Colonization potential of plantaricin-producing Lactobacillus plantarum SF9C and S-layer carrying Lactobacillus brevis SF9B among gut microbiota

Background: The influence of an S-layer carrying Lactobacillus brevis SF9B and plantaricin-producing Lactobacillus plantarum SF9C on the gut microbiota composition was evaluated in the rats. Considering the probiotic potential of Lb. brevis SF9B, this study aimed to examine the antimicrobial activity of Lb. plantarum SF9C and potential for their in vivo colonization , which could be the basis for the investigation of their synergistic functionality. Results: We identified plantaricin encoding cluster in Lb. plantarum SF9C, a strain which efficiently inhibited Listeria monocytogenes ATCC®19111™ and Staphylococcus aureus 3048. Contrary to a plantaricin-producing SF9C strain S-layer carrying SF9B strain excluded Escherichia coli 3014 and Salmonella enterica serovar Typhimurium FP1 from Caco-2 cells. DGGE analysis of the V2-V3 region of the 16S rRNA gene confirmed the transit of the two selected lactobacilli through the gastrointestinal tract. Microbiome profiling via the Illumina MiSeq platform revealed the prevalence of Lactobacillus spp. in the gut microbiota of rats suggesting their colonization potential in GIT. Conclusion: The combined application of the two strains could influence intestinal microbiota composition, which is reflected through the increased abundance of Lactobacillus genus, but also along with the abundances of other bacterial genera, either in the model of health or aberrant microbiota. Obtained results contribute to the functional aspect of SF9C and SF9B strains which could be incorporated in the probiotic-containing functional foods to beneficially influence gut microbiota composition.

members dominate the microbiota of the sauerkraut and are under the constant competition from other strains for nutrients and space (Collins et al., 2018). The antibacterial activity of Lactobacillus strains is an important factor for the pathogen elimination in the complex microbial communities. Some Lactobacillus strains are able to produce bacteriocins. Bacteriocin synthesis is shown to be one means by which strain producer can gain a competitive advantage and therefore is attractive in the terms of food biopreservation (Collins et al., 2018). Their applications are even expanding to the health since bacteriocin production is recognised as an important probiotic trait and bacteriocins have even been proposed as alternatives to antibiotics (Chikindas et  Lactobacillus brevis SF9B was isolated from the respective spontaneous fermentation. This strain showed desirable functional and technological properties and expresses the S-layer proteins (Slps), which have a functional role in conveying the in vitro survival of SF9B in the intestinal tract (IT) stress conditions (Banić et al., 2018). Besides SF9B, autochthonous isolate SF9C, was isolated in the final stage of spontaneous fermentation. This strain was identified as Lb. plantarum, which is a prevalent species in sauerkraut fermentation, probably due to its competitiveness among autochthonous microbiota (Beganović et al., 2014). Therefore, the aim was to evaluate the competitive advantage potential of this strain, targeting on its possibility to produce bacteriocin. SF9C strain and S-layer carrying Lb. brevis SF9B can act synergistically and the use of substrates occurs by their combined metabolic activity. Next, a possible bacteriocinogenic activity of SF9C strain against Grampositive pathogens Listeria monocytogenes ATCC®19111™ and Staphylococus aureus 3048 was tested. To observe whether this strain has a broader spectrum of antimicrobial activity, antagonistic activity against Gram-negative Escherichia coli 3014 and Salmonella Typhimurium FP1 was evaluated. Apart from evaluating the antimicrobial activity, the induction of plantaricin production in SF9C producing strain was performed by coculturing with common food pathogens. Since the importance of the gut microbiota, both health and disease, is recognized, with preclinical evidence indicating that probiotic Lactobacillus strains may provide a means to ameliorate different disorders (Distrutti et al., 2014;Chen et al., 2016), the goal was to assess colonization potential and the capacity of the two Lactobacillus strains, SF9C and SF9B to induce microbiome alterations in vivo, after joint application, either in healthy or AlCl 3 exposed rats as a model of disturbed microbiota.
PCR-DGGE and sequencing analysis of faeces content was employed to demonstrate if Lb. brevis SF9B and Lb. plantarum SF9C have the potential for in vivo colonization and to influence the microbiota of rat's intestinal tract (IT).

