Bioprospecting of probiotics with antimicrobial activities against Salmonella Heidelberg and that produce B-complex vitamins as potential supplements in poultry nutrition

The demand for animal protein for human consumption has been risen exponentially. Modern animal production practices are associated with the regular use of antibiotics, potentially increasing the emerging multi-resistant bacteria, which may have a negative impact on public health. In poultry production, substances capable of maximizing the animals’ performance and displaying an antimicrobial activity against pathogens are very well desirable features. Probiotic can be an efficient solution for such a task. In the present work, lactic acid bacteria (LAB) were isolated from chicken cecum and screened for their antagonistic effect towards many pathogens. Their capacity of producing the B-complex vitamins folate and riboflavin were also evaluated. From 314 isolates, three (C43, C175 and C195) produced Bacteriocin-Like Inhibitory Substances (BLIS) against Staphylococcus aureus (inhibition zones of 18.9, 21.5, 19.5 mm, respectively) and also inhibited the growth of Salmonella Heidelberg. The isolate C43 was identified as Enterococcus faecium, while C173 and C195 were both identified as Lactococcus lactis subsp. lactis. Moreover, the isolates L. lactis subsp. lactis strains C173 and C195 demonstrated high potential to be used as probiotic in poultry feed, in addition to their advantage of producing folate (58.0 and 595.5 ng/mL, respectively) and riboflavin (223.3 and 175.0 ng/mL, respectively).


Materials and methods
Sampling of chicken cecum and LAB isolation. The screening for interesting LAB was conducted based on the protocol described by Messaoudi et al. 41 . Broiler chickens (Gallus gallus domesticus) in our study were fed with animal protein and AGP-free commercial feed and water ad libitum. They were maintained under controlled management conditions (diet, room temperature, cleaning). When animals reached 12-weeks of age, they were sacrificed with an anesthetic overdose (barbiturate). The cecum was carefully removed from the carcasses under aseptic conditions and transported to the laboratory in a refrigerated thermal box (~4 °C). Inside a laminar airflow biological cabinet, 10 g of it was transferred to a sterile plastic bag containing 90 mL of 0.1% (w/v) bacteriological peptone water (Sigma-Aldrich, Missouri, USA). It was homogenized for 2 min in a Stomacher device (SPlabor, São Paulo, Brazil), and the resulting product was serially diluted in sterile 0.85% (w/v) NaCl solution. One hundred μL of each dilution was transferred to Petri dishes (90 × 15 mm) containing Man, Rogosa and Sharpe (MRS) agar medium (Difco, Michigan, USA), Bifidus Selective Medium (BSM) agar (Sigma-Aldrich), and M17 agar medium (Difco, Michigan, USA) supplemented with 0.1 g/L of cycloheximide (Inlab, São Paulo, Brazil), as selective culture media for Lactobacillus, Bifidobacterium, Lactococcus genus, respectively. Using a sterile Drigalski handle, the dilutions were spread over the surface of the agar medium plates and, anaerobically incubated for 48 h at 30, 37, or 45 °C in a microaerophilic jar using a GasPak EZ Container System (BD Diagnostic Systems, Maryland, USA) to ensure hypoxia. After this period 314 colonies showing different morphologies were randomly selected and subcultured onto the aforementioned agar media plates, which were used in the following steps.
This study The study was approved by Ethics Committee on the use of animals of School of Veterinary Medicine and Animal Science in the University of São Paulo (registration number 7843030717). All experiments were performed in accordance with relevant guidelines and regulations.
Screening for LAB with antimicrobial activity against S. Heidelberg. S. Heidelberg strains IOC 969/17, isolated from chicken carcass sample was provided by FioCruz (Rio de Janeiro, RJ, Brazil) and used as bioindicator strain to evaluate the antimicrobial activity of the 314 isolates. The bioindicator strain was cultured in Brain Heart Infusion (BHI; Difco, Michigan, USA) broth for 18 h at 37 °C and 100-fold diluted (to approximately 10 6 CFU/mL). This suspension was used to inoculate BHI soft agar (0.75% [w/v] of agar) at 1/10 ratio (v/v) and the resulting mixture was used to overlay the agar media plates containing the 314 isolates. The plates were incubated at 37 °C for 24 h. Isolates showing the largest inhibition zones against S. Heidelberg IOC 969/17 were selected and subcultured in the appropriated medium (BHI, M17 or BSM) and temperature (30, 37 or 42 °C), according to the isolates preferences. They were first characterized morphologically by microscopy, Gram staining, and the detection of catalase activity. Gram-positive, catalase negative isolates were considered as potential LAB and used in further studies. All the isolates were stored at −70 °C using 20% (v/v) glycerol as cryopreservative.
