Molecular Characterization of Staphylococcus aureus Isolated from Raw Milk and Humans in Eastern Tanzania: Genetic Diversity and Inter-Host Transmission

Staphylococcus aureus is a common cause of infection in humans and animals, including bovine mastitis, globally. The objective of this study was to genetically characterize a collection of S. aureus isolates recovered from milk and nasal swabs from humans with and without animal contact (bovine = 43, human = 12). Using whole genome sequencing (NextSeq550), isolates were sequence typed, screened for antimicrobial resistance and virulence genes and examined for possible inter-species host transmission. Multi locus sequence typing (MLST) and single nucleotide polymorphism (SNP)-based phylogeny revealed 14 different sequence types, including the following six novel sequence types: ST7840, 7841, 7845, 7846, 7847, and 7848. The SNP tree confirmed that MLST clustering occurred most commonly within CC97, CC5477, and CC152. ResFinder analysis revealed five common antibiotic resistance genes, namely tet(K), blaZ, dfrG, erm©, and str, encoding for different antibiotics. mecA was discovered in one human isolate only. Multidrug resistance was observed in 25% of the isolates, predominantly in CC152 (7/8) and CC121 (3/4). Known bovine S. aureus (CC97) were collected in humans and known human S. aureus lineages (CC152) were collected in cattle; additionally, when these were compared to bovine-isolated CC97 and human-isolated CC152, respectively, no genetic distinction could be observed. This is suggestive of inter-host transmission and supports the need for surveillance of the human–animal interface.


Introduction
Staphylococcus aureus is a commensal bacterium found on human and animal skin as well as mucous membranes. As a pathogen, S. aureus is associated with causing a wide range of infections ranging from mild skin infections to more life-threatening infections such as pneumonia, endocarditis, bloodstream infections, and food poisoning [1,2]. S. aureus is also the leading cause of mastitis, one of the most important and economically costly infections in dairy cattle [3]. The success of S. aureus as a pathogen might have emerged as a result of massive control programs against Streptococcus agalactiae, a bacterium associated with subclinical mastitis, resulting in low milk quality and low yield [4]. The elimination of Streptococcus agalactiae in dairy production farms has notably facilitated the rise of S. aureus,

Ethical Approval
The Vice-Chancellor of the University of Dar es Salaam issued a research permit letter on behalf of the National Institute of Medical Research (NIMR) and the Tanzania Commission for Science and Technology (COSTECH) with reference no. AB3/12 (B). Further permission to carry out this study was granted by the Executive Directors of the Handeni, Bagamoyo, and Morogoro urban districts. In addition, written consent was provided by all human participants prior to being sampled.

Study Site
Three eastern Tanzanian districts were included in this study. Bagamoyo (Pwani region) and Morogoro urban (Morogoro region) were categorised as semi-urban districts, and Handeni (Tanga region) was categorised as a rural district.

Study Design
This was a cross-sectional study conducted between September 2017 and December 2017 in the Eastern zone of Tanzania. The study involved the Tanga, Coast, Morogoro, and Dar es Salaam regions. Tanga, Coast, and Morogoro regions are categorized as moderately dense livestock-keeping communities (Ministry of Livestock and Fisheries report 2015/2016 [27]). Eighty-four S. aureus isolates were collected from the raw milk of seemingly healthy cattle from 16 dairy farms in the study area (out of 43 included in the study). Six human S. aureus nasal swab isolates from humans with animal contact who worked or resided on the sampled dairy farms were taken and all were included in the study. Further samples from humans with no animal contact were taken in Dar es Salaam, the commercial capital of Tanzania. Out of 20 S. aureus isolates identified from this group, 6 were included in this study.

