Mutation of the Surface Layer Protein SlpB Has Pleiotropic Effects in the Probiotic Propionibacterium freudenreichii CIRM-BIA 129

Propionibacterium freudenreichii is a beneficial Gram-positive bacterium, traditionally used as a cheese-ripening starter, and currently considered as an emerging probiotic. As an example, the P. freudenreichii CIRM-BIA 129 strain recently revealed promising immunomodulatory properties. Its consumption accordingly exerts healing effects in different animal models of colitis, suggesting a potent role in the context of inflammatory bowel diseases. This anti-inflammatory effect depends on surface layer proteins (SLPs). SLPs may be involved in key functions in probiotics, such as persistence within the gut, adhesion to host cells and mucus, or immunomodulation. Several SLPs coexist in P. freudenreichii CIRM-BIA 129 and mediate immunomodulation and adhesion. A mutant P. freudenreichii CIRM-BIA 129ΔslpB (CB129ΔslpB) strain was shown to exhibit decreased adhesion to intestinal epithelial cells. In the present study, we thoroughly analyzed the impact of this mutation on cellular properties. Firstly, we investigated alterations of surface properties in CB129ΔslpB. Surface extractable proteins, surface charges (ζ-potential) and surface hydrophobicity were affected by the mutation. Whole-cell proteomics, using high definition mass spectrometry, identified 1,288 quantifiable proteins in the wild-type strain, i.e., 53% of the theoretical proteome predicted according to P. freudenreichii CIRM-BIA 129 genome sequence. In the mutant strain, we detected 1,252 proteins, including 1,227 proteins in common with the wild-type strain. Comparative quantitative analysis revealed 97 proteins with significant differences between wild-type and mutant strains. These proteins are involved in various cellular process like signaling, metabolism, and DNA repair and replication. Finally, in silico analysis predicted that slpB gene is not part of an operon, thus not affecting the downstream genes after gene knockout. This study, in accordance with the various roles attributed in the literature to SLPs, revealed a pleiotropic effect of a single slpB mutation, in the probiotic P. freudenreichii. This suggests that SlpB may be at a central node of cellular processes and confirms that both nature and amount of SLPs, which are highly variable within the P. freudenreichii species, determine the probiotic abilities of strains.


INTRODUCTION
Probiotic bacteria are defined as "living microorganisms which when administered in adequate amounts confer a health benefit on the host" (Food and Agriculture Organization of the United Nations and World Health Organization, 2002). This term was further used by International Scientific Association for Probiotics and Prebiotics (ISAP) (Hill et al., 2014). Clinical proofs of efficiency were indeed obtained, in the context of antibioticand Clostridium difficile-associated diarrhea (Rondanelli et al., 2017), lactose intolerance (Oak and Jha, 2018), irritable bowel syndrome (IBS) (Ford et al., 2014), and ulcerative colitis, one of the disorders that constitute Inflammatory bowel disease (IBD) (Plaza-Díaz et al., 2017). The mechanisms underpinning these effects mainly belong to three categories: (i) metabolic effects, (ii) modulation of the gut microbiota, and (iii) probiotic/host molecular interactions. Although lactobacilli and bifidobacteria were mainly considered for probiotic usage, promising effects were also reported for dairy propionibacteria .
The probiotic properties of dairy propionibacteria are strain-dependent and include microbiota modulation, apoptosis modulation in colonic cells and immunomodulation. Some of these probiotic abilities were validated at the clinical level. Microbiota modulation by dairy propionibacteria result in a bifidogenic effect (Roland et al., 1998;Seki et al., 2004;Suzuki et al., 2006). The corresponding molecular mechanisms were elucidated, and two molecules are shown to be involved in bifidogenic effect: 1,4-dihydroxy-2-naphtoic acid (DHNA) and 2-amino-3-carboxy-1,4-naphthoquinone (ACNQ) (Isawa et al., 2002;Furuichi et al., 2006). The pro-apoptotic effect of dairy propionibacteria was evidenced using in vitro cellular models (Jan et al., 2002) and animals models (Lan et al., 2008). This effect is mainly due to the production of the short chain fatty acids (SCFA) acetate and propionate by dairy propionibacteria (Lan et al., 2007;Cousin et al., 2016). The anti-inflammatory effect was suggested in IBD patients (Mitsuyama et al., 2007) and confirmed in animal colitis models (Foligné et al., 2010;Plé et al., 2015Plé et al., , 2016. Immunomodulatory properties are due to several metabolites as SCFAs and to cells wall component . Indeed, surface proteins considered as microorganism-associated molecular patterns (MAMP) play a pivotal role in interaction with host's immune system (Deutsch et al., 2012;Le Maréchal et al., 2015). This includes SlpB and SlpE, surface proteins anchored to the cell wall via surfacelayer homology (SLH) domains (Deutsch et al., 2017;do Carmo et al., 2018). Indeed, mutation of slpB and slpE genes clearly affected the immunomodulatory properties of P. freudenreichii (Deutsch et al., 2017). We have recently shown that SlpB is involved both in immunomodulation and in adhesion to cultured human intestinal epithelial cells .
In probiotic bacteria, extractable surface proteins play several role in bacterium/host interaction, protection against environmental stresses, inhibition of pathogens, survival within the host digestive tract, and determination or maintenance of cell shape (Hynönen and Palva, 2013;do Carmo et al., 2018). In this study, we investigated the impact of slpB gene mutation on the physiology of P. freudenreichii CIRM-BIA 129 using a proteomic approach. In this purpose, we investigated alterations in extractable surface proteins and in the whole-cell proteome. We compared wild-type CIRM-BIA 129 with mutant CB129 slpB. We report pleiotropic effects of this single mutation on physicochemical properties of this propionibacteria.

