Specimens of different types of fermented food products (sauerkraut, fermented green tomatoes, fermented cucumbers, kefir, white cheesе, and Izvara (curdled milk)) were prepared according to traditional Bulgarian recipes and used as sources of LAB isolation. Among these products were homemade and artisanal markets. The samples were collected in sterile containers from different regions in Bulgaria. The potential LAB in the samples were enriched via the cultivation of 1 g (mL) of each product in de Man, Rogosa, and Sharpe (MRS) broth (Merck, Darmstadt, Germany), separately. The resulting enriched cultures were diluted ten-fold, and aliquots of 100 µL were spread onto the MRS agar surface. The Petri dishes were incubated at 30–37 °C in an anaerobic jar (utilizing the BBL GasPack anaerobic system) for 48 hours. Colonies with different morphological characteristics were selected and re-streaked to obtain pure bacterial cultures. Preliminary selection of potential LAB isolates was performed according to results obtained for Gram-staining (Gram-staining Kit, Sigma Aldrich, St. Louis, MO, USA) to determine oxidase (Oxidase strips, Sigma-Aldrich) and catalase activities [26].

The total DNA of all isolates was extracted from overnight MRS broth cultures using a Tissue and Bacterial DNA Purification kit (EURx Ltd., Gdansk, Poland), following the manufacturer’s instructions. The DNA Purification kit purified the isolated DNA (PCR/DNA Clean-up, GeneMatrix, EURx Ltd., Gdansk, Poland). The quantity of purified DNA was measured by BioDrop µLITE+ (100 ng/µL $\pm$ 20 ng/µL). The extracted DNA samples were stored at –20 °C and used for all PCR analyses in this study. PCR amplification of the entire length of 16S rDNA was performed using universal primers 27F and 1492R [27] and ready-to-go PCR mixtures iProof HF MasterMix (BioRad, Laboratories Inc., Hercules, CA, USA). The reactions were carried out using a PCR thermal cycler (Techne® Prime, VWR International, Radnor, PE, USA) following the protocol: 94 °C, 5 min; 35 cycles of 94 °C, 45 s; 56 °C, 45 s; 72 °C, 45 s. The final elongation steps were at 72 °C (10 min). The PCR amplification products were analyzed by 1.5% gel agarose electrophoresis. Purified PCR products were sequenced by Macrogen Europe (Meibergdreef 57 1105 BA, Amsterdam, The Netherlands). The obtained sequences were subjected to comparative analyses using BLASTN (NCBI).

The antimicrobial activity of the isolated LAB was tested against the following pathogenic test microorganisms: Listeria innocua F 4078, Staphylococcus aureus ATCC 6538, Escherichia coli NBIMCC 3548, Salmonella enterica subsp. enterica serotype Enteritidis NBIMCC 8691, and Candida albicans ATCC 10231 using the agar–well diffusion method [28, 29]. The pathogenic bacteria and yeast were cultivated overnight on Brain Heart Infusion (BHI) broth (Merck KGaA, Darmstadt, Germany) and Yeast Peptone Dextrose (YPD) broth (Merck KGaA, Darmstadt, Germany), respectively, to obtain log cultures. The LAB isolates were cultivated in MRS broth at 37 °C for 24 hours, and after that, cell-free supernatants (CFS) were obtained by centrifugation (at 6000 $\times$g and 5 °C for 15 min). The experiments were conducted using three CFS variants—obtained by following the methodology described before [30, 31]: (1) CFS, (2) neutralized cell-free supernatants (NtCFS), and (3) heat-treated (100 °C for 10 min) neutralized cell-free supernatant (htNtCFS). Then, these supernatant variants were introduced into the prepared agar wells for diffusion. The plates were incubated at 37 °C for 24 hours. The presence of clear zones surrounding the wells confirmed antibacterial activity, and the diameters of the inhibition zones were measured and expressed in millimeters. All experiments were performed in triplicate, and mean values were calculated.

