INTRODUCTION Staphylococcal food poisoning is one of the most common food-borne diseases worldwide. The causative agents are enterotoxins produced by enterotoxigenic strains of Staphylococcus aureus during its growth in favorable conditions in food. Epidemiological data show that S. aureus is often found in raw milk cheeses and as a result cheeses are often implicated in food poisoning outbreaks (DE BUYSER et al., 2001; DELMAS et al., 2006). The monitoring of domestic cheeses from the markets of the city of Dubrovnik so far has shown very poor microbiological quality (LJEVAKOVIĆ-MUSLADIN et al., 2014; LJEVAKOVIĆ- MUSLADIN et al., 2016). The authors state that as many as 80% of the cheeses examined were of unsatisfactory microbiological quality, and 70% of the cheeses contained high levels of S. aureus. Although none of the cheese samples contained staphylococcal enterotoxins (SE) (LJEVAKOVIĆ-MUSLADIN et al., 2016), the high frequency and high level of contamination with S. aureus indicated the need for further investigation of enterotoxogenicity of isolates and assessment of the real risk of staphylococcal poisoning. It should be emphasized that the absence of SE in cheeses does not exclude the possibility of their production since it has not been investigated whether S. aureus isolates possess SE genes (LITTLE et al., 2008; ROSENGREN et al., 2010; ROLA et al., 2016). Many studies have shown that although SE was not detected in cheeses, S. aureus isolates carried one or more SE genes (MORANDI et al., 2007; CREMONESI et al., 2007; NORMANNO et al., 2007a; CARFORA et al., 2015; ROLA et al., 2016). Numerous enterotoxogenicity studies of S. aureus strains isolated from cheeses have shown remarkable diversity in the prevalence of enterotoxin genes worldwide. Since there are no data on the nature of S. aureus isolated from cheese produced in Croatia, the aim of this study is to determine: i) the occurrence of S. aureus and staphylococcal enterotoxins SEA-SEE in domestic fresh cheeses; ii) phenotypic characteristics of isolates, including antibiotic susceptibility; iii) the ability to produce SEA-SEE enterotoxins in vitro by immunological methods; iv) the presence of SEA-SEE enterotoxin genes by molecular methods; v) the efficiency of newly established Real-time PCR method, including a comparison with other methods used. MATERIAL AND METHODS Cheese samples. A total of 30 fresh cheese samples (26 bovine, two caprine, and two bovine/caprine) were collected from the Dubrovnik city markets during 2018 and 2019. Cheeses were produced by 17 small-scale unregistered domestic producers or small family farms, with a production size of 1.5-4 kg of cheese per week and 1-2 cows and/or goats. None of the samples were refrigerated during sales on the market and many were exposed to direct sunlight during warm months, being sold on the stand. Cheese temperature was measured with a probe thermometer immediately after collection and samples were transported to the laboratory in a cooler within 15 minutes. Microbiological analyses of cheese. All cheese samples were analyzed according to HRN EN ISO 6888-1:2004 using Baird-Parker agar for the enumeration of coagulase-positive staphylococci. Ten colonies per cheese sample were used in further examinations. Colonies were identified as S. aureus based on the coagulase test (Merck Bactident® Coagulase plasma EDTA) and latex agglutination test for detection of fibrinogen affinity antigen, protein A, and capsular polysaccharides (Bio-Rad PastorexTM Staph Plus test). Physico-chemical analyses of cheese. Besides temperature, for all cheese samples, water activity and pH were measured. Water activity was measured by HygroPalm23-AW (Rotronic, Switzerland) according to HRN EN ISO 21808:2005, while pH was measured by pH meter Schott Lab 850 with BlueLine 27 electrode. Bacterial strains. A total of 175 isolates were collected from 18 cheese samples. All S. aureus isolates were tested for enterotoxin production and antibiotic susceptibility and their phenotypic characteristics were determined. Determination of phenotypic characteristics. Included Dnase and catalase test, Gram staining, hemolysis test, “egg yolk” reaction on Baird-Parker agar, and biochemical identification by bioMerieux API STAPHTM. Antibiotic susceptibility testing. It was conducted according to the EUCAST disk diffusion method on Mueller-Hinton agar. The following antibiotics were tested: azithromycin, cefoxitin, clindamycin, erythromycin, gentamicin, moxifloxacin, mupirocin, oxacillin, and trimethoprimsulphamethoxazole. The interpretation of the results was conducted according to the EUCAST manual and criteria. Enterotoxin detection in cheese samples. Detection was performed according to HRN EN ISO 19020:2017, which included extraction followed by a concentration based on dialysis principle, and an immunoenzymatic detection using bioMerieux VIDASTM SET2 method.Detection of enterotoxin production in vitro. All S. aureus isolates were examined for the production of SEA-SEE enterotoxins. Over-night TSB culture of well-isolated colonies from the Baird-Parker agar was examined for the presence of SEA-SEE by VIDASTM SET2 and Thermo ScientificTM OxoidTM RPLA methods. Detection of SEA-SEE genes. Gene detection was preformed by Real-time PCR and confirmed by classical PCR at Department of Veterinary Medicine, Laboratorio di Ispezione degli alimenti di