Comparative genome analyses of clinical and non-clinical Clostridioides difficile strains

Abstract

The pathogenic bacterium Clostridioides difficile is a worldwide health burden with increasing morbidity, mortality and antibiotic resistances. Therefore, extensive research efforts are made to unravel its virulence and dissemination. One crucial aspect for C. difficile is its mobilome, which for instance allows the spread of antibiotic resistance genes (ARG) or influence strain virulence. As a nosocomial pathogen, the majority of strains analyzed originated from clinical environments and infected individuals. Nevertheless, C. difficile can also be present in human intestines without disease development or occur in diverse environmental habitats such as puddle water and soil, from which several strains could already be isolated. We therefore performed comprehensive genome comparisons of closely related clinical and non-clinical strains to identify the effects of the clinical background. Analyses included the prediction of virulence factors, ARGs, mobile genetic elements (MGEs), and detailed examinations of the pan genome. Clinical-related trends were thereby observed. While no significant differences were identified in fundamental C. difficile virulence factors, the clinical strains carried more ARGs and MGEs, and possessed a larger accessory genome. Detailed inspection of accessory genes revealed higher abundance of genes with unknown function, transcription-associated, or recombination-related activity. Accessory genes of these functions were already highlighted in other studies in association with higher strain virulence. This specific trend might allow the strains to react more efficiently on changing environmental conditions in the human host such as emerging stress factors, and potentially increase strain survival, colonization, and strain virulence. These findings indicated an adaptation of the strains to the clinical environment. Further, implementation of the analysis results in pairwise genome comparisons revealed that the majority of these accessory genes were encoded on predicted MGEs, shedding further light on the mobile genome of C. difficile. We therefore encourage the inclusion of non-clinical strains in comparative analyses.

Keywords: Clostridioide difficile; clinical; genome comparison; mobile genetic element; non-clinical; virulence.

Figure 1

C. difficile-associated virulence factors in the analyzed strains. (A) Presence of the examined virulence factors is indicated as the protein sequence query coverage to the reference VFDB dataset, by color and stated coverage value. White coverage values highlight deviating sequences in proteins of the same query coverage between clinical and non-clinical strain. Virulence factors are labelled with their names as obtained from the VFDB dataset, and their related functions are stated on top. *flagellar operon comprising 41 CDSs obtained from the VFDB dataset, and its query coverage calculated as the relative number of present CDSs of the total 41. (B) Sequence similarity as percentage identity between clinical and non-clinical strains of the proteins highlighted in (A) as white.

Figure 2

Gene cluster comparisons of toxin loci. Genes within and next to the toxin-encoding loci were compared on nucleotide sequence level between all analyzed strains for (A,B) the PaLoc with 630 as reference, and (C,D) the CdtLoc with R20291 as reference sequence. (A,C) depict 100% sequence identity, while (B,D) represent identities above 80%.

Figure 3

Core and accessory genome sizes of the analyzed C. difficile strains. Venn diagrams depicting the shared and unique genes among (A) all eight strains, or (B) pairwise between ST-corresponding clinical and non-clinical strain. The relative proportions of unique genes with regard to the total number of CDS per genome are indicated in parentheses below each absolute number of unique genes.

Figure 4

Relative abundance of COGs assigned to the unique genes from pairwise pan genome analyses. (A) The relative proportions of unique genes to total number of CDSs per genome in pairwise comparisons was transferred to their assigned COGs, which are designated with COG category name and function. (B) Individual COG proportions of unique genes of non-clinical strains were subtracted from the corresponding clinical strain to see if specific COGs are more frequent among accessory genes of a certain clinical background.

Figure 5

Predicted ARGs in the analyzed C. difficile strains. The number of identified ARGs and AR-conferring mutations as predicted with RGI-CARD (Alcock et al., 2023) and AMRFinderPlus (Feldgarden et al., 2021) were indicated by color and respective value, with white/no value meaning gene absence. Heatmap-tiles are missing for genes that were not part of the analysis tools. The total number of predicted ARGs for each program is additionally stated. ARGs are grouped according the associated antibiotic class: AG, aminoglycosides; BL, beta-lactams; DA, disinfecting agent and antiseptics; FQ, fluorquinolones; GP, glycopeptides; MLS, macrolides/lincosamides/streptogramins; RF, rifamycin; ST, streptothricins; TET, tetracyclines.

Figure 6

Predicted MGEs in the analyzed strains. The number of the analyzed MGEs prophages, integrative elements, GIs, and IS elements were indicated by color and the respective number, with white/no value meaning no prediction. Prophages are described as incomplete or intact as predicted by PHASTEST (Wishart et al., 2023). Integrative elements are categorized by the assigned superfamily and grouped into ICE and IME. Presence of IS elements is described in the context of the identified families, and additionally given as total number of IS elements.

Figure 7

Pairwise genome comparisons complemented with the predicted ARGs, MGEs, and accessory genes. Genome comparisons of (A) ST1-strains and (B) ST3-strains are depicted with different tracks for each visualized feature in the clinical strain at the top (track letter a) and non-clinical strain at the bottom (track letter b). The tracks represent: 1 unique genes with genes assigned to COG S, K, L highlighted, and multiple genes of the same COG grouped together if necessary for better visibility, 2 AMRFinderPlus (Feldgarden et al., 2021) predicted ARGs, 3 RGI+CARD (Alcock et al., 2023) predicted ARGs, 4 predicted IS elements, 5 predicted integrative elements labelled with assigned superfamily, 6 predicted GIs, 7 prophage prediction with completeness color-coded according to PHASTEST (Wishart et al., 2023), 8 MUMmer alignment (Kurtz et al., 2004), 9 replicons.

Figure 8

Pairwise genome comparisons complemented with the predicted ARGs, MGEs, and accessory genes. (A) ST8-strains, and (B) ST11-strains are depicted with different tracks for each visualized feature in the clinical strain at the top (track letter a) and non-clinical strain at the bottom (track letter b). The tracks represent: 1 unique genes with genes assigned to COG S, K, L highlighted, and multiple genes of the same COG grouped together if necessary for better visibility, 2 AMRFinderPlus (Feldgarden et al. 2021) predicted ARGs, 3 RGI+CARD (Alcock et al. 2023) predicted ARGs, 4 predicted IS elements, 5 predicted integrative elements labelled with assigned superfamily, 6 predicted GIs, 7 prophage prediction with completeness color-coded according to PHASTEST (Wishart et al. 2023), 8 MUMmer alignment (Kurtz et al. 2004), 9 replicons.