Bacterial and Genetic Features of Raw Retail Pork Meat: Integrative Analysis of Antibiotic Susceptibility, Whole-Genome Sequencing, and Metagenomics

The global antibiotic resistance crisis, driven by overuse and misuse of antibiotics, is multifaceted. This study aimed to assess the microbiological and genetic characteristics of raw retail pork meat through various methods, including the isolation, antibiotic susceptibility testing (AST), whole-genome sequencing (WGS) of selected indicator bacteria, antibiotic residue testing, and metagenomic sequencing. Samples were purchased from 10 pre-selected retail stores in Gauteng, South Africa. The samples were aseptically separated, with portions sent to an external laboratory for isolating indicator bacteria and testing for antibiotic residues. Identification of the isolated bacteria was reconfirmed using matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS). AST was performed using the Microscan Walkaway system (Beckman Coulter, Brea, CA, USA). WGS and metagenomic sequencing were performed using the Illumina NextSeq 550 instrument (San Diego, CA, USA). The isolated E. coli and E. faecalis exhibited minimal phenotypic resistance, with WGS revealing the presence of tetracycline resistance genes. Both the isolated bacteria and meat samples harboured tetracycline resistance genes and the antibiotic residue concentrations were within acceptable limits for human consumption. In the metagenomic context, most identified bacteria were of food/meat spoilage and environmental origin. The resistome analysis primarily indicated beta-lactam, tetracycline and multidrug resistance genes. Further research is needed to understand the broader implications of these findings on environmental health and antibiotic resistance.


Introduction
Since 2005, pork consumption in South Africa has seen a significant increase [1,2].The country has a well-established pork production industry, with the largest contributions coming from the Limpopo and North West provinces, accounting for 24% and 20% of the total production, respectively [1].In South Africa, meat is sold through formal and informal channels, including butcheries, supermarkets, farms, and open markets [2].The formal sector adheres to safety checks before selling meat to consumers; informal outlets often skip these checks, posing risks of microbial contamination, antibiotic resistance, and poor meat quality [2].
Antibiotic resistance is multifaceted and has been linked to overuse and misuse of antibiotics in humans and animals, improper prescribing practices, and the widespread use of antibiotics in livestock production [3,4].Antibiotic resistance is a critical concern in pork production and other food production systems, as it can negatively impact human health [5].Antibiotic-resistant bacteria and antibiotic resistance genes (ARGs) can contaminate food at any stage, from field to retail [2,5].Previous studies have found that food products not only serve as a reservoir for antibiotic-resistant bacteria and ARGs but also act as a mediator, facilitating the transfer of antibiotic-resistant bacteria and ARGs between the environment and humans through the consumption of contaminated foods [4,[6][7][8].ARGs can also be shared among different bacterial species through horizontal gene transfer [9].Among these species are commensal flora and pathogenic foodborne pathogens, including Campylobacter spp., Enterococcus spp., Escherichia coli, and non-typhoidal Salmonella spp.(NTS), which can cause diseases in both humans and animals [2,9].
In animals, antibiotics such as sulphonamides, tetracyclines, and fluoroquinolones are widely used for therapeutic, metaphylactic, and prophylactic purposes or as growth promotors in animal feed [8,10,11].Understanding the diversity and abundance of ARGs, virulence factor (VF) genes, and antibiotic residues in food, especially retail pork meat, is important for controlling antibiotic resistance [4,10].This involves implementing effective antibiotic stewardship programs, regulating the use of antibiotics in food production, improving hygiene and sanitation practices, and promoting responsible antibiotic usage in both human and veterinary medicine [8].To address this issue, the Food and Agriculture Organization and the European Union have established tolerance levels, known as maximum residue limits (MRLs), for antibiotic residues in animal-derived food products [12].In South Africa, these MRLs are regulated by the Foodstuffs, Cosmetics and Disinfectants Act (Act No. 54 of 1972) and the Meat Safety Act (Act No. 40 of 2000) [13,14].
The aim of this study was (1) to isolate and characterise four common indicator bacteria, Campylobacter spp., Enterococcus spp., E. coli, and NTS, that overlap between humans and animals; (2) to test for antibiotic residues; and (3) to assess the microbial community as well as the resistome present in purchased raw retail pork meat.

