Metagenomic analysis reveals distinct patterns of gut microbiota features with diversified functions in C. difficile infection (CDI), asymptomatic carriage and non-CDI diarrhea

Abstract

Clostridioides difficile infection (CDI) has been recognized as a leading cause of healthcare-associated infections and a considerable threat to public health globally. Increasing evidence suggests that the gut microbiota plays a key role in the pathogenesis of CDI. The taxonomic composition and functional capacity of the gut microbiota associated with CDI have not been studied systematically. Here, we performed a comprehensive shotgun metagenomic sequencing in a well-characterized human cohort to reveal distinct patterns of gut microbiota and potential functional features associated with CDI. Fecal samples were collected from 104 inpatients, including : (1) patients with clinically significant diarrhea and positive nucleic acid amplification testing (NAAT) and received CDI treatment (CDI, n = 47); (2) patients with positive stool NAAT but without diarrhea (Carrier, n = 17); (3) patients with negative stool NAAT but with diarrhea (Diarrhea, n = 14); and (4) patients with negative stool NAAT and without diarrhea (Control, n = 26). Downstream statistical analyses (including alpha and beta diversity analysis, differential abundance analysis, correlation network analysis, and potential functional analysis) were then performed. The gut microbiota in the Control group showed higher Chao1 index (p < 0.05), while Shannon index at KEGG module level was higher in CDI than in Carrier and Control (p < 0.05). Beta diversity for species composition differed significantly between CDI vs Carrier/Control cohorts (p < 0.05). Microbial Linear discriminant analysis Effect Size and ANCOM analysis both identified 8 species (unclassified_f_Enterobacteriaceae, Veillonella_parvula, unclassified_g_Klebsiella and etc.) were enriched in CDI, Enterobacter_aerogenes was enriched in Diarrhea, Collinsella_aerofaciens, Collinsella_sp_4_8_47FAA, Collinsella_tanakaei and Collinsella_sp_CAG_166 were enriched in Control (LDA >3.0, adjusted p < 0.05). Correlation network complexity was higher in CDI with more negative correlations than in other three cohorts. Modules involved in iron complex transport system (M00240) was enriched in CDI, ABC-2 type transport system (M00254), aminoacyl-tRNA biosynthesis (M00359), histidine biosynthesis (M00026) and inosine monophosphate biosynthesis (M00048) were enriched in Carrier, ribosome (M00178 and M00179) was enriched in Diarrhea, fluoroquinolone resistance (M00729) and aminoacyl-tRNA biosynthesis (M00360) were enriched in Control (LDA > 2.5, adjusted p < 0.05). Resistance functions of acriflavine and glycylcycline were enriched in CDI, while resistance function of bacitracin was enriched in Carrier (LDA > 3.0, adjusted p < 0.05), and the contributions of phylum and species to resistance functions differed among the four groups. Our results reveal alterations of gut microbiota composition and potential functions among four groups of differential colonization/infection status of Clostridioides difficile. These findings support the potential roles of gut microbiota and their potential functions in the pathogenesis of CDI.

Keywords: CDI; antibiotic resistance; gut microbiota; metabolomics; module.

Figure 1.

Alpha diversities of the fecal microbiota in CDI, carrier, diarrhea and control participants. Chao1 indices reflect the abundance species, module and antibiotic resistance gene (ARG) in samples, and Shannon indices reflect the diversity of species, module and antibiotic resistance gene in samples. Alpha diversity at species level, Chao1 index (a) and Shannon index (b). Alpha diversity at module level, Chao1 index (c) and Shannon index (d). Alpha diversity at antibiotic resistance gene level, Chao1 index (e) and Shannon index (f). Abbreviations: ARG, antibiotic resistance gene. ns: p > 0.05, *p < 0.05, ***p < 0.001.

Figure 2.

Beta diversities of the fecal microbiota in CDI, carrier, diarrhea and control participants. Principal coordinates analysis (PCoA) with Bray-Curtis dissimilarity was performed to assess the community structure at the species level (a), module level (b) or antibiotic resistance gene (ARG) level (c) in four groups. The ordinate and abscissa represent the two main coordinate axes, and the percentage values of the coordinate axes represent interpretations of the differences in sample composition. The closer the two sample points are, the more similar their bacteria composition. The ellipses represent the 95% of the samples belonging to each group. Dissimilarity was analyzed using adonis statistical tests with 999 permutations based on Bray-Curtis dissimilarity at the species level {CDI vs carrier (R² = 0.125, p = 0.037), CDI vs diarrhea (R² = 0.142, p = 0.037) and CDI vs control (R² = 0.089, p = 0.030)} at the ARG level CDI vs control (R² = 0.136, p = 0.003) while no significant differences were found among four cohorts at the modules level (p > 0.05). Abbreviations: ARG, antibiotic resistance gene; PCoA, principal coordinates analysis.

Figure 3.

Different trends of species from the fecal microbiomes of CDI, carrier, diarrhea and control participants. Linear discriminant analysis (LDA) effect size (LEfSe) is a tool to identify biomarkers from high dimensional data of two or more groups using. This tool clarifies about statistical significance and biological correlation and can identify statistically different biomarkers between groups. Identified biomarkers ranked by effect size, and only species meeting an LDA significant threshold > 3.0 and adjusted p < 0.05 were shown. Abbreviations: LDA, linear discriminant analysis; LEfSe, linear discriminant analysis effect size.

Figure 4.

Different trends of functional modules from the fecal microbiomes of CDI, carrier, diarrhea and control participants. Linear discriminant analysis (LDA) effect size (LEfSe) is a tool to identify biomarkers from high dimensional data of two or more groups using. This tool clarifies about statistical significance and biological correlation and can identify statistically different biomarkers between groups. Identified biomarkers ranked by effect size, and only KEGG modules meeting an LDA significant threshold > 2.5 and adjusted p < 0.05 were shown (a). Relative contributions for modules to microbial phylum (b, top 10) or species (c, top 10) was estimated. Abbreviations: LDA, linear discriminant analysis; LEfSe, linear discriminant analysis effect size.

Figure 5.

Different trends of antibiotic resistance gene (ARG) from the fecal microbiomes of CDI, carrier, diarrhea and control participants. LDA is a tool to identify biomarkers from high dimensional data of two or more groups using. This tool clarifies about statistical significance and biological correlation and can identify statistically different biomarkers between groups. Identified biomarkers ranked by effect size, and only taxa meeting an LDA significant threshold > 3.0 and adjusted p < 0.05 were shown (a). Relative contributions for ARG to microbial phylum (b, top 10) or species (c, top 10) was estimated. Abbreviations: LDA, linear discriminant analysis; ARG, antibiotic resistance gene.

Figure 6.

Network analysis from the fecal microbiomes of CDI (a), carrier (b), diarrhea (c) and control (d) participants. Each node represents species and is colored according to phylum-level taxonomy. Node size indicates relative abundance of a species, red lines indicate a positive association between nodes, and blue lines indicate a negative association between nodes. Line thickness indicates magnitude of the Spearman correlation coefficient, where thickness increases with magnitude.