Probiotic potential characterization and clustering using unsupervised algorithms of lactic acid bacteria from saltwater fish samples

This research aimed to isolate lactic acid bacteria from the bowel of saltwater fish to assess their potential probiotic properties. Nineteen isolates of LAB including Lactiplantibacillus plantarum, Lactiplantibacillus pentosus, Lactobacillus acidophilus, Levilactobacillus brevis, Pediococcus pentosaceus, and Pediococcus acidilactici were recognized using molecular tools. All the isolates survived in the simulated conditions of the GI tract. Auto-aggregation ranged from 01.3 ± 0.5 to 82.6 ± 1.4% and hydrophobicity with toluene ranged from 3.7 ± 1.6 to 69.4 ± 1.3%, while the range of hydrophobicity with xylene was from 02.2 ± 1.6 to 56.4 ± 2.1%. All the isolates of lactobacilli, pediococci, enterococci, and lactococci indicated variable sensitivity and resistance towards clinical antibiotics. Non-neutralized cell free supernatant of isolates F12 and F15 showed antimicrobial activity against all the 8 evaluated enteric pathogens. Cluster analysis of identified potential probiotic bacteria based on heat-map and PCA methods also highlighted the priority of isolates F3, F7, F12, and F15 as bio-control agents in fishery industry. The findings of this study may essentially contribute to the understanding of the probiotic potential of LAB in saltwater fish, in order to access their probiotic characterization for use as biocontrol in fishery.

Acid and bile resistance. The MRS broth with pepsin enzyme (3 mg/mL) was utilized as a medium to assess acid resistance. The pH of broth was adjusted to 2.5 with 1.0 N HCl and broth (pH 7.0) was used as a control. Moreover, the broth was inoculated for 3 h and the optical density (OD) was measured at 600 nm. The acid resistance was determined by the following equation: Survival Rate (%): [OD (After treatment)/OD (Before treatment)] × 100%. Strains with viability more than 80% were selected for further analysis.
The ability of isolates to grow in presence of bile salt was measured. This was carried out with 0.3% w/v oxgall and Control was maintained employing MRS broth. The samples were inoculated at 37 °C for 4 h and OD 620 of samples was measured to check the viability of cells. The oxgall resistance was determined by the following equation: Survival Rate (%): [OD (After treatment)/OD (Before treatment)] × 100%.
The viability of selected strains in simulated gastrointestinal conditions was measured based on method described by Nami et al. (2019) 17 . For this purpose, cells cultured overnight were collected by centrifugation at 1500 × g at 4 °C for 15 min. The cell pellet was washed twice with neutralized PBS solution and resuspended in simulated gastric juice (SGF) and simulated intestinal juice (SIF) with 10 8 colony forming units CFU ml −1 . Sterilized PBS (pH 3) supplemented with pepsin (3 mg ml −1 ) and sterile PBS (pH 8) supplemented with SGF and pancreatin (1 g L −1 , Sigma Aldrich, St. Louis, Missouri).) Was prepared., USA) and bovine bile (0.3 g L −1 ) were used to prepare SIF. The resistance profile was examined by counting colonies on the MRS agar plate after 0, 1, 2, 3 h for SGF and after 0, 1, 2, 3, 4 h for SIF experiments.
Antimicrobial activity. The antimicrobial features of isolates were performed by agar well diffusion assay.
In this method, plates containing Mueller-Hinton agar medium impregnated with different indicator bacteria was used (Table 1). A sterile pasteurized pipette was used to create wells in medium. Finally, 100 µL of supernatants of isolates were placed inside each wells and plates were then incubated at 37 °C for 24 h. After incubation time, the measured inhibition halo zone diameters were statistically analyzed 22 . Antibiotic susceptibility. The disc diffusion method was performed for evaluation of the susceptibility of isolates to some high-consumption and clinically important antibiotics listed in Table 2. For this purpose, isolates were streaked over the solidified MRS medium and antibiotic disks were placed on the medium and incubated overnight at 37 °C. Finally, the diameters of inhibition zone around disks were measured by digital caliper 23 .
