Subclinical cholestasis is a hallmark of gut dysbiosis causing resistance to cancer immunotherapy Authors: Anne-Laure Mallard de La Varende, Ai-Ling Tian, Simon Thomas, Imran Lahmar, Meriem Messaoudene, Sijing Li, Omar Motiño, Hortense Guillaume-dit-Taunière, Valerio Iebba, Yoan Hurtado, Thao-Nguyen Pham, Cassandra Thélémaque, Miguel Araujo-Voces, Deborah Suissa, Pierre Ly, Ella Reich, Giacomo Vitali, Bryan Thierry Arlunno, Sylvere Durand, Fanny Aprahamian, Marion Leduc, Sabrina Forveille, Oliver Kepp, Carlos de la Calle-Fabregat, Angela Schippers, Norbert Wagner, Pierre-Edouard Fournier, Kenya Honda, Federica Marmorino, Chiara Cremolini, Saman Maleki Vareki, John G Lenehan, Aurélien Marabelle, François-Xavier Danlos, Margaux Deligne, Caroline Truntzer, François Ghiringhelli, Anne H Ree, Paula A Bousquet, Sebastian Meltzer Published: 2026 ABSTRACT SummaryGut dysbiosis compromises cancer immunosurveillance by downregulating ileal mucosal addressin cell adhesion molecule 1 (MAdCAM-1), but the metabolic landscape associated with gut dysbiosis remains elusive. Here, we show that antibiotics (ABX) or ABX-associated Enterocloster species lead to the loss of secondary bile acids (BAs) including deoxycholic acid (DCA) and the accumulation of tauro-conjugated primary BAs (tauro-chenodeoxycholic acid [TCDCA] and tauro-β-muricholic acid [T-βMCA]) from the alternative pathway in the plasma of patients and mice. Fecal microbial transplantation (FMT), the ileum-specific farnesoid X receptor (FXR) agonist fexaramine, or glycodeoxycholic acid (GDCA) compensated dysbiosis-associated BA abnormalities and circumvent primary resistance to PD-1 blockade. GDCA curtailed ABX-induced MAdCAM-1 downregulation and T cell exhaustion in tumors. Subclinical cholestasis defined by elevation of γ-glutamyl transferase (γGT) correlated with increased TCDCA and decreased soluble MAdCAM-1 in plasma and predicted poor survival in multivariate analyses in six cohorts of patients who received immunotherapy. Hence, subclinical cholestasis accompanies gut dysbiosis, paving the way to immunoresistance.Keywords: cancer, PD-1, immune checkpoint blockade, bile acids, mucosal addressin cell adhesion molecule 1, MAdCAM-1, antibiotics, gut microbiota, microbiome, immunotherapy Introduction Among metabolic pathways, the lipid metabolism has emerged as a key determinant of T cell responses, with extracellular lipid uptake contributing to their activation status, differentiation, and adaptation to environmental stress.1,2,3 Bile acids (BAs) are abundant cholesterol-derived metabolites essential for lipid absorption. Primary BAs such as cholic acid (CA) and chenodeoxycholic acid (CDCA) are synthesized in the liver from cholesterol via the classical pathway under the action of hepatic enzymes (CYP7A1 and CYP8B1) and conjugated with taurine, glycine or other amino acids and then excreted into the small intestine.1 Under physiological conditions, the liver-specific classical pathway is dominant, while the alternative (or acidic) pathway dependent on the enzymes CYP27A1 and CYP7B1 is marginal and not liver specific.1 BA composition is determined by not only the activity of hepatic microsomal enzymes but also that of microbial enzymes in the small and large intestines.2 Bacterial bile salt hydrolases (BSHs) mediate the deconjugation of BAs, whereas bacterial hydroxysteroid dehydrogenases (HSDHs) mediate oxidation and epimerization of the 3-, 7-, and 12-hydroxy groups of BA. Liver-derived BAs are further modified by the gut microbiota to generate intestinal secondary BAs, such as CA-derived deoxycholic acid (DCA), CDCA-derived lithocholic acid (LCA), and ursodeoxycholic acid (UDCA).1 In mice, but not humans, an additional C6 hydroxylation of CDCA (by murine Cyp2c70) gives rise to muricholic acids (MCAs) and gut microbiota-modified derivatives.1 In germ-free mice, the absence of microbiota leads to an increase in the BA pool and farnesoid X receptor (FXR) antagonists, including tauro-βMCA (T-βMCA).1 These hepatic and intestinal BAs along with their conjugated or hydroxylated derivatives profoundly influence T cell functions. Intestinal BAs regulate Th17 and peripheral Foxp3+ regulatory T cell differentiation directly or indirectly through dendritic cells.3,4,5,6,7 Liver cancers exhibit elevated levels of BAs in their parenchyma, presumably as a result of deregulated BA synthesis or defects in BA export.8 A recent report unveiled the immunosuppressive effect of primary conjugated and secondary BAs accumulating in hepatocellular carcinomas on T cell responses and cancer immunosurveillance during PD-1 blockade,9 echoing a pioneering study showing the inhibition of hepatic natural killer T (NKT) cell chemotaxis by secondary BAs during liver carcinogenesis.10 Beyond their local effects on the gut-liver axis, BAs may exert systemic immunomodulation.11,12,13 Given the critical role of intestinal commensalism in the metabolic transformation of hepatic BAs, dysbiosis, which is a prolonged deviation of the taxonomic composition of the gut microbiota,14 may affect both local and systemic adaptive T cell functions.15,16,17 Carcinogenesis18 and cancer treatments, co-morbidities, and comedications may lead to dysbiosis.19,20 Typically, cessation of antibiotics (ABX) induces the relative dominance of the genus Enterocloster, associated with immunosuppression and resistance to PD-1 blockade.21 This immunosuppressive effect of Enterocloster species involves the downregulation of mucosal vascular addressin cell adhesion molecule 1 (MAdCAM-1), expressed on high endothelial venules of the ileal lamina propria and gut lymphoid organs. Specifically, Enterocloster clostridioformis reshuffles the composition of BAs.21 However, deviations in BA metabolism and their links with gut dysbiosis and immune dysfunction in cancer patients during immunotherapy remain poorly characterized and were the focus of the present study. Results Gut dysbiosis-associated shifts in BA metabolism ABX alter the BA composition, but their effects in mice bearing extraintestinal tumors have not been comprehensively studied.22 Mass spectrometry-based targeted metabolomics was performed on ileal and colonic contents, liver tissue, and plasma from C57BL/6 mice inoculated with subcutaneous MCA205 fibrosarcoma. Mice received a broad-spectrum ABX cocktail (ampicillin, colistin, and streptomycin) orally for two weeks prior to analysis (Figures 1A and S1A–S1C) in two geographically distinct animal facilities (France and Canada) but harbored similar plasma BA profiles at baseline (Figures S1A and S1B). ABX caused significant perturbations in BA metabolism (Figures 1B and S1D; Table S1). At both sites, ABX increased the plasma BA pool (Figures S1E and S1F) and drastically decreased secondary BA plasma concentrations, including DCA (Figure 1C, upper; Figure S1G, left), UDCA, ω-MCA, hyo-DCA (HDCA), as well as their conjugated forms (tauro-DCA [TDCA]) (Figure 1C, lower; Figures S1D and S1G, right), 7-ketoDCA, and nor-UDCA (Figures 1B and S1D). In plasma, the levels of primary unconjugated BAs (CA and CDCA) dropped (Figures 1B and S1D), while the concentrations of conjugated primary BAs (tauro-α-β-ω-muricholic acid [T-α-β-ω-MCA], tauro-cholic acid [TCA], and tauro-chenodeoxycholic acid [TCDCA]) rose with ABX, together with the tauro-conjugated UDCA (TUDCA) for about 2 weeks (Figures 1D, S1D, and S1H). In the liver, ABX caused a significant reduction in α-MCA, ω-MCA, TDCA, and tauro-lithocholic acid (TLCA), as well as a rise in β-MCA, its tauro-conjugated derivative (T-α-β-ω-MCA), and TUDCA (Figure 1B). Similarly, in the colon and to a lesser extent in the ileum, ABX depleted most BAs, such as DCA, glyco-DCA (GDCA), UDCA, and primary BA, but increased tauro-conjugated primary BAs, T-α-β-ω-MCA, TCA, TCDCA, and TUDCA (Figure 1B). To establish a causal link between gut dysbiosis and BA perturbations, we subjected ABX-pretreated mice to oral gavage with fecal microbial transplantation (FMT) from four cancer patients responding to immunotherapy in the ONCOBIOTICS cohort to revert dysbiosis (Figures S1I–S1K). Each of the four FMTs prevented the loss of DCA, HDCA, 7-ketoDCA, nor-UDCA (Figures S1I and S1J) and unconjugated primary BAs (CDCA, CA) after ABX treatment, while blunting the recirculation of tauro-conjugated primary BAs, T-α-β-ω-MCA, TCDCA, and TCA (Figures S1I and S1K). Gut dysbiosis perturbs BA composition (A) Experimental settings at two different centers; D, day. (B) Supervised hierarchical clustering of BAs significantly different between ABX and water-fed tumor bearers assessed by metabolomics. (C and D) Kinetics of the percentages of distinct circulating BAs among the total BA. (E) Id. in the plasma of naive mice at GRCC and orally gavaged with E. clostridioformis or water controls. (F and G) Metabolite levels in tumor beds after 15 days of ABX. Volcano plot comparing ABX vs. drinking water (F), red dots indicate metabolites significantly increased after ABX. Dot plots depicting the percentages of distinct BA (G). Error bars indicate mean ± SEM (G). One representative experiment containing 8 mice/group (B, C, D, F, and G) and 6 and 7 mice/group (E) is depicted. Comparisons between groups were performed using non-parametric Mann-Whitney U test (2 groups), followed by multiple-comparisons test controlling the false discovery rate (FDR) (B, F, and G), or non-parametric Friedman test (>2 paired groups for longitudinal metabolomics analysis), followed by two stage set-up method of Benjamini, Krieger, and Yekutieli (C, D, and E). See also Figures S1–S3 and Table S1. Next, we replaced ABX by iterative gavages with the ABX-associated Enterocloster clostridioformis strain 141.21 In such conditions, we generated a “gut dysbiosis” as revealed by 16S rRNA targeted sequencing, unveiling a profound shift in the stool taxonomic composition, characterized by the enrichment of strains from the Enterocloster genus (Figures S2A and S2B), increased alpha diversity (Figure S2C), and significant alterations in the dynamics of beta diversity (Figure S2D), along with a relative overrepresentation of Eggerthellaceae and Clostridiaceae family members, coinciding with the depletion of immunogenic Alistipes spp. (Figure S2E) over 4 days after gavaging with E. clostridioformis. In these conditions, E. clostridioformis tended to mimic some of the effects of broad-spectrum ABX on BA composition, inducing the accumulation of β-MCA and its derivates, such as HDCA (Figures S3A and S3B), as well as T-α-β-ω-MCA and TUDCA in the plasma of both naive (Figure 1E) and tumor-bearing mice treated with anti-PD1 antibodies (Abs) (Figures S3A, S3C, and S3D). Importantly, such tauro-conjugated primary BAs reached the tumor microenvironment (TME), as indicated by their increased relative abundance among all BAs (Figures 1F and 1G). We next analyzed the transcriptional profile of BA rate-limiting enzymes in the liver by quantitative reverse-transcription PCR (RT-qPCR) in animals receiving ABX, followed by 4 days of recolonization with or without two gavages of E. clostridioformis in naive mice. The expression of the enzymes of the classical pathway, Cyp7a1 and Cyp8b1, was drastically decreased by ABX and did not recover after ABX cessation (Figures S3E and S3F). However, ABX +/− E. clostridioformis tended to increase the expression of enzymes of the alternative pathway, Cyp27a1 and Cyp7b1 (Figure S3G), in line with the increase in CDCA derivatives, such as TCDCA, β-MCA, and T-α-β-ω-MCA after oral gavage with E. clostridioformis or ABX. Of note, the mouse-specific enzyme converting CDCA into α-MCA (Cyp2c70) was upregulated in the liver post oral gavage with ABX and E. clostridioformis (Figure S3H). The activity of bile acid-CoA:amino acid N-acyltransferase (Baat) enzyme involved in the glyco- or tauro-conjugation tended to increase following ABX and E. clostridioformis in the liver and the ileum, albeit at low levels (Figure S3I). We confirmed the downregulation of the enzymes of the BA classical pathway and the upregulation of Cyp2c70 using bulk RNA sequencing of cell-sorted hepatocytes as well as the upregulation of Baat gene product in ileal enterocytes (albeit at very low expression levels) (Figures S3J–S3M). Gut dysbiosis increased Vdr transcription in ileal enterocytes but not that of other BA receptors in hepatocytes, cholangiocytes, or enterocytes (Figures S3N–S3P). Of note, the sodium/BA co-transporter family member Slc10a1, which is involved in the uptake of BA for the enterohepatic cycle, was not significantly affected by ABX (Figures S3Q–S3S). Altogether, ABX caused a metabolic shift in the BA composition, typically featured by an abnormal accumulation of primary BAs derived from CDCA, such as the tauro-conjugated primary BAs (TCDCA), and β-MCA and its tauro-conjugated derivative (T-α-β-ω-MCA), as well as the loss of intestinal and plasma secondary BAs derived from CA, i.e., DCA and G/TDCA. These ABX-induced metabolic shifts in the BA composition were prevented by FMT and partially mimicked by intestinal colonization with E. clostridioformis. GDCA circumvents resistance to PD-1 blockade during dysbiosis To draw causal links between gut dysbiosis-associated BA abnormalities and resistance to immunotherapy, we attempted to compensate dysbiotic tumor-bearing mice with oral delivery of secondary BAs during PD-1 blockade. C57BL/6 mice were orally gavaged with E. clostridioformis, then inoculated with MCA205 sarcoma cells or RET melanoma cells, and treated five days later intraperitoneally (i.p.) with α-PD-1 monoclonal antibody (mAb) alone or together with anti-CTLA4 Abs. At immunotherapy start, mice were fed with selected secondary BAs (GDCA, LCA, and UDCA) at a physiological concentration four times every other day (Figure 2A). Tumor growth retardation induced by immunotherapy was markedly inhibited by E. clostridioformis (Figure 2B left, p = 0.023) and restored by GDCA (Figure 2B right, p = 0.021) but neither by UDCA nor by LCA (Figure 2C). Similarly, the Ret transgenic melanoma model responded to the combination of α-CTLA-4 + α-PD-1 mAbs, but the gut colonization with E. clostridioformis tended to compromise the efficacy of immunotherapy (Figure 2D, p = 0.06), unless GDCA or UDCA (but not LCA) was co-administered (Figure 2D, p = 0.002; Figure 2E left, p = 0.009; right, p = 0.044). Even in the context of full-fledged gut dysbiosis provoked by ABX followed by two E. clostridioformis oral gavages, other secondary BAs from the classical pathway, TDCA and DCA, could control the progression of MCA205 sarcomas to some extent (Figure 2F, p = 0.035 with TDCA; p = 0.065 with DCA on day 13). GDCA rescues gut dysbiosis-associated resistance to immunotherapy (A) Experimental setting at GRCC. (B–F) Effect of E. clostridioformis with or without GDCA on the response to PD-1 blockade (B and C) or PD-1+CTLA-4 blockade (D and E) against MCA205 fibrosarcomas (B and C) or RET (D and E). Tumor size at day 13–14 (B and D, left) or tumor growth kinetics (B right and E left) in three (C) or two (D) concatenated experiments comprising 25–29 (B), 9–10 (C), and 12–14 or 5 mice/group (E) and one representative experiment (E). (F) MCA205 tumor growth kinetics in mice treated with ABX (2 weeks) and orally gavaged twice with E. clostridioformis and BA (TDCA and DCA) treated with α-PD-1. Two pooled independent experiments of 4–5 animals/group. Comparisons between groups were performed using non-parametric Kruskal-Wallis H test (>2 groups), followed by multiple-comparisons test controlling the false discovery rate (FDR). p values were calculated at each time point comparing each group with E. clostridioformis + corn oil (red curves) with the Mann-Whitney U test (2 groups) (C, E right, and F). The tumor growth kinetics were analyzed using the online tool “TumGrowth” (see STAR Methods) (B right and E left). Mean ± SEM (B, D, and E left) or individual tumor growth kinetics (C, E four right panels, and F) are depicted. (G) Heatmaps representing the log2FC of BA levels between before (at D-1) and after oral BA at day 7 (left) and day 12 (right) in 5 mice/group. p value (∗, p < 0.05; ∗∗, p < 0.01) on the log2FC between oral BA versus corn oil (Mann-Whitney U test, 2 groups). (H) Volcano plot contrasting plasma metabolites after GDCA versus LCA in dysbiotic mice at day 7 (n = 5 mice/group). Green and red dots are significant metabolites (Mann-Whitney U test, 2 groups). See also Figures S4 and S5 and Tables S1 and S2. To investigate the pharmacokinetics of these exogenously delivered BA, we performed longitudinal mass spectrometry-based metabolomics of plasma and liver on day 7, 24 h after the 3rd administration, or on day 12, 4 days after the 4th oral administration of BA. Oral gavage with GDCA and TDCA (but not with LCA or UDCA) increased or tended to increase the recirculation of GDCA and/or DCA but decreased that of TCDCA (Figure 2G; Figures S4A and S4B; Table S1). Oral LCA and/or UDCA (but not GDCA) increased the plasma concentrations of LCA and UDCA as well as those of MDCA and microbial-CA or TLCA, TUDCA, and tauro-HDCA (THDCA) (Figure 2G; Figures S4A and S4B; Table S1). In the liver, GDCA decreased α-MCA, while UDCA increased T/G-UDCA, T/G-HDCA, and TLCA (Figure S4C). The direct comparison of the BA metabolic profile of animals gavaged with GDCA versus LCA confirmed the dominance of GDCA in the plasma at day 7 in the former group, while the LCA-fed group exhibited not only a dominance of LCA and TLCA but also an ABX-like profile (TCDCA, THDCA, and TUDCA) in the liver and plasma (Figures 2G and 2H; Figure S4C). Tauro-conjugated primary BAs, TCDCA and TCA, were dampened by GDCA (Figure 2G, left) and UDCA, respectively (Figure 2G, right) in the plasma. To better understand the kinetics of orally fed BA and their downstream derivatives, we analyzed the bile salt hydrolase/aminoacyltransferase (bsh) sequences and activity of our E. clostridioformis strain 141 that are required to deconjugate acyl-bound amines from the sterol core of the primary BAs.23 Bsh sequencing of our strain revealed that its enzyme is more closely related to Penicillin V acylases, a group of N-terminal nucleophile hydrolases that are similar to classical cholylglycine hydrolases. This is somewhat in contrast to many other Enterocloster species, which contain classical enzymes (Figure S4D). Activity analysis of the E. clostridioformis strain 141 showed that it has a limited capacity to hydrolyze tauro-conjugated BAs, such as TCDCA or T-β-MCA, compared with other Enterocloster species but drastically hydrolyzes GDCA (Figures S4E and S4F), in accordance with the dysbiosis-associated BA abnormalities. Moreover, using 16S rRNA targeted sequencing, we did not observe major differences in the fecal taxonomic composition across BA recipient groups (Figures S5A–S5C). GDCA and TDCA triggered similar microbial shifts (Figures S5A and S5B), and GDCA significantly induced a mild relative overrepresentation of immunogenic Bifidobacterium pseudolongum24,25 and Erysipelatotrichaceae family members26 (Figures S5C and S5D; Table S2). We conclude that oral GDCA supplementation of dysbiotic mice causes a rise in GDCA and DCA levels and a drop in TCDCA level in the plasma while decreasing α-MCA level in the liver, culminating in the restoration of sensitivity to PD-1 blockade, though without correcting dysbiosis. Inhibitory effects of experimental cholestasis on PD-1 blockade Attempting to draw causal links between the accumulation of tauro-conjugated primary BAs and resistance to immunotherapy, we used two procedures. First, we conducted a surgical bile duct ligation (BDL) (versus sham surgery) in mice transplanted with MCA205 fibrosarcoma three days before (Figure 3A), which resulted in cholestatic liver inflammation and injury (Figure 3B) associated with a brisk rise in plasma Tα-β-ω MCA and TCA (Figure 3C; Table S1), paralleled by a drastic drop in CA and DCA (Figure 3D; Table S1). Consequently, BDL culminated in complete cancer resistance to PD-1 blockade (Figure 3E). Secondly, using fexaramine, an ileum-selective FXR agonist (Figure 3F), we markedly rescued the efficacy of α-PD-1 Ab in ABX-treated tumor bearers (Figure 3G), restoring T cell functions in tumor-resident lymphocytes (Figures 3H and S6A) and decreasing plasma Tα-β-ω-MCA compared with ABX alone treatment (p = 0.015 without vs. p = 0.422 with fexaramine on day 7) and TCA (p = 0.001 without vs. p = 0.422 with fexaramine on day 7; Figure 3I; Table S1), albeit with interindividual variability (Figure S6B). Indeed, fexaramine induced the ileal and colonic rise in the FXR antagonist Tα-β-ω-MCA,27 potentially for a subsequent elimination (Figure S6C) but did not act on ABX-induced liver expression of Baat, the enzyme inducing tauro-conjugation of primary BAs (Figure S6D). In parallel, fexaramine tended to increase UDCA (p = 0.052) and DCA (p = 0.054) (Figures 3J and S6B; Table S1) as well as other secondary BAs (Figure S6E). Driven by a recent report showing that TCDCA can promote exhaustion of T lymphocytes infiltrating liver cancers12 and by the capacity of fexaramine to restrain LAG3 expression on CD8+ tumor-infiltrating lymphocytes (TILs) (Figure 3H, right), we assessed whether cholestasis and dysbiosis-associated BAs could directly affect T cell functions in vitro. Indeed, we confirmed that TCDCA and T-β-MCA (but not all the other BAs) markedly reduced the viability of T cell receptor (TCR)-engaged splenic CD3+ T cells, while GDCA maintained the proliferation of CD8+ T cells (Figures 3K, S6F, and S6G). Hence, mechanical cholestasis or gut dysbiosis could