The ability of the LAB isolates to produce H${}_2$O${}_2$ was examined by the agar plate method, described by Maldonado et al. [32]. Horseradish peroxidase (Sigma Chemical Co, St. Louise, MO, USA) and tetramethyl–benzidine (TMB) (Sigma Chemical Co, St. Louise, MO, USA) were added to MRS agar medium. The LAB isolates were surface spot-inoculated in the TMB–MRS plates and anaerobically cultivated at 37 °C in an anaerostat for 48 hours. The results were interpreted after storing at room temperature outside the anaerostat for 2 hours, according to the intensity of the blue color of the LAB cultures in the plates as negative (–), weakly positive ($\pm$), moderately positive (+), and strongly positive (++).

Autoaggregation (AA) and co-aggregation (CA) assays were carried out according to Tuo et al. [33]. The LAB isolates were grown on MRS broth at 37 °C overnight. For AA, the bacterial cells were harvested by centrifugation at 10,000 $\times$g and 4 °C for 10 min, washed twice with phosphate-buffered saline (PBS) (pH 7.2), and then resuspended in PBS until the required cell concentration (1 $\times$ 10${}^8$ CFU/mL, optical density, OD${}_{\text{600nm}}$ = 0.25) was obtained. The AA degree was measured at 600 nm (A${}_{\text{600 nm}}$) at three time points: 0, 4, and 24 hours. The autoaggregation percentage was calculated using the formula: [1 – A${}_{\text{t}}$/A${}_0$] $\times$ 100, where A${}_{\text{t}}$ represents the absorbance at time t = 4 or 24 hours, and A${}_0$ the absorbance at t = 0. The CA was carried out with mixtures of LAB and suspensions of the pathogenic microorganism C. albicans (1 $\times$ 10${}^8$ CFU/mL) in equal volumes (5 mL), incubated at 37 °C without agitation. The co-aggregation percentage was measured at 600 nm (A${}_{\text{600 nm}}$) at three time points: 0, 4, and 24 hours and calculated according to the following formula [(A${}_0$ – A${}_{\text{t}}$)/A${}_0$] $\times$ 100, where A${}_0$ represents the absorbance of the mix immediately after mixing, and A${}_{\text{t}}$ represents the absorbance of the mix at 4 or 24 hours [2].

The degree of hydrophobicity on the surface of bacteria was evaluated by measuring their adherence to hexane using the modified protocol of Collado et al. [34]. LAB isolates were cultivated in MRS at 37 °C for 24 hours, washed twice in PBS, and resuspended in 3 mL of PBS buffer to achieve approximately 1 $\times$ 10${}^8$ CFU/mL (OD${}_{\text{600 nm}}$ = 0.25). The absorbance of each suspension was measured at 600 nm (A${}_0$). Three milliliters of hexane were added to each cell suspension to form two-phase systems, and after incubating at room temperature for 10 min, the tubes were mixed by vortexing for 3 min. After incubating at room temperature (approximately 23 °C) for 1 hour (t = 1), the aqueous phase was carefully removed, and its absorbance was measured at 600 nm (A). The percentage of cell surface hydrophobicity (H, %) was expressed as adhesion percentage according to the following formula: H (%) = [(A${}_0$ – A)/A${}_0$] $\times$ 100, where A represents the absorbance at t = 1 and A${}_0$, the absorbance at t = 0 [34].

With a few minor adjustments, an evaluation of the capacity of each LAB isolate to produce a biofilm was conducted in accordance with earlier protocols by Bujnakova et al. [35] and Gómez et al. [36]. The isolated LAB strains were cultivated in 5 mL MRS broth and incubated at 37 °C for 24 hours. The resulting cultures were centrifuged (10,000 $\times$g, 10 min), and the pellets were resuspended in PBS (McFarland 1.0, corresponding to 3 $\times$ 10${}^8$ CFU/mL). A mixture of 180 µL MRS and 20 µL cell suspension was added dropwise into 96-well polystyrene microtitre plates (Nunc, Roskilde, Sjælland) and incubated at 37 °C for 24 hours. The analyses were conducted separately in triplicate for each LAB strain. The wells in which only MRS broth was added were used as the negative control. After incubation, the wells were carefully rinsed three times with 2 mL of sterile deionized water. Two milliliters of methanol (Romyl, Leics, UK) were used to fix the attached bacteria for 15 min. The methanol was extracted, and the plates were left to dry at room temperature. Subsequently, the cells in the wells were stained with 2 mL of a 0.1% (v/v) crystal violet solution for 5 min. After that, the excess staining was eliminated by running a gentle stream of tap water over the plate. The addition of 50 µL of 33% (v/v) glacial acetic acid removed the stain from the adhering cells. The OD of each well was measured at 595 nm using a plate reader ELIZA (Multiskan EX, Cat. no. 51118170 (200–240 V), Thermo Fisher Scientific, Vantaa, Finland). The ODc was defined as the mean OD value of the negative control. Based on the OD, strains can be classified as non-biofilm producers (OD $\le$ OD${}_{\text{C}}$), weak biofilm producers (OD${}_{\text{C}}$ OD $\le$ 2 $\times$ OD${}_{\text{C}}$), moderate (2 $\times$ OD${}_{\text{C}}$ OD $\le$ 4 $\times$ OD${}_{\text{C}}$), or strong biofilm producers (4 $\times$ OD${}_{\text{C}}$ OD) [37].

