Role of the LytSR Two-Component Regulatory System in Staphylococcus lugdunensis Biofilm Formation and Pathogenesis

Staphylococcus lugdunensis is a coagulase negative Staphylococcus recognized as a virulent pathogen. It is responsible for a wide variety of infections, some of which are associated with biofilm production, such as implanted medical device infections or endocarditis. However, little is known about S. lugdunensis regulation of virulence factor expression. Two-component regulatory systems (TCS) play a critical role in bacterial adaptation, survival, and virulence. Among them, LytSR is widely conserved but has variable roles in different organisms, all connected to metabolism or cell death and lysis occurring during biofilm development. Therefore, we investigated here the functions of LytSR in S. lugdunensis pathogenesis. Deletion of lytSR in S. lugdunensis DSM 4804 strain did not alter either susceptibility to Triton X-100 induced autolysis or death induced by antibiotics targeting cell wall synthesis. Interestingly, ΔlytSR biofilm was characterized by a lower biomass, a lack of tower structures, and a higher rate of dead cells compared to the wild-type strain. Virulence toward Caenorhabditis elegans using a slow-killing assay was significantly reduced for the mutant compared to the wild-type strain. By contrast, the deletion of lytSR had no effect on the cytotoxicity of S. lugdunensis toward the human keratinocyte cell line HaCaT. Transcriptional analyses conducted at mid- and late-exponential phases showed that lytSR deletion affected the expression of 286 genes. Most of them were involved in basic functions such as the metabolism of amino acids, carbohydrates, and nucleotides. Furthermore, LytSR appeared to be involved in the regulation of genes encoding known or putative virulence and colonization factors, including the fibrinogen-binding protein Fbl, the major autolysin AtlL, and the type VII secretion system. Overall, our data suggest that the LytSR TCS is implicated in S. lugdunensis pathogenesis, through its involvement in biofilm formation and potentially by the control of genes encoding putative virulence factors.

Staphylococcus lugdunensis is a coagulase negative Staphylococcus recognized as a virulent pathogen. It is responsible for a wide variety of infections, some of which are associated with biofilm production, such as implanted medical device infections or endocarditis. However, little is known about S. lugdunensis regulation of virulence factor expression. Two-component regulatory systems (TCS) play a critical role in bacterial adaptation, survival, and virulence. Among them, LytSR is widely conserved but has variable roles in different organisms, all connected to metabolism or cell death and lysis occurring during biofilm development. Therefore, we investigated here the functions of LytSR in S. lugdunensis pathogenesis. Deletion of lytSR in S. lugdunensis DSM 4804 strain did not alter either susceptibility to Triton X-100 induced autolysis or death induced by antibiotics targeting cell wall synthesis. Interestingly, ΔlytSR biofilm was characterized by a lower biomass, a lack of tower structures, and a higher rate of dead cells compared to the wild-type strain. Virulence toward Caenorhabditis elegans using a slow-killing assay was significantly reduced for the mutant compared to the wild-type strain. By contrast, the deletion of lytSR had no effect on the cytotoxicity of S. lugdunensis toward the human keratinocyte cell line HaCaT. Transcriptional analyses conducted at mid-and late-exponential phases showed that lytSR deletion affected the expression of 286 genes. Most of them were involved in basic functions such as the metabolism of amino acids, carbohydrates, and nucleotides. Furthermore, LytSR appeared to be involved in the regulation of genes encoding known or putative virulence and colonization factors, including the fibrinogen-binding protein Fbl, the major autolysin AtlL, and the type VII secretion system. Overall, our data suggest that the LytSR TCS is implicated in S. lugdunensis pathogenesis, through its involvement in biofilm formation and potentially by the control of genes encoding putative virulence factors.

