Guild-level signature of gut microbiome for diabetic kidney disease

Shasha Tang,Guojun Wu,Yalei Liu,Binghua Xue,Shihan Zhang,Weiwei Zhang,Yifan Jia,Qinyuan Xie,Chenghong Liang,Limin Wang,Hongyan Heng,Wei Wei,Xiaoyang Shi,Yimeng Hu,Junpeng Yang,Lingyun Zhao,Xiaobing Wang,Liping Zhao,Huijuan Yuan

DOI: https://doi.org/10.1128/mbio.00735-24

IF: 6.4

2024-06-01

mBio

Abstract:ABSTRACT Current microbiome signatures for chronic diseases such as diabetic kidney disease (DKD) are mainly based on low-resolution taxa such as genus or phyla and are often inconsistent among studies. In microbial ecosystems, bacterial functions are strain specific, and taxonomically different bacteria tend to form co-abundance functional groups called guilds. Here, we identified guild-level signatures for DKD by performing in-depth metagenomic sequencing and conducting genome-centric and guild-based analysis on fecal samples from 116 DKD patients and 91 healthy subjects. Redundancy analysis on 1,543 high-quality metagenome-assembled genomes (HQMAGs) identified 54 HQMAGs that were differentially distributed among the young healthy control group, elderly healthy control group, early-stage DKD patients (EDG), and late-stage DKD patients (LDG). Co-abundance network analysis classified the 54 HQMAGs into two guilds. Compared to guild 2, guild 1 contained more short-chain fatty acid biosynthesis genes and fewer genes encoding uremic toxin indole biosynthesis, antibiotic resistance, and virulence factors. Guild indices, derived from the total abundance of guild members and their diversity, delineated DKD patients from healthy subjects and between different severities of DKD. Age-adjusted partial Spearman correlation analysis showed that the guild indices were correlated with DKD disease progression and with risk indicators of poor prognosis. We further validated that the random forest classification model established with the 54 HQMAGs was also applicable for classifying patients with end-stage renal disease and healthy subjects in an independent data set. Therefore, this genome-level, guild-based microbial analysis strategy may identify DKD patients with different severity at an earlier stage to guide clinical interventions. IMPORTANCE Traditionally, microbiome research has been constrained by the reliance on taxonomic classifications that may not reflect the functional dynamics or the ecological interactions within microbial communities. By transcending these limitations with a genome-centric and guild-based analysis, our study sheds light on the intricate and specific interactions between microbial strains and diabetic kidney disease (DKD). We have unveiled two distinct microbial guilds with opposite influences on host health, which may redefine our understanding of microbial contributions to disease progression. The implications of our findings extend beyond mere association, providing potential pathways for intervention and opening new avenues for patient stratification in clinical settings. This work paves the way for a paradigm shift in microbiome research in DKD and potentially other chronic kidney diseases, from a focus on taxonomy to a more nuanced view of microbial ecology and function that is more closely aligned with clinical outcomes.

microbiology

What problem does this paper attempt to address?

The paper aims to address the issue of early diagnosis and clinical intervention of diabetic kidney disease (DKD). Traditional microbiome studies mainly rely on taxonomic units (such as genus or phylum), which may not reflect the functional dynamics or ecological interactions of microbial communities. Therefore, this study conducted an in-depth metagenomic sequencing and genome-centric and guild-based analysis of fecal samples from 116 DKD patients and 91 healthy controls. The study found 54 high-quality metagenome-assembled genomes (HQMAGs) with differential distribution among young healthy controls, elderly healthy controls, early-stage DKD patients, and late-stage DKD patients. These HQMAGs were further divided into two guilds, with guild 1 containing more short-chain fatty acid synthesis genes, while guild 2 contained more indole biosynthesis, antibiotic resistance, and virulence factor-related genes. The guild index was able to distinguish DKD patients from healthy individuals and was associated with different severities of DKD. Additionally, the random forest classification model validated its effectiveness on an independent dataset. In summary, this genome- and guild-based microbial analysis strategy may help in the early identification of DKD patients with varying severities, thereby guiding clinical interventions.