The involvement of oncobiosis and bacterial metabolite signaling in metastasis formation in breast cancer

The breast tissue has its own microbiome that has higher diversity than that of the vagina, but has lower diversity than that of the oral cavity or the gut [96]. The microbiomes of the breast skin and the inner breast tissues are different [56]. The microbiome of the exterior surface of the breast (e.g., the nipple) does not change in breast cancer [57]. Nevertheless, the composition of the microbiome of the milk ducts changes in the presence of a malignant process, evidenced in nipple aspirates [57].

Next-generation sequencing methods identified bacterial DNA in breast tumors [66] that was confirmed later by alternative methods [64, 65, 67]. Breast tumor has a higher bacterial load than melanoma, lung tumors, ovarian cancer, and glioblastoma and has similar counts to pancreas and bone cancers [67]. The carcinoma tissue is colonized by bacteria [54, 58‐60, 62, 65, 67, 68, 70], and differences were detected among the subtypes of breast cancer as a function of hormone receptor status [54, 59, 60, 63‐65, 70], HER2 receptor status [65], invasiveness [54], grade [54, 58, 60, 66, 70], stage [59, 60, 70], and immunological signatures [60]. Tumor-associated bacteria are culturable and class among Proteobacteria, Firmicutes, and Actinobacteria [67]. In a fraction of breast cancer cases, intracellular and perinuclear bacteria were identified [67]. Furthermore, fungi, parasites, and viruses [64, 70, 91, 92] were detected in tumor tissue. Racial differences were identified in the breast tissue and tumor microbiome [59, 63]. Of note, certain components of the intratumoral and tissular microbiome correlate with patient survival [70].

What are the functional contributions of the breast microbiome to tumorigenesis and tumor progression? As LPS + , Gram negative bacteria were detected intratumorally [67], intratumoral bacteria likely have a key role in the local immune response (see in a later chapter). In agreement with this concept, the expression of bacterial LPS biosynthetic genes were upregulated in nipple fluid aspirate [57]. Lipoteichoic acid (LTA), a marker for Gram positive bacteria, was absent in breast cancer [67].

Imputed pathway analysis revealed relevant functions for the breast microbiome (Table 2). As with other features, there are discrepancies between the studies. Nevertheless, the identified pathways can be classified into logical categories that can be linked to tumorigenesis. Furthermore, certain predicted metabolic pathways are similar to the pathway associated with carcinogenesis in breast tissue [62].

DNA coding for beta-glucuronidase enzymes (KEGG ortholog Beta-Glucuronidase K01195), responsible for conjugated estrogen reactivation, was elevated in the nipple aspirate fluid from breast cancer patients [57], suggesting a possible local reactivation of estrogens. In patients, Roseburia, Rikenellaceae, Bacteroides uniformis, Paenibacillus amylolyticus, and Ellin6075 were the OTUs contributing to increased beta-glucuronidase abundance [57]. The same study showed an increase in the abundance of genes coding for steroid hormone biosynthesis [57].

Breast cancer cells are characterized by lower oxidative stress than the parent terminal lobular-ductal unit (TLDU) tissue [97]. The low level of oxidative and nitrosative stress is a key feature of successful tumor growth [98, 99]. Tissular and intratumor bacteria can support low tumoral oxidative and nitrosative stress by upregulating L-ascorbate biosynthesis II (L-glucose pathway) [67]. This is further strengthened by increased mycothiol biosynthesis in ER + tumors [67]. Mycothiol is used by bacteria to detoxify reactive oxygen species [100]. In addition to oxidative/nitrosative stress, the physical presence of bacteria can also induce DNA damage, at least in part, due to the overexpression of colibactin [55]. Among Enterobacteriaceae, Escherichia coli or Staphylococcus can produce a toxin, colibactin [55, 101, 102]. In agreement with this concept, Klann and colleagues [62] demonstrated a correlation between the expression of base excision repair genes and the bacterial colonization of the breast tissue.

