Hiding in Plain Sight: A Novel Microbial Driver for Colorectal Cancer?

Emerging evidence suggests that the gut microbiome, a specialized ecological niche in our gastrointestinal tract spanning commensal bacteria, viruses, fungi, and protozoa, plays an integral role in determining cancer growth and response to therapy (1). Although not limited to gastrointestinal malignancies, this “oncomicrobiotic” relationship has been explored extensively in colorectal cancer due to the natural proximity of the commensal microbiota to the colonic mucosa. Several important studies over the past two decades have led to the recognition of microbial drivers of colorectal cancer such as Fusobacterium nucleatum, enterotoxigenic Bacteroides fragilis (ETBF), and pks⁺ Escherichia coli (2). The tumor­igenic activities of these microbes occur in the background of a “dysbiotic” gut microbiome (1), which refers to the state of altered composition and/or function from the normal homeostatic levels. Despite a plethora of evidence characterizing the dysbiotic states associated with various malignancies, including colorectal cancer, it has not been possible to identify an overarching phenotype distinguishing the “healthy” from the “dysbiotic” gut microbiome, leaving open the possibility of yet unidentified microbial drivers of the tumorigenic process.

In the current issue of Cancer Discovery, Drewes and colleagues (3) reported some unexpected observations while attempting in vivo characterization of bacterial pathogens associated with human colorectal cancer. To investigate possible microbial drivers, the authors focused on colorectal cancer–associated mucosal biofilms, which they have previously shown to be a characteristic feature of microbial organization in colonic adenomas and adenocarcinomas (4). Mucosal biofilms harbor dense bacterial collections in contact with colonic epithelial cells, where they can influence cell proliferation and induce inflammatory reactions. Using oral gavage of mucosal slurries containing the bacterial component of biofilms obtained from patients with colorectal cancer, the authors created humanized gut microbiome AVATAR mice by repopulating the gut microbiome of germ-free (GF) or standard pathogen–free (SPF) mice. The recipient mice harbored mutations in the Apc gene (GF or SPF Apc^Min/+ mice), predisposing them to the spontaneous development of small and large intestinal adenomas. The authors observed a significant interpatient heterogeneity in the induction of tumor­i­genic response, with slurries from some patients inducing robust distal colonic tumors, while others failed to do so. Using rigorous culturomics and metagenomics methodologies, the authors were able to generate a 30-bacteria isolate from the stool of AVATAR mice that showed increased adenoma formation, and this isolate was able to capture the tumor-promoting effects of the original patient slurry in vivo (3). Interestingly, 16s rRNA sequencing indicated that none of the known microbial drivers of colorectal cancer were enriched in the AVATAR mice, but Clostridioides difficile could be consistently isolated from all of them. Moreover, toxigenic C. difficile was enriched in the protumorigenic patient slurries, although it was undetectable in nontumorigenic slurries, leading the authors to hypothesize that C. difficile could be driving the tumorigenic response.

C. difficile is a common human pathogen that causes colitis in an opportunistic manner, such as in patients with recent antibiotic exposure or those undergoing chemotherapy (5). The spectrum of clinical presentations includes asymptomatic carriers, diarrheal illness, or fulminant colitis, which can be a medical emergency. An estimated half a million cases of C. difficile infections (CDI) occur every year in the United States (5). However, there is scant literature indicating the association of C. difficile with colorectal cancer. Therefore, it was quite surprising that C. difficile was the only strain showing up consistently in mice with increased tumor formation (3). The authors further established the tumorigenic potential of C. difficile by isolating a toxigenic strain from the AVATAR mice that was able to significantly increase tumor formation upon inoculation into a fresh set of mice. Importantly, the tumor-promoting effects were evident in diverse microbiological environments, including a monocolonization experiment, indicating that C. difficile may not be dependent upon specific microbial partners to demonstrate its tumorigenic effects. When vancomycin treatment was initiated 1 week after inoculation for bacterial clearance (3), it led to abrogation of the tumor-promoting effects of C. difficile, making a “hit-and-run” phenomenon (6), in which a single insult is sufficient to induce and sustain the carcinogenic cascade, unlikely. Rather, the number of observed colonic microadenomas increased at later time points after inoculation, suggesting that chronic colonization with C. difficile might be an important determinant of tumor growth.

