Characterization of Enterococcal Community Isolated from an Artisan Istrian Raw Milk Cheese : Biotechnological and Safety Aspects

In this study, prevalence, biotechnological and safety profi les of 588 Enterococcus isolates isolated from raw milk and Istrian cheese during diff erent stages of ripening were analyzed. Despite the low and variable presence of enterococci in milk ((3.65±2.93) log CFU/mL), highly comparable enterococcal populations were established aft er 30 days of cheese ripening ((7.96±0.80) log CFU/g), confi rming Enterococcus spp. as a major part of the core microbiota of Istrian cheese. The dominant species were E. faecium (53.8 %) and E. faecalis (42.4 %), while minor groups, consisting of E. durans (2.84 %) and E. casselifl avus (0.95 %), also occurred. A pronounced intraspecies variability was noticed based on molecular fi ngerprinting, with 35 strains (genotypes) detected. Most of the genotypes were farm-specifi c with one third being shared between the farms. This genotype variability refl ected particular diff erences of Istrian cheese production, mainly variable salt concentration, ripening temperature and air humidity as well as microclimatic or vegetation conditions. There was considerable variation between the strains of the same species regarding wide range of biotechnologically important traits as well as their ability to survive in simulated gastrointestinal conditions. A considerable number of strains were resistant to critically important antibiotics such as tetracycline (43.56 %), erythromycin (35.79 %) and vancomycin (23.48 %). Polymerase chain reaction-based detection did not identify any of the common genetic determinants for vancomycin and erythromycin resistance; for tetracycline tetM gene was detected. The presence of virulence genes including agg, efaAfs, gelE, cylM, cylB, cylA, esp, efaAfm, cob and cpd was frequently recorded, especially among E. faecalis strains.


Introduction
Enterococci are one of the most common groups of bacteria in diverse ecosystems, mainly due to their high adaptability and resistance to adverse environmental conditions.They have been frequently isolated from natural sources such as soil, water, plants and gastrointestinal (GI) tract of mammals (1).Also, they are an integral part of many spontaneously fermented, traditional cheese varieties in which they contribute to the fermentation and ripening process (2,3).Thanks to their proteolytic activity, hydrolysis of milk fat by esterase and production of typical fl avour components such as acetaldehyde, acetoin and diacetyl (2,4), they are responsible for the development of particular sensory characteristics of many cheese varieties.As indigenous members of the GI tract, enterococci can have benefi cial eff ect on human health by balancing gut microbiota; therefore, they may also act as probiotic bacteria (1).
Enterococcus spp.are nowadays recognized for their benefi cial eff ects on the development of cheese fl avour and are considered as an integral part of the natural microbiota of many traditional types of cheese; however, some strains may present an emerging threat due to the enterococcal virulence determinants that are able to cause infections in humans (3,5).Over the last three decades, epidemiological data have noted the emergence of particular ecotypes of enterococci as leading nosocomial pathogens and they have arisen to become a major cause of human diseases, especially in immunocompromised patients (6).Their plasmid and transposon-mediated resistance to tetracycline, erythromycin, chloramphenicol and glycopeptides is considered particularly problematic, since it leads to the increased number of infections related to the acquired antimicrobial resistance of enterococci (6).Their capacity to exchange genetic material by conjugation is well documented (7) and it can take place in GI tract, where genes coding for resistance to antibiotics and virulence factors can be transferred (5).As enterococci are usually present at fairly high levels in ready-to-eat cheese products (8,9), they may have signifi cant infl uence on consumers' health and as such, their safety aspect should be carefully evaluated.
Istrian cheese is a hard type of artisan raw milk cheese that is manufactured at home, on small scale, on the whole Istrian peninsula, without the addition of starter cultures (10).It is characterized by ripening of 90-120 days.However, because of its popularity in the region and due to the commercial pressure, the ripening time has been shortened and it has become market practice to sell the cheese as soon as possible.During the ripening of Istrian cheese, the number of enterococci reached up to 10 8 per g when assessed by real-time PCR (11).They are also found to be one of the dominant bacterial populations during ripening when denaturing gradient gel electrophoresis (DGGE) or next-generation sequencing approach were applied (11,12).Although total microbial diversity of Istrian cheese is analysed in details by culture-independent methods, such as molecular, sequencing-based assays did not allow distinguishing among Enterococcus species (E.faecalis, E. faecium and/or E. durans) or providing information on their intraspecies variability.Due to the bivalent eff ect of enterococci on food and human health, there is a need for their detailed characterization, not only at the species but also at the strain level.For example, strains of the species E. faecalis and E. faecium have been recognized as the most dominant species in fermented cheese (4), but also strains of the same species have been most frequently associated with the human and animal infections (13).
The purpose of this work is to identify and characterize members of the enterococcal community isolated during the ripening of Istrian cheese and to assess their role in cheese ripening as well as to select novel strains with interesting features that could be used as starter or adjunct cultures for fermented foods.Special focus is set towards the surveillance rate of indigenous cheese enterococci in simulated gastrointestinal conditions as well as detection of their antibiotic resistance and virulence traits.

