Sunteți pe pagina 1din 12

Molecular Microbiology (2015) ■ doi:10.1111/mmi.

13272
Molecular Microbiology (2016) 99(5), 897–908 䊏 doi: 10.1111/mmi.13272
First published online 1 December 2015

Escherichia coli ClbS is a colibactin resistance protein


Nadège Bossuet-Greif,1,2,3,4 Damien Dubois,1,2,3,4,5 island is also found in certain strains of Enterobacter
Claude Petit,1,2,3,4,6 Sophie Tronnet,1,2,3,4 aerogenes, Citrobacter koseri and Klebsiella pneumoniae
Patricia Martin,1,2,3,4,5 Richard Bonnet,7 (Nougayrède et al., 2006; Johnson et al., 2008; Putze
Eric Oswald1,2,3,4,5 and et al., 2009; Lai et al., 2014). Epidemiological studies
Jean-Philippe Nougayrède1,2,3,4* suggest that the prevalence of pks+ E. coli is increasing in
1
INRA, USC 1360, Toulouse, France. the human microbiota in developed countries. There is a
2
Inserm, UMR 1043, Toulouse, France. shift in the E. coli population from the phylogenetic group A
3
CNRS, UMR 5282, Toulouse, France. to the newly dominant group B2, and up to 40% of B2 E. coli
4
Université de Toulouse, UPS, Toulouse, France. strains host the pks island (Nougayrède et al., 2006;
5
CHU Toulouse, Service de bactériologie-Hygiène, Johnson et al., 2008; Putze et al., 2009; Payros et al.,
Toulouse, France. 2014). Carriage of the pks island could provide a fitness
6
INP-ENVT ESC, Toulouse, France. advantage for long-term persistence in the human gut
7
Université d’Auvergne, Inserm UMR 1071, INRA USC (Nowrouzian and Oswald, 2012). Colibactin is also a viru-
2018, Clermont-Ferrand, France. lence determinant of extraintestinal pathogenic E. coli, as
its production aggravates the lymphopenia and lethality in
mouse and neonatal rat systemic infection models (Martin
Summary et al., 2013; Marcq et al., 2014; McCarthy et al., 2015).
The genomic pks island codes for the biosynthetic Exposure of mammalian cells to live pks+ bacteria induces
machinery that produces colibactin, a peptide- DNA double strand breaks (DSB) in vitro and in vivo
polyketide metabolite. Colibactin is a genotoxin that (Nougayrède et al., 2006; Putze et al., 2009; Cuevas-
contributes to the virulence of extra-intestinal patho- Ramos et al., 2010; Marcq et al., 2014). This DNA damage
genic Escherichia coli and promotes colorectal leads to gene mutations, chromosomal instability and pre-
cancer. In this work, we examined whether the pks- mature senescence that ultimately drive tumorigenesis
encoded clbS gene of unknown function could par- (Cuevas-Ramos et al., 2010; Secher et al., 2013;
ticipate in the self-protection of E. coli-producing Cougnoux et al., 2014). It was recently shown in different
colibactin. A clbS mutant was not impaired in the mouse models that pks+ E. coli strains promote intestinal
ability to inflict DNA damage in HeLa cells, but the tumor progression (Arthur et al., 2012; Cougnoux et al.,
bacteria activated the SOS response and ceased to 2014; Dalmasso et al., 2014) and modulate gut homeosta-
replicate. This autotoxicity phenotype was markedly sis (Olier et al., 2012; Payros et al., 2014; Secher et al.,
enhanced in a clbS uvrB double mutant inactivated 2015). Such pks+ B2 E. coli strains are overrepresented in
for DNA repair by nucleotide excision but was sup- human colon cancer biopsies (Arthur et al., 2012; Buc
pressed in a clbS clbA double mutant unable to et al., 2013; Cougnoux et al., 2014). In Taiwan, more than
produce colibactin. In addition, ectopic expression of 25% of K. pneumoniae isolates are pks+, two-third of which
clbS protected infected HeLa cells from colibactin. belongs to the highly virulent K1 type causing pyogenic
Thus, ClbS is a resistance protein blocking the geno- liver abscesses that are associated with a higher risk of
toxicity of colibactin both in the procaryotic and the colorectal cancer (Lai et al., 2014).
eucaryotic cells. The 54 kb pks gene cluster is highly conserved in
Enterobacteriaceae with ∼100% identical nucleotide
sequences and preserved syntheny (Putze et al., 2009).
Introduction
The chromosomal insertion at asn tRNA loci, the flanking
The genomic pks (polyketide synthase) island present in by direct repeats and by genes for integration and conju-
pathogenic but also commensal and probiotic Escherichia gation indicate it was shared recently by horizontal gene
coli of the phylogenetic group B2 encodes a biosynthetic transfer. The pks gene cluster is physically associated with
machinery allowing the production of colibactin, a geno- the High Pathogenicity Island (HPI) coding for the sidero-
toxic secondary metabolite (Nougayrède et al., 2006). This phore yersiniabactin, another polyketide metabolite (Putze
et al., 2009; Martin et al., 2013). Similar biosynthetic gene
Accepted 9 November, 2015. *For correspondence. E-mail jean-
philippe.nougayrede@toulouse.inra.fr; Tel. +33 (0)5 62 74 45 25; Fax clusters were recently identified in the honeybee gut com-
+33 (0)5 62 74 45 25. mensal Frischella perrara (43–81% protein sequence

© 2015 John Wiley & Sons Ltd


C 2015 John Wiley & Sons Ltd
V
2 N. N.
898 Bossuet-Greif et alet. al.■ 䊏
Bossuet-Greif