evaluating the nature of the antimicrobial activity. To evaluation of the nature of the antimicrobial activity demonstrated by the selected isolates against S. Heidelberg IOC 969/17 was performed using previously established protocols 42 . The isolates were cultured in their preferential media broth and temperature for 24 h, then centrifuged (Boeco, Hamburg, Germany) at 12,000 g at 4 °C for 15 min. From each culture broth, the supernatant was recovered and the pH adjusted to 6.0 -6.5 with 1.0 M NaOH. The samples were filtered through 0.22-μm hydrophilic PVDF membranes (Millipore, Maryland, USA), resulting in cell-free supernatants labeled as CFS-A. Then, these CFS-A samples were investigated regarding their antimicrobial effect against S. Heidelberg IOC 969/17.
The potential inhibitory effect of CFS-A samples against S. Heidelberg IOC 969/17 was investigated by the agar diffusion method, specifically spot-on-the-lawn methodology 42 . For that, the bioindicator strain was precultivated at 37 °C in BHI broth and after 18 h it was 100-fold diluted (approximately 10 6 CFU/mL) and 1 mL of this suspension was transferred to a Petri dish (90 × 15 mm) containing 9 mL of BHI melted soft agar (supplemented with 0.75% [w/v] of agar). After solidification, 20 μL of each CFS-A was pipetted onto the agar surface and, finally, the plates were incubated at 37 °C for 24 h. The inhibition zone formed after such period by the CFS-A was measured using a digital caliper and the antimicrobial activity was associated with BLIS production.
Additionally, attempting to verify the inhibitory spectrum caused by BLIS present in the CFS-A, the samples were also tested against other bioindicator strains, such as Listeria monocytogenes CECT-934, Salmonella enterica CECT-724, S. aureus CECT-237, Escherichia coli ATCC 25922, Carnobacterium piscicola CECT-4020, Listeria innocua CLIST 2052. All strains were cultured under conditions similar to S. Heidelberg IOC 969/17, except for C. piscicola CECT-4020, which was cultured in MRS broth. The antimicrobial activity was determined in technical triplicate.
To see if BLIS were of proteinic nature CFS-A samples which exhibited inhibition zones against the bioindicator strains were treated with 3 mg/mL (w/v) of the proteolytic enzyme protease XIV (Sigma-Aldrich, Missouri, USA) which was added to 1 mL of each CFS-A and incubated at 30 °C for 2 h 42 . After that, the solutions were evaluated regarding their remaining antimicrobial activity against S. aureus CECT-237, following the protocol described above. The solutions demonstrating no inhibition zones were considered, in this study, as being BLIS.
The isolates producing CFS-A with antimicrobial activity against most bioindicator strains were cultured and after 24 h incubation the supernatants were once again recovered. At this time, the aim was to investigate the antimicrobial activity of the organic acids supposedly produced against S. Heidelberg IOC 969/17, so no pH adjustment were performed, resulting in CFS-B samples. To carry out this assay, the bioindicator strain was cultured in 10 mL of BHI broth at 37 °C for 24 h and diluted in 2× BHI broth to achieve a cell concentration of 0.2 optical density (OD) units at 600 nm per milliliter. After that, 100 μL of this solution was placed into wells of a 96-well sterile covered microplates (TPP, Trasadingen, Switzerland) together with 100 μL of each CFS-B sample from the Identification of LAB producing B-complex vitamins. The isolates, which displayed the largest inhibition zones against most of bioindicator strain, were selected to evaluate their ability to produce the B-complex vitamins riboflavin and folate. The protocol was followed according to Laiño et al. 43 . Briefly, the selected isolates, stored at −70 °C, were first activated in the above-mentioned medium and conditions, according to the isolates preferences. After 24 h incubation, 1 mL-aliquots were centrifugated at 12,000 g at 4 °C for 15 min. The supernatants were discarded and the pellet cells were washed 3 times with sterile 0.85% (w/v) saline, being resuspended at the original culture volume after this 3-washes steps. The resulted solutions were used to inoculate at 2% (v/v) the Folic Acid Casei Medium (FACM; Himedia, Mumbai, India), and the Riboflavin Assay Medium (RAM; Difco, Maryland, USA), which are free from folates and riboflavin, respectively. The cultures were incubated without agitation at 37 °C for 18 h and after growth those 3-washes and resuspending step was repeated. The resulting solutions were used to inoculate at 2% (v/v) fresh FACM or RAM. In order to deplete the reserves of folates or riboflavin from the isolates, this last step was repeated 7 times with the cultures showing good growth (observed by increased turbidity). The isolates which did not grow in FACM or RAM were not used in further studies.
