Microbiota-derived bile acid metabolic enzymes and their impacts on host health

Abstract

Bile acids are amphipathic sterol molecules regulated by both the host and gut microbiota, serving as classical mediators for deciphering host-microbiota interactions. Synthesized primarily in the liver and undergoing extensive structural modifications along the gastrointestinal tract, bile acids are dynamically shaped by diverse bile acid metabolic enzymes, especially from gut microbiota. Beyond their canonical detergent-like functions, bile acids act as receptor modulators, immune regulators, and microbiota sculptors, profoundly involved in regulating host metabolic processes, maintaining immune homeostasis, and contributing to metabolic disorders when dysregulated. The modifications of bile acids by microbial enzymes critically influence their functional diversity. However, despite the vast array of bile acid modifications observed, significant gaps remain in the systematic identification and characterization of microbial bile acid metabolic enzymes. This review underscores the urgency of exploring the biosynthetic pathways for the production of key bile acids and highlights its potential to advance precision therapeutic strategies targeting gut microbiota and their enzymatic machinery.

1. Introduction

As the terminal products of cholesterol catabolism, bile acids (BAs) are amphipathic molecules characterized by their tetracyclic steroid cores conjugated with the carboxy tail. The panorama of mammalian BA pool is jointly shaped by host hepatic synthesis and gut microbial metabolism, which comprises primary BAs (host-derived) and secondary BAs (microbiota-processed), though this dichotomy is increasingly blurred (Molinaro et al., 2018). BA pool exhibits remarkable structural diversity with concentrations spanning from picomoles (pM) to hundreds of millimoles (mM). Recent advancements in high-resolution mass spectrometry, development of reverse metabolomics and novel bioinformatic pipelines, have expanded the catalog of known BA derivatives from ∼20 classical forms to over 1000 structurally distinct variants (Gentry et al., 2024; Mohanty, Allaband, et al., 2024). Paradoxically, the explosive growth in BA structural characterization contrasts starkly with the stagnation in metabolic enzyme discovery — only a few kinds of microbial bile acid metabolic enzyme-associated biotransformation (conjugation/deconjugation, 7α-dehydroxylation, epimerization, oxidation/reduction, 3-acylation and 3-sulfation) have been mechanistically characterized to date (Guzior & Quinn, 2021; Mohanty, Allaband, et al., 2024; Ridlon & Gaskins, 2024).

Historically recognized as biological detergents facilitating lipid absorption and regulators of cholesterol homeostasis (Kuipers et al., 2014), the physiological significance of BAs has underwent a conceptual revolution, following a seminal discovery that some BAs serve as endogenous ligands for the farnesoid X receptor (FXR), a nuclear receptor previously classified as an orphan receptor (Makishima et al., 1999). This work precipitated a renaissance in BA-related research, fundamentally expanding our understanding of their systemic roles beyond digestive physiology. Contemporarily, BAs have evolved into a diverse molecular repertoire that not only facilitates lipid emulsification but orchestrates systemic regulation of host metabolism, immunity, and signaling via activation of a series of nuclear receptors and membrane receptor networks (Cai et al., 2022; Lee et al., 2024; Sun et al., 2021). BA pool disorder is causally linked to inflammatory bowel disease (IBD), obesity, type 2 diabetes mellitus (T2DM), metabolic dysfunction-associated steatohepatitis (MASH, formerly known as NASH) and polycystic ovary syndrome (PCOS) (Collins et al., 2023; Di Lorenzo et al., 2023; Jia et al., 2018). Acting as dynamic liaisons between the gut microbiota and distal organs, BAs constitute a critical nexus for inter-organ crosstalk, serving as indispensable mediators in maintaining host metabolic-immune homeostasis.

In this review, we will systematically elaborate on the functional advancements of bile acid metabolic enzyme systems, with an emphasis on the use of emerging methodologies to identify novel bile acids and their corresponding synthetases. We further provide a comprehensive overview of the multi-dimensional mechanisms of BA action in maintaining host health, including their downstream signal transduction pathways, modulation of the gut microbiota, and their influence on metabolic diseases.

2. Synthesis and transformation of BAs

The biosynthesis of BA originates from cholesterol and is mediated by the cytochrome P450 enzyme family through two major pathways: the classical (neutral) pathway and the alternative (acidic) pathway (Molinaro et al., 2018). The classic bile acid synthesis pathway is initiated by the rate-limiting enzyme cholesterol 7α-hydroxylase (CYP7A1), which converts cholesterol into 7α-hydroxycholesterol (Schwarz et al., 1996). Subsequent enzymatic steps involving sterol 12α-hydroxylase (CYP8B1) and sterol 27-hydroxylase (CYP27A1) lead to the production of cholic acid (CA) and chenodeoxycholic acid (CDCA). In addition to the classic pathway, CDCA can also be synthesized via an alternative pathway that begins with CYP27A1-mediated hydroxylation of cholesterol to 27-hydroxycholesterol, followed by further conversion to CDCA catalyzed by 24-hydroxycholesterol-7α-hydroxylase (CYP7B1) (Fuchs & Trauner, 2022). The ratio of CA to CDCA is primarily determined by CYP8B1, which is essential for CA synthesis (Molinaro et al., 2018).

In mice, however, the presence of cytochrome P450 2C70 (CYP2C70) enables additional bile acid synthesis, producing α- and β-muricholic acids (α/β-MCA) from CDCA and ursodeoxycholic acid (UDCA) (de Boer et al., 2020). Bile acids synthesized from cholesterol in the liver are referred to as primary bile acids. Within the body, these bile acids undergo conjugation with glycine (G) or taurine (T) to form conjugated BAs (GCA, GCDCA, TCA, TCDCA and so on), through the action of bile acid Coenzyme A: amino acid N-acyltransferase (BAAT). These primary BAs are stored in the gallbladder and released into the intestinal lumen when food consumption to aid in the digestion and absorption of dietary lipids (Molinaro et al., 2018). Approximately 95% of BAs are reabsorbed in the jejunum and ileum, returning to the liver via the portal vein in a process known as the enterohepatic circulation (Dawson & Karpen, 2015). This efficient recycling mechanism ensures minimal loss of BAs while maintaining their critical role in lipid metabolism and systemic homeostasis. Approximately 5% of BAs reach the colon, where they undergo diverse microbial modifications—such as conjugation/deconjugation, dehydroxylation, oxidation/reduction, epimerization, sulfation, and acylation—to generate numerous secondary BAs (Guzior & Quinn, 2021). The human secondary BA pool is remarkably diverse, with concentrations of different BAs spanning an extraordinary range from pM to mM. While deoxycholic acid (DCA) and lithocholic acid (LCA) usually dominate this pool (Sinha et al., 2020), a subset of BAs likely remains undetected due to current analytical challenges or technical limitations in identification and quantification.

