Thirty-one strains of Lactobacillus pentosus isolated by Abriouel et al. (2012) from Aloreña green table olives naturally fermented by four small–medium enterprises (SMEs) from Malaga (Spain) were used in this study (Supplementary Table S1). Selection of lactobacilli was done on the basis of the preliminary functional screening done by Abriouel et al. (2012). These strains were routinely cultured at 30°C in de Man Rogosa and Sharpe (MRS) broth (Fluka, Madrid, Spain) or agar under aerobic conditions for 24–48 h. Strains were kept in 20% glycerol at -80°C for long term storage.

DNAse and gelatinase activities were determined as described by Lavilla-Lerma et al. (2013). Mucin degradation ability of lactobacilli was carried out as reported by Muñoz-Atienza et al. (2013). Production of biogenic amines from tyrosine, histidine, ornithine, or lysine (all of them from Sigma, Madrid) was done as described by Yousif et al. (2005) on MRS agar plates supplemented with the corresponding amino acids. With respect to hemolytic activity, overnight cultures of lactobacilli were streaked on the agar blood (Scharlab, Spain) and then incubated anaerobically at 37°C for 24 h.

Survival under gastric conditions including low pH (1.5–3) and bile salt concentration (0–4% with increments of 1%) was done according to the methods described by Millette et al. (2008). For acid tolerance, 1 ml of overnight MRS broth cultures was inoculated onto 19 ml of simulated gastric fluid (3.2 g/l pepsin and 2 g/l NaCl) adjusted at different pH values (with 5M HCl) and then incubated for 30 min at 37°C. Viable counts (CFU/ml) were determined after incubation plating 1 ml of the mixture on MRS-agar. As reference, viable bacteria without simulated gastric was used (Millette et al., 2008). Regarding bile salt tolerance, MRS-agar plates added with different concentrations of bile salt mixture (Sigma B-3426) were inoculated onto the surface by overnight MRS broth cultures (100 μl). Then, the plates incubated at 37°C for 72 h were examined visually for bacterial growth (Millette et al., 2008).

Auto-aggregation capacity of lactobacilli was determined as reported by Vizoso Pinto et al. (2007). Overnight cultures (2 ml) of lactobacilli in MRS broth were harvested, washed and resuspended in sterile phosphate buffered saline (PBS). After 2 h at room temperature, 100 μl were removed from the top of the suspension and were transferred to a cuvette containing 900 μl PBS. The auto-aggregation percentage is expressed as: (1 -A₁/A₀) × 100, where A₀ and A₁ represent the absorbance measured at 580 nm at time = 0 and time = 2 h, respectively.

Co-aggregation capacity of lactobacilli with pathogenic bacteria (Listeria innocua CECT 910, Staphylococcus aureus CECT 4468, Escherichia coli CCUG 47553, and Salmonella Enteritidis UJ3449) was carried out according to Vlková et al. (2008). Overnight cultures (10 ml) of lactobacilli in MRS broth and pathogenic bacteria in TSB broth at 37°C were harvested, washed, resuspended in sterile PBS and their OD₆₀₀ was adjusted to 1. Cell suspension was prepared mixing 3 ml of each bacteria (L. pentosus and one pathogenic strain) and then the OD₆₀₀ of upper suspension was measured at time 0 and after 1 h incubation at room temperature. The percentage of co-aggregation was expressed as: Co-Agg% = [1 - (A₆₀₀of upper suspension at time 1 h/A₆₀₀of total bacterial suspension at time 0)] × 100.

Biofilm formation by lactobacilli was tested as described by Toledo-Arana et al. (2001). The OD₆₂₀ was measured in microplate reader (Varioskan Flash Reader, Thermo Scientific) using 1% crystal violet.

