2.1 Four-kingdom microbiome in healthy subjects

The human microbiome was composed of bacteria, fungi, viruses, protozoa, and related genes and metabolites.^{17, 18} Of those, the bacteria, which is dominated by three major phyla: Firmicutes, Bacteroidetes, and Actinobacteria, has been extensively studied.¹⁹ In contrast, fungi only account for approximately 0.2% of microorganisms in the human gut, predominated by three phyla, including Ascomycota, Basidiomycota, and Chytridiomycota.²⁰ Archaea is the single-cell prokaryote with distinct cellular characteristics from either bacterium or eukaryotes, such as the lack of peptidoglycan and d-glycerol esters or fatty acids.²¹ The major components of human gut archaeome are species from Methanobacteriales and Methanomassiliicoccales.²² The human gut virome includes double-stranded DNA viruses, single-stranded DNA viruses, as well as RNA viruses.²³ Particularly, intestinal bacteriophages account for nearly 90% of human virome.²⁴

2.2 Bacterial microbiota dysbiosis in CRC

Disruption of gut bacterial microbiome homeostasis induces chronic inflammation of the GI tract and production of carcinogenic metabolites, which leads to colon damage.²⁵ In studies comparing the composition of gut microbiota between CRC patients and healthy subjects, the abundance of Bacteroides fragilis, Fusobacterium nucleatum (Fn), Parvimonas micra, Porphyromonas asaccharolytica, Prevotella intermedia, Alistipes finegoldii, and Thermanaerovibrio acidaminovorans in CRC patients were increased, which can also be used as diagnostic markers.^{26, 27} However, the abundance of Roseburia, Clostridium, Faecalibacterium, and Bifidobacterium were significantly reduced.^{28, 29} It was demonstrated that B. fragilis downregulated miR-149-3p, which has been linked to intestinal inflammation and malignancy.²⁵ FadA adhesin in Fn can help to promote CRC cells proliferation through E-cadherin/β-catenin signaling.³⁰ In addition, Fn can also impact tumor microenvironment recruiting tumor-infiltrating myeloid cells and potentiated CRC metastasis by modulating miR1322/CCL20 axis, miR-1246/92b-3p/27a-3p, CXCL16, and KRT7-AS/KRT7.^31-33 Besides, several bacterial species known as probiotics were also absent in CRC patients. The abundance of Akkermansia muciniphila was reduced in dextran sulfate sodium (DSS) treat mice, whereas Lactobacillus reuteri and Clostridium butyricum were reduced in CRC patients.^34-36 Detailed information regarding CRC-associated bacterial microbiome alteration reported by previous studies is summarized (Table 1). Taken together, these studies suggested that gut microbiota was significantly altered in CRC compared with healthy control, and certain species (like B. fragilis and F. nucleatum) were commonly changed across studies.

2.3 Non-bacterial microbiota dysbiosis in CRC

Despite the high abundance of bacteria species in the gut microbiome, the relationship between nonbacterial microbiota dysbiosis and the progression of CRC should not be underestimated. Characterization of mycobiome, archaeome, and virome of CRC patients was also performed in the past few years. In one study, researchers characterized the gut mycobiota in 184 patients with CRC, 197 patients with adenoma, and 204 healthy subjects in Hong Kong and found that the Ascomycota/Basidiomycota ratio was decreased with the enrichment of Malasseziomycetes in CRC patients. Depletion of Saccharomycetes and Pneumocystidomycetes were also found in CRC patients compared with healthy control.⁶¹ Multi-cohort analysis of 1368 samples from 8 geographically distinct cohorts identified an increased abundance of 93 fungal species, including Candida pseudohaemulonis, Aspergillus ochraceoroseus, Aspergillus rambellii, and Malassezia globosa, whereas another 15 fungal species, including Aspergillus niger, Macrophomina phaseolina, Talaromyces islandicus, and Sistotremastrum niveocremeum, were decreased in CRC patients.⁶⁷ In another study, the molecular evidence is mostly associated with inflammatory bowel disease, whereas CRC-related evidence was limited.⁶⁸ Multi-cohort metagenomics suggested that A. rambellii was enriched in CRC patient and promoted tumor growth and cell proliferation.⁶⁹ The abundance of Candida albicans was reported to increase in mice lacking the C-type lectin Dectin-3 upon chemical induction and may promote epithelial cell proliferation dependent on Dectin-1 and Wnt signaling.^{70, 71} Gut fungal commensals promote inflammasome activation and depletion of CARD9 (caspase recruitment domain 9), or kinase spleen tyrosine kinase increases the susceptibility of mice to CRC.⁷² Taken together, alterations in mycobiome are closely associated with CRC progression.⁷³

