letters to nature

15. Ota, Y. & Samelson, L. E. The product of the proto-oncogene c-cbl: a negative regulator of the Syk
tyrosine kinase. Science 276, 418±420 (1997).
16. Holm, L. & Sander, C. Dali: a network tool for protein structure comparison. Trends Biochem. Sci. 20,
478±480 (1995).
17. Kretsinger, R. H. EF-hands embrace. Nature Struct. Biol. 4, 514±516 (1997).
erratum
18. Essen, L. O., Perisic, O., Cheung, R., Katan, M. & Williams, R. L. Crystal structure of a mammalian
phosphoinositide-speci®c phospholipase Cd. Nature 380, 595±602 (1996).

Protein translation and folding

19. de Beer, T., Carter, R. E., Lobel-Rice, K. E., Sorkin, A. & Overduin, M. Structure and Asn-Pro-Phe
binding pocket of the Eps15 homology domain. Science 281, 1357±1360 (1998).
20. Becker, S., Groner, B. & Muller, C. W. Three-dimensional structure of the Stat3b homodimer bound
to DNA. Nature 394, 145±151 (1998).
21. Chen, X. et al. Crystal structure of a tyrosine phosphorylated STAT-1 dimer bound to DNA. Cell 93, are coupled by an
827±839 (1998).
22. Hatada, M. H. et al. Molecular basis for the interaction of the protein tyrosine kinase ZAP-70 with the
T-cell receptor. Nature 377, 32±38 (1995).
endoplasmic-reticulum-
resident kinase
23. Eck, M. J., Shoelson, S. E. & Harrison, S. C. Recognition of a high-af®nity phosphotyrosyl peptide by
the Src homology-2 domain of p56lck. Nature 362, 87±91 (1993).
24. Otwinowski, Z. & Minor, W. Processing of X-ray diffraction data collected in oscillation mode.
Methods Enzymol. 276, 307±326 (1997).
25. Collaborative Computational Project Number 4. The CCP4 suite: Programs for protein crys- Heather P. Harding, Yuhong Zhang & David Ron
tallography. Acta Crystallogr. D 50, 760±776 (1994).
26. Jones, T. A., Zhou, J. Y., Cowan, S. W. & Kjeldgaard, M. Improved methods for building protein Nature 397, 271±274 (1999)
..................................................................................................................................
models in electron density maps and the location of errors in these models. Acta Crystallogr. A 47,
110±119 (1991). As a result of a production error, the arrows in Fig. 3d were
27. Brunger, A. X-PLOR Version 3.0: A System for Crystallography and NMR (Yale University Press, New inappropriately aligned. The correct ®gure is reproduced below.
Haven, 1992).
28. Navaza, J. (ed.) AMoRe: A New Package for Molecular Replacement (SERC, Daresbury, UK, 1992).
29. Kraulis, P. J. MOLSCRIPT: a program to produce both detailed and schematic plots of protein d
structures. J. Appl. Crystallogr. 24, 946-950 (1991).
30. Nicholls, A., Sharp, K. A. & Honig, B. Protein folding and association: insights from the interfacial and
thermodynamic properties of hydrocarbons. Proteins Struct. Funct. Genet. 11, 281-296 (1991).

Acknowledgements. We thank C. Dahl for synthesis and puri®cation of the ZAP-70 phosphopeptide, the
staff at MacCHESS for assistance with data collection, and S. Harrison and T. Roberts for comments on
the manuscript. M.J.E. is a recipient of a Burroughs±Wellcome Career Award in the Biomedical Sciences.
Diffraction data were recorded at the Cornell High Energy Synchrotron Source (CHESS), which is Perk ∆C µ-chain
supported by grants from the NSF and NIH.

Correspondence and requests for materials should be addressed to M.J.E. (e-mail: eck@red.dfci.harvard.
edu). Atomic coordinates have been deposited with the Protein Data Bank (Brookhaven National
Laboratory) under accession numbers 2cbl and 1b47.

CHOP CHOP

90
© 1999 Macmillan Magazines Ltd NATURE | VOL 398 | 4 MARCH 1999 | www.nature.com

