Contribution of sortase SrtA2 to Lactobacillus casei BL23 inhibition of Staphylococcus aureus internalization into bovine mammary epithelial cells

Probiotics have been considered as a promising strategy to prevent various diseases in both humans and animals. This approach has gained interest in recent years as a potential means to control bovine mastitis. In a previous study, we found that several L. casei strains, including BL23, were able to inhibit the internalization of S. aureus, a major etiologic agent of mastitis, into bovine mammary epithelial cells (bMEC). This antagonism required a direct contact between L. casei and bMEC or S. aureus, suggesting the inhibition relied on interactions between L. casei cell surface components and bMEC. In this study, we have investigated the impact of some candidates which likely influence bacteria host cell interactions. We have shown that L. casei BL23 fbpA retained its inhibitory potential, indicating that L. casei BL23 antagonism did not rely (solely) on competition between S. aureus and L. casei fibronectin-binding proteins for adhesion to bMEC. We have then investigated the impact of four sortase mutants, srtA1, srtA2, srtC1 and srtC2, and a double mutant (srtA1-srtA2) on L. casei BL23 inhibitory potential. Sortases are responsible for the anchoring on the bacterial cell wall of LPXTG-proteins, which reportedly play an important role in bacteria-host cell interaction. All the srt mutants tested presented a reduced inhibition capacity, the most pronounced effect being observed with the srtA2 mutant. A lower internalization capacity of L. casei srtA2 into bMEC was also observed. This was associated with several changes at the surface of L. casei BL23 srtA2 compared to the wild type (wt) strain, including altered abundance of some LPXTG- and moonlighting proteins, and modifications of cell wall structure. These results strongly support the role of sortase A2 in L. casei BL23 inhibition against S. aureus internalization. Deciphering the contribution of the cell surface components altered in srtA2 strain in the inhibition will require further investigation.


Introduction
The increasing amount of data on the relationship between host health and microbiota composition has raised a real interest in the development of probiotic solutions for human and animal health, considering that these probiotics could restore a balanced microbiota and, consequently, ecosystem homeostasis. They can also exert or contribute to a barrier effect with regard to pathogen colonization and its associated symptoms [1][2][3]. In this context, lactic acid bacteria (LAB) are candidates of choice due to their GRAS status (Generally Recognized as Safe) and to the well-documented beneficial effects they exert on intestinal and vaginal human ecosystems [4][5][6].
Probiotic solutions have been considered as a promising strategy for the control of various diseases in humans [7,8] and in animals [9] as well. They are of special interest in animal health and are regarded as an alternative means to reduce massive antibiotic use against infectious diseases, including ruminant mastitis in dairy farms [9][10][11][12][13]. Staphylococcus aureus, a Gram-positive opportunistic pathogen, is one of the main pathogens involved in mastitis, responsible for great economic losses [14,15].
To explore the probiotic potential of LAB in a mastitis context, we recently tested the ability of three Lactobacillus casei strains, including the well-characterized probiotic strain BL23, to compete with S. aureus for bovine mammary epithelial cell (bMEC) adhesion and internalization [16]. In this study, we found that L. casei antagonism with regard to S. aureus internalization into bMEC required a direct contact between L. casei and bMEC or S. aureus. We thus postulated that internalization inhibition relied on interactions between L. casei cell surface components and bMEC.
This study aimed at demonstrating the involvement of cell surface components in the inhibition capacities of L. casei with regard to S. aureus internalization and at identifying some candidates that might be involved in this phenomenon. Many cell surface components are known to interact with host cells in L. casei. It would therefore be impossible to comprehensively investigate each one of them. One of the main mechanisms involved in S. aureus internalization relies on the interaction between S. aureus fibronectin-binding protein and integrin α5 β1 via fibronectin bridging [17,18]. L. casei produces a fibronectin binding protein (FbpA). We thus first postulated that inhibition could rely on a competition for fibronectin attachment through the interaction between L. casei FbpA and fibronectin. To test this hypothesis, we evaluated the inhibition capacity of L. casei BL23 fbpA [19]. Secondly, we evaluated the inhibition capacities of L. casei BL23 sortase mutants [20]. These enzymes are involved in the processing of cell wall-anchored (CWA) proteins. Sortases recognize the LPXTG motif characteristic of CWA proteins, cleave and covalently bind the mature moiety of the protein to the peptidoglycan. Four genes encoding sortases have been identified in L. casei BL23, srtA1, srtA2, srtC1 and srtC2. They potentially control the anchoring of 23 proteins that harbor the LPXTG motif in their C-terminal region. Eight of these proteins exhibit adhesion-related functions. Furthermore, the srtA1-srtA2 double mutant and the srtA2 mutant show a reduced ability of adhesion to Caco-2 cells [20]. We thus hypothesized that sortases and their substrates could contribute to the inhibition capacities of L. casei against bMEC colonization by S. aureus. The impact of each single sortase mutant and the srtA1-srtA2 double mutant on the inhibition of S. aureus colonization and on the colonization capacities of L. casei BL23 on bMEC was explored. The impact of sortase mutations on the cell surface was monitored through analysis of the cell surface proteome and cell shape, revealing several bacterial surface components that could contribute to the inhibitory potential of L. casei.

