The Bacteroides species are Gram-negative, obligate anaerobic bacteria that represent approximately 25% of all anaerobes in the gut microbiota (Wexler, 2007). Bacteroides spp. mostly exist in a commensal relationship with their host and serve key roles in nutrition and immunity (Wexler, 2007; Sears, 2009; Zafar and Saier, 2021). However, some strains can be opportunistic pathogens, causing widespread disease if they escape the gut. B. fragilis is one of the most widely studied of these strains due to its opportunistic nature and high virulence. Furthermore, B. fragilis is the most commonly isolated anaerobe from clinical cases of diarrhea, sepsis, and extra-intestinal infections (Haghi et al., 2019).

Strains of B. fragilis are characterized based on the presence or absence of a 20-kDa protein toxin called B. fragilis toxin (BFT), or fragilysin (Moncrief et al., 1995; Wu et al., 1998; Wu et al., 2004). Enterotoxigenic B. fragilis (ETBF) contain this toxin, while non-enterotoxigenic B. fragilis (NTBF) do not. BFT is transcribed from the B. fragilis toxin gene (bft) and is one of three different isotypes, BFT-1, BFT-2, and BFT-3 (Franco et al., 2002). BFT functions as a zinc-dependent metalloprotease that cleaves E-cadherin and causes its degradation (Wu et al., 1998; Wu et al., 2004). E-cadherin is necessary for maintaining the mechanical integrity of the IEC barrier and proper maturation of Paneth and goblet cells, and reduced levels of E-cadherin has been linked to both ulcerative colitis and Crohn’s disease (Schneider et al., 2010). BFT-induced cleavage of E-cadherin stimulates the opportunistic nature of ETBF and promotes disease pathogenesis. ETBF and its toxic properties were initially described in lamb diarrheal disease and later spread to other livestock, such as piglets, foals, and calves (Myers et al., 1984; Myers et al., 1985; Myers and Shoop, 1987; Myers et al., 1987a; Collins et al., 1989). It was further described in cases of diarrhea in children (Myers et al., 1987b; Sack et al., 1992; Sack et al., 1994; Durmaz et al., 2005). It has since been established as a global cause of diarrhea (Sears, 2009).

Bacteroides fragilis toxin induces morphological changes in IECs including loss of cell-cell adhesions, rounding, swelling, and alterations to the actin cytoskeleton (Weikel et al., 1992; Donelli et al., 1996). It enhances expression of chemokines (interleukin 8 (IL-8), transforming growth factor beta (TGF-β), epithelial neutrophil-activating protein 78 (ENA-78), and growth-regulated alpha protein (GRO-α)) (Sanfilippo et al., 2000; Kim et al., 2001); secretory response factors in the gut (cyclooxygenase-2 (COX-2) and prostaglandin E2) (Kim et al., 2006); leukocyte recruitment proteins (intercellular adhesion molecule 1 (ICAM-1)) (Roh et al., 2011); and anti-inflammatory mediators (heme oxygenase-1 (HO-1)) (Ko et al., 2016; Ko et al., 2020). BFT activates NF-κB, β-catenin, and MAPK signaling (KIM et al., 2002; Kim et al., 2005) and plays a role in the upregulation of the DNA damage marker, phosphorylated H2AX (γ-H2AX), and production of ROS through spermine oxidase (SMO) (Goodwin et al., 2011). Recently, Allen et al. (2022) utilized whole-genome and whole-exome sequencing on mouse tumors after colonization with ETBF and found that ETBF did not produce a unique mutational profile (Allen et al., 2022). ETBF-induced tumors followed a similar mutational pattern as spontaneous tumors, resulting from errors in DNA mismatch repair machinery and homologous recombination. Interestingly, B. fragilis was significantly enriched in mismatch repair-deficient (dMMR) CRC but not mismatch repair-proficient (pMMR) CRC (Hale et al., 2018).

B. fragilis has also been shown to modulate the anti-tumor immune response to immunotherapy (Vétizou et al., 2015). Mouse tumors grown in germ-free conditions or with broad-spectrum antibiotics were unresponsive to cytotoxic T-lymphocyte-associated antigen 4 (CTLA-4) blockade; however, when B. fragilis was recolonized in mice, these tumors regained responsiveness to CTLA-4 treatment in a manner consistent with an elevated T helper 1 (Th1) response and maturation of intratumoral dendritic cells (DCs) (Vétizou et al., 2015). These findings not only suggest the need of gut microbiota for the efficacy of some immunotherapies, but also the protective, anti-cancer properties of B. fragilis in bolstering immunotherapy.

