1 Host metabolism balances microbial regulation of bile acid signaling 1 2 Tae Hyung Won1,10,$, Mohammad Arifuzzaman2,3,$, Christopher N. Parkhurst2,$, Isabella C. 3 Miranda2, Bingsen Zhang1, Elin Hu2,3, Sanchita Kashyap2,3, Jeffrey Letourneau4, Wen-Bing Jin2, 4 Yousi Fu5, Douglas V. Guzior5,6, JRI Live Cell Bank2, Robert A. Quinn5, Chun-Jun Guo2,3,7, 5 Lawrence A. David4,8, David Artis2,3,7,9*, and Frank C. Schroeder1* 6 7 1Boyce Thompson Institute and Department of Chemistry and Chemical Biology, Cornell 8 University, Ithaca, NY, USA. 9 2Jill Roberts Institute for Research in Inflammatory Bowel Disease, Division of Gastroenterology 10 and Hepatology, Joan and Sanford I. Weill Department of Medicine, Weill Cornell Medicine, 11 Cornell University, New York, NY, USA. 12 3Friedman Center for Nutrition and Inflammation, Weill Cornell Medicine, Cornell University, 13 New York, NY, USA. 14 4Department of Molecular Genetics and Microbiology, Duke University, Durham, NC, USA. 15 5Department of Biochemistry and Molecular Biology, Michigan State University, East Lansing, 16 MI, USA. 17 6Department of Microbiology, Genetics, and Immunology, Michigan State University, East 18 Lansing, MI, USA. 19 7Department of Microbiology and Immunology, Weill Cornell Medicine, Cornell University, New 20 York, NY, USA. 21 8Program in Computational Biology and Bioinformatics, Duke University School of Medicine, 22 Durham, NC, USA. 23 9Allen Discovery Center for Neuroimmune Interactions, Weill Cornell Medicine, Cornell 24 University, New York, NY, USA. 25 10Current address: College of Pharmacy and Institute of Pharmaceutical Sciences, CHA 26 University, 120 Haeryong-ro, Pocheon-si, Gyeonggi-do 11160, Republic of Korea. 27 $These authors contributed equally 28 *Correspondence to: dartis@med.cornell.edu and schroeder@cornell.edu 29 2 Abstract (228 words) 30 Metabolites derived from the intestinal microbiota, including bile acids (BAs), extensively 31 modulate vertebrate physiology, including development1, metabolism2-4, immune responses5-7, 32 and cognitive function8. However, to what extent host responses balance the physiological 33 effects of microbiota-derived metabolites remains unclear9,10. Employing untargeted 34 metabolomics of mouse tissues, we identified a family of bile acid-methylcysteamine (BA-MCY) 35 conjugates that are abundant in the intestine and dependent on VNN1/Vanin 1, a pantetheinase 36 highly expressed in intestinal tissues. This host-dependent MCY conjugation inverts BA function 37 in the hepatobiliary system. Whereas microbiota-derived free BAs function as agonists of the 38 farnesoid X receptor (FXR) and negatively regulate BA production, BA-MCYs act as potent 39 antagonists of FXR and promote expression of BA biosynthesis genes in vivo. Supplementation 40 with stable-isotope labeled BA-MCY increased BA production in an FXR-dependent manner and 41 BA-MCY supplementation in a mouse model of hypercholesteremia decreased lipid 42 accumulation in the liver, consistent with BA-MCYs acting as intestinal FXR antagonists. BA-43 MCYs levels were reduced in microbiota-deficient mice and restored by transplantation of 44 human fecal microbiota. Dietary intervention with inulin fiber further increased levels of both free 45 BAs and BA-MCY levels, indicating that BA-MCY production by the host is regulated by levels of 46 microbiota-derived free BAs. We further show that diverse BA-MCYs are also present in human 47 serum. Taken together, our results indicate that BA-MCY conjugation by the host balances host- 48 and microbiota-dependent metabolic pathways that regulate FXR-dependent physiology. 49 50 51 3 MAIN 52 Vertebrates harbor diverse communities of bacteria, protozoa, fungi and viruses, collectively 53 referred to as the microbiota. Co-evolution with these microbes has resulted in the 54 establishment of host-microbiota cross-talk via complex metabolic networks9-11. One such 55 example is the metabolism of bile acids2-4,12,13, which play a central role in vertebrate physiology 56 as ligands of the farnesoid X receptor (FXR), a conserved nuclear hormone receptor that 57 controls cholesterol and BA biosynthesis as well as fat metabolism and glucose 58 homeostasis3,4,14-16. BAs are produced in the liver where they are conjugated to taurine and 59 amino acids by BA-CoA:amino acid N-acyltransferase (BAAT)17,18. Conjugated BAs are then 60 secreted via the biliary system into the intestine where they undergo extensive modification by 61 microbial metabolism4,19-23, including diverse dehydroxylation and oxidation reactions as well as 62 deconjugation, which recovers free, unconjugated BAs that are efficiently reabsorbed and 63 transported back to the liver. While BA-taurine conjugates are inactive or have weak agonistic or 64 antagonistic activities24-27, free BAs act as potent FXR agonists that negatively regulate BA 65 production2,3,14,28. BA-taurine conjugation by the host and microbial deconjugation thus form a 66 negative feedback loop that tightly regulates BA levels and BA-dependent physiology. However, 67 the requirement for adaptation to changing metabolic demands and feeding states and the 68 ubiquity of feed-forward signaling in biological systems29 raises the question whether there exist 69 FXR antagonists that act as positive regulators of BA production. Here we utilize untargeted 70 metabolomics to compare germ-free (GF) and microbiota-replete specific pathogen-free (SPF) 71 mice to uncover a host-mediated BA modification generating potent FXR antagonists that act as 72 intestinal regulators of BA metabolism. 73 74 Comparative metabolomics reveals BA-MCYs 75 Lack of microbial deconjugation results in increased levels of BA-taurine conjugates in GF mice, 76 while abundances of free BAs are dramatically reduced24. We hypothesized that abundances of 77 yet unannotated BA derivatives may be similarly affected by the absence of the microbiota. In 78 order to discover such unannotated compounds, we first obtained a comprehensive overview of 79 microbiota-dependent changes in the mouse serum metabolome via high-resolution mass 80 spectrometry- (HRMS-) based comparative metabolomics of samples from microbiota-replete 81 specific pathogen-free (SPF) and GF mice (Fig. 1a). The resulting datasets were processed 82 using the xcms30-based Metaboseek platform31, which facilitates identification of MS features 83 whose abundances differ significantly between different conditions (Fig. 1b). This untargeted 84 4 comparison revealed stark differences between GF and SPF mice. In total we detected more 85 than 40,000 MS features from combined analyses of serum samples in positive and negative 86 ionization modes, of which ~10% were significantly differential (at p≤0.05) between GF and SPF 87 mice. To prioritize among the large number of differential features, we focused on compounds 88 that were robustly detected in all replicates (see Methods) and at least five-fold different in GF 89 relative to SPF samples. Using these stringent criteria, we detected several hundred microbiota-90 dependent metabolites in the serum metabolome (Supplementary Table 1). For further 91 characterization, we acquired tandem mass spectrometry (MS2) data for all differential 92 metabolites. To focus on BAs and BA derivatives, we applied a permissive molecular formula 93 filter that required the presence of MS2 fragments containing a complete 24-carbon backbone, 94 which would suggest the presence of a steroid backbone (see Methods). MS2 networking 95 revealed three major clusters of microbiota-dependent metabolites with fragmentation patterns 96 indicating they represent BAs or BA-derivatives (Fig. 1c and Supplementary Figs. 1-3). Two of 97 these clusters represented BA-taurine conjugates and free BAs, whose abundances were 98 greatly increased and decreased in GF mice, respectively, as expected, given the lack of 99 microbial taurine deconjugation. In contrast, the third cluster appeared to represent a family of 100 previously unannotated BA derivatives, whose abundances, similar to free BAs, were reduced in 101 GF mice. 102 Detailed analysis of the MS isotope patterns and MS2 spectra of the putative BA derivatives 103 suggested the presence of an S-methylcysteamine (MCY) moiety (Fig. 1d and Extended Data 104 Fig. 1). The MS2 spectra further indicated that these MCY derivatives belong to three different 105 series, representing putative bile acid methylcysteamides (BA-MCY), corresponding sulfoxides 106 (BA-MCYO) and sulfodioxides (BA-MCYO2) (Extended Data Fig. 1). Structures of these BA-107 MCYs were proposed based on the relative abundances and retention times of free bile acids in 108 the analyzed samples and confirmed via synthesis of authentic standards, which led to the 109 identification of a total of 18 BA derivatives, including the MCY, MCYO, and MCYO2 derivatives 110 of cholic acid (CA), β-muricholic acid (βMCA), chenodeoxycholic acid (CDCA), ursodeoxycholic 111 acid (UDCA), deoxycholic acid (DCA), and 7-ketodeoxycholic acid (7-KDCA) (Fig. 1e, 112 Supplementary Fig. 4, and Supplementary Table 2, see Supplementary information for synthetic 113 procedures and NMR data). Taken together, our untargeted metabolomic comparison of SPF 114 and GF mice revealed BA-MCYs as a previously unannotated family of microbiota-dependent 115 BA derivatives. 116 117 5 Microbiota-dependence of BA-MCY levels 118 Abundances of BA-MCYs were strongly reduced in GF compared to SPF serum samples, but 119 not abolished (Fig. 2a,b, Extended Data Fig. 2a, and Supplementary Table 3). Similarly, BA-120 MCY levels were decreased in feces of GF mice compared to SPF mice (Extended Data Fig. 121 2b-d). To better understand the relationship between BA-MCY levels and the presence of the 122 microbiota, we next tested whether introduction of human microbiota into GF mice would affect 123 BA-MCY production. For this purpose, we performed human fecal microbiota transfer (FMT) 124 from healthy individuals into GF mice32, and then profiled BAs and BA conjugates in serum and 125 feces. We found that abundances of both free BAs and BA-MCY conjugates were greatly 126 increased in serum and feces of mice that received human FMT (Fig. 2a,b and Extended Data 127 Fig. 2a-g). 128 The results from our comparison of GF with SPF and human FMT mice suggested that BA-MCY 129 levels are governed in part by levels of the corresponding free BAs. In previous work, we and 130 others have shown that changes in the microbiota induced by supplementation with inulin fiber 131 can dramatically increase levels of free BAs in serum of SPF mice32,33. We took advantage of 132 this to test whether such a dietary intervention-based increase of free BAs would also affect BA-133 MCY levels. We observed that BA-MCY-levels were greatly increased in SPF mice fed an inulin-134 based high fiber diet (Extended Data Fig. 3a-c), suggesting that expansion of the free-BA pool 135 leads to increased BA conjugation with MCY. Moreover, we found that levels of free BAs and 136 BA-MCY conjugates in SPF mice fed different diets are generally correlated (Fig. 2c and 137 Extended Data Fig. 3d,e). Finally, to determine whether BA-MCYs are also present in humans, 138 we analyzed human serum samples, which revealed MCY derivatives of all major bile acids 139 common in humans (Fig. 2d and Extended Data Fig. 3f,g). Collectively, these data indicate that 140 BA-MCY conjugates are present in mouse and human, and that increases of free BA levels, 141 following human FMT into GF mice or as a result of supplementation with dietary fiber, are 142 associated with parallel increases of the corresponding BA-MCY conjugates. 143 144 Biochemical origin of BA-MCY conjugates 145 To further clarify the roles of host and microbiota for the production of the BA-MCY conjugates, 146 we next investigated the in vivo origin of the cysteamine moiety. Cysteamine is produced 147 primarily via degradation of coenzyme A34, and oxidation of cysteamine in the liver and other 148 tissues produces taurine, which is then conjugated with BAs in the liver, producing BA-taurine 149 conjugates that are secreted into the intestine. Therefore, we considered two different models 150 6 for the origin of the MCY moieties in the BA-MCYs (Fig. 2e). First, these compounds could 151 originate from conjugation of BAs with cysteamine or a cysteamine derivative derived from 152 breakdown of Coenzyme A. Alternatively, the BA-MCY conjugates could be derived from 153 reduction of corresponding BA-taurine derivatives by the host or the gut microbiota (Fig. 2e). To 154 distinguish between these scenarios, we performed a series of stable-isotope labeling 155 experiments with taurine-d4 and L-cysteine-3,3-d2 in SPF mice (Extended Data Fig. 4a-c). 