Material and methods
Bacterial strains and culture conditions L. casei BL23 wild type and the previously constructed mutants are listed in S1 Table [19][20][21][22]. S. aureus Newbould 305 (hereafter referred to as N305), a well-characterized strain of S. aureus isolated from bovine mastitis [23], was used for inhibition tests.
For S. aureus, subculture was carried out in brain-heart infusion medium (BHI; pH 7.4; BD, Le Pont de Claix, France) at 37˚C under agitation (180 rpm), and L. casei strains were sub-cultured in Man Rogosa Sharpe medium (MRS; pH 6.8; BD) at 37˚C without shaking. Subcultures were washed once with phosphate-buffered saline (PBS) and suspended at different concentrations in Dulbecco's modified Eagle's medium (DMEM; pH 7.4; D. Dutscher, Brumath, France) for invasion assays, or in ultra-filtered (UF) milk medium for enzymatic shaving of surface experiments. Alternatively, subcultures for invasion assays were performed on ultra-filtered milk medium so as to be in agreement with conditions used for enzymatic shaving of surface experiments. Whenever necessary (i.e., for L. casei BL23 mutant strains), erythromycin (Sigma Aldrich, Saint Louis, USA) was added to the culture at a final concentration of 5 μg/mL.
Bacterial concentrations in subcultures were estimated by spectrophotometric measurements at 600 nm with a VWR V-1200 spectrophotometer. They were further confirmed by determination of the bacterial population using a micromethod, as previously described [24]. The S. aureus population (in CFU/mL) was determined on mannitol salt agar (MSA; D. Dutscher) after 24 h of incubation at 37˚C. The L. casei population was determined on MRS (pH 5.4) and incubated anaerobically for 48 h at 30˚C in an anaerobic jar.

Mammary epithelial cell culture
The bovine mammary epithelial cell line (bMEC) MAC-T (Nexia Biotechnologies, Quebec, Canada) was cultured in DMEM containing 10% fetal calf serum (FCS) (D. Dutscher), 100 U/ mL penicillin (Ozyme, Montigny-le-Bretonneux, France), 10 mg/mL streptomycin (Ozyme) and 5 μg/mL insulin (Sigma Aldrich) in T75 cell culture flasks (Starlab, Orsay, France). Cells were incubated at 37˚C in a humidified incubator with 5% carbon dioxide (CO 2 ). They were cultured to a confluent monolayer, treated with 0.05% trypsin (Sigma Aldrich) and resuspended in fresh DMEM to obtain a final concentration of 2 x 10 5 cells/mL. Adhesion and internalization assays were performed in 12-well plates with 2 x 10 5 cells seeded per well. Cells were incubated in a humidified incubator with 5% CO 2 at 37˚C overnight to obtain a confluent monolayer.

Internalization assays
Internalization assays were performed according to Bouchard et al. (2013) [16]. Confluent monolayers of MAC-T cells (at 2.5 x 10 5 cells/well) were washed twice with phosphate-buffered saline and incubated for 2 hours at 37˚C and 5% CO 2 with 1 mL of L. casei BL23 wt or mutant strains resuspended in DMEM, with a multiplicity of infection (MOI) of 400:1 or 2,000:1. For internalization inhibition assays, S. aureus N305 (MOI 100:1) and L. casei strains were co-incubated with cells for two hours. Cells were then washed four times with PBS and incubated with DMEM containing 100 μg/mL gentamicin (Pan Biotech, Aidenbach, Germany) for two additional hours in order to kill adhered extracellular bacteria. The MAC-T cell monolayer was then washed four times with PBS, treated with trypsin, centrifuged for 5 minutes at 800 x g and lysed with triton at 0.01%. Internalized bacterial populations were determined using a micromethod (see above). The L. casei population was determined following 48 h incubation at 30˚C on MRS (pH 5.4) in an anaerobic jar. The S. aureus population was determined on MSA after 24 h of incubation at 37˚C.
For internalization inhibition assays, internalization of S. aureus was used as a reference. The inhibition rate of internalization was defined as the internalized S. aureus population in the presence of L. casei with respect to the internalized S. aureus population in the reference experiment.

