Microbial Ecology of Sheep Milk, Artisanal Feta, and Kefalograviera Cheeses. Part II: Technological, Safety, and Probiotic Attributes of Lactic Acid Bacteria Isolates

The aim of the present study was to examine 189 LAB strains belonging to the species Enterococcus faecium, E. faecalis, Lactococcus lactis, Pediococcus pentosaceus, Leuconostoc mesenteroides, Lactiplantibacillus pentosus, Latilactobacillus curvatus, Lp. plantarum, Levilactobacillus brevis, and Weissella paramesenteroides isolated form sheep milk, Feta and Kefalograviera cheeses at different ripening stages, for their technological compatibility with dairy products manufacturing, their activities that may compromise safety of the dairy products as well as their capacity to survive in the human gastrointestinal tract. For that purpose, milk acidification and coagulation capacity, caseinolytic, lipolytic, hemolytic, gelatinolytic, and bile salt hydrolase activity, production of exopolysaccharides, antimicrobial compounds, and biogenic amines, as well as acid and bile salt tolerance and antibiotic susceptibility were examined. The faster acidifying strains were Lc. lactis DRD 2658 and P. pentosaceus DRD 2657 that reduced the pH value of skim milk, within 6 h to 5.97 and 5.92, respectively. Strains able to perform weak caseinolysis were detected in all species assessed. On the contrary, lipolytic activity, production of exopolysaccharides, amino acid decarboxylation, hemolytic, gelatinase, and bile salt hydrolase activity were not detected. Variable susceptibility to the antibiotics examined was detected among LAB strains. However, in the majority of the cases, resistance was evident. None of the strains assessed, managed to survive to exposure at pH value 1. On the contrary, 25.9 and 88.9% of the strains survived after exposure at pH values 2 and 3, respectively; the reduction of the population was larger in the first case. The strains survived well after exposure to bile salts. The strain-dependent character of the properties examined was verified. Many strains, belonging to different species, have presented very interesting properties; however, further examination is needed before their potential use as starter or adjunct cultures.


Introduction
The microbiota that drives spontaneous fermentations has been in the epicenter of intensive research for decades. The primary aim was to depict the microecosystem composition, as well as the role, symbiotic patterns, and dynamics of the key players. Product standardization essentially requires the development of starter cultures that provide the necessary metabolic capacities to carry out the desired biotransformations. However, the former are usually accompanied by undesired activities that may compromise the safety of the product. This is also the case of dairy products. The criteria used for the selection of proper starter cultures may be divided into technological, safety, and probiotic-related attributes.
tobacillus brevis (29 strains), and Weissella paramesenteroides (32 strains) were included in the present study (Table S1). All strains were stored at −20 • C, in Nutrient Broth (LAB M, Lancashire, UK) supplemented with 30% glycerol (Applichem, Darmstadt, Germany). Before experimental use, strains were subcultured twice in de Mann, Rogosa, and Sharpe (MRS) broth (LAB M) at 37 • C for enterococci and 30 • C for the rest LAB, for 24 h. Purity was examined after streaking in MRS agar. Before inoculation, the bacterial cultures were washed twice with sterile saline and resuspended in the same diluent at the initial culture volume, unless otherwise stated.

Technological Properties 2.2.1. Milk Acidification and Coagulation
Milk acidification and coagulation was examined in 74 LAB representative strains, which were selected based on their different isolation source or separate genotypic cluster [20], namely E. faecium (strains DRD DRD 2616DRD , 2634DRD , 2635DRD , 2652DRD , 2655DRD , 2657DRD , 2676DRD , 2681DRD , 2689DRD , 2694DRD , 2748DRD , 2749DRD , 2752DRD , 2753DRD , 2761. Overnight bacterial cultures of the representative strains were used to inoculate (1% inoculum) 10 mL skim milk (LAB M). Incubation took place at 37 • C for enterococci and 30 • C for the rest LAB, for 24 h. The pH value was measured immediately upon inoculation and for every 3 h until 24 h. After each measurement, the respective test tube was discarded.
The ∆pH values presented were calculated by subtracting the pH value after the given incubation time at 30 • C (37 • C for enterococci) from the respective of uninoculated skim milk incubated for the same time at the same temperature.
Milk coagulation was performed according to Bancalari et al. [21]. In brief, overnight cultures of the aforementioned strains were inoculated (2% inoculum) into 10 mL skim milk (LAB M) and incubated at 37 • C for enterococci and 30 • C for the rest of the LAB, for up to 7 days. Milk coagulation was optically evaluated at 6, 12, 24 h, 2, 3, 4, 5, 6 and 7 days. The time in which coagulation was visibly observed in presented.

