Evaluating the Probiotic Potential of Lactic Acid Bacteria Implicated in Natural Fermentation of Table Olives, cv. Cobrançosa

The probiotic features of Lactiplantibacillus (L.) pentosus and L. paraplantarum strains, endogenous in Cobrançosa table olives from northeast Portugal, were assessed in terms of functional properties and health benefits. Fourteen lactic acid bacteria strains were compared with Lacticaseibacillus casei from a commercial brand of probiotic yoghurt and L. pentosus B281 from Greek probiotic table olives, in attempts to select strains with higher probiotic performances than those references. For functional properties, the i53 and i106 strains, respectively, exhibited: 22.2 ± 2.2% and 23.0 ± 2.2% for Caco-2 cell adhesion capacity; 21.6 ± 7.8% and 21.5 ± 1.4% for hydrophobicity; 93.0 ± 3.0% and 88.5 ± 4.5% for autoaggregation ability by 24 h of incubation; and ability to co-aggregate with selected pathogens—from 29 to 40% to Gram+ (e.g., Staphylococcus aureus ATCC 25923 and Enterococcus faecalis ATCC 29212); and from 16 to 44% for Gram− (e.g., Escherichia coli ATCC 25922 and Salmonella enteritidis ATCC 25928). The strains proved to be resistant (i.e., halo zone ≤14 mm) to some antibiotics (e.g., vancomycin, ofloxacin, and streptomycin), but susceptible (i.e., halo zone ≥ 20 mm) to others (e.g., ampicillin and cephalothin). The strains exhibited health-beneficial enzymatic activity (such as acid phosphatase and naphthol-AS-BI-phosphohydrolase), but not health-harmful enzymatic activity (such as β-glucuronidase and N-acetyl-β-glucosaminidase). Additionally, the antioxidant activity and cholesterol assimilation features, respectively, of the strains were 19.6 ± 2.8% and 77.5 ± 0.5% for i53, and 19.6 ± 1.8% and 72.2 ± 0.9% for i106. This study indicated that the addition of L. pentosus strains i53 and/or i106 to Cobrançosa table olives is likely to enhance the added value of the final product, in view of the associated potential benefits upon human health.


Introduction
The food industry has grown recently in response to consumer demand for higher quality food and food bearing positive effects upon human health. The latter belongs to the so-called family of "functional foods"-defined as matrices with the addition of nutraceutical ingredients, or subjected to minimum food processing to enhance bioactivity, but always providing health benefits beyond the provision of essential nutrients if regularly Overall, the main goal of this study was to evaluate the probiotic potential of 14 selected Lactoplantibacillus strains isolated from Cobrançosa cultivar. This was achieved by running tests that were split into two categories: functional properties-related to (expected) strain performance along the human gastrointestinal tract, thus critical to ascertain their chance of exerting a probiotic function once in the colon; and health benefits-associated with specific bioactivities exhibited in vitro, thus likely to also be present in vivo in the colon. The study of the probiotic potential of LAB will eventually allow for development of a starter or adjunct culture with commercial expression, specifically adapted to fermentation of Cobrançosa olives. By doing so, a final product of higher quality is anticipated-bearing improved functional properties in addition to their pre-existing sensory and nutritional features. Therefore, probiotic Cobrançosa table olives will likely reach a premium price in the modern food market, thus contributing to sustainable socio-economic and industrial development of the northern region of Portugal.

Functional Properties
The results pertaining to functional properties, such as adhesion to Caco-2 cells, hydrophobicity, autoaggregation, and co-aggregation, for the 14 LAB native strains (belonging to genus Lactiplantibacillus (L.), and species L. paraplantarum and L. pentosus, previously collected from Cobrançosa table olive brines [24]) (see Table 1), additionally of two probiotic reference strains (from animal and plant sources), are tabulated in Tables 2 and 3. Statistical differences between LAB cultures were found at the 5% level of significance (see ANOVA p-value in Table 2). For instance, strain i106 exhibited the highest value of Caco-2 cell adhesion (denoted by superscript a ), which was statistically similar to that of strain i53 (also denoted by superscript a ), but statistically different from the others (accordingly lacking superscript a ). Comparatively, strain i24 exhibited the lowest value of Caco-2 cell adhesion (denoted by superscript i ), which was statistically similar to that of strains i21, i26, i32, i34, and i108 (also denoted by superscript i ), but statistically different from the others (accordingly lacking superscript i ). Upon comparison of our results of adhesion to those encompassing the strain isolated from Greek table olives, it was concluded that strains i106, i53, and i101, exhibit similar performance-with scores of 23, 22, and 19%, respectively. Conversely, strains i21, i24, i32, i34, and i108, exhibited adhesion significantly lower than the yoghurt probiotic strain, viz. 6.2, 5.3, 6.0, 6.1, and 4.8%, respectively. The other strains displayed significantly higher adhesion than that of commercial probiotic yoghurt, but lower than that of Greek olives.

