2A and 2B), and the metabolically active choroid plexus
(Figure 1E) had high numbers of COX-negative cells in the Mutators. SVZ was normal in Deletors (Figure 2B), but choroid plexus
showed occasional COX-negative cells (Figure 1E). Apoptosis
was not induced in the SVZ of 40- to 46-week-old Mutators
(Figure S1E).
These results indicate that even at the terminal phase of the
severe progeroid phenotype of mtDNA Mutator mice, their
neurons and muscle cells can maintain a well-preserved RC.
Mutators Show Decreased Neural Progenitors
in the Adult Brains
As the adult Mutator brains showed RC deficiency in SVZ,
we asked whether the neural progenitors were affected. We
analyzed the presence of quiescent neural progenitors by scoring
nestin positivity of adult Mutator brain (Figures 2C and 2D) and
compared it to Deletor and WT mice. Statistically significant
decrease in nestin-positive cells was seen in >40-week-old
Mutator SVZ when compared to WT littermates, whereas Deletors showed WT-like staining. Proliferative CDC47-positive cells
in Mutator (Figure 2E) and in Deletor SVZ were similar to WT in all
ages. Since the SVZ region is feeding neural progenitors to the
olfactory bulb throughout the life, we analyzed the number of
olfactory bulb periglomerular cells. The number of calbindinpositive periglomerular cells in Mutator olfactory bulb was similar
compared to WT littermates in different time points (Figure 2F).
These results show that the amount of nestin-positive NSCs is
reduced in the adult Mutator SVZ, but during the restricted lifetime of these mice, this has little effects on olfactory bulb neuronal
populations.
Our results show that even at an advanced stage of premature
aging syndrome, the central nervous system shows a minimal
phenotype, except for partial loss of quiescent nestin-positive
SVZ progenitors. Three subtypes of NSCs have been identified
from the SVZ region (Miller and Gauthier-Fisher, 2009; Suh
et al., 2009). Type B cells, resembling radial glial cells that serve
as NSCs during development, are slowly or infrequently dividing,
whereas type C cells are actively proliferating. Type B cells
express nestin and GFAP, whereas type C cells are both
nestin and GFAP negative. Our results strongly suggest that
the quiescent type B cells were decreased in the old Mutator
brains. However, in Deletors at the age when neurons were
clearly affected with RC deficiency (18 months), no indication
of abnormal SVZ was seen.
mtDNA Mutagenesis Leads to Decreased Self-Renewal
Capacity of Neural Stem Cells In Vitro, Attenuated with
NAC Treatment
To obtain understanding of the Mutator SSC phenotype, we
cultured NSCs from E11.5E15.5 Mutator embryos as freely
floating neurospheres. Spheres from E15.5 Mutator embryos
Cell Metabolism
Figure 1. Old Mutator Brain and Skeletal Muscle

Show Mild Respiratory Chain Deficiency
(A) At 13 months of age, Mutator mice manifest earlyonset progeria with alopecia and kyphosis as well as (B)
terminal-stage anemia. Hb values shown as mean SD.
WT (n = 2), 119.5 3.5; Mutator (n = 3), 52.67 24.1;
Deletor (n = 12), 114.1 12.7. (CG) Samples of Mutator,
13 months; WT, 18 months; Deletor, 18 months. Cryosections showing cytochrome c oxidase (COX, brown
staining) and succinate dehydrogenase (SDH, blue staining, visible only in cells with COX deficiency) histochemical
activity assay from (C) hippocampus (scale bar, 500 mm;
insets from CA2 region 10 mm); (D) cerebellar dentate
nucleus (scale bar, 20 mm). No COX-negative, SDH+ cells
were found in Mutator hippocampus or cerebellum at the
age of terminal manifestation, whereas Deletors at the age
of 18 months showed occasional COX-negative, SDH+
cells (blue) in all areas with large neurons. (E) In metabolically active choroid plexus, Mutators showed high
numbers of COX-negative cells (blue), similar to Deletors,
but also WT 18-month-old mice had occasional COXnegative cells (scale bars, 10 mm). (F) RC complex I
immunohistochemical staining of Purkinje cell layer.
Mutators show occasional CI-low cells, but typically excellent staining of CI (brown; counterstained with hematoxylin), whereas Deletors show high variability of CI
staining (scale bar, 40 mm). (G) COX-SDH activity assay of
frozen sections of skeletal muscle. Less than 1% of
Mutator skeletal muscle fibers were COX negative,
whereas in Deletors 5% of total muscle fibers were COX
negative (scale bar, 20 mm). (H) Mitochondrial extracts
from Mutator and WT brain at 39 weeks of age were
analyzed by western blot, and Mutator brain showed
slightly decreased amount of complexes I (80% of WT) and
IV (72% of WT). Complexes I, III, and IV were normalized
against nuclear-encoded complex II, and results are
shown as mean SD. Western blot analysis was done from
two separate protein extracts per animal, WT n = 1 and
Mutator n = 3. (I) Serum FGF21 concentration is elevated in
Mutators only at 3740 weeks of age (2425 w; p = 0.559,
3740 w; p = 0.026). See also Figure S1.
showed significantly higher mtDNA point mutation load

compared to the WT spheres (Figure 3A). At the protein level,
the increased mtDNA mutations resulted in slight reduction of
RC complexes I and IV in NSCs (Figures 3B and 3C), but complexes II and III did not differ from WT when quantified against
nuclear-encoded mitochondrial outer membrane protein, porin (Figure 3C). Also the cytochrome c oxidase activity was normal (COX
activity/mitochondrial matrix enzyme, citrate
synthase [CS] activity, WT NSCs [n = 4 lines]
0.66 0.31; Mutator NSCs [n = 3] 0.56 0.06).
We next examined whether mtDNA mutations
affected the self-renewal capacity of NSCs
in vitro. The primary proliferation characteristics
of early passage (<6) NSCs, measured by incorporation of BrdU and by flow cytometry,
showed that mtDNA mutations did not affect
the proliferative activity of NSCs (Figure 3D).
To test the NSCs ability to self-renew, cultures
were diluted to clonal density, and the singlecell ability to produce new neurospheres was monitored. Mutator NSCs formed 3-fold less spheres when compared to WT
NSCs (Figure 3E).
We next studied whether the decline in self-renewal ability was
affected by increasing redox buffer capacity and antioxidant
Cell Metabolism
Figure 2. Adult Mutator Brains Show COX-Negative Cells and Decreased Levels of Nestin-Positive NSCs in the SVZ, but the Amount of
Proliferative Cells in the SVZ, as well as Olfactory Bulb Interneurons, is Wild-Type Like
(A) Midsagittal view of adult mouse brains showing the areas of neurogenesis in red. Newly formed neural progenitors migrate from SVZ through rostral migratory
stream (RMS) to olfactory bulb (OB), where they give rise to interneurons maintaining the OB function.
(B) The SVZ (arrows) had high numbers of COX-negative cells in Mutators, visualized by COX-SDH activity from cryosections (Scale bar, 10 mm), but not in the
Deletors or WT mice.
(C) Old Mutator brains at 14 months show decreased number of nestin-positive NSCs in SVZ region, whereas in Deletors at 18 months the NSC numbers did not
differ from WT mice.
(D) The relative optical density of the nestin signal was analyzed from SVZ region (6001,200 cells per sample) and normalized against WT sample. Shown as
mean SD, ***p < 0.0001; animals per genotype, n = 2; scale bar, 20 mm.
(E) Mutators show WT-like amounts of CDC47-positive, proliferating cells in the SVZ area. Five hundred to seven hundred cells per sample were counted, CDC47positive cells shown as percentage of total cells, mean SD. Animals per genotype: 14 weeks, n = 2; 25 weeks, n = 3; >40 weeks, n = 2. Shown is a representative
picture of CDC47 immunostaining, proliferating cells seen in brown, from 46-week-old Mutator and WT brains. Scale bar, 20 mm.
(F) The number of olfactory bulb periglomerular interneurons, of which a subset is calbindin positive, was similar to WT in old Mutator brains. Three hundred to four
hundred cells per genotype were counted, calbindin-positive cells are shown as percentage of total cells, mean SD. Analyzed animals: 12 weeks, n = 1; 25 weeks,
n = 2; >40 weeks, n = 2. Shown is a representative picture of calbindin immunostaining from 25-week-old WT and Mutator olfactory bulb (calbindin, red; nuclei, blue).
action. Heterozygous females were fed with NAC throughout the

pregnancy and NSCs were extracted from E14.5 embryos, and
cultures were supplemented with NAC. Self-renewal ability of
treated and nontreated NSCs was monitored and showed that
NAC treatment restored the self-renewal ability of Mutator
NSCs but had no significant affect on the WT NSCs (Figure 3E).
When NAC was administrated only during the in vitro culture, and
not during embryonal life, it had no effect on self-renewal ability
of any NSCs (data not shown). During long-term culture, Mutator
NSCs also showed growth restriction: most lines were unable to
grow more than 10 passages, whereas WT neurospheres continued growth unaffected after 50 passages (Figure 3F). NAC
treatment was unable to reverse the growth restriction in Mutator
NSCs but instead caused high variability to WT NSCs growth
properties (Figure 3G). Next we examined NSCs multipotency
by inducing differentiation of single-cell-derived neurospheres
to progenitor lines. Mutator NSCs were able to produce morphologically similar neurons and astrocytes as WT (Figure 3H). Our
data show that Mutator NSCs have a WT-like capacity to differentiate to different cell types, but reduced self-renewal capacity,
which was attenuated by NAC supplementation. However, the
growth defect of long-term cultures was not affected by NAC.
These results suggest that the mechanism of NSC self-renewal
defect involves ROS/redox status.
Mutators Show Abnormal Fetal Erythropoiesis
and Disrupted Hematopoietic Progenitor Differentiation
in the Adult Bone Marrow
The Mutator NSC phenotype prompted us to ask whether other
stem/progenitor cell compartments were affected in the developing embryo. Adult Mutator mice have been shown to develop
severe progressive anemia after 6 months of age (Figure 1B)
(Chen et al., 2009; Trifunovic et al., 2004) with progressive
dysfunction of bone marrow HSCs (Chen et al., 2009; Norddahl
et al., 2011). We could replicate the previous findings from adult
bone marrow: despite their severe anemia, the adult Mutator
Cell Metabolism
Figure 3. Cultured Mutator NSCs Accumulate mtDNA Point Mutations and Show Decreased Self-Renewal Capacity and Growth Defect in
Long-Term Culture, which Can Be Attenuated by NAC Supplementation
(A) Cultured Mutator (n = 3) NSCs from E15.5 embryos showed increase in point mutation load in the cytochrome B gene of mtDNA compared to WT (n = 4) NSCs
(Mutator 13.7 mutations/10 kb; WT 0.2 mutations/10 kb).
(B) Mutator NSCs (n = 3) showed slight reduction of RC complexes I and IV analyzed by western blot.
(C) Mutator NSCs (n = 3) RC protein levels were compared to WT NSCs (n = 4), and the amount of RC subunits was normalized to nuclear-encoded mitochondrial
porin.
(D) Mutator NSCs (n = 4) showed BrdU incorporation similar to that of WT (n = 8) NSCs, indicating unaffected proliferation capability in flow cytometric analysis
when 120,000240,000 cells per genotype were analyzed. Analysis was performed with low-passage NSCs (p < 6).
(E) Mutator NSCs (n = 13) produced significantly less neurospheres than did WT (n = 18; ***p = < 0.0001) or Deletor (n = 6; ***p = 0.0005) NSCs in clonal expansion
analysis, indicating reduced self-renewal capacity. Mutator NSCs with NAC (n = 11) showed improved self-renewal capacity (***p < 0.0001), while the treatment
had no significant effect on the WT NSCs (n = 7; p = 0.321). Altogether 10,50027,000 cells per genotype were analyzed. Self-renewing cells are shown as
a percentage of total cells (mean SD): Mutator, 6.5% 3.4%, WT, 19.9% 7.7%, and Deletor, 16.5% 2.8%; Mutator with NAC, 15.8% 5.3%, and WT with
NAC, 23.3% 6.5%.
(F) Mutator NSCs (n = 3) showed growth restriction in continuous long-term culture compared to WT NSCs (n = 5), and the difference between WT and Mutator
lines became statistically significant by the ninth week of culture.
(G) The growth restriction of Mutator NSCs was partially attenuated by NAC supplementation: Mutator NSCs with NAC (n = 6) grew slightly better compared to
Mutator NSCs but could not reach the same passage levels as WT or WT-NAC (n = 6) lines.
(H) Cultured Mutator NSCs were able to differentiate into both neuronal (Tuj-1, green) and glial (GFAP, red) cells when induced to differentiate with 2% FBS.
Hoechst (blue) was used to stain nuclei.
bone marrow showed well-maintained total cellularity (results

shown as mean SD, WT [n = 2] 9.6 3 106 1.1 3 106 and
Mutator [n = 2] 7.2 3 106 2.3 3 106). Even at 40 weeks, all
cell populations expressing the erythroid markers typical for
each maturation stage were present, but the erythroid FACS
profiles were clearly abnormal, with lower mean intensity of
Ter119 signal compared to the WT bone marrow (Figures S2A
and S2B). Both the myeloid and B-lymphoid cell numbers were
decreased in the 40-week-old adult Mutator bone marrow
(Figure S2C).
The end stage severe HSC phenotype suggested that HSCs
could be affected already in early phase. We investigated
whether the hematopoietic system and erythropoiesis were

established during embryonic development. At E13E15.5, the
major site for mouse hematopoiesis is the liver (Ema and Nakauchi, 2000). We detected a decrease in the total amount of cells
in Mutator fetal liver extracts compared to WT littermates (Figure 4A). Based on FACS analysis, the major hematopoietic lineages, myeloid and lymphoid, were established normally in
Mutator embryos (Figure S2D). However, the relative proportions
of different erythroid precursors differed from WT already in the
early embryo: basophilic erythroblasts (CD71+,Ter119+ population) progressively decreased in the Mutators (Figure 4B), and
the most mature erythroid progenitors (Ter119+,CD71low)
Cell Metabolism
Figure 4. Mutators Show Abnormal Erythropoiesis and Lymphopoiesis Already

during Fetal Development, Attenuated by
NAC Supplementation
(A) Mutator fetal liver extracts showed slightly
decreased amount of total cells compared to WT
littermates at E15E15.5 (WT, n = 15; and Mutator,
n = 7; p = 0.099). NAC supplementation during the
fetal development increased the total fetal cell
count in both WT (n = 12) and Mutator (n = 10)
embryos, bringing the Mutator cell number to the
same level as the WT. (B) The amount of basophilic
erythroblasts (CD71+,Ter119+) progressively
decreased during the embryogenesis of Mutator
mice, whereas (C) the amount of more mature orthochromatophilic erythroblasts (Ter119+,
CD71low) increased. NAC supplementation during
the fetal development restored both cell populations in Mutator fetal liver to the WT level
(E13E13.5 WT, n = 9; Mutator, n = 8; E14E14.5
WT, n = 3; Mutator, n = 4; E15.5 WT, n = 15; WT
NAC, n = 12; Mutator, n = 7; and Mutator NAC,
n = 10). (D) The amount of B220-positive
B-lymphoid cells increased in Mutator fetal liver,
reaching statistical significance at E15.5 (p =
0.0092), while NAC supplementation during fetal
development was able to normalize the frequency of B220-positive cells (E13E13.5 WT, n = 9; Mutator, n = 5; E14E14.5 WT, n = 3; Mutator, n = 8; E15.5 WT,
n = 15; WT NAC, n = 12; Mutator, n = 7; and Mutator NAC, n = 10). Graphs in (B)(D) show the percentage of positive cells from total fetal liver cells (mean SD).
increased in the fetal liver at E15.5 (Figure 4C). The proerythroblast population (CD71+) was similar to WT (results shown as
mean SD, E13-E13.5: WT [n = 7] 9.1% 2.6% of total cells;
Mutator [n = 3] 8.1% 0.5%; E15.5: WT [n = 15] 8.4% 2.8%;
WT NAC [n = 12] 7.5% 1.8%, Mutator [n = 7] 8.9% 1.5%;
Mutator NAC [n = 10] 8.0% 0.9%).
We next tested whether we could affect the fetal hematopoietic phenotype with NAC supplementation. Heterozygous
females were fed with NAC throughout the pregnancy, and
hematopoietic cells were extracted from the liver of E15.5
embryos and subjected to FACS analysis immediately after
extraction. The total amount of cells in the fetal liver of Mutators
was restored to WT levels by NAC supplementation (Figure 4A).
Also the proportions of different erythroid progenitor populations
were normalized to WT level in the Mutator embryos treated with
NAC (Figures 4B and 4C). We then analyzed the frequency of
CD11b-positive myeloid and B220-positive B-lymphoid progenitors in fetal liver and found a progressive increase in the amount
of B-lymphoid cells in the Mutators compared to WT littermates
(Figure 4D). However, NAC-treated Mutator E15.5 embryos
showed B220+ cell amounts similar to those of WT embryos (Figure 4D). These results show that, similar to the NCS defect, the
fetal hematopoietic phenotype can be ameliorated by NAC
treatment.
To analyze the function of fetal HPCs, we analyzed their ability
to produce hematopoietic colonies when plated on methylcellulose. Mutator fetal liver HPCs were able to produce mixed myeloerythroid colonies (CFU-GEMM), whereas erythroid (BFU-E)
and granulocyte-macrophage colonies (CFU-GM) were modestly increased in number (Figure S2E). Forty-week-old Mutator
bone marrow formed fewer mixed myeloerythroid, erythroid, and
granulocyte-macrophage colonies than did WT (Figure S2F).
These results show that different lineages of the hematopoietic
system appear to be established during embryogenesis,

although distorted erythroid and lymphoid lineage differentiation
was evident already at the fetal stage. Furthermore, NAC supplementation was able to bring the frequencies of the erythroid and
lymphoid cell populations in Mutator fetal liver to WT level.
Deletors Show No mtDNA Deletion Formation in NSCs,
and Have Normal NSC Characteristics and
Hematopoiesis
Deletor mice, which accumulate mtDNA deletions and have RC
deficiency in their postmitotic tissues (Figures 1B1G) (Tyynismaa et al., 2005), displayed neither increased mtDNA point
mutation load (1.1/10 kb in Deletors, 0.7/10 kb in WT littermates)
nor mtDNA deletions (data not shown) in their NSCs, and were
also able to self-renew (Figure 3E) and differentiate normally
when compared to WT littermates. The mutant transgene was
expressed in NSCs 2-fold compared to the endogenous gene,
which is similar to the level that causes mtDNA deletions in the
Deletor skeletal muscle, brain, and heart. Hematopoiesis was
normal in the Deletors: their erythroid, lymphoid, and myeloid
progenitor amounts were similar to WT both in fetal liver and in
adult bone marrow from animals of 110 weeks of age (Figures
S2GS2J). These findings logically show that the mtDNA maintenance defect affects SSC function only when mtDNA mutation
load is increased in the stem/progenitor cells themselves.
DISCUSSION
We report here that the mtDNA Mutator mice with premature
aging-like syndrome have an embryonal-onset progressive
dysfunction of neural and hematopoietic progenitor cells,
leading to reduction of quiescent neural progenitors in the SVZ
and to severe anemia and lymphopenia in the adults. Previously,
Cell Metabolism
progeroid mice with dysfunctional nuclear DNA repair have been

