In vitro assessment of probiotic properties and gamma - aminobutyric acid production of selected lactic acid bacteria strains

This study aimed to perform an in vitro assessment of probiotic properties and to select lactic acid bacteria (LAB) strains that can produce the bioactive compound, gamma aminobutyric acid (GABA). A total of seven LAB strains of various species were examined for probiotic properties. These strains produced antimicrobial substances to inhibit the growth of target pathogenic bacteria. All selected LAB strains could withstand GI tract conditions. The strains showed bile salt hydrolase (BSH) activities. An antibiotic susceptibility test revealed that all examined strains were susceptible to ampicillin, chloramphenicol, erythromycin, gentamycin, penicillin and tetracycline, whereas they were resistant to kanamycin, norfloxacin, streptomycin and vancomycin. All strains of LAB in the current study gave high adhesion to Caco - 2 cells ranging from 23.32 - 75.48%. Moreover, all tested LAB cells could inhibit Salmonella enterica serovar Enteritidis invasion of Caco - 2 cells. Additionally, the tested LAB cells and cell - free supernatants (CFS) showed low cytotoxicity to inhibit the proliferation of Caco - 2 cells. Moreover, three LAB strains, L . curvatus FJNP12, L. plantarum SKKL1 and W. cibaria SFM5, clearly showed GABA production based on the TLC results. Additionally, these strains also yielded high amounts of GABA according to an HPLC method, 15.66, 11.76 and 5.62 g/L, respectively. The open reading frame (ORF) of the gadB gene of L. plantarum SKKL1 revealed that it was 1,407 bp in length, encoding a polypeptide of 469 amino acids, with a molecular weight of 53.56 kDa and an isoelectric point (pI) of 5.58 predicted using the ExPASy program. These strains could be used in starter cultures for value - added agricultural products.


Introduction
Probiotics are live microorganisms when consumed in adequate amounts confer health benefits (FAO/WHO, 2002). The potential advantages of probiotics include protection against diseases such as inflammatory bowel disease and cancer, mitigating lactose intolerance, lowering serum cholesterol levels, increased utilization of nutrients and decreased use of antibiotics (Guo et al., 2010). Normally, probiotic bacteria at a level of 10 6 CFU/mL or g in food and feed have been recommended for health benefits (Fuller, 1989). The favourable probiotic properties of lactic acid bacteria (LAB) starter cultures are numerous. They have to be non-pathogenic, have the ability to produce antimicrobial substances to inhibit pathogenic bacteria, and survive in the gastrointestinal (GI) tract. Additionally, they should produce bile salt hydrolase (BSH) to lower serum cholesterol levels and increase the utilization of nutrients. Moreover, they must be susceptible to antibiotics and enable decreased use of antibiotics (Guo et al., 2010), while producing no blood haemolysis, but allowing for adhesion to intestinal epithelium cells to inhibit pathogens and modulate the host immune response, as well as exhibiting in vitro hydrophobicity (Fuller, 1989;Ratanaburee et al., 2013a). Moreover, they should adhere to the intestinal mucus and epithelial cells to form a barrier to inhibit the colonization of pathogenic bacteria (Ehrmann et al., 2002). Probiotic LAB can be used as a substitute for antibiotic growth promoters to improve livestock growth and prevent disease. LAB are generally recognized as safe (GRAS) and are frequently used as probiotics. Lactobacillus and Bifidobacterium strains have been used due to their beneficial effects on host health. LAB has an important role in food fermentation and preservation either as natural microflora or as starter cultures added under controlled conditions (Pumriw et al., 2021). They can produce organic acids such as lactic acid, which lower pH and inhibit pathogenic bacteria (Halami et al., 2000;Cleveland et al., 2001). Gamma-aminobutyric acid (GABA) is a four-carbon free amino acid, a non-protein, that is widely distributed in microorganisms (bacteria, fungi, yeasts), plants, animals and vertebrates (Li et al., 2008). It has well-known physiological functions such as its function in neurotransmission, induction of hypotension, diuretic tranquillizer effects and as an inhibitory neurotransmitter in the central nervous system (Watanabe et al., 2011). Consequently, it is used for the treatment of sleeplessness, depression and autonomic disorders (Siragusa et al., 2007;Di Cagno et al., 2010). The use of GABA-producing LAB as starter cultures in fermentation processes could help to achieve the bioactivity of GABA. Currently, bio-synthetic production of GABA by LAB is done for the manufacture of fermented milk products, dairy products such as yoghurt, black raspberry juice, fermented soybeans, kimchi, Nham and cheese (Komatsuzaki et al., 2005;Park and Oh, 2007;Dhakal et al., 2012;Kim and Kim, 2012;Ratanaburee et al., 2013b;Shan et al., 2015). GABA-producing LAB has been reported. These include Lactobacillus namurensis, Pediococcus pentosaceus and Levilactobacillus brevis F064A isolated from fermented Thai foods (Ratanaburee et al., 2013b;Kanklai et al., 2021), L. futsaii isolated from Thai fermented shrimp (Sanchart et al., 2016), L. brevis isolated from Kimchi (Park and Oh, 2007), fresh milk (Huang et al., 2007), alcohol distillery lees (Yokoyama et al., 2002), yogurt (Nomura et al., 1999) and cheese (Siragusa et al., 2007). Lactococcus lactis has been isolated from cheese (Nomura et al., 1999;Park and Oh, 2007), L. paracasei from cheese (Siragusa et al., 2007) and Japanese traditional fermented fish (Komatsuzaki et al., 2005). L. delbrueckii subsp. bulgaricus and L. plantarum have been found in cheese (Siragusa et al., 2007). GABAproducing ability varies widely among the LAB strains (Yokoyama et al., 2002;Komatsuzaki et al., 2005;Huang et al., 2007;Park and Oh, 2007). However, the potential for using GABA-producing LAB probiotics has been of interest. LAB are microorganisms and thus attract more attention to GABA production (Li and Cao, 2010). Moreover, LAB can catalyze the αdecarboxylation of monosodium glutamic acid (MSG) using glutamate decarboxylase (GAD), resulting in the release of GABA and CO 2 (Higuchi et al., 1997). The production of GABA by LAB is safe and eco-friendly and this provides the possibility of the production of new naturally fermented health-oriented products enriched in GABA for use as food and feed. Selection of LAB that produces high levels of GABA and evaluation of the probiotic properties of LAB strains to identify single and mix-strains that may have potential uses as probiotics in food or animal production is undertaken. Moreover, fermentation offers the possibility to use low-cost agricultural waste substrates as raw materials for GABAproducing LAB strains. An objective of the current study is to examine the fermentative production of high levels of GABA. This is of interest for food and feed applications to obtain potential probiotic products and take advantage of the functional properties of GABA. This study aimed to assess the probiotic properties of selected LAB strains that are able to produce the bioactive compound, GABA. It further aimed to characterize the glutamate decarboxylase gene and its role in starter cultures for food and feed supplements.

