Colorectal cancer: Looking for answers in the microbiota

Summary

At a simplistic level, colorectal cancer (CRC) arises from mutations in various proto-oncogenes and tumor suppressor genes. Aside from genetically inherited factors, environmental, lifestyle and dietary habits have all been identified as risk agents promoting mutational events leading to the development of CRC. In this review, I present evidence that the intestinal endogenous bacterial community represents a risk factor for the development of CRC.

Introduction

Colorectal cancer (CRC) is the third leading cause of cancer incidence with an estimated 1,000,000 new cases per year worldwide and a significant mortality rate of nearly 33% in the developed world. Although integration of genomic, proteomic and metabolomic approaches have provided significant advances in cancer prevention, detection, and therapy translating into important improvements in patient survival and quality of life, CRC still represents a important health and socioeconomical burden in the United states. A low proportion of CRC cases are inherited [e.g., Lynch syndrome, familial adenomatous polyposis (FAP), and MUTYH-associated polyposis], but the vast majority of cases (~90%) are attributed to sporadic mutations in various genes controlling cellular proliferation, survival, differentiation, and migration. In addition to host genetics, environmental factors such as dietary habits and lifestyle influence CRC development. Understanding the contribution of these various environmental factors could provide new means to prevent development of CRC or at least to help manage the pathology. It is interesting to note that the most ubiquitous environmental factor, the luminal microbiota, has only been the subject of sparse research. A bacterial community ranging from 10¹³to 10¹⁴ organisms that contributes a collective genome evaluated at 3 × 10⁶ genes resides in the colon. This microbial ecosystem lives in relatively close proximity to the intestinal epithelium and contributes essential functions involved in the maintenance of host homeostasis. Recent evidence based on metagenomics and experimental models has highlighted the potential contribution of the microbiome in modulating CRC development. Here, I discuss the interplay between microbes and CRC development and highlight key findings supporting the role of symbionts in this pathology. Because chronic inflammation represents a significant risk factor for CRC, the literature will be discussed predominantly in light of the experimental model of colitis-associated cancer (CAC).

Microbial molecular patterns and pattern recognition receptors

The contribution of bacteria in the development of CRC could be investigated indirectly by using a proxy approach focusing on the role of receptors responsible for sensing microorganisms and their associated ligands. In the Eukaria domain, a vast repertoire of receptors termed Pattern Recognition Receptors (PRR) recognize specific conserved microbial patterns (bacterial cell walls, nucleic acids, motility apparatuses, etc.). The most studied PRRs in relation to CRC belong to the group of nucleotide-binding domain leucine-rich repeat proteins (NLR; also known as Nod-like receptors), and Toll-like receptors (TLR). Following microbial sensing, these PRRs engage a complex set of signaling proteins that shape the host immune and inflammatory response. The goal of this innate/immune response is to protect the host from any deleterious effects caused by microorganisms. Many of these sensors are expressed on intestinal epithelial cells (IEC) and on various mucosal immune cells, two important cellular compartments implicated in CRC development.

In a chemical model where tumorigenesis is initiated with the colon-specific carcinogen azoxymethane (AOM) and colitis with dextran sodium sulfate (DSS), investigators have determined the impact of various PRRs in CAC. For instance, AOM/DSS-induced colonic tumors increased in Tlr2^-/- mice, whereas they decreased in Tlr4^-/- mice compared to control wild-type mice. The differential outcome on tumor development may be related to the different microbial pattern detected by these PRRs (LPS vs. peptidoglycan). The mechanism by which TLR4 favors CRC development is still unclear but recent findings generated in Apc^Min/+ (multiple intestinal neoplasia allele) mice suggest that defective barrier function favors LPS translocation and activation of mucosal immune cell-derived cytokines such as IL-23, IL-6, and IL-17A (1). This unique inflammatory milieu favors progression of cancer-initiated cells and development of adenocarcinomas.