The hemolytic activity was evaluated following the methodology reported by Carrillo et al. [38]. Pure LAB cultures were surface spot inoculated on blood agar plates supplemented with 5% (v/v) horse blood and incubated at a temperature of 37 °C for a duration of 24 to 48 hours. The hemolytic activity of the isolated LAB strains was evaluated after the incubation period. The strains were classified according to the changes in blood agar medium around the spot culture as follows: (1) green zones around spot culture ($\alpha$-hemolysis), (2) clear zones around spot culture ($\beta$-hemolysis), and (3) no zones around spot culture ($\gamma$-hemolysis). Only strains with $\gamma$-hemolysis were considered safe [39].

The susceptibility of all isolates to different antibiotics was tested according to the agar disc diffusion procedure described before by Bauer et al. [40] and Yasmin et al. [1], with modifications. Various antibiotics (AB), belonging to different groups were used (ampicillin/sulbactam (A/S)—10 µg/disc, vancomycin (VA)—5 µg/disc, chloramphenicol (C)—30 µg/disc, erythromycin (E)—15 µg/disc, tetracycline (T)—10 µg/disc, ciprofloxacin (CP)—10 µg/disc, cephalothin (CF)—10 µg/disc, gentamicin (G)—10 µg/disc, streptomycin (S) — 10 µg/disc, and neomycin (N)—5 µg/disc). A total of 1 mL of each strain (approximately 1 $\times$ 10${}^8$ CFU/mL) was inoculated and mixed with 20 mL melted MRS agar and left to solidify at room temperature. After that, discs containing different antibiotics were placed on the agar surfaces and stored at 4 °C for 2 hours for AB diffusion. Then, the Petri dishes were cultivated at 37 °C in an anaerobic environment (utilizing the BBL GasPack anaerobic system) for 24 hours. The diameters of the inhibition zones surrounding the antibiotic discs were measured in millimeters and categorized as 8 mm—resistant (R), between 8 and 10 mm—moderately susceptible (MS), and $>$10 mm—strongly susceptible (SS) [1].

The genes related to antibiotic resistance to gentamicin (aac(6′)-aph(2′′)), chloramphenicol (cat), tetracycline (tet(M)), $\beta$-lactamase (blaZ), a macrolide (mefA), kanamycin (aph (3′)-III), erythromycin (ermA, ermB), ciprofloxacin (gyrA), and vancomycin (vanA) were examined by PCR using specific primers, according to the protocol of Guo et al. [41] and Liu et al. [42]. The PCR products were visualized using 1.5% agarose gel electrophoresis.

A total of 45 potential LAB strains (Table 1) were obtained from various traditional Bulgarian fermented foods (sauerkraut, fermented green tomatoes, fermented cucumbers, kefir, white cheese, and Izvara (curdled milk)). Of these, 25 strains were from fermented vegetable products, and 20 were from fermented milk products. Phenotypic characterization of the isolates showed that all strains formed white to creamy-colored colonies with smooth or jagged edges, predominantly with a convex profile and a uniform to slightly granular structure. Micromorphologically, the cells were rod-shaped with different lengths. All isolates were Gram-positive bacteria, catalase, and oxidase-negative. These phenotypic data are typical characteristics of the LAB group.

