Metagenomic analysis revealed the potential role of gut microbiome in gout

Abstract

Emerging evidence indicates an association between gut microbiome and arthritis diseases including gout. However, how and which gut bacteria affect host urate degradation and inflammation in gout remains unclear. Here we performed a metagenome analysis on 307 fecal samples from 102 gout patients and 86 healthy controls. Gout metagenomes significantly differed from those of healthy controls. The relative abundances of Prevotella, Fusobacterium, and Bacteroides were increased in gout, whereas those of Enterobacteriaceae and butyrate-producing species were decreased. Functionally, gout patients had greater abundances for genes in fructose, mannose metabolism and lipid A biosynthesis, and lower for genes in urate degradation and short chain fatty acid production. A three-pronged association between metagenomic species, functions and clinical parameters revealed that decreased abundances of species in Enterobacteriaceae were associated with reduced amino acid metabolism and environmental sensing, which together contribute to increased serum uric acid and C-reactive protein levels in gout. A random forest classifier based on three gut microbial genes showed high predictivity for gout in both discovery and validation cohorts (0.91 and 0.80 accuracy), with high specificity in the context of other chronic disorders. Longitudinal analysis showed that uric-acid-lowering and anti-inflammatory drugs partially restored gut microbiota after 24-week treatment. Comparative analysis with obesity, type 2 diabetes, ankylosing spondylitis and rheumatoid arthritis indicated that gout metagenomes were more similar to those of autoimmune than metabolic diseases. Our results suggest that gut dysbiosis was associated with dysregulated host urate degradation and systemic inflammation and may be used as non-invasive diagnostic markers for gout.

Conflict of interest statement

The authors declare no competing interests.

© 2021. The Author(s).

Figures

Fig. 1. Gut microbial alterations in gout patients.

a The gene rarefaction curves based on the Chao2 estimated gene counts in healthy controls (n = 63) and gout patients (n = 77) (paired Wilcoxon rank-sum test for median gene counts P = 5.6e–12). b Box and whisker plot of gene count in the healthy controls and gout patients. Wilcoxon rank-sum test was used to determine significance. **P < 0.01. c, d Box and whisker plots of alpha diversity (Shannon index) and beta diversity (Bray–Curtis distance) at the gene level. Wilcoxon rank-sum test was used to determine significance. **P < 0.01. To exclude the influence of the various data sizes among the samples, panels a, b, d and d were based on 11 M matched reads per individual. e Principal component analysis (PCA) based on the gene relative abundance profile. The 95% confidence ellipses were shown for gout and control samples. f The Bacteroidetes/Firmicutes ratio (Wilcoxon rank-sum test; **P < 0.01). The relative abundance of differential phyla (g top 3), genera (h top 30) and species (i top 30) between gout patient and healthy control groups (FDR P < 0.05, Wilcoxon rank-sum test). The color bar above genera or species names were colored according to the phylum. For all box and whisker plots, the center line represents median. The bounds of box represent the first and third quartiles. The upper whisker extends from the hinge to the largest value no further than 1.5 * interquartile range (IQR) from the hinge. The lower whisker extends from the hinge to the smallest value at most 1.5 * IQR of the hinge. The notch represents a confidence interval around the median as the median ± 1.58*IQR/sqrt(n).

Fig. 2. Gout-associated microbial gene functions related to urate degradation.

a KEGG module for urate degradation. b Relative abundance of KOs involved in urate degradation. Significantly enriched KOs were identified by Wilcoxon rank-sum test, and the boxes or KO names were colored according to the direction of enrichment. Green, enriched in healthy controls (FDR P < 0.05). Boxes with no color or KO names with black, no difference; boxes with gray, not detected in samples. c Correlations between gout-associated genera and urate degradation-associated KOs (red and purple for positive and negative correlation, respectively). Spearman correlation test: ‘plus’ denotes FDR P < 0.05; ‘asterisk’ denotes FDR P < 0.01; ‘hash’ denotes FDR P < 0.001. The enrichment direction and family classification of genera were shown in left panel and the mean Spearman’s correlation coefficient of each genus with urate degradation-associated KOs was shown in the right panel. d The associations between SUA and Enterobacteriaceae or Klebsiella. Spearman’s rank correlation was calculated by taking the species relative abundance and SUA content. An inverse correlation was observed between SUA and Enterobacteriaceae and Klebsiella. For all box and whisker plots, the center line represents median. The bounds of box represent the first and third quartiles. The upper whisker extends from the hinge to the largest value no further than 1.5 * interquartile range (IQR) from the hinge. The lower whisker extends from the hinge to the smallest value at most 1.5 * IQR of the hinge. The notch represents a confidence interval around the median as the median ± 1.58*IQR/sqrt(n). hpxO FAD-dependent urate hydroxylase, uraH 5-hydroxyisourate hydrolase, hpxQ 2-oxo-4-hydroxy-4-carboxy-5-ureidoimidazoline decarboxylase, allB and hpxB allantoinase.

Fig. 3. The three-pronged association heatmap of bacterial species, KEGG pathways and clinical parameters.