Hydrophobicity
Statistical differences between LAB cultures were found at the 5% level of significance (see ANOVA p-value in Table 2). Both reference strains performed similarly with regard to hydrophobicity-27 and 32% for yoghurt and table olives, respectively; most tested strains exhibited, in general, behaviours similar to those of the reference strains. Strain i21 showed significantly higher hydrophobicity (56%, denoted by superscript a ), followed by strain i32 (41%, denoted by superscript b ); whereas lower hydrophobicity was recorded for four strains-i19 (16%), i101 (19%), i112 (19%), and i108 (21%) (all denoted by superscript de ).

Autoaggregation
At the 5% level of significance (see ANOVA p-values in Table 2), no statistical differences between LAB cultures were apparent by 5 h, but the reverse occurred by 24 h. For instance, strain i53 exhibited the highest value of autoaggregation value by 24 h of incubation (ca. 93%, denoted by superscript a ), whereas i22 exhibited 59% as the lowest value of autoaggregation (denoted by superscript d ); the other strains exhibited autoaggregation performance statistically similar to the commercial yoghurt strain (denoted by superscript abcd ).

Health Benefits
The results pertaining to potential health benefits upon the human being, regarding the 14 LAB native strains from Cobrançosa and two probiotic reference strains (from animal and plant sources), are depicted in Tables 4 and 5. Table 4. Performance of L. pentosus and L. paraplantarum strains pertaining to antibiotic resistance.

Antibiotic Resistance
All native LAB cultures assessed exhibited resistance to vancomycin, ofloxacin, and streptomycin; and susceptibility to ampicillin, erythromycin, tetracycline, chloramphenicol, and cephalothin (see Table 4). Strain i112 and the yoghurt control strain exhibited intermediate resistance to ofloxacin, yet their inhibition zones were very close to the "Resistant score" (i.e., halo ≤ 14 mm). Therefore, the resistances of our strains were quite similar to those of either reference strain.

Antimicrobial Activity
Every LAB culture presented no significant inhibition halo around the well when neutralized cell-free supernatant was added. However, blurred halos were observed when the acidic CFS (cell-free supernatant) was used.

Antioxidant Activity
Statistical differences between LAB cultures were found at the 5% level of significance (see ANOVA p-value in Table 5); results ranged from 18% (strain i23) to 23% (strain i19). Additionally, the antioxidant activities of all our strains (except i22 and i23) were statistically higher (denoted by superscripts a or ab ) than those of the reference strains, viz. 18% and 13% for yoghurt (denoted by superscript bc ) and Greek olives (denoted by superscript cb ), respectively.

Cholesterol Assimilation
Statistical differences between cholesterol content of LAB cultures were found at the 5% level of significance (see ANOVA p-value in Table 5). Strain i23 presented the highest cholesterol assimilation percentage (83%, denoted by superscript a ), despite statistical equivalence to all strains except i24, i101, i17, i108, and i32; the lowest assimilation was exhibited by the latter (68%, denoted by superscript e ). All LAB strains exhibited significantly higher values than that of the yoghurt reference strain-53% (denoted by superscript f ). Compared to the Greek reference strain-77% (denoted by superscript abcd )-all strains presented significantly equivalent percentages of cholesterol assimilation.

Proteolytic Activity
All studied strains showed proteolytic activity, detected by clear zones around the colonies (see Table 5). Such zones were not identical for all strains, yet the variations in diameter were sufficiently small to merit report.

EPS Production
All strains of L. pentosus, except i32, tested positive for EPS production-as per the appearance of white colonies; a positive result was found only for L. paraplantarum strain i21 (see Table 5).

Enzymatic Activity
As shown in Table 6, all LAB strains exhibited high activities of aminopeptidase (regarding leucine arylamidase and valine arylamidase), acid phosphatase, naphthol-AS-BI-phosphohydrolase, β-galactosidase, and α-glucosidase; β-glucosidase activity was recorded as high for all strains, except the yoghurt reference strain. No strains exhibited lipase, trypsin, α-chymotrypsin, β-glucuronidase, N-acetyl-β-glucosaminidase, αmanosidase, or α-fucosidase activities. All strains, except i21, i17, i22, i23, i34, and i106, presented high esterase and lipase/esterase activities. Regarding the yoghurt reference strain, these two activities were found to be high, while the Greek strain exhibited low activities. Finally, α-galactosidase activity was recorded as low for strains i21, i17, i22, i23, i34, i106, and the Greek strain; and was absent in the other strains. The same strains possessed no alkaline phosphatase activity-similarly to the reference strains, and unlike the remainder (with low activities). Table 6. Performance of L. pentosus and L. paraplantarum strains pertaining to enzymatic activities.