It is noteworthy that the distinction between primary and secondary BAs is not entirely clear. Emerging research has revealed that gut microbiota harbor BAAT microbial-host-isozymes. For instance, an early study demonstrated that the marine bacterium Myroides sp. strain SM1 can produce GCA and GDCA (Maneerat et al., 2005). Subsequently, the fungus Penicillium was also shown to generate glycine-conjugated BAs (Ohashi et al., 2008). A recent large-scale screening of gut bacteria demonstrated their widespread ability to conjugate primary and secondary BAs with glycine and 15 other amino acids, significantly expanding our understanding of conjugated BAs (Lucas et al., 2021). However, the extent of gut bacterial contribution to the total conjugated BA pool remains an open question.

3. Microbiota-derived bile acid metabolic enzymes

3.1. Bile salt hydrolase/transferase

As an earliest discovered and most extensively studied bile acid metabolic enzyme, bile salt hydrolase (BSH) was firstly isolated and purified from Bacteroides fragilis in 1796, with its pH optimum and substrate specificities primarily depicted (Stellwag & Hylemon, 1976). In its initial understanding, BSH catalyzed the “gateway” reaction in the metabolism of secondary BAs, as it could hydrolyze conjugated BA (e.g., glycine-conjugated or taurine-conjugated) to release the amino acids and free BAs (Fig. 1), allowing for the amino acids utilization and further diverse BA modifications, with optimum pH approximately at 6.0 (Han et al., 2025). BSH demonstrates widespread distribution across all major gut microbiota phyla with its highly conservation, such as Bacillota (or Firmicutes), Bacteroidota (or Bacteroidetes) and Actinomycetota (or Actinobacteria) (Collins et al., 2023; Luo et al., 2025), yet displays marked differences in substrate preference spectrum. For example, Bifidobacterium species encodes three BSH subtypes (A-C) (Kim et al., 2004). Notably, Types A and C display a marked preference for glycine-conjugated BAs, whereas their catalytic efficiency toward taurine-conjugated BAs shows significant divergence (Kim et al., 2004).

Fig. 1.

Summary ofmicrobiota-derivedbile acid metabolic enzymes. The current understanding of bile acid metabolic enzymes remains limited, with only six major classes fully or partially characterized to date: bile salt hydrolases/transferase (BSH/T), the bai operon, hydroxysteroid dehydrogenases (HSDH), 5α/β-reductases (5AR/5BR), sulfotransferases (SULT), and acylated bile acid synthetases (BAS). However, the full spectrum of enzymatic diversity governing bile acid metabolism remains largely unknown. Critical gaps persist in identifying synthetic pathways for some bile acid derivatives, including FA-isoBAs, polydeoxycholate, (poly)amine bile amidates, dimethylated bile acids and so on. This knowledge gap highlights the urgent need for systematic exploration of novel bile acid metabolic enzymes. GCA, glycocholic acid; TCA, taurocholic acid; CA, cholic acid; MCBA, microbially conjugated bile acids; DCA, deoxycholic acid; LCA, lithocholic acid; CA7S, cholic acid-7-sulfate; LCA-S, lithocholic acid-3-sulfate; FA-isoBAs, fatty acid-conjugated isobile acids. This figure was created with BioRender.com.

From an evolutionary perspective, BSH in gut-associated species evolved from ancestral Ntn_CGH-like family proteins, a process driven by selection pressure imposed by conjugated BAs (Jones et al., 2008). For some bacteria, this adaptation facilitates survival by reducing BA toxicity (Jones et al., 2008).

Beyond its canonical deconjugation function, recent studies have unveiled a novel N-acyltransferase activity of BSH, enabling the transfer of amino acids and/or polyamines to the C24 position of CA, thereby generating microbially conjugated BAs (MCBAs) (Guzior et al., 2024; Quinn et al., 2020; Rimal et al., 2024). This discovery demonstrates BSH's bidirectional regulatory capacity in BA conjugation/deconjugation, blurring the precise origins of conjugated BAs (host vs. microbial). Consequently, the conceptual framework of BSH is being revised to BSH/T (bile salt hydrolase/transferase). Notably, as a bifunctional enzyme, BSH/T's balance between hydrolytic and conjugative activities can profoundly reshape the host's conjugated-to-free BA ratio, with significant physiological implications (Fu et al., 2025).

3.2. Bile-acid-induced operon

Another classical microbial transformation of BAs is the 7α-dehydroxylation pathway, through which primary BAs (CA and CDCA) undergo hydroxyl group elimination at the C7 position to generate secondary BAs (DCA and LCA), respectively. The discovery of bile-acid-induced (bai) operon—the genetic locus responsible for this transformation—took 40-year joint efforts in research. While the initial discovery dates back to the 1980 with a study demonstrating 7α-dehydroxylation of CA to DCA by a Bacillota bacterium Clostridium scindens VPI 12708 (Eubacterium species V.P.I. 12708 at that time) (White et al., 1980), the complete enzymatic pathway remained elusive for decades. It was not until 2020 that an elegant work systematically elucidated this biochemical pathway through heterologous reconstitution of the bai operon in a related species Clostridium sporogenes, revealing a multi-enzyme cascade requiring strict anaerobiosis, cofactors and ATP (Funabashi et al., 2020).

Current understanding indicates that the bai operon is phylogenetically restricted to a narrow subset of Bacillota members, including Clostridium leptum (Stellwag & Hylemon, 1979), Clostridium hylemonae (Ridlon et al., 2010), Peptacetobacter hiranonis (Kitahara et al., 2001), Faecalicatena contorta (Jin et al., 2022), the murine isolate Extibacter muris (Streidl et al., 2021), etc. Despite their limited taxonomic distribution, these bacterial species exhibit remarkable catalytic efficiency, accounting for nearly all DCA and LCA in the secondary BA pool (Ridlon et al., 2006).

