Nature - USA (2020-09-24)

510 | Nature | Vol 585 | 24 September 2020

Review

for CRC^12 , as are heritable genetic defects such as familial adenoma-
tous polyposis and Lynch syndrome^13. Beyond such inherited condi-
tions, mathematical modelling attributes 70–90% of disease risk to
environmental factors—most notably diets that are low in fibre and
high in red meat^2 ,^14. Pathogenic infections are also well-established
environmental triggers of many cancers, with a link between H. pylori
and gastric cancer first identified in 1991^15 ,^16. Although evidence for a
single disease-triggering pathogen in CRC is lacking, these observa-
tions have sparked a growing interest in the role of the microbiota,
in which cancer-initiating microorganisms might prosper owing to
ecological changes.
Irrespective of what underlies the initiation of CRC, it typically
involves the disruption of homeostatic immune and microbiota-derived
signals, stimulating responses that have evolved to permit epithelial
restitution to excess^17 ,^18. Whether such responses initiate tumorigen-
esis depends on a succession of genomic alterations within intestinal
stem cells, often in the form of mutations in genes such as APC, KRAS
or PIK3CA^19. Notably, the incidence of such genetic alterations can be
increased by a maladaptation of the host–microbiota interface (Fig.  1 ),
which can contribute to the hallmarks of cancer—including sustained
epithelial-cell proliferation, resistance to cell death, invasion and
immune evasion^20.
Like the delicate balance in protective versus pathogenic epithe-
lial responses, intestinal inflammation is considered a ‘double-edged

sword’ in CRC. The numbers of in situ T cells correlate with beneficial

CRC outcome^21 , and harnessing the immune system for antigen-specific

elimination remains the goal of many therapeutic strategies. How-

ever, tumours often escape such immune-mediated destruction by

immunoediting tumour antigens, rendering them undetectable^22.

Although innate immune cells such as neutrophils and macrophages

can aid tumour clearance, under certain circumstances they release

reactive oxygen species, which can potentially initiate a carcinogenic

cascade by damaging the genomic integrity of IECs^23. Tumour cells

thereby create a vicious cycle in which numerous cells in the micro-

environment—including resident fibroblasts—can be reprogrammed

to produce additional growth factors, cytokines and pro-angiogenic

factors that sustain unrestrained proliferation and invasion^17 ,^18 ,^24 ,^25.

Dysbiosis in CRC

Initial evidence for host–microbiota interactions in CRC emerged in

1975, when it was shown that the carcinogen dimethylhydrazine trig-

gered significantly less colonic tumorigenesis in germ-free rats than in

those with intestinal microbiota^26. More functionally, mice transplanted

with a faecal microbiota from patients with CRC developed more intes-

tinal polyps than those receiving microbiota from healthy controls^27.

Thanks to developments in microbiome profiling—including 16S

rRNA and shotgun metagenomics—it is indisputable that individuals

with CRC have a different taxonomic composition relative to healthy

controls, referred to as ‘dysbiosis’^28. These metagenomic and metataxo-

nomic studies show that patients with CRC have a greater overall taxo-

nomic diversity of faecal species and an outgrowth of certain species, the

nature of which has been described extensively elsewhere^29 –^33. In sum-

mary, a higher relative abundance of putatively pro-carcinogenic micro-

bial members—including Fusobacterium nucleatum, Escherichia coli,

Bacteroides fragilis, Enterococcus faecalis, Streptococcus gallolyticus

and Peptostreptococcus spp.—has been detected in CRC tumour tissue,

whereas so-called protective genera—including Roseburia, Clostridium,

Faecalibacterium and Bifidobacterium—are reduced^29 –^34. Some of

these differences, most consistently levels of Fusobacterium, have

been correlated with clinical outcomes and chemosensitivity, and

therefore such bacteria have potential as biomarkers^33. Recent analysis

of biopsies from 100 patients with Lynch syndrome highlighted early

microbial changes in colonic neoplasia, including a shift in flagellin

contributors^34 ,^35.

However, determining whether such changes are a cause or an effect

of cancer, and attributing the initiation and/or progression of CRC to

certain so-called ‘oncogenic’ microorganisms, remains challenging—in

part due to considerable inter-individual differences within the micro-

biota^3. Nonetheless, a recently published CRC signature identified an

enrichment of 29 species across 8 geographic locations, taking the

field closer to defining a ‘CRC microbiome’^34. Important mechanistic

questions remain as to whether the observed pathophysiology is due

to the activity of the collective ‘oncogenic’ community, for example

via metabolite production, or to single strains with certain genotoxic

and/or yet to be discovered characteristics.

Genotoxicity induced by CRC-associated bacteria

For a single microorganism or community to be considered oncogenic,

they must elicit carcinogenic effects, such as causing DNA damage.

Pathogens associated with cancer—including H. pylori—are known to

trigger cancerous mutations^36 , and a number of recent studies have

reported the genotoxicity of species associated with CRC, including

colibactin-producing polyketide synthase (pks)+ E. coli^37 –^39 , entero-

toxigenic B. fragilis (ETBF)^40 , E. faecalis^41 and cytolethal distending

toxin-producing Campylobacter jejuni^42. In particular, pks+ E. coli

can induce double-strand breaks, aneuploidy and improper cellular

division^37 , an effect driven by the mutagen colibactin^43 ,^44. This was

Microbiota

Host

Barrier

function

Barrier

maintenance

Mesenchyme Immune cells

Niche

modication

Immune

Tissuerepair activation

a b

c

d

e

f

Fig. 1 | A schematic of the host–microbiota interactions in health and in
colorectal cancer. The host–microbiota interface comprises a continuously
evolving microbial ecosystem neighbouring host immune and mesenchymal
cells, which underlie a single-layer epithelial-cell barrier. Under homeostatic
conditions, bi-directional communication between host cells and the
microbiota facilitates critical symbiotic functions and maintains the structural
integrity of the intestine (arrows). For example, microorganisms and
metabolites can trigger responses that maintain barrier function and tissue
repair. This response is simultaneously affected by the genetic makeup of host
cells, which in turn shapes microbiota niche formation. However, such
interactions can become maladapted in CRC, in which the microbiota
composition is different (a), including an increase in bacteria with genotoxic
abilities (b). These changes are accompanied by deterioration of the epithelial
cell barrier (c); this enables an inf lux of microorganisms (d) that can trigger
inf lammation (e) and promote established hallmarks of cancer, such as DNA
damage (f).