reported to have similar defects. The NSCs of prematurely aging
Atm / Terc / mice showed reduced self-renewal capacity
in vitro and decreased amount of proliferating cells in adult
SVZ region in vivo (Wong et al., 2003). Ku70 / , Ku80 / , and
TTD mice showed a decreased amount of hematopoietic progenitors in their bone marrow and defective repopulation
activity of HSCs (Gu et al., 1997; Rossi et al., 2007). Similar to
Mutators, Atm / mice showed progressive anemia and
decreased amount of granulocyte-macrophage and lymphoid
progenitors in their bone marrow (Ito et al., 2004). The close similarities between the stem cell properties of other premature
aging models and Mutators emphasize the importance of both
nuclear and mtDNA integrity in tissue progenitor homeostasis
throughout life.
Parallel to their severe SSC phenotype, the Mutator mice
developed modest late-onset signs of mitochondrial RC deficiency in brain, skeletal muscle, and heart. Although an expected
consequence in mtDNA Mutator, the surprisingly late and mild
nature of RC defect indicates that postmitotic tissues can resist
random mtDNA mutagenesis well. We compared the findings in
the Mutators to another mouse model with mtDNA maintenance
defect, the Deletors. The latter accumulate large-scale mtDNA
deletions in postmitotic tissues, leading to late-onset RC deficiency, mimicking the muscle and brain phenotype of Mutators.
However, the Deletors have normal SSCs and no premature
aging phenotype. These findings support a crucial role of SSC
dysfunction in generating the progeroid phenotype of Mutators.
Previously, Mutator brains have been reported to have
RC deficiency, based on changes in macroscopic COX-SDH
activity, with no data on protein complexes or histology on
neuronal level (Ross et al., 2010; Vermulst et al., 2008). We did
not find abnormal COX or SDH activities in our old Mutator
brains, but only a marginal reduction of complex I and COX
protein amounts by western and immunohistochemical analyses. While COX-negative neurons could readily be found in
the Deletors, they did not exist in Mutators, suggesting that
neurons can compensate random mtDNA mutagenesis well.
This is supported by the finding that mice with postnatal disruption of mitochondrial transcription factor A in cortical and hippocampal neurons, leading to loss of mtDNA, survived for months
before developing RC dysfunction and neurodegeneration (Sorensen et al., 2001). Based on these results, we propose that the
Mutator mice have two separate manifestations: (1) a progeroid
syndrome as a consequence of early-onset dysfunction of
SSC pool, and (2) a late-onset mild RC deficiency in brain,
skeletal muscle, and heart as a consequence of progressive
RC dysfunction.
SSC maintenance is dependent on a fine balance of selfrenewal and differentiation, in the regulation of which physiological ROS has recently been implicated: low concentrations
of ROS, which did not cause detectable DNA, protein, or lipid
damage, could shift quiescent SSCs toward proliferation and/
or differentiation (Hamanaka and Chandel, 2010; Le Belle
et al., 2011; Shao et al., 2011; Yoneyama et al., 2010). Interestingly, the NSC and HPC defects of Mutator could be ameliorated by NAC treatment, suggesting involvement of aberrant
ROS signaling or redox status in Mutator somatic precursor
cell maintenance. No direct evidence of oxidative damage to
proteins, lipids, or nucleic acids was found in the tissues or

cells of Mutators (Kujoth et al., 2005; Trifunovic et al., 2005),
but their cardiomyopathy was attenuated by overexpression of
mitochondrial-targeted catalase, a ROS scavenger (Dai et al.,
2010). In Atm / and FoxO1/3/4L/L mice, even a small increase
in ROS production in HSCs could affect their quiescence and
disrupt their ability to reconstitute their niches (Ito et al., 2004;
Tothova et al., 2007). Similar to Mutators, the SSC phenotypes
of Atm / and FoxO1/3/4L/L mice could be complemented with
NAC. NAC is a thiol, which is readily deacetylated to form
L-cysteine, the rate-limiting amino acid for reduced glutathione
(GSH) synthesis. Therefore, NAC increases the amount of GSH
in cells, improving their redox buffering capacity, but also has
direct antioxidant effects (De Flora et al., 1995; Murphy et al.,
2011). We suggest that a subtle increase in mitochondrial O2
and/or change in the redox status, as a consequence of mtDNA
mutagenesis, is sufficient to modify signaling in SSCs and modify
their function to severely disrupt SSC homeostasis.
Our data of NSCs and HPCs suggest that the Mutators have
a widespread dysfunction of tissue progenitors. The Mutator
NSCs showed reduced self-renewal capacity in vitro and failure
to maintain proliferation in long-term culture. In adult brains, the
number of the quiescent nestin-positive type B neural progenitors of SVZ was decreased, a logical consequence of disability
to self-renew. Considering the early-onset NSC defect in vitro
in the Mutators, the adult brain phenotype was surprisingly
modest, suggesting that the life span of Mutators may be too
short to allow full manifestation of late-onset neurodegeneration.
No increase of apoptosis in the brain was seen, even in the
neurogenic regions, which, however, showed RC deficiency. In
Mutator brains, the cells of choroid plexus and SVZ neurogenic
area showed high numbers of COX-negative cells. Choroid
plexus sometimes shows COX-negative cells even in old WT
mice, which is suggested to be caused by high metabolic activity
of these cells. However, RC deficiency in potential neural progenitors raises the question of whether these cells preferentially
accumulate mtDNA mutations and whether their mtDNA half-life
is short and turnover rapid, making them dependent on POLG
function. These questions require further attention in the future.
In the adult bone marrow, at the time when the mice suffered
from severe anemia, different erythroid progenitor populations
were present, but considerable amounts of cells were in intermediate states of maturity, supporting previous results (Chen et al.,
2009; Norddahl et al., 2011). Furthermore, we show that Mutator
hematopoiesis is affected already during fetal development,
when different progenitor populations are all established, but
present in aberrant amounts. These data suggest that mtDNA
mutagenesis affects the quality of hematopoietic progenitors
more than quantity, possibly disturbing their normal quiescent
state, which has been suggested to be crucial for reconstitution
capacity and long-term sustenance of HSCs (Arai et al., 2004).
Although we found no signs of exhaustion of undifferentiated
progenitor cells in Mutator bone marrow, our results do suggest
deranged differentiation pattern and function of erythroid progenitor populations, which may result in increased apoptosis
of hematopoietic cells, as previously suggested (Norddahl
et al., 2011), leading to life-threatening anemia. We found mtDNA
mutagenesis also to affect lymphopoiesis, leading to lymphopenia in adults, supporting a previous report (Chen et al., 2009).
Cell Metabolism
Paradoxically, Mutator embryos showed increased B-lymphoid

cells in the fetal liver, whereas the adult bone marrow showed
a low number of B-lymphoid progenitors and decreased ability
to produce myeloerythroid colonies. These data further suggest that disrupted proliferation and/or differentiation of the
hematopoietic progenitor cells underlies Mutator anemia and
lymphopenia.
POLG is part of the minimal replisome of mtDNA together with
Twinkle helicase (Korhonen et al., 2004). Because of the necessity of these two proteins for mtDNA replication, their defects
could be assumed to affect tissues with high proliferative
activity, such as the skin, the gastrointestinal tract, or the blood.
However, in humans, recessive POLG and Twinkle mutations
result in progressive childhood, juvenile, or adult-onset neurodegenerative disorders, with muscle and brain RC deficiency, but
no anemia or lymphopenia, or no apparent signs of premature
aging (Suomalainen and Isohanni, 2010). Therefore, it is somewhat surprising that exonuclease deficiency of POLG does not
mimic human POLG disorders but affects the proliferative
tissues that have high demand for mtDNA replication, especially
the progenitor populations. This suggests that catalytic POLG
functions involved in mtDNA repair are important for the longlived brain mtDNAs (Weissman et al., 2007), whereas the proofreading POLG function and faithful mtDNA replication are
especially essential in proliferating progenitors. However, no
Mutator-like phenotype with extensive random mutagenesis of
mtDNA has been characterized in humans, and therefore comparison of Mutator results to human diseases should be done
with caution.
Recently, endurance exercise was reported to ameliorate
the premature aging symptoms in Mutator mice (Safdar et al.,
2011). Exercise has been shown to affect the vitality of muscle
and NSCs (Blackmore et al., 2009; Shefer et al., 2010; Wu
et al., 2008) and also to promote hematopoiesis (Baker et al.,
2011). It remains to be studied whether increased SSC fitness
underlies the effect of exercise in Mutators.
We show here that accumulation of mtDNA mutations in
Mutator mice already leads to alterations in two distinct stem/
progenitor cell compartments during fetal development. RC
deficiency in postmitotic tissues is a late phenomenon in these
mice, starting to manifest at the time when anemia and lymphopenia are restricting the life of these animals. Our results show
that the hematopoietic compartment is especially vulnerable to
mtDNA mutagenesis, whereas the nervous system is well preserved. Our results of NAC complementation suggest that the
mechanism by which mtDNA mutations affect the homeostasis
of tissue progenitors involves physiological ROS/redox signaling
in SSCs, repressing quiescence state. Our results strongly suggest that mtDNA mutagenesis in somatic stem/progenitor cells
can contribute to aging-related phenotypes.
cDNA with a dominant human disease mutation, leading to duplication of

13 amino acids (dup353365) in the linker region of mitochondrial Twinkle
helicase (Deletor mice [Tyynismaa et al., 2005]), were used. We used two
Mutator mouse colonies: one in C57Bl6, originating from T.P., and these
mice were used in NSC and HSPC experiments and in western analysis of
the RC complexes from NSCs. Embryos for HSPC and NSC extraction, as
well as adult brains for immunohistology and histochemistry, were obtained
from Mutator colony established by A.T. and N.-G.L.
Analysis of the Respiratory Chain Complexes by Western Blot
Whole-cell protein extraction from NSCs, mitochondrial protein enrichment
from brains, SDS-PAGE, and immunodetection of RC complexes were performed as previously described (Ylikallio et al., 2010). For protein detection,
we used monoclonal antibodies against the 39 kDa subunit of complex I,
core 2 or Rieske subunit of complex III, cox1p subunit of complex IV, 70 kDa
Ip subunit of complex II (Mitosciences) and porin (VDAC) (Calbiochem) as
a loading control. All samples were analyzed in duplicates.
Respiratory Chain Enzyme Activity Measurements and FGF21
The RC enzyme activity of complex IV (cytochrome c oxidase) and the enzyme
activity of CS were determined from whole-cell lysates of cultured NSCs as
previously described (Suomalainen et al., 1992). All samples were analyzed
as duplicates. FGF21 levels in serum of the mice were analyzed by ELISA,
with Quantikine mouse FGF-21 Immunoassay kit (R&D Systems, MF2100) as
described (Tyynismaa et al., 2010).
NSCs from Embryos
NSCs were extracted from the lateral ventricular wall of E11.5E15.5 mouse
brain as previously described (Piltti et al., 2006). Neurospheres were cultured
in serum-free F12 medium (Sigma-Aldrich) supplemented with FGF2 and EGF
(Sigma-Aldrich).
mtDNA Deletion and Point Mutation Analysis
DNA from low-passage (<10) neurospheres was extracted by standard
phenol-chloroform and ethanol precipitation method. MtDNA deletions were
analyzed as previously described (Tyynismaa et al., 2005). For point mutation
analysis, the CytB and control region areas of mtDNA were amplified as
described (Trifunovic et al., 2004) with a high-fidelity DNA polymerase (Phusion, New England BioLabs). PCR products were cloned into ZeroBlunt
Topo vector (Invitrogen) and sequenced. Altogether, 30 kb of sequence was
analyzed from each genotype from both CytB and control region using
Sequencher software (Genecodes). Single mutations were counted once, to
avoid repeated calculations of clonal mutation events, but this strategy
ignored same-site mutations.
Analysis of NSC Self-Renewal Capacity
Analysis of NSC self-renewal capacity was performed as previously described
(Piltti et al., 2006).
BrdU Incorporation Assay
To determine the proliferation rate, neurospheres were incubated in 10 mM
BrdU (BD PharMingen), stained with anti-BrdU and fluorescent secondary
antibody, and analyzed using FACSAria Cell Sorter.
See also the Supplemental Experimental Procedures.
N-Acetyl L-Cysteine Supplementation

Mice were given 1 mg/ml of NAC (Sigma-Aldrich) in drinking water throughout
the pregnancy. NSCs were extracted from E14.5 embryos, and culture
medium for NSCs was supplemented with 100 mM NAC. HSCs were extracted
from E15.5 embryos and subjected to FACS analysis immediately after the
extraction.
Mouse Models
All animal experimentation was approved by the Ethical Review Board of
Finland. Mice with a knockin inactivating mutation (D257A) in the exonuclease
domain of DNA polymerase gamma, PolG (Mutator mice [Kujoth et al., 2005;
Trifunovic et al., 2005]), and mice ubiquitously overexpressing mouse Twinkle
Differentiation Analysis of NSCs

Neurospheres originating from single cells were induced to differentiate using
2% FBS and poly-D-lysine-coated chamber slides; cells were stained with
anti-Tuj-1 and anti-GFAP, and presence of neurons and astrocytes was
analyzed.
Cell Metabolism
Brain Immunohistochemistry
Brains of Mutator and WT mice aged 12, 14, 25, 40, and 46 weeks and Deletors
of 72 weeks were collected and standard hematoxylin and eosin (TissueTek)
protocol was used to determine the morphology from paraffin sections. RC
complex subunits were visualized using rabbit anti-CI ND-1 (1:500), mouse
anti-CI NDUFS3 (1:100), mouse anti-CII 70 kDa subunit (1:200), and mouse
anti-CIV II (1:20). Mouse anti-CDC47 (1:200), rabbit anti-cleaved caspase 3
(1:100), mouse anti-nestin (1:200), and mouse anti-myelin basic protein
(1:200) were used to visualize proliferating cells, apoptotic cells, neural progenitors, and oligodendrocytes, respectively, together with the biotinylated
secondary antibodies from Vector ABC Elite kit (anti-mouse and anti-rabbit,
Vector), and developed using DAB Fast Kit (Sigma-Aldrich). Hematoxylin
(Tissue Tek) was used as a counterstain. COX-SDH immunohistological analysis was performed as previously described (Tyynismaa et al., 2005). Rabbit
anti-GFAP (1:400) and rabbit anti-calbindin (1:200) were used to visualize
astrocytes and subset of the periglomerular neurons of the olfactory bulb,
respectively, together with Alexa 594 chicken anti-rabbit secondary antibody.
Hoechst was used to stain nuclei. IgG from the primary antibody hosts was
used as a negative control for the staining. Analysis of optical density
(immunoperoxidase staining) or signal intensity (fluorescence staining) was
performed using ImageJ software.
(for K.J.A.). We thank Hanna Mikkola for advice and antibodies for HPC
FACS experiments and for very helpful discussions about hematopoietic
data. The authors are grateful for Anu Harju, Ilse Paetau, Tuula Manninen,
Markus Innila, Minna Teerijoki, Ivana Bratic, Jarmo Palm, Susanna Lauttia,
and Katja Piltti for advice and help in experimental procedures. Mika Hukkanen
is thanked for help in image analysis.
Isolation of Hematopoietic Cells from Fetal Liver and Adult Bone

Marrow
Embryos were harvested at E13E15.5, and liver was dissected out and kept
on ice in 1% FBS. Fetal liver was mechanically suspended using scissors and
1825 G needles, and the suspension was filtered through 40 mm filter (Dako)
and subjected to cell counting with crystal violet (Sigma-Aldrich). Adult mice
were sacrificed; their femoral and tibial bones were cut out and mechanically
cleaned. Bone marrow was flushed out with 1 ml of D-MEM supplemented
with penicillin-streptomycin and L-glutamine using insuline syringe, filtered
through 40 mm filter (Dako), and subjected to cell counting with crystal violet
(Sigma-Aldrich). The cells were immediately forwarded to analysis.
Baker, J.M., De Lisio, M., and Parise, G. (2011). Endurance exercise training
promotes medullary hematopoiesis. FASEB J. 25, 43484357.
Flow Cytometry Analysis of Hematopoietic Tissues

The hematopoietic lineages derived from the fetal livers and adult bone
marrows were analyzed using FACSAria Cell Sorter (Beckton Dickinson) and
fluorescence-conjugated antibodies against Sca1, CD71, CD11b, CD34,
Ter119 (eBioscience), c-kit, and B220. All antibodies except Ter119 were
from BD PharMingen, and antibodies were used 1 mg/1 3 106 cells for staining.
CD16/CD32 blocker (0.25 mg/1 3 106 cells) was used to inhibit possible nonspesific binding of the antibodies, and isotype controls (1 mg/1 3 106 cells)
were used to determine the background fluorescence of the labels. Thirty
thousand cells per cell line and antigen set were analyzed.
Hematopoietic Colony-Forming Assay
Presence of clonogenic hematopoietic progenitors in the fetal liver and
adult bone marrow was analyzed by culturing HPCs in MethoCult (StemCell
Technologies) for myeloerythroid progenitors according to the manufacturers
instructions. Duplicates were made from each sample.
Statistical significance between groups was determined by using Students
t test. Statistical analysis was performed when each group had at least three
samples. p < 0.05 was considered significant.
Supplemental Information includes two figures and Supplemental Experimental Procedures and can be found with this article online at doi:10.1016/
j.cmet.2011.11.012.
ACKNOWLEDGMENTS
We wish to thank the following funding agencies for support: Sigrid Juselius
Foundation, Jane and Aatos Erkko Foundation, the Academy of Finland,
University of Helsinki, Institute of Molecular Medicine Finland (FIMM), and
Biocentrum Helsinki (for A.S.); and Helsinki Biomedical Graduate School
Received: June 7, 2011

