Comparative Genomics of Plasmid-Bearing Staphylococcus aureus Strains Isolated From Various Retail Meats

Food poisoning due to the consumption of Staphylococcus aureus contaminated food is a major health problem worldwide. In this study, we sequenced the genomes of ten plasmid-bearing S. aureus strains isolated from retail beef, chicken, turkey, and pork. The chromosomes of the strains varied in size from 2,654,842 to 2,807,514 bp, and a total of 25 plasmids were identified ranging from 1.4 to 118 kb. Comparative genomic analysis revealed similarities between strains isolated from the same retail meat source, indicating an origin-specific genomic composition. Genes known to modulate attachment, invasion, and toxin production were identified in the 10 genomes. Strains from retail chicken resembled human clinical isolates with respect to virulence factors and genomic islands, and retail turkey and pork isolates shared similarity with S. aureus from livestock. Most chromosomes contained antimicrobial resistance, heavy metal resistance, and stress response genes, and several plasmids contained genes involved in antimicrobial resistance and virulence. In conclusion, the genomes of S. aureus strains isolated from retail meats showed an origin-specific composition and contained virulence and antimicrobial resistance genes similar to those present in human clinical isolates.


INTRODUCTION
The bacterium Staphylococcus aureus can incite life-threatening infections in both humans and animals (Tong et al., 2015). Although nosocomial infections are common, the acquisition of methicillin-resistant S. aureus (MRSA) from communities and livestock is also responsible for clinical cases and the spread of antimicrobial resistant (AMR) strains (Miller, 2010;Mediavilla et al., 2012). The virulence factors, exoenzymes, and toxins produced by S. aureus genomes are responsible for pathogenesis (Tam and Torres, 2019). Staphylococcal food poisoning (SFP) related to consumption of contaminated food is a major problem worldwide, and staphylococcal enterotoxins are responsible for clinical symptoms (Hennekinne et al., 2012;Lin et al., 2016). In the United States, S. aureus remains responsible for about 241,148 annual cases of domestically acquired foodborne illness (Hedberg, 2011), which is typified by nausea and vomiting.
Many studies describe the isolation and characterization of S. aureus strains from retail food products. However, few studies have used WGS to characterize strains from food products (Mossong et al., 2015;Ge et al., 2017;Sivaraman et al., 2017), and most of these focused on MRSA or strains related to food poisoning outbreaks (Mossong et al., 2015;Ge et al., 2017;Sivaraman et al., 2017). To our knowledge, WGS comparisons are lacking for S. aureus isolated from retail meats. We previously isolated multiple S. aureus strains from retail meat products and screened them for virulence and antimicrobial susceptibility (Abdalrahman and Fakhr, 2015;Abdalrahman et al., 2015a,b). The aims of this study were to sequence the whole genome of a selected set of plasmid-bearing S. aureus strains previously isolated from various retail meats, and to perform comparative genomics analysis to explore their potential virulence, AMR, and origin.

Bacterial Cultures and DNA Isolation
The 10 S. aureus strains used in this study were previously isolated from retail beef (n = 3), chicken (n = 4), turkey (n = 2), and pork (n = 1); all strains contained plasmid DNA based on alkaline lysis and PFGE (Table 1; Abdalrahman and Fakhr, 2015;Abdalrahman et al., 2015a,b). S. aureus strains were selected to represent various retail meat sources (beef, beef liver, chicken, chicken liver, chicken gizzard, pork, and turkey), different plasmid sizes previously determined by PFGE, and various plasmid rep types (Neyaz, 2019;Neyaz et al., 2020). DNA isolation and sequencing were carried out as described previously with minor modifications Fakhr, 2016a,b,c, 2017a,b;Neyaz et al., 2019). Briefly, S. aureus cells were grown in tryptic soy agar (TSA) at 37 • C for 16-24 h and then used for genomic and plasmid DNA isolation. The DNeasy Blood and Tissue kit (QIAGEN Inc., Valencia, CA, United States) was used for genomic DNA isolation. The Qubit dsDNA HS Assay kit (Life Technologies, Carlsbad, CA, United States) was used for DNA quantification with a Qubit 2.0 fluorimeter.