Regarding the gut virome, one metagenomics analysis, including 74 CRC patients and 92 healthy controls, in Hong Kong showed that the bacteriophage diversity was increased in CRC patients; notably, 22 viral genera can be used to distinguish between patients and controls.⁶⁴ Multi-cohort analysis revealed that the abundance of 115 viral species is increased, whereas 18 viral species were decreased in CRC patients.⁶⁷ By constructing the random forest model to characterize the gut virome of 30 CRC patients, 30 adenoma patients, and 30 healthy controls, Hannigan et al. identified that temperate bacteriophage is dominant in CRC patients.⁶⁵ In addition, by comparing primary tumor, metastatic tumor, and normal tissues in 12 cases, 5 enriched phages, including Enterobacteriaceae, Bacillus, Proteus, and Streptococcus phages, in CRC patients were identified.⁷⁴ Bacteriophage can worsen colitis through TLR9 and increased IFN-γ.⁷⁵ Animal studies provided evidence from another perspective. With AOM–DSS induction of CRC in mice, virome alteration was observed, and Brunovirus and Hpunavirus were positively related to tumor growth, whereas Lubbockvirus showed negative association.⁷⁶

At last, the diverse archaeal community was detected in multiple human body sites and found associated with CRC as well.^{21, 77} Investigation of archaeome in fecal samples of CRC patients revealed the increased abundance of halophilic components, including Halorubrum, Natrinema, and Halococcus species and Euryarchaeota, associated with the depletion of methanogenic archaea, including Methanobacterium, Methanosarcina, Methanobrevibacter, Methanococcus, Methanocorpusculum, Methanococcoides, Methanosphaera, and Crenarchaeota species.^{66, 67} Differences in archaeal composition between mucosal samples from healthy and diseased tissues were observed in tubular adenoma and adenocarcinoma, Methanobacteriales and Methanobrevibacterium were also significantly higher in tumor tissues compared with control.⁷⁸ Little is known about the role of archaea in CRC.

Collectively, the nonbacterial microbiome was demonstrated to be closely associated with CRC pathogenesis and exhibited abundance alteration in major species. However, much is still unknown about their evolution in CRC progression, and mechanistic insights are still unclear, which warrants further investigation. Noteworthy, the combination of the four kingdoms of bacteria, fungi, virus, and archaea performed better in early CRC diagnosis that provided the future clinical application of microbiota in CRC, which highlighted the critical role of non-bacteria in CRC.⁶⁷

2.4 Interaction between multi-kingdom microbiota

In addition to single-kingdom microbiota analysis, interactions across multiple-kingdom microbiota can also provide valuable insights into CRC. Compared to healthy controls, higher numbers of co-occurring fungal intra-kingdom and co-exclusive bacterial–fungal correlations were found in CRC patients revealed by ecological analysis. Moreover, co-occurrences between fungi and bacteria, mostly contributed by fungal Ascomycota and bacterial Proteobacteria in control, were reverted to co-exclusive interplay in CRC.⁶¹ Besides, interactions between bacteriophages and oral bacterial commensals in patients with CRC were changed compared with controls, and the survival rate of Streptococcus species was regulated by the bacteriophage community.⁶⁴ These results provide fundamental evidence that bacteriophage communities are associated with CRC and potentially impact cancer progression by altering the bacterial communities.⁶⁵ Moreover, archaea and bacteria exhibit significant ecologic relationships in CRC patients as well. CRC-associated halophiles were positively related to the oncogenic bacterium B. fragilis and were negatively associated with butyrate-producing Clostridium species.⁶⁶ Monozygotic and dizygotic twin pairs study reported that methanogenic archaea were positively correlated with bacteria and their mutualism was supported by animal studies.^{79, 80} Although extensive research is still required, current studies suggest that inter-kingdom interactions among archaea, fungi, viruses, and bacteria contribute to colon tumorigenesis and are worthwhile to be investigated.