Microbes can affect cancer growth through direct effects on cancerous cells or through the breakdown of cancer immunosurveillance (7). Previous studies in colorectal cancer have uncovered multiple mechanisms of a microbial-driven direct effect on cancer cells, such as toxin-induced double-stranded DNA breaks in colonic epithelial cells by pks⁺ E. coli and activation of Wnt and NF-κB signaling, ROS generation, and inflammatory mediator secretion by ETBF. F. nucleatum on the other hand affects tumor growth independent of toxin production instead of adhering to epithelial cells and immune cells via surface receptors such as FadA and inducing oncogenic signaling in epithelial cells while inhibiting cytotoxic T-cell function and impairing the efficacy of immune-checkpoint blockade immunotherapy (6). During colorectal cancer progression, gut microbiota can also affect the metabolic milieu in the colonic lumen by decreasing the levels of short-chain fatty acids or upregulating polyamine metabolites in biofilms to aid cell proliferation and inflammation in the colonic mucosa (6). Using isogenic mutants for bacterial toxins, Drewes and colleagues (3) found that the protumorigenic phenotype of their C. difficile isolate was dependent on the secretion of toxin B (TcdB), although toxin A was found to be dispensable for these effects. Interestingly, biofilm formation, invasion, and tissue colonization were similar between the tcdB⁺ and tcdB⁻ strains, indicating that spatial localization and interaction with other members of the gut microbiome might be less significant compared with toxin production as the dominant driver of tumorigenesis in this model. Although further mechanistic studies exploring the role of TcdB in colorectal cancer pathogenesis are required, preliminary data from single-cell RNA sequencing (scRNA-seq) analysis of the colonic epithelial compartment showed the upregulation of known oncogenic pathway genes (8) in progenitor cells (increased Wnt/Myc, Klf5, and inhibition of Pten) as well as differentiated colonocytes (Wnt/Myc and inflammatory genes such as Stat and Irf ). The immune compartment also showed significant alterations with infiltration of CD11b⁺ myeloid cells and multiple IL17-secreting cell types in the colonic lamina propria (3).

Overall, this study raises an interesting possibility that C. difficile might be a novel microbial driver of colorectal cancer (Fig. 1). However, many questions still need to be addressed to fully understand the scope of this finding. The relationship between C. difficile and colorectal cancer has been borne out in epidemiologic studies only sporadically despite there being a plethora of such studies. One possibility could be the low abundance of C. difficile relative to some of the more well-known colorectal cancer drivers in stool or tumor samples. In the present study, the relative abundance of C. difficile in stool and biofilms was quite low (∼2%) despite the strong protumorigenic effect, indicating that the effect is strong enough even at low abundance levels. However, spatial localization, host responses, immune milieu, and the presence of other microbial colonies might modify these effects in humans—factors that need to be evaluated in future studies. The exact mechanisms underlying the tumor-promoting effects of TcdB need to be explored as well. Curiously, the authors observed upregulation of the Wnt/Myc pathway in the colonic progenitor cells as well as increased stemness and proliferation-related signatures in their scRNA-seq data set (3). Previous studies have shown that the C. difficile toxin may actually inhibit the Wnt pathway in colonic epithelial cells (9). Moreover, C. difficile TcdB is a rho GTPase inhibitor and can lead to colonic epithelial cell apoptosis (10). Future efforts will need to reconcile these conflicting observations and shed some light on the intricacies of the effects of CDI on colonic progenitor cells.

Given the burden of CDI and the increasing antibiotic resistance seen in the past decade, the protumorigenic potential of this bacterium demonstrated by Drewes and colleagues (3) may hold immense significance for clinical management. Recurrent disease and chronic colonization are seen in a significant number of patients with CDI, and those patients might need to be monitored closely for colorectal cancer development. CDI is itself accompanied by dysbiosis; therefore, it will be imperative to assess the polymicrobial relationships operating in that dysbiotic environment to dissect the actual impact of C. difficile on colorectal cancer incidence and progression. Another interesting question is what implications these findings hold for asymptomatic carriers, who presumably do not have a component of active luminal inflammation. If C. difficile indeed drives colorectal cancer in a subset of patients, antibiotic administration might become a viable anticancer therapy against colorectal cancer.