lead to plasma accumulation of tauro-conjugated primary BAs, including the strong T-β-MCA FXR antagonist, culminating in T cell dysfunction and resistance to cancer immunotherapy that the ileal FXR agonist fexaramine could antagonize. Modulations of tauro-conjugated primary BAs and efficacy of PD-1 blockade (A) Surgical bile duct ligation (BDL) in MCA205 tumor-bearing mice treated with anti-PD-1 Abs at Centre de Recherche des Cordeliers. (B) Livers were stained with hematoxylin and eosin (top) as well as Sirius Red (bottom). One representative photograph of each liver in each group of 8–10 mice is shown. (C and D) Percentages of distinct BAs influenced by BDL. (E) Mean tumor growth curves in both groups. The tumor growth kinetics were analyzed using the online tool “TumGrowth” (see STAR Methods). (F) Experimental setting of ABX + E. clostridioformis-induced gut dysbiosis compensated by fexaramine in MCA205 tumor bearers at GRCC. (G) Tumor growth kinetics were analyzed as in (E). A representative experiment comprising 5 mice/group is shown, out of two yielding similar results. (H) Flow cytometric analysis of TILs in mice receiving ABX and/or fexaramine and treated with α-PD-1 mAb. One representative experiment comprising 10 mice/group is depicted. One dot represents one tumor. (I and J) Longitudinal plasma metabolomics analyzing various BAs (in percentages of the total BA) over time. Each dot represents one mouse housed at GRCC animal facility. (K) Ex vivo stimulation of CD3+ T splenocytes exposed to various concentrations of BAs (left), followed by flow cytometry analysis of the co-culture for cell viability and proliferation. The absolute number of live CD3+ T cells or percentage of Ki67+ in CD8+ T cells recovered at 24 h of BA exposure is shown (middle and right, respectively). Also refer to Figures S6F and S6G for CD4+ T cells and fold change calculations. Each dot represents one well, and each experiment was performed in triplicate wells and performed between three to five times (n = 9–15 wells), except for T-β-MCA condition that was performed once (n = 3 wells). n = 9–10 mice/group (B–E and H–J). Error bars indicate mean ± SEM (C, D, E, G, H, and K). Comparisons between groups were analyzed using non-parametric Mann-Whitney U test (2 groups) or Kruskal-Wallis H test (>2 groups), followed by multiple-comparisons test controlling the false discovery rate (FDR) (C, D, H, and K), or non-parametric Friedman test (>2 paired groups for longitudinal metabolomics analysis), followed by two stage set-up method of Benjamini, Krieger, and Yekutieli (I and J). The tumor growth kinetics (E and G) were analyzed using the online tool “TumGrowth” (see STAR Methods). See also Figure S6 and Table S1. Gut BAs reach the subcutaneous TME to reprogram local immunity Natural and orally fed BAs reached distal tissues by day 12, as revealed by tumor metabolomics (Figures 1F and 1G and 4A–4D; Table S1). Indeed, oral gavage with GDCA increased GDCA tumoral levels only in one-third of the animals (Figure 4A, p = 0.144), while that of UDCA induced a significant rise in LCA and UDCA (Figures 4B and 4C, p = 0.004 and 0.03, respectively). This might be explained by the capacity of E. clostridioformis strain 141 to metabolize GDCA into DCA in the intestines (Figures S4E and S4F). Of note, oral GDCA could prevent the tumoral accumulation of Tα-β-ω-MCA observed with E. clostridioformis (Figure 4D, p = 0.025). GDCA acts at sites distal from the liver to reprogram the TME (A–D) Intratumoral abundance of orally gavaged BAs determined by mass-spectrometry-based metabolomics in E. clostridioformis dysbiotic mice treated with α-PD-1 on day 12 post MCA205 injection, comparing BA vs. water + corn oil controls using Z scores of AreaQCcorr. Shown is the concatenation of 2 experiments including 3–5 mice/group. (E–G) Flow cytometric analysis of LAG-3 expression on various CD4+ T subsets infiltrating the tumor of dysbiotic mice receiving BA and α-PD-1 mAb. One representative experiment comprising 5–10 mice/group is depicted. (H–K) Violin plots showing the expression levels of selected genes encoding the exhaustion signature in all CD8+ TILs in each treatment group on day 11. See also Figure S7. (L and M) In vitro screening of several BAs at 100 μM or cilofexor at 50 μM on Madcam1 relative expression by RT-qPCR in TSECs exposed to TNF-α (vs. negative control [Ctrl-]) (L) or TSECs genetically modified by means of CRISPR-Cas9 gene editing (M). Results from 3–4 independent experiments were pooled, and each dot represents one well; n = 18–42 (L) and 24 (M). (N) In vivo effects of orally gavaged BA on Madcam1 expression in the ileum of mice treated with ABX. Results from 2 independent experiments are pooled using Z score, and each dot represents one mouse; n = 10. (O–Q) Experimental setting (O) and flow cytometric determination of the percentage of CFSE+ T cell subsets reaching the tdLN in day 12 tumor-bearing mice at 24 h after CFSE injection into mLN (P and Q). Results from 2 independent experiments are pooled, and each dot represents one mouse; n = 7–11. (R and S) Differential effects of corn oil (R) versus GDCA (S) in WT vs. Madcam-1−/− MCA205-bearing mice treated with PD-1 blockade. Two pooled experiments comprising n = 9–10 mice/group, showing individual kinetics and mean + SEM tumor growths. p values were calculated at each time point, comparing the two groups. Error bars indicate mean ± SEM (A–N and P–S). Comparisons between groups were performed using non-parametric Kruskal-Wallis H test (>2 groups), followed by multiple-comparisons test controlling the false discovery rate (FDR), except written otherwise (A–G, L–N, P, and Q), or significance was calculated with Wilcoxon rank-sum test and adjusted for multiple testing using Benjamini-Hochberg correction (H–K). The tumor growth kinetics (R left and S left) were analyzed using the online tool “TumGrowth” (see STAR Methods). For R (right) and S (right), p values were calculated at each time point comparing the two groups together using the Mann-Whitney U test. See also Figures S4–S8 and Tables S1 and S3. Flow cytometric analyses of TILs and myeloid cells unveiled an E. clostridioformis-induced rise in the LAG3 exhaustion marker in CD4+ T cells (Figure 4E, p = 0.01; Figure S6A), most specifically in effector memory (TEM) CD4+ T cells (Figures 4F and 4G, p = 0.006). Conversely, GDCA significantly prevented this dysbiosis-associated T cell exhaustion, dampening LAG3 in CD4+ (Figure 4E, p = 0.035) and in TEM CD4+ T cells (Figures 4F and 4G, p = 0.013). We confirmed these findings using single-cell RNA sequencing of tumor-infiltrating CD45+ cells on day 11 post oral gavage with GDCA, DCA, or LCA (Figures 4H–4K; Figure S7). Single-cell-resolved levels of gene expression of the TME were determined after isolating 18,669 and 14,842 cells from the tumor beds and tumor-draining lymph nodes (tdLNs), respectively, across all experimental conditions. We identified 20 clusters in tumors (including 6 T cell clusters) and 19 clusters in tdLNs (including 12 T cell clusters) (Figures S7A–S7C). Focusing on GDCA versus corn oil (CO) control-imprinted TME, we did not find a quantitative impact of GDCA on the distribution of T cells within clusters (Figures S7B and S7C), but rather observed significant variations in gene expression levels in the proliferating CD4+ and/or CD8+ T cell subset in the tumor beds and in the naive/memory CD8+ T cell cluster in the tdLNs (Figure S7D, lower and upper panels). This naive/memory CD8+ T cell subset from GDCA-imprinted tdLNs was characterized by a type 1 IFN signature (Stat1, Ifit3, Ifi27l2a, Gimap7, Gbp2, and B2m) associated with the upregulation of the prototypic translation machinery (including many Rps and Rpl ribosomal proteins) and the cytoskeleton remodeling (Actb), with a downregulation of the mitochondrial oxidative metabolism (Lgals1, mt-Cytb, mt-Co-1/2/3, and mt-Atp6) (Figure S7D, upper panel). This signature has been described during the T cell clonal expansion phase, notably followed by inhibition of the translation of those transcripts in CD8+ T effectors,23 what we indeed observed in the TME (Figure S7D, lower panel). The cluster of proliferating CD4+ and/or CD8+ T cells in the GDCA-imprinted TME (defined by markers featuring in Table S3, including Nkg7, Cxcr6, Skap1, Ablim1, Sh2d2a, MKi67, Cdk1, and Top2a) that may contain progenitor exhausted T cells (Tpex), downregulated Ptpn22,28 Gzmb29 (Figure S7D, lower panel), Tox, Havcr2, Pdcd1, and Lag3 (Figures 4H–4K) gene transcripts, while upregulating the AP1 family member Fos, Il21r, Il1r2, Cd40, and Cd70 gene products (Figure S7D, lower panel), suggesting retardation in the T cell terminally differentiation status. Interestingly, one of the top gene transcripts upregulated by GDCA on this proliferating CD4+ and/or CD8+ T cell subset is the creatine transporter Slc6a8, critical for TCR-mediated activation of the mTORC1 signaling relevant in cancer immunosurveillance (Figure S7D, lower panel).30,31 The proliferating CD4+ and/or CD8+ T cells of the “GDCA tumors” also had a myeloid chemoattracting secretory profile (Figure S7D, lower panel). Intriguingly, the dominant CX3CR1 monocytic/macrophage cluster within the myeloid compartment of “GDCA tumors” was characterized by the expression of the nonsense mediated mRNA decay (NMD), a cellular quality control pathway degrading mutant transcripts, potentially involved in tumor antigenicity and immunoediting32 (Figure S7E). In the “DCA-imprinted TME”, the CD8+ cytotoxic T cell cluster was characterized by the functions associated with proficient antitumor responses, including the upregulation of Bhlhe40, Trbv2, Ccr7, Irf8, Il21r, and Sema4c33 (Table S3). The “GDCA imprinted” gene profiles contrasted with that observed with LCA-gavaged mice where NOS2+ and/or CX3CR1+ macrophages lost transcription of their antigen-processing machinery (H2-K1, H2-D1, H2-Ab1, Ctss, Ccl5, and Ifitm3) (Table S3). Of note, “LCA imprinted” tdLN-derived naive/memory CD8+ T cells upregulated Adamts17, a sheddase blunting the CD122 signaling in CD8+ T cells34 (Table S3). Moreover, the gene expression fingerprint of LCA TIL revealed a profound downregulation of the type 1 and 2 IFN responses compared with the negative control-imprinted TME (Figure S7F; Table S3). Altogether, these results showed that BAs can act at distal sites to reprogram the TME and tdLNs, influencing the outcome of tumor-specific immune responses. Because dysbiosis facilitates the exodus of terminally differentiated and exhausted α4β7-expressing CD4+ Tregs and their emigration to the tdLNs by downregulating ileal MAdCAM-1 in high endothelial venules,21 we wondered whether BAs might downregulate MAdCAM-1 expression, thereby favoring the emigration of enterotropic Tregs to tdLNs. For this, transformed endothelial sinusoidal cells (TSECs) were cultured in the presence of physiological concentrations of an array of commercially available primary and secondary, tauro- and glyco-conjugated BAs, followed by the quantification of Madcam1 mRNA by RT-qPCR.21 As reported, LCA downregulated Madcam1 expression.21 In sharp contrast, DCA and its derivatives (GDCA, HDCA, and UDCA) upregulated Madcam1. Conversely, most tauro-conjugated BAs failed to affect Madcam1 expression (Figure 4L). CRISPR-Cas9-based knockout of 3 potential BA receptors (Fxr, Vdr, and Tgr5) in TSECs led to the conclusion that Fxr/Nr1h4 is essential for the GDCA-induced upregulation of Madcam1 (Figure 4M). In vivo, GDCA efficiently increased Madcam1 mRNA in the ileum from ABX-treated mice, validating the in vitro results (Figure 4N). Next, we injected carboxyfluorescein succinimidyl ester (CFSE) into the mesenteric lymph nodes (mLNs) and then monitored the emigration of CFSE-labeled cells to distant sites at 24 h in tumor-bearing mice fed with E. clostridioformis under PD-1 blockade (Figure 4O). Oral administration of GDCA decreased the migration of CFSE+ α4β7+ CD4+ T cells (and more specifically α4β7+ CD4+ Foxp3+ CD25+ Treg cells) (Figure 4P) but not that of α4β7− CD4+ Treg to tdLNs (Figure 4Q), compared with E. clostridioformis + LCA, albeit non-significantly compared with the E. clostridioformis group without BA gavage. Finally, we compared the capacity of GDCA to restore sensitivity to PD-1 blockade in MCA205 tumor-bearing Madcam1-proficient mice versus deficient mice in settings of gut dysbiosis (as detailed above, Figure 2F). GDCA lost its adjuvant properties in the Madcam1−/− mice (Figures 4R and 4S) but not in Tgr5/Gpbar −/− animals (Figure S8A). Incidentally, to exclude possible direct effects of BAs on cancer cells, we analyzed the expression patterns of the five BA receptors in MCA205 fibrosarcoma by RT-qPCR. In comparison with ileal mucosa, MCA205 sarcoma tissues expressed low levels of Fxr/Nr1h4, Vdr, and Tgr5/Gpbar1 gene transcripts (Figure S8B). At physiological concentrations, BAs failed to kill HT29 cells, which readily succumbed to oxaliplatin or mitoxanthrone (Figure S8C). CRISPR-Cas9-mediated knockout of Fxr/Nr1h4 or Tgr5/Gpbar1 failed to affect tumor outgrowth, but the Fxr/Nr1h4 knockout sensitized MCA205 fibrosarcomas to α-PD-1 blockade (Figures S8D and S8E, p = 0.001). In summary, oral GDCA reached tumor bed and prevented the exhaustion of TIL effector memory lymphocytes. It is, at least in part, a consequence of the MAdCAM-1-dependent blockade of tumor homing of immunosuppressive enterotropic Tregs. Subclinical cholestasis in cancer patients treated with immune checkpoint blockade To investigate the clinical significance of these findings, we analyzed the effects of ABX on mass spectrometry-detectable circulating BAs in a prospective cohort (ONCOBIOTICS cohort [NCT04567446]), which included 133 patients with lung, kidney, or bladder cancer treated with immune checkpoint blockade (ICB)-based first-line regimens.35 ABX exposure prior to plasma sampling significantly decreased the plasma concentration of UDCA (Figure 5A, p = 0.021 compared with no ABX exposure), while increasing TCDCA (Figure 5A, p = 0.038 compared with no ABX exposure). These ABX effects were significant within a 30-day time window (Figures 5B–5D) and led to a disruption of the normal pattern of correlations between unconjugated primary and secondary BAs (Figure 5E). In the 61 patients with paired plasma metabolomics and stool metagenomics data, S-score (reflecting gut dysbiosis when closest to 0)36 showed significant, albeit weak, positive correlations with DCA (Figure 5F, left) and negative correlations with total tauro-conjugated BAs (Figure 5F, right). Gut dysbiosis alters BA composition in cancer patients (A–D) Percentages of each BA among the total BAs in the plasma, assessed using metabolomics in 133 cancer patients (n = 138 samples) from the ONCOBIOTICS cohort who took ABX (n = 16 + [0–30 days] (A–D); n = 9 +[>30 days] and n = 13 + [>60 days] (B–D)) or did not take ABX (n = 100) (A–D) within different time windows post-ABX cessation as well as for plasma ratio of glyco- over tauro-conjugated BAs (D, bottom). Error bars indicate mean ± SD. Comparisons between groups were performed using non-parametric Mann-Whitney U test (2 groups) in (A) and non-parametric Kruskal-Wallis H test (>2 groups) in (B)–(D). (E) Clustering of Spearman’s correlation coefficients of plasma BAs in patients who took ABX (n = 13 +[0–30 days], left) or did not take ABX (n = 85, right). (F) Spearman’s correlations between distinct plasma BAs (in percentages of total BA) and the paired stool S-score,36 calculated in 61 ONCOBIOTICS patients before and until 15 days after immunotherapy initiation. See also Figures S9–S11 and Tables S4, S5, and S6.. To dissect the patterns of circulating BAs in the ONCOBIOTICS cohort of cancer patients, we performed an unsupervised hierarchical clustering of the plasma concentrations of BAs, monitored before or within the first 42 days of PD-1/PD-L1 blockade in 111 patients. Three clusters of patients could be clearly identified (Figure 6A; Table S4). Cluster 1 was characterized by the relative dominance of unconjugated primary and secondary BAs (Figure 6B). In cluster 2, patients harbored higher levels of glyco-conjugated primary BAs than those of the others (Figure 6C), while cluster 3 was typified by the accumulation of tauro-conjugated primary and secondary BAs, leading to a ratio between glyco-/tauro-conjugated BAs of <3, as well as by elevated plasma levels of the inflammatory marker calprotectin (S100A9) (Figures 6D and 6E). A similar clustering pattern was found in metastatic melanoma (Figures S9A–S9C) and NSCLC patients (Figures S9D–S9F) in Canada (as described by Routy et al.37 and Duttagupta et al.38; Table S4). Univariate Cox regression analysis of the overall survival (OS) of ONCOBIOTICS patients revealed that cluster 3 tends to be associated with dismal outcome (Figure 6F; Table S4). Notably, these 3 clusters exhibited different immune profiles. Cluster 2 was enriched in naive CD8+ T cells, while cluster 3 showed a relative scarcity of total and memory CD27+ and CD127+ CD4+ T cells (Figures S10A and S10B). Cancer patients can be stratified based upon plasma BA composition (A) Non-supervised hierarchical clustering of the percentages of plasma BAs in 111 ONCOBIOTICS patients before and until 42 days post ICB initiation. (B–E) Dot plots of the percentages of BAs or conjugated BAs and calprotectin (S100A9) measured in the plasma according to the three clusters illustrated in (A). Also refer to Table S4 and Figures S9 and S10. Each dot represents plasma levels in one patient (n = 111). Error bars indicate mean ± SD. Comparisons between groups were performed using non-parametric Kruskal-Wallis H test (>2 groups), followed by multiple-comparisons test controlling the false discovery rate (FDR). (F) Kaplan-Meier survival curves and univariate Cox regression to assess the prognostic value of each BA cluster (Cox-Mantel test). See also Figures S9–S11 and Tables S4, S5, and S6 Given the limited effect size of ONCOBIOTICS and the demonstration of a potential clinical relevance of subclinical cholestasis1 brought up in rare retrospective studies in renal cell carcinoma patients treated with nivolumab,39,40 we determined whether the routine biomarkers of cholestasis, which are gamma-glutamyl transferase (γGT) and alkaline phosphatases (ALP), might correlate with immunotherapy resistance across six independent and pan-cancer cohorts gathering 599 patients: ONCOBIOTICS (N = 111),35 PREMIS ( N = 185),41 G2R (N = 100), MEDITREME (N = 54),42 METIMMOX (N = 38),43 and AtezoTRIBE (N = 111) amenable to PD-1/PD-L1 and/or CTLA-4 blockade, alone or combined with other therapeutic modalities (Figure 7A; Table S5). An elevation of γGT (defined as twice the upper limit of normal [ULN]: ≥45 U/L for males and ≥35 U/L for females) was observed at baseline in 25.7% ± 3.9% (mean ± SEM) of patients across these cohorts (Figure 7A; Tables S5). Among the 561 patients across the five cohorts, those above this threshold at immunotherapy initiation exhibited reduced progression-free survival (PFS) and/or OS in univariate analyses (Table S5; Figures 7B–7G; Figure S11A), except the AtezoTRIBE arm “no ICB” (Figure 7F). The Cox regression multivariate analyses adjusting for liver metastases and other prognostic factors (but not for co-medications or autoimmune hepatitis) confirmed that γGT was an independent negative prognostic parameter for PFS in ONCOBIOTICS (p = 0.024, Table S6), G2R (p = 0.015, Table S6), and MEDITREME (p = 0.013, Table S6), as well as for OS in PREMIS (p = 0.014, Table S6), MEDITREME (p = 0.048, Table S6), and AtezoTRIBE (p = 0.039, Table S6). Subclinical cholestasis reduces cancer patient survival (A) Patient cohorts amenable to immunotherapy (ICB), +/− chemotherapy (CT), radiotherapy (RT), or tyrosine kinase inhibitors (TKI) and randomized in 2 arms (AtezoTRIBE) for whom γGT and alkaline phosphatase (ALP) levels were assessed in the plasma. (B–G) Univariate analyses for overall survival (OS) (B, F, and G) and progression-free survival (PFS) (C–G), using the Kaplan-Meier estimator and Cox regression with indicated hazard ratios (HRs) and log rank (Mantel Cox) test p values to assess the prognostic value of γGT in each cohort aligned in (A). Also refer to Tables S5 and S6 for univariate and multivariate analyses, respectively. (H) Longitudinal evolution of the mean γGT levels in U/L (scale in log10) for 111 patients who progressed (PFS < 6, orange) or not (PFS ≥ 6, blue) within 6 months of ICB initiation in the ONCOBIOTICS cohort, as well as for different periods of time. Comparisons between time points (0–1, 1–6, and 6–12 months) were performed using non-parametric Wilcoxon matched-paired test for each PFS group. Comparisons between groups (PFS <6 or ≥6) were performed using non-parametric Kruskal-Wallis H test (>2 groups), followed by multiple-comparisons test controlling the false discovery rate (FDR). (I) Clustering of Spearman’s correlation coefficients between BAs (%), liver parameters, and sMAdCAM-1 (in ng/mL) in the plasma of patients from ONCOBIOTICS, FMT-LUMINATE, MIMIC, and AtezoTRIBE cohorts (p values: ∗, <0.05; ∗∗, <0.01; ∗∗∗, <0.001). High γGT levels were defined as ≥2ULN (ULN: male patients ≥ 45U/L; female patients ≥ 35U/L). See also Figures S9–S11 and Tables S4, S5, and S6. Similar results were obtained for elevated ALPs (ALP ≥ 1 ULN) that were associated with a dismal prognosis in univariate analyses (Figures S11B–S11D for ONCOBIOTICS, METIMMOX, and MEDITREME) and in multivariate analyses, for PFS and OS in ONCOBIOTICS (p = 0.002 and 0.042, respectively; Table