Raw milk and cheese sampling
Six batches of Istrian cheese produced by diff erent manu facturers (F1-F6) across the Istrian peninsula (Croatia) were analysed in the frame of this study.All manufacturers are part of a consortium for the preservation of the certifi ed product Istrian cheese and are located in diff erent areas of Istria with diff erent microclimatic conditions, relief and vegetation.At each farm, two independent cheese samples were produced from the same batch of raw, full-cream ewe's milk in accordance with the traditional cheese making procedure (10).In brief, milk is heated up to 35 °C prior to the addition of natural rennet.The milk coagulum is left for maximum 1 h at room temperature until the curd formed, which is cut into pieces and cooked at 42 °C for 5 min before moulding.Aft er 2 h of whey drainage with self-pressing, the surfaces of cheese are covered in co arse salt (various mass fractions) and left for 2 days at room temperature.The cheese is placed in a ripening chamber at controlled but variable temperature and air humidity (up to 19 °C and up to 90 %, respectively) for 90-120 days.
The cheese samples were taken with sterile cork borers (1 cm in diameter) from two individual cheese samples of each manufacturer aft er 30 (Ch30), 60 (Ch60), 90 (Ch90) and 120 (Ch120) days.In addition, milk (M) and cut curd (Cr) samples were taken at day zero.All samples were stored at 4 °C and analyzed within 24 h aft er sampling.The pH of the ripened cheese samples was 4.7 to 5.4 and their fat and salt contents ranged between 40 and 52 %, and 3.4 and 4.7 %, respectively.

Isolation of enterococci
The milk and cheese samples were subjected to mi crobiological analysis for enumeration and isolation of Enterococcus species.The cheese samples were homogenized with a Stomacher ® (BagMixer 400; Interscience; St Nom, France) for 3 min and diluted tenfold in a sterile physiological solution (0.9 % NaCl).Aliquots of 100 μL of the three highest dilutions were inoculated on two selective kanamycin aesculin azide (KAA) agar plates (Merck, Darmstadt, Germany) and incubated under aerobic conditions at 37 °C for 48 h.Ten to fi ft een colonies were randomly selected from the highest dilution of each sample and purifi ed twice on KAA agar.Overall 588 isolates were collected and further analyzed.The pure cultures were characterized using the Gram stain method and catalase test.Aft er microscopic examination, Gram-positive and catalase-negative coccoid isolates were grown overnight in 4 mL of brain-heart infusion (BHI) medium (Merck) supplemented with 20 μg/mL of kanamycin.One aliquot of overnight cultures was used for DNA extractions and another one was stored at -80 °C in BHI (Merck) broth supplemented with glycerol (80:20 by volume).