origine animale, University of Perugia, Italy. The Real-time PCR method was a modified method described by NAKAYAMA et al. (2006) for the detection of sea-see genes. Method comparison. Results from all three methods (VIDAS SET2, RPLA, and Real-time PCR) were compared whether they were in a full, partial, or negative agreement. Determination of Real-time PCR efficiency. It was determined based on the standard curves of target genes (sea-see). Real-time PCR was performed on serial dilutions 1:10 of each of the target gene DNA. Standard curve was constructed based on Ct values versus log values of DNA concentrations. Slope value was used for calculation of efficiency according to the equation 𝐸𝑓𝑓 = (10(−1/𝑠𝑙𝑜𝑝𝑒) − 1) ∙ 100 %. The intercept of the Y-axis determined the limit of detection. RESULTS Out of 30 analyzed cheese samples, 18 samples (60%) were highly contaminated with S. aureus strains. The contamination level ranged from 8.7x103 cfu/g to 1.8x106 cfu/g, with an average 6.9x105 cfu/g. The temperature of cheese samples varied between 6 °C and 23.3 °C, with an average of 12.5 °C. In thirteen cheese samples (43.3%) S. aureus number was above legal safety criteria of 105 cfu/g, at which enterotoxin production occurs. Although highly contaminated, staphylococcal enterotoxins were not detected in any of the cheese samples. A total of 180 coagulase-positive isolates were collected from 18 cheese samples, 175 of which were confirmed as S. aureus by latex agglutination test and API STAPH. All isolates possessed phenotypic characteristics typical for S. aureus (protein A, bound and free coagulase, catalase, Dnase, morphology). Variations were found in certain biochemical reactions, including the „egg-yolk“ reaction and hemolysis pattern. „Egg yolk“ reaction was characteristic of 94 (53.7%) out of 175 isolates, 32 (18.3%) isolates showed weak „egg yolk“ reaction, while in 49 (28%) isolates the reaction was completely absent. Beta-hemolysis was observed in 103 (58.8%) isolates, while 64 (36.6%) isolates produced a double hemolysis zone. On the otherhand, 8 (4.6 %) isolates were non-hemolytic. Antibiotic resistance was detected in 1.1% of isolates and to mupirocin only. Enterotoxin production was detected in 37 isolates (21.1%) by VIDAS SET2, among which 34 (19.4%) isolates produced enterotoxin SEC as detected by RPLA. Other classical enterotoxins were not detected. Three isolates detected by VIDAS SET2 were considered false positives since neither RPLA nor PCR methods confirmed enterotoxigenicity. The percentage of enterotoxin-producing strains within one sample varied between 10% and 100%, which indicates a heterogeneous population of S. aureus in cheese. Cheese samples with enterotoxigenic strains were from the same two cheese producers. Real-time PCR detected 34 (19.4%) isolates positive for enterotoxin gene sec but most of the isolates (80.6%) were not enterotoxigenic. Other classical genes were not detected. The presence of the sec gene was confirmed by the conventional PCR method as well. All strains that carried the sec gene were detected as SEC-producers by the RPLA method. The modified Real-time PCR method for five classical enterotoxins had efficiencies between 97.0% and 105.1%, which are in the optimal range. The method proved to be very efficient, reproducible, and discriminatory between positive and negative results. Comparison of three methods (VIDAS SET2, RPLA, and Real-time PCR) showed very high (98.3%) agreement between results. CONCLUSIONS Domestic cheese produced in the Dubrovnik area is highly contaminated with S. aureus as a result of inappropriate hygiene during production, as well as inappropriate temperature regime during market sales and possibly during storage altogether. Although highly contaminated, staphylococcal enterotoxins were not detected in any of the cheese samples. This is mainly because most S. aureus isolates from our study were not enterotoxigenic. All isolates possessed typical phenotypic characteristics of S. aureus. Antibiotic resistance was surprisingly very low and to mupirocin (which is used in human medicine) only. Enterotoxigenic potential was found in 19.4% of isolates. Based on their enterotoxigenicity all isolates could be divided into two phenotypes/genotypes: non-enterotoxigenic and toxigenicSEC phenotype (sec genotype) since only SEC production and sec gene were detected. Cheese samples were contaminated with the mixture of both, non-enterotoxigenic and SEC-producing strains. The modified Real-time PCR method proved to be highly efficient, reproducible, and discriminatory. The combined use of immunological and molecular methods is strongly recommended. Although the enterotoxigenic potential of S. aureus isolates from fresh cheese was lower than expected, it could be significantly increased due to horizontal gene transfer between enterotoxigenic and non-enterotoxigenic strains in the cheese population. Since certain strains possessed the sec gene and also were able to produce SEC in vitro, enterotoxin production in vivo could not be ruled out. Education of cheese manufacturers about good hygiene practices, refrigeration of cheese during market sales, alongside better sanitary control and surveillance, could significantly improve the microbiological quality and safety of cheese produced in the Dubrovnik area. Staphylococcus aureus, cheese, enterotoxigenicity, staphylococcal enterotoxins