Three thermophilic lactic acid bacteria species from laboratory collection of Laboratory of Molecular Microbiology, Institute of Molecular Genetics and Genetic Engineering, University of Belgrade, Serbia were used in this study: Lactobacillus helveticus BGRA43 (Acc. No. LN909514) isolated from human intestinal tract (Strahinic et al., 2013), Lactobacillus fermentum BGHI14 (Acc. No. LN909513) isolated from newborn feces of a breast-fed infant (Lukic et al., 2013) and Streptococcus thermophilus BGVLJ1-44 (Acc. No. LN909512) isolated from an artisanal yogurt produced from uncooked cow’s milk.

For the cultivation of S. thermophilus BGVLJ1-44 liquid and solid M17 medium (Merck GmbH, Darmstadt, Germany), supplemented with glucose (0.5% w/v) (GM17) was used, while for the cultivation of strains L. fermentum BGHI14 and L. helveticus BGRA43 liquid and solid MRS medium (Merck, GmbH) was used. The strains were grown at 37°C, in anaerobic condition using Anaerocult A (Merck). Solid medium plates were prepared by adding agar (1.7% w/v, Torlak, Belgrade) into each medium broth. Esherichia coli ATCC25922 and Salmonella typhimurium Enteriditis 654/7E (veterinary isolate, the Collection of Scientific Institute of Veterinary Medicine, Novi Sad, Serbia) were cultivated in Luria-Bertani broth (LB) at 37°C.

The strains L. helveticus BGRA43, L. fermentum BGHI14 and S. thermophilus BGVLJ1-44 were subjected to a set of tests as follows: growth at different temperatures (15°C, 30°C and 45°C); growth in broth with 2% NaCl (w/v); L-arginine and esculine hydrolysis, the use of citrate as energy source (Kempler and Mckay, 1981); CO₂ production from glucose in reconstituted MRS broth tubes containing inverted Durham bells; the production of acetoin by the Voges-Proskauer test (Zourari et al., 1991); diacetyl production was tested only qualitatively. After overnight incubation of 11% reconstituted skimmed milk (RSM) at 37°C, 1 ml of coagulated milk was mixed with 0.1 g of creatinine (Alfa Aesar, GmbH & Co KG, Karlsuhe, Germany) and 1 ml of 30% NaOH (w/v). Diacetyl production was indicated by the formation of a red halo at the top of the tubes after 2 h of incubation at room temperature. EPS production was detected visually on MRS or GM17 agar plates, depending on the strain, as long strands when the colonies were extended with an inoculation loop (Nikolic et al., 2012). Speed curdling was determined visually by time measuring from the moment of inoculation of 11% sterile RSM with each single starter culture to the moment of curd forming at the incubation temperature of 42°C. Test in litmus milk was prepared in RSM with litmus indicator. After overnight incubation of inoculated RSM at 37°C the changes of color and structure of litmus milk were detected according to Cappuccino and Sherman (2010) as follows: A-acid production (red ring); C-curdling (milk curdled due to lactose fermentation) and R-reduction of H⁺ ions (white color). Test of surviving of the strains BGRA43, BGHI14 and BGVLJ1-44 in MRS and GM17 broth after heating at 63.5°C for 30 min was performed too. Aggregation ability of tested starter cultures was detected visually after shaking of tubes with inoculated MRS or GM17 broth that were previously incubated overnight at 37°C.

The proteolytic activity of the isolates BGRA43, BGHI14 and BGVLJ1-44 were assayed as previously described (Kojic et al., 1991). The collected fresh cells (10 mg with an approximate density of 10¹⁰ cells/ml) were resuspended in 0.1 M of sodium phosphate buffer with pH 6.8 and mixed in a 1:1 ratio with 5 mg/ml of α_s1-, β- and κ- casein respectively (Sigma, St. Louis, MO, United States) dissolved in the identical buffer. The mixtures were incubated for 3 h at 37°C. The degradation of α_s1-, β- and κ- casein was analyzed on 12.5% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). The level of α_s1-, β- and κ- casein degradation was quantified by ImageQuant software (Molecular Dynamics GmbH, Krefeld, Germany).

