Colorectal cancer (CRC) is the third most prevalent cancer and ranks second in mortality worldwide among all cancer types with 1.8 million new cases and 881,000 deaths in 2018.¹ The 5-year survival rate in the United States is 64%, with early detection improving the survival significantly.² CRC incidences vary greatly with higher prevalence in developed countries and it is estimated that the global burden will increase by 60% in 2030.³ Twin and family studies suggest that genetic-predisposition causes only approximately 5% of the CRC cases, while it was estimated that about 90% of gastric cancer and CRC may be attributed to diet and other lifestyle factors.^4,5 This is supported by the finding that the prevalence of CRC is roughly 60 times higher in African Americans compared to Native Africans.⁶ A major risk factor for CRC is chronic intestinal inflammation, and patients with Crohn’s disease and ulcerative colitis are at an increased risk for CRC.^7,8 The mucosal immune system is regulated by the gut microbiome. The human gut hosts approximately 38 trillion microorganisms and has been linked to inflammatory bowel disease.⁹ More recently, the role of the microbiome in the development and progression of CRC is becoming increasingly appreciated.¹⁰ For instance, patients with CRC have decreased bacterial diversity and richness compared to healthy individuals.¹¹ Additionally, certain bacterial species, such as Fusobacterium nucleatum, Streptococcus bovis, and Bacteroides fragilis have been linked to CRC, either by producing virulence factors or by producing pathogenic microbial metabolites.^12-15 The gut microbiota has an extensive metabolic capacity, greatly exceeding that of human cells.¹⁶ It has been shown that approximately 50% of the metabolites in the feces and urine are derived from the gut microbiota.^17,18 Microbial metabolites, which are small molecules, generally defined as <1,500 Da, can influence host physiology and immune functions.^19,20 Due to their small size, microbial metabolites are readily absorbed into colonocytes, and some reach the blood circulation and peripheral organs. Microbial metabolites can be generated from dietary compounds or endogenous host products. Interestingly, it is estimated that the average human diet contains about 8,000 non-nutritious compounds, most of which are processed by the gut microbiota but not by human enzymes.²¹ Some of these metabolites have carcinogenic or anticarcinogenic effects.^22,23 The most abundant microbial metabolites in the gut are derived from dietary fiber, fat, bile acids and proteins. This review summarizes the effects of these microbial metabolites on the development and progression of CRC and discusses potential mechanisms behind the effects.

Non-digestible complex carbohydrates resistant to digestion by human enzymes are fermented in the intestine to short-chain fatty acids (SCFAs), which serve as a major energy source for colonocytes and influence the immune system.²⁴ SCFAs are short forms of fatty acids with fewer than six carbon atoms, with acetate, propionate and butyrate being the major SCFAs in the colon (Fig. 1). The concentrations of these SCFAs in the colon are in the order of acetate > propionate > butyrate with molar ratios of approximately 60:20:20.^25,26 The total luminal SCFA concentrations in humans are 70–140 mmol/kg in the proximal colon, 20–70 mmol/kg in the distal colon, and 10–20 mmol/kg in the terminal ileum.^27,28 It has been estimated that approximately 95% of the produced SCFAs are absorbed by colonocytes, and the absorbed SCFAs contribute to 5%–15% of the caloric requirement for humans.^29-31 SCFAs are taken up by cells via passive diffusion or by transporters, such as monocarboxylate transporters (SLC16A1 and SLC16A3) and sodium-coupled monocarboxylate transporters (SLC5A8 and SLC5A12).^32-34 The presence of SCFAs is sensed by intestinal epithelial cells and immune cells by G-protein coupled receptors (GPRs), such as GPR41, GPR43, GPR109a, and Olf78.^35-37 Furthermore, SCFAs act as type I and II histone deacetylase (HDAC) inhibitors, with butyrate and propionate having higher HDAC-inhibiting activities compared to acetate.^38,39 Interestingly, the peroxisome proliferator-activated receptor γ can also serve as a receptor for butyrate in colonocytes.^40,41