All 45 isolated strains from traditional homemade and artisanal Bulgarian fermented products were subjected to molecular identification. The genetic material of all potential LAB isolates was extracted, and the 16S rDNA gene was amplified using universal primers 27F/1492R [27]. The obtained PCR products (approximately 1500 bp) were purified and sequenced by Macrogen in The Netherlands. The results were processed and compared with the NCBI database. All LAB isolates with varying similarity percentages were identified at genus and species levels. Among them, representatives of three genera, according to the new reclassification within the family Lactobacillaceae proposed by Zheng et al. [43], were found (Fig. 1A). The most prevalent genera were found to be Lactiplantibacillus (47%) and Levilactobacillus (40%). Less represented was the genus Lactobacillus (13%). The LAB diversity established after species identification revealed the predominance of five bacterial species: L. plantarum, Levilactobacillus koreensis, L. brevis, L. helveticus, and Levilactobacillus yonginensis (Fig. 1B). L. plantarum was the most prominent (47%) species, as it was found in most tested products. It is interesting to note that this species represents 83% and 80% of all isolated LAB from the sauerkraut products: artisanal market and homemade, respectively. Similar results for the distribution of L. plantarum in traditional fermented foods have also been reported globally [3, 44]. Other authors have reported the presence of Leuconostoc, Weissela, and lactobacilli in sauerkraut fermentation, expressing the opinion that the composition of LAB microflora depends on the cultivar of the cabbage used in the fermentation process [45]. The only product where L. plantarum was not found was fermented green tomato (FGT), where the dominant LAB species was L. koreensis. Interestingly, L. koreensis was found only in two products (fermented green tomatoes, FGT, and fermented cucumbers, FC). Surprisingly, only one strain of L. yonginensis was isolated solely from artisanal sauerkraut. Moreover, both species (L. koreensis and L. yonginensis) are more commonly found in Asian fermented foods such as kimchi [46]. To our knowledge, this study is the first report of L. koreensis and L. yonginensis isolated from fermented vegetable foods in Bulgaria. L. helveticus was the predominant species in the white cheese and Izvara products. It seems that L. brevis is an irrevocable part of fermentative microflora, as it was found in 57% of the tested products in this study (Table 1).

Fig. 1.

Diversity of LAB in the studied foods. (A) Genus diversity of LAB strains isolated from traditional homemade and artisanal Bulgarian fermented products. (B) Species diversity of LAB isolated from traditional homemade and artisanal Bulgarian fermented products. LAB, lactic acid bacteria.

The fermentation process, which turns sugars into organic acids (lactic and acetic acids), resulting in the formation of acidic environments, is mostly responsible for the preservation behavior of LAB. However, these bacteria can generate and release inhibitory compounds in addition to lactic and acetic acids. Such substances have been reported to have antagonistic effects on various microorganisms. They include a variety of less well-defined or entirely unknown inhibitory substances, such as hydrogen peroxide, diacetyl, ammonia, ethanol, bacteriocins, antibiotics, etc. They are produced in much smaller amounts than lactic and acetic acids. Several of these compounds have demonstrated antagonistic effects against a variety of harmful bacteria found in food, such as Bacillus cereus, Clostridium botulinum, Clostridium perfringens, Listeria monocytogenes, and S. aureus [47, 48].

Our LAB isolates were evaluated for their antimicrobial potential and ability to produce hydrogen peroxide. We found that only 11 strains showed moderate ability to produce H${}_2$O${}_2$ (S2, S3, S4, S5, S14, S16, FGT2, FGT3, FGT4, FC2, FC3). The remaining isolates were negative or weak producers. The antimicrobial potential of the isolated lactic acid bacteria was studied against two Gram-positive species (L. innocua F, S. aureus), two Gram-negative species (Sal. Enteritidis, E. coli), and one yeast species (C. albicans). Antibacterial activity was evaluated using the agar well diffusion method. The size of the inhibition zones was expressed in mm. The inhibitory properties of the supernatants separated after centrifugation were studied in three variants: 1/CFS, 2/NtCFS, and 3/htNtCFS.