Caseinolytic and Lipolytic Capacity
The caseinolytic and lipolytic capacity of all strains were assessed through agar well diffusion assays. In the first case, wells were aseptically punched in freshly prepared lawns of a medium consisting of 0.5% tryptone (LAB M), 0.25% yeast extract (LAB M), 0.1% glucose (Applichem), 1% casein (Sigma-Aldrich, St. Louis, MO, USA), and 1.5% bacteriological agar (LAB M), pH 6.9. Overnight bacterial cultures were centrifuged (12,000× g; 15 min; 4 • C) and 25 µL of each cell-free supernatant was added in each well. The plates were incubated at 30 • C for 5 days. The plates were then stained with 0.05% (w/v) Coomassie Brilliant Blue G-250 (Applichem). Presence of a clarification halo around each well was perceived as an indication of caseinolytic activity. The latter was characterized as weak, if the clarification zone extended for less than 0.5 cm around the wells, and strong if the clarification zone extended for more than 0.5 cm. For the assessment of the lipolytic capacity, wells were aseptically punched in freshly prepared lawns of a medium consisting of 0.5% peptone (LAB M), 0.3% meat extract (LAB M), 0.5% lecithin (Serva, Heidelberg, Germany), 1% tributyrin (Merck, Darmstadt, Germany), 3% meat extract (Merch), and 1.5% bacteriological agar (LAB M), with the addition of 2.5 mM CaCl 2 (Applichem) and 5 mM MgSO 4 (Applichem) [22]. Overnight bacterial cultures were centrifuged (12,000× g; 15 min; 4 • C) and 25 µL of each cell-free supernatant was added in each well. The plates were incubated at 30 • C for 10 days. Presence of a clarification halo around each well was perceived as an indication of lipolytic activity. As in the case of proteolytic activity, it was characterized as weak, if the clarification zone extended for less than 0.5 cm around the wells, and strong if the clarification zone extended for more than 0.5 cm.

Production of Exopolysaccharides (EPS)
The capacity of all strains to produce exopolysaccharides was assessed according to Smitinont et al. [23]. In brief, overnight bacterial cultures were streaked on the surface of MRS agar (LAB M) containing 2% glucose, galactose, or lactose (Applichem). Occurrence of slimy phenotype after incubation at 37 • C for enterococci and 30 • C for the rest LAB for 3 days was indicative of EPS production.

Production of Antimicrobial Compounds
Production of antibacterial compounds against a mixture of five strains from each of Listeria monocytogenes, Salmonella enterica, and Escherichia coli O157:H7 pathogen was assessed through agar well diffusion assay. More accurately, wells were aseptically punched in freshly prepared BHI agar (LAB M) lawns inoculated with the mixture of the pathogens. Overnight bacterial cultures of all strains were centrifuged (12,000× g; 15 min; 4 • C), each cell-free supernatant was neutralized and treated with catalase (Sigma-Aldrich) and 25 µL of each was added in each well. The plates were incubated at 37 • C for 24 h. Growth inhibition of the pathogenic bacteria strain mixture used as indicator around the wells was indicative of the presence of antibacterial compounds in the respective cell-free supernatant. The protein nature of the antibacterial compounds was verified by treatment with proteinase (Sigma-Aldrich) and trypsin (Sigma-Aldrich), according to Syrokou et al. [24].