Figure 1. Major metabolites produced by the gut microbiota. (A) Non-digestible carbohydrates are metabolized to the short-chain fatty acids, acetate, propionate, and butyrate. (B) Primary bile acids are conjugated in several different forms and secreted into the intestine. Gut bacteria deconjugate and further modify primary bile acids by 7α-dehydroxylation to produce secondary bile acids, such as deoxycholic acid and lithocholic acid. (C) Arginine is converted to polyamines by host and bacterial enzymes. Polyamines are also present in diets. (D) Tryptophan metabolites are produced via four different pathways: The serotonin pathway, the tryptamine pathway, the kynurenine pathway, and the bacterial indole pathway. (E) Degradation of cysteine to pyruvate produces hydrogen sulfate (H₂S). (F) The animal meat-derived metabolites carnitine, choline, or phosphocholine (PC) are degraded by gut bacteria to trimethylamine (TMA), which is further metabolized in the liver to trimethylamine N-oxide (TMAO).

Dietary fiber intake is inversely correlated with CRC occurrences.^42-44 Two recent meta-analyses found that higher dietary fiber intake reduces the risk of CRC by 26% to 32%.^45,46 Butyrylated high-amylose maize resistant starch decreased red meat-derived pro-mutagenic adduct O6-methyl-2-deoxyguanosine and related CRC risk.⁴⁷ Certain butyrate-producing bacteria such as Clostridium, Ruminococcus, Faecalibacterium spp., Eubacterium spp., and Roseburia, were found to decrease in relative frequency in CRC patients.^48,49 Interestingly, it has been observed that CRC patients had lower stool concentrations of butyrate, while acetate concentrations were increased.^48,50 In a gnotobiotic mouse model colonized with either wild-type or mutant strains of the butyrate-producing bacterium Butyrivibrio fibrisolvens, high levels of dietary fiber intake increased the production of butyrate and suppressed chemically-induced CRC in animals colonized with the wild-type strain but not with its mutant form that cannot produce butyrate.⁵¹ The three major SCFAs promote the proliferation of normal crypt cells in the intestine of healthy individuals. It has been reported that butyrate inhibited the growth of CRC cells and induced their apoptosis in vitro at pharmacologically high concentrations.⁵² This appears to involve microRNA biogenesis.⁵³ A caveat is that there is no conclusive in vivo evidence showing that butyrate can suppress or kill cancer cells at physiological concentrations. Interestingly, CRC patients on a high fiber diet had relatively low levels of butyrate-producing bacteria and consequently lower butyrate concentrations in the colonic lumen compared with healthy controls on the same diet.⁴⁹ This indicates that CRC patients and healthy subjects intrinsically have different gut microflora. Overall, SCFAs, especially butyrate, appear to suppress CRC development.

However, how SCFAs, particularly butyrate, suppress colon cancer is not entirely clear at this point but there are several potential mechanisms. One potential mechanism proposed by Donohoe et al.,⁵¹ which is based on the in vitro tumor-killing phenomenon by butyrate, may involve a metabolic shift from oxidative phosphorylation to aerobic glycolysis (often referred to as “Warburg effect”). This may lead to an intracellular accumulation of butyrate, which augments HDAC inhibition, thereby decreasing cell proliferation and causing apoptosis (Fig. 2). However, it remains unclear if the butyrate-mediated apoptosis is a major mechanism to suppress CRC in vivo where the transport and intracellular concentrations of SCFAs are highly controlled.