INTRODUCTION
Staphylococcus lugdunensis is a member of the coagulase-negative staphylococci (CoNS) family. This bacterium belongs to the human skin flora; van der Mee-Marquet et al. found, for example, carriage of inguinal S. lugdunensis in 22% of 140 incoming patients (van der Mee-Marquet et al., 2003). However, its pathogenicity is closer to that of Staphylococcus aureus than that of other CoNS in terms of tissue destruction ability and clinical course (Frank et al., 2008;Heilmann et al., 2019). In particular, S. lugdunensis native valve endocarditis can be particularly aggressive and destructive, often requiring surgery (Anguera et al., 2005). It can also cause a wide range of infections such as abscesses and wound infections (Böcher et al., 2009;Heldt Manica and Cohen, 2017), bone and joint infections (Argemi et al., 2017b), and infections associated with catheters or implanted medical devices (Nesher et al., 2017). Only a few virulence factors have been identified so far to explain such pathogenicity (Heilbronner et al., 2011;Argemi et al., 2017a). Similar to other staphylococci, the pathogeny of S. lugdunensis appears in many infections to be related to biofilm formation within host tissues or indwelling medical devices (Frank and Patel, 2007;Argemi et al., 2017a). The resulting infections are usually difficult to treat because biofilm protects bacteria from both the host's immune system and the antimicrobial therapies (Stewart and Costerton, 2001;Lebeaux et al., 2014).
In staphylococci, biofilm formation appears to depend mainly on proteins (as in S. aureus), or on the exopolysaccharide poly-N-acetylglucosamine (PNAG), also termed polysaccharide intercellular adhesion (PIA), as described for Staphylococcus epidermidis (Otto, 2018). PIA is encoded by the icaADBC operon composed of four genes whose expression is regulated by the transcriptional repressor IcaR. Despite the identification of an icaADBC locus, PIA is not a major component of the S. lugdunensis biofilm matrix (Frank and Patel, 2007). Instead, biofilm consists of extracellular factors, which are mainly proteinaceous. However, not all of these proteins have been identified yet. Among the identified surface proteins that are involved in S. lugdunensis biofilm formation, the iron-regulated surface determinant IsdC under iron-limited conditions (Missineo et al., 2014) and the major autolysin AtlL (Gibert et al., 2014;Hussain et al., 2015) have been demonstrated to be important actors.
Despite their clinical significance, the production and regulation of staphylococcal biofilm remain poorly defined. It is assumed that environmental signals can influence the polysaccharide or proteinaceous composition of the biofilm matrix (Vergara-Irigaray et al., 2009). Indeed, some isolates of S. aureus and S. epidermidis are able to form biofilm without PIA (Rohde et al., 2007;Arciola et al., 2015). The ability of bacteria to adapt to environment is mediated by complex regulatory networks, including two-component signal transduction systems (TCSs), which allow a pathogen to adapt its gene expression in response to environmental stimuli (Beier and Gross, 2006;Haag and Bagnoli, 2017). Some of these TCSs are involved in controlling staphylococcal biofilm formation, such as agr (Kavanaugh and Horswill, 2016), arlRS (Burgui et al., 2018), and lytSR (Sharma-Kuinkel et al., 2009). Interestingly, the LytSR TCS plays an important role in biofilm development in S. aureus (Sharma-Kuinkel et al., 2009) as well as in S. epidermidis (Zhu et al., 2010) through an ica-independent mechanism.
The LytSR system was first identified in S. aureus, where it was shown to affect murein hydrolase activity and autolysis (Brunskill and Bayles, 1996a). Further characterization demonstrated that LytSR positively regulates the expression of the immediately downstream lrgAB operon (Brunskill and Bayles, 1996b), which, along with the cidABC operon, is implicated in the control of programmed cell death and lysis during biofilm development (Bayles, 2007;Rice et al., 2007;Sadykov and Bayles, 2012). Indeed, cidA gene encodes a holinlike protein that is an effector of extracellular murein hydrolase activity, cell lysis, and DNA release, while lrgA encodes an anti-holin-like protein, which is an inhibitor of these processes (Rice et al., 2007). Thus, the subpopulation of dead bacteria that lyses within the biofilm releases extracellular DNA (eDNA), which plays a critical role in intercellular adhesion and biofilm stability (Bayles, 2007). As lrg and cid operons are involved in autolysis and cell death (van den Esker et al., 2017), their transcription is subject to complex control. They are under the control of two overlapping regulatory networks, one in response to the metabolism of carbohydrates (Yang et al., 2005;Patel and Golemi-Kotra, 2015) and the other in response to changes in membrane potential (Patton et al., 2006).
Given the critical role of LytSR in the control of lysis and cell death in the biofilm of S. aureus and S. epidermidis, it seemed interesting to explore its functions in the particularly virulent CoNS, S. lugdunensis. In the present study, we constructed a DSM 4804 ΔlytSR strain and investigated the effects of this deletion on bacterial autolysis, biofilm formation, and in vivo virulence. Furthermore, we set out to determine the full extent of the S. lugdunensis LytSR regulon by transcriptional profile analysis.

