BSH and bile acids: from microbial deconjugation to receptors and clinical evidence

“Bile salt hydrolase” can sound like the name of an enzyme that makes deoxycholic acid and lithocholic acid. In fact, bacterial BSH cleaves the amide bond between a bile acid and glycine or taurine. Removing the steroid nucleus’s 7α-hydroxyl group is a separate, multistep activity of a narrow group of bacteria carrying functional bai genes. Once BSH releases a free bile acid, it may enter that pathway, undergo epimerization, be absorbed, or leave the body. The dominant route depends on the host, microbial community, substrate availability, and location of the reaction—not on bsh alone.

This is a narrative review of enzyme chemistry, microbial genetics, receptor physiology, and clinical conditions. On 9 October 2026, targeted checks revisited primary studies on BSH structure and distribution, BSH inhibitors, bai-dependent secondary-acid formation, colonization resistance, and trials in bile acid diarrhea. The review did not use an exhaustive search or formal risk-of-bias assessment; no GRADE assessment or external peer review was performed. Findings from purified enzymes, gnotobiotic mice, healthy volunteers, and patients with diarrhea are kept distinct because they cannot be transferred automatically from one setting to another.

The host pool: synthesis, conjugation, and return

The principal human primary bile acids are cholic acid (CA; hydroxyl groups at 3α, 7α, and 12α) and chenodeoxycholic acid (CDCA; 3α and 7α). The hepatic classical pathway begins with CYP7A1 in the endoplasmic reticulum: cholesterol becomes 7α-hydroxycholesterol, then HSD3B7 forms 7α-hydroxy-4-cholesten-3-one, also called C4. CYP8B1 controls addition of the 12α-hydroxyl group and thereby shifts synthesis toward CA; without this step, the relative contribution of CDCA is higher. The alternative pathway begins with side-chain oxidation by mitochondrial CYP27A1 and also involves CYP7B1. This is a network of compartments and controls, not a direct hepatic “conversion of cholesterol to DCA.” Fecal bile-acid loss supports new synthesis from cholesterol, but there is no fixed contribution of each pathway or universal CA:CDCA ratio for every person.

Before secretion, the liver activates the acid’s carboxyl group through SLC27A5/BACS at the cost of ATP, and BAAT transfers the acyl group to glycine or taurine. This is amide conjugation, not sulfation. The two overall stages for cholic acid can be written as:

$$\mathrm{CA+CoA+ATP\xrightarrow{BACS}CA{-}CoA+AMP+PP_i}$$ $$\mathrm{CA{-}CoA+glycine\xrightarrow{BAAT}GCA+CoA}$$

The taurine product is taurocholate (TCA); CDCA yields GCDCA or TCDCA. In the intestinal milieu, the carboxyl group of glycine conjugates and the sulfonate of taurine conjugates are strongly ionized. Their amphipathic surface forms mixed micelles with phosphatidylcholine and lipolysis products, which help carry lipids to the mucosa. Solubility, critical micelle concentration, and membrane effects depend on pH, salts, phospholipids, and the mixture. A CMC measured for a purified free acid cannot simply be applied to intestinal contents, and a hydrophobic metabolite is not a “toxin” at every concentration.

BSEP/ABCB11 exports conjugated salts across the hepatocyte canalicular membrane using ATP. MDR3/ABCB4 supplies phosphatidylcholine to bile, while ABCG5/ABCG8 contributes to cholesterol secretion. After a meal, salts enter the small intestine and are recovered mainly in the terminal ileum: apical ASBT/SLC10A2 uses the sodium gradient; FABP6/IBABP binds bile acids in the enterocyte; and basolateral OSTα–OSTβ carries them into portal blood. NTCP/SLC10A1 and OATPs mediate hepatic uptake of several conjugated and unconjugated forms. Returned acids may be conjugated again. Delivery to the colon depends on secretion, transit time, ileal function, and microbial conversion. “95%” is a rule-of-thumb for normal circulation, not a fixed fraction in a person after ileal resection.

In a healthy cycle, a relatively small pool passes through the intestine repeatedly each day. Thus the mass secreted after several meals is not the pool size and is not the amount lost in stool. This matters when interpreting “increased BSH”: deconjugation can shift the site of absorption, alter receptor signaling, and change loss, but these three masses cannot be calculated from one fecal test. Disease or resection of the terminal ileum impairs preferential active uptake of conjugates, although free acids can still be absorbed passively elsewhere. Chemical conversion and transport create different exposures in the small and large intestine.