Microorganisms and cultivation
A total of seven LAB strains were obtained from the culture collections of the Fermentation Research Center for Value Added Agricultural Products (FerVAAP) and the Department of Biotechnology, Faculty of Technology, Khon Kaen University, Thailand (Table 1). All selected LAB strains were stored in 50% (v/v) glycerol at -80°C. Each strain was reactivated by streaking it on de Man Rogosa and Sharpe (MRS) agar containing 0.5% (w/v) CaCO 3 , followed by incubation at 37°C for 24 hrs. A fresh culture was sub-cultured into MRS broth.

Antagonistic activity
An antagonistic activity test was performed using an agar well diffusion method. FULL PAPER target organisms. All of the targeted strains were grown in tryptic soy agar (TSA) at 37°C for 24 hrs. One loopful of each culture was inoculated into tryptic soy broth (TSB) followed by incubation for 24 hrs. Each culture broth was adjusted to approximately 1×10 6 cells/mL for swabbing onto TSA plates.
The selected LAB strains were separately grown in MRS broth and incubated at 37°C for 24 hrs. The pH values of the culture broths were measured. Cell-free supernatants (CFS) were obtained by centrifugation at 10,000 rpm for 10 mins and subsequent filtration through a 0.45 µm filter. A volume of 140 µL of each CFS was added to each well of the pathogen target seeded TSA plates. All plates were incubated at 37°C for 24 hrs before measuring inhibition zone diameters (mm). The results are reported indicating the degree of inhibition.

Acid, bile salt and fresh chicken bile tolerance
The method described by Vinderola and Reinheimer (2003) was used. Briefly, each strain grown in MRS broth at 37°C for 24 hrs was harvested by centrifugation at 10,000 rpm for 10 mins. The bacterial cells were washed twice and re-suspended in normal saline (0.85% (w/v) NaCl). Then, cells were tested for their ability to grow in MRS broth media at an acidic pH (pH 2.5) and in the presence of 0.3% (w/v) bile salts at 37°C for 3 and 24 hrs, to test their acid and bile tolerance, respectively. The tolerance of LAB to these factors was assessed in terms of viable colony counts on MRS agar containing 0.5% (w/v) CaCO 3 .
The fresh chicken bile tolerance method described by Gilliland et al. (1984) was used. An overnight grown culture of each LAB strain was inoculated into an MRS broth containing 3% (w/v) fresh chicken bile. Subsequently, the cultures were incubated at 37°C for 3 hrs. Growth under both a control treatment (without bile) and 3% (w/v) fresh chicken bile was determined through viable cell counts of LAB on MRS agar containing 0.5% (w/v) CaCO 3 .

Cell surface hydrophobicity
A test of adhesion between bacterial cells and hydrocarbon solvents, chloroform (a monopolar acidic solvent), xylene (nonpolar neutral solvent) and hexadecane (nonpolar solvent) was modified from Guo et al. (2010) and Pringsulaka et al. (2015). Briefly, each strain of LAB was grown in MRS broth at 37°C for 24 hrs and washed with sterile phosphate-buffered saline solution (PBS, pH 7), harvested and re-suspended in the same buffer. The suspension was then adjusted to approximately 1×10 8 CFU/mL (OD 640nm = H 1 ). Aliquots (3 mL) of each bacterial suspension were placed into 1 mL of each of the hydrocarbon solvents. The optical density of the aqueous phase (H 2 ) that resulted from thoroughly mixing the bacterial suspension with each of the solvents by vortexing for 2 mins was determined spectrophotometrically at 640 nm. The percentage of hydrophobicity was calculated using the following equation: LAB strains with greater than 70% hydrophobicity were classified as high, 36-70% were classified as moderate, and 0-36% were classified as having low hydrophobicity.

