The role of bile acids in carcinogenesis

Bile acids are soluble derivatives of cholesterol produced in the liver that subsequently undergo bacterial transformation yielding a diverse array of metabolites. The bulk of bile acid synthesis takes place in the liver yielding primary bile acids; however, other tissues have also the capacity to generate bile acids (e.g. ovaries). Hepatic bile acids are then transported to bile and are subsequently released into the intestines. In the large intestine, a fraction of primary bile acids is converted to secondary bile acids by gut bacteria. The majority of the intestinal bile acids undergo reuptake and return to the liver. A small fraction of secondary and primary bile acids remains in the circulation and exert receptor-mediated and pure chemical effects (e.g. acidic bile in oesophageal cancer) on cancer cells. In this review, we assess how changes to bile acid biosynthesis, bile acid flux and local bile acid concentration modulate the behavior of different cancers. Here, we present in-depth the involvement of bile acids in oesophageal, gastric, hepatocellular, pancreatic, colorectal, breast, prostate, ovarian cancer. Previous studies often used bile acids in supraphysiological concentration, sometimes in concentrations 1000 times higher than the highest reported tissue or serum concentrations likely eliciting unspecific effects, a practice that we advocate against in this review. Furthermore, we show that, although bile acids were classically considered as pro-carcinogenic agents (e.g. oesophageal cancer), the dogma that switch, as lower concentrations of bile acids that correspond to their serum or tissue reference concentration possess anticancer activity in a subset of cancers. Differences in the response of cancers to bile acids lie in the differential expression of bile acid receptors between cancers (e.g. FXR vs. TGR5). UDCA, a bile acid that is sold as a generic medication against cholestasis or biliary surge, and its conjugates were identified with almost purely anticancer features suggesting a possibility for drug repurposing. Taken together, bile acids were considered as tumor inducers or tumor promoter molecules; nevertheless, in certain cancers, like breast cancer, bile acids in their reference concentrations may act as tumor suppressors suggesting a Janus-faced nature of bile acids in carcinogenesis.


Background
Bile acids (BAs) belong to cholesterol-derived sterols. Due to the side chain carboxyl group and hydroxylation of their steroid ring they are more polar than cholesterol. They have an amphipatic character for which they are known as natural detergents. Majority of cholesterol is excreted by bile acids that are prone to enterohepatic circulation between the gallbladder and the liver. Cholesterol absorption in the intestine and cholesterol secretion into the bile both require bile salts, which are, together with enterohepatic circulation of BAs, crucial for balancing the plasma cholesterol level [1]. BAs are also signaling molecules. They deorphanized the farnesoid X nuclear receptor (FXR) which is now known as a ligand-inducible transcription factor responsive to BAs [2]. It is important to note that BAs are metabolized in a similar manner as xenobiotics, contributing to the crosstalk between the endogenous and xenobiotic metabolism in the liver through nuclear receptors Pregnane X receptor (PXR), constitutive androstane receptor (CAR) and others [3]. While their synthesis takes place exclusively in the liver, the homeostasis and excretion involve multiple organs and compartments in the body. After discovering their signaling role, BAs have been considered as pro-carcinogenic molecules [4][5][6]. However, recent studies have provided evidence that in certain cancers, BAs can have antineoplastic features (e.g. breast cancer [7][8][9][10][11]). This novel, context-dependent, dualistic finding prompted us to thoroughly assess the involvement of BAs in carcinogenesis and cancer progression.