Biogenic Amine Production
The capacity of all bacterial strains to decarboxylate lysine (Serva), tyrosine (Serva), ornithine (Applichem), and histidine (Applichem) was assessed by using the substrate described by Bover-Cid and Holzapfel [25]. Overnight bacterial cultures were used to inoculate the aforementioned substrate supplemented with each amino acid. Incubation took place at 37 • C for enterococci and 30 • C for the rest of the LAB for 48 h; occurrence of color change around the colonies was indicative of the respective amino acid decarboxylation.

Hemolytic Activity
Hemolytic activity was assessed with the use of Columbia blood agar (Oxoid, Hampshire, UK) supplemented with 5% defibrinized horse blood (TCS Biosciences, Buckingham, UK). Overnight bacterial cultures of all strains were used to inoculate the aforementioned substrate. Examination of the plates for α-hemolysis (green discoloration around the colonies), β-hemolysis (clarification around the colonies), or γ-hemolysis (no discoloration or clarification) took place after incubation at 37 • C for enterococci and 30 • C for the rest LAB, for 48 h.

Antibiotic Susceptibility
The resistance of all strains to eight antibiotics, namely ampicillin (Applichem), bacitracin (Sigma), chloramphenicol (Applichem), erythromycin (Sigma), novobiocin (Sigma), streptomycin (Applichem), tetracycline (Applichem), and vancomycin (Applichem) was estimated using the Kirby-Bauer disk diffusion susceptibility test [26]. In brief, 6-mm filter paper disks were impregnated with known concentration of the aforementioned antibiotics (0, 1, 2, 4, 8, 16, 32, 64, and 128 mg/mL) and placed on the surface of an MRS agar plate, on the surface of which 100 µL of 10 7 colony forming units per mL (CFU/mL) dilution of each strain included in the present study, was spread. Growth of the strains around the disks was indicative of their resistance to the respective concentration.

Gelatinase Activity
Gelatinase activity was assessed by the nutrient gelatin stab method and the nutrient gelatin plate method according to Cruz and Torres [27]. In brief, overnight bacterial cultures of all strains were used to inoculate both substrates. Incubation took place at 25 • C for 1 week in the first case and at 35 • C for 24 h in the latter. After incubation, the substrates were examined for liquefaction in the first case and clear zones around the colonies in the latter.

Acid and Bile Salt Tolerance
Acid and bile salt tolerance of all strains was assessed according to Paramithiotis et al. [28]. More accurately, overnight bacterial cultures were washed twice with sterile saline. Acid tolerance was assessed after resuspension of the biomass in phosphate-buffered saline (PBS) buffer adjusted to pH 1, 2, or 3 with HCl and incubation at 37 • C for 3 h. Bile salt tolerance was assessed after resuspension of the biomass in PBS buffer, pH 8, containing 0.5, 1.0, or 2.0% (w/v) bile bovine (Sigma) and incubation at 37 • C for 4 h. In both cases, the viable cell counts enumerated immediately upon inoculation and after the incubation period were used to estimate acid and bile salt tolerance. If the surviving population was below the enumeration limit (10 CFU/mL), the latter was used for calculating purposes.

Bile Salt Hydrolase Activity
Bile salt hydrolase activity was performed according to Paramithiotis et al. [28]. In brief, overnight bacterial cultures of all strains were streaked on MRS agar, supplemented with 0.5% (w/v) bile bovine, and incubated at 37 • C for 48 h. Bile bovine hydrolysis was indicated by precipitation zones around the colonies and altered colony morphology.

Milk Acidification and Coagulation
In Table 1, skim milk acidification (in ∆pH) and coagulation capacity of representative strains of the LAB included in the present study, is presented. In general, skim milk acidification and coagulation seemed to be a strain-rather than a species-dependent property. Strains with variable acidification and coagulation capacity were included in all species. Enterococcus faecalis was an exception, since coagulation was achieved after 24 h, by all strains. The strains that achieved faster acidification were Lc. lactis DRD 2658 and P. pentosaceus DRD 2657. In both cases, coagulation of skim milk was observed after 6 h. Fast acidification and coagulation (after 6 h) was also achieved by Lc. lactis DRD 2692, Lp. pentosus DRD 2662, P. pentosaceus DRD 2655, and Lv. brevis DRD 2664.  1 strain; DRD 2722). Since the caseinolytic activity was weak, no further assessment with SDS-PAGE took place. As far as lipolytic activity was concerned, this was not observed by the strains under study.