Protein translation and folding
are coupled by an
endoplasmic-reticulum-
resident kinase
Heather P. Harding, Yuhong Zhang & David Ron
Skirball Institute of Biomolecular Medicine, Departments of Medicine and
Cell Biology and the Kaplan Cancer Center, NYU School of Medicine, New York,
New York 10016, USA
.........................................................................................................................
Protein synthesis and the folding of the newly synthesized pro-
teins into the correct three-dimensional structure are coupled in
cellular compartments of the exocytosis pathway by a process that
modulates the phosphorylation level of eukaryotic initiation
factor-2a (eIF2a) in response to a stress signal from the endo-
plasmic reticulum (ER)1,2. Activation of this process leads to
reduced rates of initiation of protein translation during ER
stress3. Here we describe the cloning of perk, a gene encoding a
type I transmembrane ER-resident protein. PERK has a lumenal
domain that is similar to the ER-stress-sensing lumenal domain of
the ER-resident kinase Ire1, and a cytoplasmic portion that
contains a protein-kinase domain most similar to that of the
known eIF2a kinases, PKR and HRI. ER stress increases PERK's
protein-kinase activity and PERK phosphorylates eIF2a on serine
residue 51, inhibiting translation of messenger RNA into protein.
These properties implicate PERK in a signalling pathway that
attenuates protein translation in response to ER stress.
In eukaryotic cells, the folding of proteins that are destined to be
secreted or membrane-bound takes place in the ER. This process is
impeded when cells are deprived of essential nutrients or energy
sources or exposed to toxins that perturb the specialized environ-
ment of the ER4. The accumulation of incorrectly folded proteins
triggers an ER-stress response that has at least two distinct compo-
nents. The ®rst, known as the unfolded-protein response (UPR),
consists of the transcriptional induction of genes that encode ER-
resident proteins such as Grp78, Grp94 and protein disulphide
isomerase. These proteins are believed to promote the folding of
newly synthesized peptides in the ER lumen5,6. The second compo-
nent consists of a profound and rapid repression of protein
synthesis3. Both responses can be rationalized in terms of the need
to relieve ER stress: the ®rst response increases the capacity of the ER
to actively fold proteins and the second decreases the demands
made on the organelle by attenuating protein-synthesis rates2.
The mediators of the UPR have been well characterized, ®rst in
yeast7,8 and, more recently, in mammalian cells9,10. The sensor of ER
stress in both types of cell is an ER-resident transmembrane kinase,
IRE1, which transduces a lumenal signal imparted by the presence
of incorrectly folded proteins to a nuclear event that results in the
increased transcription of the previously mentioned genes11.
Reduced protein synthesis in response to ER stress is associated
with polysome disassembly and correlates with increased phosphor-
ylation of eIF2a (refs 1, 3). Phosphorylated eIF2a interferes with the
formation of an active 43S translation-initiation complex12 and
expression of a non-phosphorylatable mutant eIF2a (with a muta-
tion of residue S51 to A; S51A) attenuates translational inhibition by
ER stress13. Two different eIF2a kinases have been identi®ed in
mammalian cells, haem-regulated eIF2a kinase (HRI)14 and the
interferon-inducible RNA-dependent protein kinase (PKR)15. Ca2+
release from ER stores enhances the kinase activity of PKR16 and
dominant-negative mutant forms of PKR attenuate the inhibition
of translation mediated by Ca2+ release13. Although both of these
results are consistent with a role for PKR in translational inhibition
NATURE | VOL 397 | 21 JANUARY 1999 | www.nature.com
© 1999 Macmillan Magazines Ltd
letters to nature
by ER stress, the level of inhibition of protein synthesis is nearly the
same in wild-type and pkr-/- cells (see below). This suggests the
existence of alternative eIF2a kinases that may respond to ER stress.
A search of the nucleotide databases (GenBank) identi®ed a
Caenorhabditis elegans cosmid, CEF46C3, that contains a gene
encoding a predicted type I transmembrane protein (one with its
amino terminus in the lumen) with a cytoplasmic portion that has a
protein-kinase domain most similar to that of PKR and HRI and an
external/lumenal domain that shares blocks of identity with
C. elegans and mammalian Ire1. A human expressed sequence tag
(EST) that encodes a peptide fragment with similarity to this
C. elegans protein was used as a hybridization probe to clone a
full-length complementary DNA encoding a murine homologue.
The murine protein, of 1,114 residues, is ubiquitously expressed and
has the same predicted type I transmembrane topology and ,30%
overall identity to the C. elegans protein. Its N terminus has blocks
of identity with mammalian Ire1, totalling 20% identity, and its
carboxy terminus is ,40% identical to PKR (Fig. 1a, b). A
C-terminal Myc-tagged derivative of the full-length protein, when
expressed in COS-1 cells, co-localized with endogenous ribophorin
(an ER-membrane marker; Fig. 1c). Increased mobility of the
metabolically labelled protein doublet in immunoprecipitates
treated with either endoglycosidase H or peptide N-glycosidase F
indicates that it is a glycoprotein that is not transported to the distal
Golgi complex and resides in the ER. We do not known why there
are two species of the protein in the transfected cells (Fig. 1d). On
the basis of its similarity to PKR and its localization in the ER
membrane, we named the protein PERK (for PKR-like ER kinase).
We proposed that PERK, like Ire1, might respond to ER-stress
signals and transduce these to changes in protein-synthesis rates
through phosphorylation of eIF2a.
The puri®ed kinase domain of PERK, expressed as a fusion
protein with glutathione-S-transferase (GST) in Escherichia coli
undergoes autophosphorylation in vitro (Fig. 2a). The active
kinase phosphorylates puri®ed eIF2a effectively, indicating that
the latter is a direct substrate (Fig. 2a, middle). Phosphorylation of
eIF2a takes place on S51 (Fig. 2b), the same residue that is targeted
by PKR and HRI12. Treatment of translation-competent reticulocyte
lysate with puri®ed, bacterially expressed PERK leads to a profound
inhibition in mRNA translation. Replacing K618 of PERK with
alanine (K618A), a mutation that abolishes the ability of the protein
to undergo autophosphorylation or to phosphorylate eIF2a
(Fig. 2a, lane 2, top and middle), also abolishes its ability to inhibit
translation (Fig. 2c, lane 2). Cells transfected with PERK expression
plasmids accumulate large quantities of the protein, and when levels
of expression reach a certain threshold PERK's kinase activity is
activated (Fig. 3c, compare lanes 2, 9). This correlates with pro-
found inhibition of mRNA translation (Fig. 2d, e), indicating that
the effects of PERK on protein translation observed in vitro in the
reticulocyte lysate also occur in cells in vivo.
Mouse embryonic ®broblasts that are nullizygous for pkr are not
impaired in their ability to reduce levels of protein synthesis when
treated with agents that induce ER stress (Fig. 3a). This indicates
that another eIF2a kinase, perhaps PERK, may be involved in the
attenuation of translation under these circumstances. Results of
metabolic labelling experiments show that PERK is a phosphopro-
tein. In response to ER stress, it undergoes hyperphosphorylation
and an associated shift in mobility during SDS±polyacrylamide-gel
electrophoresis (Fig. 3b). The shift is not observed for the inactive
K618A mutant, is reduced by dephosphorylation in vitro and
correlates with the autokinase activity of immunoprecipitated
PERK (Fig. 3b, c). These ER-stress-induced changes in PERK are
consistent with intermolecular or intramolecular autophosphoryla-
tion, as has been suggested for the activation of the other known
eIF2a kinases, HRI17 and PKR15. The stress-activated conforma-
tional shift in PERK is independent of synthesis of new PERK
protein (Fig. 3c, lane 4), and treatment with arsenite, heat-shock or
271
letters to nature
irradiation with ultraviolet lightÐconditions that trigger a cyto-
plasmic but not an ER stress response2 Ðinduce only a minimal
shift in PERK's mobility (Fig. 3c, lanes 5±7). Thus the change in
conformation is a post-translational modi®cation of PERK that
correlates with the ER-stress-dependent activation of the kinase.
The structural similarity between PERK and Ire1 indicates that
both proteins may use a similar mechanism to transduce ER stress.
This theory is supported by the observation that overexpression of
the lumenal domain of PERK interferes with the induction of the
UPR marker chop18,19 by ER stress (Fig. 3d). We have found
previously that Ire1 activates the chop gene and that overexpression
of the truncated lumenal domain of Ire1 attenuates chop induction
by the UPR10. Together these results are consistent with a model in
which the lumenal domains of Ire1 and PERK share activating
upstream signals.
The UPR and translational inhibition by ER stress proceed by