Folate quantification. A 500 μL sample of isolates grown in FACM were taken and diluted with same volume of protecting buffer (0.1 mol/L phosphate buffer, pH 6.8, containing 1.5% [w/v] ascorbic acid to prevent vitamin oxidation and degradation) followed by immediate centrifugation at 12,000 g at 4 °C for 5 min. The obtained supernatants were then boiled (100 °C) for 5 min and stored at -70 °C until folate quantification.
Folate quantification was determined by a microbiological assay using Lactobacillus casei subsp. rhamnosus ATCC 7469, provided by Fiocruz (Rio de Janeiro, RJ, Brazil) as the indicator strain according to Horne and Petterson 44 . The strain was previously cultured in MRS broth at 37 °C for 24 h and cryopreserved using 20% (v/v) glycerol and stored at -70 °C. Fresh MRS broth was used to activate the strain under the same mentioned conditions. After growth, 1mL-aliquot was washed 3 times with sterile 0.85% (w/v) saline, resuspended in the original volume, and used to inoculate at 2% (v/v) fresh FACM, which was incubated for 24 h at 37 °C. This last step was repeated and the second culture was used to perform the folate determination, as follows bellow.
The frozen samples were thawed at room temperature (24 °C) and processed in light-reduced conditions 45 . The samples were then diluted with protection buffer and 100 μL of each sample was placed into wells of a 96-well sterile covered microplates (TPP, Trasadingen, Switzerland). The folate concentration of each diluted sample was determined in triplicate. One hundred microliters of the indicator strain L. casei subsp. rhamnosus ATCC 7469, grown in FACM as described above, was added to each well and mixed. The growth in the presence of the samples was compared to those with a standard curve prepared using HPLC grade folic acid (Sigma-Aldrich, Missouri, USA) diluted in the protection buffer at different concentrations (between 0 and 0.1 ng/mL).
Riboflavin determination. The determination of riboflavin was carried out according to de Arruda et al. 46 with some adaptations. Briefly, the selected isolates were cultured following above-described conditions. After 24 h incubation, using a 100 mL-Erlenmeyer flasks, 5 mL of each culture broths were added with 45 mL of 0.1 M hydrochloric acid. The resulting solutions were heated in a shaking thermoregulated bath at 100 °C/100 rpm for 30 min. The solutions were cooled to room temperature (24 °C) and had their pH adjusted to 4.6 using 2.5 M sodium acetate. After adding 0.5 g of the enzyme diastase from Aspergillus oryzae (Merck Millipore, Massachusetts, USA), the solutions were incubated for 2 h at 42 °C in a thermoregulated bath. After such enzymatic treatment and cooling at room temperature (24 °C), the solutions were transferred to 100 mL-volumetric flasks and the volumes were completed with deionized water, homogenized and filtered through 0.45-μm hydrophilic PVDF membranes (Nova Analítica, São Paulo, Brazil) for injection into the chromatographic system. This step was carried out in triplicate.

Identification of the selected LAB isolates using matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOFMS).
To identify the selected isolates, MALDI-TOFMS analysis were adopted according to Alves et al. 47 . They were cultured in plates containing MRS or BSM 1.5% agar and after 24 h incubation under anaerobic condition, the isolates were transferred from the plates to a 1.5 mL-microtube containing 200 μL of distilled water and homogenized during 1 min in a vortex device. Then, 900 μL of ethanol was added and the samples were centrifuged at 12,000 g at 4 °C for 5 min. The supernatant was discharged, and the residual alcohol was left to dry at room temperature. The resulting pellet was suspended in 50 μL of 70% formic acid and 50 μL of acetonitrile was added, and the sample was homogenized using a vortex. An α-cyano-4-hydroxycinnamic acid matrix was prepared as a saturated solution in 50% acetonitrile and 2.5% trifluoroacetic acid. Subsequently, 1 μL of samples previously treated and 1 μL of matrix solution was spotted onto a steel target plate and allowed to dry at room temperature.