156 HPLC-HRMS analysis of serum from SPF mice supplemented with taurine-d4 revealed 157 extensive labeling of taurine conjugates of BAs, as expected (Extended Data Fig. 4a). However, 158 none of the MCY conjugates were labeled (Extended Data Fig. 4a), indicating that the 159 biosynthetic pathways of taurine and MCY conjugates are distinct. Next, to test whether the 160 MCY conjugates originate from incorporation of a cysteine-derived cysteamine moiety, we 161 analyzed serum samples from SPF mice supplemented with L-cysteine-3,3-d2 (Extended Data 162 Fig. 4b,c). HPLC-HRMS analysis revealed incorporation of deuterium in both the BA-taurine and 163 BA-MCY conjugates (Extended Data Fig. 4b,c), consistent with a cysteine-origin of both taurine 164 and the MCY moiety. As expected, we also observed deuterium incorporation into the CoA-165 breakdown product pantetheine (Extended Data Fig. 4d). 166 These results support a model in which the BA-MCY conjugates are derived from acylation of a 167 cysteamine derivative other than taurine. Next, we considered whether the conjugation is likely 168 to be mediated by the microbiota or the host. While the abundances of the BA-MCY conjugates 169 were strongly microbiota-dependent, their production was not abolished in GF animals (Fig. 170 2a,b and Extended Data Fig. 2a-d), suggesting that the conjugation reaction itself does not 171 require the microbiota. Therefore we next investigated whether BA-MCY production is related to 172 the biosynthesis of BA-taurine conjugates. Conjugation of BAs with taurine and amino acids in 173 the liver is mediated by BAAT, a type-1 acyl-CoA thioesterase (ACOT). Members of this gene 174 family catalyze a wide range of acyl transfer reactions35; and putative conjugates of BAs with 175 cysteamine and methylcysteamine were recently reported to accumulate in Baat−/− mice36. We 176 confirmed that the bile acid cysteamine derivatives accumulating in Baat−/− mice are identical 177 with the compounds we identified (Extended Data Fig. 5a-d), demonstrating that BA-MCY 178 biosynthesis is distinct from BA-taurine conjugation. Lack of taurine conjugation in Baat−/− mice 179 results in greatly increased levels of both free bile acids and BA-MCY conjugates (Extended 180 Data Fig. 5e,f)36, consistent with parallel increase of free BAs and BA-MCY conjugates 181 observed in mice fed a high fiber diet. Suppression of microbiota in Baat−/− mice does not 182 diminish the elevated BA-MCY levels in this mutant strain36. Taken together, our results indicate 183 7 that BA-MCYs are derived from a host-dependent conjugation pathway that exists in parallel 184 with conjugation of BAs with taurine. 185 Next we considered potential sites of BA-MCY biosynthesis. While BA-taurine conjugates are 186 produced in the liver, the fact that BA-MCY levels are governed primarily by the abundance of 187 free BAs, which in SPF mice are predominantly derived from deconjugation of BA-taurine and 188 BA-glycine by the intestinal microbiota, suggested that the intestine may be involved in BA-MCY 189 production. Therefore, we additionally profiled BA-MCY levels in the small intestine and cecum, 190 which revealed that BA-MCY conjugates are much more abundant in intestinal tissues than liver 191 (Fig. 3a and Extended Data Fig. 6), Moreover, the ratio of BA-MCY conjugates relative to their 192 oxidation products BA-MCYO and BA-MCYO2 is higher in intestinal tissues than in liver, serum, 193 and feces (Fig. 3a and Supplementary Table 3), suggesting that BA-MCY conjugates are 194 produced from free BAs following their re-uptake in the intestine. A model in which intestinal 195 production of BA-MCYs is dependent on re-uptake of free BAs is further consistent with our 196 observation that, across different diets and conditions, abundances of free BAs and BA-MCYs 197 are correlated (Extended Data Fig. 3a-c). 198 199 BA-MCY biosynthesis depends on host VNN1 200 Since BA-MCY conjugates could plausibly be derived from pantetheine breakdown, we sought 201 to test whether BA-MCY biosynthesis may proceed via the pantetheinase VNN1/Vanin 1, which 202 is highly expressed in intestinal tissues37,38, where BA-MCY concentrations were highest. 203 VNN1/Vanin 1’s primary function is to hydrolyze pantetheine into cysteamine and pantothenic 204 acid (Fig. 3b), as part of coenzyme A recycling, and recent studies have shown that 205 VNN1/Vanin 1 plays important roles in the regulation of metabolism, inflammation, and 206 associated diseases37,39. While VNN1 could be a plausible source for the cysteamine moiety 207 required for BA-MCY production, we hypothesized that, in addition to hydrolyzing pantetheine, 208 VNN1 may also be capable of hydrolyzing BA-pantetheine or BA-CoA conjugates. The resulting 209 S-linked BA-cysteamine derivatives would then rearrange to the N-linked isomer, which 210 following S-methylation would produce the BA-MCY conjugates (Fig. 3b). To investigate this 211 hypothesis, we first tested whether recombinant VNN1 could hydrolyze a synthetic cholic acid-212 pantetheine conjugate (CA-pant). We found that VNN1 breaks down CA-pant as efficiently as 213 pantetheine (Fig. 3c and Extended Data Fig. 7a) and further showed that the resulting CA-214 cysteamine conjugate re-arranges to the corresponding cholic-acid-cysteamide (CA-CY), a 215 plausible precursor of CA-MCY that we had also detected in Baat−/− mice (Extended Data Fig. 216 8 7b). To assess whether VNN1 contributes to BA-MCY biosynthesis in vivo, we compared BA 217 profiles of WT and Vnn1−/− mice in a number of tissues (Fig. 3d). We found that BA-MCY levels 218 are dramatically reduced in the small intestine, liver, and serum, and, to a lesser extent, in feces 219 of Vnn1-/- mutant mice (Fig. 3d). These results indicate that in vivo BA-MCY biosynthesis largely 220 depends on the host enzyme VNN1. Furthermore, we found that the predicted precursor of CA-221 CY, CA-pant, accumulates in the small intestine and feces of Vnn1-/- mice, whereas this 222 compound was absent in the corresponding WT samples and also could not be detected in 223 Vnn1-/- liver and serum, suggesting that BA-pantetheine conjugates are direct precursors for BA-224 MCY biosynthesis in the intestine (Extended Data Fig. 7c,d). 225 226 Microbial and host metabolism of BA-MCYs 227 Taurine and glycine BA conjugates are efficiently deconjugated in the gut by microbial bile salt 228 hydrolases (BSHs)3,4,32,40, generating free BAs. Correspondingly, the ratio of BA-taurine 229 conjugates to free BAs was dramatically increased in feces of GF compared to SPF mice 230 (Extended Data Fig. 8a,b). Considering the possibility that the gut microbiota may play a role 231 also in the deconjugation of BA-MCYs, we noted that, even though BA-MCY levels are reduced 232 in GF mice (Extended Data Fig. 2b-d, 8c), the ratio of BA-MCY conjugates to free BAs was 233 greatly increased in GF compared to SPF fecal samples (Extended Data Fig. 8a). In fact, BA-234 MCY levels were similar to or exceeded levels of free BAs in feces of GF mice (Extended Data 235 Fig. 8d). 236 To determine whether BA-MCYs can indeed be deconjugated by the microbiota, we analyzed 237 fecal and liver samples from SPF and GF mice supplemented with stable-isotope labeled 238 CDCA-d5-MCY for the presence of free labeled CDCA and other labeled BAs that can be 239 derived from CDCA. To broadly survey metabolism of BA-MCYs, we additionally compared 240 supplemented and control mice via untargeted metabolomics using the Label Finder approach 241 in the Metaboseek platform31. Targeted analysis of fecal samples from CDCA-d5-MCY-242 supplemented SPF mice revealed CDCA-d5-MCY as well as d4- and d5-labeled free CDCA 243 (Extended Data Fig. 8e,f,g and Supplementary Fig. 5a,b), indicating that BA-MCYs can be 244 deconjugated in the gut. In addition, we detected labeled versions of other free BAs that can be 245 derived from CDCA (Extended Data Fig. 8f and Supplementary Fig. 5b), whereas CA, DCA, and 246 other CA-derived BAs remained unlabeled (Supplementary Fig. 5b), consistent with their 247 separate biosynthetic pathway23,41. Analysis of fecal and liver samples from CDCA-d5-MCY-248 supplemented SPF mice further indicated that labeled free BAs derived from supplemented 249 9 CDCA-d5-MCY are partly re-conjugated with taurine (Extended Data Fig. 8h,i,j,k,l,m and 250 Supplementary Table 4). Label Finder analysis additionally revealed that the remainder of 251 supplemented CDCA-d5-MCY that was not deconjugated was converted into its oxidized 252 derivative, CDCA-d4/5-MCYO, and, to a lesser extent, CDCA-d4/5-MCYO2. In fact, only trace 253 amounts of CDCA-d5-MCY could be detected in the liver of supplemented mice, indicating that 254 supplemented CDCA-d5-MCY is quickly oxidized to CDCA-d4/5-MCYO(2) (Extended Data Fig. 255 8k,m). 256 In contrast to SPF mice, deconjugation-derived, labeled CDCA or other labeled free BAs were 257 not detected in GF mice supplemented with CDCA-d5-MCY (Fig. 3e and Supplementary Fig. 258 5c,d), indicating that deconjugation of CDCA-MCY is microbiota-dependent. Similarly, 259 suppression of microbiota in SPF mice treated with antibiotics (ABX) resulted in significantly 260 reduced deconjugation of supplemented CDCA-d5-MCY compared to SPF mice (Fig. 3e and 261 Extended Data Fig. 8f,g,j,k). In ABX and GF mice, supplemented CDCA-d5-MCY was instead 262 primarily converted into the corresponding oxidation products, CDCA-d4/5-MCYO and CDCA-263 d4/5-MCYO2 (Extended Data Fig. 8g,k,m and Supplementary Table 4). 264 Next we demonstrated that BA-MCY conjugates are deconjugated by fecal suspensions 265 obtained from SPF mice and individual gut bacteria known to harbor the BSH gene42 (Extended 266 Data Fig. 9a,b). To determine whether BSH is required for BA-MCY deconjugation, we tested 267 gnotobiotic mice colonized with Bacteroides ovatus ATCC 8483 (WT Bo) or a B. ovatus mutant 268 strain in which we deleted the BSH-encoding gene BO_02350 (Δbsh Bo)32,40. We found that 269 CDCA-d5-MCY was partially deconjugated in the gnotobiotic mice colonized with WT Bo, but 270 was not deconjugated in mice colonized with mutant Δbsh Bo (Fig. 3f), where the supplemented 271 CDCA-d5-MCY was exclusively converted to the oxidized CDCA-d4/5-MCYO(2), as in GF mice 272 (Extended Data Fig. 9c). These results indicate that BSH of gut microbiota can deconjugate 273 BA-MCY conjugates, albeit less efficiently than the corresponding taurine conjugates (Fig. 3f), 274 and that, in the absence of microbiota, BA-MCY conjugates are metabolized by the host into the 275 corresponding BA-MCYO and BA-MCYO2 derivatives (Fig. 3g). 276 277 BA-MCYs act as FXR antagonists in vitro 278 For functional evaluation of the BA-MCY conjugates, we focused on the farnesoid X receptor 279 (FXR) one of the major endogenous targets of BAs in vertebrates. Free BAs, e.g., the broadly 280 conserved CDCA, CA, and DCA14, as well amino acid conjugates of CA43, function as potent 281 10 FXR agonists that negatively regulate BA production. In contrast, it is unclear whether there are 282 any conserved endogenous FXR antagonists that would promote BA production. 283 To test for potential FXR agonist or antagonist activity of the identified BA-MCY conjugates, we 284 selected four derivatives, CA-MCY, CA-MCYO, βMCA-MCY, and CDCA-MCY based on their 285 relative abundance in SPF mouse serum samples and considering previously reported FXR 286 agonist activity of the corresponding free BAs14. We assayed these four compounds in agonist 287 and antagonist modes using a protein-protein interaction assay between the full-length human 288 FXR protein and a Steroid Receptor Co–activator Peptide- (SRCP-) derived nuclear fusion 289 protein44. Whereas none of the tested conjugates showed agonist activity at any of the tested 290 concentrations (Fig. 4 and Extended Data Fig. 10a,b), CDCA-MCY, CA-MCY, and βMCA-MCY 291 showed potent antagonistic activity with IC50 values of 1.68, 19.9, and 104.5 μM, respectively, 292 against GW4604-mediated activation of FXR (Fig. 4a,b and Extended Data Fig. 10a)45. In 293 contrast, CA-MCYO was inactive, indicating that sulfur oxidation abolished antagonistic activity 294 (Fig. 4c). CDCA-MCY also inhibited FXR activation mediated by CDCA and the more potent 295 synthetic BA, obeticholic acid (Extended Data Fig. 10c-g). In parallel, we also tested TβMCA, 296 which had previously been reported as a weak, murine-specific FXR antagonist24. However, 297 TβMCA was inactive at the tested range of concentrations in this assay (Extended Data Fig. 298 10h). These results suggest that BA-MCY conjugates function as endogenous FXR antagonists 299 that complement the role of free BAs as FXR agonists. 300 301 BA-MCYs regulate FXR signaling in vivo 302 BA biosynthesis in the liver is controlled by a complex signaling network regulated by hepatic 303 and intestinal FXR via distinct pathways23,28,41(Fig. 5a). In the liver, FXR agonists promote 304 expression of short heterodimer partner (SHP), which in turn antagonizes expression of Cyp7a1, 305 a cytochrome P450 enzyme required for the first and rate-limiting step in BA synthesis (Fig. 306 5a)23,28,41. In addition, SHP expression suppresses Cyp8b1, which catalyzes the conversion of 307 the BA precursor 7α-hydroxy-4-cholesten-3-one into 7α,12α-dihydroxy-4-cholesten-3-one and 308 thereby controls the balance between the relative amounts of bile acids that are 12α-309 hydroxylated (such as CA) and bile acids that are not 12α-hydroxylated (such as CDCA). In 310 contrast, intestinal FXR activation promotes production of ileal hormone fibroblast growth factor 311 15 (FGF15, FGF19 in humans), a signaling peptide that via the enterohepatic circulation travels 312 to the liver to suppress expression of Cyp7a1. Conversely, FXR antagonists promote BA 313 11 synthesis by relieving repression of Cyp7a1 and Cyp8b1 by suppressing SHP and FGF15/19 314 expression 24,39-41 (Fig. 5a). 