Plantaricin-related genes andWGS of bacteriocinogenic strain SF9C
Plantaricin-related genes plnA, plnEF, plnJ were identified using PCR amplification, suggesting that SF9C could harbour a pln locus in its genome. RAST Table   1). The genome sequence of Lb. plantarum SF9C contains a cluster for biosynthesis of a putative plantaricin. In silico BAGEL4 analysis identified one area of interest (AOI) located at a contig 13. The pln locus of SF9C contains the genes encoding their cognate immunity proteins, whose location is just downstream of the bacteriocin genes, as well as ABC transporters, probably involved in the export of peptides with a double glycine leader ( Figure 2).

Induction of antimicrobial activity in Lb. plantarum SF9C through cocultivation with pathogens
Preliminary results regarding the antibacterial activity of Lb. plantarum SF9C and Lb. brevis SF9B clearly demonstrated the difference in the spectrum of antimicrobial activity among two naturally coexisting strains (data not shown). Interestingly, while SF9C strain demonstrated antibacterial activity against L. monocytogenes ATCC ® 19111™ and S. aureus 3048, strain SF9B failed in inhibition of respective pathogens by agar well-diffusion method (data not shown). Similar, the evaluation of the antibacterial activity of SF9C and SF9B against closely related LAB strains, showed that SF9C compared to SF9B, was more effective in the inhibition of particular strains, with the strongest effect observed against Enterococcus, moderate against Lactococcus and the weakest against Lactobacillus strains by agar well-diffusion method. The results were confirmed by agar spot-test (data not shown). To assess the possibility of the potential induction of bacteriocinogenic activity, Lb. plantarum SF9C was cocultivated with S. aureus 3048 and L. monocytogenes ATCC ® 19111™, respectively. The cocultivation of Lb. plantarum SF9C significantly decreased S. aureus 3048 counts below the detection limit after 2 days ( Figure 3A). The antimicrobial effect was even more pronounced when L. monocytogenes ATCC ® 19111™ was used as a cocultured strain ( Figure 3B). The log CFU/ml values were significantly reduced after 24 hs to non-detectable levels when cocultivated the L. monocytogenes

