Food and water are well-known vectors for the dissemination of zoonotic microorganisms, some of them can be extremely harmful to human health (Gallo et al., 2020). Foodborne diseases are a serious public health concern, associated with losses in productivity and high medical expenses every year (Garcia et al., 2020). Milk, as a central food in the human diet is a critical vehicle of both beneficial and pathogenic microorganisms. Several pathogens including Brucella spp., Campylobacter spp., Shiga toxin-producing Escherichia coli, Listeria monocytogenes, Mycobacterium spp., Salmonella spp., and also bacterial toxins have been associated with milk-borne diseases (Dhanashekar et al., 2012). Furthermore, there has been a drastic increase in antimicrobial resistance among zoonotic pathogens. Thus, food surveillance is a major concern for the food industry and for public health (Pérez-Rodríguez and Mercanoglu Taban, 2019).

Staphylococcus aureus is ubiquitous in the environment and a major cause of bovine mastitis. Thus, milk is a common source of contamination for the dairy supply chain, the environment, as well as for final consumers (Rola et al., 2016). In the European Union, 43 foodborne outbreaks (FBO), 402 cases and 32 hospitalizations associated to this pathogen were reported in 2020, according to the recent European Food Safety Authority report (European Food Safety Authority and European Centre for Disease Prevention and Control, 2021b). Typically, staphylococcal food poisoning (SFP) are caused by the ingestion of food contaminated with preformed staphylococcal enterotoxins (SEs) produced by coagulase-positive staphylococci (CPS) (Argudín et al., 2010). There are 24 SEs and enterotoxin like (SEl-) toxins currently identified, including the classical (SEA-SEE) and the newer (SEG-SElY), which are encoded on different pathogenicity islands (Fisher et al., 2018). Most SFP outbreaks are reported in countries where consumption of unpasteurized milk cheeses is common, such as France and Italy. Additionally, raw milk vending machines and traditionally made food products, such as cheese manufactured at local dairy farms, are becoming more popular throughout Europe, increasing the risk of SFP (Rola et al., 2016).

Staphylococcus aureus can also produce several other extracellular virulent proteins such as toxic shock syndrome toxin 1 (TSST-1), Panton-Valentine Leukocidin (PVL), hemolysins, and coagulase. These proteins can contribute to a broad spectrum of pathologies beyond food poisoning that can range from toxin-mediated syndromes to fatal systemic diseases (Chambers and DeLeo, 2009). Moreover, this pathogen is a well-known example of acquired resistance to multiple antibiotics. Methicillin-resistant Staphylococcus aureus (MRSA) are of particular concern to human health because it is virtually resistant to all available β-lactam antibiotics and represent a significant cause of morbidity and mortality throughout the world. These have often been found outside the health environment, including in farm animals (van Duijkeren et al., 2014; Wu et al., 2018). Surveillance of raw milk is therefore essential for a better understanding of the risk factors along the milk food chain and to guarantee public health safety.

In Portugal, data on S. aureus circulating in raw milk are scarce. Pereira et al. (2009) characterized several S. aureus isolates from different foods in Portugal, including a limited number of raw milk samples, and detected the presence of enterotoxigenic S. aureus. Molecular typing can be a powerful tool for improve such epidemiological studies with data about clonal relatedness, genetic diversity, and tracking the spread of pathogens. S. aureus protein A (spa) typing is a rapid, affordable, and easy molecular typing method that assigns a classification to S. aureus strains from the number/sequence variation in repeats at a specific region of the spa gene. It offers excellent discriminatory results to the study of S. aureus diversity (Sabat et al., 2013). Molecular typing and characterization of virulence factors are thus an important tool in the control of zoonotic diseases.

The aim of this study was to determine the prevalence and diversity of S. aureus and staphylococcal enterotoxins in raw cow’s milk collected from bulk cooling tanks on dairy farms from the main dairy region of mainland Portugal. Genetic determinants associated with enterotoxinogenicity (i.e., SE-encoding genes), antimicrobial resistance (i.e., mecA and mecC genes) and severe human infections (i.e., pvl and tsst-1 genes) were investigated.

The presence of S. aureus and its enterotoxins in raw milk can be a serious threat to public health given the importance of dairy products in human diet. Therefore, an estimation of the prevalence and genetic determinants of S. aureus is always important to implement rational mitigation strategies and avoid the dissemination of this pathogen through the food chain (Kadariya et al., 2014).