The left panel denotes the significant Spearman correlations (FDR P < 0.05) between bacterial species and clinical indices with or without adjustment for BMI, colored by positive (red), negative (blue), or nonsignificant correlation (gray). The top panel denotes the significant association (Wilcoxon rank-sum test, FDR P < 0.05) between KEGG pathways and clinical indices with or without adjustment for BMI. The bottom panel denotes the functional category and directionality of enrichment in gout and controls for KEGG pathways colored by healthy control-enriched (green), gout patient-enriched (red) or no significant difference (gray). The right panel denotes the family-level taxonomy and directionality of enrichment in gout and controls for species colored by healthy control-enriched (green), gout patient-enriched (red), or no significant difference (gray). The centered heatmap denotes the median Spearman correlation coefficient between each species and all KOs within a given KEGG pathway, adjusted for background distribution by subtracting the median Spearman correlation coefficient between the species and all other KOs outside the pathway (red: positive correlation; purple: negative correlation). A Wilcoxon rank-sum test was performed between Spearman correlation coefficients between each species and all KOs within a given KEGG pathway and Spearman correlation coefficients between the species all other KOs outside the pathway. Wilcoxon rank-sum test: ‘plus’ denotes FDR P < 0.05; ‘asterisk’ denotes FDR P < 0.01, ‘hash’ denotes FDR P < 0.001.

Fig. 4. The gut metagenomic classifier for gout.

a The model was trained using relative abundance of microbial genes in discovery cohort. All microbial genes were first ranked based on their variable importance and then added sequentially into the model. The error curves were plotted for the five trials of 10-fold cross-validation in random forest classification as the number of genes increased. The black curve indicates the average cross-validation error of the five trials (in gray). The minimum error in the averaged curve plus the standard deviation at that point was used as the cutoff for feature selection. The model containing the smallest number of genes with an error below that cutoff was chosen as the optimal classifier. The red line marks the number of genes in the optimized model. b The relative abundance of three microbial gene markers in discovery and validation cohorts. Wilcoxon rank-sum test: ‘asterisk’ denotes FDR P < 0.05; ‘double asterisks’ denote FDR P < 0.01; ‘triple asterisks’ denote FDR P < 0.001. c Receiver operating curve (ROC) for the discovery samples. d ROC for the validation samples (healthy control, n = 23; gout patient, n = 25). e ROCs for gout and four public case-control metagenomic datasets for ankylosing spondylitis (AS), obesity (OB), rheumatic arthritis (RA), and type 2 diabetes (T2D) using three gout-associated gene markers. The AUC for each disease was shown in parenthesis. For all box and whisker plots, the center line represents median. The bounds of box represent the first and third quartiles. The upper whisker extends from the hinge to the largest value no further than 1.5 * interquartile range (IQR) from the hinge. The lower whisker extends from the hinge to the smallest value at most 1.5 * IQR of the hinge. The notch represents a confidence interval around the median as the median ± 1.58*IQR/sqrt(n).

Fig. 5. Alternation of gut microbiota by therapeutic intervention in gout.

a PCoA based on Bray–Curtis distance at gene level of healthy controls and gout patients before and after treatment (0W, n = 77; 2W, n = 61; 4W, n = 38; 24W, n = 7). The 95% confidence ellipses were shown for all subgroups. b PCoA based on Bray–Curtis distance at gene level of healthy controls and five time point paired gout patients. The 95% confidence ellipses were shown for all subgroups. c Box and whisker plot of beta diversity of gout patients between before and after treatment. Wilcoxon rank-sum test: Double asterisks denote P < 0.01. d The relative abundance of bacterial species was modulated after 24 weeks treatment (P < 0.05, paired Wilcoxon rank-sum test, n = 7). Bacterial species in color green and red indicate healthy control-enriched and gout patient-enriched in discovery cohort, respectively. e Microbial gene functions were changed after treatment. Purple, enriched in healthy controls or patients after treatment; red, enriched in patients before treatment. Asterisk denotes reporter score of pathways > 1.65 or < −1.65. For all box and whisker plots, the center line represents median. The bounds of box represent the first and third quartiles. The upper whisker extends from the hinge to the largest value no further than 1.5 * interquartile range (IQR) from the hinge. The lower whisker extends from the hinge to the smallest value at most 1.5 * IQR of the hinge. The notch represents a confidence interval around the median as the median ± 1.58*IQR/sqrt(n).

Fig. 6. Comparison of gut microbiome between gout and other autoimmune and metabolic diseases.

a The distribution of P values by Wilcoxon rank-sum test for all microbial genes in case-control comparison within each of the AS (n = 211), gout (n = 140), OB (n = 200), RA (n = 169), and T2D (n = 268) datasets. b Comparison of differential species in AS, gout, OB, RA and T2D. Purple, enriched in healthy controls; red, enriched in patients; Wilcoxon rank-sum test: asterisk denotes FDR P < 0.05. c Comparison of microbial gene functions in AS, gout, OB, RA and T2D. Purple, enriched in healthy controls; red, enriched in patients. Asterisk denotes reporter score of pathways > 1.65 or < −1.65.