Figure 2. Basic functions and host receptors of microbial metabolites. Dietary or host-derived molecules are converted by the gut microbiota to microbial metabolites that act on host receptors or are utilized by the host. Thereby, microbial metabolites regulate host metabolism, gene expression and signaling pathways. G-protein coupled receptors (GPRs) that sense short-chain fatty acids (SCFAs) include GPR43, GPR41, GPR109A, and Olfr78. The solute carriers that transport SCFAs across the cell membrane include SLC16A1, SLC16A3, SLC5A8, and SLC5A12. The receptors for the microbial metabolites are distinctively expressed depending on cell their types (i.e., epithelial cells, macrophages, endothelial cells, T cells etc.), and the effects of the metabolites on different cell types can be distinct.
SLC, solute carrier family proteins; PPARγ, peroxisome proliferator-activated receptor γ; HDACs, histone deacetylases; TGR5, G protein-coupled bile acid receptor 1; PXR, pregnane X receptor; FXR, farnesoid X receptor; VDR, vitamin D receptor; AhR, aryl hydrocarbon receptor; PC, phosphocholine; TMA, trimethylamine; TMAO, trimethylamine-N-oxide; NLRP3, NLR family pyrin domain containing 3; ROS, reactive oxygen species.

Besides their roles as nutrients and HDAC inhibitors, SCFAs also signal via GPRs. Triggering GPR43 and GPR41 dynamically regulates the gut barrier, inducing tight junction proteins at steady state but rapidly increasing the permeability during infection, perhaps, to facilitate acute immune responses. Mice deficient in the SCFA receptors are slow to mount acute immune responses but develop chronic unresolving inflammatory responses.⁵⁴ In line with the notion that chronic inflammation increases the risk for CRC, it has been shown that mice deficient in GPR43 or GPR109a develop tumors at elevated rates.^54-56 High levels of fat consumption is a major risk factor for CRC, which is supported by the increased tumor formation in Apc^min/+ mice on high-fat, compared with regular, diets. The butyrate-producer Clostridium butyricum, when orally inoculated, decreased the number of tumors in these animals.⁵⁷ The same study also found that treatment of CRC cell lines with either C. butyricum or butyrate suppressed the Wnt signaling pathway and induced apoptosis in a GPR43-dependent manner. Furthermore, reduced expression of GPR43 and GPR109a was observed in cancerous colon tissues.^55,57 Thus, GPR triggering by SCFAs is an important mechanism to suppress CRC development.

SCFAs also regulate immune responses. SCFAs improve immunity against bacterial and viral pathogens.^58-61 These functions, together with their effects on intestinal epithelial cells, are important for maintaining intestinal immune homeostasis and preventing CRC development.⁶² Additionally, activation of GPRs on intestinal epithelial cells by SCFAs promotes barrier functions and stimulates the production of interleukin (IL)-18 and secretion of anti-microbial peptides.^63-66 Innate immune cell populations, such as neutrophils, macrophages, dendritic cells and innate lymphoid cells (ILCs), are also regulated by SCFAs.^67-69 Group 2 innate lymphoid cells (ILC2) express GPR43 and monocarboxylate transporter 1 (MCT1, encoded by SLC16A1), while group 3 innate lymphoid cells (ILC3) express GPR41 and GPR43 but not MCT1.^69,70 Deficiency of GPR43 on ILC3s decreased IL-22 production, thereby causing impaired gut epithelial function and increasing susceptibility to colonic injury and infection.⁷¹ While SCFAs increase the proliferation of both ILC3 and ILC2 via GPR43, SCFAs can also suppress ILC2, potentially by inhibiting HDACs.⁷² Additionally, neutrophils express high levels of GPR43 and its stimulation by SCFAs induces chemotaxis and activation of neutrophils.⁷³ Contradictorily, mice fed a zero fiber diet had an increase in neutrophils in the colon and were more susceptible to colitis induced by oral treatment of dextran sulfate sodium.⁷⁴ Thus, SCFA activation of neutrophils is not necessarily inflammatory. Intestinal macrophages and dendritic cells are also regulated by SCFAs. Activation of GPR109a by butyrate induces anti-inflammatory properties in colonic macrophages and dendritic cells.⁶³ SCFAs increase the expression of IL-10 and retinaldehyde dehydrogenase 1 (ALDH1A1) in myeloid cells and induce Tregs. ^63,75 SCFAs also directly promote IL-10⁺ regulatory T cell activity.³⁹ While SCFAs do not appear to affect peripherally induced FoxP3⁺ T cells in general, it has been also reported that intestinal FoxP3⁺ T cells can be increased by SCFAs.^76,77 Butyrate and propionate reduced the expression of IL-6 and IL-12 in lipopolysaccharide-stimulated dendritic cells.⁶⁸ Similar anti-inflammatory effects were observed by GPR41 activation by butyrate, which decreased the production of tumor necrosis factor (TNF-α), monocyte chemoattractant protein-1 (MCP-1), and IL-6 in macrophages by decreasing nuclear factor-κB (NF-κB) activity.⁶⁷ Thus, the anti-inflammatory function of SCFAs by regulating immune cells is another important mechanism behind the suppressive effect of SCFAs on CRC development.