a c
mPERK 1 ME R A T R PG P R A L L L L L F L L L GC A AG I S A V A P A RS L L A P AS E T V F G L G A A A A P T S A A R V P A V A T A E V T V ED 70
cePERK 1 - - - - - - - - - - - MS V Y Y I V L A G - - - - - - - - - - - - - - - - - - - - - - F L L F M A L V P F N A G Q Q Y I D D D I E V V S S C 37
mIRE1b 1 - - - - - - - - - M A R P V Q R F Q L WS - - - - - - - - - - - - - P L G F L L Q L V T L L G K - L G P Q V Q S - - - - - - - - - - - - - - 33

mPERK 71 A E A L P A A A G E P E S R A T E P D D D V E L R P R G R S L V I I S T L D G R I A A L D A E N D G K K QWD L D V G S G S L V S S S L S K 140

cePERK 38 QN G Y VQN AGC D QN S Q - - - - - - - - - - - - - VN I I I T A T L D G V V T A L D G E T - G EM I WR Y ED A P L L RG T L S T S D 93
mIRE1b 34 - - - - - - - - - - - - - - - - - - - - - - - - - V R P E S L L F V S T WD G S L H A L N K Q T - G D L KW T V K D D P I I Q G P M Y V T E 77

mPERK 141 P E V F G - - N K M I I P S L D G D L F QWD R D R E - S M E A V P F T V E S L L E S S Y K F G D D V V L V G G - KS L I T YGL S A YS G 206

cePERK 94 P I D I GG T S L Q L MP T L DG RL F S Y T HN T N - L I E P L P I T T DS L L ES T I RL GQD A V AGG - - KS V T T KGFD L F TG 160
mIRE1b 78 MA F L S - - - - - - D P A - D GS L Y V L G T H K L QG L MK L P F T I P E L VH AS PC RS S D G V F Y T G R K Q D AW F V V D P E S G 140
¥
mPERK
cePERK
207
161
K L R Y I C S A L G C R RWD S D E M E E E E D I L L L Q R T Q K T V R A V G P R S G S E K WN F S V G H F E L
E Q K Y E C S M E S C G P G D T E T P K N P - - I I L I R R T T N S I R A M D T L R G I E RWN L S T A E I G V
RY I PDME T RAGF I E
- - - - - - - T L AGG I T
276
221
PERK
mIRE1b 141 E T Q M T L T T E G L S T P - - - - - - - - - - Q L F I G R T Q Y T V S M H D L R T P A L RWN T T Y R R Y S A P - - - - - - - - - - - - - 187

mPERK 277 S T F K P G G N K E D S K I I S D V E E Q E A T M L D T V I K V S V A DW K V M A F S R K G G R L EW E Y Q F C T P I A S - - AW L V RD G 344

cePERK 222 S - - - - - - - - - - P T L V S D V K - - - - - - - - - - I L L Q P P D G V I V A V D - K Y N R E EW K T N V D G H I V S - - VWQ V Y G N 268
mIRE1b 188 - - - - - - - - - - - L L N G S P G K - - - - Y MS H L T S C G MG L L L T V D P G S - - - G I V L W T Q D L G V P V T G I Y T WH Q D G L 239

mPERK 345 - K V I P I S L F D D T S - Y T AS E E A L GD E - - - - - - - - ED I V E A A RG A T EN S V Y L GMY RGQ L Y L QS S V R VS E - - - 401