2.5 Aging-related microbiota in CRC

The vital effect of microbiota in the progression of aging has not yet been completely investigated, and Paul et al. provided evidence that the reduction of core microbiota group diversity is related with increased frailty but not significantly associated with chronological aging (Table 2).⁸¹ Several cohort studies also support the revealed depletion in core enrichment taxa (Bacteroides, Roseburia, and Faecalibacterium spp., among others) while rare taxa increased.^{15, 82, 83} In a cohort study compromised of centenarians, older people showed high levels of Firmicutes population and facultative anaerobes. Intriguingly, increased oral microbiota abundance, Bifidobacterium species, and Eubacterium limosum, which were considered beneficial microbiota, were also reported as microbial markers of aging in recent cross-cohort studies.^{84, 87} In addition to the altered diversity, increased α diversity was also reported reflecting the complex and dynamic alteration of microbiome with aging.^{85, 86} Given the altered microbial species pattern in aging is not consistent with that in CRC, more studies are needed to investigate the relationship among microbiota, aging, and CRC.

3.1 Mediterranean diet

Mediterranean-style diet (MED) is nutritionally balanced, characterized by large amounts of monounsaturated fatty acids (olive oil); fibers (cereals, vegetables, legumes, and nuts) and fruit; moderate consumption of seafood and red wine.⁹³ Case-control studies reported MED adherence is associated with decreased CRC risk^{94, 95}; however, the results of cohort studies are inconclusive.^96-98 Study design, age, ethnicity, and gender of the population may contribute to controversial observations.⁹⁹ However, meta-analysis supported the overall protective effect of MED on CRC.^{93, 100} Variable nutrients in MED positively modulate the intestinal microbial composition, which is beneficial in preventing CRC. Specifically, it helps promote microbial diversity and secretion of anti-inflammatory metabolites like short-chain fatty acids (SCFAs).^101-104 MED-associated bacterial taxa were enriched and occupied keystone interaction positions in the microbiome ecosystem network in the gut after a 12-month MED intervention.¹⁰⁵ Higher levels of Actinobacteria, as well as lower levels of Proteobacteria, Fusobacteria, and Firmicutes–Bacteroidetes ratios, were observed after MED administration.^106-108 Depletion of metabolic endotoxemia and maintenance of gut barrier integrity is, on the other side, another beneficial effect of MED in improving metabolic health.^{109, 110} Anti-inflammatory and immune-modulating activities of MED are possible molecular mechanisms for reducing the risk of CRC.^{111, 112} Typical food components in MED as well as their anti-CRC effects will be discussed subsequently.

3.2 Nordic diet

Nordic diet (ND) is another dietary habit that favors traditional, locally grown, and seasonal foods of the Nordic countries, characterized by the intake of whole grains, including oats, rye, barley, and specific fruit and vegetables, including apples/pears, berries, root vegetables, cabbages, and fish.^{145, 146} MED and ND are similar, except for the difference in oil used from each other. Olive oil is widely used in MED, whereas rapeseed oil is commonly seen in ND, leading to a higher level of monounsaturated fatty acids in MED.^{147, 148} Longitudinal epidemiological studies focusing on the role of ND in preventing chronic disease are limited and few conclusions have been drawn by far. One Danish cohort study showed that a higher level of ND consumption was related to a lower risk of CRC in women rather than in men, whereas the result from another Swedish cohort was against it, which showed no association in women.^{149, 150}

