Gut microbiota dysbiosis orchestrates vitiligo-related oxidative stress through the metabolite hippuric acid

Abstract

Background: Vitiligo, a depigmenting autoimmune skin disease characterized by melanocyte dysfunction or death, is known to be associated with an imbalance in gut microbiota. Oxidative stress plays a critical role in the pathogenesis of vitiligo. However, the complex promising interaction between abnormal accumulation of reactive oxygen species (ROS) in the skin and gut microbiota has remained unclear.

Results: Here, we compared transcriptome data of vitiligo lesions and normal skin and identified a high expression of oxidative stress-related genes in vitiligo lesions. We also established a vitiligo mouse model and found that the presence of gut microbiota influenced the expression of ROS-related genes. Depletion of gut microbiota reduced abnormal ROS accumulation and mitochondrial abnormalities in melanocytes, significantly improving depigmentation. Our findings from manipulating gut microbiota through cohousing, fecal microbiota transplantation (FMT), and probiotic supplementation showed that transferring gut microbiota from mice with severe vitiligo-like phenotypes exacerbated skin depigmentation while probiotics inhibited its progression. Targeted metabolomics of fecal, serum, and skin tissues revealed gut microbiota-dependent accumulation of hippuric acid, mediating excessive ROS in the skin. Elevated serum hippuric acid levels were also confirmed in vitiligo patients. Additionally, a microbiota-dependent increase in intestinal permeability in vitiligo mice mediated elevated hippuric acid levels, and we found that hippuric acid could directly bind to ROS-related proteins (NOS2 and MAPK14).

Conclusions: Our results suggested the important role of gut microbiota in regulating vitiligo phenotypes and oxidative stress. We identified hippuric acid, a gut microbiota-host co-metabolite, as a critical mediator of oxidative stress in vitiligo skin and its binding targets (NOS2 and MAPK14), resulting in oxidative stress. Validation in a small human cohort suggested that hippuric acid could serve as a novel diagnostic biomarker and therapeutic target for vitiligo. These findings provided new insights into how gut microbiota regulates skin oxidative stress in vitiligo and suggested potential treatment strategies for the disease. Video Abstract.

Keywords: Gut microbiota; Hippuric acid; Oxidative stress; Vitiligo.

Fig. 1

Integrative transcriptomic analysis of lesional vs. non-lesional skin in vitiligo patients from the GEO Database. A Schematic diagram of workflow of transcriptomic analysis from datasets GSE65127 and GSE75819. B Principal component analysis (PCA) was performed on normalized gene counts comparing lesional and non-lesional skin samples from vitiligo patients (n = 25). C Number of upregulated and downregulated DEGs identified in lesional versus non-lesional skin across the vitiligo patients. D Heatmap displaying a subset of DEGs (Padj < 0.05 show with a red circle) in lesional versus non-lesional skin. The list of genes with symbols (left) indicates their functional annotation (top left). Each column represents a biological replicate, and each row represents a gene. E Significant Gene Ontology (GO) terms enriched with the respective − log10(FDR). The list of terms with symbols (left) indicates their functional annotation (bottom left)

Fig. 2

Tail skin transcriptome in the melanoma-regulatory T cell (Treg)-induced vitiligo mouse model. A Schematic diagram of the melanoma/Treg-induced vitiligo mouse model (created with BioRender.com). B Representative hair coat images from control and vitiligo-induced mice post-induction at days 35 and 60. C Quantification of the percentage of depigmented skin area on the backs of vitiligo mice. Mean and SEM, 3 independent experiments including control (n = 9) and vitiligo mice (n = 9). P_adj = 0.0428 for no induction versus vitiligo day 35, P_adj < 0.0001 for no induction versus vitiligo day 35, and P_adj = 0.0428 for vitiligo day 35 versus vitiligo day 60. (D) Representative epidermal of tail skin images of vitiligo mice at day 35 post-induction, including bright field images and wholemount immunofluorescent staining showing Melan-A⁺ melanocytes and CD8⁺ T cells. E, F Quantification of Melan-A⁺ melanocytes and CD8⁺ T cells in the epidermis of control (n = 9) and vitiligo mice (n = 9). G Principal component analysis (PCA) on normalized gene counts of tail skin samples isolated from control (n = 9) and vitiligo-induced (n = 9) mice. H Number of upregulated and downregulated differentially expressed genes (DEGs) in vitiligo-induced versus control mice. I Heatmap displaying a subset of DEGs in vitiligo-induced versus control mice. Each column represents a biological replicate, and each row represents a gene. J Significant Gene Ontology (GO) terms enriched with the respective − log10(FDR) and gene percent contained. The number and percentage of genes per term are shown. Statistical significance was determined using (C) Kruskcal-Wallis followed by Dunn’s post hoc test or (E and F) unpaired Student’s t-test. *P < 0.05, ****P < 0.0001