The mass spectrometry analyses were conducted using an UltrafleXtreme MALDI-TOF mass spectrometer (Bruker Daltonics, Germany) operating in the linear positive ion mode. Mass spectra were acquired in a mass Scientific RepoRtS | (2020) 10:7235 | https://doi.org/10.1038/s41598-020-64038-9 www.nature.com/scientificreports www.nature.com/scientificreports/ range from 2 to 20 kDa with ions formed by irradiation of smart beam using a frequency of 2000 Hz, PIE 100 ns, 7 kV lens. The voltages for the first and second ion sources were 25 kV and 23 kV, respectively. Bacteria were identified by means of the Biotyper 3.1 database. The identification cut-off values higher than 2 and 1.7 were used for species and genus identification, respectively.

In vitro evaluation of desirable criteria to consider the selected isolates as probiotic LAB.
Tolerance to low pH. To verify the isolates' tolerance to low pH conditions, it was adopted the protocol developed by Zoumpopoulou et al. 48 and Argyri et al. 49 . Briefly, the isolates were cultured in the preferential medium and temperature. After 18 h incubation, the culture broths were centrifugated at 4,470 g at 4 °C for 10 min and the pellets cells were washed twice with phosphate buffered saline (PBS) (pH 7.2) before being resuspended in PBS solution adjusted to pH 2.5. To reflect the time spent by food in the stomach, the resulting solutions were incubated at 37 °C for 0, 0.5, 1.0, 2.0 and 3.0 h and the tolerance was observed in triplicate in terms of viable colony counts and enumerated on the same medium initially used to cultivate the isolates added with 1.5% (w/v) agar.
Tolerance to bile salts. The tolerance to bile salts was evaluated adapting the protocol followed by Maragkoudakis et al. 50 and Pedersen et al. 51 . The isolates were cultured in MRS broth at 37 °C for 24 h. After that, 1 mL of such cultures broths were used to inoculate, individually, 9 mL of MRS broth supplemented with 0.5% (w/v) porcine bile extract (Sigma-Aldrich, Missouri, USA). The resulting solutions were incubated at 37 °C for 4 h. The isolates demonstrating bile resistance were assessed in terms of viable colony counts, enumerated after incubation for 0, 1, 2, 3 and 4 h, which the latter being representative of the time spent by food in the small intestine.
Hemolytic activity. In order to investigate if the isolates produced hemolytic activity, they were cultured in the appropriated conditions mentioned before and after 24 h incubation, the cultures were streaked onto Mueller Hinton agar surface (Difco, Michigan, USA) previously supplemented with 5% (v/v) sheep blood (Ebefarma, Rio de Janeiro, Brazil). The plates were incubated at 37 °C under anaerobic condition and after 24 h, they were exanimated for signs of β-hemolysis (clear zones around colonies), α-hemolysis (green-hued zones around colonies) or γ-hemolysis (no zones around colonies) 49 .
Antibiotic susceptibility. Antibiotic susceptibility was determined by the modified Kirby-Bauer disc diffusion method 52 . Discs (all provided by Laborclin, São Paulo, Brazil) of the following antibiotics were used: ampicillin (10 μg); chloramphenicol (30 μg); clindamycin (2 μg); gentamicin (10 μg); streptomycin (10 μg); rifampicin (5 μg), and vancomycin (30 μg), all selected following the suggestion stated on the document Opinion of the Scientific Committee on Animal Nutrition on the criteria for assessing the safety of micro-organisms resistant to antibiotics of human clinical and veterinary importance 53 . Briefly, the isolates were cultured in the above-mentioned medium and conditions. After 24 h, the concentration of isolates' cell suspension was adjusted according to 0.5 McFarland (Probac, São Paulo, Brazil) standard and spread over the surface of Mueller Hinton agar plates. When the suspensions were dried, the discs were carefully placed onto the plates and anaerobically incubated at 37 °C for 24 h. Presence or absence of inhibition zones was defined as sensitivity or resistance, respectively, interpreted according to the cut-off levels proposed by Charteris et al. 54 .