Inhibition of pathogen adherence to Caco-2 cells by Lb. brevis SF9B
The pathogen competition and exclusion assays by S-layer carrying Lb. brevis SF9B and Lb. plantarum SF9C, respectively on Caco-2 human intestinal cells were performed. Lb.
By contrast, Lb. plantarum SF9C strain did not adhere to Caco-2 cells and did not affect the adhesion of these Gram-negative pathogens (data not shown). Strain SF9B inhibited S.
Typhimurium FP1 adherence to the significant levels in competitive exclusion assay. In exclusion assay, Caco-2 cells exposed to Lb. brevis SF9B before S. Typhimurium FP1 had significantly fewer Salmonella adhered to them (5.708±0.014 log 10 CFU) than Caco-2 cells exposed to S. Typhimurium FP1 alone (6.825 ± 0.099 log 10 CFU). Interestingly, after  Table 2). Influence of Lb. brevis SF9B and Lb. plantarum SF9C on gut microbiome composition Acetylcholinesterase (AChE) activity was assessed in the brain tissue homogenates to monitor the possible influence of AlCl 3 treatment in the rats. In our previous paper (Ledinski et al. 2017), the AChE activity and histopathological and immunohistochemical analyses of the brain, number of plaques and AChE activity was significantly higher in the brains of the AT group, compared to that of the control group (P < 0.05). According to the result, the neuropathological changes were observed in the Alexposed rat group (Figure 4.). Diffuse plaques, also called benign plaques, occurred much earlier than the neuritic plaques in the cerebellum. In the treatments cerebellum of the AT rats were negative on AT8 marker, but positive on 4G8 and Iba1 marker ( Proteobacteria (1.84±0.58%) ( Figure 5A). The Firmicutes and Bacteroidetes phyla accounted for more than 85% of total sequences, similarly to previous findings in the gut microbiota of rats. However, phylum-through genus-wide differences in bacterial abundance were observed among the two groups. In the microbiome of rats exposed to aluminium, the abundance of Firmicutes and Actinobacteria decreased, while the abundance of Bacteroidetes increased compared to the control group. In a bacterial classlevel, the most abundant classes for all groups were Bacilli, Clostridia and Bacteroidia ( Figure 5B). Bifidobacterium was also consistently detected over the samples (Fig. 5A).
Since our main goal was to evaluate the survival and colonisation potential of the two Lactobacillus strains in the model of healthy, but also of experimental animals with disturbed microbiota, the focus was on the evaluation of Lactobacillus abundance. The abundance in Lactobacillus sp. was observed in all treated rat groups, implying well adaptation of SF9B and SF9C to the IT, especially since these two strains are not of an intestinal, but sauerkraut origin. Rat's gut microbiome analysis revealed taxonomic differences of gut microbiota composition influenced by Lactobacillus treatments. The culture-independent PCR-DGGE approach was applied to verify the differences in the gut microbiota composition of faecal samples among groups of rats (data not shown). Culture Furthermore, in a DGGE profile of the healthy rat, an intensive band was consistently detected, which after the sequencing and BLAST search was assigned to Lb. animalis, while in the 3 rd day after Lactobacillus treatment, a faint band corresponding to One can speculate that the SF9C plantaricin activity is potentially enhanced in the presence of L. monocytogenes ATCC®19111™. L. monocytogenes tolerates a broad pH range, the antilisterial potential of SF9C could be assigned to the potential plantaricin synthesis. Contrariwise, SF9C did not exhibit inhibitory activity against enteric Gramnegative E. coli 3014 and S. Typhimurium FP1. This is in agreement with the feature of Lactobacillus bacteriocins which are mostly active towards Gram-positive bacteria.
Since SF9C did not prevent pathogen exclusion on the Caco-2 cells, the potential of Slayer carrying Lb. brevis SF9B, to exclude enteric pathogens was tested. S-layer proteins To our knowledge Lb. brevis SF9B is a non-producing bacteriocin strain, whose genome contains plnI encoding the bacteriocin immunity protein (Banić et al., 2018). This cooperation of coexisting Lactobacillus strains can be also exploited to control bacterial infection for the reestablishment of the disturbed microbiota associated with certain diseases (Dicks et al., 2018). Further, in an attempt to demonstrate the potential of coculture to compete among healthy or disturbed gut microbiota, or even colonize the rat s IT, plantaricin-producing SF9C and S-layer expressing SF9B were orally applied to a healthy and Al-treated rats. Al exposure can cause a variety of adverse physiological effects in humans and animals (Chen et al., 2016;Ledinski et al., 2017). In this study, we aim to outline its adverse effects on the gut microbiome. The changes in intestinal microbiota composition were observed, not only in the abundance of Lactobacillus genus, but also in abundance of other bacterial genera. According to microbiome analysis, Blautia genus was not detected in the healthy rat group but was identified in AT rat group in which its ratio decreased as revealed 3 and 10 days after the Lactobacillus treatment. The ratio of Bacteroides and Phascolarctobacterium genera before the Lactobacillus treatment was higher than in AT rat group compared to healthy groups, but 3 days after the Lactobacillus administration the ratio of this genera was reduced in AT rat group.
Furthermore, the abundance of Bifidobacterium genus reminds unchanged, before and after Lactobacillus treatment, in both healthy and AT rat group. Clostridium and Adlercruetzia genera were evenly present in both groups, before application of two Lactobacillus strains, while after the 3rd and 10th day the ratio of Adlercruetzia genus was

Conclusion
The results of this research supported an enhanced functionality potential of the joined application of SF9C and SF9B strains in vivo. The cooperation between two strains could result in a facilitated adhesion of Lb. plantarum SF9C due to the competitive pathogen exclusion by coexisting Lb. brevis SF9B. At the same, SF9B could benefit from the improved colonization due to plantaricin production by Lb. plantarum SF9C, resulting in a broader spectrum of antimicrobial activity of the coculture against the pathogens. The plantaricin and S-layer expressing Lactobacillus duo could be a promising probiotic candidate, which needs further investigation for an application in functional food or targeted application for the different disorders closely linked with a dysbiosis of gut microbiota.

Bacterial strains, culture media and cultivation conditions
Bacterial strains and cultivation conditions, used in this study, are listed in Table 1.

Human cell line, culture medium and cultivation conditions
Enterocyte-like Caco-2 cells were donated by the Rudjer Bošković Institute, Zagreb, Croatia. Caco-2 cells were grown as monolayer cultures in RPMI 1640 medium (GIBCO, USA), supplemented with 15% of the foetal bovine serum (GIBCO, USA) and 4500 mg/L of glucose. Cells were grown up to confluence at 37 °C and 5% of CO 2 in T-flasks, trypsinised and seeded into 24-multiwell plates. Prior to experiments, cells reached sub-confluence.