cePERK 269 - Q I G E I S I F D P S N I F T T Q Y E V MQ R E - - - - - - - - Q H N L Q T Q S S - - - - L L Y MG T S N G F P F I I Q S P K A K N - - - 322
mIRE1b 240 H Q L P H L T L A R D T L H F L V L RWG H I R L P A S S Y Q D T A T Q F S S L D T Q L L M T L Y V G K E E A G F Y V - S K A L V H A G V A 308

mPERK 402 - - - - - - - - - - - K F P T S P K A L E S V N G E N A I I P L P T I KW K P L I H S P S R T P V L V G S D E F D K C L S N D K Y S H E E - 459

cePERK 323 - - - - - - - - - - - N L K Q RMN A L P - - - - E L S T M T E L T N P R F C T A N E E T RS L A Y N V K D E T L R L V L H N A F RH S Q S 377
mIRE1b 309 L V P R G L T L A P MD G P T T D E V T L Q V S G E R E G S P S T A V R Y P S G S V A L P S QW L L I G Y H E P P P V L H T T M L R V H P I 378
Ribophorin
mPERK 460 - - - - - - - - YS N G A L S I L Q Y P YD N G Y Y L P Y Y K R E RN K RS T Q I T - - - - - - V R F L D S PH YS KN I R K KD P I L L L 515
cePERK 378 K A I ED KS L S GS S A R R K L Q I I AS D T E VS AQ R I G T EN L RS T S VS KS GD YG Y L V L ES E PQ R V K F K VN S P I T L M 447
mIRE1b 379 PG K VS A E T RAS ED L - - - - H A P P V F F E L L N L RRED - - P E L H P E E K AS DS Y PG L GS QD - - - - - - - - - - - - - - 428

mPERK 516 H WW K E I F G T I L L C I V A T T F I V R R L F H P Q P H R Q R K - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - 549

cePERK 448 - - - Q T I F S Y I F N P T A V VS F L AG L I G V T V A V V YN K I A KS S P RM I EH L S S T ES A E T ES AS H R T R T T S F A P T D 514
mIRE1b 429 - L L A A T F T A I L L G AW V L Y L M R 448

mPERK
cePERK
550
515
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - E S E T Q C Q T E S K Y D S V S A D V S D N S WN D M K - - - - - - - -
D E I E RF V E EGS D L S T T P I G A I H RK P L MP I E KS N I E T H T Q PQ I K P VQ RL V K T D I D T D EDS F S ND E K K RL L R
577
584
d

mPERK 578
*
- - - - - - - - Y S G Y V S R Y L T D F E P I Q C MG RG G F G V V F E A K N K V D D C N Y A I K R I R L P N R E L A R E K V M R E V K A L 639
PNGase-F: – + –
cePERK 585 N R T I S RS S L E G F T S R F A N E F E V K K V I G H G G F G V V F R A Q S I T D MN E Y A V K R I A V A D N D K A RN R V L R E A R A L 654 Endo H: – – +
mPKR 240 - - - - - - - - - - - - - - - - - S D F ED I E E I G L GG F GQ V F K A KH R I DG K RY A I K RV K YN - - - - - T E K A EH E VQ A L 287
Mr(K)
mPERK 640 A K L E H P G I V R Y F N AW L E T P P E KWQ E E MD E I W L K D - - E S T D W P L S S P S P MD A P S V K I R R MD P F S T K E Q I E V 707
cePERK 655 A M F D H P G I I R Y F Y AW E E R P P K G F Q E K E D E N L L G K I K A E K L A K L H E I K K A K K H T S E G K R V RS A D T A S F A E S 724
mPKR 288 A E L N H V N I V Q Y H S CW - E G V D Y D P E H S MS D T S R Y K - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - 320 200
mPERK 708 I A PS P E RS RS F S VG I S C GQ T S S S ES Q F S P L E F S G T D C GD N - - S D S AD A A YN L QD S C L T D C ED V ED G T VD G 775
cePERK 725 F A M P P V V G N T T D A E N S WS T S A K P Q E V G A K R T T S E S K L G L H G G S D R T T A E L K E E S V A F S E S D E E S D T T E D S 794
mPKR 321 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - 320

mPERK 776 N D EGH S F E L C PS E AS P Y T RS R EG T S S S I V F ED S GC GN AS S K E E P RGN R L H D GN H Y VN K L T D L KC S S S RS S 845

cePERK 795 S S S D ES PS S S S GS S I DD E P K K YNS S S GG I E F VDGS DD VD - - - - - - - - - - - - - N E A V K E T K K E I A V I E E L S 851
mPKR 321 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - 320 Uncleaved
mPERK 846 S E A T T L S T S P T R P T T L S L D F T K N T V G Q L Q P S S P K V Y L Y I Q MQ L C R K E N L K D WMN R R C S L E D R E H G V C L H I 915 Cleaved
cePERK 852 I H N R V RH F V T L S A M I V E T E N Q E L E V R E RN D T G D C A Y L Y I V MQ L C A E K T L E DW I R RS K T M E S R P L F T M K NW 921
mPKR 321 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - T R C L F I Q M E F C D K G T L E QWM R - N R N Q S K V D K A L I L D L 356

mPERK 916 F L Q I A E A V E F L H S K G L MH RD L K P S N I F F T MD D V V - - - - - K V G D F G L V T A MD Q D E E E Q T V L T P M P A Y A T H T 980