Zeta Potential Analysis
Electrophoretic mobility (zeta potential) was determined according to the well-described protocol of Schär-Zammaretti and Ubbink (2003). Bacteria were harvested from a 5 mL stationary phase culture by centrifugation (8.000 × g, 10 min, room temperature) and washed twice with a PBS buffer pH 7.0. Cell count of the final suspensions was approximately 10 8 CFU/ml. The pellet was resuspended in a 10 mM KH 2 PO 4 solution (pH 7.0). The electrophoretic mobility was measured by using a ZetaSizer nanoZS (Malvern Instruments, Malvern, United Kingdom) and a glass capillary Zetasizer Nanoseries DTS 1061 (Malvern Instruments, Malvern, United Kingdom) as the electrophoretic cell. Electrophoretic mobilities were converted to the ζ-potential using the Helmholtz-Smoluchowski equation (Schär-Zammaretti and Ubbink, 2003). All experiments were done in biological and technical triplicates.

Cell Surface Hydrophobicity Analysis
The Microbial Adhesion To Hydrocarbons (MATH) assay was performed as described by Kos et al. (2003). The optical density of the stationary phase bacteria was adjusted to an OD 650 nm = 1. The samples were centrifuged for 5 min, 10,000 × g at room temperature and the pellets washed twice with the same volume of PBS pH 7.0 prior to resuspension in 15 mL of 0.1M KNO 3 , pH 6.2. An aliquot of each bacterial suspension (4 ml) was mixed with 1 mL of the solvent (Xylene, chloroform and ethyl acetate), incubated for 5 min at room temperature and mixed by vortex during 120 s. Subsequently, samples were incubated during 60 min to allow phases separation, the aqueous phase was carefully removed and absorbance (OD 600 nm) was determined as above. Cell surface hydrophobicity in terms of per cent (H %) was calculated using the following formula: H % = (1-A1/A0) × 100. All experiments were done in biological and technical triplicates.

Transmission Electron Microscopy Assay
Cultures were grown on YEL medium to an OD 650 nm of 1. Transmission electron microscopy was executed after bacteria were washed with PBS and fixed overnight at 4 • C in 0.1 M sodium cacodylate buffer (pH 7.2) containing 2% glutaraldehyde. Fixed bacteria were rinsed and stored at 4 • C in cacodylate buffer containing 0.2 M sucrose. They were then postfixed with 1% osmium tetroxide containing 1.5% potassium cyanoferrate and 2% uranyl acetate in water before gradual dehydration in ethanol (30% to 100%) and embedding in Epon. Thin sections (70 nm) were collected on 200-mesh cooper grids and counterstained with lead citrate before examination. The thickness of the cell wall was determined using the imageJ software in both strains analyzed by Transmission Electron Microscopy (TEM) as described (Foligné et al., 2010;Deutsch et al., 2012).

Stress Conditions Challenge
P. freudenreichii strains in stationary phase were subjected to lethal doses of different stresses. The acid challenge was carried out at pH 2.0 for 1 h as described (Jan et al., 2000). The bile salts stress was induced by adding 1.0 g/l of bile salts for 60 s as described (Leverrier et al., 2003). For the thermic stress, bacteria were heated for 30 min at 63 • C. Viable cells were determined by serial dilutions of samples made up in peptone water (0.1% bacteriological peptone, Kasvi, Brazil), adjusted to pH 7.0 and containing 0.9% NaCl, into YEL medium containing 1.5% agar. CFU were counted after 6 days of anoxic incubation at 30 • C (Anaerocult R A -Merck Millipore). All experiments were done in biological and technical triplicates.