Mucin adhesion
The protocol for determining mucin adhesion followed Tallon et al. (2007) with modification. Briefly, mucin was dissolved in PBS (pH 7) to a final concentration of 10 mg/mL. 100 µL aliquots of this solution were immobilized in microtiter plate wells by overnight incubation at 4°C. Wells were washed twice with 200 µL of PBS and saturated with a 2% (w/v) bovine serum albumin (BSA) solution for 4 hrs at 4°C. Last, the wells were washed twice with 200 µL of PBS. A minimum of four replicates was used to estimate the adhesion of each strain.
LAB strains were grown separately in MRS broth at 37°C for 18 hrs and centrifuged at 10,000 rpm for 10 mins. The cell pellets were washed twice with sterile PBS, resuspended in PBS and adjusted to an optical density at 600 nm (OD 600nm ) = 0.1. Approximately, 100 µL of the bacterial suspension was added to each well. Then, the plates were incubated at 37°C for 1 hr. The wells were washed 12 times with 200 µL of sterile PBS to remove unbound bacteria. Each of the wells was then treated with 200 µL of a 0.5% (v/v) Triton X-100 solution to desorb the bound bacteria. Plates were then incubated at room temperature under orbital agitation for 2 hrs. Approximately 100 µL of the contents of each well were removed, diluted in PBS and plated onto MRS agar plates. The per cent adhesion was determined using Equation (2).
Where, A 0 = log number of free cells before adhesion to mucin per microtiter plate well (CFU/mL), and A 1 = log number of free cells in wells after adhesion to mucin per microtiter plate well (CFU/mL).

Haemolytic activity
The haemolytic activity of the selected LAB strains was evaluated by growing the cells in MRS broth and incubating them at 37°C for 24 hrs. Then, the cells were transferred onto blood agar plates. The plates were incubated at 37°C for 48 hrs and observed for subsequent

Bile salt hydrolase activity
Semi-qualitative bile salt hydrolase (BSH) activity of the cultures was determined with a procedure modified from Du Toit et al. (1998) and Guo et al. (2010). Briefly, 20 µL of cultures of each LAB strain grown overnight were spotted onto MRS agar plates supplemented with 0.37 (g/L) CaCl 2 and 0.5% (w/v) of human bile salts as the sodium salts of taurocholic acid (TC) and taurodeoxycholic acid (TDC). Plates were incubated at 37 o C for 48-72 hrs. The presence of precipitated bile acid around the colonies (opaque or halo) was considered indicative of a positive result.

Autoaggregation assay
A test for autoaggregation was performed with a method modified from Del Re et al. (2000). Briefly, a 0.1 mL cell suspension was mixed with 1.9 mL of PBS to prepare a 2 mL system containing approximately 1×10 9 CFU/mL. It was vortexed for 10 s and incubated at 37 o C for 24 hrs. 0.1 mL of the upper layer suspension was carefully collected, mixed with 0.9 mL of PBS and its optical density at 600 nm (OD 600nm ) was measured. Autoaggregation is expressed as 1 -(OD upper suspension/OD total bacterial suspension) × 100.

Coaggregation assay
Coaggregation between each LAB and pathogenic bacteria strain was determined following the method of Savedboworn et al. (2014). Briefly, equal volumes (2 mL) of each LAB and pathogen cell suspension were mixed in pairs by vortexing for 10 secs. Control tubes were set up at the same time, containing 4 mL of each bacterial suspension. The optical density (A) at 600 nm (OD 600nm ) of the suspensions was measured after incubation at 37 o C for 5 hrs. Samples were taken using the same procedure as in the autoaggregation assay. The percentage of coaggregation was calculated as: Where, x and y represent LAB and pathogenic bacteria, respectively and (x+y) represents the mixture of LAB and each pathogenic bacteria.

Antibiotic susceptibility
An agar disc diffusion method was used to test the antibiotic resistance of LAB strains with the following antibiotics: ampicillin (10 µg), chloramphenicol (30 µg), erythromycin (15 µg), gentamycin (10 µg), kanamycin (30 µg), norfloxacin (10 µg), penicillin G (10 U), streptomycin (10 µg), tetracycline (30 µg) and vancomycin (30 µg). An overnight culture of each LAB strain was prepared to obtain a cell suspension of approximately 1×10 6 CFU/mL. The selected LAB suspension was swabbed onto MRS agar plates. Later discs onto which various antibiotics had been absorbed were placed on the agar and incubated at 37°C for 24 hrs. The diameters of the inhibition zones were measured after incubation and compared with those in the interpretative standard chart of the Clinical and Laboratory Standards Institute (2014). The results were reported as resistant (R), intermediate (I) and susceptible (S).

Coexistence test
Coexistence among seven LAB strains was determined using a cross-streak method (modified from Guo et al. (2010) and Pringsulaka et al. (2015). Briefly, each of the four LAB strains was streaked perpendicularly from other streaked strains on MRS agar containing 0.5% (w/v) CaCO 3 . The plates were incubated at 37 o C for 48 hrs to observe the antagonism of bacteria against one another.

