The Bladder Microbiome, Metabolome, Cytokines, and Phenotypes in Patients with Systemic Lupus Erythematosus

Abstract

Emerging studies reveal unique bacterial communities in the human bladder, with alteration of composition associated to disease states. Systemic lupus erythematosus (SLE) is a complex autoimmune disease that is characterized by frequent impairment of the kidney. Here, we explored the bladder microbiome, metabolome, and cytokine profiles in SLE patients, as well as correlations between microbiome and metabolome, cytokines, and disease profiles. We recruited a group of 50 SLE patients and 50 individually matched asymptomatic controls. We used transurethral catheterization to collect urine samples, 16S rRNA gene sequencing to profile bladder microbiomes, and liquid chromatography-tandem mass spectrometry to perform untargeted metabolomic profiling. Compared to controls, SLE patients possessed unique bladder microbial communities and increased alpha diversity. These differences were accompanied by differences in urinary metabolomes, cytokines, and patients' disease profiles. The SLE-enriched genera, including Bacteroides, were positively correlated with several SLE-enriched metabolites, including olopatadine. The SLE-depleted genera, such as Pseudomonas, were negatively correlated to SLE-depleted cytokines, including interleukin-8. Alteration of the bladder microbiome was associated with disease profile. For example, the genera Megamonas and Phocaeicola were negatively correlated with serum complement component 3, and Streptococcus was positively correlated with IgG. Our present study reveals associations between the bladder microbiome and the urinary metabolome, cytokines, and disease phenotypes. Our results could help identify biomarkers for SLE. IMPORTANCE Contrary to dogma, the human urinary bladder possesses its own unique bacterial community with alteration of composition associated with disease states. Systemic lupus erythematosus (SLE) is a complex autoimmune disease often characterized by kidney impairment. Here, we explored the bladder microbiome, metabolome, and cytokine profiles in SLE patients, as well as correlations between the microbiome and metabolome, cytokines, and disease profiles. Compared to controls, SLE patients possessed a unique bladder microbial community and elevated alpha diversity. These differences were accompanied by differences in bladder metabolomes, cytokines, and patients' disease profiles. SLE-enriched genera were positively correlated with several SLE-enriched metabolites. SLE-depleted genera were negatively correlated to SLE-depleted cytokines. Alteration of the bladder microbiome was associated with disease profile. Thus, our study reveals associations between the bladder microbiome and the bladder metabolome, cytokines, and disease phenotypes. These results could help identify biomarkers for SLE.

Keywords: bladder microbiome; complement; disease profile; systemic lupus erythematosus; urinary cytokines; urinary metabolome.

FIG 1

Bacterial composition, diversity and phylum difference between controls and SLE group. (A) PCoA based on Bray-Curtis distances at species level showed different microbial compositions between groups of SLE patients and controls. The 95% confidence ellipse is drawn for each group. Permutational multivariate analysis of variance (PERMANOVA) was performed for statistical comparisons of samples in the two groups. P value was adjusted by the Benjamini and Hochberg false discovery rate (FDR). (B) Bacterial diversity measured by Shannon index was calculated at the bacterial species level. Wilcoxon rank-sum test was performed and adjusted by Benjamini and Hochberg false discovery rate (FDR). **, P_adj < 0.01. (C) Microbial profile at the phylum level. Only phyla with more than 1% average relative abundances in all samples are shown. (D) Bacterial phyla that were differentially abundant between controls and SLE patients. P value was calculated using Wilcoxon rank-sum test and adjusted by Benjamini and Hochberg FDR. *, P_adj < 0.05 and ***, P_adj < 0.001. (E) Firmicutes/Bacteroidetes ratio differed in controls and SLE patients. P value was calculated using Wilcoxon rank-sum test and adjusted by Benjamini and Hochberg FDR. ***, P_adj < 0.001.

FIG 2

Bacterial genera and species are different between controls and SLE group. (A) Bacterial genera that were more abundant in controls compared to SLE patients (P_adj < 0.05). P value was calculated using Wilcoxon rank-sum test and adjusted by Benjamini and Hochberg false discovery rate (FDR). (B) Bacterial genera that were less abundant in controls compared to SLE patients (P_adj < 0.05). P value was calculated using Wilcoxon rank-sum test and adjusted by Benjamini and Hochberg FDR. (C) Bacterial species that were more abundant in controls compared to SLE patients (P_adj < 0.05). P value was calculated using Wilcoxon rank-sum test and adjusted by Benjamini and Hochberg FDR. (D) Bacterial species that were less abundant in controls compared to SLE patients (P_adj < 0.05). P value was calculated using Wilcoxon rank-sum test and adjusted by Benjamini and FDR.

FIG 3

Urinary metabolome differed in SLE patients. (A) Separation of urinary metabolome between patients with SLE and controls, revealed by principal-component analysis (PCA). The explained variances are shown in brackets. Anosim was performed for statistical comparisons of samples in two groups. The 95% confidence ellipse is drawn for each group. (B) Partial least square-discriminant analysis (PLS-DA) plot. Scores plot between the selected PCs. The explained variances are shown in brackets. PERMANOVA was used to test statistical comparisons of ions in SLE and control groups. (C) The metabolites showing significant difference between the control and SLE groups. The metabolites described in the graph met the following criteria: P_adj < 0.05 in Wilcoxon rank-sum test variable importance in projection (VIP > 1) in PLS-DA; and fold change (FC) >2 or <0.5. (D) Receiver operating characteristic curve (ROC) curve for validation of metabolomic classification of control and SLE patients. Sensitivity is on the y axis, and specificity is on the x axis. The area-under-the-curve (AUC) is in blue.

FIG 4

Bladder microbiome was associated with metabolites The heatmap depicted the association between the taxa and metabolites that differ in SLE relative to controls. Spearman correlation analysis was performed on the abundant bacterial genera (>1% average relative abundances) and metabolites that differed between the healthy control (HC) and SLE groups. The correlation of two variables with values of |r|>0.3 and P < 0.05 are displayed. *, P < 0.05; **, P < 0.01; and ***, P < 0.001.

FIG 5

Urinary cytokines and disease profiles in SLE were associated with bladder microbiome. (A) Urinary cytokines increased and decreased in SLE group compared to controls. P value was calculated using Wilcoxon rank-sum test and adjusted by Benjamini and Hochberg false discovery rate. (B) Spearman correlation analysis was performed on the most abundant bacterial genera (>1% average relative abundances) and cytokines that differed between the controls and SLE groups. The correlation of two variables with values of |r|>0.3 and P < 0.05 are displayed. *, P < 0.05; **, P < 0.01; and ***, P < 0.001. (C) Spearman correlation analysis was performed on the abundant bacterial genera (>1% average relative abundances) and disease profiles of SLE patients. The correlation of two variables with values of |r|>0.3 and P < 0.05 are displayed. *, P < 0.05; **, P < 0.01; and ***, P < 0.001.