S6) and for OS in MEDITREME (Table S6, p = 0.017). Immunotherapy tended to decrease γGT concentrations in patients with PFS of ≥6 months, while the levels remained ≥2 ULN in patients who progressed within 6 months after ICB initiation (Figures 7H and S11E). This trend was not observed for ALP levels (Figures S11F and S11G), although patients with elevated ALP at the start of immunotherapy had significantly reduced PFS in both ONCOBIOTICS and METIMMOX cohorts (Figures S11B and S11C). Moreover, circulating liver enzymes (γGT, ALP, and transaminases) and soluble MAdCAM-1 (sMAdCAM-1) at baseline were positively and negatively correlated with tauro-conjugated BAs, respectively, in all patients from the pooled ONCOBIOTICS, as well as in the individual FMT-LUMINATE, MIMIC, and AtezoTRIBE cohorts (Figure 7I). The increased prevalence of immunosuppressive SIG1 bacteria (as defined by the S-score and Toposcore36) in patients presenting elevated γGT or tauro-conjugated BAs at baseline is in line with the observation that γGT levels are higher in patients after ABX uptake (Figures 8A–8C; Figure S11H). Moreover, γGT levels correlated with blood markers of inflammation (neutrophils) and gut permeability (soluble ST2) (Figure 8D). In accordance with our previous report21 and Figure 7I, sMAdCAM-1 and γGT baseline plasma levels exhibited a negative correlation in all three cohorts pooled together and PREMIS (Figure 8E, p = 0.008 and 0.004, respectively). Altogether, subclinical cholestasis defined by γGT can be considered as a proxy of gut dysbiosis. Subclinical cholestasis is a surrogate marker of gut dysbiosis (A) Absolute number and percentages of SIG1 bacteria among the 37 SIG136 in patient feces according to plasma γGT < or ≥ 2 ULN (n = 73 and 8 samples, respectively). (B) Plasma γGT levels (in U/L) in patients treated with ABX according to the window of blood sampling before ABX (n = 85 in no ABX and 18 samples in ABX cessation less than 60 days, respectively). (C) Spearman’s correlation between percentages of tauro-conjugated BAs and SIG1 bacteria species among 37 SIG1 at baseline.36 (D) Increase in neutrophil blood counts and plasma soluble ST2 levels in patients with γGT ≥ 2ULN versus those with γGT < 2ULN (n = 16 and 95, respectively). (E) Spearman’s correlation between γGT (U/L) level and sMAdCAM-1 (ng/mL) in ONCOBIOTICS, FMT-LUMINATE, and MIMIC pooled cohorts and PREMIS. (F) Trajectory of TCDCA (in mean and CI95% Log2AreaQCcorr) in the plasma (n = 32 patients) or in the stools (n = 34 patients) of responder vs. non-responder patients before and during ICB in pooled MIMIC and FMT-LUMINATE cohorts. Comparison between responders and non-responders were performed using a longitudinal slope estimation approach. Left: the significance of temporal changes of plasma TCDCA levels within responders and non-responders was assessed from the estimated time effect in each group, and p values are annotated (related to STAR Methods). Right: Spearman’s correlations between fecal TCDCA levels and days after FMT. (G) Boxplots of raw values of TCDCA (AreaQCcorr) at three time points in TACITO (n = 45). The box shows the interquartile range (Q1–Q3), with the median marked inside. Whiskers extend up to 1.5 × IQR beyond Q1 and Q3. Values outside the whiskers are plotted separately as outliers. (H and I) Idem as in (F) right in the 2 arms of TACITO for TCDCA (H) and TCA (I), depicted in trajectories. Also refer to Figure S11I to visualize the differential evolution of all BAs in the two treatment arms. The significance of temporal changes of plasma TCDCA (H) and TCA (I) levels within placebo and FMT arms after the treatment start was assessed using Spearman’s correlations, and p values and coefficients correlation (Rho) are annotated. Comparisons between groups were performed using non-parametric Mann-Whitney U test (2 groups) (A, B, and D) or Kruskal-Wallis H test (>2 groups) (G), followed by multiple-comparisons test, controlling the false discovery rate (FDR). Error bars represent mean ± SD in (A)–(D). See also Figures S9–S11 and Tables S4, S5, and S6. Finally, we addressed whether FMT might avoid the accumulation of tauro-conjugated BAs associated with gut dysbiosis and resistance to immunotherapy in cancer patients. Using a linear mixed-effect model approach to predict trajectory slope, we analyzed the evolution of each plasma BA pre- and post-FMT in responders versus non-responders from the MIMIC and FMT-LUMINATE trials, using three time points of plasma metabolomics.37,38,44 Notably, tauro-CDCA (TCDCA) followed an ascending slope in non-responders but a stable or descending slope in responders (Figure 8F, left). Concomitantly, the levels of TCDCA were increased in the stool of the non-responder patients (Figure 8F, right; p = 0.009). We validated this observation in the TACITO trial randomizing metastatic renal cell carcinoma (mRCC) patients into FMT versus placebo, in addition to the combination of tyrosine kinase inhibitor and ICB, showing a 30% increase in patients without progression at 12 months in the FMT arm.45 Plasma levels of TCDCA increased in the first week of therapy in the placebo but not in the FMT arm (Figure 8G, p = 0.03), and a descending slope of TCDCA and TCA was observed in the FMT group over the first year of follow-up, but not in placebo (Figures 8H and 8I; Figure S11I). Altogether, FMT can be considered as a therapy preventing dysbiosis-associated accumulation of tauro-conjugated BAs over the course of the disease and/or treatment. Discussion This study links gut dysbiosis with abnormalities in BA recirculation to the tumor, compromising tumor immunosurveillance and hence causing resistance to ICB in murine malignancies distant from the enterohepatic cycle. In particular, the abnormal synthesis or recirculation of TCDCA or Tα-β-ω-MCA (FXR antagonists) and the loss of secondary BAs of the classic pathway (DCA, an FXR agonist, and GDCA/TDCA derivatives) act in concert to compromise cancer immunosurveillance during ABX uptake or E. clostridioformis-enforced gut dysbiosis. Our data extend previous findings using BA oral gavage in mice19,21 and others demonstrating the aberrant activity of the BAAT enzyme in liver cancers culminating in TCDCA synthesis and exhaustion of TILs.9 In mice, GDCA (and its derivative DCA, an FXR agonist), FMT, or the ileal FXR agonist fexaramine could dampen plasma levels of TCDCA and/or Tα-β-ω-MCA, notoriously FXR antagonists, while surgical BDL did the contrary, phenocopying the effects of ABX. Accordingly, fexaramine and GDCA supplementation of dysbiotic mice prevented CD4+ effector T cell and Tpex exhaustion (characterized by Lag-3, Tox, and Tim-3 expression) in TILs. Hence, the balance between FXR agonists and antagonists in the ileum might regulate immunosurveillance of tumors located at distal sites from the intestines. The prognostic value of baseline γGT levels had long been underestimated until Winter et al. reported their independent negative impact on OS.46 Our results across six independent cohorts of cancer patients corroborated this original study, suggesting that γGT elevations may be added to the routine panels that identify patients unlikely to respond to immunotherapy, presumably as a consequence of a disabled gut-liver axis. Therapeutic strategies (such as UDCA47 or obeticholic acid1) to curtail liver cholestasis aim at increasing BA flow, reducing BA reabsorption from the intestine, shutting down BA synthesis in the liver, and/or triggering intestinal FGF15/FGFR-βKlotho signaling. However, in our hands, UDCA inconsistently circumvented primary resistance to α-PD-1 mAb induced by ABX/E. clostridioformis. In contrast, oral GDCA supplementation restored immunotherapy responses in dysbiotic mice. A proof-of-concept phase 1 trial showed that GDCA-containing capsules repressed primary BA synthesis without impacting glucose and energy metabolism.48 FMT represents another option for dysbiotic cancer patients. Indeed, subclinical cholestasis correlated with low Toposcore or ABX uptake, downregulation of sMAdCAM-1, and systemic inflammation in our study, in accordance with previous reports suggesting that intestinal microbiota plays an important role in the pathophysiology of cholestatic liver disease.49,50 Pioneering studies have evaluated the interceptive function of probiotics, such as Lactobacillus rhamnosus GG, in cholestatic liver diseases.51,52 Probiotics could decrease the T-β-MCA FXR antagonist, unleashing liver FXR activity. Similarly, gut microbiota not only regulates secondary BA metabolism but also inhibits BA synthesis in the liver by alleviating FXR inhibition in the ileum.53 Therefore, restoring a healthy intestinal commensalism may avoid cancer-associated subclinical cholestasis to facilitate adaptive anticancer immune responses. Prospective clinical trials may stratify cancer patients based on various proxies of gut dysbiosis including subclinical cholestasis to predict primary resistance to cancer immunotherapy. Limitations of the study In murine models, Enterocloster spp.-induced gut dysbiosis was not stricto sensu mimicking that mediated by ABX. The use of stools from ABX-treated patients for a transfer into avatar mouse models would have been an additional line of evidence that gut dysbiosis is clinically significant to dampen ICB efficacy. Demonstration of a causal link between GDCA compensatory effects on dysbiosis and endothelial MAdCAM-1 function would have required endothelium-restricted Nr1h4/Fxr gene deletion. Therefore, this study cannot exclude alternative BA sensing routes. Multivariable analyses testing the prognostic value of γGT in cancer patients took into account only ABX and liver metastases but failed to consider other confounding factors such as hepatic autoimmune reactions,46 alcohol consumption, several co-medications, or the socioeconomical status. Prospective studies including these extrinsic factors should confirm these findings. Resource availability Lead contact Request for further information and resources should be directed to and will be fulfilled by the lead contact, Laurence Zitvogel (laurence.zitvogel@gustaveroussy.fr). Materials availability This study did not generate new unique reagents. Data and code availability The bulk RNA sequencing of murine hepatocytes, cholangiocytes, and enterocytes (refer to Figure S3) and the 16S sequencing data of murine fecal samples (refer to Figures S2 and S5) have been deposited in the ENA (European Nucleotide Archive) database under submission number ERA35991104, with project number ENA: PRJEB109926. Raw and processed data of the single-cell RNA sequencing of murine immune cells from tumors and tdLNs (refer to Figures 4 and S7) have been deposited in the Gene Expression Omnibus (GEO) under accession number GEO: GSE326943. Human and mouse metabolomic data supporting the claims of the manuscript are available in Mendeley Data: https://doi.org/10.17632/wsydn677rw.1. Human metagenomic data for S-score and Toposcore calculations for Figures 5F, 8A, and 8C are available in Mendeley Data: https://doi.org/10.17632/wsydn677rw.1. No new code was generated in this study. Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request. Acknowledgments L.Z., G.K., L.D., and M.F. were supported by the European Research Council (ERC) advanced grant under agreement number 101052444 and 101142062, the ANR-23-RHUS-0010 (LUCA-pi), the European Union’s Horizon Europe research and innovation program under grant agreement no. 101095604 (project acronym: PREVALUNG-EU, project title: Personalized lung cancer risk assessment leading to stratified Interception), “ANR-21-5 RHUS-0017” IMMUNOLIFE,” and MAdCAM INCA_ 16698, Ligue contre le cancer. L.D. received the ARC Made-IT (2023). L.Z., G.K., L.D., A.E., M.F., B.R., and G.I. received funding from the SEERAVE Foundation. G.K. is supported by ANR under the France 2030 program (ESR/Equipex+ Onco-Pheno-Screen, 21-ESRE-0028), European Research Council (ERC-2021-ADG, grant no. 101052444), Institut National du Cancer (INCa), PAIR-Obésité INCa 18713, and SIRIC CARPEM (INCa-DGOS-Inserm-ITMO Cancer_18006, IdEx Université de Paris Cité ANR-18-IDEX-0001). A.E. was supported by The Fonds de recherche Quebec – Santé (FRQS) Clinician Scientist Award, the Cancer Research Society Next Generation of Scientists, and the Society of Immunotherapy of Cancer-Melanoma Research Alliance Women in Melanoma Award. B.R. was supported by the FRQS Clinician Scientist Award. PRIMM was supported by the European Research Council (ERC-STG project MetaPG to N.S.), MIUR ‘Futuro in Ricerca’ (grant RBFR13EWWI_001 to N.S.), the European Horizon 2020 program (ONCOBIOME-825410 project and MASTER-818368 project to N.S.), the National Cancer Institute of the National Institutes of Health (grant 1U01CA230551 to N.S.), and the Premio Internazionale Lombardia e Ricerca 2019. O.K. received funding from INCa, ARC, the European Innovation Council (EIC), and the Fondation Gustave Roussy. A.L.M.d.L.V., S.T., E.R., and D.S. were all supported by Paris-Saclay University, Doctoral School CBMS582. IHU Mediterranée infection was supported by the National Research Agency under the “Investissements d’Avenir program with the reference ANR-10-IAHU-03, by the Contrat Plan Etat Region and the European funding FEDER IHUPERF. The PRISM project was funded by the Agence Nationale de la Recherche (ANR) under France 2030, grant number ANR-23-IAHU-0002. ATEZOTRIBE was partially supported by the European Union - NextGenerationEU through the Italian Ministry of University and Research (PNRR - M4C2-I1.3 Project PE_00000019 “HEAL ITALIA” CUP): I53C22001440006 to C.C. The METIMMOX study was supported by the Norwegian Cancer Society (grants nos. 182496 and 215613). The PREMIS was funded by Fondation Malakoff Médéric/Humanis and Sanofi. The MEDITREME study was supported by AstraZeneca. A.S. and N.W. were funded by DFG (German Research Foundation) – Project-ID 403324012 – SFB 1382. S.M.V. received salary support from the Ontario Institute of Cancer Research (OICR) and The London Health Sciences Foundation Helen and Andy Spriet funds. The S.M.V. laboratory was funded by grants from the Lotte and John Hecht Memorial Foundation and the Weston Family Foundation and a project grant from the Canadian Institute of Health Research (203745). C.Cremolini. received support at UNIPI by European Union - NextGenerationEU through the Italian Ministry of University and Research under PNRR - M4C2-I1.3 Project PE_00000019 “HEAL ITALIA” - CUP: I53C22001440006. F.G. received support from an ARC Leader de Demain grant and Fondation Gustave Roussy. S.P. is supported by the My First AIRC Grant 2025 of the AIRC (project 32405). G.I. is supported by the PNRR 2023 (project PNRR–POC–2023–12377319) of the Italian Ministry of Health, by the Next Gen Clinician Scientist 2024 of the AIRC (project 30203), by the Fondo Italiano per la Scienza of the Italian Ministry of Research (project FIS00001711), by the European Research Council (project ERC-StG MicroRestore-101221279), and by the Joint Programming Initiative on Antimicrobial Resistance (JPIAMR, project DONATE ERP-2022-23682517). Fondazione Policlinico Gemelli IRCCS staff thank the Fondazione Roma and was also supported by the Ricerca Corrente 2026 of the Italian Ministry of Health. R.A.Q. acknowledges support from NIH grant R01DK140854. C.d.l.C.-F was supported by ARC (ARCPOST-DOC2022070005398) and Fondation Gustave Roussy. C.Ciccarese was supported by My First AIRC Grant MFAG 29275, AIRC - Fondazione AIRC per la Ricerca sul Cancro. R.I. was supported by the Ricerca Finalizzata Giovani Ricercatori 2018 (project GR–2018–12365734) of the Italian Ministry of Health. We thank Dr. Kristina Schoonjans, MacKenzie Brasseur, Nabila TOUAZI, Dr. Clara Sehl, and Dr. Chiara Maiuri for their collaborative efforts. All icons, experimental settings, and the graphical abstract were generated using BioRender. Author contributions Conceptualization, L.Z.; data curation, A.L.M.d.L.V., M.F., M.A.V., C.d.l.C.-F., F.-X.D, Y.H., and M.D.; formal analysis, A.L.M.d.L.V., M.F., C.T., I.L., A.-L.T., S.T., M.A.V., Y.H., T.-H.P., D.S., V.I., S.J., G.V., D.K., Y.F., and R.A.Q.; investigation, A.L.M.d.L.V., M.F., C.T., I.L., M.M., A.-L.T., S.T., S.L., O.M., H.G.d.T., M.A.V., P.L., E.R., B.T.A., S.D., G.V. F.A., M.L., S.F., O.K., C.d.l.C.-F., P.-E.F., F.G., D.K., Y.F., R.A.Q., C.A.C.S., L.D., G.K., L.Z., F.-X.D, C. Ciccarese, and R.I.; methodology, L.Z., G.K., A.L.M.d.L.V., M.F., C.T., I.L., M.M., A.-L.T., S.T., S.L. S.D., F.A., M.A.V., M.L., S.F., O.K., C.d.l.C.-F., A.S., N.W., D.K., Y.F., and R.A.Q.; resources, F.M., C. Cremolini, S.M.V., J.G.L., A.M., F.-X.D., M.D., C.T., F.G., A,H.R., P.A.B., S.M., S.P., A.E., B.R., C. Ciccarese, R.I., G.I., A.S., and, N.W.; software, M.F., M.A.V., C.d.l.C.-F., Y.H., T.-H.P., D.S., and V.I.; supervision, L.Z., G.K., M.F., and L.D.; validation, all authors; visualization, A.L.M.d.L.V., M.F., A.-L.T., S.T., M.A.V., S.L., S.F., Y.H., M.L., C.d.l.C.-F., D.K., Y.F., R.A.Q., and V.I.; writing – review & editing, L.Z., G.K., M.F.; A.L.M.d.L.V., A.-L.T., S.T., L.D. T.-H.P., and Y.H.; project administration, L.Z., G.K., A.L.M.d.L.V., M.F., A.-L.T., and S.T.; funding acquisition, L.Z. and G.K. Declaration of interests L.Z. is the main founder and the president of the everImmune scientific advisory board and had a research contract with everImmune. A.H.R. received research support from Bristol-Myers Squibb (on behalf of the METIMMOX sponsor Akershus University Hospital). A.E. declares research support from GMT Science, Merck, Astrazeneca, and BMS; declares consulting fees from Merck, Astrazeneca, BMS, and everImmune; and honoraria/speaking fees from Merck, Astrazeneca, BMS, and EMD Soreno. B.R. is a co-founder of Curebiota. A.E., B.R., and S.M.V. received support from Kanvas Bioscience. O.K. is a co-founder of Samsara Therapeutics. C.A.C.S. has received honoraria from JNJ, BMS, Ipsen, and AstraZeneca. S.M.V. is a co-founder of LND Therapeutics Inc. and reports grants from Microviable Therapeutics and mbiomics. C.Cremolini holds a consulting or advisory role with Roche, MSD, Amgen, Pierre Fabre, Nordic Pharma, Takeda, Lilly, AstraZeneca, Revolution Medicine, and Rottapharm. Speaker’s Bureau: Bayer, Servier, and Merck Serono. Coordinating PI: Roche, Bayer, Servier, Merck, Tempus Local PI: Seagen, Hutchinson, GSK, Amgen, Pfizer, Revolution Medicine. C.Ciccarese declares advisory board, invited speaker, and meetings traveling from MSD, Eisai, J&J, Recordati, AstraZeneca, Novartis, Astellas, Pfizer, and BMS. R.I. served as an advisory board member or consultant for Astellas, AstraZeneca, Pfizer, Janssen, Sanofi, Ipsen, MSD, BMS, Novartis, GSK, and EISAI and received institutional support for research project form Pfizer and BMS. STAR★Methods Key resources table REAGENT or RESOURCESOURCEIDENTIFIERAntibodiesAlexa Fluor(R) 700 anti-mouse CD45BiolegendCat# 103128; RRID:AB_493715PerCP anti-mouse CD45BiolegendCat# 103130; RRID:AB_893339CD3ε Antibody, anti-mouse, APC-Vio® 770, REAfinity™Miltenyi BiotecCat# 130-117-676; RRID:AB_2733115CD4 Antibody, anti-mouse, VioBlue®, REAfinity™Miltenyi BiotecCat# 130-118-568; RRID:AB_2733943PE/Dazzle(TM) 594 anti-mouse CD8aBioLegendCat# 100762; RRID:AB_2564027Brilliant Violet 650(TM) anti-mouse CD25BioLegendCat# 102038; RRID:AB_2563060ROR gamma (t) Monoclonal Antibody (B2D), PE-Cyanine7, eBioscienceThermo Fisher ScientificCat# 25-6981-82; RRID:AB_2784671FOXP3 Monoclonal Antibody (FJK-16s), APC, eBioscienceThermo Fisher ScientificCat# 17-5773-82; RRID:AB_469457APC/Cyanine7 anti-mouse CD3BiolegendCat# 100222; RRID:AB_2242784Brilliant Violet 605(TM) anti-mouse CD4BioLegendCat# 116027; RRID:AB_2800581FITC anti-mouse CD279 (PD-1)BioLegendCat# 135214; RRID:AB_10680238Brilliant Violet 421(TM) anti-mouse CD186 (CXCR6)BioLegendCat# 151109; RRID:AB_2616760Brilliant Violet 650(TM) anti-mouse CD183 (CXCR3)BioLegendCat# 126531; RRID:AB_2563160TIGIT Monoclonal Antibody (GIGD7), PerCP-eFluor™ 710, eBioscienceThermo Fisher ScientificCat# 46-9501-82; RRID:AB_11150967TOX Antibody, anti-human/mouse, APC, REAfinity™Miltenyi BiotecCat# 130-118-335; RRID:AB_2751485PE anti-mouse TCF-1 (C63D9)Cell signaling TechnologyCat# 14456; RRID:AB_2798483PE anti-TCF1 (TCF7)BioLegendCat# 655208; RRID:AB_2728492Brilliant Violet 421(TM) anti-mouse CD223 (LAG-3)BioLegendCat# 125221; RRID:AB_2572080Brilliant Violet 650(TM) anti-mouse/human CD44BioLegendCat# 103049; RRID:AB_2562600CD137 (4-1BB) Monoclonal Antibody (17B5), APC, eBioscienceThermo Fisher ScientificCat# 17-1371-82; RRID:AB_2573162Alexa Fluor(R) 700 anti-mouse CD62LBioLegendCat# 104426; RRID:AB_493719CD278 (ICOS) Monoclonal Antibody (C398.4A), PE-Cyanine7, eBioscienceThermo Fisher ScientificCat# 25-9949-82; RRID:AB_2573566FITC anti-mouse Ki-67BioLegendCat# 652410; RRID:AB_2562141Brilliant Violet 421(TM) anti-human CD197 (CCR7)BioLegendCat# 353208; RRID:AB_11203894Human Integrin alpha 4 beta 7/LPAM-1 (Research Grade Vedolizumab Biosimilar) Alexa Fluor® 647-conjugated AntibodyR and D SystemsCat# FAB10078R; RRID:AB_3644725PE/Dazzle(TM) 594 anti-human CD274 (B7-H1, PD-L1)BioLegendCat# 329732; RRID:AB_2616889Hu CD273 RB780 MIH18 50 μgBD BiosciencesCat# 755831; RRID:AB_3688138CD194 (CCR4)BD BiosciencesCat# 562906; RRID:AB_2737882Hu CD196 (CCR6) BUV661 11A9 50 μgBD BiosciencesCat# 750696; RRID:AB_2874817PE/Fire™ 640 anti-human CD103 (Integrin αE)BioLegendCat# 350243; RRID:AB_2924550Hu CD1c BUV737 F10/21A3 50 μgBD BiosciencesCat# 748723; RRID:AB_2873127PE/Fire(TM) 810 anti-human CD183 (CXCR3)BioLegendCat# 353759; RRID:AB_2894484Hu CXCR5 BUV615 RF8B2 50 μgBD