REFERENCES
Ameur, A., Stewart, J.B., Freyer, C., Hagstrom, E., Ingman, M., Larsson, N.G.,
and Gyllensten, U. (2011). Ultra-deep sequencing of mouse mitochondrial
DNA: mutational patterns and their origins. PLoS Genet. 7, e1002028. 10.
1371/journal.pgen.1002028.
Arai, F., Hirao, A., Ohmura, M., Sato, H., Matsuoka, S., Takubo, K., Ito, K., Koh,
G.Y., and Suda, T. (2004). Tie2/angiopoietin-1 signaling regulates hematopoietic stem cell quiescence in the bone marrow niche. Cell 118, 149161.
Barlow, C., Hirotsune, S., Paylor, R., Liyanage, M., Eckhaus, M., Collins, F.,
Shiloh, Y., Crawley, J.N., Ried, T., Tagle, D., and Wynshaw-Boris, A. (1996).
Atm-deficient mice: a paradigm of ataxia telangiectasia. Cell 86, 159171.
Blackmore, D.G., Golmohammadi, M.G., Large, B., Waters, M.J., and Rietze,
R.L. (2009). Exercise increases neural stem cell number in a growth hormonedependent manner, augmenting the regenerative response in aged mice.
Stem Cells 27, 20442052.
Chen, M.L., Logan, T.D., Hochberg, M.L., Shelat, S.G., Yu, X., Wilding, G.E.,
Tan, W., Kujoth, G.C., Prolla, T.A., Selak, M.A., et al. (2009). Erythroid
dysplasia, megaloblastic anemia, and impaired lymphopoiesis arising from
mitochondrial dysfunction. Blood 114, 40454053.
Dai, D.F., Chen, T., Wanagat, J., Laflamme, M., Marcinek, D.J., Emond, M.J.,
Ngo, C.P., Prolla, T.A., and Rabinovitch, P.S. (2010). Age-dependent cardiomyopathy in mitochondrial mutator mice is attenuated by overexpression of
catalase targeted to mitochondria. Aging Cell 9, 536544.
de Boer, J., Andressoo, J.O., de Wit, J., Huijmans, J., Beems, R.B., van Steeg,
H., Weeda, G., van der Horst, G.T., van Leeuwen, W., Themmen, A.P., et al.
(2002). Premature aging in mice deficient in DNA repair and transcription.
Science 296, 12761279.
De Flora, S., Cesarone, C.F., Balansky, R.M., Albini, A., DAgostini, F.,
Bennicelli, C., Bagnasco, M., Camoirano, A., Scatolini, L., Rovida, A., et al.
(1995). Chemopreventive properties and mechanisms of N-acetylcysteine.
The experimental background. J. Cell. Biochem. Suppl. 22, 3341.
Ema, H., and Nakauchi, H. (2000). Expansion of hematopoietic stem cells in the
developing liver of a mouse embryo. Blood 95, 22842288.
Frappart, P.O., Tong, W.M., Demuth, I., Radovanovic, I., Herceg, Z., Aguzzi, A.,
Digweed, M., and Wang, Z.Q. (2005). An essential function for NBS1 in the
prevention of ataxia and cerebellar defects. Nat. Med. 11, 538544.
Gu, Y., Seidl, K.J., Rathbun, G.A., Zhu, C., Manis, J.P., van der Stoep, N.,
Davidson, L., Cheng, H.L., Sekiguchi, J.M., Frank, K., et al. (1997). Growth
retardation and leaky SCID phenotype of Ku70-deficient mice. Immunity 7,
653665.
Hamanaka, R.B., and Chandel, N.S. (2010). Mitochondrial reactive oxygen
species regulate cellular signaling and dictate biological outcomes. Trends
Biochem. Sci. 35, 505513.
Ito, K., Hirao, A., Arai, F., Matsuoka, S., Takubo, K., Hamaguchi, I., Nomiyama,
K., Hosokawa, K., Sakurada, K., Nakagata, N., et al. (2004). Regulation of
oxidative stress by ATM is required for self-renewal of haematopoietic stem
cells. Nature 431, 9971002.
Korhonen, J.A., Pham, X.H., Pellegrini, M., and Falkenberg, M. (2004).
Reconstitution of a minimal mtDNA replisome in vitro. EMBO J. 23, 24232429.
Cell Metabolism
Kujoth, G.C., Hiona, A., Pugh, T.D., Someya, S., Panzer, K., Wohlgemuth, S.E.,
Hofer, T., Seo, A.Y., Sullivan, R., Jobling, W.A., et al. (2005). Mitochondrial DNA
mutations, oxidative stress, and apoptosis in mammalian aging. Science 309,
481484.
Le Belle, J.E., Orozco, N.M., Paucar, A.A., Saxe, J.P., Mottahedeh, J., Pyle,
A.D., Wu, H., and Kornblum, H.I. (2011). Proliferative neural stem cells have
high endogenous ROS levels that regulate self-renewal and neurogenesis in
a PI3K/Akt-dependant manner. Cell Stem Cell 8, 5971.
Miller, F.D., and Gauthier-Fisher, A. (2009). Home at last: neural stem cell
niches defined. Cell Stem Cell 4, 507510.
Murphy, M.P., Holmgren, A., Larsson, N.G., Halliwell, B., Chang, C.J.,
Kalyanaraman, B., Rhee, S.G., Thornalley, P.J., Partridge, L., Gems, D.,
et al. (2011). Unraveling the biological roles of reactive oxygen species. Cell
Metab. 13, 361366.
Narasimhaiah, R., Tuchman, A., Lin, S.L., and Naegele, J.R. (2005). Oxidative
damage and defective DNA repair is linked to apoptosis of migrating neurons
and progenitors during cerebral cortex development in Ku70-deficient mice.
Cereb. Cortex 15, 696707.
Nijnik, A., Woodbine, L., Marchetti, C., Dawson, S., Lambe, T., Liu, C.,
Rodrigues, N.P., Crockford, T.L., Cabuy, E., Vindigni, A., et al. (2007). DNA
repair is limiting for haematopoietic stem cells during ageing. Nature 447,
686690.
Norddahl, G.L., Pronk, C.J., Wahlestedt, M., Sten, G., Nygren, J.M., Ugale, A.,
Sigvardsson, M., and Bryder, D. (2011). Accumulating mitochondrial DNA
mutations drive premature hematopoietic aging phenotypes distinct from
physiological stem cell aging. Cell Stem Cell 8, 499510.
Piltti, K., Kerosuo, L., Hakanen, J., Eriksson, M., Angers-Loustau, A., Leppa,
S., Salminen, M., Sariola, H., and Wartiovaara, K. (2006). E6/E7 oncogenes
increase and tumor suppressors decrease the proportion of self-renewing
neural progenitor cells. Oncogene 25, 48804889.
Ross, J.M., Oberg, J., Brene, S., Coppotelli, G., Terzioglu, M., Pernold, K.,
Goiny, M., Sitnikov, R., Kehr, J., Trifunovic, A., et al. (2010). High brain lactate
is a hallmark of aging and caused by a shift in the lactate dehydrogenase A/B
ratio. Proc. Natl. Acad. Sci. USA 107, 2008720092.
Rossi, D.J., Bryder, D., Seita, J., Nussenzweig, A., Hoeijmakers, J., and
Weissman, I.L. (2007). Deficiencies in DNA damage repair limit the function
of haematopoietic stem cells with age. Nature 447, 725729.
Safdar, A., Bourgeois, J.M., Ogborn, D.I., Little, J.P., Hettinga, B.P., Akhtar,
M., Thompson, J.E., Melov, S., Mocellin, N.J., Kujoth, G.C., et al. (2011).
Endurance exercise rescues progeroid aging and induces systemic mitochondrial rejuvenation in mtDNA mutator mice. Proc. Natl. Acad. Sci. USA 108,
41354140.
Schlessinger, D., and Van Zant, G. (2001). Does functional depletion of stem
cells drive aging? Mech. Ageing Dev. 122, 15371553.
Shao, L., Li, H., Pazhanisamy, S.K., Meng, A., Wang, Y., and Zhou, D. (2011).
Reactive oxygen species and hematopoietic stem cell senescence. Int. J.
Hematol. 94, 2432.
Sharpless, N.E., and DePinho, R.A. (2007). How stem cells age and why this
makes us grow old. Nat. Rev. Mol. Cell Biol. 8, 703713.
DNA deletions associated with mutations in the gene encoding Twinkle,

a phage T7 gene 4-like protein localized in mitochondria. Nat. Genet. 28,
223231.
Suh, H., Deng, W., and Gage, F.H. (2009). Signaling in adult neurogenesis.
Annu. Rev. Cell Dev. Biol. 25, 253275.
Suomalainen, A., and Isohanni, P. (2010). Mitochondrial DNA depletion
syndromesmany genes, common mechanisms. Neuromuscul. Disord. 20,
429437.
Suomalainen, A., Majander, A., Haltia, M., Somer, H., Lonnqvist, J.,
Savontaus, M.L., and Peltonen, L. (1992). Multiple deletions of mitochondrial
DNA in several tissues of a patient with severe retarded depression and familial
progressive external ophthalmoplegia. J. Clin. Invest. 90, 6166.
Tothova, Z., Kollipara, R., Huntly, B.J., Lee, B.H., Castrillon, D.H., Cullen, D.E.,
McDowell, E.P., Lazo-Kallanian, S., Williams, I.R., Sears, C., et al. (2007).
FoxOs are critical mediators of hematopoietic stem cell resistance to physiologic oxidative stress. Cell 128, 325339.
Trifunovic, A., Wredenberg, A., Falkenberg, M., Spelbrink, J.N., Rovio, A.T.,
Bruder, C.E., Bohlooly, Y.M., Gidlof, S., Oldfors, A., Wibom, R., et al. (2004).
Premature ageing in mice expressing defective mitochondrial DNA polymerase. Nature 429, 417423.
Trifunovic, A., Hansson, A., Wredenberg, A., Rovio, A.T., Dufour, E.,
Khvorostov, I., Spelbrink, J.N., Wibom, R., Jacobs, H.T., and Larsson, N.G.
(2005). Somatic mtDNA mutations cause aging phenotypes without affecting
reactive oxygen species production. Proc. Natl. Acad. Sci. USA 102, 17993
17998.
Tyynismaa, H., Mjosund, K.P., Wanrooij, S., Lappalainen, I., Ylikallio, E.,
Jalanko, A., Spelbrink, J.N., Paetau, A., and Suomalainen, A. (2005). Mutant
mitochondrial helicase Twinkle causes multiple mtDNA deletions and a lateonset mitochondrial disease in mice. Proc. Natl. Acad. Sci. USA 102, 17687
17692.
Tyynismaa, H., Carroll, C.J., Raimundo, N., Ahola-Erkkila, S., Wenz, T.,
Ruhanen, H., Guse, K., Hemminki, A., Peltola-Mjosund, K.E., Tulkki, V., et al.
(2010). Mitochondrial myopathy induces a starvation-like response. Hum.
Mol. Genet. 19, 39483958.
Vermulst, M., Wanagat, J., Kujoth, G.C., Bielas, J.H., Rabinovitch, P.S., Prolla,
T.A., and Loeb, L.A. (2008). DNA deletions and clonal mutations drive premature aging in mitochondrial mutator mice. Nat. Genet. 40, 392394.
Weissman, L., de Souza-Pinto, N.C., Stevnsner, T., and Bohr, V.A. (2007). DNA
repair, mitochondria, and neurodegeneration. Neuroscience 145, 13181329.
Williams, S.L., Huang, J., Edwards, Y.J., Ulloa, R.H., Dillon, L.M., Prolla, T.A.,
Vance, J.M., Moraes, C.T., and Zuchner, S. (2010). The mtDNA mutation spectrum of the progeroid Polg mutator mouse includes abundant control region
multimers. Cell Metab. 12, 675682.
Wong, K.K., Maser, R.S., Bachoo, R.M., Menon, J., Carrasco, D.R., Gu, Y., Alt,
F.W., and DePinho, R.A. (2003). Telomere dysfunction and Atm deficiency
compromises organ homeostasis and accelerates ageing. Nature 421,
643648.
Shefer, G., Rauner, G., Yablonka-Reuveni, Z., and Benayahu, D. (2010).

Reduced satellite cell numbers and myogenic capacity in aging can be alleviated by endurance exercise. PLoS ONE 5, e13307. 10.1371/journal.pone.
0013307.
Wu, C.W., Chang, Y.T., Yu, L., Chen, H.I., Jen, C.J., Wu, S.Y., Lo, C.P., and
Kuo, Y.M. (2008). Exercise enhances the proliferation of neural stem cells
and neurite growth and survival of neuronal progenitor cells in dentate gyrus
of middle-aged mice. J. Appl. Physiol. 105, 15851594.
Sorensen, L., Ekstrand, M., Silva, J.P., Lindqvist, E., Xu, B., Rustin, P., Olson,
L., and Larsson, N.G. (2001). Late-onset corticohippocampal neurodepletion
attributable to catastrophic failure of oxidative phosphorylation in MILON
mice. J. Neurosci. 21, 80828090.
Ylikallio, E., Tyynismaa, H., Tsutsui, H., Ide, T., and Suomalainen, A. (2010).
High mitochondrial DNA copy number has detrimental effects in mice. Hum.
Mol. Genet. 19, 26952705.
Spelbrink, J.N., Li, F.Y., Tiranti, V., Nikali, K., Yuan, Q.P., Tariq, M., Wanrooij,
S., Garrido, N., Comi, G., Morandi, L., et al. (2001). Human mitochondrial
Yoneyama, M., Kawada, K., Gotoh, Y., Shiba, T., and Ogita, K. (2010).
Endogenous reactive oxygen species are essential for proliferation of neural
stem/progenitor cells. Neurochem. Int. 56, 740746.
Cell Metabolism
Article
Glucose-Independent Glutamine
Metabolism via TCA Cycling
for Proliferation and Survival in B Cells
Anne Le,1,* Andrew N. Lane,6,8,* Max Hamaker,2 Sminu Bose,1 Arvin Gouw,3 Joseph Barbi,4 Takashi Tsukamoto,5
Camilio J. Rojas,5 Barbara S. Slusher,5 Haixia Zhang,9 Lisa J. Zimmerman,9 Daniel C. Liebler,9 Robbert J.C. Slebos,9
Pawel K. Lorkiewicz,6 Richard M. Higashi,6,7 Teresa W.M. Fan,6,7,8,* and Chi V. Dang10,*
1Division
of Gastrointestinal and Liver Pathology, Department of Pathology

of Hematopathology, Department of Pathology
3Graduate Program in Pathobiology, Department of Pathology
4Division of Immunology and Hematopoiesis, Department of Oncology
5Neurology and Brain Science Institute
Johns Hopkins University School of Medicine, Baltimore, MD 21205, USA
6Center for Regulatory and Environmental Analytical Metabolomics
7Department of Chemistry
8J.G. Brown Cancer Center
University of Louisville, Louisville, KY 40202, USA
9Biochemistry Jim Ayers Institute for Precancer Detection and Diagnosis, Vanderbilt University Medical School, Nashville,
TN 37232-8575, USA
10Abramson Cancer Center of the University of Pennsylvania, Philadelphia, PA 19104, USA
*Correspondence: annele@jhmi.edu (A.L.), anlane01@louisville.edu (A.N.L.), twmfan@gmail.com (T.W.M.F.), dangvchi@exchange.upenn.edu
(C.V.D.)
DOI 10.1016/j.cmet.2011.12.009
2Division
SUMMARY
Because MYC plays a causal role in many human

cancers, including those with hypoxic and nutrientpoor tumor microenvironments, we have determined
the metabolic responses of a MYC-inducible human
Burkitt lymphoma model P493 cell line to aerobic
and hypoxic conditions, and to glucose deprivation,
using stable isotope-resolved metabolomics. Using
[U-13C]-glucose as the tracer, both glucose consumption and lactate production were increased by
MYC expression and hypoxia. Using [U-13C,15N]glutamine as the tracer, glutamine import and metabolism through the TCA cycle persisted under hypoxia, and glutamine contributed significantly to
citrate carbons. Under glucose deprivation, glutamine-derived fumarate, malate, and citrate were
significantly increased. Their 13C-labeling patterns
demonstrate an alternative energy-generating glutaminolysis pathway involving a glucose-independent
TCA cycle. The essential role of glutamine metabolism in cell survival and proliferation under hypoxia
and glucose deficiency makes them susceptible to
the glutaminase inhibitor BPTES and hence could
be targeted for cancer therapy.
INTRODUCTION
Mutations of genes involved in the tricarboxylic acid (TCA) cycle,
such as fumarate hydratase, succinate dehydrogenase, or isoci-
trate dehydrogenase 1 or 2, are causally linked to familial cancer

