Human Intestinal Microbiota and Colorectal Cancer: Moving Beyond Associative Studies

Human surfaces and cavities are populated by numerous microbial communities, including bacteria, viruses, archaea, and fungi, which form a complex interactive network between themselves and the host. These inter-kingdom interactions are the result of millions of years of co-evolution, and are an intrinsic part of host health and disease balance. For example, intestinal bacterial communities have been associated with pathologic conditions such as inflammatory bowel diseases, colorectal cancer, obesity, and liver cirrhosis by 16S rRNA gene or whole genome metagenomic sequencing analysis.¹ An important concept emerging from these correlative studies is that a homeostatic equilibrium must exist among the bacterial community and the host to maintain health. According to this hypothesis, disruption of intestinal microbial equilibrium has the capacity to alter the homeostatic network, thereby eliciting deleterious host responses as observed in inflammatory bowel disease and CRC.

In support of this hypothesis, the microbial community varies between tumor and normal flanked tissue in CRC patients,^2,3 distal versus proximal tumors, and adenoma to adenocarcinoma progression.^4,5 Interestingly, differences in luminal intestinal biota may potentially serve as noninvasive CRC biomarkers when paired with either whole genome metagenomic,^6,7 or 16S rRNA sequencing analysis.⁸ Thus, a core microbial component likely drives homeostatic signaling, and the identification of these microorganisms could prove invaluable for both prevention and therapeutic intervention. A number of studies using preclinical models colonized with selected bacterial candidates (eg, Fusobacterium nucleatum, Bacteroides fragilis, adherent invasive Escherichia coli) identified from microbial genomic work, have shed light on mechanisms (eg, inflammation, genotoxicity) by which bacteria could promote intestinal carcinogenesis^9–11 or even modulate therapeutic response in preclinical models.¹² Despite this, however, the evidence for the microbial community as a whole playing a functional role in intestinal carcinogenesis is unclear. In this issue of Gastroenterology, Wong et al¹³ demonstrated the carcinogenic properties of microbial communities obtained from CRC patients, using fecal microbiota transfers into preclinical models (Figure 1).

Figure 1.

Colorectal cancer (CRC) biota promotes neoplastic lesions after oral transfer into mice. Introduction of stools from CRC patients into either axozymethane (AOM)-exposed or germ-free mice promotes the development of intestinal polyps or proliferative response respectively. Microbial composition present in the stools of mice reveals increased relative abundance of bacterial species forming networks in CRC patients. Colonic tissues from CRC biota associated mice showed increased immune cell recruitment, as well as immune and oncogenic gene expression.

In this study, the authors used 2 different approaches to dissect the impact of biota on intestinal carcinogenesis: wild-type mice treated with wide spectrum antibiotics and then exposed to the procarcinogenic compound axozymethane (AOM) and germ-free mice; both models colonized with stools from a pool of either 5 CRC patients or 5 healthy controls. In the antibiotic experiment, both the prevalence and number of colonic polyps were significantly higher in mice associated with CRC biota compared with healthy biota or control (no human biota) after 9 weeks. Interestingly, germ-free mice associated with CRC or healthy biota for up to 32 weeks failed to develop polyps, although increased intestinal proliferation was observed in the CRC biota condition when compared with healthy controls as measured by PNCA staining. Taxonomic analysis using 16S rRNA gene sequencing, performed in colonized AOM or germ-free mice, showed decreased microbial diversity and increased relative abundance of F nucleatum, Peptostreptococcus anaerobius, Peptostreptococcus stomatis, Parvimonas micra, Solobacterium moorei, and Gemella morbillorum in CRC biota compared with healthy biota.

On the host side, CRC-biota increased interleukin (IL)-17a, IL-22, and IL-23a messenger RNA accumulation, suggesting a Th17 profile, and Cxcr1 and Cxcr2, which are indicative of immune cell recruitment. Accordingly, abundance of CD4⁺ interferon-γ⁺ (Th1), and CD4⁺ IL-17⁺ (Th17) immune cells in intestinal tissues was significantly increased in mice colonized with CRC biota compared with healthy biota control mice. Finally, a number of oncogenic factors such as aurora kinase A, cell division cycle 20, and B lymphoma Mo-MLV insertion region 1 homolog (BMI1) were induced in colonic tissues of mice associated with CRC biota compared with healthy biota.