3.3 Western diet

In contrast to MED, the western diet is characterized by a high intake of red or processed meat, refined grains, and sugar-sweetened beverages, whereas a low consumption of fresh fruits, vegetables, and legumes.¹⁵¹ Red meat has been demonstrated to increase the risk of CRC.¹⁵² N-Glycolylneuraminic acid (Neu5Gc) and heme are specifically enriched in red meat, and the overconsumption of red meat leads to the accumulation of Neu5Gc that disrupts the homeostasis of gut microbiota characterized by the upregulation of Bacteroidales and Clostridiales.¹⁵³ Mice fed with a heme-supplemented diet lead to a significant alteration of gut microbiota, characterized by the decrease in α-diversity, reduction of Firmicutes, increase of Proteobacteria, particularly Enterobacteriaceae, and reduction in fecal butyrate levels.¹⁵⁴ In human studies, red meat diet has been shown to increase fecal Bacteroides, which increases the risk of CRC together with Streptococcus bovis, Fusobacterium, Clostridium, and Helicobacter pylori; meanwhile, L. plantarum, Lactobacillus acidophilus, and Bifidobacterium longum that can restrain the process were reduced.¹⁵⁵ Another characteristic of the western diet is high fat. The high-fat diet (HFD) promotes the hepatic synthesis of BAs and increases its delivery efficiency to the colonic lumen, which stimulates the growth and activity of 7α-dehydroxylation bacteria, and promotes the conversion of primary BAs into secondary BAs, especially deoxycholic acid (DCA).¹⁵⁶ The influence of diet on BA conjugation varies. For instance, vegetarian diets favor glycine conjugation, whereas diets high in animal protein favor taurine conjugation. Digestion of taurine-conjugated BAs by gut microbes generates hydrogen sulfide, which was shown to promote CRC progression as a genotoxic compound.¹⁵⁶ Compared with NHWs, AAs have a significantly higher abundance of sulfidogenic bacteria and B. wadsworthia-specific dsrA along with higher fat and protein intake and daily servings of meat of AAs.¹⁵⁷ By using a sulfur microbial dietary score to investigate the relationship between diet mode and CRC incidence, it was found that long-term adherence to a particular dietary mode favoring sulfur-metabolizing bacteria in stool was associated with an increased risk of distal CRC,¹⁵⁸ which was further supported by animal studies comparing the microbiota composition of rats feeding with a normal diet or HFD, which showed an altered microbiota composition in HFD group characterized by the reduction in Ruminococcus, Candida, Saccharomyces, Enterobacter, Clostridium IV, Enterococcus, Enterobacter, Vibrioaceticus, and so on, which leads to an inflammatory condition of host.¹⁵⁹ High salt intake is another feature of western diet, whereas high salt diet was reported to inhibit colonic inflammation, in-line with the reduction of enterotoxigenic B. fragilis (ETBF)-mediated tumorigenesis through decreasing the expression of IL-17A and inducible nitric oxide synthase.¹⁶⁰ Taken together, the western diet–induced microbiota pattern is close to the one under CRC-associated microbiota, with the increase of Fusobacterium, Bacteroid, and ETBF and decrease of Bifidobacterium and Lactobacillus. However, the impact of diets on gut bacteria and CRC is complicated, and more studies are required to clarify their relationships.

3.4 Interaction between nonbacterial microbiome and diet

In contrast to gut bacteria, few studies have shown the effect of diet on the nonbacterial microbiome.¹⁶¹ The field of nonbacterial microbiome lags gut bacterial research, and most of the data were collected from healthy participants rather than CRC patients. According to a study of 98 healthy adults, the consumption of carbohydrate-rich food was correlated with an increase in Candida spp. and Methanobrevibacter, whereas diets high in amino acids, protein, and fatty acids had opposite effects.¹⁶² Analysis of healthy participants with a vegetarian diet showed that Fusarium was the most abundant genus, followed by Malassezia, Penicillium, Aspergillus, and Candida.¹⁶³ The HFD was correlated with a reduced abundance of Alternaria, Saccharomyces, Septoriella, Tilletiopsis genera, Saccharomyces cerevisiae, and Tilletiopsis washingtonensis in mice.¹⁶⁴ Halophiles are enriched in salty fish and seafood.¹⁶⁵ Research revealed that the intake of salty foods promotes the enrichment of halophilic archaea in the gut, which is associated with a higher rate of colon cancer incidence.¹⁶⁶ The abundance of Methanobrevibacter was higher in healthy people, associated with the high intake of fiber, whereas it was relatively low in patients with rheumatoid arthritis and was reversed by MED diet consumption.^{102, 167} For virome, dietary interventions consistently affect the composition and proportions of viral species in the gut.¹⁶⁸ HFD leads to an increase of temperate phages under Caudovirales order.^{169, 170} These results demonstrated that fungi, archaea, and viruses are affected in response to diet intervention. However, further studies are still required to understand the mechanisms of different diets in modulating nonbacterial microbes.