315 To determine whether BA-MCYs affect FXR-dependent regulation of BA biosynthesis in vivo, 316 we supplemented SPF mice via oral gavage with CDCA-MCY, which had shown the highest 317 potency in our in vitro FXR antagonist assay (Fig. 4a). Gene expression analysis indicated that 318 Cyp7a1 and Cyp8b1 expression in the liver was significantly increased (Fig. 5b). In addition, we 319 found that ileal Shp mRNA and serum FGF15 levels were significantly decreased in mice 320 supplemented with CDCA-MCY (Fig. 5c,d), suggesting that increased expression of Cyp7a1 is 321 in part due to antagonism of the intestinal FXR-FGF15 pathway. Increased Cyp8b1 expression, 322 which is largely independent of the intestinal FXR-FGF15 pathway46-48, suggests that liver FXR 323 may also be affected by CDCA-MCY supplementation, or that other pathways contribute to 324 FXR-dependent ileum-to-liver signaling49. Because we found that BA-MCY production may be 325 dependent on reabsorption of free BAs from the ileum, we additionally tested whether CDCA-326 MCY supplementation affects expression of Slc10a2, the transporter mediating BA re-uptake 327 from the gut; however, Slc10a2 expression was unchanged (Fig. 5e). 328 To measure effects of CDCA-MCY on BA production in vivo, we conducted additional 329 supplementation studies using stable isotope-labeled CDCA-d5-MCY. The use of labeled 330 CDCA-d5-MCY avoided potentially confounding effects arising from deconjugation and further 331 metabolism of the supplemented CDCA-MCY, as it allowed us to distinguish unambiguously 332 between BAs derived from the supplemented, labeled CDCA-d5-MCY and de novo-produced, 333 unlabeled BAs (Extended Data Fig. 8e). Quantification of BA levels from CDCA-d5-MCY-334 supplemented animals showed a strong increase of unlabeled CDCA-derived BAs in fecal 335 samples (Fig. 5f and Extended Data Fig. 11a). Similarly, levels of CA-derived BAs were 336 increased in fecal samples of animals supplemented with CDCA-d5-MCY or CDCA-MCY (Fig. 337 5g and Extended Data Fig. 11b). Fecal BA levels were also increased by CDCA-d5-MCY-338 supplementation in ABX mice (Extended Data Fig. 11c). Given that levels of unlabeled BAs in 339 liver and serum of supplemented mice were not significantly changed (Extended Data Fig. 11d-340 g), the large increase in fecal excretion of both CA- and CDCA-family BAs indicates strong 341 upregulation of BA production in CDCA-MCY-supplemented animals, in line with the increased 342 expression of BA biosynthesis genes (Fig. 5b)47,48,50,51. 343 Next we tested whether upregulation of BA synthesis by CDCA-MCY supplementation is in fact 344 FXR dependent. We found that CDCA-MCY-supplementation increased fecal BA abundance in 345 WT mice but not in FXR-deficient (Nr1h4−/−) mice (Fig. 5h). BA levels in liver and serum of WT 346 12 and Nr1h4−/− mice were not significantly affected by CDCA supplementation (Extended Data Fig. 347 12a,b), consistent with results from our initial supplementation study (Extended Data Fig. 11d-g). 348 These data demonstrate that CDCA-MCY supplementation increases BA biosynthesis in an 349 FXR-dependent manner. 350 Given that intestinal FXR antagonists have been shown to alleviate hepatic steatosis in mouse 351 models of obesity46,52, we asked whether CDCA-MCY supplementation could improve lipid 352 accumulation in the liver of mice fed a high cholesterol diet (HCD). Liver histology and oil red-O 353 staining revealed greatly decreased hepatic lipid accumulation in CDCA-MCY-supplemented 354 HCD-fed mice compared to untreated HCD-fed mice (Fig. 5i,j), consistent with previous studies 355 of synthetic compounds acting as intestinal FXR antagonists46,52. We observed similar effects 356 when CDCA-MCY was supplied at a 10-fold lower dose (Extended Data Fig. 12c,d). 357 Taken together, our results support a model in which host-derived BA-MCY conjugates act as 358 intestinal FXR antagonists that balance the FXR agonistic activity of microbiota-derived free 359 BAs, as part of a regulatory circuitry that fine tunes BA signaling within the hepatobiliary system 360 (Fig. 5k). 361 362 DISCUSSION 363 BAs represent gut microbiota-dependent metabolites whose pervasive effects on human 364 physiology are among the most well studied4,7,43,53,54. Correspondingly, the biochemical 365 mechanisms by which the host may regulate their activities are of significant interest in the 366 context of human health and disease. In the case of BAs, their taurine conjugation by the host 367 and subsequent deconjugation by intestinal microbiota provide a classical example for the 368 regulation of metabolite abundance via opposing host- and microbiota-dependent pathways. 369 Our identification of VNN1/Vanin 1-dependent BA-MCY conjugates as FXR antagonists reveals 370 a previously unrecognized host-dependent layer of BA metabolism that counteracts the 371 physiological functions of free BAs (Fig. 5k). In vitro protein-protein interaction assays 372 demonstrated strong FXR antagonistic activity for CA-MCY, CDCA-MCY, and βMCA-MCY (Fig. 373 4a,b and Extended Data Fig. 10a), whereas free BAs generally act as FXR agonists14. 374 Supplementation of mice with CDCA-MCY increased total BA production and expression of the 375 enzymes that catalyze the rate limiting steps in BA biosynthesis, consistent with CDCA-MCY 376 functioning as an FXR antagonist in vivo. This is in contrast to its parent compound CDCA, 377 which acts as an FXR agonist and reduces overall BA levels55. BA-MCYs were most abundant 378 13 in intestinal tissues (Fig. 3a), where VNN1 is highly expressed, and appear to get oxidized 379 quickly into the inactive BA-MCYO derivatives when entering the general circulation. Similar to 380 other intestinal FXR antagonists46,52, CDCA-MCY supplementation alleviated hepatic lipid 381 accumulation in mice fed a high cholesterol diet (Fig. 5i,j). Thus it seems likely that, following 382 their re-uptake in the intestine, conversion of free BAs (FXR agonists) into BA-MCYs (FXR 383 antagonists) via the pantetheinase VNN1 represents an important component of the feedback 384 mechanisms regulating BA biosynthesis and other FXR-dependent phenotypes, including 385 diverse aspects of fatty acid metabolism15,56,57. Given the dysregulation of BA levels in type II 386 diabetes, metabolic syndrome, and the cholestatic diseases12,58,59, BA-MCYs may have 387 significant therapeutic potential. We further demonstrate that dietary fiber can upregulate 388 production of BA-MCY conjugates in mice, suggesting that the levels of these conjugates can 389 be regulated by diet and prebiotic or probiotic supplements, which may have translational 390 potential in conditions of dysregulated immune or metabolic homeostasis. 391 The profound effects of MCY-conjugation on the biological activity of BAs led us to investigate 392 the roles of both host and microbiota in the production of the MCY conjugates. Levels of 393 unconjugated, free BAs are largely microbiota-dependent, since the vast majority of free BAs in 394 SPF mice is derived from microbial deconjugation of the corresponding taurine conjugates. 395 Using a series of stable isotope supplementation studies, we show that BA-MCYs are derived 396 from conjugation with cysteamine or another CoA-derived metabolite, instead of reduction of 397 corresponding taurine derivatives. Consistent with the idea that BA-taurine and BA-MCY 398 conjugates have distinct biosynthetic origins, BA-MCY conjugates accumulate in mice deficient 399 in the enzyme conjugating BAs with taurine (BAAT)36. The intriguing connection to CoA 400 breakdown metabolism led us to uncover the role of the pantetheinase VNN1 for BA-MCY 401 production, demonstrating that a host enzyme that is highly expressed in the intestine, the site 402 of BA re-uptake, plays a central role for BA-MCY production (Fig. 3d and Extended Data Fig. 403 7d). BA-MCY biosynthesis is strongly reduced, but not fully abolished in Vnn1-/- mice, 404 suggesting that other pantetheinases (VNN3 in mouse or VNN2 in humans) may also contribute. 405 Furthermore, although bacteria have no close homologs of vertebrate pantetheinases, it is 406 possible that other bacterial hydrolases or peptidases have similar activity and also contribute to 407 the residual amounts of BA-MCYs observed in Vnn1-/- mice. 408 VNN1/Vanin 1 serves diverse functions in lipid metabolism and forms an important link between 409 lipid accumulation and hepatic diseases37,39, which is of particular interest in light of our finding 410 that BA-MCY supplementation alleviated lipid accumulation in the liver of HCD mice (Fig. 5i,j). 411 14 Clarifying the role of VNN1 for BA-MCY production and other aspects of BA metabolism in the 412 intestine and other tissues may provide new insights in associated phenotypes. More generally, 413 the identification of the role of VNN1 in BA-MCY production reveals an intriguing connection 414 between BA signaling and CoA breakdown pathways, which are extensively regulated by 415 nutritional state, feeding back on many other aspects of metabolism34. BA-MCY conjugates can 416 be hydrolyzed by microbial BSH, albeit perhaps less efficiently than BA-taurine conjugates. To 417 what extent microbial deconjugation of BA-MCYs is physiologically relevant is unclear, however, 418 it may represent an additional mechanism by which the microbiota contribute to regulating the 419 balance of FXR agonists and antagonists. 420 Taken together, our results suggest that MCY conjugation of BAs by the host balances 421 microbiota-dependent taurine deconjugation, as part of a metabolic network integrating host- 422 and microbiota-derived pathways that regulates FXR-dependent BA production, fat metabolism, 423 CoA metabolism and possibly other processes downstream of BAs. 424 425 15 METHODS 426 427 Mice 428 C57BL/6 (Jax, 000664) and Nr1h4-/- (Jax, 007214) mice were originally purchased from The 429 Jackson Laboratories and bred at Weill Cornell Medicine (WCM). Vnn1-/- mice60 were bred at 430 the Yale School of Medicine (gift of Dr. Phillipe Nasquet (CIML, France) and Dr. Ruslan 431 Medzhitov (Yale University, USA). GF C57BL/6J mice were bred and housed in flexible PVC 432 isolators (Park Bioservices) at WCM. Gnotobiotic mice were maintained in Sentry sealed 433 positive pressure cages (SPP, Allentown) for the duration of the experiments. All other mice 434 were maintained under specific pathogen-free condition. All mice used were between 6 and 12 435 weeks old, age- and sex-matched for each experiment, maintained on a 12-h light–dark cycle, 436 an average ambient temperature of 21 °C and an average humidity of 48%, and provided food 437 and water ad libitum. When studying the effects of dietary fiber32, mice were given an inulin fiber 438 diet (D16052309, Research Diets, Inc.) supplemented with 30% fiber (26% inulin and 4% 439 cellulose) or a calorie-matched control diet (D12450J-1.5, Research Diets, Inc.) containing 4.7% 440 cellulose. The duration of the fiber dietary intervention was two weeks unless otherwise stated. 441 For the experimental model of hypercholesteremia, mice were given a high cholesterol diet 442 (C23041301, Research Diets, Inc.) supplemented with 1% cholesterol, or standard mouse chow. 443 The duration of the high cholesterol dietary intervention was four weeks. All protocols were 444 approved by the Weill Cornell Medicine Institutional Animal Care and Use Committees (IACUC), 445 and all mice were used in accordance of governmental and institutional guidelines for animal 446 welfare. 447 448 Antibiotic treatment 449 Mice were provided autoclaved drinking water supplemented with a cocktail of broad-spectrum 450 antibiotics as previously described61: ampicillin (0.5 mg/ml, Santa Cruz), gentamicin (0.5 mg/mL, 451 Gemini Bio Products), metronidazole (0.5 mg/mL Sigma), neomycin (0.5 mg/ml, Sigma), 452 vancomycin (0.25 mg/ml, Chem-Impex International), and saccharin (4 mg/mL, Sweet’N Low, 453 Cumberland Packing). Mice in the control group were provided with autoclaved drinking water 454 supplemented with saccharin alone. Saccharin was added to make the antibiotic cocktail more 455 palatable. Antibiotic treatment was started two weeks before the experiments and continued for 456 the duration of the experiments with antibiotic and saccharin only control water replaced weekly. 