The bai operon comprises eight genes: seven encoding genes (BaiB, BaiCD, BaiE, BaiA2, BaiF, BaiH, BaiI) and one transporter genes (BaiG), with high conservation across nearly all known 7α-dehydroxylating species (Wise & Cummings, 2022). Functional studies reveal that all proteins except BaiI and BaiG are strictly required for the 7α-dehydroxylation of CA (Funabashi et al., 2020). The cascade initiates with BaiB-mediated CoA activation of CA to form cholyl-CoA. This substance subsequently undergoes two sequential oxidation steps catalyzed by BaiA2 and BaiCD to yield 3-oxo-Δ⁴-cholyl-CoA, followed with dehydration mediated by BaiE and CoA deconjugation by BaiF to generate 3-oxo-Δ^4,6-DCA. Ultimately, the reductive arm of this pathway completes the transformation to yield DCA as the terminal metabolite (Funabashi et al., 2020). Notably, the rate-limiting step involves C7 dehydration catalyzed by BaiE, a dedicated 7α-dehydratase (Guzior & Quinn, 2021). Although non-essential for core activity, BaiG enhances bacterial BA uptake (Mallonee & Hylemon, 1996), while BaiI, a putative Δ-ketosteroid isomerase, are hypothesized to process non-BA substrates. Intriguingly, genomic analyses have identified some enzymes—including the baiJKL (Ridlon & Hylemon, 2012) and baiN (Harris et al., 2018)—may be functionally redundant enzymes to participate in this transformation, suggesting evolutionary fine-tuning of this pathway.

The bai operon-mediated 7α-dehydroxylation of CA and CDCA serves as a pivotal bridge converting primary to secondary BAs, representing a critical step in microbial BA modification. This biochemical transformation confers increased hydrophobicity to the resultant BAs while endowing them with distinct receptor regulatory capacities (Fleishman & Kumar, 2024), thereby manifesting profound physiological significance in host physiology.

3.3. Hydroxysteroid dehydrogenases

The microbial epimerization of BAs by hydroxysteroid dehydrogenases (HSDHs) further expands the structural diversity of secondary BAs. Current evidence has revealed epimeric modifications at 3-, 7-, and 12-hydroxyl positions of CA, and 3- and 12-hydroxyl positions of CDCA, with distinct nomenclature: “iso” denoting 3β-hydroxyl configuration, “urso” denoting 7β-hydroxyl, and “epi” denoting 12β-hydroxyl (Doden et al., 2021). This stereochemical interconversion typically involves a pair of position-specific HSDHs. For instance, the conversion of CDCA to UDCA proceeds via coordinated action of 7α-HSDH and 7β-HSDH (Hirano & Masuda, 1981). The reaction initiates with 7α-HSDH-mediated oxidation of CDCA to 7-oxolithocholic acid (7-oxoLCA), followed by NAD(P)⁺-dependent reduction catalyzed by 7β-HSDH to yield UDCA, achieving α→β hydroxyl inversion through sequential oxidoreduction steps (Huang et al., 2019).

HSDHs are predominantly distributed across Bacteroidota, Bacillota, and Actinomycetota phyla (Doden & Ridlon, 2021). An example is the baiA2 gene within the bai operon, encoding 3α-HSDH essential for C7-dehydroxylation reactions (Funabashi et al., 2020). Notably, the HSDH-pairs frequently reside in distinct gut species, suggesting potential interspecies metabolic cooperation within gut microbiome (Fleishman & Kumar, 2024). Particular bacterial strains demonstrate remarkable catalytic versatility, exemplified by Clostridium perfringens (Macdonald et al., 1976), Eggerthella lenta (Devlin & Fischbach, 2015), and some 7α-dehydroxylating species (C. scindens, C. hylemonae, and P. hiranonis) (Baron et al., 1991; Doden et al., 2018), all of which collectively exhibit 3α-, 7α-, and 12α-HSDH activities. Remarkably, E. lenta and Ruminococcus gnavus (possessing both 3α- and 3β-HSDH activities) serve as principal producers of iso-BAs in the gut ecosystem (Devlin & Fischbach, 2015). Emerging evidence suggests that current understanding substantially underestimates HSDH prevalence, with >40% of gut bacterial species harboring at least one HSDH homolog (Jin et al., 2025). Through systematic functional characterization, 59 previously unannotated HSDHs and 56 novel BA stereoisomers have been identified (Jin et al., 2025).

Phylogenetic clustering of HSDHs reveals substantial sequence divergence across enzyme subtypes, with substrate selectivity principally governed by the stereochemical arrangement of hydroxyl groups (Jin et al., 2025). Notably, E. lenta 3α-HSDH demonstrates catalytic activity toward taurine-/glycine-conjugated BAs, thereby challenging the prevailing recognition that microbial metabolism of conjugated BAs strictly requires prior deconjugation by BSHs (Mythen et al., 2018). Mechanistically, microbial species harboring multiple HSDH activities display separate genetic organization (Doden et al., 2018). Unlike the clustered biosynthetic gene clusters (BGCs) typically observed in nature product synthesis, HSDH-encoding genes frequently reside in disparate genomic loci (Doden et al., 2018), suggesting evolutionary acquisition through independent genetic events and possible autonomous regulatory circuits.

3.4. 5α-reductases and 5β-reductases

Utilizing nicotinamide adenine dinucleotide phosphate (NADPH) as hydride donor, steroid 5α-reductases (5AR, also known as 3-oxo-5α-steroid 4-dehydrogenases), catalyze the reduction of the Δ^4,5 bond in Δ⁴-3-ketosteroids (Xiao et al., 2020). 5β-reductase (5-BR) executes similar Δ^4,5 bond reduction but generates distinct stereochemical outcomes through differential hydrogen atom configuration at C5 during the reaction (Ridlon et al., 2016). 5AR was historically characterized for its endocrine functions in androgen metabolism, as it primarily mediated the conversion of testosterone (the most abundant circulating androgen) to 5α-dihydrotestosterone (DHT), the latter one is a more effective androgen (Nacusi & Tindall, 2011).