Isolation of Indicator Bacteria
All 10 raw meat samples were tested for the selected four indicator bacteria (Table 1).E. coli was detected and isolated from one butchery raw meat sample (PC9-B4).E. faecalis were detected and isolated from two supermarket raw meat samples (PC3-S3 and PC4-S4) and one butchery raw meat sample (PC10-B5).Campylobacter spp.and Salmonella spp.were not detected in any of the collected raw meat samples (Supplementary Table S1).
Endocarditis and biofilm-associated pili genes A total of eight VF genes were identified in E. coli, and nine VF genes were identified in E. faecalis.VF genes associated with E. coli included colicins (cba, cea, cia, and cma), as well as the toxin gene (astA).VF genes associated with E. faecalis isolates included adhesins (ace and efaAfs), toxins (cyl-A, -L, and -M), and genes associated with biofilm formation (ebp-A and -B) and pheromone production (cad, camE, cCF10, and cOB1).

Antibiotic Residue Testing
The average antibiotic residue concentration detected in all the raw meat samples was less than 50 µg/kg for the majority of the tested antibiotics (Table 4).Sample PC5-S5 had an antibiotic residue concentration of 71.5 µg/kg for chlortetracycline.All tested residue levels were below the acceptable limits set by Codex/SA MRL [12].Approximately 7 million host reads (ranging from 5,797,550-8,900,728 sequencing reads) were removed from each sample prior to the metagenomic analysis (Table 5).The remaining reads were subject to further bacterial community profiling and ARG prediction analyses.

Estimated Relative Abundance
The distribution of the bacterial community structures in the raw meat samples reveals functional diversity among bacteria (Figure 1).The relative abundance was similar between raw meat samples from supermarkets and butcheries, but PC9 was different.

Resistome Prediction
Using the assembled contigs generated by MEGAHIT from the 10 raw meat samples, the PathoFact pipeline predicted a total of 61,665 ORFs.Among these, 89 ORFs were associated with ARGs, 138 with VF secretion, and 40 with toxin secretion (Supplementary Figure S1).
VF genes from all categories were present in the raw meat samples collected from supermarkets (Supplementary Table S5).No exotoxin, invasion, post-translational modification, or stress survival VF genes were detected in the raw meat samples collected from butcheries.Fifteen types of toxin genes were detected in the raw meat samples collected from supermarkets and butcheries (Figure 4B).4A; Supplementary Table S4; Supplementary Figure S3).

5; Figure
VF genes from all categories were present in the raw meat samples collected from supermarkets (Supplementary Table S5).No exotoxin, invasion, post-translational modification, or stress survival VF genes were detected in the raw meat samples collected from butcheries.Fifteen types of toxin genes were detected in the raw meat samples collected from supermarkets and butcheries (Figure 4B).

Ethical Clearance and Study Definitions
The study was conducted in accordance with the Declaration of Helsinki and approved by the Human Research Ethics Committee of the University of the Witwatersrand (M190244; 10 May 2019).
Supermarkets were defined as stores that sold raw meat and various other grocery items, while butcheries were defined as stores selling primarily raw meat commodities.Raw meat samples were defined as pork chops (alternative names included loin, rib, sirloin, top loin, and blade chops).Indicator bacteria in this study were defined as Campylobacter spp., Enterococcus spp., E. coli, and NTS spp.

Study Setting and Sampling
Raw meat samples (i.e., pork chops) were purchased on the 4 January 2022, from 10 pre-selected supermarkets and butcheries in Johannesburg and Pretoria, Gauteng.Convenience sampling was used based on the location of the laboratory, Centre for Healthcare-Associated Infections, Antimicrobial Resistance and Mycoses (CHARM), National Institute for Communicable Diseases (NICD).The stores selected for sample collection included both supermarkets (n = 5) and butcheries (n = 5).Pork chops were chosen as it is the most popular cut among consumers [21].Raw meat samples (containing at least two pork chops in the same container/packet) were randomly selected from the store shelves.All the raw meat samples were collected within the recommended dates for human consumption.Store demographics were captured by the study investigators on the day of sample collection (Supplementary Table S6).A unique number was assigned to each raw meat sample as well as the sampled stores to maintain anonymity and to ensure that the results cannot be linked back to a specific store.
The purchased raw meat samples were transported on ice to CHARM, NICD.The outsides of all the meat containers/packets were wiped with 70% ethanol before segregation and processing to avoid cross-contamination.The raw meat samples were separated aseptically.One raw meat sample from each store was sent on ice within 24 h after collection to a subcontracted laboratory in Gauteng, South Africa, for (1) the isolation of four selected indicator bacteria, and (2) antibiotic residue testing using liquid chromatographytandem mass spectrometry (LC-MS/MS).The remaining raw meat samples underwent metagenomics sequencing at the Sequencing Core Facility (SCF), NICD.All the raw meat samples were processed within 24 h to 48 h after sample collection.