syndromes (Bensaad et al., 2006) or spontaneous low-grade
gliomas and acute myelogenous leukemia (Dang et al., 2010).
Together with the well-known Warburg effect (Koppenol et al.,
2011; Vander Heiden et al., 2009; Warburg, 1956; Warburg
et al., 1924) and numerous other alterations in the central metabolism of cancers (King et al., 2006; Samudio et al., 2009), these all
point to the important role of metabolism in the development of
many cancers and its therapeutic opportunities (Vander Heiden,
2011). Further, tumor suppressors such as p53 and oncogenes
such as MYC and RAS have been directly linked to regulating
metabolic pathways (Dang et al., 2009a; Telang et al., 2007) for
initiating tumorigenesis and tumor progression. These genetic
changes correlate with an increase in glucose consumption
and lactate production when MYC is high. However, solid tumors
contain regions that are both hypoxic and glucose depleted
(Schroeder et al., 2005), requiring alternative strategies for survival and/or proliferation.
Here, we report stable isotope-resolved metabolomic (SIRM)
studies of MYC-induced alterations in glucose and glutamine
metabolism, in which we find persistent, MYC-dependent hypoxic metabolism of glutamine, even in the absence of glucose.
Using NMR and MS, we have traced the fates of individual atoms
from uniformly 13C-labeled glucose ([U-13C]-Glc) or 13C,15Nlabeled glutamine ([U-13C,15N]-Gln) in human B cell P493 cells
carrying an inducible MYC vector. Overexpressed MYC resulted
in the concurrent conversion of glucose to lactate and the oxidation of glutamine via the TCA cycle. Under hypoxic conditions
with high MYC, a substantial fraction of the glucose consumed
was converted to excreted lactate, and glutamine continued to
be utilized by the TCA cycle, which was used for cell survival.
We further document the finding of a fully 13C-labeled citrate
isotopologue that contained carbons deriving completely from
Cell Metabolism
Glucose-Independent Glutamine-Dependent TCA Cycle
labeled glutamine, suggesting the existence of a glucose-independent TCA cycle. We also found under glucose-depleted
culture conditions that a glutamine-dependent and glucoseindependent TCA cycle may operate under both aerobic and
hypoxic conditions. Moreover, we observed an enhanced conversion of glutamine to glutathione under hypoxia; glutathione
is an important reducing agent for controlling the accumulation
of mitochondrial reduced oxygen species (ROS). Under moderate hypoxia, excess ROS is generated at complex II owing to
a mismatch between NADH production and terminal oxidase
activity (Wu et al., 2007). We therefore tested whether inhibition
of glutamine metabolism could induce oxidative stress under
hypoxia. We found that inhibition of glutaminase (GLS) by the
glutaminase-selective inhibitor BPTES (Robinson et al., 2007)
elevated ROS levels and diminished ATP levels in hypoxic cells.
In fact, we found that inhibition of glutaminase effectively kills
hypoxic cancer cells in vitro and delays tumor xenograft growth
in vivo.
RESULTS
Coexistence of Oxidative and Aerobic Glycolysis
Our genomic analysis of MYC target genes indicates that
the expression of genes involved in glycolysis and in mitochondrial respiration is coregulated by MYC (Dang, 2010; Kim
et al., 2007, 2008; Li et al., 2005). We thus determined the
metabolic consequences of MYC activation in a model cell line
(P493) of human Burkitt lymphoma grown in uniformly labeled
[U-13C]-Glc under aerobic (21% O2) or hypoxic (1% O2) conditions. Although the levels of metabolites at steady state are
the result of the balance between production and consumption,
the use of SIRM enabled us to determine not only steady levels of
metabolites but also the isotopomer and isotopologue distributions of metabolites derived from 13C-labeled glucose for reconstruction of metabolic pathways. Time course isotopomer data
on extracellular metabolites further provided flux measurement
for substrate import and product release (Figure 1A, see Figures S2C and S3B and Table S1 available online). P493 cells
contain a tetracycline-repressible MYC construct, such that
tetracycline withdrawal results in rapid induction of MYC and
tetracycline treatment results in MYC suppression (Figure S1A).
Induction of MYC resulted in an increase of 13C-glucose consumption and 13C-lactate production, which were further accentuated by hypoxia (Figure 1A). NMR analysis of 13C-labeled
metabolites derived from [U-13C]-Glc in cell extracts also corroborated the finding that overexpressed MYC resulted in lactic
fermentation even under aerobic conditions (Figure 1B). The
same cell extracts were further analyzed by GC-MS to quantify
glucose-derived 13C isotopologues of lactate (lactate with different number of 13C atoms) (Figure 1D). Levels of the m+3 isotopologue of lactate (triply 13C-labeled lactate or 13C3-lactate)
derived from glucose shown with MYC ON or OFF in aerobic
(A) or hypoxic (H) conditions (Figure 1D) also support the ability
of MYC to increase aerobic glycolysis (Figure 1C). Under hypoxic
conditions, glucose-derived lactate was increased but was
less dependent on MYC. The production of lactate (Table S1)
accounts for only part of the glucose consumed. Although
a complete carbon inventory has not been achieved, we estimate that a significant fraction of the glucose enters new
biomass, which with respiration is associated with carbon loss

as CO2.
As shown in Figure 2, glucose-derived TCA cycle intermediates under aerobic condition displayed a dependence on
MYC, such that the doubly 13C-labeled isotopologue of citrate
(13C2-citrate), succinate, fumarate, and malate (m+2 forms,
circled red) increased when MYC was ON. Glucose-derived
a-ketoglutarate (m+2), which was at very low cellular concentration, also demonstrated a dependence on MYC; it was only
detectable when MYC was ON. Hypoxia decreased the MYCinduced conversion of glucose to citrate (m+2) and to other
m+2 isotopologues of TCA cycle intermediates (malate, fumarate, and succinate), but these activities were independent of
MYC expression (MYC ON-H versus MYC OFF-H). In addition
to the synthesis of 13C2-citrate, there was a significant production of 13C5-citrate (m+5, circled green) with MYC ON under
aerobic conditions. This citrate isotopologue could be produced
from m+2 acetyl-CoA plus m+3 oxaloacetate (product of pyruvate carboxylation, green arrow and circles, Figure 2) (Fan
et al., 2010), and its level appeared to be attenuated by hypoxia
and the absence of ectopic MYC (Tet treatment). It is notable that
a large fraction (up to 70%) of these TCA metabolites (m+0) were
not derived from the labeled glucose, suggesting an alternative
source and/or prolonged half-lives of these metabolites that
could have existed prior to the administration of labeled glucose.
Persistence of Glutamine Oxidation via the TCA Cycle
under Hypoxia
The attenuation of glucose entry into the TCA cycle under
hypoxia (Figure 2) is consistent with the hypoxia inducible factor
(HIF)-mediated diversion of pyruvate to lactate (away from
acetyl-CoA) through the induction of LDHA (which increases
the relative flux from pyruvate to lactate) and PDK1 (which
decreases the relative flux from pyruvate to acetyl-CoA) (Kim
et al., 2006). Because it was previously documented that MYC
induces glutamine metabolism under aerobic conditions (Gao
et al., 2009; Wise et al., 2008), we sought to determine whether
glutamine entry into the TCA cycle would also be compromised
by hypoxia.
Using [U-13C,15N]-Gln (13C515N2-Gln) as the tracer with SIRM
analysis, the fates of glutamine as a function of MYC induction
and oxygen availability were determined (Figures 3B3D). Glutamine is transported into cells by transporters, such as the direct
targets of MYC SLC1A5 or ASCT2 (Figure S1C), and then converted to glutamate by glutaminase (GLS, kidney isoform, which
is also a target of MYC) (Figure S1D). 13C515N2-Gln (m+7) is converted by glutaminase into 13C515N-Glu (m+6) plus 15NH4+
(Figure 3A). NMR studies of biological replicate experiments
revealed a MYC-dependent conversion of labeled glutamine to
glutamate, which unexpectedly persisted in hypoxia (Figure 3B).
This result was corroborated by the GC-MS analysis of the same
set of polar extracts. Intracellular glutamine was converted to
glutamate (m+6 isotopologue or 13C515N1-Glu, Figure 3D) in a
MYC-dependent fashion that persisted under hypoxia. A large
fraction of the m+5 glutamate isotopologue was also present
and displayed a similar MYC and hypoxia dependence as
the m+6 isotopologue. The m+5 isotopologue of glutamate
was largely 13C5-Glu as determined by high-resolution FTICR-MS (Figure S1E), which resolved the neutron mass from
Cell Metabolism
Figure 1. MYC Induces Aerobic Glycolysis that Is Heightened in Hypoxia

(A) Time course of glucose consumption and lactate secretion into the medium. P493 cells were treated with 0.1 mg/ml tetracycline (MYC OFF) or without
tetracycline (MYC ON) for 48 hr in hypoxic (H) or aerobic (A) conditions and were grown in RPMI containing 10 mM [U-13C] glucose (13C-Glc6) for 24 hr. Media
metabolites and 13C enrichments were measured by 1D 1H NMR. The metabolite amounts are expressed as mmole. Each time data point is an average of
duplicate samples. The rates were calculated from linear regression of the time courses and normalized to cell mass as mmole/h/g cells. Open symbols are
13
C lactate, filled symbols are 13C glucose. Filled red circles, MYC ON aerobic; blue diamonds, MYC OFF aerobic; green squares, MYC ON hypoxic; black
triangles, MYC OFF hypoxic.
(B) 1-D 1H{13C} HSQC NMR spectra of cell extracts of P493 MYC ON versus MYC OFF. Intracellular 13C-lactate synthesized from 13C glucose was attenuated
when MYC is off, which was also observed by GC-MS analysis of the same extracts (D). NMR spectra were recorded at 800 MHz and 20 C. Each 1H peak arose
from protons directly attached to 13C, and the peak assignment denotes the 13C-carbon. Thus, the peak intensity reflects 13C abundance of the attached carbon.
(C) Diagram of 13C-labeling patterns of glycolytic products with 13C6-Glc as tracer. Glucose with carbons (red circles) labeled at all six positions (blue) generates
13
C3-lactate via pyruvate (three carbons), which also produces 13C2-acetyl CoA.
(D) GC-MS analysis of MYC and/or hypoxia effect on 13C-Glc6 metabolism to different 13C isotopologues of lactate. Overexpressed MYC enhanced lactate
production (most notably 13C3-lactate; m+3), and hypoxia further increased labeled incorporation into lactate. Each value is an average of duplicate samples. The
error bars represent SEM.
13
C and 15N. 13C5-Glu should be a transamination product of

C-labeled glutamine-derived a-ketoglutarate (a-KG) with unlabeled nitrogen sources (Figure 3A). a-ketoglutarate levels also
tracked MYC expression (Figure 4). Furthermore, the dependence of glutaminase activity on MYC expression measured in
extracts suggests that the intracellular conversion of labeled
glutamine to glutamate is at least partly regulated by GLS1
activity in response to MYC (Figure S2A). This was further supported by the higher level of ammonium ionsthe other product
of the glutaminase reactionthat were released into the medium
under MYC ON conditions (Figure S2B).
Levels of fully labeled glutamine in the media were measured to determine the rate of consumption of glutamine. Glutamine consumption rates were in the following order: MYC ON
aerobic zMYC ON hypoxic > MYC OFF hypoxic > MYC OFF
aerobic (Figure 3C, Figure S2C). The m+5 and m+6 isotopologues of glutamate were also present in the medium (Figure 3C),
13
which reflects exchange of intracellular glutamine-derived glutamate for other amino acids such as cystine (see below).
As depicted in Figure 3A, labeled glutamine catabolism by
glutaminase led to the production of 13C5-aKG, which can enter
the TCA cycle for further oxidation. As shown in Figure 4, the synthesis of 13C4-succinate, -fumarate, and -malate (m+4) is consistent with the oxidation of 13C5-aKG via the forward reactions of
the TCA cycle (red circles, Figure S2D). These labeled TCA intermediates all responded to MYC status by increasing 60%
to >100% when MYC was ON, regardless of O2 availability (Figure 4). The levels of 13C4-citrate, which is synthesized in the
second turn from 13C4-oxaloacetate (OAA, derived from labeled
glutamine in the first turn) and acetyl-CoA (from unlabeled glucose or other unlabeled sources) by citrate synthase (CS), also
responded to MYC expression under both aerobic and hypoxic
conditions (Figure 4). These results show that MYC can drive
glutamine metabolism around the TCA cycle even under hypoxia.
Cell Metabolism
Figure 2. Glucose Entry into the TCA Cycle Is Induced by MYC and Suppressed by Hypoxia
The cycle reactions are depicted without or with pyruvate carboxylation (green arrow), and the 13C isotopomer patterns are the result of one cycle turn. The
isotopologue distributions were determined by GC-MS. The incorporation of 13C atoms from 13C6-Glc into citrate, a-ketoglutarate (a-KG), succinate, fumarate,
and malate are denoted as m+n, where n is the number of 13C atoms. The m+5 (13C5)-citrate (green circle) is produced by condensation of m+2 acetyl-CoA with
m+3 OAA (from pyruvate carboxylation), while m+2 (13C2)-citrate is synthesized without input of pyruvate carboxylation. Red and green circles, 13C atoms derived
from 13C6-Glc without or with pyruvate carboxylation, respectively. GPT2, glutamate-pyruvate transaminase; PC, pyruvate carboxylase; CO2 indicates where
carbon dioxide is released. A, aerobic; H, hypoxic. The error bars represent SEM.
Glucose-Independent Oxidation of Glutamine

for Survival and Proliferation
In addition to the production of the 13C4-citrate isotopologue,
there was a significant formation of other citrate isotopologues,
e.g., 13C3- (m+3), 13C5- (m+5), and 13C6-citrate (m+6) (Figure 4).
These labeled species indicate that the production of labeled
acetyl-CoA and OAA isotopologues from the glutamine tracer
are using pathways external to the TCA cycle. Figure S2D
depicts the pathways that can lead to the synthesis of 13C3-,
13
C5-, and 13C6-citrate. These include the cytoplasmic ATPcitrate lyase (ACL) plus malic enzyme (ME) reactions, which
produce, respectively, 13C4/13C2-OAA and 13C3/13C2-pyruvate;
the latter yields 13C2/13C1-acetyl-CoA via pyruvate dehydrogenase (PDH). Condensation of the labeled OAA and acetyl-CoA
species derived from the ACL-ME pathway, glycolysis, and the
TCA cycle produces 13C3 to 13C6-citrate (light blue circles, Figure S2D). In particular, the presence of the 13C6-citrate isotopo-
logue unambiguously confirmed the production of fully labeled

acetyl-CoA from the glutamine tracer via the ACL-ME-PDH
pathway. Labeled acetyl-CoA production from the ACL reaction
is further supported by 13C label incorporation into lipids such
as triacylglycerides (TAGs) and phosphatidylcholines (PCs)
(Figure S2E).
The 13C3- and 13C5-citrate isotopologues can also be formed
from the ACL-ME plus pyruvate carboxylase (PC) reactions
(green circles, Figure S2D). Pyruvate carboxylase is active in
P493 cells, as evidenced from the labeled glucose tracer experiment described above (Figure 2). The operation of the ACLME1-PDH and ACL-ME1-PC pathways is further corroborated
by the production of 13C3-succinate, -fumarate, and -malate,
which cannot be formed from labeled glutamine via the TCA
cycle activity alone. Finally, the production of 13C5-citrate from
the glutamine tracer can also be explained by reductive carboxylation of aKG to form citrate (orange circles, Figure S2D),
Cell Metabolism
B
D
Figure 3. MYC Induces Glutamine to Glutamate Conversion that Persists in Hypoxia

(A) Diagram of 13C labeling patterns products with 13C5,15N2-Gln as tracer. Glutamine is labeled at all five carbon and both nitrogen atoms. Glutaminase (GLS)
activity produces glutamate 13C15N1 with the loss of the amido nitrogen as ammonia. Red and green circles, respective 13C and 15N labeling of glutamate from the
glutaminase reaction; light blue circles, 13C labeling of glutamate from the transamination of a-ketoglutarate; black, unlabeled N; GS, glutamine synthetase.
(B) 1-D 1H{13C} HSQC NMR spectra of cell extracts of P493 MYC ON versus MYC OFF. Intracellular 13C-4-glutamate production from glutamine was reduced
when MYC is off in aerobic condition. The conversion of glutamine to glutamate in hypoxia was as high as aerobic condition regardless of MYC expression. NMR
spectra were recorded at 800 MHz and 20 C. Each 1H peak arose from protons directly attached to 13C, and the peak assignment denotes the 13C-carbon. The
peak intensity reflects 13C abundance of the attached carbon.
(C) Glutamine levels in the media. P493 cells were treated with 0.1 mg/ml tetracycline (MYC OFF) or without tetracycline (MYC ON) for 48 hr in hypoxic (H) or
aerobic (A) conditions and were grown in 13C5,15N2-Gln (m+7) medium in the presence or absence of tetracycline for 24 hr. Media metabolites were analyzed by
GC-MS. Under aerobic conditions, glutamine consumption was decreased when MYC is off. Under hypoxia, glutamine consumption was decreased. The
metabolite concentration was expressed as mmole/g protein dry weight.
(D) GC-MS analysis of intracellular conversion of 13C5,15N2-glutamine to glutamate. The m+5 glutamate was primarily 13C5-glutamate (cf. Figure S1E) with no
15
N label, which is derived from 13C5-aKG via transamination. The data shown in (B)(D) were performed three times. The error bars represent SEM.
a reversal of the citrate to aKG reaction catalyzed by aconitase

and IDH as recently reported in other cells (Metallo et al., 2011;
Mullen et al., 2011; Wise et al., 2011; Yoo et al., 2008) and driven
by the hydrolysis of ATP via citrate lyase and ACC. There is abundant CO2/HCO3 in cell culture for reductive carboxylation and
presumably in tissue from a number of decarboxylation reactions. However, the relative proportion of the m+5 versus m+3,
m+4, and m+6 species, which are characteristic of the forward
reactions in the Krebs cycle plus pyruvate carboxylase activity,
indicates that reductive carboxylation is not the major pathway
in P493 cells, especially under aerobic conditions in which the
ratio of aKG to citrate concentration was very low (Figure 4);
a high ratio is important for driving this thermodynamically uphill
reaction.
We were also intrigued by the apparent upregulation of such

TCA cycle-mediated glutamine metabolism by MYC and its
persistence under hypoxia (Figure 4). Such a glucose-independent TCA cycle activity would be advantageous for cancer cells
subjected to glucose deficiency and/or hypoxia in the tumor
microenvironment. Thus, we next determined whether the glutamine-mediated TCA cycle can operate in the absence of glucose
and whether glutamine metabolism alone can sustain cell growth
and survival.
P493 cells were grown in the absence of glucose using the
tracer [U-13C,15N]-Gln to determine whether TCA cycle intermediates could be derived solely from glutamine (Figure 5). In the
absence of glucose, the P493 cells completed one doubling in
3 days under aerobic conditions with MYC ON compared with
Cell Metabolism
Figure 4. P493 Cells Were Grown in 13C5,15N2-Glutamine Medium with Glucose

MYC induces glutamine entry into the TCA cycle, which persists in hypoxia. 13C5,15N2 glutamine enters the Krebs cycle to produce a-ketoglutarate, succinate,
fumarate, and malate. Citrate can be generated by reductive carboxylation (RedCarb) of glutamine-derived a-ketoglutarate (m+5) (orange circles) or via the
forward reactions of the cycle. Malic enzyme (ME) catalyzes the oxidative decarboxylation of malate to pyruvate (light blue). This pyruvate pool (m+3) can be
a precursor for both m+2 acetyl-CoA and m+3 OAA (from pyruvate carboxylation, green circles) to make m+5 citrate. The citrate m+6 isotopologue could only
arise from glutamine-derived m+4 oxaloacetate (from forward cycle reactions, red circles) and m+2 acetyl-CoA (mediated by malic enzyme). The isotopologue
distributions were determined by GC-MS. Each value is an average of duplicate samples. The data shown were performed three times. GPT2, glutamatepyruvate transaminase; ME, malic enzyme; PC, pyruvate carboxylase; GLS, glutaminase; CO2 indicates where carbon dioxide is released. A, aerobic; H, hypoxic.
The error bars represent SEM.
doubling every 34 2 hr in the presence of glucose (Figure S3A). The cells consumed glutamine (Figure S3B) to produce
13
C5-aKG (m+5) in a MYC-dependent fashion (circled red, Figure 5). The continued functioning of the TCA cycle under
glucose-deprived conditions was identified by the production
of various isotopologues of fumarate, malate, and aspartate,
particularly the 13C4-isotopologues (m+4, Figure 5). However,
these labeled isotopologues accumulated to much higher levels
(>100-fold for 13C4-Asp) than under glucose-replete conditions
(compare Figures 4 and 5 or Figure 1B and Figure S3C). This
could result from a lower supply of acetyl-CoA under glucosedeprived conditions such that excess glutamine-derived OAA
was transaminated to form aspartate. This is also consistent
with the lower levels of labeled citrate and aKG isotopologues
in glucose-deprived cells than those found in glucose-replete
cells (Figures 4 and 5). In the absence of glucose, it is also
notable that 13C incorporation from the glutamine tracer into all
TCA cycle intermediates decreased under hypoxia but increased
with MYC OFF (Figure 5). Furthermore, the 13C alanine isotopologues (e.g., 13C3-Ala or m+3 in Figure 5) showed the opposite
behavior in response to MYC expression, regardless of the O2

conditions. The significant buildup of labeled alanine was in
contrast to a small production of lactate from glutamine (Figure S3C), which again argues against the operation of the canonical glutamine to lactate pathway in P493 cells.
The relatively low accumulation of labeled TCA metabolites
with MYC ON under glucose deprivation could be caused by
a combination of limited TCA cycling and the demands for cell
proliferation and cell maintenance. The same argument could
also apply to aerobic versus hypoxic conditions. In Figure S3A,
the highest proliferation rate was observed under aerobic conditions with MYC ON, followed by MYC OFF under aerobic
conditions, whereas under hypoxia there was little proliferation
regardless of the MYC status. With a slowdown of TCA cycling
due to glucose deficiency, a higher consumption rate of labeled
TCA intermediates for cell proliferation could deplete these
metabolites for both MYC ON and MYC OFF, but more so for
MYC ON than MYC OFF and with more depletion under
aerobic than hypoxic conditions (Figure 5). Diversion of glutamine for maintenance purposes, e.g., glutathione synthesis for
Cell Metabolism
Figure 5. Glutamine-Driven Glucose-Independent TCA Cycle