DNA Extraction and Polymerase Chain Reaction Amplification
Isolates were grown overnight at 37°C on tryptic soy agar (TSA; Bio-Rad) supplemented with 5% horse blood. DNA was extracted using the InstaGene Matrix kit (Bio-Rad) according to the manufacturer's instructions. PCRs were performed using a Veriti Thermal Cycler (Applied Biosystems, Foster City, CA, USA) in a final volume of 25 μl containing 12.5 μl GoTaq G2 Green Master Mix (Promega, Charbonnières-les-Bains, France), 0.50 μM of each primer, and 5 μl of DNA. When Frontiers in Microbiology | www.frontiersin.org necessary, PCR products were purified and verified by sequencing by Eurofins GATC Biotech SARL (Konstanz, Germany).

Construction of the S. lugdunensis DSMΔlytSR Mutant and Complementation Strains
The lytSR deletion mutant was constructed by homologous recombination using pMAD plasmid (Arnaud et al., 2004). The fragment containing the upstream region of lytS (with respect to the lytS translation initiation site) and the downstream region of lytR (with respect to the lytR stop codon) was amplified by a two-step overlap PCR reaction using the oligonucleotide pairs lytS_EcoRI_F/ lytS_R and lytR_ F/lytR_ BamHI_R ( Table 2). The purified PCR product was digested with EcoRI and BamHI restriction enzymes (New England BioLabs, Évry, France) and ligated into the pMAD vector with T4 DNA ligase (New England BioLabs) following the manufacturer's instructions and then cloned into E. coli TOP10. Clones containing the resulting recombinant pMADΔlytSR vector were selected on TSA supplemented with ampicillin. The recombinant plasmid pMADΔlytSR was extracted (NucleoBond Xtra Midi endotoxin free, Macherey Nagel, Hoerdt, France) and used to transform the wild-type (WT) S. lugdunensis DSM 4804 strain (Freney et al., 1988) by protoplast formation as previously described (Szabados et al., 2011;Marlinghaus et al., 2012). After incubation at 30°C for 5 days, clones containing the plasmid were selected, and the double-crossover event was performed by a shift of temperature to 43°C and in the presence of erythromycin (Arnaud et al., 2004). The loss of the plasmid was obtained after successive shifts between 30 and 43°C and in the absence of antibiotic. Successful deletion of the lytSR operon via homologous recombination and loss of the plasmid was verified by PCR and sequencing.
To construct the complemented strain, the lytSR operon including its predicted promoter and ribosome binding site was amplified by PCR using primers 5′_lytS and 3′_lytR (282 bp upstream lytS and 441 bp downstream lytR; Table 2) and was ligated into the pCU1 plasmid (Augustin et al., 1992) at the EcoRI/BamHI restriction sites and subcloned into E. coli TOP10. After verification by PCR and sequencing of the vector pCU1: lytSR, this vector and the empty vector pCU1 were transformed into S. lugdunensis DSMΔlytSR by protoplast formation.

Triton X-100-Induced Autolysis
Triton X-100-induced autolysis assay was performed as previously described (Brunskill and Bayles, 1996a). Briefly, cells were grown in TSB containing 1 M NaCl to mid-exponential phase (OD 600 nm = 0.7), and 50 ml was pelleted by centrifugation. Cells were washed twice with 50 ml of ice-cold sterile water and resuspended in 50 ml of 0.05 M Tris-HCl (pH 7.2) containing 0.05% (vol/vol) Triton X-100. Cells were then incubated at 37°C under shaking (180 rpm). OD 600 nm was measured at 15-min intervals for 1 h and then at 30-min intervals for 2 h to evaluate autolysis. All experiments were conducted using at least three biological replicates.