With the exception of TLR3, most TLRs including TLR2 and TLR4 signal through the adaptor protein myeloid differentiation factor 88 (MyD88). Deletion of Myd88 diminishes development of spontaneous colorectal cancer in Apc^Min/+ mice, reducing both intestinal tumor size and multiplicity. In contrast, AOM/DSS-induced colonic tumors increased in Myd88^-/- mice compared to WT mice, an effect linked to an impaired wound healing response due to defective IL-18/MyD88 signaling. These data suggest that host recognition of microbial entities through TLRs differentially impact CRC development. Since functional polymorphisms in the human TLR2 and TLR4 genes have been shown to be associated with CRC risk (2,3), these findings may have important translational implications.

PRRs of the NLR family including NOD1, NOD2, NLRP3, NLRP6, and NLRP12 have also been investigated for their involvement in CRC. Importantly, polymorphisms in some of these PRR genes such as NLRP3 and NOD2 have been linked to development of human IBD or CRC, respectively. NOD2, an intracellular NLR detecting both gram positive and gram negative bacteria appears essential for a proper healthy host/bacterial interaction. Indeed, Nod2-deficient mice have a dysbiotic microbiota, a microbial state characterized by an unbalanced ratio of bacteria compared to eubiosis, the default normal microbial state found in healthy host. Interestingly, Nod2-deficient mice are more susceptible to AOM/DSS-induced CRC (4). The intracellular NLR NOD1, which detect peptidoglycan, has also been shown to play a protective role against the development of CRC. In AOM/DSS or spontaneous CRC (Apc^Min/+) models, tumor development was augmented in Nod1^-/- mice compared with WT mice, a phenomenon attenuated with broad-spectrum antibiotic treatment. Although both NOD1 and NOD2 are protective against development of experimental CRC, it is not clear whether these NLRs operate through a unifying mechanisms. The intracellular sensors NLRP3 and NLRP6 also appear to be protective against colitis-associated colorectal cancer, a role attributed to the production of IL-18. A recent report demonstrated that NLRP3/6-induced IL-18 expression was essential in regulating expression of IL-22 binding protein (IL-22BP) , a critical negative regulator of the pro-proliferative cytokine IL-22 (5). Finally, NLRP12 signaling from the hematopoietic compartment protects against development of colitis-associated CRC. From these findings, it appears that PRRs are important modulators of CRC development, with intracellular NLRs being mostly protective whereas TLRs, with the exception of TLR2, may be pro-carcinogenic. The mechanism responsible for the protective function is not entirely clear but production of IL-18, a protein controling cellular proliferative response, appears to be a central component of the PRR-mediated effect.

Although these findings have greatly contributed to our understanding of the role of innate sensors in development of CRC, they provide minimal insight into the impact of microorganisms. Indeed, some of the PRRs discussed above have no identified ligands (NLRP6, NLRP12) or alternatively detect a wide range of patterns which make an association with specific microorganism a daunting task. For example, TLR2 and TLR4 respond to non-microbial ligands such as heat shock proteins and histones, endogenous components often released in the environment in response to stress and cellular injury as seen in inflammatory conditions. Therefore, assessing the contribution of microorganisms to CRC development using mice with various PRRs gene deletion is fraught with limitation.