BiosciencesCat# 751293; RRID:AB_2875306Hu Fc Epsl R1 Alp BV480 AER-37 50 μgBD BiosciencesCat# 747779; RRID:AB_2872243Spark Red(TM) 718 anti-human/mouse/rat CD278 (ICOS)BioLegendCat# 313563; RRID:AB_2936693IgA Antibody, anti-human, APC-Vio® 770, REAfinity™Miltenyi BiotecCat# 130-116-884; RRID:AB_2727742IgG BV480BD BiosciencesCat# 746341; RRID:AB_2743662Brilliant Violet 510(TM) anti-human IgMBioLegendCat# 314522; RRID:AB_2562916Brilliant Violet 750(TM) anti-human CD279 (PD-1)BioLegendCat# 329965; RRID:AB_2810504Spark Violet(TM) 423 anti-human CD123BioLegendCat# 396719; RRID:AB_2910457CD148 BUV395BD BiosciencesCat# 745674; RRID:AB_2743163Brilliant Violet 570(TM) anti-human CD16BioLegendCat# 302035; RRID:AB_10915988Hu CD21 BUV805 B-ly4 50 μgBD BiosciencesCat# 742008; RRID:AB_2871306CD24 (Heat Stable Antigen)BD BiosciencesCat# 564521; RRID:AB_2738834PE/Fire(TM) 700 anti-human CD25BioLegendCat# 356145; RRID:AB_2876678Brilliant Violet 650(TM) anti-human CD3BioLegendCat# 300468; RRID:AB_2629574Brilliant Violet 711(TM) anti-human CD38BioLegendCat# 303528; RRID:AB_2563811Brilliant Violet 650(TM) anti-human CD40BioLegendCat# 334338; RRID:AB_2566209Spark Blue(TM) 574 anti-human CD45BioLegendCat# 368557; RRID:AB_2904402Hu CD5 BUV563 UCHT2 50 μgBD BiosciencesCat# 741354; RRID:AB_2870855Hu CD71 BUV496 M-A712 50 μgBD BiosciencesCat# 750652; RRID:AB_2874780CD88 (C5a Receptor)BD BiosciencesCat# 742320; RRID:AB_2740700Cytek® 25-Color Immunoprofiling Assay, cFluor®Reagent Kit (18C)Cat# R7-40002IgD anti-human cFluor® BYG667Cytek BioscienceCat# R7-40002CD11c anti-human cFluor® BYG781Cytek BioscienceCat# R7-40002CD14 anti-human cFluor® B548Cytek BioscienceCat# R7-40002CD19 anti-human cFluor® R685Cytek BioscienceCat# R7-40002CD4 anti-human cFluor® YG584Cytek BioscienceCat# R7-40002CD45RA anti-human cFluor® V450Cytek BioscienceCat# R7-40002CD8 anti-human cFluor® B532Cytek BioscienceCat# R7-40002HLA-DR anti-human cFluor® B690Cytek BioscienceCat# R7-40002CD127 anti-human cFluor® R659Cytek BioscienceCat# R7-20078CD27 anti-human cFluor® R840Cytek BioscienceCat# R7-20082TruStain FcX(TM) PLUS (anti-mouse CD16/32)BiolegendCat# 156604; RRID:AB_2783138Pacific Blue(TM) anti-mouse CD4BioLegendCat# 100531; RRID:AB_493374Brilliant Violet 650(TM) anti-mouse CD8aBioLegendCat# 100742; RRID:AB_2563056FITC anti-human/mouse Granzyme BBioLegendCat# 515403; RRID:AB_2114575PE/Cyanine7 anti-mouse IFN-γBioLegendCat# 505826; RRID:AB_2295770Ki-67 Monoclonal Antibody (SolA15), APC, eBioscienceThermo Fisher ScientificCat# 17-5698-82; RRID:AB_2688057InVivoMAb anti-mouse PD-1 (CD279)BioXCellCat# BE0146; RRID:AB_10949053InVivoMAb rat IgG2a isotype control, anti-trinitrophenolBioXCellCat# BE0089; RRID:AB_1107769InVivoMAb anti-mouse CTLA-4 (CD152)BioXCellCatalog #BE0164; RRID:AB_10949609InVivoMAb mouse IgG2b isotype control, unknown specificityBioXCellCat# BE0086; RRID:AB_1107791Anti-mouse CD16/CD32eBioscienceCat# 14-0161-85; RRID:AB_467134TruStain FcX™ PLUS (anti-mouse CD16/32) antibodyBiolegendCat# 156604; RRID:AB_2783137Bacterial and virus strainsEnterocloster clostridioformisThis labFidelle, 2023EB BAA613 - Enterocloster bolteaeATCCBAA-613EC 25537 - Enterocloster clostridioformisATCC25537Chemicals, peptides, and recombinant proteinsZombie Aqua™ Fixable Viability KitBiolegendCat# 423102LIVE/DEAD™ Fixable Aqua dyeThermo Fisher ScientificCat# L34966CellTrace™ CFSE Cell Proliferation Kit, for flow cytometryThermo Fisher ScientificCat# C34554LIVE/DEAD™ Fixable Blue Dead Cell StainInvitrogen™Cat# L23105BD Horizon™ Brilliant Stain Buffer PlusBD BiosciencesCat# 566385PE StreptavidinBiolegendCat# 405203Anti-SynCAM (TSLC1/CADM1) mAb-BiotinMBLCat# CM004-6Cholic acid, CAAvantiCat# 700212PChenodeoxycholic acid, CDCASigma-AldrichCat# C8261α-muricholic acid, α-MCAAvantiCat# 700232Pβ-muricholic acid, β-MCAAvantiCat# 700233PDeoxycholic acid, DCASigma-Aldrich/Thermo Fisher ScientificCat# 30960/148140250Hyocholic acid, HCAAvantiCat# 700159PHyodeoxycholic acid, HDCAAvantiCat# 700214P12-dehydrocholic acid, 12-dDCAAvantiCat# 700215PLithocholic acid, LCAThermo Scientific ChemicalsCat# 229090050Ursodeoxycholic acid, UDCAThermo Scientific ChemicalsCat# 208590250Tyrosine-conjugated cholic acid, Tyro-CAGenerously provided by Kenya Honda, Keio University School of Medicine, Tokyo, JapanNAGlycocholic acid, GCAAvantiCat# 700265PGlycodeoxycholic acid, GDCASigma-AldrichCat# G9910Glycohyocholic acid, GHCAMedChemExpressCat# HY-122141Glycohyodeoxycholic acid, GHDCAAvantiCat# 700264PTaurocholic acid sodium salt hydrate, TCASigma-AldrichCat# T4009Sodium taurochenodeoxycholate, TCDCASigma-AldrichCat# T6260Taurodeoxycholic acid sodium salt, TDCAMilliporeCat# 580221Taurohyodeoxycholic acid sodium salt, THDCAAvantiCat# 700248PTaurolithocholic acid sodium salt, TLCAAvantiCat# 700252PTauroursodeoxycholic acid sodium salt, TUDCAMilliporeCat# 580549CilofexorMedChemExpressCat# HY-109083Murideoxycholic acid, MDCAAvantiCat# 700270Pω-muricholic acid, ω-MCAAvantiCat# 700231PTauro-α-muricholic acid sodium salt, T-α-MCAAvantiCat# 700243PTauro-β-muricholic acid sodium salt, T-β-MCAAvantiCat# 700244PTauro-ω-muricholic acid sodium salt, T-ω-MCAAvantiCat# 700245Tauro-γ-muricholic acid sodium salt, THCAAvantiCat# 700246PRecombinant mouse IL-2BioLegendCat# 575406AmpicillinSigma-AldrichCat# A9518StreptomycinSigma-AldrichCat# A9137ColistinSigma-AldrichCat# C4461FexaramineMedChemExpressCat# 574013-66-4LiberaseRocheCat#05 401 119 001DNase1RocheCat# 04 716 728 001Recombinant murine TNF-αPeproTechCat# 315-01AMitoxantroneSigma-AldrichCat# M6545OxaliplatineAccordNAHoechst 33342Thermo ScientificCat# R37165Propidium iodideSigma-AldrichCat# P4864Critical commercial assayseBioscience™ Foxp3/Transcription Factor Staining Buffer SetThermo Fisher ScientificCat# 00-5523-00MojoSort™ Mouse CD3 T cell Isolation KitBioLegendCat# 480031PureLink™ Genomic DNA Mini KitThermo Fisher ScientificCat# K182002Platinum™ SuperFi II PCR Master MixThermo Fisher ScientificCat# 12368010PureLink™ PCR Micro KitThermo Fisher ScientificCat# K310250CD45 microbeadsMiltenyiCat# 130-052-30BD™ Mouse Single Cell Sample Multiplexing KitBDCat# 633793BD Rhapsody™ Whole Transcriptome Analysis (WTA) Amplification KitBDCat# 633801BD Rhapsody™ Whole Transcriptome Analysis (WTA) Reagent KitBDCat# 665915Lamina Propria Dissociation kitMiltenyiCat# 130-097-410CD326 (EpCAM) MicroBeadsMiltenyiCat# 130-105-958Liver Perfusion kitMiltenyiCat# 130-128-030RNAeasy Mini kitQiagenCat# 74106Dynabeads® Magnetic BeadsGibcoCat# 11452DeBioscience™ Cell Stimulation CocktailThermo Fisher ScientificCat# 00-4975-93Deposited dataThe bulk RNA sequencing of murine hepatocytes, cholangiocytes, and enterocytes from the murine experimentThis paperENA (European Nucleotide Archive): PRJEB10992616S rRNA sequencing data of murine fecal samples from the murine experimentThis paperENA (European Nucleotide Archive) : PRJEB109926Single cell RNA sequencing of immune cells from tumor and tumor draining lymph nodes from the murine experimentThis paperGEO: GSE326943Metabolomic data in mouse from the murine experimentsThis paperMendeley Data: https://doi.org/10.17632/wsydn677rw.1Metabolomic data in human from the ONCOBIOTICS, MIMIC and FMT-LUMINATE trialsThis paperMendeley Data: https://doi.org/10.17632/wsydn677rw.1Experimental models: Cell linesMCA205 fibrosarcomaAmerican Type Culture CollectionN/ARET melanomaProvided by Viktor Umansky (DKFZ—German Cancer Research CenterN/ATransformed sinusoidal endothelial cells (TSECs)Provided by B.P. Fennimore (University of Colorado Anschutz Medical Campus, Aurora, CO, USA)N/ATSECs with knockout of three bile acid receptors-TGR5 (gene: Gbpar1), FXR (gene: Nr1h4), and VDR (gene: Vdr)This paperN/AHT29ATCCHTB-38OligonucleotidesNr1h4-/- detection primer AEurofinsATGGGCCTTTCAGAATGCCGNr1h4-/- detection primer BEurofinsTCAAGGCTCATCCGGGTGATGpbar1-/- detection primer AEurofinsACCATCAGGGCTACTGGTCCGpbar1-/- detection primer BEurofinsTGCTGCATTGGCTACTGGTGVdr-/- detection primer AEurofinsGCCAGAACCTCAGCTTGGTCVdr-/- detection primer BEurofinsCATGATAAGGCATCGGCGCTSoftware and algorithmsStatistical environment Rhttp://www.R-project.org/N/APrism 10GraphPad, San Diego, CA, USAN/AOtherEdit-R All-in-one Lentiviral (knockout) Mouse Nr1h4 sgRNAHorizon DiscoveryCat# catalog no. VSGM11942-247929487Target sequence: AGTGCCAGGAGTGCCGGCTAEdit-R All-in-one Lentiviral (knockout) Mouse Gpbar1 sgRNAN/ACat# VSGM11942-247972009Target sequence: ATGAGCGTCGCCCACCACTAEdit-R All-in-one Lentiviral (knockout) Mouse Vdr sgRNAN/ACat# VSGM11942-248142193Target sequence: TTCGTGCAGACGTAAGTACA Experimental model and study participant details Human Study participant details Data were derived from eight prospective clinical trials, comprising both observational and interventional design. 1.ONCOBIOTICS◦ClinicalTrials.gov ID: NCT04567446◦Type: Observational design◦Description: ONCOBIOTICS is a multicenter observational study conducted across 14 sites in France and Canada. It investigates the impact of the microbiome on clinical outcomes in patients with advanced non-small cell lung cancer (NSCLC), renal cell carcinoma (RCC), and bladder cancer (BC) treated with anti-PD-(L)1 therapies, either alone or combined with chemotherapy or tyrosine kinase inhibitors. Patients were followed according to standard care until disease progression, intolerable toxicity, or up to two years of immune checkpoint inhibitor (ICI) treatment. Detailed eligibility criteria, baseline data, and laboratory tests including baseline gamma-glutamyl transferase (GGT), are provided in the trial protocol and recorded via electronic case report forms. All participants provided written informed consent. Blood and stool samples were collected at baseline. Available samples were used in this analysis, with availability varying based on patient consent, sample pairing, volume constraints, and site-level data reporting. Blood samples were used for metabolomics analysis, specifically for TDCA quantification in the MVA (multivariate analysis), following Alves Costa Silva et al., 2024.54 Stool samples were used to calculate the Toposcore via metagenomic analysis, following the methodology described in Derosa et al., Cell, 2024.36 ONCOBIOTICS◦ClinicalTrials.gov ID: NCT04567446◦Type: Observational design◦Description: ONCOBIOTICS is a multicenter observational study conducted across 14 sites in France and Canada. It investigates the impact of the microbiome on clinical outcomes in patients with advanced non-small cell lung cancer (NSCLC), renal cell carcinoma (RCC), and bladder cancer (BC) treated with anti-PD-(L)1 therapies, either alone or combined with chemotherapy or tyrosine kinase inhibitors. Patients were followed according to standard care until disease progression, intolerable toxicity, or up to two years of immune checkpoint inhibitor (ICI) treatment. Detailed eligibility criteria, baseline data, and laboratory tests including baseline gamma-glutamyl transferase (GGT), are provided in the trial protocol and recorded via electronic case report forms. All participants provided written informed consent. Blood and stool samples were collected at baseline. Available samples were used in this analysis, with availability varying based on patient consent, sample pairing, volume constraints, and site-level data reporting. Blood samples were used for metabolomics analysis, specifically for TDCA quantification in the MVA (multivariate analysis), following Alves Costa Silva et al., 2024.54 Stool samples were used to calculate the Toposcore via metagenomic analysis, following the methodology described in Derosa et al., Cell, 2024.36 ◦ClinicalTrials.gov ID: NCT04567446 ClinicalTrials.gov ID: NCT04567446 ◦Type: Observational design Type: Observational design ◦Description: ONCOBIOTICS is a multicenter observational study conducted across 14 sites in France and Canada. It investigates the impact of the microbiome on clinical outcomes in patients with advanced non-small cell lung cancer (NSCLC), renal cell carcinoma (RCC), and bladder cancer (BC) treated with anti-PD-(L)1 therapies, either alone or combined with chemotherapy or tyrosine kinase inhibitors. Patients were followed according to standard care until disease progression, intolerable toxicity, or up to two years of immune checkpoint inhibitor (ICI) treatment. Detailed eligibility criteria, baseline data, and laboratory tests including baseline gamma-glutamyl transferase (GGT), are provided in the trial protocol and recorded via electronic case report forms. All participants provided written informed consent. Blood and stool samples were collected at baseline. Available samples were used in this analysis, with availability varying based on patient consent, sample pairing, volume constraints, and site-level data reporting. Blood samples were used for metabolomics analysis, specifically for TDCA quantification in the MVA (multivariate analysis), following Alves Costa Silva et al., 2024.54 Stool samples were used to calculate the Toposcore via metagenomic analysis, following the methodology described in Derosa et al., Cell, 2024.36 Description: ONCOBIOTICS is a multicenter observational study conducted across 14 sites in France and Canada. It investigates the impact of the microbiome on clinical outcomes in patients with advanced non-small cell lung cancer (NSCLC), renal cell carcinoma (RCC), and bladder cancer (BC) treated with anti-PD-(L)1 therapies, either alone or combined with chemotherapy or tyrosine kinase inhibitors. Patients were followed according to standard care until disease progression, intolerable toxicity, or up to two years of immune checkpoint inhibitor (ICI) treatment. Detailed eligibility criteria, baseline data, and laboratory tests including baseline gamma-glutamyl transferase (GGT), are provided in the trial protocol and recorded via electronic case report forms. All participants provided written informed consent. Blood and stool samples were collected at baseline. Available samples were used in this analysis, with availability varying based on patient consent, sample pairing, volume constraints, and site-level data reporting. Blood samples were used for metabolomics analysis, specifically for TDCA quantification in the MVA (multivariate analysis), following Alves Costa Silva et al., 2024.54 Stool samples were used to calculate the Toposcore via metagenomic analysis, following the methodology described in Derosa et al., Cell, 2024.36 2.MEDITREME◦ClinicalTrials.gov ID: NCT03202758◦Type: Interventional design◦Description: Clinical data were obtained from the phase II clinical trial NCT03202758, which enrolled 54 patients with histologically confirmed metastatic colorectal cancer harboring a KRAS mutation. All patients received a combination treatment consisting of Durvalumab (750 mg every 2 weeks), Tremelimumab (75 mg every 4 weeks), and FOLFOX chemotherapy. Eligible patients had an Eastern Cooperative Oncology Group (ECOG) performance status of 0 or 1. All participants provided written informed consent and any locally required authorizations prior to undergoing protocol-related procedures, including screening assessments. Eligibility criteria, baseline characteristics, and concomitant medications were recorded in eCRFs and are detailed in the trial protocol available on ClinicalTrials.gov For this study, we used de-identified clinical data collected from the trial, focusing in particular on GGT levels. These data were used for Kaplan–Meier survival analyses and multivariable analyses (MVA) to explore associations between GGT levels and clinical outcomes. MEDITREME◦ClinicalTrials.gov ID: NCT03202758◦Type: Interventional design◦Description: Clinical data were obtained from the phase II clinical trial NCT03202758, which enrolled 54 patients with histologically confirmed metastatic colorectal cancer harboring a KRAS mutation. All patients received a combination treatment consisting of Durvalumab (750 mg every 2 weeks), Tremelimumab (75 mg every 4 weeks), and FOLFOX chemotherapy. Eligible patients had an Eastern Cooperative Oncology Group (ECOG) performance status of 0 or 1. All participants provided written informed consent and any locally required authorizations prior to undergoing protocol-related procedures, including screening assessments. Eligibility criteria, baseline characteristics, and concomitant medications were recorded in eCRFs and are detailed in the trial protocol available on ClinicalTrials.gov For this study, we used de-identified clinical data collected from the trial, focusing in particular on GGT levels. These data were used for Kaplan–Meier survival analyses and multivariable analyses (MVA) to explore associations between GGT levels and clinical outcomes. ◦ClinicalTrials.gov ID: NCT03202758 ClinicalTrials.gov ID: NCT03202758 ◦Type: Interventional design Type: Interventional design ◦Description: Clinical data were obtained from the phase II clinical trial NCT03202758, which enrolled 54 patients with histologically confirmed metastatic colorectal cancer harboring a KRAS mutation. All patients received a combination treatment consisting of Durvalumab (750 mg every 2 weeks), Tremelimumab (75 mg every 4 weeks), and FOLFOX chemotherapy. Eligible patients had an Eastern Cooperative Oncology Group (ECOG) performance status of 0 or 1. All participants provided written informed consent and any locally required authorizations prior to undergoing protocol-related procedures, including screening assessments. Eligibility criteria, baseline characteristics, and concomitant medications were recorded in eCRFs and are detailed in the trial protocol available on ClinicalTrials.gov For this study, we used de-identified clinical data collected from the trial, focusing in particular on GGT levels. These data were used for Kaplan–Meier survival analyses and multivariable analyses (MVA) to explore associations between GGT levels and clinical outcomes. Description: Clinical data were obtained from the phase II clinical trial NCT03202758, which enrolled 54 patients with histologically confirmed metastatic colorectal cancer harboring a KRAS mutation. All patients received a combination treatment consisting of Durvalumab (750 mg every 2 weeks), Tremelimumab (75 mg every 4 weeks), and FOLFOX chemotherapy. Eligible patients had an Eastern Cooperative Oncology Group (ECOG) performance status of 0 or 1. All participants provided written informed consent and any locally required authorizations prior to undergoing protocol-related procedures, including screening assessments. Eligibility criteria, baseline characteristics, and concomitant medications were recorded in eCRFs and are detailed in the trial protocol available on ClinicalTrials.gov For this study, we used de-identified clinical data collected from the trial, focusing in particular on GGT levels. These data were used for Kaplan–Meier survival analyses and multivariable analyses (MVA) to explore associations between GGT levels and clinical outcomes. 3.METIMMOX◦ClinicalTrials.gov ID: NCT03388190◦Type: Interventional design◦Decription: Clinical data were obtained from the NCT03388190 study, a multicenter, open-label, randomized phase II trial evaluating first-line treatment in patients with microsatellite stable (MSS)/mismatch repair-proficient (pMMR) metastatic colorectal cancer. Patients were randomized to receive either the oxaliplatin-based standard-of-care Nordic FLOX regimen (control arm) or a sequential treatment with Nordic FLOX followed by nivolumab (experimental arm). The primary objective was to determine whether the experimental arm improved progression-free survival (PFS), with secondary endpoints including safety, tolerability, and quality of life. The trial was approved and conducted as per Norwegian legislation, and all participants provided written informed consent. Eligibility criteria, baseline characteristics, and concomitant medications were documented in eCRFs, with full criteria available in the trial protocol on ClinicalTrials.gov. For the purpose of this study, we used de-identified clinical data from this trial, focusing particularly on the ALP variable for Kaplan–Meier survival analyses. METIMMOX◦ClinicalTrials.gov ID: NCT03388190◦Type: Interventional design◦Decription: Clinical data were obtained from the NCT03388190 study, a multicenter, open-label, randomized phase II trial evaluating first-line treatment in patients with microsatellite stable (MSS)/mismatch repair-proficient (pMMR) metastatic colorectal cancer. Patients were randomized to receive either the oxaliplatin-based standard-of-care Nordic FLOX regimen (control arm) or a sequential treatment with Nordic FLOX followed by nivolumab (experimental arm). The primary objective was to determine whether the experimental arm improved progression-free survival (PFS), with secondary endpoints including safety, tolerability, and quality of life. The trial was approved and conducted as per Norwegian legislation, and all participants provided written informed consent. Eligibility criteria, baseline characteristics, and concomitant medications were documented in eCRFs, with full criteria available in the trial protocol on ClinicalTrials.gov. For the purpose of this study, we used de-identified clinical data from this trial, focusing particularly on the ALP variable for Kaplan–Meier survival analyses. ◦ClinicalTrials.gov ID: NCT03388190 ClinicalTrials.gov ID: NCT03388190 ◦Type: Interventional design Type: Interventional design ◦Decription: Clinical data were obtained from the NCT03388190 study, a multicenter, open-label, randomized phase II trial evaluating first-line treatment in patients with microsatellite stable (MSS)/mismatch repair-proficient (pMMR) metastatic colorectal cancer. Patients were randomized to receive either the oxaliplatin-based standard-of-care Nordic FLOX regimen (control arm) or a sequential treatment with Nordic FLOX followed by nivolumab (experimental arm). The primary objective was to determine whether the experimental arm improved progression-free survival (PFS), with secondary endpoints including safety, tolerability, and quality of life. The trial was approved and conducted as per Norwegian legislation, and all participants provided written informed consent. Eligibility criteria, baseline characteristics, and concomitant medications were documented in eCRFs, with full criteria available in the trial protocol on ClinicalTrials.gov. For the purpose of this study, we used de-identified clinical data from this trial, focusing particularly on the ALP variable for Kaplan–Meier survival analyses. Decription: Clinical data were obtained from the NCT03388190 study, a multicenter, open-label, randomized phase II trial evaluating first-line treatment in patients with microsatellite stable (MSS)/mismatch repair-proficient (pMMR) metastatic colorectal cancer. Patients were randomized to receive either the oxaliplatin-based standard-of-care Nordic FLOX regimen (control arm) or a sequential treatment with Nordic FLOX followed by nivolumab (experimental arm). The primary objective was to determine whether the experimental arm improved progression-free survival (PFS), with secondary endpoints including safety, tolerability, and quality of life. The trial was approved and conducted as per Norwegian legislation, and all participants provided written informed consent. Eligibility criteria, baseline characteristics, and concomitant medications were documented in eCRFs, with full criteria available in the trial protocol on ClinicalTrials.gov. For the purpose of this study, we used de-identified clinical data from this trial, focusing particularly on the ALP variable for Kaplan–Meier survival analyses. 