Recent gerontological research has identified elevated fecal levels of isoallolithocholic acid (isoalloLCA) in centenarians, a secondary BA biosynthesized through collaborative action of 5AR and 3β-HSDH (Sato et al., 2021). The metabolic cascade initiates with 5AR-mediated reduction of the Δ^4,5 bond in 3-oxo-Δ⁴-lithocholic acid (3-oxo-Δ⁴-LCA) to yield 3-oxoallolithocholic acid (3-oxoalloLCA), followed by stereospecific 3β-hydroxylation catalyzed by 3β-HSDH to generate isoalloLCA (Sato et al., 2021). Phylogenetic profiling implicates Parabacteroides, Bacteroides, Alistipes, and Odoribacter species as major 5AR-active commensals, with alkaline pH stress potentially upregulating both enzymatic activity and gene expression in this biotransformation pathway (Sato et al., 2021). Notably, isoalloLCA exerts potent bactericidal activity against multidrug-resistant pathogens including Clostridioides difficile and Enterococcus faecium, and modulates T_reg activity through a hormone receptor nuclear receptor subfamily 4 group A member 1 (NR4A1), suggesting its critical role in pathogen suppression, and host immunity homeostasis maintenance (Li et al., 2021; Sato et al., 2021).

3.5. Sulfotransferase

Nowadays, two predominant sulfation types of BAs have been identified: 3-sulfation and 7-sulfation (Alnouti, 2009; Ito et al., 2024). Cholic acid-7-sulfate (CA7S) and lithocholic acid-3-sulfate (LCA-S) could be generated by host sulfotransferase 2A1 (SULT2A1), though this process is modulated by gut microbiota, as germ-free conditions significantly deplete host CA7S production (Wahlström et al., 2016). Recent studies revealed that Bacteroides thetaiotaomicron encodes a microbial sulfotransferase (BtSULT, BT0416) demonstrating functional homology to host SULT2A1, participating in the production of 3-sulfated BAs (Le et al., 2022; Yao et al., 2022). BT0416 homologs are widely distributed in gut Bacteroidota species (e.g., Parabacteroides merdae), with limited prevalence (<2%) in Bacillota and Pseudomonadota (or Proteobacteria) (Yao et al., 2022). The enzyme exhibits substrate preference for 5α-trans-fused BAs, such as isoalloLCA and allolithocholic acid (alloLCA), while showing no activity towards CA, DCA, CDCA, or UDCA (Yao et al., 2022). One exception for 5α-cis-fused BA is isolithocholic acid (isoLCA), which may undergo sulfation through prior conversion to isoalloLCA (Yao et al., 2022).

Sulfated metabolites were historically regarded as metabolic waste products, but recent advances have challenged this concept (D'Agostino et al., 2024). For instance, LCA-S functions as an antagonistic ligand for retinoid-related orphan receptor γt (RORγt), indicating its roles in mediating host immunity through potential modulation of T helper cells that express IL-17a (T_H17 cells) (Xiao et al., 2022). Concurrently, elevated levels of 3-sulfated BAs observed in cholestatic patients with pruritus symptoms have been mechanistically linked to itch sensation via activation of human Mas-related G-protein coupled receptor member X4 (MRGPRX4) (Yang et al., 2024).

3.6. BA acyl synthetase

Recent studies have identified 3-acylated BAs in fecal samples (Takei et al., 2022), which can be metabolically produced by limited members of the Bacteroidota (e.g., Bacteroides uniformis) and Bacillota (e.g., Christensenella minuta) phyla (Liu et al., 2024). Currently characterized 3-acylated cholic acid derivatives include 3-acetylated CA (3-aceCA), 3-propionylated CA (3-proCA), 3-butyrylated CA (3-butCA), 3-valerylated CA (3-valCA), 3-succinylated CA (3-sucCA), 3-malonylated CA (3-malCA) and so on. However, individual bacterial strains exhibit mutually exclusive biosynthetic capacities for these modifications. For instance, C. minuta preferentially generates monoacid acylated variants, particularly 3-butCA and 3-aceCA (Liu et al., 2024), whereas B. uniformis uniquely produces 3-sucCA. However, only in B. uniformis have identified BA acyl synthetase for succinyl (BAS-suc) (Nie et al., 2024), an enzyme previously annotated as a β-lactamase. These 3-acylated BAs demonstrate potential therapeutic effects against T2DM and MASH via pleiotropic mechanisms, such as receptor modulation and microbiota regulation (Liu et al., 2024; Nie et al., 2024).

3.7. More metabolic enzymes remain undiscovered

While biosynthesis pathways have been fully or partially elucidated for a limited subset of BA modifications, the majority remain unmapped, with low-throughput and inefficient approaches continuing to impede progress in this field. For instance, fecal analyses and microbial culture studies have identified a group of esterified BAs, including polydeoxycholate (Benson et al., 1993), fatty acid-conjugated isoBAs (FA-isoBAs) (Takei et al., 2022), valolithocholate ester (Garcia et al., 2022) and so on. Other uncharacterized modifications include methylation variants (Li et al., 2009) and host-derived N-acetylglucosaminidation (Marschall et al., 1992; Pellock & Redinbo, 2017).

In recent years, the development and application of novel technological workflows have significantly accelerated the discovery of bile acid diversity. Especially, the application of mass spectrometry (MS)-based metabolomics has generated vast amounts of data, providing a valuable resource for uncovering previously unrecognized bile acid derivatives. Building upon this foundation, techniques such as molecular networking (Quinn et al., 2020), reverse metabolomics (Gentry et al., 2024), and Mass Query Language (MassQL) (Mohanty, Mannochio-Russo, et al., 2024) have emerged as powerful tools to mine these complex datasets.

Molecular networking, which leverages the recognition of chemical transformations based on parent mass shifts between related spectra, enables meta–mass shift chemical profiling even without prior knowledge of molecular structures (Quinn et al., 2020). This approach has led to the identification of several amino acid conjugated bile acids and BA-MCY (Quinn et al., 2020; Won et al., 2025). In contrast, another technology, reverse metabolomics, begins with the tandem MS (MS/MS) spectrum of a known molecule of interest and searches public untargeted metabolomics datasets to confirm its presence and uncover potential phenotypic associations (Charron-Lamoureux et al., 2025; Gentry et al., 2024). Recent advancements have enabled systematic interrogation of over 1.2 billion publicly available MS/MS spectra, resulting in the construction of a modification-centric bile acid MS/MS library (Mohanty, Mannochio-Russo, et al., 2024). This effort has led to the discovery of novel polyamine bile amidates, greatly expanding the known chemical diversity of bile acid modifications (Mohanty, Mannochio-Russo, et al., 2024).