It should be noted that Kordahi et al. (2021) recently discovered a possible role of NTBF in CRC development, despite being the so-called non-enterotoxigenic form. Researchers found that B. fragilis isolates from patients with colonic polyps were colonized mostly with NTBF, not ETBF, and NTBF was positively correlated with levels of interleukin 12p40 (IL-12p40) and polyp size (Kordahi et al., 2021). These NTBF isolates were enriched in genes related to lipopolysaccharide (LPS) synthesis and were able to induce TLR4 signaling and a pro-inflammatory response (Kordahi et al., 2021). Based on the findings by Kordahi et al. (2021), NTBF should not be excluded as a plausible contributor to CRC development solely due to its lack of BFT.

Escherichia coli is a Gram-negative, facultative anaerobe present in the gut microbiome of over 90% of people (Martinson and Walk, 2020). Aside from a few select pathogenic strains, E. coli has a commensal relationship with its host and maintains a myriad of critical roles in the human gut, such as vitamin K production (Suvarna et al., 1998) and protection against pathogenic microbes in the gut (Richter et al., 2018). Furthermore, E. coli uses up oxygen present in the gut, thus bolstering an anaerobic environment for obligate anaerobes such as B. fragilis to thrive (Mueller et al., 2015). The essential physiological roles and relative abundance of E. coli in the human gut make it a great candidate for investigating microbial-induced CRC.

Escherichia coli strains are divided into four phylotypes, A, B1, B2, and D (Friesen, 1988; Herzer et al., 1990; Clermont et al., 2000). Most commensal strains belong to phylotype A, while most pathogenic strains belong to phylotype B2 and D (Bingen et al., 1998; Boyd and Hartl, 1998; Picard et al., 1999; Johnson and Stell, 2000). Pathogenic strains of E. coli are common causes of urinary tract infections, sepsis, meningitis, etc. (Johnson and Russo, 2002a; Johnson and Russo, 2002b). The pathogenicity of certain E. coli strains can be attributed to the presence of different virulence factors and the ability to induce a genotoxic effect in cells. The bacterial toxins that stimulate this genotoxic effect are termed cyclomodulins (Nougayrède et al., 2005). Cyclomodulins interfere with the eukaryotic cell cycle through induction of DNA damage and genomic instability, both of which can contribute to carcinogenesis (Gagnière et al., 2016).

The B2 phylotype of E. coli possesses a genomic island called the polyketide synthetase (pks) island (Nougayrède et al., 2006; Nowrouzian and Oswald, 2012) which carries the information to produce a polyketide-peptide toxin called colibactin (Nougayrède et al., 2006; Nowrouzian and Oswald, 2012). The structure of colibactin has recently been confirmed as containing an α-dicarbonyl group and two electrophilic spirocyclopropyldihydro-2-pyrrolone groups that can undergo ring-opening to attach to DNA (Xue et al., 2019). Although the exact signaling mechanism of colibactin has yet to be uncovered, its genotoxic activity in eukaryotic cells has been examined. Like the other cyclomodulins of E. coli, colibactin translocates to the nucleus where it induces interstrand crosslinks and double-stranded DNA breaks (Nougayrède et al., 2006; Dougherty and Jobin, 2021). Colibactin has also been shown to trigger alkylation and formation of adenine adducts in vitro and in vivo through the electrophilic cyclopropane groups (Wilson et al., 2019).

It has been established that pks+ E. coli can induce a unique mutational profile, providing evidence of the causality between colibactin and mutations in IECs (Pleguezuelos-Manzano et al., 2020). Whole genome sequencing (WGS) revealed a specific mutational signature of colibactin characterized by single base substitutions (SBS-pks), specifically T>N substitutions at ATA, ATT, and TTT sites, as well as a small indel signature (ID-pks) characterized by single T deletions within T homopolymers (Pleguezuelos-Manzano et al., 2020). These same SBS-pks and ID-pks mutational signatures were found enriched in CRC-derived metastases compared to other solid cancer metastases (Pleguezuelos-Manzano et al., 2020). Although further research is necessary, taken altogether, these data may suggest that in respect to the proposed models of the relationship between the microbiota and CRC, pks+ E. coli may be the driver that initiates CRC development while ETBF further promotes it via extra-genomic mechanisms.