Primary bile acids (PBAs), such as cholic acid (CA) and chenodeoxycholic acid (CDCA), are produced from cholesterol in hepatocytes and are conjugated to either taurine or glycine via an amide linkage at the C24 carboxyl group to generate conjugated bile acids. Conjugated PBAs are released into the duodenum to support the absorption of lipids and fat-soluble vitamins (Fig. 1). The majority of secreted PBA are reabsorbed in the terminal ileum.⁷⁸ However, a small portion (~5%) of the PBAs (about 200 to 800 mg daily in humans) escapes the reabsorption and reaches the colon where gut bacteria convert them into secondary bile acids (SBAs), such as deoxycholic acid (DCA), ursodeoxycholic acid (UDCA), and lithocholic acid (LCA).⁷⁹ SBA concentrations in the colon can range from 200 to 1,000 µM, and the majority (90% to 95%) of SBAs are reabsorbed into colonocytes to return to the liver for detoxification and recycling.^80-82

Bile acids trigger multiple cellular receptors, such as the farnesoid X receptor (FXR), the pregnane X receptor (PXR), the vitamin D receptor, the G protein-coupled bile acid receptor 1 (TGR5), α5β1 integrin, and the sphingosine-1-phosphate receptor 2 (Fig. 2).⁸³ Bile acids differ in their receptor binding affinity and activation efficiency. For example, FXR shows bile acid affinities in the order of CDCA>DCA>LCA>CA, while the TGR5 shows affinities in the order of LCA>DCA>CDCA>CA.²⁰ Interestingly, TGR5 is upregulated in lamina propria mononuclear cells in the inflamed mucosa of inflammatory bowel disease patients,⁸⁴ and DCA instillation into the rat colon induced low-grade colitis.⁸⁵ These findings suggest that bile acids promote inflammatory responses in the gut.

Elevated levels of fecal SBAs, especially DCA, are associated with an increased risk for CRC.^86,87 CRC incidence is extremely high in Alaska Natives, and Alaska Natives have elevated levels of fecal DCA levels, suggesting that high DCA levels correlate with increased CRC risk.⁸⁸ Moreover, African Americans have higher DCA levels (~5 times) and greatly increased incidence of CRC (~60 times) compared to Native Africans. This appears to be due, in part, to their high fat but low fiber diet.⁸⁹ In the high-risk populations for CRC, increased levels of SBA-producing 7α-dehydroxylating bacteria were found.⁸⁸ Using biopsy specimen from patients with or without colorectal adenoma, a positive correlation was found between serum DCA levels and intestinal epithelial cell proliferative activity.⁹⁰ While bile acids are associated with increased CRC risk in humans, it has been also reported that the treatment with another SBA, UDCA, decreased recurrence of adenomas with high-grade dysplasia.⁹¹ UDCA has microbiome-changing and DCA‐lowering functions and therefore this further supports the pro-tumorigenic effects of DCA and other SBA.^92,93

Animal studies further indicate that SBAs promote the development of experimental CRC. Intrarectal instillation of DCA increased intestinal tumors in azoxymethane-treated animals.⁹⁴ Surprisingly, DCA alone in the diet for 8 to 10 months without the application of carcinogens was sufficient to induce tumors in most animals.⁹⁵