Discussion
Adhesion performance to Caco-2 cells in vitro-which mimics in vivo epithelial adhesion on the colon-is crucial for a microorganism to exhibit a probiotic effect. Probiotics must anchor in the colon to exert their action and provide an opportunity to interact with local bacteria and the host's own cells. The results of our strains (from 1% to 23%) are similar to those of Luz et al. [25], with values between 3% and 19%, and of Maragkoudakis et al. [26], ranging from 0.2% to 25.5%; the latter considered adhesion capacity above 13% to be sufficient.
The ability of bacteria to stick to hydrocarbon moieties constrains their extent of adhesion to epithelial cells in the colon, known as cell surface hydrophobicity [27]; the higher the hydrophobicity, the higher the adherence ability [28]. The values of hydrophobicity of our strains in contact with xylene, as model hydrocarbon, varied significantly among them-16-56%; likewise, Abouloifa et al. [17] reported viz. 15-35%. Similar to Benítez-Cabello et al. [29], our study was directed at an opposite relation between hydrophobicity and adhesion to cell lines. This is, probably, a consequence of the fact that attachment to surfaces depends not only on hydrophobicity of the cell surface, but also on Brownian movement, van der Waals attraction, and surface electrostatic charges. Another explanation is the possibility that microorganisms switch between hydrophobic and hydrophilic phenotypes, in response to changes in environmental conditions and growth phases [29]. Motey et al. [30] and Joghataei et al. [31]-who also found a peculiar anticorrelation-claimed that electrostatic power and cell surface charges originating from proteins, glycoproteins, teichoic and lipoteichoic acids, and exopolysaccharides on the cell wall, may in fact be expressed differently from strain to strain [32].
The data of autoaggregation performance by 5 h of incubation, viz. 19-32%, were similar to those of Abouloifa et al. [17], viz. 10-41%; whereas data collected by 24 h of incubation, viz. 59-93%, presented similarities to those of Margalho et al. [4], viz. 69-99%. Reasoning for having tested those two incubation times arises from the experimental conditions selected among similar studies, e.g., 5 h [17,33] or 24 h [34,35]. However, the former appears to not be sufficiently discriminatory between strains. Autoaggregation ability ensures that the probiotic strain reaches a high cell density in the gut, thus contributing to the adhesion mechanism by forming a protective barrier on the intestinal mucosa and epithelial cells, and promoting temporary settlement [4,36].
Co-aggregation of a strain with a pathogen is expected to play an important role in its elimination from the gastrointestinal tract-namely, via formation of a barrier that prevents colonization by pathogenic bacteria, and enhances production of antimicrobial substances [31,37]. Our values for co-aggregation with pathogenic strains ranged from 16 to 45%. These results agree with Luz et al. [25], viz. 9-31% for LAB strains vs. Escherichia coli, Salmonella enterica, and Listeria monocytogenes; with Joghataei et al. [31], viz. 28-37% vs. E. coli and 25-51% vs. Salmonella typhimurium; and with Peres et al. [18], viz. 10%-50% vs. E. coli, S. aureus, S. typhimurium, L. monocytogenes, and Helicobacter pylori. Regarding Candida albicans, similar results were reported by Jørgensen et al. [38]. Despite consistent results, this test failed to differentiate between strains, given their statistical similarity.
Resistance towards antibiotics associated with probiotic food consumption-together with evidence of horizontal gene transfer from beneficial bacteria to pathogenic bacteria in host's gut [36]-has become a public health concern; this accordingly justified inclusion of this test in our study. Resistance of our strains to streptomycin and vancomycin, and susceptibility to chloramphenicol, have also been reported by Ozkan et al. [28]; susceptibility to erythromycin, ampicillin, chloramphenicol, and cephalothin, as well as resistance to vancomycin, have also been reported by Abouloifa et al. [17]. Ofloxacin resistance has been claimed by Zarour et al. [39]; Vergalito et al. [40] and Luz et al. [25] have shown susceptibility to tetracycline, as well. Regarding the other antibiotics, our results did not always agree with the literature-probably due to strain-dependent behaviours [40]. Absence of antibiotic resistance has been outlined in FAO/WHO probiotic selection guidelines as a crucial safety parameter, as reported by Ozkan et al. [28]. Transmissible resistance to antibiotics, e.g., to tetracycline encoded by host mobilizable plasmids, would be a threat; however, our strains did not convey this [39]. Although resistance to vancomycin has been recorded, it is known to be intrinsic and due to chromosomal genes, with no expected transfer of antibiotic resistant genes from LAB to pathogens; furthermore, Ozkan et al. [28] showed that resistance to vancomycin and streptomycin is naturally present in LAB strains, rather than acquired through gene transfer. Therefore, the risk of lactobacillemia caused by probiotic Lactobacillus has been considered unequivocally negligible [17]. Conversely, probiotic strains with intrinsic resistance to antibiotics may prove useful in restoring intestinal microbiota, following treatment therewith [39].
Antimicrobial activity against pathogens is deemed a desirable trait, although not mandatory for probiotic performance [41]. Our results did not support clear conclusions in this regard, consequently this test was discarded at early stages in the selection process of probiotic strains.
Antioxidant compounds produced by probiotics protect the human body against high levels of oxygen radicals-known to cause damage to lipids, proteins, and DNA-and accordingly help prevent many diseases, e.g., cardiovascular diseases, diabetes, and ulcers in the gastrointestinal tract [17,36]. Compared to data conveyed by Abouloifa et al. [17], viz. 43-93% (via the same method), our results were considerably lower, viz. 16-23%; comparatively, Oliveira et al. [20] reported even lower values, 3-17%. Ayyash et al. [42] claimed antioxidant values of 5-20% on the day of the experiment, depending on the strain; but an increasing trend until 21 days of incubation-when values ranged within 10-50%. Therefore, antioxidant activity may increase with residence time in the gut.
Recent discoveries have also linked probiotic ingestion to prevention of heart disease, by lowering serum levels of cholesterol and decreasing its solubility which reduces uptake in the gut [36]. Our study produced cholesterol assimilation values between 68 and 82%, thus revealing significantly higher percentages for all strains than those of the yoghurt control strain. Our values were higher than those of Vergalito et al. [40], viz. 20-50%, despite the use of similar methodology.
It is important that probiotic strains possess proteolytic activity, as this facilitates nutrient procurement once in the gastrointestinal tract; additionally, bioactive peptides may be produced [18,43]. Although all strains exhibited proteolytic activity, the qualitative data were not conclusive enough to support discrimination between them.
LAB strains presented a broad spectrum of enzymatic activities, yet our results were essentially similar to those in the literature [17,35]. β-glucosidase activity was high for all our strains, except the yoghurt reference strain; this was expected, as this enzyme is responsible for degrading oleuropein, a natural component of table olives. This activity is very important to remove natural fruit bitterness, or reduce the time needed in olive debittering-known to enhance food flavour and accelerate ripening [45][46][47]. Other enzymatic activities can enhance food flavour; some strains showed positive results for esterase and esterase/lipase, which support lipolyzed flavours. Other examples include α-glucosidase activity, which catalyses degradation of saccharose into sweeter fructose via hydrolysis of α-glycosidic linkages of the non-reducing end of oligosaccharides and polysaccharides [35]; and aminopeptidase activity, which plays an important role toward hydrolysis of bitter peptides, and concomitant release of amino acids [17,25]. Additionally, acid phosphatase and naphthol-AS-BI-phosphohydrolase activities were found to be high; in the human digestive system, both are necessary to free phosphoryl groups from other molecules during digestion, aside from improving food maturation [25,47]. Contrarily, some enzymes should be absent in probiotic strains. This is notably the case of N-acetylβ-glucosaminidase activity, which may raise a safety concern for application in foods, as it releases toxic products, and transforms many pro-mutagens and pro-carcinogens into their mutagenic and carcinogenic forms, respectively [48]; in addition, being associated to the development of several gastrointestinal diseases [40,49]. Furthermore, trypsin, αchymotrypsin, and β-glucuronidase, are also associated with intestinal diseases [17], thus should be absent in potential probiotic strains. Low activities toward carbon sources such as α-mannose and α-fucosidase, as well as high β-galactosidase and αand β-glucosidase activities, suggest that our strains have a preference for lactose and glucose as carbon and energy sources [47]. Alkaline phosphatase (responsible for dephosphorylation) and lipase (bearing lipolytic activity) activities were quite low, even absent, in our strains. β-Galactosidase activity was found to be high among our strains and is relevant for its ability to alleviate the symptoms of lactose intolerance following dairy product ingestion, or intolerance to galactooligosaccharides that serve as specific nutrients to probiotics already established in the gut [25].
In attempts to select the best adventitious (potentially) probiotic strain in Cobrançosa, analysis should depart from the functional properties of the strains under scrutiny. LAB isolates bearing low values for hydrophobicity exhibited high values for adhesion and autoaggregation, and vice versa. The variable hydrophobicity is thus redundant, consequently it was not considered as a selection criterion.
The best candidates to probiotic LAB strains, based on functional properties that include RAPD profiles, are i17, i22, i23, i53, and i106. Considering Margalho et al. [4] labelled 24 h-aggregation percentages as "high" when above 70%, and "moderate" if between 20% and 70%, our strains i17, i23, i53, and i106, would accordingly rank as high, unlike strain i22 (59%). Therefore, we decided to discard i22 as a best candidate for a probiotic culture tailor-designed in the addition to Cobrançosa table olives in a future standardized process.
The best candidates to probiotic LAB strains, based on functional properties including processing time, were strains i17, i24, i53, i101, i106, and i108; this is probably a consequence of ecological dominance of the most resistant strains by the final time of each stage.
Of the two aforementioned analyses regarding functional properties, the isolates selected for further consideration were i17, i53, and i106. This selection is consistent with Table 2, as these isolates exhibited good values for autoaggregation (84, 93, and 89%, respectively); the same did not occur with adhesion (14,22, and 23%, respectively), for which i17 exhibited significantly lower adhesion performance. Therefore, strain i17 was discarded on statistical grounds, thus leaving only i53 and i106 as statistically identical for both criteria.
The next screening step relied on health benefit features, as per in vitro assessment. Antibiotic resistance and proteolytic activity could not be used as criterion since performance of isolates were very similar to each other ( Table 4). The isolates i53 and i106 produced EPS (+), exhibited antioxidant performance (ca. 20%), and cholesterol assimilation (ca. 78.5%), statistically equivalent to each other (see Table 5); hence, no strain was discarded from further analysis based on these criteria. Finally, such isolates revealed positive activity of advantageous enzymes (see Table 6): aminopeptidases, acid phosphatase, naphthol-AS-BI-phosphohydrolase, β-galactosidase, and αand β-glucosidase; along with negative results pertaining to enzymes associated with safety concerns for human health: trypsin, α-chymotrypsin, β-glucuronidase, and N-acetil-β-glucosaminidase.
Since differences were not substantial between strains, i53 and i106 remained the best candidates as a starter or adjunct culture possessing a high probiotic potential-on average, gathering the best probiotic features. Finally, the addition of strain i106 to Cobrançosa table olives had previously been tested in loco (at pilot scale) to ascertain its viability and effects upon the technological, microbiological, and physicochemical profiles along the fermentation process, as well as upon the sensory features thereof. Such experiments were run in parallel with a control using only the spontaneous microflora and supported the conclusion that it is technologically feasible to use those potentially probiotic strains as additives, without compromising the expected quality of the final product [50]. However, addition of that strain is likely to enhance the added value of the final product, for the associated potential benefits upon human health.