DNA extraction and identifi cation at the genus and species level
Genomic DNA from isolates was extracted from overnight BHI cultures using NucleoSpin Tissue Kit (Macherey-Nagel, Düren, Germany) according to the manu facturer's instructions.The quantity of DNA extracts was checked with a Nanodrop spectrophotometer (NanoDrop Technologies, Wilmington, DE, USA).
To determine the affi liation of isolates to genus level, PCR assay was carried out with Enterococcus genus-specific primers as described by Deasy et al. (14).A subsequent multiplex PCR reaction with E. faecalis and E. faecium species-specifi c primers was carried out in accordance with the PCR protocol of Dutka-Malen et al. (15).The primers are listed in Table 1 (5,14-23).Isolates that were not assigned to Enterococcus genus were not characterized further.Enterococci that were not affi liated to E. faecalis or E. faecium species were additionally amplifi ed and sequenced using the universal bacterial primer sets 27f and 1401R designed by Lane ( 16) and Nübel at al. (17).The PCR mixture and amplifi cation program were as previously described by Schreiner et al. (24).The PCR products were separated on 1.5 % agarose gels (100 V, 1 h) and purifi ed by NucleoSpin Gel and PCR Clean-up Kit (Macherey-Nagel).PCR products were sequenced on an ABI PRISM ® 3730 DNA Sequencer (Applied Biosystems, Foster City, CA, USA) using a BigDye Terminator v. 3.1 Cycle Sequencing Kit (Applied Biosystems).Reaction conditions and thermal cycle program were as previously described by Mrkonjić Fuka et al. (25).Unincorporated dye terminators were removed by ethanol precipitation.

Fingerprinting of Enterococcus spp.
For a strain-specifi c characterization, PCR assay based on randomly amplifi ed polymorphic DNA (RAPD) was used.Amplifi cation reactions were performed in a 25-μL reaction volume containing 50 ng of template DNA, 1× PCR buff er, 0.1 mM dNTP, 1.5 mM MgCl 2 , 2 μM primer and 1.25 U of Taq DNA polymerase (Invitrogen, Thermo Fisher Scientifi c, Waltham, MA, USA).The used primers (M13, D8635 and P2) are listed in Table 1.Thermal cycler was programmed to: initial denaturation at 94 °C for 30 s, 94 °C for 5 min for complete denaturation, 37 °C for 5 min for low stringency annealing and 72 °C for 5 min for extension for 2 cycles.This was followed by 40 cycles of 1 min at 94 °C, 1 min at 37 °C, 2 min at 72 °C.Final extension was programmed at 72 °C for 10 min.Amplifi ed PCR products were separated by electrophoresis on 2.5 % (by mass per volume) agarose gels.Clustering was carried out in Bionumerics v. 7.6 soft ware platform (26) using the unweighted pair group method with arithmetic average (UPGMA) clustering algorithm (27) and band-based Dice correlation coeffi cient for creating the similarity matrix.Based on their RAPD-PCR patt ern, representative strains were selected and tested for further phenotypic, technological and safety properties.Monophyletic groups were omitt ed from further analysis.

Acidifying activity and tolerance to diff erent environmental conditions
Overall 35 representative strains were tested for their acidifying activity.The diff erences in the pH were measured aft er 2, 6 and 24 h of incubation at 37 °C using a pH--meter.Combined pH electrodes (InPro ® 3030; Mett er Toledo, Greifensee, Switzerland) were standardized using two buff ers (pH=4.0 and 7.0) and disinfected aft er each run using a 3 % HCl solution.Strains were inoculated in duplicates at a level of 2 % in reconstituted sterile nonfat dry milk (10 % by mass per volume; Fluka, Sigma-Al drich, St. Louis, MO, USA).The acidifi cation rate (as ΔpH) was calculated according to Jamalay et al. (28): The ability of representative strains to grow at 10 and 45 °C, in 4 and 6.5 % NaCl and at pH=4 and 9.6 was tested in duplicates in BHI broth and recorded aft er 72 h of growth at 37 °C.

