REVIEW ARTICLE Do bacterial genotoxins contribute to chronic inflammation, genomic instability and tumor progression? Lina Guerra, Riccardo Guidi and Teresa Frisan Department of Cell and Molecular Biology, Karolinska Institutet, Stockholm, Sweden Introduction Epidemiological evidence has linked chronic bacterial infections with an increased risk of tumor development. Heleicobacter pylori is associated with gastric cancers and has been classified as a type I carcinogen by the World Health Organization [1]. Other known associa- tions between bacterial infection and human cancers are Salmonella enterica serovar Typhi and carcinoma of the gallbladder in chronic carriers; Streptococcus bovis and colon cancer; persistent Chlamydia pneumoniae and lung cancer; as well as Bartonella species and vascular tumor formation [2]. The acquisition of genomic instability is a crucial feature in tumor initiation and progression. Because the baseline mutation rate is insufficient to account for the multiple genetic changes required for cancer pro- gression, tumor cells must acquire a ‘mutator pheno- type’ that enhances the mutation frequency, and allows the evolution from a pre-malignant to an inva- sive cancer cell [3]. This phenotype can be caused by a failure to repair damaged DNA and ⁄ or altered activa- tion of the DNA damage-induced checkpoint responses that selectively eliminate irreversibly dam- aged cells. In the case of bacterial infection, several events (e.g. the establishment of chronic inflammation, as well as the production of genotoxins or bacterial products that interfere with regulation of cell cycle Keywords bacterial genotoxin; chronic inflammation; colibactin, cytolethal distending toxin; DNA damage; DNA damage response; genomic instability; tumor induction ⁄ progression Correspondence T. Frisan, Department of Cell and Molecular Biology, Karolinska Institutet, Box 285, S-171 77, Stockholm, Sweden Fax: +46 8 337412 Tel: +46 8 52486385 E-mail: teresa.frisan@ki.se (Received 1 March 2011, revised 4 April 2011, accepted 13 April 2011) doi:10.1111/j.1742-4658.2011.08125.x Cytolethal distending toxin, produced by several Gram-negative bacteria, and colibactin, secreted by several commensal and extraintestinal patho- genic Escherichia coli strains, are the first bacterial genotoxins to be described to date. Exposure to cytolethal distending toxin and colibactin induces DNA damage, and consequently activates the DNA damage response, resulting in cell cycle arrest of the intoxicated cells and DNA repair. Irreversible DNA damage will lead to cell death by apoptosis or to senescence. It is well established that chronic exposure to DNA damaging agents, either endogenous (reactive oxygen species) or exogenous (ionizing radiation), may cause genomic instability as a result of the alteration of genes coordinating the DNA damage response, thus favoring tumor initia- tion and progression. In this review, we summarize the state of the art of the biology of cytolethal distending toxin and colibactin, focusing on the activation of the DNA damage response and repair pathways, and discuss the cellular responses induced in intoxicated cells, as well as how prolonged intoxication may lead to chronic inflammation, the accumulation of geno- mic instability, and tumor progression in both in vitro and in vivo models. Abbreviations AaCDT, Aggregatobacter actinomycetemcomitans CDT; CDT, cytolethal distending toxin; DDR, DNA damage response; DSB, double strand break; EcCDT, Escherichia coli CDT; ER, endoplasmic reticulum; H2AX, histone 2AX; HdCDT, Haemophilus ducreyi CDT; HR, homologous recombination; MRN, Mre11-Rad50-Nbs1; NF, nuclear factor; NRPS, nonribosomal peptide megasynthetase; PARP, poly(ADP-ribose) polymerase; PKS, polyketide megasynthetase; ROS, reactive oxygen species; StCDT, serovar Typhi CDT; Th, T helper. FEBS Journal 278 (2011) 4577–4588 ª 2011 The Authors Journal compilation ª 2011 FEBS 4577 progression and apoptosis), in association with host genetic factors, may contribute to the acquisition of the mutator phenotype. In this review, we focus on the two bacterial prod- ucts that act as genotoxins and can directly damage the host DNA: cytolethal distending toxin (CDT) and colibactin. We discuss the biology of these two geno- toxins, as well as their contribution to induction of chronic infection ⁄ inflammation, genomic instability and tumor progression. CDT CDT is the first bacterial genotoxin to be described and is produced by a variety of Gram-negative bacteria, such as Escherichia coli, Aggregatobacter actinomycetemcom- itans, Haemophilus ducreyi, Shigella dysenteriae, Cam- pylobacter sp. and Helicobacter sp, and S. enterica [4]. CDT is generally a exotoxin, and the active holotox- in is a tripartite complex encoded by a single operon [5,6], formed by the CdtA, CdtB and CdtC subunits (Fig. 1A). Nesic et al. [7] solved the crystal structure of the H. ducreyi CDT (HdCDT) and demonstrated that the CdtA and CdtC subunits are lectin-like molecules, sharing structural homology with the B-chain repeats of the plant toxin ricin [7] (Fig. 1B). CDT is an exotoxin secreted by the pathogen at the infection site. Functional studies have identified CdtA and CdtC as comprising essential proteins for mediating toxin bind- ing to the membrane and internalization into target cells [8–10]. The CdtB subunit adopts the canonical four-layered fold of the DNase I family: a central 12-stranded b- sandwich packed between outer a-helices and loops on each side of the sandwich [7]. The crystal structure confirms previous data demonstrating that the CdtB subunit is functionally homologous to the mammalian DNase I, and also possesses DNase capacity both in vitro and when ectopically expressed or microinjected in eukaryotic cells. Mutation in any conserved residue important for the catalytic activity or the Mg 2+ bind- ing abolishes the ability of CdtB to cleave DNA in vitro and to induce DNA damage responses (DDRs) in model cell lines [11–14]. CDT is therefore defined as an A-B 2 toxin, where CdtA and CdtC are required for binding the holotoxin to the plasma membrane of target cells, allowing entry of the active CdtB, which can translocate to the nucleus and induce DNA lesions. In addition to the well-characterized DNA damaging activity of the CdtB subunit, Shenker et al. [15] reported that the active