Bacterial Strains and Growth Conditions
Fourteen LAB strains belonging to genus Lactiplantibacillus, and species paraplantarum and pentosus were tested in this study (see Table 1); they were previously collected from Cobrançosa table olive brines [24]. These strains were identified by combining RAPD-PCR and 16S rRNA sequencing, according to the method described by Reis et al. [24]; 16S rRNA sequences of these strains were deposited in the NCBI GenBank database, under convenient accession numbers (see Table 1) [51].
In our laboratory, they were first screened for technological characteristics (i.e., ability to survive/grow at different salt concentrations, ability to survive/grow at high and low pH, capacity to degrade/assimilate oleuropein, and tendency to produce CO 2 ) (unpublished data); food safety features (i.e., mucin degradation ability, and haemolytic and DNase activities); and gastrointestinal survival. Such strains, stored in 15% glycerol at −80 • C, when not in use, were revived in Man, Rogosa, and Sharpe (MRS) agar (VWR Chemicals, Leuven, Belgium), at 30 • C for 48 h, then maintained at 4 • C for upward of one week. Prior to the experiments, they were subcultured in MRS broth (VWR), at 30 • C for 15 h, without shaking (i.e., overnight incubation), to attain the stationary phase. A strain of Lacticaseibacillus (Lc.) casei isolated from a well-known commercial brand of probiotic yoghurt was used as the probiotic reference strain; in addition, a Lactiplantibacillus (ex. Lactobacillus) pentosus strain B281 bearing probiotic properties, previously isolated from Greek table olives, kindly provided by the Laboratory of Microbiology and Biotechnology of Foods, Department of Food Science and Human Nutrition, Agricultural University of Athens [8].