identity with that in Enterobacteriaceae) and in the marine defense function of the E. coli c2450 protein, hereby
sponge commensal Pseudovibrio sp. (28–65% protein called ClbS. We observed that a clbS mutant E. coli is not
identity with that in F. perrara) (Bondarev et al., 2013; Engel impaired for colibactin production but exhibits a marked
et al., 2014). It is not known whether the pks island in autotoxicity phenotype. In addition, clbS ectopic expres-
Pseudovibrio induces genotoxicity, but despite genetic sion within HeLa cells confers protection against
divergence, infection with F. perrara induces DNA DSB in colibactin-inflicted DNA DSB, demonstrating that ClbS is a
HeLa cells. This conservation of function together with the colibactin resistance protein.
spread by horizontal transfer suggest that colibactin allows
similar interactions in distinct ecological communities.
The pks island clusters the clb genes encoding a hybrid Results
nonribosomal peptide synthase (NRPS)-polyketide syn- clbS is not required for genotoxicity of pks+ E. coli
thase (PKS) multi-modular assembly line related to that
synthesizing many microbial secondary metabolites of To examine whether the clbS gene could play a role in the
polyketide and non-ribosomal peptide origin (Fischbach genotoxin synthesis or trafficking, we quantified the geno-
and Walsh, 2006). These multimodular enzymes are acti- toxicity of a wild-type pks and an isogenic clbS mutant E.
vated by ClbA, a phosphopantetheinyl transferase that coli. The laboratory E. coli strain DH10B hosting the
primes the NRPS–PKS enzymes to allow ketide–peptide BACpks, a clbS isogenic mutant or DH10B hosting the
chain elongation (Beld et al., 2014). The primed assembly empty BAC vector as a negative control, were grown to
line utilizes malonyl-coA and amino acids, including the reach the exponential growth phase (OD600 = 0.3), and
unusual nonproteinogenic aminocyclopropane-carboxic then inoculated to HeLa cells at various multiplicity of
acid, to synthesize the non-ribosomal peptide-polyketide infection (moi). After a 4 h infection, the HeLa cells were
hybrid metabolites (Bian et al., 2015; Brotherton et al., further incubated for 4 h, and the host DSBs were demon-
2015; Vizcaino and Crawford, 2015). These inactive strated by immunostaining S139-phosphorylated histone
metabolites (called precolibactin) contain a prodrug motif H2AX (pH2AX) (Rogakou et al., 1998) (Fig. 1A). HeLa
that is hydrolyzed to produce the active genotoxin colibac- cells exposed to the wild-type pks and clbS mutant strains
tin. This prodrug activation is achieved by ClbP, a exhibited similar levels of pH2AX in relation to the moi
membrane-bound protein with a periplasmic D-amino (Fig. 1B). However, when the bacteria were grown over-
peptidase activity (Dubois et al., 2011; Cougnoux et al., night before infection, the clbS mutant strain displayed a
2012; Brotherton and Balskus, 2013). The high instability marked decrease of genotoxicity compared with the wild-
of the activated colibactin has precluded its purification type pks and with the clbS mutant in exponential culture
and characterization during a decade. The relatively (Fig. 1D–E). This defect could be fully reversed by reintro-
stable intermediate biosynthesis derailment products ducing a single copy of the clbS gene with its cognate
were recently characterized, shedding light on the active promoter region at the chromosomal attTn7 site (Fig. 1E).
colibactin ‘warhead’, an unusual spirobicyclic ring system The decreased genotoxicity of the clbS mutant grown
and ultimate electrophilic species that could alkylate DNA overnight was correlated with lower colony-forming unit
or proteins (Bian et al., 2015; Brotherton et al., 2015; (CFU) counts in the cell culture medium at the end of the
Vizcaino and Crawford, 2015). infection (Fig. 1C and F). The genotoxicity of the clbS
Colibactin was recently shown to possibly induce cross- mutant grown overnight could be restored to wild-type
links on purified DNA in vitro (Vizcaino and Crawford, levels with a 10-fold increase of the moi (Supplementary
2015). This raises the question of colibactin autotoxicity Fig. S1), indicating that the decreased genotoxicity of the
and the self-resistance mechanisms in the producing bac- clbS mutant grown overnight resulted from the reduced
terium. The inactive prodrug assembly and activation by number of infecting bacteria. Together, these results
ClbP cleavage during periplasmic export was previously confirm that clbS is not required for colibactin synthesis or
proposed as a resistance mechanism (Dubois et al., 2011; trafficking, but for the fitness of the bacteria producing the
Brotherton and Balskus, 2013). In a screen using activity- genotoxin.
based probes, Kunzmann and Sieber (2012) discovered
that the c2450 protein, a pks island-encoded protein with
A clbS mutant exhibits decreased CFU numbers in the
previously unknown function (Nougayrède et al., 2006),
stationary growth phase
could bind a α-methylene-γ-butyrolactone probe with
intrinsic reactivity found in potent anticancer and antibac- To investigate the fitness of the clbS mutant, we deter-
terial drugs. Homologue c2450 proteins are encoded on mined its growth kinetics. The wild-type pks, clbS mutant
the genomic islands of Pseudovibrio and Frischella (53% and complemented mutant strains were grown in LB
and 64% amino-acid identity respectively), suggesting a (inoculated with pre-cultures in the exponential phase)
conserved function. In this work, we uncover the self- and plated at indicated times (0–24 h) to determine the
C 2015 John ©
V 2015
Wiley & John
Sons Wiley & Sons Ltd,
Ltd, Molecular Molecular 99,
Microbiology, Microbiology
897–908
ClbSClbS protects
protects fromfrom colibactin8993
colibactin

Fig. 1. Colibactin production and growth of a clbS mutant. E. coli DH10B hosting the BAC vector, or the BACpks (pks), the isogenic clbS
mutant or the clbS mutant complemented with the clbS gene inserted at the chromosomal attTn7 site, were pre-cultivated in the exponential
growth phase (panels A–C) or overnight to reach stationary phase (panels D–F) then inoculated to HeLa cells with given multiplicities of
infection (moi, number of bacteria per cell at the onset of infection). After a 4 h infection followed by 4 h post-incubation, host DNA double
strand breaks were quantified by staining S139-phosphorylated histone H2AX (pseudocolored green) relative to DNA (red) (panels A and D).
Panels B and E: Genotoxic indexes (phosphorylated-H2AX fold-induction relative to control cells). Panels C and F: Colony-forming units (CFU)
per milliliter in the interaction medium were determined after the 4 h infection.
This figure is available in colour online at wileyonlinelibrary.com.

numbers of CFUs. The clbS mutant showed an exponen- NRPS–PKS machinery completely abolishes production
tial growth phase closely similar to that of the wild-type of colibactin and a clbS clbP double mutant deficient for
pks (Fig. 2). Similar CFU numbers were observed up to the ClbP peptidase that cleaves precolibactin to release
the late exponential phase. In the stationary phase, in the active genotoxin. The drop in CFU counts was not
contrast to the wild type that showed stable counts, the observed in the clbS clbA double mutant (Fig. 3). Com-
clbS mutant exhibited a significant reduction of CFU plementation of the clbS clbA double mutant with a
counts (Fig. 2). Chromosomal complementation of the plasmid-encoded clbA restored and even increased the
clbS mutant with the wild-type gene restored CFU levels growth defect, possibly as a result of ClbA overexpression
similar to that of the wild-type pks strain (Fig. 2). Similarly, and increased synthesis of colibactin (Fig. 3). Similarly,
a plasmid-encoded clbS gene cloned with its cognate the clbS clbP double mutant did not show reduced CFU
promoter region abrogated the growth defect of the clbS counts and was fully complemented by a plasmid-
mutant (Fig. 3). It was previously observed that clb gene encoded clbP gene (Fig. 3). Thus, the reduction of CFU in
transcription peaked in the late exponential to early sta- the clbS mutant in the stationary growth phase was tightly
tionary phase, suggesting that this transition phase sup- associated with production of active colibactin.
ports maximal genotoxin production (Homburg et al.,
2007). We thus next investigated whether the decreased
Autotoxicity of colibactin in a clbS mutant induces a
numbers of CFU in the clbS mutant culture in the station-
viable-non replicating state
ary phase was indeed associated to production of colibac-
tin. We constructed a clbS clbA double mutant in which In line with an autotoxicity of colibactin, we noted a distinct
removal of the ClbA PPTase required for maturation of the colony morphology in the clbS mutant grown on agar