Antibiotic Time-Kill Assays
For time-kill kinetic assays, bacteria were subcultured in Mueller-Hinton broth (MHB) to early-exponential phase (3 h), diluted to approximately 5 × 10 5 CFU/ml in MHB containing antimicrobial agents, and incubated for 24 h at 37°C (National Committee for Clinical Laboratory Standards, 1999). Penicillin G, vancomycin, and teicoplanin were added to a final concentration equivalent to 10 times their MICs (i.e. final concentration of 0.64 mg/L for penicillin G, 10 mg/L for vancomycin, and 2.5 mg/L for teicoplanin). Bacterial counts were performed just before, at 6 and 24 h after antibiotic addition. A 1 ml aliquot was taken and serially diluted in 0.9% sodium chloride, and viable cells were quantified by plating dilutions on TSA. Tolerance was defined as a <3-log 10 reduction of the bacterial count after 24 h according to CLSI guidelines (National Committee for Clinical Laboratory Standards, 1999). We also carried out studies, in which bacteria approaching the stationary phase (10 h) were treated with penicillin G (final concentration 0.64 mg/L). Cell viability was assessed at 2, 4, 6, and 8 h after antibiotic addition. All experiments were repeated at least three times.

Biofilm Assays and Confocal Laser Scanning Microscopy
Overnight cultures of S. lugdunensis WT and mutant strains grown in TSB supplemented with 1% glucose (TSB 1%glc ) were adjusted to an OD 600nm of 0.01 in TSB 1%glc . Cultures were distributed in triplicate into six-well flat-bottom plate. After 24 h incubation at 37°C, cells were washed gently two times with sterile distilled water and stained with SYTO ® 9 and propidium iodide (PI) (Live/Dead kit, Thermo Fisher Scientific). The architecture of biofilms was observed with a confocal laser Frontiers in Microbiology | www.frontiersin.org scanning microscope (Leica TCS SP2). CLSM z-stack processing was performed using Leica Confocal Software. Measurements of the biofilms produced were performed using IMARIS software (Bitplane, Belfast, Ireland), calculating the biomass and the percentage of cells labeled by SYTO ® 9 and PI. Each confocal experiment was repeated a minimum of three times. Statistical significance was calculated by the chi-squared test with Yates continuity correction; p < 0.05 was considered as significant.

Caenorhabditis elegans Virulence Assays
The Caenorhabditis elegans WT Bristol strain N2 was obtained from the Caenorhabditis Genetics Center (Minneapolis, MN, USA). C. elegans worms were maintained under standard culturing conditions at 23°C on nematode growth medium (NGM; 3 g NaCl, 2.5 g peptone, 17 g agar, 5 mg cholesterol, 1 ml 1 M CaCl 2 , 1 ml 1 M MgSO 4 , 25 ml 1 M KH 2 PO 4 , H 2 O to 1 L) agar plates with Escherichia coli OP50 as a food source. Synchronous cultures of worms were generated as previously described (Blier et al., 2011) and used for both slow-killing assay and liquid-killing assay. Bacterial lawns used for C. elegans survival slow-killing assays were prepared by spreading 50 μl of WT or ∆lytSR strains on 35 mm NGM conditioned Petri dishes supplemented with both 0.05 mg/ml of 5-fluoro-2′-deoxyuridine and 5 μmol/L of tryptophan. Plates were incubated overnight at 37°C and then placed at room temperature for 4 h. Fifteen to twenty L4 (48-h old) synchronized nematodes were harvested with M9 solution (3 g KH 2 PO 4 , 6 g NaHPO 4 , 5 g NaCl, 1 ml 1 M MgSO 4 , H 2 O to 1 L), placed on 35 mm assay Petri dishes and incubated at 23°C. C. elegans survival was scored at 1, 24 h and each subsequent day until death, using an Axiovert S100 optical microscope (Zeiss, Oberkochen, Germany) equipped with a Nikon digital Camera DXM 1200F (Nikon Instruments, Melville, NY, USA). The nematodes were considered dead when they remained static without grinder movements for 20 s or did not respond to light flashes. The results are expressed as the percentage of living nematodes and correspond to the mean of two independent assays. Nematode survival was calculated by the Kaplan-Meier method, and differences in survival kinetics were tested for significance by using the log rank test (GraphPad Prism version 4.0; GraphPad Software, San Diego, CA, USA).
A liquid killing assay (Chua et al., 2014) was also performed in order to evaluate the toxicity of bacterial released factors toward C. elegans. Volumes (80 μl) of 5-h old bacterial cultures were distributed onto microtiter plate wells. Immediately after, 20 μl of M9 solution containing 15-20 L4 (48 h old) synchronized worms was added to bacteria in each well and incubated at 23°C. Exact worm number was immediately determined (t = 0 h), and worm survival was scored at different time points (24, 48, and 120 h) using an Axiovert S100 optical microscope (Zeiss). Each value reported for the assays is the mean measurement for six replicates of six independent preparations. The nonparametric Mann-Whitney U test was used to compare the means within the sets of experiments, using Past 3.x software.