Production of Exopolysaccharides (EPS)
None of the strains under study presented the slimy phenotype when glucose, galactose, or lactose were supplied as carbon sources.

Biogenic Amine Production
None of the strains under study presented with the necessary color change that was indicative of decarboxylation of lysine, tyrosine, ornithine, or histidine that were provided in the growth medium.

Hemolytic Activity
None of the strains under study presented hemolytic activity and thus can be classified as non-hemolytic or γ-hemolytic.

Antibiotic Susceptibility
The antibiotic susceptibility of the strains under study is presented in Figure 1A-J. A variable degree of susceptibility to the antibiotics examined was observed, with the exception of streptomycin and bacitracin, the concentration of which required to inhibit growth of the majority of the strains under study was above 128 mg/mL. Tetracycline and erythromycin were also relatively ineffective since the concentration required to inhibit growth of the majority of the strains under study was equal or above 64 mg/mL, with the exception of Lc. lactis and Lp. pentosus. In the first case, growth of the majority of the strains was inhibited by 32 mg/mL tetracycline, whereas growth of half of the Lp. pentosus was inhibited by 32 mg/mL and the other half by more than 128 mg/mL erythromycin. Regarding novobiocin, growth of the majority of the strains belonging to E. faecalis, Lp. pentosus, Lt. curvatus, E. faecium, P. pentosaceus, and Lv. brevis was inhibited by concentration equal or above 64 mg/mL. On the contrary, the strains belonging to Lc. lactis, Ln. mesenteroides, Lp. plantarum, and W. paramesenteroides seemed to be more susceptible to novobiocin since a concentration equal or less than 32 mg/mL was required to inhibit growth of the majority of them. As far as vancomycin was concerned, growth of the majority of the strains belonging to E. faecalis, Lt. curvatus, P. pentosaceus, Ln. mesenteroides, Lp. plantarum, Lv. brevis, and W. paramesenteroides was inhibited by concentration equal or above 64 mg/mL. On the contrary, half of the Lp. pentosus strains were inhibited by 8 mg/mL and the other half by more than 128 mg/mL, whereas growth of the majority of Lc. lactis and E. faecium strains was inhibited by concentration equal or less than 16 and 32 mg/mL, respectively. Finally, ampicillin seemed to be the most effective among the antibiotics examined. Growth of the majority of Lt. curvatus and Lv. brevis strains was inhibited by concentration equal or less than 32 mg/mL, Lp. pentosus, P. pentosaceus and Ln. mesenteroides equal or less than 16 mg/mL, E. faecalis, E. faecium, Lp. plantarum and W. paramesenteroides equal or less than 8 mb/mL and Lc. lactis equal or less than 4 mg/mL.

Gelatinase Activity
None of the strains under study presented gelatinase activity, by any of the methods employed.