Lipolysis, haemolysis, caseinolytic and proteolytic activity
For screening of lipolytic and caseinolytic activity, the representative strains were cultured overnight at 37 °C in BHI broth.Overnight cultures were subcultured in fresh BHI broth and grown until an absorbance (A) of 0.3 was achieved.A volume of 10 μL of cultures was transferred on sterile cellulose discs (Biorad, Berkeley, CA, USA) previously placed on Tributyrin Agar (Oxoid LTD, Basingstoke, UK) to screen for lipolytic activity, or on BHI agar containing 1.5 % of skimmed milk for caseinolytic activity.Plates were incubated for 3 days at 30 °C for lipolytic or at 37 °C for caseinolytic activity and observed daily for halo formation around the discs.The radius of the halo formation (in mm) at the end of incubation was used as measure of the lipolytic or caseinolytic activity.The sizes of clear zones (including the diameter of disk) on the agar surface around the discs were measured in mm and expressed as -(no halo observed or less than 10 mm) or + (≥10 mm).The haemolysis was tested by streaking overnight cultures on BHI agar plates containing 5 % cow's blood.The plates were incubated for 72 h at 37 °C.Zones of clearing around colonies indicated haemolysin production.Proteolytic activity was evaluated by the addition of chromogenic peptide N-succinyl-Ala-Ala-Pro-Phe p-nitroanilide (S-Ala, Sigma-Aldrich).Release of p-nitroanilide (pNA) by the action of enterococcal proteinase is measured at 410 nm as described by Savoy de Giori and Hébert (29).Results were expressed in μM of pNA.

Survival in simulated gut conditions and autoaggregation assay
To estimate the transit tolerance through a simulated in vitro digestion, all representative enterococcal strains were tested as described by Doleyres et al. (30) and Nueno-Palop and Narbad (31) with slight modifi cations.To simulate the possible hydrolysis of bacteria in the oral cavity, enterococcal strains were grown overnight in skimmed milk (10 %, by mass per volume) and diluted 1:10 in phosphate-buff ered saline (PBS), aft er which lys-ozyme (Sigma-Aldrich) was added to a fi nal concentration of 100 mg/L.Samples were then incubated for 5 min at 37 °C.In order to estimate the cell survival rate in simulated gastric and intestinal conditions, overnight enterococcal cultures (1 mL) were harvested by centrifugation (Eppendorf 5415R; Eppendorf, Hamburg, Germany) at 10 000×g for 10 min at room temperature.The pellets were washed twice with 0.1 % peptone water, resuspended in 100 μL of 0.1 % peptone and stored on ice until use.To simulate the gastric digestion, 0.5 % NaCl and 0.3 % pepsin solution (Sigma-Aldrich) were used.The samples were adjusted to pH=2.5 (with 1 mol/L of HCl).The solu- tion containing 270 μL of simulated gastric juices and 30 μL of cell suspension was mixed and incubated at 37 °C for 30 min.For cell survival in simulated intestinal conditions, 270 μL of 0.4 % bile salts and 0.2 % pancreatin solution (Sigma-Aldrich) were mixed with 30 μL of cell suspension and incubated for 30 min at 37 °C.Samples were removed for determination of cell counts before and aft er oral, gastric and intestinal digestion.Aliquots were serially diluted in PBS buff er (pH=7.2) and viable counts were determined aft er incubation at 37 °C for 48 h.Each strain was tested twice on BHI agar.
Autoaggregation assay was performed according to Fortina et al. (32) and Del Re et al. (33) with slight modifications.Aggregation was evaluated using 4 mL of cell suspension corresponding to viable counts of approx.10 8 CFU/mL.The cell suspension was vortexed for 10 s.The absorbance (A) was measured at 610 nm aft er 3 and 5 h of incubation at room temperature.The aggregation percentage was expressed as: where A t represents the absorbance measured at 610 nm at time t=3 and 5 h, and A 0 the absorbance at t=0.

Detection of antibiotic resistance genes
Detection of the genes responsible for resistance to tetracycline (tetK, tetL, tetM, tetO and tetS), vancomycin (vanA and vanB) and erythromycin (ermA and ermB) was investigated for all strains exhibiting the respective phenotypic resistance patt ern by specifi c PCR assays.Primers used are listed in Table 1.The DNA amplifi cation was performed according to Ng et al. (22) and Sutcliff e et al. (23) in 25 μL of reaction mixture containing 100 ng of DNA template.The amplifi ed PCR products were visualized on 1.5 % agarose gel aft er staining with ethidium bromide.