In this study, a prevalence of 53.0% of S. aureus was established in raw cow’s milk collected from the bulk cooling tanks of 100 dairy farms located in the main dairy region of mainland Portugal. To date, this is the first significant study of S. aureus in raw milk carried out in Portugal. A similar level of contamination (approximately 40–70%) in raw cow’s milk has been found in other countries such as Poland, Italy, New Zealand, Norway, India and China (Jørgensen et al., 2005b; Jakobsen et al., 2011; Hill et al., 2012; Bianchi et al., 2014; Sudhanthiramani et al., 2015; Rola et al., 2016; Liu et al., 2017; Wang et al., 2018; Kou et al., 2021). In fact, the presence of S. aureus in bulk tank milk is not surprising given that this species is ubiquitous in nature and one of the main causes of bovine mastitis. Even so, the EU regulation established a criterion for CPS in cheeses made from raw milk (m = 10⁴ CFU/g – M = 10⁵ CFU/g) during the manufacturing process. If values are higher than 10⁵ CFU/g cheese, the batch has to be tested for staphylococcal enterotoxin (EU Commission, 2007). In the present study, the CFU counts obtained on the CPS positive samples were all below the imposed limits. Low levels of CPS at the farms tanks are expected since milk storage is controlled at low temperatures (≈4°C) and for a short time after milking (12 – 24 h), as a way to prevent the growth of pathogens, including S. aureus (Owusu-Kwarteng et al., 2020).

Still, S. aureus presence in foods can represent a risk for human health since this pathogen produces an array of exoproteins with toxicological effects that are highly stable. Although the conditions used during the pasteurization process are suitable to destroy harmful microorganisms, they are not capable of eliminating some proteins, such as toxins. They will keep their activity in pasteurized milk, even at few micrograms, and be a serious threat to end consumers. From those toxins, SEs are the most relevant in the context of foodstuff. In fact, SEs are responsible for several SFP outbreaks reported every year worldwide (Argudín et al., 2010). Thus, their presence in foods is strictly prohibited in Europe (EU Commission, 2007; European Food Safety Authority and European Centre for Disease Prevention and Control, 2021a). In this study, one raw milk sample was positive for staphylococcal enterotoxins (SEA-SEE). Data on SEs in raw milk are scarce, as direct analysis on the raw material is out of scope of the European regulations. Although the presence of enterotoxins in raw milk has never been studied in Portugal, it has been found in cheeses during official controls, and also an outbreak affecting 13 persons due to an unknown food source (European Food Safety Authority [EFSA], 2019; European Food Safety Authority and European Centre for Disease Prevention and Control, 2021a). In Poland, the presence of SEs in raw milk was assessed without any positive result (Rola et al., 2016). Furthermore, SEs production usually requires the presence of S. aureus in high amounts, at least 5–6 log10 CFU.mL^–1 (Argudín et al., 2010). None of the bulk tank milk samples analyzed had a level of contamination near or above this value, contributing to a lower probability of detection of SEs in the analyzed samples. Surprisingly, in the SE-positive sample found in this study no CPS was detected. Several hypotheses can be made to explain this result. Contamination of milk directly with enterotoxin from the udder of an animal is one of them. In fact, if S. aureus was in the udder of dairy cows with enough number of cells able to produce SEs, then enterotoxins could be transferred to the milk during the milking process, even after the bacteria have been eliminated (Rola et al., 2016). However, neither clinical nor subclinical mastitis were indicated in the animals from the dairy farm. Another possible explanation might be related with a dilution effect in the bulk milk, if only one or a few caws are infected. This dilution effect might affect more significantly culture techniques than SE determination, considering that the detection limit reported for VIDAS SET 2 is as low as 0.25 ng toxin per gram of food (Schultz et al., 2004). Enterotoxigenic coagulase-negative staphylococci (CNS) cannot be ruled out as several virulence determinants, usually associated with S. aureus, such as SE-encoding genes, have been detected in CNS genomes (Park et al., 2011; Podkowik et al., 2013) and different studies have identified several enterotoxigenic CNS species in dairy animals, capable of producing enterotoxins, predominantly SEC (Bergonier et al., 2003; Taponen et al., 2006; Ünal and Çinar, 2012). Since the methodology followed, in this study, for the isolation of S. aureus reject CNS strains, it is not possible to evaluate if CNS are responsible for the SE-positive sample. Finally, we should consider the possibility of a false positive result. Nonetheless, replicates were tested to confirm the sample result.