Survival of probiotic strains after the chemically simulated gastrointestinal tract (GIT) transit was essentially performed as described previously (Nikolic et al., 2012), with minor modifications. The strains were grown in appropriate medium for 24 h and cultures were harvested by centrifugation (10,000 × g for 10 min), washed twice with 0.85% NaCl and concentrated 10-times in reconstituted (10%) sterilized skimmed-milk (Difco, Becton Dickinson, Franklin Lakes, NJ, United States) or in saline buffer (0.85% NaCl). Afterward, bacterial suspensions were diluted 10-times with gastric juice (GJ: 125 mM NaCl, 7 mM KCl, 45 mM NaHCO₃, 0.3% Pepsine [Sigma, St. Louis, MO, United States] adjusted to pH 2 with HCl), incubated for 90 min at 37°C in aerobic conditions under shaking (≈180 rpm). Then, bacterial suspensions were centrifuged (2050 × g, 15 min), resuspended in duodenal juice (DJ: 1% bovine bile [Sigma] adjusted with 10 M NaOH to pH 8.0) and incubated for 10 min at 37°C in anaerobic condition using Anaerocult A (Merck). Finally, harvested cell suspensions were resuspended in intestinal juice (IJ: 0.3% bovine bile, 0.1% Pancreas acetone powder porcine Type I [Sigma], pH 8.0) and incubated for 120 min at 37°C in anaerobic conditions. Two samples were collected during IJ challenge, after 60 and 120 min. Determination of viable counts was carried out in the initial cultures, as well as after each of the challenges. Serial dilutions of the samples were made in 0.85% NaCl and 10 μl were spotted onto an appropriate medium. Plates were incubated for 24 h at 37°C and results were expressed as CFU/ml. The percentage of survival was calculated from the viable counts recovered after each chemically simulated GIT step with respect to the initial counts (% CFU recovered bacteria/CFU initial bacteria). The experiments were performed in triplicate.

Adhesion to Caco-2 intestinal epithelial cells (IEC) was performed as previously described (Nikolic et al., 2012). The colonocyte-like cell line Caco-2 was used to determine the adhesion ability of bacterial strains. Caco-2 was purchased from the European Collection of Cell Cultures (ECACC No. 86010202). The culture and maintenance of the cells was carried out following standard procedures (Sánchez et al., 2010) using DMEM medium for Caco-2. Media and reagents were purchased from Sigma (St. Louis, MO, United States). Caco-2 cells (passage 5–6) were seeded in 24-well plates and cultivated until a confluent differentiated state was reached. For adhesion experiments, 13 ± 1 day-old cellular monolayers were used. All strains were cultured for 24 h and after washing twice with PBS solution were resuspended in the corresponding cell-line media without antibiotics at a concentration of about 10⁸ CFU/ml. Cellular monolayers were also carefully washed and bacterial suspensions were added at a ratio of about 10:1 (bacteria : eukaryotic cell). Adhesion experiments were carried out for 1 h at 37°C, 5% CO₂ and, afterward, wells were gently washed to release unattached bacteria before proceeding with the lysis of cellular monolayers using 0.25% Trypsin–EDTA solution (Sigma, St. Louis, MO, United States). Dilutions of samples, before and after adhesion, were made in Ringer’s solution and bacterial counts were performed on GM17 and MRS plates. The adhesion was calculated as: % CFU adhered bacteria/CFU added bacteria. Experiments were carried out in two replicated plates and in each plate three wells were used per sample.

