Bacterial protein toxins in human cancers

Many bacteria causing persistent infections produce toxins whose mechanisms of action indicate that they could have a role in carcinogenesis. Some toxins, like CDT and colibactin, directly attack the genome by damaging DNA whereas others, as for example CNF1, CagA and BFT, impinge on key eukaryotic processes, such as cellular signalling and cell death. These bacterial toxins, together with other less known toxins, mimic carcinogens and tumour promoters. The aim of this review is to fulfil an up-to-date analysis of toxins with carcinogenic potential that have been already correlated to human cancers. Bacterial toxins-induced carcinogenesis represents an emerging aspect in bacteriology, and its significance is increasingly recognized.

INTRODUCTION

Cancer development is a multifactorial process (Shappell et al. 2004; Lakritz et al. 2014), its evolution depending on the micro and macro environment (Grivennikov, Greten and Karin 2010; Hanahan and Weinberg 2011; Hanahan and Coussens 2012). In fact, the genetic as well as the life style and immune and stromal cells, cytokines, proteases and hormones are acknowledged as major contributors in the natural history of cellular malignant transformations (Grivennikov, Greten and Karin 2010; Hanahan and Weinberg 2011). Nowadays, the link between some infectious diseases and cancer development is emerging. In this context, it is known that viruses, such as hepatitis B virus, Epstein–Barr virus and human papilloma virus, play a direct role in carcinogenesis, mechanistic effects of viral single genes resulting in cell transformation (Kuper, Adami and Trichopoulos 2000). In the last decades, starting from the discovery of the link between Helicobacter pylori and gastric cancer (IARC Working Group 1994), the involvement of bacteria in carcinogenesis has been recognized, with a consequent outburst of the research in the field. Pathogenic bacteria produce virulence factors that enable them to interact with the host in order to colonize it or to alter its equilibrium, thus causing disease. Among virulence factors, bacterial protein toxins, including those somehow connected to the development of cancer, have been the targets of a large number of studies.

To suit their own needs, pathogenic bacteria engage protein toxins to manipulate crucial cellular processes of the host, such as protein synthesis, cytoskeleton organization, intracellular signalling pathways and trafficking of vesicles. Thanks to their highly specific enzymatic activity on molecular targets, some bacterial toxins have represented useful tools for molecular biology studies and, in recent years, have been considered for medical applications (Fabbri et al. 2015). Unfortunately, these bacterial defensive factors can lead, as a side effect, to mutational events that contribute to carcinogenesis and can drive transformation by affecting genomic stability, resistance to cell death and proliferative signalling (Candela et al. 2014).

In this review, up-to-date information on the involvement of bacterial protein toxins in cancer promotion and/or development is reported. We will gather such toxins into two subgroups, the first one comprising toxins that exert the carcinogenic activity by damaging DNA, and the second one characterized by toxins that specifically perturb cell signalling pathways somehow linked to transformation. For each toxin, the main actions related to carcinogenesis will be briefly explained.

DNA-DAMAGING TOXINS

A number of bacterial protein toxins damage DNA and, by means of this action, are thought to be related to cancer development. Up to now, only toxins produced by intestinal bacteria have been reported to exert such activities. In this context, Escherichia coli is, to date, the dominant producing bacterium (Elliott et al. 2013) as well as the most frequently found in colorectal cancer biopsies (Buc et al. 2013) although other species and genera have been uncovered (Putze et al. 2009).

Colibactin

Certain E. coli strains of the phylo-genetic group B2 harbour the polyketide synthetase (pks) island, which is involved in the synthesis of the non-ribosomal peptide-type genotoxin colibactin (Nougayrède et al. 2006; Putze et al. 2009). Colibactin has been suggested as having an impact on cancer development. In fact, pks-harbouring bacteria are able to induce DNA double-strand breaks in eukaryotic cells both in vitro and in vivo, thus causing the appearance of phosphorylated H2AX foci in mouse enterocytes (Arthur et al. 2012). These lesions activate the DNA damage checkpoint pathway ATM-CHK2 and cells present signs of incomplete DNA repair, marked by megalocytosis, G2/M cell cycle arrest, chromosomal instability and anchorage-independent colony formation, all features supporting the carcinogenic potential of colibactin (Cuevas-Ramos et al. 2010; Nougayrède et al. 2006). Very recently, Cougnoux and coworkers (2014) demonstrated that tumour growth induced by colibactin-producing bacteria is sustained by cellular senescence via the secretion of the growth factor HGF.