Whole-Cell Protein Extraction and Preparation of Total Bacterial Lysates
The optical density of the stationary phase bacteria was adjusted to an OD 650 nm = 1. The cultures were centrifuged for 5 min, 10,000 × g at room temperature and the bacterial pellets from biological triplicates were resuspended in 1 mL of lysis buffer containing 42% urea, 15% thiourea, 4% SDC (sodium deoxycholate), 12.5 mM Tris-HCl pH 7.5 and 1.5% dithiothreitol (DTT) with 10 µL of protease inhibitor (GE HealthCare, Pittsburgh, USA). Next, whole-cell proteins were extracted as described  and quantified by Qubit 2.0 fluorometer (Invitrogen, Carlsbad, USA). 100 µg of each protein extract were denatured with 0.2% of RapiGest SF solution (Waters, Milford, USA) at 80 • C for 15 min, reduced with 100 mM DTT at 60 • C for 30 min, and alkylated with 300 mM iodoacetamide at room temperature in a dark room for 30 min (Leibowitz et al., 2017). Subsequently, proteins were enzymatically digested with 10 µl of trypsin at 0.5 µg.µL −1 (Promega, Madison, USA), and the digestion stopped with the addition of 10 µL of 5% trifluoroacetic acid (TFA) (Sigma Aldrich, Saint Louis, USA) . Tryptic peptides were subjected to SDC removal (Lin et al., 2010), desalted using C18 MacroSpin Columns (Harvard Apparatus, Holliston, USA), according to the manufacturer's instructions, and dried under vacuum in the Eppendorf TM Vacufuge TM Concentrator (Eppendorf, Hamburg, Germany) (Wong et al., 2013). Prior to injection, the peptides were resuspended in 20 mM ammonium formate (Sigma Aldrich) and transferred to Waters Total Recovery vials (Waters).
The measurements for all samples by mass spectrometer was operated in resolution mode with a typical m/z resolving power of at least 25,000 Full Width at Half Maximum (FWHM) and an ion mobility cell that was filled with helium gas and a cross-section resolving power at least 40 / . The effective resolution with the conjoined ion mobility was 25,000 FWHM. Analyses were performed using nano-electrospray ionization in positive ion mode nanoESI (+) and a NanoLock-Spray (Waters) ionization source. The multiplexed data-independent acquisition (DIA) scanning with added specificity and selectivity of a non-linear "T-wave" ion mobility (HDMS E ) device was performed with a Synapt G2-Si HDMS mass spectrometer (Waters) (Giles et al., 2011). Synapt G2-Si HDMS was automatically planned to switch between standard MS (3 eV) and elevated collision energies HDMS E (19-45 eV) applied to the transfer "T-wave" collision-induced dissociation cell with nitrogen gas. The trap collision cell was adjusted to 1 eV, using a millisecond scan time that was previously adjusted based on the linear velocity of the chromatographic peak that was delivered through nanoACQUITY UPLC (Waters). A minimum of 20 scan points was generated for each single peak, both in low-energy and highenergy transmission at an orthogonal acceleration time-of-flight (oa-TOF) and a mass range from m/z 50 to 2,000.
Mass spectrometric analysis of tryptic peptides was performed using a mass spectrometer equipped with a T-Wave-IMS device (Waters) in MS E mode following the method previously described (Distler et al., 2014). Stoichiometric measurements based on scouting runs of the integrated total ion account prior to analysis were performed to ensure standardized molar values across all samples. Therefore, the tryptic peptides of each strain were injected with the same amount on the column. The radio frequency (RF) offset (MS profile) was adjusted such that the nanoESI-HDMS E data were effectively acquired from m/z 400 to 2000, which ensured that any masses less than m/z 400 that were observed in the high energy spectra with arose from dissociations in the collision cell .
The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE (Vizcaíno et al., 2016) partner repository with the dataset identifier PXD009804.

Proteins Identification and Quantification
HDMS E raw data were processed using Progenesis QI for Proteomics (QIP) v.2.0 (Nonlinear Dynamics, Newcastle, UK) as described by Kuharev et al. (2015). For proteins identification, the peptides were searching against a P. freudenreichii strain CIRM-BIA 129 database as described above. The reversed sequences were joined together to the original sequences using ProteinLynx Global Server (PLGS) v 3.0.2 (Waters) database management tool. The reversed sequences were used to calculate the false positive rate during identification process. Next, the following parameters were used for peptide identification: digest reagent = trypsin; maximum missed cleavage = one; maximum protein mass = 600 kDa; modifications: carbamidomethyl of cysteine (fixed), acetyl N-terminal (variable), phosphoryl (variable), oxidation of methionine (variable); search tolerance parameters: peptide tolerance = 10 ppm, fragment tolerance = 20 ppm, maximum false discovery rate (FDR) = 1%.
The protein-level quantitation was performed with Relative Quantitation using Hi-N algorithm. Proteins identified with at least two peptides and presents in at least two of the three biological replicates were considered . The proteins list was exported by the function "export protein measurements" and was used to subsequent bioinformatics analysis. Proteins were considered to be differentially expressed between mutant and wild type if there were a significant (p < 0,05, ANOVA) change in expression ≥ 2-fold (log2 ratio ≥ 1.0). A volcano plot was generated to visualize the differentially expressed proteins across these strains.