4.G2R◦ClinicalTrials.gov ID: NCT04169841◦Type: Interventional design◦Description: The NCT04169841 phase II Study Evaluate the Efficacy of a Double Immunotherapy by Durvalumab and Tremelimumab Combined With Olaparib in Patients With Solid Cancers and Carriers of Homologous Recombination Repair Genes Mutation in Response or Stable After Olaparib Treatment. Eligibility criteria and baseline data, including recent medications, are in the trial protocol and recorded in electronic case reports. Full eligibility criteria are listed in the trial protocol (available at www.clinicaltrials.gov). All patients must give signed informed consent For this study, we used de-identified clinical data collected from the trial, focusing in particular on GGT levels. These data were used for Kaplan–Meier survival analyses and multivariable analyses (MVA) to explore associations between GGT levels and clinical outcomes. G2R◦ClinicalTrials.gov ID: NCT04169841◦Type: Interventional design◦Description: The NCT04169841 phase II Study Evaluate the Efficacy of a Double Immunotherapy by Durvalumab and Tremelimumab Combined With Olaparib in Patients With Solid Cancers and Carriers of Homologous Recombination Repair Genes Mutation in Response or Stable After Olaparib Treatment. Eligibility criteria and baseline data, including recent medications, are in the trial protocol and recorded in electronic case reports. Full eligibility criteria are listed in the trial protocol (available at www.clinicaltrials.gov). All patients must give signed informed consent For this study, we used de-identified clinical data collected from the trial, focusing in particular on GGT levels. These data were used for Kaplan–Meier survival analyses and multivariable analyses (MVA) to explore associations between GGT levels and clinical outcomes. ◦ClinicalTrials.gov ID: NCT04169841 ClinicalTrials.gov ID: NCT04169841 ◦Type: Interventional design Type: Interventional design ◦Description: The NCT04169841 phase II Study Evaluate the Efficacy of a Double Immunotherapy by Durvalumab and Tremelimumab Combined With Olaparib in Patients With Solid Cancers and Carriers of Homologous Recombination Repair Genes Mutation in Response or Stable After Olaparib Treatment. Eligibility criteria and baseline data, including recent medications, are in the trial protocol and recorded in electronic case reports. Full eligibility criteria are listed in the trial protocol (available at www.clinicaltrials.gov). All patients must give signed informed consent For this study, we used de-identified clinical data collected from the trial, focusing in particular on GGT levels. These data were used for Kaplan–Meier survival analyses and multivariable analyses (MVA) to explore associations between GGT levels and clinical outcomes. Description: The NCT04169841 phase II Study Evaluate the Efficacy of a Double Immunotherapy by Durvalumab and Tremelimumab Combined With Olaparib in Patients With Solid Cancers and Carriers of Homologous Recombination Repair Genes Mutation in Response or Stable After Olaparib Treatment. Eligibility criteria and baseline data, including recent medications, are in the trial protocol and recorded in electronic case reports. Full eligibility criteria are listed in the trial protocol (available at www.clinicaltrials.gov). All patients must give signed informed consent For this study, we used de-identified clinical data collected from the trial, focusing in particular on GGT levels. These data were used for Kaplan–Meier survival analyses and multivariable analyses (MVA) to explore associations between GGT levels and clinical outcomes. 5.ATEZOTRIBE◦ClinicalTrials.gov: NCT03721653◦Type: Interventional design◦Decsription: AtezoTRIBE is a phase II trial enrolling 218 patients with unresectable stage IV colorectal cancer. Participants were randomized 2:1 to receive FOLFOXIRI plus bevacizumab with or without atezolizumab (anti–PD-L1). Eligibility criteria and baseline data, including recent medications, are in the trial protocol and recorded in electronic case reports. Full eligibility criteria are listed in the trial protocol (available at www.clinicaltrials.gov). All patients signed a written informed consent to study procedures For this study, we used de-identified clinical data collected from the trial, focusing in particular on GGT levels. These data were used for Kaplan–Meier survival analyses and MVA to explore associations between GGT levels and clinical outcomes. ATEZOTRIBE◦ClinicalTrials.gov: NCT03721653◦Type: Interventional design◦Decsription: AtezoTRIBE is a phase II trial enrolling 218 patients with unresectable stage IV colorectal cancer. Participants were randomized 2:1 to receive FOLFOXIRI plus bevacizumab with or without atezolizumab (anti–PD-L1). Eligibility criteria and baseline data, including recent medications, are in the trial protocol and recorded in electronic case reports. Full eligibility criteria are listed in the trial protocol (available at www.clinicaltrials.gov). All patients signed a written informed consent to study procedures For this study, we used de-identified clinical data collected from the trial, focusing in particular on GGT levels. These data were used for Kaplan–Meier survival analyses and MVA to explore associations between GGT levels and clinical outcomes. ◦ClinicalTrials.gov: NCT03721653 ClinicalTrials.gov: NCT03721653 ◦Type: Interventional design Type: Interventional design ◦Decsription: AtezoTRIBE is a phase II trial enrolling 218 patients with unresectable stage IV colorectal cancer. Participants were randomized 2:1 to receive FOLFOXIRI plus bevacizumab with or without atezolizumab (anti–PD-L1). Eligibility criteria and baseline data, including recent medications, are in the trial protocol and recorded in electronic case reports. Full eligibility criteria are listed in the trial protocol (available at www.clinicaltrials.gov). All patients signed a written informed consent to study procedures For this study, we used de-identified clinical data collected from the trial, focusing in particular on GGT levels. These data were used for Kaplan–Meier survival analyses and MVA to explore associations between GGT levels and clinical outcomes. Decsription: AtezoTRIBE is a phase II trial enrolling 218 patients with unresectable stage IV colorectal cancer. Participants were randomized 2:1 to receive FOLFOXIRI plus bevacizumab with or without atezolizumab (anti–PD-L1). Eligibility criteria and baseline data, including recent medications, are in the trial protocol and recorded in electronic case reports. Full eligibility criteria are listed in the trial protocol (available at www.clinicaltrials.gov). All patients signed a written informed consent to study procedures For this study, we used de-identified clinical data collected from the trial, focusing in particular on GGT levels. These data were used for Kaplan–Meier survival analyses and MVA to explore associations between GGT levels and clinical outcomes. 6.PREMIS◦ClinicalTrials.gov: NCT03984318◦Type: Observational design◦Description: Patients enrolled in PREMIS were ≥18 years of age, with histologically proven solid malignancy and at least one tumor evaluation by imaging after immunotherapy onset. Eligibility criteria and baseline data, including recent medications, are in the trial protocol and recorded in electronic case reports. Full eligibility criteria are listed in the trial protocol (available at www.clinicaltrials.gov). Patient should understand, sign, and date the written informed consent form prior to any protocol-specific procedures performed. For this study, we used de-identified clinical data collected from the trial, focusing in particular on GGT levels. These data were used for Kaplan–Meier survival analyses and MVA to explore associations between GGT levels and clinical outcomes. PREMIS◦ClinicalTrials.gov: NCT03984318◦Type: Observational design◦Description: Patients enrolled in PREMIS were ≥18 years of age, with histologically proven solid malignancy and at least one tumor evaluation by imaging after immunotherapy onset. Eligibility criteria and baseline data, including recent medications, are in the trial protocol and recorded in electronic case reports. Full eligibility criteria are listed in the trial protocol (available at www.clinicaltrials.gov). Patient should understand, sign, and date the written informed consent form prior to any protocol-specific procedures performed. For this study, we used de-identified clinical data collected from the trial, focusing in particular on GGT levels. These data were used for Kaplan–Meier survival analyses and MVA to explore associations between GGT levels and clinical outcomes. ◦ClinicalTrials.gov: NCT03984318 ClinicalTrials.gov: NCT03984318 ◦Type: Observational design Type: Observational design ◦Description: Patients enrolled in PREMIS were ≥18 years of age, with histologically proven solid malignancy and at least one tumor evaluation by imaging after immunotherapy onset. Eligibility criteria and baseline data, including recent medications, are in the trial protocol and recorded in electronic case reports. Full eligibility criteria are listed in the trial protocol (available at www.clinicaltrials.gov). Patient should understand, sign, and date the written informed consent form prior to any protocol-specific procedures performed. For this study, we used de-identified clinical data collected from the trial, focusing in particular on GGT levels. These data were used for Kaplan–Meier survival analyses and MVA to explore associations between GGT levels and clinical outcomes. Description: Patients enrolled in PREMIS were ≥18 years of age, with histologically proven solid malignancy and at least one tumor evaluation by imaging after immunotherapy onset. Eligibility criteria and baseline data, including recent medications, are in the trial protocol and recorded in electronic case reports. Full eligibility criteria are listed in the trial protocol (available at www.clinicaltrials.gov). Patient should understand, sign, and date the written informed consent form prior to any protocol-specific procedures performed. For this study, we used de-identified clinical data collected from the trial, focusing in particular on GGT levels. These data were used for Kaplan–Meier survival analyses and MVA to explore associations between GGT levels and clinical outcomes. 7.MIMic◦ClinicalTrials.gov ID: NCT03772899◦Type: Interventional design◦Description: The NCT03772899 study is a phase I open-label single-arm trial for 20 patients with a confirmed diagnosis of unresectable or metastatic cutaneous melanoma (BRAF wild type or mutant) treated with an association of ICB (pembrolizumab or nivolumab) and FMT from healthy donors. The purpose of this study is to examine the safety of combining these two therapies in melanoma patients. The investigator will use fecal material from a healthy donor selected via our stringent protocol that is Health Canada approved. In addition to assessing the safety of the combination, the investigator will also study the effect of FMT on the immune system and microbial ecosystem of the gut. Full eligibility criteria are listed in the trial protocol (available at www.clinicaltrials.gov). Patients must be able to provide written informed consent prior to enrollment in the study. Blood samples were collected at baseline and used for metabolomics analysis. MIMic◦ClinicalTrials.gov ID: NCT03772899◦Type: Interventional design◦Description: The NCT03772899 study is a phase I open-label single-arm trial for 20 patients with a confirmed diagnosis of unresectable or metastatic cutaneous melanoma (BRAF wild type or mutant) treated with an association of ICB (pembrolizumab or nivolumab) and FMT from healthy donors. The purpose of this study is to examine the safety of combining these two therapies in melanoma patients. The investigator will use fecal material from a healthy donor selected via our stringent protocol that is Health Canada approved. In addition to assessing the safety of the combination, the investigator will also study the effect of FMT on the immune system and microbial ecosystem of the gut. Full eligibility criteria are listed in the trial protocol (available at www.clinicaltrials.gov). Patients must be able to provide written informed consent prior to enrollment in the study. Blood samples were collected at baseline and used for metabolomics analysis. ◦ClinicalTrials.gov ID: NCT03772899 ClinicalTrials.gov ID: NCT03772899 ◦Type: Interventional design Type: Interventional design ◦Description: The NCT03772899 study is a phase I open-label single-arm trial for 20 patients with a confirmed diagnosis of unresectable or metastatic cutaneous melanoma (BRAF wild type or mutant) treated with an association of ICB (pembrolizumab or nivolumab) and FMT from healthy donors. The purpose of this study is to examine the safety of combining these two therapies in melanoma patients. The investigator will use fecal material from a healthy donor selected via our stringent protocol that is Health Canada approved. In addition to assessing the safety of the combination, the investigator will also study the effect of FMT on the immune system and microbial ecosystem of the gut. Full eligibility criteria are listed in the trial protocol (available at www.clinicaltrials.gov). Patients must be able to provide written informed consent prior to enrollment in the study. Blood samples were collected at baseline and used for metabolomics analysis. Description: The NCT03772899 study is a phase I open-label single-arm trial for 20 patients with a confirmed diagnosis of unresectable or metastatic cutaneous melanoma (BRAF wild type or mutant) treated with an association of ICB (pembrolizumab or nivolumab) and FMT from healthy donors. The purpose of this study is to examine the safety of combining these two therapies in melanoma patients. The investigator will use fecal material from a healthy donor selected via our stringent protocol that is Health Canada approved. In addition to assessing the safety of the combination, the investigator will also study the effect of FMT on the immune system and microbial ecosystem of the gut. Full eligibility criteria are listed in the trial protocol (available at www.clinicaltrials.gov). Patients must be able to provide written informed consent prior to enrollment in the study. Blood samples were collected at baseline and used for metabolomics analysis. 8.FMT-LUMINATE◦ClinicalTrials.gov ID: NCT04951583◦Type: Interventional design◦Description: The NCT04951583 study is a multi-center single-arm, open-label Simon two-stage, phase II trial in 45 patients with metastatic or unresectable non-small cell lung cancer (NSCLC) and melanoma treated with ICB. Patients will receive standard-of-care ICB therapy in combination with fecal microbial transplantation (FMT). Eligibility criteria, baseline characteristics, and concomitant medications were documented in eCRFs, with full criteria available in the trial protocol on ClinicalTrials.gov. All patients signed an informed consent prior to enrollment in the study. Blood samples were collected at baseline and used for metabolomics analysis. FMT-LUMINATE◦ClinicalTrials.gov ID: NCT04951583◦Type: Interventional design◦Description: The NCT04951583 study is a multi-center single-arm, open-label Simon two-stage, phase II trial in 45 patients with metastatic or unresectable non-small cell lung cancer (NSCLC) and melanoma treated with ICB. Patients will receive standard-of-care ICB therapy in combination with fecal microbial transplantation (FMT). Eligibility criteria, baseline characteristics, and concomitant medications were documented in eCRFs, with full criteria available in the trial protocol on ClinicalTrials.gov. All patients signed an informed consent prior to enrollment in the study. Blood samples were collected at baseline and used for metabolomics analysis. ◦ClinicalTrials.gov ID: NCT04951583 ClinicalTrials.gov ID: NCT04951583 ◦Type: Interventional design Type: Interventional design ◦Description: The NCT04951583 study is a multi-center single-arm, open-label Simon two-stage, phase II trial in 45 patients with metastatic or unresectable non-small cell lung cancer (NSCLC) and melanoma treated with ICB. Patients will receive standard-of-care ICB therapy in combination with fecal microbial transplantation (FMT). Eligibility criteria, baseline characteristics, and concomitant medications were documented in eCRFs, with full criteria available in the trial protocol on ClinicalTrials.gov. All patients signed an informed consent prior to enrollment in the study. Blood samples were collected at baseline and used for metabolomics analysis. Description: The NCT04951583 study is a multi-center single-arm, open-label Simon two-stage, phase II trial in 45 patients with metastatic or unresectable non-small cell lung cancer (NSCLC) and melanoma treated with ICB. Patients will receive standard-of-care ICB therapy in combination with fecal microbial transplantation (FMT). Eligibility criteria, baseline characteristics, and concomitant medications were documented in eCRFs, with full criteria available in the trial protocol on ClinicalTrials.gov. All patients signed an informed consent prior to enrollment in the study. Blood samples were collected at baseline and used for metabolomics analysis. 