Colibactin has also been shown to exert its genotoxic effects on the local microbial community. Infection of pregnant mice with the genotoxic B2 phylotype of E. coli revealed microbial changes in the guts of mouse offspring at two different time points: 15 and 35 days. At day 15, the mice displayed increased abundance of Lachnospiraceae, decreased abundance of Proteobacteria, and significant alterations to the overall microbial profile despite no change in overall diversity (Tronnet et al., 2020). At day 35, the mice displayed decreased abundance of Firmicutes and related taxa, increased abundance of Deferribacteres and Tenericutes, increased overall diversity, and enrichment in microbial DNA replication and repair pathways (Tronnet et al., 2020). The findings from Tronnet et al. (2020) reveal a significant role for colibactin in influencing and modifying the surrounding microbial community. Further findings from Chen et al. (2022) revealed that the genotoxic B2 phylotype of E. coli reduced populations of Vibrio cholerae by three to five orders of magnitude, inhibited the growth of other Vibrio species, of Enterobacter aerogenes and Staphylococcus, and provided direct evidence of colibactin-mediated reduction of B. fragilis (Chen et al., 2022), thus emphasizing the role of colibactin in shaping the gut microbiota and influencing the interactions with invading pathogens.

Before discussing the synergistic involvement of B. fragilis and E. coli in the development of CRC, it is important to first discuss the evidence for involvement of each bacterium in CRC individually. B. fragilis has been extensively linked to CRC in vivo. Gnotobiotic mice developed intestinal ulcerations, edema, and inflammatory infiltration after infection with BFT-producing ETBF, but not NTBF (Nakano et al., 2006). Furthermore, ETBF was shown to cause colitis independently (Rhee et al., 2009; Goodwin et al., 2011; Wick et al., 2014; Hwang et al., 2020), worsen colitis induced by dextran sodium sulfate (DSS) (Rabizadeh et al., 2007), induce colon tumorigenesis in a Stat3/T helper 17 cell (Th17)-mediated manner (Wu et al., 2009; Chung et al., 2018; Liu et al., 2020), increase median amounts of colon tumors with increased duration of ETBF colonization (DeStefano Shields et al., 2016), and promote polyp formation and tumorigenesis in an azoxymethane (AOM)/DSS mouse model (Hwang et al., 2020).

Several human observational studies have also demonstrated a link between B. fragilis and CRC. A study by Toprak et al. (2006) first provided the evidence for this association, with bft being detected at a higher rate in stool samples of CRC patients compared to the control group (Toprak et al., 2006). A later study by Boleij et al. (2015) also detected bft at higher rates in colon mucosal samples of CRC patients compared to the control group (Boleij et al., 2015) and more so in late- versus early-stage CRC, a finding that was supported by another study (Viljoen et al., 2015). Other studies have shown an association between ETBF and CRC; ETBF was detected at higher rates in tumors compared to normal tissues and controls in 97 CRC patients from South China (Zhou et al., 2016). Also, an association was found between ETBF positivity and low-grade dysplasia, tubular adenomas, and serrated polyps, indicating a possible role of ETBF in CRC development (Purcell et al., 2017). Further studies have successfully reproduced these findings of increased detection of B. fragilis in CRC patients. Despite this, some studies have shown results contradictory to the proposition that B. fragilis expression is enhanced in CRC (Dutilh et al., 2013; Zeller et al., 2014; Lennard et al., 2016; Wirbel et al., 2019; Piciocchi et al., 2021; Shariati et al., 2021). Whether due to method design, confounding variables, etc., more research is needed to compare studies and truly uncover the causative role of B. fragilis in CRC development.

Escherichia coli has also demonstrated a role in colon tumorigenesis in vivo. Germ-free Il10^-/- mice colonized with pks+ or pks- E. coli developed severe colitis, and AOM-treated germ free Il10^-/- mice colonized with pks+ E. coli developed increased colonic tumorigenesis (Arthur et al., 2012). Interestingly, AOM-treated germ-free Il10^-/-Rag2^-/- mice, lacking functional T and B cells, did not develop colonic inflammation or tumorigenesis when colonized with pks+ E. coli, suggesting that pks+ E. coli requires inflammation to promote tumorigenesis initiated by AOM. AOM/DSS-treated mice also developed enhanced colonic tumorigenesis when colonized with pks+ E. coli (Cougnoux et al., 2014). Furthermore, multiple intestinal neoplasia (Min) mice, harboring a mutant allele in the APC gene, colonized with pks+ E. coli developed increased colonic polyps compared to uncolonized controls (Bonnet et al., 2014).