Extraction of Genomic DNA of the CB129 slpB Strain
Genomic DNA was extracted from CB129 slpB culture grown in YEL medium supplemented with chloramphenicol (10 µg ml −1 ), during the phase (76 h at 30 • C). Samples was centrifuged at 4 • C and 8,000 × g for 10 min. Bacterial pellets were resuspended in 1 ml Tris/EDTA/RNase [10 mM Tris/HCl (pH 7.0), 10 mM EDTA (pH 8.0), 300 mM NaCl, 50 µg RNase A ml −1 ] with 50 mg of Glass beads VK01 and cell lysis occurred in Precellys R 24 by 2 cylces of 15 s at 6,500 rpm. DNA was purified using phenol/chloroform/isoamyl alcohol and precipitated with ethanol according with Sambrook and Russell (2001). DNA concentrations were determined spectrophotometrically in Thermo Scientific NanoDrop 1000.
Genome Sequencing, Assembly and Annotation of the CB129 slpB Strain CB129 slpB strain sequencing libraries were constructed using 100 ng of genomic DNA. The gDNA was sheared with the Ion Shear TM Plus Reagents Kit and barcoded using the Ion Xpress Fragment Library kit and Ion Xpress TM Barcode Adapters (Life Technologies, USA), according to the manufacturer's protocol. Size selection of ∼400 bp was performed with 2% E-Gel R SizeSelect TM Agarose Gels (Invitrogen, USA) and quantified with the Ion Library Quantitation Kit. The libraries were amplified with the OneTouch Template 400 kit on the Ion One Touch TM 2 (Life Technologies) and enriched on the Ion OneTouch TM ES (Life Technologies). Genomic libraries were enriched using Ion PI TM Hi-Q TM Sequencing Polymerase in the Ion 318 TM v2 Chip, according to the manufacturer's protocols, and they were sequenced using Ion Torrent Personal Genome Machine (PGM). The amplification processes were performed using Ion PGM TM Hi-Q TM Sequencing 400 Polymerase with required 1,100 flows. Finally, signal processing was performed using Torrent Suite 4.2.1 to conclude the sequencing process.
The P. freudenreichii CIRM-BIA 1 strain (without the slpB gene) was used for comparative analysis as it is a reference from INRA strain collection strain and P. freudenreichii JS17 strain was used due to the presence of the s-layer gene slpB. CLC Genomics Workbench 7.0 (Qiagen, USA) was used to map the raw reads against the reference genome and to generate the consensus sequence used to the gap filling. The plasmid that integrated within and disrupted the slpB gene was not found in the scaffold, but its sequence was found within the contigs that were excluded during the scaffold generation. It was manually inserted to the scaffold by mapping its ends on the slpB gene and using the overlap sequences as coordinates for the insertion. The insertion was validated by mapping the reads on the assembly and checking for mismatches on the regions flanking the plasmid. The genome of CB129 slpB strain was annotated automatically using RAST pipeline (Aziz et al., 2008;Brettin et al., 2015).

Bioinformatics Analyses
The predicted proteins of CB129 slpB and WT strain were analyzed using the SurfG+ v1.0 tool (Barinov et al., 2009) to predict sub-cellular localization. It enabled the classification of proteins within the following categories: cytoplasmic (CYT), membrane (MEM), potentially surface-exposed (PSE) and secreted (SEC). The prediction of orthologous groups by functional category the sequences was performed using Cluster of Orthologous Genes (COG) database version 2014db (Galperin et al., 2015). The COG database search was performed using an in-house script (available at https://github.com/aquacen/blast_ cog). The number of predicted proteins in relation to subcellular localization and functional category were visualized in plots generated using TIBCO SpotFire software 7.0 (TIBCO, Boston, USA) from the protein list exported of QIP. The Interactivenn web-based tool (Heberle et al., 2015) was used to evaluate the shared proteins among strains through Venn diagram.
Protein-protein interaction (PPI) network was constructed using interolog mapping methodology and metrics according to Folador et al. (2014). To generate a preview of the interaction network was generated using Cytoscape version 2.8.3 (Shannon et al., 2003) with a spring-embedded layout. To indicate the reliability of our predicted PPIs in the database STRING, the network was selected using the score 500 (0.5). In the PPI network, the interactions with score close to 500 are with red or yellow lines and, above 700 in dark green lines. The score indicating how much the pair of proteins in the interaction is similar (homologous) to the interaction according to the database. In the PPI, they interact with at least 65% identity with at least 65% coverage.

Statistical Analyses
Growth curve, MATH assay, Zeta potential measure, and stress challenges were performed with three technical replicates and three biological replicates. The results were expressed as means ± standard deviations. Statistical analyses were performed in GraphPad Prism Software version 7 (GraphPad Software) using Student's t-test, one-way or two-way ANOVA with SIDAK's or Tukey post-hoc analyses for multiple comparisons. Asterisks represent statistically significant differences and were indicated as follows: * p < 0.05; * * p < 0.01; * * * p < 0.001.