Additionally, compound-based approaches allow for the direct observation of microbial transformations of BAs, but they rely on the labeling of the substrate. To enable BA labeling and further investigate BA modifications, a click-chemistry-based enrichment strategy has been developed (Nie et al., 2024). By adding an alkyne group to the 24-carboxyl position of CA, an alkyne-tagged probe (alkCA) was synthesized (Nie et al., 2024). When co-incubated with human stool-derived ex vivo communities (SECs), and followed by enrichment via click chemistry, a series of CA derivatives were identified, revealing new microbial modifications, 3-acylation, of bile acids by gut communities (Nie et al., 2024).

However, merely identifying novel BA modifications is not sufficient. Elucidating the biosynthetic pathways of these newly discovered BAs is crucial for the development of targeted interventions aimed at precision therapies. In the early stages of enzymology research, the discovery of gut microbial enzymes primarily relied on enzyme activity-based, low-throughput screening approaches, leading to a considerable disconnect between the identification of novel BAs and that of their corresponding biosynthetic enzymes. As the field has progressed, the development of sequence-based tools (e.g., BLAST (Johnson et al., 2008)) and structure-based approaches (e.g., Foldseek-Multimer (Kim et al., 2025)) has invigorated the search for bile acid metabolic enzymes by enabling more systematic and scalable analyses. Looking ahead, the integration of artificial intelligence (AI), deep learning, and advanced protein structure prediction algorithms will likely facilitate large-scale identification and functional annotation of candidate enzymes from massive sequence datasets, particularly those capable of catalyzing the formation of specific BA derivatives (Abramson et al., 2024; Yu, Cui, et al., 2023). Moreover, continuous algorithmic innovation is paving the way for the rational design of novel enzymes, which may serve as templates or guides in the discovery of previously uncharacterized microbial enzymes in the gut (Lauko et al., 2025). These advances hold great promise for liberating bile acid enzyme discovery from traditional, labor-intensive workflows, accelerating our understanding of BA metabolism and its role in host–microbe interactions.

4. BAs and host health

4.1. BAs regulate downstream signaling

Current research has demonstrated that BAs function as hormone-like signaling molecules capable of activating a series of nuclear and membrane receptors (Fig. 2). FXR, the first discovered and most extensively studied BA nuclear receptor, is activated by multiple BAs with efficacy following the order: CDCA > DCA > LCA > CA (Collins et al., 2023; de Aguiar Vallim et al., 2013). Tissue-specific activation of FXR would elicit divergent regulatory outcomes. The activation of FXR induces the expression of small heterodimer partner (SHP) in liver, initiating the FXR-SHP axis to suppress the expression of cholesterol 7α-hydroxylase (CYP7A1), thereby inhibiting hepatic BA synthesis, and repress sodium/taurocholate cotransporting polypeptide (NTCP), thereby reducing hepatic BA uptake (Dawson et al., 2009; Hoeke et al., 2009). Simultaneously, FXR increases the expression of intestinal fibroblast growth factor 15 (FGF15) in mice and FGF19 in human, and downstream receptor fibroblast growth factor receptor 4 (FGFR4), forming the FXR-FGF15/19-FGFR4 axis to further inhibit BA biosynthesis (Hoeke et al., 2009; Inagaki et al., 2005). Additionally, FXR enhances BA efflux in hepatocytes and enterocytes by upregulating the expression of canalicular bile salt export pump (BSEP), multidrug resistance-associated protein 2 (MRP2), and organic solute transporter α/β (OSTα/β) (Jiang et al., 2021; Modica et al., 2010).

Fig. 2.

Downstream signaling of bile acids. Bile acids exert pleiotropic signaling effects through an intricate network of downstream targets, including a series of membrane receptors (e.g., TGR5, S1PR2) and nuclear receptors (e.g., FXR, VDR). In this manner, on the one hand, bile acid maintenance is kept via negative feedback mechanisms. On the other hand, bile acid-associated signaling plays a crucial role in various physiological processes, including glycolipid metabolism, detoxification, inflammation regulation, and immune homeostasis. Some vital bile acid receptors are listed in this figure with annotated physiological processes and pathophysiological manifestations. TGR5, Takeda-G protein-coupled receptor 5; MRGPRX4, Mas-related G-protein coupled receptor member X4; S1PR2, sphingosine-1-phosphate receptor 2; MRGPRE, Mas-related G protein-coupled receptor family member E; FXR, farnesoid X receptor; PXR, pregnane X receptor; VDR, vitamin D receptor; RORγt, retinoid-related orphan receptor γt; CAR, constitutive androstane receptor; AR, androgen receptor; LIP, liver-enriched inhibitory protein; GLP-1, glucagon-like peptide-1; S1P, sphingosine-1-phosphate. This figure was created with BioRender.com.

Beyond these roles, the activation of FXR inhibit the expression of pro-inflammatory transcription factors activator protein-1 (AP-1) and nuclear factor-κB (NF-κB) to repress the inflammatory response (Fuchs & Trauner, 2022), and increase the phosphorylation level (thus inactivate) of glycogen synthase kinase 3β (GSK3β) to elevate hepatic glycogen levels (Han et al., 2004; Zhang et al., 2006). The activation of gut FXR also elevates the production of ceramide synthetic enzymes including sphingomyelin phosphodiesterase 3/4 (SMPD3/4) and ceramide synthase 2/4 (CERS2/4). These ceramide-related pathways reduce mitochondrial acetyl-CoA levels, diminish pyruvate carboxylase (PC) activity, inhibit hepatic gluconeogenesis, and ultimately impair glucose tolerance in mice (Xie et al., 2017). Metabolic disorder models have demonstrated that inhibiting intestinal FXR signaling can reduce obesity, insulin resistance, and hepatic steatosis by regulating BA metabolism as well as the synthesis of intestinal ceramides (Sun et al., 2021). Notably, several BAs exhibit FXR antagonistic properties, such as tauro-β-muricholic acid (T-β-MCA, the first identified BA-derived FXR inhibitor) (Jiang, Xie, Lv, et al., 2015; Li et al., 2013), glycoursodeoxycholic acid (GUDCA) (Sun et al., 2018), and the recently discovered BA–methylcysteamine (BA-MCY) (Won et al., 2025), all of which demonstrate therapeutic potential in ameliorating metabolic disorders.