The capacity of L. pentosus strains (1 × 10⁶ CFU/ml) to grow in MRS broth (Scharlab, Spain) under different conditions of temperature (4, 10, 30, and 37°C) and in the presence of salt (6.5% NaCl at 30°C) was tested and quantified determining viable cell number (CFU/ml) after 0, 1, 3, and 6 days of incubation. Survival capacity of lactobacilli to freezing temperature (at -80°C) was checked during 0, 1, 3, and 6 days of storage. In all cases, cell counts were done in triplicate on MRS agar (Scharlab, Spain) for 48 h at 30°C.

Screening of α-amylase, protease, bile salt hydrolase (BSH), haeme-dependent catalase and carboxymethyl cellulase (CMC) were tested as described by Knauf et al. (1992), Franz et al. (2001), Lucas et al. (2001), Ben Omar et al. (2004), and Yousif et al. (2005), respectively.

Regarding the utilization of non-digestible compounds, the α-galactoside sugars tested were stachyose or raffinose as described by Yousif et al. (2005). The plates were incubated at 30°C and observed for acid production every day over a 3-day period. With respect to oxalate degradation, lactobacilli were screened as reported by Gomathi et al. (2014) using the agar well-diffusion method. For this, 20 mM calcium oxalate plates were inoculated by 0.1 ml of lactobacilli overnight cultures and then incubated at 37°C for 12 h for clear zone observation around the wells.

On the other hand, growth of lactobacilli on prebiotics was done as described by Makras et al. (2005) using the agar plate assays. In this sense, modified MRS broth without glucose and supplemented with 0.5 g/l of L-cysteine hydrochloride (Sigma) (mMRS) was added with 2% (w/v) of different energy sources (glucose, fructose, galactose, lactose, saccharose, lactulose, or inulin) and 300 mg/l of bromocresol purple (Sigma) as a color indicator. Lactobacilli suspensions were prepared as described by Makras et al. (2005) and spotted on mMRS agar plates which were incubated anaerobically at 37°C for 48 h. Plates performed in triplicate were checked for color changes around the colonies.

Production of hydrogen peroxide was performed according to the method of Marshall (1979). Bacteriocin screening was done by the spot-on-a-lawn method as described by Abriouel et al. (2012). PCR screening of plantaricin genes was carried out as described by Ben Omar et al. (2006).

Tolerance of selected L. pentosus strains -on the basis of their probiotic profile obtained by means of statistical methods (Principal Component Analysis) explained below- to simulated human gastrointestinal tract was carried out as reported by Chen et al. (2014) under simulated gastric juice (pH 3.0) and intestinal gastric juice (pH 8.0). Furthermore, we studied the effect of nitrate (5 mM) or glucose (500 mM) in both simulated gastrointestinal conditions.

Selected L. pentosus strains with the best probiotic profile (simulated gastro-intestinal juice in standard conditions and in the presence of nitrate or glucose) were tested for their capacity to adhere to Enterocyte-like Caco-2 ECACC 86010202 (from colon adenocarcinoma) and HeLa 229 ECACC 86090201 (from vaginal cervix carcinoma) (both from the Scientific Instrument Services of the University of Granada, Spain). Eukaryotic cells were cultured as described by Lavilla-Lerma et al. (2013). Adhesion assays were carried out following the method of Moroni et al. (2006) by adding 250 μl (10⁸ CFU/mL) of each bacterial strain to a monolayer of differentiated cells (Lavilla-Lerma et al., 2013). Plates were then incubated at 37°C for 30 min and free bacteria were eliminated by washing the cell layers twice with phosphate-buffered saline (PBS, Sigma). To determine the CFU/ml of lactobacilli adhered to cells, those were harvested with EDTA-trypsin, centrifuged, and serially diluted in PBS before plating on agar-MRS.

All analyses were done in triplicate. Statistical analysis of data was accomplished using Excel 2007 program to determine the average data ± standard deviations. Statistical treatment of adhesion data was conducted by analysis of variances (ANOVA) in Statgraphics Centurion XVI, software using Shapiro–Wilk test and the Levene test to check data normality and the 2-sided Tukey’s test to determine the significance of differences between strains, where a P-value of <0.05 was considered statistically significant.

