Safety Evaluation Of Sjenica Cheese With Regard To Coagulase-Positive Staphylococci And Antibiotic Resistance Of Lactic Acid Bacteria And Staphylococci

Abstract Sjenica cheese is an artisanal cheese stored in brine, traditionally produced from raw sheep’s milk in the southern part of Serbia - Sjenica Pester plateau. The aim of this study was to perform the safety evaluation of Sjenica cheese. As one of the safety criteria we considered the number of coagulase positive staphylococci and their enterotoxigenic potential. Antibiotic susceptibility/resistance patterns of autochthonous lactic acid bacteria and coagulase-positive staphylococci isolated from Sjenica cheese was also investigated. During the monitoring period of the cheese-making process, coagulase positive staphylococci did not reach the value of 105 cfu/g. Among coagulase positive staphylococci, 12 (46,15%) isolates showed enterotoxigenic potential and were identified as Staphylococcus intermedius (11 isolates) and Staphylococcus aureus (1 isolate). Vancomycin resistance was the most prevalent phenotypic resistance profile in coagulase positive staphylococci. Lactococci present the most dominant population among lactic acid bacteria. The most prevalent resistance phenotype in lactococci was resistance to streptomycin (83.33%), ampicillin and penicillin (70.83%); lactobacilli were characterized by resistance to vancomycin (62.5%) and tetracycline (54.17%), while resistance to streptomycin (82.46%) was the most prevalent phenotypic profile in enterococci. All coagulase positive staphylococci and lactic acid bacteria isolates that showed resistance to tetracycline on disc diffusion and E-test, were tested for the presence of ribosomal protection proteins, tet(M) and tet(K) genes. All isolates were positive for ribosomal protection proteins genes; 14 (60.87%) isolates showed tet(M) gene presence, while 2 lactobacilli isolates revealed the presence of tet(K) gene.


INTRODUCTION
Serbia has a strong tradition in regional cheese making. Brined cheeses are the most important family of Serbian artisanal cheeses, and among them, Sjenica cheese received Protected Geographical Indication (PGI) designation on the national level [1]. Sjenica cheese is an artisanal cheese stored in brine, traditionally produced from raw sheep's milk in the southern part of Serbia -Sjenica Pester plateau. It is highly appreciated for its unique fl avor and classifi ed as a soft cheese which requires a ripening period of minimum 5 months for developing specifi c structure and aroma, but also for stabilizing the cheese matrix from the microbiological point of view. The aroma/fl avor and textural characteristics of cheese depend strongly on the type of milk, production method, diversity and metabolic activity of lactic acid bacteria (LAB) strains as thermal treatment of raw milk is not done. The specifi c pedioclimate microenvironments of Sjenica Pester plateau and selection pressure achieved by traditional technology maintained throughout centuries contribute to the establishment of a unique cheese ecosystem as a fermented microbial society with prevalence of lactic acid bacteria.
On our knowledge very few studies have been conducted on Sjenica cheese mainly regarding the technological characteristics of the traditional cheese making process [2,3] and potential of autochthonous strains of lactic acid bacteria isolated from Sjenica cheese to be used as starter cultures [4]. No literature data exist on safety evaluation of Sjenica cheese.
Traditionally, white-brined cheeses have been manufactured from raw milk and there is a risk that certain pathogens can survive, multiply in restrictive cheese matrix and contaminate the fi nal product [5][6][7]]. Furthermore, traditional cheese-making is open to contamination, so there is a growing need to monitor the process hygiene. Staphylococcus aureus is one of the major causes of food-borne intoxication due its' enterotoxigenic potential [8,9]. Sjenica cheese ripens in brine with 6-7% salt, and the temperature profi le of curd handling techniques are not restrictive to staphylococcal growth, therefore conditions support the growth and multiplication of mesophilic halotolerant staphylococci, highlighting the risk of enterotoxins synthesis.
There is a growing scientifi c evidence that commensal microbiota in food could become a reservoir of antibiotic resistance genes disseminated via the food chain by ubiquitously demonstrated mechanism of lateral gene transfer [10][11][12]. The most frequent antibiotic resistance genes documented to widely occur in lactic acid bacteria are tet genes [13] -mostly due to their localization on mobile genetic elements such as plasmids and transposons and extensive use of tetracycline in human and animal therapy. Also, tetracycline resistance (TCr) is one of the most prevalent resistance phenotypes among staphylococci isolated from various food substrates due to the fact that the genetic base of TCr is encoded on transmissible plasmids and conjugative transposons.