BAs could additionally activate other nuclear receptors including the pregnane X receptor (PXR), constitutive androstane receptor (CAR), and vitamin D receptor (VDR) (Chiang, 2013; Fuchs & Trauner, 2022). These receptors are typically activated by limited BA kinds (e.g., LCA) and participate in regulating lipid/glucose homeostasis, detoxification, innate immunity, and xenobiotic metabolism (Jia et al., 2018; Li & Chiang, 2013). Current evidence suggests LCA and its derivatives seem to exhibit broader receptor activation profiles. Notably, 3-oxoLCA (3-oxolithocholic acid) and isoLCA—metabolites generated through 3-HSDH-mediated conversion of LCA—suppress T_H17 cell differentiation via inhibition of RORγt, with both their concentrations and 3α-HSDH gene expression being significantly diminished in patients with IBD (Hang et al., 2019; Paik et al., 2022). 3-oxoLCA is an even more potent agonist of PXR than LCA itself. Activation of PXR plays a protective role against LCA-induced toxicity (Thibaut & Bindels, 2022) and promotes de novo lipogenesis in hepatocytes by upregulating the expression of the lipogenic regulator hyroid hormone–responsive SPOT14 homolog (S14) (Moreau et al., 2009).

Furthermore, LCA derivatives 3-oxo-Δ5-LCA, 3-oxo-Δ4-LCA, and 3-oxo-Δ4,6-LCA demonstrate potent androgen receptor (AR) antagonistic activity. Particularly, 3-oxo-Δ4,6-LCA exhibits therapeutic effects by suppressing tumorigenesis and potentiating the efficacy of immune checkpoint inhibitors (e.g., anti-PD-1) in cancer treatment (Guan et al., 2022; Jin et al., 2025).

BAs also function as versatile modulators of membrane receptor activity, most prominently exemplified by their agonistic effects on Takeda-G protein-coupled receptor 5 (TGR5). The relative efficacy of common BAs in activating TGR5 follows a descending order: LCA > DCA > CDCA > CA (Collins et al., 2023), suggesting that bai-related transformation may act as an activation enhancer to increase the activation efficacy of TGR5 by BAs. TGR5 activation triggers adenylate cyclase (AC)-cyclic adenosine 3′,5′-monophosphate (cAMP) signaling pathway, wherein cAMP serve as a secondary messenger to coordinate systemic metabolic adaptations (Yang et al., 2023). An important metabolic outcome involves cAMP-mediated induction of type 2 iodothyronine deiodinase (D2), which enhances energy expenditure through local thyroid hormone activation in brown adipose tissue (Watanabe et al., 2006). Beyond metabolic regulation, TGR5 signaling exhibits multifaceted immunomodulatory effects by intersecting with various molecular pathways, highlighting the role of BAs as immune modulators. For instance, the TGR5-AC-cAMP signaling pathway induces ubiquitination and inactivation of the NLR family pyrin domain containing 3 (NLRP3) inflammasome, thereby suppressing its pro-inflammatory activity (Guo et al., 2016). Additionally, TGR5 activation enhances the protein kinase B (Akt)-mammalian target of rapamycin (mTOR) pathway to promote the production of liver-enriched inhibitory protein (LIP), which reduces macrophage migration, cell differentiation and the production of proinflammatory cytokine within adipose tissue (Perino et al., 2014; Sorrentino et al., 2020). Furthermore, TGR5 activation inhibits the NF-κB pathway, alleviating hepatic inflammatory responses (Wang et al., 2011). Beyond these classical functions, TGR5 also modulates pancreatic islet cells to stimulate the secretion of incretins such as glucagon-like peptide-1 (GLP-1) (Fleishman & Kumar, 2024). In intestinal stem cells, TGR5 triggers the activation of yes-associated protein (YAP) and its upstream regulator tyrosine kinase SRC, facilitating epithelial repair and regeneration (Sorrentino et al., 2020).

Beyond classical membrane receptors such as TGR5, BAs are also capable of activating certain non-classical membrane receptors, such as MRGPRX4, G protein-coupled receptor 39 (GPR39), sphingosine-1-phosphate receptor 2 (S1PR2) and recently found Mas-related G protein-coupled receptor family member E (MRGPRE) (Lin et al., 2025; Nagahashi et al., 2016; Yu et al., 2019; Zi & Rao, 2024). MRGPRX4 has gained attention for its role in promoting human cholestatic pruritus, with DCA demonstrating superior agonistic potency compared to CDCA, CA, and LCA (Meixiong et al., 2019; Yu et al., 2019). Certain bile acids with 3-sulfation exhibit enhanced agonistic activity toward MRGPRX4. For instance, 3-sulfated deoxycholic acid (DCA-3S) shows significantly stronger activation of MRGPRX4 compared to its unsulfated counterpart, DCA (Yang et al., 2024). Building upon this finding and structure-guided drug design, researchers successfully engineered C7, a novel BA derivative that lacks the 3-hydroxyl moiety to avoid MRGPRX4-mediated pruritus while retaining therapeutic efficacy in hepatic disorders (Yang et al., 2024). 3-sulfated BAs, particularly taurolithocholic acid 3-sulfate (TLCA-S), glycolithocholic acid 3-sulfate (GLCA-S), and LCA-S, are also shown to activate the orphan receptor GPR39, thereby contributing to pancreatic injury in biliary acute pancreatitis through mechanisms involving intracellular calcium elevation and necrotic cell death (Zi & Rao, 2024). Some conjugated BAs engage S1PR2, triggering accumulation of nuclear sphingosine-1-phosphate (S1P) (Nagahashi et al., 2016). Further effects include the inhibition of histone deacetylases 1/2 (HDAC1/2) and induced expression of hepatic enzymes and transporters governing lipid/sterol metabolism (Nagahashi et al., 2015). The S1PR2-S1P-S1PR signaling axis emerges as a promising therapeutic target for inflammatory and metabolic disorders, with receptor modulation demonstrating efficacy in preclinical models of hepatobiliary diseases (Nagahashi et al., 2016). Additionally, a recent work found a kind of MCBA, tryptophan-conjugated cholic acid (Trp-CA) was the physiological ligand of MRGPRE, which improved T2DM through boosting GLP-1 secretion (Lin et al., 2025).