What bacterial BSH actually does

BSH, or choloylglycine hydrolase (EC 3.5.1.24), belongs to the N-terminal nucleophile family of amidases. In the characterized Clostridium perfringens structure, the protein forms a tetramer, and its active site positions both the steroid moiety and substrate amide bond. The numbering of its catalytic N-terminal cysteine depends on whether the initiator methionine is removed, so Cys1 and Cys2 in different descriptions may refer to the same functional residue. Its thiol attacks the amide carbonyl. After a tetrahedral intermediate, glycine or taurine leaves and an acyl-enzyme remains; water hydrolyzes the thioester and releases the bile acid. The net reaction does not remove a hydroxyl group from the steroid rings:

$$\mathrm{GCA+H_2O\xrightarrow{BSH}CA+glycine}$$ $$\mathrm{TCDCA+H_2O\xrightarrow{BSH}CDCA+taurine}$$

The structural study of C. perfringens showed recognition of the steroid and amino-acid portions; it does not establish one set of $K_m$ and $k_{cat}$ values for every BSH. Isoenzymes from different strains prefer different combinations of steroid nucleus, glycine, and taurine. For example, a biochemical study of purified Bacteroides thetaiotaomicron BSH found hydrolysis of TUDCA but not TCA; it would be incorrect to generalize that every Bacteroides BSH removes taurine faster. $K_m$ values are measured at specified pH, temperature, substrate, and assay conditions, such as detection of released amino acid. $k_{cat}/K_m$ compares enzymes only when conditions are comparable. Fecal bsh copy number is not a measure of the rate of hydrolysis of each bile salt in the ileum.

A typical kinetic experiment uses purified recombinant protein, a defined concentration of one conjugate, and sampling during the near-initial-rate interval. A ninhydrin assay detects released amino acid; chromatography with mass spectrometry can distinguish substrate from free bile acid. Both need controls: free glycine may already be present in bacterial medium, some isomers are difficult to resolve without standards, and substrate loss can reflect adsorption or downstream conversion. Several competing salts are present at millimolar concentrations in the intestine, so Michaelis–Menten hydrolysis of one substrate does not describe the whole mixture. The fraction of active protein, N-terminal processing, bile-acid concentration, and water availability determine whether a laboratory rate can be transferred to intestinal contents.

Genes with BSH function occur across bacterial lineages. Jones and colleagues’ functional metagenomics found broad and redundant distribution and showed that BSH contributed to bile tolerance in some bacteria. A homologous gene without a substrate-specific assay does not prove activity, and a strain-level property cannot be assigned to the whole genus Lactobacillus, Bifidobacterium, or Bacteroides. Enzyme location—inside the cell, at the surface, or released—changes access to bile salts. Gene presence, transcript, active protein, and final bile-acid profile are distinct measurements. Unsupported percentages such as “80% of bacteria have BSH,” or a universal four-type scale of substrate preference, do not establish an individual’s functional state.

Deconjugation may help a microbe tolerate bile, but the direction of the effect depends on the chemical environment: at one pH a free acid may precipitate more readily; at another it may enter membranes and become more toxic. Released taurine may feed other organisms, but its release does not mean that every BSH-positive strain itself uses taurine as an electron acceptor. The microbial community, mucus, host bile acids, and diet jointly shape the result. “High BSH always protects the gut” is as simplistic as “BSH always makes a toxin.”

Deconjugation opens several microbial routes

Free CA and CDCA can serve as substrates for bacteria carrying out 7α-dehydroxylation, but substrate availability does not make conversion inevitable. A narrow group of anaerobes with functional bai genes, including the studied Clostridium scindens, performs oxidative, dehydration, and reduction steps. After transport of free acid, activation by the BaiB CoA ligase, and modification of the 3α-hydroxyl by BaiA, the pathway forms a conjugated unsaturated intermediate and removes 7α-OH with the BaiE dehydratase. Further reductases and CoA transfer restore a saturated structure. The precise order of intermediates and roles of individual Bai proteins vary by organism and experiment; a clinical stool test does not reveal each step. The net chemistry is safer to show than an unsupported detailed arrow-by-arrow scheme:

$$\mathrm{CA\xrightarrow{bai\ pathway}DCA}\qquad\mathrm{CDCA\xrightarrow{bai\ pathway}LCA}$$

DCA loses CA’s 7α-hydroxyl but retains 12α-OH; LCA comes from CDCA and retains only 3α-OH. Both are called secondary bile acids. Without a bai carrier or suitable growth conditions, BSH merely creates free primary acids. Even when both gene systems are present, transit time, substrate amount, pH, oxygen, competing microbes, and absorption matter. Therefore “inhibiting BSH completely switches off DCA” and “a small fraction of bai bacteria converts almost the entire pool” are not universal physiological laws.