5.1 Radiotherapy

Radiotherapy is one of the most frequently used treatments in CRC, and microbiota plays important role in CRC radiotherapy.^{185, 186} An animal study revealed the increased abundance of commensal Alistipes along with the decreased amount of Mucispirillum genus in mouse gut after radiation.¹⁸⁷ β-diversity significantly differed between complete response (CR) and non-CR patients. Specifically, a higher abundance of Bacteroidales (Bacteroidaceae, Rikenellaceae, and Bacteroides) was observed in non-CR than CR patients after concurrent chemoradiation in rectal cancer. Duodenibacillus massiliensis was associated with an improved CR rate.¹⁸⁸ It showed that GF mice are resistant to lethal radiation that suggests the role of microbiota in regulating intestinal radiosensitivity.¹⁸⁹ Meta-analysis revealed notable changes in intestinal microbiota characterized by the decrease in Bifidobacterium, Clostridium cluster XIVa, and Faecalibacterium prausnitzii and an increase in Enterobacteriaceae and Bacteroides after radiotherapy.¹⁹⁰ Taken all, altered bacterial composition after radiotherapy has implied the potential effects of bacteria in radiotherapy while the exact mechanism is still under-explored. Other components of intestinal microbiota are also related to radiotherapy effectiveness. Polyphenol has been discovered as a natural radioprotective product.¹⁹¹ In addition to gut bacteria, distal bacteria can also translocate to CRC sites and affect the efficiency of treatment, as it showed that F. nucleatum can migrate to CRC tissue and impairs radiotherapy efficacy.¹⁹²

Commensal fungi also regulate the responsiveness of tumor cells to radiation therapy for lung and breast cancers through the modulation of T cell activity. Depletion of commensal fungi enhances the efficacy of radiation therapy, whereas commensal bacteria had exactly the opposite effect.¹⁹³ Mice orally administrated with S. cerevisiae-derived-beta-d-glucan before irradiation showed decreased ROS level, mitigated DNA damage, and apoptosis, which was associated with an increased concentration of radioprotective cytokines in plasma. Supplementation of S. cerevisiae derived-beta-d-glucan had radioprotective effects, and the side effects were reduced accordingly.¹⁹⁴ Overall, the specific mechanism of fungi species in regulating radiotherapy efficiency needs to be addressed.

