Microbiome-derived bile acid signatures in early life and their association with islet autoimmunity

Abstract

Emerging studies reveal that gut microbes can conjugate diverse amino acids to bile acids, known as microbially conjugated bile acids. However, their regulation and health effects remain unclear. Here, we analyzed early-life microbially conjugated bile acid patterns and their link to islet autoimmunity. We quantified 110 microbial bile acids in 303 stool samples collected longitudinally (3-36 months) from children who developed one or more islet autoantibodies and controls who remained autoantibody-negative. We identified distinct age-dependent trajectories of these bile acid amidates and correlated them with gut microbiome composition. We found that altered levels of ursodeoxycholic and deoxycholic acid conjugates were linked to islet autoimmunity as well as modulated monocyte activation in response to immunostimulatory lipopolysaccharide and Th17/Treg cell balance. These findings suggest that microbially conjugated bile acids influence immune development and type 1 diabetes risk.

Fig. 1. Outline of the study.

We analyzed MCBAs in a longitudinal series of stool samples collected at 3, 6, 12, 18, 24, and 36 months of age from children at HLA-conferred risk for TD, who later developed (i) multiple islet autoantibodies (P2Ab), (ii) single islet autoantibody (P1Ab), or (iii) remained autoantibody negative (CTRs) during the follow-up. The graphical illustration for the in vitro assay, microbes (marked in red), colon, mass spectrometer instrument was created in BioRender. Lamichhane, S. (2025) https://BioRender.com/ crck5np.

Fig. 2. Presence of microbial conjugated bile acids (MCBAs) in fecal DIABIMMUNE samples.

The bar in the color map corresponds to the number of fecal samples (n = 303) in which MCBAs were detected. CA cholic acid, CDCA chenodeoxycholic acid, DCA deoxycholic acid, HDCA hyodeoxycholic acid, UDCA ursodeoxycholic acid. The amino acids are Alanine (Ala), Arginine (Arg), Asparagine (Asn), Aspartic Acid (Asp), Cysteine (Cys), Glutamic Acid (Glu), Glycine (Gly), Histidine (His), Isoleucine (Ile), Leucine (Leu), Lysine (Lys), Methionine (Met), Phenylalanine (Phe), Proline (Pro), Serine (Ser), Threonine (Thr), Tryptophan (Trp), Tyrosine (Tyr), and Valine (Val). L-DOPA L-3,4-dihydroxyphenylalanine.

Fig. 3. Trajectories of MCBA in early life.

The loess curve plot of MCBAs over time (n = 303, fecal samples over 3, 6, 12, 18, 24, and 36 months) for a CA-conjugates, b CDCA-conjugates, c DCA-conjugates, and d UDCA-conjugates. The shaded area around each smooth line 3a–d is the 95% confidence interval for the smoothed curve. CA cholic acid, CDCA chenodeoxycholic acid, DCA deoxycholic acid, HDCA hyodeoxycholic acid, UDCA ursodeoxycholic acid.

Fig. 4. MCBAs in progression to islet autoimmunity.

Forest plot illustrating the coefficient estimate of a linear mixed-effects model for individual MCBAs species, with fixed covariates of case a CTR (n = 131) vs b P1Ab (n = 73), CTR (n = 131) vs P2AB (n = 38), age, sex, and length of breastfeeding accounting for random effects within individual samples (n = 242). Filled circles with corresponding confidence intervals (error bars) represent significant MCBA species (p < 0.05), while faded circles depict non-significant MCBAs. The p-values shown are nominal; adjusted p-values (corrected for multiple comparisons using the Benjamini–Hochberg method) are available in Supplementary Table S2. c–f The loess curve plot of MCBAs over time for significant MCBA species obtained in the linear mixed-effects model (p < 0.05). The shaded area around each smooth line 3a–d is the 95% confidence interval for the smoothed curve. CA cholic acid, CDCA chenodeoxycholic acid, DCA deoxycholic acid, HDCA hyodeoxycholic acid, UDCA ursodeoxycholic acid. The amino acids are Alanine (Ala), Asparagine (Asn), Cysteine (Cys), Isoleucine (Ile), Leucine (Leu), Lysine (Lys), Proline (Pro), Serine (Ser), Tyrosine (Tyr), and Valine (Val).