Metabolic changes were described in the imputed pathway analyses. Of note, some reports show contradictory results. Changes affect core metabolic pathways, including anaerobic respiration [67], oxidative phosphorylation [57], and central carbon metabolism [62]; stage I tumors were enriched in genes of energy metabolism [59]. Another set of changes affected lipid metabolites, such as sphingolipid metabolism [57], synthesis and degradation of ketone bodies [57], linoleic acid biosynthesis [57], and choline metabolism [62]. In benign cases, fatty acid biosynthesis and branched dibasic acid metabolism increased [56], while in stage I tumors, genes of fat digestion and absorption were enriched [59]. A large set of amino acid metabolic pathways were affected (tryptophan glycine, serine, threonine, phenylalanine biosynthesis [57]), as well as nitrogen metabolism [57]. In benign cases, cysteine and methionine metabolism genes were enriched [56]. Changes were described in detoxification processes, such as polycyclic aromatic hydrocarbon degradation [57] and benzoate degradation [57] in the oncobiome. In malignant cases, local, oncobiome-mediated drug metabolism may increase [56]. Interestingly and importantly, signal transduction pathways such as PI3K-Akt [62], HIF-1 [62] and the AMPK pathway [62], microRNAs involved in carcinogenesis [62], and inositol phosphate metabolism [56] changed in the breast cancer oncobiome. In stage 2 tumors, the microbiome was enriched in phosphotransferase system protein genes [59]. Furthermore, sulfur metabolism genes changed in the breast cancer oncobiome [57] and glycosyltransferases increased in benign cases [56]. Breast cancer is characterized by changes in cell metabolism that is actionable for disease treatment and management [103‐112]. Furthermore, the intricate supportive metabolic circuit between cancer cells and non-cancerous stroma cells can facilitate tumor growth and lead to worse clinical outcomes [103‐105, 113]. Giallourou and colleagues [71] showed that breast bacteria interfere with the biosynthesis of ceramide, cholesterol, oxidized cholesteryl esters, diacylglycerol, lysophosphatidylcholine, phosphatidylethanolamines, and phosphatidylcholines to modulate the lipid composition of tumors. Of note, cholesterol and lipid homeostasis also play a role [4, 114‐116]. The widespread changes to the oncobiome metabolism suggest that the breast cancer oncobiome participates in the metabolic support of rapidly dividing breast cancer cells.

In terms of cell death and cell division, the breast cancer oncobiome is associated with necroptosis [62]. Haemophilus influenzae is associated with mitotic spindle formation and G2M checkpoint regulation [68]. The breast cancer oncobiome is also associated with movement and metastasis-related processes, such as dermatan sulfate degradation [67], peptidoglycan biosynthesis [57], and proteoglycan homeostasis [62]. Listeria fleischmannii in the breast cancer oncobiome is associated with epithelial-to-mesenchymal transition [68].

Tzeng and colleagues have shown that elements of the microbiome covaried with different markers of bad clinical outcomes. Namely, Staphylococcus negatively covaried with TRAF4, Pelomonas positively covaried and Bradyrhizobium negatively covaried with VEGF-A, and Propionibacterium negatively covaried with PDGF-AA and PDGF-BB. In addition to these data, circumstantial data suggest bacterial secretory proteins [41] are probably also involved in communication between the microbiome and tumor tissue.

The gut microbiome undergoes oncobiotic transformation in breast cancer [21, 51, 52, 73‐82]. In terms of diversity indices, Goedert and colleagues [73] and Byrd and colleagues [74] reported lower or trends towards lower diversity indices in three different cohorts, while Zhu and colleagues [75] and Bobin-Dubigeon and co-workers [84] reported increased alpha diversity indices. Howe and co-workers [76] reported increases in alpha diversity indices in a Pten-deficient mouse model backing the observations of Zhu and colleagues [75] and Bobin-Dubigeon and co-workers [84]. No differences in alpha diversity were reported in [87] and [89]. However, the multiple observations make lower alpha diversity more likely. As noted earlier, risk factors for breast cancer lead to decreases in diversity, such as high-density breast [50], early menarche [51], low physical activity [51], and increases in BMI [51, 52]. Furthermore, antibiotic overdose, which leads to lower diversity, increases the risk for breast cancer [22, 23, 25‐32], while probiotics that increase diversity have a protective effect [37‐45].