SBAs appear to regulate cancer progression via multiple mechanisms in bile acid receptor-dependent and independent manners. Bile acids, including SBAs, are hydrophobic and therefore can perturb cell membrane integrity.⁹⁶ DCA also induces phosphorylation of β-catenin and upregulates the expression of cyclin D1 and urokinase-type plasminogen activator receptor (uPAR), thereby increasing cell proliferation and invasiveness.⁹⁷ Furthermore, DCA stimulates extracellular signal-regulated kinases, which counteracts the tumor-suppressive activity of the tumor suppressor p53.^98,99 DCA impaired p53 signaling in a cancer cell line exposed to DNA-damaging reagents in vitro.⁹⁸ The bile acid receptor TGR5 is overexpressed in gastric adenocarcinoma, and bile acids can increase cell proliferation via TGR5 signaling.¹⁰⁰ However, FXR, another receptor for bile acids, is downregulated in CRC and low levels of FXR expression correlate with worse clinic outcomes.^101,102 This raises the possibility of bidirectional regulation of CRC by bile acids through distinct receptors. Overall, DCA is linked to increased CRC development and multiple BA receptors with diverse functions are implicated in this process. Interestingly, DCA also stimulates the uptake of polyamines in CRC cells, which also have tumorigenic effects.¹⁰³

Amino acids that reach the colon are converted by the microbiome to polyamines or indole derivates. Polyamines are molecules that contain more than two amino groups and are synthesized from arginine and ornithine by mammalian and bacterial enzymes (Fig. 1). Additionally, certain foods, such as fruits, cheeses and meat, contain relatively high levels of polyamines.¹⁰³ The most common polyamines are putrescine, spermidine, and spermine, which play important roles in cell growth, proliferation, and migration. It is considered that diets provide a major source of luminal polyamines in the small intestine, while the gut microbiota are responsible for the luminal polyamine concentrations in the colon.¹⁰⁴ Colonocytes are exposed to putrescine at concentrations of 100 to 800 mM and spermidine and spermine at concentrations of 50 to 60 µM.¹⁰⁵ Correlations between the gut microbiota composition and fecal polyamine concentrations have been reported, which suggests a significant role of gut microbes in polyamine production.¹⁰⁵

Polyamines can feed cancer cells for growth. Tumor cells require more polyamines for growth than healthy cells, and cancer patients have increased polyamine levels in the blood and urine.¹⁰⁶ Generally, intracellular polyamine concentrations are associated with epithelial cell proliferation and transformation.^107-109 Polyamine concentrations appear to be increased in CRC tissues compared to healthy tissues.¹¹⁰ Polyamines also promote oncogenic signaling. Spermidine/spermine-N(1)-acetyltransferase is a catabolic enzyme that decreases the level of intracellular polyamines. This enzyme can suppress the levels of phosphorylated protein kinase B and β-catenin and decrease the proliferation and migration of tumor cells.¹¹¹ While the relative contributions remain unclear, the currently available body of evidence suggests that any polyamines, regardless of their sources (i.e., dietary, microbial, and tissues) can drive tumorigenesis.

The rate-limiting enzyme in polyamine synthesis is the ornithine decarboxylase (ODC), which is highly expressed in tumors.¹¹² Interestingly, the chemotherapeutic agent difluoromethylornithine (DFMO), which targets the mammalian ODC has been efficient in the inhibition of cancer cell growth in vitro but failed in vivo, presumably due to its inability to block microbial polyamine production. Indeed, suppression of microbial production of polyamines enhanced the efficiency of DFMO.¹¹³ Polyamines are transported into cells by endocytosis and mice treated with a combination of DFMO and the polyamine transport inhibitor AMXT 1501 showed a reduction in tumor growth, while either drug alone did not show a significant effect.^114,115 Similarly, a combination of DFMO and the polyamine uptake inhibitor Trimer PTI also reduced tumor growth.¹¹⁶ Interestingly, T cells are implicated in the anti-cancer effect of the inhibitors. While polyamines are required for T and B cell activation and proliferation, blocking polyamine synthesis and transport increased tumor infiltration of T cells.^114,117 This is presumably due to the anti-inflammatory effect of polyamines on myeloid cells.¹¹⁸ These findings suggest that both endogenous and exogenous (dietary or microbial) polyamines can contribute to cancer growth and immune system evasion (Fig. 2).