Adhesion of selected lactic acid bacteria to Caco-2 cells
Cell lines and culture conditions: Cells of a human intestinal epithelial cell line (Caco-2) are routinely grown in Dulbecco's modified Eagle's minimal essential medium (DMEM) supplemented with 10% (v/v) fetal calf serum (FBS) and 1% (v/v) penicillin-streptomycin (10,000 IU/mL). The cells were incubated at 37°C in a humidified atmosphere containing 5% CO 2 . Intestinal epithelial cells were seeded with 1 mL of a culture medium containing 1×10 5 cells/mL in 24-well tissue culture plates. The cell culture medium was changed every 2 days, and the cells were subcultured at 80% confluence. The intestinal epithelial cells were used at 14 days post-confluence after becoming fully differentiated. A medium of non-supplemented DMEM was replaced at least 1 hr before these assays.
Preparation of bacterial cells and cell-free culture supernatants: LAB strains were grown in MRS broth at 37°C for 24 hrs. Cell-free culture supernatant (CFS) was collected by centrifugation at 10,000 rpm for 10 mins. It was then filter-sterilized using a 0.45 μm pore size filter. The pelleted cells at concentrations of 1×10 8 CFU/mL were resuspended in an equal volume of serum-free DMEM.
Adhesion of selected LAB to Caco-2 cells: The LAB cultures in MRS broth were harvested by centrifugation at 10,000 rpm for 10 mins, washed twice with PBS and diluted to a concentration of approximately 1×10 8 CFU/ mL in serum-free DMEM (with no antibiotics). Concurrently, Caco-2 cells were also washed twice with PBS to remove antibiotics before any bacterial suspension was added. One millilitre of each LAB at a specified bacterial concentration was added to each well of a 24-well plate. The plates were incubated at 37°C for 1 hr under 5% CO 2 . Then, the supernatants were removed and wells were gently washed three times with PBS to remove unattached bacteria. Cells were lysed with 0.1% (v/v) Triton-X 100 for 5 mins. The adhering bacterial cells were placed in MRS agar and then incubated at 37°C for 48 hrs. Adhesion data were expressed as the percentage of bacteria adhering compared to the total of bacteria added (CFU bacteria adhered/CFU bacteria added).

Infection inhibition activity of selected lactic acid bacteria against Salmonella enterica serovar Enteritidis invasion of Caco-2 cells
The methods of Hudault et al. (1997) and Thirabunyanon and Hongwittayakorn (2013) were used with minor modifications. The enteropathogenic bacterium, S. enterica ser. Enteritidis and LAB strains were cultured overnight in TSB or MRS broth. Bacteria were harvested by centrifugation at 10,000 rpm for 10 mins at 4°C, washed twice with PBS and diluted to a concentration of 1×10 8 CFU/mL in DMEM (with no antibiotics). After that, non-supplemented DMEM was replaced at least 1 hr before the invasion assay. The intestinal epithelial cells were washed twice with PBS. Approximately, 1 mL of S. enterica ser. Enteritidis at approximately 1×10 8 CFU/mL was loaded into the intestinal epithelial cell culture monolayers in 24-well tissue culture plates. Then, one mL of each LAB culture in DMEM was added at a concentration of approximately 1×10 8 CFU/mL. These were mixed into the wells and incubated at 37°C for 2 hrs under 5% CO 2 . After that, the supernatants were removed and wells were gently washed three times with PBS to remove non -attached bacteria. Intestinal epithelial cells were lysed with 0.1% (v/v) Triton X-100. The adhering bacteria were counted on xylose lysine deoxycholate agar (XLD agar) plates. The viability of S. enterica ser. Enteritidis was used to estimate the invasion rates. Adhesion data were expressed as the percentage of bacteria adhered compared to the total bacteria added (CFU bacteria adhered/CFU bacteria added).

Cytotoxicity assay
The activity of LAB on the antiproliferation of Caco-2 cells was evaluated using MTT assays, as previously described by Thirabunyanon and Hongwittayakorn (2013). The MTT assay used in this experiment employed 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium (MTT). Caco-2 cells were prepared in 100 μL of a suspension at a density of 10 5 cells/well in a 96well plate. The cell suspensions were incubated in a CO 2 incubator at 37°C for 24 hrs. After incubation, 100 μL of CFS was added to each well and incubation continued for another 24 hrs under the same conditions. Cells were then washed twice with PBS. Next, 10 μL of MTT solution with 0.5 mg/mL dimethyl sulfoxide (DMSO) was added to each well and incubated at 37°C for 4 hrs. The precipitate of formazan was then solubilized by adding 100 μL of DMSO with continued incubation at 37°C for 5 mins. Optical density was measured at 595 nm (OD 595nm ) using a microplate reader for the cell viability calculation, while the absorbance of the control group, which used MRS broth, was taken as 100% cell viability. The percentage of cell viability = (sample OD/ control OD) × 100.

Selection of Gamma-aminobutyric acid-producing lactic acid bacteria 2.2.1 Determination of Gamma-aminobutyric acidproducing lactic acid bacteria
Each strain of LAB was inoculated into an MRS broth supplemented with 5% (w/v) monosodium glutamate (MSG). LAB strains were incubated at 37°C for 24 hrs and then GABA production of each LAB culture broth was determined using thin-layer chromatography (TLC) and high-performance liquid chromatography (HPLC) (Choi et al., 2006;Thwe et al., 2011;Ratanaburee et al., 2013a).