Cell surface hydrophobicity. The adhesion ability of isolates to xylene and toluene was determined as previously described by 24 . The test was performed in triplicate and expressed as: where A 0 represents the initial absorbance at 600 nm before adding xylene and toluene; A 1 represents absorbance after 4 h inoculated with xylene and toluene.
Autoaggregation assay. The ability of isolates to auto-aggregate was performed according to the method described by 25 . Auto-aggregation percentage was determined using the following equation:  F1  0  17  14  17  19  0  21  22   F2  8  0  21  20  0  17  19  13   F3  14  18  10  0  16  13  22  20   F4  0  15  17  9  21  16  16  19   F5  16  0  11  12  9  18  13  14   F6  11  13  20  8  0  21  17  14   F7  22  21  19  0  12  0  10  18   F8  13  11  11  11  17  12  18  0   F9  0  0  16  15  19  9  22  14   F10  9  10  14  0  0  8    www.nature.com/scientificreports/ Adhesion ability to human intestinal cells. Selected isolates were explored for their attachment capacity to the human colon carcinoma cell line Caco-2. RPMI medium supplemented with 10% heat-inactivated fetal bovine serum and 1% penicillin-streptomycin blend were utilized and cells were refined on 24-well tissue culture plates and incubated at 37 °C in 5% CO 2 under a relatively humidified atmosphere until a confluent monolayer was formed. Sometime recently the attachment measure, the media within the wells containing a Caco-2 cell monolayer were evacuated and supplanted with new antibiotic-free RPMI. From that point, 1 × 10 7 CFU/mL of isolates was included to each well with a add up to volume of 1 mL and after that incubated for 3 h at 37 °C under an atmosphere of 5% (v/v) CO 2 . The wells were washed twice with a sterile pre-warmed PBS solution to evacuate non-attached bacterial cells. One mL of 1% (v/v) Triton X-100 was included to each well to withdraw the cells from the wells and the blend was mixed for 10 min. To measure the viable cell count, the cell suspension was plated onto MRS agar and hatched at 37 °C 27 .
Cholesterol assimilation. The cholesterol-removing capacity of the isolate was determined using the method previously described by Nami et al. 28 . Isolates were inoculated for 20 h into MRS broth supplemented with water-soluble cholesterol (polyoxyethanyl cholesteryl sebacate; Sigma) and 0.3% bovine bile at a concentration of 150 μg/mL at 37 °C. After the incubation period, the cells were centrifuged at 4250 × g for 15 min to obtain cell pellets, and the cholesterol remaining in the upper layer was measured by the o-phthaldehyde method.
Biofilm formation. The ability of isolates to form biofilm was determined according to method described by 29 with some modifications. For this purpose, the wells of a sterile 6-well tissue culture plate were filled with 5 mL of MRS broth supplemented with 500 µL of overnight isolates (10 7 -10 8 CFU/mL). Cultures were incubated anaerobically for 48 h at 37 °C. After this period, the wells were gently washed three times with 5 mL of sterile distilled water. Then, 3 mL of methanol was used for 15 min to fix the attached bacteria and then plates were emptied and dried at room temperature. Afterwards, 3 mL of a 2% (v/v) crystal violet solution was poured in the wells and held at room temperature for 5 min. Finally, to release stain from Adherent cells, 2 mL of 33% (v/v) glacial acetic acid was used and the optical density (OD) of each well was measured at 595 nm using a platereader (Microplatereader, Bio-Rad, Hercules; CA, USA).

Exopolysaccharide (EPS) production.
To assess EPS production ability of isolates, method described by 30 with slight modification was used. Briefly, modified MRS agar medium (replacing glucose with 100 g/L of sucrose) was used and isolates were streaked on cultures and incubated at 37 °C for 24 h. Metal loop was used to drag up formed colonies. If the length of slime was above 1.5 mm, the isolate was considered positive slimy producers.