Impact of slpB Mutation on P. freudenreichii Extractable Surface Proteins
SLPs play a key role in probiotic/host interactions and we have shown that such interactions are impaired in an slpB mutant. Electrophoretic analysis of guanidine extracts confirmed the disappearance of the corresponding SlpB protein . In the present study, we further investigated these extractable fractions in order to decipher the impact of such a single mutation on the inventory of SLPs, and more widely, of extractable surface proteins, including surface layer associated proteins (SLAPs). Using nanoLC-MS/MS, we identified 40 surface extractable proteins in CB129 slpB strain, yet 33 in the parental wild-type CIRM BIA 129 one ( Table 1). The core of extractable proteins, non-covalently bound to the cell wall, common to mutant and parental strains, was composed of 23 proteins, including solute-binding protein of the ABC transport system (BopA), internalin A (InlA), surface protein with SLH Domain E (SlpE), and surface-Layer Protein A (SlpA). Moreover, it comprised a series of cytoplasmic proteins involved in different biological processes like Heat shock 70 kDa protein 1 (HSP70 1), Clp chaperone, GroL1 and GroL2, Elongation factor Tu, and subunits of Methylmalonyl-CoA mutase and subunits of Methylmalonyl-CoA carboxytransferase. Among extractable proteins specific of the CB129 slpB, we identified proteins involved in metabolic processes like Coenzyme A transferase involved in acetyl-CoA metabolic process and Pyruvate phosphate dikinase Pyruvate synthase involved in pyruvate metabolic process. Furthermore, this specific subset also comprised another protein involved in stress response (HSP70 2). As expected, the SlpB protein was found only in the parental wild type CIRM BIA 129, yet not in the CB129 slpB mutant.

Impact of slpB Mutation on P. freudenreichii ζ-Potential and Cell Surface Hydrophobicity
Propionibacterial SLPs, with a low isoelectric point, confer negative charges to the cell surface. In order to identify if the net surface charge was altered in the mutant strain, we conducted ζ-potential and cell surface hydrophobicity assays in both strains. As shown in the Figure 1A, the WT strain exhibited a zeta potential of −21.73 ± 1.63 mV, reflecting a negative net charge, in accordance with the low isoelectric point of P. freudenreichii SlpB protein. By contrast, mutation of slpB gene significantly affected the zeta potential of the CB129 slpB strain, which        Frontiers in Microbiology | www.frontiersin.org was −6.75 ± 0.55 mV, showing a reduced electronegativity, in accordance with a disorganization of the S-layer at the bacterial cell surface. As shown in Figure 1B, the wild type strain also showed a high affinity to the hydrocarbons tested, whereas the CB129 slpB mutant showed a decreased adhesion, whatever the hydrocarbon used in the assay. Adhesion, respectively to mutant and WT strains, were as follow: to Xylol, 0.33 ± 0.52% and 43.67 ± 6.31%, to Chloroform 16.5 ± 10.7% and 75 ± 5.88, and to Ethyl Acetate 5.33 ± 7.17% and 43.83 ± 5.74%. Cell surface properties being drastically affected, we then sought morphological changes caused by the mutation (Figure 2). Both strains exhibited a similar cell wall thickness, 24.33 ± 0.4154 nm and 24.90 ± 0.4154 nm, respectively. No significant difference in term of bacteria morphology, cell wall thickness and shape was observed between the two strains using transmission electron microscopy.

Impact of slpB Mutation on P. freudenreichii Growth and Stress Tolerance
A single mutation, inactivating a key gene, may affect bacterial fitness and thus probiotic efficacy. We therefore monitored P. freudenreichii growth and tolerance toward acid, bile salts and heat challenges, in the wild type and in the mutant. The growth curves showed a similar pattern for both strains ( Figure 3A). The bacterial count at the stationary phase end was also equivalent for both strains, with a viable population count of 1.63 × 10 9 CFU.mL −1 and 1.75 × 10 9 CFU.mL −1 for the wild type and the mutant strains, respectively. Tolerance toward stress challenges is reported in Figure 3B. In the case of acid stress, we observed a significant decrease in viability for the CB129 slpB strain 0.71 ± 0.13% (7.3 × 10 6 CFU.mL −1 ) compared to the WT strain 5.76 ± 1.48% (5.76 × 10 7 CFU.mL −1 ). During the bile salts stress, we observed the same trend in the tolerance. Indeed, the survival rate for the CB129 slpB strain was significantly decreased 0.37 ± 0.24% (3.71 × 10 6 CFU.mL −1 ), compared to the WT strain 2.19 ± 1.01% (2.19 × 10 7 CFU.mL −1 ). The same stands for heat challenge, with a reduced survival in CB129 slpB 0.71 ± 0.16% (9.01 × 10 6 CFU.mL −1 ) compared to WT strain 5.76 ± 1.35% (5.86 × 10 7 CFU.mL −1 ).