Coexistence test. The compatibility among the selected isolated was performed according to Guo et al. 55 . The selected isolates were cultured in the appropriate conditions and after 24 h they were streaked perpendicularly and across each other on MRS 1.5% (w/v) agar plates. The plates were anaerobically incubated at 37 °C and after 24 h antagonism or absence of was observed.
Cell surface hydrophobicity. The cell surface hydrophobicity assay was conducted according to García-Hernández et al. 56 . In this sense, the hydrophobicity was determined as the ability of the isolates to adhere to hydrocarbons (MATS: Microbial Adhesion to Solvents), using toluene as solvent. This assay was carried out culturing the isolates in appropriated medium and temperature for 24 h. A 3 mL-aliquot from each culture broth were centrifugated at 4.470 g at 4 °C for 10 min and the pellets were washed twice with PBS (pH 7.2) and resuspended in same buffer until achieve the concentration of 1.0 OD at 560 nm (A b0 ). To three milliliter of these solutions, 0.6 mL of toluene (Synth, São Paulo, Brazil) was added and gently mixed for 2 min. After an incubation period of 1 h at 37 °C, the aqueous phase (bottom phase) was carefully collected and the OD 560 nm was determined again (A b1 ). Percentage of MATS was calculated using the following equation: Isolates with MATS above 50% were considered as hydrophobic.
Bacterial adhesion to intestinal epithelial cells. The adhesion capacity of the isolates was investigated according to Jensen et al. 57  www.nature.com/scientificreports www.nature.com/scientificreports/ antibiotics from the original cell media and 1 mL of each bacterial cells solution (10 7 CFU/mL) were individually added to the wells. Subsequently, the well-plates were incubated at 37 °C for 1, 2 or 4 h in order to determine the best adhesion period. After that, the bacterial solutions were removed from each well and the monolayers were washed twice with PSB to remove non-adherent bacteria. Finally, the monolayers were lysed by addition of PBS added with 0.1% Triton-X100 (Sigma-Aldrich, St. Louis, USA). The resulting suspension with viable adhered bacteria were serially diluted and plated onto MRS agar by pour plate method. After 48 h incubation in anaerobic jar, the number of colony forming units per mL was counted. Bacterial adhesion capacity was expressed as a percentage, which was calculated using the ratio of the number of bacterial cells that remained attached in the monolayer to the total number of bacterial cells added initially to each well. The experiment was performed in biological triplicate.

Statistical analyses.
The results were expressed as the mean and standard deviation (S.D.). The counts of viable bacteria were transformed to log values. The values in the tolerance test were compared by Student's t-test. P-values less than 0.05 were regarded as a significant difference.

Results and discussion
evaluation of antimicrobial activities in LAB isolates. From the cecum content of broiler chickens, 314 colonies were aleatory selected and their antimicrobial activity against S. Heidelberg IOC 969/17 were evaluated. From these assays 35 isolates stood out compared to the others, producing large inhibition zones against the bioindicator strain ( Fig. 1) and these were characterized as being Gram-positive with coccus-and rod-shaped, non-sporulating and catalase-negative, which are predominant features of the LAB group 58 .
The antimicrobial activity of pH neutralized samples (CFS-A) was evaluated against S. Heidelberg IOC 969/17 and did not present any antimicrobial activity. Considering this result and the fact that BLIS antimicrobial activity is target-specific 59 , the samples were tested against other pathogens including L. monocytogenes CECT-934, S. enterica CECT-724, S. aureus CECT-237, E. coli ATCC 25922, C. piscicola CECT-4020 and L. innocua CLIST 2052.
Interestingly, the highest antimicrobial activities were observed against Gram-positive species, especially S. aureus CECT-237 (see Table 1). None of the neutralized samples were capable of producing inhibition zones against S. enterica strain CECT-724 or E. coli ATCC 25922, both being Gram-negative. According to Prudêncio et al. 60 , bacteriocins produced by Gram-positive microorganisms are not effective against Gram-negative ones since the latter are resistant to their action. This occurs because Gram-negative bacteria have are very complex outer membrane that make them resistant to certain antibiotics, dyes, and detergents 61 . Certain treatments can alter this unique outer membrane and make it permeable and, consequently, making the bacterial cells sensitive to bacteriocins 62 . Particularly, Salmonella species have been frequently pre-exposed to chelating agents, for example EDTA which acts by destabilizing the complex membrane of Gram-negative bacteria and cause the release its structural components or disrupt the lipopolysaccharide layer (LPS), allowing the bacteriocin to act 60 . In this sense, this strategy could be a useful alternative to overcome the Salmonella's resistance to the BLIS produced by our isolates, being therefore a test that will be carried out in the future. Based on the results presented in Table 1, the isolates labeled as C43, C173 and C195 were chosen for further study.