Escherichia coli harbouring the colibatin-producing genes was recently shown to be overrepresented in biopsies of patients with human colorectal tumours (Buc et al. 2013). It is important to underline that experiments in colitis-susceptible interleukin 10−/− mice demonstrate that the presence of E. coli pks is able to promote invasive carcinoma. In fact, deletion of the pks genotoxic island from the E. coli NC101 decreases tumour multiplicity and invasion in AOM/Il10(−/−) mice (Arthur et al. 2012). However, inflammation has been reported to be an important player in such a process since it targets the intestinal microbiota; thus, inducing expansion of microbes that influence CRC in mice and altering the expression of a small subset of microbial genes that are crucial for tumour development (Arthur et al. 2014).

Cytolethal distending toxin

Cytolethal distending toxin (CDT) was initially isolated from both E. coli and Campylobacter jejuni strains, although subsequently other Gram-negative bacterial pathogens obtained from human clinical isolates were shown to produce CDT (Smith and Bayles 2006). The CDT holotoxin is made of three different subunits, CdtA, CdtB and CdtC, encoded by three genes organized in one operon (Thelestam and Frisan 2006). CdtB is the enzymatic subunit whereas the binding to the host cell membrane is dependent on the CdtA and CdtC subunits. CDT internalization in host cells occurs through receptor-mediated endocytosis, via dynamin-dependent endocytosis and requires intact lipid rafts (Cortes-Bratti et al. 2000; Guerra et al. 2005, 2009; Boesze-Battaglia et al. 2006). After cell entry by endocytosis, CdtB follows a retrograde endosome-Golgi traffic to the endoplasmic reticulum (Cortes-Bratti et al. 2000; Guerra et al. 2009) and subsequently is translocated to the nucleus where it exerts its cytotoxic activity (Guerra et al. 2009). Important domains for CdtB nuclear transport have been characterized in the CdtB N-terminal region, and studies on this sequence have suggested that CdtB nuclear localization is crucial for CDT cytotoxicity (Nishikubo et al. 2003).

The enzymatically active subunit, CdtB, is structurally and functionally homologous to the mammalian deoxyribonuclease I (DNase I) (Elwell and Dreyfus 2000; Lara-Tejero and Galan 2000) and in fact, in eukaryotic cells, CDT intoxication leads to distinctive features characterized by DNA damage and activation of the DNA damage responses (DDR) (Frisan et al. 2003; Hassane, Lee and Pickett 2003). These cytotoxic effects result in the G1 and/or G2 phase arrest of the target cell cycle and in the activation of DNA repair mechanisms. Cells that fail to repair the damage undergo senescence or apoptosis (Cortes-Bratti, Frisan and Thelestam 2001a; Cortes-Bratti et al. 2001b; Blazkova et al. 2010; Liyanage et al. 2010; for a comprehensive review see Guerra et al. 2011). However, in some circumstances, the DDR is counteracted by the concomitant generation of survival signals mediated by activation of the guanine nucleotide exchange factor Net1, the RhoA GTPase and its downstream effector p38 MAP kinase (MAPK) (Guerra et al. 2008; for a review see Frisan 2015).

It is well-known that chronic exposure to DNA damaging agents, either endogenous (such as reactive oxygen species) or exogenous (such as ionizing radiation), may cause genomic instability because of the alteration of key genes involved in DDR, cell cycle regulation or DNA repair mechanisms, therefore favouring tumour initiation and progression (Janssen and Medema 2013). Interestingly, when cells are exposed for a long period of time to sublethal doses of CDT, they show increased frequency of mutations, accumulation of chromosomal aberrations, impairment of the DDR, partial failure to activate cell cycle checkpoints in response to genotoxic stress and activation of p38 MAPK-dependent survival signals (Guidi et al. 2013). These data suggest that long-term exposure to sublethal doses of CDT allows the selection of cells that have bypassed the tumorigenesis barrier, thus favouring tumour progression.