© 2015 John
C 2015
V John Wiley
Wiley && Sons
Sons Ltd,
Ltd, Molecular
Molecular Microbiology,
Microbiology 99, 897–908
4 N. N.
900 Bossuet-Greif et alet. al.■ 䊏
Bossuet-Greif

investigated clbS mutant viability by assessing bacterial


membrane integrity with cell impermeant dyes. The wild-
type pks, clbS and clbS clbA mutants, the complemented
mutants and the vector control strain were grown over-
night in LB, then propidium iodide (PI) was added, and the
PI-uptake by membrane-compromised (i.e. dead) bacteria
was assessed by flow cytometry. In contrast to control
heat-killed bacteria, none of the strains tested was found
to take up the dye (Fig. 4A and Supplementary Table S1).
Similar results were obtained using the lower molecular
weight dye ethidium bromide (Supplementary Table S1).
We also monitored the bacteria metabolic state, by quan-
tification of cellular ATP levels using a luciferase-based
luminescence assay. All the overnight cultures of wild-type
pks, clbS and clbS clbA mutants, and the complemented
Fig. 2. Decreased numbers of clbS mutant culturable cells in the mutants showed similar levels of ATP (Fig. 4B). Thus,
stationary growth phase. E. coli DH10B hosting the BACpks (pks),
the isogenic clbS mutant or the mutant complemented by clbS mutant bacteria producing colibactin exhibit reduced
reinsertion of the clbS gene at the chromosomal attTn7 site (clbS numbers of CFU together with largely conserved mem-
attTn7::clbS) were pre-grown in LB to reach exponential phase,
brane integrity and metabolic activity, suggesting that
2 × 106 bacteria ml−1 were inoculated in LB and grown for indicated
times before plating on LB to determine colony-forming unit (CFU) these bacteria remained viable but did not proliferate. We
numbers. Each point represents the mean and standard error of thus monitored the replication of clbS mutant bacteria in a
the mean (SEM) of three to six independent experiments. Two-way
fluorescence dilution assay. The wild-type pks and clbS
ANOVA and Bonferroni post-test, ***P < 0.001.
mutant strains transformed with a plasmid encoding an
IPTG-inducible gfp gene were grown overnight in the
plates. The clbS mutant colonies exhibited flattened or
presence of IPTG, further incubated in fresh LB without
depressed centers, giving the colonies a ringed ‘draughts-
inducer for 0–7 h, then the dilution of the preformed pool
man’ appearance. This phenotype was suppressed by
of fluorescent GFP protein in dividing cells was monitored
complementation (Supplementary Fig. 2). We thus next
by flow cytometry. Although fluorescence intensities of the
pks bacterial cell populations were gradually reduced as a
result of replication, the clbS mutant population remained
highly fluorescent with no dilution of GFP, indicating that it
was non-replicating (Fig. 4C). Thus, clbS mutant bacteria
producing colibactin remain viable but cease to replicate.
We next asked whether arrested clbS mutant cells could
be rescued by expression of the clbS gene cloned under
the control of an inducible promoter. The clbS mutant was
transformed with a gfp-clbS gene cloned on an IPTG
inducible expression vector. Following a culture without
IPTG up to the stationary phase to express the autotox-
icity phenotype, the bacteria were induced (or not) with
IPTG. Although the induced cells expressed GFP fluores-
cence (but not the control non-induced cells, not shown),
no significant increase of CFU levels was observed (Sup-
plementary Fig. 3). Thus, colibactin-induced stasis could
not be reversed following expression of ClbS.

The SOS-response is activated in a colibactin-producing


Fig. 3. Decreased numbers of CFU in overnight cultures of clbS
but not in clbS clbA and clbS clbP mutants. E. coli DH10B hosting clbS mutant
the BACpks (pks), the clbS mutant, the clbS clbA and clbS clbP
double mutants, and the complemented mutants, were pre-grown in Induction of a senescent-like state in which viable cells
LB to reach exponential phase, 106 bacteria ml−1 were inoculated in are not able to form colonies has been described in E. coli
LB and grown overnight before plating on LB to determine treated with a DNA-damaging antibiotic that activates the
colony-forming unit (CFU) numbers. The median and individual
results of independent experiments are shown. One-way ANOVA co-ordinated cellular SOS response (Pennington and
and Bonferroni post-test, ***P < 0.001. Rosenberg, 2007; Dörr et al., 2009). We examined the
C 2015 John ©
V 2015
Wiley & John
Sons Wiley & Sons Ltd,
Ltd, Molecular Molecular 99,
Microbiology, Microbiology
897–908
ClbSClbS protects
protects fromfrom colibactin9015
colibactin

Fig. 4. The clbS mutant exhibits a viable but


non-replicating state. Panel A: DH10B hosting
the BAC vector, the BACpks (pks) and the
clbS mutant were grown overnight in LB, then
propidium iodide (PI) was added and the dye
uptake by membrane-compromised bacteria
was quantified by flow cytometry. Heat-killed
(95°C) bacteria were used as a positive
control. Percentages of various strains in the
positive gate (bar) are shown in
Supplementary Table S1. Panel B: The
bacteria were grown overnight in LB, then
bacterial ATP levels were determined with a
luminescence assay. Relative light units
(RLUs) per optical density units (OD) at
600 nm were recorded with a microplate
reader. The median and individual results of
three independent experiments are shown.
Panel C: DH10B pBACpks (pks) and the clbS
mutant hosting the plasmid pDIGi coding for
an IPTG-inducible GFP were grown overnight
in LB + IPTG, then without inducer for the
indicated times. The dilution of GFP
fluorescence (caused by bacterial replication;
Helaine et al., 2010) was analyzed by flow
cytometry.

induction of the SOS response in the clbS mutant, using clbA mutant cells did not show activation of the recA
the plasmid pRecA-GFP coding for the recA promoter promoter. Interestingly, the clbS mutant exhibited a sig-
controlling the gfp reporter gene. The wild-type pks, clbS nificant increase of fluorescence (Fig. 5). No increase of
and clbS clbA mutants and vector control strains hosting fluorescence was observed in the clbS clbA double
pRecA-GFP were grown overnight in LB and GFP fluo- mutant, showing that the SOS-response activation in the
rescence was measured (Fig. 5). In contrast to positive clbS mutant was associated to colibactin production
control bacteria treated with ciprofloxacin, wild-type pks or (Fig. 5). Similar results were obtained with bacterial colo-
nies grown overnight on LB plates (Supplementary
Fig. 4). These results suggest that the clbS mutant pro-
ducing colibactin activates the SOS response following
self-inflicted DNA damage.