The reduction of adhesion of E. coli ATCC25922 and Salmonella 654/7E strains in the presence of probiotic strains was tested on Caco-2 cells as described previously (Živković et al., 2016). The capability of E. coli ATCC25922 and Salmonella 654/7E strains to be associated to the Caco-2 cells in the presence and absence of probiotic strains, as well as in the presence of probiotic combination were tested. Bacterial cultures were washed twice with PBS and resuspended in DMEM without antibiotics at a concentration of ∼1 × 10⁷ CFU/ml; this number was corroborated by plate counting in the agar medium specific for each bacterium. The bacterial suspensions containing one of the pathogens or a combination of pathogens and probiotic strains (ratio 1:1) were independently added to the Caco-2 monolayers at a ratio of 10:1, (bacteria : eukaryotic cells; in the case of pathogens – probiotic combination, each bacterial type was added at a ratio of 10:1) and incubated at 37°C, with 5% CO₂ for 1 h. Afterward, the monolayers were gently washed twice with PBS to remove the unattached bacteria, and the eukaryotic cells were released using 0.25% Trypsin–EDTA solution (Sigma, St. Louis, MO, United States). The samples were diluted in 0.85% (w/v) NaCl buffer and plated to LB medium to enumerate the associated E. coli ATCC25922 and Salmonella 654/7E strains. The percentage of E. coli ATCC25922 and Salmonella 654/7E strains association was calculated as follows: 100 × CFU/ml bacteria associated/CFU/ml bacteria added (the dilution of bacteria was taken into account). Each combination was tested in triplicate. To determine the capability of probiotic strains to decrease the association of E. coli ATCC25922 and Salmonella 654/7E strains to Caco-2 monolayers, data were referred to that obtained with the E. coli ATCC25922 and Salmonella 654/7E strains alone, respectively (i.e., 100% association). Each combination of pathogens and probiotic strains, and the pathogens strain alone, respectively, were tested in three replicates.

Preparation of fermented probiotic culture includes the milk fermentation process by use of starter and probiotic bacteria in defined ratio. Incubation of inoculated milk was carried at 42°C for approximately 5 h until the pH 4.8 was reached. After that, fermented milk was transferred to the refrigerator at 4°C where the pH of fermented milk is further slowly lowered to 4.6 during cooling. The pH value and total viable count of probiotic bacteria was estimated immediately after the inoculation of milk as well as each hour during first 5 h of the milk fermentation and after 24 h, 7, 14, and 21 days of storage of fermented dairy probiotic at 4°C. Experiments are performed in three independent measurements.

The effect of the probiotic combination (including strains: BGVLJ1-44, BGHI14 and BGRA43) on the health status of farm animals was monitored in vivo in farm trial. The experiments were performed in accordance to the European convention for protection of vertebrate animals used for experimental and other scientific purposes (Directive 2010/63/EU) as well as International Guiding Principles for Biomedical Research involving Animals (C.I.O.M.S., c/o WHO, CH 1211 Geneva 27, Switzerland), and approved by Ethical Committee of Faculty of Biology, University of Belgrade, Serbia; No EK-BF-2013/09). The probiotic strains were prepared in reconstituted (10%) sterilized skimmed-milk (Difco, Becton Dickinson, Franklin Lakes, NJ, United States). Out of 50 pregnant sows 25 were treated with probiotic combination for 10 days before predicted farrowing term (at a time when the natural immunity of pregnant sows usually declines). Pregnant sows have been received the 200 ml of mixed probiotic culture (10⁸ CFU/ml) every day during the first morning feeding. In parallel, 25 control sows, not treated with the probiotic, were followed. Before and after the probiotic treatment, the general health of treated and control sows and the total number of bacteria (LAB and Enterobacteriaceae) in piglets’ faces was followed as described previously (Anonymous, 1996; Terzic-Vidojevic et al., 2007).

