Applications of next generation sequencing for the assessment of microbiological safety of fresh produce

Abstract

Fresh produce, which is often minimally processed and consumed raw, is increasingly recognised as a route of contamination leading to foodborne illness. This study focussed on the potential of next generation sequencing (NGS) to detect and characterise foodborne pathogens and elucidate the microbiome and potential influences on the survival and transmission of human pathogens within the fresh produce supply-chain. Initial research assessed the ability of several NGS methodologies to screen the microbiome of fresh produce and identify the limit of detection of bacterial and viral pathogens, using a mock sample set created with known levels of contamination. The limit of detection of human pathogens was found to be dependent upon enrichment method, sequencing approach and bioinformatics analysis utilised. The most sensitive approach tested involved sequence preparation using ribosomal depletion followed by RNAseq and analysis of the microbiome using Kraken; yielding a limit of detection of 105 colony forming units / plaque forming units (CFU/PFU) per extraction. Subsequent in silico work showed the differing read lengths obtained from the MiSeq, HiSeq and NovaSeq has no influence on the limit of detection of human pathogens within a mock community. Methodological work was applied to study the microbiome associated with commercial fresh produce samples to identify species that may confer a positive or negative effect on the survival of human pathogens. Sequences were also screened for antimicrobial resistance (AMR) associated genes. Data revealed the microbiome to be dominated by potential spoilage organisms and plant pathogens. Four unique AMR-associated genes, with over 80% identity and coverage, were found (CRP, H-NS, MexF, MexB). Nine taxa, including Pectobacterium and Dickeya - soft rot causing bacteria previously linked in the literature to survival of Salmonella on fresh produce - were found to be positively correlated with the microbiological detection of Enterobacteriaceae and Listeria species. Pseudonocardia was the only taxon detected in a large proportion of samples that was inversely correlated with detection of Enterobacteriaceae. A collection of 48 strains of Listeria monocytogenes originating from the fresh produce supply chain was subject to a combination of phenotypic and genotypic methods to characterise the resistome, virulome, and biofilm forming ability. The genes present and phylogenetic identity of these isolates was then compared to those of 80 isolates from meat and clinical origin, to identify signatures of fresh produce contaminating L. monocytogenes. Phenotypic screens revealed no evidence of AMR within L. monocytogenes isolated from UK fresh produce. All isolates of L. monocytogenes tested were capable of forming biofilms and displayed an increased propensity for biofilm formation in nutrient broth than in brain heart infusion broth (suggesting that biofilm formation may be induced by stress). Whole genome sequencing data from fresh produce isolates, meat isolates and clinical isolates were indistinguishable based on phylogeny, resistome and virulome. When screened using genome wide association study (GWAS) many potential genes of interest were highlighted.