Microbial Bile Acid Metabolism Shapes Effector T Cell Responses during Inflammation in Mouse and Human

Microbial transformation of bile acids (BAs) regulates intestinal immune homeostasis but their effect on T cell-driven pathologies is unknown. We previously reported that gut dysbiosis in patients undergoing allogeneic hematopoietic cell transplantation (allo-HCT) is associated with increased graft-versus-host disease (GVHD), in which graft-derived T cells recognize the host as foreign. We hypothesized that the dramatic changes in intestinal microbiome composition during GVHD would lead to changes in the BA pool, which would alter immunoregulatory mechanisms governing T cell responses. Using a preclinical MHC-disparate GVHD model (B6 into BALB/c) and quantifying BAs with liquid chromatography-mass spectrometry, we found that GVHD was associated with a significant decrease in total BA concentration with a preferential loss of microbiota-dependent unconjugated primary BA (PBAs) and secondary BA (SBAs) (p=0.0002 and p<0.001, n=10/group). Metagenomic sequencing revealed changes in microbiota composition, including reduced abundance of Lactobacillus johnsonii and expansion of Enterococcus faecalis in GVHD mice (FDR=0.03, n=5/group) as previously reported. These changes were accompanied by a reduction in the abundance of bile salt hydrolase (BSH) genes (p=0.008) encoded by the gut microbiome (mainly Lactobacillus johnsonii). BSH catalyzes the deconjugation of PBAs, which is the gateway reaction leading to SBAs formation, suggesting that GVHD affects the BA pool by compromising early steps in microbial BA metabolism. To investigate the effects of GVHD on host metabolism, we profiled the liver transcriptome by bulk RNA-sequencing analysis. Pathways involved in BA metabolism were reduced in the livers of GVHD mice, which was associated with reduced levels of transcripts for genes involved in BA synthesis (Abcd3, Hsd17b11, Cyp7b1, Hsd17b6) and secretion (Abcg8, Slc22a18). As BAs regulate their synthesis via farnesoid X receptor (FXR) in a negative feedback loop, these results suggested that FXR activation may be increased because of loss of SBAs. Indeed, SBAs counteracted the activation of FXR by the natural agonist chenodeoxycholic acid, in FXR reporter assay and FXR TR-FRET coactivator assay, suggesting that bacterial transformation of BAs generally decreases FXR activity in vivo. Pharmacological activation of FXR (B6 into BALB/c, treatment with selective FXR agonist GW4064 day -1 to day 7, p<0.0001, n=20/group) showed increased mortality, whereas genetic ablation of FXR in donor T cells decreased GVHD and improved OS in two preclinical models (B6 into BALB/c, p<0.0001, n=30/group, Fig. 1a and B6 into 129, p<0.0001, n=20/group). Mice transplanted with TΔFXRcells had reduced GVHD-related pathology compared to recipients of TWT on day 28 post allo-HCT (p=0.006, n=10/group). Furthermore, we found reduced IFNγ production by CD4+ and CD8+ TΔFXR cells in the lamina propria of the large intestine on day 14 post-transplant (p=0.01 and p=0.02, n=10/group). Finally, we set out to corroborate our preclinical findings in allo-HCT patients. We compared 63 fecal samples from allo-HCT patients with lower gastrointestinal GVHD obtained at peri-GVHD onset (+/- 10 days) with 61 samples from control allo-HCT patients without GVHD, who were matched for age, graft type and HLA compatibility. Samples from controls were selected +/-10 days from pseudo-GVHD day that was assigned based on their paired GVHD patient. Most of the samples were exposed to the SBA ursodeoxycholic acid (UDCA, 93%). Similar to our findings in mice, we found that SBAs (non-UDCA, Fig. 1b) and unconjugated PBAs, as well as BSH abundances were decreased in GVHD patients compared to controls (p=0.007, p=0.008, and p=0.02 respectively). Notably, we found that important immunomodulatory SBAs were reduced in GVHD patients compared to controls, including 3-oxolithocholic acid, isolithocholic acid and ω-muricholic acid (p=0.01, p=0.004, and p<0.001). Lastly, in 747 allo-HCT recipients, we found that UDCA exposure was associated with a lower incidence of Enterococcus sp. domination (p=0.001), lower aGVHD cumulative incidence (p=0.005) and improved OS (p=0.017) as previously reported. Our results in mice and humans suggest that loss of SBA and increased FXR activation may amplify T-cell driven disease such as GVHD and provide a rationale to target the host-microbiota BA network to improve clinical outcomes in these patients. Figure 1View largeDownload PPTFigure 1View largeDownload PPT Close modal