Sample Collection
Dairy farms housing between 20 and 500 cattle were included. The sample size for each region was calculated by using a stratified method, whereby the three regions of interest, i.e., Morogoro, Coast, and Tanga, acted as strata. The estimated population of dairy cattle was established, and an assumption was made of constant prevalence of S. aureus of 50% across all regions with a precision within 10% of the true prevalence. A 95% confidence interval was used.
At district level, extension officers provided lists of farms in the given area. Purposive sampling was used to choose qualifying farms to be interviewed and sampled. The study team visited the chosen farms to ask for consent and to determine the exact number of animals on each farm. This information was used to calculate the exact sample size to be collected on each farm. This was calculated considering the number of animals on the farm, the number of animals on all farms to be sampled, and the calculated sample size in the particular region. About 8-10 mL of midstream milk was collected in sterile Falcon tubes from individual cattle during the milking process. Nasal swabs from people residing or working on the same farm were collected consecutively. After obtaining their consent for participation, sterile cotton swabs were dipped in distilled water and gently rubbed against both inner nares of the participant's nose. Samples for the group of people with no animal contact were collected from drug addicts residing in the commercial city of Dar es Salaam. After obtaining consent and completing a questionnaire to ensure the participant had had no contact with any animal in the past 12 months, sampling took place using the same procedures as in people with animal contact. The samples were then dipped into Stuart Transport Medium in 15 mL Falcon tubes. All types of samples were kept in a cooler box at temperatures ranging from 4-8 • C before being transported to the laboratory for further analysis.

Identification
Milk samples were pre-enriched with buffered peptone water at a 9:1 ratio (9 mL of peptone water to 1 mL of milk sample), and the mixture was incubated for 24 h at 37 • C [28]. After the milk enrichment step, all samples were treated the same, and were cultured aerobically at 35 • C for 18-24 h on Columbia 5% Sheep Blood Agar (Biorad, Marnes-la-Coquette, France) and Mannitol Salt Agar (Oxoid Ltd., Basingstoke, UK). S. aureus colonies were presumptively identified by morphology and haemolysis on Blood Agar, Mannitol fermentation (yellowish colonies), Gram staining and catalase production. Identification of S. aureus was confirmed by a tube coagulase test (BD BBL Coagulase plasma, Rabbit) [25,29].

DNA Extraction and Whole Genome Sequencing
Genomic DNA was extracted using a QIAamp DNA min kit (Qiagen GmbH, Hilden, Germany). The quality and quantity of genomic DNA were confirmed using a Qubit 2.0 fluorometer, (Thermal Fisher Scientific, Waltham, MA, USA). Library preparation (dual indexing) was performed using an Illumina DNA prep kit, (Illumina Inc., San Diego, CA, USA). Whole genome sequencing of the library was completed on an Illumina NextSeq 500 platform using a paired-end 2 × 150 bp protocol.

Bioinformatic Analysis
Species identification, multilocus sequence typing (MLST), Spa typing, and identification of antimicrobial resistance and virulence genes were carried out using Center for Genomic Epidemiology web-based tools (

Results
A total of 43 S. aureus were isolated from cows' milk samples from the main local milksupplying farms in different Tanzanian regions. Twelve of the S. aureus isolates included were of human origin, six of which originated from people residing at or working on the same dairy farms as the sampled cattle and six from people with no animal contact (drug addicts) residing in Dar es Salaam. The isolates were sequenced using NextSeq 550 in order to determine sequence types, screen for antimicrobial resistance and virulence genes. Further genetic relatedness between the animal-and human-isolated S. aureus and evidence of possible inter-host transfer was evaluated. A summary of the findings can be found in Tables 1 and 2.