The obtained results showed that antibacterial activity was observed only for CFS (Table 2). No inhibitory activity was detected for NtCFS and htNtCFS. The lack of such activity means the isolated LAB does not produce bacteriocins or bacteriocin-like substances. The observed inhibition activity of CFS against some of the test microorganisms is probably due to the production of lactic acid, hydrogen peroxide, or other substances with an inhibitory effect. Different percentages of the strains of L. plantarum showed apparent antagonistic effects against E. coli and Sal. Enteritidis (57% and 19%, respectively). All these strains were isolated from artisanal markets and homemade sauerkraut. Very weak inhibitory activity (inhibition zones 1–2 mm) against Listeria innocua F and S. aureus was detected in 24% and 29% CFS of all L. plantarum isolates, respectively. Similar weak, barely detectable activity against Sal. Enteritidis, E. coli, and S. aureus were also established for two strains of L. koreensis (FGT2 and FC2). Very weak antimicrobial activity by CFS, obtained from lactobacilli isolated from sauerkraut, was previously reported against Listeria monocytogenes [45]. Only one of our strains L. brevis (S15—isolated from artisanal sauerkraut) showed activity against E. coli (Table 2). None of our isolates showed inhibitory properties against C. albicans. According to our results and those reported by other authors, we can conclude that no correlation between antibacterial activity and H${}_2$O${}_2$ production was found [49]. However, three strains isolated from sauerkraut (S4, S5, and S14) were the exceptions since they demonstrated antibacterial activity against E. coli and are moderate producers of H${}_2$O${}_2$.

The ability to autoaggregate is an essential property of LAB to form a barrier on the GIT mucosa, which prevents pathogenic bacteria from attaching to it. Some authors claim that lactobacilli have a modest to moderate capability for autoaggregation [50]. To verify the autoaggregation potential of our strains, all isolates were tested for their ability to autoaggregate, and the results were analyzed at 4 and 24 hours. From the group of isolates identified as L. plantarum, four strains (S8, WC6, I1, and I2) showed very good autoaggregation properties after 4 hours—more than 40% of the cells autoaggregated (Fig. 2). However, the best autoaggregation ability after 4 hours was observed for L. koreensis FC4 (48%) and L. brevis S2 (44%). The lowest AA activity was observed for two strains of L. helveticus isolated from white cheese (WC4, 8%, and WC5, 11%). The percent of AA in the rest of the LAB isolates varied from approximately 15% to 25%.

Fig. 2.

Autoaggregation ability (in percentage, %) of the isolated LAB strains, measured at 4 and 24 hours.

The autoaggregation activity increased alongside the incubation time, and after 24 hours, it reached between 29 and 77%. The highest AA ability was detected for L. koreensis FGT2 (77%), L. plantarum S13 and FC1 (both 76%), L. helveticus WC2 (72%), and L. brevis K7 (71%). However, for FC1 and WC2, the AA ability detected after 4 hours was relatively low, at 18%. Interestingly, the two strains with the highest autoaggregation ability after 4 hours did not possess the maximum AA activity after 24 hours (S2: 63%; FC4: 68%). Fifteen of the isolates (S2, S3, S4, S8, S11, S16, FGT1, FC2, FC3, FC4, K3, WC6, I2, I4, and I5) showed over 60% autoaggregation after 24 hours. Recently, numerous authors have investigated the autoaggregation abilities of lactobacilli as part of their probiotic potential. There is still no single scale for determining the degree of autoaggregation. However, most authors accept 35–40% as sufficient for lactobacilli to be defined as strains with a high aggregation potential. Notably, the strains of the L. plantarum species show different degrees of autoaggregation. Ramos et al. [51] reported that one L. plantarum and one L. fermentum strain reached autoaggregation values of 61.83% and 55.61%, respectively. L. plantarum strain 557, isolated from vegetables, exhibited 34% autoaggregation [49]. Comparing these reported results with those obtained by our study, we can conclude that our L. plantarum isolates showed similar AA abilities. Seven strains of the L. fermentum species isolated from fermented Chinese products presented highly varying autoaggregation limits ranging from 0.86% to 65.15% [52]. In a study of nine strains isolated from milk and milk products (cheese, yogurt), vegetables, and the intestinal tract, Ren et al. [49] described a high rate of autoaggregation only in Lactobacillus salivarius subsp. salicinius CICC 23174 (46%), and Lactobacillus acidophilus CGMCC 1.1854 (45%), compared to the reference strain L. rhamnosus LGG (33%). Our results are similar to those for the strains isolated from milk and milk products, as three L. plantarum strains isolated from white cheese and Izvara (WC6, I1, and I2) showed high autoaggregation potential.