457 458 16 Tissue collection and processing 459 Mice were euthanized by CO2 narcosis according to institutional policies. Just prior to 460 euthanasia, fecal samples were collected and frozen on dry ice. After euthanasia, whole blood 461 was collected by subxiphoid cardiac puncture using a 1 mL syringe fitted with a 25G x 5/8” 462 needle (both BD). Blood was immediately transferred to a SST Microtainer tube (BD) and 463 allowed to clot for 30 minutes prior to centrifugation according to the manufacturer’s instructions. 464 Serum was then collected, transferred to a new tube, and frozen on dry ice. After blood 465 collection, animals were perfused through the left ventricle with 20 mL of ice-cold Ca2+/Mg2+ free 466 DPBS (Corning) to flush the remaining vascular contents. Liver and ileum samples were then 467 collected and either snap frozen on dry ice for metabolomics or placed into RNAprotect (Qiagen) 468 for RNA extraction. 469 470 Human serum samples 471 Serum samples were obtained from a cohort described previously62. The original study protocol 472 was approved by the Duke Health Institutional Review Board (IRB) at Duke University under 473 protocol number Pro00093322, and registered on ClinicalTrials.gov, with the identified number 474 NCT04055246. Informed consent was obtained from all participants. 475 476 Human fecal microbiota transplantation 477 For FMT studies, donor fecal samples were collected from healthy subjects and re-suspended 478 in PBS with 10% glycerol in an anaerobic chamber63. Samples were obtained following 479 Institutional Review Board-approved protocols from the JRI IBD Live Cell Bank Consortium at 480 Weill Cornell Medicine and informed consent was obtained from all participants. Fecal 481 suspensions from individual donors were administered to recipient germ-free mice by oral 482 gavage (100 μL per mouse). Transplanted animals were maintained in sterile isocages for two 483 to four weeks. Animals were evaluated for successful transplantation by comparing 16S 484 sequencing between human donors and recipient mice. 485 486 Metabolite extraction from mouse liver, small intestine, and cecum 487 Intact mouse liver was frozen and stored at -80°C before processing. Frozen tissues were 488 crushed and grinded using a pre-chilled mortar and pestle. Dry ice was added to mortar and 489 pestle throughout the homogenization process to prevent tissues from thawing. The resulting 490 powdered samples were sonicated for 1 min with 5 mL methanol in 20 mL glass scintillation 491 vials at a ratio with 10 μL solvent per mg, followed by another 10 min of vigorous stirring. 492 17 Extracts were pelleted at 5,000 g for 5 min, and supernatants were transferred to another 20 mL 493 glass vials. Remaining pellets were further extracted with another 10 min of vigorous stirring in 5 494 mL ethanol. The supernatants were combined and then dried in a SpeedVac (ThermoFisher 495 Scientific) vacuum concentrator. Dried materials were resuspended in 300 μL of methanol. 496 Samples were pelleted at 5,000 g for 5 min and clarified extracts were transferred to fresh 497 HPLC vials and stored at −20 °C until analysis. 498 499 Metabolite extraction from mouse serum samples 500 800 µL of methanol were added to 200 µL of serum in 1.7 mL Eppendorf tubes. The tubes were 501 sonicated for 1 min followed by another 10 min of vigorous stirring. Extracts were pelleted at 502 5,000 g for 5 min, and supernatants were transferred to 2 mL HPLC vials. Remaining pellets 503 were further extracted with another 10 min of vigorous stirring in 500 µL ethanol. Extracts were 504 pelleted at 5,000 g for 5 min, and the combined supernatants were dried in a SpeedVac 505 (ThermoFisher Scientific) vacuum concentrator. Samples were resuspended in 100 μL of 506 methanol and pelleted at 5,000 g for 5 min. Clarified extracts were transferred to fresh HPLC 507 vials and stored at −20 °C until analysis. 508 509 Metabolite extraction from mouse feces 510 600 µL of methanol were added to 30 mg of feces in 1.7 mL Eppendorf tubes. The tubes were 511 sonicated for 1 min, followed by another 10 min of vigorous stirring. Extracts were pelleted at 512 5,000 g for 5 min, and supernatants were transferred to 2 mL HPLC vials. Remaining pellets 513 were further extracted with another 10 min of vigorous stirring in 600 µL ethanol. Extracts were 514 pelleted at 5,000 g for 5 min, and the combined supernatants were then dried in a SpeedVac 515 (ThermoFisher Scientific) vacuum concentrator. Samples were then resuspended in 100 μL of 516 methanol and pelleted at 5,000 g for 5 min, and the clarified extracts were transferred to fresh 517 HPLC vials and stored at −20 °C until analysis. 518 519 Analytical methods and equipment overview 520 (a) Mass spectrometry: High resolution LC−MS was performed on a ThermoFischer Scientific 521 Vanquish UHPLC system coupled with a Thermo Q-Exactive HF hybrid quadrupole-orbitrap 522 high-resolution mass spectrometer equipped with a HESI ion source. Metabolites were 523 separated using acetonitrile containing 0.1% formic acid (organic phase) and 0.1% formic acid 524 in water (aqueous phase) as solvents on a ThermoFisher Scientific Hypersil GOLD C18 column 525 (150 mm × 2.1 mm, particle size 1.8 μm). The gradient started at 1% organic for 3 min after 526 18 injection and increased linearly to 100% organic over 20 min, then 100% organic for 5 min, and 527 down to 1% organic for 3 min at a flow rate of 0.5 mL/min. Mass spectrometer parameters: 528 spray voltage 3.5 kV, capillary temperature 380 °C, probe heater temperature 400 °C; 60 sheath 529 flow rate, 20 auxiliary flow rate, and one spare gas; S-lens RF level 50, resolution 240,000, AGC 530 target 3 × 106. The instrument was calibrated weekly with positive and negative ion calibration 531 solutions (ThermoFisher). Each sample was analyzed in negative and positive ionization modes 532 using a m/z range of 100 to 800. (b) NMR spectroscopy: NMR spectroscopy was performed on 533 a Varian INOVA 600 MHz NMR spectrometer (600 MHz 1H reference frequency, 151 MHz for 534 13C) equipped with an HCN indirect-detection probe. Non-gradient phase-cycled dqfCOSY 535 spectra were acquired using the following parameters: 0.6 s acquisition time; 400–600 complex 536 increments; 8, 16 or 32 scans per increment. HSQC and HMBC spectra were acquired with 537 these parameters: 0.25 s acquisition time, 200–500 increments, 8–64 scans per increment. 1H, 538 13C-HMBC spectra were optimized for JH,C = 6 Hz. HSQC spectra were acquired with or without 539 decoupling. NMR spectra were processed and baseline corrected using MestreLabs MNOVA 540 software packages. 541 542 Feature detection and characterization 543 LC−MS RAW files for all serum samples were converted to mzXML format (centroid mode) 544 using MSconvert (ProteoWizard version 3.0.18250-994311be0), followed by analysis using the 545 XCMS analysis feature in Metaboseek version 0.9.7 (metaboseek.com)31 based on the 546 centWave XCMS algorithm to extract features64,65. Peak detection values were set as: 4 ppm, 3 547 to 20 peakwidth, 3 snthresh, 3 and 100 prefilter, FALSE fitgauss, 1 integrate, TRUE 548 firstBaselineCheck, 0 noise, wMean mzCenterFun, -0.005 mzdiff. XCMS feature grouping 549 values were set as: 0.2 minfrac, 2 bw, 0.002 mzwid, 500 max, 1 minsamp, FALSE usegroup. 550 Metaboseek peak filling values set as: 5 ppm_m, 5 rtw, TRUE rtrange. Resulting tables of all 551 detected features were then processed with the Metaboseek data explorer. To remove 552 background derived features, we first applied filters that only retained entries with a retention 553 time window of 1 to 20 min, and then applied maximum intensity (at least one repeat > 10,000), 554 and Peak Quality (>0.98) thresholds, as calculated by Metaboseek31. To select differential 555 features, we applied a filter retaining entries with peak area ratios more than 5-fold reduced or 556 5-fold increased in GF mice compared to SPF mice, as calculated by Metaboseek31. We 557 manually curated the resulting list to remove false positive entries, i.e., features that upon 558 manual inspection of raw data were not differential. For verified differential features, we 559 examined elution profiles, isotope patterns, and MS1 spectra to find molecular ions and remove 560 19 adducts, fragments, and isotope peaks. Remaining masses were put on the inclusion list for 561 MS/MS (ddMS2) characterization. Positive and negative ionization mode data were processed 562 separately. To acquire MS2 spectra, we ran a top-10 data dependent MS2 method on the 563 Thermo Q-exactive-HF mass spectrometer with MS1 resolution 60,000, AGC target 1 × 106, 564 maximum IT (injection time) 50 ms, MS2 resolution 45,000, AGC target 5 × 105, maximum IT 80 565 ms, isolation window 1.0 m/z, stepped NCE (normalized collision energy) 10 and 30 for positive 566 and negative ionization mode, dynamic exclusion 3 s. To focus on BAs and BA derivatives, we 567 selectively analyzed differential features whose MS/MS spectra contain fragment ions for 568 complete 24-carbon backbones (Supplementary Table 1). 569 570 MS2-based molecular networking 571 A MS2 molecular network was created using Metaboseek version 0.9.7 31 and visualized in 572 Cytoscape version 3.9.166. Features were matched with their respective MS2 scan within an m/z 573 window of 5 ppm and a retention time window of 15 s, using the MS2scans function. To 574 construct the molecular network, tolerance of the fragment peaks was set to m/z of 0.001 or 3 575 ppm, minimum number of peaks was set to 1, with a 2% noise level. Once the network was 576 constructed, a cosine value of 0.7 was used, the number of possible connections was 577 constrained to 6, and the maximum cluster size was restricted to 200 for both negative ion mode 578 and positive ion mode. 579 580 General procedures for chemical syntheses 581 Unless noted otherwise, all chemicals and reagents were purchased from Sigma-Aldrich. 582 Solutions and solvents sensitive to moisture and oxygen were transferred via standard syringe 583 and cannula techniques. Acetic acid, acetonitrile, dichloromethane, and methanol used for 584 chromatography and as a reagent or solvent were purchased from Fisher Scientific. Flash 585 chromatography was performed using Teledyne Isco CombiFlash systems using Teledyne Isco 586 RediSep Rf silica and C18 columns. See Supplementary Information for synthetic schemes, 587 protocols, NMR spectroscopic data, and spectra. NMR spectra were processed and baseline 588 corrected using Mestrelab Mnova (v14.2.1-27684) software packages. 589 590 Generation of correlation plots 591 Correlation plots were created using GraphPad Prism version 10.2.0. The R² values from non-592 linear regression analyses were calculated using GraphPad Prism and used as a measure of 593 goodness-of-fit. Quantification was performed via integration in Xcalibur 4.1 Qualbrowser 594 20 (v4.1.31.9, Thermo Scientific). Cholic acid, -muricholic acid, -muricholic acid, -muricholic 595 acid, chenodeoxycholic acid, ursodeoxycholic acid, 7-ketodeoxycholic acid, deoxycholic acid 596 were quantified and included in “total free BAs”, and their corresponding MCY, MCYO, and 597 MCYO2 conjugates were quantified and included in “total BA-MCYs”. The average total amount 598 of free BAs in groups of SPF mice used as control for different experiments was used to 599 normalize data for total free BAs and total BA-MCYs for all groups. 600 601 Administration of metabolites 602 Taurine-d4 (Cambridge isotope laboratories) dissolved in saline was delivered intraperitoneally 603 daily at a rate of 100 mg/kg body weight in a volume of 100 μL. L-cysteine-3,3-d2 was delivered 604 at a rate of 100 mg/kg body weight per day by oral gavage. The metabolites were administered 605 for a duration of 5 days. CDCA-MCY or CDCA-d5-MCY (dissolved in DMSO and further diluted 606 in corn oil) was delivered by oral gavage at a rate of 50 mg/kg body weight per day in a volume 607 of 100 μL for two weeks. The vehicle control group received DMSO and corn oil. In the high 608 cholesterol diet model, mice were treated under the same conditions for a total of six weeks 609 (two weeks prior to initiation of the high cholesterol diet, and throughout the diet exposure 610 period. In some experiments, a 10-fold lower dose (5 mg/kg) was used for CDCA-MCY or 611 CDCA-d5-MCY. 612 613 VNN1 in vitro assays with pantetheine derivatives 614 Reactions were initiated by adding 10 μL 0.02 μM recombinant human ΔN490aa VNN1 protein 615 (Sino Biological, with C-terminal poly His tag, reconstituted in PBS pH 7.0) to 10 μL pantetheine 616 (or CA-pantetheine derivative) diluted with PBS (pH 7.0) so that the final concentrations of 617 pantetheine (derivative) were 20, 40, 80, 160, 320, and 640 μM. The mixtures were quenched 618 with 180 μL methanol after allowing the reaction to proceed for 15 min at 37 °C and directly 619 injected for HPLC-HRMS analysis. The HPLC-HRMS analysis method was the same as 620 described in the ‘Analytical methods and equipment overview’ section, except that a shorter 621 HPLC method was used. The gradient was as follows: The gradient started at 1% organic for 2 622 min, linearly increasing to 98% organic over 2.5 min, then 98% organic for 2.5 min, shifting back 623 to 1% organic in 0.1 min until 9 min. Pantothenic acid was detected in ESI+ mode at m/z 624 220.1180 ± 5 ppm. Initial velocity of each reaction was calculated by dividing the concentration 625 of produced pantothenic acid determined using a standard curve of D-calcium pantothenate 626 (Acros Organics) with 15 min. Michaelis-Menten kinetic regression was performed using 627 GraphPad Prism version 10.2.0. 