The microbiome

For decades, the field of host/bacterial interaction in the gastrointestinal (GI) tract has mostly focused on the study of enteropathogenic bacteria. Indeed, tremendous efforts have been directed at understanding the molecular mechanisms by which certain pathogens cause GI illnesses. Although of prime public health importance, enteropathogenic bacteria represent a small proportion (3%) of the total microbial community (~7,000 strains). Moreover, most of these microorganisms are not natural residents of the GI tract and the transient nature of their passage in the intestine rarely impacts long-term health status once eliminated. With more than 100 trillion microorganisms in the GI tract, most of them found in the colon, researchers have begun to question the potential implication of these symbionts in health and disease. This field of research is relatively new and stems from a worldwide effort to catalogue the microbiome at different human biological sites, including the GI tract. Using next-generation sequencing and toxonomic studies based on ribosomal 16S bacterial genes, a clearer picture has emerged regarding the identity of the microorganisms inhabiting our intestine. At the phylum level, 95% of the GI tract is dominated by Firmicutes (~75%) and Bacteroidetes (~20%) followed by Proteobacteria and Actinobacteria. Although a healthy core microbiome has not been clearly identified, studies using cases and controls have provided new insights into changes in microbial composition, a phenomenon refered to as dysbiosis. Applying microbiome analysis of tissues and fecal materials, researchers have identified various microbial groups associated with CRC. Although these studies have not identified a consensus group of bacteria associated with CRC, these investigation have systemically showed differences between the microbiome of patients and the one present in healthy subjects. For example, the stool of CRC patients harbors a higher population of bacteria belonging to the group Bacteroides-Prevotella compared with normal controls (6). Another study showed that the genera Enterococcus, Escherichia/Shigella, Klebsiella, Streptococcus, and Peptostreptococcus were significantly more prevalent in the luminal compartment of CRC patients than controls, whereas the family Lachnospiraceae, which contains butyrate-producing bacteria thought to exert intestinal protection function were less abundant (7). At the intestinal mucosal surface, an increased abundance of Firmicutes, Bacteroidetes, and Proteobacteria was observed in patients with adenoma compared with non-adenoma subjects (8). Using resected tissues from adenocarcinoma patients and adjacent non-malignant sites, another group showed expansion of the phyla Bacteroidetes and reduction of Firmicutes in tumors compared with controls (9). At the genera level, increases in Coriobacteridae, Roseburia, Fusobacterium, and Faecalibacterium were observed in this study, whereas the family Enterobacteriaceae decreased. Interestingly, an expansion of Fusobacterium in rectal swab samples from CRC patients compared with healthy controls was also reported by another group (10). In addition, using whole-genome sequencing and RNA-sequencing approaches, two independent groups have demonstrated that Fusobacterium is highly prevalent in colonic tissues of CRC patients compared with normal controls (11,12). As the presence of Fusobacterium nucleatum correlates with development of inflammatory bowel diseases (IBD), this bacterium could provide a potential link between the development of IBD and CRC. Unfortunately, despite the extensive amount of effort dedicated at surveying the microbiomes of CRC and normal subjects, these findings remain essentially correlative and caution is warranted about interpreting these findings. In the absence of longitudinal analysis and functional experiments using experimental models, microbiome changes in CRC patients could simply represent a consequence of environmental changes caused by neoplastic lesions. In addition, it is not clear which environmental factors contribute to the expansion or contraction of these various microbial entities.

To circumvent the limitations of human microbiome surveys, novel studies using germ-free mice and gnotobiotic technology have been initiated and the functional impact of certain bacteria on the development of CRC has been reported. The Apc^Min/+ mouse represents a popular model of human familial adenomatous polyposis and develops tens to hundreds of intestinal adenomas when maintained in regular housing conditions. Interestingly, Apc^Min/+ mice raised in germ-free conditions show a reduction in tumor load in both the small intestine and colon compared with mice housed in specific pathogen free (SPF) conditions (2). It would be interesting to investigate the functional impact of bacteria associated with human CRC (e.g Fusobacterium, E.coli, etc) using germ-free Apc^Min/+ mice. The enterotoxigenic Bacteroides fragilis (ETBF) promotes the development of CRC in Apc^Min/+ mice, an effect caused by the production of B. fragilis toxin. The prevalence of ETBF appears higher in the stool of patients with CRC compared with controls, suggesting a potential role of this bacterium in CRC pathology. Patients with IBD or CRC have an increased number of adherent-invasive E.coli on their mucosal surface compared with normal controls. Interestingly, a recent study demonstrated that luminal proteobacteria, especially E.coli, expand during the development of experimental colitis in Il10^-/- mice (13). Using a murine adherent-invasive E.coli strain (NC101), the authors demonstrated that this bacterium promotes development of CRC through the action of a genotoxic island named polyketide synthases (pks). Another important observation from this study is that although intestinal inflammation developed in mono-colonized Enterococcus feacalis Il10^-/- mice, these mice failed to develop tumors as observed in E.coli-associated Il10^-/- mice. This finding showed in an experimental model that 1) chronic intestinal inflammation alone, an established risk factor for CRC, is not sufficient to promote tumors and 2) microbial composition is an important environmental factor in the pathology of CRC.