subunit from A. actinomyce- Fig. 1. Structure of CDT holotoxin and the psk genomic island. (A) Schematic representation of the CDT genes from H. ducreyi and S. enter- ica, serovar Typhi. In all CDT-producing bacteria identified to date, excluding S. enterica, the three cdt genes are organized in an operon and are transcribed as monocistrons. Conversely, in S. enterica, the three genes required for an active holotoxin are present as two separate units: one unit containing the cdtB gene, encoding the active subunit, and the pltB ⁄ pltA unit, encoding proteins possibly required for the proper traffic of CdtB to the nucleus of target cells. No homologous genes for the cdtA and cdtC subunits have been identified in this bacte- rium. (B) Crystal structure of the HdCDT, adapted from Nesic et al. [7]. Protein data bank code: 1SR4. (C) Schematic representation of the pks genomic island that encodes the enzymes and accessory proteins required for synthesis of an active colibactin in the E. coli strain Nissle 1917 [30]. Bacterial genotoxins and genomic instability L. Guerra et al. 4578 FEBS Journal 278 (2011) 4577–4588 ª 2011 The Authors Journal compilation ª 2011 FEBS temcomitans exhibits PI-3,4,5-triphosphate phosphatase activity [15]. However, there is evidence suggesting that the DNA damaging activity alone is sufficient to con- fer CdtB toxicity. Indeed, the G2 arrest and cell death induced by conditional expression of CdtB in Saccha- romyces cerevisiae depend exclusively on its DNase- catalytic residue because yeast does not harbor the substrate for CdtB phosphatidylinositol-3,4,5-triphos- phate phosphatase activity [16] and specific CdtB mutations that inhibit the phosphatase activity (but retain DNase activity) are sufficient to induce the cell death in proliferating U937 monocytes [17]. The discrepancy between the requirements of the dif- ferent enzymatic activities of CDT may depend on the cell type used as model. It is conceivable that T lym- phocytes are more susceptible to the phosphatase activity of CDT compared to all of the other cell lines tested. The only exception to the general structure of the CDT family described so far is represented by the S. enterica, serovar Typhi CDT (StCDT). In this intra- cellular pathogen, the three genes required for an active cytotoxin are not part of a single operon because the gene for cdtB is not associated with genes encoding for the CdtA and CdtC subunits. No homo- logs for cdtA and cdtC genes have been found within the complete Salmonella typhi genome [18]. However, the toxicity of the StCDT on target cells requires the transcription of two other genes: pltB (pertussis-like toxin B) and pltA (pertussis-like toxin A) (Fig. 1A). In vitro reconstitution experiments have shown that the products of these three genes form an tripartite com- plex inducing DNA damage in intoxicated eukaryotic cells [19]. Interestingly, expression of the cdtB, pltB and pltA genes is switched on upon bacterial uptake by the host cells, and it is conceivable that the PltB and PltA sub- units are required to transport CdtB from its site of production within the cells to the extracellular med- ium, from where StCDT can also intoxicate cells that have not been infected, in a paracrine manner [19]. Several details of CDT binding to the plasma mem- brane of target cells and its intracellular trafficking to the nucleus have been elucidated (Fig. 2). Interaction of the A. actinomycetemcomitans CDT (AaCDT) occurs within GM1-enriched regions of the plasma membrane, which are characteristic of mem- brane rafts [14,20], and cholesterol depletion by methyl-b-cyclodextrin reduces the ability of both AaCDT and HdCDT to bind to Jurkat and HeLa cell lines, respectively, and prevents intoxication [14,20]. Furthermore, inactivation of mutations within the SGMS1 gene that reduce the levels of sphingomyelin (a key component of lipid rafts) confers resistance to the E. coli CDT (EcCDT) [21]. The identity of the toxin receptor is still unknown. Fucose may represent the binding determinant for the EcCDT-II [10], whereas another study indicated that the AaCDT holotoxin binds to surface glycosphingoli- pids and that inhibitors of glycosphingolipid synthesis can prevent intoxication of the human monocytic U937 cell line [22]. Site-directed mutagenesis of a human cell line haploid for all chromosomes except chromosome 8 identified mutants for the membrane- expressed protein TMEM181 that were resistant to EcCDT [21]. On the basis of such evidence, it is conceivable that each individual CDT exhibits different receptor speci- ficity. In line with this evidence, Eshraghi et al. [23] reported that CDTs from H. ducreyi, A. actinomyce- temcomitans, E. coli and Campylobacter jejuni differ in their abilities to intoxicate host cells. The binding of Aa, Hd and EcCDT-III, but not CjCDT, is dependent on the presence of cholesterol. Unexpectedly, mutant Chinese hamster ovary cells that lack N-linked com- plex and hybrid carbohydrates, as well as cells that lack glycosphingolipids or are deficient in fucose bio- synthesis, are as similarly sensitive as the wild-type to intoxication by all four CDTs tested, indicating that N- and O-glycan, or fucosylated structures are dispens- able for mediating toxin binding [23]. Upon binding to the plasma membrane, the HdCDT is internalized in HeLa cells by dynamin-dependent endocytosis, and it further transits to the endosomal compartment [24]. Biochemical and imaging experi- ments have demonstrated that the toxin is then retro- gradely transported via the Golgi complex to the endoplasmic reticulum (ER) [14]. Other bacterial and plant toxins have been described to traffic from the plasma membrane to the ER (Shiga, cholera and ricin), and they subsequently reach their targets in the cytosol by retrograde transport via the ER degradation pathway (Fig. 2). However, HdCDT could not be detected by biochemical assays in the cytosol of intoxi- cated cells [25]: this opens the possibility that the CDT active subunit may translocate directly from the ER to the nucleus. Two studies have identified specific nuclear localization signals (NLS) within AaCdtB and EcCdtB-II. In AaCdtB, a nonconventional NLS is localized at the N-terminus of the protein, and the deletion of 11 amino acids within the this sequence abolishes intoxication [26]. Conversely, two NLS