P493 cells were grown in 13C5,15N2-glutamine medium without glucose. The isotopologue distributions were determined by GC-MS. Under glucose-deprived
conditions, the m+5 a-ketoglutarate from labeled glutamine was produced in a MYC-dependent fashion. Labeled glutamine was also incorporated into m+4
isotopologues of fumarate and malate downstream of a-ketoglutarate. These isotopologues decreased under hypoxia but not with decreased MYC. The m+3
alanine isotopologue, which is derived from the transamination of glutamine-derived m+3 pyruvate (via malic enzyme), was highly elevated under high MYC and
aerobic conditions. Each value is an average of duplicate samples. Colored circles denote the same sets of reactions as in Figure 4. GPT2, glutamate-pyruvate
transaminase; ME, malic enzyme; PC, pyruvate carboxylase; GLS, glutaminase; CO2 indicates where carbon dioxide is released. A, aerobic; H, hypoxic. The error
bars represent SEM.
ROS detoxification (inferred by increased ROS production with

BPTES inhibition of glutaminolysis, Figure S4A) could also lead
to less flux through the TCA cycle. This is consistent with a higher
buildup of glutamine-derived glutathione with aerobic and MYC
ON conditions (Figure S3C).
When TCA cycling is faster, as in glucose-replete cells, the
production rate for the labeled TCA intermediates may be higher
than their consumption rate, leading to a higher buildup of these
labeled metabolites when MYC is overexpressed (Figure 4). The
bottleneck in TCA cycling could also contribute to the significant
buildup of 13C3-Ala (m+3) alanine under MYC ON and glucose
deprivation (Figure 5) via excess production of glutaminederived pyruvate (by way of the ACL-ME1 pathway, Figure S4),
which is transaminated to form alanine via glutamic pyruvic
transaminase 1 or 2. Again, this did not occur under glucosereplete conditions, in which glucose, not glutamine, was the
main source of alanine production (Figure 2). In the absence
of glucose and under hypoxia, P493 cells did not proliferate
but continued to consume glutamine and remained viable
(Figure S3A).
Under aerobic conditions with MYC ON, the percentages of
viable (78% versus 76%) and proliferating (R2, 30.6% versus
30.3%) cell populations were similar between glucose-replete

(Figure S6) and -deplete conditions (Figure S7). This can be
mediated by the ability of glutamine metabolism to alleviate
oxidative stress (Figure S4A) and to support cell bioenergetics
(Figure S4B). These data show that when driven by MYC, glutamine plays a crucial role for both cell survival and proliferation
under glucose deprivation.
Thus, the above observations confirm a reprogrammed glutamine-dependent TCA cycle that functions in the absence of
glucose. Figure S4C outlines the pathways by which 13C carbons
of glutamine are converted to labeled acetyl-CoA and reenter
the TCA cycle. The significant presence of 13C5- (m+5) and
13
C6-citrate (m+6, Figure 5) is consistent with an ACL-MEmediated production of acetyl-CoA and citrate solely from glutamine. The 13C labeling of lipid acyl chains, although at relatively
low levels (Figure S4D), and the increased labeling under aerobic
with MYC ON confirm the activity of ACL and are consistent with
the higher rate of proliferation under these conditions (Figure S3A). However, since there was a significant level of unlabeled and 13C4-citrate (m+4, Figure 5) under glucose deprivation, there must also be a source(s) of unlabeled acetyl-CoA
that contributes to the continued operation of the TCA cycle.
Cell Metabolism
Figure 6. Glutamine Contributes to Glutathione Synthesis and Redox Homeostasis

(A) Diagram of glutamine metabolism to glutathione. Glutamine, when converted to glutamate, is involved in the import of cystine through the antiporter SLC7A11.
Cystine is converted to cysteine for the synthesis of glutathione, which also requires glycine and glutamine-derived glutamate.
(B) 1-D 1H{13C} HSQC NMR analysis of 13C-reduced (*GSH) plus oxidized (*GSSG) glutathiones indicates that the fraction of glutamine contributing to glutathione
production was sustained under hypoxia (H) as compared to aerobic condition (A). 13C incorporation was estimated from the 13C-4 peak of the glutamate residue
in GSH+GSSG by 1D HSQC NMR.
(C) FT-ICR-MS determination of glutathione oxidation. The 13C5 and 13C10 isotopologues of GSSG were most likely derived from the 13C5,15N2-glutamine tracer
via respective incorporation of one and two units of 13C5-Glu. The error bars in (B) and (C) represent SEM.
(D) Glutamine-dependent redox homeostasis. Flow cytometric histograms of ROS levels were determined by DCFDA fluorescence in P493 cells treated with
2, 10, and 20 mM BPTES for 24 and 48 hr.
(E) Mean fluorescence intensity (MFI). MFI for DCFDA fluorescence is shown for the different concentrations of BPTES under aerobic and hypoxic conditions.
MFI was determined from triplicate experiments with SD and p values (t test) shown.
We have observed that fatty acid oxidation, a possible source for

unlabeled acetyl-CoA, increased in the absence of ectopic MYC
in P493 cells (Figure S5B). This could account for the higher
levels of 13C4-citrate when MYC was off (Figure 5). However, it
cannot explain the comparable levels of unlabeled citrate (m+0
Figure 5) between aerobic MYC ON and MYC OFF. Therefore
other sources of unlabeled acetyl CoA, such as oxidation of
amino acids, may also exist.
Glutamine-Dependent Bioenergetics and Redox
Homeostasis, a Cell Survival Pathway in Hypoxia
The dependence of P493 cells on glutamine for proliferation and
maintenance under aerobic and hypoxic conditions suggests
a key role for glutamine in driving anaplerotic and bioenergetic
needs of both dividing and resting cells. We found that the interruption of glutamine metabolism with the glutaminase inhibitor
BPTES decreased ATP levels under aerobic conditions. In

hypoxia, cells maintained a lower ATP level that was further
diminished by BPTES treatment (Figure 7B). These results
suggest that glutamine metabolism via the glucose-independent
TCA cycle supports cellular bioenergetics for cell survival (and
proliferation) under both aerobic and hypoxic conditions.
In addition, both dividing and resting cells require redox
homeostasis, not only for continuing glycolysis but also for
detoxifying ROS. Glutamine, when converted to glutamate,
could also be involved in the import of cystine through the antiporter SLC7A11 (Figure 6A). This is consistent with the significant excretion of labeled glutamate into the culture medium
(Figure 3C). Cystine, when converted to cysteine, contributes
to the synthesis of glutathione together with glycine and glutamate derived from glutamine. We thus determined the contributions of labeled glutamine and glucose to total de novo
Cell Metabolism
Figure 7. Effects of a Glutaminase Inhibitor on Neoplastic Cells In Vitro and in a Tumor Xenograft Model In Vivo
(A) Effect of glutaminase inhibitor, BPTES, on aerobic and hypoxic P493 cell growth. Growth of control cells compared with cells treated with BPTES under both
aerobic and hypoxic conditions. All cells were grown at 1 3 105 cells/mL. Cell counts were performed in triplicate and shown as mean SD. Live cells were
counted daily in a hemocytometer using trypan blue dye exclusion.
(B) Effect of BPTES on steady-state ATP levels. P493 cells were treated with 2 mM BPTES for 20 hr and counted. ATP levels (mean SD, n = 3 experiments) were
determined by luciferin-luciferase-based assay on aliquots containing an equal number of live cells. *p = 0.0006; **p = 0.01 (t test).
(C) Effects of hypoxia and glucose deprivation on oxygen consumption rates. Oxygen consumption rates of aerobic (A) or hypoxic (H) P493 cells were determined
by a Clark-type oxygen electrode after culture in the presence or absence of glucose for 24 hr. Data are from duplicate experiments with SD and p values (t test)
shown.
(D) In vivo efficacy of BPTES. 2.0 3 107 P493 human lymphoma B cells were injected subcutaneously into tumor-bearing SCID mice. When the tumor volume
reached 100 mm3, 200 mg BPTES was injected every other day by intraperitoneal (i.p.) administration (12.5 mg/kg bodyweight) for 20 days. Control animals were
treated with daily i.p. injection of vehicle (2% [vol/vol] DMSO). BPTES inhibited lymphoma xenograft growth as compared to control. The tumor volumes were
measured using digital calipers every 4 days and calculated using the following formula: (length [mm] 3 width [mm] 3 width [mm] 3 0.52). The results represent
the average SEM (n = 7 each).
glutathione synthesis by quantifying 13C-labeled reduced plus

oxidized glutathione levels derived from respective glucose or
glutamine tracers (Figure 6B; *GSH-*GSSG). Relative to glucose,
glutamine carbons were more readily incorporated into the glutamate moiety of the glutathione, indicating that glutamine, not
glucose, is the main precursor for glutathione synthesis. While
the glucose carbon contribution to glutathione diminished in
hypoxia, the contribution from glutamine carbons persisted.
Furthermore, the 13C fractional distribution in oxidized glutathione (GSSG) isotopologues (e.g., m+5 and m+10 in Figure 6C)
was in fact higher under hypoxia with MYC ON than under
aerobic with MYC OFF, which could reflect a higher ROS production or oxidation of de novo-synthesized glutathione in
response to MYC expression and hypoxia. This observation is
supported by a lower level of reduced glutathione (GSH) (Figure S5D) and higher ROS production (Figure 6D) in P493 cells
under hypoxia and inhibition of glutaminase by BPTES. The
order of GSH levels was aerobic control > hypoxic control >
aerobic BPTES > > hypoxic BPTES, while that of ROS production (measured by DCFDA fluorescence) was reversed, which
is expected for removal of ROS by oxidation of GSH to GSSG.
To determine further the role of glutaminase in redox homeostasis, we isolated T cells from mice that are heterozygous for
mitochondrial glutaminase 1 (Gls+/ ) (Masson et al., 2006), which

is the homolog of human kidney type glutaminase rather than
liver-specific glutaminase 2. When compared to aerobic and
hypoxic wild-type (WT) T cells, the Gls+/ T cells had higher basal
ROS levels, which were further increased under hypoxia (Figure S5E). Altogether, these data suggest that MYC expression
can enhance aerobic glutathione biosynthesis from glutamine
to maintain redox homeostasis. Glutamine-derived glutathione
production was also sustained under hypoxia to cope with
heightened ROS production from the perturbed mitochondrial
electron transport chain (Chandel et al., 1998; Guzy et al., 2005).
Our findings that MYC-induced glutamine metabolism persisted in hypoxia and in the absence of glucose for cell maintenance led us to test whether targeting glutaminase is feasible
for cancer therapy. We determined the consequences of BPTES
treatment on aerobic and hypoxic P493 cell proliferation (Figure 7A). While BPTES decreased glutaminase activity (Figure S5C) and the proliferation of aerobic P493 cells, inhibition
of glutaminase killed hypoxic P493 cells (Figure 7A). Under
hypoxia, the cells proliferated less but continued to import and
metabolize glutamine (Figure S2C and Figure 4) as well as
survive (Figure S6), even under glucose-deprived conditions
(Figure S7). The killing effect of BPTES under hypoxia can be
Cell Metabolism
ascribed to the crucial role of glutamine metabolism for survival

by supporting cell bioenergetics (Figure 7B) and alleviating
oxidative stress (Figure 6). This notion is consistent with the
persistent, although diminished, rate of oxygen consumption
by hypoxic cells as compared with aerobic cells (Figure 7C).
Glucose deprivation increases the rate of oxygen consumption
(Figure 7C) as previously reported (Gao et al., 2009). To determine the in vivo significance of our findings, P493 tumor xenograft-bearing mice were treated with intraperitoneal injections
of BPTES (Figure 7D). As compared with DMSO vehicle-treated
mice, the BPTES-treated mice demonstrated a significantly
diminished tumor progression.
DISCUSSION
Although increased glucose metabolism has been hypothesized
to drive the bioenergetic needs of cancer cells, it is clear that
highly proliferative cancer cells require additional supplies of
biosynthetic precursors not met by glucose metabolism; thus
there is the requirement of anaplerosis via glutaminolysis and/
or pyruvate carboxylation as well as other sources such as fatty
acids (DeBerardinis et al., 2007; Deberardinis et al., 2008; Fan
et al., 2009; Gao et al., 2009; Wise et al., 2008). A significant
proportion of the biosynthetic needs under aerobic conditions
may be met by glutamine metabolism, which is regulated by
the activity of the MYC oncogene (Gao et al., 2009; Wise et al.,
2008). In this regard, our previous studies documented the ability
of MYC to induce mitochondrial biogenesis, function, and
enhanced oxygen consumption (Li et al., 2005), which persisted
under hypoxia (Figure 7C). Our data show that the import rates
of both glucose and glutamine greatly exceed those of other
amino acids under these various experimental conditions, suggesting that they are the major sources for energy and anaplerosis. However, it is a long-standing question how cancer cells
survive or even proliferate under the hypoxic and nutrientdeprived conditions encountered in the tumor microenvironment
(Schroeder et al., 2005).
Solid tumors are often poorly vascularized, leading to regions
of hypoxia and glucose deficiency (Goel et al., 2011; Schroeder
et al., 2005), which must be overcome for cancer cells to continue proliferation and survive. Although MYC-overexpressing
cells appear to become addicted to glutamine, the heightened
addiction to glutamine under hypoxia had not been previously
established (Yuneva et al., 2007). We therefore determined
how limited oxygen availability would influence glutamine
metabolism in vitro to gain insights into what might happen
in vivo. It should be noted, however, that our well-controlled
cell line studied in vitro may have adapted epigenetically and
metabolically to prolonged cultures in nutrient-rich conditions,
and hence metabolic changes under our in vitro experimental
conditions may be instructive, but may not fully represent what
happens in vivo in the tumor microenvironment. Notwithstanding
this caveat, we used uniformly labeled [U-13C,15N]-Gln and
traced the fate of glutamine carbon and nitrogen atoms in both
aerobic and hypoxic conditions with MYC turned OFF or ON in
a human B cell model of Burkitt lymphoma (P493). The P493
system is strictly dependent on MYC function for cell proliferation; as such, this experimental model does not conclusively allow us to distinguish the metabolic consequences of
MYC-mediated transcriptional changes versus those that occur

due to changes in cell size increase or proliferative status as
previously reported (Schuhmacher et al., 1999). We surmise
that most of the metabolic changes between MYC ON and
MYC OFF states are consequential to MYC regulation of genes
involved in both glucose and glutamine metabolism as previously documented (Dang et al., 2009a). The entry of glutamine,
after its conversion to glutamate and then to a-ketoglutarate,
into the TCA and oxidation to succinate, fumarate, and malate
were highly facilitated by MYC expression. More intriguingly,
glutamine metabolism via the TCA cycle persisted under hypoxia
and remained responsive to MYC. Based on detailed analyses of
13
C isotopologues of citrate, succinate, fumarate, and malate
(Figure 4), we uncovered a glucose-independent TCA cycle
solely supported by glutamine, which involved the supply of
acetyl-CoA and OAA to the TCA cycle by the combined activity
of ATP citrate lyase, ME, PDH, and/or pyruvate carboxylase (Figure 4). This pathway was operative even when glucose was
abundant and remained active under hypoxia. These data collectively define a more efficient pathway of glutaminolysis, in
which glutamine can be used either for anabolic purposes for
proliferation, or for producing energy. This pathway can produce
up to 17.5 mol ATP/mol glutamine oxidized to CO2. This is in
contrast to the canonical glutaminolysis pathway from glutamine
to lactate (Helmlinger et al., 2002; Newsholme et al., 1985a,
1985b), in which the glutamine carbons are not oxidized via the
TCA cycle and produce only 5 ATP/mol glutamine converted to
lactate + CO2.
To further explore the role of the glucose-independent TCA
cycle, we traced glutamine metabolism under glucose-deprived
culture conditions. SIRM analysis of TCA cycle metabolites revealed a relative increase in this glutamine-mediated pathway
in glucose-deprived conditions, as evidenced by the increase
in the ratios of 13C6-citrate and 13C5-citrate to 13C4-citrate (Figure 5). The metabolites produced through this glutamine pathway provided the anabolic precursors (citrate) for lipid synthesis
(Figure S4D) and cell survival precursor (Glu) for glutathione
production (Figure 6B, Figure S3C), which was commensurate
with the aerobic growth demand (Figure S3A) and the need for
ROS quenching under hypoxia (Figure 6). These observations
are consistent with our previous findings that knockdown of
glutaminase by siRNA resulted in increased ROS and slowed
growth that could be partially rescued by N-acetylcysteine
(Gao et al., 2009). It is also notable that pyruvate carboxylase
plays a significant role in this glutamine pathway (Figures 4
and 5) and that glutamine-independent tumor cells use pyruvate
carboxylase to drive a glucose-dependent TCA cycle in the
absence of glutamine (Cheng et al., 2011). Thus, the flexibility
of the TCA cycle to use either or both anaplerotic pathways
may be important for tumor adaptation. More practically, therapeutic approaches targeting metabolism must consider these
adaptive strategies.
As a proof of concept, we tested the effect of a glutaminase
inhibitor (BPTES) on growth and ROS production of P493 cells
and showed that blocking glutamine metabolism not only inhibited tumor cell growth under aerobic conditions but also led
to cell death under hypoxia and reduction of tumor xenograft
growth in vivo (Figures 7A and 7D). The specificity of BPTES
was previously documented by our studies in glioma cells, which
Cell Metabolism
revealed BPTES-induced metabolic changes, such as diminished glutamate, a-ketoglutarate, succinate, fumarate, and
malate levels, consistent with an on-target effect on glutaminase
(Seltzer et al., 2010). Further, the uncompetitive nature of BPTES
inhibitory activity was recently underscored by the crystal structure of glutaminase (GAC form) cocrystallized with BPTES, which
sits at the oligomerization interface of the glutaminase tetramer
(Delabarre et al., 2011). Our finding of a glutamine-driven TCA
cycle and other recent discoveries of cancer-related metabolic
pathways (Dang et al., 2009b; Frezza et al., 2011; Possemato
et al., 2011) suggest that metabolic flexibility may be a common
feature of tumor cell metabolism. Hence, a broader and deeper
understanding of cancer cell metabolism and their ability to
reprogram canonical biochemical pathways under metabolic
stress can be a rich ground for uncovering strategies for therapeutic targeting of tumors.
(to T.W.M.F.) and 20061 (to A.N.L.); and National Institutes of Health National
Center for Research Resources (NIH NCRR) grant 5P20RR018733, Kentucky
Challenge for Excellence, the Brown Foundation, and Kentucky Lung Cancer
Research Program (postdoctoral fellowship to P.K.L.). The FT-ICR-MS instrumentation was supported by the National Science Foundations Office of
Experimental Program to Stimulate Competitive Research (NSF/EPSCoR)
grant number EPS-0447479. We thank Ramani Dinavahi, Julie Tan, and
Radhika Burra for excellent technical assistance and Stephen Rayport for the
heterozygous glutaminase mice. A.L., A.N.L., T.W.M.F., and C.V.D. designed
research and wrote the manuscript. A.L., A.N.L., M.H., S.B., J.B., C.J.R.,
H.Z., L.J.Z., R.J.C.S., P.K.L., R.M.H., and T.W.M.F. performed research; all
authors interpreted the data and commented on the manuscript.
Bensaad, K., Tsuruta, A., Selak, M.A., Vidal, M.N., Nakano, K., Bartrons, R.,
Gottlieb, E., and Vousden, K.H. (2006). TIGAR, a p53-inducible regulator of
glycolysis and apoptosis. Cell 126, 107120.
Detailed materials and methods are available online in the Supplemental