Library Preparation, Sequencing and Assembly
Sequencing libraries of genomic DNA were prepared and normalized with the Nextera XT DNA Library Preparation Kit (Illumina Inc., San Diego, CA, United States) as recommended by the manufacturer. Sequencing of prepared libraries was conducted in the Illumina MiSeq platform using the Illumina MiSeq V2 Reagent kit and 2 × 250 cycles. Sequence assembly was conducted using the CLC Genomic Workbench v. 7.0 and the microbial genome finishing module. Identification, clustering and segregation of plasmid sequences from chromosomal sequences was performed with plasmidSPAdes (Antipov et al., 2016) and the PHASTER web server (Arndt et al., 2016). Several contigs were joined manually by consulting reference sequences. Assembled genomic and plasmid sequences were deposited in GenBank as listed in Table 1. Sequences were annotated using the NCBI Prokaryotic Genome Annotation pipeline. RAST 1 (Overbeek et al., 2014) and PATRIC v. 3.5.39 2 (Wattam et al., 2014) tools were used to annotate whole genomic sequences of S. aureus for comparative genomic analysis.

Identification of Virulence Factors and AMR Genes
Assembled FASTA files of whole genomic sequences derived from the CLC workbench were used for virulence gene prediction using VFanalyzer 9 (Liu et al., 2019), and the results were tallied with RAST-and PATRIC-annotated genomic features. Curated S. aureus virulence gene sequences from VFDB 10 (Liu et al., 2019) were manually blasted against whole genome sequences of all 10 S. aureus strains. RAST-and PATRIC-annotated virulence genes that were identified in the virulence, defense and diseases subsystems and stress response genes (stress response subsystem) were listed and compared. Raw FASTA files or assembled contigs of whole genome sequences were used in RGI 11 , ResFinder 3.1 12 (Zankari et al., 2012), and CARD analysis (PATRIC annotation pipeline) for the prediction of AMR genes. Island viewer 4 13 (Bertelli et al., 2017) was used for the identification and visualization of genomic islands. Genomic arrangements in νSaα, νSaβ, and SaPI genomic islands and the Type VII secretion system were illustrated using genomic features from Seed Viewer (RAST) 14 . S. aureus NCTC8325 was used as a reference strain for studying the Type VII secretion system.

Phylogenetic Analysis Among Virulence Genes
Protein sequences for clfA, fnbA, and spa genes were downloaded from the RAST server. Protein sequence alignments were conducted with MUSCLE (Edgar, 2004) 15 and analyses for maximum likelihood were conducted in Mega X (Kumar et al., 2018) with a bootstrap value of 100.

Features of Genomic and Plasmid DNA Sequences
Details for the ten sequenced S. aureus strains including size, contig number, and G+C ratio are shown in Table 1. Chromosomes ranged from 2,654,842 to 2,807,514 bp, and the average G+C ratio was 32.26-33.07. Twenty-five plasmids were detected and sequenced from the 10 S. aureus strains, and all harbored one or more plasmids ( Table 1). Plasmids ranged from 1.4 to 118 kb with a mean G+C ratio of 26.46-36.19. The mean ORF length ranged from 2447 to 2704 bp. Between 1 and 6 ORFs were identified in small plasmids (<5 kb), and 6-184 ORFs were present in larger plasmids (>5 kb).