The role of E. coli has also been reported by human observational studies of CRC. In a study by Arthur et al. (2012), the abundance of pks+ E. coli was significantly increased in 35 IBD and 21 CRC patients (Arthur et al., 2012). Other studies investigating the abundance of pks+ E. coli in patients with CRC or diverticulosis found higher percentages of E. coli in CRC patients compared to patients with diverticulitis, although the percentages varied slightly between studies (Buc et al., 2013; Bonnet et al., 2014; Raisch et al., 2014). Elevated levels of pks+ E. coli markers have also been detected in stool samples of patients diagnosed with CRC (Eklöf et al., 2017). Also, higher levels of pks+ E. coli were detected in CRC patients compared to healthy controls (Prorok-Hamon et al., 2014; Iyadorai et al., 2020). Some studies suggest E. coli B2 strains heavily associate with cancerous lesions more so due to a result of tumorigenesis as opposed to a cause of tumorigenesis (Brader et al., 2008; Stritzker et al., 2010; Kocijancic et al., 2016). Again, more research and longitudinal studies are needed to uncover the causative roles of pks+ E. coli in the development of CRC.

The synergistic potential of B. fragilis and E. coli was initially described in intra-abdominal infection (Rotstein et al., 1985). Bacteroides fragilis-E. coli coinfection of intraperitoneal fibrin clots in rats resulted in increased mortality compared to either bacterium alone. Similarly, B. fragilis enhanced the severity and persistence of E. coli-induced abscesses and vice versa (Rotstein et al., 1989).

Based on the proposed models of the relationship between the gut microbiota and CRC, it is unlikely for one microorganism to individually affect the colon in the transition to an oncogenic state. Considering both B. fragilis and E. coli are plausible “Alpha-bugs,” the possibility that these two bacteria work together to contribute to the development of CRC should not be excluded. Furthermore, biofilms found in the colons of patients diagnosed with CRC are commonly composed of both ETBF and E. coli, providing further evidence of this profound interaction (Dejea et al., 2018).

Dejea et al. (2018) provided the first evidence of a possible synergistic role between B. fragilis and E. coli in the development of CRC (Dejea et al., 2018). Examination of the colonic mucosa of five patients with familial adenomatous polyposis (FAP) revealed the presence of biofilms. These FAP biofilms were found throughout the colon and consisted of mainly B. fragilis and E. coli. Moreover, the mucosal samples from 25 patients with FAP were significantly associated with ETBF and pks+ E. coli compared to the mucosal samples from healthy controls.

To investigate whether ETBF and pks+ E. coli coinfection enhances tumorigenesis compared to either bacterium alone, Dejea et al. utilized Apc^MinΔ716/+ and AOM-injected wild-type (WT) mice. Compared to infection with either bacterium alone, AOM-treated WT mice developed more tumors when coinfected with ETBF and pks+ E. coli, and coinfected Apc^MinΔ716/+ mice had shortened survival times. Interestingly, AOM-treated WT mice developed few to no tumors when infected with ETBF or pks+ E. coli alone, indicating the necessity for both bacteria to synergistically induce carcinogenesis (Dejea et al., 2018). Coinfected mice also displayed significantly increased colonic hyperplasia and micro-adenomas compared to monoinfection with either bacterium, and significantly increased inflammation compared to monoinfection with pks+ E. coli but not monoinfection with ETBF (Dejea et al., 2018).

Dejea et al. also discovered that coinfected mice had a significantly increased fecal IgA response to pks+ E. coli compared to mice infected with pks+ E. coli alone, and displayed an increase in mucosal-associated pks+ E. coli in the distal colon compared to only the colonic lumen in monoinfection with pks+ E. coli (Dejea et al., 2018). This indicates that increased tumorigenesis may be a result of a shift in the spatial arrangement of pks+ E. coli from the lumen to the colonic mucosa. ETBF alone or with pks+ E. coli significantly reduced the mucous depth like the well-known colonic mucin-degrading bacterium, Akkermansia muciniphila. However, coinfection of AOM-treated WT mice with A. muciniphila and pks+ E. coli resulted in reduced tumorigenesis, suggesting that alterations to the mucosa alone is insufficient in promoting carcinogenesis and hinting at a potential additive effect of B. fragilis in carcinogenesis. Additionally, DNA damage in IECs of coinfected mice was significantly increased compared to mice infected with each bacterium alone. Combined, these results indicate that ETBF-induced mucosal degradation is necessary for enhanced pks+ E. coli colonization, possibly allowing colibactin to further damage IECs. The data from Dejea et al. (2018) suggest a role for ETBF and pks+ E. coli interactions in promoting carcinogenesis during coinfection of the gut, and the direct evidence provided by Chen et al. (2022) for colibactin-mediated reduction of B. fragilis further emphasize the complex interactions between ETBF and E. coli and their potential influence on gut pathogenesis. Further investigation into the targeted killing of B. fragilis by colibactin-producing E. coli is needed to elucidate these interactions and their possible effects on the gut microbiome.