9.TACITO◦ClinicalTrials.gov ID: NCT04758507◦Type: Interventional design◦Decsription : The NCT04758507 trial is an investigator-initiated, randomized, double-blind, placebo-controlled phase 2a study. It aims to evaluate whether fecal microbiota transplantation (FMT) from donors who achieved a complete response to immune checkpoint inhibitors (ICB) can improve the efficacy of first-line therapy with pembrolizumab combined with axitinib in patients with metastatic renal cell carcinoma (mRCC). The trial was carried out in accordance with the principles of the Declaration of Helsinki and the International Conference on Harmonisation Good Clinical Practice (ICH-GCP) guidelines, while also complying with applicable local and institutional regulations. The study protocol received approval from the institutional review board/local ethics committee (ID: 2664). Written informed consent was obtained from all participants prior to enrollment in the study. Full eligibility criteria are listed in the trial protocol (available at www.clinicaltrials.gov). TACITO◦ClinicalTrials.gov ID: NCT04758507◦Type: Interventional design◦Decsription : The NCT04758507 trial is an investigator-initiated, randomized, double-blind, placebo-controlled phase 2a study. It aims to evaluate whether fecal microbiota transplantation (FMT) from donors who achieved a complete response to immune checkpoint inhibitors (ICB) can improve the efficacy of first-line therapy with pembrolizumab combined with axitinib in patients with metastatic renal cell carcinoma (mRCC). The trial was carried out in accordance with the principles of the Declaration of Helsinki and the International Conference on Harmonisation Good Clinical Practice (ICH-GCP) guidelines, while also complying with applicable local and institutional regulations. The study protocol received approval from the institutional review board/local ethics committee (ID: 2664). Written informed consent was obtained from all participants prior to enrollment in the study. Full eligibility criteria are listed in the trial protocol (available at www.clinicaltrials.gov). ◦ClinicalTrials.gov ID: NCT04758507 ClinicalTrials.gov ID: NCT04758507 ◦Type: Interventional design Type: Interventional design ◦Decsription : The NCT04758507 trial is an investigator-initiated, randomized, double-blind, placebo-controlled phase 2a study. It aims to evaluate whether fecal microbiota transplantation (FMT) from donors who achieved a complete response to immune checkpoint inhibitors (ICB) can improve the efficacy of first-line therapy with pembrolizumab combined with axitinib in patients with metastatic renal cell carcinoma (mRCC). The trial was carried out in accordance with the principles of the Declaration of Helsinki and the International Conference on Harmonisation Good Clinical Practice (ICH-GCP) guidelines, while also complying with applicable local and institutional regulations. The study protocol received approval from the institutional review board/local ethics committee (ID: 2664). Written informed consent was obtained from all participants prior to enrollment in the study. Full eligibility criteria are listed in the trial protocol (available at www.clinicaltrials.gov). Decsription : The NCT04758507 trial is an investigator-initiated, randomized, double-blind, placebo-controlled phase 2a study. It aims to evaluate whether fecal microbiota transplantation (FMT) from donors who achieved a complete response to immune checkpoint inhibitors (ICB) can improve the efficacy of first-line therapy with pembrolizumab combined with axitinib in patients with metastatic renal cell carcinoma (mRCC). The trial was carried out in accordance with the principles of the Declaration of Helsinki and the International Conference on Harmonisation Good Clinical Practice (ICH-GCP) guidelines, while also complying with applicable local and institutional regulations. The study protocol received approval from the institutional review board/local ethics committee (ID: 2664). Written informed consent was obtained from all participants prior to enrollment in the study. Full eligibility criteria are listed in the trial protocol (available at www.clinicaltrials.gov). Ethics approval and consent to participate. The study design and conduct complied with all relevant regulations regarding the use of human participants and were conducted in accordance with the Declaration of Helsinki and Good Clinical Practice (GCP) guidelines. All participants provided written informed consent. Consent to publish clinical information potentially identifying individuals (e.g., age and sex) was also obtained. Data and sample collection adhered to applicable regulatory and ethical requirements, as well as the ICH E6(R2) GCP guidelines. Ethic committees. -ONCOBIOTICS “IML1” (NCT04567446): Translational analyses were performed under the study ONCOBIOTICS (sponsor protocol N: CSET 2017/2619, ID800 RCB N: 2017-A02010-53) according to the ethical guidelines and approval of the local ethical committee (Comité Consultatif de Protection des Personnes dans la Recherche Biomédicale (CCPPRB) of the Kremlin Bicêtre Hospital). ONCOBIOTICS “IML1” (NCT04567446): Translational analyses were performed under the study ONCOBIOTICS (sponsor protocol N: CSET 2017/2619, ID800 RCB N: 2017-A02010-53) according to the ethical guidelines and approval of the local ethical committee (Comité Consultatif de Protection des Personnes dans la Recherche Biomédicale (CCPPRB) of the Kremlin Bicêtre Hospital). -MIMIC (NCT03772899): Patients were enrolled and treated at three academic centers in Canada to include the London Regional Cancer Program (LRCP, Ontario), the Center hopitalier de l'Université de Montréal (CHUM, Québec), and the Jewish General Hospital (JGH, Québec). The research ethics board (REB) at each institution approved the conduct of the trial in accordance with the current federal regulations to include the Canadian Food and Drug Regulations (C.05.001); the US Code of Federal Regulations (21CFR part 56); International Conference on Harmonisation of Technical Requirements for Registration of Pharmaceuticals for Human Use Good Clinical Practice Guidelines (ICH GCP); and the Declaration of Helsinski. Refer to https://doi.org/10.1038/s41591-023-02453-x for this information. MIMIC (NCT03772899): Patients were enrolled and treated at three academic centers in Canada to include the London Regional Cancer Program (LRCP, Ontario), the Center hopitalier de l'Université de Montréal (CHUM, Québec), and the Jewish General Hospital (JGH, Québec). The research ethics board (REB) at each institution approved the conduct of the trial in accordance with the current federal regulations to include the Canadian Food and Drug Regulations (C.05.001); the US Code of Federal Regulations (21CFR part 56); International Conference on Harmonisation of Technical Requirements for Registration of Pharmaceuticals for Human Use Good Clinical Practice Guidelines (ICH GCP); and the Declaration of Helsinski. Refer to https://doi.org/10.1038/s41591-023-02453-x for this information. -FMT-LUMINATE (NCT04951583): The clinical trial and correlative analyses were approved by the CHUM Ethics Review Board and each participating center, ethics number: MP-02-2022-10121/21.173. Refer to https://doi.org/10.1038/s41591-025-04186-5 for this information. FMT-LUMINATE (NCT04951583): The clinical trial and correlative analyses were approved by the CHUM Ethics Review Board and each participating center, ethics number: MP-02-2022-10121/21.173. Refer to https://doi.org/10.1038/s41591-025-04186-5 for this information. -ATEZOTRIBE (NCT03721653): The study has been approved by 25 ethics committees and is currently ongoing at 25 Italian oncology units. The protocol is conducted in accordance to the standards of Good Clinical Practice, in agreement with the principles laid down by the 18th Word Medical Assembly (Helsinki 1964) and subsequent amendment (Tokyo 1975, Venice 1983, Hong Kong 1989, Somerset West 1996, Edinburgh 2000).The protocol version 1.2 (6th October 2017) was approved on 20th September 2018 by the Ethics Committee of Coordinating Center (Comitato Etico Area Vasta Nord Ovest, CEAVNO) and by the local Ethics Committees of participating centers. The list is provided as Additional file 1. Then, the study was amended in order to regulate additional tissue, blood and fecal samples collection and the protocol version 2.1 (19th February 2019) was approved on 13th June 2019 by the Ethics Committee of Coordinating Center (Comitato Etico Area Vasta Nord Ovest, CEAVNO) and by the local Ethics Committees of participating centers. Refer to https://doi.org/10.1186/s12885-020-07169-6 for this information (Supplementary information). ATEZOTRIBE (NCT03721653): The study has been approved by 25 ethics committees and is currently ongoing at 25 Italian oncology units. The protocol is conducted in accordance to the standards of Good Clinical Practice, in agreement with the principles laid down by the 18th Word Medical Assembly (Helsinki 1964) and subsequent amendment (Tokyo 1975, Venice 1983, Hong Kong 1989, Somerset West 1996, Edinburgh 2000).The protocol version 1.2 (6th October 2017) was approved on 20th September 2018 by the Ethics Committee of Coordinating Center (Comitato Etico Area Vasta Nord Ovest, CEAVNO) and by the local Ethics Committees of participating centers. The list is provided as Additional file 1. Then, the study was amended in order to regulate additional tissue, blood and fecal samples collection and the protocol version 2.1 (19th February 2019) was approved on 13th June 2019 by the Ethics Committee of Coordinating Center (Comitato Etico Area Vasta Nord Ovest, CEAVNO) and by the local Ethics Committees of participating centers. Refer to https://doi.org/10.1186/s12885-020-07169-6 for this information (Supplementary information). -TACITO (NCT04758507): The study was performed in compliance with the Declaration of Helsinki. Ethical approval was granted by Ethics Committee of the Fondazione Policlinico Gemelli IRCCS (ID #2664). Written informed consent was obtained from all participants. Refer to https://doi.org/10.1038/s41591-025-04189-2 for this information. TACITO (NCT04758507): The study was performed in compliance with the Declaration of Helsinki. Ethical approval was granted by Ethics Committee of the Fondazione Policlinico Gemelli IRCCS (ID #2664). Written informed consent was obtained from all participants. Refer to https://doi.org/10.1038/s41591-025-04189-2 for this information. -The PREMIS study was approved by an ethical committee (Comité de Protection des Personnes Sud-Ouest et Outre Mer I; ID-RCB: 2018-A01257-48) and the institutional review board (CSET no. 2018/2728). All enrolled patients provided a signed informed consent. The PREMIS study was approved by an ethical committee (Comité de Protection des Personnes Sud-Ouest et Outre Mer I; ID-RCB: 2018-A01257-48) and the institutional review board (CSET no. 2018/2728). All enrolled patients provided a signed informed consent. -The MEDITREME study was approved by the Ethics Committee CPP TOURS Région Center – Ouest 1 on 27 March 2017 under the number 2017T1-03 and was registered with the French national health products agency (ANSM). The French ethical authorities asked us to include only RAS-mutated metastatic CRC because RAS WT patients must receive an anti-EGFR in first-line of treatment, and it would be unethical to give a protocol without anti-EGFR for these patients. The study was conducted in accordance with the Declaration of Helsinki and International Conference on Harmonization for Good Clinical Practice guidelines and the CONSORT 2010 guidelines. All patients provided written informed consent to study procedures before enrollment. The last version of the protocol is available as supplementary information. The MEDITREME study was approved by the Ethics Committee CPP TOURS Région Center – Ouest 1 on 27 March 2017 under the number 2017T1-03 and was registered with the French national health products agency (ANSM). The French ethical authorities asked us to include only RAS-mutated metastatic CRC because RAS WT patients must receive an anti-EGFR in first-line of treatment, and it would be unethical to give a protocol without anti-EGFR for these patients. The study was conducted in accordance with the Declaration of Helsinki and International Conference on Harmonization for Good Clinical Practice guidelines and the CONSORT 2010 guidelines. All patients provided written informed consent to study procedures before enrollment. The last version of the protocol is available as supplementary information. -Protocol approval for the METIMMOX trial was obtained from the Regional Committee for Medical and Health Research Ethics of South-East Norway (2017/1850), the Norwegian Medical Agency (17/12752) and the institutional review boards. The trial was conducted in accordance with the guidelines of Good Clinical Practice and the Declaration of Helsinki. All patients provided written informed consent. Protocol approval for the METIMMOX trial was obtained from the Regional Committee for Medical and Health Research Ethics of South-East Norway (2017/1850), the Norwegian Medical Agency (17/12752) and the institutional review boards. The trial was conducted in accordance with the guidelines of Good Clinical Practice and the Declaration of Helsinki. All patients provided written informed consent. -The G2R (GUIDE2REPAIR) study was approved by the Ethics Committee CPP île de France 8 on 18 July 2019 under the EudraCT number 2018-002971-17 and was registered with the French national health products agency (ANSM). The G2R (GUIDE2REPAIR) study was approved by the Ethics Committee CPP île de France 8 on 18 July 2019 under the EudraCT number 2018-002971-17 and was registered with the French national health products agency (ANSM). Human participant details. -ONCOBIOTICS “IML1” (NCT04567446): Biological sex and gender were considered in the study, and we had 69% male and 31% female participants as referred in Table S5. ONCOBIOTICS “IML1” (NCT04567446): Biological sex and gender were considered in the study, and we had 69% male and 31% female participants as referred in Table S5. -MIMIC (NCT03772899) and FMT-LUMINATE (NCT04951583: Biological sex was considered in the study and we had 62% male and 38% female participants, and refer to Table S4. MIMIC (NCT03772899) and FMT-LUMINATE (NCT04951583: Biological sex was considered in the study and we had 62% male and 38% female participants, and refer to Table S4. -ATEZOTRIBE (NCT03721653): Biological sex and gender were considered in the study, and refer to Table S5. ATEZOTRIBE (NCT03721653): Biological sex and gender were considered in the study, and refer to Table S5. -The PREMIS : Biological sex and gender were considered in the study, and refer to Table S5. The PREMIS : Biological sex and gender were considered in the study, and refer to Table S5. -METIMMOX : Biological sex and gender were considered in the study, and refer to Table S5. METIMMOX : Biological sex and gender were considered in the study, and refer to Table S5. -MEDITREM : Biological sex and gender were considered in the study, and refer to Table S5. MEDITREM : Biological sex and gender were considered in the study, and refer to Table S5. -TACITO (NCT04758507): Biological sex and gender were considered in the study. TACITO (NCT04758507): Biological sex and gender were considered in the study. -G2R (GUIDE2REPAIR): Biological sex and gender were considered in the study, and refer to Table S5. G2R (GUIDE2REPAIR): Biological sex and gender were considered in the study, and refer to Table S5. Race, ethnicity, or other socially relevant groupings were not included in this study. OS and PFS were assessed from the date of treatment start to the date of death or disease progression/death, respectively. Experimental model Mice All animal experiments were performed in compliance with French and European laws and regulations or the Canadian Council on Animal Care guidelines. The institutional animal ethics committee (Direction départementale de la protection des populations (DDPP) du Val-de-Marne ) and the French Ministère de la Recherche approved all procedures in this study (permission number: 2023-024-42156 (all mice experiments except written otherwise) and 2024_010_48794 (FMT experiments) or #46833–202401111607908 v6 (BDL experiment)). Experiments were performed in accordance with Government and institutional guidelines and regulations. Seven-weeks old female C57BL/6 mice were purchased from Harlan or Envigo (France). Canadian animal studies were approved by the Institutional Animal Care Committee (CIPA) (Ethics numbers: C18025BRs). Murine experiments were conducted using seven-week-old female C57BL/6 mice, obtained from Charles River. All mouse experiments were performed at the animal facility at the GRCC or at CHUM in Montreal (Figure 1 except D; Figures S1A and S1C) except for the bile duct ligation performed at the Center de Recherche des Cordeliers. The animals were housed in SPF conditions with a 12 hour-day/night rhythm and naive at baseline. Before every experiment, we randomized the mice to avoid bias. Each experiment was at least duplicated. No samples were excluded from the experiments. The plasma and the stool were drawn or harvested in the afternoon and morning respectively. If not otherwise indicated, mice were treated with a broad-spectrum ABX solution containing ampicillin (1mg/mL), streptomycin (5 mg/mL), and colistin (1 mg/mL) (Sigma-Aldrich) (ABX) added to sterile drinking water. Solutions and bottles were changed three times weekly. ABX were administered during 7–14 days in the drinking water of naive or MCA205 tumor-bearing mice. In accordance with our institutional ethics committee guidelines, the maximal permitted tumor burden was 1500 mm3 corresponding to 300 mm2 (calculation was L x l). Tumor sizes were measured and controlled three times a week and when the tumor size exceeded 250 mm2, mice were euthanized immediately to avoid reaching the permitted size of 300 mm2. Mice were also controlled by technicians every day. Predefined humane endpoints included tumor ulceration or body weight loss greater than 20% relative to baseline. If either endpoint was reached, mice were euthanized within 24 h. Madcam1−/− (B6.129-Madcam1tm1.2Nwag/J) mice were a kind gift from A. Schippers (University Hospital Aachen, Aachen, Germany). Tgr5/Gpbar1 −/− mice were a kind gift from Kristina Schoonjans, PhD, Laboratory of Metabolic Signaling (LMS), EPFL, Switzerland. The mice were breed at Gustave Roussy in SPF conditions with a 12 hour-day/night rhythm and naive at baseline. The genotyping was performed by western-blot. Method details Blood sample collection for ONCOBIOTICS, MIMic and FMT-LUMINATE Blood samples were drawn from patients. Plasma was collected after centrifugation of lithium heparin tubes without EDTA and peripheral blood mononuclear cells (PBMC) were separated using a density gradient medium. Blood samples were used for the detection of bile acids by UHPLC-MS. Stool samples collection for ONCOBIOTICS Stool samples were self-collected by the participants at home according to the IHMS_SOP 004 (https://human-microbiome.org/index.php?id=Sop&num=004). Immediately after the collection, they were stored at −20°C and transported on dry ice. Upon arriving at the biobank of the recruitment center (between 4 and 24 h later), they were stored at -80°C in plastic tubes. Two aliquots of the same sample were used for metaproteomic and metagenomic analyses. All the samples were processed according to International Human Microbiome Standards (IHMS) guidelines (SOP 03 V1). DNA was extracted from an aliquot of the same stool samples used for metaproteomic analysis using the IHMS SOP 07 V2 H and sequenced with an Ion proton sequencer (Thermo Fisher Scientific). The sequencing was performed at a single site (MetaGenoPolis). Reads were cleaned using Alien Trimmer (version 0.4.0) to (1) remove resilient sequencing adapters and (2) trim low-quality nucleotides at the 3′ side using a quality cutoff of 20 and a length cutoff of 45 bp. Cleaned reads were subsequently filtered from human and other possible food contaminant DNA using human genome (GRCh38-p13), with an identity score threshold of 95%. For each metagenome we profiled the taxonomic composition with MetaPhlAn-4.55 For alpha diversity, we computed the per-sample Richness and Shannon indexes56 using the vegan R package.57 Differences between groups were assessed using the Wilcoxon-Mann-Whitney test. For beta-diversity, we computed the between-samples Bray-Curtis dissimilarities using the implementation available in the Vegan R package 51. Differential abundance analysis on taxonomic profiles between groups were performed using the LefSE software. Human and mouse plasma sample preparation and bile acid detection by UHPLC-MS For each patient, 50 μL of human plasma was mixed with 500 μL of ice-cold extraction mixture (methanol/water, 9/1 spiked with a cocktail of internal standards) and centrifuged (15 min at 10,000g, 4°C). Regarding murine plasma, 25 μL of biological sample was mixed with 250 μL of the same ice-cold extraction mixture and treated as described above. Mice plasma was harvested at the same time for each time point. Supernatants were collected and several fractions were split to be analyzed by different liquid and gaz chromatographies coupled with mass spectrometers (LC-MS and GC-MS).58 For human samples, 150 μL of supernatant (50 μL for murine) were transferred to glass vials and evaporated to be used for GC-MS. Then, 50 μL of methoxyamine (20 mg/mL in pyridine) was added on dried extracts, which were stored at room temperature in the dark, overnight. The day after, 80 μL of N-Methyl-N- trimethylsilyl-trifluoroacetamide were added, and the final derivatization was made at 40°C for 30 min. Samples were then transferred into vials and directly injected into the GC-MS system – 7890A gas chromatograph coupled with a triple quadrupole 7000C (Agilent Technologies) – equipped with a high sensitivity electronic impact source (EI) operating in positive mode. Others aliquots (polyamines, bile acids and short chain fatty acids) were analyzed by LC-MS 1290 UHPLC (Ultra-High Performance Liquid Chromatography) coupled to a QQQ 6470 (Agilent Technologies). Biliary acids and polyamines were analyzed by MRM analysis.54 Small organic acids and ketones were analyzed by MRM analysis in both positive and negative polarity, alternatively, gas temperature was set to 300°C with a gas flow of 12 L/min. Capillary voltage was set to 4 kV in positive and 5 kV in negative polarity mode. Ten (10) μL of sample were injected on a Zorbax Eclipse XDB- column (P/N 981758-902, 100 mm × 2.1 mm particle size 1.8 μm 1200 bars) from Agilent technologies, protected by a guard column (XDB-C18; 5 mm × 2.1 mm particle size 1.8 μm) and heated at 50°C by a Pelletier oven. Gradient mobile phase consisted of water with 0.1% of formic acid (A) and 0.1% of formic acid in acetonitrile (B) freshly made. Flow rate was set to 0.7 mL/min, and gradient as follow: initial condition was 80% phase A and 20% phase B for 3 min, changing to 45% phase B over 4 min before rinsing phase and equilibrium before next injection. Peak detection and integration of the analytes were performed using the Agilent Mass Hunter quantitative software (B.07.01). Pseudo-targeted analysis by UHPLC-HRAM (Ultra-High Performance Liquid Chromatography – High Resolution and Accurate Mass) was performed on a U3000 (Dionex)/Orbitrap q-Exactive (Thermo) coupling, previously described in Durand et al. 202159 and Abdelatif et al.60 All targeted treated data were merged and cleaned with a dedicated R package (version 4.0) (https://github.com/kroemerlab/GRMeta). sST2, sMAdcAM-1 and S100A9 quantification in patients sST2, sMAdCAM-1 and S100A9 were quantified in patient’s