sequences, designated as NLS1 and NLS2, have been identified in the carboxy-terminal region of the EcCdtB-II [8]. Interestingly, the deletion of each of these sequences produces a differential localization of L. Guerra et al. Bacterial genotoxins and genomic instability FEBS Journal 278 (2011) 4577–4588 ª 2011 The Authors Journal compilation ª 2011 FEBS 4579 the active toxin subunit, suggesting that they play dif- ferent roles in the intracellular trafficking of EcCDT- II. Cells intoxicated with a holotoxin containing the EcCdtB-II-DNLS1 display a perinuclear distribution, which is consistent with trapping of the active toxin component in the late endosome and ⁄ or ER com- partment, whereas a diffuse cytoplasmic staining is observed in cells exposed to the EcCdtB-II-DNLS2- containing toxin. It is conceivable that the NLS1 promotes the ER-to-nucleus translocation, whereas NLS2 may function as an ER compartimentalization signal preventing the escape of the CdtB molecules to the cytosol, allowing their transit to the nucleus. It is still not known why the presence of the putative NLS domains are so divergent in AaCDT and EcCDT-II because the structure of CdtB subunits presents a high degree of conservation among different bacteria [27]. On the basis of such evidence, it is clear that the nuclear translocation of the CdtB subunit into the nucleus remains an open issue. Another interesting question is whether the CdtA and CdtC subunits assist the active component in its trafficking within the host cells. Colibactin This toxin has been recently characterized and very little information is available regarding its biology. Colibactin is a putative hybrid peptide–polyketide genotoxin found in both commensal and pathogenic bacteria. Colibactin has been mostly characterized in extraintestinal pathogenic E. coli strains of the phylo- genetic group B2 [28]. The enzymes required for the synthesis of colibactin are present in a 54 kb genomic island, referred to as the pks island, located in the asnW tRNA locus. This region contains 23 putative ORFs, including three nonribosomal peptide megasyn- thetases (NRPS), three polyketide megasynthetases Fig. 2. Summary of the CDT internalization pathway and cellular responses induced by bacterial genotoxins. Binding of CDT is dependent on the presence of intact lipid rafts, and the toxin is internalized via dynamin-dependent endocytosis into early and late endosomes. At this stage, it is not known whether the CdtA (violet) and ⁄ or the CdtC (pink) subunits are internalized and follow the active CdtB subunit (green) into the nucleus. The CdtB subunit further transits to the Golgi complex, and is then retrogradely transported to the ER. The mechanisms of nuclear translocation have not yet been fully elucidated. Once in the nucleus, the CdtB subunit causes DNA damage and activates the ATM- dependent DNA damage response, characterized by recruitment of the MRN complex, and full activation of ATM at the site of the damage, which also requires a functional c-MYC. Activation of ATM promotes phosphorylation of histone H2AX and activation of the DNA damage checkpoint responses via activation of: (a) the tumor suppressor p53 and its downstream effector p21, which results in G1 arrest, and (b) the kinase CHK2 that blocks cell proliferation in the G2 phase of the cell cycle by inactivating the CDC25 phosphatase, resulting in hyper- phosphorylation and inactivation of the cyclin-dependent kinase CDK1 (pCDK1). CDT intoxication also activates RhoA-dependent survival sig- nals. This effect requires a functional ATM and is dependent on dephosphorylation of the guanine nucleotide exchange factor Net1. Activation of RhoA regulates two distinct pathways: (a) induction of actin stress fibers, which requires the RhoA kinases ROCKI and ROCKII, and (b) activation of p38 mitogen-activated protein kinase, associated with delayed cell death. Bacterial genotoxins and genomic instability L. Guerra et al. 4580 FEBS Journal 278 (2011) 4577–4588 ª 2011 The Authors Journal compilation ª 2011 FEBS (PKS), two hybrid NRPS ⁄ PKS megasynthetases, and ten accessory, tailoring and editing enzymes (Fig. 1C). Mutation analysis demonstrated that all the PKS and NRPS and eight of the accessory genes are required for production of an active genotoxin [28]. Screening studies have demonstrated that the pks island is also present in other members of the Entero- bacteriaceae family, such as Klebsiella pneumoniae, Enterobacter aerogenes and Citrobacter koseri isolates [29]. Gene expression of the ORFs required for the colib- actin synthesis has been studied, using the nonpatho- genic E. coli strain Nissle 1917 as a model, by Homburg et al. [30], who identified seven transcripts, four of which are polycistrons. The polycistronic tran- scripts comprise the genes: (a) clbC to clbG; (b) clbI to clbN; (c) clbO to clbP; and (d) clbR to clbA, whereas the other ORFs are transcribed as monocistrons [30]. Luciferase reporter assays performed on the clbA, clbB, clbQ and clbR genes demonstrated that their expression increased during late logarithmic and early stationary phase of the bacteria growth course. The levels and the duration of expression depend on the culture medium used, with DMEM being the best con- dition compared to LB or minimal medium supple- mented with 0.2% glucose [30]. Interestingly, the transcription of these ORFs was not induced by direct contact with the mammalian cell line HeLa [30]. It remains to be determined whether the expression of the other clb genes is dependent on interaction with eukaryotic cells. DDRs and genomic instability CDT possesses DNase activity in vitro, and both CDT and colibactin cause DNA damage in intoxicated cells [11–14,28]. Thus, before discussing the cellular responses to these genotoxins, we briefly review the state of the art of the DNA damage sensing and repair pathways in mammalian cells, as well as the conse- quences of an altered DDR in the promotion of geno- mic instability. On the basis of the type of DNA damage induced by CDT and colibactin, we focus mainly on the cellular responses to DNA double strand breaks (DSB). DDRs are essential for preserving the genetic infor- mation and maintain genomic integrity in cells exposed to the damaging activity of endogenous [e.g. reactive oxygen species (ROS) produced by the cellular metab- olism] and