The CdtB subunit is also part of the typhoid toxin produced by Salmonella enterica serovar Typhi and by non-typhoidal (Song, Gao and Galan 2013; Mezal, Bae and Khan 2014). In this form, it is connected to PltA, which possesses ADP ribosyl transferase activity, and to the PltB subunit, which represents the binding component of the holotoxin. CdtB-encoding Salmonellae induces, in human epithelial Henle-407 cells, cell cycle arrest in the G2/M phase a phenomenon absent in cells infected with corresponding isogenic cdtB null mutants (Rodriguez-Rivera et al. 2015). According to the model suggested by Spanò, Ugalde and Galan (2008), the typhoid toxin could be implicated in the persistent infection in humans, characteristic of this pathogen. In fact, S. typhi establishes residence within the gall bladder and produces the typhoid toxin that could act on other cells and at systemic sites. It is noteworthy that CDT through the action of its active subunit CdtB is known to hamper the activity of immune cells (Lara-Tejero and Galan 2000; Mooney et al. 2001). Therefore, the typhoid toxin activities may be critical for S. Typhi to establish persistent infection in the human host with the aim of avoiding immune defences. In this context, it is of importance to underline that chronic infections with S. Typhi have been linked to cancer of the gallbladder (Caygill et al. 1994).

Thus, the tumorigenic effect of colibactin and CDT are characterized by the induction of a response generated from DNA damage. This response that impinge on the cellular machinery characteristic of DDR, together with the involvement of proliferating signals, pushes cells towards a transformed phenotype (Fig. 1).

CELL-SIGNALLING-DISRUPTING TOXINS

The majority of bacterial protein toxins influencing tumour onset and progression interfere with pathways related to cell proliferation. However, a deregulated inflammatory response is emerging as an important mechanism entailed in bacteria-induced cancer. In fact, bacterial protein toxins, as well as the producing bacteria, often act on immune cells in diverse ways, thus stimulating a persistent inflammatory state characterized by specific immune cells activation and by the production of cytokines and metabolites. The association between the stimuli derived from toxins and from the inflammatory state can push more easily cells towards cancer development (Candela et al. 2014). Moreover, inflammation allows the proliferation of pathogenic opportunistic bacteria to the detriment of symbionts (Guigox, Doré and Schiffrin 2008; Sansonetti 2011), further creating an environment predisposed to cancer progression.

Cytotoxin-associated gene-A (CagA) and Vacuolating cytotoxin A (VacA)

Helicobacter pylori is a microaerophilic Gram-negative bacterium, whose ideal habitat is the gastric mucus of the human stomach. Therefore, the bacterium is able to live in an acid environment and needs small amounts of oxygen. It selectively colonizes the gastric epithelium of approximately 50% of the population and evidence suggests that humans have been infected with H. pylori for at least 88 000 years (Moodley et al. 2012).

Helicobacter pylori is classified as a class I carcinogen (International Agency for Research on Cancer (IARC) Biological agents. 2012) and represents the strongest known risk factor for severe gastric diseases, ranging from chronic gastritis and ulceration to neoplastic changes in the stomach (Cid et al. 2013; De Martel et al. 2012). Eradication of H. pylori infection has been reported as a valuable strategy for both the treatment of peptic ulcers and gastric mucosa-associated lymphoid tissue (MALT) lymphoma as well as for the prevention of gastric cancer (Fukase et al. 2008; Suzuki, Nishizawa and Hibi 2010). Gastric carcinoma is the result of a long, multistep and multifactorial process, and is influenced by host genetic susceptibility factors, environmental factors, and H. pylori virulence (for a review see Ferreira, Machado and Figueiredo 2014). The major pathogenic factors harboured by H. pylori are the cytotoxin VacA, the pathogenicity island cag encoding a type 4 secretion system and the first bacterial oncoprotein CagA.