Mutating the nucleotide excision repair system strongly


enhances the autotoxicity of colibactin in a clbS
mutant background

The SOS induction can be driven by primary damage to


DNA such as covalent binding of adducts or cross-links,
and colibactin was recently proposed to cross-link DNA
(Vizcaino and Crawford, 2015). As these damages are
repaired mainly by the nucleotide excision repair uvrABC
Fig. 5. Induction of the SOS response in the clbS mutant repair system (Truglio et al., 2006), we reasoned that an
producing colibactin. DH10B pBACpks (pks), the clbS, clbA and
clbS clbA mutants, hosting the plasmid pRecA-GFP (coding for gfp uvrB mutant would be highly sensitive to colibactin in a
reporter gene under control of the DNA damage-inducible recA clbS mutant background. We thus constructed an uvrB
promoter) were grown 17 h in LB broth and the GFP fluorescence single mutant and a clbS uvrB double mutant, and as a
per OD600 units was measured with a fluorescence plate reader
and normalized to that in the DH10B pBAC vector control strain. control an uvrB mutant in the E. coli strain hosting the
Individual results of four independent experiments and the median BAC vector. Whereas the single uvrB mutants did not
are shown. As a DNA damage positive control, ciprofloxacin (cipro, exhibit an altered growth, the clbS uvrB double mutant
1 μg ml−1) was added to a DH10B pBAC vector overnight culture
and incubated 1 h at 37°C. One-way ANOVA with Bonferonni shown a marked reduction of CFU numbers compared
multiple comparison test, *P < 0.05, **P < 0.01, ***P < 0.001. with that of the clbS single mutant (Fig. 6). Complemen-

© 2015 John
C 2015
V John Wiley
Wiley && Sons
Sons Ltd,
Ltd, Molecular
Molecular Microbiology,
Microbiology 99, 897–908
902N. Bossuet-Greif
6 . 䊏
et alet. al■
N. Bossuet-Greif

only for GFP as a control. Microscopic examination of the


transfected HeLa cells showed that both the GFP control
and GFP-ClbS proteins exhibited similar diffuse localiza-
tion in the host cell cytoplasm and nucleus and that
expression of GFP-ClbS did not induce signs of toxicity
(Fig. 7A). Following transfection with GFP or GFP-ClbS
expressing vectors, the HeLa cells were infected 4 h with
the wild-type pks and DNA DSBs were demonstrated by
staining pH2AX 4 h later (Fig. 7A). In HeLa cells that were
transfected with the GFP control plasmid, the cells
expressing or not GFP exhibited similar nuclear pH2AX
foci. In contrast, HeLa cells that were transfected with
GFP-ClbS showed a non-homogeneous pH2AX staining
pattern: GFP-ClbS positive cells exhibited less pH2AX
staining compared with that in nearby cells not expressing
GFP-ClbS (Fig. 7A, arrows). To quantify this effect, the
pH2AX signal in relation with GFP fluorescence was
recorded by flow cytometry (Fig. 7B). The GFP/pH2AX
scatter graphs showed that pH2AX levels did not change
Fig. 6. Enhanced growth defect in a clbS uvrB double mutant. E. with the GFP signal in pks-infected and GFP-transfected
coli DH10B hosting the BAC vector, BACpks (pks), the clbS and
uvrB single mutants, the clbS uvrB double mutants, and the double cells. In GFP-ClbS transfected cells, however, pH2AX
mutant complemented with the empty vector plasmid (pASK) or the levels diminished in conjunction with increased GFP-ClbS
plasmid coding for clbS or uvrB (pClbS or pUvrB) were pre-grown expression (Fig. 7B). Following infection with wild-type
in LB to reach exponential phase, 106 bacteria ml−1 were inoculated
in LB and grown 17 h before plating on LB to determine pks at various moi the numbers of cells with a pH2AX
colony-forming unit (CFU) numbers. The median and individual signal above a defined threshold in GFP-ClbS expressing
results of independent cultures in five experiments are shown. cells were significantly reduced compared with that in
One-way ANOVA and Bonferroni post-test, ns: not significant,
**P < 0.01, ***P < 0.001. GFP control cells (Fig. 7C). In contrast, GFP-ClbS
expressing cells were not protected against the DNA-
damaging peptide-polyketide bleocin (Supplementary
tation of the clbS uvrB mutant with pClbS (but not with the Fig. 5). These results show that clbS gene expression in
empty pASK75 vector) restored CFU levels similar to that mammalian cells confers a specific protection against the
of the single uvrB mutant. Similarly, complementation with genotoxicity of colibactin, confirming that ClbS is a coli-
the uvrB gene restored a growth level similar to that of the bactin resistance protein.
single clbS mutant (Fig. 6).
Together these results support a model where ClbS is a
colibactin resistance protein that, coupled with the Discussion
prodrug assembly and efflux strategies, prevents self- Drug-producing bacteria protect themselves against the
inflicted DNA damage. toxicity of their biosynthesized products by using multiple
resistance mechanisms in combination, including drug
sequestering proteins, which are typically encoded on the
Ectopic expression of clbS protects HeLa cells
biosynthesis gene cluster (Cundliffe and Demain, 2010).
from colibactin
We thus tested whether the pks-encoded ClbS protein,
To provide additional support to the hypothesis that ClbS which binds reactive α-methylene-γ-butyrolactones
is a colibactin resistance protein, we tested whether its (Kunzmann and Sieber, 2012), could participate in the
expression within mammalian HeLa cells could protect self-protection of E. coli producing colibactin. A clbS
the host cell DNA against colibactin. To allow ectopic mutant displayed an autotoxicity phenotype characterized
expression and visualization of expressing eukaryotic by a marked growth defect with a decrease of CFU
cells, the clbS gene was cloned in a mammalian expres- numbers in overnight cultures. This was linked to produc-
sion vector as a translational fusion with the gfp gene. The tion of colibactin, as this phenotype was expressed at the
GFP-ClbS fusion protein was functional for complemen- transition between the exponential and stationary phases,
tation of the clbS mutant, as verified with a bacterial the time that supports maximal clb gene expression (and
expression vector (Supplementary Fig. 3). The mamma- supposedly colibactin production) (Homburg et al., 2007).
lian GFP-ClbS expression vector was then transfected in In addition, the growth defect was abrogated in a clbS
HeLa cells, in parallel with the expression vector coding clbA double mutant inactivated for the ClbA PPTase that
C 2015 John ©
V 2015& John
Wiley SonsWiley & Sons Ltd,
Ltd, Molecular Molecular Microbiology
Microbiology, 99, 897–908
ClbSClbS protects
protects fromfrom colibactin9037
colibactin

Fig. 7. clbS ectopic expression protects HeLa cells from colibactin. HeLa cells were transfected with plasmids coding for a GFP-ClbS fusion
protein or for GFP alone and 24 h later, infected 4 h with DH10B pBACpks (pks) or with a clbA mutant at given multiplicities of infection (moi,
number of bacteria per cell at the onset of infection). Host DNA double strand breaks were demonstrated 4 h later by staining phosphorylated
histone H2AX (pH2AX). Panel A: epifluorescence optical section (Apotome) images of pH2AX (red), GFP (green) and DNA (blue) in cells
expressing GFP or GFP-ClbS and infected with DH10B pBACpks (MOI = 50). Bar = 20 micrometer. The green arrow points to a GFP-ClbS
expressing cell with low pH2AX, and the red arrow a cell that does not express GFP-ClbS but shows high pH2AX. Panel B: HeLa cells were
transfected and infected as in panel A and analyzed by flow cytometry to generate density plots of GFP and pH2AX fluorescence. Panel C:
cell counts of H2AX positive and negative cells in the GFP and GFP-ClbS-expressing population, based on the quadrant gates shown in panel
B. The experiment was performed three times in duplicates, with similar results. Chi-square test, *P < 0.001.
This figure is available in colour online at wileyonlinelibrary.com.