Cloning of the gadB gene from Lactobacillus
LAB that produced high amounts of GABA were selected for further study of the gadB gene. LAB genes were amplified using a primer set based on gadB from L. plantarum gadB LP pET26b, gadB_F (5ʹ-CATATGGCAATGTTATACGGTAA-3ʹ) and gadB_R (5ʹ GTCGACGTGTGTGAATCCGTATTTCT-3ʹ). Amplified fragments were ligated with a pT7blue vector. E. coli DH5α cells were transformed with the ligation mixture by a standard protocol (Sambrook et al., 1989). DNA sequences were analyzed using BLAST (http:// blast.ncbi.nlm.nih.gov/Blast.cgi). The molecular mass and the predicated isoelectric point (pI) of the corresponding GAD protein were obtained using the ExPASy program.

Statistical analysis
The data were expressed as a mean ± standard deviation. The experimental data were analyzed using a one-way analysis of variance (ANOVA) with SPSS.

Antagonistic activity
Selected LAB strains of various species (Lactobacillus plantarum SKKL1, L. curvatus FJNP12, L. pentosus SKKP2, Pediococcus pentosaceus SFM1, Weissella cibaria SFM5, W. halotolerans FFJSS5 and Enterococcus sp. FeHuE3) were tested for inhibitory action against pathogenic bacteria using an agar well diffusion assay. All LAB strains produced culture pH values in the range of 3.6-3.7. Antagonistic activities of their culture-free supernatants (CFS) are presented in Additionally, E. coli and S. aureus exhibited moderate tolerance (inhibition zone >15 mm). It has been recognized that some LAB produces antimicrobial substances, for instance, bacteriocins, diacetyl, hydrogen peroxide and short-chain fatty acids (Cleveland et al., 2001;Foulquié Moreno et al., 2006;Kantachote et al., 2010). These antimicrobial substances can inhibit the growth of other microorganisms. Accordingly, these LAB were subjected to antibacterial activity assays against representative intestinal pathogens. As expected, all LAB strains exhibited strong inhibitory activity against each of the tested pathogenic bacteria.
Antibacterial activity against pathogenic bacteria is an attractive property of starter cultures because it could help to achieve safe fermented foods (Kantachote et al., 2010). All LAB strains in these tests had final pH values below 4. They all could inhibit the target microorganisms, whereas culture filtrates at pH 6.5 of these strains did not inhibit pathogens. This indicates pH and organic acids might be the sole origin of antibacterial activity in culture filtrates. Therefore, it would be interesting to use these strains as starter cultures, as doing so could help to reduce the postacidification of products such as fermented ground pork.

Acidic pH, bile salt and fresh chicken bile tolerance
The results of acidic pH, bile salt and fresh chicken bile tolerance of selected challenge LAB strains are shown in Figure 1. All of the selected LAB strains could survive in the highly acidic condition (pH 2.5) in the range of 88.47-96.72%, and tolerate up to 0.3% (w/v) bile salt in the range of 67.42-78.74%. These LAB could tolerate 3% (w/v) fresh chicken bile in the range of 89.15 -97.28%. Evaluation at an acidic condition of pH 2.5 showed that almost all isolates had high survival and tolerance under these circumstances. Additionally, all isolates were shown to have good survival and tolerance in 0.3% (w/v) bile salt and 3% (w/v) fresh chicken bile in  The tests used for in vitro functionality screening were a determination of survival under conditions resembling the pH and digestive enzymes of the human oral cavity and GI tract (Saarela et al., 2000). LAB strains should survive in harsh GI tract environments of gastric juice at pH 2.5 (Kimoto et al., 2000) and bile (Begley et al., 2006) secreted into the human GI tract. Other beneficial functional aspects include adherence to an intestinal mucin layer, production of antimicrobial compounds and health-promoting enzymes such as βgalactosidase and BSH activity (Collado et al., 2008;Pithva et al., 2014). It is a precondition for a strain to become a part of the GI microflora to extend health benefits to a host (Shekh et al., 2016). Based on our results, all of the isolates survived under acidic conditions and high concentrations of bile salt treatments. The acidity (pH 2.5 to 3.5) of gastric secretions and bile salts destroys the majority of bacteria that enter the GI tract . However, many LAB showed an ability to survive in simulated GI tract fluid, which was not unexpected. These bacteria are well-known indigenous flora of the GI tracts of humans and animals. In this ecological niche, these LAB encounter bile salts (Bhardwaj et al., 2010).

Cell surface hydrophobicity
The percentage of cells adhering to chloroform (monopolar acidic solvent), xylene (nonpolar neutral solvent) and hexadecane (nonpolar solvent) demonstrated hydrophobic cell surface properties, as shown in Figure 2. LAB isolates with greater than 70% adhesion were classified as highly hydrophobic, 36-70% were classified as moderately hydrophobic, and 0-36% were classified as having low hydrophobicity. Results of the hydrophobicity tests revealed that the percentages of hydrophobicity of each solvent were chloroform at 4.88-95.93%, xylene at 0.76-44.72%, and hexadecane at 5.99-47.57%. It was demonstrated that high hydrophobicity was found only in chloroform.
High hydrophobicity (>70%) of cell surfaces was found in all strains of Lactobacillus (SKKL1, FJNP12 and SKKP2) when tested with chloroform. These strains showed moderate and low hydrophobicity when tested with xylene and hexadecane, respectively. The strain, P. pentosaceus SFM1, showed moderate hydrophobicity in all tested solvents, while the strains, W. cibaria SFM5 and Enterococcus sp. FeHuE3, exhibited low hydrophobicity in all tested solvents. W. halotolerans FFJSS5 also showed low hydrophobicity in the tested solvents, except in chloroform where moderate hydrophobicity was obtained. Interestingly, the tested strains showed similar hydrophobicity results in xylene and hexadecane. Bacterial adhesion to hydrocarbons indicates cell surface hydrophobicity plays a key role in non-specific interactions between microbial cells and mucus or epithelial cells (Osmanagaoglu et al., 2010). Hydrophobicity is generally thought to be correlated with bacterial adhesion to the intestinal mucosa. It has been suggested that bacterial cells with a high hydrophobicity usually form strong interactions with mucosal cells or adhere strongly to epithelial cells or mucus. Mechanisms of adherence to an epithelial surface involve receptor-specific binding, charge and hydrophobic interactions. These differences in cell surface hydrophobicity could be due to variations in the levels of expression of cell surface proteins among strains of a species, as well as due to environmental conditions that could affect the expression of surface proteins (Savedboworn et al., 2014). Bacterial adhesion to a host is the initial step for their colonization and can subsequently lead to commensalism or infectious diseases. This adhesion to host cell surfaces initially involves nonspecific binding such as hydrophobic interactions (Collado et al., 2007;Piwat et al., 2015), followed by specific interactions between bacterial surface components, such as lipoteichoic acid and surface layer proteins as well as their receptors on the host epithelial cells (Vélez et al., 2007;Zhang et al., 2013).