Comparative Genomics
Sequenced genomes from the same retail meat product showed greater similarity (%) to each other than sequences from different meats ( Figure 1A). Sequence similarity values obtained from BLASTn analysis ranged from 97.93 to 99.99% ( Figure 1A). Despite their variable origin (retail chicken, chicken liver, and chicken gizzard), all chicken strains shared similar MLST and spa types (ST5, t002). Similar results were found among beef (ST1159, t091), pork (ST9, t3446), and turkey isolates (ST398, t034). Nucleotide difference (ND) trees were created from genomic sequences, and NDtree v. 1.2 revealed clusters specific for meat sources (Figure 1B). Genome tree report (Supplementary Figure S1) showed that retail beef strains from our laboratory clustered together and most of the strains in neighboring clusters were clinical isolates from human host (Supplementary Figure S2 and Supplementary Table S1). Only two strain G08M and G11F in the neighboring clusters were from cattle (mastitis) which were isolated from milk. All retail chicken isolates from our study clustered with S. aureus strains mainly from broiler chicken (Supplementary Figure S3 and Supplementary Table S2). Strain B4-59C was found in the same cluster with S. aureus ED98 which has been studied in a previous report (Lowder et al., 2009). Blast analysis showed 99.98% similarity between genomic sequences of B4-59C and ED98 (Supplementary Figure S3). Retail pork isolate B9-22D was found to be clustered with mainly pork and cow isolates in Genome Tree report (Supplementary Figure S4 and Supplementary Table S3). Neighboring clusters included strains mainly from poultry and livestock. Likewise, most strains found in cluster with S. aureus strain B3-14B and B6-55A (retail turkey isolates) were from turkey source (Supplementary Figure S5 and Supplementary Table S4). Sequences were annotated and then analyzed for virulence, AMR and stress response genes ( Figure 2B). Pangenome analysis showed differences between the S. aureus genomes from retail meats and genomes from reference clinical isolates (Figure 2A), most notably in hypothetical proteins, mobile elements and phage proteins (Supplementary Table S5).

Virulence Factors
Virulence and toxin genes were predicted and identified using VFanalyzer and RAST annotation, and the resulting patterns reflected the meat origin of the isolates (Figures 3A,B and Supplementary Table S6). Among genes related to adherence, differences were seen for collagen adhesion (cna), sdrD, and sdrE. The sdrD and sdrE genes were harbored by both beef and chicken isolates, whereas cna was only present in turkey isolates. Genes encoding clfA and eap/map were present in all RASTannotated sequences but were not observed with VFanalyzer. Comparative analysis of protein sequences derived from cflA, fnbA, and spa genes revealed differences that were inferred from maximum likelihood analysis (Figures 4A-C). spa gene has been annotated as pseudogene in NCBI Prokaryotic Genome Annotation Pipeline (PGAP) for all retail chicken isolates. Retail chicken isolates grouped with clinical reference strains; however, beef, pork and turkey isolates were genetically distinct from most clinical strains (Figures 4A-C). Comparative analysis of the genomic sequences on the individual gene level supports the results obtained from the nucleotide difference tree created using whole genome sequences. Other adherence-related genes, including atl, ebh, clfB, ebp, efb, fnbB, icaA-D, icaR, and sdrC, were present in all sequenced strains. Among the exoenzymes, serine protease genes (splA, splB, splC, splD, and splF) were present in beef and chicken isolates; however, splE was only identified in beef isolates. Genes for cysteine protease, staphylocoagulase, and thermonuclease were present in all strains, but staphylokinase was absent.
Genes encoding sphingomyelinase (Supplementary Table S6) and hemolysin (hly/hla, hld, and hlgA-C) were identified in all strains, and a truncated hlb was present in chicken isolates (RAST annotation). Exotoxin genes (set6, set7 set10, set11, set13, and set14) were found in all strains, and chicken and pork isolates harbored one or more enterotoxin genes such as seg, sei, ent1/2 selm, seln, slo, and selq. Leukotoxin genes lukD and lukE were present in beef and chicken isolates, but absent in pork and turkey isolates. lukF-PV and lukS-PV were identified in all 10 RASTannotated sequences; however, BLASTp annotated these as LukG and LukH leukotoxins in retail beef, chicken, and pork isolates. Isolates from retail turkey contained genes for β-pore-forming cytolysin and leucocidin S.
All 10 genomes contained genes related to the Type VII secretion system (e.g., ssaA, esxA, esaA, esaB, and essA-C). Beef strains also contained esaC and esxB and higher numbers of repeats than reference strain S. aureus NCTC8325 (Figure 5A). Although chicken and pork isolates shared similarity with reference strain NCTC8325, they contained proteins with unidentified functions instead of esaE, esxD, or essD ( Figure 5A). Retail turkey isolates lacked esxC, esxB and the SAV0291 homolog, and exhibited differences in the repeat regions.