The aims of the present work were therefore to perform the safety evaluation of Sjenica cheese regarding the risk of coagulase positive staphylococci presence; also to determine the current susceptibility/resistance patterns of autochthonous LAB and staphylococci originated from Sjenica cheese using phenotypic methods. The genetic base of tetracycline resistance in resistant staphylococcal and LAB population was revealed by applying molecular techniques.

Material
The material for investigation consisted of 5 samples of raw sheep milk, 5 samples of curd before salting and 9 samples of Sjenica cheese from 1 to 90 days of ripening, collected from three households in the southern part of Serbia -Sjenica Pester plateau.

Microbiological examination
Microbiological examination was carried out in order to estimate the safety of the product and characterize the main microbial groups of lactic acid bacteria. As one of the safety criteria, coagulase-positive staphylococci (CoPS) were enumerated. Furthermore, the microbiota of LAB was studied in order to estimate their succesion during the ripening process.

Plating and enumeration
About 25 ml (g) of the sample were aseptically homogenized in 225 ml of a 2% (w/v) sodium citrate solution prewarmed at 45 o C using a stomacher. Sample homogenates were tenfold diluted in buffered peptone water and the dilutions were plated in duplicate on selective culture media. The isolation and enumeration procedure of coagulase positive staphylococci followed the ISO standard [14]. Also, for enumeration of the main group of LAB, serial dilutions were plated on specifi c agar media: thermophilic lactobacilli on MRS Agar (Merck) at 37 o C for 48h, mesophilic lactococci on M17 Agar (Merck) at 30 o C for 48h; enterococci on Kanamycin Aesculin Azide agar (Oxoid) at 37 o C for 24h.

Determination of enterotoxigenic potential of coagulase-positive staphylococci and their identifi cation
VIDAS® Staph enterotoxin II (SET2; bioMeriéux, REF 30 705, 2004) Enzyme Linked Fluorescent Assay (ELFA) technique was used to determine the enterotoxigenic potential of CoPS isolates. API Staph biochemical test kit (bioMeriéux REF 20 500) identifi cation system was used for the identifi cation of enterotoxigenic coagulase positive staphylococci.

Antimicrobial susceptibility testing
Antimicrobial susceptibility testing was accomplished with the commercial test BBL Sensi-Disc Antimicrobial Susceptibility Test Discs performing the standard agar disc diffusion method in accordance with the Clinical Laboratory Standards Institute guidelines [15].
Testing the antimicrobial susceptibility/resistant patterns, enterococci were exposed to the same antibiotics as lactobacilli, except for ciprofl oxacin and additionally gentamicin discs were used in a concentration of 10 and 120 μg.
The tetracycline resistant subpopulation of staphylococci and LAB screened by disc diffusion test was subjected to Minimal Inhibition Concentration (MICs) determination by applying E test strips (AB Biodisk, Sweden) following the manufacturer's recommendations. Evaluation of MIC values for CoPS was done according to manufacturer's instructions; and for LAB according to FEEDAP [17].

Genetic characterization of tetracycline resistance
Isolates displaying tetracycline MICs equal to or higher than the breakpoints were considered resistant and therefore were submitted for PCR-based detection of genes encoding resistance to tetracycline (tet genes). In the fi rst PCR assay, tet genes encoding tetracycline resistance through ribosomal protection proteins (RPP) were detected using universal primers. If positive for RPP genes, additional PCRtest was performed using gene-specifi c primer for tet(K) and tet (M). The genomic DNA of tested strains was isolated by using MasterPure Complete DNA and RNA Purifi cation kit (Epicentre, USA) and amplifi cation procedure was performed in a thermocycler (Applied Biosystems AB2720). The PCR reaction mixture consisted of Amplitaq Gold PCR Master Mix 2x (Invitrogen, USA), 400 nm each of the primers tested and 100 ng of respective DNA. Primer sequences, annealing temperatures and amplicon sizes are listed in Table 1. PCR-based detection of tet genes was performed under the conditions given in Table 2. Amplifi cation products were detected by electrophoresis in 1% agarose gel (Invitrogen, USA), stained with ethidium bromide and as a fi nal step were visualized and documented by using UV transluminator and GelDoc system (Eppendorf, Germany).