Information.
13
C-Labeled Glutamine or Glucose and Hypoxic Exposure

Briefly, Human lymphoma B P493 cells were grown in [U-13C]-labeled glucose
or [U-13C,15N]-labeled glutamine for 24 hr. Nonhypoxic cells (21% [(vol/vol] O2)
were maintained at 37 C in a 5% (vol/vol) CO2 and 95% (vol/vol) air incubator.
Hypoxic cells (1% O2) were maintained in a controlled atmosphere chamber
(PLAS-LABS) with a gas mixture containing 1% O2, 5% (vol/vol) CO2, and
94% (vol/vol) N2 at 37 C for 48 hr. Metabolites were extracted from cells
and media as previously described (Fan et al., 2011).
Bromobimane and carboxy-H2DCFDA were used to measure reduced
glutathione ROS levels, respectively, using flow cytometry.
ATP Levels
ATP levels were determined by luciferin-luciferase-based assay (Promega).
Glutaminase Activity
Cells treated with BPTES and/or tetracycline were then incubated with [3H]
glutamine to assess glutaminase activity.
The b-oxidation of 14C labeled oleic acid was used to assess fatty acid
oxidation.
Cell Viability Assay
The presence of dead cells was determined by propidium iodide staining at
2 mg/ml using flow cytometry.
Values are shown as mean standard deviation (SD) or standard error of the
mean (SEM). Data were analyzed using the Students t test, and significance
was defined as p < 0.05.
Supplemental Information includes seven figures, one table, Supplemental
Experimental Procedures, and Supplemental References and can be found
with this article online at doi:10.1016/j.cmet.2011.12.009.
ACKNOWLEDGMENTS
This work was supported by in part by grant 1R21NS074151-01 (to T.T.);
The Sol Goldman Pancreatic Cancer Research fund (to A.L.); grants
5R01CA051497-21 and 5R01CA057341-20, Leukemia Lymphoma Society
grant LLS-6363-11, and Stand-Up-to-Cancer/American Association for
Cancer Research translational grant (to C.V.D.); grants 1R01CA118434-01A2
and 3R01CA118434-02S1, University of Louisville CTSPGP grants 20044
Received: November 3, 2011

Revised: December 8, 2011
REFERENCES
Chandel, N.S., Maltepe, E., Goldwasser, E., Mathieu, C.E., Simon, M.C., and
Schumacker, P.T. (1998). Mitochondrial reactive oxygen species trigger
hypoxia-induced transcription. Proc. Natl. Acad. Sci. USA 95, 1171511720.
Cheng, T., Sudderth, J., Yang, C., Mullen, A.R., Jin, E.S., Mates, J.M., and
DeBerardinis, R.J. (2011). Pyruvate carboxylase is required for glutamineindependent growth of tumor cells. Proc. Natl. Acad. Sci. USA 108, 8674
8679.
Dang, C.V. (2010). Rethinking the Warburg effect with Myc micromanaging
glutamine metabolism. Cancer Res. 70, 859862.
Dang, C.V., Le, A., and Gao, P. (2009a). MYC-induced cancer cell energy
metabolism and therapeutic opportunities. Clin. Cancer Res. 15, 64796483.
Dang, L., White, D.W., Gross, S., Bennett, B.D., Bittinger, M.A., Driggers, E.M.,
Fantin, V.R., Jang, H.G., Jin, S., Keenan, M.C., et al. (2009b). Cancer-associated IDH1 mutations produce 2-hydroxyglutarate. Nature 462, 739744.
Dang, L., Jin, S., and Su, S.M. (2010). IDH mutations in glioma and acute
myeloid leukemia. Trends Mol. Med. 16, 387397.
DeBerardinis, R.J., Mancuso, A., Daikhin, E., Nissim, I., Yudkoff, M., Wehrli, S.,
and Thompson, C.B. (2007). Beyond aerobic glycolysis: transformed cells can
engage in glutamine metabolism that exceeds the requirement for protein and
nucleotide synthesis. Proc. Natl. Acad. Sci. USA 104, 1934519350.
Deberardinis, R.J., Sayed, N., Ditsworth, D., and Thompson, C.B. (2008). Brick
by brick: metabolism and tumor cell growth. Curr. Opin. Genet. Dev. 18, 5461.
Delabarre, B., Gross, S., Fang, C., Gao, Y., Jha, A., Jiang, F., Song, J.J., Wei,
W., and Hurov, J.B. (2011). Full-length human glutaminase in complex with an
allosteric inhibitor. Biochemistry 50, 1076410770.
Fan, T.W., Lane, A.N., Higashi, R.M., Farag, M.A., Gao, H., Bousamra, M., and
Miller, D.M. (2009). Altered regulation of metabolic pathways in human lung
cancer discerned by 13C stable isotope-resolved metabolomics (SIRM).
Mol. Cancer 8, 41.
Fan, T.W.-M., Yuan, P., Lane, A.N., Higashi, R.M., Wang, Y., Hamidi, A., Zhou,
R., Guitart-Navarro, X., Chen, G., Manji, H.K., et al. (2010). Stable isotope
resolved metabolomic analysis of lithium effects on glial-neuronal interactions.
Metabolomics 6, 165179.
Fan, T.W.-M., Lane, A.N., Higashi, R.M., and Yan, J. (2011). Stable isotope
resolved metabolomics of lung cancer in a scid mouse model. Metabolomics
7, 257269.
Frezza, C., Zheng, L., Folger, O., Rajagopalan, K.N., MacKenzie, E.D., Jerby,
L., Micaroni, M., Chaneton, B., Adam, J., Hedley, A., et al. (2011). Haem oxygenase is synthetically lethal with the tumour suppressor fumarate hydratase.
Nature. 10.1038/nature10363.
Gao, P., Tchernyshyov, I., Chang, T.C., Lee, Y.S., Kita, K., Ochi, T., Zeller, K.I.,
De Marzo, A.M., Van Eyk, J.E., Mendell, J.T., et al. (2009). c-Myc suppression
Cell Metabolism
of miR-23a/b enhances mitochondrial glutaminase expression and glutamine

metabolism. Nature 458, 762765.
Functional genomics reveal that the serine synthesis pathway is essential in

breast cancer. Nature 476, 345350.
Goel, S., Duda, D.G., Xu, L., Munn, L.L., Boucher, Y., Fukumura, D., and Jain,
R.K. (2011). Normalization of the vasculature for treatment of cancer and other
diseases. Physiol. Rev. 91, 10711121.
Robinson, M.M., McBryant, S.J., Tsukamoto, T., Rojas, C., Ferraris, D.V.,
Hamilton, S.K., Hansen, J.C., and Curthoys, N.P. (2007). Novel mechanism
of inhibition of rat kidney-type glutaminase by bis-2-(5-phenylacetamido1,2,4-thiadiazol-2-yl)ethyl sulfide (BPTES). Biochem. J. 406, 407414.
Guzy, R.D., Hoyos, B., Robin, E., Chen, H., Liu, L., Mansfield, K.D., Simon,
M.C., Hammerling, U., and Schumacker, P.T. (2005). Mitochondrial complex
III is required for hypoxia-induced ROS production and cellular oxygen
sensing. Cell Metab. 1, 401408.
Helmlinger, G., Sckell, A., Dellian, M., Forbes, N.S., and Jain, R.K. (2002). Acid
production in glycolysis-impaired tumors provides new insights into tumor
metabolism. Clin. Cancer Res. 8, 12841291.
Kim, J.W., Tchernyshyov, I., Semenza, G.L., and Dang, C.V. (2006). HIF-1mediated expression of pyruvate dehydrogenase kinase: a metabolic switch
required for cellular adaptation to hypoxia. Cell Metab. 3, 177185.
Kim, J.W., Gao, P., Liu, Y.C., Semenza, G.L., and Dang, C.V. (2007). Hypoxiainducible factor 1 and dysregulated c-Myc cooperatively induce vascular
endothelial growth factor and metabolic switches hexokinase 2 and pyruvate
dehydrogenase kinase 1. Mol. Cell. Biol. 27, 73817393.
Kim, J., Lee, J.H., and Iyer, V.R. (2008). Global identification of Myc target
genes reveals its direct role in mitochondrial biogenesis and its E-box usage
in vivo. PLoS One 3, e1798. 10.1371/journal.pone.0001798.
King, A., Selak, M.A., and Gottlieb, E. (2006). Succinate dehydrogenase and
fumarate hydratase: linking mitochondrial dysfunction and cancer.
Oncogene 25, 46754682.
Koppenol, W.H., Bounds, P.L., and Dang, C.V. (2011). Otto Warburgs contributions to current concepts of cancer metabolism. Nat. Rev. Cancer 11,
325337.
Li, F., Wang, Y., Zeller, K.I., Potter, J.J., Wonsey, D.R., ODonnell, K.A., Kim,
J.W., Yustein, J.T., Lee, L.A., and Dang, C.V. (2005). Myc stimulates nuclearly
encoded mitochondrial genes and mitochondrial biogenesis. Mol. Cell. Biol.
25, 62256234.
Masson, J., Darmon, M., Conjard, A., Chuhma, N., Ropert, N., Thoby-Brisson,
M., Foutz, A.S., Parrot, S., Miller, G.M., Jorisch, R., et al. (2006). Mice lacking
brain/kidney phosphate-activated glutaminase have impaired glutamatergic
synaptic transmission, altered breathing, disorganized goal-directed behavior
and die shortly after birth. J. Neurosci. 26, 46604671.
Metallo, C.M., Gameiro, P.A., Bell, E.L., Mattaini, K.R., Yang, J., Hiller, K.,
Jewell, C.M., Johnson, Z.R., Irvine, D.J., Guarente, L., et al. (2011).
Reductive glutamine metabolism by IDH1 mediates lipogenesis under
hypoxia. Nature. Published online November 20, 2011. 10.1038/nature10602.
Mullen, A.R., Wheaton, W.W., Jin, E.S., Chen, P.H., Sullivan, L.B., Cheng, T.,
Yang, Y., Linehan, W.M., Chandel, N.S., and Deberardinis, R.J. (2011).
Reductive carboxylation supports growth in tumour cells with defective mitochondria. Nature. Published online November 20, 2011. 10.1038/nature10642.
Newsholme, E.A., Crabtree, B., and Ardawi, M.S. (1985a). Glutamine metabolism in lymphocytes: its biochemical, physiological and clinical importance.
Q. J. Exp. Physiol. 70, 473489.
Newsholme, E.A., Crabtree, B., and Ardawi, M.S. (1985b). The role of high
rates of glycolysis and glutamine utilization in rapidly dividing cells. Biosci.
Rep. 5, 393400.
Possemato, R., Marks, K.M., Shaul, Y.D., Pacold, M.E., Kim, D., Birsoy, K.,
Sethumadhavan, S., Woo, H.-K., Jang, H.G., Jha, A.K., et al. (2011).
Samudio, I., Fiegl, M., and Andreeff, M. (2009). Mitochondrial uncoupling and
the Warburg effect: molecular basis for the reprogramming of cancer cell
metabolism. Cancer Res. 69, 21632166.
Schroeder, T., Yuan, H., Viglianti, B.L., Peltz, C., Asopa, S., Vujaskovic, Z., and
Dewhirst, M.W. (2005). Spatial heterogeneity and oxygen dependence of
glucose consumption in R3230Ac and fibrosarcomas of the Fischer 344 rat.
Cancer Res. 65, 51635171.
Schuhmacher, M., Staege, M.S., Pajic, A., Polack, A., Weidle, U.H.,
Bornkamm, G.W., Eick, D., and Kohlhuber, F. (1999). Control of cell growth
by c-Myc in the absence of cell division. Curr. Biol. 9, 12551258.
Seltzer, M.J., Bennett, B.D., Joshi, A.D., Gao, P., Thomas, A.G., Ferraris, D.V.,
Tsukamoto, T., Rojas, C.J., Slusher, B.S., Rabinowitz, J.D., et al. (2010).
Inhibition of glutaminase preferentially slows growth of glioma cells with
mutant IDH1. Cancer Res. 70, 89818987.
Telang, S., Lane, A.N., Nelson, K.K., Arumugam, S., and Chesney, J.A. (2007).
The oncoprotein H-RasV12 increases mitochondrial metabolism. Mol. Cancer
6, 77.
Vander Heiden, M.G. (2011). Targeting cancer metabolism: a therapeutic
window opens. Nat. Rev. Drug Discov. 10, 671684.
Vander Heiden, M.G., Cantley, L.C., and Thompson, C.B. (2009).
Understanding the Warburg effect: the metabolic requirements of cell proliferation. Science 324, 10291033.
Warburg, O. (1956). On the origin of cancer cells. Science 123, 309314.
Warburg, O., Posener, K., and Negelein, E. (1924). Ueber den Stoffwechsel der
Tumoren. Biochem. Z. 152, 319344.
Wise, D.R., DeBerardinis, R.J., Mancuso, A., Sayed, N., Zhang, X.Y., Pfeiffer,
H.K., Nissim, I., Daikhin, E., Yudkoff, M., McMahon, S.B., et al. (2008). Myc
regulates a transcriptional program that stimulates mitochondrial glutaminolysis and leads to glutamine addiction. Proc. Natl. Acad. Sci. USA 105, 18782
18787.
Wise, D.R., Ward, P.S., Shay, J.E., Cross, J.R., Gruber, J.J., Sachdeva, U.M.,
Platt, J.M., Dematteo, R.G., Simon, M.C., and Thompson, C.B. (2011).
Hypoxia promotes isocitrate dehydrogenase-dependent carboxylation of
alpha-ketoglutarate to citrate to support cell growth and viability. Proc. Natl.
Acad. Sci. USA 108, 1961119616.
Wu, W., Platoshyn, O., Firth, A.L., and Yuan, J.X. (2007). Hypoxia divergently
regulates production of reactive oxygen species in human pulmonary and
coronary artery smooth muscle cells. Am. J. Physiol. Lung Cell. Mol. Physiol.
293, L952L959.
Yoo, H., Antoniewicz, M.R., Stephanopoulos, G., and Kelleher, J.K. (2008).
Quantifying reductive carboxylation flux of glutamine to lipid in a brown adipocyte cell line. J. Biol. Chem. 283, 2062120627.
Yuneva, M., Zamboni, N., Oefner, P., Sachidanandam, R., and Lazebnik, Y.
(2007). Deficiency in glutamine but not glucose induces MYC-dependent
apoptosis in human cells. J. Cell Biol. 178, 93105.
Cell Metabolism
Short Article
Coordination of Triacylglycerol
and Cholesterol Homeostasis by DHR96
and the Drosophila LipA Homolog magro
Matthew H. Sieber1,2 and Carl S. Thummel1,*
1Department
of Human Genetics, University of Utah School of Medicine, 15 N 2030 E, Room 2100, Salt Lake City, UT 84112-5330 USA
address: Department of Embryology, Carnegie Institution, 3520 San Martin Dr., Baltimore, MD 21218 USA
*Correspondence: carl.thummel@genetics.utah.edu
DOI 10.1016/j.cmet.2011.11.011
2Present
SUMMARY
Although transintestinal cholesterol efflux has been