Fig. 3

Gut microbiota contributes to vitiligo-related oxidative stress in the skin and mitochondrial dysfunction in melanocytes. A Schematic diagram showing oral administration of antibiotics (ABX) for 2 weeks followed by the vitiligo induction procedure. Subsequent experiments were conducted on day 35 (created with BioRender.com). B Representative confocal images showing dihydroethidium (DHE) fluorescence in tail skin sections from vitiligo mice treated with water and ABX. Scale bars, 200 µm. C Quantification of mean DHE fluorescence intensities indicating reactive oxygen species (ROS) levels. Mean and SEM, 3 independent experiments including vitiligo mice treated with water (n = 6) and ABX (n = 6). P_adj = 0.0372 for non-induction-water versus vitiligo-water, P_adj = 0.0001 for non-induction-ABX versus vitiligo-water, and P_adj = 0.0481 for vitiligo-water versus vitiligo-ABX. D Quantification of depigmentation in back skin. Mean and SEM, 3 independent experiments including vitiligo mice treated with water (n = 9) and ABX (n = 9). E Representative electron micrographs of tail skin from vitiligo mice treated with water and ABX. Blue arrowheads, healthy; red arrowheads, abnormal. Scale bar, 500 nm. F, G Principal component analysis (PCA) on normalized gene counts of tail skin samples isolated from non-induction and vitiligo mice treated with water (n = 6, 6) and ABX (n = 6, 6). H Number of upregulated and downregulated differentially expressed genes (DEGs) in vitiligo mice treated with water versus ABX. I Heatmap showing ROS-related genes in the tail skin of non-induction and vitiligo mice treated with water and ABX. Each column represents a biological replicate. Statistical significance was determined using. C Two-way ANOVA followed by Tukey’s post hoc test or D unpaired Student’s t-test. *P < 0.05, ***P < 0.001

Fig. 4

Signature and regulation of gut microbiota in the melanoma-Treg-induced vitiligo mouse. A Alpha diversity indices of the genus level for non-induction (n = 9) and vitiligo (n = 9) mice, assessed using the Shannon index. B Principal coordinate analysis (PCoA) of overall gut microbiota based on unweighted UniFrac distance for non-induction (n = 6) and vitiligo (n = 9) mice. C Relative abundance of the top 3 taxa at the class and order levels for non-induction (n = 6) and vitiligo (n = 9) mice. D Bar chart showing the relative abundance of gut microbiota at the order level in fecal samples. Each column represents a biological replicate. E Linear discriminant analysis effect size (LEfSe) analysis indicating significantly different bacterial taxa abundance between non-induction and vitiligo mice. F Schematic diagram of the co-housing and fecal microbiota transplantation (FMT) experiments (created with BioRender.com). G Schematic diagram of the probiotic supplement treatment experiments (created with BioRender.com). H Bar chart showing the relative concentration of taxa at the class level in FMT (n = 6) and donor mice (n = 4). Each column represents a biological replicate. I Analysis of depigmentation of the back skin from sep-2 M and co-2 M mice after the vitiligo induction procedure (from days 14 to 42, n = 5). Data are presented in kinetic line plots showing mean and SEM. J Analysis of depigmentation of the back skin from control and FMT mice after the vitiligo induction procedure (from days 14 to 42, n = 5). Data are presented in kinetic line plots showing mean and SEM. K Analysis of depigmentation of the back skin from vehicle and probiotic-supplement-treated mice after the vitiligo induction procedure (from days 35 to 63, n = 5). Data are presented in kinetic line plots showing mean and SEM. Statistical significance was determined using (D, I, J, and K) Mann–Whitney U test. *P < 0.05, **P < 0.01, ***P < 0.001

Fig. 5

Microbiota and vitiligo-related regulation of fecal, serum, and skin metabolites. A Schematic diagram illustrating the targeted quantitative metabolomics analysis conducted on fecal, serum, and skin tissue from non-induction and vitiligo mice treated with water and ABX (created with BioRender.com). B Bar chart showing the relative concentration of metabolite classes in fecal, serum, and skin tissue from non-induction (n = 6) and vitiligo mice (n = 6). C Volcano plots depicting differentially abundant metabolites from targeted quantitative metabolomics analysis of fecal (left: n = 6, 6), serum (middle: n = 6, 6), and skin tissue (right: n = 6, 6) samples from control and vitiligo mice. x-axis: − log10(FC). y-axis: − log10(P). Purple, upregulated; green, downregulated. D Venn diagram of differentially abundant metabolites from fecal (67), serum (99), and skin (40) samples, P < 0.05; intercept (8). E Heatmap of intercepted differentially abundant metabolites in skin samples from non-induction and vitiligo mice treated with water (n = 6, 6) and ABX (n = 3, 3). Each column represents a biological replicate. F Bar chart showing the relative concentration of 2-phenylpropionate, hippuric acid, and indoleacetic acid in fecal, serum, and skin samples from non-induction (n = 6) and vitiligo mice (n = 6). G Quantification of 2-phenylpropionate, hippuric acid, and indoleacetic acid in skin tissues of vitiligo mice treated with water (n = 6) and ABX (n = 3). H Quantification of 2-phenylpropionate, hippuric acid, and indoleacetic acid in fecal samples of vitiligo mice treated with water (n = 6) and ABX (n = 3). Statistical significance was determined using (G and H) Mann–Whitney U test. *P < 0.05

