Gut microbial β-glucuronidases influence endobiotic homeostasis and are modulated by diverse therapeutics

Abstract

Hormones and neurotransmitters are essential to homeostasis, and their disruptions are connected to diseases ranging from cancer to anxiety. The differential reactivation of endobiotic glucuronides by gut microbial β-glucuronidase (GUS) enzymes may influence interindividual differences in the onset and treatment of disease. Using multi-omic, in vitro, and in vivo approaches, we show that germ-free mice have reduced levels of active endobiotics and that distinct gut microbial Loop 1 and FMN GUS enzymes drive hormone and neurotransmitter reactivation. We demonstrate that a range of FDA-approved drugs prevent this reactivation by intercepting the catalytic cycle of the enzymes in a conserved fashion. Finally, we find that inhibiting GUS in conventional mice reduces free serotonin and increases its inactive glucuronide in the serum and intestines. Our results illuminate the indispensability of gut microbial enzymes in sustaining endobiotic homeostasis and indicate that therapeutic disruptions of this metabolism promote interindividual response variabilities.

Keywords: FDA-approved drugs; X-ray crystallography; hormones; interindividual response variability; metabolism; metagenomics; metaproteomics; multi-omics; neurotransmitters; β-glucuronidase.

Figure 1.. Differences glucuronide and aglycone profiles in germ-free mice and conventional mice.

(A) Over-arching mechanism by which GUS enzymes are hypothesized to modulate endobiotic homoeostasis. (B) Workflow for comparing profiles of inactive glucuronides and active aglycones between germ-free (GF) and conventional mice. (C) Heatmap reflecting relative levels of inactive glucuronides and active aglycones in GF and conventional mice. (D) Volcano plot reflecting significance of differences in levels of inactive glucuronides and active aglycones between GF and conventional mice. Levels of neurotransmitters (E-F) and hormones (G-H) are diminished in GF mice relative to conventional mice. (I) Neurotransmitter- and hormone-glucuronides chosen for further examination. (J) Catalytic efficiencies for reactivation of neurotransmitter- and hormone-glucuronides by a panel of diverse purified bacterial GUS enzymes organized by structural class. ** P ≤ 0.01, *** P ≤ 0.001. See also Figure S2.

Figure 2.. Neurotransmitter and hormone reactivation by human fecal lysates.

(A) Probe-enabled activity-based proteomics pipeline and substrate reactivation pipelines for GUS enzymes in fecal samples. (B) Proteomic abundance of GUS enzymes in human fecal samples by structural class. (C) Heatmap reflecting average rate of hormone and neurotransmitter reactivation by human fecal samples in nanomolar per second, arranged from left to right in order of greatest average endobiotic processing rate. N=3 biological replicates. (D) Cladogram of GUS proteins identified across cohort with branch colors indicating structural class. Protein clusters that comprise correlations are denoted by branch-ending shapes. (E) Linear regression demonstrating association between proteomic abundance of Loop 1 GUS and rate of reactivation of estradiol. (F-H) Linear regressions demonstrating association between proteomic abundance of FMN GUS and rate of reactivations for estradiol (F), estrone (G), and thyroxine (H). (I) Linear regression demonstrating association between proteomic abundance of FMN GUS enzymes denoted as triangles on cladogram (D) and rate of reactivation of dopamine. (J) Dopamine reactivation versus Lachnospiraceae metagenomic abundance. (K) Linear regression demonstrating association between proteomic abundance of FMN GUS enzymes denoted as squares on cladogram (D) and rate of reactivation of serotonin. (L) Serotonin reactivation versus Clostridia metagenomic abundance. P values represent confidence in a slope that is significantly non-zero as determined by the Wald Test, ρ represents Spearman’s rank correlation coefficient. See also Figures S1, S3–S4 and Tables S1–S12.

Figure 3.. Inhibition of Purified GUS Enzymes by Piperazine and Piperidine Compounds.

(A) Schematic of synthetic targeted GUS inhibitors with (purple) and without (blue) a piperazine moiety. (B) Inhibition of Purified Loop 1 and FMN GUS Enzymes by FDA-approved drugs and synthetic GUS inhibitors containing cyclic piperazine or piperidine moieties. N = 3 biological replicates. (C) Co-crystal structures of 3-OH desloratidine (C; green), ceritinib (D; purple), and norquetiapine (E; coral) covalently linked to glucuronic acid (blue) bound to E. eligens GUS with catalytic residues shown in gray. (F) Oligomers of co-crystal structures of norquetiapine (coral) covalently linked to glucuronic acid (blue) bound to E. eligens GUS and R. hominis 2 GUS with catalytic residues shown in gray with unique active site features labelled. See also Figure S5 and Table S11.

Figure 4.. Inhibition of GUS Enzymes in Human Fecal Samples.

(A) Inhibition of GUS activity in fecal lysates collected from fourteen healthy donors by a panel of synthetic GUS inhibitors and FDA-approved drugs, arranged from left to right in order of greatest average percent inhibition. N = 3 average of biological replicates. (B) sPLS-DA plot of metaproteomic GUS profile as a function of average inhibition. (C) Regression relating total metaproteomic GUS abundance to percent inhibition. (D) Regression statistics for Roseburia GUS metaproteomic abundance versus percent inhibition. (E) sPLS-DA plot comparing metagenomic GUS profiles as a function of average inhibition. (F) Average inhibition versus Roseburia metagenomic abundance. (G) Overlay of AlphaFold monomers for Roseburia GUS enzymes across the cohort metagenome. (H) Workflow for GUS inhibition study in healthy mice. Cecum (I) and Serum (J) levels of serotonin glucuronide for each treatment group. See also Figure S6.