Impact of slpB Mutation on P. freudenreichii Qualitative and Quantitative Proteome
Considering the major alterations in surface extractable proteins, bacteria cell surface physicochemical properties, and stress tolerance, a qualitative and quantitative analysis of the total proteome was performed to elucidate the impact of the slpB gene knockout in the mutant strain. A total of 1,288 quantifiable proteins (53.26% of predicted proteome) wherein 1,253 proteins (reported in Figure 4A) were identified (Table S1). In the WT strain 1,227 proteins were found, whereas in the CB129 slpB strain, we detected 1,252 proteins. Comparative analysis revealed a core-proteome, composed by 1,226 proteins, shared by both strains (Figure 4A). Differences in protein abundance were observed by proteomic quantitative analysis ( Figure 4B). A Wild-type (WT) and mutant CB129 slpB strains were compared. Bar represents the mean SD of three biological replicates and three technical replicates. The asterisks (****) denotes the statistical significance of the represented value between CB 129 WT and CB 129 slpB (p > 0.0001).
total of 97 proteins (4.2% of the predicted proteome) of these common proteins showed differences in the level of expression among strains, including 36 up-regulated and 61 down-regulated proteins in CB129 slpB in comparison with the WT strain ( Table 2). According to the predicted subcellular localization of the 1,253 proteins identified, 1,081 proteins are CYT (61% of predicted proteome), 71 are MEM (18% of predicted proteome), 77 are PSE (41% of predicted proteome) and 24 are SEC (38% of predicted proteome). In the analysis of non-differentially expressed proteins, we classified 1,001 as CYT proteins, 67 as MEM proteins, 70 as PSE proteins and 22 as SEC proteins ( Figure 4C). Meanwhile, between the P. freudenreichii WT and the CB129 slpB strains, from 97 proteins differentially expressed, the subcellular localization were predicted as follow: 81 CYT, 2 MEM, 7 PSE, and 7 SEC proteins ( Figure 4C).
According to COG functional classifications, the differentially expressed proteins were classified into 20 biological processes ( Figure 5A). A general category of differentially regulated proteins in CB129 slpB strain core proteome showed 27 proteins involved in information storage and processing, 25 associated to metabolism and, 18 proteins related to cellular processes and signaling ( Figure 5A). Proteins that mediate different biological process were dysregulated in the mutant strain. As seen in Figure 5B, 11 proteins were classified as having general functions, 10 proteins related to process of replication, recombination and repair, other 10 proteins linked to posttranslational modification, chaperones, protein turnover, and 9 proteins involved in the transcription process. The differentially expressed proteins between wild-type and mutant strains detected in each functional category are shown in Table 2. In addition, we detected proteins exclusive to the proteome of each strain. WT strain exhibits a unique exclusive protein, the Putative carboxylic ester hydrolase, which is involved in The growth curve of Wild-type (WT) and mutant CB129 slpB strains was determined at 30 • C in YEL broth until stationary phase (72 h). Growth was monitored by OD 650 nm as a function of time. No statistically significant difference was found in growth curve between strains. (B) Wild-type (WT) and mutant 129 slpB strains were subjected to acid, bile salts and thermal challenges. Viable propionibacteria were enumerated by plate counting before and after each challenge. Asterix represent statistically significant differences between strains and were indicated as follows: **p < 0.01. metabolism, especially in hydrolase activity. Interestingly, 27 proteins were found exclusively in the mutant strain, they are involved in several processes like metabolism and replication, recombination and repair (Table S1).
slpB Gene Mutagenesis and Whole-Genome Co-localization Complete genome of CB129 slpB (BioProject -PRJNA476583, Accession -CP030279) strain was sequenced and assembled in a circular chromosome, which exhibits a length of 2.6815.18 bp, with a G+C content of 67.28%, and a total of 2,479 CDSs, 6 rRNA genes (5S, 16S, and 23S), and 45 tRNA genes. The circular map showed a high similarity when comparing CB129 slpB with the CIRM-BIA 1 and the JS17 reference strains (Figure 6A). Figure 6B shows the localization of the plasmid inserted within the slpB gene during its knockout and Figures S1, S2 shows the read mapping before and after the insertion. Genomic analyses of genetic context, i.e., the sequences upstream and downstream the slpB gene, confirmed that this locus is not part of an operon and thus should not affect the expression of downstream genes or upstream genes. Complete genome sequence of CB129 slpB strain further ruled out any homologous recombination (HR) in other genome sites.

Protein-Protein Interaction (PPI)
We performed a PPI network to evaluate the interactions among the proteins differentially regulated in WT and CB129 slpB strains (Figure 7). The interactome analysis revealed 118 interactions between identified proteins. In PPI network, we observed that upregulated proteins, such as DNA-directed RNA polymerase alpha chain (PFCIRM129_08045), and 50S ribosomal protein L2 (PFCIRM129_08225), which exhibit high interaction, are involved in Transcription and Translation, respectively. Moreover, downregulated proteins such as GTP binding signal recognition particle protein (PFCIRM129_00245), DNA polymerase III alpha subunit (PFCIRM129_04260) and Enolase 2 (PFCIRM129_06035) showing high interaction, are involved in metabolism, DNA repair and main glycolytic pathway, respectively.