PCR detection of genes coding for virulence factors
The indices of genes encoding for following virulence factors were investigated: aggregation substance (agg), gela-tinase (gelE), cytolysin (cylM, cylB), cytolysin activator (cylA), enterococcal surface protein (esp), cell wall adhesin (efaAfs from E. faecalis and efaAfm from E. faecium) and sex pheromones (cpd and cob).PCR amplifi cation was performed with primers and conditions described by Eaton and Gasson (5), except for cob and agg genes, whose 29 cycles of elongation were prolonged to 45 s as described by Čanžek Majhenić et al. (35).The primers are listed in Table 1.

Statistical analysis
Statistical analysis was performed using the R software programming language and R soft ware environment for statistical computing (36).The level of significance was p<0.05.Student's t-test was used to determine whether there were signifi cant diff erences in results when E. faecalis and E. faecium were compared.Post hoc Tukey's honestly signifi cant diff erence (HSD) test was used to ascertain whether there were signifi cant diff erences between the results of samples from diff erent stages of cheese manufacturing and ripening.

Enumeration and identifi cation of enterococci
Despite the available data in other countries, studies on prevalence and safety issue of enterococci in spontaneously fermented cheese in Croatia are limited.In order to improve the knowledge on the role of enterococci in the cheese ripening process and their possible infl uence on public health as well as to select for potential adjunct/ starter cultures, total viable count and isolation of enterococci from 6 diff erent farms during the ripening of Istrian cheese were carried out.The lowest count of enterococci was noticed, as expected, in milk ((3.65±2.93)log CFU/ mL) and aft er 30 days of ripening it reached the maximum value ((7.96±0.80)log CFU/g).Aft erwards, the number was 10-fold lower and remained stable until 90 days of ripening.At the end of ripening (120 days) the average number of enterococci was (6.21±0.98)log CFU/g (Fig. 1).
As nowadays Istrian cheese is rarely ripened for 120 days and it is mostly sold and consumed during the fi rst two months, the average load of enterococci within that ripening period is almost 10 8 CFU/g, exceeding the limit of maximum 10 6 CFU/g recommended by International Commission on Microbiological Specifi cations for Foods  (37).The number of enterococci higher than 10 6 CFU/g and even up to 10 8 CFU/g has oft en been reported in different raw milk products (9,38) and in some cases irrespective of raw milk or pasteurized/thermally treated milk cheese (8).
As particular species in the same microbial group, or even the particular strain of a given species specifi cally affect the organoleptic profi le of the fi nal product and may have unique technological or safety traits, a total of 588 presumptive enterococci isolated from KAA agar in this study were identifi ed by PCR-based approach to the genus, species and strain level.In PCR reaction with genus--specifi c primers, 92.19 % of isolates were confi rmed as Enterococcus members (data not shown).PCR identifi cation at the species level identifi ed 53.8 % of Enterococcus isolates as E. faecium and 42.4 % as E. faecalis (Fig. 2).They were found at the high level in all samples and their presence remained stable during the fermentation and ripening.The predominance of E. faecium and E. faecalis over other enterococci is in agreement with other studies on artisan cheese, where high abundance of these two species has been reported (1).E. durans and E. casselifl avus were identifi ed based on the sequencing of 16S rRNA gene and were detected in much lower percentage (2.84 and 0.95 % respectively).
The genetic diversity of the identifi ed species was analyzed by RAPD-PCR-based fi ngerprinting (Fig. 3).The RAPD-PCR allowed detailed characterization of isolates at the strain (ecotype) level.Thus, we used this approach to select representative strains for further technological and safety analysis.RAPD-PCR fi ngerprinting based on M13, D8635 and P2 primers discriminated 35 clusters that comprised two or more isolates.From each of these clusters, one representative strain was selected and further analyzed (a total of 35 strains).The enterococcal isolates (2.4 %) that displayed unique banding patt erns were omitt ed from further analysis.Although the same distribution profi le of Enterococcus species was noticed among six farms in this study, most of the detected genotypes were farm specifi c.Of 35 detected genotypes, fi ve dominant ones (approx.30 % of total isolates) were shared among the farms.Such a genotype distribution refl ected the particular diff erences in traditional Istrian cheese production.Mainly, variable salt concentration, ripening temperature and air humidity as well as microclimatic or vegetation conditions may be the reason for specifi c strain distribution.

