Long-read metagenomics gives a more accurate insight into the microbiota of long-ripened Gouda cheeses

Hannes Decadt Cristian Díaz-Muñoz Louise Vermote Inés Pradal Luc De Vuyst Stefan Weckx ^*

• Vrije Universiteit Brussel, Brussels, Belgium

Metagenomic studies of the Gouda cheese microbiota and starter cultures are scarce. During the present study, short-read metagenomic sequencing (Illumina) was applied on 89 Gouda cheese and processed milk samples, which have been investigated before concerning their metabolite and taxonomic composition, the latter applying amplicon-based, high-throughput sequencing (HTS) of the full-length 16S rRNA gene. Selected samples were additionally investigated using long-read metagenomic sequencing (Oxford Nanopore Technologies, ONT). Whereas the species identified by amplicon-based HTS and metagenomic sequencing were identical, the relative abundances of the major species differed significantly. Lactococcus cremoris was more abundant in the metagenomics-based taxonomic analysis compared to the amplicon-based one, whereas the opposite was true for the non-starter lactic acid bacteria (NSLAB). This discrepancy was related to a higher fragmentation of the lactococcal DNA compared with the DNA of other species when applying ONT. Possibly, a higher fragmentation was linked with a higher percentage of dead or metabolically inactive cells, suggesting that full-length 16S rRNA gene amplicon-based HTS might give a more accurate view on active cells. Further, fungi were not abundantly present in the Gouda cheeses examined, whereas about 2 % of the metagenomic sequence reads was related to phages, with higher relative abundances in the cheese rinds and longripened cheeses. Intraspecies differences found by short-read metagenomic sequencing were in agreement with the amplicon sequence variants obtained previously, confirming the ability of fulllength 16S rRNA gene amplicon-based HTS to reach a taxonomic assignment below species level. Metagenome-assembled genomes (MAGs) were retrieved for 15 species, among which the starter cultures Lc. cremoris and Lactococcus lactis and the NSLAB Lacticaseibacillus paracasei, Loigolactobacillus rennini and Tetragenococcus halophilus, although obtaining MAGs from Lc. cremoris and Lc. lactis was more challenging because of a high intraspecies diversity and high similarity between these species. Long-read metagenomic sequencing could not improve the retrieval of lactococcal MAGs, but, overall, MAGs obtained by long-read metagenomic sequencing solely were superior compared with those obtained by short-read metagenomic sequencing solely, reaching a highquality draft status of the genomes.