Bile acid biosynthesis
The excess of free cholesterol is toxic to cells and needs to be excreted, primarily through conversion to more polar BAs. The introduction of a hydroxyl group in cholesterol reduces the half-life and directs the oxidized molecule to excretion [12]. BA synthesis is thus the main cholesterol detoxification pathway where multiple cytochrome P450 (CYP) enzymes are involved in the classical or alternative pathways (Fig. 1). The two major primary BAs in humans are cholic acid (CA) and chenodeoxycholic acid (CDCA). They are synthesized in the liver and secreted into the gallbladder as glycine or taurine conjugates [13]. The BA composition in mice substantially differs from the humans which has to be taken into account when using mouse as a model for BA related diseases. The mouse Cyp2c70 metabolizes CDCA to more hydrophilic primary muricholic acids (MCAs) [14].
The first enzyme of the classical BA synthesis pathway is cholesterol 7α-hydroxylase (CYP7A1), leading to 7α-cholesterol in a rate-limiting reaction step, followed by several enzymatic conversions. This enzyme is prone to the negative feedback regulation by BAs and FXR [2]. Sterol 12α-hydroxylase (CYP8B1) lies at the branching point that leads to CA. Sterol 27-hydroxylase (CYP27A1) is needed for both CA and CDCA. In the alternative pathway, cholesterol is first metabolized by CYP27A1 to form 27-hydroxycholesterol that is a substrate for 25-hydroxycholesterol 7α-hydroxylase (CYP7B1) and later other enzymes [15]. The alternative pathway leads majorly to CDCA. The ratio of CA to CDCA is determined by the expression level of CYP8B1, which transforms a di-hydroxylated BA to trihydroxylated BA. The alternative pathway is estimated to account for about 10% of cholesterol conversion [16]. Of importance, there are major differences in individual BA synthesis genes in mouse and in humans which may be due also to different biological roles of human and mouse BA species (reviewed in [15]).
lipases in the small bowel. BAs are then reabsorbed through the enterocytes and get to the liver for reuptake and reuse through the portal circulation. This circle is termed the enterohepatic circulation of BAs. A fraction of the reabsorbed BAs enter the systemic circulation (total BA concentration in the serum is < 5 µM in a healthy individual) and exert hormone-like effects [7,[17][18][19][20]. The reference concentrations of the serum, tissue and fecal bile acids are in Tables 1, 2, 3.  BAs are very powerful surfactants [21]; therefore, bacteria, mostly in the large bowel, need to protect themselves against being disintegrated by BAs. For example, lipopolysaccharides serve as membrane components in Gram-negative bacteria to passively ward off external toxins or BAs [22]. In addition to that, bacteria have a more sophisticated enzymatic system to cope with BAs termed BA conversion [23].
The hydroxyl groups and the tauryl or glycyl conjugate on BAs are crucial elements of the molecular structure of BAs for their strong surfactant properties. Therefore, the removal, modification or substitution of these molecular elements diminishes the potentially toxic features of primary BAs and renders them largely apolar.   [57], further underlining that notion BAs act as potent drivers of the early intestinal microbiota maturation [58]. Oncobiosis (dysbiosis associated with cancers) [59] can alter the secondary BA pool that may contribute to carcinogenic effects [4,5,7,18]. It is of note that several other non-BA bacterial metabolites are known that play role in carcinogenesis [60][61][62][63][64].

Bile acid transporters
The enterohepatic circulation of BAs depends on BA transporters in the gastrointestinal system. Almost 90% of BAs are involved in circulation due to efficient active transport [65]. Different uptake and efflux BAs transporters are present in the hepatic and intestinal cells (Fig. 2). After BAs are synthesized in the liver they are transported into the bile mainly by the ATP-dependent cassette transporter (BSEP) [65], but also minor transporters, the multidrug resistanceassociated protein 2 (MRP2, ABCC2) and the multidrug resistance protein 1 (MDR1, ABCB1) [65]. From the intestinal lumen, BAs are uptaken into the intestinal cells by the major apical sodium-dependent bile acid transporter (SLC10A2, ASBT), which transports BAs also across the canalicular membrane in cholangiocytes and renal tubule apical membrane from glomerular filtrate [66]. BAs are then effluxed into the portal circulation by two Solute Carrier Family members, SLC51A or OSTα and SLC51B or OSTβ. The bile acids are then taken back up into hepatocytes by the major transporter the solute carrier family 10 (SLC10A1, NTCP), [65].
BAs can enter the systemic circulation via export across the hepatic sinusoidal membrane by OSTα/OSTβ, the multidrug resistance-associated protein 3 (MRP3, ABCC3) and the multidrug resistance-associated protein 4 (MRP4, ABCC4) [67]. The MRP transporters have a role in reducing hepatic BA concentration in cholestatic conditions. MRP3 and MRP4 are also present in cholangiocytes, where they efflux BAs to portal circulation and are part of the cholehepatic shunt together with ASBT [66]. Several transporters are expressed in the kidney, where they participate in BA elimination via urine (Fig. 2) [66,68,69]. The Solute Carrier Organic Anion Transporter Family, OATP1B1 or SLCO1B1 and OATP1B3 or SLCO1B3 contribute to the systemic clearance of BAs via liver [70]. Other cells also express BA transporters and can, therefore, uptake BAs from the systemic circulation [68,69,71].