Enzymatic shaving of surface proteins
Enzymatic shaving of L. casei was adapted from Le Marechal et al. (2015) [25]. L. casei was cultured in UF milk medium [26] until the stationary growth phase (48 h). This medium was retained for technical reasons since cultures of L. casei on MRS led to contaminating signals in nano-LC-MS analysis in relation to the presence of Tween 20. Moreover, UF milk medium contains fewer proteins, which resulted, once again, in fewer "contaminating signals". The bacteria were centrifuged (8,000 x g, 10 min, 4˚C), washed three times in one volume of PBS (pH 8.5) containing 5 mM dithiothreitol (DTT) (Sigma Aldrich) and resuspended in 0.5 mL of the same buffer. The 0.5 mL suspension containing 2 x 10 10 bacteria was treated with 4 mg/mL Sequencing Grade Modified Trypsin (V5111, Promega, Madison, USA) for 1 hour at 37˚C with gentle agitation (180 rpm) ("shaving"). Bacteria were then removed by centrifugation (8,000 x g, 10 minutes, 4˚C) and the supernatant was filtered (0.2 μm filter, Nalgene, Rochester, USA). The viability of L. casei was checked after shaving using a micromethod. For a complete digestion of the peptides released by shaving, the supernatant was further treated with 1 μg of trypsin for 16 hours at 37˚C and 120 rpm. Trypsin digestion was stopped by adding trifluoroacetic acid (TFA) to a final concentration of 0.15% (v/v), and the supernatants containing peptides were then concentrated in a Speed-Vac concentrator prior to nano-LC-MS/MS analysis. Shaving was performed in triplicate on three independent cultures.

Nano LC-MS analyses
Experiments were performed using a nano RSLC Dionex U3000 system fitted to a Q-Exactive mass spectrometer (Thermo Scientific, San Jose, USA) equipped with a nano-electrospray ion source. A preliminary sample concentration step was performed on a C18 pepMap100 reverse-phase column (C18 column: inner diameter (i.d.): 300 μm; length: 5 mm; particle size: 5 μm; pore size: 100 Å; Dionex, Amsterdam, the Netherlands). Peptide separation was performed on a reversed-phase column (PepMap RSLC C18: i.d.: 75 μm; length: 150 mm; particle size: 3 μm; pore size: 100 Å; Dionex), with a column temperature of 35˚C, using solvent A (2% (v/v) acetonitrile, 0.08% (v/v) formic acid and 0.01% (v/v) TFA in deionized water) and solvent B (95% (v/v) acetonitrile, 0.08% (v/v) formic acid and 0.01% (v/v) TFA in deionized water). Peptides were separated using a gradient of 5 to 35% of solvent B over 70 min, followed by 35 to 85% of solvent B over 5 min at a flow rate of 0.3 μL/min. Eluted peptides were directly electrosprayed into the mass spectrometer operating in positive ion mode with a voltage of 2 kV. Spectra were recorded in full MS mode and selected in a mass range 230-2000 m/z for MS spectra with a resolution of 70,000 at m/z 200. For each scan, the ten most intense ions were selected for fragmentation. MS/MS spectra were recorded with a resolution of 17,500 at m/z 200, and the parent ion was subsequently excluded from MS/MS fragmentation for 20 s. The instrument was externally calibrated according to the supplier's instructions.
Peptides were identified from the MS/MS spectra using X!Tandem pipeline software (Plateforme d'Analyse Protéomique de Paris Sud-Ouest (PAPPSO), INRA, Jouy-en-Josas, France, http://pappso.inra.fr) [27]. The search was performed against a database composed of (i) a portion of the UniProtKB database corresponding to L. casei BL23 (http://www.uniprot.org/ taxonomy/543734), and (ii) an in-house database composed of major milk and egg proteins taken from www.uniprot.org (207 proteins in total). The latter was used to remove signals associated with milk proteins since cultures were performed on milk-derived medium. Database search parameters were specified as follows: trypsin cleavage was used and the peptide mass tolerance was set to 10 ppm for MS and 0.05 Da for MS/MS. Oxidation of methionine and phosphorylation on threonine, serine and tryptophan were selected as a variable modification. For each peptide identified, a minimum score corresponding to an e-value below 0.05 was considered as a prerequisite for peptide validation.
Transmission electron microscopy L. casei was grown in UF milk medium until the stationary phase of growth so as to be in accordance with growth conditions used for enzymatic shaving of surface experiments. Bacteria were recovered by centrifugation (6,000 x g, 5 min) and fixed for three hours with gentle agitation (150 rpm) in 0.1 M sodium cacodylate buffer (pH 7.2) containing 2.5% glutaraldehyde. Bacteria were then washed four times and stored at 4˚C overnight in a cacodylate buffer. Post-fixation of the bacteria was performed in a cacodylate buffer containing 1% osmium tetroxide (Electron Microscopy Sciences, Hatfield, USA) for 1 hour, and bacteria were then embedded in 1.5% agar (Biokar Diagnostics, Beauvais Cedex, France). Progressive dehydration was performed with ethanol at concentrations of 50-100% and cells were progressively embedded in Epon (Electron Microscopy Sciences).
Transmission electron microscopy was performed at the MRic platform (University of Rennes 1, Rennes, France). Thin sections of 90 nm were prepared with diamond knives on an Ultracut, placed on cupper grids, stained with 2% uranyl acetate, and subsequently analyzed with JEOL 1400 TEM (Jeol, Tokyo, Japan) operating at 120 kv accelerating voltage.
Digital images were acquired using the Gatan SC1000 Orius1 CCDcamera (4008 x 2672), set up with Gatan Digital Micrograph™ imaging software (Gatan, Pleasanton, USA). TEM images were performed at the Microscopy Rennes Imaging Center platform (MRic-MET, University of Rennes 1, Rennes, France). The same procedure was used to visualize L. casei and/or S. aureus internalized into bMEC, starting from 2.5 x 10 5 cells. In addition to the presence of internalized bacteria in infected MAC-T cells compared to non-infected cells, degradation vesicles, as defined as vesicles containing degraded bacteria, were observed.
Image analysis was performed using ImageJ, version 1 [28]. Two measurements of the bacterial wall thickness were made, with a total of ten measurements in each of the three biological replicates.

