Bioprotective potential of lactic acid bacteria for Salmonella biocontrol in vitro

Lactic acid bacteria (LAB) are an important option for Salmonella control in animal production, resulting in lower antibiotic use. The objective of this research was to isolate LAB from meat products and from commercial probiotics sold as nutritional supplements for in vitro verification of their bioprotective potential. Eleven bacteria were identified as Pediococcus acidilactici, two as Lacticaseibacillus rhamnosus, one as Lacticaseibacillus paracasei paracasei, one as Limosilactobacillus fermentum, and one as a consortium of Lactobacillus delbrueckii bulgaricus and L. fermentum. All bacteria showed inhibitory activity against Salmonella, with emphasis on the inhibition of P. acidilactici PUCPR 011 against Salmonella Enteritidis 33SUSUP, S. Enteritidis 9SUSP, S. Enteritidis 56301, S. Enteritidis CRIFS 1016, Salmonella Typhimurium ATCC™ 14,028®, and Salmonella Gallinarum AL 1138, with inhibition halos of 7.3 ± 0.5 mm, 7.7 ± 1.0 mm, 9.0 ± 1.8 mm, 7.3 ± 0.5 mm, 7.7 ± 1.0 mm, and 7.3 ± 0.5, respectively. The isolates P. acidilactici PUCPR 011, P. acidilactici PUCPR 012, P. acidilactici PUCPR 014, L. fermentum PUCPR 005, L. paracasei paracasei PUCPR 013, and L. rhamnosus PUCPR 010 showed inhibition greater than 2 mm against at least 3 Salmonella and were used for encapsulation and in vitro digestion. The encapsulation efficiency ranged from 76.89 ± 1.54 to 116.48 ± 2.23%, and the population after 12 months of storage was from 5.31 ± 0.17 to 9.46 ± 0.09 log CFU/g. When simulating swine and chicken digestion, there was a large reduction in bacterial viability, stabilizing at concentrations close to 2.5 log CFU/mL after the analyses. The analyzed bacteria showed strong in vitro bioprotective potential; further analyses are required to determine in vivo effectiveness.


Introduction
Salmonella is one of the main pathogens associated with animal production and public health issues, responsible for important economic losses. Salmonella has a high prevalence in swine production with a prevalence of 29%, 14.2%, and 39% in the European Union (EU), the United States (US), and Brazil, respectively (Kich et al. 2011;Snary et al. 2016;Bjork et al. 2018). Furthermore, it is an important pathogen in chicken production; the EU and the US have presented a prevalence of 27% and 15.6%, respectively (Alali et al. 2016;Koutsoumanis et al. 2019).
This genus also covers bacteria of public health relevance and is the main cause of foodborne diseases in several countries. In the US, Salmonella is responsible for 1.2 million cases of illness, 23,000 hospitalizations, and 450 deaths every year (Robertson et al. 2019).
This pathogen is highly related to antimicrobial resistance; isolates with cephalosporins, ampicillin, tetracyclines, amoxicillin, chloramphenicol, quinolone, sulfonamides, and aminoglycosides resistance were found in several countries, mainly in China, EU, US, and Brazil, which are the main animal producers in the world (Shigemura et al. 2018;Xu 1 3 et al. 2018;Khan et al. 2019;Zhu et al. 2019;Procura et al. 2019). Several strategies were developed to mitigate risks associated with these bacteria and their antimicrobial resistance, such as probiotics, prebiotics, bacteriophages, and bioprotective agents (Heithoff et al. 2015;Grant et al. 2016;Evangelista et al. 2021Evangelista et al. , 2022Danielski et al. 2022).
Lactic acid bacteria (LAB) excel at bioprotective use by several mechanisms, such as competitive exclusion and the production of peptides and/or antimicrobial substances (Corrêa et al. 2019;Danielski et al. 2022). Thus, the objective of this research was to evaluate the in vitro bioprotective potential of LAB isolates against Salmonella.