Extraction of bacterial DNA from fecal samples was done using the ZR Genomic DNA Tissue Mini Prep kit (Zymo Research, Irvine, CA, United States). PCR reaction with isolated bacterial DNA as a template, denaturing gradient gel electrophoresis (DGGE) analysis and gel manipulation after electrophoresis was performed essentially as described previously (Lukic et al., 2013). The dominant bacterial communities were determined by PCR using primers U968-GC and L1401-r (Invitrogen, Paisley, United Kingdom) complementary to the V6 to V8 region of eubacterial 16S rDNA (Randazzo et al., 2006), while lactobacilli diversity was detected by lactobacilli-specific primers Lab-0159f and Uni-0515-GCr (Invitrogen, Paisley, United Kingdom) (Lukic et al., 2013). The reaction was performed in thermal cycler Gene AmpR PCR System 2700 (Applied Biosystems, Foster City, CA, United States). Bacterial DNA was then set on denaturing gradient gel prepared according to Lukic et al. (2013) and glass plates for DGGE apparatus DGGE-2001 (C.B.S. Scientific, San Diego, CA, United States) were used.

To detect possible presence of the most common intestinal pathogens in piglets’, PCR reactions were performed using as a template 100 ng of DNA extracted from fecal samples [as described in Section “Denaturing Gradient Gel Electrophoresis (DGGE) Analysis”]. Primers used in the study were complementary to eltA gene and invA gene, specific for enterotoxigenic Escherichia coli (Youmans et al., 2014) and Salmonella sp. (Li and Chen, 2013), respectively. PCR reactions were prepared using 10 pmols of each primer, 0.2 mM dNTP and 0.6 U of Taq Polymerase in a volume of 30 μl and run under the following conditions: 94°C for 5 min and then 35 cycles of 94°C for 30 s, 57°C for 60 s and 72°C for 30 s with a final extension at 72°C, 7 min.

Differences between treatments were examined for significance by Student’s t-test after analysis of variance. For comparison of proportions of individual bands on DGGE profiles between piglets from different treatment groups, Pearson’s chi-squared test (χ2) was applied. Wilcoxon Signed Ranks test was used to compare the number of total bands in DGGE profiles in piglets with the number of bands in their mother sows. P > 0.05 was considered to be statistically insignificant.

The survival of probiotic strains S. thermophilus BGVLJ1-44, L. fermentum BGHI14 and L. helveticus BGRA43 after chemically simulated GIT transit was tested. The results revealed that only the strain BGVLJ1-44 successfully survived the passage through simulated GIT conditions in saline buffer, while the strains L. helveticus BGRA43 survived in simulated GIT conditions only when administered in milk (Figure 1). The results showed that the number of viable cells administered in milk after 90 min was similar to that at the beginning of the experiment in all probiotic strains, indicating that all strains were resistant to acidic conditions. However, 10 min exposure to high bile salts concentration decreased the survival of the strain BGHI14 both in saline (Δlog CFU/ml 1.1) and milk (Δlog CFU/ml 2.3), while the strain BGRA43 survived only when administred in milk (Δlog CFU/ml 2.65). Finally, 120 min exposure to lower bile salt concentration (0.3%) and pancreatic digestion considerably reduced the viability of BGHI14 (Δlog CFU/ml from 0.9 to 1.1 in saline buffer and milk, respectively), as well as the number of viable cells of BGVLJ1-44 (Δlog CFU/ml 1). Interestingly, the number of viable cells of BGRA43 in milk remained the same. The best survival rate showed BGRA43 (3.2%; Δlog CFU/ml 1.5) in milk, while the survival rate of BGVLJ1-44 was 1%; Δlog CFU/ml 2 in saline. The lowest survival rate showed BGHI14 (0.01%; Δlog CFU/ml 4.2 when administered in milk and 0.001%; Δlog CFU/ml 4.9 when administered in saline) (Figure 1).