Bile acids as signaling molecules
In addition to their role in digestion, BAs act as signaling molecules. BAs can activate membrane receptors (Fig. 3), such as G protein-coupled bile acid receptor 1 (GPBAR1, also known as TGR5), sphingosine-1-phosphate receptor 2 (S1PR2), muscarinic receptors (CHRM2 and CHRM3) and nuclear receptors (NRs), such as farnesoid X receptor (FXR, NR1H4), PXR (NR1H2), vitamin D receptor (VDR, NR1H1), CAR (NR1H3) and liver X receptor (LXR, NR1H2-3). Each BA can interact with more than one receptor. Receptors are differentially activated by BAs. For example, FXR is activated by CDCA > DCA > LCA > CA [72], while TGR5 is activated by LCA > DCA > CDCA > CA [73,74], respectively. VDR and PXR are mainly activated by LCA. BAs mediate immune responses [75], gastrointestinal mucosal barrier function, gestation [76], carcinogenesis [11,18,56] and metabolic diseases [20]. The activation of BA receptors may lead to the induction of signaling pathways involved in the regulation of several physiological functions, such as glucose, lipid and energy metabolism, as well as, in cancers. Below, we review the mode of action of BA receptors and highlight those receptor-mediated functions that have a key role in regulating the behavior of cancer cells.

G protein-coupled bile acid receptor 1 (GPBAR1, TGR5)
TGR5 is a member of the G protein-coupled receptor superfamily, highly expressed in the epithelium of the gallbladder [77], the intestine [74], the brown adipose tissue and the skeletal muscle [20], as well as in the brain [78]. TGR5 is also expressed in human monocytes/macrophages [73]. TGR5 is not expressed by hepatocytes, while Kupffer cells and liver sinusoidal cells can express the receptor [79].

Farnesoid X receptor (FXR, NR1H4)
FXR is a member of the nuclear hormone receptor superfamily. There are two FXR genes, encoding FXRα and FXRβ of which only FXRα is expressed, FXRβ is present as a non-expressed pseudogene in humans. The FXR receptor heterodimerizes with retinoid X receptor (RXR) and binds to FXR response elements (FXREs) within the regulatory regions of its target genes [107]. BAs are physiological ligands for FXR (with decreasing affinity: CDCA, DCA, LCA, CA) [72]. FXR is expressed mainly in the liver, intestine, kidney and adrenal glands [107].
FXRα controls BA synthesis, transport and detoxification. The activation of FXR receptor by BAs reduces the expression of Cyp7a1 and Cyp8b1, key enzymes of BA biosynthesis pathway. In the liver, FXRα induces the transcription of its target gene encoding small heterodimer partner (SHP, NR5O2), an orphan nuclear hormone receptor (see in detail later) that lacks a DNA binding domain and acts as a transcriptional repressor [108]. SHP inhibits the expression of Cyp7a1 through the inhibition of the interaction with liver receptor homolog-1 (LRH-1, NR5A2) [109]. In addition to LRH-1, SHP also prevents the function of hepatocyte nuclear factor-4α (HNF4α), a positive regulator of Cyp7a1 and Cyp8b1 [110]. In the intestine, FXRα induces the expression of fibroblast growth factor 19 (FGF19) in humans and its mouse homolog fibroblast growth factor 15 (FGF15). The secreted growth factor via portal blood reaches the liver where it binds to its receptor, fibroblast growth factor receptor 4 (FGFR4) and induces JNK and ERK pathways and causes repression of Cyp7a1, thus reducing BA synthesis [111]. In addition to Cyp7a1, Cyp8b1 is also repressed by FXRα via SHP-dependent mechanism involving HNF4α [110]. FXRα is also a key regulator of BA transport by influencing the expression of BA transporters. FXRα activation suppresses BA reuptake to hepatocytes through repressing the expression of NTCP via SHP dependent mechanism [112]. At the same time, FXRα facilitates the efflux of BAs from hepatocytes into bile by enhancing the expression of BSEP and into the systemic circulation via OSTα/β transporter [113]. FXR also upregulates MRP2, which promotes BA secretion into the gallbladder. Finally, FXRα activates the expression of intestinal BA-binding protein (I-BABP) in the ileum which promotes transport of BAs from enterocytes into portal blood [114] whereas limits enterocyte uptake of BAs by reducing ASBT expression. FXRα increases the expression of enzymes involved in the detoxification of BAs, such as cholesterol 25-hydroxylase or cytochrome P450 family 3 subfamily A4 (CYP3A4) [115], dehydroepiandrosterone-sulfotransferase (SULT) 2a1 [116] and uridine 5′-diphosphate-glucuronosyltransferase 2B4 (UGT2B4) [117]. Many studies have reported the relationship between FXR and inflammation. NF-kB activation suppressed FXRmediated gene expression, indicating that there is a negative crosstalk between the FXR and NF-kB signaling [118].