Assessment of oxidative stress resistance
Resistance of L. casei wt and mutant strains to oxidative stress was evaluated through exposure to hydrogen peroxide (H 2 O 2 ). Strains were grown on MRS and were harvested in the exponential phase (OD600 nm:~0.8) or stationary phase (24 h) of growth. H 2 O 2 (30% w/v, Sigma Aldrich) was added to 14 mL of culture to a final concentration of 0.1% and 0.2% for the exponential phase and 0.25% and 0.5% for the stationary phase, and incubated at 37˚C without agitation. Samples were collected at time 0 (without addition of H 2 O 2 ), and at 10, 20 and 30 minutes of exposure in order to evaluate the surviving L. casei population. H 2 O 2 was eliminated by adding catalase (Sigma Aldrich) at a final concentration of 10 U/mL and the L. casei population was determined by plating serial dilution on MRS and incubating plates for 48 h at 37˚C.

Analysis of the auto-aggregation capability
The auto-aggregation test was performed according to Ocaña and Nader-Macias (2002) [29] with some modifications. L. casei was cultured in UF milk medium for 48 hours at 37˚C and the percentage of auto-aggregation was calculated as follows: where OD suspension is the optical density (OD) of the cell suspension at 600 nm after 48 hours of L. casei growth under static conditions, and OD total is the cell suspension OD measured on the same culture after homogenization. The total bacterial population was also determined by CFU counting to see if differences in OD were related to differences in auto-aggregation or growth capacities in these conditions.

Statistical analysis
Statistical analysis of the bacterial surface proteome was performed as follows: every peptide identified by tandem mass spectrometry after enzymatic shaving of surface proteins was quantified using free MassChroQ software [30] before data treatment and statistical analysis within R software (R 3.2.2, Project for statistical computing). A specific R package called 'Mass-ChroqR' was used to automatically filter dubious peptides for which the standard deviation of retention time was greater than 40 s and to regroup peptide quantification data into proteins. Three different and complementary methods of analysis were used based on peak counting, spectral counting and XIC (Extracted Ion Chromatogram).
For peak counting and spectral counting analysis, a non-parametric Kruskall-Wallis H test was performed on proteins. A minimum difference of two quantified peaks between strains was retained for peak counting. Proteins with a p-value < 0.05 were considered to be significantly different.
For XIC-based quantification, normalization was performed to take possible global quantitative variations between LC-MS runs into account. For each LC-MS run, the ratio of all peptide values to their value in the chosen reference run was computed. Normalization was performed by dividing peptide values by the median value of peptide ratios. Peptides shared between different proteins were automatically excluded from the dataset, as were peptides present in less than 80% of the samples. Missing data were then imputed from a linear regression based on other peptide intensities for the same protein [31]. Analysis of variance was used to determine proteins with significantly different abundances between mutant strains.
Other statistical analyses were performed with GraphPad Prism software, version 5.01 [32]. Differences between groups were assessed by one-way ANOVA, followed by Bonferroni's Multiple Comparison Test and Student's t-test, considering a P value of less than 0.05. Each experiment was conducted in biological triplicate.

Inhibition of S. aureus internalization into bMEC by L. casei mutant strains
The inhibitory potential of L. casei BL23 wt and fbpA and srt mutant strains against S. aureus N305 internalization into bMEC was evaluated using MOIs of 400:1 and 2,000:1 for L. casei and 100:1 for S. aureus. No significant difference in the S. aureus internalization rate was observed in the presence of L. casei BL23 wt and mutant strains at an MOI of 400:1 (data not shown). A significant reduction of 58% of the internalization rate of S. aureus into bMEC was observed in the presence of L. casei BL23 wt at an MOI of 2,000:1. A similar reduction of the internalization rate of S. aureus into bMEC was observed in the presence of L. casei BL23 fbpA (Fig 1, panel A). On the contrary none of the sortase mutants showed a significant inhibition of S. aureus internalization (Fig 1, panel B). A significant and maximum release of inhibition was observed with the srtA2 mutant strain when compared to the inhibition observed with the control L. casei BL23 wt. In order to confirm that the release of inhibition was due to srtA2 inactivation and not related to a secondary mutation, a second srtA2 mutant strain, named ΔsrtA2 and obtained by double crossing over (S1 Table), was tested. A significant release of inhibition was also observed with this second srtA2 mutant strain compared to L. casei BL23 wt strain as illustrated by an S. aureus internalization rate of 86% +/-26% relative to the control in the presence of L. casei ΔsrtA2.
In order to be in accordance with enzymatic shaving experiments that were done using UFmedium grown cultures for technical reasons (see Material and Methods for details), we further confirmed that similar results were obtained when L. casei subcultures were performed on UF-medium instead of MRS prior to invasion assays (data not shown). Results were further confirmed by transmission electron microscopy (TEM) observation (Fig 2). A greater number of internalized S. aureus was observed when cells were infected by S. aureus alone (Fig 2,