Genomic Islands
The genomic island νSaα, containing exotoxin set genes and a tandem lipoprotein cluster, was present in all genomes; however, differences were observed in set gene arrangement ( Figure 5B). The νSaβ island, which encodes the enterotoxin gene cluster (egc), was present in chicken and pork isolates ( Figure 5C). Beef isolates lacked an enterotoxin gene cluster but possess a serine protease cluster. Interestingly, the toxic shock syndrome toxin gene (tsst) was encoded by the SaPI genomic island in strains B4-59C ( Figure 5D) and B3-17D.

AMR and Stress Response Genes
Various AMR genes were identified in the sequenced genomes ( Figure 6A). The tetracycline resistance (Tet R ) gene tetM was present in the chromosome of turkey isolates, and tetK, tetL, and tetT were plasmid-encoded. Beef and chicken isolates harbored similar AMR genes ( Figure 6A). All strains encoded bacitracin stress response genes bceA, bceB, bceR, and bceS ( Figure 3B).
Genes related to multidrug resistance, fluoroquinolone resistance (parC parE, gyrA gyrB) and multidrug reistance efflux pumps (acrB, MATE family MDR pump) were present in all strains ( Figure 3B). The fosfomycin-resistance gene fosB was present in chicken and pork isolates but absent in beef and turkey isolates ( Figure 3B). All strains harbored genes related to cobalt/zinc/cadmium resistance (CzcD, CzrB, and TR) and the stress response subsystem ( Table 2). Genes encoding nitric oxide reductase (EC1.7.99.7) and a quinol-dependent gene were only present in turkey isolates.
pSALNT46 encoded β-lactamase (BlaZ family) and arsenate resistance genes (Supplementary Figure S9A). Similarly, blaI, blaR, and Tn552 transposase genes were encoded by pSALNT46 and pSALNP58. Plasmid pSALNP58 harbored heavy metal transporting ATPase, aminoglycoside N(6 )-acetyltransferase, FIGURE 5 | (A) Organization of gene clusters encoding the Type VII secretion system in S. aureus strains according to RAST annotation. The functional map for reference strain S. aureus NCTC8325 was adapted from a previous study (Warne et al., 2016). Cross-hatched and gray arrows indicate proteins with unknown functions. S. aureus strain B1-4A is a beef isolate that shares similarity with beef strains B2-7A and B3-4A, and the chicken isolate B4-59C is similar to the other three chicken strains (B3-17D, B8-13D, and B2-15A) and the pork isolate B9-22D. The map for the turkey isolate B3-14B is shown and shares similarity with B6-55A, which is also from turkey. (B) Organization of the νSaα genomic island and exotoxin cluster (set genes) in S. aureus strains. Genes were predicted from VFanalyzer and drawn to scale based on the genomic arrangement in RAST seed viewer. Arrows represent the following: red, coding sequence (CDS) for exotoxins; yellow, CDS for tandem lipoproteins; purple boundary, transposase; and gray, hypothetical proteins with unknown functions. S. aureus strains include representative isolates from beef (B1-4A), chicken (B3-17D), pork (B9-22D), and turkey (B3-14B). (C) νSaβ genomic island with serine proteases and enterotoxin gene cluster (seg, sei, ent1/2, selm, seln, selo, selq) in S. aureus strains. Genes were predicted from VFanalyzer and drawn to scale using the arrangement observed with the RAST seed viewer. Arrows represent the following: red, CDS for serine proteases; green, CDS for enterotoxins; purple, putative transposase; and gray, hypothetical proteins. S. aureus strains include representative isolates from beef (B1-4A), chicken (B4-59C), and pork (B9-22D). (D) S. aureus pathogenicity island (SaPI) carrying the gene for toxic shock syndrome toxin-1 and seb in S. aureus strain B4-59C. Genes were predicted from VFanalyzer and drawn to scale using the genomic arrangement observed with RAST seed viewer. and traC/D/L and trsB. The small multi-drug resistance (SMR) family protein gene was encoded by pSALNBL75 (Supplementary Figure S9B). The pSALNT20 sequence was comprised primarily of unique phage genes (Supplementary Figure S10B), and other phage-related genes were encoded by pSALNB22, pSALNB86, pSALNT106, and pSALNBL118 (Supplementary Figure S10 and Supplementary Table S8).