Phenotypic characterization and biotechnological properties of selected strains
During cheese manufacturing, lactic acid bacteria are exposed to diff erent temperatures, pH and salt concentration.The recovery and persistence of enterococci in some types of cheese during ripening can be att ributed to their wide range of growth temperatures and their tolerance to pH and salt (3).All 35 representative strains tested in this study were able to grow at 10 and 45 °C in the presence of 4 and 6.5 % NaCl as well as at pH=5 and 9.6, which is in line with the high adaptability of enterococci to harsh environmental conditions.
A rapid decrease of milk pH is essential for coagulation and is therefore a crucial step in the cheese manufacturing process.In general, the investigated strains in this study showed low acidifying activity.None of the isolates were able to reduce the milk pH by more than 1.51 units even aft er 24 h of incubation (Table 2).Only three strains belonging to E. faecalis, one E. faecium, one E. casselifl avus and one E. durans strain (19.4,0.38, 0.95 and 2.84 % of total identifi ed enterococci) were able to lower the pH of 10 % skimmed milk below pH=5 (initial pH=6.33).No significant diff erences were found between E. faecalis and E. faecium strains aft er 2 and 6 h of incubation, whereas aft er 24 h E. faecalis strains exhibited bett er acidifying potential by lowering the pH of the milk by 1.09 units on average, while E. faecium lowered the pH of the milk by 0.87 units (p<0.01) on average.The poor acid production by enterococci has also been noticed in other studies and most strains reduce the pH of the milk by only 0.4-0.8units.Likewise, in our study, overall bett er acidifi cation properties of E. faecalis strains than other enterococcal species were found (39,40).
Although isolates that show low acidifi cation capacity are not ideal candidates for use as part of starter cultures, such strains can take part in the development of fl avour and texture, by proteolytic and lipolytic activity or as bioprotective cultures by producing antimicrobial substances.
The degradation of proteins and lipids plays an important role in the development of texture and taste of the cheese.Almost 40 % of tested enterococci in our study showed caseinolytic activity and lipolytic activity of nearly one third of strains (28.60 %) was measured by agar disc diff usion method (Table 2).Sarantinopoulos et al. (39) demonstrated the ability of majority of 129 enterococal species isolated from diverse environments to hydrolyse tributyrin and as such to be presumably involved in lipid conversion during the cheese maturation.
Overall, E. faecium strains had higher caseinolytic activity than E. faecalis and for lipolytic activity opposite trend was noticed.With regard to peptidase activity, as a result of measuring the amount of pNA released from the chromogenic peptide (S-Ala), E. faecalis had signifi cantly higher (p<0.01)activity than E. faecium (6641.91 and 6028.90 μM, respectively).It seems that E. faecium isolates from Istrian cheese are mainly involved in casein degradation, whereas lipolysis and peptidase activity are mainly due to the activity of E. faecalis strains, at least under the applied laboratory conditions.

Surveillance in simulated gut conditions
The ability of entereococcal strains to survive in the GI tract and their autoaggregation ability were tested in order to evaluate their colonization potential and passage through the GI tract, which is a prerequisite for the estimation of their possible infl uence on human health and their use as probiotics.
The 35 representative enterococci tested in this study were lysed in varying percentage in the conditions of human oral cavity, in the acidic environment of the stomach and in the intestinal conditions (Table 2).The survival rate in the fi rst barrier (human oral cavity) was between 38.52 and 98.30 % (on average 58.68 %).Although low pH and antimicrobial activity of pepsin are known to be very eff ective in decreasing viability of bacteria, survival rates of the tested strains in simulated environment of the stomach (pH=2 and presence of pepsin) were higher than in the conditions of human oral cavity, and was on average 75.45 % (3.19-99.32%).While the pH of the small intestine is more favourable for bacterial survival, the presence of pancreatin and bile salts may have detrimental eff ect.In simulated intestinal conditions (bile salts and pancreatin), survival rate among the tested strains was between 2.75-95.95% (on average 55.58 %).
Signifi cant diff erences in the ability to survive in simulated GI tract conditions were noted among Enterococcus species and among the strains of the same species (Table 2).E. faecalis showed higher survival rate in the conditions of human oral cavity than E. faecium (p<0.01), while E. faecium showed higher survival rate in gastric conditions (p<0.01).These fi ndings are in agreement with those reported by other authors for E. faecium probiotic strains that show higher resistance to low pH than other enterococci (41).No signifi cant diff erences were found between the survival rates of E. faecalis and E. faecium in intestinal conditions.
All tested strains exhibited poor autoaggregation ability, with values aft er 5 h ranging from 4.64 to 46.66 % (on average 13.22 %), therefore exhibiting low capacity of adhesiveness.