628 21 629 Analysis of CDCA-d5-MCY metabolism in vivo 630 Use of the Label Finder tool in Metaboseek31 facilitated untargeted detection of any D4-labeled 631 (Δm/z, 4.0251) and D5-labeled (Δm/z, 5.0314) metabolites derived from administered CDCA-d5-632 MCY in SPF, GF, ABX, or GF mice monocolonized with B. ovatus strains. CDCA-d5-MCY was 633 administered by oral gavage at a rate of 50 mg/kg body weight as described above. The vehicle 634 control group received DMSO and corn oil. Peak detection values were set as follows: RT 635 tolerance, 3 s; peak width tolerance factor, 10; m/z tolerance, 5 ppm; RT window, 5 s; minimum 636 intensity ratio unlabelel peak/labeled peak in unlabeled samples, 10; maximum intensity ratio 637 unlabelel peak/labeled peak in unlabeled samples. 5. Resulting tables of all detected features 638 were then processed with the Metaboseek data explorer. To remove background derived 639 features, we first applied filters that only retained entries with a retention time window of 5 to 20 640 min and m/z between 300 to 600, and then applied a Peak Quality (>0.98) threshold, as 641 calculated by Metaboseek31. The resulting table was then processed with Microsoft Excel to 642 generate bubble plots67. 643 644 DiscoverX assay for activity against human FXR 645 Tauro-β-muricholic acid (Sigma) and synthetic MCY conjugates including CA-MCY, CA-MCYO, 646 CDCA-MCY, and βMCA-MCY were tested in a cell-based assay on FXR with known ligands, 647 GW4064 and DY268, in both agonist (item # 86-0003P-2453AG) and antagonist modes (item # 648 86-0003P-2453AN), respectively (assays were performed by DiscoverX/Eurofins). 649 CHO-K1 cells tested negative for mycoplasma and were expanded from freezer stocks 650 according to standard procedures. Cells were not authenticated. Cells were seeded in a total 651 volume of 20 L into white walled, 384-well microplates and incubated at 37 °C for the 652 appropriate time prior to testing. Assay media contained charcoal-dextran filtered serum to 653 reduce the level of hormones present. For agonist determination, cells were incubated with 654 sample to induce response. Intermediate dilution of sample stocks was performed to generate 655 5X sample in assay buffer. 5 L of 5X sample was added to cells and incubated at 37 °C or 656 room temperature for 3-16 hr. Final assay vehicle concentration was 1 L. For antagonist 657 determination, cells were pre-incubated with antagonist followed by agonist challenge, GW4064 658 (0.37 µM, EC80), obeticholic acid (25 µM), or CDCA (25 µM). Intermediate dilution of sample 659 stocks was performed to generate 5X sample in assay buffer. 5 L of 5X sample was added to 660 cells and incubated at 37 °C or room temperature for 60 min. Vehicle concentration was 1%. 5 661 L of 6X EC80 agonist in assay buffer was added to the cells and incubated at 37 °C or room 662 22 temperature for 3-16 hr. Compound activity was detected by chemiluminescent signals 663 indicating ligand binding that induces FXR activation, translocation, and co-activator interaction. 664 For signal detection, assay signal was generated through a single addition of 12.5 or 15 L (50% 665 v/v) of PathHunter Detection reagent cocktail, followed by a 1 hr incubation at room temperature. 666 Microplates were read following signal generation with a PerkinElmer EnvisionTM instrument for 667 chemiluminescent signal detection. Agonist mode measures percentage activity relative to 668 maximum value activated by GW4064 (100% activation)45. Antagonist mode measures 669 percentage activity relative to maximum value activated by DY268 (100% activation). For 670 agonist mode assays, percentage activity was calculated using the following formula: activity (%) 671 =100 x (mean RLU of test sample - mean RLU of vehicle control) / (mean MAX control ligand - 672 mean RLU of vehicle control). For antagonist mode assays, percentage inhibition was 673 calculated using the following formula: inhibition (%) =100 x (1 - (mean RLU of test sample - 674 mean RLU of vehicle control) / (mean RLU of EC80 control - mean RLU of vehicle control). 675 676 DiscoverX assay for cell viability against human primary hepatocytes 677 Chenodeoxycholic acid (Sigma) and synthetic CDCA-MCYO were tested in a cell-based assay 678 using human primary hepatocytes (assays were performed by DiscoverX/Eurofins, item # 5137). 679 Cryopreserved human hepatocytes were thawed and plated into collagen-coated 96-well plates 680 in the Plating Medium (serum-containing) at a density of 0.2 × 106 viable cells/mL. Hepatocytes 681 were cultured at 37 °C and 5% CO2 for 18-24 h. After 18-24 h of plating, the hepatocytes were 682 incubated with the test compound in the Incubation Medium (serum-free). Cell viability is 683 measured by CellTiter-Glo (Promega) after 24 h compound incubation. Compounds were tested 684 in duplicates at multiple concentrations (0.15, 0.45, 1.5, 4.5, 15, 45, 150, and 450 µM) for IC50 685 determinations. Chlorpromazine was tested at multiple concentrations to obtain an IC50 value. 686 The final DMSO concentration was 0.5%. For data analysis, the percent of control activity was 687 calculated by comparing the readings in the presence of the test compound to the vehicle 688 control. IC50 values (concentration causing a half-maximal inhibition of the control value) are 689 determined by non-linear regression analysis of the concentration response curves using the 690 Hill equation. 691 692 Quantitative real-time PCR and ELISA 693 Total RNA was isolated from tissues using the RNeasy Plus mini kit (Qiagen), and cDNA was 694 synthesized using the High-Capacity cDNA Reverse Transcription Kit with Multiscribe™ 695 Reverse Transcriptase (Thermo Fisher), according to the protocol provided by the manufacturer. 696 23 Quantitative PCR reactions were set up using the Power SYBR Green PCR Master Mix 697 (Thermo Fisher) and run on a QuantStudio 6 Flex Real-Time PCR System (Applied Biosystems) 698 using QuantStudio Real-Time PCR software v1.0. The primers are listed in Supplementary 699 Table 543. FGF15 ELISA was performed using the mouse FGF15 ELISA Kit (LSBio) as per the 700 manufacturer’s protocol. 701 702 Histological analysis 703 After euthanasia and perfusion, a single lobe of the liver was carefully removed and fresh-frozen 704 in OCT using a dry-ice/isobutane bath. Tissue blocks were stored at -80 ºC until sectioning. 705 Sectioning was performed on a Leica CM3050S cryostat to a thickness of 10 µm and collected 706 on Superfrost Plus Slides (VWR). Sections were then again stored at -80 ºC prior to staining. 707 For Oil Red O staining, slides were stained according to the manufacturer’s instructions using 708 an Oil Red O Stain Kit (abcam) omitting the nuclear counterstaining step. Stained slides were 709 imaged on a Nikon Eclipse Ti microscope using a 20X Plan APO 0.45 NA lens using identical 710 illumination and detection settings. Four random 20X fields were captured per section in two 711 separate tissue sections for a total of 8 fields per animal. Total Oil Red O pixel area in each 20X 712 field was calculated using a custom Fiji script (https://github.com/cnp9004/Fiji-ORO-macro). 713 Pixel area was then averaged over the 8 captured sections and converted to area in µm2 using 714 the known pixel:µm calibration of the imaging system. 715 716 Monocolonization of GF mice 717 WT or mutant B. ovatus strains generated previously32 were grown in a TYGB/Mega medium at 718 37 °C overnight inside an anaerobic chamber. GF mice were monocolonized by oral gavage of 719 the bacterial culture (200 μL; ~1 × 107 colony-forming units (CFU)) and maintained in Sentry 720 sealed positive pressure cages (SPP, Allentown) for the duration of the experiments. The level 721 of colonization was determined by quantifying the bacterial load in mouse fecal pellets (CFU per 722 gram of feces). In brief, approximately 5 mg of fecal material was resuspended in 200 μL of 723 prereduced Gibco PBS buffer (pH 7.4). Then a 10-fold serial dilution (to 10−4) was made in the 724 same buffer on a 96-well plate, and 50 μL from each 10−4-diluted well was plated onto 725 anaerobically prereduced TSAB agar at 37 °C. After 24 h, colonies appeared and the CFUs for 726 fecal samples collected from WT and mutant colonized GF mice were calculated after 727 normalizing to fecal weight. 728 729 https://github.com/cnp9004/Fiji-ORO-macro 24 In vitro experiments with gut bacteria and mouse fecal suspensions 730 50 M of Enterococcus faecalis, Lactiplantibacillus plantarum, Ruminococcus gnavus, 731 Enterocloster bolteae, or Clostridium scindens (OD600 0.5-0.6) were inoculated with 100 M of 732 CA-MCY, CA-MCYO or CA-MCYO2 in 450 L of Reinforced Clostridial Medium (RCM) medium 733 and incubated at 37 °C for 48 h. Fecal samples were collected from 3 SPF mice in different 734 cages. Approximately 50 mg of mice fecal samples were homogenized in 4 mL Brain Heart 735 Infusion (BHI) medium and incubated at 37 °C for 24 h. Subsequently, passaging was then 736 performed in 15 ml tubes at a dilution of 1:4 every 24 h for a total of 2 passages. 50 L of 737 passaged bacteria community (OD600 0.5-0.6) was inoculated with 100 M of CA-MCY, CA-738 MCYO or CA-MCYO2 in 450 L of BHI medium and incubated at 37 °C for 48 h. After in vitro 739 culture, 200 μL of culture was added to 300 μL ice-cold methanol at a final concentration of 60% 740 methanol followed by overnight incubation at 4 °C. Extracted metabolites were stored at -80 °C. 741 Before the mass spectrometry analysis, samples were spun down at 12,000 g for 5 min, and the 742 supernatant was transferred to a 96-well plates followed by dilution 1:1 (v:v) in 50% methanol 743 containing 2.5 μg/mL phenolsulfonphthalein for liquid chromatography-tandem mass 744 spectrometry (LC-MS/MS) analysis. Detailed methods for the comparative metabolomic 745 analyses have been described previously42. Briefly, raw files were converted to mzXML format 746 via GNPS Vendor Conversion and submitted the Global Natural Products Social Molecular 747 Networking Database (GNPS, University of California at San Diego) for spectral identification. 748 749 Statistical analysis 750 All statistical analysis were performed with GraphPad Prism version 10.2.0 or Metaboseek 751 version 0.9.7. The statistical details of experiments can be found in the figure legends. Reported 752 n values are the total samples per group. The P values of datasets were determined by 753 unpaired two-tailed Student’s t-test with 95% confidence interval, unless specified otherwise. 754 Normal distribution was assumed. If equal variances between two groups could not be assumed, 755 Welch’s correction was performed. No statistical methods were used to predetermine sample 756 size. The experiments were not randomized, and the investigators were not blinded to allocation 757 during experiments and outcome assessment. No animals were excluded from the analysis 758 unless clearly indicated. The entity n represents biologically independent samples and not 759 technical replicates unless specified otherwise. Error bars show the s.e.m., unless specified 760 otherwise. 761 762 25 Data availability 763 Raw sequencing reads were uploaded to the Sequence Read Archive (SRA), and MS data for 764 all mouse metabolome samples analyzed in this study were uploaded to the GNPS Web site 765 (massive.ucsd.edu) under MassIVE ID number MSV000090974 and are publicly available. 766 767 Code availability 768 The custom Fiji script used for the analysis of liver lipid accumulation imaging is available at 769 https://github.com/cnp9004/Fiji-ORO-macro. 770 771 Acknowledgements 772 We thank members of the Schroeder and Artis laboratories for discussion and critical reading of 773 the manuscript. We thank Bennett Fox for assistance with mass spectrometry. We thank all 774 contributing members of the JRI IBD Live Cell Bank consortium, which is supported by the Jill 775 Roberts Institute for Research in IBD, the Jill Roberts Center for IBD, Cure for IBD, the Rosanne 776 H. Silbermann Foundation, the Sanders Family, and Weill Cornell Medicine Division of Pediatric 777 Gastroenterology and Nutrition. The cartoons in Figure 1a was created with BioRender.com. All 778 chemical structures were created with ChemDraw. 779 780 Funding 781 This work was supported by the Crohn's & Colitis Foundation (to M.A.), the Thomas C. King 782 Pulmonary Fellowship, the Weill Cornell Medicine Fund for the Future, the Sackler Brain and 783 Spine Institute Research Grant, and a Brain and Behavior Research Foundation (NARSAD) 784 Young Investigator Award (all to C.N.P.), Office of Naval Research grant N00014-18-1-2616 (to 785 L.A.D), the Howard Hughes Medical Institute (to F.C.S.), Kenneth Rainin Foundation and the W. 786 M. Keck Foundation (all to C.