environmental agents (ionizing radiation and UV radiation). Sensing and activation of the DDR is coordinated by proteins of the phosphatidyl- inositol 3-kinase-like protein kinase family: ATM, ATR and DNA-PK [31]. The outcome is a block of cell cycle progression and activation of the DNA dam- age repair pathways. Successful repair will allow the cell to resume the normal cell cycle progression, whereas damage beyond the possibility of repair will promote either apoptosis or senescence, precluding the survival and replication of cells that can accumulate genomic instability [32]. A key effector protein that regulates activation of cell death or senescence in the presence of chronic and unrepaired DNA damage is the tumor suppressor gene p53, which acts as a barrier for cancer initiation ⁄ progression [33]. Many DNA repair pathways have been evolved to cope with all the possible insults to which the cellular DNA is exposed. A mismatch DNA base is replaced with the correct one by the mismatch repair [34], whereas small base alterations, such as alkylation, are repaired by the base excision repair, which removes the altered base [35]. More complex lesions, such the UV-induced pyrimidine dimers, require a longer exci- sion (approximately 30 bp) and are repaired by the nucleotide excision repair pathway [36]. DSB will be processed by nonhomologous end joining or homolo- gous recombination (HR) [37]. Nonhomologous end joining occurs throughout the cell cycle and is based on the identification of the break and subsequent rejoining of the two ends. Consequently, there is a loss or addition of nucleotides at the site of the lesion, and the repair mechanism per se can contribute to a certain degree of genomic alteration. Conversely, HR per- forms an error-free repair of the lesion because the sis- ter chromatid is used as template, restricting this type of repair mechanism to the late S and G2 phases of the cell cycle. Recognition of DSB is mediated by several sensor complexes: members of the poly(ADP-ribose) polymer- ase (PARP) family, specifically PARP1 and PARP2, the Mre11-Rad50-Nbs1 (MRN) complex and the Ku70 ⁄ Ku80 hetorodimer [31]. PARP1 and 2 are activated by DNA DSB and catalyze the addition of poly(ADP-ribose) chains on histones and nuclear proteins. This step is essential for the recruitment of the MRN complex, which initiates a resection of the DNA ends to produce a 3¢ tail, and promotes the accumulation and full activation of the ATM kinase at the site of the damage [38]. In turn, ATM coordinates the full DNA resection to promote HR, and activates the checkpoint responses to block cell cycle progression, allowing repair and preservation of the genomic integrity. This is achieved by: (a) recruitment and phosphorylation of c-histone 2AX (H2AX) important to sustain the DDR; (b) activation of effectors such as CtlP, BRCA1, ARTEMIS L. Guerra et al. Bacterial genotoxins and genomic instability FEBS Journal 278 (2011) 4577–4588 ª 2011 The Authors Journal compilation ª 2011 FEBS 4581 (DNA resection); and (c) phosphorylation ⁄ activation of CHK2 and p53 that lead to cell cycle arrest and, in ultimate instances, to apoptosis or senescence [31,32,39]. Upon extensive DNA resection, the RPA complex binds to the 3¢ ssDNA ends and modulates the activity of effectors, such as BRCA2 and Rad51, which execute the recombination process by performing strand inva- sion and Holliday junction formation and resolution to produce two completed undamaged sister chromat- ids that preserve the original genetic information [31]. Figure 3 summarizes the main events in the ATM- dependent activation of the DDR. In nonhomologous end joining, the end of the breaks are recognized and bound by the Ku79 ⁄ Ku80 heterodimer, which in turn recruits the DNA-PK and initiates the DNA resection process. After binding to the DNA, the DNA-PK is autophosphorylated and provides access to the site of the resecting protein ARTEMIS, and also the ligase complex XRCC4 ⁄ Lig4, which promotes re-ligation of the broken ends [40]. Genomic instability is a characteristic of almost all human tumors. The major genomic alterations described in cancer cells include chromosomal instabil- ity and microsatellite instability [41]. Chromosomal instability is characterized by losses of entire or large portions of chromosomes, resulting in aneuploidy, translocation and loss of heterozygosity, whereas microsatellite instability is defined as an expansion or contraction of the number of oligonucleotide repeats present in microsatellite sequences apart from the nucleus. The importance of the DDR in the maintenance of genomic integrity is highlighted by the demonstration that the acquisition of genomic instability is linked to mutations in genes controlling DNA repair and mitotic checkpoint pathways in hereditary cancers [42]. Conversely, high throughput analysis demonstrated that, in sporadic cancers, the most frequently altered genes are the TP53 tumor suppressor and genes that regulate cell growth either positively (e.g. the oncoge- nes EGFR and RAS) or negatively (e.g. PTEN and CDKN2A) [42]. This pattern is not unexpeceted because deregulation ⁄ overexpression of oncogenes will lead to DNA replication stress and stalled replication, resulting in the formation of DNA damage and activa- tion of DDR. One of the consequences of this will be the activation of p53 and the promotion of cell death or senescence, a response that will pose a barrier to tumor initiation ⁄ progression. Therefore, only cells in which the deregulation of oncogenes is accompanied by alteration of the p53 tumor suppression pathway will have the possibility of overcoming this barrier. Cellular responses to colibactin and CDT We now discuss the key cellular responses activated in cells exposed to CDT or colibactin-producing bacteria, focusing on the DDR and cell survival, which are both relevant in the context of maintaining genome integ- rity. For a more detailed analysis of the CDT-induced cellular responses, several comprehensive reviews on CDT biology are available [4,43,44]. Fig. 3. Summary of the ATM-dependent DNA damage responses to DNA DSBs. Induction of DNA DSBs activates PARP1, which mediates the initial recruitment of the MRN complex and promotes the full activa- tion of ATM. In turn, ATM coordinates the DNA damage repair resulting in: (1) sus- tained DDR by