Caco-2 Cell Adhesion
The method of Argyri et al. [52] was hereby followed. Caco-2 cells (2 × 10 5 ) (from European Collection of Cell Cultures, n. 86010202) were seeded in 12-well plates with Dulbecco's modified Eagle's medium-composed of 2 mM glutamine, 1% (w/v) nonessential amino acids, and 20% (v/v) fetal bovine serum, supplemented with 100 U/mL penicillin/streptomycin (Sigma Aldrich, St. Louis, MO, USA). Medium in the wells was replaced by fresh medium every 2-3 days, and monolayers maintained for 13 days in 7% CO 2 at 37 • C until lack of further (visible) differentiation. Before proceeding to Caco-2 monolayers, the growth medium in the tissue culture plates was withdrawn, cells were washed twice with 3 mL of Phosphate-Buffered Saline (PBS, i.e., 10 mM Na 2 HPO 4 , 1 mM KH 2 PO 4 , 137 mM NaCl, and 2.7 mM KCl (VWR), pH 7.2), and incubated at 37 • C for 1 h with 2 mL of RPMI-1640 without serum or antibiotics. Before performing the adhesion assay, all bacterial cultures were grown overnight until the stationary phase in MRS at 37 • C, centrifuged for 10 min at 4 • C and 3211× g (Model 1580R, GYROZEN, Gimpo, Korea), washed twice with sterile PBS, and finally diluted in RPMI-1640 medium (without serum or antibiotics) to 10 7 CFU/mL. Subsequently, 1 mL of bacterial RPMI suspension was added to each well (technical replicates), with each strain being assessed for adherence in triplicate for each experiment. Following co-incubation for 2 h at 37 • C under 5% CO 2 /95% air, the medium was removed, and the monolayer washed three times with 1 mL of sterile PBS. The Caco-2 cells were detached with 2 mL of 1% of Triton X-100 (v/v) (Sigma Aldrich, St. Louis, MO, USA) in PBS. Following incubation for 5 min at 37 • C, cell lysates were serially diluted and plated on MRS agar. The experiments were repeated three times (biological replicates) for all strains. Adherence performance, expressed as percentage, Ad(%), was calculated via Equation (1): where CFU f = number of bacterial cells that remained attached and CFU i = total number of bacterial cells initially added to each well.