Fig. 6

Hippuric acid leading to microbiota-modulated skin ROS accumulation in vitiligo. A Schematic diagram illustrating the metabolite treatment protocol. Non-induction and vitiligo mice were treated intraperitoneally with 2-phenylpropionate (2-PP), hippuric acid (HA), or indoleacetic acid (IAA) daily from days 12 to 35 during the vitiligo induction procedure (created with BioRender.com). B, C Representative confocal images and quantification of mean dihydroethidium (DHE) fluorescence intensities showing reactive oxygen species (ROS) levels in tail skin sections from non-induction and vitiligo mice treated with vehicle, 2-PP, HA, or IAA, n = 4 mice/group. Scale bars, 200 µm. Mean and SEM, 3 independent experiments. D Quantification of depigmentation of the back skin from vitiligo mice treated with vehicle, 2-PP, HA, or IAA, n = 4 mice/group. Mean and SEM, 3 independent experiments. E Quantification of HA levels in skin tissues of wild-type mice treated intraperitoneally with vehicle (n = 6) and HA (n = 6). F Heatmap of ROS-related genes in the tail skin of wild-type mice treated intraperitoneally with vehicle (n = 3) and HA (n = 5). Each column represents a biological replicate. G Significant Gene Ontology (GO) terms enriched with the respective − log10(FDR) and gene percent contained. The number and percentage of genes per term are shown. H Volcano plot of differentially expressed genes (DEGs) of tail skin from non-induction versus vitiligo mice in Fig. 2 with HA-specific genes (orange labeled). Statistical significance was determined using (C and D) one-way ANOVA followed by Tukey’s post hoc test or E unpaired Student’s t-test. *P < 0.05, **P < 0.01

Fig. 7

Disruption of the gut–blood barrier in vitiligo aggravates HA rise in the skin and direct binding to protein NOS2 and MAPK14. A Representative histological images of the small intestine using H&E and alcian blue staining in non-induction and vitiligo mice. Scale bars, 50 µm. B Quantification of small intestine thickness in non-induction (n = 6) and vitiligo mice (n = 6). Mean and SEM. C Quantification of the number of goblet cells per unit length in the small intestine from non-induction (n = 6) and vitiligo mice (n = 6). Mean and SEM. D Intestinal permeability was assessed by measuring the percentage of fluorescent FITC-dextran (4 kDa) translocation into the circulation following oral gavage. Mean ± SEM from three independent experiments, including non-induction and vitiligo mice treated with water (n = 4) or ABX (n = 4). E Measurement of HA levels translocated into circulation 4 h post-oral gavage in non-induction and vitiligo mice treated with water and ABX, n = 4 mice/group. Mean and SEM. P_adj = 0.0004 for non-induction-water versus vitiligo-water, P_adj < 0.0001 for non-induction-ABX versus vitiligo-water, and P_adj < 0.0001 for vitiligo-water versus vitiligo-ABX. F Schematic diagram illustrating the reverse virtual molecular docking using whole and part databases (created with BioRender.com). G Ridgeline plot showing the smoothed density distribution of binding proteins with their binding scores. H Significant Gene Ontology (GO) terms enriched with the respective − log10(FDR). I The proteins with binding scores (absolute value > 7) from the whole and part databases based on simulated molecular docking are shown. J Network diagram illustrating oxidative stress-related proteins with binding scores (absolute value > 7) from the simulated molecular docking. K Differential scanning fluorimetry (DSF) analysis showing the effects of various concentrations of hippuric acid on the melting temperature (T_m) of proteins (NOS2, MPO, and MAPK14) with protein concentrations ranging from 0 to 500 µM. L Surface plasmon resonance (SPR) analysis demonstrated the direct interaction between hippuric acid and the proteins (NOS2, MPO, and MAPK14). Statistical significance was determined using unpaired t test (B, C) or two-way ANOVA test (D, E) followed by Tukey’s post hoc test. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001