Pregnane X receptor (PXR, NR1I2)
In humans, PXR is mainly expressed in the liver and intestine [119]. Among BAs, the most potent ligand of PXR is LCA, and the oxidized, 3-keto form of LCA. PXR acts as a xenobiotic sensor and regulates the expression of genes involved in the detoxification and metabolism of BAs [120]. Upon ligand binding, PXR binds to the promoter of its target gene as a heterodimer with RXR. Activation of PXR induces the uptake of xenobiotics, their modification by phase I enzymes (CYPs, including CYP3A, CYP2B, CYP2C), conjugation by phase II enzymes, such as glutathione S-transferases, UDP-glucuronosyl-transferases (UGTs) and sulfotransferases, and finally elimination by phase III drug transporters including MDR1, MRP2 and organic aniontransporting polypeptide (OATP2) [120]. The activation of PXR prevents cholesterol gallstone disease by regulating BA biosynthesis and transport [121] and protects the liver against LCA-induced toxicity [122][123][124][125]. PXR activation disrupts the interaction between HNF4α and peroxisome proliferator-activated receptor gamma coactivator 1 alpha (PGC-1α, PPARGC1A), which is required for the activation of CYP7A1 gene expression, thus reducing the expression of CYP7A1 and inhibiting the synthesis of BAs

Constitutive androstane receptor (CAR, NR1I3)
CAR is the closest relative to the PXR and is expressed primarily in the liver. First studies identified that CAR has constitutive transcriptional activity in the absence of its ligand [134]. Later, it was reported that the constitutive transcriptional activity of CAR is reversed by androstane metabolites, which are inverse agonists [135]. CAR can be activated by direct ligand binding and indirect activation [136]. In the absence of ligand binding, CAR forms a heterodimer with RXR and transactivates its target genes [137]. CAR recruits coactivators in the nucleus, such as steroid receptor coactivator 1 (SRC-1, NC0A1) and PGC-1 [138]. Similar to PXR, CAR controls the expression of drug-metabolizing enzymes and transporters, thereby supporting the detoxification of xenobiotics [120,139]. In contrast to PXR, it remains unclear whether BAs can function as natural ligands for CAR; nevertheless, there are reports underscoring the involvement of CAR in BA signaling [11].

Vitamin D receptor (VDR, NR1I1)
In humans, VDR is highly expressed in the kidney, intestine, bone as well as in hepatocytes but expressed at low levels in other tissues [140][141][142]

Liver X receptor (LXR, NR1H2-3)
LXRs are activated by naturally occurring cholesterol metabolites such as oxysterols and bind to DNA as heterodimers with the RXR [153]. LXRα (NR1H3) and LXRβ (NR1H2) share a high structural homology [154]. LXRβ is ubiquitously expressed, while LXRα is primarily expressed in the liver, the adipose tissue, the intestine and macrophages. Upon ligand activation LXRs regulate gene expression via binding to LXR response elements in the promoter regions of the target genes. LXRα promotes the conversion of cholesterol into BAs through the induction of CYP7A1 expression in the liver. LXRs enhance the efflux of cholesterol from cells [155] and have an anti-inflammatory response in the adipose tissue and macrophages [156]. Hyodeoxycholic acid (HDCA), a naturally occurring secondary BA generated by bacterial C-6 hydroxylation of LCA, is a weak LXRα agonist [157].

