Quantitative metagenomics reveals unique gut microbiome biomarkers in ankylosing spondylitis

Abstract

Background: The assessment and characterization of the gut microbiome has become a focus of research in the area of human autoimmune diseases. Ankylosing spondylitis is an inflammatory autoimmune disease and evidence showed that ankylosing spondylitis may be a microbiome-driven disease.

Results: To investigate the relationship between the gut microbiome and ankylosing spondylitis, a quantitative metagenomics study based on deep shotgun sequencing was performed, using gut microbial DNA from 211 Chinese individuals. A total of 23,709 genes and 12 metagenomic species were shown to be differentially abundant between ankylosing spondylitis patients and healthy controls. Patients were characterized by a form of gut microbial dysbiosis that is more prominent than previously reported cases with inflammatory bowel disease. Specifically, the ankylosing spondylitis patients demonstrated increases in the abundance of Prevotella melaninogenica, Prevotella copri, and Prevotella sp. C561 and decreases in Bacteroides spp. It is noteworthy that the Bifidobacterium genus, which is commonly used in probiotics, accumulated in the ankylosing spondylitis patients. Diagnostic algorithms were established using a subset of these gut microbial biomarkers.

Conclusions: Alterations of the gut microbiome are associated with development of ankylosing spondylitis. Our data suggest biomarkers identified in this study might participate in the pathogenesis or development process of ankylosing spondylitis, providing new leads for the development of new diagnostic tools and potential treatments.

Keywords: Ankylosing spondylitis; Biomarkers; Human gut microbiome; Pathogenesis.

Fig. 1

Differences of phylogenetic abundance between AS patients and healthy controls. The phylotypes that were increased (a) or decreased (b) in the AS patients at the phylum, genus, and species levels. Red and blue indicate the AS patients and healthy controls, respectively. The phylogenetic abundance of phyla that had mean values less than 1% and that of genera and species that were less than 0.01% were excluded. After exclusion, Wilcoxon rank-sum tests were applied to identify the differentially abundant phyla, genera, and species. Among these, the highest medians of the phylogenetic abundance in the enriched cohort were drawn as boxplots

Fig. 2

MGSs in AS patients and healthy controls and association with clinical indices. a The abundance of 12 MGSs are shown as heatmaps: the discovery set (n = 156) is on the left and the validation set (n = 55) is on the right. The colors denote the variation in abundance (white indicates zero; black indicates the highest abundance). b The association between MGSs and clinical index is shown in the middle: the darker the color, the greater the intensity. Violet indicates a positive correlation with each index (the index partition in first row), green indicates a negative correlation with each index (the index partition in first row). The 25 genes in each MGS for which the mean abundance values were the highest are shown in the heatmap. c On the right of the heatmap, the Wilcoxon rank-sum test p values for the mean abundance of the 25 “marker genes” are indicated. Above the heatmap, the color key shows how the color variation indicates the abundance. d The networks of the 12 MGSs reflect the interaction between them. The notes represent the MGSs and the note size is proportional to the mean abundance of the genes in the MGS. The red lines represent the negative correlation between the two notes and the blue lines represent the positive correlation between the two notes. e The Venn diagram of the MGS in AS Project, LC Project, T2D Project, T2D European Women Project, and Obesity Project (Additional file 1: Table S11)

Fig. 3

Receiver-operating characteristic (ROC) curves of the sequenced reference genome markers, gene markers, and cluster markers. a Classifier based on 25 sequenced reference genome markers and the ROC curves for the discovery and validation cohorts. b Classifier based on 35 gene markers and the ROC curves for the discovery and validation cohorts. c Classifier based on 62 cluster markers and the ROC curves for the discovery and validation cohorts. The discovery cohort was the 156 samples that were used to identify the markers; the validation cohort was the 55 samples that were used to validate the markers such as those shown in Fig. 2

Fig. 4

A schematic diagram of the main functions of the gut microbes associated with AS. The red text denotes enriched in the AS patients; the blue text denotes depleted in the AS patients; the orange lines and arrows denote the actions initiated by the gut microbes or functional in the gut environment in this study; the black line and arrows denote the known actions and mechanisms functional in the host tissues as previously reported; the blue dashed line and arrows denote the inferred actions and mechanisms in the host tissues that were associated with the gut microbes. Here, we present some information regarding the influence of the gut microbiota on both the innate and adaptive immune responses. With respect to the innate immune responses (see Fig. 1), RegIIIγ hyposecretion caused by decreased levels of LPS and flagellin and accompanied by depletion of bacterial chemotaxis, regulation of the actin cytoskeleton, Fc gamma R-mediated phagocytosis, and NOD-like receptor signaling result in the dysbiosis of the gut microbiome and the onset of AS. With respect to the adaptive immune responses (see Fig. 2), a reduction in the levels of Polysaccharide A (PSA), which is mainly produced by the Bacteroides, may directly or indirectly influence the differentiation of the Treg cells and thereby contribute to AS