The samples were treated with proteolytic enzymes and no inhibition zones was observed against S. aureus CECT-237, suggesting that the antimicrobial effect is provided by BLIS as suggested by previous works [63][64][65][66] .
Considering that neutralized samples from C43, C173 e C195 did not produced antimicrobial activities against Salmonella species, untreated samples (CFS-B) were evaluated against S. Heidelberg IOC 969/17. As can be observed in Fig. 2, all three samples demonstrated a satisfactory growth inhibition of S. Heidelberg IOC 969/17, especially the CFS-B extracted from C173 and C195. These results suggest that the antimicrobial activity   Table 1. Antimicrobial activity of CFS-A extracted from the culture of the isolates against bioindicator strains. "-" corresponds assays without inhibition zone. www.nature.com/scientificreports www.nature.com/scientificreports/ produced by C43, C173 and C195 isolates against S. Heidelberg IOC 969/17 is not provided by BLIS, but by organic acids produced during their growth. When the culture media without bacterial growth was used as a control, S. Heidelberg inhibition was not observed (data not shown).
The use of such substances to combat Salmonella species in poultry products or products is very well documented [67][68][69][70] . The antimicrobial activity of organic acids can be explained because these substances can pass through the membranes of the target cells which will be dissociated in the more alkaline interior, acidifying the cell cytoplasm 70 . Many authors have shown that organic acids produced by probiotic LAB have antimicrobial effects and can be used as feed additive [71][72][73][74] . Furthermore, it has been shown that LAB can produce short and medium chain fatty acids (SCFA and MCFA) and other metabolites in the gastrointestinal tract of poulty that possess antagonism properties 67 against different Salmonella species and on other microorganisms found in the avian microbiome 67,70,73 . Our results concur with these previous studies and suggest that our isolates could be useful to be used as supplements for poultry feed.
Screening for vitamin producing isolates. In poultry nutrition, the deficiency of B-complex vitamins, especially riboflavin and folates can directly affect the health and well-being of birds.
Riboflavin are present in numerous co-enzymes that are involved in oxidation-reduction reactions for cell respiration 75 . When there is riboflavin deficiency in poultry, their growth is reduced caused by the loss of appetite and the appearance of diarrhea is very common 75 . Frequently, riboflavin deficient chicks cannot walk due to leg or curled-toe paralysis 76 . In addition, riboflavin deficiency in hens has been shown to lower egg production, cause the death of embryos, and increase the fat content in their livers 75,77 .
Folate also acts as a co-enzyme by accepting or donating single-carbon in amino acids and nucleic acids synthesis 31 . This vitamin is essential for purine and methyl groups synthesis which is why folates are essential for cell multiplication 75 . Folate deficiencies can reduce growth, cause megaloblastic anemia, produce poor feathering, achromat of feathers in colored poultry, and chondrodystrophy 31,75 .
Riboflavin and folate producing LAB can be an interesting option to supplement the diets of animal diets since they could release the vitamin in their feed or produce these essential nutrients in the gastrointestinal tract (although it is still unknown how much vitamins can be produce in situ).
In our study, among the 35 isolates selected in the previously step, those displaying the largest inhibition zones against the bioindicator strains were used to study their ability to produce folates and/or riboflavin. Eight isolates were selected, as means C43, C81, C83, C173, C195, C197 and C206. Posteriorly, these isolates were cultured in media culture broths free from folic acid and riboflavin and after 7 subcultures, from 8 isolates, 3 of them (C43, C173, and C195) grew well in the riboflavin free-culture medium and 5 isolates (C43, C81, C83, C173 C195) presented a good growth in the folic acid free-culture medium. Three of these 8 isolates (C43, C173, and C195) showed good growth in both culture media, suggesting riboflavin and folates are produced simultaneously.
As can be observed in Table 2, the isolates C195 and C173 showed the higher folate production (595 and 58 ng/ mL respectively).