Milk
, gel, curd and cheese samples in the early phase of Sjenica cheese manufacturing process were subjected to conventional microbiological procedures in order to estimate the number of CoPS (Table 3). Milk coagulation and curd formation conditions were favorable to CoPS growth but as the cheese matrix became more restrictive due to establishment of LAB in high numbers and during subsequent competition and lactic acid development we noticed a decrease in CoPS population.
As coagulase positive staphylococci, 26 out of 36 isolates suspected to belong to the CoPS group by observation on Baird Parker plates, were determined. Enterotoxigenic potential was detected in 12 (46,15%) isolates.
By applying the API identifi cation system, enterotoxigenic isolates of staphylococci were identifi ed as Staphylococcus intermedius (11) and Staphylococcus aureus (1).
Sjenica cheese is made from raw milk, therefore, the main biochemical activities responsible for conversion of milk to the end product with desired and unique sensory properties are governed by autochthonous lactic acid bacteria. Taking into account this fact, the biodiversity of autochthonous LAB can be considered as the crucial factor for the maintenance of typical features of artisanal cheeses. The mean count (expressed as log cfu/g) of LAB in Sjenica cheese throughout the 90 days of ripening is presented in Table 4. Results of the LAB study showed that Lactococcus spp. and Lactobacillus spp. were the dominant populations, followed by Enterococcus spp. The number of Lactococcus spp. reached level of 10 7 cfu/g and Lactobacillus spp. and Enterococcus spp. attained 10 6 and 10 4 cfu/g, respectively. Enterococci have been reported to be one of the most resistant microbial groups to restrictive conditions in the cheese matrix such as salt and acidity which may be the explanation for the slight increase in the enterococcal number.
CoPS isolates were screened for antimicrobial susceptibility/resistance profi le by applying the agar disc diffusion method (Figure 1.) In our study, we noticed that all tested staphylococci showed phenotypic resistance to at least one antibiotic. Multiresistance profi le which was defi ned as resistance to 2 or more of antibiotics tested was described in 33.33 % isolates of coagulase positive staphylococci. None of the CoPS isolates showed resistance to oxacillin. It is worth mentioning that 83.33% CoPS isolates were characterized by vancomycin resistance. LAB microbiota of Sjenica cheese were screened for antimicrobial resistance phenotypes by disc diffusion test (Figure 2, 3, 4). The screened LAB population demonstrated different profi les of phenotypic antibiotic resistance and multiple resistance to the tested antibiotics.
The most prevalent resistance phenotype among lactococci was resistance to streptomycin (83.33%), ampicillin and penicillin (70.83%). Resistance to vancomycin was detected in 16 (66.66%) isolates of tested lactococci.  The tested lactobacilli isolates were susceptible to erythromycin, clindamycin, chloramphenicol and ampicillin. The most prevalent resistance phenotype was resistance to vancomycin (62.5%), followed by resistance to tetracycline (54.17%).
Out of 57 tested isolates of enterococci, 47 (82.46%) isolates exhibited resistance to streptomycin. Vancomycin resistance was detected in 17 (29.82%) enterococcal isolates. None of the tested isolates of enterococci showed resistance to clindamycin and chloramphenicol.
As phenotypic screening of CoPS and LAB by agar disc-diffusion method showed the tetracycline resistant subpopulation, our next step was to perform the MIC evaluation (Table 5). Currently, there is no offi cally recognized cut-off value for LAB, so as a reference, we used the breakpoints established by the FEEDAP Panel of the European Food Safety Authority [20].  Figure 5).

