Distinct Responses of Gut Microbiota to Jian-Pi-Yi-Shen Decoction Are Associated With Improved Clinical Outcomes in 5/6 Nephrectomized Rats

Abstract

Gut dysbiosis contributes to the development and progression of chronic kidney disease (CKD) and its complications. However, the effect of drugs on the gut microbiota of CKD patients and its influence on treatment outcomes remains to be explored. Here, we assessed whether the response of gut microbiota to the traditional Chinese medicine Jian-Pi-Yi-Shen (JPYS) decoction differed from that to piperazine ferulate (PF), a kidney-targeted drug, by 16S rDNA sequencing, and whether the difference could be linked with drug-specific clinical outcomes. We showed that both JPYS and PF improved renal function, but only JPYS was able to restore the blood reticulocyte counting and serum calcium level in CKD rats. We also found that weighted UniFrac beta-diversity of the gut microbiome of the JPYS treated rats was significantly different from that of PF. Microbiome markers of drug-specific response were identified and subjected to correlation network analysis, together with clinical parameters and KEGG pathways. Among the microbiome markers of CKD, Corynebacterium was found to form a network hub that was closely correlated with the JPYS responder Enterococcus, suggesting a potential indirect impact of JPYS on Corynebacterium via interspecies interactions. We also identified two network hubs of the PF responder Blautia and the JPYS-only marker Coprococcus, which were connected with many genera and clinical parameters. They might serve as keystone taxa driving the response of gut microbiota to the drugs and influence host outcomes. Moreover, the JPYS-only marker Clostridium_XIVb was found to be connected to many pathways that are associated with CKD progression and might account for the improved outcomes in the JPYS treated rats. At last, the identified keystone markers of drug response were validated by qPCR for their differential abundance between CKD and the two drugs. Taken together, our study revealed that the responses of gut microbiota to JPYS were distinct from that to PF, and pinpointed drug-specific keystone microbiome markers closely correlated to clinical parameters, which could serve as candidate microbiome targets for further studies on their roles in medicating the drug efficacy of TCM in CKD.

Keywords: 5/6 nephrectomized rats; Jian-Pi-Yi-Shen decoction; chronic kidney disease; gut microbiota; piperazine ferulate.

Figure 1

Distinct host responses to JPYS and PF in the CKD rats. (A) Study design to evaluate the responses of the CKD rats and their gut microbiota to JPYS or PF. (B) Principal component analysis (PCA) of clinical parameters of the sham-operated, CKD, JPYS and PF rats after 12 weeks of treatment (Sham: n = 9, CKD: n = 9, JPYS: n = 10 and PF: n = 10). Overall clinical variation of the four treatment groups was assessed along PC1 and PC2 respectively. (C) Variation of clinical parameters of kidney function in each group. (D) Variation of clinical parameters of CKD complications in each group. Statistical significance of variation between groups was tested by Wilcoxon rank-sum test with FDR correction. Only significant differences between groups (p < 0.05) were annotated. JPYS, Jian-Pi-Yi-Shen decoction; PF, Piperazine ferulate; CKD, chronic kidney disease; BUN, blood urea nitrogen; U-ALB, urinary albumin; RET, reticulocyte; S-Ca, serum calcium levels.

Figure 2

Distinct responses of gut microbiome to JPYS and PF in the CKD rats. (A) Weighted UniFrac-based principal coordinates analysis (PCoA) of the gut microbiome of the sham-operated, CKD, JPYS, and PF rats after 12 weeks of treatment. Statistical significance of variation of the microbiomes was assessed along PC1 and PC2 respectively, tested by Wilcoxon rank-sum test with FDR correction. (B) Relative abundances of phyla across the four treatment groups. (C) Selection of genus-level microbiome markers by the Boruta algorithm and Venn diagram of the selected markers that are common or specific to disease or treatment. JPYS, Jian-Pi-Yi-Shen decoction; PF, Piperazine ferulate; CKD, chronic kidney disease; PC1, first principal component; FDR, false discovery rate.

Figure 3

Gut microbiome markers correlated with host clinical outcomes. (A) Correlation network of microbiome marker, clinical parameter, and KEGG pathway. Circles represent markers of genus; Squares represent KEGG pathways; and diamonds represent clinical parameters. Node colors indicate treatment groups. Node sizes are scaled according to their degrees of connections. The larger the node is, the more connections it has. Two nodes are connected if their spearman correlation is significant (p < 0.05, FDR-corrected) between genera, or between genera and KEGG pathways. Connections are drawn between genera and clinical parameters only when their spearman correlation is significant (p < 0.05, FDR-corrected) and the absolute of correlation coefficient is larger than 0.6. Thickness of edges between nodes is scaled according the absolute value of spearman correlation coefficient. Edge colors indicate correlations between genera (or between genera and KEGG pathways) in different groups (green: CKD group; pink: treat groups of PF and JPYS). Orange edges represent correlations between genera and clinical parameters. Dotted and solid lines indicate negative or positive correlations respectively. (B) Relative abundances of keystone genus-level microbiome markers in each group. Keystone markers are manually selected from the network hubs, with regard to their interspecies interactions and clinical associations. Variations between groups were tested by Wilcoxon rank-sum test with FDR correction. Only significant differences between groups (p < 0.05) were annotated. FDR, false discovery rate; JPYS, Jian-Pi-Yi-Shen decoction; PF, Piperazine ferulate; CKD, chronic kidney disease.

Figure 4

qPCR validation of the identified keystone microbiome markers. (A) Pearson correlation between relative abundance of the genera Methanobrevibacter, Enterococcus and Blautia from 16S rDNA sequencing and qPCR data. Colors and shapes of the plot symbols indicate samples from different treatment groups. Shaded area represents the 95% confidence interval for the regression line. (B) Relative abundance of the genera Methanobrevibacter, Enterococcus, and Blautia across treatment groups, as detected by qPCR. Variations between groups were tested by Wilcoxon rank-sum test with FDR correction. Only significant differences between groups (p < 0.05) were annotated. FDR, false discovery rate.