Hydrophobicity
The hydrophobicity of the strain cell surface was evaluated by measuring cell adhesion to hydrocarbons such as o-xylene. As described by Abouloifa et al. [17], overnight cultures were centrifuged (3211× g, 10 min, 5 • C), washed twice in sterilized PBS, and re-suspended in 0.1 M sterilized KNO 3 (VWR) to an optical density at 600 nm, adjusted to 10 8 CFU/mL. Next, 3 mL of cell suspension was mixed with 1 mL of o-xylene (Sigma Aldrich), and the mixture left at room temperature (25 • C) for 10 min. It was then vortexed for 2 and 20 min to promote phase separation. The aqueous phase was collected, and its absorbance recorded at 600 nm. This assay was performed in triplicate for all strains (biological replicates). The cell surface hydrophobicity was expressed as percentage, H(%), using Equation (2): where OD i = optical density at start of assay and OD f = optical density at end of assay.

Autoaggregation
Following Abouloifa et al. [17], LAB cultures were harvested by centrifugation (3211× g, 10 min, 5 • C), after overnight incubation in MRS broth at 30 • C. The pellet was washed twice and re-suspended in PBS to obtain 10 8 CFU/mL. The initial absorbance at 600 nm was measured, and the bacterial suspension was then incubated at 30 • C for 5 h. An aliquot was collected from the surface of the sample, and its absorbance measured at 600 nm; this measurement was repeated by 24 h. All aforementioned experiments were repeated three times (biological replicates) for all strains. Autoaggregation performance was expressed as percentage, A(%), according to Equation (3): where OD i = optical density measured at start of assay (time 0) and OD f = optical density measured at end thereof (by 5 or 24 h).

Co-Aggregation
The methods by Montoro et al. [53] and Taheur et al. [54] were followed. LAB cultures were incubated overnight in MRS broth at 30 • C, without agitation. The pathogenic strains were incubated at 37 • C with agitation, in appropriate media: Bacillus cereus ATCC 11778 in Tryptic Soy broth with Yeast extract (TSY, VWR); Candida albicans ATCC 10231 in Yeast extract Peptone Dextrose broth (YPD, VWR); Escherichia coli ATCC 25922 and Salmonella enteritidis ATCC 25928 in Luria Broth (LB, VWR); and Enterococcus faecalis ATCC 29212, Listeria innocua ATCC 33090, and Staphylococcus aureus ATCC 25923 in Brain Heart Infusion (BHI, VWR). All cultures were harvested by centrifugation (3211× g, 10 min, 5 • C), washed and re-suspended in PBS, and adjusted to 10 8 CFU/mL. Then, 2.4 mL of a LAB suspension was mixed to 2.4 mL of pathogen suspension. As controls, 4.8 mL of each LAB strain, and 4.8 mL of each pathogenic strain were used. Absorbance at 600 nm of each bacterial suspension was measured at time 0. By 24 h of incubation at 37 • C, absorbance of the upper suspension was recorded. This assay was performed in triplicate for all strains (biological replicates). Co-aggregation performance, CA (%), was calculated via Equation (4): (4) where OD t0 = optical density measured at start of assay and OD t = optical density measured at end of assay, for the "mix" (of lactic acid bacteria and a pathogenic strain), "LAB" (lactic acid bacteria)", or "pat" (pathogenic strain).