DISCUSSION
Raw milk cheeses are typical dairy products associated with foodborne intoxication caused by staphylococcal enterotoxins (SE). It is generally considered that enterotoxinogenic staphylococci must reach levels of at least 10 5 -10 6 cfu/g to produce detectable amounts of SE which signifi cantly correlate with a toxic dose of less than 1μg required for clinical manifestation of intoxication [27]. This critical level of staphylococci may be reached in the early phase of cheese manufacturing (from 5 to 48 hours) when the cheese matrix is not acidifi ed to a growth-restricting pH value and the competing LAB population has not reached a high number. Therefore, we followed the staphylococcal population (coagulase positive staphylococci-CoPS) throughout the hazardous period from the initial phase of Sjenica cheese manufacturing from raw milk to cheese aged for 7 days.
The slight increase in number of CoPS from milk to gel is partly due the physical phenomenon of bacterial entrapment by curd particles during milk coagulation [21,22].
During the initial phase of cheese manufacturing (coagulation, curd formation and handling) we noticed an increase in staphylococcal number as acidifi cation rate governed by LAB fermentation activity was slow. The pH value of curd amounts 5.77±0.15 (results not shown). Cogan et al. [23] showed that the majority of LAB isolates originated from 25 European artisanal cheeses were not good acid producers and reduced the pH of milk to 5.3 in 6h at 30 o C. This moderate acidifi cation is not suffi cient to inhibit the staphylococcal growth. It was found that the number of CoPS decreased from day 3 to day 7 and was only present in low numbers (10 2 cfu/g) by day 90 (results not shown). The similar dynamic of CoPS population in raw milk cheese matrix was reported by other authors [24][25][26] who pointed out that inhibitory effects of natural LAB microbiota, reduced water activity (a w ) resulting from whey drainage, increased salt concentration during brining and temperature profi le of the ripening process generate a hostile environment suboptimal for staphylococcal growth [27].
Out of 26 CoPS isolates, 12 (46.16%) were determined as enterotoxigenic. Similar prevalence of enterotoxigenic CoPS isolates was detected by a number of authors [28][29][30]. We have to bear in mind the scientifi cally well argumented fact that even the coagulase positive staphylococci characterized as SE producers, will not produce toxins in every cheese matrix. It is attributed to the fact that SE synthesis is infl uenced by a number of intrinsic cheese parameters such as pH, water activity, redox potential, and to a great extent by bacterial antagonism as staphylococci are quite sensitive to microbial competition. Furthermore, the technological parameters such as acidifi cation kinetics, intensity of curd operation, temperature profi le of curd handling and ripening process, as an interplay, may generate a restrictive microenvironment which does not support enterotoxins synthesis [27].
The enterotoxigenic isolates of staphylococci were identifi ed as Staphylococcus intermedius (11) and Staphylococcus aureus (1). Lamprell et al. [31] pointed out that all CoPS counted on Baird Parker agar do not necessarily belong to the S. aureus species, as some colonies may be identifi ed as S. intermedius. Staphylococcus intermedius is the predominant non-S. aureus species isolated from food; some strains were characterized by enterotoxigenic potential [32] and shown to be clearly involved in staphylococcal food poisoning outbreaks [33]. Capurro et al. [34] described the presence of S. intermedius in bovine milk in a range of 0,2 to 2%. Staphylococcus aureus and Staphylococcus intermedius were identifi ed as dominant species among enterotoxigenic CoPS isolates from Turkish artisanal white cheeses [35].
Antibiotic resistance is a major public health concern as it is an ecological phenomenon with resistant bacteria persisting and circulating in the environment with possible transmission to humans via contaminated food.
In our study, 12 cheese-associated CoPS isolates were investigated for their resistance to antibiotics (Figure 1.) Multiresistance is a common characteristic of 33.33 % CoPS isolates. None of the tested CoPS isolates were characterized by oxacillin resistance. Nevertheless, the scientifi c community agrees that accurate detection of oxacillin/methicillin resistance can be diffi cult due to the presence of two subpopulations -one susceptible and the other resistant-coexisting within a culture of staphylococci. This phenomenon is called heteroresistance and frequently occurs in oxacillin-resistant populations of staphylococci [36]. Vancomycin resistance is coded by genes carried on genetic mobile elements such as plasmids and transposons. The transmissive nature of vancomycin resistance contributes to the dissemination of this resistance profi le among staphylococcal populations. The incidence of vancomycin resistant Staphylococcus aureus in raw milk, cheeses and biofi lms has been reported worldwide [37][38][39]. Abulreesh and Organji [40] noticed remarkable resistance to β-lactams and vancomycin among Staphylococcus aureus and coagulase negative staphylococci recovered from raw milk, cheese samples, potable water and biofi lms. Our results indicated vancomycin resistance in 83.33% CoPS isolates originated from autochthonous Sjenica cheese. On the contrary, Spanu [41] pointed out that all tested staphylococci isolated from raw sheep's milk cheese were susceptible to vancomycin.