Molecular identification.
Genomic DNA extraction. The total genomic DNA of the isolates was extracted through the procedure designed in our laboratory. For this purpose, a single colony of each isolate was placed in 0.2 tube and 20 µL lysis buffer was poured inside the tube. The tubes were gently vortexed until the contents were completely homogeneous and were placed for one hour at room temperature. Then, they were incubated in thermal cycler PTC 200 (MJC research, Waltham, USA) for 10 min at 85 °C. After this period, 150 µL deionized water was added to tubes and they were centrifuged at 8000 × g for 5 min. Finally, the upper phase containing genomic DNA was removed and poured in new tubes. The genomic DNA was stored at refrigerator until using.
Amplification of 16S-rRNA gene by polymerase chain reaction (PCR). The amplification of genomic DNA samples of the isolates was performed in a thermal cycler PTC 200 by using a pair of primers (Hal-6F: 5'-AGA GTT TGATCMTGG CTC AG-3') and Hal-R6: 5'-TAC CTT GTT AGG ACT TCA CC-3') 31,32 . The following temperature profile was used to amplify DNA: an initial denaturation at 95 °C for 5 min, followed by 32 cycles of denaturation at 94 °C for 60 s, annealing at 59 °C for 60 s, extension at 72 °C for 60 s, and a final extension step at 72 °C for 5 min. Electrophoresis in a 0.8% (w/v) agarose gel was used to resolve the PCR products and visualized via ethidium bromide staining.
16S-rRNA gene sequencing. The PCR products of the 16S-rRNA gene (1544 bp) were amplified using the mentioned primer set. The PCR products were sequenced at Sinaclone Corporation, Tehran, Iran. The sequences were then analyzed using the BLAST program of the National Center for Biotechnology Information (http:// www. ncbi. nlm. nih. gov/ BLAST).

Statistical analysis.
To determine the significant differences between the parameters of each isolates Unsupervised clustering algorithms. The heat-map was generated using Euclidean distance measure and Ward clustering methods implemented in ggplot2 package of R software as prescribe in 33 . For unsupervised clustering of samples based on desired probiotic characteristics, principal component analysis (PCA) method was used. This algorithm provides a means to achieve unbiased dimensionality reduced structure of studies strains 34 .

Results
Morphological and biochemical assays. A total of 31 rod-and cocci-shaped colonies ( Fig. 1) were grown and isolated on culture media. Each colony was separately propagated for assessments. Based on Grampositive and catalase-negative assays, 19 isolates were identified as LAB isolates and selected for further analysis.
Acid and bile tolerance. Figure 2 represents the survival rates of 19 rod-and cocci-shaped isolates after 3 h of incubation at pH 3.0. Based on the findings, all 19 isolates retained their viability and percent survival rates of evaluated isolates ranged from 26 to 100% after 3 h of incubation at 37 °C (data not shown). The isolates with the most efficient resistance to low pH conditions were isolates F3, F12, F7, and F15 with survival rates of 99, 96, 94, and 92%, respectively. Conversely, isolates F4, F9, and F11 indicated the lowest viability after 3 h, with survival rates of 26, 29, and 32%, respectively. The survival rates of 19 isolates in 0.3% oxgall are indicated in Fig. 2. All 19 isolates appeared resistant to high bile salt condition, despite variations in the degree of viability. Percent survival rates of evaluated isolates ranged from 31 to 100% after 4 h of incubation at 37 °C. The highest tolerance was observed for isolate F7, followed by isolates F3, F15, and F12 with survival rates of 100, 95, 93, and 92%, respectively. Isolates F11, F17, and F9 showed the lowest viability after 4 h of incubation in media containing 0.3% oxgall, with survival rates of 31, 33, and 36%, respectively. Antimicrobial activity. Table 1 Table 2, isolate F12 was susceptible to all the tested antibiotics, while isolate F18 was resistant to all the tested antibiotics. The most widespread resistance among the isolates was observed against gentamycin, penicillin, and ciprofloxacin which were detected in 57.9% of the isolates, while the least resistance was observed against tetracycline which was detected only in isolate F18.
