A multiomics approach to identify host-microbe alterations associated with infection severity in diabetic foot infections: a pilot study

Abstract

Diabetic foot infections (DFIs) are a major cause of hospitalization and can lead to lower extremity amputation. In this pilot study, we used a multiomics approach to explore the host-microbe complex within DFIs. We observed minimal differences in the overall microbial composition between PEDIS infection severities, however Staphylococcus aureus and Streptococcus genera were abundant and highly active in most mild to moderate DFIs. Further, we identified the significant enrichment of several virulence factors associated with infection pathogenicity belonging to both Staphylococcus aureus and Streptococcus. In severe DFIs, patients demonstrated a greater microbial diversity and differential gene expression demonstrated the enrichment of multispecies virulence genes suggestive of a complex polymicrobial infection. The host response in patients with severe DFIs was also significantly different as compared to mild to moderate DFIs. This was attributed to the enrichment of host genes associated with inflammation, acute phase response, cell stress and broad immune-related responses, while those associated with wound healing and myogenesis were significantly depleted.

Fig. 1. Analysis of metagenomic shotgun sequencing data using Humann2.

a Bar chart representing the 30 most highly abundant species (%) for 36 individuals with DFIs grouped by infection severity (PEDIS). Each bar represents individual species level identification. b Boxplots identifying metabolic pathways which are significantly different in abundance between infection severity groups (LDA score >2 (p < 0.05)).

Fig. 2. Bar chart showing the top 20 most metabolically active taxa across each sample following metratranscriptomic analysis with the SqueezeMeta pipeline.

Activity was based on normalized read counts (LogCpm) annotated to individual taxa and grouped according to infection severity (PEDIS group). Only taxa which comprises >1% of the total annotated microbiome are shown.

Fig. 3. Functional analysis of annotations isolated by the keyword “virulence”.

a Bar chart showing the top 10 taxa by with virulence annotations. Annotations were extracted using the “subsetTax” tool included with the SQMtools R package. Patients 16 and 11 had no annotations related to “virulence” within these samples. b Heat map of the top ten most enriched virulence proteins across all PEDIS infection severities. Read counts were normalized prior to applying a multilevel analysis (ANOVA) fitting all contrasts between groups to identify the top ten enriched genes. Patients are grouped by infection severity (PEDIS) with features clustered using Ward clustering.

Fig. 4. Heat map of the top 50 most enriched host genes across all PEDIS infection severities.

Read counts were normalized prior to applying a multi-level analysis (ANOVA) fitting all contrasts between groups to identify the top 50 enriched genes. Patients are grouped by infection severity (PEDIS) with features clustered using Ward clustering.