There is a growing scientifi c evidence that commensal bacteria, especially lactic acid bacteria established at high numbers in fermented dairy ecosystems, may serve as reservoirs of antibiotic resistance genes potentially transferable to human and animal pathogens [42,43]. Hence, the investigation of LAB antibiotic resistance patterns may represent an effi cient tool in predicting the antibiotic resistance among clinical pathogens. Extensive literature data pointed out that genes conferring resistance to several antimicrobials (i.e., chloramphenicol, erythromycin, streptomycin, tetracycline, and vancomycin), hosted on mobile genetic elements, have been characterized in lactococci [44], lactobacilli [45,46] and enterococci [47,48] from food.
Pioneer antibiotic susceptibility studies have showed Lactococcus strains to be susceptible to most antimicrobial agents [49][50][51]. Conversely, most recent studies have confi rmed isolates of Lactococcus lactis resistant to chloramphenicol, erythromycin, streptomycin and tetracycline [10,52]. In our study, the majority of lactococci isolates were resistant to streptomycin (83.33%), ampicillin and penicillin (70.83%). Noteworthy, vancomycin resistance has been shown by 66.66% of tested lactococci. Contrary to our results, no Lactococcus isolates originated from traditionally fermented Indian food showed phenotypic resistance to tested antibiotics including the most representative ones among aminoglycosides, beta-lactams, cephalosporins, chloramphenicol, glycopeptides, lincosamides, macrolides and tetracyclines [53]. Toomey et al. [54] reported on phenotypic resistance screen of Lactococcus lactis isolates to 6 common antibiotics (ampicillin, chloramphenicol, erythromycin, streptomycin, tetracycline, and vancomycin), although molecular characterization of streptomycin resistance failed to give amplicons corresponding to any of the known streptomycin resistance genes.
Our results showed a high level (62.5%) of phenotypic resistance to vancomycin among lactobacilli isolated from Sjenica cheese. This type of resistance associated to many Lactobacillus spp. has been often described as intrinsic [13]. However, in spite of the propulsion of vancomycin resistance among lactobacilli, it has also been suggested that it cannot be an intrinsic feature of the species due to the variability occurring among L. delbrueckii subsp. bulgaricus, L. acidophilus, L. johnsonii, and L. crispatus strains [55,56].
All tested lactobacilli showed susceptibility to ampicillin and also high susceptibility toward penicillins which coincides with the general observation that lactobacilli are sensitive to the inhibitors of cell wall synthesis [13].
Enterococci are considered intrinsically resistant to beta-lactams [57], but our results do not support this observation since only two isolates of enterococci were characterized by penicillin and ampicillin resistance. Similar fi ndings were recorded by other authors [58,59]. We observed a high prevalence of streptomycin resistance (82.46%). Using a 120 μg gentamicin disc, as a reliable indicator of high level gentamicin (HLG) resistance [60], we also found out that 12 (21.05%) enterococcal isolates showed HLG resistance. Similar results were obtained by other authors [61][62][63]. Prevalence of vancomycin resistance among enterococcal isolates was 29.82%. Bulajić & Mijačević [64] pointed out that among enterococcal strains isolated from autochthonous Sombor cheese, only one strain showed vancomycin resistance. In contrast, Citak et al. [65] have shown resistance to vancomycin among the population of enterococci isolated from Turkish white cheeses and was found in 96.8% of E. faecalis isolates, and 76% of E. faecium strains. Having in mind the fact that aminoglycosides (in combination with glycopeptides) are considered antimicrobials of choice for the treatment of enterococcal infections, the worst case scenario related to possible dissemination of this resistance through the food chain highlights the biological hazard.