phosphorylating histone H2AX; (2) resection of the damaged DNA to allow activation of the HR process; and (3) activation of the checkpoint responses that block cell proliferation in distinct phases of the cell cycle (G1 or G2) to allow repair. If the damage is beyond repair, this response will result in elimination of the altered cell by apoptosis or the induction of cellular senescence, a p53-dependent process defined as the tumorigenesis barrier. Bacterial genotoxins and genomic instability L. Guerra et al. 4582 FEBS Journal 278 (2011) 4577–4588 ª 2011 The Authors Journal compilation ª 2011 FEBS It is now well established that CDT acts as a nucle- ase cleaving DNA substrates in vitro, inducing nuclear fragmentation and chromatin disruption when trans- fected in cultured mammalian cells or Saccharomy- ces cereviase, and promoting DNA fragmentation in cells exposed to purified soluble toxin [11–13,45]. Simi- larly exposure of mammalian cells to E. coli strains expressing the pks island promotes DNA fragmenta- tion, as detected by the comet assay [28]. As a result of the DNA damaging activity, intoxi- cated cells activate the classical DDR, which includes recruitment of the DNA damage sensor complex MRN and the ATM kinase, phosphorylation of his- tone H2AX, activation of p53 and its transcriptional target, the cyclin-dependent kinase inhibitor p21, and phosphorylation of checkpoint kinase CHK2. Tran- scriptional upregulation of p21 leads G1 arrest, whereas CHK2-dependent inactivation of the CDC25 phosphatase leads to an accumulation of the hyper- phosphorylated form of cyclin-dependent kinase 1 (CDK1, also known as CDC2), and consequent induc- tion of G2 arrest [9,46–51]. The prompt activation of the ATM-dependent response to CDT or ionizing radiation-induced DNA damage also requires a functional proto-oncogene MYC [52]. As a consequence of the activation of ATM-depen- dent checkpoint responses, cells exposed to CDT or colibactin stop proliferating [28,46,51]. Furthermore, CDT-intoxicated normal or tumor cells express the hallmarks of cellular senescence, such as persistently- activated DNA damage signaling (detected as 53BP1 ⁄ cH2AX-positive foci), enhanced senescence- associated b-galactosidase activity, and expansion of promyelocytic nuclear compartments [53]. In support of the DNA damaging activity of CDT, a genome wide analysis performed in S. cerevisae identified HR and activation of the DNA damage checkpoints as comprising essential mechanisms for the response to damage induced by the conditional expression of the active CdtB subunit from C. jejuni [54]. Another important aspect in the context of bacterial genotoxins and carcinogenesis is the activation of sur- vival signaling pathways because the survival of cells with damaged DNA enhances the risk of acquiring genomic instability and favors tumor initiation and ⁄ or progression [55,56] (Fig. 4). The survival of CDT intoxicated cells is dependent on the activation of the small GTPase RhoA [45], which induces actin stress fiber formation via the RhoA kinases, ROCKI and ROCKII, and prevents cell death via activation of the mitogen-activated protein kinase p38 and its down- stream target mitogen-activated protein kinase-acti- vated protein kinase 2 [57] (Fig. 2). The activation of RhoA is dependent on the dephosphorylation on ser- ine152 of the RhoA-specific guanine nucleotide exchange factor Net1, and it appears to be part of the DDR response because it requires a functional ATM [57]. The cellular responses to the two bacterial toxins are summarized in Fig. 2. Infection with CDT-producing bacteria: chronic inflammation and tumor progression Over the past 10 years, chronic inflammation has been shown to be associated with an enhanced risk of tumor development. How can inflammation favor the acquisi- tion of the mutator phenotype? The inflammatory environment is characterized by the production of ROS and reactive nitrogen intermediates. These com- pounds are potent genotoxic agents that may increase the mutation rate and promote the accumulation of genomic instability, thus altering the crucial biological processes such as the regulation of DNA repair and DDRs, allowing tumor initiation ⁄ progression [58,59]. Chronic inflammation is also associated with constit- utive activation of pleiotropic nuclear factor (NF)-jB, Fig. 4. Possible role of bacterial genotoxins in cancer development. Chronic infection with CDT or colibactin-producing bacteria can pro- mote the induction of genomic instability by direct secretion of bac- terial genotoxins and activation of a chronic inflammation, which is associated with the production of endogenous DNA damaging agents, such as ROS. Persistent activation of the transcription fac- tor NF-jB, via the pro-inflammatory cytokine TNF-a or sustained triggering of pathogen recognition receptors (e.g. Toll-like recep- tors), in combination with survival signals induced by cellular intoxi- cation with genotoxins, may allow cells carrying genomic instability to break through the tumorigenesis barrier, resulting in an increased risk of tumor development. L. Guerra et al. Bacterial genotoxins and genomic instability FEBS Journal 278 (2011) 4577–4588 ª 2011 The Authors Journal compilation ª 2011 FEBS 4583 which promotes cell survival, and has been demon- strated to contribute to tumor formation in models of colitis-associated cancer and hepatocellular carcinoma [60,61]. Bacteria that cause persistent infections associated with chronic inflammation have a higher risk of pro- moting carcinogenesis [58]. The best-studied example is H. pylori and its association with gastric carcinoma and mucosa-associated lymphoid tissue lymphoma [62]. Chronic infection with this bacterium is associ- ated with the downregulation of mismatch repair and base excision repair proteins both in vitro and in vivo [63], and these effects correlate with a five-fold increased mutation frequency and also the induction of microsatellite instabilities in the gastric epithelium of mice 6 months after infection [64]. Epidemiological evi- dence also demonstrate an increased risk of carcinoma of the gallbladder in chronic carriers of S. enterica, ser- ovar Typhi [2] and colon cancer in individuals colo- nized by Bacteroides fragilis [65]. Bacterial products can contribute to a sustained and deregulated inflam- matory microenvironment