The co-aggregation between probiotic microorganisms and harmful microorganisms creates an unsuitable environment for developing pathogens. This reduces their growth, eliminates the pathogen, and restores the normal gastric microbiota [53]. In the present study, the pathogenic yeast species C. albicans and the studied lactic acid isolates were co-inoculated to monitor their co-aggregation.

In the studied population of LAB isolates, four strains showed the highest percentage of CA with the pathogen after 4 hours: FGT1 (70%), S2 (54%), S13 (51%), and S6 (50%). The FGT1, K7, S2, and FC2 strains reached the greatest CA after 24 hours: 95%, 89%, 72%, and 70%, respectively (Fig. 3). Thus, based on our results, we can conclude that L. koreensis FGT1 could be considered a candidate for potential probiotic application, according to this criterion. Interestingly, another group of strains, including FGT2, I6, FC4, S9, and S6, showed great CA activity at the moment of contact between the LAB and the pathogen cells (t${}_0$): 66%, 65%, 60%, 58%, and 56%, respectively. The high CA level at t${}_0$ of L. koreensis FC4 with C. albicans may be due to its high AA percentage. The rest of the strains had relatively low AA activities. The observed high CA activity of these isolates may be due to the affinity of the surface structures of the LAB with those of the pathogen. Mastromarino et al. [54] observed a high level of co-aggregation of L. salivarius and L. gasseri with Gardnerella vaginalis. In other studies, L. fermentum and L. plantarum strains efficiently co-aggregate with E.coli, Shigella flexneri, Sal. Enteritidis, Listeria monocytogenes, and S. aureus [51, 55].

Fig. 3.

Co-aggregation ability (in percentage, %) of the isolated LAB strains, measured at three time points: 0, 4, and 24 hours.

Bacterial attachment and colonization to the inner side of the GIT depend on cell surface hydrophobicity and AA. The LAB strain must possess a minimum value of 40% hydrophobicity to be called a probiotic strain [56]. Based on this criterion, 84% of our isolates could not be considered presumptive probiotic strains (Fig. 4). However, seven of the isolated LAB strains showed hydrophobicity above 40%. The great ability to adhere to hexane was detected in three LAB strains isolated from milk—fermented products (kefir and white cheese): L. helveticus WC5, 86%; L. brevis K2, 72%, and L. plantarum K6, 52%. The other four LAB strains were isolated from fermented vegetable products (sauerkraut and fermented cucumbers): L. yonginensis S16, 50%; L. plantarum S11, 46%; L. plantarum S12, 44%, and L. koreensis FC4, 43%. The last four stains, except strain S12, also showed good AA ability and were detected after 24 hours. Microbial cell adhesion is affected by a combination of electrostatic and van der Waals forces, as well as the hydrophobicity of the surfaces [57]. Hydrophobicity is important in the initial interaction between bacterial cells and mucosal or epithelial cells [58]. Lactobacilli adhesion is also connected with particular interactions involving the receptor on the bacterial surface recognizing a particular area or ligand [59]. It turns out that the ability of LAB to attach to hydrocarbons is encountered relatively rarely among LAB. For example, Ren et al. [49] investigated nine strains, of which only two strains (22%) showed relatively high levels of hydrophobicity (59% and 43%, respectively). Hoxha et al. [3] reported that none of the twelve LAB strains tested had hydrophobicity above 40%.

Fig. 4.

Cell surface hydrophobicity (in percentage, %) of the isolated LAB strains.

Bacterial biofilms are crucial in understanding how bacteria adapt to environmental stress and colonize diverse habitats. To investigate the ability of the LAB strains isolated from traditional Bulgarian fermented food to form biofilms, they were cultivated in a 96-well plate for 24 hours. Based on the obtained results, the strains were grouped as follows: Non-biofilm producers, 18% of the strains (S2, S3, S4, S6, S7, S7’, S8, S14); weak biofilm producers, 13% of the strains (S11, S12, S15, K2, K4, K8); moderate biofilm producers, 53% (S5, S9, S10, S13, K1, K6, WC1, WC5, I1, I2, I4, I5, I6, WC3, WC4, FC3, FGT1, FC1, FC4, FGT2, FGT3, FGT4, S16, FC2); strong biofilm producers, 16% of the strains (S1, K3, K5, K7, WC6, I3, WC2). Based on these results, we can conclude that the ability to form a biofilm is rather strain-dependent.