Characteristic changes in the microbiome were observed between clinical stages [51, 77‐79, 81] and grades [51, 89], MIB positivity [51], receptor status [51, 89], and proliferative capacity [89]. The most drastic changes were observed among in situ carcinoma and stage I patients, which were gradually rediversified in subsequent stages [77‐79, 81]. Characteristic changes in taxa between patients and controls include Clostridiales [51, 73, 77, 82], Blautia [51, 77], and Akkermansia muciniphila [52, 76].

The gut microbiome is distant from the breast tumor; hence, signaling pathways are needed to connect the two distant compartments. Multiple pathways cross-connect the oncobiome and the tumor tissue. The direct immunomodulatory effects of the microbiome will be discussed in the next chapter.

Intestinal bacteria expressing beta galactosidases (gus and BC genes [117‐119]) can deconjugate conjugated estrogens. The gus gene is widespread among bacteria, while changes in BC include Bacteroidetes and Firmicutes [119]. Collinsella, Edwardsiella, Alistipes, Bacteroides, Bifidobacterium, Citrobacter, Clostridium, Dermabacter, Escherichia, Faecalibacterium, Lactobacillus, Marvinbryantia, Propionibacterium, Roseburia, and Tannerella were shown to express β-glucuronidases [120]. Goedert and colleagues provided strong evidence for the involvement of Clostridiales in estrogen reactivation in breast cancer patients [73, 121, 122]. The oncobiome has increased capacity to reactivate estrogens [21, 50, 73, 74, 83, 121‐124] enabling their reuptake and supporting the growth of estrogen-dependent, estrogen receptor–positive (ER +) breast tumors. Of note, the capacity for estrogen reactivation was identified by pathway analysis in the breast and nipple aspirate microbiomes [57].

Bioactive metabolites, synthesized by the microbiome or the oncobiome, can act in a similar fashion to hormones and can link up the microbiome and the distantly located cancer cells [2‐4, 125]. As the gut microbiome is the biggest in the body in terms of the number of bacteria, its metabolic capacity is considerable. The biosynthetic capacity of the oncobiome is suppressed compared to the eubiome [126, 127]. Multiple bioactive bacterial metabolites were identified that can modulate the behavior of breast cancer cells (Table 3). The importance of changes to the metabolome in breast cancer is further highlighted by the large number of metabolomic studies that point towards the role of the metabolome in breast cancer incidence and evolution [128‐132]. The bioactive bacterial metabolites are very chemically diverse. We will briefly highlight the most important bacterial metabolites that have cytostatic features. A set of bacterial toxins contributes to the oncogenic property of the oncobiome (Table 4) similar to other neoplasias [133‐135].

Lithocholic acid (LCA) is a secondary bile acid derived from primary bile acids. Mostly Clostridia in the large intestines are responsible for the production of LCA and secondary bile acids in general [82]; however, other taxons are also involved [191]. The enzymes involved in secondary bile acid production are assembled in the bile acid inducible (bai) operon in bacteria [191]. The production of LCA from its precursors involves the deconjugation of primary bile acids and the oxidation, epimerization, and dehydroxylation of the gonane core [2, 4, 167, 191, 192]. Secondary bile acids, such as LCA, have pleiotropic roles in the microbiome, including regulation of the microbiome composition [193‐200], facilitation of bacterial translocation into tissues [201], and quorum sensing [202, 203].

The bile acids in the breast are of the gut origin [204, 205]. Total bile acid levels in the serum of healthy individuals are lower than 5 µM. LCA serum reference concentrations are low (30–50 nM). However, tissue levels in the breast may be substantially higher [78, 206]. In breast cancer, both the hepatic synthesis of primary bile acids and the bacterial conversion to secondary bile acids in the large intestine are suppressed, and this suppression is the most dominant in in situ carcinoma and stage I patients [78, 82]. In good agreement with this concept, serum LCA levels negatively correlate with the Ki67 labeling index in breast cancer [207]. The composition of the serum bile acid pool in patients with benign breast disease is different from breast cancer patients; breast cancer patients had higher serum chenodeoxycholic acid levels and lower dihydroxytauro-conjugated bile acids (Tdi-1) and sulfated dihydroxyglyco-conjugated bile acids (Gdi-S-1) [208]. Another secondary bile acid deoxycholic acid (DCA) may act as a procarcinogenic agent [209, 210] and may be responsible for the procarcinogenic character of secondary bile acids [211].