Antibiotic Resistance
Susceptibility to antibiotics was measured by the agar disc diffusion method, using representatives from different groups (e.g., β-lactams, aminoglycosides), and comprising diverse mechanisms of action (e.g., inhibition of cell wall, inhibition of protein synthesis). The final selection was 30 µg/mL of chloramphenicol, 30 µg/mL of tetracycline, 10 µg/mL of ampicillin, 30 µg/mL of cephalothin, 15 µg/mL of erythromycin, 10 µg/mL of streptomycin, 30 µg/mL of vancomycin, and 5 µg/mL of ofloxacin (Oxoid, Thermo Fisher, Basingstoke, UK). Our testing followed Abouloifa et al. [17] and Szutowska et al. [55], with modifications. Lactic acid bacteria revived in MRS agar were inoculated in ca. 3 mL of 0.85% NaCl solution, to obtain a bacterial suspension of 10 5 CFU/mL. Then, 1 mL of each bacterial suspension was pour-plated on MRS agar, and discs containing the antibiotics were deposited on the plates after 15 min of diffusion; these were incubated at 37 • C for 24/48 h. The results were consubstantiated by diameter of inhibition zone around each antibiotic disc and the strains were categorized based thereon. To ascertain reliability of antibiotic discs used, the same procedure was repeated for E. coli and S. aureus, pour-plated in Muller-Hinton agar (VWR).

Antimicrobial Activity
The agar well diffusion method of LAB cell-free MRS broth was followed, as per Abouloifa et al. [56] and Mulaw et al. [57]. For this assay, overnight cultures of LAB strains were incubated in MRS broth with 1% (w/v) glucose (D+; Sigma Aldrich) at 30 • C. The pathogenic strains and growth media were the same as for the co-aggregation ability test, including 37 • C and stirring. Overnight cultures were centrifuged (3211× g, 10 min, 5 • C), washed twice in sterilized PBS, and re-suspended in the same buffer. All bacterial pathogenic strains were seeded in Mueller Hinton agar, to an initial optical density of ca. 0.22 at 600 nm. Pathogenic yeast Candida albicans was seeded at an initial optical density of ca. 0.4 at 600 nm. Afterwards, each well was loaded with 100 µL of sterile CFS of LAB cultures and incubated at 30 • C for 24 h for bacteria, or 48 h for yeast. To compare the effect of pH on this assay, all LAB cultures CFS were also neutralized with NaOH (5 M; VWR) and pour-plated with pathogens. The qualitative result of this test was reported as inhibition zones of indicator strains around the wells.

Antioxidant Activity
To assess the antioxidant activity, the decrease in concentration of 1,1-diphenyl-2picrylhydrazyl (DPPH; Sigma Aldrich) radical brought about by LAB strains was selected, following Abouloifa et al. [17]. LAB strains were grown overnight in MRS broth at 30 • C. The bacterial cells were harvested by centrifugation (3211× g, 10 min, 5 • C), and washed twice with a sterile solution of 0.9% NaCl; the resulting pellets were adjusted to 10 8 CFU/mL and re-suspended in the same solution. The cell suspension (0.5 mL) was transferred into a fresh tube, to which 1 mL of methanolic DPPH radical solution (0.1 mM) was also added. The mixture was mixed vigorously and incubated for 30 min at room temperature in the dark. The resulting solution was centrifuged at 9600× g (Model SorvallTM LegendTM Micro 17R, Thermo Fisher Scientific, Frankfurt, Germany) for 3 min at 5 • C; and 300 µL of the supernatant was then transferred into 96-well plates, to measure absorbance at 517 nm. The controls included plain sterile 0.9% NaCl and DPPH solution; the blanks contained only ethanol and cells. The experiments were repeated for all strains as three biological replicates, each with two technical replicates. To calculate the antioxidant activity AA(%), the percent reduction in DPPH was determined via Equation (5): where OD sample = optical density measurement of samples, OD blank = optical density measurement of blank (ethanol mixed with bacterial supernatant), and OD control = optical density measurement of control (DPPH and distilled water solutions) by the end of the assay.

Cholesterol Assimilation
The selection of strains with lowering-cholesterol ability was performed by in vitro tests using water-soluble cholesterol, Cholesterol-PEG 600-polyoxyethylene cholesteryl sebacate (Sigma Aldrich)-and resorted to colorimetry, with a commercial kit for cholesterol quantification (Sigma Aldrich). According to Benítez-Cabello et al. [46], overnight cultures of LAB were centrifuged (3211× g, 10 min, 5 • C), washed twice with a 0.85% NaCl solution, and re-suspended in PBS. After 2 h at room temperature (to simulate starvation), 20 µL of suspension was inoculated in 200 µL of MRS broth and supplemented with 3 g/L of oxgall (Sigma Aldrich) and 0.100 g/L of cholesterol. After incubation at 37 • C for 24 h, samples were centrifuged (9600× g, 3 min, 5 • C) and pellet was discarded. The cholesterol was quantified according to manufacturer's instructions, using a standard curve of absorbance at 500 nm vs. cholesterol concentration. MRS and oxgall without cholesterol solution were used as controls. The experiments were repeated for all strains as two biological replicates, each with two technical replicates. After knowing cholesterol concentration, according to said kit, cholesterol assimilation CA(%) was determined via Equation (6): where C i = initial cholesterol concentration and C f = final cholesterol concentration in the samples.