Many researchers stated the folate and riboflavin production by lactic acid bacteria 30,78-81 , but few have already reported the simultaneous production of both B-complex vitamins by a single strain, which represents an important advantages of our findings, especially for the poultry industries, since the LAB producing B-complex vitamins are mainly focused in human consumptions, instead of animals nutrition 82 .
According to the document published by National Research Council in 1994, the daily intake requirement of folic acid and riboflavin is about 1 and 5 mg per kg of diet, respectively 83 . The daily feed consumption varies considerably between gender and also the poultry specification, i.e. laying hens and broiler chicken, for examples. For broiler chicken, a 1-day old male chicken consumes about 13 g of feed 84 . When the same chicken reaches 45 days of life, its daily consumption is of approximately 186 g of feed. The same estimative can be done about water intake. A 1-week broiler chicken can drinks about 40 mL of water per day, which increase 5 times when this chicken is 3-weeks old 84 .
Feed for poultry are formulated to make sure that they contain more than enough micronutrients (vitamins and minerals) to avoid deficiencies and take into account the amounts that are lost during their processing, storage and transport 75 . In fact, this is sometimes questioned as being too costly 84 . Thus, including our isolates in commercial poultry nutrition can be an economic alternative. Another advantage is associated to the fact that they could be more stable to storage conditions, one of the major causes of loss of vitamins bioavailability 75 .  www.nature.com/scientificreports www.nature.com/scientificreports/

Identification of the selected isolates by MALDI-TOFMS Biotyper.
Considering the best performance of the isolates C43, C173 and C195, the genus and species of such microorganisms were identified by MALDI-TOFMS Biotyper that identifies microorganisms by using protein profiles of microorganisms and comparison with known patterns 47,85 . This technique is fast, economic and is a reliable alternative to conventional 16 s rRNA PCR amplification and sequencing to identify microorganisms 85,86 .
The isolate C43 was identified as E. faecium, whereas C173 and C195 were both identified as being strains of L. lactis subsp. lactis.
The presence and the isolation of E. faecium from chicken microbiota is very well documented in the literature [87][88][89] , but this is not the case for L. lactis strains. This species is normally found in dairy, vegetable, sausages, raw pork, vacuum-packed seafood 90 and from chicken cecum 91 .
According to the Food and Agriculture Organization 92 , both species have been used as probiotics in poultry diets. E. faecium has shown to be able to increased growth performance, improved intestinal morphology, and possess antagonism effects against important pathogens in poultry production 93,94 . Its usage in poultry nutrition is so widely defunded that are many commercial feed composed with E. faecium, for example PoultryStar ® , Protexin ® , and Biomin 92 .
Although the use of L. lactis in poultry is not very common, Maiorano et al. 95 demonstrated increased feed conversion ratio (FCR) when a symbiotic mixture of L. lactis ssp. cremoris IBB SC1 and raffinose were injected in chickens' eggs in order to evaluate its effect on productivity, the quality of the meat produced, and on the incidence of pectoral muscle pathologies in broiler chickens. On the other hand, Pruszynska-Oszmalek et al. 96 reported that of L. lactis subsp. lactis added with a prebiotic mixture of galacto-oligosaccharide and inulin did not result in improved FCR, but significantly increased final body weight of the chickens tested. Another interesting approach is a lactococcal vaccine proposed by Reese and collaborators 97 . In this study, the author demonstrated the protected effect of the vaccinated birds against the avian influenza virus.
As can be observed, both species, i.e. E. faecium and L. lactis subsp. lactis have being proven their benefits to chickens health when administrated as probiotic bacteria. evaluating the probiotic potential of the selected isolates. In order for orally ingested bacteria to be able to exert a beneficial effect on their host, they must survive the passage through the stomach and the presence of bile salts in the intestinal lumen and arrive alive in the large intestines 98 . As can be observed in Fig. 3, except the C43, the isolates C173 and C195 showed significantly loss of viability in low pH conditions; however, their encapsulation or addition in a food matrix could increase their survival 50 .
Although bovine bile is normally used to evaluate the tolerance of bacteria, bile salts from pork are more similar to that produced by humans 99 , justifying its using in the present study.