primes the NRPS–PKS assembly line and therefore does mutants are unable to repair cross-links as well as
not produce (pre)colibactin. Complementation with a adducts (Truglio et al., 2006), our results also suggest that
plasmid-encoded clbA gene restored and even increased colibactin can produce such DNA damages in bacterio.
the growth defect, suggesting that overexpression of the This is in line with the recent finding that colibactin might
activating PPTase resulted in increased colibactin synthe- cross-link DNA, based on the in vitro reactivity of a can-
sis and enhanced autotoxicity. The growth defect was also didate purified precolibactin (Vizcaino and Crawford,
suppressed in a clbS clbP double mutant, inactivated for 2015). More work is required to establish the mode of
the periplasmic ClbP peptidase that cleaves precolibactin action of the mature colibactin, the binding to DNA, the
to generate mature active colibactin. nature of the DNA damage(s) inflicted and how it ulti-
The altered growth of the clbS mutant was associated mately results in DNA DSB and mutagenesis in the mam-
with the activation of the SOS response, the bacterial malian cell.
response to DNA damage. The severe growth defect The clbS mutant producing colibactin appeared to stop
observed in the clbS uvrB double mutant, in which the replication without showing significant cell death, as exam-
nucleotide excision repair UvrABC system is inactivated, ined by fluorescence dilution, membrane impermeant dyes
further indicates that colibactin inflicted DNA damage in uptake and levels of intracellular ATP. The growth arrest is
the absence of the self-defense ClbS protein. As uvrB reminiscent of persistence, a phenotype where a subpopu-

© 2015 John
C 2015
V John Wiley
Wiley && Sons
Sons Ltd,
Ltd, Molecular
Molecular Microbiology,
Microbiology 99, 897–908
8 N. N.
904 Bossuet-Greif et alet. al.■ 䊏
Bossuet-Greif

lation of non-growing bacteria escapes the lethal action of


antibiotics without acquiring heritable resistance (Balaban
et al., 2004). It was shown that persistence can be medi-
ated by the SOS response, providing a mechanism of
survival upon DNA-damage (Dörr et al., 2009). The SOS
response has also been shown to induce a prolonged
senescence-like state in E. coli cells sustaining spontane-
ous DNA damage (Pennington and Rosenberg, 2007).
Senescence is triggered in mammalian cells exposed to
colibactin (Secher et al., 2013; Cougnoux et al., 2014;
Dalmasso et al., 2014). Senescence is regarded in eukary-
otes as a programmed cellular response to DNA damage
that suppresses proliferation of potentially mutated cells. In
E. coli enduring colibactin-inflicted damage, induction of
SOS followed by replication arrest could represent a similar
mechanism to cease replication of genetically damaged
bacterial cells. The analogous responses of the clbS
mutant E. coli and the eukaryotic cells may also relate to
the type of DNA damage induced by colibactin. Indeed
DNA cross-links present a unique challenge for the bacte-
rial or eukaryotic cell to repair and a major senescence Fig. 8. Putative model of suicide avoidance in
inducer in mammalian cells (Grillari et al., 2007). colibactin-producing E. coli. The NRPS-PKS enzymes,
The drug-binding property of the ClbS protein post-translationally modified by the ClbA PPTase, synthesize
colibactin as a prodrug. The multidrug and toxic compound
(Kunzmann and Sieber, 2012), the SOS-response activa- extrusion (MATE family) efflux pump ClbM secretes the prodrug
tion, the autotoxicity phenotype and increased sensitivity into the bacterial periplasm, where the peptidase ClbP cleaves the
of a uvrB mutant in a clbS background, together with the prodrug to release the active colibactin ‘warhead’. ClbS could
sequester or modify colibactin and thereby prevent DNA-damage.
protection of HeLa cells expressing GFP-ClbS, support Ultimately, the SOS-response and DNA nucleotide excision repair
the hypothesis that ClbS is a colibactin resistance protein. system stands as the last-line of defense.
We propose that ClbS can sequester, or modify colibactin
and thereby prevent self-inflicted DNA-damage. Such colibactin ‘warhead’ in the periplasm, away from the
resistance proteins are common in antibiotic-producing genomic DNA. This prodrug assembly strategy with cleav-
microorganisms. For example, two resistance proteins age of an acylated D-asparagine to produce active drugs
have been characterized in Streptomyces sp. producing upon secretion is also found in other NRPS and hybrid
the peptide-polyketide antitumor antibiotic bleomycin, the PKS-NRPS antibiotic biosynthesis pathways, such as
BlmA sequestering protein and the N-acetyl-transferase xenocoumacin, amicoumacin and zwittermicin (Reimer
BlmB that acetylates and inactivates bleomycin (Galm et al., 2011). Bacterial resistance during the biosynthesis
et al., 2005). Kunzmann and Sieber (2012) could not of intracellular toxic compounds typically involves multiple
detect any esterase, peptidase or protease activity in the mechanisms in combination (Cundliffe and Demain,
ClbS protein, suggesting that ClbS would rather act as a 2010). Indeed the producing bacteria have also to cope
colibactin-sequestering protein. Our results also indicate with the previously exported drug that might reenter into
that ClbS protective function is specific to colibactin, as the producer, and with exogenous drug produced by
ectopic expression of GFP-ClbS did not protect HeLa sibling bacteria. ClbS could thus function as a comple-
cells against the DNA-cleaving activity of bleocin. In addi- mentary protection component, where the prodrug activa-
tion, the pClbS plasmid did not confer any resistance to a tion mechanism is ineffective against active genotoxin
variety of antibiotics, including the DNA-damaging drugs (re)entry. Microorganisms that synthesize DNA-damaging
bleocin, ciprofloxacin and mitomycin-C (data not shown). drugs develop such complementary self-protection strat-
We propose a putative model in which ClbS functions egies relying on drug sequestration or modification
as a backup for the prodrug assembly and efflux mecha- together with efflux systems. For example, the antitumor
nisms, to confer a robust protection system to the produc- antibiotic mitomycin-C produced by Streptomyces laven-
ing bacteria (Fig. 8). Following synthesis by the NRPS– dulae is synthesized as an inactive prodrug that is
PKS enzymes, the inactive precolibactin equipped with an sequestered by the small soluble resistance protein Mrd,
N-terminal acylated D-asparagine is secreted to the peri- delivered to the Mct efflux pump, and deactivated by the
plasm, possibly by the putative efflux pump ClbM. The McrA resistance protein (Sheldon et al., 1997; 1999). Fol-
peptidase ClbP cleaves the prodrug to release the active lowing efflux in the periplasm and cleavage by ClbP, the