identified as an important means of clearing excess
sterols, the mechanisms that underlie this process
remain poorly understood. Here, we show that
magro, a direct target of the Drosophila DHR96
nuclear receptor, is required in the intestine to maintain cholesterol homeostasis. magro encodes a LipA
homolog that is secreted from the anterior gut into
the intestinal lumen to digest dietary triacylglycerol.
Expression of magro in intestinal cells is required to
hydrolyze cholesterol esters and promote cholesterol clearance. Restoring magro expression in the
intestine of DHR96 mutants rescues their defects in
triacylglycerol and cholesterol metabolism. These
studies show that the central role of the intestine in
cholesterol efflux has been conserved through evolution, that the ancestral function of LipA is to coordinate triacylglycerol and cholesterol metabolism,
and that the region-specific activities of magro correspond to the metabolic functions of its upstream
regulator, DHR96.
INTRODUCTION
Coordinate regulation of lipid metabolism is central to human
health, and disruption of this process leads to a range of metabolic disorders, including obesity and cardiovascular disease.
Normal lipid homeostasis is maintained by balancing dietary lipid
uptake and synthesis with lipid catabolism and excretion. Under
normal feeding conditions, dietary lipids, such as triacylglycerol
(TAG) and cholesterol esters, are broken down into free fatty
acids, monoacylglycerols, and free sterols in the lumen of the
intestine. These digested lipids can then be absorbed by the
intestinal cells, where TAG is resynthesized and packaged
together with cholesterol, cholesterol esters, and carrier proteins
to form lipoprotein particles that are trafficked throughout the
body. These lipids can be either utilized by cells or deposited
in storage tissues, such as the adipose and liver. Under conditions of excess lipids, TAG and cholesterol esters are broken
down and free fatty acids can be utilized for energy, whereas
excess cholesterol is excreted from the body (Lusis and Pajukanta, 2008; van der Velde et al., 2010).
Nuclear receptors (NRs) are ligand-regulated transcription
factors that play essential roles in multiple aspects of lipid
homeostasis. Many NRs bind small lipophilic compounds,
such as fatty acids, sterols, and other metabolic intermediates,
and coordinate multiple aspects of metabolism by directing
specific changes in gene expression. One example of this is
LXRa (NR1H3), which binds oxysterols and promotes the modification and clearance of excess sterols (Kalaany and Mangelsdorf, 2006). In addition, LXRa is required to maintain proper
TAG levels, at least in part through the regulation of SREBPmediated fat synthesis (Schultz et al., 2000). Thus, LXRa
activity is central to both TAG and cholesterol homeostasis,
although much remains to be learned about the roles of specific LXR target genes in mediating these key metabolic
functions.
We have been studying a Drosophila homolog of LXR, DHR96,
as a simple system to understand the physiological and molecular roles for this family of NRs and their target genes. Biochemical and genetic studies of DHR96 have shown that it
shares the central metabolic functions of its mammalian counterpart. DHR96 binds cholesterol and is required for normal cholesterol homeostasis, with DHR96 null mutants exhibiting an 20%
increase in whole animal cholesterol levels due, at least in part, to
increased npc1b expression (Horner et al., 2009; Bujold et al.,
2010). In addition, DHR96 mutants display an 50% decrease
in whole animal TAG levels that can be attributed to an inability
to break down dietary TAG due to reduced expression of the
intestinal lipase Magro (CG5932) (Sieber and Thummel, 2009).
Interestingly, magro transcription responds to dietary cholesterol levels and this regulation is dependent on DHR96 function,
providing a potential link between cholesterol levels, DHR96,
and TAG homeostasis (Horner et al., 2009; Bujold et al., 2010).
Moreover, although Magro protein is most similar to mammalian
gastric lipase (38% identity, 56% similarity), the second most
similar protein is LipA (32% identity, 50% similarity), which has
both TAG lipase and cholesterol esterase activities (Ameis
et al., 1994). Genetic studies have demonstrated a central role
for LipA in maintaining proper cholesterol levels in mice (Du
et al., 2001). Similar phenotypes are seen in human LipA mutants
suffering from cholesterol ester storage disease (CESD) and
Wolmans disease (Burke and Schubert, 1972). These observations raise the possibility that magro, in addition to controlling
TAG homeostasis, may regulate cholesterol homeostasis and
Cell Metabolism
Regulation of Lipid Homeostasis
140
120
120
CE activity
100
80
80
60
40
40
magro RNAi
66
5
44
3
22
1
00
WT
!"
10
!"
15
l extract
20
20
00
Figure 1. magro Maintains Proper Cholesterol

Levels and Has Cholesterol Esterase Activity
Act-GAL4/+
88
7
D
H
R
96
c
1
U
t
-G
A
SA
L4
m
ag
/+
-R
A
ct
N
A
>m
i/+
ag
-R
N
A
i
% cholesterol
A 160
GST
GST-magro
66
g sterol
CE activity
88
44
Mex-GAL4/+
magro RNAi
6
6
4
4
2
2
22
00
00
55
10
15
10
15
ng protein
20
20
25
free
chol
!"
total
chol
DHR96 may function through magro to help coordinate TAG and

cholesterol metabolism.
In this study we show that loss of magro function leads to an
increase in cholesterol levels similar to that seen in DHR96
mutants. We show that Magro, like LipA, has cholesterol
esterase activity, and this enzyme is required in intestinal cells
to maintain cholesterol homeostasis. In contrast, the TAG lipase
activity of Magro arises from the anterior end of the gut and acts
in the intestinal lumen to facilitate dietary fat uptake. Restoring
magro expression in the intestine of the DHR96 mutant is sufficient to rescue their lean phenotype and elevated levels of
cholesterol. Our data support the model that DHR96 functions
through magro in the intestine to coordinate both dietary TAG
breakdown and the clearance of excess sterols.
RESULTS
magro Is Required for Normal Cholesterol Homeostasis
The regulation of magro transcription by dietary cholesterol
combined with its significant homology to LipA prompted us to
test whether magro function is required for cholesterol homeostasis. DHR961 null mutants grown on a normal diet display
elevated levels of cholesterol compared to genetically matched
wild-type controls (Figure 1A), similar to the results seen when
DHR96 mutant larvae are subjected to a high cholesterol diet
(Horner et al., 2009). Interestingly, ubiquitous RNA interference
(RNAi)-mediated silencing of magro expression using ActGAL4, as done previously (Sieber and Thummel, 2009), leads
to a similar phenotype (Figure 1A). Taken together with our earlier
work, which showed that both DHR96 mutants and magro RNAi
animals have significantly lower levels of TAG, these data
suggest that DHR96 functions through transcriptional regulation
of magro to coordinate TAG and cholesterol homeostasis in
Drosophila.
chol
esters
(A) RNAi for magro results in elevated cholesterol levels

similar to those seen in DHR96 mutants. Wild-type (WT),
Act-GAL4/+, and UAS-magro-RNAi/+ control flies, along
with DHR961 mutants and Act-GAL4/UAS-magro-RNAi
animals (Act>mag-RNAi), were assayed for total cholesterol levels. The data were normalized to protein levels and
are presented relative to a wild-type level of 100%.
(B) RNAi for magro results in reduced cholesteryl ester
hydrolase activity (CE) in intestines. Intestines dissected
from both Act-Gal4/+ control and Act-GAL4/UAS-magroRNAi (magro RNAi) animals were homogenized and
increasing amounts of lysate were tested for cholesteryl
ester hydrolase activity by assaying for the release of free
cholesterol from a cholesterol acetate substrate. The
y-axis shows mg/ml of free glycerol released.
(C) Purified GST-Magro protein has CE activity. Recombinant GST and GST-Magro were purified as described
(Sieber and Thummel, 2009) and increasing amounts of
protein were assayed for CE activity. The y-axis shows mg/
ml of free glycerol released.
(D) Intestine-specific RNAi for magro results in elevated
levels of esterified cholesterol. Free, esterified, and total
cholesterol levels were measured from Mex-GAL4/+
control and Mex-GAL4/UAS-magro-RNAi (magro RNAi)
animals. Error bars represent SEM (*p < 0.05,
**p < 0.0001).
magro Regulates Cholesterol Homeostasis by Breaking

Down Stored Cholesterol Esters
In order to determine whether Magro can act like LipA by
cleaving cholesterol esters in addition to its TAG lipase activity,
we assayed for cholesterol esterase activity in control and magro
RNAi animals (Ameis et al., 1994). Although control intestinal
lysates exhibit a high level of cholesterol esterase activity in
a dose-dependent manner, the lysates from magro RNAi animals
display an 50% decrease in enzymatic function (Figure 1B).
Moreover, purified recombinant GST-Magro efficiently cleaves
a cholesterol ester substrate in vitro, demonstrating that these
effects are a direct result of Magro enzymatic activity (Figure 1C).
Taken together with our earlier biochemical studies, this result
shows that Magro is a bifunctional enzyme that can act as both
a TAG lipase and a cholesterol esterase (Sieber and Thummel,
2009).
If decreased intestinal cholesterol esterase activity causes the
elevated cholesterol levels seen in the magro RNAi animals, then
we should see an increase in stored cholesterol esters when
magro expression is specifically silenced in the intestine. Consistent with this proposal, we see a significant increase in the levels
of both total cholesterol and cholesterol esters in Mex>magro
RNAi animals, whereas free cholesterol levels remain unchanged
(Figure 1D). Elevated cholesterol levels can also be detected
clearly in isolated intestines, supporting a role for magro in maintaining cholesterol levels in this tissue (Figure S1). These data
support the model that Magro maintains cholesterol homeostasis through its ability to directly break down stored cholesterol
esters in the intestine.
magro Is Expressed throughout the Digestive Tract
The similarity between lipase sequences complicates our ability
to raise specific antibodies against Magro. Accordingly, we
determined the pattern of Magro expression using a genomic
Cell Metabolism
the intestinal lumen (Figure S3). Lower levels of Magro-EGFP

protein can also be seen in the major cell type of the intestine,
the enterocytes, in a punctate cytoplasmic pattern (Figure 2B).
Interestingly, we do not observe Magro-EGFP in large Lysotracker-positive vesicles in these cells. Taken together, these
observations suggest that Magro is trafficked differently in
different cell types of the digestive tract, raising the possibility
that the cholesterol esterase and TAG lipase activities of this
enzyme are regionally specified.
130 m
Prov
24 m
Lysotracker
magro-GFP
magro-GFP
75 m
Overlay
Lysotracker
20 m
Overlay
Figure 2. Magro Is Expressed in the Proventriculus and Midgut

(A) Magro-EGFP (green) expression in mid-third instar larvae is restricted to the
intestine and accumulates to high levels in the anterior region of the proventriculus (Prov) (A, arrow in C). Antibody staining for EGFP also revealed lower
levels of punctate expression in the cytoplasm of enterocytes.
(B) Higher resolution images of the proventriculus revealed Magro-EGFP in
large vesicles that extend from the abundant expression at the anterior end of
the proventriculus toward the junction with the midgut lumen (yellow boxes
in C, D, and E, shown in panels F, G, and H). Lysotracker Red stains the large
acidic vesicles that contain Magro-EGFP protein (D, E, G, and H).
magro-EGFP transgenic line. This construct contains 7 kb of

genomic DNA spanning the magro locus with 5 kb of upstream
promoter sequences and the EGFP gene fused in-frame to the
30 -end of the magro protein coding region. Magro-EGFP is expressed specifically in the intestine with the highest levels of
protein accumulation in the anterior region of the proventriculus
(Figure 2A). The proventriculus is a bulb shaped structure at the
anterior end of the gut that consists of three distinct cell layers
(Figure S2). Magro-EGFP is expressed in the anterior half of
the outermost layer of cells in the proventriculus (Figures 2C,
arrow; and S2). In addition, protein is clearly visible in large vesicles that lie posterior to this region of expression (Figures 2C2E,
yellow boxes). These are acidic vesicles that stain positive for
Lysotracker Red (Figures 2F2H), consistent with the acid
lipase activity of Magro and LipA. Interestingly, visualization of
vesicles using CD8-EGFP in this region reveals that they move
in a posterior direction toward the intestine, providing a potential
mechanism to deliver digestive enzymes, such as Magro, into
magro Acts within the Intestinal Lumen to Maintain

Triacylglycerol Homeostasis
As a first step toward determining whether the enzymatic activities of Magro localize regionally within the intestine, we raised
DHR96 mutant and magro RNAi animals on a diet supplemented
with pancreatin. This enzymatic mixture of pancreatic enzymes,
including TAG lipase and cholesterol esterase, should specifically restore TAG lipase and cholesterol esterase activity in the
lumen of the gut. Although pancreatin supplementation had no
effect on TAG levels in control animals, this diet restored wildtype levels of TAG in both DHR96 and magro RNAi animals
(Figures 3A and 3B). Conversely, pancreatin supplementation
had no impact on the elevated cholesterol levels in these animals
(Figure 3C and 3D). These results are consistent with those seen
when wild-type flies are treated with Orlistat (tetrahydrolipstatin),
which acts as a competitive inhibitor of secreted lipases and cholesterol esterases inside the lumen of the intestine (Borgstrom,
1988). Although Orlistat treatment is sufficient to decrease whole
animal TAG levels, as seen previously (Sieber and Thummel,
2009), it has no significant effect on total cholesterol levels (Figure S4). These data confirm our earlier studies indicating that
Magro functions in the intestinal lumen to maintain appropriate
levels of TAG, and demonstrate that its effects on cholesterol
homeostasis are conferred within the cells of the intestinal tract.
magro Functions in the Proventriculus to Promote
the Breakdown of Dietary Triacylglycerol
The apparent vesicular trafficking of Magro protein from the
proventriculus toward the body of the midgut provides a mechanism to explain its delivery into the lumen of the intestine. If this
model is correct, then disrupting magro function specifically in
the proventriculus should have an effect on TAG levels but little
or no effect on cholesterol homeostasis. To test this hypothesis,
we used the bab1-GAL4 driver, which is expressed highly in the
proventriculus and weakly near the midgut-hindgut junction (Figure 4A) (Cabrera et al., 2002). Using this construct to direct
magro RNAi resulted in a significant decrease in whole animal
TAG levels (Figure 4B), similar to that seen with the panintestinal
Mex-GAL4 driver (Sieber and Thummel, 2009). In contrast,
proventriculus-specific magro RNAi has no effect on total
cholesterol levels (Figure 4C). We conclude that the TAG metabolic functions of Magro are restricted to the proventriculus,
whereas its cholesterol regulatory function resides outside the
proventriculus in the intestine.
DHR96 Regulates Major Aspects of Lipid Metabolism
through Its Target Gene magro
If decreased magro expression is physiologically relevant to the
cholesterol accumulation phenotype seen in DHR96 mutants,
Cell Metabolism
%total TAG
120
80
40
0
+
WT
120
80
40
0
Panc
DHR961
+
Act-GAL4/+
+
Act>mag
RNAi
Panc
Wild-type (WT) and Act-GAL4/+ controls, along with DHR961 mutants

and Act-GAL4/UAS-magro-RNAi (Act>mag-RNAi) animals, were
raised on medium supplemented with 2 mg/mls pancreatin, an enzymatic mixture of pancreatic TAG lipase and cholesterol esterase.
(AD) Mature adults were collected and assayed for TAG (A and B) and
cholesterol (C and D) levels. The data were normalized to protein
levels and are presented relative to a wild-type level of 100%. Error
bars represent SEM (**p < 0.0001).
D
160
% cholesterol
Figure 3. Magro Enzymatic Activity Is Not Required in the

Intestinal Lumen to Maintain Cholesterol Homeostasis
120
160
% cholesterol
%total TAG
120
disease (Burke and Schubert, 1972). Patients with Wolmans disease also have digestive dysfunction, which
40
40
may be related to the defects in lipid uptake that we
0
0
observe in magro RNAi animals. In addition to these
+
+ Panc
+
+ Panc
shared phenotypes, however, mammalian LipA mutants
DHR961
Act-GAL4/+
Act>mag
WT
RNAi
display massive accumulations of lipid in the liver, spleen,
and intestinedefects that are not apparent in magro
RNAi animals (Du et al., 2001). Nonetheless, the parallels
then restoring magro function in these animals should rescue between Magro and LipA function in flies and humans establish
their defect in cholesterol homeostasis. Consistent with this Drosophila as a system to further our understanding of CESD
hypothesis, expressing wild-type magro specifically in the intes- and Wolmans disease and define the ancestral function for
tine of the DHR96 mutant is sufficient to reduce the cholesterol this class of acid lipases, demonstrating their central role in the
levels in these animals (Figure 4D). This rescue, however, is not intestine to coordinate TAG and cholesterol homeostasis.
Interestingly, the dual enzymatic functions of Magro appear to
complete, which is consistent with the multiple levels of cholesterol metabolism that are regulated by DHR96 (Horner et al., arise from distinct regions of the intestine. Disruption of magro
2009; Bujold et al., 2010). Moreover, specific expression of function specifically in the proventriculus blocks its TAG lipolytic
magro in the proventriculus of DHR96 mutants effectively activity but does not affect cholesterol levels in these animals
rescues their lean phenotype (Figure 4E) but has no significant (Figures 4B and 4C). In contrast, magro RNAi throughout the
effect on their elevated cholesterol levels (Figure 4F). Taken intestine affects both TAG and cholesterol homeostasis (Figtogether with our other data, these results indicate that the ure 1D). These region-specific activities are consistent with our
region-specific enzymatic activities of Magro correspond to the dietary supplementation studies with pancreatin and Orlistat
major lipid metabolic functions of DHR96. DHR96 regulation of (Figures 3 and S4). They are also consistent with the expression
magro expression in the proventriculus maintains an appropriate pattern of Magro-EGFP protein, providing insights into how the
level of TAG lipase activity in the intestinal lumen to facilitate the dual functions of this enzyme are manifested. Magro-EGFP is
breakdown of dietary lipid whereas DHR96 regulation of magro expressed abundantly in the large outer columnar cells in the
in the body of the intestine promotes the clearance of excess anterior half of the proventriculus (Figures 2A and S2). We also
see a stream of large acidic vesicles that originate from this
sterols.
region and move in a posterior direction toward the lumen of
the intestine (Figures 2F2H and S3). This apparent vesicular
DISCUSSION
trafficking of Magro is consistent with the cells at the anterior
end of the proventriculus having secretory functions, depositing
Magro and LipA Share Conserved Functions
peritrophic matrix components into the lumen that lies between
in Maintaining Lipid Homeostasis
Relatively little is known about the mechanisms that regulate the outer and inner cell layers of the proventriculus (King, 1988)
cholesterol metabolism in Drosophila. Most studies have (Figure S2). The peritrophic matrix is a meshwork of chitin and
focused on DHR96 and the Niemann-Pick (NPC) disease gene glycoproteins that provides a protective lining within the gut,
homologs, which play important roles in dietary cholesterol much like the mucosal layer of the mammalian intestine (Hegeabsorption and intracellular cholesterol trafficking (Niwa and dus et al., 2009). The observation that the Magro-EGFP vesicles
Niwa, 2011). In this paper, we identify the intestine as a key tissue reside in the same region of the proventriculus as the developing
for maintaining cholesterol homeostasis, and we define a central peritrophic matrix suggests that they are synthesized and exrole for the LipA homolog, Magro, in mediating this function. Like ported into the lumen of the gut in a similar manner. This also raiLipA, Magro has dual enzymatic activities, cleaving both TAG ses the possibility that the digestive enzymes that are embedded
and cholesterol esters, consistent with their common fatty acid in the peritrophic matrix may originate from vesicular trafficking
ester bonds (Ameis et al., 1994). Mouse LipA mutants display in the proventriculus. Like magro, many genes with predicted
a lack of stored fat in the form of white adipose tissue along digestive functions, including glucosidases, mannosidases, and
with excess cholesterol esters (Du et al., 2001), reflecting the endopeptidases, are regulated by DHR96 and expressed in the
major defects in magro RNAi animals. Similar phenotypes are intestine (Sieber and Thummel, 2009). Several genes that conseen in human LipA mutants suffering from CESD and Wolmans tribute to the peritrophic matrix are also regulated by DHR96.
80
80
Cell Metabolism
B120120
%total TAG
100100
80 80
60 60
40 40
20 20
0 0
Sm
ag
-R
N
A
ba
i/
+
b1
-G
A
ba
L4
b1
/+
>m
ag
-R
N
A
i
!"
120
ag
/+
al
ba
b1
ex
>m
S-
-G
ag
-R
ag
m
S-
!"
>m
WT
ex
-R
i/
+
!"
L4
A
-G
b1
A
DHR961
E 120
120
F 160
140
140
% cholesterol
100
100
120
120
8080
DHR961
ag
S-
>m
/+
/+
!"
b1
ba
9
HR
WT
L4
ag
2020
ag
/+
ag
ba
S-
>m
/+
L4
-G
A
A
U
ba
b1
DH
4040
0
b1
6
R9
!"
WT
6060
-G
2020
8080
4040
% TAG
100
100
6060
b1
ba
2020
0
/+
4040
20
6060
40
80 80
ag
60
4/
% cholesterol
% cholesterol
100
100
80
00
140
140
120
120
100
ba
DHR961
Figure 4. DHR96 Regulation of magro in Distinct Regions of the