cePERK 922 I KQ L AS G L E Y L H N KG F I H RD L K PGN V F F S L D S T H GH Q I L K I GD L G L A T K T D G A P K - I T V RQD S D S S A KH T 990
mPKR 357 Y EQ I V T G V E Y I H S KG L I H RD L K PGN I F L - VD E RH - - - - I K I GD F G L A T A L E - - - - - - - - - N D G KS - - - R T 409 97
mPERK 981 G Q V G T K L Y MS P E Q I H G N N Y S H K V D I F S L G L I L F E L L Y P F S T Q M E R V R I L T D V RN L K F P L L F T Q K Y P Q E H M 1050
cePERK 991 K N V G T RS Y MS P E Q L K H Q Q Y T E K V D I F A L G L V A T E L I I S F S T A S E R I H T F A D F Q K G D I P - A I L D N V P E S RD 1059
mPKR 410 R R T G T L Q Y MS P E Q L F L K H Y G K E V D I F A L G L I L A E L L H T C F T E S E K I K F F E S L R K G D F S N D I F D N K E K S - - 477
1 2 3
mPERK 1051 M V Q D M L S P S P T E R P E A T D I I E N A I F E N L E F P G K T V L RQ RS RS MS S S G T K H S RQ P S C S Y S P L P G N 1114
cePERK 1060 F L L QL TS L E PS E RP - T AH E V A TH K F L Q 1085
mPKR 478 L L K K L L S E K P K D R P E T S E I L K T L A EW RN I S E K K K RN T C 515

32 521 536 589 1,114

b PERK
SP TM
18 444 468
Ire1
mPKR
Sensor domain, ER-lumenal
Figure 1 perk encodes an ER-resident protein with similarity to Ire1 and PKR. a,
Alignment of the peptide sequence of mouse PERK, its putative C. elegans
orthologue, mouse Ire1-b and mouse PKR. Identical residues are shaded. The
predicted N-linked glycosylation site, conserved in PERK and Ire1-b, is identi®ed
by a `¥' symbol. The invariant lysine (in kinase domain 2 of PERK) is marked by an
asterisk and the predicted transmembrane domain of PERK is underlined. b, The
predicted peptide features of PERK, Ire1 and PKR. SP, signal peptide; TM,
transmembrane domain. The lumenal domain is cross-hatched and the predicted
Kinase
977
242 515
Effector domain, cytoplasmic
two lobes of the kinase domain, conserved between PERK and PKR and
separated by a large insert in the former (white box), are black. c, Photomicro-
graphs of COS-1 cells co-stained for Myc-tagged PERK and endogenous
ribophorin (a marker for rough ER). d, Autoradiogram of Myc-tagged PERK
immunoprecipitated from metabolically labelled COS-1 cells and treated in vitro
with the glycosidases endoglycosidase H (Endo H) or peptide N-glycosidase F
(PNGase-F). The migration of the normal (uncleaved) and deglycosylated
(cleaved) PERK doublets is indicated.

272
© 1999 Macmillan Magazines Ltd NATURE | VOL 397 | 21 JANUARY 1999 | www.nature.com
 letters to nature
a e 1000

Forward scatter
b WT KA
GST-PERK: WT KA WT KA PERK: – – WT WT KA KA
AP: – – + + eIF2α: WT 51 WT 51 WT 51 PERK
32
P kinase P-2α 0 Non-transfected
2α 100 101 102 103
32
P eIF2α 1 2 3 4 5 6
1000

Forward scatter
Western
H33258
1 2 3 4 d – WT KA
c Wild-type
PERK 35 0
S pulse
Time (min): 15 15 1 15 100 101 102 103 – WT KA
GST-PERK: – KA WT WT PERK 1000

Forward scatter
GFP
mRNA mRNA
35S- Tubulin Tubulin
Luciferase
Haemoglobin PERK Western K618A mut PERK Western
0
1 2 3 4 100 101 102 103
1 2 3 GFP activity 1 2 3

Figure 2 PERK is an eIF2a kinase that inhibits translation. a, Top, autoradiogram of
the in vitro autokinase activity of the wild-type (WT) and K618A mutant (KA) PERK,
puri®ed as GST-fusion proteins from E. coli. Middle, autoradiogram of eIF2a that
had been added to the kinase reaction and immunoprecipitated with a speci®c
antiserum. Bottom, PERK western blot of same lysate untreated or treated with
alkaline phosphatase (AP). b, Autoradiogram of 35S-labelled wild-type and S51A
mutant (51) eIF2a that had been incubated with puri®ed wild-type and mutant
PERK. The phosphorylated (p-2a) and non-phosphorylated (2a) forms of eIF2a
were resolved by isoelectric focusing. c, Autoradiogram of 35S-labelled luciferase
translated from mRNA in reticulocyte lysates that had been pre-incubated with
puri®ed wild-type and mutant PERK for the indicated times. Haemoglobin staining
serves as a loading control. d, Reduced synthesis rates of wild-type compared
with mutant PERK. Top, autoradiogram of 35S-labelled PERK, immunoprecipitated
from pulse-labelled COS-1 cells 48 h after transfection with the indicated PERK
expression plasmids. Middle, northern blot of PERK and a-tubulin mRNA from
identical plates, showing similar levels of wild-type and mutant PERK mRNA.
Bottom, immunoblot showing similar levels of both PERK proteins and revealing
that all the wild-type PERK in these cells had been activated. e, Interference with
translation of a co-expressed green ¯uorescent protein (GFP) reporter gene by
PERK. Cells transfected with a reporter plasmid expressing both GFP and wild-
type or K618A mutant PERK were analysed by ¯uorescence-activated cell sorting
for GFP activity (left panels), stained for PERK expression (top right four panels)
and analysed for GFP reporter mRNA by northern blot and PERK protein by
western blot (lower left panels).