Bacterial strains and isolates, and culture conditions
Seventeen LAB were isolated from autochthonous microbiome of meat products naturally fermented, stored at environmental conditions, and from human commercial probiotics using de Man, Rogosa, and Sharpe (MRS) culture media (Merck, Darmstadt, Germany), according to Endo et al. (2019) with modifications.
Briefly, 1 g of samples was inoculated in MRS broth. The materials were incubated at 37 ºC for 48 h, followed by obtaining pure cultures on MRS agar and initial characterization by Gram stain. To avoid the isolation of the same microorganism, only one microorganism of the same species was kept in the same isolation cycle. The isolates were identified by matrix-assisted laser desorption ionization-time of flight (MALDI-TOF) analyses (Bruker MALDI Biotyper ® ).
Six Salmonella strains (Salmonella Enteritidis 33SUSUP, S. Enteritidis 9SUSP, S. Enteritidis 56301, S. Enteritidis CRIFS 1016, Salmonella Typhimurium ATCC ™ 14028 ® , and Salmonella Gallinarum AL 1138) were used, obtained from American Type Culture Collection, and from the University of Manitoba and Canadian Research Institute for Food Safety (Canada) and Universidade Federal do Paraná (Brazil) microbiology collections, originally isolated from animal production.
Bacterial strains and isolates were stored at -20 ºC in 50% glycerol in culture media until assays. Prior to the experiments, all bacteria were subcultured twice at 37 °C for 24 h in MRS broth -for LAB -or in brain and heart infusion (BHI) broth (Sigma-Aldrich, St. Louis, MO) -for Salmonella.

Antimicrobial activity assay
In vitro antimicrobial activity was evaluated by agar spot tests (Jacobsen et al. 1999). Initially, LAB were grown individually in MRS broth, incubated at 37 ºC for 24 h. After bacterial growth, 3 µL were inoculated as spots on the surface of MRS agar plates and incubated at 37 ºC for 24 h. Then, 100 µL of an overnight fresh culture of each Salmonella was inoculated into 7 mL of unhardened BHI agar at 40 ºC and poured onto the MRS agar with the spots. The final population in BHI agar was ~ 10 7 UFC/mL. A positive control was performed with Salmonella inoculated on MRS agar without LAB spots, and a negative control with Salmonella inoculated on MRS agar with spots of MRS broth only, without LAB. The plates were then incubated at 48 °C for 24 h, and the inhibition zones were measured. An inhibition halo greater than or equal to 1 mm was considered positive. This breakpoint was determined through analysis of the scientific literature in the area, in order to be in accordance with reference articles (Jacobsen et al. 1999;Edalati et al. 2019;Mulaw et al. 2019;Meena et al. 2022). Each test was performed three times in duplicate.

Preparation of inocula
LAB with an inhibition zone of more than 2 mm against at least 3 Salmonella strains in the spot analyses were selected for individual microencapsulation. The breakpoint point was established to select LAB with action 2 × greater than the minimum determined against ≥ 50% of the Salmonella used in the study.
The selected LAB were individually inoculated in 800 mL of MRS broth and incubated under agitation (120 rpm) for 24 h at 37 ºC. After bacterial growth, the material was centrifuged (5,000 × g, 10 min), the supernatant was discarded, and the bacterial biomass was washed with 20 mL of sterile 0.9% saline solution followed by centrifugation (5,000 × g, 10 min) 3 times to remove all growth medium residue. In the end, the bacteria were resuspended in 400 mL of sterile 0.9% saline solution at a concentration of 10 9 CFU/mL, adjusted by optical density (Arepally and Goswami 2019).

Preparation of drying media and microencapsulation
The encapsulation agents used were maltodextrin (Globe ® A 1910, Ingredion ™ , Mogi Guaçu, Brazil) and gum Arabic (Biotec ® , São José dos Pinhais, Brazil), at concentrations of 20% and 7.5%, respectively. The agents were dispersed in the 400 mL-inocula at room temperature under magnetic stirring, until complete homogenization. The solutions were submitted to a non-lethal treatment at 52 ºC for 10 min to acclimatize the bacteria to the spray dryer process.
Immediately after the non-lethal treatment, solutions were spray-dried (MSD/EV, Elettronica Veneta, Motta di Livenza, Italy), at a constant flow rate of 12 mL/min and outlet air temperature of 70.0 ± 5.0 ºC. Before the process, the spray dryer was operated for 30 min, to stabilize the operational conditions.
The powder samples were collected in ethylene-vinyl alcohol bags and immediately stored at 4 ºC for further analysis (Arepally and Goswami 2019). Each bacterium was microencapsulated twice.