Hydroxysteroid dehydrogenases (HSDHs) support different branches. A pair of 7α-HSDH and 7β-HSDH enzymes can invert CDCA’s hydroxyl configuration through a 7-oxo intermediate to form ursodeoxycholic acid (UDCA). A pair of 3α/3β-HSDHs makes 3β epimers such as isoDCA. Devlin and Fischbach described isoDCA formation by Ruminococcus gnavus and other organisms; in experiments, this changed the effect of DCA on Bacteroides growth. Experiments with individual metabolites have also found effects on T-cell differentiation. This does not mean that the entire pool of secondary acids is “anti-inflammatory” or that an HSDH-positive strain treats colitis.

HSDH reactions depend on oxidized and reduced NAD(P)⁺/NAD(P)H cofactors, reaction direction, and hydroxyl position. Removal of an oxygen through bai and α-to-β inversion through a ketone intermediate have different atom balances. UDCA should not be called a “hydrophilic BSH product”: BSH may release its substrate, but other enzymes carry out the stereoinversion. Likewise, isoDCA is not synonymous with a lower total amount of DCA: the hydroxyl configuration has changed, potentially altering bacterial or receptor interactions.

Free acids and conjugates differ in charge, solubility, and ability to bind calcium or mixed micelles. Yet a simple statement that “low pH precipitates all DCA with calcium” ignores actual concentrations, ionic composition, dietary fat, and heterogeneity of intestinal contents. The balance between dissolved and bound fractions can be studied, but it cannot establish an individual’s mucosal injury risk from stool pH alone.

From molecule to signal: FXR, TGR5, and protective responses

Bile acids bind receptors as well as lipids. FXR/NR1H4 is a nuclear transcription factor in intestine and liver. Ligand binding changes its interaction with RXR, coactivators, and DNA response elements. CDCA is a strong natural agonist in several cell systems; responses to other acids depend on species, conjugation, membrane transport, and assay system. A list of $EC_{50}$ values cannot predict that raising DCA will switch the whole body “from FXR to TGR5”: both signals may occur at once in different cells and concentrations.

In ileal enterocytes, FXR stimulates FGF19 production in humans (Fgf15 in mice). Portal FGF19 signals through hepatic FGFR4/β-Klotho to suppress CYP7A1 transcription and new bile-acid synthesis. Hepatic FXR also affects export, conjugation, and the SHP/NR0B2 regulator; a change in CYP7A1 cannot be assigned to one pathway alone. If the terminal intestine is damaged or FGF19 feedback is insufficient, the liver may overproduce acids that reach the colon. This mechanism matters in bile acid diarrhea. In Walters’ small study, 25 mg of obeticholic acid daily for two weeks increased FGF19 and reduced C4 in 10 people with primary bile acid diarrhea, with improvement in some symptoms; there were also 10 participants with secondary diarrhea and 8 with idiopathic diarrhea. This short open-label study without placebo does not establish FXR stimulation as a universal treatment.

TGR5/GPBAR1 is a membrane receptor coupled to $G_s$. Activation raises cAMP and initiates PKA and other effector pathways. DCA and LCA can strongly activate TGR5 in model systems. In enteroendocrine L cells this has been linked to GLP-1 release; effects on inflammatory signals, including NLRP3, have been described in macrophages. GPBAR1 studies show metabolic and immune possibilities, but a higher fecal metabolite does not measure receptor occupancy on the basolateral membrane of an L cell. GLP-2 release, restoration of intestinal villi, and diabetes prevention are not inevitable links in a chain from BSH.

In Watanabe’s mouse experiments, bile acids activated TGR5–cAMP–DIO2: the deiodinase converted T4 to T3, increasing tissue thermogenesis. The model, acid availability, and species-specific brown adipose tissue matter. It does not follow that a BSH-containing supplement will meaningfully increase energy expenditure or UCP1 expression in humans. LCA can also activate VDR and PXR, which are linked to bile-acid biotransformation and clearance. LCA sulfation and other pathways reduce exposure to the free hydrophobic molecule; actual flows differ among intestine, liver, and blood.