Overall, bile salts at 0.5% concentration did not affect the survival of the isolates C43, C173 and C195 (Fig. 4) Bile salt concentrations are crucial as a defense mechanisms by inhibiting the survival of microorganisms 100 . The resistance to bile salts is this critical for probiotic strains selection in order for them to arrive in a live form after its passage in the small intetines 49 .
Hemolytic activity and antibiotic resistance. None of the selected isolates demonstrated β or α-hemolytic activities, essential for the possible use of the strains 101 .
Because of the increasing problems concerning antibiotic resistance, it is important that probiotics that are to be used in poultry feed additive not be reservoirs for antibiotic resistance genes 102,103 .
According to zones diameters breakpoints established by Charteris et al. 54 , all isolates were sensitive to the studied antibiotics except for L. lactis subsp. lactis C173, which showed resistant to vancomycin. This result is similar to those of other studies 50,82,98 . Although it was previously reported certain strains of L. casei, L. rhamnosus, L. plantarum, pediococci and Leuconostoc spp. are resistant to vancomycin, this property is of much concern because this antibiotic is is one of the last antibiotics that is useful in infections caused by multidrug-resistant pathogens 104,105 . Based in the study reported by Temmerman et al. 106 , it would be appropriate to use other methods to confirm the presence of vancomycin resistance in L. lactis subsp. lactis C173, like Minimal Inhibitory Concentration (MIC) or PCR assay, which will be considered in a future study. www.nature.com/scientificreports www.nature.com/scientificreports/ Coexistence test. With the aim of obtaining multi-strain probiotics, and due to their antimicrobial properties, strain compatibility must be evaluated 107 .
As can be observed in Fig. 5, the isolate L. lactis subsp. lactis C173 and L. lactis subsp. lactis C195 reveled their antagonistic effect against E. faecium C43. Based on this data and assuming that this effect could also occur in vivo, the E. faecium C43 were excluded from the following tests.
Hydrophobicity property and adhesion of the selected isolates to Caco-2 cells. In order to evaluate their potential action in the gastrointestinal tract, adhesion to human colon carcinoma and the evaluation of hydrophobicity must be performed 107 . Hydrophobicity has been associated with bacterial adhesion properties 108 reason for which the isolates' adhesion to toluene was measured. The L. lactis subsp. lactis C173 showed a moderate hydrophobicity (44%), while L. lactis subsp. lactis C195 hydrophilic surface (9.2%).
Adherence to Caco-2 assay showed that both isolates presented high value of adhesion to this cells (Fig. 6). For L. lactis subsp. lactis C173, after 1, 2 and 4 h incubation, the bacterial solution presented values of adhesion of 13.2, 27.9 and 35.6%, respectively. L. lactis subsp. lactis C195 in turn, resulted in adhesion values of 7.3, 20.3 and 46.3% after 1, 2 and 4 h incubation, respectively. According to Gharbi et al. 98 , Laiño et al. 109 and Jensen et al. 57 an adhesion rate of 11% is considered a high adhesion capacity. In this study, it is clear that a longer period of  www.nature.com/scientificreports www.nature.com/scientificreports/ incubation results higher adhesion percentages, but independently of this incubation time, both cells showed strong adhesion with Caco-2 cells. This high adhesion percentage suggests that these strains are likely to colonize the GI tract. It is important that probiotics adhere to the intestinal epithelium in order to facilitate their colonization and prevent pathogen binding 110 .
As mentioned before, bacterial cell hydrophobicity is frequently associated to the adherence to the human epithelium cells; however there is conflicting reports on this direct association 98,111 . Our findings are in agreement with these latter authors that there is no direct association between hydrophobicity and the binding habilities of our isolates.
conclusion From 314 isolates from chicken cecum, the isolates labeled as C43, C175 and C195 demonstrated antagonistic effect towards many pathogens. Using proteolytic enzymes, it was demonstrated that the antimicrobial effect against Listeria species, C. piscicola and S. aureus was caused by BLIS and by organic acids against S. Heridelberg. The isolates, identified as E. faecium and L lactis subsp. lactis, also produced the B-complex vitamins folate and riboflavin, which is very interesting considering the diseases related to the lack of these nutrients in the feeding of birds. After in vitro probiotic assays, the L. lactis subsp. lactis strains C173 and C195 demonstrated high potential to be used as probiotic in poultry feed being a useful alternative to replace antibiotics in poultry husbandry. In order to demonstrate this potential, in-vivo trials will be required and are currently underway.