Mucin adhesion
The results of mucin adhesion testing are presented in Figure 3. All tested LAB strains showed high mucin adhesion, in the range of 73.17-84.89%. W. halotolerans FFJSS5 gave the highest value of 84.89%, followed by L. plantarum SKKL1 (78.69%), L. pentosus SKKP2 curvatus FJNP12 (75.87%), P. pentosaceus SFM1 (75.50%) and W. cibaria SFM5 (73.17%). The epithelial cells of the GI tract are covered by a layer of mucus that protects the epithelium from damage and pathogens. It also provides habitat and nutrients for the intestinal microflora (Tuomola et al., 1999). Mucus is the first barrier that ingested microorganisms confront in the gut. Thus, adhesion to this matrix is considered a prerequisite for colonization of the gut (Ouwehand et al., 2001). The major components of mucus are polymeric glycoproteins called mucins. They have been widely used for in vitro adhesion tests of probiotics (Tuomola et al., 2000;Matsumoto et al., 2002). Mucus is continuously renewed. Therefore bacteria that are unable to bind to the epithelium are eliminated.

Blood haemolysis
Results showed that all LAB strains showed γhaemolysis. None displayed α or β blood haemolysis. These LAB are not harmful to human red blood cells since they produce no haemolysis. This is a further sign of their potential utility for use as probiotic microorganisms.

Bile salts hydrolase activity
The results of BSH activities are shown that all LAB strains exhibited the same BSH activity. Almost all deconjugated taurocholic acid (TC) and taurodeoxycholic acid (TDC). Only W. cibaria SFM5 gave a negative result for TDC. One of the reasons that the tested LAB strains could survive in such conditions was their ability to hydrolyze various bile salts (TC and TDC) via BSH, decreasing the solubility of these salts. It is well-recognized that undissociated forms of bile salts can act as antimicrobial agents, such as organic acids. In this way, microbial cell membranes are altered by bile salts (Ratanaburee et al., 2013a). The above results suggest that all tested LAB strains might be able to compete in the human intestine. The selected LAB strains showed BSH activity in all strains tested. Guo et al. (2010) suggested that microbial BSH functions in the detoxification of bile salts as well as increasing the intestinal survival and persistence of the producing strains (Begley et al., 2006). Beneficial effects of BSHpositive LAB strains (such as lowering serum cholesterol levels) in vivo were also reported (Du Toit et al., 1998).

Autoaggregation ability
The autoaggregation ability of selected LAB strains was in the range of 69.07-84.81%, as shown in Figure 4. W. halotolerans FFJSS5 and Enterococcus sp. FeHuE3 gave the highest aggregation value, 84.81%, followed by W. cibaria SFM5 (80.00%), L. pentosus SKKP2 (78.12%), L. curvatus FJNP12 (74.59%), P. pentosaceus SFM1 (69.27%) and L. plantarum SKKL1 (69.07%). In the current study, all tested LAB strains showed high autoaggregation. As shown in Figure 4, most isolates exhibited strong autoaggregation after 24 hrs of incubation. Autoaggregation of probiotic strains refers to the clumping of bacterial cells from the same strain . This phenomenon appears necessary for adhesion to intestinal epithelial cells. Furthermore, autoaggregation could be useful in forming biofilms during GI tract colonization, which is a way to form a barrier against colonization by pathogens (Ramirez-Chavarin et al., 2013;Savedboworn et al., 2014). Aggregation and cell surface hydrophobicity was used to preliminarily screen for probiotic properties. This procedure has been proposed as an indirect method for the evaluation of the adhesion ability of LAB.