HaCaT Cell Cytotoxicity Tests
The cytotoxicity activity of bacteria was studied using the human keratinocyte cell line HaCaT (Eppelheim, Germany) as previously described (N'Diaye et al., 2016). Briefly, HaCaT cells were grown at 37°C under 5% CO 2 atmosphere, in Dulbecco's modified Germany), as previously described (Dahyot et al., 2019). To validate the differential gene expression obtained by microarray, qRT-PCR was carried out using a CFX96 real-time PCR detection system (Bio-Rad) with SYBR ® Green PCR Master Mix (Bio-Rad), 1 μl of diluted cDNA template, and primers listed in Table 2 in 20 μl PCR mixtures. Cycling conditions were one initial denaturation step at 95°C for 10 min, followed by 40 cycles of amplification including denaturation at 95°C for 30 s, annealing at Tm depending on the primers for 30 s, and extension at 72°C for 30 s. Relative expression of genes was quantified using the gene expression analysis module of CFX Manager™ software, with 16S rRNA gene as reference to normalize the results. All experiments were performed in at least three independent RNA preparations, and results are presented as the mean ± the standard error of the mean.

RNA Extraction and Sample Preparation for Microarray
To decipher the mechanisms affected by the deletion of lytRS genes, the transcriptome of ΔlytSR and WT strains was compared by DNA microarray at 6 and 8 h of growth. S. lugdunensis was grown in TSB, and cells were collected at 6 and 8 h of growth. Total RNA was extracted using the RNeasy Mini kit (Qiagen) following the manufacturer's instructions. Two DNase treatments were successively performed using DNase I (Sigma-Aldrich) and Turbo DNA free (Thermo Fisher Scientific). The quality of RNA was evaluated using the RNA 6000 Nano with a Bioanalyzer (Agilent, Les Ulis, France). The preparation of samples and array processing were performed using the Two-color Microbased Exon Analysis kit (Agilent) following the manufacturer's instructions. Briefly, cRNA was labeled with either Cy3 or Cy5 before being quantified and co-hybridized on the microarray. The custom-made S. lugdunensis GeneChips (Agilent) used in this study were generated based on DNA sequences from two publicly available genomes, S. lugdunensis N920143 and HKU09-01 strains (Tse et al., 2010;Heilbronner et al., 2011). Then, the microarray was washed and scanned. Genes with significantly different expression (p ≤ 0.05) were selected based on at least a two-fold change in expression level. The genes identified were grouped following the classification established in the KEGG pathway database 1 .

In silico and Transcription Analyses of the lytSR locus
We first amplified and sequenced the lytSR region of the S. lugdunensis DSM 4804 strain. Sequence analysis revealed a 1755-nucleotide lytS open reading frame (ORF) immediately followed by a 762-nucleotide lytR ORF. BLAST searches indicated 67% homology with the lytSR operon of the S. aureus NCTC5661 strain and 69% with the S. epidermidis ATCC 12228 strain (Zhang et al., 2003). The transcription start site was tentatively located 63 bp upstream of the lytS gene, preceded by a canonical −10 element (TGTTAAAAT) and near-canonical −35 element (TTAACA). A consensus ribosome-binding site (GGGAG) was found 9 bp upstream of the lytS ATG start codon. The predicted amino acid sequence of lytS gene product (LytS) contains 584 amino acids corresponding to a 64.83 kDa protein. This protein shares 77 and 71% amino acid sequence identity with S. aureus and S. epidermidis, respectively. lytR gene encodes a putative 253 amino acid protein (29.4 kDa), which shares 64 and 63% amino acid identity with the LytR proteins of S. aureus and S. epidermidis, respectively. The predicted LytS and LytR proteins share putative conserved domains with members of a two-component regulatory system, namely a sensor histidine kinase protein and a response regulator transcription factor, respectively. Sequence analysis revealed at 510 bp downstream lytR the presence of lrgA and lrgB genes predicted to encode two anti-holin-like proteins.
As it has been shown that lytSR and lrgAB are temporally regulated in S. aureus (Groicher et al., 2000), the expression of these genes was analyzed by qRT-PCR assays at various culture times of S. lugdunensis DSM 4804. Transcription of lytS and lytR genes appeared maximal at 6 h growth (mid-exponential phase), whereas that of lrgA and lrgB genes reached its maximum at 8 h (late-exponential phase) (data not shown). A fragment overlapping lytS and lytR genes was obtained by RT-PCR analysis, indicating that these genes are co-transcribed (data not shown).