DISCUSSION
The S. aureus strains used in this study were previously isolated from retail meat products and characterized for enterotoxin genes and antimicrobial susceptibility (Abdalrahman and Fakhr, 2015;Abdalrahman et al., 2015a,b). Although the number of S. aureus strains used here was relatively small (n = 10), an origin-specific genomic composition for the strains was clearly observed. Genomic comparisons ( Figure 1A) and clustering of S. aureus strains ( Figure 1B) indicate that strains from the same meat origin (e.g., beef, chicken, turkey or pork) shared a similar genomic composition. Genome report tree showed the clustering of genomes of retail chicken isolate from our study with poultry isolates, retail pork isolate with pork isolates, and retail turkey isolates with turkey isolates from GenBank (Supplementary Figure S1-S5 and Supplementary Tables S2-S5). This support that these isolates originated from respective animal sources either due to systemic infection, colonization or during slaughtering and processing. Interestingly, retail beef isolates didn't cluster with any other strains, but neighboring cluster mainly includes strains from human sources. Most of the whole genome sequenced S. aureus sequences in GenBank are from human sources. Hence, it is not surprising to see most of the strains from retail meat sources clustering with human strains. Meanwhile, previous reports from our laboratory had also showed that S. aureus strains mainly originated from retail beef (retail beef liver) were similar to S. aureus strains from human sources (Abdalrahman and Fakhr, 2015;Abdalrahman et al., 2015a,b).
NDtree constructed using our strains and few reference clinical strains has shown the proximity of retail chicken isolates with JH1, Mu50, and N315 clinical strains ( Figure 1B). Retail chicken isolates and these clinical strains are also found in neighboring clusters in Genome report tree (Supplementary Figure S1). The arrangement of the νSaα genomic island in the four retail chicken isolates was similar to the Type I νSaα genomic island in clinical strain S. aureus N315 (Baba et al., 2002(Baba et al., , 2008. Retail chicken isolates B3-17D and B4-59C harbored the toxic shock syndrome toxin gene (tsst), which was 100% similar to tsst in clinical strain N315. Interestingly, prior studies suggested a human origin for S. aureus isolated from retail chicken (Abdalrahman and Fakhr, 2015;Abdalrahman et al., 2015a). However, the origin of poultry ST5 clade has been suggested to be due to the recent human to poultry host jump with adaptation in poultry hosts (Lowder et al., 2009). The Genome tree report showed the clustering of WGS retail chicken isolates from our laboratory with genomes of S. aureus strains from chicken in GenBank (Supplementary Figure S3 and Supplementary  Table S2). Strain B4-59C shared higher genomic similarity with strain ED98 (Supplementary Figure S3). Many functional genes, found in human clinical strains which play role in human host pathogenesis, had been documented to have mutation to create pseudogenes among poultry associated S. aureus strains (Lowder et al., 2009). Acquisition of novel mobile genetic element had also played major role for genetic diversification among poultry specific ST5 S. aureus strains. Meanwhile, enhanced resistance toward the chicken heterophils was found by poultry specific S. aureus ED98 (Lowder et al., 2009). Many genes including spa and asp1 were found to be pseudogenes in annotation from PGAP for all our retail chicken isolates similar like S. aureus ED98. Presence of poultry specific plasmids (pAvY-B1, pAvY, pAVX) Retail beef isolates: B1-4A, B2-7A, and B3-4A; retail chicken isolates: B3-17D, B4-59C, B8-13D, and B2-15A; retail pork isolate: B9-22D; and retail turkey isolates: B3-14B and B6-55A. All genes were present among all S. aureus isolates except Nitric-oxide reductase (EC 1.7.99.7), quinolo-dependent which was found absent among S. aureus isolates from retail beef, retail chicken and retail pork.  (Table 3) in retail chicken isolates from our study which supports their poultry associated origin. Although the presence of similar virulence genes is predicted in genomes of retail chicken isolates from this study with reference clinical strains, their virulence on chicken or human host have not been assayed which remains a limitation of this study and warrants for future investigation. The pork isolate was assigned to ST9, a MLST group prevalent in pigs (Neela et al., 2009), which supports a livestock origin for pork isolate B9-22D. Many S. aureus strains found in cluster with B9-22D were isolated from pork skin and pig source (Supplementary Figure S4 and Supplementary Table S3). The turkey isolates were assigned to ST398, that also includes a dominant type of livestock-associated (LA) MSRA (Smith and Pearson, 2011) in CC398. Differences in virulence and AMR genes were previously observed in LA-MRSA (CC398) and community-associated (CA)-MRSA strains in a mouse model (Randad et al., 2019). Although ST398 MRSA had been well documented for its involvement in clinical cases in livestock and human, methicillin susceptible ST398 (MSSA) has also been reported as emerging public health threat (Valentin-Domelier et al., 2011). ST398 strains causing human infection are found not only as livestock associated but also found in livestock free environment (Diene et al., 2017). Previous studies had shown major factor influencing for pathogenesis in human hosts is the presence of prophages in S. aureus genome carrying virulence related genes (Diene et al., 2017;Kashif et al., 2019). Although, two retail turkey isolates used in this study (B3-14B and B6-55A) lack mec genes, phenotypic resistance toward oxacillin and cefoxitin was observed for these strains (Abdalrahman and Fakhr, 2015;Abdalrahman et al., 2015a,b). In this study, we observed that virulence factors in LA-CC398 S. aureus strains IHW398-1 and IHW398-2 (Randad et al., 2019) were similar to B3-14B and B6-55A. Presence of tetM gene, Tn916 and lack of ϕ3 prophage indicate livestock origin of these strains. In genome report tree, both retail turkey isolates cluster with S. aureus strains (MRSA as well as MSSA) from turkey source (Supplementary Figure S5). Interestingly, strain B3-14B from our study is found highly similar to SA33924 strain that had been described as human adapted ST398 (blood stream infection isolate) in a previous report (Diene et al., 2017).
The S. aureus strains sequenced herein encoded virulencerelated genes associated with adherence, enzyme secretion, immune system evasion, toxins, and leucocidins. Lipoproteins are known to function in immune system stimulation and invasion in human cells (Nguyen et al., 2015), which were also harbored by all our strains. Furthermore, the type VII secretion system genes esxA, esaA, esaB, and essA-C (Warne et al., 2016) were present in all strains. Meanwhile, these strains also lack many virulence related genes and phage proteins found in studied reference clinical strains (Supplementary Table S6). Presence of virulence related genes and AMR genes might play role in pathogenesis and adaptation in animals and human hosts. However, the virulence potential of these strains has not been assayed in this study which limits our assertion for their pathogenic nature in animals or human hosts. Enterotoxin genes were identified in retail chicken and pork strains (Figure 5C), which suggests a potential role in enterotoxin-mediated food poisoning.
The S. aureus strains in this study were resistant to antimicrobial compounds ( Figure 6B; Abdalrahman and Fakhr, 2015;Abdalrahman et al., 2015a,b). Despite having mecA, mecB or mecC in their genome, all sequenced strains were resistant to ampicillin. Several strains (B8-13D, B9-22D, B3-14B, and B6-55A) found to be phenotypic MRSA and were resistant to oxacillin (Abdalrahman and Fakhr, 2015;Abdalrahman et al., 2015a,b). Antimicrobial resistance genes detected in plasmids sequenced in our study (Figures 6A,B) correlate with the phenotypic antibiotic resistance patterns seen in the S. aureus strains harboring them ( Figure 6B). Ampicillin resistance might be partially mediated by pSALNT46 and pSALNP58, which contain Tn552-encoded blaI/blaR genes. The pork isolate B9-22D was kanamycin resistant, and the aminoglycoside 4nucleotidyltransferase gene borne on pSALNP9 might impart Km resistance. Strains B3-4A, B4-59C, B8-13D, B9-22D, B3-14B, and B6-55A were Tet R and encode tet(38) in their genome (Figures 3B, 6A). The gene tetM is commonly encoded by Tn916 and Tn5801 in S. aureus from humans or by Tn916 in animals (De Vries et al., 2009). The two turkey isolates harbor tetM in Tn916; furthermore, pSALNT4.9 and pSALNP9 plasmids in B9-22D and B3-14B were shown to encode tetK, tetL, and tetT. Plasmids pSALNB2.8 and pSALNBL75 from retail beef isolates harbored genes encoding SMR proteins; this is significant because SMR family proteins confer resistance