a pkr +/+ pkr -/- c Time (h): 20 40

Treatment: – Tun Tg – Tun Tg
200 Protein: – WT KA WT KA WT KA
Treatment: – – Tg Tg* As HS UV Tg – – AP AP
97 P-PERK
PERK
69 1 2 3 4 5 6 7 8 9 10 11 12

45
d
30
Labelling (%): 100 56 13 100 67 7
1 2 3 4 5 6

b 3T3 COS1
Antibody: PI anti-PERK anti-PERK
Protein: – – – – – WT KA
Treatment: Tun – Tun Tg – – –
200 Perk ∆C µ-chain

PERK
in vivo
32
P

in vitro
(IP kinase)
1 2 3 4 5 6
-/-
Figure 3 PERK is activated by ER stress. a, pkr cells are capable of attenuating
translation in response to ER stress. Autoradiogram of total proteins synthesized
in wild-type (+/+) and mutant (-/-) mouse embryo ®broblasts that had been
untreated or exposed to tunicamycin (Tun) or thapsigargin (Tg) before a 30-min
35
pulse label with S-methionine. b, Top, autoradiogram of endogenous PERK
immunoprecipitated from NIH3T3 cells (lanes 1±4) or of exogenous wild-type or
K618A mutant PERK from COS-1 cells transfected with the indicated expression
vectors (lanes 5±7). Cells had been preloaded with 32P-labelled orthophosphate
and exposed to a brief treatment with tunicamycin or thapsigargin as indicated.
Lane 1 was immunoprecipitated with preimmune serum (PI). Bottom, in vitro
autophosphorylation of PERK immunoprecipitated (IP) from cells treated as
above. The left and right panels are exposures of the same gel that differ by a
factor of 10 (left . right). c, Immunoblot of PERK in lysates of COS-1 cells
97
7 CHOP CHOP
transfected with wild-type (lanes 2±7, 9, 11) or K618A mutant (lanes 8, 10, 12)
PERK and collected 20 (lanes 1±8) or 40 (lanes 9±12) h after transfection. Lanes 3,
4 and 8 were treated with thapsigargin; lane 4 was pretreated with anisomycin
(asterisk) to block protein synthesis. As, arsenite; HS, heat-shock; UV, ultraviolet
irradiation. Extracts from lanes 11 and 12 were treated with alkaline phosphatase
(AP) before being loaded on the gel. P-PERK, phosphorylated PERK on the gel. d,
Immunostaining of CHOP in tunicamycin-treated NIH3T3 cells. Upper panels,
photomicrographs of cells expressing the lumenal domain of PERK (PERKDC) or
the human immunoglobulin m-chain, a protein that is retained in the ER and serves
as a control for the effects of PERKDC. Lower panels, photomicrographs of the
same cells stained with an antibody to CHOP. The arrows point to the cells
expressing the exogenous PERKDC or m-chain.