by stimulation of the host pathogen recognition receptors, leading to a constant supply of ROS, reactive nitrogen intermediates and cytokines. The most well-characterized family of PPRs is the Toll-like receptor family. The majority of Toll- like receptor signaling converges to the adaptor molecule MyD88 and activates the transcription factor NF-jB. MyD88 knockdown was shown to strongly reduce the development of spontaneous colorectal carcinoma in mice carrying heterozygous mutation of the tumor suppressor genes APC (Apc Min ⁄ + mouse model) [66]. In addition to the contribution of the innate immune responses in inflammation-induced car- cinogenesis via stimulation of PPR or pro-inflamma- tory cytokine production, there is now evidence linking T cell-mediated immune responses in infection-induced carcinogenesis. The production and secretion of the B. fragilis toxin induces colitis, which further develops into colon cancer in the Apc Min ⁄ + mouse model, and the carcinogenic capacity of B. fragilis toxin-producing strains is directly associated with the recruitment of the highly pro-inflammatory subset of T helper (Th) 17 lymphocytes [67]. Figure 4 summarizes the effectors that may contrib- ute to carcinogenesis in chronic bacterial infections. Considering the importance of establishing a chronic infection as risk for tumor development, several studies have assessed whether CDT may contribute to persis- tent bacterial colonization of the gastrointestinal tract. Fox et al. [68] demonstrated that a functional toxin favors colonization of the stomach and lower bowel of C57BL ⁄ 129 mice infected with C. jejuni. The C57BL ⁄ 129 mouse model was not suitable for studying the effect of a persistent infection with CDT-producing bacteria in the induction of inflammation because bac- teria were detected only in a proportion of the mice. Conversely, a persistent colonization of the stomach and the lower bowel for 100% of animals was achieved in C57BL ⁄ 129 mice carrying a homozygous deletion of p50 and heterozygous deletion of the p65 subunits of the transcription factor NF-jB (p50 ) ⁄ ) p65 + ⁄ ) ). In this model of chronic infection, the presence of CDT- producing bacteria was associated with significantly enhanced severity of the gastritis and a greater induc- tion of gastric hyperplasia and dysplasia, which is an indication of an early neoplastic process [68]. Colonization of the Swiss Webster mice with Heli- cobacter hepaticus was also dependent on expression of a wild-type CDT [69]. This persistent infection was associated with a highly inflammatory response, char- acterized by the production of Th1-associated IgG2a, Th2-associated IgG1 and mucosal IgA [69]. Similarly, a strong inflammatory response was described in the liver of male A ⁄ JCr mice infected with H. hepaticus carrying a wild-type CDT 10 months after infection compared to mice infected with an isogenic strain carrying a mutant toxin, where the cdtB gene was inactivated by transposon mutagenesis. This response was characterized by an increase in transcrip- tion levels of pro-inflammatory (TNF-a, IFN-c and Cox-2, IL-6 and TGF-a) and anti-apoptotic (Bcl-2 and Bcl-X L ) genes, as well as upregulation of hepatic mRNA levels of components of the NF-jB pathway (p65 and p50) [70]. The presence of CDT was further associated with a progression of inflammation to dys- plasia. The dysplastic lesions were characterized by the presence of hepatocytes with a marked variation in cell and nuclear size and shape, as well as a loss of hepatic architecture [70]. An interesting notion, providing fuel for future stud- ies, is to assess whether these conditions of dysplasia are associated with the CDT-dependent induction of DNA damage, chronic activation of the DDRs, acqui- sition of genomic instability and alteration of cellular pathways that regulate senescence, allowing cells to break through the tumorigenesis barrier. Indeed, very little is known about the ability of CDT to induce genomic instability. Effects of colibactin on genomic instability The effects of colibactin in induction of DNA damage in vivo and genomic instability in vitro were studied by Cuevas-Ramos et al. [50]. These authors reported that Bacterial genotoxins and genomic instability L. Guerra et al. 4584 FEBS Journal 278 (2011) 4577–4588 ª 2011 The Authors Journal compilation ª 2011 FEBS the expression of the clbA gene, a key gene for the syn- thesis of this genotoxin, was detected in a mouse intes- tinal loop model and in the colons of antibiotic-treated BALB ⁄ cJ mice 6 h and 5 days, respectively, after infec- tion with an E. coli strain harboring the psk genetic island. In each case, an isogenic strain carrying a clbA mutant gene (and therefore unable to produce colibac- tin) was used as negative control. The expression of the clbA gene was further associated with the induction of DNA damage, as assessed by increased phosphory- lation of the histone H2AX [50]. Short-term exposure of the Chinese hamster ovary cell line to psk+ E. coli at low multiplicity of infection (in the range five to 20 bacteria per cell) induced DNA damage that could still be observed, although at a very low level, up to 24 h post-infection in actively cycling cells, indicating that the DNA repair process was not completed. As a consequence of the partial DNA repair, the infected cells presented anaphase bridges, accumulated chromosomal aberrations in approxi- mately 7% of the chromosomes, with ring chromo- somes and translocations being the most common alterations, and aneuploidy (a loss or gain of chromo- somes) 72 h post-infection. Such aberrations were maintained in a proportion of cells up to 21 days post- infection. The chromosomal instability induced by infection with colibactin-producing E. coli was further associated with an enhanced rate of mutation fre- quency and an increased ability of the cells to grow in soft agar, which is a feature of anchorage-independent growth [50]. Future perspectives Our knowledge on how bacterial infections can con- tribute to carcinogenesis has begun to be unraveled. The journey started from epidemiological data and several molecular mechanisms have been identified to date, with special focus on the oncogenic role of H. pylori infection. The identification of bacterial genotoxins opens yet another new avenue. We have come a long way in