AST and WGS of Isolated Indicator Bacteria from Raw Meat Samples
The bacteria isolated from the raw meat samples were transported at room temperature (20 • C to 25 • C) from the subcontracted laboratory to CHARM, NICD within 24 h after isolation.Organism identification was reconfirmed at CHARM, NICD, with matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS) (Microflex, Bruker Daltonics, Bremen, Germany).AST was performed using the Microscan Walkaway System with the Gram-negative NM44 (Beckman Coulter, Brea, CA, USA) and Grampositive PM33 (Beckman Coulter, USA) cards.The AST results were interpreted using the 2023 European Committee on antimicrobial susceptibility testing (EUCAST) guidelines [15].

Metagenomics
Total gDNA was extracted using the QIAamp Fast DNA stool mini kit (Qiagen, Hilden, Germany) and host depletion was performed using the NEBNext Microbiome DNA enrichment kit (New England Biolabs, Ipswich, MA, USA) with the inclusion of controls (the ZymoBIOMICS Gut Microbiome Standard was used as a positive control).Sequencing was performed on the NextSeq 550 (2 × 150 bp and 10 M reads) (Illumina, USA).

Discussion
The overuse and misuse of antibiotics in livestock production systems have led to residual contamination in food, resulting in antibiotic-resistant bacteria and ARGs [3,4].In this study, we investigated and categorised four indicator bacteria (Campylobacter spp., Enterococcus spp., E. coli, and NTS spp.) that overlap between human and animals.We examined the diversity and abundance of bacterial communities, as well as ARGs and antibiotic residues in raw pork meat samples.
Out of the 10 raw meat samples collected, six did not contain any of the tested indicator organisms.A single E. coli isolate was obtained from sample PC9-B4, while one E. faecalis isolate was obtained from each of the samples PC3-S3, PC4-S4, and PC10-B5.The isolated E. coli and E. faecalis isolates showed minimal phenotypic resistance.The E. coli isolate showed resistance to tigecycline and trimethoprim/sulfamethoxazole, while all three E. faecalis isolates showed resistance to moxifloxacin only.WGS data for the E. coli (PC9-B4) and E. faecalis isolates (PC3-S3, PC4-S4, and PC10-B5) showed mainly the presence of tetracycline resistance genes.
The WGS data further revealed eight and nine different VF gene categories associated with the sequenced E. coli (PC9-B4) and E. faecalis (PC3-S3, PC4-S4, and PC10-B5) isolates, respectively.VF genes enable pathogenic bacteria to colonise host niches and establish infections, contributing both directly and indirectly to disease processes [35].Aslam et al. (2012) reported that cytolysin (cyl-A, -B, -L and -M) and aggregation substances (agg) are associated with ARGs in retail meats, although the latter were not detected in our study [37,46].Other VF genes, such as ace, ebp, and efaA have been reported in E. faecalis isolates from livestock and raw meat in Ghana and hospitalised patients in Iran [40,47].
The E. coli ST10 detected in this study is a known reservoir for ARG and mobile genetic elements, such as class 1 integrons and plasmids [45,48,49].The IncB/O/K/Z and IncFII (pCoo) plasmids detected in the E. coli isolates have also been reported in E. coli from hospitalised patients in South Africa and in carbapenem-resistant Enterobacterales from hospitalised patients in Thailand [50,51].To our knowledge, E. faecalis ST30 has only been identified in hospitalised patients [52][53][54].The repUS43, repUS11, and rep9a plasmids detected in the E. faecalis isolates have been reported in other E. faecalis isolates from livestock and raw meat in Ghana as well as hospitalised patients in the USA [40,55].The repUS43 plasmid is integrated chromosomally near the tetracycline resistance gene (tetM), while the rep9a plasmid encodes the cytolysin VF gene [55].Another study also reported that the tet-M and -L and ermB resistance genes were harboured on the repUS43 and rep9a plasmids [40].Therefore, sequencing and plasmid monitoring are essential for future surveillance studies to track the spread and evolution of these genetic elements in humans, animals, and food products.
The antibiotic residue concentrations in the 10 tested raw meat samples were within acceptable limits for human consumption according to the published guidelines [12].Ramatla et al. (2017) reported higher antibiotic residue concentrations in raw meat samples collected in North West, South Africa [10].The discrepancy between their findings and ours could be due to differences in sampling frequency and sample types.
The metagenomic data revealed that most of the identified bacteria were associated with food/meat spoilage and environmental sources [16][17][18][19][20]. Their pathogenicity or ability to cause disease was not confirmed in this study.Resistome analysis revealed the presence of beta-lactam, tetracycline, and multidrug resistance genes in the 10 raw meat samples, consistent with previous local and internal studies [44,56,57].
Further research and analysis are needed to fully understand the potential implications of these findings in terms of environmental health and antibiotic resistance.This study has several limitations.Firstly, a high number of host reads (i.e., pig DNA) in the metagenomic data limited our ability to recover bacterial metagenome-assembled genomes (MAGs) from the raw meat samples.Consequently, this limited our ability to validate ARG, VF, and toxin gene predictions.Using the MAGs approach could provide more comprehensive information about the bacterial species in the meat samples.Secondly, increasing the sequencing depth or using alternative microbiome enrichment methods is recommended.Thirdly, the study's limited sample size may affect the generalisability of the findings to other supermarkets and butcheries across South Africa.

