Detection of Antimicrobial Resistance Using Proteomics and the Comprehensive Antibiotic Resistance Database: A Case Study

Abstract

Purpose: Antimicrobial resistance (AMR), especially multidrug resistance, is one of the most serious global threats facing public health. The authors proof-of-concept study assessing the suitability of shotgun proteomics as an additional approach to whole-genome sequencing (WGS) for detecting AMR determinants.

Experimental design: Previously published shotgun proteomics and WGS data on four isolates of Campylobacter jejuni are used to perform AMR detection by searching the Comprehensive Antibiotic Resistance Database, and their detection ability relative to genomics screening and traditional phenotypic testing measured by minimum inhibitory concentration is assessed.

Results: Both genomic and proteomic approaches identify the wild-type and variant molecular determinants responsible for resistance to tetracycline and ciprofloxacin, in agreement with phenotypic testing. In contrast, the genomic method identifies the presence of the β-lactamase gene, bla_OXA-61 , in three isolates. However, its corresponding protein product is detected in only a single isolate, consistent with results obtained from phenotypic testing.

Keywords: Campylobacter jejuni; Comprehensive Antibiotic Resistance Database; antimicrobial resistance; shotgun proteomics; whole genome sequencing.

Figure 1

Genomic and proteomic searches of CARD for AMR screening. A) Heatmap of AMR protein sequences identified in WGS data of Campylobacter isolates using RGI v4.0.1. Each cell is labeled and shaded from red to pink representing their percent identity match to CARD as indicated in the legend. Only perfect (**) and strict (*) matches as reported by RGI are coloured; entries without a protein sequence match in an isolate are uncoloured. B) Heatmap of AMR protein abundance detected using FDR = 0.01 and an additional presence cut‐off in isolates with six biological replicates. The log₂ relative intensities are labeled in each cell and colored in a gradient from red to yellow indicating higher and lower abundance, respectively. Proteins with significant abundance variation among groups as tested with ANOVA (Benjamini–Hochberg adjusted p ≤ 0.05) are marked with an asterisk.

Figure 2

Conserved peptide and peptide variants of GyrA and their relative peptide abundance in isolates. The three examples provided show a peptide shared among isolates, and two wild type/mutant peptide‐pairs. A) Examples of peptide sequences appear in bold with the mutations in red and underlined. All identified mutations are harbored by the 00‐1597 isolate; all other isolates are wild type using the Swiss‐Prot database as a reference. B) Heatmap of peptide abundance using FDR = 0.01 and an additional length‐dependent presence cut off in isolates with six biological replicates. The log₂ relative peptide abundance is labeled and the shading from red to yellow indicates higher and lower abundance, respectively. Peptides that are undetected or do not pass the additional presence cut off are uncolored.