Acid and Bile Salt Tolerance
In Figure S1, the microbial survival after exposure of the strains to PBS buffer adjusted at pH values 2 and 3 and incubation at 37 • C for 3 h, is presented. A box and whisker plot, summarizing the survival of the strains, at species level, is shown in Figure 2. After exposure to PBS buffer adjusted at pH value 1 and incubation at 37 • C for 3 h, no surviving population was enumerated; therefore, these results were not included in Figures 2 and S1. Exposure to PBS buffer adjusted at pH values 2 and 3 and incubation at 37 • C for 3 h resulted in reduction of the population of all strains. In the former case, the reduction of the population was larger (p < 0.05) in all strains, with the exception of the ones in which no surviving population was detected in both pH values.
The median values of the microbial survival after exposure to pH 2 ranged from −6.56 log N/N 0 in Lv. brevis to −4.38 log N/N 0 in Lt. curvatus and after exposure to pH 3 from −4.24 log N/N 0 in Ln. mesenteroides to −0.81 log N/N 0 in Lc. lactis.
Bile tolerance of the strains included in the present study is presented in Figure S2. A box and whisker plot, summarizing the survival of the strains after incubation, at species level, is shown in Figure 3. The lowest survival at 0.5 and 1.0% referred to P. pentosaceus strain DRD 2761, while at 2% bile bovine concentration to Ln. mesenteroides strain DRD 2604. Growth was also observed in many cases; it reached 0.66, 1.13, and 1.10 log N/N 0 at 0.5, 1.0, and 2.0% bile bovine, respectively. In the first case, such a growth was observed in the case of Lv. brevis strain DRD 2632, in the second case in E. faecium strain DRD 2723 and in the third case in E. faecium strain 2667. The median values of the microbial survival ranged from −0.17 to 0.03, from −0.22 to 0.23, and from −0.37 to 0.06 log N/N 0 at 0.5, 1.0 and 2.0% bile bovine, respectively. The lowest median values were observed for Lp. plantarum at 0.5% bile bovine, for Lc. lactis at 1.0%, and for Lp. pentosus at 2.0% bile bovine. On the other hand, the highest median values were observed in all cases for E. faecalis.

Bile Salt Hydrolase Activity
None of the strains under study presented with the necessary precipitation zones around the colonies or the altered colony morphology, when grown on the surface of MRS agar, which was indicative of bile salt hydrolase activity. Bile tolerance of the strains included in the present study is presented in Figure S2. A box and whisker plot, summarizing the survival of the strains after incubation, at