Importantly, bile acid metabolic enzymes critically modulate receptor activation profiles through structural modifications that enhance, attenuate, or even reverse the intrinsic activity of parent BAs. As previously exemplified by T-β-MCA, GUDCA, and BA-MCY, enzymatic transformations profoundly alter ligand-receptor interaction dynamics. Notably, LCA functions as a potent TGR5 agonist, yet its sulfonation at the 3-hydroxyl position completely abrogates receptor-ligand binding affinity (Sato et al., 2008). In contrast, 3-oxo modification of CDCA moderately enhances TGR5 activation affinity (Sato et al., 2008). The differential effects of structural analogs are further illustrated by CA exhibiting weak TGR5 agonism, while its 7-sulfated derivative CA7S demonstrates superior receptor activation capacity (D'Agostino et al., 2024). These findings underscore the enzymatic capacity to pharmacologically “reprogram” BA signaling through regio-specific transformation.

4.2. BAs shape gut microbiota

BAs exert profound regulatory effects on the composition and functionality of the gut microbiota (Larabi et al., 2023). Compared with specific pathogen free controls, germ-free rats exhibited decreased levels of BAs in various tissues, depleted unconjugated and glycine-conjugated BAs, increased taurine-conjugated BAs and altered hepatic expression of FXR-regulated genes (Swann et al., 2011). Clinical observations also reveal significantly reduced bacterial diversity in cholestasis patients compared to healthy controls (Yu, Liu, et al., 2023). Experimental studies demonstrate BA-dependent modulation of microbial populations: dietary CA supplementation increases Bacillota abundance in mice (Islam et al., 2011), whereas DCA administration reduces gut BSH activity, diminishes Bacillota populations, and increases Bacteroidota proportions (Larabi et al., 2023). BA exposure also disrupts bacterial metabolic pathways involving amino acid, nucleotide, and carbohydrate utilization (Tian et al., 2020).

Furthermore, the biotransformation of BAs by specific gut bacteria mediates intricate microbial interactions that influence both self-regulation and cross-species dynamics. For example, E. lenta converts DCA to the less toxic isoDCA derivative, thereby creating an ecological niche favoring Bacteroides species’ growth (Devlin & Fischbach, 2015). B. uniformis generates 3-sucCA from CA, which enhances Akkermansia muciniphila growth through enhancement of Amuc-NagB activity to promote glucosamine amination (Nie et al., 2024). A. muciniphila can alleviate the progression of MASH by restoring intestinal barrier integrity and suppressing the lipopolysaccharide (LPS)–Toll-like receptor (TLR4) signaling pathway (Nie et al., 2024). Another classic example is the prevention of C. difficile (formerly called Clostridium difficile) infection (CDI), a global healthcare challenge characterized by severe diarrhea and intestinal inflammation (Schnizlein & Young, 2022). Evidence has accumulated that primary BAs, such as TCA and GCA, promote C. difficile spore germination and toxin production (Sorg & Sonenshein, 2008), while secondary BAs, such as DCA and LCA, exhibit potent inhibitory effects on C. difficile growth (Thanissery et al., 2017). The colonization resistance against C. difficile observed in healthy individuals is likely mediated through coordinated biotransformation of BAs by BSH and bai-encoded enzymes (Buffie et al., 2015; Foley et al., 2023; Schnizlein & Young, 2022). BSH initiates this process by hydrolyzing conjugated primary BAs to free BAs, which are subsequently converted by bai-encoded enzymes into DCA and LCA, both of which exhibit inhibitory effects against C. difficile proliferation. Concomitantly, bacteria harboring the bai operon could synthesize tryptophan-derived antibiotics targeting C. difficile, with enhanced inhibitory efficacy when combined with DCA/LCA (Kang et al., 2019). Clinically, UDCA administration also has been validated as an effective therapeutic strategy for alleviating C. difficile-associated intestinal inflammation in human trials (Weingarden et al., 2016).

4.3. BAs modulate metabolic diseases

BAs, as pivotal regulators of host metabolic processes, have been implicated in the pathogenesis of various metabolic diseases (e.g., obesity, T2DM, MASH and PCOS) through impaired BA signaling (Collins et al., 2023; Di Lorenzo et al., 2023; Jia et al., 2018) (Fig. 3). In the early 2000s, an elegant work has demonstrated that transplantation of gut microbiota from obese to lean mice could induce weight gain in the recipient animals (Bäckhed et al., 2004). In individuals with metabolic syndrome, vancomycin treatment significantly reduced fecal secondary BAs, with observed decreased abundance of Bacillota, compensatory expansion of Pseudomonadota and insulin resistance (Vrieze et al., 2014). Patients with metabolic associated fatty liver disease (MAFLD, formerly known as NAFLD) exhibit enrichment of taurine- and glycine-metabolizing bacteria, leading to increased levels of DCA and decreased CDCA, indicating enhanced microbial conversion of primary to secondary BAs (Jiao et al., 2018). Collectively, these findings underscore the vital role of microbial bile acid metabolic enzymes participating in the development and progression of metabolic diseases through dynamic remodeling of the BA landscape.

Fig. 3.

Roles of bile acids in human diseases. Bile acids influence host disease progression through two major mechanisms. On the one hand, they act as signaling molecules by activating or inhibiting a range of host receptors, thereby regulating metabolic and immune homeostasis. On the other hand, certain bile acids can modulate the growth of specific microbial taxa, indirectly influencing disease development. (A) A simplified overview of gut microbial hydrolysis and further modifications of bile acids is shown, along with the associations between altered bile acid profiles and various diseases. In many disease states, a cascade of events can be observed—beginning with altered microbial enzyme activity, leading to disrupted bile acid homeostasis, and ultimately resulting in dysregulated host signaling. (B) Highlights how specific microbiota-derived enzymes modify cholic acid (CA)—such as 3-acylation or 24-amidation—generating a series of chemically modified CA derivatives that influence metabolic disease progression through diverse mechanisms. (C) Emphasizes the unique role of hyocholic acid (HCA), a bile acid predominantly found in pigs but also present in humans, which exerts a dual action in the host: activating TGR5 and inhibiting FXR on enteroendocrine cells, thereby synergistically enhancing GLP-1 secretion. T2DM, type 2 diabetes mellitus; MAFLD, metabolic associated fatty liver disease; MASH, metabolic dysfunction-associated steatohepatitis; CDI, Clostridioides difficile infection; IBD, inflammatory bowel disease; PCOS, polycystic ovary syndrome; GATA3, GATA binding protein 3; IL-22, interleukin-22; LPS, lipopolysaccharide; TLR4; Toll-like receptor 4. This figure was created with BioRender.com.