Powder morphology
The microcapsule morphology was evaluated by scanning electron microscopy (SEM). Samples were fixed with conductive carbon tapes in SEM stubs and sputter-coated with gold by cathode spraying (Q150R ES, Quorum, Laughton, United Kingdom). The images were obtained in a scanning electron microscope (VEGA 3 LMU, TESCAN, Kohoutovice, Czech Republic) at 4,000 × magnification and a voltage of 5 kV (Arepally and Goswami 2019). Five images of each sample were obtained for morphological evaluation.

Enumeration of the bacterial population in powder, encapsulation efficiency, and stability
The LAB population was evaluated just after the spray dryer process and after each 30 d for 12 months. In aseptic conditions, 1 g of the spray-dried powder was homogenized with a vortex in a sterile 0.9% saline solution for 3 min at room temperature. To determine the population, serial dilutions were inoculated in MRS agar plates, incubated at 37 ºC for 48 h. Each test was performed twice in duplicate.
To estimate the encapsulation efficiency (%) immediately after spray drying, the following formula was used, where N is log CFU/g after the encapsulation process, and 9 represents 9 log CFU/mL of the initial inoculum (Arepally and Goswami 2019):

Encapsulated LAB stability under in vitro digestion
Before the in vitro digestion, 500 g/L of encapsulated LAB were reconstituted in sterile deionized water at 25 ºC for 30 min, under magnetic stirring until complete homogenization.

Swine in vitro digestion
The swine in vitro digestion was performed in Falcon-type tubes at 39 ºC, 200 rpm, in a water bath with agitation Encapsulation efficiency (%) = N∕9 × 100 (Dubnoff bath, TE-053, Tecnal, Sao Paulo, Brazil), simulating the stomach and intestine action. In the stomach phase, 500 µL of the reconstituted powder was used, added with 20 mL of phosphate buffer pH 6.0, 8 mL of HCl 0.8 M, and 50 U of pepsin (Sigma-Aldrich, St. Louis, MO), and incubated for 2 h. After incubation, the intestinal phase in the same Falcon-type tubes proceeded with the addition of 7 mL of phosphate buffer pH 6.8, 4 mL of NaOH 0.6 M, and 100 mg of pancreatin (4 × USP, Sigma-Aldrich, St. Louis, MO), and incubation for 4 h (Boisen and Fernández 1997). The analysis was made twice, in triplicate.

Chicken in vitro digestion
The chicken in vitro digestion was performed in Falcon-type tubes at 42 ºC, 200 rpm, in a water bath with agitation, simulating the crop, proventriculus, gizzard, and small and large intestines action. In the crop phase, 500 µL of reconstituted powder was used, added with 10 mL of sterile deionized water at final pH 6.0 (adjusted with HCl 1.0 M). The tubes were incubated for 30 min. In the proventriculus phase, the tubes received 25 mL of sterile deionized water and 3,850 U/mL of pepsin, and the final pH was adjusted to 2.5. The tubes were re-incubated for 30 min. In the gizzard phase, glass pearls were added to simulate mechanical digestion. The pH was adjusted to 3.0 with NaHCO 3 1.0 M, and tubes were returned to incubation for 1 h. For the small intestine phase, 8 U/mL of pancreatin and 135 mg of bile salts (Sigma-Aldrich, St. Louis, MO) were added and the pH was adjusted to 6.2. This phase required a 2 h-incubation period. Finally, the large intestine phase was simulated through pH adjustment to 7.0 and 20 min of incubation (Oliveira et al. 2019). The analyses were made twice, in triplicate.

Lactic acid bacteria recovery after in vitro digestion
LAB populations were analyzed in three times: i) baseline, ii) after the acid phase (stomach and gizzard phases for swine and chicken simulation, respectively), and iii) after the alkaline phase (intestinal phases in both swine and chicken simulation). The samples were inoculated in MRS agar plates, incubated at 37 ºC for 24-48 h. Each digestion tube was analyzed twice.