plasma samples with the Bio-Plex 200 systems (Bio-Rad) and sMAdCAM-1/ST2/S100A9 kit from R&D Systems (Human Luminex Discovery Assay LXSAHM). Spectral flow cytometry on PBMC Aliquots of 5–10×106 PBMC cryopreserved in fetal calf serum (FCS)/DMSO solution were defrost, washed and resuspended in FACS Buffer (Brilliant Stain Buffer Plus BD + 5% FCS + 2mM EDTA). Dead cells were excluded using LIVE/DEAD™ Fixable Blue Dead Cell Stain Kit from Invitrogen. CCR7 (BV421), LPAM1 (Alexa 647), PDL1 (PE-Dazzle594), PDL2 (RB780), CADM1 (PE), CCR4 (BV605), CCR6 (BUV661), CD103 (PE-Fire640), CD1c (BUV737), CXCR3 (PE-Fire810), CXCR5 (BUV615), FceRIa (BV480), ICOS (Spark R718), IgA (APC-Vio700), IgD (cF BYG667), IgG (BV480), IgM (BV510), PD1 (BV750), CD11c (cF BYG781), CD123 (Spark V432), CD127 (cF R659), CD14 (cF B548), CD148 (BUV395), CD16 (BV570), CD19 (cF R685), CD21 (BUV805), CD24 (BB515), CD25 (PE-Fire700), CD27 (cF R840), CD3 (BV650), CD38 (BV711), CD4 (cF YG584), CD40 (BV650), CD45 (Spark B574), CD45RA (cF V450), CD5 (BUV563), CD71 (BUV496), CD8 (cF B532), CD88 (BV786), HLA-DR (cF B690), Streptavidin (PE) were purchased from BD Biosciences, R&D systems, BioLegend, Miltenyi Biotec and Cytek and used to stain cells. (see STAR Methods) Stained samples were acquired on Cytek Aurora 5-laser spectral analyzer and analyses were performed on FlowJo v10.8.1. Slope-prediction method Responders and non-responders were compared using a longitudinal slope estimation approach. For each metabolite, individual longitudinal slopes were estimated using Best Linear Unbiased Predictors (BLUP) derived from linear mixed-effects models fitted to repeated measurements over time. Estimated slopes were subsequently compared between responders and non-responders to evaluate differences in metabolite trajectories. The significance of temporal changes within responders and non-responders was assessed from the estimated time effect in each group. (Related to Figure 8F, left). Clinical cox models Kaplan-Meier survival curves were generated using Prism 10 (GraphPad, San Diego, CA, USA) and the statistical difference between groups was compared using the Log rank test. Multivariate analysis and hazard ratio evaluation were performed using the Cox proportional hazard regression model. Patients clustering based on circulating bile acids composition Unsupervised hierarchical clustering was performed to identify groups of patients with similar circulating bile acid profiles. A matrix containing bile acid percentages or log2foldchange values for each patient was used as input. Clustering analysis was conducted in R using the “pheatmap” package (version 1.0.13). Hierarchical clustering was applied to both patients and bile acid variables, or only for bile acid variables. Pairwise distances between patients were calculated using the Euclidean distance metric. Clusters were generated using Ward’s minimum variance method (Ward.D2 linkage), which minimizes the total within-cluster variance at each step of the agglomerative clustering process. No data scaling or normalization was applied prior to clustering. The resulting dendrograms and heatmaps were generated using the “pheatmap” function, with patient annotations displayed according to the available clinical metadata. The clustering output was used to define patient subgroups exhibiting distinct circulating bile acid signatures. (related to Figure 6A; Figures S9A, S9D, and S11I). Oral bacterial gavage with commensal species or BAs E. clostridioformis was isolated from the ileum of mice that received ABX in drinking water for 7 days followed by 4 days ABX-free at GRCC.21 E. clostridioformis, were grown on Columbia agar with 5% sheep blood (COS) plates in an anaerobic atmosphere created using three anaerobic generators (Biomerieux) at 37°C for at least 48 h. Colonization of SPF C57BL/6 mice was performed by oral gavage with 100 mL of suspension containing 1×108 to 1×109 bacteria. The optical density of bacterial solution in in 0.9% NaCl was measured using a fluorescence spectrophotometer (Eppendorf) at wave lengths of 600 nm in 0.9% NaCl. Mice were gavaged one to eight times. The identification of specific bacteria was accomplished using a matrix-assisted laser desorption/ionization– time-of-flight (MALDI-TOF) mass spectrometer (Brucker, France). Mice were gavaged four times every 2 days with 120 mg/kg body weight of LCA (Thermo-Fisher Scientific, 229090050), UDCA (Thermofisher, 208590250), GDCA (Sigma Aldrich, G9910-5G), TDCA (EMD Millipore Corp, 580221-5GM), DCA (Thermofisher, 148140250) diluted in corn oil. For Fexaramine treatment (MCE, 574013-66-4), mice were gavaged every day during 10 days (starting at day 0 of ABX treatment) and then every 2 days for one more week (after ABX cessation) at the dose of 120mg/kg of mice. FMT experiments In FMT experiments, mice received 3 days of ABX before undergoing FMT (from 4 different cancer patients) the next day by oral gavage. Fecal samples collected from patients in the ONCOBIOTICS study were used for FMT. FMT was performed by thawing fecal material. As described before, frozen fecal samples were brought back to room temperature and resuspended in NaCl at a density of 0.1g/mL. Mice were placed in a new cage after the procedure. Two hundred microliters of the suspension were then transferred by oral gavage into each ABX-pretreated recipient using animal-feeding needles. In addition, another 100μL was applied to the fur of each animal. Subcutaneous cancer mouse models MCA205 fibrosarcoma was purchased from the American Type Culture Collection and. The RET melanoma cell line, generated by transgenic expression of the RET proto-oncogene under the control of the metallothionein-1 promoter driving spontaneous melanogenesis, was provided by Viktor Umansky (DKFZ—German Cancer Research Center; syngeneic from C57BL/6J mice). Both cell lines (syngeneic to C57BL/6J mice) were cultured at 37°C in the presence of 5% CO2 in RPMI 1640 containing 10% fetal calf serum (FCS), 2 mM L-glutamine, 100 IU/mL penicillin–streptomycin, 1 mM sodium pyruvate, and minimum essential medium nonessential amino acids (henceforth referred to as complete RPMI 1640). Cell lines were regularly tested for Mycoplasma contamination and were not used after more than 10 passages. MCA205 sarcoma and RET syngeneic to C57BL/6J mice were implanted subcutaneously with 0.8×106 and 0.5 × 106 cells/mouse for MCA205 and RET respectively and treated intraperitoneally when tumors reached 25 to 35 mm2 in size with αPD-1 mAb (250 μg/mouse, clone RMP1-14) or isotype control (250 μg/mouse, clone 2A3) for MCA205 and αPD-1 + αCTLA-4 mAb (250 μg/mouse, clone RMP1-14 and 100μg/mouse, clone 9D9 respectively) or isotype control (250 μg/mouse, clone 2A3 and 100μg/mouse, clone MPC-11 respectively) for RET. Mice were injected four times at a 3-day interval with αPD-1/αCTLA-4 (Bio X Cell, clone RPMI-14, 250 μg per mouse; clone 9D9, 100μg/mouse respectively) or isotype control (Bio X Cell, clone 2A3, 250 μg per mouse; clone MPC-11, 100μg/mouse respectively) mAbs. Tumor length and width were routinely monitored three times a week by means of a caliper. All antibodies were purchased from Bio X Cell. Statistical analysis of tumor growth Longitudinal tumor growth data were analyzed using the online tool “TumGrowth” available at https://kroemerlab.shinyapps.io/TumGrowth/, which is an open-access platform specifically developed for the statistical analysis of experimental tumor growth curves. Tumor volume measurements collected over time were modeled using linear mixed-effects models, allowing incorporation of repeated measurements within individual animals and inter-animal variability. Tumor growth kinetics were fitted using a two-phase piecewise linear regression model (linear-linear model). When appropriate, the breakpoint between growth phases was estimated automatically by the software. Treatment group and time were included as fixed effects, while individual animals were modeled as random effects. Model adequacy was assessed using the diagnostic plots provided by the application. Differences in tumor growth between treatment groups were assessed by pairwise comparisons of the estimated growth slopes derived from the mixed-effects model. Statistical significance was evaluated using Wald tests, and p-values were adjusted for multiple comparisons when appropriate. A two-sided p-value <0.05 was considered statistically significant. Surgical bile duct ligation Cell culture. MCA205 murine fibrosarcoma cells were maintained in DMEM supplemented with 10% fetal bovine serum (FBS) and 1% penicillin–streptomycin (PS). Animal experiments. Nine-week-old female C57BL/6J mice were housed under specific pathogen–free conditions with ad libitum access to food and water on a 12 h light/dark cycle. All animal experiments were approved by the institutional animal care and use committee (ethical protocol #46833) and conducted in accordance with applicable guidelines. On day 0 (D0), mice were subcutaneously injected in the right posterior flank with 0.8 × 106 MCA205 cells suspended in 50 μL 1× PBS under 2% isoflurane anesthesia. Tumor dimensions were measured with calipers at the time points indicated in the scheme, and tumor area (mm2) was calculated as length × width. On D3–D4 after tumor inoculation, mice underwent bile duct ligation (BDL) or sham surgery under 2% isoflurane anesthesia. For BDL, the common bile duct was exposed and doubly ligated. For sham surgery, the bile duct was exposed and manipulated without ligation. Perioperative care and postoperative monitoring were performed in accordance with institutional guidelines. Anti-PD-1 treatment was initiated when tumors reached approximately 25 mm2. Mice then received intraperitoneal injections of anti–PD-1 or isotype control (Iso) antibody under 2% isoflurane anesthesia on D6, D8, and D11, at 250 μg per mouse in 100 μL PBS (1×) per injection. Entire tumor growth curves were analyzed using TumGrowth (https://kroemerlab.shinyapps.io/TumGrowth/) with independent sample t test. A p value <0.05 was considered statistically significant. Data are presented as mean ± SEM unless otherwise stated. MCA205 fibrosarcoma knockout mouse model Establishment of BA receptors knockout cell lines Edit-R All-in-one Lentiviral (knockout) Mouse Nr1h4 sgRNA (target sequence: AGTGCCAGGAGTGCCGGCTA; catalog no. VSGM11942-247929487), Gpbar1 sgRNA (target sequence: ATGAGCGTCGCCCACCACTA; catalog no. VSGM11942-247972009), and Vdr sgRNA (target sequence: TTCGTGCAGACGTAAGTACA; catalog no. VSGM11942-248142193) were purchased from Horizon Discovery. For transduction, MCA205 fibrosarcoma cells were expanded and seeded at 4 × 105 cells per well in tissue culture-treated 6-well plate with complete RPMI-1640 medium. TSECs were expanded and seeded at 1.4 × 105 cells per well in 6-well plate with complete DMEM/F-12 medium supplemented with ECGS, and cultured overnight. The following day, the culture supernatant was removed, and the transduction mixture was prepared by combining 10 μg/mL polybrene (catalog no. TR-1003-G, Sigma-Aldrich) with 200 μL of lentiviral particles and 800 μL of complete medium (without serum and antibiotics) per well, corresponding to a multiplicity of infection (MOI) of 0.3. The mixture was added to the cells and incubated for 6 h at 37°C in a 5% CO2 incubator. Subsequently, 1 mL of complete medium (containing 2× serum and 2× antibiotics) was added, and cells were cultured for an additional two days. An additional control well, treated identically but without lentiviral particles, was included. Forty-eight hours post-transduction, cells were detached with trypsin and transferred to T75 tissue culture flasks. Selection was performed using puromycin: 2 μg/mL for TSEC knockouts and 6 μg/mL for MCA205 knockouts, maintained for 7 days until all cells in the control flask were eliminated. Surviving cells were collected and subjected to single-cell sorting using a BD FACSAria™ Fusion flow cytometer (BD, Franklin Lakes, NJ, USA). Immediately prior to sorting, 1 μg/mL DAPI was added to the cells to assess viability. Viable (DAPI-negative), puromycin-resistant single clones were isolated and cryopreserved at -80°C. These KO tumor cells were inoculated into mice, which were subsequently treated with either αPD-1 or an isotype control antibody as previously described. Mice received four intraperitoneal injections at a 3-day interval with αPD-1 (Bio X Cell, clone RPMI-14, 200 μg per mouse) or isotype control (Bio X Cell, clone 2A3, 200 μg per mouse) monoclonal antibodies. Tumor length and width were measured three times per week using a caliper. Flow cytometry Tumors and tdLNs, were harvested at different time points as indicated in the individual experiments. Excised tumors were cut into small pieces and digested in RPMI medium containing Liberase at 25 mg/mL (Roche) and DNase1 at 150 IU/mL (Roche) for 30 min at 37°C, and then crushed and filtered twice using 70μm and 30μm cell strainers (Miltenyi Biotec). Lymph nodes and spleens were crushed in RPMI medium and subsequently filtered twice through a 70-μm cell strainer. Four million cells from each sample were pre-incubated with purified anti-mouse CD16/CD32 (clone 93; eBioscience) for 10 min at 4°C before membrane staining (30 min at 4°C). For intracellular staining, the FoxP3 staining kit (eBioscience) was used. Dead cells were excluded using the LIVE/DEAD Fixable Aqua Dead Cell Stain Kit (Biolegend). Information on the antibodies used to stain murine cells is found in STAR Methods. Stained samples were acquired on CytoFLEX S 13 colors (Beckman Coulter), and analyses were performed with Kaluza software 2.1 (Beckmann Coulter). Treg cells were defined after gating on FoxP3+ CD25+ CD4+ cells. Tracing migration of leukocytes from the gut to the TME: CFSE staining of mLN cells up to TME or tdLNs C57BL/6J tumor-bearing mice were anesthetized with 2–2.5% isoflurane and administered meloxicam (1mg/kg) intraperitoneally for analgesia. Abdominal skin and peritoneum were cut at the midline to access the mLNs. The mLNs were gently mobilized through the midabdominal incision onto a sterile plastic-coated surgical drape. Ileum draining mLNs were visually identified according to their vasculature. The two most prominent mLNs were injected with 100 μM CFSE diluted in 5 μL of PBS using a 30-Ga insulin syringe. After repositioning the mLNs, the peritoneal membrane was closed by continuous clips. RNA extraction and RT-qPCR lysis and extraction protocols for mouse samples Tumor or intestinal samples were snap-frozen in liquid nitrogen. On the day of extraction, samples were thawed at 4°C and put in RLT Plus buffer containing 0.1% beta mercaptoethanol then homogenized in a microtube homogenizer (Benchmark Scientific) in RNA-free glass bead tubes (Dutscher). Total RNA extraction and genomic DNA removal were performed with the RNeasy Mini kit (Qiagen) following the manufacturer’s recommendations. A maximum of 1 μg of RNA, measured using a NanoDrop Spectrophotometer (ThermoFisher Scientific), was reverse transcribed into cDNA with a mixture composed of SuperScript III Reverse Transcriptase (Life Technologies), RNaseOUT Recombinant Ribonuclease Inhibitor (Life Technologies), random primers (Promega), and Deoxynucleoside Triphosphate Set, PCR grade (Roche Diagnostics). Quantitative gene expression assay The expression of mouse Ppia, Madcam1, Nr1h4, Gpbar1, Vdr, Nr1i2, Nr1i3 genes (all from Life Technologies) was analyzed with the TaqMan Gene Expression Assay using the Universal Master Mix II on a QuantStudio™ 3 Real-Time PCR System (Life Technologies). Amplifications were performed using the following ramping profile: one cycle at 95°C for 10 min, followed by 45 cycles of 95°C for 30 s, and 60°C for 1 min. RT-qPCR data were normalized to the expression levels of the housekeeping genes Ppia, as indicated in each figure, by means of the 2−DCt method multiplied by 106. Bulk RNA-sequencing for hepatocytes, cholangiocytes and enterocytes For enterocytes isolation, we used the Lamina Propria Dissociation kit from Miltenyi followed by magnetic separation using the CD326 (EpCAM) MicroBeads mouse from Miltenyi. For hepatocytes and cholangiocytes, we used the Liver Perfusion kit from Miltenyi. The hepatocytes were obtained from the pellet at the end of the protocol. We used the supernatant and performed a magnetic separation using the CD326 (EpCAM) MicroBeads mouse from Miltenyi to obtain the cholangiocytes. We extracted the RNA from each population with the RNAeasy Mini kit form Quiagen. The Bulk RNA-sequencing was performed at the Genomic Plateform at Gustave Roussy, France. RNA-seq data generation and analysis RNA-seq libraries were generated in five replicates for each condition and sequenced in paired-end mode with a read length of 100 bp. Raw sequencing reads were processed with fastp (v0.23.2)61 to trim adapters and remove low-quality bases (Phred score <30), and, when necessary, downsampled using seqkit (v2.5.1)62 to obtain comparable read depths across samples (20–60 million high-quality reads per sample). Fastq files were subsequently assessed with FastQC (v0.12.1)63 and MultiQC (v1.16),64 and pseudoaligned to the GRCm38 transcriptome reference with kallisto (v0.50.1)65 using default parameters. Genes with an absolute expression of less than 20 reads across all samples were filtered out. Differential expression analysis was performed with DESeq266(v1.40.2). Logarithm fold changes were modified by the lfcShrink algorithm with the “Ashr” option. Differentially expressed gene (DEG) cutoff was defined at an FDR less than 0.05 and an absolute log2 fold change (log2FC) value greater than 0.5 in pairwise comparisons between experimental conditions. For visualization purposes, variance stabilizing transformation (VST) values and transcripts per million (TPM) were obtained with DESeq2 functions and used where indicated in the figures. Single cell RNA-sequencing for tdLN and tumors scRNA-sequencing whole transcriptome using BD rhapsody BD Rhapsody experiments were performed following manufacturer’s (BD Biosciences) protocol. Prior to experiment, tumor cells were enriched using mouse CD45 microbeads from Miltenyi (130-052-30) following the Miltenyi’s protocol. Two independent cartridges were used to capture 77,821 cells (tumor) and 76,529 cells (lymph nodes). Experimental conditions were pooled prior to analysis using the BD™ Mouse Single Cell Sample Multiplexing Kit (Cat No: 633793), with 9 and 11 sample barcodes per cartridge, respectively. Libraries were processed according to the BD Rhapsody™ Whole Transcriptome Analysis (WTA) Amplification Kit (Cat No: 633801) and the BD Rhapsody™ Whole Transcriptome Analysis (WTA) Reagent Kit (Cat No: 665915). Libraries were then subjected to indexed paired-end sequencing, 2 × 100bp setting, on a DNBSEQ-T7 (BGI Genomics, Shenzhen, China) with 30% PhiX spike-in control. A total of 1,415 million reads and 2,068 million reads were sequenced for the tumor and lymph node libraries, respectively. Alignment and pre-processing of single-cell RNA sequencing data Transcriptomic FASTQ files were processed using the standard Rhapsody analysis pipeline (BD Biosciences), following the manufacturer’s recommendations. For read quality control, recursive substitution error correction (RSEC) and distribution-based error correction (DBEC) were applied; these manufacturer-developed algorithms correct for PCR and sequencing errors. A cell was defined as a singlet if the minimum read count associated with a single sample tag exceeded a threshold of 75%. A cell was classified as a multiplet if it exceeded this threshold for more than one sample tag. Cells that did not meet the threshold were labeled as undetermined. Both multiplets and undetermined cells were excluded from downstream analysis. Samples were demultiplexed using the Sample Tag Call matrix generated by the Seven Bridges BD Rhapsody alignment pipeline, and metadata were manually assigned to each tag. Analysis of single cell RNA-sequencing Filtered gene-expression matrices were subjected to processing using Seurat R Package (v 5.4.0). Quality control measures were implemented to remove debris, damaged cells and doublets (min_genes_per_cell = 600; max_genes_per_cell = 5000; max_%mitochondrial_reads = 20; max_%ribosomal_reads = 30). After filtering, each dataset was normalized using Seurat’s log-normalization method, and the top 3,000 highly variable genes were identified. The data were then scaled while regressing out technical sources of variation, including mitochondrial and ribosomal content and total UMI counts. Principal component analysis (PCA) was subsequently performed using the selected variable genes, and the first 30 principal components were retained for downstream analyses. UMAP dimensionality reduction was generated based on these 30 components, followed by neighbor identification and clustering using the Louvain algorithm (resolution = 0.5). Differential expression analysis was performed using FindAllMarkers to identify cluster-specific upregulated genes, which were used for manual annotation of each cell population. For T cells, FindSubCluster (resolution = 0.1, Louvain algorithm) was applied to distinguish CD4 and CD8 subsets. Statistical comparisons were performed using Wilcoxon tests with FDR correction. For the differential expression analysis of total CD8 lymphocytes in tumor single-cell data, all clusters containing CD8 lymphocytes were merged into a single cluster and filtered to remove cells containing Cd4 reads (Cd3e > 0 & Cd8a >0 & Cd4 = = 0). Gene expression was extracted using the