Construction of the S. lugdunensis ΔlytSR Strain and the Complemented Strains
To investigate the roles of LytSR in S. lugdunensis, a ΔlytSR strain was constructed by homologous recombination. We verified by PCR that the ΔlytSR strain presented the 1,200-bp deletion at lytSR locus. In order to exclude a polar effect of this deletion, we carried out complementation of the ΔlytSR strain using pCU1:lytSR plasmid. The ΔlytSR strain 1 http://www.kegg.com Frontiers in Microbiology | www.frontiersin.org FIGURE 1 | Growth curves of S. lugdunensis WT, ΔlytSR, and complemented strains. Bacterial cultures were grown in TSB at 37°C. Growth was monitored by measuring OD 600 nm , and viable cell counts were determined by plating diluted aliquots on TSA. Datum points represent the means of three independent experiments for WT and mutant strains.
FIGURE 2 | Triton X-100 autolysis assays of S. lugdunensis WT and derivative (ΔlytSR and ΔatlL) strains. Bacterial cells were collected from early-exponential growth (OD 600 nm = 0.7) and resuspended in an equal volume of buffer containing 0.05% Triton X-100. The rate of autolysis was monitored at OD 600 nm . The S. lugdunensis atlL deletion mutant was used as a control for resistance to autolysis. Error bars represent the SD of three independent experiments. containing pCU1 plasmid was used as a negative control. All four strains showed a similar growth rate in TSB medium with no difference in cell viability (i.e. no difference in CFU/ ml counts when incubated 48 h at 37°C; Figure 1).

Implication of LytSR in Autolysis
To determine the contribution of LytSR in the regulation of autolysis, Triton X-100-induced autolysis of S. lugdunensis WT and ΔlytSR strains was measured. Both strains showed a similar trend of autolysis, while the S. lugdunensis ΔatlL strain used as a negative control (Gibert et al., 2014) showed increased resistance to autolysis (Figure 2).

Impact of lytSR Deletion on Antibiotic-Induced Killing
The possible role of LytSR in antibiotic-induced cell death was assessed by time-kill kinetic assays. A β-lactam (penicillin G, Figure 3A) and two glycopeptides (vancomycin, Figure 3B and Frontiers in Microbiology | www.frontiersin.org teicoplanin, Figure 3C), which interfere with peptidoglycan biosynthesis, were used. The MICs of these antibiotics were the same for WT and mutant strains (data not shown). Penicillin G or glycopeptides were added to early-exponential growth cultures at a lytic concentration (10 times their MICs), and the rate of living cells over time was determined. The decrease in bacterial counts was not significantly different between the WT and the mutant strains, exposed to either penicillin G or glycopeptides, for 6 and 24 h (Figure 3 and Table 3). Penicillin G was bactericidal with a 5-log CFU/ml reduction after 24 h exposure for both strains. On the contrary, the WT strain was tolerant to glycopeptides, a phenomenon also observed for the ΔlytSR strain (i.e. ΔlogCFU/ml < 3 at 24 h). Furthermore, lytSR deletion did not promote penicillin G-induced killing of cells entering stationary phase of growth (data not shown).