Proteolytic Activity
The proteolytic activity of isolates was assayed according to Peres et al. [18]. Overnight cultures were spot inoculated on skim milk agar, composed of 5 g/L casein (Sigma Aldrich), 2.5 g/L yeast extract (Merck), 1 g/L glucose (Sigma Aldrich), and 28 g/L skim milk. After 48 h of incubation at 37 • C, the clear zones around the colonies were checked and taken as indicator of proteolytic activity.

Exopolysaccharide (EPS) Production
The ruthenium red staining method, by Abouloifa et al. [17] and Anagnostopoulos et al. [58], was chosen to assess EPS synthesis. Starting with overnight cultures of LAB, 5 µL was deposited on the surface of plates containing ruthenium red milk medium composed of 0.5% yeast extract (Merck), 10% skim milk powder (Sigma Aldrich), 1% sucrose (Sigma Aldrich), 0.08 g/L ruthenium red (Sigma Aldrich), and 15 g/L bacteriological agar (VWR). The plates were incubated at 37 • C for 48 h, then colony colour was checked. Ruthenium red is a carbohydrate-binding dye used to stain biofilms formed by EPS-producing bacteria; as a result, the non-ropy strains should develop red colonies due to bacterial cell wall staining, while EPS strains should appear as white colonies.

Enzymatic Activity
The procedure followed to assay for enzymatic activity included API ZYM system (BioMérieux, Marcy-l'Étoile, France), for its coverage of a wide range of enzymes; the procedure was previously described by Abouloifa et al. [17] and Taheur et al. [54]. The overnight LAB cultures were centrifuged (3211× g, 10 min, 5 • C), washed, and re-suspended in 0.85% of NaCl solution, at initial optical density of 0.08 at 600 nm. Each well of API ZYM system was loaded with 65 µL of LAB suspension. After 4 h of incubation at 30 • C, reagents ZYM-A and ZYM-B were added to each well, according to manufacturer's instructions [47]. The enzymatic activity was evaluated based on APY ZYM system scale, as per the manufacturer; the results were first graded from 0 to 5, depending on colour intensity developed within 5 min vis-à-vis with colour reaction chart; then, the values obtained (0-5) were approximately converted to nanomole of hydrolysed substrate (0-40). Each strain was tested three times with the API ZYM kit, to ensure reproducibility of experimental results.

Statistical Analysis
One-way analysis of variance (ANOVA), followed by Tuckey Post-Hoc, equates to multiple comparisons being performed for epithelial adhesion capacity, autoaggregation and co-aggregation abilities, hydrophobicity degree, antioxidant activity, and cholesterol assimilation tests-with the aid of IBM SPSS 27.0 software (IBM, Armonk, NY, USA), at the 5% level of significance. The same software was used for Principal Component Analysis (PCA) with oblique rotation (direct oblimin), applied to Caco-2 cell adhesion, autoaggregation, and hydrophobicity performance; as additionally, determination of Kaiser-Meyer-Olkin (KMO) measure to check that sample size was adequate for analysis, and Bartlett's test of sphericity to unfold inter-correlations between variables. Furthermore, Canoco™ V5.0 (Microcomputer Power, Ithaca, NY, USA) was used to produce both plot of loadings and plot of scores from PCA applied to adhesion, autoaggregation and hydrophobicity, LAB RAPD profile, table olive fermentation time, and isolate identification.

Conclusions
All 14 strains of Lactiplantibacillus paraplantarum and Lactiplantibacillus pentosus tested, adventitious in Cobrançosa table olives, exhibited probiotic potential. Comparison with strains from a commercial brand of yoghurt or strains retrieved from Greek table olives unfolded a better performance of L. pentosus strains i53 and i106 in terms of functional properties. Antibiotic resistances were found to be intrinsic-thus discarding the possibility of (undesired) gene exchange with pathogens. The aforementioned strains were able to produce exopolysaccharides and exhibited antioxidant and cholesterol-decreasing activities-likely to bring health advantages to the host.
Limitations of this work were the assessment of antimicrobial and proteolytic activitieswhich proved not useful for screening, in view of the statistically similar results obtained; more quantitative analytical methodologies are thus recommended.
In future work, probiotic performance of L. pentosus i53 and/or L. pentosus i106 endogenous lactic acid bacteria strains should be confirmed by resorting to animal models; their inclusion in a starter culture or adjunct culture should be tested at pilot-scale, aiming at a more standardized commercial process.

Data Availability Statement:
The data presented in this study are available upon request from the corresponding author. The data are not publicly available due to generation thereof from samples from a commercial manufacturer; the original data and their matching to batches, accordingly, requires prior discussion with company CEO prior to topical release.