Autoaggregation assay. Results of the autoaggregation of the 19 selected isolates are illustrated in Table 3.
Values of autoaggregation ranged from 1.3 ± 0.5% to 82.6 ± 1.4% after 4 h of incubation. The highest autoaggregation values was observed for isolate F7, with value of 82.6 ± 1.4%, while the least value was obtained for isolate 11, with 1.3 ± 0.5% autoaggregation after 4 h of incubation.
Coaggregation assay. Biofilm formation. The ability of isolates to form biofilm in MRS medium was illustrated in Fig. 3.

Molecular identification.
The 16S-rRNA gene sequencing was used to validate the phenotypic characterization of the selected LAB isolates. The PCR-amplified 1544 bp fragments of the 16S-rRNA gene (Fig. 1)  Clustering analysis using heat map and PCA methods. Identified and characterized strains with desired probiotic features such as acid and bile tolerance, cell surface hydrophobicity, auto-aggregation and coaggregation were subjected to cluster analysis based on unsupervised methods. Unsupervised methods are types of learning algorithm used to knowledge discovery from datasets that are neither classified nor labeled [35][36][37] . As depicted in Fig. 4, Identified strains clustered in three distinct groups by Ward and PCA methods. It is obvious www.nature.com/scientificreports/ cluster I in heat map included the isolates F3, F7, F12, and F15, whereas; remaining nine strains created the other two clades. The projection of identified strains diversity based on the PC1 and PC2 highlighted the diversity of identified bacteria. In total, about 97% of the variation in the characterized bacteria was explained by two principal components (Table 3). More interestingly, congruence was displayed between the results of two applied methods in clustering the identified bacteria based on probiotic characteristics.

Discussion
There are several criteria for bacteria to recognize as probiotic. One of the most important features is tolerance to low pH and high bile salt concentrations 6 . They must overcome physical and chemical barriers in the gastrointestinal tract to be able to transit through the stomach. Probiotics to be able to survive and grow and could exert their action in the small intestine must be tolerant to high bile salt conditions 38 . In this work, the evaluated isolates indicated varying values of viability in pH 3 and 0.3% bile oxgall. The isolates F3, F12, F15, and F7 showed more than 90% viability at pH 3 and 0.3% bile oxgall. High survival rates of LAB especially L. plantarum 24,39 , and P. pentosaceus and P. acidilactici [40][41][42] at gastric pH of 3.0 and 0.3% bile concentration have been reported. The antimicrobial properties of probiotics are attributed to their metabolite productions including organic acids, hydrogen peroxide, and bacteriocin 43 . In this study, isolates were investigated against two Gran-positive (St. iniae and St. agalactia) and six Gram-negative (V. harveyi, V. cholera, V. alginolyticus, V. parahaemolyticus, and Cl. cochlearium) indicators. Evaluated isolates showed better antibacterial effects on Gram-positive than Gram-negative indicators, which this could be due to the outer membrane of Gram-negative bacteria. All the 19 isolates inhibited the growth of St. iniae and St. agalactia. Isolates F12 and F15 showed the best results and were able to inhibit the growth of all the 8 evaluated indicators.
Another strange property of probiotics is their ability to form biofilms. It can promote colonization within the inner wall of the intestine and longer lasting quality of LAB, thus maintaining a strategic distance from colonization by pathogenic microorganisms 44 . The ability of form biofilm is of great economic importance to food-processing companies, as subsequent microorganisms attached to industrial devices can form highly resistant biofilms, and their strong tendency to adhere to the surface facilitates the separation of biomass during the fermentation process 45 . Isolates F3, F7, F12, and F15 showed the strong biofilm formation ability; cause them to be suitable for use as biocontrol agents.