Discussion
In recent years, consumer awareness regarding food safety and nutritional status is increasing. In addition, the demand for functional food, i.e., food that may provide more than just nutritional elements [29] is also increasing. Dairy industry is addressing these demands through the selection of microorganisms that may serve as starter or adjunct cultures that apart from the product-dependent sensorial and technological characteristics, may provide with additional benefits. In the present study, the technological, safety and probiotic properties of lactic acid bacteria isolated from sheep milk and during ripening of Feta and Kefalograviera cheeses are assessed. To our knowledge, there is a knowledge gap, especially regarding the properties of Kefalograviera wild microbiota.
Milk acidification and coagulation, caseinolytic and lipolytic activity, as well as production of exopolysaccharides are among the critical technological properties for LAB that are evaluated as starter or adjunct cultures in cheesemaking. The rapid pH reduction contributes to safety, shaping of appropriate environment for lipolytic, and proteolytic enzymes and it is essential for milk coagulation [30,31]. Caseinolytic and lipolytic activity as well as EPS production play a major role in aroma, flavor, and texture development of final products [32,33].
Rapid milk acidification is one of the most important properties of LAB that are used as starter cultures, but it is not desirable in the adjunct cultures, since it could negatively affect the quality of the final products [34]. Therefore, strains with poor acidifying activity but with interesting technological or functional properties, can be used as adjunct cultures and combined with more acidifying strains, in mixed starter cultures [30,35]. According to Beresford et al. [36], the starter bacteria should reduce the pH value of milk to 5.3 within 6 h at 30-37 • C. In the present study, the faster acidifying strains were Lc. lactis DRD 2658 and P. pentosaceus DRD 2657, which reduced the pH value of skim milk, within 6 h, to 5.97 and 5.92, respectively, failing to meet this criterion. The results obtained in the present study highlighted the strain-dependent character of this property. Indeed, in all species, faster and slower acidifying strains were included, concurring with the results presented in the literature [30][31][32][33][34][37][38][39]. The only exception was E. faecalis, in which a uniformity was observed in the acidifying capacity, which may be assigned to the small number of strains assessed. Milk coagulation is encouraged by acid production from LAB strains, as well as coagulant enzymes activity [36,38]. This property is also strain-dependent and strains able to coagulate skim milk within 24 h were detected in all species included in the present study. Similar results were also presented by Fusieger et al. [37], Teixeira et al. [39], and Picon et al. [40].
Proteolytic LAB strains, through the activity of cell wall-associated proteinases and intracellular peptidases, degrade casein leading to the development of aroma, texture, and flavor [37,41,42]. Therefore, caseinolysis is of great importance for starter or adjunct LAB. Nevertheless, extensive caseinolysis could result in organoleptic and structural defects [37,43]. In the present study, weak caseinolytic strains were detected in all species assessed. Interestingly, the majority of them were isolated from the same source, namely Feta cheese in early ripening stage [20]. Caseinolysis does not seem to be a property shared among the majority of dairy LAB; in addition, the extend of proteolytic activity is strain-dependent [30,32,34,[37][38][39][44][45][46][47][48].
Exopolysaccharides production by LAB used in cheese-making is important from both a technological and functional perspective. More specifically, these molecules may function as thickeners and stabilizers, improving texture and structure of products. Moreover, EPS may have a positive effect on consumer health [48,50,52]. In the present study, no EPS producing strain was detected, concurring with Camara et al. [46], who characterized this property as rare among dairy LAB. However, EPS producing strains have been isolated from milk and dairy products [35,48,50,52,53].
Regarding the safety traits that were assessed in the present study, none of the isolates exhibited amino acid decarboxylation, hemolytic and gelatinase activity, while their response to antibiotic varied widely; finally, 14 out of 189 strains were found to produce antibacterial compounds able to inhibit growth of the L. monocytogenes strain mixture that was used as indicator.
LAB may decarboxylate amino acids and produce biogenic amines during the production and storage of fermented products. The consumption of biogenic amines may lead to series of adverse health effects [18]. Biogenic amines production seems to be a rather common property among dairy LAB [40,46,48,50,53]. However, in the present study no such strains were detected. This could be attributed to the method used for their detection and should be verified by analytical techniques with improved limit of detection, such as High-Performance Liquid Chromatography (HPLC).
Hemolytic and gelatinase activities are also very important; strains possessing such activities constitute a hazard to consumers health [45,52,54,55]. In general, no such activities have been reported for dairy LAB with the exception of enterococci, in which occurrence of hemolytic and gelatinolytic strains has been reported [33,48,52,56,57]. No such strains were detected in the present study.