Fig. 5. Cross-correlation between the microbes and stool levels of selected MCBAs.

Heatmap showing the correlation coefficients of association between microbes and stool levels of MCBAs that were found altered in progression to islet autoimmunity, in a subset of the sample with available metagenomics data (n = 110). Red color represents positive correlations, while blue represents negative correlations, as determined by Spearman's rank correlation. Only correlation coefficients with nominal p-values less than 0.05 are shown. The adjusted p-values are shown in Supplementary Table S7.

Fig. 6. Effects of secondary bile acids on LPS-induced signaling in human monocytes.

a Time-course analysis of LPS-stimulated pp38 and pERK and degradation of IkBa, markers of canonical NF-kB activation, in primary human monocytes (n = 6). LPS (1 ng/mL) induced robust activation of pp38/pERK and IkBa degradation within 60 min. Geometric MFI was measured by flow cytometry, and the average of n = 6 was plotted in time curves. Quantitative differences among treatments were shown by AUC analysis. Geometric MFI values were used for AUC calculations. “Inhibitors” serve as a positive control for pathway inhibition, and “No treatment” represents the baseline LPS response. b UDCA and its amino-acid conjugates (Trp-UDCA, Asp-UDCA, Glu-UDCA, Asn-UDCA, Cit-UDCA) show limited or no inhibition of these pathways. c DCA consistently inhibits signaling across all three pathways. DCA demonstrates significant suppression of LPS-induced signaling. AUC values were analyzed by one-way ANOVA followed by post-hoc pairwise comparisons vs no treatment using the general linear hypothesis test (glht, R package multcomp). All tests were two-sided, with p-values adjusted for multiple comparisons using the Benjamini–Hochberg false-discovery-rate (FDR) method. Significant effects included inhibition of pp38 by Cit-UDCA (p = 0.0065), UDCA (p = 0.00011), and DCA (p = 7.37 × 10⁻⁵) and activation by Asn-UDCA (p = 7.20 × 10⁻¹⁰); activation of pERK by Asn-UDCA (p = 0.00031) and inhibition by DCA (p = 0.0378); elevation of IκBα by Asn-UDCA (p = 3.11 × 10⁻¹³) and reduction by UDCA (p = 0.0309) and DCA (p = 0.0185). Significance: ^***p < 0.001; ^**p < 0.01; ^*p < 0.05. Illustrations were Created in BioRender. Tadepally, L. (2025) https://BioRender.com/5e1kt0e. DCA deoxycholic acid, UDCA ursodeoxycholic acid. The amino acids are Citrulline (Cit), Asparagine (Asn), and Glutamic Acid (Glu).

Fig. 7. MCBAs modulate Th17 and Treg cell differentiation.

a Schematic of the Th17 and iTreg differentiation protocol for primary human naïve CD4⁺CD25⁻ T cells isolated from the umbilical cord blood of healthy neonates. CD4⁺CD25⁻ T cells were activated with anti-CD3/anti-CD28 and differentiated into Th17 or iTreg cells in the presence of corresponding cytokines for 3 days. DMSO control or conjugated bile acids (Asn-UDCA, Ser-CDCA, Tyr-CDCA, and unconjugated UDCA) at 100 µM were added on day 0 of differentiation. Illustrations in a panel were created with BioRender.com (licensed). b–e IL-17a secretion in the supernatant of Th17 cultures treated with Asn-UDCA (b), Ser-CDCA (c), Tyr-CDCA (d), and UDCA (e) was quantified on day 3 of differentiation from four biological replicates using ELISA. f–i. Intracellular Foxp3 protein expression in iTregs cultured with Asn-UDCA (f), Ser-CDCA (g), Tyr-CDCA (h), or UDCA (i) was assessed on day 3 of differentiation by flow cytometry. Geometric mean fluorescence intensity (MFI) values are shown for six biological replicates. Statistical significance was determined using a paired, two-tailed Student’s t-test. Illustrations were Created in BioRender. Hirvonen, K. (2025) https://BioRender.com/5s3bnh7. Asn–UDCA asparagine–ursodeoxycholic acid, Ser–CDCA serine–chenodeoxycholic acid, Tyr–CDCA tyrosine–chenodeoxycholic acid, UDCA ursodeoxycholic acid.