An important requirement for probiotic strains is their safety. An important initial step in developing or selecting new beneficial microorganisms is to evaluate the lack of hemolytic activity or cytotoxicity [60]. Although hemolytic activity is not inherent to lactobacilli and is relatively rarely found in enterococci, evaluating each case is necessary. Even though many LAB strains have been used for a long time due to their technological properties for food preparation or as probiotics in humans and animals, they are not directly granted GRAS status [61, 62]. This makes verifying that the isolated LAB meets this criterion mandatory. For this reason, the hemolytic activity of all isolated 45 strains was evaluated on blood agar. Hemolytic activity was not detected in any of the tested strains, as evidenced by the absence of hydrolysis zones around the colonies of the isolates. Thus, the tested strains cover one of the safety criteria, such as the absence of hemolytic activity.

Natural resistance to ciprofloxacin, bacitracin, streptomycin, cefoxitin, fusidic acid, kanamycin, nitrofurantoin, gentamicin, metronidazole, norfloxacin, sulphadiazine, teicoplanin, trimethoprim/sulphamethoxazole, and vancomycin has been widely reported for some LAB. [63]. A significantly elevated rate of spontaneous mutation to nitrofurazone (10${}^{-5}$), kanamycin, and streptomycin has been documented for various lactobacilli [64]. Some authors suggested that genes encoding antibiotic resistance can be transferred between bacteria of diverse origins [65]. Concerning the potential consequences for human health, probiotic bacteria should not be able to transfer antibiotic resistance genes [66]. This ability must be regarded as a crucial factor in selecting beneficial strains.

The antibiotic susceptibility of our LAB isolates to 10 antibiotics (ampicillin, tetracycline, erythromycin, gentamicin, streptomycin, vancomycin, neomycin, cephalothin, ciprofloxacin, and chloramphenicol) was evaluated using the agar disc diffusion method. The obtained results are summarized in Table 3, where the isolates were categorized according to Yasmin et al. [67]. Variable susceptibility to the tested antibiotics was observed even at the intraspecies level. All tested strains showed resistance to vancomycin and streptomycin (except L. plantarum S13, which showed strong sensitivity). Our results differ from those reported by Ren et al. [49], showing that all LAB isolated from fermented food were susceptible to streptomycin. A large proportion of strains showed resistance or moderate susceptibility to cephalothin (71%), neomycin (58%), and ciprofloxacin (80%). Three strains (I2, I3, and I4) showed resistance to chloramphenicol; the remaining strains were strongly sensitive. The group of strains (47%) isolated from all fermented products showed resistance or moderate susceptibility to gentamicin. The strains were strongly susceptible to three of the tested antibiotics: Ampicillin, tetracycline, and erythromycin. L. yonginensis S16 was the exception since it showed resistance to tetracycline.

Distinguishing between intrinsic (non-specific, non-transferable) and acquired resistance by the beneficial strains is also necessary [68]. Molecular techniques that directly screen bacterial strains for the presence of antibiotic resistance genes could be considered an important addition to phenotypic testing. Usually, these include PCR-based methods [69, 70] or DNA microarrays that combine various antibiotic resistance genes [71].

Phenotypic analyses showed that all LAB isolates in this study, belonging to the family Lactobacillaceae, showed resistance to one or more antibiotics. To clarify the nature of this antibiotic resistance, PCR amplification was performed with specific primers for acquired resistance genes: Gentamicin, aac(6′)-aph(2′′); chloramphenicol, cat; tetracycline, tet(M); $\beta$-lactamase, blaZ; macrolide, mefA; kanamycin, aph (3′)-III; erythromycin, ermA, ermB; ciprofloxacin, gyrA, and vancomycin, vanA. No amplification products were observed in any strains, confirming that the isolates did not possess these resistance genes. These experiments conclude that the lactic acid strains isolated from Bulgarian fermented foods possess intrinsic resistance based on different mechanisms for neutralizing the activity of some of the antibiotics studied and are not potential vectors for antibiotic resistance genes.