Whole genome sequencing and identification of genesencodingbacteriocins
Genomic DNA was prepared according to Frece et al. (2009). Genome sequencing was done using a paired-end approach as essentially described in Banić et al. (2018). Briefly, the Nextera DNA Library Preparation Kit (Illumina, San Diego, CA, USA) was used to construct a library. The library was processed with the Illumina cBot and sequenced on the MiSeq2500 (Illumina, San Diego, CA) pair-end with 300 cycles per read. Contigs were classified as belonging to Lb. plantarum when obtaining the best blastn v2.2.27 hit (Altschul et al., 1990) in the NCBI nt database. RAST server, which identifies proteinencoding, rRNA and tRNA genes, assigns functions to the genes, and predicts which subsystems are represented in the genome (Aziz et al., 2008), was used for the annotation, and the distribution and categorization of all sequenced genes were done. The assembled contigs were compared with so far identified bacteriocins in the NCBI using the tblastn v2.2.27. To further supplement the annotation, BAGEL4 was used to predict genes related to bacteriocin synthesis (van Heel et al., 2018). The input file was the genome sequence of Lb. plantarum SF9C in a fasta file. Conserved genes associated with the bacteriocin synthesis were retrieved using the RAST server (Aziz et al., 2008).

Testing of antimicrobial activity
Antimicrobial activity of Lb. plantarum SF9C and Lb. brevis SF9B against L. monocytogenes ATCC ® 19111™ and S. aureus 3048 was tested by agar-spot test. The agar spot test was strain. Statistical analysis was carried out using ANOVA and the results are reported as mean values of three individual experiments ± standard deviation. One-way analysis of variance (ANOVA) and Tukey tests were performed using VassarStats software to determine significant group differences and means were considered as statistically significant if P < 0.05.

Induction of antibacterial activity
In order to induce bacteriocins synthesis of Lb. plantarum SF9C, cocultivation with L. monocytogenes ATCC ® 19111™ and S. aureus 3048 was performed according to Kos et al. (2008) with slight modifications. The number of viable cells was determined by spot-plate method using the corresponding selective media for each strain: MRS for lactobacilli; Baird-Parker (Oxoid, Hampshire, UK) for S. aureus and ChromoBio (Biolab Diagnostic Laboratory, Budapest, Hungary) for L. monocytogenes, during every 2 hours for the first 10 hours, and after 22, 24 and 48 h of the incubation. Plates were incubated for 24 hours at 37 °C and the number of viable cells was expressed as log CFU/ml. Also, during the experiment, the antimicrobial activity of SF9C strain, in monoculture and coculture, was tested by agar spot method as described above. Experiments were conducted in triplicate and values were expressed as the mean ± standard deviation. One-way analysis of variance (ANOVA) and Tukey tests were performed for statistical analysis.

SF9C on Caco-2 cell line
For exclusion and competition assay experiments, Caco-2 cells were routinely grown in 24well culture plates until confluent differentiated monolayers were obtained. Cellular monolayers were carefully rinsed three times with PBS (pH 7.4) before addition of the bacterial cells. Two separate protocols were followed to assess the ability of viable lactobacilli strains to inhibit E. coli 3014, S. Typhimurium FP1 adhesion to Caco-2 cells. For both assays, Lactobacillus strains and pathogens were routinely cultivated; the cells were harvested and prepared in PBS (pH 7.4) to reach A 620 = 1 (approximately 1 x 10 9 CFU/ml).
The competition assay was performed according to the procedure described by Uroić et al. broth. After overnight incubation at optimal conditions, the cells were harvested by centrifugation at 5000 g for 10 min, resuspended in saline solution and the presence of both strains was microscopically examined. The bacteria suspensions were prepared daily to ensure viability and the CFU was controlled to maintain strictly the number of CFU administered by a rat as it is described in the next chapter.

Experimental animals
Three-months-old male highly inbred Y59 strain rats, weighing 200 to 250 g,