Reduced internalization of L. casei srtA2 into bMEC
Adhesion and internalization capacities of L. casei BL23 wt and sortase mutant strains were determined on bMEC with an MOI of 400:1 and 2,000:1. No significant difference in the The internalization assay of S. aureus alone was used as a reference (control). Internalization rates were then defined as the internalized S. aureus population in the presence of the different L. casei strains relative to the internalized S. aureus population of the reference experiment. Data are presented as means ± standard deviations. Each experiment was done in triplicate, and differences between groups were compared using one-way ANOVA with Bonferroni's Multiple Comparison Test. *: P < 0.05. adhesion rate was observed in sortase mutant strains compared to the BL23 wt strain (data not shown). However, a significant reduction of internalization rates was obtained for srtA2 (37%) and srtA2-srtA1 (24%) strains compared to BL23 wt with an MOI 2,000: 1 (Fig 3). Of note, TEM observation of bMEC infected with BL23 wt or srtA2 showed degradation vesicles in greater abundance in the bMEC infected with the srtA2 mutant strain (Fig 4). The observed reduction in the srtA2 mutant internalization may thus result either from a lower capability of internalization into bMEC and or from a lower survival rate of this strain once internalized into the bMEC.

L. casei srtA2 is less resistant to oxidative stress
Oxidative stress is a major stress encountered by bacteria once they have been internalized into eukaryotic cells. Resistance to oxidative stress was evaluated by measuring the resistance of L. casei BL23 wt and srtA2 strains to H 2 O 2 exposure in the exponential and stationary phases of growth. No significant difference in survival to H 2 O 2 exposure was observed between strains when tested in the exponential phase (data not shown). In contrast, when tested in the stationary phase, L. casei BL23 srtA2 was more sensitive to H 2 O 2 exposure than BL23 wt (Fig 5). The difference was more pronounced after 20 minutes of exposure to 0.5% H 2 O 2 , as illustrated by a Altered profile of surface proteins in L. casei srtA2 The impact of srtA2 disruption on the cell surface proteome of L. casei BL23 was determined by enzymatic shaving. The absence of cell lysis during trypsin treatment was confirmed by measuring L. casei population in the cell lysis buffer before and after shaving. The total  population was 2.7 x 10 10 CFU/mL before trypsin treatment for both BL23 wt and srtA2 mutant strain. Trypsin treatment did not change significantly L. casei population. Hence, following 1h of incubation in cell lysis buffer, L. casei BL23 wt population was 3.20 x 10 10 CFU/ mL and 3.18 x 10 10 CFU/mL in the presence and absence of trypsin respectively. Likewise, L. casei srtA2 population was 3.05 x 10 10 CFU/mL and 3.28 x 10 10 CFU/mL in the presence and absence of trypsin respectively. Data analysis was performed using three different and complementary methods of analysis, based on peak counting, spectral counting and XIC (Extracted Ion Chromatogram). While the two former methods are based on the detection of spectra and thus reveal proteins that exhibit strong differences between conditions, the latter reveals proteins that are present in both conditions but with differential abundance. Complete datasets are presented in S2, S3 and S4 Tables. Proteomic data analysis revealed several changes in the cell surface proteome of BL23 srtA2. Hence, 57, 144 and 196 proteins were differentially abundant at the cell surface of L. casei BL23 srtA2 compared to the BL23 wt strain using the peak counting, spectral counting and XIC methods, respectively. A selection of proteins differentially abundant at the cell surface between wt and srtA2 mutant strains and discussed in this work is presented in Table 1. They include the proteins containing the LPXTG motif, those presumably located at the cell surface or involved in cell surface component biosynthesis, or those known to be involved in the interaction with the host, including moonlighting proteins and proteins involved in cellular stress response.
Two proteins containing the LPXTG motif were identified. Internalin J (InlJ) was found to be more abundant in the BL23 srtA2 surfaceome, whereas Beta-N-acetylglucosaminidase (BnaG), was more abundant in BL23 wt.
Several proteins other than CWA proteins exhibited differential abundance in the cell surface proteome in BL23 srtA2 and are involved in cell wall synthesis or degradation, exopolysaccharide biosynthesis, as well as in cell division processes. A greater abundance was observed in Contribution of SrtA2 to L. casei inhibition of S. aureus internalization into bMEC Contribution of SrtA2 to L. casei inhibition of S. aureus internalization into bMEC BL23 srtA2 for the bifunctional dimerization/transpeptidase penicillin-binding protein 2B (Pbp2B2), UDP-N-acetylmuramyl-tripeptide synthetase (MurE), two proteins involved in exopolysaccharide biosynthesis (RmlB and Wzr), the septation ring formation regulator EzrA, and the signal recognition particle-docking protein (FtsY). On the contrary, a greater abundance was observed in the wt strain for N-acetylmuramoyl-L-alanine amidase, the cell-division proteins DivIVA and FtsX, and four proteins involved in exopolysaccharide biosynthesis (Wzb, RmlC, Wze and Wzd). In addition, three proteins known as moonlighting proteins were found to be less abundant at the cell surface of BL23 srtA2. They included the fructose bisphosphate aldolase and two glyceraldehyde 3-phosphate dehydrogenases. Of note, the N-acetylmuramoyl-L-alanine amidase mentioned above is also known as a moonlighting protein [33].
Finally, another set of differentially abundant proteins included proteins involved in the interaction with the host cell and in the fitness of bacteria within their host. This group is comprised of proteins involved in stress response. Both thioredoxin reductase and thiol peroxidase, involved in oxidative stress response, were more abundant in BL23 srtA2, as were the ClpB chaperone, ClpC and ClpP protease. On the contrary, the glutathione peroxidase and the GrpE chaperone were more abundant in BL23 wt. In addition, three proteins involved in glycogen biosynthesis, GlgB, GlgC and GlgP, were also more abundant in the surfaceome of BL23 wt. Contribution of SrtA2 to L. casei inhibition of S. aureus internalization into bMEC L. casei srtA2 has a thinner cell wall The impact of srtA2 disruption on L. casei cell morphology was investigated using TEM on BL23 grown on UF milk medium until the stationary phase (48 hours). Although no alteration of the cell morphology (e.g., rod shape and length) was observed, a significant reduction (Student's t-test; pval = 3.0 x 10 -15 ) of the cell wall thickness was measured in BL23 srtA2 compared to the BL23 wt control (Fig 6). Cell wall thickness was found to be 26.5 +/-4.6 nm and 16.0 +/-2.6 nm in L. casei BL23 wt and srtA2 strains, respectively.