Tetracycline resistance in most bacteria is governed by acquisition of new genetic material. Currently, there are more than 40 different genes coding for tetracycline resistance [66], mediated mainly by two mechanisms: protection of ribosomes and energy-dependent effl ux of tetracycline. The variety of tetracycline genes and their localization on the mobilome, mainly conjugative plasmids and transposons [67,68] explains the wide distribution of TCr among foodborne LAB [42,13] and other bacterial genera.
Literature data showed the existence of intergenus and interspecies differences and also species dependency of lactobacilli resistance to various antibiotics [69]. Hence in performing MIC evaluation of lactobacilli, we applied group-specifi c or species specifi c MICs values. By applying the API identifi cation system, the lactobacilli isolates were identifi ed as Lactobacilus paracasei subsp. paracasei, Lactobacillus plantarum and Lactobacillus brevis (results not shown), so we used the cut-off value specifi c for corresponding species.
The MIC evaluation of presumptive tetracycline resistant populations of CoPS and LAB showed that majority of 30 tested isolates should be considered resistant to tetracycline (23 isolates).The discrepancy between the results obtained by discdiffusion and E-test may be due to several factors: insuffi cient growth of LAB on standard test media; undesirable interaction between complex media components and tested antibiotics, but also a recognizable effect of inoculum size, temperature and time of incubation. At last but not least, no validated breakpoints for the discrimination of resistant and susceptible strains are defi ned, which additionally complicates the antimicrobial susceptibility assessment.
In order to clearly identify the risk of antibiotic resistance through the food chain, we need to characterize the genetic base of tetracycline resistance and therefore to complete the picture about the nature of antibiotic resistance. As the fi rst step, we used tetracycline resistance genes encoding the ribosomal protection proteins (RPPs). As phylogenetic analysis confi rmed the monophyletic origin of these determinants, it is possible to design a set of PCR primers which in positive samples give the amplicons of RPP genes in general [70]. After the isolates were subjected to PCR amplifi cation with the universal primer set, we exposed the tested isolates to PCR screening with classspecifi c primers Tet M and Tet K. Tet K protein revealed the different mechanisms of tetracycline resistance-i.e. energy-dependent effl ux of tetracycline.
All isolates were positive for RPP genes; out of the 23 tested isolates, 14 (60.87%) were characterized by tet(M) gene presence, while 2 lactobacilli isolates revealed the presence of tet(K) resistance determinant. In our study, PCR amplicons of some isolates revealed the presence of RPP genes, but not amplicons for tet(M) suggesting that other tetracycline resistance genes are present. There is an observation that tet(M) gene is the most frequently found tetracyline resistance determinant in LAB [71]. The widespread distribution of tet(M) among members of L. plantarum, L. curvatus, L. casei, L. acidophilus, L. gasseri and L. crispatus was reported by several authors [46,71,72]. Ammor et al. [73] confi rmed coexistence of two tetracycline resistance genes tet(M) and tet(K) in a foodborne strain of Lactobacillus sakei. Our results did not support this fi nding. Molecular characterization of tetracycline resistance genes among staphylococci and LAB population originated from Spanish and Italian retail cheeses detected tet(S), tet(W) and tet(M) as the most common resistance genes (12). Florez et al. [12] reported the presence of the tet(M) gene in two L. lactis strain isolated from an artisanal starter-free cheese. We confi rmed the presence of the tet(M) gene in 2 out of 5 tested lactococcal isolates. The general observation pointed out that the presence of tet(M) has not been as frequently associated with lactococci as to other LAB such as E. faecalis and Lactobacillus spp. [74][75][76]. The transmissible nature of tet(M) and tet(K) genes originated from food-borne lactobacilli were confi rmed as it has been shown their association to host a Tn916 transposon [76] and a plasmid [73] respectively.
In conclusion, the presence of coagulase positive enterotoxigenic staphylococci in Sjenica cheese does not present the objective risk as compliance with the principles of Good Hygienic Practice and Good Manufacturing Practice allows reliable control of contamination level and growth of staphylococci. Although the small proportion of CoPS and LAB isolated from Sjenica cheese were characterized by the presence of tet genes, the well-documented transmissible nature of these resistance determinants justifi ed the need for further investigation in order to completely characterize the molecular mechanism of possible lateral transfer and consequently resistance dissemination through the food chain.