C 2015 John ©
V 2015& John
Wiley Sons Wiley & Sons Ltd,
Ltd, Molecular Molecular 99,
Microbiology, Microbiology
897–908
ClbSClbS protects
protects fromfrom colibactin905
colibactin 9

colibactin warhead is further trafficked to the eukaryotic Carbenicillin (50 μg ml−1), kanamycin (50 μg ml−1), gen-
cell through a yet unexplored pathway. The protection of tamicin (15 μg ml−1) or chloramphenicol (25 μg ml−1) were
HeLa cells by ectopic expression of GFP-ClbS indicates added to the medium as required. For enumeration and
kinetic experiments, each strain was pre-grown in LB from
that the colibactin warhead likely enters the eukaryotic
a −80°C glycerol stock to reach exponential growth
cells to inflict DNA damage. (DO600 = 0.4) and then inoculated at 2 × 106 bacteria ml−1 in
In conclusion, we have gained a novel insight into how 5 ml LB and grown overnight (17 h) or for indicated times. For
toxigenic E. coli protects itself against colibactin. Ulti- CFU enumeration, the culture broth was serial-diluted in
mately, this knowledge could help to further decipher coli- PBS, plated on LB agar plates and incubated overnight at
bactin mode of action and to design new means to 37°C.
prevent colibactin-mediated carcinogenesis.
Eukaryotic cell culture, transfection and infection
Experimental procedures HeLa cells (ATCC) were grown in Dulbecco’s modified
Bacterial strains, mutants and plasmids Eagle’s medium (DMEM glutamax, Invitrogen) supplemented
with 10% (vol/vol) Foetal Calf Serum (FCS; Eurobio) and 1%
Strains used in this study were Escherichia coli DH10B (vol/vol) Non Essential Amino Acid (Invitrogen) at 37°C in a
hosting the BACpks (Nougayrède et al., 2006) and isogenic 5% CO2 atmosphere. Cells were confirmed mycoplasma-free
mutants and complemented mutants that are listed in Sup- by PCR and maintained by serial passage. HeLa cells were
plementary Table S2. Gene mutagenesis was performed by infected as described previously (Nougayrède et al., 2006).
using the lambda Red recombinase method (Datsenko and Briefly, the day before transfection or infection, HeLa cells
Wanner, 2000), with primers used listed in Supplementary were trypsinized and seeded in cell culture plates or slides.
Table S2. The double mutants were constructed sequentially. The following day, cells were washed three times with HBSS
The mutations and deletion of FRT cassettes were verified by (Invitrogen), inoculated with a defined multiplicity of infection
PCR using primers upstream and downstream of the target (moi; number of bacteria per cell at the onset of infection) in
genes. For clbA and clbP complementation, the mutants infection medium (DMEM with 25 mM Hepes, Invitrogen) and
were transformed with the previously described plasmids incubated with the bacteria for 4 h at 37°C in a 5% CO2
pClbA (Cuevas-Ramos et al., 2010) and pClbP (Dubois et al., atmosphere. Cells were then washed three times with HBSS
2011). To construct plasmid pClbS, the clbS gene (with and incubated 4 h in cell culture medium supplemented with
207 bp upstream sequence) was PCR-amplified using 200 microgram ml−1 gentamicin, and then processed for
primers CPTPNG9/10, digested by XhoI and XmaI, and analysis. For ectopic expression experiments, HeLa cells
ligated into pASK75 (Skerra, 1994). The resulting plasmid were transfected with plasmids pEGFP-C2 or pEGFP-ClbS
pClbS was verified by sequencing. The clbS chromosomal (EndoFree plasmid preparation, Qiagen) using TurboFect
complementation was constructed by mini-Tn7 insertion of (ThermoScientific), following the manufacturer protocol.
the clbS gene at the attTn7 site (Choi and Schweizer, 2006); Twenty-four hours later, cells were infected as above and
the clbS gene was subcloned (XmaI/XhoI) from pClbS into processed for fluorescence microscopy or flow cytometry.
pUC18R6K-mini-Tn7T-Gm and the resulting plasmid was
electroporated in the clbS mutant together with the pTNS2
helper plasmid. The insertion of the clbS gene at the attTn7 Phospho-H2AX staining and quantification by
site was selected on LB medium supplemented with microscopy, flow cytometry and in-cell western
15 μg ml−1 gentamicin and verified by PCR using the
PTn7R and PglmS-down-Ecoli primers. To construct plasmid For the imaging and quantification of pH2AX and GFP, HeLa
pEGFP-ClbS for the eukaryotic cell transfection experiments, cells were grown in 8-chamber slides or 6-wells plates, trans-
the clbS gene was PCR-amplified using primers fected and infected as described above. As a control, cells
CPTPNG1/2; the PCR product was digested with XhoI/XmaI were treated 4 h with 120 microgram ml−1 bleocin (Calbio-
and ligated in the pEGFP-C2 vector (Clontech). The con- chem). The cells were fixed, and immunostained as
struction was verified by sequencing. Plasmid pGFP-ClbS described previously (Nougayrède et al., 2006). The antibod-
was constructed by PCR amplifying the GFP-ClbS gene ies used were rabbit monoclonal anti pH2AX (20E3, Cell
fusion from pEGFP-ClbS with primers CPTPNG2/11 and Signaling Technology) and TexasRed or AlloPhycoCyanin
ligating the PCR product into pSC-B-amp/kan vector under (APC) conjugated secondary antibodies (Jackson). DNA was
control of the IPTG-inducible promoter (Stratagene). Plasmid labeled with DAPI. Optical section images were acquired
pUvrB was constructed by PCR amplifying uvrB from DH10B using an ApoTome microscope (Zeiss). The GFP and APC
with primers CPTPJPN34/35; the PCR product was digested fluorescence of individual cells was quantified with a FACS-
with XmaI/BamHI and ligated into pASK75. Calibur flow cytometer (Becton). FL1 and FL4 data of at least
50 000 events (gated on SSC/FSC) and 10 000 GFP-positive
cells were recorded and analyzed with FlowJo 7.6 software.
Bacterial growth, enumeration and kinetics The in-cell western (ICW) procedure was performed as
described previously (Audebert et al., 2010; Martin et al.,
Escherichia coli strains were routinely grown at 37°C with 240 2013). Briefly, the cells were grown and infected in 96-wells
r.p.m. shaking in 5 ml of Lennox L broth (LB, Invitrogen) in cell culture plate. Following fixation with formaldehyde, per-
closed 16 × 100 mm tubes (Kima), or on LB agar plates. meabilization and blocking, the cells were incubated over-

© Sons Ltd,
C 2015 John Wiley & Sons
V Ltd, Molecular
Molecular Microbiology,
Microbiology 99, 897–908
Bossuet-Greifetetalal. . ■䊏
10 N.N.Bossuet-Greif
906