Statistical analysis
The quantitative variables will be presented as mean ± standard deviation. The variables were submitted to the D'Agostino-Pearson normality test, and ANOVA for variance analysis, followed by Tukey HSD to mean comparison. The significance level p ≤ 0.05 was used. The analyzes were performed in GraphPad® Prism 8.0 for Windows (San Diego, CA).

MALDI-TOF identification of LAB
The Bruker MALDI Biotyper ® software compares each sample mass spectrum to the reference mass spectra in the database and calculates an arbitrary unit score value between 0 and 3, which reflects the similarity between the sample and the reference spectra. Scores above 2 indicate satisfactory identification, with greater precision according to the proximity to score 3 (Schulthess et al. 2014). From the 17 isolates obtained, 11 were identified as Pediococcus acidilactici, two as Lacticaseibacillus (Lactobacillus) rhamnosus, one as Lacticaseibacillus (Lactobacillus) paracasei paracasei, one as Limosilactobacillus (Lactobacillus) fermentum, and one as a consortium of Lactobacillus delbrueckii bulgaricus and L. fermentum (Table 1).
Other research indicate that bacteria of the same species have bioprotective potential, producing health benefits -for instance, P. acidilactici modulates the intestinal microbiota and stimulates intestinal immunity in rainbow trout (Al-Hisnawi et al. 2019); L. rhamnosus protects the intestinal barrier by stimulating both mucus production and cytoprotective response (Martín et al. 2019); L. paracasei modulates cholesterol biosynthesis (Dehkohneh et al. 2019); L. fermentum improves cognitive behavior and modulates immune response with gut microbiota (Park et al. 2020); and L. delbrueckii bulgaricus presents immunomodulatory capacity, increasing the percentage of natural killers-cells, improving the parameters of the immune risk profile, and increasing in T-cell subsets that are less differentiated (Oyeniran et al. 2020). Thus, the bacterial species isolated in the present study have a scientific history to be used in research as bioprotective agents.
Several researchers showed the inhibitory potential of LAB against Salmonella (Adetoye et al. 2018;Hernández-Aquino et al. 2020;Hai et al. 2021). This action can be mediated by competitive exclusion and the production of antimicrobial substances, such as organic acids (e.g. acetic, lactic, and propionic acids), bacteriocins, and antimicrobial peptides (Adetoye et al. 2018;Corrêa et al. 2019). Among the metabolites produced, lactic acid is cited as the most 1 3 important, mainly due to the capacity of reducing the environmental pH (Adetoye et al. 2018). In scientific databases, P. acidilactici exhibits several antimicrobial effects against Salmonella that can explain the results obtained in this study. Seo and Kang (2020) have demonstrated that bacteriocins produced by P. acidilactici inhibit S. Typhimurium biofilm; when in combination with L. plantarum and Pediococcus pentosaceus, P. acidilactici reduced Salmonella population in alfalfa seeds and sprouts, probably due to the reduction of pH, the accumulation of lactic acid, and the production of bacteriocins (Rossi and Lathrop 2019); in both HT-29 and Caco-2 cells P. acidilactici inhibited the adhesion of S. Typhimurium and S. Enteritidis by competitive exclusion, manifesting strong adhesive properties (Kim et al. 2019);and Pei et al. (2021) have reported that P. acidilactici inhibits S. Typhimurium in vitro possibly due to the production of bacterial substances, such as organic acids, hydrogen peroxide, and bacteriocins.
As P. acidilactici, the species L. fermentum has several action mechanisms against Salmonella that justify the obtained results. The antimicrobial activity of L. fermentum can also be attributed to its production of organic acids (primarily lactic and acetic acids), antimicrobial peptides, and hydrogen peroxide (Zhao et al. 2019). Khan (2019) has demonstrated that L. fermentum inhibits the development of S. Typhimurium by co-aggregation and by the production of both acetic and lactic acids. Heredia-Castro et al. (2021) also reported that L. fermentum inhibitory potential can be reached by the production of bacteriocins.
L. paracasei strains have exhibited several mechanisms against Salmonella, such as the production of bacteriocins (Ye et al. 2021), the reduction of environmental pH (Gomaa et al. 2020), and the production of organic acids (Fadare et al. 2022), which is in accordance to the metabolites produced by other LAB species. Ye et al. (2021) showed that L. paracasei bacteriocins kill pathogens by forming pores on the surface of the cell membrane, and Gomaa et al. (2020) and Fadare et al. (2022) demonstrated that Salmonella has only moderate growth around 4.4 to 5.2 pH, level reached by the acid metabolites of L. paracasei.
The species L. rhamnosus is one of the main bacterial probiotics used in both human and animal health, with several anti-Salmonella mechanisms. Nalle et al. (2021) proposed that the main bactericidal effect was caused by the production of antimicrobial compounds, such as organic acids (lactic acid and acetate acid), hydrogen peroxide, carbon dioxide, diacetyl, and bacteriocins, and by the competitive exclusion. Also, Liu et al. (2021) postulated that the production of various organic acids and antimicrobial peptides, including bacteriocins, might be the underlying mechanisms of the bioprotective potential.
It can be observed that several authors corroborate the inhibitory effects of the LAB evaluated against Salmonella, with extremely similar mechanisms of action. The diffusion of acidic metabolites in the culture medium, in addition to the production of other substances with antimicrobial potential, appears as a key point for the bioprotective potential of the isolated bacteria.
Excluding the values in which the inhibition was considered below the established reference (P. acidilactici PUCPR 006 against S. Enteritidis 9SUSP and P. acidilactici PUCPR 003 against S. Typhimurium ATCC ™ 14028 ® ), the inhibitory halos produced in agar plates ranged from 1.0 ± 0.0 to 9.0 ± 1.8 mm. Some studies have indicated that the analysis of halo inhibition in agar is directly related to the ability to inhibit Salmonella in intestinal lineage cells and in vivo. Mohanty et al. (2019) demonstrated that L. plantarum DM 69 with a high Salmonella inhibition capacity in agar also showed a high Salmonella anti-adhesive activity in HCT-116 cells. Kim et al. (2022) demonstrated that the supply of different LAB with high inhibitory activity in agar to 1to 20-day-old chickens led to a reduction of Salmonella in intestinal samples, in addition to zootechnical improvement.