-J.G.), AGA Research Foundation, WCM-RAPP Initiative, CURE 787 for IBD, the Jill Roberts Institute for Research in IBD, Kenneth Rainin Foundation, the Sanders 788 Family Foundation, Rosanne H. Silbermann Foundation, the Glenn Greenberg and Linda Vester 789 Foundation, the Allen Discovery Center Program, a Paul G. Allen Frontiers Group advised 790 program of the Paul G. Allen Family Foundation (all to D.A.), and the National Institutes of 791 Health (K99AI173660 to MA, K08MH130773 and NIAID Mucosal Immunology Studies Team 792 Young Investigator Award to C.N.P., GM131877 to F.C.S., DK116187 to L.A.D, DP2 HD101401 793 and DK135816 to C.-J.G., and DK126871, AI151599, AI095466, AI095608, AR070116, 794 AI172027, DK132244, all to D.A.). 795 796 https://github.com/cnp9004/Fiji-ORO-macro 26 Author contributions 797 D.A. and F.C.S. supervised the study. T.H.W. carried out metabolomics and chemical synthesis 798 and analyzed most of the data. C.N.P. and M.A. conducted all animal experiments. W.-B.J. and 799 C.-J.G. provided the bacterial strains and helped with monocolonization experiments. S.K., E.H., 800 and I.M. helped with mouse experiments and various other assays. B.Z assisted with 801 metabolomic analyses and conducted VNN1 assays. J.L. and L.D. provided the human samples. 802 Y.F., D.V.G., and R.A.Q. provided BAAT knockout mouse metabolome samples and data 803 analysis as well as microbial analyses. T.H.W., M.A., C.N.P., D.A., and F.C.S. analyzed data 804 and wrote the manuscript with input from all co-authors. 805 806 Consortia 807 The members of the JRI IBD Live Cell Bank consortium are: David Artis, Randy Longman, 808 Gregory F. Sonnenberg, Ellen Scherl, Robbyn Sockolow, Dana Lukin, Vinita Jacob, Laura 809 Sahyoun, Michael Mintz, Lasha Gogokhia, Thomas Ciecierega, Aliza Solomon, Arielle Bergman, 810 Kimberley Chein, Elliott Gordon, Michelle Ramos, Kenny Joselin Castro Ochoa, Victoria Ribeiro 811 de Godoy, Adriana Brcic-Susak, Seun Oguntunmibi, Dario Garone, Caitlin Mason. 812 813 Competing interests 814 D.A. has contributed to scientific advisory boards at, Pfizer, Takeda, Nemagene and the 815 Kenneth Rainin Foundation. F.C.S. is a cofounder of Ascribe Bioscience and Holoclara Inc. The 816 other authors declare no competing interests. 817 818 Correspondence and requests for materials 819 All correspondence and material requests should be addressed to D.A. or F.C.S. 820 821 27 Main text references 822 1 Robertson, R. C., Manges, A. R., Finlay, B. B. & Prendergast, A. J. The Human Microbiome and 823 Child Growth - First 1000 Days and Beyond. Trends Microbiol 27, 131-147 (2019). 824 https://doi.org:10.1016/j.tim.2018.09.008 825 2 Fuchs, C. D. & Trauner, M. Role of bile acids and their receptors in gastrointestinal and hepatic 826 pathophysiology. Nature reviews. Gastroenterology & hepatology 19, 432-450 (2022). 827 https://doi.org:10.1038/s41575-021-00566-7 828 3 Thomas, C., Pellicciari, R., Pruzanski, M., Auwerx, J. & Schoonjans, K. Targeting bile -acid 829 signalling for metabolic diseases. Nat Rev Drug Discov 7, 678-693 (2008). 830 https://doi.org:10.1038/nrd2619 831 4 Wahlstrom, A., Sayin, S. I., Marschall, H. U. & Backhed, F. Intestinal Crosstalk between Bile Acids 832 and Microbiota and Its Impact on Host Metabolism. Cell Metab 24, 41-50 (2016). 833 https://doi.org:10.1016/j.cmet.2016.05.005 834 5 Belkaid, Y. & Hand, T. W. Role of the microbiota in immunity and inflammation. Cell 157, 121-835 141 (2014). https://doi.org:10.1016/j.cell.2014.03.011 836 6 Blander, J. M., Longman, R. S., Iliev, I. D., Sonnenberg, G. F. & Artis, D. Regulation of 837 inflammation by microbiota interactions with the host. Nat Immunol 18, 851-860 (2017). 838 https://doi.org:10.1038/ni.3780 839 7 Hang, S. Y. et al. Bile acid metabolites control T(H)17 and T-reg cell differentiation. Nature 576, 840 143-+ (2019). https://doi.org:10.1038/s41586-019-1785-z 841 8 Morais, L. H., Schreiber, H. L. t. & Mazmanian, S. K. The gut microbiota-brain axis in behaviour 842 and brain disorders. Nat Rev Microbiol (2020). https://doi.org:10.1038/s41579-020-00460-0 843 9 Fan, Y. & Pedersen, O. Gut microbiota in human metabolic health and disease. Nat Rev Microbiol 844 19, 55-71 (2021). https://doi.org:10.1038/s41579-020-0433-9 845 10 Nicholson, J. K. et al. Host-gut microbiota metabolic interactions. Science 336, 1262-1267 (2012). 846 https://doi.org:10.1126/science.1223813 847 11 Martin, A. M., Sun, E. W., Rogers, G. B. & Keating, D. J. The Influence of the Gut Microbiome on 848 Host Metabolism Through the Regulation of Gut Hormone Release. Front Physiol 10 (2019). 849 https://doi.org:ARTN 42810.3389/fphys.2019.00428 850 12 Molinaro, A., Wahlstrom, A. & Marschall, H. U. Role of Bile Acids in Metabolic Control. Trends 851 Endocrinol Metab 29, 31-41 (2018). https://doi.org:10.1016/j.tem.2017.11.002 852 13 Ramírez-Pérez, O., Cruz-Ramón, V., Chinchilla-López, P. & Méndez-Sánchez, N. The Role of the 853 Gut Microbiota in Bile Acid Metabolism. Annals of hepatology 16, s15-s20 (2017). 854 https://doi.org:10.5604/01.3001.0010.5494 855 14 Wang, H., Chen, J., Hollister, K., Sowers, L. C. & Forman, B. M. Endogenous bile acids are ligands 856 for the nuclear receptor FXR/BAR. Mol Cell 3, 543-553 (1999). https://doi.org:10.1016/s1097-857 2765(00)80348-2 858 15 Claudel, T., Staels, B. & Kuipers, F. The Farnesoid X receptor - A molecular link between bile acid 859 and lipid and glucose metabolism. Arterioscl Throm Vas 25, 2020-2031 (2005). 860 https://doi.org:10.1161/01.ATV.0000178994.21828.a7 861 16 Jiang, L., Zhang, H., Xiao, D., Wei, H. & Chen, Y. Farnesoid X receptor (FXR): Structures and 862 ligands. Computational and structural biotechnology journal 19, 2148-2159 (2021). 863 https://doi.org:10.1016/j.csbj.2021.04.029 864 17 Johnson, M. R., Barnes, S., Kwakye, J. B. & Diasio, R. B. Purification and characterization of bile 865 acid-CoA:amino acid N-acyltransferase from human liver. J Biol Chem 266, 10227-10233 (1991). 866 18 Killenberg, P. G. & Jordan, J. T. Purification and characterization of bile acid-CoA:amino acid N-867 acyltransferase from rat liver. J Biol Chem 253, 1005-1010 (1978). 868 https://doi.org:10.1016/j.tim.2018.09.008 https://doi.org:10.1038/s41575-021-00566-7 https://doi.org:10.1038/nrd2619 https://doi.org:10.1016/j.cmet.2016.05.005 https://doi.org:10.1016/j.cell.2014.03.011 https://doi.org:10.1038/ni.3780 https://doi.org:10.1038/s41586-019-1785-z https://doi.org:10.1038/s41579-020-00460-0 https://doi.org:10.1038/s41579-020-0433-9 https://doi.org:10.1126/science.1223813 https://doi.org:ARTN https://doi.org:10.1016/j.tem.2017.11.002 https://doi.org:10.5604/01.3001.0010.5494 https://doi.org:10.1016/s1097-2765(00)80348-2 https://doi.org:10.1016/s1097-2765(00)80348-2 https://doi.org:10.1161/01.ATV.0000178994.21828.a7 https://doi.org:10.1016/j.csbj.2021.04.029 28 19 Funabashi, M. et al. A metabolic pathway for bile acid dehydroxylation by the gut microbiome. 869 Nature 582, 566-570 (2020). https://doi.org:10.1038/s41586-020-2396-4 870 20 White, B. A., Lipsky, R. L., Fricke, R. J. & Hylemon, P. B. Bile acid induction specificity of 7 alpha-871 dehydroxylase activity in an intestinal Eubacterium species. Steroids 35, 103-109 (1980). 872 https://doi.org:10.1016/0039-128x(80)90115-4 873 21 Jones, B. V., Begley, M., Hill, C., Gahan, C. G. & Marchesi, J. R. Functional and comparative 874 metagenomic analysis of bile salt hydrolase activity in the human gut microbiome. Proc Natl 875 Acad Sci U S A 105, 13580-13585 (2008). https://doi.org:10.1073/pnas.0804437105 876 22 Ridlon, J. M., Harris, S. C., Bhowmik, S., Kang, D. J. & Hylemon, P. B. Consequences of bile salt 877 biotransformations by intestinal bacteria. Gut Microbes 7, 22-39 (2016). 878 https://doi.org:10.1080/19490976.2015.1127483 879 23 Pandak, W. M. & Kakiyama, G. The acidic pathway of bile acid synthesis: Not just an alternative 880 pathway(). Liver Res 3, 88-98 (2019). https://doi.org:10.1016/j.livres.2019.05.001 881 24 Sayin, S. I. et al. Gut microbiota regulates bile acid metabolism by reducing the levels of tauro-882 beta-muricholic acid, a naturally occurring FXR antagonist. Cell Metab 17, 225-235 (2013). 883 https://doi.org:10.1016/j.cmet.2013.01.003 884 25 Sun, L. et al. Gut microbiota and intestinal FXR mediate the clinical benefits of metformin. 885 Nature medicine 24, 1919-1929 (2018). https://doi.org:10.1038/s41591-018-0222-4 886 26 Makishima, M. et al. Identification of a nuclear receptor for bile acids. Science 284, 1362-1365 887 (1999). https://doi.org:10.1126/science.284.5418.1362 888 27 Reschly, E. J. et al. Evolution of the bile salt nuclear receptor FXR in vertebrates. J Lipid Res 49, 889 1577-1587 (2008). https://doi.org:10.1194/jlr.M800138-JLR200 890 28 Chiang, J. Y. Bile acids: regulation of synthesis. J Lipid Res 50, 1955-1966 (2009). 891 https://doi.org:10.1194/jlr.R900010-JLR200 892 29 Brandman, O. & Meyer, T. Feedback loops shape cellular signals in space and time. Science 322, 893 390-395 (2008). https://doi.org:10.1126/science.1160617 894 30 Tautenhahn, R., Patti, G. J., Rinehart, D. & Siuzdak, G. XCMS Online: A Web-Based Platform to 895 Process Untargeted Metabolomic Data. Analytical Chemistry 84, 5035-5039 (2012). 896 https://doi.org:10.1021/ac300698c 897 31 Helf, M. J., Fox, B. W., Artyukhin, A. B., Zhang, Y. K. & Schroeder, F. C. Comparative 898 metabolomics with Metaboseek reveals functions of a conserved fat metabolism pathway in C. 899 elegans. Nature communications 13, 782 (2022). https://doi.org:10.1038/s41467-022-28391-9 900 32 Arifuzzaman, M. et al. Inulin fibre promotes microbiota-derived bile acids and type 2 901 inflammation. Nature (2022). https://doi.org:10.1038/s41586-022-05380-y 902 33 Singh, V. et al. Dysregulated Microbial Fermentation of Soluble Fiber Induces Cholestatic Liver 903 Cancer. Cell 175, 679-694.e622 (2018). https://doi.org:10.1016/j.cell.2018.09.004 904 34 Naquet, P., Kerr, E. W., Vickers, S. D. & Leonardi, R. Regulation of coenzyme A levels by 905 degradation: the 'Ins and Outs'. Progress in lipid research 78, 101028 (2020). 906 https://doi.org:10.1016/j.plipres.2020.101028 907 35 Hunt, M. C., Siponen, M. I. & Alexson, S. E. The emerging role of acyl-CoA thioesterases and 908 acyltransferases in regulating peroxisomal lipid metabolism. Biochim Biophys Acta 1822, 1397-909 1410 (2012). https://doi.org:10.1016/j.bbadis.2012.03.009 910 36 Neugebauer, K. A. et al. BAAT gene knockout alters early life development and the gut 911 microbiome and reveals unusual bile acids in mice. J Lipid Res, 100297 (2022). 912 https://doi.org:10.1016/j.jlr.2022.100297 913 37 Bartucci, R., Salvati, A., Olinga, P. & Boersma, Y. L. Vanin 1: Its Physiological Function and Role in 914 Diseases. Int J Mol Sci 20 (2019). https://doi.org:10.3390/ijms20163891 915 https://doi.org:10.1038/s41586-020-2396-4 https://doi.org:10.1016/0039-128x(80)90115-4 https://doi.org:10.1073/pnas.0804437105 https://doi.org:10.1080/19490976.2015.1127483 https://doi.org:10.1016/j.livres.2019.05.001 https://doi.org:10.1016/j.cmet.2013.01.003 https://doi.org:10.1038/s41591-018-0222-4 https://doi.org:10.1126/science.284.5418.1362 https://doi.org:10.1194/jlr.M800138-JLR200 https://doi.org:10.1194/jlr.R900010-JLR200 https://doi.org:10.1126/science.1160617 https://doi.org:10.1021/ac300698c https://doi.org:10.1038/s41467-022-28391-9 https://doi.org:10.1038/s41586-022-05380-y https://doi.org:10.1016/j.cell.2018.09.004 https://doi.org:10.1016/j.plipres.2020.101028 https://doi.org:10.1016/j.bbadis.2012.03.009 https://doi.org:10.1016/j.jlr.2022.100297 https://doi.org:10.3390/ijms20163891 29 38 Kaskow, B. J., Proffitt, J. M., Blangero, J., Moses, E. K. & Abraham, L. J. Diverse biological 916 activities of the vascular non-inflammatory molecules - the Vanin pantetheinases. Biochem 917 Biophys Res Commun 417, 653-658 (2012). https://doi.org:10.1016/j.bbrc.2011.11.099 918 39 Yu, H. et al. Vanin1 (VNN1) in chronic diseases: Future directions for targeted therapy. Eur J 919 Pharmacol 962, 176220 (2024). https://doi.org:10.1016/j.ejphar.2023.176220 920 40 Yao, L. et al. A selective gut bacterial bile salt hydrolase alters host metabolism. Elife 7 (2018). 921 https://doi.org:10.7554/eLife.37182 922 41 Al-Dury, S., Wahlstrom, A., Stahlman, M., Backhed, F. & Marschall, H. U. Cyp3a11 Is Dispensable 923 for the Formation of Murine Bile Acids. J Hepatol 64, S436-S436 (2016). https://doi.org:Doi 924 10.1016/S0168-8278(16)00714-5 925 42 Guzior, D. V. et al. Bile salt hydrolase acyltransferase activity expands bile acid diversity. Nature 926 (2024). https://doi.org:10.1038/s41586-024-07017-8 927 43 Quinn, R. A. et al. Global chemical effects of the microbiome include new bile-acid conjugations. 928 Nature 579, 123-129 (2020). https://doi.org:10.1038/s41586-020-2047-9 929 44 Voegel, J. J., Heine, M. J., Zechel, C., Chambon, P. & Gronemeyer, H. TIF2, a 160 kDa 930 transcriptional mediator for the ligand-dependent activation function AF-2 of nuclear receptors. 