In recent years, the emergence of antimicrobial resistant strains, particularly MRSA, in livestock animals that is readily transferable to humans, has also become a growing public health concern (Sharma et al., 2018). In the present study, only 35.5% of the S. aureus isolates showed resistance to at least one antimicrobial agent. Similar results were reported on studies carry out in Italy (39.4%), Poland (23.0%) and China (38.5%) (Rola et al., 2015; Giacinti et al., 2017; Wang et al., 2018). Penicillin resistance was the most prevalent (32.3%) among the antimicrobial agents tested in accordance with previous studies in raw milk (Rola et al., 2015; Wang et al., 2018). Tetracycline, a broad-spectrum antimicrobial agent frequently used in the treatment of infections in cattle, was the second most prevalent antimicrobial resistance (24.2%), a level similar found to other studies (Liu et al., 2017). It was notable that ciprofloxacin- and chloramphenicol-resistant S. aureus were the third most frequently detected antimicrobial resistance. Ciprofloxacin is a third generation fluoroquinolone used at clinical level, while chloramphenicol is an antibiotic not authorized for use in food-producing animals in the European Union (Benford et al., 2014; Wang et al., 2018). Moreover, five isolates (8.1%) were identified as MRSA by antimicrobial susceptibility, revealing resistance to cefoxitin and all other β-lactams, except ceftaroline (Clinical and Laboratory Standards Institute [CLSI], 2020). Despite the low percentage of resistance observed, antibiotics should be used with caution in dairy animals because they may compromise the treatment of future infections. The presence of isolates with the MRSA phenotype is of major concern for human public health.

Although the presence of SEs was not detected in samples carrying S. aureus, enterotoxins’ production is still possible under appropriate environmental conditions (e.g., temperature) if the SE-encoding genes are part of the S. aureus genome. In our study, we have identified the presence of five different SE-encoding genes among 46.8% of S. aureus isolates, including two classical SE genes (sea and sec) and three non-classical SE genes (seh, seg and sei). Only five (8.1%) S. aureus isolates had classical SE genes (sea-see), although the literature suggested that about 95% of SFP outbreaks are caused by classical SEs (Argudín et al., 2010). Based on our results, the official method recommended for the detection of SEs in foodstuff does not reflect the complete diversity of enterotoxins found in nature. In fact, three of the SEs genes identified in this study, including the two most prevalent SEs (seg and sei), are not covered by ISO 19020:2017 methodology. The low prevalence of sea-see genes explains the low prevalence of SEA-SEE found in the bulk tank raw milk samples. Moreover, since the presence of non-classic SEs is not covered by the recommended methodologies, their prevalence may be underestimated. In Portugal, the characterization of S. aureus from a small number of raw milk samples had already verified the predominance of the seg and sei genes (Pereira et al., 2009). High prevalence of seg and sei was also observed in raw milk in Italy, and its co-existence in most isolates is also consistent with previous reports (Blaiotta et al., 2006; Bianchi et al., 2014; Johler et al., 2018). In fact, these two SE genes are typically located in tandem on the ecg locus and have already been implicated in scarlet fever, toxic shock and neonatal intractable diarrhea cases (Naik et al., 2008). However, these genes do not exist strictly together, as verified in one isolate of this study (only seg) and suggested by Jørgensen et al. (2005b). Other studies on raw milk have observed a higher prevalence of sea-see genes in Italy, Poland and China, yet most of them are in agreement with our study on the predominance of non-classic SE genes (Bianchi et al., 2014; Rola et al., 2016; Wu et al., 2018). Classical SE genes seems to be more prevalent in isolates of human and non-animal food origin than in animal food origin, such as S. aureus from raw milk (Chao et al., 2015). Furthermore, it should be mentioned that despite the prevalence of certain enterotoxin genes, they may have different genomic localization that may differentiate the virulence potential of the strains. Enterotoxins can be encoded in prophages, pathogenicity islands, genomic islands or plasmids, which can have different levels of regulation and expression of these virulence factors (Malachowa and Deleo, 2010).

Regarding the presence of resistance genes, the detection of mecA (8.1%) shows that raw cow’s milk can also be an antimicrobial resistance vehicle. The positive mecA-MRSA isolates correspond with the isolates classified as MRSA by the antimicrobial susceptibility test, demonstrating a complete correspondence between the genomic and phenotype results. The presence of MRSA strains have been reported in raw milk and dairy cattle in Europe (Tenhagen et al., 2014, 2018; Cortimiglia et al., 2016; Hansen et al., 2019; Schnitt et al., 2020; Lienen et al., 2021), but also in Algeria, Uganda, Brazil, and China (Asiimwe et al., 2017; Guimarães et al., 2017; Dai et al., 2019; Titouche et al., 2019). In Europe, MRSA prevalence in raw milk was 3–10%, which is in accordance with our results (Schnitt et al., 2020). In Portugal, mecA-MRSA strains have also been reported in bovine mastitis (Pereira et al., 2009), while mecA is also the predominant gene in most MRSA isolates worldwide. No association between MRSA strains and enterotoxigenic determinants was found in our analysis. In contrast, TSST-1 genetic determinant was found in enterotoxigenic isolates (sec/sec/sei-positive). Such relation may result in an increase in the toxigenic consequences of these strains. No pvl or mecC genes were detected, consistent with previous studies demonstrating the low prevalence of these genes among S. aureus isolated from dairy products (van Duijkeren et al., 2014; Johler et al., 2018). However, three isolates were identified as spa types (t843 and t528) commonly associated to CC130, the frequent genetic background of mecC (Bortolami et al., 2017). Given the epidemiological relevance of this genetic determinant of MRSA, the negative result for the presence of the mecC gene was confirmed. Furthermore, these isolates revealed a complete susceptibility to the tested antimicrobials, including cefoxitin (mecC-positive strains are typically cefoxitin resistant).