to antispetics and quaternary ammonium products (Bay and Turner, 2009;Costa et al., 2013). The presence of genes on plasmids for arsenic efflux pump protein, cobalt zinc cadmium resistance protein, Pb, Cd, Zn, and Hg transporting ATPase, multicopper oxidase, SMR proteins might play possible roles in AMR and heavy metal resistance mechanisms of their hosts. It also emphasizes the role of plasmids in dissimination of resistance genes among S. aureus strains. The unique genomic composition of pSALNP2.8 and pSALNP9 plasmids when compared to all available plasmid sequences in GenBank warrants further studies in regards to plasmids from retail pork isolates. Phage or phage-like sequences are prevalent in S. aureus strains and are important for horizontal gene transfer (Goerke et al., 2009). Plasmids pSALNB22, pSALNT20, pSALNB86, pSALNBL75, pSALNT46, and pSAlNT106 contained phage-related proteins, and these plasmids might be important in the transmission of virulence factors.
Lineage specific distribution of plasmids has been suggested in regards to the presence of resistance and virulence determinants (McCarthy and Lindsay, 2012). Plasmids in S. aureus have been categorized according to rep gene families (McCarthy and Lindsay, 2012) as well as the presence of functional genes (Bukowski et al., 2019). Poultry specific plasmids (Group I) (Bukowski et al., 2019) like pAvY-B1 (pSALNCL1.4, pSALNC1.4-1, pSALNC1.4-2, and pSALNCG1.5) and pAvY plasmids (pSALNC1.6) are cryptic plasmids. Likewise, another small plasmid, pSALNT2.2, from a retail turkey isolate bears only a gene coding for a replication protein. Beneficiary roles of these cryptic plasmids for the S. aureus host strains are yet to be identified. One study has hypothesized that cryptic plasmids serve as flexible vectors to acquire mobile genomic elements which leads to the origination of plasmids with known functional genes (Attéré et al., 2017). Staphopain A (Laarman et al., 2012) and lysophospholipase (Daugherty and Low, 1993) are well documented virulence factors which are found prevalent in poultry specific pAvX plasmids sequenced in our study. pSALNC2.8 shares 100% genomic similarity with pSALNC17 and pSALNCG17 and other pAvX plasmids available in GenBank despite the discrepencies on their sizes. Although, some plasmids (pSALNB2.8, pSALNP2.8, pSALNC2.8, and pSALNBL2.8) have similar size (∼2.8 kb), origin specific differences could be seen among these plasmids (Supplementary Table S3).
In summary, staphylococcal food poisoning (SFP) remains a major health problem, and outbreaks have been documented around the globe (Argudín et al., 2010;Kadariya et al., 2014;Johler et al., 2015;Mossong et al., 2015;Vitale et al., 2015;Stewart, 2017;Thapaliya et al., 2017). Contamination of retail food products with S. aureus and the presence of enterotoxins in foodborne strains has increased the potential for SFP outbreaks (Crago et al., 2012;Abdalrahman et al., 2015a,b;Jamali et al., 2015;Mossong et al., 2015;Al-Ashmawy et al., 2016;Ge et al., 2017). The 10 S. aureus strains isolated from retail meat products in this study contain various virulence factors and AMR genes. Sequencing more strains from various retail meat products will be ideal to represent the population of S. aureus circulating in retail meat. As concluding remarks, genomic analysis supported that all sequenced S. aureus strains from retail meat products in this study originated from respective animal sources. Source specific differences in virulence factors and genomic composition was found among the strains in comparative genomic analysis. As S. aureus strains with potential viurlence factors and stress response genes are retained in retail meat products even after processng and storage conditions, these strains might play role in possible food poisioning outbreaks and other clinical cases. Hence, enhancement in food safety measures for reduction of foodborne pathogens including S. aureus contamination is required. The pathogenicity of these strains studied in this study has not been tested but remains possible and warrants further investigation.

DATA AVAILABILITY STATEMENT
The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found in the article/Supplementary Material.

AUTHOR CONTRIBUTIONS
MF: research design. AK and LN: experimental procedures, sequence assembly, and comparative genomics analysis. AK, LN, and MF: manuscript preparation. All authors contributed to the article and approved the submitted version.