Coaggregation ability
The coaggregation between LAB strains and pathogens showed great variation. All LAB strains could coaggregate to all pathogens, event at the low percentages, as shown in Figure 5. Clearly, all LAB strains could coaggregate with S. aureus TISTR 1466 better than S. enterica ser. Typhimurium TISTR 292 and E. coli TISTR 780 (in the range of 21.99-26.70%, 7.71-12.37% and 5.84-9.67%, respectively). Interestingly, L. curvatus FJNP12 gave the highest coaggregation to all pathogenic bacteria, while the lowest coaggregation to all pathogens was found in L. plantarum SKKL1. These results were similar to Lapsiri et al. (2011) who reported that 13 strains of L. plantarum exhibited low coaggregation to E. coli O157:H7 DMST 12743 (3.68-19.17%), S. enterica ser. Typhimurium ATCC 13311 (3.85-12.16%). Alternatively, Savedboworn et al. (2014) revealed that 23 strains of LAB isolated from fermented vegetables showed coaggregation ability with E. coli (1.92-53.12%), and S. enterica ser. Typhimurium ATCC 13311 (3.25-53.28%) that were higher than in the current study. It has been suggested that probiotic microorganisms that have the ability to coaggregate with pathogens may be better able to kill undesirable bacteria. This is because they can produce antimicrobial substances in very close proximity to their target (Savedboworn et al., 2014). Coaggregation mechanisms between probiotic and pathogenic bacteria could be involved in a reduction in pathogen adhesion to mucus. Additionally, it has been suggested that the coaggregation of probiotic bacteria may enable the formation of a barrier to prevent colonization by pathogens.

Antibiotic susceptibility
All tested strains were susceptible to ampicillin, chloramphenicol, erythromycin gentamycin penicillin G and tetracycline. They were resistant to four antibiotics (kanamycin, norfloxacin, vancomycin and streptomycin) as reported in Table 3. Some strains (L. pentosus SKKP2, P. pentosaceus SFM1 and W. halotolerans FFJSS5) showed an intermediate level of streptomycin susceptibility. The susceptibility and resistance to antibiotics of various strains were variables depending on the species. LAB must show an ability to resist various antibiotics and subsequently exhibit beneficial effects on the health of the host to be used as probiotics (Lim and Im, 2009). Antibiotics are utilized by the medical and pharmacological industries to fight pathogenic bacteria. Resistance of probiotic strains to some antibiotics could be useful for both preventive and therapeutic purposes in controlling intestinal infections.

Adhesion to intestinal epithelial cells
The adhesion of seven LAB isolates (L. plantarum  Adhesion to intestinal epithelia is the main criterion for the selection of probiotic strains. Additionally, when probiotic cells adhere, they have beneficial effects on their host, such as when they are involved in preventing colonization of the gut by pathogens and maintaining gut mucosal immunity (Tancrède, 1992). In our study, all LAB exhibited an excellent ability to adhere to Caco-2 cells. The high adhesion of these strain cells indicates their ability to colonize the intestinal epithelium and represents a barrier that protects intestinal tissue from pathogens (Pringsulaka et al., 2015).

Antiproliferation activity (Cytotoxicity)
The potential actions of cells and cell-free supernatant (CFS) were correlated with the antiproliferation activity of Caco-2 cells by MTT assays. CFSs and cells gave similar results for antiproliferation activity. CFS activities of all strains to inhibit proliferation of Caco-2 cells were found significant (P<0.05) in each active isolate. The observed values ranged from 29.50-43.62% (Figure 8) with a low cytotoxicity. The highest antiproliferation activity was exhibited by L. plantarum SKKL1 (43.62%) followed by W. cibaria SFM5 (39.56%), L. curvatus FJNP12 (37.22%), P. pentosaceus SFM1 (36.03%), W. halotolerans FFJSS5 (36.00%) and Enterococcus sp. FeHuE3 (33.10%) (P<0.05). L. pentosus SKKP2 showed the lowest rate of cytotoxicity, 29.50%. Alternatively, no significant (P>0.05) antiproliferation activity was observed in cells of LAB strains. The observed values ranged from 23.60-33.15%. Higher rates were exhibited by L. plantarum SKKL1 (33.15%), followed by W. cibaria SFM5 (28.20%), L. curvatus FJNP12 (27.93%), W. halotolerans FFJSS5 (27.57%), P. pentosaceus SFM1 (26.04%), Enterococcus sp. FeHuE3 (25.41%) and L. pentosus SKKP2 (23.60%). The implication is that LAB strains may have anticancer actions, such as the antiproliferation of colon cancer cells. This may be important in interrupting the process of colon cancer cell formation. These probiotic LAB strains might be useful in colon cancer prevention and treatment. The mechanism for proliferative inhibition of colon cancer cells may be induced when probiotic LAB adhere FULL PAPER directly to these colon cancer cells and/or induce bioproduction of antimicrobial substances such as bacteriocins, reuterin, hydrogen peroxide, organic acids, and biosurfactants (Thirabunyanon and Thongwittaya, 2012;Thirabunyanon and Hongwittayakorn, 2013). This growth inhibition of enteric pathogens suggests that these LAB isolates might have a protective role against the initiation of colon cancer.

Gamma-aminobutyric acid production by selected lactic acid bacteria
Only three LAB strains showed clear GABA production based on the TLC chromatogram results. They included L. plantarum SKKL1, L. curvatus FJNP12 and W. cibaria SFM5 (Figure 9). The cell-free supernatants of P. pentosaceus SFM1, W. halotolerans FFJSS5, L. pentosus SKKP2, and Enterococcus sp. FeHuE3 strains did not show clear chromatograms indicating the presence of GABA. The amount of GABA after cultivation of LAB strains in MRS broth supplement with 5% (w/v) MSG detected by HPLC was statistically significant (P<0.05). The results of GABA production by selected LAB strains are shown in Figure  10. L. curvatus FJNP12 was the best GABA producer (15.66 g/L) followed by L. plantarum SKKL1 (11.76 g/ L) and W. cibaria SFM5 (5.62 g/L). The four poorest GABA-producing strains produced this substance in concentrations ranging from 0.30-0.99 g/L including Enterococcus sp. FeHuE3 (0.99 g/L), L. pentosus SKKP2 (0.41 g/L), W. halotolerans FFJSS5 (0.38 g/L), and P. pentosaceus SFM1 (0.30 g/L). These results were similar to those of other research groups (Park and Oh, 2007;Kim et al., 2009;Ratanaburee et al., 2013a), who previously reported that Lactobacillus and Lactococcus spp. produced high amounts of GABA. Since extracellular GABA production was observed in LAB culture broth, it may be possible to use these GABAproducing LAB as starter cultures in fermented foods. All GABA-producing LAB strains can potentially be used as starter cultures.