There are many studies, which investigated the cholesterol reduction ability of LAB isolates 28,46,47 . The reduction of serum cholesterol values is important to prevent the coronary heart disease due to the main risk factor for the progression of coronary heart disease is hypercholesterolemia 48 . Isolates F12, F3, and F7 showed remarkable cholesterol removal ability, with values of more than 59%.
For selection of functional LAB isolates, two groups of antibiotics are recommended; first those are inhibitors of cell wall synthesis such as vancomycin and ampicillin, and second those are inhibitors of protein synthesis including erythromycin, chloramphenicol, clindamycin, gentamycin, kanamycin, tetracycline, and streptomycin 49 . In this study, the LAB isolates indicated susceptibility to chloramphenicol (89.47%), tetracycline (94.73%), erythromycin (84.21%), ampicillin (84.21%), azithromycin (52.63%), gentamycin (42.11%), streptomycin (57.89%), penicillin (42.11%), vancomycin (52.63%), and ciprofloxacin (42.11%). The results of this study is consistent with the previous results which showed that LAB are susceptible to antibiotics inhibiting the proteins synthesis, such as chloramphenicol, tetracycline, and erythromycin, and are resistance towards glycopepticides   50 , and aminoglycosides antibiotics including streptomycin, kanamycin, and gentamycin 51 . Several LAB species belonged to especially Lactobacillus, Pediococcus and Leuconostoc have been reported to exhibit high levels of natural resistance to vancomycin, which is useful to separate them from other Gram-positive bacteria 41 . A desirable property of probiotic bacteria is colonization in intestinal wall 28 . Hydrophobicity, auto-aggregation and adhesion to epithelial cells could help probiotics to colonize in intestinal cells. The hydrophobicity of the bacterial cell surface is one of the most important factors for probiotics to be colonized in human and animal gastrointestinal tracts. In this study, the cell surface hydrophobicity percentages of the isolates ranged from 3.7 ± 1.6 to 70.2 ± 2.0% with toluene and 2.2 ± 1.6% to 76.4 ± 2.1% with xylene. This result confirms the results of several previous researches reported by [52][53][54] . Four out of 19 isolates showed both considerable cell surface hydrophobicity and high auto-aggregation (> 67%), which indicated good colonization potential 55 . According to Vallejo et al. (2008), strains with values of autoaggregation higher than 65% are strongly autoaggregating. Isolates F3, F7, F12, and F15 showed remarkable autoaggregation ability; making these isolates interesting candidates for potential probiotic applications. The capacity to attach to epithelial cells and mucosal surfaces may be a critical feature of bacterial isolates to be utilized as probiotics. Results showed that only isolates F3, F7, F8, F12, and F15 were able to adhere to Caco-2 cells. The adhesion ability of these 4 isolates to Caco-2 cell had also significant coherence to auto-aggregation and hydrophobicity. The adhesion rates of these isolates to Caco-2 cells were between 23 and 38%. It seems that adhesion ability is correlated with auto-aggregation 46 . These results also support the correlation between adhesion and auto-aggregation.
In conclusion, several isolates evaluated in this study showed potential probiotic properties. The majority of evaluated isolates showed high antimicrobial activities against potentially pathogenic Gram-positive and Gram-negative indicator bacteria, which deserve them to be used as biocontrol agents in fishery. The highest values of autoaggregation and hydrophobicity as well as a high value of coaggregation and a high capacity for biofilm formation, isolates F3, F7, F12, and F15 could be candidates to be used as bio-control agents in fishery to protect against pathogenic microorganisms and improve health care and food safety by avoiding the use of additives. Cluster characterization of identified potential probiotic bacteria based on two above-mentioned methods also highlighted the priority of isolates F3, F7, F12, and F15 as bio-control agents in fishery industry. In line with our study, prior finding also highlighted the effectiveness of above-mentioned methods for characterization and prioritization of newly identified isolates 33,56 . In addition, our empirical and cluster analysis based on unsupervised methods prioritized some isolates as biocontrol agent candidates in fishery industry, however; further analysis should be performed.