The CagA factor resides in the Cag pathogenicity island (cagPAI) which is a 40 kb stretch of DNA acquired by a horizontal DNA transfer event from a yet unknown donor and only present in a subset of highly virulent H. pylori strains (Covacci et al. 1999). The cagPAI also encodes for a type 4 secretion system that is used to inject CagA into a target cell upon H. pylori attachment. After translocation, CagA localizes to the inner surface of the cell membrane and undergoes tyrosine phosphorylation by Src family kinases (SFKs) (Hatakeyama and Higashi 2005). The Abl kinase members, c-Abl and Arg, directly phosphorylate CagA within Glu-Pro-Ile-Tyr-Ala (EPIYA) motifs (Tammer et al. 2007). To date, four different EPIYA motifs (A, B, C and D) have been described, based on the surrounding amino acid sequences (Hatakeyama 2011). CagA from Western strains are characterized by EPIYA A and B motifs followed by one or more EPIYA C motifs. CagA from East Asian strains contains one EPIYA D motif instead of EPIYA C (Hatakeyama 2011). Upon phosphorylation in the EPIYA-C and EPIYA-D motifs, CagA interacts with host molecules, leading to morphological changes resulting in cell elongation (Higashi et al. 2002; Backert, Tegtmeyer and Selbach 2010). A higher number of copies of the EPIYA C motif and the presence of the EPIYA D motif, are associated with increased levels of CagA phosphorylation that lead to a higher extent of cytoskeletal changes in epithelial cells (Naito et al. 2006; Argent et al. 2004). Moreover, increased motif numbers increase the proinflammatory and carcinogenic potential of the toxin. The interaction with SFKs partners play complex roles in H. pylori-induced actin–cytoskeletal rearrangements, scattering and elongation of infected host cells in culture (Backert, Tegtmeyer and Selbach 2010) a phenomenon referred to as the ‘hummingbird phenotype’ (Segal et al. 1999). In addition, several additional binding partners were identified including Shp1, PI3-kinase, Grb2, Grb7 and Ras-GAP (Backert, Tegtmeyer and Selbach 2010), so phosphorylated CagA can influence GTPases. Not all cellular effects exerted by CagA depend on tyrosine phosphorylation. Host interaction partners of non-phosphorylated CagA have been described, and these interactions have been reported to induce the transcription factor NF-κB (Rieder, Fischer and Haas 2005), the disruption of cell-to-cell junctions by interaction with the transmembrane protein E-cadherin (Murata-Kamiya et al. 2007) or the loss of cell polarity (Zeaiter et al. 2008). Moreover, in gastric epithelial cells, CagA stimulates increased levels of spermine oxidase (SMO) that results in both apoptosis and DNA damage (Chaturvedi et al. 2012). However, although perturbing multiple host signalling pathway and despite its decisive role in the development of gastric cancer, CagA is not required for the maintenance of a neoplastic phenotype in established cancer cells. CagA-conducted gastric carcinogenesis progresses through a hit-and-run mechanism in which prooncogenic actions of CagA are successively taken over by a series of genetic and/or epigenetic alterations compiled in cancer-predisposing cells during long-standing infection with CagA-positive H. pylori (Hatakeyama 2014). Studies with transgenic expression of CagA uncovered a genetic interaction between CagA and a number of phenomena linked to cancer progression, including JNK signalling (Wandler and Guillemin 2012), increase in cell proliferation, cytoplasmic and nuclear accumulation of the Wnt effector β-catenin, and activation of known Wnt target genes (Neal et al. 2013). In this context, it has been reported that long-term expression of wild-type CagA, but not the phosphorylation-resistant form, is sufficient to induce pathologic intestinal hyperplasia in adults (Neal et al. 2013). It is important to underline that transgenic expression of CagA in mice is sufficient to cause gastric and intestinal carcinomas, but only in less than 5% of the animals (Ohnishi et al. 2008). Conversely, in the transgenic zebrafish model, high rates of intestinal neoplasia develop when CagA is expressed with a mutant allele of the tumour suppressor p53 (Neal et al. 2013); thus, suggesting the need for an oncogenic cooperation between CagA and oncoproteins.