Principal Component Analysis (PCA) was used to emphasize variation and bring out strong differences in co-aggregation capacity of L. pentosus strains with Gram-negative and Gram-positive pathogens. On the other hand, we also used PCA analysis for selection of the best probiotic L. pentosus strains by using the following discriminating variables: survival at low pH of 1.5, auto-aggregation and co-aggregation with different pathogens.

None of the strains analyzed in the present study showed positive results for safety aspects tested.

Under gastric conditions, different viability rates were shown depending on the L. pentosus strain (Table 1). All L. pentosus strains were able to survive (>85–100%) at low pH (2–3), however at pH 1.5 only 8 of 31 strains showed high and statistically significant survival rates (86–97%). All L. pentosus strains were able to survive in the presence of 4% bile salt (Table 1).

Auto-aggregation of lactobacilli belonging to the same strain is an important feature especially in the human gut. Table 2 showed that 6, 13, and 12 of L. pentosus strains exhibited different auto-aggregation abilities ranging from high (50–77.92%), medium (35–50%), and low (16–35%), respectively (Table 2), taking as control L. johnsonii CECT 289 (35%). Variability in auto-aggregation ability was obtained among the tested strains (p < 0.05) belonging to the three groups mentioned above, indicating that auto-aggregation is a strain specific property.

Co-aggregation of lactobacilli with pathogenic bacteria was variable and statistically significant depending on the lactobacilli and pathogenic strains used (Table 2). High co-aggregation capacity (40–67%) of lactobacilli (nine strains) was detected with E. coli and S. aureus, while five and six lactobacilli strains highly co-aggregated with Listeria innocua and Salmonella, respectively (Table 2). However, the other lactobacilli strains (39–48%) showed variable co-aggregation capacities ranging from 20 to 38% with all pathogenic bacteria. Furthermore, 23–39% of lactobacilli showed less than 20% of co-aggregation capacity (Table 2). In general, L. pentosus strains showed higher and statistically significant co-aggregation capacity with Gram-negative bacteria as compared to Gram-positive bacteria as shown by a multivariate analysis (PCA) with three components explaining 93.88% of total variation (Figure 1).

Regarding biofilm formation, 20 of 31 of L. pentosus strains were able to form biofilms although with different degree being 9 of 20 strains with high capacity (Table 3).

All L. pentosus strains were able to grow in the presence of 6.5% NaCl (data not shown). Supplementary Table S1 showed that all lactobacilli generally showed good survival capacity under different temperature conditions being growth capacity mainly dependent on the incubation temperature and the L. pentosus strain. Generally, under temperatures of 4, 10, 30, and 37°C, all strains showed growth after 1–6 days incubation by almost 2.74 Log₁₀ units reaching the maximum after 1 day incubation at 30 or 37°C (except few cases) and 3 days at 10°C (Supplementary Table S1). However, at freezing temperature (-80°C) no growth was recorded and survival of almost all L. pentosus strains during storage for 6 days was shown (Supplementary Table S1). In this sense, almost all L. pentosus strains showed high survival capacity of 100%, however six strains showed a slight decrease in viable cell counts by 1.04–1.65 Log₁₀ units after 6 days storage at -80°C (Supplementary Table S1).

The results obtained showed that all lactobacilli strains were able to produce BSH, 58 and 39% of strains were able to produce haeme-dependent catalase and cellulolytic activity, respectively (Supplementary Table S2). However, none of lactobacilli strains produced α-amylase nor protease (data not shown). Regarding fermentation of human non-digestible α-galactoside sugars, 52% of L. pentosus strains exhibited the capacity to ferment raffinose, but not stachyose except L. pentosus MP-10 (Supplementary Table S2). Furthermore, no oxalate degradation ability was found in L. pentosus strains (data not shown). Concerning growth of lactobacilli on prebiotics, all strains fermented the monosaccharides glucose, fructose, and galactose (except L. pentosus CF2-12 for galactose) (Supplementary Table S2). Moreover, all lactobacilli ferment saccharose and lactulose and almost all lactobacilli ferment lactose except three L. pentosus strains (CF2-12, Lp-7N, and 5C3) but none of the strains ferment inulin (Supplementary Table S2).