These findings have established the functional impact of the microbial community on the development of intestinal carcinogenesis, and have identified a potential microbial carcinogenic core. This important contribution reveals a series of new questions. For example, because luminal CRC biota promotes a low level of polyps in AOM-treated mice, and only proliferative signals in GF mice, it may be important to investigate host response to carcinogenic microorganisms obtained from mucosal tissues. The compositional profile of the mucosal community varies along the carcinogenic progression,⁴ and is different than that in the luminal compartment.¹⁴ Moreover, microbial organization, such as biofilm forming communities, may also be an important component of carcinogenesis, because 90% of right-sided tumors contain a biofilm positive community compared with <15% of left-sided tumors.¹⁵

Therefore, understanding the functional impact of these microorganisms in carcinogenesis would be important. Animal model selection is another important element to consider when studying host responses to mucosal or luminal bacterial communities. The authors used antibiotic-pretreated mice exposed to AOM to study host response to human biota. A confounding element in this approach is the presence of the murine biota in the housing facility environment, which likely resulted in the generation of a hybrid human–mouse biota ecosystem that may have influenced outcomes. Because a longitudinal microbial composition study was not performed in this cohort, the stability of the transplanted microbial community is unknown. Although the germ-free approach avoided this cross-contamination problem, the experimental setting did not result in neoplastic changes, only a pro-proliferative response. Interestingly, a previous report using wild-type germ-free mice exposed to AOM and the inflammatory agent dextran sodium sulfate (DSS) showed that tumor burden was higher in mice colonized with feces obtained from healthy subjects compared with CRC fecal materials.¹⁶

Moreover, intestinal tumor burden is higher in germ-free mice exposed to AOM/DSS than those given the same treatment but housed in a conventional environment.¹⁷ These findings clearly highlight how models could impact the phenotypic presentation of a host exposed to carcinogenic human biota. Using a mouse model of spontaneous intestinal neoplasia, such as Apc^min/+, housed in a germ-free environment would be useful in studying the impact of CRC-derived biota on carcinogenesis.¹⁸ Equally important is the identification of the mechanisms responsible for CRC biota-induced carcinogenesis. It is likely that the microbial consortia impacts various steps along the carcinogenesis process. For example, Wong et al¹³ showed that relative abundance of P anaerobius and F nucleatum increased in the CRC carcinogenic biota, with the former taxa activating Toll-like receptor 2 and Toll-like receptor 4 signaling to promote dysplastic lesions in AOM-exposed wild-type mice,¹⁹ and the latter altering CRC chemotherapeutic response.²⁰

Finally, limited understanding is available regarding the mechanisms by which microbial composition differs between CRC patients and healthy subjects. Environmental and dietary factors in conjunction with host genetics are likely contributing to microbial function. In this issue of Gastroenterology, Liu et al²¹ showed that conditional deletion of both polycomb group (PcG) genes BMI1 and MEL18 in intestinal epithelial cells significantly decreased AOM/DSS induced colitis-associated colon cancer compared with wild-type mice. Importantly, Wong et al¹³ showed that BMI1 expression is induced following fecal transplant of CRC biota in mice. Liu et al²¹ identified REG3B, a protein that exerts bactericidal activity, as a target of these PcG genes. Administration of recombinant REG3B significantly reduced polyp formation in both the early and late stages of AOM/DSS exposure compared with control mice. Furthermore, the authors showed that REG3B interfered with STAT3 activation, thereby providing a potential antitumorigenesis mechanism. However, microbiota balance and function may also be impacted as a result of REG3B bactericidal activity, and this possibility should be investigated. It is also important to determine whether CRC biota-induced carcinogenesis is attenuated in mice deficient for BMI1 and MEL18 in intestinal epithelial cells.

In summary, the studies by Wong et al¹³ and Liu et al²¹ highlight the complex interaction between host genetics, microbiota and intestinal neoplasia (Figure 1). Clearly, more information is needed about mechanisms through which the microbial community both promotes and prevents carcinogenesis. Understanding the events leading to a disrupted microbial-host interaction and development of intestinal carcinogenesis will provide novel insight into this pathology, and potentially lead to new therapeutic modalities targeting the microbiota.