L. casei srtA2 presents a greater auto-aggregation ability
The auto-aggregation capacity of L. casei BL23 wt and srtA2 strains was assessed after a growth of 48 h in UF milk medium (stationary phase). The auto-aggregation rate of BL23 srtA2 was slightly but significantly higher compared to that of the wt control, as illustrated by auto-aggregation rates of 50% and 66.67% in L. casei BL23 wt and srtA2 strains, respectively (Fig 7). The total bacterial population counts were similar in both strains.

L. casei bnaG retained its ability to inhibit S. aureus internalization
One of the LPXTG motif containing proteins whose abundance was higher in L. casei BL23 wt compared to srtA2 mutant strain was the Beta-N-acetylglucosaminidase BnaG. In order to determine BnaG contribution to L. casei BL23 ability to inhibit S. aureus internalization into bMEC, a bnaG mutant strain was used in competition assays. L. casei BL23 bnaG retained its capacity to inhibit S. aureus internalization, at a level similar to the one obtained with BL23 wt strain (Fig 8).

Discussion
In this work, we investigated the involvement of cell surface components in the capacity of L. casei BL23 to inhibit S. aureus internalization into bMEC. We first focused on FbpA as one of the main mechanisms involved in S. aureus internalization relies on the interaction between S. aureus fibronectin-binding protein and integrin α5 β1 via fibronectin bridging [17,18]. However, we found that L. casei fbpA mutant strain inhibited S. aureus internalization at a level similar to that of the wt strain, suggesting that FbpA was not involved in inhibition mechanisms. Alternatively, we could not exclude the possibility that other BL23 proteins exhibiting fibronectin-binding activities contributed to this competition or that inhibition rely on several mechanisms and bacterial determinants that can compensate each other. It is interesting to note that the contribution of a fibronectin-binding protein (FnbA) to the inhibition of S. Contribution of SrtA2 to L. casei inhibition of S. aureus internalization into bMEC aureus adhesion to bMEC has been demonstrated for another LAB species, namely Weissella cibaria [34]. Such apparent discrepancies more likely point out the species-dependence or even strain dependence of such property. We then focused on the L. casei sortase genes. Sortases specifically ensure the anchoring of CWA proteins in the cell wall of Gram-positive bacteria. The disruption of sortase genes thus directly affects the bacterial surfaceome in terms of CWA protein display and also indirectly affects the surfaceome through other changes in surface components [35]. The effect of sortase gene disruption in L. casei BL23 was thus evaluated with regard to surface properties, colonization capacities and inhibition potential against S. aureus bMEC invasion. All sortase mutant strains resulted in a reduced L. casei BL23 ability to inhibit S. aureus internalization into bMEC. The most pronounced effect was observed with the BL23 srtA2 mutant. A significant release of inhibition was observed with a second srtA2 mutant strain (ΔsrtA2), confirming that the loss of inhibition most likely relied on srtA2 inactivation than on a secondary mutation. The internalization capacity of L. casei BL23 srtA2 alone into bMEC was also reduced. We thus focused on this BL23 srtA2 mutant to investigate changes in surface properties that may contribute to the loss of inhibition, by combining i) surface proteome analysis, ii) TEM observations of bacterial cell shape and of L. casei-infected bMEC, and iii) phenotypic characterizations. The srtA2 mutation resulted in several changes at the bacterial surface, which, for some of them or all together, probably contribute to the loss of inhibition properties. These changes affect cell surface-exposed proteins beyond the direct substrates of sortases (i.e., proteins containing an LPXTG motif and a C-terminal cell wall anchor structure), moonlighting protein abundance, and they include modifications in the cell wall thickness and biosynthesis as well as in the oxidative stress response. Modification of auto-aggregative capacities of L. casei BL23   Fig 7. Auto-aggregation capacities of L. casei BL23 wt and srtA2 strains. Strains were grown in UF milk medium for 48 h at 37˚C. Auto-aggregation was evaluated by spectrophotometry (600 nm) and expressed as the auto-aggregation percentage. Cell suspension OD after growth (48 h) and homogenization was used as a reference (100%). Data are presented as means ± standard deviations. Each experiment was done in triplicate, and differences between groups were compared using Student's t-test. *: P < 0.05.
https://doi.org/10.1371/journal.pone.0174060.g007 srtA2 points out significant changes at the bacterial cell surface that may modify interaction with bMEC and thus competition with S. aureus.