A multitude of receptors is involved in bile acid signaling, including Takeda G protein-coupled receptor 5 (TGR5) and farnesoid X receptor (FXR), which are important for the current discussion. One study [212] suggested the use of bile acid-tamoxifen conjugates for breast cancer therapy.

Multiple amino acid catabolic products derived from lysine and tryptophan have cytostatic properties in breast cancer. Indole derivatives are made from tryptophan, while lysine decarboxylation yields cadaverine.

The microbiome accounts for 4–6% of tryptophan catabolism to yield indol derivatives [213], of which indolepropionic acid (IPA) and indoxyl sulfate (IS) have cytostatic properties in breast cancer [81, 175]. The serum reference concentration of IPA is submicromolar [176, 214, 215], while IS concentrations are low micromolar [216]. The bacterial enzyme responsible for IPA and IS biosynthesis, called tryptophanase (TnaA), can be found in the tryptophanase operon [176]. Tryptophanase expression is widespread among bacteria [217, 218]. IPA and IS, similar to other indole derivatives, can activate the aryl hydrocarbon receptor (AHR) and bind to the pregnane-X receptor (PXR) receptor [219‐221]. Indole derivatives have a strong immunostimulatory effect [222‐224]; IPA and IS can induce antitumor immunity in breast cancer [81, 175] and modulate the composition of the microbiome [219, 225‐227]. Evidence from human studies supports the role of tryptophan and indole metabolism in breast cancer. Elevated extracellular tryptophan levels decrease survival in breast cancer (Table S8 [228]). Ki67 positivity negatively correlates with 3-indoxyl sulfate levels ([207] Additional file 9, Table S8 line 130), and 3-indoxyl sulfate levels are downregulated in both estrogen receptor–positive and negative cases ([207] Additional file 3, Table S3 line 44). TnaA DNA is downregulated in the breast cancer microbiome, and the most drastic changes were observed in in situ and stage I cases [81].

In the human body, cadaverine can be of bacterial, human, or nutritional origin. Nevertheless, bacterial cadaverine production seems to dominate [79]. In bacteria, the LdcC and CadA genes are responsible for cadaverine biosynthesis [172, 229], while diamino-oxidase eliminates cadaverine [230]. The capacity for cadaverine biosynthesis was identified in a number of bacteria [231‐233]. Human serum reference concentration of cadaverine is submicromolar (100–800 nM) [234, 235]. Cadaverine can activate trace amine-associated receptors (TAAR1, 2, 3, 5, 8, 9), and these receptors are associated with breast cancer [79, 174]. Fecal TnaA protein content was reduced in E-cadherin-negative breast cancer cases compared to E-cadherin-positive cases [80].

Short-chain fatty acids (SCFAs), such as formate, acetate, propionate, butyrate, and lactate, are generated by a large set of bacterial species from non-digestible carbohydrates and a minor fraction from amino acids [236]. SCFAs are formed at multiple points in bacterial metabolism and are then released into the environment. Therefore, a wide circle of bacteria can synthesize SCFAs. The reference concentration of SCFAs in the human serum is in the range of 10–100 µM [237‐239] and may reach up to 1 mM locally [240]. The receptors for SCFAs are the free fatty acid receptors (FFARs) and AHR [151].

Oncobiosis-related changes to bacterial metabolite production support oncogenesis and not tumor initiation through multi-faceted effects. In certain cases, as for SCFAs, effects are context-dependent; SCFAs can have positive (e.g., [241]) or negative (e.g., [242]) effects in breast cancer. Many studies used metabolites in superphysiological concentrations that may render the interpretation of these studies difficult [243‐246]. In superphysiological concentrations, metabolites can induce cell death [171, 244‐247] and other features (e.g., in superphysiological concentrations, LCA blocked fatty acid biosynthesis [171] or induced cell death and the expression of multidrug resistance proteins [247]) that is usually mediated by secondary receptors, such as FXR in the case of LCA [247]. The metabolites assessed in the context of breast cancer were ineffective on untransformed primary human fibroblasts at their reference concentration, suggesting selectivity towards transformed cells [78, 79, 81, 97, 175].