Small heterodimer partner (SHP, NR5O2)
SHP is a unique nuclear receptor that contains a ligand-binding domain but lacks the conserved DNA-binding domain. SHP acts as a transcriptional corepressor regulating different metabolic processes, including lipid, glucose, energy homeostasis and BA synthesis via interaction with multiple transcription factors and nuclear receptors (reviewed in [158]). BAs or FGF19 signaling enhances posttranslational modifications of SHP, which modulates the regulatory function of SHP protein [159,160]. SHP acts as an inhibitory regulator in Hedgehog/Gli signaling pathway [161].

Effects of bile acids in cancers
The role of BAs was implicated in a wide variety of neoplasias (Fig. 4, Tables 5, 6, 7). When assessing the effects of BAs, one has to keep in mind that the concentrations applied in the experiments need to correspond to the reference concentrations in serum or the compartment in question (e.g. parts of the gastrointestinal tract). However, several reports are using substantially higher concentrations than the reference. These studies need to be considered as ones using "therapeutic" concentrations. In the forthcoming chapters, we will review those neoplasias where BAs were implicated in pathogenesis.

Oesophageal carcinoma
The development of Barrett's esophagus (BE) and its progression to oesophageal adenocarcinoma (EAC) are linked to gastroesophageal reflux disease (GERD). Conjugated BAs, mainly taurocholic acid (TCA) and glycocholic acid (GCA) are the main BA constituents in GERD refluxate [162]. Conjugated BA levels in the refluxate from patients with advanced BE or EAC are significantly higher than from patients with benign BE [163]. Conjugated BAs, as TCA or taurodeoxycholic acid (TDCA), promote EAC progression [164,165] (Table 7). Unconjugated BAs, including DCA and CDCA, induce oxidative stress, DNA damage and inflammation contributing to EAC carcinogenesis, while UDCA protects against DCA-induced injury (Tables 5 and 7).
Apparently, numerous BA receptors as TGR5, S1PR2, FXR and VDR are activated in EAC cells in response to BAs in the refluxate [164][165][166][167]. In good agreement with that, the inhibition of the FXR receptor suppresses tumor cell viability in vitro and reduced tumor formation in nude mouse xenografts [168]. Furthermore, TGR5 is highly expressed in the EAC and precancerous lesions and is associated with worse overall survival [169] suggesting that these observations can be translated to the human situation.

Gastric cancer
Carcinogenesis in gastric cancer is a sequential process that includes chronic superficial gastritis, intestinal metaplasia (IM), atrophic gastritis, intramucosal carcinoma, dysplasia and invasive neoplasia [174]. IM is considered a risk factor for gastric tumorigenesis. The concentrations of BAs in gastric juice positively correlate with the degree of intestinal metaplasia [175] and BAs serve a critical multipronged role in the induction of intestinal metaplasia. BAs can enhance caudal-related homeobox family 2 (CDX2) and mucin 2 (MUC2) expression via FXR/NF-κB signaling [176,177] and cyclooxygenase-2 (COX-2) expression via induction of SHP [178], all promoting gastric intestinal metaplasia. Acidic bile salts can induce telomerase activity in a c-Myc-dependent fashion [179,180], while DCA can induce the metaplastic phenotype of gastric cancer cells [181] (see Tables 6 and 7). TGR5 is a key factor in BA-induced gastric metaplasia via HNF4α [181], EGFR and mitogen-activated protein kinase (MAPK) [182] activation and promotes EMT in gastric carcinoma cells [183]. TGR5 is overexpressed in gastrointestinal adenocarcinomas, and moderate to strong TGR5 staining is associated with decreased patient survival [184]. Nevertheless, there anticarcinogenic effects of bile acids in gastric cancer, as UDCA (Table 5) or DCA in supraphysiological concentrations [185,186] or 23(S)-mCDCA [187].
Interestingly, senescence-associated secretory phenotype has crucial role in promoting obesity-associated HCC development in mice. Administration of high-fat diet to mice induces alterations in the gut microbiota and increases the levels of DCA. Increased DCA level promotes SASP   [194] Neuroblastoma (NB) SK-n-MCIXC, BE(2)-m17, SK-n-SH, Lan-1 LCA (100 µM) LCA selectively kills the NB cell lines while sparing normal neuronal cells. LCA triggers intrinsic and extrinsic pathways of apoptosis [8] phenotype in hepatic stellate cells (HSCs), which in turn secretes various tumor-promoting factors in the liver, thus facilitating HCC development in mice exposed to chemical carcinogen [6]. SHP has a pleiotropic role in HCC, regulates cell proliferation [215], apoptosis [216], epigenetic changes [217] and inflammation [200,218], which are associated with the antitumor role of SHP in the development of liver cancer.