NATURE | VOL 397 | 21 JANUARY 1999 | www.nature.com

© 1999 Macmillan Magazines Ltd 273
letters to nature
parallel yet distinct pathways20. Our ®ndings suggest considerable
homology between the two processes: both are initiated by trans-
membrane ER-resident kinases, Ire1 or PERK, which have similar
lumenal domains, and both kinases undergo ER-stress-induced
activation, a step that may involve an activating autophosphoryla-
tion event21,22 (Fig. 3). Once activated, however, the two kinases
appear to deliver distinct signals. On the basis of comparisons with
yeast, mammalian Ire1 probably acquires endonuclease activity to
cleave speci®c mRNA substrates9 and effect enhanced expression of
upstream activators of the UPR23. PERK, on the other hand,
phosphorylates eIF2a and attenuates translation. Thus, in mam-
malian cells two signalling pathways that are responsive to ER stress
appear to have evolved by appending different effector functions to
otherwise similarly structured upstream transducers.
Note added in proof : The rat orthologue of PERK has recently been
identi®ed as an elF2a kinase that is expressed at high levels in
the pancreas, an organ that is heavily engaged in protein secretion
and ER traf®cking26. This ®nding underscores the potential
physiological signi®cance of PERK's role in coupling folding to
translation. M 3. Wong, W. L., Brostrom, M. A., Kuznetsov, G., Gmitter-Yellen, D. & Brostrom, C. O. Inhibition of
.........................................................................................................................
Methods
Cloning of Perk cDNA. We used the insert from human EST clone IMAGE
number 746909 as a hybridization probe to screen 3 3 106 clones from an
unampli®ed NIH3T3 ®broblast cDNA library in l-Zap.express (Stratagene).
We identi®ed 19 clones.
Expression plasmids. A C-terminal Myc tag and K618A mutation were
incorporated into PERK by altering the Perk cDNA by polymerase chain
reaction. The modi®ed cDNAs were ligated into the pcDNA1/Amp expression
vector (Invitrogen). PERK lacking the C terminus (PERKDC) was created by an
internal SnaBI±SmaI deletion (removing amino acids 582±1,081 of the
protein) of the Myc-tagged full-length cDNA. A bacterial expression plasmid
encoding a GST-fusion protein with the cytoplasmic domain of PERK (amino
acids 537±1,114) was constructed in pGEX 4T1 (Pharmacia). The puri®ed
fusion protein, of relative molecular mass ,93,000 (Mr ,93K), was used in the
in vitro assays described below and to raise rabbit antisera.
+/+
Cell culture and transfection. COS-1, NIH3T3 and pkr and pkr-/- mouse
embryonic ®broblasts were maintained in DMEM supplemented with 10%
fetal bovine serum. Unless otherwise indicated, cells were treated with 2 mg ml-1
tunicamycin for 4 h, 1 mM thapsigargin for 30 min, 100 mM sodium arsenite for
2 h, and heat-shock at 42 8C for 2 h; pretreatment was with 27 J m-2 ultraviolet
light or 10 mg ml-1 anisomycin for 45 min. COS-1 cells were transfected with
3 mg plasmid DNA per 60-cm dish by the DEAE-Dextran method. NIH3T3
cells were transfected with 1 mg DNA per 35-mm dish using Lipofectamine Plus
(Life Technologies). Immunoblots were performed with rabbit anti-PERK
serum diluted 1:10,000 nearly as described18 except that cell lysates were
prepared in a HEPES±Triton-X100 buffer (see below) and contained 10 mg
total protein. Immunostaining was done as described10.
Cell labelling and immunoprecipitation. Transfected COS-1 cells were
starved for 30 min in methionine-free DMEM containing 10% dialysed fetal
calf serum, then labelled for 2 h with 250 mCi ml-1 35S-Translabel (ICN) in the
same medium. Cells were lysed in RIPA buffer immediately after labelling
(Fig. 2d) or after a 4 h cold-chase in complete media (Fig. 1d). The labelled
protein was immunoprecipitated with the indicated antibodies and resolved by
6% SDS±PAGE. Where indicated, the immunopuri®ed PERK was digested for
16 h at 37 8C with 0.025 U endoglycosidase H or 0.25 U peptide N-glycosidase F
according to the manufacturer's instructions (Boehringer). In vivo 32P-labelling
of PERK was performed by preloading 10-cm plates of NIH3T3 cells with
125 mCi ml-1 32P-orthophosphate for 14 h before stress treatment and lysis in
HEPES±Triton-X100 buffer (20 mM HEPES, pH 7.5, 150 mM NaCl, 1%
Triton-X100, 10% glycerol, 1 mM EDTA, 10 mM sodium diphosphate, 100 mM
NaF, 17.5 mM B-glycerophosphate, 1 mM phenylmethylsulphonyl ¯uoride,
4 mg ml-1 aprotonin, and 2 mg ml-1 pepstatin A). Nuclei were removed by brief
centrifugation and the extracts were adjusted to a ®nal concentration of 0.5%
sodium deoxycholate and 0.1% SDS before immunoprecipitation with anti-
PERK antiserum (1 ml).
In vitro phosphorylation. These reactions, containing immunopuri®ed or
274
© 1999 Macmillan Magazines Ltd
recombinant bacterially expressed wild-type or K618A mutant GST±PERK
(50 ng), were done in 20 ml kinase buffer24 at 30 8C for 30 min with either 5 mCi
[g-32P]ATP (4,500 Ci mmol-1; ICN; Fig. 2a, 3b) or 0.1 mM unlabelled ATP
(Fig. 2b). Where indicated, the reactions contained ,10 ng partially puri®ed
human eIF2a, which was immunoprecipitated from the supernatant after
incubation (Fig. 2a), or 0.5 ml in vitro-translated 35S-methionine-labelled wild-
type or mutant (S51A) eIF2a (TNT reticulocyte lysate, Promega; Fig. 2b). The
products in Fig. 2b were analysed by isoelectric focusing using Pharmalytes, pH
4±6.5, as described25. To measure translational inhibition by PERK, glu-
tathione±Sepharose beads (1 ml bed volume) containing ,50 ng GST±PERK
(wild-type or K618A) were pre-incubated for 1±15 min at room temperature
with 12 ml of an in vitro translation mixture (Promega) before addition of