The bacterial antibiotic resistance is a matter of great concern for the food industry. The phenotypical assessment of LAB antibiotic susceptibility is the first step to evaluate the presence of antibiotic resistance genes [32]. The majority of the strains included in the present study, exhibited resistance to the antibiotics examined. According to the cut-off values recommended by EFSA [58], all isolates, with only a few exceptions, seemed to be resistant to chloramphenicol, erythromycin, streptomycin, and tetracycline. Similar was the case of ampicillin, but more strains could be characterized as sensitive. Regarding vancomycin, all E. faecalis, E. faecium, Lc. lactis, and W. paramesenteroides strains, can be characterized as resistant. For the remaining species, no cut-off values have been proposed for vancomycin.
In recent years, the capacity of LAB as natural biopreservatives, especially in fermented foods, has been exhibited. This capacity is attributed to the production of various metabolites with antimicrobial activity, such as organic acids (lactic, acetic, propionic, and formic acids), hydrogen peroxide, diacetyl, carbon dioxide, ethanol, and bacteriocins. The latter are ribosomally synthesized proteinaceous molecules that target mainly the bacterial cytoplasmic membrane [33,39,52]. L. monocytogenes, E. coli, and Salmonella are among the most frequent foodborne pathogens and the antimicrobial potential of LAB against them has been extensively assessed. In the present study, 14 strains exhibited such activity against the L. monocytogenes strains mixture. Noticeably, ten of those strains belonged to the genetically closely related E. faecium, which explains this activity [64]. In line with present data, Albayrak and Duran [32] reported that the majority of Enterococcus spp. isolated from a ripened white cheese sample, inhibited growth of L. monocytogenes. In addition, Ispirli et al. [65] reported that 12 Enterococcus spp., isolated from Turkish white cheese, produced bacteriocins against Gram-negative bacteria, such as S. Typhimurium and E. coli. In the present study, one Lp. plantarum, one Lv. brevis, and two Ln. mesenteroides strains also exhibited antilisterial activity. Such activity from strains belonging to these species is commonly reported in literature [33,52,56].
Regarding the probiotic potential of LAB isolates, acid and bile tolerance, as well as bile salt hydrolase (BSH) activity were tested since resistance to gastric acidity and bile salts is essential for the survival of microorganisms in the gastrointestinal tract (GIT). The capacity of LAB to survive acidic conditions is essential, since in the stomach, they may confront pH values less than 3.0 [47]. In the present study, none of the 189 LAB strains assessed, managed to survive to exposure at pH value 1. On the contrary, 25.9 and 88.9% of the strains survived after exposure at pH values 2 and 3, respectively. The population reduction was greater in the first case. Similarly, tolerance to exposure at pH 3.0, but population reduction greater than 2 log CFU/mL of 116 LAB strains isolated from Serpa cheese was reported by Ruiz-Moyano et al. [63]. Zoumpopoulou et al. [56] reported that 32 isolates belong to 14 different species, isolated from Kalathaki and Melichloro cheeses presented with population reduction larger than 2 log CFU/mL after exposure to pH 2.5 for 2 h. Survival rates ranging from 70-90% were reported by Islam et al. [48] when 11 strains belonging to different species were exposed to pH 2 for 3 h.
The transit time of LAB in human GIT is estimated from 4 to 6h, and the bile concentration from 0.3 to 0.5% w/v, depending on the person [32,49]. In the present study, the maximum population reduction observed was 1.5 log CFU/mL, in any bile salt concentration studied. In some cases, growth enhancement was observed. Similarly, Albayrak and Duran [32] observed enhancement of Enterococcus spp. growth, the population increase of which, in some cases exceeded 1 log CFU/mL. In general, good survival of dairy LAB is reported [47,56], with just a few exceptions [34].
Bile salts are antimicrobial agents that interact with the cell membrane, proteins, DNA, and chelate iron and calcium. The deconjugation of bile salts from BSH enzyme contribute to bacterial survival in the GIT and has an important role in intestinal homeostasis and in the reduction of serum cholesterol levels. Therefore, BSH activity is a desirable property. In the present study, no such strains were detected. This concurs with the conclusion that such a property is rather uncommon among dairy LAB [38,66].
Overviewing the results obtained in the present study, many strains, belonging to different species presented very interesting properties. However, these properties were scattered among the strains and no strain would present more than one. More specifically, the strains able to produce antilisterial compounds were poor acidifiers and among the least susceptible to antibiotics. Therefore, selection could only be based on fast acidification capacity among the strains that were more susceptible to antibiotics; these strains were Lc. lactis DRD 2692 and P. pentosaceus DRD 2655.

Conclusions
The properties assessed in the present study are strain-rather than species-dependent. It was further supported that proteolysis, lipolysis, EPS and bacteriocin production, amino acid decarboxylation, and hemolytic and gelatinase activity are rather uncommon traits among dairy LAB. However, resistance to antibiotics seems very common. Many strains, belonging to different species, have presented very interesting properties, which deserve further assessment. included in the present study to PBS buffer adjusted to pH 2 (blue) or 3 (green) with HCl and incubation at 37 • C for 3 h. N: microbial population after incubation; N 0 : initial microbial population. Figure S2: Microbial survival (in log N/N 0 ) after exposure of E. faecalis (A), E. faecium (B), Lc. lactis (C), Ln. mesenteroides (D), Lp. pentosus (E), Lp. plantarum (F), Lt. curvatus (G), Lv. brevis (H), P. pentosaceus (I) and W. paramesenteroides (J) strains included in the present study to PBS buffer, pH 8, containing 0.5% (blue), 1.0% (green) or 2.0% (yellow) (w/v) bile bovine and incubation at 37 • C for 4 h. N: microbial population after incubation; N 0 : initial microbial population. Table S1: Origin of the strains included in the present study.