931 The EMBO journal 15, 3667-3675 (1996). 932 45 Willson, T. M., Jones, S. A., Moore, J. T. & Kliewer, S. A. Chemical genomics: functional analysis 933 of orphan nuclear receptors in the regulation of bile acid metabolism. Medicinal research 934 reviews 21, 513-522 (2001). https://doi.org:10.1002/med.1023 935 46 Jiang, C. et al. Intestine-selective farnesoid X receptor inhibition improves obesity-related 936 metabolic dysfunction. Nature communications 6, 10166 (2015). 937 https://doi.org:10.1038/ncomms10166 938 47 Inagaki, T. et al. Fibroblast growth factor 15 functions as an enterohepatic signal to regulate bile 939 acid homeostasis. Cell Metabolism 2, 217-225 (2005). 940 https://doi.org:10.1016/j.cmet.2005.09.001 941 48 Kim, I. et al. Differential regulation of bile acid homeostasis by the farnesoid X receptor in liver 942 and intestine. J Lipid Res 48, 2664-2672 (2007). https://doi.org:10.1194/jlr.M700330-JLR200 943 49 Gonzalez, F. J., Jiang, C., Xie, C. & Patterson, A. D. Intestinal Farnesoid X Receptor Signaling 944 Modulates Metabolic Disease. Dig Dis 35, 178-184 (2017). https://doi.org:10.1159/000450908 945 50 Felton, J. et al. Diminished Gallbladder Filling, Increased Fecal Bile Acids, and Promotion of Colon 946 Epithelial Cell Proliferation and Neoplasia in Fibroblast Growth Factor (Fgf)15-Deficient Mice. 947 Gastroenterology 154, S582-S582 (2018). 948 51 Ito, S. et al. Impaired negative feedback suppression of bile acid synthesis in mice lacking 949 betaKlotho. J Clin Invest 115, 2202-2208 (2005). https://doi.org:10.1172/JCI23076 950 52 Kuang, J. L. et al. Hyodeoxycholic acid alleviates non-alcoholic fatty liver disease through 951 modulating the gut-liver axis. Cell Metabolism 35, 1752-+ (2023). 952 https://doi.org:10.1016/j.cmet.2023.07.011 953 53 Song, X. Y. et al. Microbial bile acid metabolites modulate gut ROR gamma(+) regulatory T cell 954 homeostasis. Nature 577, 410-+ (2020). https://doi.org:10.1038/s41586-019-1865-0 955 54 Campbell, C. et al. Bacterial metabolism of bile acids promotes generation of peripheral 956 regulatory T cells. Nature 581, 475-+ (2020). https://doi.org:10.1038/s41586-020-2193-0 957 55 Einarsson, C., Hillebrant, C. G. & Axelson, M. Effects of treatment with deoxycholic acid and 958 chenodeoxycholic acid on the hepatic synthesis of cholesterol and bile acids in healthy subjects. 959 Hepatology (Baltimore, Md.) 33, 1189-1193 (2001). https://doi.org:10.1053/jhep.2001.23790 960 56 Li, C. et al. Farnesoid X Receptor Agonists as Therapeutic Target for Cardiometabolic Diseases. 961 Front Pharmacol 11 (2020). https://doi.org:ARTN 124710.3389/fphar.2020.01247 962 https://doi.org:10.1016/j.bbrc.2011.11.099 https://doi.org:10.1016/j.ejphar.2023.176220 https://doi.org:10.7554/eLife.37182 https://doi.org:Doi https://doi.org:10.1038/s41586-024-07017-8 https://doi.org:10.1038/s41586-020-2047-9 https://doi.org:10.1002/med.1023 https://doi.org:10.1038/ncomms10166 https://doi.org:10.1016/j.cmet.2005.09.001 https://doi.org:10.1194/jlr.M700330-JLR200 https://doi.org:10.1159/000450908 https://doi.org:10.1172/JCI23076 https://doi.org:10.1016/j.cmet.2023.07.011 https://doi.org:10.1038/s41586-019-1865-0 https://doi.org:10.1038/s41586-020-2193-0 https://doi.org:10.1053/jhep.2001.23790 https://doi.org:ARTN 30 57 Ali, A. H., Carey, E. J. & Lindor, K. D. Recent advances in the development of farnesoid X receptor 963 agonists. Ann Transl Med 3, 5 (2015). https://doi.org:10.3978/j.issn.2305-5839.2014.12.06 964 58 Ferrell, J. M. & Chiang, J. Y. L. Understanding Bile Acid Signaling in Diabetes: From 965 Pathophysiology to Therapeutic Targets. Diabetes Metab J 43, 257-272 (2019). 966 https://doi.org:10.4093/dmj.2019.0043 967 59 Li, T. & Apte, U. Bile Acid Metabolism and Signaling in Cholestasis, Inflammation, and Cancer. 968 Adv Pharmacol 74, 263-302 (2015). https://doi.org:10.1016/bs.apha.2015.04.003 969 970 https://doi.org:10.3978/j.issn.2305-5839.2014.12.06 https://doi.org:10.4093/dmj.2019.0043 https://doi.org:10.1016/bs.apha.2015.04.003 31 Method references 971 972 60 Pitari, G. et al. Pantetheinase activity of membrane-bound Vanin-1: lack of free cysteamine in 973 tissues of Vanin-1 deficient mice. FEBS Lett 483, 149-154 (2000). https://doi.org:10.1016/s0014-974 5793(00)02110-4 975 61 Hepworth, M. R. et al. Immune tolerance. Group 3 innate lymphoid cells mediate intestinal 976 selection of commensal bacteria-specific CD4(+) T cells. Science 348, 1031-1035 (2015). 977 https://doi.org:10.1126/science.aaa4812 978 62 Letourneau, J. et al. Ecological memory of prior nutrient exposure in the human gut microbiome. 979 ISME J 16, 2479-2490 (2022). https://doi.org:10.1038/s41396-022-01292-x 980 63 Goc, J. et al. Dysregulation of ILC3s unleashes progression and immunotherapy resistance in 981 colon cancer. Cell 184, 5015-5030 e5016 (2021). https://doi.org:10.1016/j.cell.2021.07.029 982 64 Tautenhahn, R., Bottcher, C. & Neumann, S. Highly sensitive feature detection for high 983 resolution LC/MS. BMC bioinformatics 9, 504 (2008). https://doi.org:10.1186/1471-2105-9-504 984 65 Wang, M. et al. Sharing and community curation of mass spectrometry data with Global Natural 985 Products Social Molecular Networking. Nature biotechnology 34, 828-837 (2016). 986 https://doi.org:10.1038/nbt.3597 987 66 Shannon, P. et al. Cytoscape: a software environment for integrated models of biomolecular 988 interaction networks. Genome Res 13, 2498-2504 (2003). https://doi.org:10.1101/gr.1239303 989 67 Zhang, B. et al. Acylspermidines are conserved mitochondrial sirtuin-dependent metabolites. 990 Nat Chem Biol 20, 812-822 (2024). https://doi.org:10.1038/s41589-023-01511-2 991 992 https://doi.org:10.1016/s0014-5793(00)02110-4 https://doi.org:10.1016/s0014-5793(00)02110-4 https://doi.org:10.1126/science.aaa4812 https://doi.org:10.1038/s41396-022-01292-x https://doi.org:10.1016/j.cell.2021.07.029 https://doi.org:10.1186/1471-2105-9-504 https://doi.org:10.1038/nbt.3597 https://doi.org:10.1101/gr.1239303 https://doi.org:10.1038/s41589-023-01511-2 32 FIGURES AND FIGURE LEGENDS Figure 1. Identification of MCY conjugates of bile acids. a, Primary role of gut microbiota in bile acids (BAs) metabolism and schematic overview of analytical strategy for comparison of GF and SPF mice. b, Volcano plots of differential metabolites detected in serum of GF (n = 12) and SPF (n = 9) mice. Blue and red dots represent metabolites five- or more-fold downregulated or upregulated in GF relative to SPF mice at P < 0.05, as calculated by unpaired two-sided t-test. c, Partial representation of MS2 network (cosine>0.7) for mouse serum in ESI+ and ESI- showing clusters representing free BAs, BA-taurine conjugates, and previously unannotated BA-MCY conjugates. Shown red and blue nodes are downregulated and upregulated, respectively, in serum of GF compared to SPF mice. See Supplementary Figs. 1-3 for full MS2 networks, including m/z values for all nodes. d, MS2 spectrum of CA-MCY. Red MS2 fragments represent the MCY group; green MS2 fragments are derived from water loss. e, Structures of BA- MCY conjugates identified in mouse serum. 33 Figure 2. Microbiota dependence and biosynthesis of BA-MCY conjugates. a,b, Relative abundances of CA-MCY conjugates (a) and MCA-MCY conjugates (b) as well as corresponding free BAs in serum of SPF mice (n = 11), GF mice (n = 12), and mice that received FMT (n = 10, the three donors are represented by triangles, circles, and crosses, n = 3-4 for each donor). Data are mean ± s.e.m. P values were calculated by unpaired two-sided t-test with Welch’s correction. c, Relationship between abundances of free BAs and BA-MCY conjugates in liver of SPF mice (n = 11), SPF control mice for the inulin fiber diet study (n = 4), and inulin fiber diet fed SPF mice (n = 5). d, Relative abundance of CA-MCY conjugates or CDCA-MCY conjugates in human serum (n = 19). Data are mean ± s.e.m. e, Pathways considered for the origin of BA-MCY conjugates. MCY conjugates could originate either from reduction of BA- taurine conjugates produced by the liver enzyme BAAT or conjugation of BAs with cysteamine or another cysteamine derivative from coenzyme A and pantetheine degradation. Oxidation of MCY conjugates produces the corresponding MCYO and MCYO2 conjugates; BSH: bile salt hydrolase. 34 Figure 3. Host-dependent production of BA-MCYs and microbial deconjugation. a, Total amounts of BA-MCY conjugates and corresponding free BAs in liver, small intestine, and cecum of SPF (n = 11) and GF (n = 12 for liver and n = 13 for small intestine and cecum) mice. Data are mean ± s.e.m. P values were calculated by unpaired two-sided t-test with Welch’s correction. b, Established function of VNN1 in pantetheine hydrolysis (box) and proposed role of VNN1 in CA-pantetheine hydrolysis, followed by re- arrangement and methylation to form CA-MCY. c, Production of pantothenic acid from a range of concentrations of CA-pant and pantetheine incubated with recombinant VNN1 in vitro. Reactions with both substrates follow saturation kinetics. Enzyme concentration was 0.01 µM. The reactions were incubated at 37 °C for 15 min. Number of independent assays using the same batch of enzyme (n = 3). Data are mean ± s.d. d, Relative abundances of BA-MCY conjugates in small intestine, liver, serum and feces of WT (n = 5) or Vnn1-/- (n = 5) mice. Data are mean ± s.e.m. P values were calculated by unpaired two-sided t-test with Welch’s correction. e, Deconjugation of supplemented CDCA-d5-MCY in feces of SPF (n = 8), GF (n = 3), and ABX (n = 15) mice. Data are mean ± s.e.m. P values were calculated by unpaired two-sided t-test with Welch’s correction. f, Deconjugation of MCY or taurine conjugates of BA in feces of GF monocolonized with WT (n = 3) or BSH-deleted B. ovatus (n = 3) (WT Bo or Δbsh Bo, respectively). Data are mean ± s.e.m. P values were calculated by unpaired two-sided t-test with Welch’s correction. g, Roles of host and microbiota in the biosynthesis and metabolism of BA taurine and MCY conjugates. 35 Figure 4. BA-MCY conjugates are FXR antagonists. a-c, BA-MCYs act as FXR antagonists in vitro. Compounds were tested against a cell-based protein-protein interaction assays in both agonist and antagonist modes. CDCA-MCY (a) and CA-MCY (b) showed strong FXR antagonistic effects to GW4604-mediated activation of FXR, whereas CA-MCYO (c) showed no FXR antagonistic effects. None of the tested BA-MCY conjugates showed FXR agonistic effects in this assay (also see Extended Data Fig. 10). Assays were performed in duplicate for each concentration. 36 Figure 5. BA-MCYs regulate BA biosynthesis in vivo and a model for the role of BA-MCYs in BA metabolism. a, BA biosynthesis scheme, highlighting separate CA and CDCA pathways. Compounds highlighted in blue and green belong to the CA and CDCA pathways, respectively. b, Gene expression ratio of Shp, Cyp8b1 and Cyp7a1 to Hprt1 control in liver of mice administered CDCA-MCY (n = 4) or control (corn oil, n = 3). Bar graphs in b-h represent mean ± s.e.m., and P values were calculated by unpaired two-sided t-test with Welch’s correction. c, Serum FGF15 levels of mice administered CDCA-d5-MCY or control (n = 12). d, Gene expression ratio of Shp to Hprt1 control in small intestine of mice administered CDCA-d5-MCY or control (n = 7). e, Gene expression ratio of Slc10a2 to Hprt1 control in small intestine of mice administered CDCA-d5-MCY or control (n = 3). f,g, Abundances of endogenously produced BAs in feces of mice administered CDCA-MCY or CDCA-d5-MCY for 14 days. Shown are total amounts of CDCA-derived BAs (f, n = 7 for control and n = 3 for CDCA-d5-MCY fed mice) and CA-derived BAs (g, n = 7 for control and n = 7 for CDCA-MCY (n = 4) or CDCA- d5-MCY (n = 3) fed mice). h, Abundances of endogenously produced BAs in feces of WT or Nr1h4−/− mice administered CDCA-d5-MCY daily for 14 days. Shown are total amounts of BAs (n = 8 for control and n = 8 for CDCA-MCY fed mice). i, Representative photomicrographs of oil red O staining of liver sections of mice treated with the indicated conditions. Mice were fed control (n = 4 for vehicle and n = 4 for CDCA-MCY) or high cholesterol diet (n = 4 for vehicle and n = 4 for CDCA-MCY). CDCA-MCY was delivered by oral gavage at a rate of 50 mg/kg body weight daily for two weeks. Scale bar, 50 m. j, Average measured oil red O area in i. Data are mean ± s.e.m. P values were calculated by one-way ANOVA with Tukey’s correction. k, Proposed model for FXR-dependent regulation of BA metabolism by BA-MCYs. 37 Extended Data Figures Extended Data Figure 1. MS2 spectra of MCY conjugates of BAs. a-e, MS2 spectra of CA-MCY (a), CDCA-MCYO (b), CA-MCYO (c), 7KDCA-MCYO (d), and CA-MCYO2 (e). The listed BA-MCY conjugates in each panel produced MS2 spectra very similar to the shown examples. Blue arrows indicate inferred fragmentation. MS2 fragments and structure parts highlighted in red represent MCY groups. Green fragments are derived from water loss. 38 Extended Data Figure 2. Microbiota dependent production of BA-taurine and BA-MCY conjugates. a, Relative abundances of BA-MCYO conjugates of less abundant BAs in serum of SPF (n = 11) and GF (n = 12), and mice that received FMT (n = 10, the three donors are represented by triangles, circles, and crosses, n = 3-4 for each donor). Data are mean ± s.e.m. P values were calculated by unpaired two-sided t-test with Welch’s correction. b,c, Relative abundances of CA-MCY conjugates (b) and MCA-MCY conjugates (c) as well as the corresponding free BAs in feces of SPF mice (n = 3), GF mice (n = 3), and mice that received FMT (n = 10, the three donors are represented by triangles, circles, and crosses, n = 3-4 for each donor). Data are mean ± s.e.m. P values were calculated by unpaired two-sided t-test with Welch’s correction. N.D., not detected. d, Relative abundances of BA-MCYO conjugates of less abundant BAs in feces of SPF (n = 3) and GF (n = 3), and mice that received FMT (n = 10, the three donors are represented by triangles, circles, and crosses, n = 3-4 for each donor). Data are mean ± s.e.m. P values were calculated by unpaired two-sided t- test with Welch’s correction. e, Relative abundances of BA-taurine conjugates in serum or feces of SPF (n = 11 for serum and n = 3 for feces) and GF (n = 12 for serum and n = 3 for feces), and mice that received FMT (n = 10, the three donors are represented by triangles, circles, and crosses, n = 3-4 for each donor). Data are mean ± s.e.m. P values were calculated by unpaired two-sided t-test with Welch’s correction. f,g, Relative abundances of free BAs in serum (f) or feces (g) of SPF (n = 11 for serum and n = 3 for feces) and GF (n = 12 for serum and n = 3 for feces), and mice that received FMT (n = 10, the three donors are represented by triangles, circles, and crosses, n = 3-4 for each donor). Data are mean ± s.e.m. P values were calculated by unpaired two-sided t-test with Welch’s correction. 