our understanding of the biology of CDT, although many questions still remain. We do not know: (a) when and under what conditions the toxin is produced in vivo; (b) what is the extent of the DNA damage and genomic instability in in vivo models; and (c) whether there is a correlation between chronic infection of CDT-producing bacteria and an increased risk of cancer development. A retro- spective study demonstrated that infection with the enteropathogenic C. jejuni, where 99% of the strains harbour cdt genes, did not correlate with an increased risk of developing a tumor in the gastrointestinal tract at least during the first 10 years after the detection of infection [71]. However, this bacterium is rarely associ- ated with the establishment of a chronic infection and therefore may not represent a suitable model, despite the fact that C. jejuni was found in tissue specimens derived from intestinal mucosa-associated lymphoid tissue lymphoma patients [72]. The biology of colibactin is still at its infancy because this toxin was only characterized recently and cannot yet be produced as a synthetic molecule in vitro. Several interesting questions remain: (a) how does it induce DNA damage; (b) how does it enter and traffic within the host cells to reach the nuclear compartment; and (c) does it contribute to long-term bacterial colonization in vivo. As a more general evolutionary aspect, we still do not know how bacteria benefit from producing such genotoxins. The experimental work in the field of bacteria and cancer, and specifically on bacterial genotoxins, has been hampered by the complexity of the host–bacteria interaction, although the development of suitable ani- mal models and the implementation of high through- put screenings will provide conditions that allow the pursuit of this exciting issue in the field of medical science. Acknowledgements This work was supported by the Swedish Research Council, the Swedish Cancer Society, the A ˚ ke-Wiberg Foundation, the Magnus Bergvall Foundation, the Karolinska Institutet to T.F., and the Robert Lund- berg Memorial Foundation to L.G. T.F. is supported by the Swedish Cancer Society. References 1 Humans IWGotEoCRt (1994) Schistosomes, Liver Flukes and Helicobacter pylori. Lyon, France. 2 Vogelmann R & Amieva MR (2007) The role of bacte- rial pathogens in cancer. Curr Opin Microbiol 10, 76–81. 3 Raptis S & Bapat B (2006) Genetic instability in human tumors. EXS 96, 303–320. 4 Smith JL & Bayles DO (2006) The contribution of cyto- lethal distending toxin to bacterial pathogenesis. Crit Rev Microbiol 32, 227–248. 5 Scott DA & Kaper JB (1994) Cloning and sequencing of the genes encoding Escherichia coli cytolethal dis- tending toxins. Infect Immun 62, 244–251. 6 Lara-Tejero M & Galan JE (2001) CdtA, CdtB and CdtC form a tripartite complex that is required for L. Guerra et al. Bacterial genotoxins and genomic instability FEBS Journal 278 (2011) 4577–4588 ª 2011 The Authors Journal compilation ª 2011 FEBS 4585 cytolethal distending toxin activity. Infect Immun 69, 4358–4365. 7 Nesic D, Hsu Y & Stebbins CE (2004) Assembly and function of a bacterial genotoxin. Nature 429, 429–433. 8 McSweeney LA & Dreyfus LA (2004) Nuclear localiza- tion of the Escherichia coli cytolethal distending toxin CdtB subunit. Cell Microbiol 6, 447–458. 9 Hassane DC, Lee RB & Pickett CL (2003) Campylobac- ter jejuni cytolethal distending toxin promotes DNA repair responses in normal human cells. Infect Immun 71, 541–545. 10 McSweeney LA & Dreyfus LA (2005) Carbohydrate- binding specificity of the Escherichia coli cytolethal distending toxin CdtA-II and CdtC-II subunits. Infect Immun 73, 2051–2060. 11 Elwell CA & Dreyfus LA (2000) DNAase I homologous residues in CdtB are critical for cytolethal distending toxin-mediated cell cycle arrest. Mol Microbiol 37, 952–963. 12 Lara-Tejero M & Galan JE (2000) A bacterial toxin that controls cell cycle progression as a deoxyribonucle- ase I-like protein. Science 290, 354–357. 13 Hassane DC, Lee RB, Mendenhall MD & Pickett CL (2001) Cytolethal distending toxin demonstrates geno- toxic activity in a yeast model. Infect Immun 69, 5752–5759. 14 Guerra L, Teter K, Lilley BN, Stenerlow B, Holmes RK, Ploegh HL, Sandvig K, Thelestam M & Frisan T (2005) Cellular internalization of cytolethal distending toxin: a new end to a known pathway. Cell Microbiol 7, 921–934. 15 Shenker BJ, Dlakic M, Walker LP, Besack D, Jaffe E, LaBelle E & Boesze-Battaglia K (2007) A novel mode of action for a microbial-derived immunotoxin: the cytolethal distending toxin subunit B exhibits phospha- tidylinositol 3,4,5-triphosphate phosphatase activity. J Immunol 178, 5099–5108. 16 Matangkasombut O, Wattanawaraporn R, Tsuruda K, Ohara M, Sugai M & Mongkolsuk S (2010) Cytolethal distending toxin from Aggregatibacter actinomycetem- comitans induces DNA damage, S ⁄ G2 cell cycle arrest, and caspase- independent death in a Saccharomyces cerevisiae model. Infect Immun 78, 783–792. 17 Rabin SD, Flitton JG & Demuth DR (2009) Aggrega- tibacter actinomycetemcomitans cytolethal distending toxin induces apoptosis in nonproliferating macrophag- es by a phosphatase-independent mechanism. Infect Immun 77, 3161–3169. 18 Haghjoo E & Galan JE (2004) Salmonella typhi encodes a functional cytolethal distending toxin that is delivered into host cells by a bacterial-internalization pathway. Proc Natl Acad Sci USA 101, 4614–4619. 19 Spano S, Ugalde JE & Galan JE (2008) Delivery of a Salmonella typhi exotoxin from a host intracellular com- partment. Cell Host Microbe 3, 30–38. 20 Boesze-Battaglia K, Besack D, McKay T, Zekavat A, Otis L, Jordan-Sciutto K & Shenker BJ (2006) Cho- lesterol-rich membrane microdomains mediate cell cycle arrest induced by Actinobacillus actinomycetem- comitans cytolethal-distending toxin. Cell Microbiol 8, 823–836. 21 Carette JE, Guimaraes CP, Varadarajan M, Park AS, Wuethrich I, Godarova A, Kotecki M, Cochran BH, Spooner E, Ploegh HL et al. (2009) Haploid genetic screens in human cells identify host factors used by pathogens. Science 326, 1231–1235. 22 Mise K, Akifusa S, Watarai S, Ansai T, Nishihara T & Takehara T (2005) Involvement of ganglioside GM3 in G(2) ⁄ M cell cycle arrest of human monocytic cells induced by Actinobacillus actinomycetemcomitans cytole- thal distending toxin. Infect Immun 73, 4846–4852. 