Regarding antimicrobial activity, none of lactobacilli strains produced hydrogen peroxide (data not shown), however bacteriocin activity was detected in all strains by means of phenotypic methods (Table 3). However, genotypic screening of plantaricin genes indicated the presence of plnA and plnD genes in 45 and 23% of L. pentosus strains, respectively (Table 3). Concerning other plantaricin genes, 6–10% of strains showed the presence of plnJ, planNC8, or plnW genes. Nevertheless, none of the strains exhibited the presence of plnB, plnC, plnEF, plnG, plnI, plnK, plnN, or plnS genes (Table 3).

To carry out this test, 9 of 31 L. pentosus strains with the best probiotic profile were selected by using PCA analysis as described in “Materials and Methods” section. Figure 2 represents the distribution of variables in a three dimensional analysis of the Principal Component (84.19% total variance) and also the position of L. pentosus strains in the space of three dimensions being organized in three main groups. Strains with the best scores were selected as the most representative strains (nine strains in total) to be used in further studies (Figure 2; Table 4). Under gastric conditions (pH 3.0), nine selected L. pentosus strains exhibited different survival rates depending on the strain and the exposure time (1–3 h) (Table 4A). After 3 h incubation in standard conditions (pH 3.0), L. pentosus AP2-15N, CF1-39, CF2-10N, CF2-12, and MP-10 strains showed high and statistically significant survival capacity of 96.96–99.76% (Table 4A). However, the rest of L. pentosus strains (AP2-16N, CF1-6, CF2-5, and 5C2) showed 58.62–81.85% survival under standard conditions (Table 4A). Similar results were obtained under simulated intestinal conditions (pH 8.0) for all strains (Table 4B).

When simulated gastric juice was supplemented with 5 mM nitrate, different results were obtained depending on the L. pentosus strain and simulated gastrointestinal conditions. Generally, reduction of survival capacity was observed in the same lactobacilli strains which showed poor survival capacity under gastric and intestinal conditions (Table 4).

To evaluate the effect of glucose on survival capacity of lactobacilli under gastric and intestinal conditions, 500 mM glucose was added to simulated gastric (pH 3.0) and intestinal (pH 8.0) juices (Table 4). The results obtained showed that glucose plays a protective role of lactobacilli under both conditions since all strains reached almost 100% (97–100%) survivability after 3 h incubation in simulated gastric juice (pH 3.0) and simulated intestinal juice (pH 8.0) (Table 4).

In conclusion, L. pentosus strains (CF2-5 and 5C2) showed less survival capacity in simulated gastro-intestinal juice in the presence of 5 mM nitrate, thus they were discarded from further analysis.

Selected L. pentosus strains with the best probiotic profile (seven strains) were tested for their capacity to adhere to Enterocyte-like Caco-2 ECACC 86010202 (from colon adenocarcinoma) and HeLa 229 ECACC 86090201 (from vaginal cervix carcinoma). The results obtained showed a high variability in adhesion capacity depending on the strain and also on the cellular line (Figure 3) since the adhesion to HeLa 229 (up to 57.88%) was more important and statistically significant than to Caco-2 (30.02%) cells as shown in Figure 3 except for L. pentosus CF2-12. Thus, L. pentosus strains showed decreasing adhesion capacity to Hela cells as follows: CF2-10N > CF1-6 > AP2-16N > group of MP-10, CF2-12, AP2-15N and CF1-39 strains (Figure 3). However, in the case of Caco-2 cells, two groups were defined: one comprising L. pentosus AP2-16N, CF1-6, and CF2-10N strains and the other group with the rest of strains being statistically different (Figure 3). In conclusion, L. pentosus CF2-10N, CF1-6, and AP2-16N strains exhibited the best adhesion profile (33.55–57.88% and 18.11–30.02% for HeLa 229 and Caco-2 cells, respectively).