Mice are particularly useful for FXR mechanisms, but their bile-acid pool is rich in muricholic acids, which are not major human bile acids. In Sayin’s study, tauro-α- and tauro-β-muricholates acted as FXR antagonists in mice; microbial conversion changed ileal Fgf15 and hepatic Cyp7a1. This mechanism cannot be transferred as “TUDCA is the human equivalent of TβMCA and a required FXR antagonist.” Yao studied a B. thetaiotaomicron BSH with selectivity for certain substrates; genetic deletion in colonized mice changed their bile-acid profile and body weight. This was a mutant in an animal model, not a clinical trial of a BSH-inhibiting drug in people.

Why “secondary acid” does not mean “poison”

Removing 7α-hydroxyl generally increases hydrophobicity: free DCA and especially LCA can disrupt membranes and intracellular organelles at sufficiently high bioavailable concentrations. In cell culture, DCA has caused mitochondrial oxidative stress and changes in NF-κB, apoptosis, or proliferative signaling depending on cell line and dose. Experiments involving β-catenin and other pathways offer mechanisms to investigate; a tumor-cell culture does not measure a safe amount of dietary fat in humans or prove an independent causal role for fecal DCA in colorectal cancer.

Dose and molecular form determine exposure. One must know whether the measured acid is free or conjugated, dissolved in the aqueous phase, incorporated into mixed micelles, bound to calcium, and how long it contacts the epithelium. After absorption, the liver reconjugates some DCA; LCA may be sulfated and excreted. A “universal 0.2 mM lysis threshold” for the living intestine cannot be inferred: a number from a model medium omits mucus, phospholipid protection, and dynamic fecal loss. At the same time, the physiological presence of DCA does not rule out harm with a different exposure or tissue susceptibility.

Experimental studies link DCA to reactive oxygen species, mitochondrial potential, caspases, β-catenin, EGFR/MAPK, and NF-κB. These events need not form one linear “path to cancer”: under some conditions a cell dies; under others it survives and changes proliferation. Epidemiological associations between secondary acids, diet, or neoplasia are difficult to separate from diet composition, transit, inflammation, microbial community, and reverse causation. A metabolite observation, mechanistic plausibility, and evidence that a specific BSH inhibitor prevents cancer are three different levels of evidence.

Secondary acids can also have opposing effects. In C. difficile cultures, taurocholate with glycine triggered spore germination; DCA could also stimulate germination while inhibiting subsequent vegetative growth. In Buffie’s work, loss of the 7α-dehydroxylating C. scindens was associated with susceptibility after antibiotics, and experimental introduction of the bacterium strengthened protection with secondary acids involved. Thus “remove all DCA to improve the microbiome” could miss protection against infection. This is not a reason to seed people with C. scindens or treat infection with a BSH capsule: a model of colonization resistance is not a clinical trial.

In bile acid diarrhea, the key question is often different: too many bile acids reach the colon because of terminal ileal disease or resection, or inadequate FGF19 feedback with excessive synthesis. Water secretion, accelerated transit, and receptor effects may contribute to symptoms; the effect of one DCA does not establish the diagnosis. “Malabsorption” does not fully cover the primary form driven by excess synthesis. The clinical setting differs after cholecystectomy, with some intestinal diseases, and with certain medicines. A BSH profile cannot automatically assign a “type 1–3” diagnosis without assessment.

Free CDCA and DCA in the colon may increase secretion and motility, but excess incoming conjugated salts also matter. Excretion changes with stool volume: accelerated transit itself changes the time available for BSH and bai; it does not only arise as a result of bile acids. Causality runs both ways. A high fecal level of one acid may reflect greater delivery, reduced absorption, or less downstream conversion. Clinical assessment starts with the pattern of chronic diarrhea and causes of ileal injury, not antimicrobial treatment aimed at “excess BSH.”

Bile acids are being studied in steatohepatitis and the intestinal vascular barrier in human profiles and models. Evidence on cultured endothelium, PV-1, cytokines, and LCA does not establish a human chain of “BSH → leaky vascular barrier → MASH/metastases.” Exposure in tissue and blood differs from fecal exposure; a lower fecal abundance of bai genes is not prevention of liver fibrosis. Liver disease and infection cannot be assessed with one universal measure of “toxic secondary bile acids.”

Interventions: where a human outcome exists

Bile-acid binding. Cholestyramine and colesevelam are nonabsorbed polymer sequestrants: positively charged groups bind bile-acid anions and increase fecal loss. They act on the host pool, not specifically on BSH’s active site. In bile acid diarrhea they may reduce the acid load available to colonocytes, but response depends on tolerability, dose, and the correct diagnosis. In Camilleri’s 10-day open study of diarrhea-predominant IBS with bile-acid malabsorption, colesevelam was given at 1,875 mg twice daily; 48-hour fecal acids, C4, and stool measures were recorded. Without a blinded control, observed changes cannot be attributed to all patients; even established benefit from a sequestrant would not show that excess BSH caused the problem. Resins can interfere with absorption of other medicines and sometimes cause constipation; the specific product and clinical context determine dose and spacing.