Figure 2 .
Figure 2. Composition of predicted AMR categories in 10 raw meat samples as pred PathoFact database.MLS = macrolides, lincosamides, and streptogramins; AMR category assignment by the PathoFact pipeline; PC6 = no AMR was detected.

Figure 2 .
Figure 2. Composition of predicted AMR categories in 10 raw meat samples as predicated by PathoFact database.MLS = macrolides, lincosamides, and streptogramins; AMR category = default assignment by the PathoFact pipeline; PC6 = no AMR was detected.

Figure 2 .
Figure 2. Composition of predicted AMR categories in 10 raw meat samples as predicated PathoFact database.MLS = macrolides, lincosamides, and streptogramins; AMR category = de assignment by the PathoFact pipeline; PC6 = no AMR was detected.

Figure 3 .
Figure 3. Relative abundance of individual ARGs of 10 raw meat samples (normalized as reads kilobase of reference sequence per million mapped sample reads (RPKMs)), grouped by the A category (A,C), and by AMR resistance mechanisms (B,D), stratified by supermarket and butc (C,D).Each circle indicating the relative abundance of an ARG gene in a single sample and g within the same AMR category or AMR resistance mechanism are represented with the same co

Figure 3 .
Figure 3. Relative abundance of individual ARGs of 10 raw meat samples (normalized as reads per kilobase of reference sequence per million mapped sample reads (RPKMs)), grouped by the AMR category (A,C), and by AMR resistance mechanisms (B,D), stratified by supermarket and butchery (C,D).Each circle indicating the relative abundance of an ARG gene in a single sample and genes within the same AMR category or AMR resistance mechanism are represented with the same colour.

Figure 4 .
Figure 4. Relative abundance of the individual predicted virulence factors (VFs) and toxin genes in 10 raw meat samples (normalized as reads per kilobase of reference sequence per million mapped sample reads (RPKMs)), grouped by the VF category according to VFDB (A), and by toxin gene description according to the PathoFact database (B).Each circle indicates the relative abundance of a VF/toxin gene in a single sample.Genes within the same VF category or toxin gene description are represented with the same colour.

Figure 4 .
Figure 4. Relative abundance of the individual predicted virulence factors (VFs) and toxin genes in 10 raw meat samples (normalized as reads per kilobase of reference sequence per million mapped sample reads (RPKMs)), grouped by the VF category according to VFDB (A), and by toxin gene description according to the PathoFact database (B).Each circle indicates the relative abundance of a VF/toxin gene in a single sample.Genes within the same VF category or toxin gene description are represented with the same colour.

Table 1 .
Indicator organisms isolated from raw meat samples.

Table 2 .
Antibiotic susceptibility testing of one Escherichia coli isolate and three Enterococcus faecalis isolates isolated from four raw meat samples.

Table 4 .
Antibiotic residue testing in 10 raw meat samples using liquid chromatography-tandem mass spectrometry (LC-MS/MS).

Table 5 .
Metagenomics read statistics of 10 raw meat samples.