Microbiota-mediated transformation of host-derived metabolites is a key mechanism by which the gut microbiome impacts human health. Bilirubin (BR), the byproduct of host heme degradation, accumulates in the gut and undergoes multi-step metabolism by intestinal bacteria. BR is protective in murine models of colitis, and its microbial metabolism is linked to reduced inflammation in humans with IBD. However, the organisms, genes, and metabolites in this pathway remain largely unknown. We hypothesized that the bilirubin-derived family of metabolites (BFMs) are host-derived, microbiota-modified molecules important to human health. We developed a combined computational and experimental pipeline to identify the bacteria, genes, and metabolites responsible for BR metabolism in the gut and to screen BFMs for mucosal immunoregulatory activity. Additionally, we screened the metabolites for immunoregulatory properties.

Our computational approach has three steps: identify candidate organisms by screening for covariance between species abundance and BFM-associated metabolic features; pinpoint genes using comparative genomics by comparing gene family carriage between species positively versus negatively associated with BFM abundance; and predict identities of unannotated, class-associated metabolites using molecular networking.

Our experimental pipeline validates and expands these predictions. Metabolically active bacteria were screened by LC-MS analysis of whole-cell assays. Gene candidates were identified by gene-trait matching from the phenotype screen and comparative transcriptomics to identify substrate responsive genes. Novel metabolites were identified from bacterial cultures by tandem mass spectrometry and structures confirmed by NMR.

Using this pipeline, we identified multiple novel metabolic capabilities: (1) Gut-associated bacteria broadly possess the ability to reduce BV to BR; (2) Unexpectedly clostridia containing BilR, a putative BR reductase, reduce BR to a novel intermediate, divinyl urobilinogen (dvUBg) instead of the canonical pathway intermediate mesobilirubin (MB); And (3) M. gnavus possesses a unique capacity to reduce dvUBg to a second novel intermediate, monovinyl urobilinogen (mvUBg), which is subsequently reduced to the canonical pathway metabolite urobilinogen (UBg). This demonstrates a novel, selective, and non-canonical two-step enzymatic route for BR reduction to UBg. Among the Bacteroides and Bacillota species studied we did not observe the canonical reduction pathway previously described of BR reduction to UBg via a mesobilirubin intermediate, or urobilin reduction to the downstream metabolites stercobilinogen (SBg) and stercobilin (SBn)

To evaluate immunoregulatory properties of BFMs, we treated murine bone marrow-derived monocytes with BFMs and measured cytokine expression with and without LPS. BR increased expression of pro-inflammatory cytokines, whereas BV and other downstream pathway metabolites (UBg, UBn, SBg, and SBn) suppressed many of these signals.

Our work identified novel bacterial species, genes, and metabolites driving bilirubin metabolism, a pathway strongly associated with gut inflammation. These results suggest BFM metabolism occurs via a distributed pathway across multiple species and is a more chemically diverse space than previously appreciated. The discovery that BR stimulates pro-inflammatory signals is counterintuitive, suggesting microbiota-mediated BFM metabolism is a key component of mucosal immunoregulation. Future work will validate candidate genes and use these findings to build bacterial consortia and genetically engineered organisms to mechanistically probe microbiota-modulated BFMs in gut inflammation. This integrated computational-experimental pipeline can also be applied to characterize similar microbiota-mediated metabolic pathways amenable to a semi-targeted metabolomics approach.