The amino acid tryptophan, which is abundant in meat, dairy products, soy and seeds, is metabolized by the kynurenine (Kyn), serotonin, indole, and tryptamine pathways in host and/or bacterial cells (Fig. 1).¹¹⁹ The host Kyn and bacterial indole pathways each process ~90% and ~5% of the available tryptophan, respectively. Many common gut bacteria are capable of producing indoles and the fecal indole concentration ranges between 1 and 4 mM.^120,121 Common indole metabolites include indole-3-pyruvate, indole-3-acetaldehyde, indole-3-lactic acid, indole-3-acetic acid, indole-3-acetamide, and indole-3-propionic acid.^119,122 Indoles are recognized by PXR and the acryl hydrocarbon receptor (AhR), which are expressed by various cell types including intestinal epithelial and immune cells. Dietary indoles, found in cruciferous vegetables, also function as AhR ligands.

Indole derivates support gut homeostasis by promoting epithelial barrier function and immune tolerance, which are important to suppress carcinogenesis in the colon.^119,123,124 Germ-free mice, colonized with a mutant Clostridium sporogenes defective in indole-3-propionic acid production, had increased gut permeability and signs of inflammation, such as immune cell infiltration, in the colon.¹²² Furthermore, indole-3-carbinol supplementation reduced colitis in mice by inducing the expression of IL-22, a pleiotropic cytokine that boosts epithelial immunity and inflammation in the intestine.¹²⁵ Thus, indole metabolites are linked to decreased colitis and CRC development.

Tryptophan metabolism is altered in CRC with a shift towards increased Kyn production by host cells but decreased indole production by gut microbes. Elevated fecal Kyn to tryptophan ratios but decreased indole to tryptophane ratios were found in patients with colorectal neoplastic lesions.¹²⁶ Interestingly, colon cancer cells have an increased ability to take up tryptophan compared to healthy cells and generate Kyn, which is a potential oncometabolite.¹²⁷ Additionally, Kyn acts as an endogenous ligand for AhR and can be secreted by cancer cells into the tumor microenvironment, suppressing anti-tumor immunity and promoting tumor cell survival through AhR via autocrine and paracrine signaling.^128,129 Conversely, mice lacking AhR developed more tumors upon azoxymethane and dextran sulfate sodium treatment compared to wild-type mice, while supplementation of wild-type mice with indole-3-cabinol reduced tumor burden by 92%.¹³⁰ Furthermore, supplementation of Apc^Min/+ mice with indole-3-carbinol reduced tumor burden.¹³¹ The plant-derived indole-3-carbinol has similar anticarcinogenic effects.¹³² The protective effect of AhR is explained, in part, by ubiquitylation and degradation of β-catenin induced by AhR triggering by indole metabolites. Interestingly, a constitutively active form of AhR induced stomach tumors in transgenic mice.¹³³ This suggests that AhR can either increase or decrease tumorigenesis.

The metabolites from the Kyn and bacterial indole pathways appear to trigger distinct effects of AhR. Kyn exerts pathogenic effects, whereas indole metabolites have protective effects.^{126,127,130,131} Interestingly, AhR also mediates the toxicity of the carcinogen dioxin.¹³⁴ Further research is needed to understand the mechanisms for the divergent effects of different AhR ligands on CRC pathogenesis.