UDCA and TUDCA. Ursodeoxycholic acid is the 7β epimer of CDCA. Prescription UDCA is used for specific cholestatic indications; doses around 13–15 mg/kg/day are used depending on diagnosis and product. This is not a general “DCA detox” dose for healthy people. UDCA/TUDCA experiments describe changes in apoptosis, mitochondria, and endoplasmic-reticulum stress, but cellular protection does not mean treatment of every case of bile-acid diarrhea, MASH, or colorectal risk. UDCA enters the circulating pool, is conjugated, and may be modified by microbes. Therapy requires an indication and monitoring, particularly with biliary obstruction or serious liver disease.

BSH-positive probiotics. A strain that hydrolyzes GCA on a plate will not necessarily persist in the small intestine, alter fecal bile-acid loss, lower LDL, or improve a clinical outcome. Probiotic studies often combine BSH with cholesterol binding, dietary changes, and other strain properties. A species name cannot establish a uniform effect, and “hyperactive BSH” is not a universal lipid-lowering therapy. For a specific product, the strain, viable dose, duration, baseline LDL, comparator, and measured clinical outcome matter. See also the discussion of the difference between strain properties and proven outcomes in probiotics and safety.

Experimental BSH inhibition. Yao and colleagues altered the BSH gene of B. thetaiotaomicron in a colonized-mouse model and observed changes in bile-acid profiles and metabolic measures; this was not pharmacological inhibition. Adhikari and colleagues later developed a covalent chemical probe with a fluoromethyl ketone group that targets catalytic Cys. A single dose in mice reduced fecal BSH activity and the fraction of deconjugated acids, but the paper did not demonstrate prevention of obesity, steatosis, or diabetes in people. Long-term effects on pathogen resistance, microbial ecology, and lipid absorption require study. Suppressing BSH solely to minimize DCA would be too narrow a goal.

Diet, fiber, and polyphenols. In David and colleagues’ brief human comparison, ten volunteers followed animal-based or plant-based diets; the animal-based diet changed microbial transcripts, increased fecal bile acids and DCA, and was accompanied by an increase in Bilophila. This short exposure did not measure future cancer and does not prove that every high-animal-fat diet raises DCA three- to fivefold. Soluble dietary fiber may affect the bile-acid pool through binding, transit, and fermentation; the result depends on type and dose. The rationale for adequate dietary fiber extends beyond a BSH hypothesis. Berberine, curcumin, and silymarin act on many targets in experimental systems; there is no basis to prescribe them as proven selective BSH or FXR regulators based on one fecal ratio. The pharmacology of berberine helps distinguish these mechanisms from clinical claims.

What to measure and which question to ask

LC–MS/MS can distinguish CA, CDCA, their glycine and taurine conjugates, DCA, LCA, UDCA, epimers, and sulfates. Blood, ileal contents, and 48-hour stool reflect different flows. Time since food, antibiotics, sequestrants, transit, sampling method, and analytical calibration alter results. The ratio of free to conjugated forms is only an indirect marker because free acid may be absorbed, converted further, or missed by the sample. A DCA+LCA to CA+CDCA ratio also mixes BSH, bai, absorption, and hepatic feedback; “>80% = carcinogenic risk” is not a validated clinical threshold.

When bile acid diarrhea is suspected, relevant clinical measures may include SeHCAT retention where available, 48-hour fecal bile acids, fasting serum C4 as a synthesis surrogate, and FGF19 interpreted according to the assay. No marker equals direct BSH activity. C4 varies with time of day, liver state, and medicines, so a number such as 47 ng/mL without laboratory context cannot be applied to every method or population. When liver or colorectal disease is in question, the appropriate clinical pathways remain decisive, not a home “dehydroxylation index.”

The practical conclusion from this biochemistry is not to maximize or suppress BSH. Specify which acid changed, in which compartment and concentration, how it was measured, and which human outcome has actually been shown. BSH opens the route from conjugates to free acids; bai and HSDH make other molecules; host receptors and transporters add another layer of response. Keeping these boundaries intact makes bile-acid profiling a useful research tool without turning it into a standalone diagnosis or universal supplement plan.