Regarding the diversity of S. aureus, we have identified 25 spa types among the 62 S. aureus isolates from the 100 bulk tank milk samples. Wang et al. (2018) found seven spa types of S. aureus in 96 isolates from raw milk collected on 2 dairy farms in China. The lower diversity obtained on that study might be due to the limited number of farms enrolled. Furthermore, the most predominant types observed in our study, t529 and t1403, differed from the t127 observed in China. A better correspondence is observed when compared to studies in European countries. Higher diversity and t529 and t1403 types were also predominant in raw milk and bovine isolates in Denmark and Switzerland (Johler et al., 2011; Ronco et al., 2018). Most of the other spa types detected in our study were also associated with bovine S. aureus from different European countries (Boss et al., 2016). Thus, the diversity of S. aureus found in raw milk in Portugal seems to be in line with that found in the rest of Europe, suggesting a geographic predominance of some spa types. In terms of human health, spa types t002 and t127, detected in one isolate each, are frequently observed in human invasive infections (Grundmann et al., 2010). Combining spa typing with virulence factors, 32 distinct S. aureus types can be identified in this study. Some spa types were exclusively associated to enterotoxigenic S. aureus, mainly t002, t117, t127, t528, t543, t899, t2873, and t9216. In contrast, S. aureus t1403 was also exclusively associated with the absence of virulence/resistance factors. The MRSA-t011 type have recently emerged in European countries, such as Germany and Denmark, associated with livestock-associated MRSA (LA-MRSA) strains including from raw milk isolates (van Duijkeren et al., 2014; Tenhagen et al., 2018; Hansen et al., 2019; Schnitt et al., 2020; Lienen et al., 2021). Accordingly, our results support that this trend is also present in Portugal. MRSA-t2383 type is a rare relative of t011 and was never reported in raw milk or dairy cattle, but in human and other animals as pigs and seabream (Fanoy et al., 2009; Salgueiro et al., 2020). Thus, spa type/virulence/resistance factors can be a good way to assess the variation in S. aureus diversity over time. In most instances, a spa types are highly associated with a specific multilocus sequence type (MLST), which is usually related with a specific clonal lineage (Tegegne et al., 2019). As an example, identified spa-types t002, t108, t117, t529, t1334, and t1207 are usually assigned to either sequence type (ST) ST151 or ST504, both belonging to clonal complex (CC) CC70, while t1403 is usually associated with ST97-CC97 or ST133-CC133. In the case of MRSA isolates, the spa-types here identified are usually assigned to ST398-CC398, which is the prevalent LA-MRSA in Europe, occasionally involved in human disease (Bortolami et al., 2017). Overall, most of the ST/CC associated with the spa-types identified in this study are commonly associated with livestock infection or carriage, especially of bovine origin (Smith et al., 2005; Hasman et al., 2010; Boss et al., 2016).

In conclusion, the high prevalence of enterotoxigenic S. aureus and the detection of MRSA strains in raw cow’s milk collected from bulk cooling tanks on dairy farms from the main dairy region of mainland Portugal is of particular concern. Furthermore, a high diversity of S. aureus was found in a relatively small geographical area, however, most with genomic lineages associated with livestock infection or carriage. This study also points out the predominance of SE-encoding genes that are not currently covered by the gold standard methodology (ISO 19020:2017) applied in the control of food samples. The higher prevalence of non-classical SEs, mainly SEG, SEH and SEI, should not be ignored, as these have been implicated in food poisoning outbreaks (Jørgensen et al., 2005a; Tang et al., 2011; Ciupescu et al., 2018; Hait et al., 2018). Commercial tests with proven effectiveness for the non-classic SEs are available on the market, which may facilitate their integration in food safety control standard for SEs analysis of foods (Hait et al., 2018). As SFP outbreaks linked to raw cow’s milk consumption and raw milk products have been increasing, the presented data on the characterization of S. aureus and its virulence determinants are important to improve risk assessment and develop solutions to limit the dissemination of this pathogen in Portugal.

The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author.

RO, EP, GA, NA, and CA designed the experiments. RO performed the experiments. RO, EP, and CA analyzed the data. All authors contributed to the writing of the manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.