Almost all H. pylori strains contain the vacA gene that encodes VacA, so called because of its ability to induce vacuolation in cell lines. VacA inserts into the cell membrane and forms anion-conducting channels (Szabo et al. 1999), but also causes cytochrome C release from mitochondria with consequent apoptosis, binding to cell membrane receptors that result in the initiation of a proinflammatory response, and inhibition of T-cell activation and proliferation (Yamaoka 2010). Like for CagA, a number of association studies have been carried out, relating differences in vacA gene structure and clinical outcomes. Differences have been described in the signal (s) region (s1 and s2) and the middle (m) regions (m1 and m2) that contribute to variations in the in vitro vacuolating activity (Atherton et al. 1995). The region of VacA between the s region and the m region (the intermediate or i region) has been separated into i1, i2 and i3 subtypes (Rhead et al. 2007). Helicobacter pylori strains producing more active s1/i1 form of VacA is strongly associated with gastric adenocarcinoma and is known that the vacA i1 allele is strongly associated with precancerous intestinal metaplasia, in contrast to i2-type strains that shown complete absence of intestinal metaplasia (Winter et al. 2014). Furthermore, one effect of acute VacA exposure is the induction of autophagy but a prolonged exposure to the toxin disrupts autophagy by preventing maturation of the autolysosome (Winter et al. 2014). This mechanism suggests that H. pylori-suppressed autophagy facilitates intracellular survival and persistence of the pathogen, while also generating an environment favouring carcinogenesis (Winter et al. 2014). In summary, despite innumerable papers attempting to relate vacA genotypes to outcome or disease pathogenesis, no truly consistent associations or demonstrable biologic basis for the putative associations has appeared. In vivo, the only consistent association with risk of a clinically important outcome has been the extent and severity of inflammation (Atherton et al. 1997).

It is important to underline that cancer occurs through the joint effects of H. pylori virulence factors and the corresponding responses of epithelial cells. Several recent studies revealed that three separate and sequential steps are essential for H. pylori to exert its virulence on the colonized stomach: (i) adhering to and colonizing the surface of epithelial cells; (ii) escaping and reducing the host defence; and (iii) damaging the epithelium (Sheu et al. 2010). Different virulence factors play a role in the distinct steps, the protein toxins CagA, VacA exploiting their action preferentially on the damage of the epithelium. Other factors are implicated in this activity, such as Tipα that stimulates the production of TNFα and the activation of NF-κB (Tosi et al. 2009) or the outer membrane protein OipA, which is associated with severe pathologic outcomes. Interestingly, OipA inactivation decreases beta-catenin nuclear localization in vitro and reduces the incidence of cancer in animal models (Franco et al. 2008). The other steps are accomplished by several factors implicated in the pathogenesis of H. pylori such urease, BabA (blood group antigen-binding adhesion), SabA (sialic acid binding adhesion), PGN (peptidoglican) and LPS (Liposaccharide) (for a review see He et al. 2014). In particular, adaptation of H. pylori to the hostile environment of the host's stomach is dependent on the secreted extracellular enzymes or adhesions urease, BabA and SabA among the others, whereas PGN and LPS influence the immune evasion step. Additional H. pylori virulence factors that cannot be classified into a precise step exists, but they are beyond the scope of this review.

Cytotoxic necrotizing factor 1

Certain pathogenic E. coli strains produce the Cytotoxic necrotizing factor 1 (CNF1) protein toxin that permanently activates proteins belonging to the Rho GTPase family (Fiorentini et al. 1997), molecular switches that oscillate between an inactive GDP-bound form and an active GTP-bound form. These GTPases encompass three groups of proteins (Rho, Rac and Cdc42) that are differently involved in the actin cytoskeleton organization. Rho induces stress fibre assembly and Rac membrane ruffling activity, whereas Cdc42 is involved in filopodia formation (Hall 1998). Rho, Rac,and Cdc42 are all activated by CNF1 through the deamidation of a pivotal glutamine residue in the switch two domain, which is involved in GTP hydrolysis (glutamine 63 in Rho; Flatau et al. 1997; Schmidt et al. 1997) or 61 in Cdc42 and Rac (Lerm et al. 1999). CNF1 following Rho GTPase activation induces, in cultured cells, a number of different phenomena that suggest a reprogramming of cells, epithelial cells acquiring different and unexpected abilities. Rho GTPases are the master regulators of the actin cytoskeleton and, in this respect, CNF1 generates a number of actin-dependent events in cells, such as multinucleation (Fiorentini et al. 1988), which arises from unsuccessful cytodieresis, nuclear constriction and budding or multipolar metaphases (Malorni and Fiorentini 2006), micropynocytosis (Fiorentini et al. 2001) and increase in cellular motility (Doye et al. 2002). Nevertheless, via the activation of Rho GTPases, CNF1 also triggers events not directly linked to the actin cytoskeleton. It actually causes the activation of NF-κB (Falzano et al. 2003; Boyer et al. 2004; Giamboi Miraglia et al. 2007) via the stimulation of the Akt/IκB kinase pathway and, by doing so, hijacks the host cell fate towards survival. In fact, CNF1 is able to protect from apoptotic stimuli by increasing the amount of anti-apoptotic proteins and by pushing Rho-dependent cell spreading (Fiorentini et al. 1998b,a). Finally, CNF1 induces the production of proinflammatory cytokines, of COX-2 and push quiescent cells to enter the cell cycle (Thomas et al. 2001). It is interesting to note that most of the effects induced by CNF1 are characteristic of transformed cells (Fabbri et al. 2013). Intriguingly, the gene coding for CNF1 has been associated with intestinal mucosa of patients with Crohn's disease (Darfeuille-Michaud et al. 2004). More strikingly, the presence of colonic mucosa-associated E. coli in biopsies from patients with CRC or diverticulosis indicates that E. coli strains producing CNF1 colonize more frequently colon cancers than diverticulosis samples (Buc et al. 2013).