Intestine Coordinates TAG and Cholesterol Homeostasis
(A) bab1-GAL4 drives high levels of UAS-NLS-DsRed expression in the
proventriculus (red arrow) with no detectable expression in the body of the
midgut (black arrow) and low levels near the midgut-hindgut junction (white
arrow).
(B and C) Proventriculus-specific RNAi for magro results in reduced levels of
TAG but does not affect cholesterol. bab1-GAL4/+ and UAS-magro-RNAi/+
control flies, along with bab1-GAL4/UAS-magro-RNAi (bab1>mag-RNAi)
animals were assayed for TAG (B) and cholesterol (C). The data were normalized to protein levels and are presented relative to a wild-type level
of 100%.
(D) Midgut-specific expression of magro is sufficient to partially rescue the
elevated cholesterol levels in DHR96 mutants. Wild-type (WT) and DHR96
mutants carrying the Mex-GAL4 driver alone, UAS-magro transgene alone, or
both Mex-GAL4 and UAS-magro (Mex>mag) to express magro in the midgut
were assayed for total cholesterol levels. The data were normalized to protein
levels and are presented relative to a wild-type level of 100%.
(E and F) Proventriculus-specific expression of magro is sufficient to rescue
the lean phenotype of DHR96 mutants, but has no effect on its elevated
cholesterol levels. Wild-type (WT) and DHR96 mutants carrying the bab1GAL4 driver alone, UAS-magro transgene alone, or both bab1-GAL4 and
UAS-magro (bab1>mag) to express magro in the proventriculus were assayed
for TAG (E) and cholesterol (F). The data were normalized to protein levels and
are presented relative to a wild-type level of 100%. Error bars represent SEM
(*p < 0.01 **p < 0.0001).
It would be interesting to determine whether the proventriculus

synthesizes and secretes these proteins in a coordinated
manner.
In addition to its abundant expression in the proventriculus,
Magro-EGFP is also present at lower levels throughout the intestine, visible as punctate cytoplasmic staining in the enterocytes (Figure 2B). This expression pattern provides a possible
mechanism to explain the role of Magro in maintaining cholesterol homeostasis. We propose that Magro acts as a cholesterol
esterase in the enterocytes, breaking down stored cholesterol to
facilitate its elimination from the intestine. This model is consistent with the neutral lipid stores that are known to reside in the
Drosophila intestine, second only to the fat body. It is also
consistent with recent evidence that LipA acts as cholesterol
esterase in macrophage foam cells to promote ABCA1-mediated cholesterol efflux (Ouimet et al., 2011). These data suggest
that LipA and Magro may break down stored cholesterol esters
upstream of reverse cholesterol transport and transintestinal
cholesterol efflux to promote the clearance of excess sterols.
Tissue-specific studies of LipA function in the pancreas and
intestine may provide a clearer understanding of its relationship
to the apparent exocrine role of Magro in the proventriculus and
its ability to promote cholesterol clearance in the intestine.
Finally, our data provide new directions in understanding the
roles of LXR family members in lipid metabolism. Both DHR96
mutants and magro RNAi animals display reduced levels of
TAG and elevated levels of cholesterol, and restoring magro
expression in the intestines of DHR96 mutant animals largely
rescues these defects, establishing magro as a key target for
DHR96 regulation (Figures 1 and 4C) (Sieber and Thummel,
2009). These functions for DHR96 parallel those of its mammalian homolog, LXR. LXR activation, specifically in the intestine, results in a dramatic increase in fecal sterol excretion that
correlates with increased expression of the ABCG5/ABCG8
sterol transporter (van der Veen et al., 2009; Lo Sasso et al.,
2010). This observation suggests that LXR promotes reverse
cholesterol transport in this tissue, which represents the best
characterized mechanism for eliminating excess cholesterol
from the body. Reverse cholesterol transport involves HDLmediated transport of cholesterol from peripheral tissues to the
hepatobiliary tract, leading to the removal of excess sterol by
biliary excretion from the body. However, genetic studies of
key components in biliary cholesterol excretion, such as abcb4
mutants and abcg5/abcg8 double mutants, have challenged
the importance of reverse cholesterol transport for cholesterol
excretion and have led to the proposal that the intestine plays
a more direct role in this process (van der Velde et al., 2010).
These studies shift the focus of cholesterol efflux toward the
intestine and implicate a central role for LXR in regulating intestinal cholesterol clearance, not only through regulation of reverse
cholesterol transport but also through novel targets involved in
transintestinal cholesterol efflux (Kruit et al., 2005). In addition,
our evidence that intestinal cholesterol esterase activity is critical
for clearing excess sterol from the body suggests that acid
lipases such as LipA may function downstream from LXR to
maintain cholesterol homeostasis. Although there is no direct
evidence that LXR regulates LipA expression, a recent study
showed that elevated levels of oxidized LDL can repress LipA
expression in endothelial cells, an effect that can be reversed
Cell Metabolism
by treatment with LXR agonists (Heltianu et al., 2011). Further

studies are required to determine whether the regulatory links
between LXR, LipA, and cholesterol homeostasis have been
conserved through evolution and whether Drosophila can be
used as a simple model system to better define the mechanisms
of transintestinal cholesterol efflux.
Fly Stocks
The following stocks were used in this study: Canton S, DHR961 (King-Jones
et al., 2006), Mex-Gal4 (Phillips and Thomas, 2006), Act-Gal4/CyO (Bloomington # 25374), UAS-magro (Sieber and Thummel, 2009), UAS-DHR96 (Horner
et al., 2009), and bab1-GAL4/TM3 (Cabrera et al., 2002). Flies were maintained
on standard Bloomington Stock Center medium with malt at 25 C.
Metabolite Assays
Newly eclosed adult male flies were aged 57 days prior to use for all experiments. TAG and cholesterol assays were conducted as described (Horner
et al., 2009; Sieber and Thummel, 2009) All results shown are derived from
12 samples of 5 animals collected from each genotype under each condition,
and repeated at least 3 times. A representative experiment is shown in each
figure.
Statistical Analyses
Statistical analysis was done for each experiment using an unpaired two-tailed
Students t-test with unequal variance. All quantitative data are reported as the
mean SEM. The n and SEM for each data point is derived from the 12
samples of 5 animals collected from each genotype under each condition.
Supplemental Information include Supplemental Experimental Procedures
and four figures and can be found with this article online at doi:10.1016/
j.cmet.2011.11.011.
ACKNOWLEDGMENTS
We thank the Bloomington Stock Center for providing stocks, FlyBase for critical information that made these studies possible, and M. Babst for helpful
discussions. We also thank D. Bricker, J. Misra, and J. Tennessen for critical
comments on the manuscript. M.H.S. was supported by an NIH Developmental Biology Predoctoral Training Grant (5T32 HD07491). This research
was supported by NIH grant 2R01DK075607.
Burke, J.A., and Schubert, W.K. (1972). Deficient activity of hepatic acid lipase
in cholesterol ester storage disease. Science 176, 309310.
Cabrera, G.R., Godt, D., Fang, P.Y., Couderc, J.L., and Laski, F.A. (2002).
Expression pattern of Gal4 enhancer trap insertions into the bric a` brac locus
generated by P element replacement. Genesis 34, 6265.
Du, H., Heur, M., Duanmu, M., Grabowski, G.A., Hui, D.Y., Witte, D.P., and
Mishra, J. (2001). Lysosomal acid lipase-deficient mice: depletion of white
and brown fat, severe hepatosplenomegaly, and shortened life span. J. Lipid
Res. 42, 489500.
Hegedus, D., Erlandson, M., Gillott, C., and Toprak, U. (2009). New insights
into peritrophic matrix synthesis, architecture, and function. Annu. Rev.
Entomol. 54, 285302.
Heltianu, C., Robciuc, A., Botez, G., Musina, C., Stancu, C., Sima, A.V., and
Simionescu, M. (2011). Modified low density lipoproteins decrease the activity
and expression of lysosomal acid lipase in human endothelial and smooth
muscle cells. Cell Biochem. Biophys. 61, 209216.
Horner, M.A., Pardee, K., Liu, S., King-Jones, K., Lajoie, G., Edwards, A.,
Krause, H.M., and Thummel, C.S. (2009). The Drosophila DHR96 nuclear
receptor binds cholesterol and regulates cholesterol homeostasis. Genes
Dev. 23, 27112716.
Kalaany, N.Y., and Mangelsdorf, D.J. (2006). LXRS and FXR: the yin and yang
of cholesterol and fat metabolism. Annu. Rev. Physiol. 68, 159191.
King, D.G. (1988). Cellular organization and peritrophic membrane formation in
the cardia (proventriculus) of Drosophila melanogaster. J. Morphol. 196,
253282.
King-Jones, K., Horner, M.A., Lam, G., and Thummel, C.S. (2006). The DHR96
nuclear receptor regulates xenobiotic responses in Drosophila. Cell Metab. 4,
3748.
Kruit, J.K., Plosch, T., Havinga, R., Boverhof, R., Groot, P.H., Groen, A.K., and
Kuipers, F. (2005). Increased fecal neutral sterol loss upon liver X receptor
activation is independent of biliary sterol secretion in mice. Gastroenterology
128, 147156.
Lo Sasso, G., Murzilli, S., Salvatore, L., DErrico, I., Petruzzelli, M., Conca, P.,
Jiang, Z.Y., Calabresi, L., Parini, P., and Moschetta, A. (2010). Intestinal
specific LXR activation stimulates reverse cholesterol transport and protects
from atherosclerosis. Cell Metab. 12, 187193.
Lusis, A.J., and Pajukanta, P. (2008). A treasure trove for lipoprotein biology.
Nat. Genet. 40, 129130.
Niwa, R., and Niwa, Y.S. (2011). The fruit fly Drosophila melanogaster as
a model system to study cholesterol metabolism and homeostasis.
Cholesterol 2011, 16.
Ouimet, M., Franklin, V., Mak, E., Liao, X., Tabas, I., and Marcel, Y.L. (2011).
Autophagy regulates cholesterol efflux from macrophage foam cells via lysosomal acid lipase. Cell Metab. 13, 655667.

Published online: December 22, 2011
Phillips, M.D., and Thomas, G.H. (2006). Brush border spectrin is required for
early endosome recycling in Drosophila. J. Cell Sci. 119, 13611370.
REFERENCES
Ameis, D., Merkel, M., Eckerskorn, C., and Greten, H. (1994). Purification,
characterization and molecular cloning of human hepatic lysosomal acid
lipase. Eur. J. Biochem. 219, 905914.
Borgstrom, B. (1988). Mode of action of tetrahydrolipstatin: a derivative of the
naturally occurring lipase inhibitor lipstatin. Biochim. Biophys. Acta 962,
308316.
Bujold, M., Gopalakrishnan, A., Nally, E., and King-Jones, K. (2010). Nuclear
receptor DHR96 acts as a sentinel for low cholesterol concentrations in
Drosophila melanogaster. Mol. Cell. Biol. 30, 793805.
Schultz, J.R., Tu, H., Luk, A., Repa, J.J., Medina, J.C., Li, L., Schwendner, S.,
Wang, S., Thoolen, M., Mangelsdorf, D.J., et al. (2000). Role of LXRs in control
of lipogenesis. Genes Dev. 14, 28312838. 10.1101/gad.850400.
Sieber, M.H., and Thummel, C.S. (2009). The DHR96 nuclear receptor controls
triacylglycerol homeostasis in Drosophila. Cell Metab. 10, 481490.
van der Veen, J.N., van Dijk, T.H., Vrins, C.L., van Meer, H., Havinga, R.,
Bijsterveld, K., Tietge, U.J., Groen, A.K., and Kuipers, F. (2009). Activation of
the liver X receptor stimulates trans-intestinal excretion of plasma cholesterol.
J. Biol. Chem. 284, 1921119219.
van der Velde, A.E., Brufau, G., and Groen, A.K. (2010). Transintestinal cholesterol efflux. Curr. Opin. Lipidol. 21, 167171.
Cell Metabolism
Erratum
Enteric Neurons and Systemic Signals
Couple Nutritional and Reproductive Status
with Intestinal Homeostasis
Paola Cognigni,1 Andrew P. Bailey,2 and Irene Miguel-Aliaga1,*
1Department
of Zoology, University of Cambridge, Downing Street, Cambridge CB2 3EJ, UK

of Developmental Neurobiology, MRC National Institute for Medical Research, The Ridgeway, Mill Hill, London NW7 1AA, UK
*Correspondence: im307@hermes.cam.ac.uk
DOI 10.1016/j.cmet.2011.12.010
2Division
(Cell Metabolism 13, 92104; January 5, 2011)

In the Experimental Procedures of the above paper (under Defecation and Intestinal Assays using Food Dyes), the long chemical
name for Fluoropoo (CholEsteryl BODIPY FL C12) was incorrectly stated as that of BODIPY alone. The correct name and source are as
follows: CholEsteryl 4,4-Difluoro-5,7-Dimethyl-4-Bora-3a,4a-Diaza-s-Indacene-3-Dodecanoate (CholEsteryl BODIPY FL C12),
Invitrogen (catalog number C-3927MP).
Although this mistake does not change any of the results or conclusions of the study, the authors regret any confusion it may have
caused.

Metabolismo Celular

Încărcat de

Informații document

Titlu original

Drepturi de autor

Formate disponibile

Partajați acest document

Partajați sau inserați document

Opțiuni de partajare

Vi se pare util acest document?

Este necorespunzător acest conținut?

Drepturi de autor:

Formate disponibile

Metabolismo Celular

Încărcat de

Drepturi de autor:

Formate disponibile

Cell Metabolism

adipose tissue, the truth is that healthy

that promote weight loss and improved

Cell Metabolism 15, January 4, 2012 2012 Elsevier Inc. 1

need presented by increasing rates of

Kim, D.H., Woods, S.C., and Seeley, R.J. (2010).

Kolonin, M.G., Saha, P.K., Chan, L., Pasqualini, R.,

Barnhart, K.F., Christianson, D.R., Hanley, P.W.,

Pajvani, U.B., Trujillo, M.E., Combs, T.P., Iyengar,

Reactive Oxygen Species Resulting

Metabolism, suggests that accumulating

2 Cell Metabolism 15, January 4, 2012 2012 Elsevier Inc.

mtDNA lacks protective histones. Accumulation of damage in the mitochondrial

need presented by increasing rates of

Kim, D.H., Woods, S.C., and Seeley, R.J. (2010).

Kolonin, M.G., Saha, P.K., Chan, L., Pasqualini, R.,

Barnhart, K.F., Christianson, D.R., Hanley, P.W.,

Pajvani, U.B., Trujillo, M.E., Combs, T.P., Iyengar,

Reactive Oxygen Species Resulting

Metabolism, suggests that accumulating

2 Cell Metabolism 15, January 4, 2012 2012 Elsevier Inc.

mtDNA lacks protective histones. Accumulation of damage in the mitochondrial

after a threshold level of mtDNA mutation

Cell Metabolism 15, January 4, 2012 2012 Elsevier Inc. 3

Jefferson Stem Cell Biology and Regenerative Medicine Center

mine) for cancer cells to burn. In response,

4 Cell Metabolism 15, January 4, 2012 2012 Elsevier Inc.

b-oxidation. Utilization of adipocytederived fatty acids was related to the

Cell Metabolism 15, January 4, 2012 2012 Elsevier Inc. 5

of Cell and Developmental Biology

Major questions in the field of epigenetics

molecule modulators; such therapeutics

6 Cell Metabolism 15, January 4, 2012 2012 Elsevier Inc.

Thus, there appears to be chromatin

Cell Metabolism 15, January 4, 2012 2012 Elsevier Inc. 7

that led to improved b cell function, insulin

8 Cell Metabolism 15, January 4, 2012 2012 Elsevier Inc.

skeletal muscle is an endocrine organ,

Cell Metabolism 15, January 4, 2012 2012 Elsevier Inc. 9

promoting activation of inflammasome-dependent caspase-1,

high-fat diet feeding (Wen et al., 2011; McGillicuddy et al., 2011).

the deleterious effects of high-fat diet feeding. For example,

example, palmitate acts as a ligand for TLR4 and induces IL-6

Figure 2. Activation of the Inflammasome

16 Cell Metabolism 15, January 4, 2012 2012 Elsevier Inc.

Gregor, M.F., and Hotamisligil, G.S. (2011). Inflammatory mechanisms in

Mandrup-Poulsen, T., Pickersgill, L., and Donath, M.Y. (2010). Blockade of

Cell Metabolism 15, January 4, 2012 2012 Elsevier Inc. 17

18 Cell Metabolism 15, January 4, 2012 2012 Elsevier Inc.

more than 40% of patients diagnosed with diabetes do not

Efficacy Profile (D Hb A1c*)

Adverse Effects/Tolerability (other)

Sulfonylureas (glyburide, glipizide,

K+-ATP channel modulator

Moderate-good - but lacking

Hypoglycemia; weight gain;

Weight gain, edema, exacerbation

GLP-1 receptor agonist

Daily or twice-daily injection required;

Hypersensitivity reactions (rare);

could be followed by confirmation via pharmacologic modulation

A recent decline in total new drug approvals is worrisome.

Figure 1. Roadmap for Common Practices in Current Drug Discovery/Development