In both pathogenic and probiotic bacteria, some surface proteins are important molecules for colonization and persistence in the ecosystem, as well as for cross-talk with the host cells and with the immune system [36]. Among these surface proteins, some CWA proteins have been described as important proteins since they promote the adhesion or invasion process in eukaryotic cells [37]. In this study, only two proteins bearing the LPXTG motif were found to be differentially abundant in the BL23 wt and srtA2 surfaceome. Internalin J (InlJ) was more abundant in L. casei srtA2, whereas Beta-N-acetylglucosaminidase (BnaG) was more abundant in L. casei BL23 wt. In Listeria monocytogenes, InlJ reportedly contributes to virulence but its precise function is not known and it is not directly involved in cell invasion, contrarily to Internalin A and B [38,39]. Internalin-like proteins have been described in other bacteria, including food-grade bacteria like L. casei or Propionibacterium freudenreichii [25], but their biological function remains unclear [39,40]. Of note, L. casei InlJ does not contain an N-terminal leucinerich repeat (LRR) domain, which is characteristic of internalin family members [39]. Whether InlJ in L. casei BL23 contributes to cell invasion remains to be determined. BnaG, the second LPXTG motif-containing protein identified here, harbors a GH20_DspB_LnbB-like domain (glycosyl hydrolase family 20 (GH20) catalytic domain of dispersin B (DspB), lacto-N-biosidase (LnbB) and related proteins) [41,42]. L. casei BL23 BnaG was recently characterized. It is an extracellular enzyme involved in the metabolism of lacto-N-triose [22], a compound found in human milk oligosaccharides as well as in the glycan moieties of glycoproteins. The GH20_DspB_LnbB-like catalytic domain is also found in dispersin B, a glycoside hydrolase that hydrolyzes the beta-1,6-linkages of PNAG (poly-beta-(1,6)-N-acetylglucosamine), a major component of the extracellular polysaccharide matrix. This polysaccharide is notably produced by several staphylococcal species, including S. aureus and Staphylococcus epidermidis [43]. It is considered as an important adhesin that facilitates adhesion to biomaterials [41,44]. As BnaG was a promising candidate that could account for the srtA2 mutant phenotype, we tested the impact of bnaG deletion on L. casei BL23 capacity to inhibit S. aureus internalization. However, L. casei bnaG mutant strain retained an inhibition capacity similar to L. casei BL23 wt. This indicates that either bnaG was not involved in the inhibition phenotype or that inhibition relied on several surface components, resulting in a limited effect of bnaG deletion.
Moonlighting proteins are a special class of multifunctional proteins, some of which have functions related to adhesion [45]. Here, three proteins, mainly known for their intracellular role in glycolysis and gluconeogenesis and that exhibit moonlighting functions in other bacteria [46], were identified with greater abundance in the BL23 wt strain. They include two glyceraldehyde-3-phosphate dehydrogenase (GAPDH) and a fructose-bisphosphate aldolase (FBA). GAPDH was found extracellularly and shown to bind to human colonic mucin in Lactobacillus plantarum LA 318. Likewise, in the group A Streptococci, GAPDH displayed multiple binding activities to fibronectin, lysozyme, myosin and actin [47,48]. Likewise, FBA has been found to be surface-exposed in several bacteria, including pathogens such as Neisseria meningitidis, which causes meningitis and septicemia [49]. In this species, FBA was shown to contribute to the adhesion to human brain microvascular endothelial and HEp-2 cells.
An additional moonlighting protein, N-acetylmuramoyl-L-alanine amidase, was found in greater abundance in L. casei BL23 wt compared to BL23 srtA2. N-acetylmuramoyl-L-alanine amidase is an enzyme that cleaves the amide bond between N-acetylmuramoyl residues and Lamino acid residues in bacterial cell walls. This hydrolase was also identified in Mycobacterium tuberculosis as an adhesin capable of binding to fibronectin and laminin [50]. The lower relative abundance of all these moonlighting proteins in BL23 srtA2 may contribute to its altered colonization properties and inhibition capacity.