night at 4°C with anti pH2AX antibodies (20E3, Cell Signaling ANOVA followed by Bonferroni post-tests. CFU/ml and
Technology; 1:200). An infrared fluorescent (800 nm) second- RLU/OD600 were log-transformed for the analyses. P
ary antibody (Rockland; 1:500) was used to detect pH2AX. values < 0.05 were considered significant and are denoted by
DNA was counterstained with RedDot2 (Biotium; 1:500). *, P < 0.01 is denoted by ** and P < 0.001 by ***. For cell
DNA and pH2AX were visualized simultaneously using an counts by flow cytometry, a Chi-square test was used with the
Odyssey Infrared Imaging Scanner (LI-COR Biosciences) at number of events per region, a conservative value of
680 nm (pseudocolored red) and 800 nm (pseudocolored P < 0.001 was considered significant.
green). Relative fluorescent units for pH2AX per cell (as
determined by the 800 nm signal divided by the 700 nm
signal) were divided by untreated controls to determine the Acknowledgements
genotoxic index.
The authors have no conflict of interest to declare. We thank
Michèle Boury for the initial cloning of clbS, Sophie Allart,
Bacterial viability, ATP content and fluorescence Astrid Canivet and Fatima-Ezzahra L’Faqihi-Olive for techni-
dilution assay cal assistance at the cellular imaging and flow cytometry
facilities of INSERM UMR1043, Toulouse. We thank David
To examine the bacterial membrane integrity, 25 μg ml−1 of Holden for the gift of pDIGi and Chris Michiels for the pRecA-
propidium iodide (Sigma) or 30 μg ml−1 ethidium bromide GFP plasmid. This work was supported by grants ANR-13-
(Sigma) was added to 8 μl of 17 h LB culture diluted in 1 ml of BSV1-0028, ANR-13-BSV3-0015 and INCA-PLBIO13-123.
PBS and incubated in the dark during 20–30 min at room
temperature. Dead control bacteria were prepared by heating
at 95°C for 10 min. To monitor bacterial division by fluores- References
cence dilution, bacteria hosting pDIGi were cultured 17 h in
LB 0.5 mM IPTG to induce production of GFP (Helaine et al., Aertsen, A., Van Houdt, R., Vanoirbeek, K., and Michiels,
2010), centrifuged, resuspended in fresh LB medium without C.W. (2004) An SOS response induced by high pressure in
the inducer and further incubated during 0 to 7 h. GFP fluo- Escherichia coli. J Bacteriol 186: 6133–6141.
rescence in bacteria was measured by flow cytometry, with a Arthur, J.C., Perez-Chanona, E., Mühlbauer, M., Tomkovich,
FACSCalibur (Becton). PI uptake and GFP fluorescence was S., Uronis, J.M., Fan, T.-J., et al. (2012) Intestinal inflam-
recorded in the FL2 and FL1 channels respectively, in at least mation targets cancer-inducing activity of the microbiota.
50 000 bacteria (gated on SSC/FSC) and analyzed with Science 338: 120–123.
FlowJo 7.6 software. Bacterial cell metabolic activity (ATP Audebert, M., Riu, A., Jacques, C., Hillenweck, A., Jamin,
content) following 17 h growth in LB was measured using the E.L., Zalko, D., and Cravedi, J.-P. (2010) Use of the γH2AX
BacTiter-Glo assay (Promega) following the manufacturer assay for assessing the genotoxicity of polycyclic aromatic
recommendations. Luminescence and optical density at hydrocarbons in human cell lines. Toxicol Lett 199: 182–
600 nm (OD600) was recorded with a Tecan Infinite Pro 192.
microplate reader. Balaban, N.Q., Merrin, J., Chait, R., Kowalik, L., and Leibler,
S. (2004) Bacterial persistence as a phenotypic switch.
Science 305: 1622–1625.
SOS response monitoring Beld, J., Sonnenschein, E.C., Vickery, C.R., Noel, J.P., and
Burkart, M.D. (2014) The phosphopantetheinyl trans-
The activation of the SOS response pathway was monitored
ferases: catalysis of a post-translational modification
with a GFP reporter under control of the recA promoter
crucial for life. Nat Prod Rep 31: 61–108.
(Aertsen et al., 2004). Bacteria hosting the pRecA-GFP
Bian, X., Plaza, A., Zhang, Y., and Muller, R. (2015) Two
plasmid were grown 17 h in LB, then 100 μl samples were
more pieces of the colibactin genotoxin puzzle from
transferred to a microplate and GFP fluorescence was meas-
Escherichia coli show incorporation of an unusual 1-
ured with a Tecan Infinite M200 microplate reader. As a posi-
aminocyclopropanecarboxylic acid moiety. Chem Sci 6:
tive control, ciprofloxacin (1 μg ml−1) was added to a 16 h
3154–3160.
culture of DH10B hosting the BAC vector and incubated for
Bondarev, V., Richter, M., Romano, S., Piel, J., Schwedt, A.,
1 h at 37°C. The OD600 was measured, and fluorescence at
and Schulz-Vogt, H.N. (2013) The genus Pseudovibrio
505 nm (bandwidth 20 nm) was measured using an excita-
contains metabolically versatile bacteria adapted for sym-
tion wavelength of 475 nm (bandwidth 9 nm). Fluorescence
biosis. Environ Microbiol 15: 2095–2113.
was expressed per unit of OD600 (Aertsen et al., 2004).
Brotherton, C.A., and Balskus, E.P. (2013) A prodrug resist-
Alternatively, the strains were grown overnight on LB plates,
ance mechanism is involved in colibactin biosynthesis and
isolated bacterial colonies were resuspended in 1 ml PBS
cytotoxicity. J Am Chem Soc 135: 3359–3362.
and GFP fluorescence was analyzed with a FACSCalibur
Brotherton, C.A., Wilson, M., Byrd, G., and Balskus, E.P.
(Becton) flow cytometer, in at least 50 000 bacteria (gated on
(2015) Isolation of a metabolite from the pks island pro-
SSC/FSC) and analyzed with FlowJo 7.6 software.
vides insights into colibactin biosynthesis and activity. Org
Lett 17: 1545–1548.
Statistical analyses Buc, E., Dubois, D., Sauvanet, P., Raisch, J., Delmas, J.,
Darfeuille-Michaud, A., et al. (2013) High prevalence of
Statistical analyses were carried out using GraphPad Prism mucosa-associated E. coli producing cyclomodulin and
5.0b. P values were calculated using one-way or two-way genotoxin in colon cancer. PLoS ONE 8: e56964.

C 2015 John ©
V 2015
Wiley & John
Sons Wiley & Sons Ltd,
Ltd, Molecular Molecular 99,
Microbiology, Microbiology
897–908
ClbSprotects
ClbS protectsfrom
fromcolibactin
colibactin 907
11