Hydrogen sulfate (H₂S) is important for normal regulation of blood pressure but is also implicated in a variety of pathological conditions, such as inflammatory bowel diseases, neurodegeneration, and CRC.^135,136 Sulfate-containing amino acids, such as cysteine, serve as a source of H₂S (Fig. 1). Both, host enzymes and gut microbes generate H₂S. Fecal H₂S concentrations are elevated in CRC patients compared with healthy controls.¹³⁷ Increased levels of H₂S can decrease tissue integrity by reducing disulfide bonds in mucosal tissues.¹³⁸ H₂S can also directly cause DNA damage (Fig. 2).¹³⁹ H₂S increased the in vitro proliferation of human cells obtained from mucosal biopsies.¹⁴⁰ Interestingly, this effect was inhibited by butyrate. Fusobacterium nucleatum, a Gram negative, anaerobic bacterium over-represented in CRC tissues, is a H₂S-producer.^141,142 F. nucleatum colonizes the tumor microenvironment and promotes tumor growth. F. nucleatum also produces the metabolite formate, which triggers AhR and increases cancer cell stemness and invasiveness.¹⁴³

The microbiota metabolizes phosphatidylcholine, choline, and carnitine to produce trimethylamine (TMA) (Fig. 1), which is further metabolized in the liver to trimethylamine-N-oxide (TMAO). Plasma choline and TMAO levels correlate with an increased risk for CRC and serum TMAO levels are significantly increased in CRC patients.^144,145 While the mechanism by which TMAO promotes colon carcinogenesis remains unclear, a recent meta-analysis including 13,783 subjects found that the level of circulating TMAO correlates with that of C-reactive protein, an inflammation indicator.¹⁴⁶ Thus, TMAO is linked also to inflammation. TMAO activates NLRP3 and increases reactive oxygen species formation.¹⁴⁷ In line with this, reduced production of TMA decreased proinflammatory cytokine and reactive oxygen species production in animals.¹⁴⁸ Thus, H₂S, formate, and TMAO are strongly linked to CRC development.

The information presented and discussed in this article supports the important regulatory roles of major microbial metabolites in CRC development. While certain metabolites, such as butyrate, UDCA, and indoles, appear protective, other metabolites, such as DCA, polyamines, H₂S, TMAO, and formate, are associated with colon carcinogenesis (Table 1). We propose that metabolite balance and dysbiosis created by our lifestyle choices, including diets, health status and underlying conditions, profoundly regulate the cellular metabolism, activation, proliferation and differentiation of tissue and immune cells (Fig. 3). Metabolite balance maintains the integrity of the intestine and other tissues including the tissue and immune compartments, whereas metabolite dysbiosis with over-produced harmful metabolites undermines their basic maintenance and functions. Metabolite dysbiosis promotes dysregulated immune functions, weakened barrier functions, microbial invasion, and increased inflammation, all of which can contribute to CRC development. Conversely, beneficial microbiome and microbial metabolites suppress metabolite dysbiosis. This concept is supported by the decreased fecal concentrations of SBAs, DCA and LCA by the butyrate-producer C. butyricum and the negative correlation between fecal butyrate and DCA levels.^57,88 Because even the harmful metabolites have certain beneficial functions in the body, the balanced presence of the beneficial versus harmful metabolites (e.g., butyrate vs DCA) is considered important for the development of CRC. As discussed, proteins can be metabolized to polyamines and indoles with distinct effects on CRC development. Diets rich in proteins, regardless whether plant or animal-based, appear to increase CRC risk in animal models.^149,150 The amino acid tryptophan can be converted to pro-tumorigenic Kyn, as well as to anti-tumorigenic indoles by the gut microbiota.

Table 1 . Studies on the Impact of Microbial Metabolites on CRC Development.