Antimicrobial Resistance, Virulence, Leukocidin Genes, and Spa Typing
ResFinder analysis revealed a limited number of antibiotic resistance genes among the S. aureus in this study, including tet(K), blaZ, dfrG, erm(C), str, qacG and fosB encoding for tetracycline, penicillin, trimethoprim, macrolide, lincosamide, streptogramin B (MLSB), and multidrug efflux pumps. mecA was discovered in one of the 55 isolates included in the study. Specificity of resistance genes, virulence factors, toxin genes, and leukocidin genes in different sequence types could also be observed (Table 2). ST7846 was isolated from cattle and humans (BH00403) in the Morogoro and Pwani regions. The isolates uniformly conferred the penicillin resistance blaZ gene, whereas the majority also conferred the str and tet(K) genes, exhibiting either the blaZ/tetK or blaZ/str combination. None of the isolates exhibited all three genes together. The sequence type also homogeneously possessed the lukD/E leukocidin genes, as well as the aur, hlgA, hlgB, hlgC, and splA/B virulence genes. All isolates in this sequence type belonged to Spa type t1236.  ST97 isolates were also collected in humans (DADST084, DADST088) and milk in Tanga, Pwani, and Dar es Salaam. All isolates belonging to this sequence type possessed the blaZ resistance gene, further the blaZ/tetK combination was also frequently observed. Isolates in this group exhibited lukD/E leukocidin genes as well as aur, hlgA, hlgB, hlgC, and splA/B virulence genes which were Spa typed t9432. Four within the ST97 sequence type possessed the splE virulence gene, which were also exclusively screened with the sak and scn toxic genes (Spa typed t267).
The ST5477 isolates in this study were of bovine origin, collected in the Tanga and Pwani regions, all of which encoded for the blaZ gene. Moreover, all ST5477 possessed aur, edinB, hlgA/C, and splA/B virulence genes, and a few exhibited an additional hlgB gene. This ST was further prone to the egc-cluster enterotoxin genes (seg, sei, sem, seo, and sen) and homogeneously encoded for the toxin shock tst gene. Further, all ST5477 encoded for only the lukE leukocidin gene, with the exception of one that exhibited the lukD/E combination. Only two isolates in this group were assigned Spa types, namely t18852 and t18853.
The novel ST7841 (Spa type t042) was exclusive to bovine isolates collected in the Pwani region. This sequence type exhibited blaZ and str resistance as well as aur, hlgA, hlgB, hlgC, splA, splB, and splE virulence genes. All were screened with lukD/E leukocidin genes and not toxin genes.
Novel ST7840 (Spa type t1398) isolates were also of bovine origin and exclusively isolated in the Morogoro region. All three exhibited the lukD leukocidin gene as well as aur, edinB, hlgA, hlgB, hlgC, splA, and splB virulence genes coupled with a range of egc-cluster enterotoxin genes. Two of the three ST7841 isolates were not detected with any resistance genes, and the one that was detected with one was screened with the erm(C) resistance gene. Furthermore, the novel ST7845 was another bovine-exclusive ST collected in the Pwani region. Two isolates belonged to this ST, and neither presented any toxin genes. One of the two was screened with the blaZ gene, whereas the other was detected with blaZ/str/tet(K) resistance genes. Both presented aur, edinB, hlgA, hlgB, and hlgC virulence genes and lukE as their leukocidin gene.
ST152 was observed in both human-and cattle-originating samples collected in the Tanga, Pwani, and Morogoro regions. Regardless of origin, the most common resistance genes in this sequence type included blaZ, dfrG, erm(C), and tet(K). The majority of the isolates in this ST encoded for edinB, hlgA, and hlgB virulence genes and exclusively for the sak and scn toxic genes. Additionally, all ST152 were PVL-positive screened with the lukF/S-PV genes. All of the ST152 were Spa typed t355, with exception of one that was typed t11429.
Interestingly, although three of the four ST121 isolates were of human origin collected in Dar es Salaam, and one of bovine origin (TC00402) that was collected in Tanga, all exhibited similar characteristics. The majority encoded for blaZ, dfrG, and erm(C) resistance genes. The sequence type was homogeneously Spa typed t272, whereby most encoded for aur, edinC, hlgA, hlgB, hlgC, and splA virulence genes, and all four encoded for a combination of sak/scn and egc-cluster (seg, sei, sem, seo, and sen) genes as well as eta/b exfoliating genes. Moreover, none of the ST121 encoded for any of the leukocidin genes ( Table 2).

Single Nucleotide Polymorphisms (SNPs)
The mapping of the raw reads of these 55 genomes to the reference genome, S. aureus 55-99-44 (NCBI ID CP024998.1), detected 20,614 qualified SNPs which were used to construct a maximum likelihood SNP tree (Figure 1). The SPN tree is composed of several different clusters of closely related isolates. The ST152 strains composed a stand-alone cluster of eight isolates originating from both humans and animals. Two larger branches emerging from a common ancestral point were also observed, each consisting of several smaller clusters with different sequence types. STs 7840, 7845, 7848, and 5477, which belong to the same cluster complex (CC5477), formed a cluster emerging from a common point. The cluster comprised exclusively of closely related bovine strains. These CC5477 strains further clustered with a ST121 branch that consisted of three human and one bovine isolate collected in Dar es Salaam and Tanga, respectively (Figure 1).

Discussion
To establish targeted interventions in human and animal health and reduce the effects of S. aureus, it is important to understand its occurrence, evolution and transmission The subsequent cluster comprised strains belonging to CC97, composed of ST97, ST7841 and ST7846. The majority of the isolates in the CC97 cluster were of bovine origin, but a number of human-originating isolates were also represented. ST97 was represented twice in the same cluster. In the first instance, five isolates of bovine and human origin clustered together in a sub-cluster. In the second instance, four bovine isolates clustered together in a sub-cluster. The largest sub-cluster was ST7846, that is composed of 12 bovineand human-originating isolates; sub-cluster ST7841 comprised of four bovine strains, all collected in Bagamoyo (Pwani). Generally, the SNP difference among clustered strains was between 2 and 10 ( Supplementary Material (SNP matrices)).