The metabolic regulatory effects of BAs primarily arise from their receptor-mediated signaling, though studies on FXR activation are self-contradictory. Study has revealed that inhibition of intestinal FXR reduces hepatic steatosis by suppressing C16:0 ceramide synthesis and downregulating the expression of sterol regulatory element-binding protein 1C (SREBP1C), suggesting intestinal FXR antagonism as a viable strategy against MASH (Jiang, Xie, Li, et al., 2015). This corresponds with the metabolic benefits of T-β-MCA, a natural FXR antagonist that ameliorates obesity through gut-liver axis modulation (Li et al., 2013). However, contradictory evidence emerges from studies demonstrating that mice BSH overexpression alleviates metabolic dysfunction via T-β-MCA deconjugation, paradoxically with reduced weight gain, plasma cholesterol and liver triglyceride levels, despite enhanced FXR signaling (Joyce et al., 2014). These discrepancies likely attribute to tissue-specific FXR responses, underscoring the necessity for compartmentalized therapeutic targeting to minimize off-tissue effects. Notably, certain BAs exhibit pleiotropic receptor cross-talk. For example, hyocholic acid (HCA) simultaneously activates TGR5 and antagonizes FXR, collaboratively enhancing GLP-1 secretion by enteroendocrine cells to improve glucose homeostasis (Zheng et al., 2021).

Emerging evidence reveals that other intestinal signals and xenobiotics (e.g., drugs) can modulate metabolic disease progression through microbiota-BA-receptor axis interactions. For example, metformin ameliorates hyperglycemia and metabolic dysfunction via a mechanism involving decreased B. fragilis abundance and subsequent GUDCA elevation, which antagonizes intestinal FXR signaling (Sun et al., 2018). Another research employing intestine-specific hypoxia-inducible factor-2α (HIF-2α) knockout mice demonstrated altered microbial ecology characterized by reduced Phocaeicola vulgatus abundance and expanded Ruminococcus torques abundance (Wu et al., 2021). These ecological shifts accompany with decreased luminal lactate concentrations and subsequent elevation of TCA and DCA levels (Wu et al., 2021). The accumulated BAs activated TGR5, driving adipose thermogenesis through coordinated upregulation of uncoupling protein 1 (UCP1) and mitochondrial creatine kinase 2 (CKMT2) expression in white adipose tissue (Wu et al., 2021).

PCOS, a leading cause of female infertility, is characterized by the pathognomonic triad of hyperandrogenism, ovulatory dysfunction and polycystic ovarian morphology (Azziz et al., 2016). Although less frequently mentioned, emerging evidence repositions PCOS as a metabolic disorder intricately linked to gut microbiota dysbiosis (Di Lorenzo et al., 2023). Comparative analyses reveal significantly reduced α-diversity in the gut microbiota of PCOS patients, concomitant with overrepresentation of P. vulgatus and depletion of BA species including GDCA and tauroursodeoxycholic acid (TUDCA) (Lindheim et al., 2017; Qi et al., 2019). Mechanistic studies demonstrate that GDCA ameliorates PCOS phenotypes through GATA binding protein 3 (GATA3)-mediated induction of interleukin-22 (IL-22) secretion from intestinal type 3 innate lymphoid cells (ILC3s), with both GDCA levels and IL-22 production being markedly decreased in affected individuals (Qi et al., 2019). This microbiota-BA-immune axis disruption underscores the metabolic cornerstone of PCOS pathogenesis, suggesting therapeutic potential in targeting BA signaling pathways.

5. Conclusion

Contemporary research has fundamentally reshaped our understanding of BAs, transcending their classical definition as biological detergents to recognize their pivotal role as host-microbial cometabolites orchestrating metabolic regulation and immune homeostasis. The remarkable biochemical journey of BAs along the gastrointestinal tract involves intricate enzymatic modifications by host and microbial enzymes, generating unprecedented structural diversity through conjugation/deconjugation, dehydroxylation, epimerization, oxidation/reduction, and so on. However, current analytical methodologies still rely on fecal BA profiling, which fails to capture critical spatiotemporal dynamics including intestinal segment-specific variations and circadian rhythmicity. Emerging in situ sampling technologies promise to revolutionize BA mapping by developing an ingestible device to collect samples from different gut intestinal segments (Shalon et al., 2023). Furthermore, the exponential growth in cataloged BA derivatives starkly contrasts with the limited repertoire of characterized microbial enzymes, highlighting an urgent need for high-throughput enzyme screening platforms combining metagenomic mining, structural proteomics, and AI-driven catalytic activity prediction. Addressing this knowledge gap will accelerate the development of precision therapies targeting microbiota-derived enzymes, such as engineered probiotic consortia for BA-directed metabolic reprogramming.

Microbiota-driven biotransformations not only remodel the BA pool composition but also regulate host energy metabolism and immune homeostasis through downstream receptor signaling, particularly via FXR and TGR5. This bidirectional host-microbial crosstalk has been mechanistically linked to metabolic disorders including obesity, T2DM, MASH, and PCOS, establishing the microbiota-BA axis as a promising therapeutic target. However, the inherent complexity of this interaction network poses significant challenges in identifying and targeting critical mechanistic nodes. Future investigations should prioritize developing precision modulation strategies focusing on in vivo engineering of microbial enzymes or BA profiles, and discrimination of microbial-host-isozymes for selective pathway manipulation (Wang et al., 2023). Such approaches may unveil novel therapeutic paradigms for metabolic disease interception.

CRediT authorship contribution statement

Haohan Ma: Writing – review & editing, Writing – original draft, Investigation. Kai Wang: Writing – review & editing, Writing – original draft, Supervision, Project administration. Changtao Jiang: Writing – review & editing, Writing – original draft, Supervision, Project administration, Funding acquisition.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.