Choi, K.H., and Schweizer, H.P. (2006) mini-Tn7 insertion in Kunzmann, M.H., and Sieber, S.A. (2012) Target analysis of
bacteria with single attTn7 sites: example Pseudomonas α-alkylidene-γ-butyrolactones in uropathogenic E. coli. Mol
aeruginosa. Nat Protoc 1: 153–161. Biosyst 8: 3061–3067.
Cougnoux, A., Gibold, L., Robin, F., Dubois, D., Pradel, N., Lai, Y.-C., Lin, A.-C., Chiang, M.-K., Dai, Y.-H., Hsu, C.-C., Lu,
Darfeuille-Michaud, A., et al. (2012) Analysis of structure- M.-C., et al. (2014) Genotoxic Klebsiella pneumoniae in
function relationships in the colibactin-maturating enzyme Taiwan. PLoS ONE 9: e96292.
ClbP. J Mol Biol 424: 203–214. McCarthy, A.J., Martin, P., Cloup, E., Stabler, R.A., Oswald,
Cougnoux, A., Dalmasso, G., Martinez, R., Buc, E., E., and Taylor, P.W. (2015) The genotoxin colibactin is a
Delmas, J., Gibold, L., et al. (2014) Bacterial genotoxin determinant of virulence in Escherichia coli K1 experimen-
colibactin promotes colon tumour growth by inducing a tal neonatal systemic infection. Infect Immun 83: 3704–
senescence-associated secretory phenotype. Gut 63: 3711.
1932–1942. Marcq, I., Martin, P., Payros, D., Cuevas-Ramos, G., Boury,
Cuevas-Ramos, G., Petit, C.R., Marcq, I., Boury, M., Oswald, M., Watrin, C., et al. (2014) The genotoxin colibactin exac-
E., and Nougayrède, J.-P. (2010) Escherichia coli induces erbates lymphopenia and decreases survival rate in mice
DNA damage in vivo and triggers genomic instability in infected with septicemic Escherichia coli. J Infect Dis 210:
mammalian cells. Proc Natl Acad Sci USA 107: 11537– 285–294.
11542. Martin, P., Marcq, I., Magistro, G., Penary, M., Garcie, C.,
Cundliffe, E., and Demain, A.L. (2010) Avoidance of suicide Payros, D., et al. (2013) Interplay between siderophores
in antibiotic-producing microbes. J Ind Microbiol Biotechnol and colibactin genotoxin biosynthetic pathways in Escheri-
37: 643–672. chia coli. PLoS Pathog 9: e1003437.
Dalmasso, G., Cougnoux, A., Delmas, J., Darfeuille-Michaud, Nougayrède, J.-P., Homburg, S., Taieb, F., Boury, M.,
A., and Bonnet, R. (2014) The bacterial genotoxin colibac- Brzuszkiewicz, E., Gottschalk, G., et al. (2006) Escherichia
tin promotes colon tumor growth by modifying the tumor coli induces DNA double-strand breaks in eukaryotic cells.
microenvironment. Gut Microbes 5: 675–680. Science 313: 848–851.
Datsenko, K.A., and Wanner, B.L. (2000) One-step inactiva- Nowrouzian, F.L., and Oswald, E. (2012) Escherichia coli
tion of chromosomal genes in Escherichia coli K-12 using strains with the capacity for long-term persistence in the
PCR products. Proc Natl Acad Sci USA 97: 6640–6645. bowel microbiota carry the potentially genotoxic pks island.
Dörr, T., Lewis, K., and Vulić, M. (2009) SOS response Microb Pathog 53: 180–182.
induces persistence to fluoroquinolones in Escherichia coli. Olier, M., Marcq, I., Salvador-Cartier, C., Secher, T.,
PLoS Genet 5: e1000760. Dobrindt, U., Boury, M., et al. (2012) Genotoxicity of
Dubois, D., Baron, O., Cougnoux, A., Delmas, J., Pradel, N., Escherichia coli Nissle 1917 strain cannot be dissociated
Boury, M., et al. (2011) ClbP is a prototype of a peptidase from its probiotic activity. Gut Microbes 3: 501–509.
subgroup involved in biosynthesis of nonribosomal pep- Payros, D., Secher, T., Boury, M., Brehin, C., Ménard, S.,
tides. J Biol Chem 286: 35562–35570. Salvador-Cartier, C., et al. (2014) Maternally acquired
Engel, P., Vizcaino, M.I., and Crawford, J.M. (2014) Gut sym- genotoxic Escherichia coli alters offspring’s intestinal
bionts from distinct hosts exhibit genotoxic activity via homeostasis. Gut Microbes 5: 313–325.
divergent colibactin biosynthetic pathways. Appl Environ Pennington, J.M., and Rosenberg, S.M. (2007) Spontaneous
Microbiol 81: 1502–1512. DNA breakage in single living Escherichia coli cells. Nat
Fischbach, M.A., and Walsh, C.T. (2006) Assembly-line enzy- Genet 39: 797–802.
mology for polyketide and nonribosomal peptide antibiot- Putze, J., Hennequin, C., Nougayrède, J.-P., Zhang, W.,
ics: logic, machinery, and mechanisms. Chem Rev 106: Homburg, S., Karch, H., et al. (2009) Genetic structure and
3468–3496. distribution of the colibactin genomic island among
Galm, U., Hager, M.H., Van Lanen, S.G., Ju, J., Thorson, members of the family Enterobacteriaceae. Infect Immun
J.S., and Shen, B. (2005) Antitumor antibiotics: bleomy- 77: 4696–4703.
cin, enediynes, and mitomycin. Chem Rev 105: 739– Reimer, D., Pos, K.M., Thines, M., Grün, P., and Bode, H.B.
758. (2011) A natural prodrug activation mechanism in nonribo-
Grillari, J., Katinger, H., and Voglauer, R. (2007) Contribu- somal peptide synthesis. Nat Chem Biol 7: 888–890.
tions of DNA interstrand cross-links to aging of cells and Rogakou, E.P., Pilch, D.R., Orr, A.H., Ivanova, V.S., and
organisms. Nucleic Acids Res 35: 7566–7576. Bonner, W.M. (1998) DNA double-stranded breaks induce
Helaine, S., Thompson, J.A., Watson, K.G., Liu, M., Boyle, histone H2AX phosphorylation on serine 139. J Biol Chem
C., and Holden, D.W. (2010) Dynamics of intracellular bac- 273: 5858–5868.
terial replication at the single cell level. Proc Natl Acad Sci Secher, T., Samba-Louaka, A., Oswald, E., and Nougayrède,
USA 107: 3746–3751. J.-P. (2013) Escherichia coli producing colibactin triggers
Homburg, S., Oswald, E., Hacker, J., and Dobrindt, U. (2007) premature and transmissible senescence in mammalian
Expression analysis of the colibactin gene cluster coding cells. PLoS ONE 8: e77157.
for a novel polyketide in Escherichia coli. FEMS Microbiol Secher, T., Payros, D., Brehin, C., Boury, M., Watrin, C.,
Lett 275: 255–262. Gillet, M., et al. (2015) Oral tolerance failure upon neonatal
Johnson, J.R., Johnston, B., Kuskowski, M.A., Nougayrede, gut colonization with Escherichia coli producing the geno-
J.-P., and Oswald, E. (2008) Molecular epidemiology and toxin colibactin. Infect Immun 83: 2420–2429.
phylogenetic distribution of the Escherichia coli pks Sheldon, P.J., Johnson, D.A., August, P.R., Liu, H.W., and
genomic island. J Clin Microbiol 46: 3906–3911. Sherman, D.H. (1997) Characterization of a mitomycin-

© Sons Ltd,
C 2015 John Wiley & Sons
V Ltd, Molecular
Molecular Microbiology,
Microbiology 99, 897–908
Bossuet-Greifetetalal. . ■䊏
12 N.N.Bossuet-Greif
908

binding drug resistance mechanism from the producing Truglio, J.J., Croteau, D.L., Van Houten, B., and Kisker, C.
organism, Streptomyces lavendulae. J Bacteriol 179: (2006) Procaryotic nucleotide excision repair: the UvrABC
1796–1804. system. Chem Rev 106: 233–252.
Sheldon, P.J., Mao, Y., He, M., and Sherman, D.H. (1999) Vizcaino, M.I., and Crawford, J.M. (2015) The colibactin
Mitomycin resistance in Streptomyces lavendulae includes warhead crosslinks DNA. Nat Chem 7: 411–417.
a novel drug-binding-protein-dependent export system. J
Bacteriol 181: 2507–2512. Supporting information
Skerra, A. (1994) Use of the tetracycline promoter for the
tightly regulated production of a murine antibody fragment Additional supporting information may be found in the online
in Escherichia coli. Gene 151: 131–135. version of this article at the publisher’s web-site.

C 2015 John ©
V 2015
Wiley & John
Sons Wiley & Sons Ltd,
Ltd, Molecular Molecular 99,
Microbiology, Microbiology
897–908

S-ar putea să vă placă și