Antimicrobial resistance (AMR) is a growing global health concern, primarily driven by the spread of antimicrobial resistance genes (ARGs). These genes can be horizontally transferred between bacteria, even from non-pathogenic to pathogenic strains. Fermented foods, such as kefir and yoghurt, contain high concentrations of bacteria, raising concerns about their potential role in ARG dissemination. This study aims to characterize the ARG diversity in various kefir and yoghurt samples to assess the potential risk associated with their consumption. The context is the increasing urgency to identify sources of ARG acquisition and prevent the emergence of multi-drug resistant pathogens. The study's importance lies in the potential to identify ways to minimize ARG intake through common food sources and contribute to AMR mitigation strategies. Understanding the resistome of fermented dairy products offers valuable insights into the potential for horizontal gene transfer and the development of AMR in the human gut. The daily use of antibiotics in human and animal healthcare is a significant factor increasing selective pressure on bacterial ARGs, and therefore understanding the resistome of commonly consumed foods is critical.

Previous research has highlighted the presence of ARGs in various environments, including raw milk and animal-based foods. Studies have also shown that the bacterial resistome can increase during fermentation processes. The specific ARG profiles of kefir and yoghurt, and the potential for horizontal gene transfer within these products, remain incompletely understood. This study builds upon existing literature by focusing specifically on the diversity and abundance of ARGs in kefir and yoghurt, using a comprehensive metagenomic approach. The study expands on existing knowledge by examining not just the final products but also the grains and individual bacterial strains used in fermentation, providing a more complete picture of the ARG landscape in these probiotic products.

The study analyzed a total of 35 samples encompassing kefir grains (n=4), kefir products (n=15), kefir strains (n=7), one yoghurt grain, and yoghurt products (n=5). Samples were obtained from the NCBI SRA repository and included the authors' own shotgun sequencing data of a kefir and a yoghurt starter culture. The sequencing data from PRJNA644779 was generated using Illumina NextSeq, with read numbers of 22,044,496 and 20,895,112 for kefir and yoghurt, respectively. DNA extraction and library preparation for these samples utilized the UltraClean Microbial DNA Isolation Kit and the NEBNext Ultra II DNA Library Prep Kit. Raw reads underwent quality control using Trimmomatic, retaining reads longer than 50 bp and having a quality threshold of 15. Taxonomic classification utilized Kraken2 (k=35) against the NCBI non-redundant nucleotide database. Kraken2 was also used with a confidence of 0.5 for more precise species-level identification. MEGAHIT assembled the bacterial reads into contigs, and Prodigal identified open reading frames (ORFs). Resistance Gene Identifier (RGI, v5.1.0) with Diamond was used to align the ORFs to ARGs in the CARD v.3.0.9 database. Strict filtering criteria (90% identity and 60% coverage) were employed to analyze ARGs, both with and without considering “nudged” hits. For the analysis of ARG abundance changes during kefir fermentation, only ARGs with maximal coverage greater than 75% were included. Bowtie2 (-very-sensitive-local parameter) aligned bacterial reads to contigs. ARG abundance was expressed as fragments per kilobase per million fragments (FPKM). HMMER and TnpPred were used to identify mobile genetic element (MGE) homologs. PlasFlow analyzed contig plasmid origins. Bray-Curtis dissimilarity matrix was used to assess ARG abundance diversity, with principal coordinate analysis (PCOA) visualizing sample distances. Permutational multivariate analysis of variance explored the association between dissimilarity and independent variables (sample type, source, and BioProject). Linear models assessed the relationship between detected ORF length and sequencing depth.

The study revealed a significant diversity of ARGs across the various kefir and yoghurt samples. Kefir samples exhibited substantially higher ARG abundance compared to yoghurt samples. Kefir strains showed the highest average ARG abundance (282 FPKM), followed by kefir products (240 FPKM) and kefir grains (209 FPKM). Yoghurt samples displayed significantly lower abundances (yoghurt grain: 17.9 FPKM; yoghurt products: 45.7 ± 32.2 FPKM). The identified ARGs conferred resistance to a broad spectrum of antibiotics, including aminocoumarins, aminoglycosides, carbapenems, cephalosporins, cephamycins, diaminopyrimidines, elfamycins, fluoroquinolones, fosfomycins, glycylcyclines, lincosamides, macrolides, monobactams, nitrofurans, nitroimidazoles, penams, penems, peptides, phenicols, rifamycins, tetracyclines, and triclosan. The most abundant ARGs were *poxtA* and APH(3')-IIb, present in both kefir and yoghurt samples. Analysis of ARG abundance changes during kefir fermentation demonstrated a positive association between the abundance of specific ARGs and the relative abundance of their most likely bacterial hosts (*L. mesenteroides* for APH(3')-IIb and *L. kefiranofaciens* for *poxtA*). Only one gene, *lmrd*, showed evidence of potential mobility based on its genomic environment. Even with stringent filtering (excluding nudged hits), several ARGs were detected, indicating a robust presence of resistance mechanisms. Analysis revealed that sequencing depth significantly impacts the detection of ARGs, suggesting a need for sufficient sequencing depth in similar studies (at least 20 million reads). The dominant resistance mechanisms in kefir products were antibiotic target protection (50.73%) and antibiotic inactivation (45.45%). In kefir grains, the main mechanism was antibiotic target protection (91.98%).

The findings support the hypothesis that ARG content can increase during fermentation due to bacterial multiplication. The higher ARG abundance in kefir compared to yoghurt might be attributed to the presence of fungi in kefir seed cultures, which could create a selective advantage for bacteria carrying ARGs. The identified ARGs originate from bacteria commonly used in fermented dairy product production, highlighting the potential for ARG transfer within the gut. The presence of ARGs conferring resistance to clinically significant antibiotics raises concerns about the potential impact of consuming fermented dairy products on human health. The observation of ARGs associated with antibiotics not commonly used in dairy cattle suggests that the source of these ARGs may not solely be related to direct antibiotic use on farms. Future research should focus on clarifying the origin and transmission pathways of ARGs in fermented foods to develop strategies to mitigate the risk of ARG spread.

This study provides valuable insights into the ARG diversity in kefir and yoghurt, highlighting the potential for ARG transmission through the consumption of fermented dairy products. The higher ARG abundance in kefir compared to yoghurt warrants further investigation. Stricter monitoring and selection of starter cultures are recommended to minimize ARG intake via food. Future research should focus on the mechanisms of ARG transfer, the impact of different fermentation conditions, and the long-term health consequences of ARG consumption.

The study's reliance on publicly available metagenomic data may limit the generalizability of the findings to other geographical regions or production methods. The study did not directly investigate the potential for horizontal gene transfer of identified ARGs in the human gut. Furthermore, the functional impact of the identified ARGs remains to be fully elucidated.