23 Eshraghi A, Maldonado-Arocho FJ, Gargi A, Cardwell MM, Prouty MG, Blanke SR & Bradley KA (2010) Cy- tolethal distending toxin family members are differen- tially affected by alterations in host glycans and membrane cholesterol. J Biol Chem 285, 18199–18207. 24 Cortes-Bratti X, Chaves-Olarte E, Lagerga ˚ rd T & Thelestam M (2000) Cellular internalization of cytole- thal distending toxin from Haemophilus ducreyi. Infect Immun 68, 6903–6911. 25 Guerra L, Nemec KN, Massey S, Tatulian SA, Thele- stam M, Frisan T & Teter K (2009) A novel mode of translocation for cytolethal distending toxin. Biochim Biophys Acta 1793, 489–495. 26 Nishikubo S, Ohara M, Ueno Y, Ikura M, Kurihara H, Komatsuzawa H, Oswald E & Sugai M (2003) An N-terminal segment of the active component of the bacterial genotoxin cytolethal distending toxin B (CDTB) directs CDTB into the nucleus. J Biol Chem 278, 50671–50681. 27 Pickett CL & Whitehouse CA (1999) The cytolethal dis- tending toxin family. Trends Microbiol 7, 292–297. 28 Nougayrede JP, Homburg S, Taieb F, Boury M, Brzuszkiewicz E, Gottschalk G, Buchrieser C, Hacker J, Dobrindt U & Oswald E (2006) Escherichia coli induces DNA double-strand breaks in eukaryotic cells. Science 313, 848–851. 29 Putze J, Hennequin C, Nougayrede JP, Zhang W, Homburg S, Karch H, Bringer MA, Fayolle C, Carniel E, Rabsch W et al. (2009) Genetic structure and distri- bution of the colibactin genomic island among members of the family Enterobacteriaceae. Infect Immun 77, 4696–4703. 30 Homburg S, Oswald E, Hacker J & Dobrindt U (2007) Expression analysis of the colibactin gene cluster coding for a novel polyketide in Escherichia coli. FEMS Microbiol Lett 275, 255–262. 31 Ciccia A & Elledge SJ (2010) The DNA damage response: making it safe to play with knives. Mol Cell 40, 179–204. Bacterial genotoxins and genomic instability L. Guerra et al. 4586 FEBS Journal 278 (2011) 4577–4588 ª 2011 The Authors Journal compilation ª 2011 FEBS [...]... distending toxin induces cell cycle arrest and apoptosis via the DNA damage checkpoint pathways J Biol Chem 276, 5296–5302 47 Li L, Sharipo A, Chaves-Olarte E, Masucci MG, Levitsky V, Thelestam M & Frisan T (2002) The Haemophilus ducreyi cytolethal distending toxin activates sensors of DNA damage and repair complexes in proliferating and non-proliferating cells Cell Microbiol 4, 87–99 Bacterial genotoxins and. .. Frost JA & Frisan T (2008) A bacterial cytotoxin identifies the RhoA exchange factor Net1 as a key effector in the response to DNA damage PLoS ONE 3, e2254 58 Grivennikov SI, Greten FR & Karin M (2010) Immunity, inflammation, and cancer Cell 140, 883–899 59 Karin M, Lawrence T & Nizet V (2006) Innate immunity gone awry: linking microbial infections to chronic inflammation and cancer Cell 124, 823–835... cells Cell Microbiol 4, 87–99 Bacterial genotoxins and genomic instability 48 Yamamoto K, Tominaga K, Sukedai M, Okinaga T, Iwanaga K, Nishihara T & Fukuda J (2004) Delivery of cytolethal distending toxin B induces cell cycle arrest and apoptosis in gingival squamous cell carcinoma in vitro Eur J Oral Sci 112, 445–451 49 Sato T, Koseki T, Yamato K, Saiki K, Konishi K, Yoshikawa M, Ishikawa I & Nishihara... of the cytolethal distending toxin Toxins 3, 172–190 45 Frisan T, Cortes-Bratti X, Chaves-Olarte E, Stenerlow ¨ B & Thelestam M (2003) The Haemophilus ducreyi cytolethal distending toxin induces DNA double strand breaks and promotes ATM-dependent activation of RhoA Cell Microbiol 5, 695–707 46 Cortes-Bratti X, Karlsson C, Lagergard T, Thelestam M & Frisan T (2001) The Haemophilus ducreyi cytolethal... actinomycetemcomitans cytolethal distending toxin Infect Immun 70, 528–534 50 Cuevas-Ramos G, Petit CR, Marcq I, Boury M, Oswald E & Nougayrede JP (2010) Escherichia coli induces DNA damage in vivo and triggers genomic instability in mammalian cells Proc Natl Acad Sci USA 107, 11537–11542 51 Sert V, Cans C, Tasca C, Bret-Bennis L, Oswald E, Ducommun B & De Rycke J (1999) The bacterial cytolethal distending toxin (CDT)... Kagnoff MF & Karin M (2004) IKKbeta links inflammation and tumorigenesis in a mouse model of colitis-associated cancer Cell 118, 285–296 61 Pikarsky E, Porat RM, Stein I, Abramovitch R, Amit S, Kasem S, Gutkovich-Pyest E, Urieli-Shoval S, FEBS Journal 278 (2011) 4577–4588 ª 2011 The Authors Journal compilation ª 2011 FEBS 4587 Bacterial genotoxins and genomic instability 62 63 64 65 66 67 L Guerra et al Galun... evokes cellular senescence with persistent DNA damage and cytokine signaling J Cell Mol Med 14, 357–367 54 Kitagawa T, Hoshida H & Akada R (2007) Genomewide analysis of cellular response to bacterial genotoxin CdtB in yeast Infect Immun 75, 1393–1402 55 Kastan MB & Bartek J (2004) Cell-cycle checkpoints and cancer Nature 432, 316–323 56 Shiloh Y (2003) ATM and related protein kinases: safeguarding genome... Collaboration and competition between DNA double-strand break repair pathways FEBS Lett 584, 3703–3708 38 Stracker TH & Petrini JH (2011) The MRE11 complex: starting from the ends Nat Rev Mol Cell Biol 12, 90–103 39 You Z & Bailis JM (2010) DNA damage and decisions: CtIP coordinates DNA repair and cell cycle checkpoints Trends Cell Biol 20, 402–409 40 Lieber MR (2010) The mechanism of double-strand DNA break... Negrini S, Gorgoulis VG & Halazonetis TD (2010) Genomic instability – an evolving hallmark of cancer Nat Rev Mol Cell Biol 11, 220–228 42 Pino MS & Chung DC (2010) The chromosomal instability pathway in colon cancer Gastroenterology 138, 2059–2072 43 Thelestam M & Frisan T (2006) Cytolethal distending toxins In The Comprehensive Sourcebook of Bacterial Protein Toxins (Alouf J & Popoff M eds), pp 448–467... gastric cancer and beyond Nat Rev Cancer 10, 403–414 Machado AM, Figueiredo C, Seruca R & Rasmussen LJ (2010) Helicobacter pylori infection generates genetic instability in gastric cells Biochim Biophys Acta 1806, 58–65 Touati E, Michel V, Thiberge JM, Wuscher N, Huerre M & Labigne A (2003) Chronic Helicobacter pylori infections induce gastric mutations in mice Gastroenterology 124, 1408–1419 Toprak NU, . REVIEW ARTICLE Do bacterial genotoxins contribute to chronic inflammation, genomic instability and tumor progression? Lina Guerra, Riccardo Guidi and Teresa. geno- toxins, as well as their contribution to induction of chronic infection ⁄ inflammation, genomic instability and tumor progression. CDT CDT is the first bacterial