Morphological and microbiological characteristics of spray-dried powder
The initial counting of viable bacteria, just after the production of the spray-dried powder (D0), ranged from 6.92 ± 0.14 to 10.48 ± 0.20 log CFU/g (Table 3) and, considering the initial bacterial population used in the encapsulation solution, encapsulation efficiency was 76.89 ± 1.54 to 116.48 ± 2.23% (Table 4). The decrease in viability may be due to cellular injuries like the denaturation of DNA and RNA, dehydration of cytoplasmic membranes, rupture, and collapse of the cell membrane due to water removal (Arepally and Goswami 2019); however, the different LAB isolates had experienced similar stresses during drying, and any difference in the survival ratio could be considered the different intrinsic tolerance of cells (Hao et al. 2021). Several authors present different efficiency indexes; Andrade et al. (2019) exhibited high encapsulation efficiency (> 86%) by spray drying using whey powder, whey powder with inulin, and whey powder with maltodextrin to encapsulate Levilactobacillus (Lactobacillus) brevis and L. plantarum; Zhang et al. (2020) obtained about 50% Two bacteria, P. acidilactici PUCPR 014 and P. acidilactici PUCPR 011, reached effectivity > 100%. Although uncommon, this may be due to the auto-aggregation capability of some LAB. Previous research indicates that the composition of the cell membrane can lead to the formation of bacterial granules by homophilic adhesion, which creates an underestimation of the initial bacterial population in the inoculum. The spray drying procedure tends to partially separate the aggregates, increasing the bacterial population in the final product (Doherty et al. 2010;Burgain et al. 2014). Thus, the process applied in this research was considered adequate, with encapsulation effectiveness equal to or greater than that found in the scientific literature databases.
During the analysis of the encapsulated bacterial population, conducted for 12 months, a significant reduction in viable bacteria can be observed, with reductions between 91.16 and 99.54% (Table 3). However, a high bacterial population remained (5.31 ± 0.17 to 9.46 ± 0.09 log UFC/g), with potential to cause beneficial effects after consumption. Research indicates that the consumption of probiotics containing a population above 5 log CFU/g or mL already leads to benefits for productivity and bioprotective action (Lambo et al. 2021).
During 110 days of storage at 4 ºC, Hao et al. (2021) reported 1.0 to 2.2 log CFU/g of Lactobacilli viability losses. Jamilah et al. (2018) found 0.6 to 1.5 log CFU/g of viability losses in both 4 ºC and room temperature storage for 30 days. Thus, the findings in previous studies support the results obtained, with the observed bacterial reduction being considered within normal parameters. Considering that different countries have different regulations on the minimum bacterial population in products intended for feed, an increase in the initial population used in encapsulation can be considered. In addition, as previously mentioned, the concentration at the end of 12 months in this research remained above the minimum levels cited as a guarantor of bioprotective benefits and for animal health and productivity.
SEM was used to visualize the powder morphology. Predominantly spherical morphology of variable size was observed. The absence of free bacteria demonstrates the effectiveness of the encapsulation process (Fig. 1). The encapsulation agents used were chosen due to the maltodextrin non-toxicity, low cost, good solubility, low viscosities even at high solid content, and easy availability; and due to the gum Arabic potential to prevent complete dehydration of cell components and stabilize bacterial cells during drying and storage (Arepally and Goswami 2019). Other encapsulating agents were evaluated, according to the material available on the scientific bases, but they were disregarded, mainly because of the granulometry of the capsules, which is often unfeasible for incorporation into feed (data not shown).