Pancreatic adenocarcinoma
BAs are involved in the induction and development of pancreatic adenocarcinoma at multiple stages. Gallstone formation can block bile flow and, therefore, can induce and sustain pancreatitis [219], a risk factor for pancreatic adenocarcinoma [220][221][222]. In fact, several BA species showed a drastic increase in pancreatic adenocarcinoma patients [223]. Treatment of pre-malignant pancreas ductal cells with bile induced carcinogenic transformation [224,225]. In pancreatic adenocarcinoma cells BAs decrease susceptibility to apoptosis, boost cell cycle progression, the expression of inflammatory mediators and cellular movement, and, in high concentrations, may perturb biomembranes (Table 7) [220,226]. UDCA, similar to its previously discussed beneficial properties, prevents EMT in pancreatic adenocarcinoma cell lines and, therefore, has antineoplastic properties (Table 5) [227].

Colorectal carcinoma (CRC)
The western diet has tumor promoting activity associated with elevated concentrations of colonic BA (mainly LCA and DCA) and increased fecal BA levels, as detected in samples from CRC patients [228]. In animals, a high-fat diet stimulates bile discharge and results in elevated BA levels in the colon [229]. Moreover, cholecystectomy, through prolonging BA exposure of the intestinal mucosa, has been suggested as a risk factor for the development of CRC [230]. BAs induce genetic instability marked by genomic instability and DNA damage via oxidative stress, defects in mitotic checkpoints, cell cycle arrest, improper chromosome alignment and multipolar division [231,232]. Genomic instability caused by BAs is coupled with apoptosis resistance due to the degradation of p53 and the inhibition of caspase-3 activity [233]. Furthermore, secondary BAs perturb cell membranes and modulate signaling cascades [234,235].  These all lead to colonic cell hyperproliferation, survival and invasion [236,237]. The disruptive effect of BAs on colon epithelium evokes a compensatory cell renewal mechanism by inducing colonic epithelial cells to become cancer stem cells (CSCs) through β-catenin signaling (Table 7) [238]. In the CRC rodent model, both LCA and DCA have tumor promoter role on colonic crypt cells in the early stages of colon carcinogenesis [239]; however, it is important to note that BAs are suggested as tumor promoters, but not as mutagenic agents, since they can not induce tumor formation without a carcinogen/mutagen or a genetic alteration [240,241]. It should be noted that DCA in low concentrations (0.05-0.3 mM) inhibit colonic cell proliferation via cell cycle block and apoptosis pathways ( Table 6) [242].
UDCA can reduce the concentration of toxic BA in stool and blood [243] and has shown to protect against CRC by inhibiting CSC and CRC cell formation and proliferation [244,245], oncogenic signaling pathways [246], as well as, inducing tumor surveillance [247] (Table 5). Moreover, UDCA can reduces CRC recurrence [248], as well as the risk to develop CRC in patients with pre-cancerous conditions, as colitis [249] or primary biliary cirrhosis [250].
Sustained inflammation was implicated in the pathogenesis of colorectal cancer due to barrier breach, and bacterial translocation leading to inflammation and neoplastic transformation of colonic epithelial cells [251][252][253]. TGR5 activation by UDCA and LCA may also exert anti-inflammatory responses through TLR4 activation or by reducing pro-inflammatory cytokine production in the colon that can decrease the frequency of developing CRC [254]. BAs can change the gut microbial community [255,256], suggesting that BAs may also interfere with bacterial translocation.