Future directions

Although infectious microorganisms such as Streptococcus bovis, Chlamydia trachomatis, and Helicobacter pylori have been associated with certain types of cancer, the potential implication of symbionts, or non-pathogenic bacteria, in the development of tumorigenesis is a relatively new concept. Human microbiome studies performed in CRC patients at different stages have provided novel insights into microbial communities living at various ecological sites. These studies have offered potential links between microbial entities and the development of cancer. In addition, proof-of-principle studies in preclinical models have shown that bacteria impact the development of CRC. Although the findings are sparse, novel mechanisms have been proposed to explain the pro-carcinogenic effect of some of these microbes.

Although it is still in its infancy, cancer microbiome investigation opens a new era for CRC research and a new set of questions has been generated from these investigations. For example, is there a specific group of microorganisms responsible for cancer initiation, progression, and even metastasis? How do microorganisms influence cancer development? Could microbial-derived metabolities or proteins affect cancer susceptibility and progression? How do environmental factors such as stress, diet, and lifestyle modulate microbial activities to influence CRC development? Could biomarkers be generated from microbiome research? Is efficacy of current therapeutic modalities (radiation, immunotherapy, surgery) influenced by the microbiome? Could the microbiome be manipulated (e.g., by pre- and probiotics or bacteriophages) for therapeutic purposes? Addressing these questions would undeniably contribute to our understanding of the interplay between the microbiome and cancer development. Nevertheless, from the current state of knowledge, one could hypothezise that the microbiota is subjected to internal/external pressure (e.g: diet, stress, inflammation, etc) affecting cancer development/progression through production of microbial-derived carcinogenic products (e.g: Colibactin, H₂S etc) (Fig.1). This bacterial contribution occurs in the background of a host responses to these environmental cues, resulting in the production of different immune and stromal cell-mediated pro-carcinogenic mediators such as cytokines (Il-17A, IL-23, IL-6, etc), radical oxygen species and radical nitrogen species.

Figure 1.

Environmental factors such as diet, inflammation, stress influence microbial composition and cause microbial dysbiosis. Expansion of cancer promoting bacteria with procarcinognic activities (toxin, metabolites) promotes epithelial cell DNA damage, which in conjunction with host responses (cytokines, DNA damaging products) from the underlining tissues, foster dyslastic events and cancer progression.

Although still a work in progress, existing data linking microbes to CRC provides a strong rationale to pursue investigation into the role of the microbiota in cancer development. However, the microbiome encompasses numerous microscopic entities beside bacteria, including viruses and fungi. It is reasonable to speculate that components of this complex ecosystem interact with each other, their environment, and the host to impact intestinal homeostasis, which poses an enormous challenge to researchers. Metagenomic studies on virus and fungi have already shown their potential effect on gastrointestinal health (14,15). Nevertheless, it has become clear that taxonomy-based investigations using an ribosomal 16S approach will not be sufficient to provide a clear understanding of the bacterial contribution to cancer. Microorganisms are closely attuned to their environment and respond to various changes (diet, stress, inflammation, etc.) by inducing complex transcriptional responses that lead to the production of various molecules (toxin, metabolites, enzymes, etc.) and future studies would need to incorporate various approaches where the bacterial meta-transcriptome, metagenome, and metabolome are investigated in conjunction with the host response. This comprehensive and integrative approach would capture the complex interaction between the microbiota and the host and could reveal novel pathways that contribute to the development of CRC. Mechanistic studies using experimental models and microbial genetic manipulation would be required to prove functional relevance. Together, such studies have the potential to reveal new paradigms that could significantly impact our understanding of CRC.