Effects in in vitro digestion
Digestion simulations were performed to evaluate the resistance of encapsulated bacteria to enzymes and other chemical compounds present in the gastrointestinal tract, in addition to the pH variations. At the end of the process, the bacteria must remain viable, with the potential for colonization of the intestinal epithelium. The bacterial population varied significantly throughout the swine digestion simulation, stabilizing at concentrations close to 2.5 log CFU/mL, regardless of the initial population of the analysis, which ranged from 2.26 ± 0.34 to 5.63 ± 0.14 log CFU/mL (Table 5). Similar results were found in the chicken digestion simulation, with an average bacterial concentration of approximately 2.5 log CFU/mL at the end of the analysis, regardless of the initial concentration, which ranged from 2.54 ± 0.15 to 3.62 ± 0.21 log CFU/mL (Table 6).
Differences in the initial population between the bacteria may be related to the different bacteria existing in each powder, with intrinsic characteristics related to the rehydration potential (Sáez-Orviz et al. 2021). Similar results in in vitro digestion were reported by Sáez-Orviz et al. (2021), with a significant reduction in the encapsulated LAB population in the stomach phase, and relative stability in the passage to the intestinal phase. Nguyen et al. (2020) also showed a reduction in the population of encapsulated LAB submitted to gastric fluid simulation, with a reduction of 42.26% of the initial population, and relative stability when submitted to intestinal fluid simulation, with a reduction of 1,68% of the residual population after the gastric phase.
In order to assess whether the remaining population of LAB will have a real bioprotective effect, it is mandatory to carry out in vivo tests. Due to physiological factors and the health condition of individuals in a herd, the effectiveness of  3.01 ± 0.16 eA 2.61 ± 0.10 cB 2.30 ± 0.27 bC a bioprotective agent can be extremely variable. Also, it must be considered the sanitary level and production management techniques, indispensable variables for the validation of a possible bioprotective product (Gardiner et al. 2004). Based on in vivo tests, possible adjustments can be made in the dosage or preparation of the bioprotective product, focusing on promoting animal health and welfare, and improving the production of food intended for human consumption.

Conclusion
The LAB used in this study showed effective potential for the biocontrol of Salmonella in vitro, and are presented as a possibility for application in animal production. The use of maltodextrin and gum Arabic proved to be a good method of encapsulation, ensuring the survival of the bacteria at adequate levels and generating a product with characteristics similar to those already existing and used in the animal production chain. Among the analyzed bacteria, P. acidilactici PUCPR 011, P. acidilactici PUCPR 012, P. acidilactici PUCPR 014, L. fermentum PUCPR 005, L. paracasei paracasei PUCPR 013, and L. rhamnosus PUCPR 010 stand out, with greater inhibitory potential in vitro, in addition to better efficiency in encapsulation and survival rates after simulations of the gastrointestinal tract of production animals.
Further analyses are required to determine the effectiveness of the isolates in vivo, as well as possible improvements in bacterial viability during storage and the increase in viable bacteria in the gastrointestinal tract.