Since the results of in vitro GIT survival showed that the best survival rate of the strains was obtained in milk, the strains S. thermophilus BGVLJ1-44, L. fermentum BGHI14 and L. helveticus BGRA43 were used as starter cultures for preparation of fermented dairy probiotic. The general physiological, biochemical and technological characteristics of the strains are summarized in Table 1. Only L. fermentum BGHI14 strain showed ability to grow at all three tested temperatures (15°C, 30°C and 45°C) and none of them grew in broth with 2% NaCl. Hydrolysis of arginine and production of CO₂ showed L. fermentum BGHI14 strain, while production of acetoin and diacetyl were characteristic for S. thermophilus BGVLJ1-44 strain. All three LAB strains survived at 63.5°C for 30 min. L. fermentum BGHI14 as heterofermentative strain could not coagulate the milk. On the other hand, the strains S. thermophilus BGVLJ1-44 and L. helveticus BGRA43 showed the ability of fast milk curdling and they completely hydrolyze α_s1-, β- and κ-casein after 3 h of incubation at 37°C. S. thermophilus BGVLJ1-44 strain was an excellent EPS producer (Table 1).

Appart from milk, the rate of acid development was also followed in whey. The pasteurized milk and way were inoculated each with 3% probiotic starter culture. Measuring of pH values during 0, 1, 2, 3, 4, and 5 h of fermentation at 42°C, as well as after 24 h, 7, 14, and 21 days of storage at 4°C was performed. In parallel, total count of viable probiotic LAB was followed (Figure 2). The pH of non-inoculated milk and whey were 6.81 and 6.65, respectively.

Probiotic starter culture inoculated in whey decreased pH value faster than probiotic starter culture inoculated in milk. The pH value of fermented whey was 4.80 after 4 h at 42°C and pH of fermented milk was 4.78 after 5 h of incubation at 42°C. During the cooling, the pH values of fermented milk and whey slowly decreased and after 21 days of storage at 4°C the pH values of fermented milk and whey were 4.47 and 4.38, respectively (Figure 2).

In parallel, total counts of viable probiotic starter cultures during fermentation of milk and whey were followed. The total number of viable LAB increased in milk and whey from 7.5 log CFU/ml and 7.4 log CFU/ml, respectively, at the beginning of incubation period to 8.9 log CFU/ml in milk and 8.2 log CFU/ml in whey at the end of fermentation at 42°C. After 21 days of storage of the fermented milk and whey the number of probiotic bacteria in fermented milk and whey was 8.2 log CFU/ml and 7.9 log CFU/ml, respectively (Figure 2).

In order to determine the ability of the strains S. thermophilus BGVLJ1-44, L. fermentum BGHI14 and L. helveticus BGRA43 to counteract the adhesion of pathogens, the adhesion properties of the probiotic strains to IEC Caco-2 were determined. The strain BGRA43 showed significantly (p < 0.05) higher adhesion of 1.7 to 2.1-fold to Caco-2 cells comparing to BGVLJ1-44 and BGHI14, respectively (Figure 3).

Further, the potential of probiotic strains BGVLJ1-44, BGHI14 and BGRA43, as well as the mixed probiotic culture, to compete with E. coli ATCC25922 and Salmonella 654/7E and to reduce their adhesion to Caco-2 cells was studied. The adhesion of E. coli ATCC25922 strain was significantly (p < 0.05) lower in the presence of BGRA43 (1.5-fold) and BGHI14 (1.2-fold) lactobacilli strains (Figure 3). The highest reduction (2.6-fold) of E. coli ATCC25922 adhesion to Caco-2 cells was in the presence of the mixed probiotic culture (p < 0.01) (Figure 3).

In addition, the adhesion of Salmonella 654/7E to Caco-2 cells was significantly reduced (16-fold) in the presence of BGVLJ1-44 (p < 0.01) and 1.5-fold in the presence of BGRA43 (p < 0.05) (Figure 3). Interestingly, there was no reduction of Salmonella 654/7E adhesion in the presence of BGHI14 and probiotic mixture.

Since the results given in Section “In Vitro GIT Survival of Probiotic Strains” showed that the number of probiotic bacteria was higher in fermented milk than in whey, the fermented milk containing probiotic strains L. fermentum BGHI14, L. helveticus BGRA43 and S. thermophilus BGVLJ1-44 was used in farm trial.