Bacteroides fragilis toxin

Bacteroides fragilis are obligate anaerobes that inhabit and flourish along the entire length of the colon, where they are minority members of the normal colonic microbiota with a propensity for mucosal adherence. However, B. fragilis is also an important opportunistic pathogen, as it is the most common anaerobe isolated from clinical infections despite comprising only a small portion (1–2%) of the total microbial community (for a review see Dejea, Wick and Sears 2013).

There are two molecular subtypes depending on the presence of the bft gene: nontoxigenic B. fragilis (NTBF) and enterotoxigenic B. fragilis (ETBF) (Sears, Geis and Housseau 2014). ETBF strains are implicated in diarrhoeal disease, inflammatory bowel disease and CRC (Basset et al. 2004; Toprak et al. 2006; Sears 2009; Wu et al. 2009), and their pathogenicity is due to a metalloprotease protein toxin called B. fragilis toxin (BFT) (Rhee et al. 2009). In vitro, BFT binds to an intestinal epithelial cell receptor, stimulating a number of pathways that lead to cleavage of the tumour suppressor protein E-cadherin (Wu et al. 2009), activation of Wnt and NF-κB signalling pathways with increased cell proliferation, release of proinflammatory mediators and DNA damage (Sears 2009; Wu et al. 1998; Wu et al. 2004). In vivo, ETBF triggers BFT-dependent acute and chronic colitis in C57BL/6 mice and induces colonic tumours in multiple intestinal neoplasia (Min^Apc+/−) mice, a model for human CRC. The ability of ETBF of increasing tumorigenesis is mediated by an augmented STAT3 expression that leads to the recruitment of the highly proinflammatory subset of T helper type 17 lymphocytes, suggesting that the procarcinogenic role of BFT is to promote a deregulated inflammatory response (Wu et al. 2009). All in all, it is now recognized that persistent ETBF infection may increase the risk of colon carcinogenesis via perturbation of apical junctional complexes and the induction of a proinflammatory response, phenomena that increase the risk of premalignant transdifferentiation. In fact, recent studies found that carriage of toxigenic, as opposed to non-toxigenic, B. fragilis increases in people with CRC, ETBF being only present in approximately 10–20% of the healthy population whereas in CRC patients the faecal carriage of ETBF is increased of about 40% (Toprak et al. 2006; Sears 2009). It is interesting to note that the bft gene is more abundantly found in the colonic mucosa than in luminal samples of colon cancer in humans, being its presence significantly higher in late stage CRCs (Boleij et al. 2015).