DISCUSSION
Propionibacterium freudenreichii CIRM-BIA 129 has emerged as a probiotic strain with a great immunomodulatory potential in the context of inflammatory bowel disease, according to promising results obtained in animal models (Plé et al., , 2016. Recently, our group has studied the role of the surface SlpB protein of P. freudenreichii CIRM-BIA 129 in adhesion to the intestinal epithelial cells, a probiotic property linked to beneficial effects. Knocking-out of the slpB gene evidenced a direct involvement of this protein in the adhesion to HT-29 cells. Electrophoretic analysis of guanidine extracts confirmed the disappearance of the corresponding SlpB protein . Surface layer proteins are associated to several functions (do Carmo et al., 2018). Therefore, in order to better understand the impact of this mutation, we performed a more thorough proteomic analysis by applying nanoLC-MS/MS to these extracts. Differences were found between the parental wild type CIRM BIA 129 and the isogenic CB129 slpB mutant strains of P. freudenreichii, in terms of surface extractable proteins. As shown in Table 1, proteins previously identified in CB 129 WT strain guanidine-extracted proteins (Le Maréchal et al., 2015) were detected in both strains, including in particular, surface proteins anchored in the peptidoglycan cell wall via surface layer homology (SLH) domains, such as SlpA, SlpB, SlpE, and InlA like as previously reported by Carmo and collaborators . However, this set of SLH domain-containing proteins was reduced in the mutant strain guanidine-extracted proteins, with the expected absence of SlpB protein, thus validating the directed mutagenesis. Analysis of CB129 slpB strain guanidine-extracted proteins, identified several proteins, including chaperones, such as ClpB, DnaK, and GroEL, and Enolase (carbohydrate metabolism) involved in stress tolerance, as previously reported for Propionibacterium ssp. strains by enzymatic shaving of the surface proteins using trypsin (Jan et al., 2000;Gagnaire et al., 2015;Huang et al., 2016). Another noticeable difference was the higher number of guanidine-extracted proteins, in the mutant strain, compared to the wild type strain. This included proteins usually described as cytoplasmic: enzymes of the central carbon metabolic pathways, such as pyruvate synthase, or the two subunits of the methylmalonyl-CoA mutase, a recognized cytoplasmic marker, previously described as an extracellular marker of autolysis (Valence et al., 2000). Interestingly, the HSP 70 cytoplasmic stress-related protein present at the surface of the mutant strain could be responsible for preventing protein denaturation. It is as such considered a factor of virulence and pathogenesis in some specific pathogens (Ghazaei, 2017), in Neisseri meingitidis (Knaust et al., 2007) and in Mycobacterium spp. (Das Gupta et al., 2008). This appeals further investigation, as it suggests a profound modification of the envelope structure and cell surface properties of the mutant strain.
SLAPs are known to determine key parameters of the surface layer of bacteria, in terms of charge and hydrophobicity (Wilson et al., 2001). Not only amino acid residues, but also covalent modification may endow the S-layer lattice with a strong negative charge. Thus, we determined the surface charge in both P. freudenreichii WT and CB129 slpB strains by measuring the zeta potential, which reflects the mobility rate of cells within an electric field. A lower negative value is reportedly linked with higher hydrophobicity, consequently improving adhesion (de Wouters et al., 2015). Likewise, considering the presence of surface proteins and their role in zeta potential, van der Mei et al. have shown that some wild type strains, like the L. acidophilus ATCC4356, with SLPs, are more negatively charged at pH 7 than strains without SLPs, such as L. johnsonii LMG9436 and L. gasseri LMG9203 (van der Mei et al., 2003). We thus further investigated the hydrophobicity of the cell surface, a parameter thought to be correlated with in vitro adhesion of bacteria to mucin, collagen, fibronectin, and to human epithelial cells (Duary et al., 2011). The cell surface hydrophobic and hydrophilic properties have been studied in lactic acid bacteria (Sandes et al., 2017) and can be correlated to the adhesion process to intestinal epithelial cells of apolar surface proteins (Guo et al., 2010). Using the MATH assay, we showed that the CB129 slpB strain has a strongly decreased ability to adhere to xylol, as well as to chloroform and to ethyl acetate solvents, indicating a change in the global properties of the cell surface, affecting adhesion to surfaces.      These results corroborate with the previous study showing a decreased adhesion to HT-29 human intestinal epithelial cells . Hydrophobicity and ζ-potential are factors correlated with bacterial adhesion to the epithelial cells, which are guided by charge and hydrophobicity of the bacterial surface.
The presence of surface layers being reportedly linked to tolerance toward stresses (do Carmo et al., 2018), we decided to investigate the impact of such a mutation on the CB129 slpB strain tolerance toward stress challenges that are relevant for the selection of new probiotics. The ability to survive acid stress in the stomach and bile salts stress in the duodenum during the passage FIGURE 7 | Protein-protein interactions of the proteins identified as differentially expressed in CB 129 slpB. The sizes of the nodes represent the degree of interaction for each gene/protein; the major nodes demonstrate greater interactions. Red, up-regulated; Blue, unchanged; Green, down-regulated; Yellow, Exclusive identified at WT strain; Purple, Exclusive identified at CB129 slpB strain.
through the digestive tract, is important for probiotic interaction with the host . Accordingly, in vitro assays can be used to simulate digestive stresses, mimicking the exposure to acidic conditions (pH 2.0) or to biliary salts (1 g.L −1 ) (Jan et al., 2000). For P. freudenreichii, commonly used as a cheese starter, the heat stress tolerance constitutes a relevant technological ability of this strain . Overall, we observed a large decrease in tolerance to the environmental stresses, confirming a role of SlpB in toughness. In the guanidineextracted proteins of the mutant strain, the chaperones and heat shock proteins, DnaK1, DnaK2, ClpB 2, GroE1, and GroE2 were found. Inside the cell, they are responsible for protein folding and are correlated to acid and bile adaptation (Leverrier et al., 2005;Gagnaire et al., 2015). Here, they were found at the surface of the CB129 slpB mutant, which was more susceptible to extreme acid stress and temperature, compared to wild type strain. Previous work showed that L. acidophilus ATCC 4356 adapts to harsh environments by increasing the expression of the s-layer SlpA protein upon bile, acidic pH and heat stress exposition (Khaleghi et al., 2010;Khaleghi and Kasra, 2012). Moreover, changes in the cell surface properties could alter the transmembrane protein complex responsible for the extrusion of protons from the cytoplasm, which are responsible for surviving environmental stresses (Ruiz et al., 2013;Rosa do Carmo et al., 2017).
Profound modifications of P. freudenreichii physiology and surface properties suggested that modifications, wider than the disappearance of a single protein, occurred as a result of slpB gene inactivation. To understand this impact of the mutation, a comparative proteomic analysis was performed to identify significant alterations in the whole proteome profile of the mutant strain, using label-free quantitative proteomic analysis. Prediction of sub cellular localization using the SurfG+ tool (Barinov et al., 2009) evidenced changes in all the categories (CYT, MEM, PSE and SEC) in the differential proteome of CB129 slpB. In addition, differential proteome was functionally classified using COG, showing a functional implication of differential proteins in cellular processes such as signaling, information storage, processing, and metabolism. Specifically, this study showed that the moonlighting enolase and NlpC/P60 are both exported (Frohnmeyer et al., 2018), as it was recently observed in the cutaneous Propionibacterium acnes strain (Jeon et al., 2017). These moonlighting proteins were downregulated in CB129 slpB. Interestingly, in the Bifidobacterium and Lactobacillus genera, moonlighting proteins, such as enolase, also play a role in immunomodulation and adhesion (Sánchez et al., 2010;Kainulainen and Korhonen, 2014;Vastano et al., 2016). Furthermore, in the PPI network we observed high interactions between the downregulated Enolase (PFCIRM129_06035), reportedly involved in human gut colonization and stress adaptation (Ruiz et al., 2009), with other proteins involved in several other processes, including metabolism and DNA repair. Moreover, all surface layer-associated proteins SlpA, SlpD, SlpE, and InlA were downregulated in CB129 slpB. These proteins form a protective layer on the surface of the bacteria, and have been associated with environmental stress tolerance (Fagan and Fairweather, 2014). As seen previously, a decreased amount of these proteins could be directly associated with stress susceptibility and with altered hydrophobicity. SLAPs can directly influence these properties (Pum et al., 2013), and consequently alter adhesion to epithelial cells .
We performed the complete genome DNA sequencing of the CB129 slpB, which, in turn, allowed us to evaluate whether the slpB gene disruption had major consequences on the mutant strain genome. The slpB gene is not part of an operon, which suggests that homologous recombination using the suicide plasmid pUC: slpB:CmR (do Carmo et al., 2017) did not affect the expression of upstream and downstream genes. Analysis of the genetic context, upstream and downstream, revealed that the homologous recombination process was sitespecific, and not affecting other genes in the genome of the mutant strain CB129 slpB. However, we were unable to evaluate possible rearrangements in the genome of CB129 slpB, which could have affected the transcription of other genes. Therefore, more studies are necessary to explore whether any probiotic potential was lost after the single mutation of the slpB gene in Propionibacterium freudenreichii CIRM-BIA 129 strain.