Nucleotide and deduced amino acid sequences of gadB
Lactobacillus plantarum SKKL1 and L. curvatus FJNP12, which produced high amounts of GABA, were selected to further study the gadB gene. gadB genes in LAB strains have been previously reported . The gadB gene was only found in L. plantarum SKKL1. Sequence analysis revealed that a cloned fragment of the gadB gene from L. plantarum SKKL1 contained a complete open reading frame (ORF) of 1,407 nucleotides, encoding a protein of 469 amino acids ( Figure 11) with a predicted molecular weight of 53.56 kDa and an isoelectric point (pI) of 5.58, as estimated using the ExPASy program. Moreover, the deduced amino acid sequence corresponding to the L. plantarum SKKL1 gad gene belongs to the gadB family of FULL PAPER pyridoxal dependent decarboxylases. Alignment of the deduced amino acid sequence with other GAD sequences revealed a highly conserved catalytic domain that belonged to the pyridoxal 5-phosphate (PLP)-dependent decarboxylase superfamily. The deduced amino acid sequence of the gadB gene from L. plantarum SKKL1 revealed a conserved lysine residue (Lysine280) that is known to be crucial for the binding of PLP, as well as the proposed active site residues (Threonine215 and Aspartate247) that promote decarboxylation. Additionally, the motif HVDAASGG, which is highly conserved in PLP-dependent decarboxylases, was found in the L. plantarum SKKL1 gadB gene sequence ( Figure  11). Lin et al. (2013) and Sanchart et al. (2016) reported that GABA synthesis was related to bacterial GAD, which plays a major role in the acid-resistance mechanism. Cytoplasmic decarboxylation resulted in the consumption of an intracellular proton after the uptake of glutamate by its specific transporter. The reaction product, GABA, was exported from cells by an antiporter, and the net result was an increase in the pH of the cytoplasm, owing to the removal of hydrogen ions, and a slight increase in the extracellular pH. This was due to the exchange of extracellular glutamate for the more alkaline GABA (Cho et al., 2014). Of the three acid resistance systems known in E. coli, the gad system is by far the most effective in conferring acid resistance to bacteria in the stationary phase. It gives them a survival capacity for at least 2 hrs in a strongly acidic environment (pH<2.5), such as that of the stomach (Capitani et al., 2003). It is hence conceivable that the gad system plays a similar role in lactobacilli. The presence of gad genes in LAB has been shown in several previous reports (Park and Oh, 2007;Fan et al., 2012;Sanchart et al., 2016). Its presence suggests an increased ability to produce GABA, since GAD must be active to convert glutamate into GABA. We cloned a DNA fragment from L. plantarum using primers for conserved regions of published gad genes. These results might provide useful information for the potential expression of a Lactobacillus gadB gene. The gadB gene characterization of L. plantarum SKKL1 should be useful in future work with this enzyme.

Conclusion
A total of seven LAB strains of various species were selected to determine their probiotic properties and production of the bioactive compound, GABA. They should be considered for further use as novel probiotic starters, either alone or in combination for food or feed production. All selected LAB strains studied contained probiotic properties such as inhibiting the growth of pathogenic bacteria, tolerance to GI tract properties, bile salt hydrolase (BSH) activities, none of haemolysis and adhesion to Caco-2 cells. Additionally, three strains, L. curvatus FJNP12, L. plantarum SKKL1 and W. cibaria SFM5, showed clear GABA production measured using a TLC method. Consequently, HPLC indicated that these strains produce high amounts of GABA, 15.66, 11.76 and 5.62 g/L, respectively. One of these strains, L. plantarum SKKL1, contained gadB gene. The ORF of the GAD gene was 1,407 bp in length, encoding a polypeptide of 469 amino acids with a predicted molecular weight of 53.56 kDa and an isoelectric point (pI) of 5.58. All selected LAB strains were found to have probiotic properties. Three strains produced high amounts of GABA and only one lactobacillus, which contained the gadB gene, would be further used as a starter culture for value-added agricultural products. The use of these three strains as probiotics in food is under investigation.

Conflict of interest
The authors declare no conflict of interest.

Acknowledgments
The first author received a student fellowship from the Rajamangala University of Technology Suvarnabhumi. The authors are grateful to Fermentation Research Center for Value Added Agricultural Products Figure 11. Nucleotide and deduced amino acid sequence of gadB gene from L. plantarum SKKL1. The box represents the amino acid residues HVDAASGG motif, which is highly conserved in PLP-dependent decarboxylases. The presumed catalytic amino acid residues are Threonine215, Aspartate247 and Lysine280. The amino acid residues in the dashes box are the PLP binding domain.