Discussion
To establish targeted interventions in human and animal health and reduce the effects of S. aureus, it is important to understand its occurrence, evolution and transmission within and between humans and animals. The present study was conducted to provide information on sequence types, virulence and antimicrobial associated genes found in S. aureus isolated from cows' milk and humans working on the same dairy farms, as well as in humans with no animal contact who were sampled in eastern Tanzania.
We observed a high diversity of sequence types; however, we could categorize them into three main clusters. The largest cluster was ST97, which clustered with novel STs belonging to the clonal complex CC97. Isolates of this cluster were found in humans with no animal contact sampled in Dar es Salaam as well as in milk sampled from Tanga, Bagamoyo (Pwani) and Morogoro. The fact that humans with no animal contact were colonized by a known bovine ST strain (ST97) may be explained by a secondary manifestation from a human with animal contact. The human isolates in question are clustered with milk samples from Bagamoyo (BC00118, BC00117, and BC00105), which is located about 60 km from where the human samples were collected. Human contact between the two towns is very high, further supporting the theory of secondary manifestation from a human with animal contact. CC97 has long been established as a bovine-specific lineage; however, there have also been reports of the clonal complex being found in humans [30,31].
ST5477 clustered with novel STs ST7840, ST7845, and ST7848, and the cluster was exclusively populated by bovine originating S. aureus, indicating that it might be a bovinespecific clonal complex. The isolates in the cluster were collected in all three regions (i.e., Pwani, Morogoro, and Tanga), which is suggestive of the wide spread of the clonal complex in Tanzania. ST5477 has previously been reported once in bovine-associated isolates in Rwanda [12]. The recent emergence of this sequence type in two neighboring countries cannot be explained; however, the most probable reason may be attributed to cattle trade between Rwanda and Tanzania. We have managed to build on the information gathered by the Rwandese study, and more sequence types closely related to ST5477 were gathered. This observation calls for further investigation into the ecology, evolution, and epidemiology of this and other clonal complexes, in order to establish the most dominant African bovine S. aureus lineages.
ST152 was a standalone branch that did not cluster with the rest of the isolates collected in the study. The eight isolates represented in this cluster stemmed from humans and dairy cattle and were collected from all three participating regions. Unfortunately, the study did not manage to obtain isolates from humans and cattle residing on the same farm; nevertheless, no genetic distinction was observed within the cluster, which is suggestive of host transfer between the two reservoirs. The majority of the ST152 were categorized as Spa type t355 and PVL+ MSSA, which is consistent with the majority of reports from the African continent [14,32].
In general, the data suggest the existence of contagious transmission within, as well as between, regions. For instance, novel STs 7841 and 78,445 were found only in Bagamoyo and Morogoro, respectively, while other STs were observed in more than one region. The highest diversity among the isolates was observed in Bagamoyo (Pwani), although the region contributed the majority of the isolates; high diversity in the area (Figure 2a,b) may also be attributed to the region's near proximity to the commercial capital city, as well as being the gateway to all northern regions. To cater for the high demand for milk and red meat in Dar es Salaam, a large influx of cattle from different parts of the country may pass through Bagamoyo to get to Dar es Salaam, creating a reservoir of different S. aureus sequence types in the area.  Five resistant conferring genes were predominantly observed in the dataset. blaZ, the gene conferring for penicillin resistance, was prevalent, at 84%, which is a common occurrence among bovine S. aureus as the antibiotic is frequently used to treat intramammary infections in cattle [12,33]. The streptomycin-resistant gene str was only detected in bovine S. aureus and tetK, in which it was prevalent at 23% and 24%, respectively. This can be explained by the high use of Penistrep and tetracycline to treat various animal infectionsin the country [34]. Multidrug resistance, which was defined as harboring three or more resistance genes, was identified in 25% of the collected samples, most of which were observed in ST152 and ST121. Both ST152 and ST121 were populated by human-and bovine-originating isolates and exhibited ermC and drfG genes conferring for erythromycin and trimethoprim resistance. The two STs have previously been isolated from both human and animal sources, supporting the hypothesis that the sequence types might have moved from human to animals, and hence presenting resistance to antibiotics more frequently used in humans [12,35]. Furthermore, none of the bovine-originating isolates were detected with mecA or mecC, hence all samples were categorized as MSSA. The only MRSA was detected in a dairy farm worker, which was categorized as ST8, a rather successful S. aureus lineage from which a number of MRSA have emerged. Additionally, the ST has also been observed in animals such as horses and cattle, posing potential threat of cross infection between humans and animals [31,36]. Virulence genes play a pivotal role in determining and developing infection, as well as the subsequent spread of infection within the same host species or between different hosts. Clear sequence type specificity in virulence factors could be observed. Hemolysin encoding genes were very frequently detected across regions, sequence types, and host species (hlgA 100%, hlgB 83.6%, and hlgC 83.6%). Leukotoxin encoding genes were also detected in the majority of the isolates, mostly exhibiting the lukD/E combination (50.9%), which is associated with the ability to target lymphocytes of a broad host range, markedly facilitating S. aureus pathogenicity [37]. Moreover, the lukED/hlgAB combination observed in 49% of the isolates is reported to be highly functional in erythrocyte binding and hemolysis, further facilitating the strain's pathogenic capacities.
ST97, ST7846, ST7841, ST152, ST7845, and ST7847 were not detected with enterotoxins. Nevertheless, none of the sequence types that did exhibit enterotoxins (65%) were detected with classic sea-see genes, all exhibited seg to seo enterotoxins, regardless of human or bovine origin. It is well known that 95% of food poisoning is caused by enterotoxins, which have the ability to retain their biological and immunological activities following pasteurization, as well as exposure to gastrointestinal protease [38]. Toxin shock encoding gene was detected in 16% in both human and bovine isolates; however, in bovine-originating samples, the gene was exclusively observed in ST5477 (12.7%). High detection levels of enterotoxins in the isolates increases the possibility of food poisoning through consumption of milk or milk products. Enterotoxins are further considered as a potential environmental pathogenic contaminant, even after leaving the human body [16,38]. In examining the genetic relatedness and potential host transfer, nine of the 12 human samples clustered with bovine isolates. Isolates from both hosts, when in a cluster, were genetically indistinguishable, exhibiting the same STs and Spa types. The predominance of lukD/E virulence factors in the isolates is also indicative of them being of bovine origin. Although lukD/E is not limited to bovine-associated lineages, they are known to have selective advantage in the bovine host [11]. Further, when bovine-originating isolates clustered with humanassociated lineages such as CC152 and CC121, they were also genetically indistinguishable, further exhibiting immune evasion clusters (scn, sak) that are uncommonly found in animals and are primarily needed for immune evasion in humans [39,40]. These observations strongly suggest S. aureus host transfer from bovine to human and vice versa. Some evidence proposes that some S. aureus are only capable of colonizing and infecting certain host species; nonetheless other lineages are non-specific. It has further been proven that S. aureus is highly adaptive to new environments through gene transfer and mutation [30]. It is therefore important that the interface between animals and humans is under constant surveillance in order to detect any population change in a timely manner.

Conclusions
The study presents an insight into antimicrobial conferring genes, virulence genes, and genetic diversity of S. aureus collected from milk and human samples in different areas around eastern Tanzania. We observed high diversity of S. aureus among human and bovine-originating isolates. Fourteen ST were documented, of which six were novel STs. Most isolates could, however, be grouped into three dominant clonal complexes, namely CC152, CC5477 and CC97. The low occurrence of antibiotic resistance conferring genes and the lack of PTSA genes in the majority of the bovine-originating isolates suggests that human health risk caused by bovine S. aureus is low. Nevertheless, the findings did suggest inter-host transmission, as known human lineages were collected in bovine samples and known bovine lineages were collected in humans. These isolates were genotypicaly indistinguishable when clustered with human-and bovine-specific clonal complexes. These finding therefore suggest that the two hosts can act as each other's reservoirs for antibiotic resistance and virulence genes, adopting pathogenic factors which, in turn, supports the need for rigorous surveillance of the bovine-human interface to track S. aureus population change in a timely manner.
Supplementary Materials: The following supporting information can be downloaded at: https:// www.mdpi.com/article/10.3390/microorganisms11061505/s1, Accession numbers assigned to the sequence after their submission to the European Nucleotide Archive has been provided.