A metagenomic framework for rapid Listeria monocytogenes surveillance in food production environments

Listeria monocytogenes remains a major foodborne pathogen with high mortality and costly persistence in food-processing environments. Established diagnostics rely on selective enrichment and single-colony isolation, which could introduce strong biases by favouring fast-growing strains or those more tolerant to enrichment broth inhibitors, while suppressing slow-growing, viable-but-nonculturable, and other co-occurring strains. This can obscure true pathogen diversity and may contribute to discrepancies between strains detected in food production environments and those associated with disease. To quantify the bias introduced by established culture-based diagnostics and to assess the potential advantage of metagenomics-based pathogen detection directly from the original sample matrix, we developed and evaluated a rapid nanopore sequencing-based metagenomic framework. We designed an artificial metagenomic community of several Listeria strains, comprising L. monocytogenes lineages I-III (including hypervirulent, persistent, and low-virulence strains), other Listeria spp., and a realistic background microbiome representative of food-processing environments. We then used this mock community to spike standard surveillance sponges and compared three workflows: (i) direct nanopore metagenomic sequencing of the original sample matrix, (ii) quasi-metagenomic sequencing after 4 h, 12 h, 24 h, or 48 h of selective enrichment, and (iii) ISO-based culture followed by whole-genome sequencing of a single presumptive L. monocytogenes isolate. We found that the culture-based approach recovered only a limited subset of strains, consistently underrepresenting diversity and failing to detect multi-strain contamination. These findings were reflected by the quasi-metagenomic results, where we found relative L. monocytogenes enrichment to be strain-dependent, indicating selective enrichment bias favouring specific strains. Metagenomics captured the full spectrum of spiked Listeria strains, enabling comprehensive strain-level resolution at all inoculation levels. We only observed relative enrichment of the L. monocytogenes strains by quasi-metagenomics compared with metagenomics after 48 h of selective enrichment. While driven primarily by the additional enrichment of L. innocua, these results suggest that quasi-metagenomics improves L. monocytogenes recovery only at the cost of a substantial reduction in speed. We finally showed that the sensitivity and accuracy of metagenomics could be improved by utilising different environmental sampling materials. We did not find any significant performance improvements from nanopore sequencing-based enrichment of L. monocytogenes through adaptive sampling approaches. We conclude that integrating long-read metagenomics into routine surveillance shows great promise to improve detection and source attribution in food safety systems.