Staphylococcal Enterotoxins

Staphylococcus aureus, named for the golden colour of its colonies, is the major virulent bacteria belonging to the genus of staphylococci. It is a Gram-positive bacterium which especially colonizes nasopharyngeal mucous membranes, skin and skin glands. Its infections may be minor if limited to the skin (impetigo, abscesses and boils), but it can also be lethal if spreads through the skin and into the blood or in the heart (Wisplinghoff et al. 2004). Staphylococcus aureus expresses a wide array of cell-associated and secreted virulence factors that render it a multifaceted pathogen capable of a wide range of infections. The virulence factors of pathogenic act differently, some of them favouring its multiplication, others damaging cells and allowing the bacterium to spread to surrounding tissues or distant to the site of primary infection. The toxins produced are cytolysine or haemolysins α (the most widely produced), β, γ and δ and leukocidin-PV (Humphreys et al. 1989). The secreted factors also include various enzymes, cytotoxins, exotoxins and exfoliative toxins. The chief function of these enzymes is to turn host components into nutrients that the bacteria may use for growth. Among the other secreted factors are exotoxins that include the toxic shock syndrome toxin and the staphylococcal enterotoxins (SEs), broadly classified as superantigens since they have the ability to stimulate large populations of T cells leading to the production of a cytokine array. Although there are more than 20 distinct SEs, only a few of them have been studied in depth and the most common are SEA and SEB (For a review see Pinchuk, Beswick and Reyes 2010).

Staphylococcus aureus frequently colonizes patients with cutaneous T-cell lymphoma (CTCL). It has long been suspected that SA, via expression of SEs, can influence the malignant expansion and the evolution of immune dysregulation in patients with CTCL (Tokura et al. 1995; Duvic, Hester and Lemak 1996). The mechanism of oligoclonal expansion of premalignant T cells is confounded by the fact that several studies on T cells from CTCL patients show that malignant T cells often display deficient function and/or deficient expression of CD3 TCR complex (Morice et al. 2006; Klemke et al. 2009). It has been reported that SEs can, in some cases, stimulate the proliferation of malignant T cells (Woetmann et al. 2007). In fact, as illustrated by Woetmann and coworkers (2007), whereas malignant T cells did not respond directly to bacterial superantigens, they proliferate vigorously in response to SEs in cocultures with non-malignant T cells, thus indicating an indirect mechanism for growth promotion by SEs. In fact, SE-treated CTCL cells induce vigorous proliferation of the SE-responsive non-malignant T cells that in turn enhance proliferation of the malignant cells. Recent evidences demonstrate that SEs generate a crosstalk between benign and malignant T cells causing a Stat3-mediated interleukin-10 (IL-10) production by the malignant T cells. The SEs did not stimulate the malignant T cells directly but triggers a cascade of events involving interactions between malignant and benign T cells finally stimulating the malignant T cells to express high levels of IL-10. Much evidence supports that malignant activation of the Stat3/IL-10 axis plays a key role in driving the immune dysregulation and severe immunodeficiency that characteristically develops in CTCL patients (Krejsgaard et al. 2014). It's known that IL-10 promotes CTCL immune dysregulation by several means, including direct inhibition of benign T cells (Wilcox et al. 2009; Krejsgaard et al. 2012). The SEs then modulate the interactions between malignant and benign immune cells leading to enhanced suppression of the host's cellular immunity and increased disease activity, thereby contributing to further deterioration of the skin barrier (Krejsgaard et al. 2014). This weakening of the host's defences stabilizes the S. aureus colonization, facilitates its propagation and increases the susceptibility to secondary infections. It has been demonstrated the link between colonization with SE-producing S. aureus and immune dysregulation in CTCL, thus strengthening the rationale for antibiotic treatment of colonized patients with severe or progressive disease. Moreover, these results highlight the fact that bacterial toxins may contribute to cancer by influencing the crosstalk between tumour and immune cells (Krejsgaard et al. 2014).

Avirulence protein A

Salmonella typhi strains that maintain chronic infections are associated with hepato-biliary cancers (Dutta et al. 2000; Lazcano-Ponce et al. 2001; Wistuba and Gazdar 2004). They secrete Avirulence protein A (AvrA), a protein toxin that influences eukaryotic cell pathways by altering ubiquitination and acetylation of target proteins to modulate inflammation, epithelial cells apoptosis and proliferation (Hardt and Galan 1997; Collier-Hyams et al. 2002; Sun et al. 2004; Liao et al. 2008; Du and Galan 2009; Liu et al. 2010). Also, it increases β-catenin signalling, thus enhancing intestinal epithelial cell proliferation (Sun et al. 2004; Ye et al. 2007;). Very recently, Lu and coworkers (2014) reported a clear role for AvrA on colon cancer development using a mouse model. They examined the effects of chronic infection with AvrA-expressing bacteria on inflammation-associated colon cancer and demonstrated that AvrA enhances proliferation and promotes colonic tumorigenesis and tumour progression, concomitantly activating the protooncogene β-catenin and its downstream targets c-Myc and cyclin D1.