Metabolite	Effect on CRC development	Mechanism in CRC	References
SCFAs and their receptors in general	Suppressive	Dietary fiber is inversely correlated with CRC and colorectal adenoma incidences	42-46
		SCFA receptors GPR43 and GPR109a are protective for CRC development	55-57
Acetate	Promoting	Increased in CRC patients	50
Butyrate	Suppressive	Decreased in CRC patients	48-50
		Lower butyrate metabolism in CRC tumors causing apoptosis	52
		Suppression of cancer proliferation	53,55,57
Secondary bile acids	Promoting	Positive correlation with CRC risk	86,87
DCA	Promoting	Levels correlate with CRC risk	86,87
		Promotes colon carcinogenesis via ROS production, stimulation of cell proliferation and p53 inhibition	97-99
UDCA	Suppressive	UDCA treatment reduced colorectal adenoma recurrence	91
Polyamines	Promoting	Increased polyamine concentrations in blood and urine of cancer patients	106
		Polyamines promote cancer cell growth	107-109
Indoles	Suppressive	Decreased fecal indole/tryptophane ratio but increased fecal Kyn/tryptophane ratio in CRC patients	126
		Indole derivates protect gut barrier functions and reduce tumor formation	123-127
H2S	Promoting	Increased fecal H2S concentrations in patients with CRC	137
		Causes damage to mucosal layer and DNA	138,139
TMAO	Promoting	Positive correlation between blood TMAO levels and CRC	144,145
		TMAO promotes inflammation and ROS formation	146,147
Formate	Promoting	Induces cancer stemness and increases invasiveness	143

CRC, colorectal cancer; SCFAs, short-chain fatty acids; GPR, G-protein coupled receptor; DCA, deoxycholic acid; ROS, reactive oxygen species; UDCA, ursodeoxycholic acid; H₂S, hydrogen sulfate; TMAO, trimethylamine-N-oxide..

Figure 3. Impact of microbial balance and dysbiosis on cell functions and colorectal cancer development. The risk for developing colorectal cancer is influenced by the concentrations and relative balance between the protective (butyrate, indoles, and UDCA) and potentially harmful (DCA, formate, TMAO, polyamines, and H₂S) metabolites produced by the gut microbiota. The absolute and relative levels of these metabolites in the gut and tissues are determined by diet, microbial composition, other lifestyle factors and underlying health conditions. Metabolite dysbiosis undermines tissue and immune cell homeostasis and barrier immunity and affects cell growth, leading to microbial invasion, inflammation, and tumorigenesis.
UDCA, ursodeoxycholic acid; DCA, deoxycholic acid; TMAO, trimethylamine-N-oxide; H₂S, hydrogen sulfate.

The host condition and microbial composition along with diet composition can determine the type and concentrations of metabolites. The fact that CRC patients have lower butyrate levels even with sufficient consumption of dietary fiber suggests that host condition is a key determining factor for microbial metabolite production. Simple dietary modifications without normalizing the microbial community would be insufficient to significantly alter gut metabolite composition and CRC risk.⁴⁹ Effective treatments of CRC patients should involve transferring beneficial microbes and/or reducing potential pathobionts. In this regard, probiotics and fecal microbiota transplants are currently considered for CRC.¹⁵¹ Additional regimens to change host inflammatory conditions could be important as well.

A caveat to consider in interpreting the data from human and animal studies is that there are considerable differences in microbiome, diet, and metabolites between the two species. While high-fat diet and associated increases in bile acid secretion promote CRC formation in mouse studies, human studies failed to find such a clear correlation.¹⁵² This reflects, in part, the diversity in human genetics and diet. Multiple factors should be considered to achieve a state of “balanced microbial metabolites” in the gut.

Finally, the impact of microbial metabolites on the mucosal immune system, barrier immunity, and chronic inflammation in the colon is a major factor to consider in CRC development.^7,8 The microbiota produces a treasure trove of metabolites that directly interact with colonocytes and the immune system. Although correlations and potential mechanisms between microbial metabolites and colon carcinogenesis have been identified, further research is needed to gain more mechanistic insights and curve the expected increase in the global burden of CRC on society.

This study was supported, in part, from grants from NIH (NIH R21AI14889801, R01AI074745, 1R01AI080769, and R01AI121302) to C.H.K. And he has been supported by Kenneth and Judy Betz Professorship in food allergy research at the University of Michigan.

The authors thank the current and past members of the Lab of Immunology and Hematopoiesis at the University of Michigan and Purdue University for helpful insights.

No potential conflict of interest relevant to this article was reported.