Breast cancer
The BAs in the breast are of gut origin [257,258]. Hepatic production of BA is reduced in breast cancer patients as marked by decreasing levels of serum and fecal BAs [7,259]. Furthermore, bacterial conversion of BAs to secondary BAs is also suppressed, which is the most dominant in in situ and stage I patients [7]. The serum bile acid composition of breast cancer and benign breast disease patients is different; specifically, breast cancer patients had higher serum chenodeoxycholic acid levels and lower dihydroxy tauroconjugated BA (Tdi-1) and sulfated dihydroxy glyco-conjugated bile acids (Gdi-S-1) [260]. Total fecal bile acid levels are lower in breast cancer patients as compared to controls [259]. LCA concentrations in the breast can be higher than the serum levels [261] (Table 6). Reports showed increased DCA levels in the serum [262] and the breast cyst fluid [263] of breast cancer patients.
CDCA in supraphysiological concentrations induces MDRs through FXR [265] and modulates estrogen and progesterone receptor-mediated gene transcription [267]. Furthermore, CDCA inhibits tamoxifen-resistant breast cancer cell proliferation through the activation of the FXR receptor [268] (Table 6). In contrast to that, a report by Journe and colleagues [269] showed that FXR activation has a positive correlation with estrogen receptor expression and luminal characteristics, as well as supported cancer cell proliferation.

Prostate cancer
Among the BAs LCA, UDCA and CDCA exerted antiproliferative effects in prostate cancer. Activation of FXR by CDCA inhibits proliferation of prostate cancer cells, reduces lipid anabolism via inhibiting Sterol Regulatory Element Binding Transcription Factor 1 (SREBF1) [270] and induces the expression of the tumor suppressor phosphatase and tensin homolog (PTEN) [271] (Table 6). Interestingly, FXR signaling also controls androgen metabolism in prostate cancer cells, its activation reduces the expression of UDPglucuronosyltransferase (UGT) 2B15 and UGT2B17 within cells and causes a reduction of androgen glucuronidation [272]. Similar to CDCA, LCA has antiproliferative effects in prostate cancer and induces apoptosis, endoplasmic reticulum stress, autophagy and mitochondrial dysfunction [9,273] (see Table 6). UDCA induces death receptor-mediated apoptosis in human prostate cancer cells [274] (Table 5).

Ovarian cancer
In the serum of ovarian cancer patients, 3b-hydroxy-5-cholenoic acid, GUDCA, DCA and TCDCA levels decreased [275,276]; importantly, taurochenodeoxycholic acid levels decreased in early-stage epithelial ovarian cancer [276]. Zhou and colleagues have shown that sulfolithocholylglycine and TCA showed changes in the serum of ovarian cancer patients [277]. Changes to the BA pool are so characteristic that Guan and colleagues suggested [278] a set of 12 BAs, including glycolithocholic acid, to be used as markers to separate healthy controls from ovarian cancer patients.
The available studies assessed the effects of BAs at supraphysiological concentrations. These concentrations of BAs are cytotoxic and induce apoptosis likely due to changes to membrane damage [279,280] that is unlikely at physiological concentrations of BAs [7]. DCA can modulate the expression of breast cancer type 1 susceptibility protein (BRCA1) and the estrogen receptor and, through these, can control drug sensitivity of ovarian cancer cells (Table 6) [281]. Furthermore, cholylglycinate interferes with the transport of cisplatin [282] and TCDC sensitizes ovarian carcinoma cells to doxorubicin and Mitomycin [280].
LXR [283][284][285], PXR [286], VDR [287][288][289][290][291][292][293][294][295][296] or CAR [297,298] activation was shown to exert protective features against ovarian cancer, similar to BA-elicited effects suggesting that BAs may have a more profound role in protecting against ovarian cancer. These protective effects involved the suppression of proliferation [283,284,286], invasion [291], EMT [288], de novo fatty acid biosynthesis [295], the proportions of the cancer stem cell population [289], and the improvement of the efficacy of chemotherapy [285,297,298] culminating in better patient survival [292,293]. Conflicting with these observation on report provided evidence that under certain conditions PXR may support proliferation [299]. BAs can influence the expression and the activity of multiple PARP enzymes [300]; therefore, it is likely that BAs could modulate the efficacy of PARP inhibition that is a novel modality in the chemotherapy of ovarian cancer.