39 Extended Data Figure 3. Relationship between abundances of free BAs and BA-MCY conjugates. a,b, Relative abundances of CA-MCY (a) and MCA-MCY conjugates (b) as well as corresponding free BAs in serum of mice fed control (n = 8) or inulin fiber diet (n = 7). Data are mean ± s.e.m. P values were calculated by unpaired two-sided t-test with Welch’s correction. c, Relative abundances of BA-MCYO conjugates as well as corresponding free BAs in serum of mice fed control (n = 8) or inulin fiber diet (n = 7). Data are mean ± s.e.m. P values were calculated by unpaired two-sided t-test with Welch’s correction. d, Relationship between abundances of free BAs and BA-MCY conjugates in feces of SPF mice used as control for the FMT study (n = 25), FMT mice (n = 10), SPF mice fed a control diet for the inulin fiber diet study (n = 6), and inulin fiber diet fed SPF mice (n = 8). e, Relationship between abundances of free BAs and BA-MCY conjugates in serum of SPF as control for the FMT study (n = 11), FMT mice (n = 10), SPF mice fed a control diet for the inulin fiber diet study (n = 8), and inulin fiber diet fed SPF mice (n = 7). f,g, Relative abundances of free BAs (f) and corresponding BA-MCY conjugates (g) in serum of human (n = 19). Data are mean ± s.e.m. 40 Extended Data Figure 4. Analysis of stable-isotope feeding experiments. a, Administration of taurine-d4 in SPF mice resulted in deuterium incorporation in all detected taurine conjugates, but not in any MCY conjugates. Shown are EICs for the m/z of molecular ions of unlabeled (black) and deuterium-labeled versions (red) of the different conjugates in serum of mouse fed taurine-d4. b, Administration of deuterium-labeled L-cysteine (L-cys-d2) in SPF mice resulted in deuterium incorporation in the MCY conjugates of BAs. Shown are EICs for the m/z of the molecular ions of the unlabeled (black) and the deuterium-labeled versions of BA-MCY conjugates (red) detected in serum of mice fed L-cys-d2. c,d, Administration of deuterium-labeled L-cysteine (L-cys-d2) in SPF mice (see Fig. 2i) resulted in deuterium incorporation in taurine conjugates of BAs (c) and pantetheine (d). EICs for molecular ion peaks (black) and deuterium isotope peaks (red) of taurine conjugates of BAs (c) and pantetheine (d) in serum of mouse fed L- cys-d2. 41 Extended Data Figure 5. Analysis of Baat−/− mice. a-d, Extracted ion chromatograms (EICs) of BA-MCY and BA-MCYO conjugates (a,b,c) and BA-MCYO2 conjugates (d) in liver of Baat−/− mice and comparison with synthetic standards analyzed in ESI+. e,f, Relative abundances of BA-MCY conjugates (e) and corresponding free BAs (f) in liver of WT (n = 4) or Baat−/− (n = 5) mice. Data are mean ± s.e.m. P values were calculated by unpaired two-sided t-test with Welch’s correction. N.D., not detected. 42 Extended Data Figure 6. Abundances of BA-MCY conjugates in different tissues. a,b, Abundances of CA-MCY conjugates, TCA and CA (a), and MCA-MCY conjugates, TMCA and MCA (b) in liver, small intestine, and cecum of SPF (n = 11) and GF (n = 12 for liver and n = 13 for small intestine and cecum) mice. Data are mean ± s.e.m. P values were calculated by unpaired two-sided t-test with Welch’s correction. c,d,e,f, Abundances of UDCA-MCY conjugates and UDCA (c), CDCA-MCY conjugates and CDCA (d), DCA-MCY conjugates and DCA (e), and 7-KDCA-MCY conjugates and 7-KDCA (f) in liver, small intestine, and cecum of SPF (n = 11) and GF (n = 12 for liver and n = 13 for small intestine and cecum) mice. Data are mean ± s.e.m. P values were calculated by unpaired two-sided t-test with Welch’s correction. N.D., not detected. 43 Extended Data Figure 7. The role of VNN1 in production of BA-MCY conjugates. a, Steady-state kinetic analysis of CA-pant and pantetheine hydrolysis catalyzed by recombinant human VNN1 (ΔN490aa truncated) revealed both reactions follow saturation kinetics. The steady-state kinetic parameters Km and Vmax are determined by HPLC-HRMS for pantothenic acid formation to be 39.78 ± 20.31 μM and 1.53 ± 0.20 min−1 for CA-pant, and 74.07 ± 41.52 μM and 2.13 ± 0.37 min−1 for pantetheine. The reaction mixtures contain 0.01 μM VNN1. Number of independent assays using the same batch of enzyme (n = 3). Data are mean ± s.d. b, EICs of CA-CY in ileum of Baat−/− mice, extracts of in vitro reaction of VNN1 hydrolyses CA- pantetheine, and comparison with a synthetic standard analyzed in ESI+. c, EICs of CA-pant in small intestine of Vnn1−/− mice and comparison with a synthetic standard analyzed in ESI+. d, Relative abundances of CA- pant in small intestine, feces, liver, and serum of WT (n = 5) and Vnn1−/− (n = 5) mice. Data are mean ± s.e.m. P values were calculated by unpaired two-sided t-test with Welch’s correction. N.D., not detected. 44 Extended Data Figure 8. Microbial deconjugation of BA-MCYs in SPF, GF, and ABX mice. a, Ratio of total BA-taurine or BA-MCY conjugates to corresponding free BAs in feces of SPF (n = 14) and GF (n = 16) mice. Data are mean ± s.e.m. P values were calculated by unpaired two-sided t-test with Welch’s correction. b,c, Total amounts of free BAs (b) or BA-MCY conjugates (c) in feces of SPF (n = 14) and GF (n = 16) mice. Data are mean ± s.e.m. P values were calculated by unpaired two-sided t-test with Welch’s correction. d, Total amounts of free BAs and BA-MCY conjugates in feces of GF (n =16) mice. Data are mean ± s.e.m. P values were calculated by paired two-sided t-test. e, HRMS analysis of feces of mice fed CDCA-d5- MCY revealed deconjugation of supplemented CDCA-d5-MCY, represented by peaks in the CDCA mass spectrum highlighted in red. Endogenously produced CDCA can be distinguished, highlighted in green. CA remained unlabeled. f,g,h, Total amounts of labeled free BAs (f), BA-MCY conjugates (g), and BA-taurine conjugates (h) in feces of SPF (n = 14), ABX (n = 15), and GF (n = 3) mice administered CDCA-d5-MCY. i, Volcano plot of differential metabolites detected in liver of SPF mice administered control (corn oil) (n = 4) or CDCA-d5-MCY (n = 5). Bubble sizes reflect peak areas. See Supplementary Table 4 for compounds derived from supplemented CDCA-d5-MCY. P values were calculated by unpaired two-sided t-test. j,k,l, Total amounts of labeled free BAs (j), BA-MCY conjugates (k), and BA-taurine conjugates (l) in liver of SPF (n = 5) and ABX (n = 5) mice administered CDCA-d5-MCY. m, Volcano plot of differential metabolites detected in liver of ABX mice administered control (corn oil) (n = 4) or CDCA-d5-MCY (n = 5). Bubble sizes reflect peak areas. See Supplementary Table 4 for compounds derived from supplemented CDCA-d5-MCY. P values were calculated by unpaired two-sided t-test. 45 Extended Data Figure 9. Microbial deconjugation of BA-MCYs in vitro and gnotobiotic mice. a,b, Deconjugation of CA-MCY conjugates in fecal suspensions obtained from SPF mice (a) (n = 3) and cultured gut bacteria (b) (n = 3). c, Relative abundances of CDCA-d5-MCY conjugates and corresponding free BA in feces of GF monocolonized with WT (n = 3) or BSH-deficient B. ovatus (n = 3) (WT Bo and Δbsh Bo, respectively). Data are mean ± s.e.m. P values were calculated by unpaired two-sided t-test with Welch’s correction. 46 Extended Data Figure 10. FXR-related activity of known ligands and BA-MCYs. a, βMCA-MCY was tested against a cell-based protein-protein interaction assays in both agonist and antagonist modes. βMCA-MCY showed strong FXR antagonistic effects to GW4604-mediated activation of FXR. βMCA-MCY showed no FXR agonistic effects in the assay. Assays were performed in duplicate for each concentration. b, FXR agonistic effect of CDCA as measured in the protein-protein interaction assays. Data were normalized to the maximal and minimal response observed in the presence of control compound (GW4064) and vehicle (DMSO), respectively. Assays were performed in duplicate for each concentration. c,d, CDCA-MCY showed FXR antagonistic effects to obeticholic acid (25 M) (c) or CDCA (25 M) (d) mediated activation of FXR. Data were normalized to the maximal and minimal response observed in the presence of control compound (DY268) and vehicle (DMSO), respectively. Assays were performed in duplicate for each concentration. e,f,g, Cytotoxicity assays for CDCA-MCY (e), CDCA-MCYO (f), and CDCA (g) in a cell-based assay on human primary hepatocytes. Assays were performed in duplicate for each concentration. h, TβMCA did not show FXR antagonistic effects in protein-protein interaction assays at the tested concentrations. DY268 a synthetic FXR antagonist was used as a positive control. Data were normalized to the maximal and minimal response observed in the presence of control compound (DY268) and vehicle (DMSO), respectively. Assays were performed in duplicate for each concentration. 47 Extended Data Figure 11. Regulation of BA biosynthesis by BA-MCYs in vivo. a,b, Abundances of endogenously produced BAs in feces of mice administered CDCA-MCY or CDCA-d5-MCY daily for 14 days. Shown are individual amounts of CDCA-derived BAs (a) and CA-derived BAs (b) in feces. Data are mean ± s.e.m. with control (corn oil) (n = 7) and CDCA-MCY fed mice (n = 7 for CDCA-derived pathway and n = 3 for CA-derived pathway). P values were calculated by unpaired two-sided t-test with Welch’s correction. c, Total endogenously produced BAs in feces of ABX mice administered CDCA-d5-MCY. Shown are total amounts of BAs (n = 13 for control and n = 14 for CDCA-MCY fed mice). Data are mean ± s.e.m. P values were calculated by unpaired two-sided t-test with Welch’s correction. d,e, Abundances of CDCA-derived BAs (d) and CA-derived BAs (e) in liver of mice administered CDCA-MCY or CDCA-d5-MCY daily for 14 days. Data are mean ± s.e.m. with control (corn oil) (n = 6) and CDCA-MCY fed mice (n = 3 for CDCA-derived pathway and n = 7 for CA-derived pathway). P values were calculated by unpaired two-sided t- test with Welch’s correction. f,g, Abundances of CDCA-derived BAs (f) and CA-derived BAs (g) in serum of mice administered CDCA-MCY or CDCA-d5-MCY daily for 14 days. Data are mean ± s.e.m. with control (corn oil) (n = 6) and CDCA-MCY fed mice (n = 3 for CDCA-derived pathway and n = 7 for CA-derived pathway). P values were calculated by unpaired two-sided t-test with Welch’s correction. 48 Extended Data Figure 12. FXR-related activity of BA-MCYs in vivo. a,b, Abundances of total BAs in liver (a) and serum (b) of WT and Nr1h4−/− mice administered CDCA-d5-MCY daily for 14 days. Data are mean ± s.e.m. with control (corn oil) (n = 4) and CDCA-d5-MCY fed mice (n = 4). P values were calculated by unpaired two-sided t-test with Welch’s correction. c, Representative photomicrographs of oil red O staining of liver sections of mice treated with the indicated conditions. Mice were fed control (n = 4 for vehicle and n = 4 for CDCA-MCY) or high cholesterol diet (HCD) (n = 4 for vehicle and n = 4 for CDCA-MCY). CDCA-MCY was delivered by oral gavage at a rate of 5 mg/kg body weight per day for two weeks. Scale bar, 100 m. d, Average measured oil red O area of liver sections of mice in c. Data are mean ± s.e.m. P values were calculated by one-way ANOVA with Tukey’s correction.