It is worth noting that S. typhi also produce CDT that, as above reported (see CDT section), possesses a carcinogenic potential. To our knowledge, no studies so far exist that correlate the activity of CDT and AvrA in tumour development, although we cannot rule out a cooperation or a synergism between the two toxins.

FadA

Fusobacterium nucleatum, a Gram-negative anaerobe, is an emerging pathogen, which attracts the attention of medical and research communities. It is ubiquitous in the oral cavity but absent or exceptionally detected in other part of the body under normal conditions (Aagaard et al. 2012; Segata et al. 2012). However, under disease conditions, F. nucleatum is one of the most prevalent species found in extraoral sites (Han 2015). Fusobacterium nucleatum is coming out as a potential candidate for CRC susceptibility and, interestingly, has been discovered to be enriched in the carcinomas of CRC patients (Castellarin et al. 2012; Kostic et al. 2012; Rubinstein et al. 2013). Subsequent studies disclose that altered levels of F. nucleatum are also found in the different stages of colorectal neoplasia development, (Kostic et al. 2013; McCoy et al. 2013). The FadA adhesin (a bacterial cell surface adhesion component) seems to be the factor responsible for the carcinogenic effect of F. nucleatum., FadA binds host E-cadherin and activates β-catenin signalling, leading to increased expression of transcription factors, oncogenes, Wnt genes and inflammatory genes, as well as growth stimulation of CRC cells (Rubinstein et al. 2013). Interestingly, the fadA gene levels are >10–100 times higher in the colon tissue from patients with adenomas and adenocarcinomas compared to normal individuals. In the same vein, augmented FadA expression in CRC correlates with an increase in the expression of inflammatory and oncogenic genes (Rubinstein et al. 2013).

All in all, the signalling pathways touched by bacterial toxins to induce cancer are multiple. They comprise activation of different target proteins that in general lead to uncontrolled cell proliferation. An interesting aspect is the induction of immune de-regulation induced by certain toxins that is probably a more general response that has so far been underestimated (Fig. 2).

CONCLUSIONS

The number of microbial cells is 10 times larger than the number of eukaryotic cells in the human body and the majority of them reside in the gut, in a continuum of dynamic ecological communities. Due to the high number and to the complexity of the bacterial species, identification of microbial organisms or bacterial products contributing to cancer remains challenging and deciphering the roles of specific bacterial products in cancer is still in its infancy. However, progress has been made, and the future is clearly in a deeper definition of the symbiotic relationships between microbiota and host that form together a complex ‘superorganism’. Such symbiotic relationship confers benefits to the host but, on the opposite side, defects in the regulatory circuits of the host that control bacterial sensing and homeostasis, or alterations of the microbiota can promote disease by various mechanisms, including the boost of toxigenic bacteria.

This review deals with this last aspect, focusing on bacterial protein toxins, with the aim to offer an updated reading of their involvement in tumour onset and progression. We highlighted the two main processes engaged by such a toxins, i.e. DNA damage and enrolment of host pathways involved in carcinogenesis. All in all, tumour development seems to be a side effect of the bacterial infection. In fact, all the bacterial toxins herein described are essential to establish the infectious disease. However, due to the central role of the pathways touched, their subversion or, particularly, their prolonged stimulation generates responses different from those of the infectious process. As cancer onset is a multifactorial process, we have underlined that additional factors, such as chronicity of infection, inflammation and environmental changes, are most probably involved in cancer-related bacterial activity. This is the case, for example, of CDT that at low doses and in a chronic exposure is responsible for mutagenic and carcinogenic effects (Guidi et al. 2013).

Although the majority of toxins so far described as involved in cancer development are produced by gastrointestinal bacteria, other toxins acting in different body districts are currently coming out as potential cancer inducers or reinforcers.

Nevertheless, considering the complexity of the phenomenon, it is only a comprehensive understanding of the relationships between host and microbiota that could furnish in the near future new therapeutic targets to be utilized for improving human health.

The authors are grateful to Rossella Di Nallo for her invaluable technical assistance in preparing the manuscript.

FUNDING

This work was supported by the Ministry of Health [grant number PE-2011-02347510 to A.F.].

Conflict of interest. None declared.

REFERENCES