FetchData function from Seurat. Pairwise statistical comparisons were performed between experimental conditions using Wilcoxon tests with FDR correction, comparing each bile acid-treated group against the EC control. Results were visualized as violin plots overlaid with boxplots and individual data points using ggplot2. GSEA was performed using the fgsea R package (v1.32.4) on ranked gene lists derived from differential expression results. Genes were ranked by average log2 fold change, with a small random noise term (mean = 0, sd = 1 × 10−6) added to break ties. Genes with non-finite log2 fold change values were excluded prior to ranking. Pathways from the msigdb_v2024.1.Mm_GMTs collection were used as the reference gene sets, with a minimum gene set size of 15 and a maximum of 500. GSEA was run with eps = 0.0 and 100,000 permutations. Pathways with an adjusted p-value <0.05 were considered significant, and results were visualized as dot plots displaying the top 50 pathways ranked by absolute normalized enrichment score (|NES|). Ileal, colonic, liver, tumor tissues preparation and bile acid detection by UHPLC-MS Ileal, colonic, liver, tumor mouse tissues. Approximately 30 mg of collected tissue was weighed and placed in a 2-mL homogenizer tube with ceramic beads (Precellys, Bertin Technologies, France) with 1 mL of ice-cold extraction mixture (9:1 methanol–water, -20°C, with labeled 13C-glycocholic acid as internal standard). To facilitate the extraction of endogenous metabolites, samples were then completely homogenized (3 cycles of 20 s/2380g; Precellys 24, Bertin Technologies, Montigny-le-Bretonneux, France) and centrifuged (10 min at 15,000g, 4°C). To detect bile acids, 300 mL of supernatant was collected and treated following a previously described protocol.58 Bile acids were analyzed by LC-MRM (Multiple Reaction Monitoring) with a 1290 UHPLC (Ultra-High Performance Liquid Chromatography, Agilent Technologies) coupled to a mass spectrometer 6470 TQ (Triple Quadrupole, Agilent Technologies), as previously described.59 Targeted data were cleaned with a R (version 4.0) dedicated package (https://github.com/kroemerlab/GRMeta). Mouse clustering based on bile acids composition Metabolomic profile analysis and visualization were performed in R (version ≥4.4), with data normalization and heatmap organization adapted to the biological objective of each figure using the “pheatmap” package (version 1.0.13). For multi-compartment analyses (plasma, liver, ileum, and colon; related to Figures 1B and S3A), raw metabolite intensities were normalized using row-wise Z score scaling and hierarchically clustered were done across rows using Ward’s minimum variance method (ward.D2). Pairwise group comparisons were performed using two-sided non-parametric Wilcoxon rank-sum tests. Statistical significance was represented as lateral row annotations using discretized p-value categories (≥0.05, <0.05, <0.01, and <0.001). Assessment of the impact of antibiotic treatment in Figure S1D was based on an initial selection of metabolites significantly altered relative to the Water group (Wilcoxon test, p ≤ 0.05). Selected metabolites were subsequently normalized using row-wise Z score scaling and hierarchically clustered across rows using Ward’s minimum variance method (ward.D2). Bile acids were highlighted in red, and p-value categories from the Water versus ABX comparison were displayed as lateral annotations. Unsupervised exploration of the bile acid composition across the Water, ABX, and FMT groups (related to Figure S1I) or with Fexaramine treatment (related to Figure S6B) was performed after inverse hyperbolic sine (asinh) transformation of metabolite intensities and removal of zero-variance variables. Hierarchical clustering was applied to both metabolites and samples using Euclidean distance and complete linkage (ward.D2). In Figure S6B, lateral annotation of p-value classes derived from a two-sided Wilcoxon test between the two groups. (related to Figures 1B; S1D and S1I, and S3A, S6B). Cell lines and treatment Culture media and supplements were obtained from Life Technologies (Carlsbad, CA, USA), and plasticware was sourced from Corning (Corning, NY, USA). Transformed sinusoidal endothelial cells (TSECs) were generously provided by B.P. Fennimore (University of Colorado Anschutz Medical Campus, Aurora, CO, USA). TSECs and TSECs with knockout of three bile acid receptors-TGR5 (gene: Gbpar1), FXR (gene: Nr1h4), and VDR (gene: Vdr)-were cultured in DMEM/F-12 medium supplemented with GlutaMAX™, 10% fetal bovine serum (FBS), 10 U/mL penicillin sodium, 10 mg/mL streptomycin sulfate, and Endothelial Cell Growth Supplement (ECGS; catalog no. 1052, ScienCell). Cells were maintained at 37°C in a humidified incubator with 5% CO2. Recombinant murine TNF-α (catalog no. 315-01A, PeproTech) was diluted into culture medium from a 40 μg/mL stock solution immediately prior to treatment. Various bile acids and related compounds were prepared as follows: cholic acid (CA; catalog no. 700212P, Avanti), chenodeoxycholic acid (CDCA; catalog no. C8261, Sigma-Aldrich), α-muricholic acid (α-MCA; catalog no. 700232P, Avanti), β-muricholic acid (β-MCA; catalog no. 700233P, Avanti), deoxycholic acid (DCA; catalog no. 30960, Sigma-Aldrich), hyocholic acid (HCA; catalog no. 700159P, Avanti), hyodeoxycholic acid (HDCA; catalog no. 700214P, Avanti), 12-dehydrocholic acid (12-dDCA; catalog no. 700215P, Avanti), lithocholic acid (LCA; catalog no. 229090050, Thermo Scientific Chemicals), ursodeoxycholic acid (UDCA; catalog no. 208590250, Thermo Scientific Chemicals), and tyrosine-conjugated cholic acid (Tyro-CA, generously provided by Kenya Honda, Keio University School of Medicine, Tokyo, Japan), as well as glycocholic acid (GCA; catalog no. 700265P, Avanti), glycodeoxycholic acid (GDCA; catalog no. G9910, Sigma-Aldrich), glycohyocholic acid (GHCA; catalog no. HY-122141, MedChemExpress), glycohyodeoxycholic acid (GHDCA; catalog no. 700264P, Avanti), taurocholic acid sodium salt hydrate (TCA; catalog no. T4009, Sigma-Aldrich), sodium taurochenodeoxycholate (TCDCA; catalog no. T6260, Sigma-Aldrich), taurodeoxycholic acid sodium salt (TDCA; catalog no. 580221, Millipore), taurohyodeoxycholic acid sodium salt (THDCA; catalog no. 700248P, Avanti), taurolithocholic acid sodium salt (TLCA; catalog no. 700252P, Avanti), and tauroursodeoxycholic acid sodium salt (TUDCA; catalog no. 580549, Millipore). The FXR agonist Cilofexor (catalog no. HY-109083, MedChemExpress) was also included. These compounds, dissolved in DMSO, were diluted 1:1000 into the culture medium to a final concentration of 100 μM (or 50 μM for Cilofexor). Additionally, murideoxycholic acid (MDCA; catalog no. 700270P, Avanti), ω-muricholic acid (ω-MCA; catalog no. 700231P, Avanti), tauro-α-muricholic acid sodium salt (T-α-MCA; catalog no. 700243P, Avanti), tauro-β-muricholic acid sodium salt (T-β-MCA; catalog no. 700244P, Avanti), tauro-ω-muricholic acid sodium salt (T-ω-MCA; catalog no. 700245P, Avanti), and tauro-γ-muricholic acid sodium salt (THCA; catalog no. 700246P, Avanti) were similarly prepared and diluted to a final concentration of 10 μM. Cells were treated with the bile acids and/or FXR agonists, together with TNF-α (final concentration of 10 ng/mL), for 48 h. Following treatment, cell pellets were harvested for RNA extraction and subsequent RT-qPCR analysis to assess Madcam1 expression as previously described. PCR amplification and TIDE analysis Genomic DNA was isolated from transduced TSECs or MCA205 fibrosarcoma cell clones using the PureLink™ Genomic DNA Mini Kit (catalog no. K182002, Invitrogen). Polymerase chain reaction (PCR) was conducted to amplify the relevant regions using the Platinum™ SuperFi II PCR Master Mix (catalog no. 12368010, Invitrogen) with primers designed to generate approximately 500–800 bp fragments encompassing the target SNP (see STAR Methods). PCR products were purified with the PureLink™ PCR Micro Kit (catalog no. K310250, Invitrogen) and assessed by agarose gel electrophoresis to confirm clean, single-band amplification prior to Sanger sequencing (Eurofins Genomics). DNA editing efficiency was evaluated by uploading sequencing chromatograms to the Tracking of Indels by DEcomposition (TIDE) online platform (https://tide.nki.nl/)67 to calculate indel percentages. Decomposition windows and left boundaries were optimized according to the distance from the cut sites to maximize alignment quality for each target. Indel analysis was limited to a maximum of 10 indel sizes, with a significance threshold set at p < 0.001 for all analyses. In vitro T cell culture and bile acid exposure Mouse T cells were isolated from the spleens of C57BL/6J mice using the MojoSort™ Mouse CD3 T cell Isolation Kit (catalog no. 480031, BioLegend) and activated with Dynabeads Magnetic Beads (catalog no. 11452D, Gibco) according to the manufacturer’s instructions. Cells were cultured in complete RPMI 1640 medium supplemented with 50 μM 2-mercaptoethanol (catalog no. ES-007-E, MERCK) and 30 U/mL recombinant mouse IL-2 (carrier-free; catalog no. 575406, BioLegend). On day 3, activated T cells were resuspended and cultured for an additional 2 days in fresh medium containing 30 U/mL IL-2. Subsequently, CD3+ T cells were plated at 3 × 105 cells per well in 48-well plates and exposed overnight to bile acids dissolved in sterile H2O, including glycodeoxycholic acid (GDCA; catalog no. G9910, Sigma-Aldrich), taurodeoxycholic acid sodium salt (TDCA; catalog no. 580221, Millipore), sodium taurochenodeoxycholate (TCDCA; catalog no. T6260, Sigma-Aldrich), taurocholic acid sodium salt hydrate (TCA; catalog no. T4009, Sigma-Aldrich), or tauro-β-muricholic acid sodium (T-β-MCA; catalog no. HY-135103, MedChemExpress). Cells were then restimulated with eBioscience™ Cell Stimulation Cocktail (plus protein transport inhibitors; catalog no. 00-4975-93, Invitrogen) for 5 h, followed by Fc receptor blocking using TruStain FcX™ PLUS (anti-mouse CD16/32) antibody (catalog no. 156604, BioLegend) and viability staining with LIVE/DEAD™ Fixable Aqua dye (catalog no. L34966, Invitrogen). Surface and intracellular staining were performed using the following antibodies: Pacific Blue™ anti-mouse CD4 (catalog no. 100531, BioLegend), Brilliant Violet 650™ anti-mouse CD8a (catalog no. 100742, BioLegend), FITC anti-human/mouse Granzyme B (catalog no. 515403, BioLegend), PE/Cyanine7 anti-mouse IFN-γ (catalog no. 505826, BioLegend), and APC-conjugated Ki-67 (clone SolA15; catalog no. 17-5698-82, Invitrogen). Intracellular cytokine staining was carried out using the eBioscience™ Foxp3 Transcription Factor Staining Buffer Set (catalog no. 00-5523-00, Invitrogen) according to the manufacturer’s protocol. Cells were resuspended and analyzed on a CytoFLEX S flow cytometer (Beckman Coulter). Enterocloster spp. treatment with BA All three bacteria were incubated with 500 μM of either TCDCA, TBMCA, or GDCA. Each bacteria was grown in 2 mL of RCM medium overnight at 37 C. Each overnight culture’s OD600 was measured and for each strain, the OD was standardized to 0.05 in 1 mL of RCM. Samples were prepared in a 24 well plate containing 1 mL of RCM and 500 μM of one of the bile acids. The 24 well plates were incubated at 37 C for 24 h. After 24 h, each sample’s endpoint OD600 was measured and 300 μL of each culture was incubated in ice-cold methanol, vortexed briefly, and then sat on the benchtop overnight. The methanol extracts were then centrifuged and the supernatant was frozen at -80 C until it was run on the LC-MS/MS. Sterile RCM was used as a control and also received the bile acid treatment(s). Samples used for LC-MS/MS were prepared in a 96 well plate. Each well contained a 50 μL aliquot of each extracted culture and 50 μL of 50% methanol/phenol red (2.5 μg/mL). Injection blanks were prepared as 100% methanol and were the first well in each row. Quality control wells were located in column 12 of each row and contained 10 μL from each non-injection blank well in each row. Genome assembly and BSH identification of strain Enterocloster clostridioformis 141 For the Enterocloster clostridioformis strain 141, genomic sequences underwent rigorous quality control by Trimmomatic (version 0.27) before being assembled using MEGAHIT68 (version 1.2.9). Protein-coding sequences were predicted by Prodigal (version 2.6.1) from the resulting contigs. To identify bile salt hydrolase (bsh) genes, a BLASTp search was performed against the predicted protein sequences using known BSH sequences as queries. Distribution and diversity analysis of BSH in Enterocloster genomes To investigate the distribution of BSH isoforms across the Enterocloster genus, a total of 151 E. clostridioformis and 158 E. bolteae genomes were retrieved from the NCBI database (2026.03.01). Potential bsh gene sequences were initially identified using Bakta69 (version 1.6.1) genome annotation and subsequently filtered based on protein length, retaining sequences between 300 and 400 amino acids to ensure functional integrity. Representative BSH sequences were then clustered using MMseqs2 with a 95% identity threshold to analyze the diversity and prevalence of specific isoforms within each species. For phylogenetic reconstruction, multiple sequence alignment of the identified BSH sequences was performed using MUSCLE within the MEGA (version 11), followed by the construction of a Maximum Likelihood (ML) tree to delineate the evolutionary relationships between these BSH isoforms. Cancer cell viability after BA in vitro treatment HT29 cells were purchased and grown as per ATCC. HT29 cells were left untreated or were exposed to BA (LCA, CDCA, TDCA, GDCA, UDCA, DCA) at different dosages as indicated for 48 h. Mitoxantrone (1 and 3 μM; Sigma-Aldrich; catalog No. M6545) and oxaliplatine (250 and 500 μM; Accord), known ICD inducers, were the positive control.70 Cells were treated with the bile acids for 48 h, then stained with Hoechst (33342; H3570, Thermofisher; 1/5000) + propidium iodide (PI; P4864-10mL; Sigma; 1/1000) for 30 min at 37°C before acquisition with an ImageXpress XL automated microscope (Molecular Devices, Sunnyvale, CA, US), equipped with a 20× PlanApo objective (Nikon, Tokyo, Japan) with a minimum of 6 images per well. Nuclei were segmented using Hoechst fluorescence, and cell parameters (nuclear area, Hoechst and PI fluorescence intensities) were measured. Cell data were cleaned using a gating based on nuclear area and Hoechst intensity to separate “cells”, “healthy cells” and “debris”, and a threshold was determined on PI intensity to identify PI+ cells. Data were then summarized to a cell count and a percentage of PI cells by image. A one way unpaired Mann-Whitney U-test was performed. DNA extraction and 16S rRNA targeted sequencing of mouse stools Preparation and sequencing of mouse fecal samples were performed at IHU Méditerranée Infection, Marseille, France. Briefly, DNA was extracted using two protocols. The first protocol consisted of physical and chemical lysis, using glass powder and proteinase K, respectively, then processing using the Macherey-Nagel DNA Tissue Extraction Kit.71 The second protocol was identical to the first protocol, with the addition of glycoprotein lysis and deglycosylation steps.72 The resulting DNA was sequenced,73,74 targeting the V3–V4 regions of the 16S rRNA gene. Bioinformatics, statistics and reproducibility of 16S rRNA targeted sequencing data Microbial species which did not have a biological meaning (environmental, rumen, extra-mammals, food, etc.) were excluded from raw data. In addition, a strengthened prevalence cutoff of 5% was employed in order to enrobust the differential analysis (DA) thus resulting in 172 species that were considered for subsequent statistical analyses. Measurements of α diversity (within sample diversity) such as Richness, Shannon and Inverse Simpson indexes, were calculated at species level using the SciKit-learn package v1.0.1, starting from raw reads counts. For beta-diversity, data matrices were first transformed with pseudocount and centered-log-ratio (CLR), then normalized and standardized using QuantileTransformer and StandardScaler methods from Sci-Kit learn package v1.0.1. Normalization using the output_distribution = ‘normal’ option transforms each variable to a Gaussian-like shaped distribution, whilst the standardization results in each normalized variable having a mean of zero and variance of one (10.1080/2162402 × .2025.2484880). Exploratory analysis of β-diversity (between sample diversity) was calculated using the Bray-Curtis measure of dissimilarity and represented in principle Coordinate Analyses (PCoA), along with methods to compare groups of multivariate sample units (analysis of similarities - ANOSIM, permutational multivariate analysis of variance - PERMANOVA) to assess significance in data points clustering. ANOSIM and PERMANOVA were automatically calculated after 999 permutations, implemented with custom scripts (Python v3.8.2, Seaborn v0.11.2, SciKit-learn v1.0.1) (10.1080/2162402 × .2023.2237354). We implemented Partial Least Square Discriminant Analysis (PLS-DA) and the subsequent Variable Importance Plot (VIP) as a supervised analysis, wherein the VIP values (equal or higher than 1, default settings) are used to identify the most discriminant bacterial species among the cohorts. Mann–Whitney U test and Kruskal–Wallis test were employed to assess significance for pairwise or multiple comparisons, respectively, considering a p value ≤ 0.05 as significant. Clustermaps of microbial abundances were generated with Euclidean distance and Ward linkage algorithms by means of custom scripts (Python v3.8.2, Seaborn v0.11.2, SciKit-learn v1.0.1). Where clearly stated, all p values were corrected for multiple hypothesis testing using a two-stage Benjamini-Hochberg FDR at 10% (https://doi.org/10.1016/j.ejca.2023.112948). Quantification and statistical analysis Overall survival and progression-free survival analyses of patients were performed using the Kaplan-Meier estimator, and Cox regression in uni-or multivariate analyses (Table S4) and the patient groups were compared using the log rank (Mantel-Cox) test. In in vitro and in vivo experiments, unless otherwise stated, statistical significance was determined by non-parametric Mann-Whitney U test (two unpaired groups), Wilcoxon matched-pairs signed-rank test (two groups of paired samples), Kruskal-Wallis H test (more than two groups), correcting the statistical significance for multiple comparisons and FDR using the two-stage step-up method of Benjamini and Hochberg using GraphPad Prism 10 (GraphPad, San Diego, CA, USA) and the statistical environment R (http://www.R-project.org/). For tumor growth analysis, we used Tumor Growth available at https://kroemerlab.shinyapps.io/TumGrowth/. p-value and FDR p-value adjusted <0.05 was considered statistically significant. All values are mean ± (SD) or (SEM) as stated in the legends. The number of independent biological replicates for each experiment and the sample size of each experimental group or condition are provided in the figure legends. Published: July 29, 2026 Footnotes Supplementary data related to this article can be found online at doi:https://doi.org/10.1016/j.ccell.2026.06.022. Contributor Information Lisa Derosa, Email: lisa.derosa@gustaveroussy.fr. Guido Kroemer, Email: kroemer@orange.fr. Marine Fidelle, Email: marine.fidelle@gustaveroussy.fr. Laurence Zitvogel, Email: laurence.zitvogel@gustaveroussy.fr. Supplemental information Document S1. Figures S1–S11 Table S1. Metabolomics in mice, related to Figures 1–4 Table S2. Metagenomics changes after E. clostridioformis or BA treatment, related to Figure 2 Table S3. Single-cell RNA sequencing, related to Figure 4 and STAR Methods Table S4. Clinical clusters based on BA composition, related to Figures 5–8 and STAR Methods Table S5. Clinical table, related to Figures 5–8 and STAR Methods Table S6. MVA GGT ALP, related to Figures 5–8 and STAR Methods Document S2. Article plus supplemental information References Associated Data Supplementary Materials Document S1. Figures S1–S11 Table S1. Metabolomics in mice, related to Figures 1–4 Table S2. Metagenomics changes after E. clostridioformis or BA treatment, related to Figure 2 Table S3. Single-cell RNA sequencing, related to Figure 4 and STAR Methods Table S4. Clinical clusters based on BA composition, related to Figures 5–8 and STAR Methods Table S5. Clinical table, related to Figures 5–8 and STAR Methods Table S6. MVA GGT ALP, related to Figures 5–8 and STAR Methods Document S2. Article plus supplemental information Data Availability Statement The bulk RNA sequencing of murine hepatocytes, cholangiocytes, and enterocytes (refer to Figure S3) and the 16S sequencing data of murine fecal samples (refer to Figures S2 and S5) have been deposited in the ENA (European Nucleotide Archive) database under submission number ERA35991104, with project number ENA: PRJEB109926. Raw and processed data of the single-cell RNA sequencing of murine immune cells from tumors and tdLNs (refer to Figures 4 and S7) have been deposited in the Gene Expression Omnibus (GEO) under accession number GEO: GSE326943. Human and mouse metabolomic data supporting the claims of the manuscript are available in Mendeley Data: https://doi.org/10.17632/wsydn677rw.1. Human metagenomic data for S-score and Toposcore calculations for Figures 5F, 8A, and 8C are available in Mendeley Data: https://doi.org/10.17632/wsydn677rw.1. No new code was generated in this study. Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.