CONCLUSION
This study evidenced the pleiotropic impact of the surface layer protein slpB mutation in the probiotic strain Propionibacterium freudenreichii CIRM-BIA 129 in relation to its physicochemical proprieties, stress challenges, surfaceome and whole cell quantitative proteome. It confirmed the key role of SLPs and strongly suggests that expression of specific ones, such as P. freudenreichii SlpB, should be used as criteria for selecting strains with probiotic potential.

AUTHOR CONTRIBUTIONS
FC performed in vitro assays, microscopy, proteomic assays and data interpretation. WS, FP, GT, and ROC performed proteomic assays, data interpretation and bioinformatics analyses. BC, EO, and SS performed in vitro assays. II and HR data interpretation. EF performed PPI network. CC performed microscopy. MC, AC, and RS performed genomics and data interpretation. VA, GJ, HF, and YL contributed to the supervision, analysis, and interpretation of data and were major contributors to revising the manuscript. All authors contributed in writing the manuscript.

SUPPLEMENTARY MATERIAL
The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fmicb. 2018.01807/full#supplementary-material Figure S1 | Verification of assembly error by read mapping. The plasmid pUC: slpB:CmR was not inserted in the slpB gene during de novo genome assembly. The read mapping on the slpB gene shows misalignments upstream and downstream insertion site, confirming the assembly error. The read mapping was performed using in CLC Genomics Workbench 7.0. Figure S2 | Assembly curation and validation by read mapping. The manual insertion of plasmid pUC: slpB:CmR in the slpB gene was validated by read mapping. The correct read alignments upstream (A) and downstream (B) the plasmid validate the manual insertion. The read mapping was performed using in CLC Genomics Workbench 7.0.
Table S1 | Total list of proteins identified in the core-proteome of CB 129 wild-type and CB129 slpB.