Rat study design and sample collection
Male rats belonging to the Y59 inbred strain were randomly divided into 2 equally sized trial groups and housed three per cage in stainless-steel cages, under the same controlled conditions. The rats were treated daily for four weeks as follows: (a) first trial group represented a model of induced aluminium toxicity (AT) which was established by intraperitoneally injecting AlCl 3 (10 mg/kg) and D-galactose (60 mg/kg) as described by Ulusoy et al. (2015) and (b) second group served as healthy (control) group and was injected with saline solution in the same manner. The rats from each group were further assigned to three groups based on the Lactobacillus treatment. AT model group was divided into: AT1 group (n=3, no Lactobacillus treatment), AT2 group (n=3, orally-cannulated every third day during the four weeks of treatment with a single dose (3x10 9 CFU) of Lactobacillus strains resuspended in saline solution) and AT3 group (n=3, orallycannulated for five consecutive days with a single dose (3x10 9 CFU/mL) of Lb. brevis SF9B and Lb. plantarum SF9C strains resuspended in saline solution, starting 24 h after the last treatment). The control group of rats was also dived into three groups (C1-C3) in the same manner and treatments as the AT model. No side effects were reported following Lactobacillus administration. Before the sacrifice rats were anesthetized using a mixture of ketamine (Narketan ® 10, Vetoquinol AG, Belp Bern, Switzerland) at dose of 75 mg/kg with xylazine (Xylapana ® Vetoquinol Biowet Sp., Gorzow, R. Poland) at dose of 10 mg/kg.
The intestinal mucosal content from each sacrificed rat was scraped and specimens were kept frozen at -80°C until the analysis. The brain was removed and frozen at -80°C or kept in buffered formaldehyde until the analysis. The brain tissue homogenates were used to assess acetylcholinesterase (AChE) activity by colorimetric method. AchE activity is expressed in mol/min/g tissue. The brain samples are prepared according to standard paraffin procedure. Changes related to early-stage Alzheimer's disease were also Photomicrographs were recorded using a digital camera (AxioCam ERc5s, Zeiss) and processed by a computer program morphometric image analysis (AxioCam ERc5s-ZEN2).
The samples were collected from the cages within two hours in after the animals were transferred into clean cages: (a) before the starting treatment (AT1-AT3 and C1-C3 groups); (b) 24 hours after the last treatment (AT1-AT3 and C1-C3 groups) and (c) on third and seventh day following the last probiotic administration (AT3 and C3 group) in triplicates. Faeces samples were stored at -80°C until analysis as described in the next chapter.
Bacterial 16S rRNA sequencing and processing using QIIME faeces of healthy rats sampled before feeding (control), and 3 rd and 10 th day after application of Lactobacillus SF9B and SF9C strains. In both cases, DNA was isolated using instrument (Promega, USA). The V2-V3 region of the 16S ribosomal DNA gene of bacteria in the faeces contents or from pure cultures of lactobacilli was amplified with primers HDA1-GC and HDA2. To identify the lactobacilli, recovered from rat faeces, the V2-V3 region of the 16S rRNA gene of the strains was amplified. The amplicons were sequenced using ABI PRISM ® 3100-Avant Genetic Analyzer (Applied Biosystems). A search of sequences deposited in the GenBank DNA database was conducted by using the BLAST algorithm. The identities of the isolates were determined based on the highest score.

Statistical analysis
All the experiments were repeated three times and the results were expressed as means of three independent trials ± standard deviation (SD). Statistical significance was appraised by one-way analysis of variance. Pairwise differences between the means of groups were determined by the Tukey HSD test for post-analysis of variance pairwise comparisons (http://vassarstats.net/test). Statistical differences between groups were considered significant when P values were less than 0.05.

Funding
This work has been supported by Croatian Science Foundation through the projects IP-2014-09-7009 and IP-2019-04-2237. Authors also acknowledge financial support of Zagreb University, Croatia. The authors declare that there is no conflict of interest.

Author contributions
All authors performed the analysis, prepared the manuscript, and contributed to editing and critical reviewing.

Availability of data and materials
The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request. 6.308 ± 0. * SF9B strain completely lost the adhesion ability to Caco-2 cell lines (Banić et al., 2018) after the removal of the S-layer, and therefore probably did not impair S. Typhimurium FP1 and E. coli 3014 adhesions (the values were the same as in control). **a means statistically significant difference (P < 0.01) of adhered E. coli 3014 and S.    A) The four most abundant phyla detected in the faecal microbiota of control and Al exposed rats, both fed with Lb. plantarum SF9C and Lb. brevis SF9B. B) The distribution of the bacterial classes in the faeces of control and Al exposed rats consuming SF9B and SF9C Lactobacillus strains. Sample designations: first ordered number indicates the day of faeces sampling (0-before; 1-3rd day; 2-10th day after the Lactobacillus administration, respectively); second ordered number indicates rat labels (A(AP)-Al exposed rats; K(KP)-control group). PCR-DGGE analysis of 16S DNA fragments generated with the universal bacterial primers HDA1 and HDA2 from pooled DNA samples of the Lactobacillus species, isolated on MRS agar. Lanes: C -rats before gavage with SF9C and SF9B strains; day 3-3 days after the end of administration SF9C and SF9B of strains; day 10-10 days after the end of administration of SF9C and SF9B strains, S-the ladder of sequences from the pure cultures of Lb. plantarum SF9C and Lb. brevis SF9B, respectively. Bands indicated by symbols were excised and, after amplification, subjected to sequencing.