In addition to surface proteins, other components of the bacterial surface, including peptidoglycan and exopolysaccharides, are known to interact with the host. Peptidoglycan plays an important role in protecting the bacterial structural integrity and allows the covalent or noncovalent anchoring of various structures such as teichoic acids, polysaccharides and proteins [51]. Peptidoglycan fragments, when released after degradation, can induce an immune response in the host cells [52]. Exopolysaccharides (EPS), including capsular polysaccharide, wall PS or secreted ones, have also been shown to be involved in interactions with the host and, notably, interactions with the immune system in both probiotic and pathogenic strains [51,53]. Electron microscopy analysis showed that the cell wall was thinner in L. casei BL23 srtA2. When considering the other changes identified here in the BL23 srtA2 surfaceome, this suggests a possible imbalance in the bacterial cell wall turnover and/or EPS biosynthesis. Three proteins involved in the cell wall synthesis process were differentially abundant on the cell surface in L. casei BL23 srtA2 compared to the wt strain. These include the bifunctional dimerization/transpeptidase penicillin-binding protein 2B (Pbp2B2) and a UDP-N-acetylmuramyltripeptide synthetase (MurE) that were more abundant on the BL23 srtA2 surface. On the contrary, an N-acetylmuramoyl-L-alanine amidase was more abundant in the wt surfaceome. This peptidoglycan hydrolase cleaves specific bonds in peptidoglycan, thus contributing to the insertion of newly synthesized peptidoglycan subunits and to the separation of daughter cells following division [36]. We did not microscopically observe changes in cell size or cell separation, but it can be noted that four proteins involved in cell separation were also differentially abundant on the cell surface on wt and srtA2 mutant strains. In addition, several proteins encoded within the EPS biosynthesis gene cluster displayed modified abundance in the srtA2 mutant compared to L. casei BL23 wt. They include two proteins involved in NDP-sugar (EPSprecursors) biosynthesis (RmlB, RmlC) and three proteins involved in polysaccharide-chainlength determination (Wzd, Wze, Wzb), which were less abundant in L. casei BL23 srtA2 than in the wt strain, with the exception of RmlB [53,54]. Wzr was only detected in the L. casei BL23 srtA2 surfaceome. The wzr gene is organized in the opposite transcriptional sense and probably encodes a transcriptional regulator whose exact function is unknown. EPSs have been shown to influence intercellular interactions and adhesion to biotic and abiotic surfaces by contributing to the cell surface physicochemical properties and/or by hiding some of the surface-exposed proteins, including some adhesins [55]. EPSs thus play a role in the formation of microcolonies and biofilms. Inactivation of wzb, which is involved in polysaccharide-chain-length determination in Lactobacillus rhamnosus GG, resulted in altered biofilm formation capacities [56]. Whether modulations of peptidoglycan and /or EPS biosynthesis account for the thinner cell wall of L. casei BL23 srtA2 remains to be determined. Similarly, the impact of these modifications on L. casei BL23 interactions with bMEC is subject to further investigation.
Interestingly, when observing internalized L. casei BL23 wt and srtA2 strains by TEM, a higher number of degradation vesicles were observed in bMEC infected with the mutant strain, suggesting that the lower internalization rate may be due to a lower capacity to invade cells, combined with a lower capacity to survive intracellularly. This was supported by a lower resistance of BL23 srtA2 to oxidative stress, one of the harmful mechanisms used by eukaryotic cells to reduce bacterial viability [57]. In agreement with an impact of the srtA2 mutation on oxidative stress resistance, three oxidative stress-related proteins, thioredoxin reductase, thiol peroxidase and glutathione peroxidase, were found to be differentially abundant in BL23 wt and srtA2 surfaceomes. Several additional proteins related to stress response, including chaperones and proteases, were also differentially abundant in BL23 wt and srtA2 surfaceomes. ClpC was notably shown to contribute to the persistence capacity of L. plantarum WCFS1 in murine GIT [58]. Finally, three proteins involved in glycogen metabolism were more abundant in the surfaceome of the wt strain. Glycogen metabolism is associated with energy storage and various physiological functions, including colonization and persistence [59]. Alteration of stress response capacity and, consequently, of intracellular survival of L. casei BL23 srtA2 may also result in modifications of its inhibitory capacities against bMEC invasion by S. aureus.
In conclusion, these results strongly support a major role, either direct or indirect, of SrtA2 in L. casei BL23 inhibition capacity with regard to S. aureus internalization. The analysis of srtA2 disruption on BL23 showed that it resulted in pleiotropic effects, including several changes in the surface proteome, beyond the LPXTG substrates, and changes at the cell wall level. Altogether, these results provide numerous presumptions about the functions (cell wall biosynthesis, oxidative stress and adhesion) putatively involved in this inhibition. They therefore open avenues for continuing research into this phenomenon since the candidates identified here could be targets for further characterization of this promising S. aureus-L. casei-bMEC tripartite interaction.
Supporting information S1