Safety issues
Enterococci in food of animal origin have been shown to harbour diff erent virulence factors, suggesting that this type of food could act as an important environmental reservoir for human infections.The presence of 10 genes coding for virulence factors and potential pathogenicity of enterococcal strains was evaluated by PCR (Table 3).Many of virulence genes, e.g.aggregation substances, cytolysin and gelatinase genes are known to have a silent state, making their genotypic study more reliable than phenotyping one because they might remain undetected under the in vitro conditions but be expressed in vivo, so the pathogenic potential of such strains might be underestimated (42).
PCR amplifi cation of the aggregation substance (agg) and surface protein (esp) genes revealed 27.65 and 9.85 % of positive strains in the current study.The esp gene is known to be associated with the biofi lm production, endocarditis and nosocomial infections and is usually detected at low frequency (43,44), quite opposite to a study of Semedo et al. (45) who detected 62 % of esp-positive strains, 57 % of which being of food origin versus 22 % from clinical samples (N=164).
Virulence traits involved in the production of the cytolysin (cylA, cylB and cylM) are one of the most studied virulence traits associated with enterococci and are usually detected in low number of isolates (46).Of all entero-  (44,46).
The genes gelE and cpd, associated with toxin production and sex pheromones, were also reported to be commonly found in commensal isolates (31).The sex pheromone determinants cpd and cob occurred with the frequency of 61.17 and 48.48 % respectively and gelatinase (gelE) gene was present in 40.34 % of our enterococcal isolates.
The widespread occurrence of adhesion-associated protein in enterococci, irrespective of origin and species allocation, is reported in many food and clinical isolates (45) and was confi rmed in our study with 77.65 % efaAfsand 81.25 % efaAfm-positive enterococci.
Overall, E. faecalis strains harboured 56.70 % more virulence genes than the tested E. faecium strains.Especially high incidence of sex pheromone (cpd and cob) and efaAfs determinants was noticed among E. faecalis isolates (100, 99 and 100 % respectively).Among E. faecium, 0.70 % of strains were lacking all virulence determinants and among 34.86 % of strains only one virulence gene was detected (efaAfm).The cylA and esp genes were absent from all E. faecium strains.A signifi cantly diff erent patt ern was noticed in E. faecium isolated in our study compared to strains tested by Eaton and Gasson (5), who did not detect agg, cylM, cylB or cylA and sex pheromone genes among the analyzed E. faecium strains.Although several studies have reported agg gene only in E. faecalis (5,47), in this study the presence of agg was demonstrated also in E. faecium.The aggregation substance (agg) promotes the conjugative transfer of sex pheromone plasmids by formation of mating aggregates between donor and recipient cells (48) and may support the strain probiotic characteristics (49).
For cob and cpd genes our data are in line with the literature on E. faecalis strains; however, the respective genes were also detected among E. faecium isolates (14.08 and 33.09 %, respectively).The detection of sex pheromones in species other than E. faecalis is confi rmed by the work of Heaton and Handwerger (50), who suggested that sex pheromone cross talk between E. faecium and E. faecalis has been established, which is also indicated in our study.
The number of virulence determinants was the highest in milk samples and was decreasing until the 30th day of cheese ripening, when it rapidly increased.Aft er 30 days of ripening, the number of virulence determinants continued to decrease and was signifi cantly lower (p<0.05)at the end of ripening.This implies a reduction of the number of strains possessing multiple virulence genes or virulence genes in general, rather than correlation to the total number of enterococci, as in milk samples the lowest CFU count and the highest number of virulence determinants was noticed.
Although enterococci can induce haemolytic reaction, none of the 35 tested strains in this study were found to cause haemolysis.Food that is not heat treated before consumption, e.g.spontaneously fermented cheese may contain viable antibiotic-resistant bacteria that are assumed to contribute to the transmission of antibiotic resistance genes into the GI tract.Species of the genus Enterococcus are intrinsically resistant to a wide variety of antimicrobials including cephalosporins, low levels of aminoglycosides, lincomycine and quinupristin/dalfopristin (7,44).However, enterococci effi ciently acquire and transmit antibiotic-resistant genes via mobile genetic elements as conjugative plasmids and transposons (51), which is pivotal in the dissemination and persistence of antimicrobial resistance in enterococci.
In the current study, only 0.76 % of strains were nonresistant to any of the tested antibiotics, 7.57 % of strains were resistant to one antibiotic and 7.95 % of the strains were resistant to two antibiotics, while others (83.72 %) were multiresistant (Table 4).None of the strains were resistant to penicillin (10 μg), ampicillin (10 μg) and ampicillin/sulbactam (30 μg).The vast majority of strains were resistant to clindamycin (2 and 10 μg; 92.80 and 63.07 %, respectively), streptomycin (10 μg; 82.00 %) and rifampicin (5 μg; 72.35 %), while total of 28.41 % of strains were resistant to chloramphenicol (10 μg).A considerable number of strains were resistant to critically important antibiotics such as tetracycline (5 and 30 μg; 43.56 and 17.99 %, respectively), erythromycin (5 and 15 μg; 35.79 and 29.35 %, respectively) and vancomycin (5 μg; 23.48 %).A high frequency of resistance to erythromycin and tetracycline in enterococci from various foods was also reported in Europe, Canada, Turkey and Tunisia (35,(52)(53)(54).The frequency of vancomycin-resistant enterococci in food is mostly low and varied between 0-25 % (6) and higher resistance to vancomycin in our study (although of low concentration) has to be considered as vancomycin is one of the few alternatives in treating enterococcal infections.
With respect to the genes responsible for resistance to tetracycline (tetK, tetL, tetM, tetO and tetS), vancomycin (vanA and vanB) and erythromycin (ermA and ermB), only tetM gene could be detected in the analyzed strains.The tetM gene is the most frequently detected in tetracycline--resistant enterococci worldwide and it is known to be mainly based on the transfer of Tn916-type transposons (55).