5.2 Chemotherapy

Platinum derivatives (oxaliplatin), antimetabolites (5-fluorouracil; 5-FU), cyclophosphamide (CTX), and topoisomerase inhibitors (irinotecan) are the most used agents in CRC chemotherapy, which also shown effects on microbiome.¹⁹⁵ Disruption of microbiota through antibiotics reduced the effectiveness of oxaliplatin and cyclophosphamide in MC38 tumor-bearing mice.¹⁹⁶ However, the therapeutic efficacy was affected by F. nucleatum through autophagy, which contributes to resistance to oxyplanin and 5-FU.¹⁹⁷ Conversely, 5-FU also affected microbiome composition characterized by the depletion of Eubacterium and Ruminococcus. Viaud et al. showed that CTX administration induced the translocation of Gram-positive bacteria (mainly Lactobacillus johnsonii and Enterococcus hirae) from the small intestine to secondary lymphoid organs enriched with effector pathogenic Th17 cells compared with control.¹⁹⁸ Consistently, Daillère et al. also showed E. hirae translocation associated with increased intratumoral CD8/Treg ratio during CTX therapy. Barnesiella intestinihominis is also involved in anticancer immune responses and promoted the infiltration of IFN-γ-producing γδT cells.¹⁹⁹ Intratumor bacteria are also important in CRC therapy. Γ-Proteobacteria within tumor metabolized gemcitabine into an inactive form, which results in drug resistance, whereas it can be reversed by co-treatment with antibiotic ciprofloxacin.²⁰⁰ Besides, β-glucuronidase produced by intestinal bacteria disrupted irinotecan metabolism by turning inactive SN-38G into the active form of SN-38 and causing intestinal damage.²⁰¹ Microbiota can be used as a biomarker to predict the efficiency of chemotherapy as it exhibits distinct profiles between nonresponders and responders. According to a study of patients receiving neoadjuvant chemoradiotherapy, Shuttleworthia was found enriched in responders, whereas several other bacteria taxa under the order of Clostridiales and so on were enriched in nonresponders.²⁰² In-line with this, certain fungi species are also important in regulating chemotherapy efficacy. For instance, Candida tropicalis increased the resistance of colon cancer to oxaliplatin by producing lactate and inhibiting the expression of mismatch repair system effectors.²⁰³

5.3 Immunotherapy

Immunotherapy is approved for the treatment of CRC patients with high microsatellite instability.²⁰⁴ Prevotella and Akkermansia are beneficial in maintaining the efficacy of PD-1 treatment by regulating the metabolism of glycerophospholipid after antibiotic treatment in CRC tumor-bearing mice.²⁰⁵ Likewise, the alteration of microbiota promoted IL-1β and IL-6 signaling pathways to induce protective intestinal Th17 cells in Tak1-deficient (transforming growth factor-β-activated kinase-1, Tak1^ΔM/ΔM) mice. Besides, Odoribacter splanchnicus, which was enriched in Tak1^ΔM/ΔM mice, exhibited protective effects on wild-type mice by inducing Th17 cells.²⁰⁶ A total of 11 bacteria strains isolated from healthy human feces induced interferon-γ-producing CD8 T cells and enhanced immune checkpoint inhibitors (ICI) effectiveness in mice colon cancer models.²⁰⁷ Although The associations between anti-inflammatory bacterial metabolites SCFA and the efficiency of ICI remain unclear,^{186, 208} studies have shown that higher fecal and plasma levels of SCFA were related to longer progression-free survival (PFS) in patients with solid tumor who were treated with anti-PD-1 antibody.²⁰⁹ Another metabolite associated with ICI efficiency is inosine, which is produced by Bifidobacterium pseudolongum and is proven to enhance immunotherapy efficiency in animal models.²¹⁰ Total membrane fractions and peptidoglycans (PDGs) in Porphyromonas gingivalis induced the upregulation of PD-L1 expression in colon cancer cells.²¹¹ On the other side, Bacteroides thetaiotaomicron and B. fragilis were associated with the efficacy of CTLA-4 blockade therapy. Oral gavaging with B. fragilis, immunization with B. fragilis polysaccharides, or adoptive transfer of B. fragilis–specific T cells efficiently resensitized antibiotic-treated or germ-free mice in response to anti-CTLA treatment.²¹² However, anti-CTLA-4 antibody treatment cohorts in French and Italian show propionate concentrations were correlated with shorter PFS.²¹³

Prophage is also associated with the efficiency of immune therapy. Fluckiger et al. reported that the administration of Enterococci harbor bacteriophage enhanced T cell responses after PD-1 blockade treatment in a mouse model. To expand the discovery to humans, they also compared the metagenomics data of advanced renal and lung cancer patients with healthy controls and revealed that detectable fecal prophage was related to a prolonged overall survival rate after anti-PD-1 therapy.²¹⁴ As above, differences in the composition or activity of nonbacterial microbes may contribute to different responsiveness in immunotherapies against CRC, although more in-depth studies are still needed.