Metabolites elicit effects by reducing oxidative and nitrosative stress through induction of NRF2 expression and its downstream effector genes and suppressing inducible nitric oxide synthase (iNOS). These effects may also involve the catechol-quinone metabolites of estrone and estradiol [81, 97, 175, 248]. Furthermore, the oncobiome contributes to the induction of Warburg metabolism in tumor cells [78, 79, 81, 97, 175]. These basic processes support epithelial-to-mesenchymal transition [78, 79, 81, 249], migration and invasion [79, 81], increased aldehyde dehydrogenase-1 (ALDH1)–positive cancer stem cells [79, 81], and suppression of tumor immunity [22, 24, 78, 81, 175, 250, 251]. These processes together support metastasis formation [24, 78, 79, 250, 251] and disease recurrence [29]. In addition, most SCFAs can act as energy sources in cells [233] and may inhibit histone deacetylases to modulate epigenetics [152, 154‐157, 241, 252‐257]. LCA supplementation can reduce VEGF production by the implanted breast cancer cell in an animal model of breast cancer [78]. Furthermore, the oncobiome correlates with omega-3 polyunsaturated fatty acid homeostasis [85].

Our current understanding indicates that these metabolites only represent the tip of the iceberg. Multiple studies [75, 87, 89] identified imputed pathways that were differentially regulated in the oncobiome and eubiome in breast cancer patients (Table 5). In addition to the metabolic changes, a large set of transport systems are dysregulated in bacteria [75]. These findings highlight the widespread effects of bacterial bioactive metabolite production.

As mentioned earlier, the suppression of the biosynthetic capacity of the microbiome is most pronounced in the early stages of the disease [79, 81, 97]. Nevertheless, studies on the expression of receptors for the metabolites suggest that receptors on the surface of tumor cells are downregulated as tumor stage or grade increases [81, 97]. The receptors for microbiome-derived metabolites and the components of the downstream signaling pathways in the tumor correlate with the receptor status, grade, and stage. The intratumoral expression of AHR, PXR, and TGR5 decreases in TNBC cases compared to ER + cases [81, 97]. These findings were mirrored in relapse-free survival rates, where high expression of the receptors and their downstream signaling components provided better survival for patients that was abrogated in TNBC cases [81, 97]. In Figs. 2 and 3, we show that a subset of TAAR receptors and FFAR1 receptors have similar properties. Similarly, in LCA-elicited downstream signaling events, PXR and AHR receptor expression decreased as a function of stage, grade, or high mitotic rate [81, 97].

The impact of changes in the urinary microbiome is unexplored. However, recently, An and colleagues showed that bacterial extracellular vesicles can be isolated from the urine of healthy subjects and breast cancer patients and these extracellular vesicles can affect breast cancer cells differently [90].

In general, the diverse status of the microbiome supports normal immune responses that are crucial for antitumor immunity [259‐261]. Both the breast and the gut oncobiome have altered interactions with the immune system that we will review here. Whether changes to the bacterial community lead to changes to the immune system or the immune system causes changes to the microbiome remains an open question. Furthermore, myeloid and lymphoid infiltrations in tumors have differential effects; lymphoid infiltration is generally considered to have antitumor effects, while neutrophils are considered proneoplastic [262, 263]. We will discuss the microbiome-mediated changes to the immune system separately for the breast microbiome and the gut microbiome.

Before going into details, we would like to briefly introduce two bacterial immunogenic toxins, which are classified as pathogen-associated molecular patterns (PAMP). First, lipopolysaccharides (LPS) (Table 4), lypoglycans, or endotoxins are constituents of the outer membrane of Gram negative bacteria [177]. Toll-like receptor 2 and 4 respond to LPS stimulation [177, 180]. Second, lysophosphatids (Table 4) are generated in reactions related to bacterial membrane homeostasis [184, 264]. Cells of the host organism can also generate lysophosphatids. Gram negative bacteria tend to generate lysophosphatids [184, 264]. Endogenous phospholipase A2 or other exogenous lipases can generate lysophosphatids [184]. Lysophosphatids exert their effects through lysophosphatidic acid receptors (LPAR1-6) [185].