Conclusion
Our study confi rmed the presence of enterococci as a major part of the indigenous core microbiota of Istrian cheese during fermentation and ripening, with Enterococcus faecalis and E. faecium being the predominant species.Considerable variation between strains of the same species was recorded in their ability to exhibit a wide range of biotechnologically important traits as well to survive in gastrointestinal (GI) tract.Even though probably none of enterococcal strains were able to colonize the GI tract, they can survive in harsh GI environment to diff erent extents.The high rate of multiresistant enterococci and high load of virulence determinants detected in our study should be taken into consideration due to the high frequencies of transmission of genes encoding for antibiotic resistance and virulence factors in GI tract.The role of ente rococci in cheese ripening remains questionable, because several isolates possess virulence or antibiotic resistance genes and thus have po ten tial of endangering Table 4. Susceptibility of 35 representative enterococcal strains to antibiotics using the disc diff usion method Species Representative strain (N(strain)/N(total enterococci))/% Disc potency/μg TET (30) TET ( 5) RIF ( 5) ERY ( 15) ERY ( 5) VAN ( 5) PEN ( 10) AMP ( 10) STR ( 10) CLI ( 10) CLI ( 2) CHL ( 10) SAM (30) human health.In agreement with the literature, the majority of enterococci isolated from ripened Istrian cheese cannot fulfi l the strict safety rules needed for starter or adjunct cultures.Based on their good technological properties, absence of most of the virulence determinants and antibiotic resistance genes, only one strain, E. durans M2_7, is a good candidate for the application as starter or adjunct culture for fermented food production.

Table 1 .
List of primers used in this study

Table 2 .
Technological characteristics and probiotic potential of representative enterococcal strains Species

Table 3 .
Detection of virulence determinants