0.5 mg in vitro-transcribed capped luciferase mRNA. The labelled luciferase
protein was resolved by 10% SDS±PAGE.
Received 5 October; accepted 16 November 1998.
1. Prostko, C. R., Brostrom, M. A. & Brostrom, C. O. Reversible phosphorylation of eukaryotic initiation
factor 2 alpha in response to endoplasmic reticular signaling. Mol. Cell Biochem. 127±128, 255±265
(1993).
2. Brostrom, C. O. & Brostrom, M. A. Regulation of translational initiation during cellular response to
stress. Prog. Nucleic Acid Res. Mol. Biol. 58, 79±125 (1998).
protein synthesis and early protein processing by thapsigargin in cultured cells. Biochem. J. 289, 71±79
(1993).
4. Lee, A. Mammalian stress response: induction of the glucose-regulated protein family. Curr. Biol. 4,
267±273 (1993).
5. Kozutsumi, Y., Segal, M., Normington, K., Gething, M. J. & Sambrook, J. The presence of malfolded
proteins in the endoplasmic reticulum signals the induction of glucose-regulated proteins. Nature
332, 462±464 (1988).
6. Gething, M. J. & Sambrook, J. Protein folding in the cell. Nature 355, 33±45 (1992).
7. Cox, J. S., Shamu, C. E. & Walter, P. Transcriptional induction of genes encoding endoplasmic
reticulum resident proteins requires a transmembrane protein kinase. Cell 73, 1197±1206 (1993).
8. Mori, K., Ma, W., Gething, M. J. & Sambrook, J. A transmembrane protein with a cdc2+/CDC28-
related kinase activity is required for signaling from the ER to the nucleus. Cell 74, 743±756 (1993).
9. Tirasophon, W., Welihinda, A. A. & Kaufman, R. J. A stress response pathway from the endoplasmic
reticulum ot the nucleus requires a novel bifunctional protein kinase/endoribonuclease (Ire1p) in
mammalian cells. Genes Dev. 12, 1812±1824 (1998).
10. Wang, X.-Z. et al. Cloning of mammalian Ire1 reveals diversity in the ER stress responses. EMBO J. 19,
5708±5717 (1998).
11. Shamu, C. E. Splicing together the unfolded-protein response. Curr. Biol. 7, R67±R70 (1997).
12. Hinnebusch, A. G. The eIF-2 alpha kinases: regulators of protein synthesis in starvation and stress.
Semin. Cell Biol. 5, 417±426 (1994).
13. Srivastava, S. P., Davies, M. V. & Kaufman, R. J. Calcium depletion from the endoplasmic reticulum
activates the double-stranded RNA-dependent protein kinase (PKR) to inhibit protein synthesis. J.
Biol. Chem. 270, 16619±16624 (1995).
14. Chen, J. J. & London, I. M. Regulation of protein synthesis by heme-regulated eIF-2 alpha kinase.
Trends Biochem. Sci. 20, 105±108 (1995).
15. Berry, M. J., Knutson, G. S., Lasky, S. R., Munemitsu, S. M. & Samuel, C. E. Mechanism of interferon
action. Puri®cation and substrate speci®cities of the double-stranded RNA-dependent protein kinase
from untreated and interferon-treated mouse ®broblasts. J. Biol. Chem. 260, 11240±11247 (1985).
16. Prostko, C. R., Dholakia, J. N., Brostrom, M. A. & Brostrom, C. O. Activation of the double-stranded
RNA-regulated protein kinase by depletion of endoplasmic reticular calcium stores. J. Biol. Chem. 270,
6211±6215 (1995).
17. Fagard, R. & London, I. M. Relationship between phosphorylation and activity of heme-regulated
eukaryotic initiation factor 2 alpha kinase. Proc. Natl Acad. Sci. USA 78, 866±870 (1981).
18. Wang, X.-Z. et al. Signals for the stressed endoplasmic reticulum induce C/EBP homologous protein
(CHOP/GADD153). Mol. Cell. Biol. 16, 4273±4280 (1996).
19. Price, B. & Calderwood, S. Gadd45 and Gadd153 messenger RNA levels are increased during hypoxia
and after exposure of cells to agents which elevate the levels of glucose-regulated proteins. Cancer Res.
52, 3814±3817 (1992).
20. Brostrom, M. A., Prostko, C. R., Gmitter, D. & Brostrom, C. O. Independent signaling of grp78 gene
transcription and phosphorylation of eukaryotic initiator factor 2 alpha by the stressed endoplasmic
reticulum. J. Biol. Chem. 270, 4127±4132 (1995).
21. Shamu, C. E. & Walter, P. Oligomerization and phosphorylation of the Ire1p kinase during
intracellular signaling from the endoplasmic reticulum ot the nucleus. EMBO J. 15, 3028±3039
(1996).
22. Welihinda, A. A. & Kaufman, R. J. The unfolded protein response pathway in Saccharomyces cerevisiae.
Oligomerization and trans-phosphorylation of Ire1p (Ern1p) are required for kinase activation. J.
Biol. Chem. 271, 18181±18187 (1996).
23. Sidrauski, C. & Walter, P. The transmembrane kinase Ire1p is a site-speci®c endonuclease that initiates
mRNA splicing in the unfolded protein response. Cell 90, 1031±1039 (1997).
24. Wu, S. & Kaufman, R. J. A model for the double-stranded RNA (dsRNA)-dependent dimerization and
activation of the dsRNA-activated protein kinase PKR. J. Biol. Chem. 272, 1291±1296 (1997).
25. Savinova, O. & Jagus, R. Use of vertical slab isoelectric focusing and immunoblotting to evaluate
steady-state phosphorylation of eIF2 alpha in cultured cells. Methods 11, 419±425 (1997).
26. Shi, Y. et al. Identi®cation and characterization of pancreatic eukaryotic initiation factor 2 alpha-
subunit kinase, PEK, involved in translational control. Mol. Cell. Biol. 18, 7499±7509 (1998).
Acknowledgements. We thank R. Schneider for the eIF2a expression plasmids, antisera and puri®ed
protein; M. Green for the immunoglobulin m-chain expression plasmid; B. Williams for the pkr-/- cells;
G. Kreibich for the antisera against ribophorin; and X.-Z. Wang for the cDNA libary. This work was
supported by NIH grants and an NIH training grant (to H.P.H.). D.R. is a Stephen Birnbaum Scholar of
the Leukemia Society of America.
Correspondence and requests for materials should be addressed to D.R. (e-mail: ron@saturn.med.nyu.
edu). The nucelotide sequence of perk has been submitted to GeneBank under the accession number
AF076681.
NATURE | VOL 397 | 21 JANUARY 1999 | www.nature.com