Conclusions
Primary and secondary BAs are long-standing players in carcinogenesis. Although these molecules were considered as initiators of neoplasias, recent advances have shown that the pro-or anticarcinogenic activity of BAs varies among neoplasias [301], most probably due to differences in the expression of BA receptors, transporters and cell-specific differences in the outcome of receptor activation. Key pathways activated in neoplasias by BAs are regulated by nuclear receptors, FXR, CAR, SHP, PXR, LXR and VDR and other membrane receptors such as S1PR2, TGR5, CHRM2 and CHRM3. They activate numerous downstream signaling pathways such as EGFR, STAT3, MAPK, HNF4α, NF-κB, TLR4, SOCS3 and β-catenin just to name some. Furthermore, BAs regulate all aspects of tumor development and progression, the EMT, invasion, metabolism, apoptosis, proliferation, senescence, immune environment and response to chemotherapy.
The effect of BAs on neoplasias also depends on the concentrations used in the studies. While in certain models BAs in low concentration have anti-cancer effects, in superphysiological concentrations BAs have pro-cancer effects. This phenomenon is related to their amphipathic structure and the activation of additional off-target pathways not tiggered at physiological concentration. At high concentrations, BAs may perturb membranes and activate signaling pathways that sense disturbance of membranes, such as PLA2 and PKC. At high concentrations, they are also toxic and activate the detoxifying pathways, which regulate the activity of transporters of steroid hormones and chemotherapeutics. Therefore, we would urge the community to carry out studies where the concentrations of BAs correspond to the reference concentrations established for the tissue or, as a proxy, to the serum reference concentrations. As a continuation of that, in the case of UDCA the therapeutic serum concentrations can also be used as a guide. These data are summarized in Table 1. Such studies would be invaluable to understand the (patho)physiological roles of BAs and would give a good frame for the therapeutic applicability.
Along the same lines, it is apparent that BAs can be considered as possible treatment options in certain cancers. Foremost, UDCA, that is a therapeutically available drug, has beneficial effects in multiple neoplasias (e.g. [227,248,302], Table 5) pointing towards the possibility for repurposing UDCA. The picture for other BAs is hazier due to frequent contradictions making it hard to outline applicability. However, before the application of BAs in neoplasias we would need to decipher the cross-talk between BAs and drug metabolism, the effect on drug efficacy and drug availability, and discover the possible adverse effects of BAs, that is currently largely missing. Moreover, it is tempting to consider the manipulation of the intestinal microbiome to affect the levels of selected secondary bile acids in humans. Finally, the modulators of BA receptors should be considered as therapeutic options as well. Given the emerging evidence on the potential anti-cancer effects of BAs, further studies are vital in order to develop novel therapeutic strategies using BAs.

Search strategy and selection criteria
References to this review were identified through the prior knowledge of the authors that was complemented by systematic search of PubMed by using the combinations "Prostate cancer AND (bile acid)", "Gastric cancer AND (bile acid)", "Hepatocellular carcinoma AND (bile acid)", "Oesophageal cancer AND (bile acid)", "(bile acid) receptors AND cancer", "(bile acid) receptors AND prostate cancer", "(bile acid) receptors AND gastric cancer", "(bile acid) receptors AND hepatocellular carcinoma", "(bile acid) receptors AND oesophageal cancer", "(bile acid) AND ABC AND transporter", "(bile acid) AND SLC AND transporter", "(bile acid) AND SLCO AND transporter", "(bile acid) AND transport AND review", "Farnesoid X receptor (FXR) AND the cancer types assessed in the study", "Pregnane X receptor (PXR) AND the cancer types assessed in the study", "Constitutive androstane receptor (CAR) AND the cancer types assessed in the study", "Vitamin D receptor (VDR) AND the cancer types assessed in the study" "Liver X receptor (LXR) AND the cancer types assessed in the study", "Small heterodimer partner (SHP) AND the cancer types assessed in the study". Articles published in English were included with no restriction on publication date. All references were checked at Pub Peer, two papers were flagged ( [215] and [156]), but when reviewing the reports we decided that the issues raised do not impact on the main message and kept the references.
Acknowledgements Not applicable. Availability of supporting data Not applicable.

Conflict of interests
The authors declare no conflict of interest.
Ethical approval and Consent to participate Not applicable.

Consent for publication All authors aggreed on the publication of the current version of manuscript.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.