Impact of lytSR Deletion on Biofilm Formation
We examined the biofilm architecture and the cell viability of WT, mutant, and complemented strains using CLSM. With the Live/Dead viability staining method, bacteria with intact cell membranes were stained in green, whereas bacteria with damaged membranes were stained in red. First, we observed that the mature biofilm of the WT strain contained typical tower structures ( Figure 4A) that were not observed in the biofilm of either the ΔlytSR strain (Figure 4B) or the ΔlytSR strain containing empty pCU1 ( Figure 4D). Complementation with lytSR partially restored these structures (Figure 4C). The total biomass of the biofilm was two-fold lower for the mutant strain than the WT strain (Supplementary Table S1). Moreover, the percentage of dead cells was significantly increased in the biofilm of the ΔlytSR strain (5.7%) compared to the parental strain (1.1%; p < 0.05, chi-squared test; Supplementary Table S1). Both complemented strains exhibited higher levels of dead cell probably because of antibiotic selection pressure. However, the ΔlytSR strain complemented with lytSR exhibited a significantly lower percentage of dead cells (10.4%) than the mutant complemented with the empty pCU1vector (15.2%; p < 0.05, chi-squared test; Supplementary Table S1).

Impact of LytSR on Virulence Toward Caenorhabditis elegans
We assessed the effect of the lytSR deletion on pathogenesis of S. lugdunensis using the C. elegans infection model. Slowkilling test showed that when nematodes were in contact with the WT strain as the sole source of food, it took 15 days to kill 50% of C. elegans ( Figure 5A). Interestingly, the survival of C. elegans in the presence of the ΔlytSR strain was significantly increased (p < 0.0001), with 19 days required for 50% of the nematodes to die (Figure 5A). By contrast, no difference in the survival kinetics of C. elegans was observed between the mutant and the WT strains placed in liquid medium (Figure 5B; p = 0.777 at 24 h, p = 0.660 at 48 h, and p = 0.935 at 72 h).

Cytotoxicity Toward HaCaT Cells in vitro
We evaluated the cytotoxicity of S. lugdunensis WT and ΔlytSR strains toward the human keratinocyte cell line HaCaT. The release of LDH by HaCaT reflecting cell necrosis was identical between the two strains 24 h post infection (36.31 ± 2.46% for WT and 42.41 ± 2.48% for ΔlytSR).

Transcriptional Profiling of the S. lugdunensis ΔlytSR Strain
To identify genes regulated by the LytSR two-component regulatory system, a transcriptional profile analysis of WT and ΔlytSR strains was performed. RNA samples were isolated from both strains at mid-exponential phase (6 h) and late-exponential phase (8 h) of growth, corresponding, respectively, to the maximal expression of lytSR and lrgAB determined by qRT-PCR analyses as described in the section "In silico and Transcription Analyses of the lytSR locus". Overall, deletion of lytSR affected the expression of 286 genes compared to the WT strain with a threshold of at least two-fold change (full list in Supplementary  Table S2). The expression of 125 genes (80 upregulated and 45 downregulated) and of 177 genes (112 upregulated and 65 downregulated) was affected at 6 and 8 h, respectively. Some genes were upregulated (n = 7) or downregulated (n = 9) at both times. Transcription of lrgAB operon was the most dramatically impacted by the deletion, with lrgA and lrgB expression at 8 h in the ΔlytSR strain decreased by 195-and 162-fold, respectively.
Moreover, genes encoding transcriptional regulators were upregulated (SLUG_00760 and SLUG_01360) or downregulated (mgrA and sarR) in the mutant strain.
Results of microarray data were confirmed by qRT-PCR for three genes (lrgA, lrgB, and cidA) among those whose transcription was impacted by the deletion (Table 4).

DISCUSSION
In any biofilm, there is a variable proportion of cells that lyse and release genomic DNA. Because of its adhesive properties, Frontiers in Microbiology | www.frontiersin.org eDNA plays a key role in biofilm matrix (Moormeier and Bayles, 2017). In S. aureus and S. epidermidis, this process is regulated by the operons cidABC and lrgAB, under the control of two regulators, CidR and LytSR TCS, respectively (Sadykov and Bayles, 2012). Because LytSR TCS has been identified as a regulator of cell death and lysis in staphylococcal biofilms, we assumed that it could also be involved in the control of these processes in S. lugdunensis, which is a particularly virulent species (Frank and Patel, 2007;Pereira et al., 2012;Argemi et al., 2017b). In order to explore the role of this TCS, a lytSR knock-out mutant of the WT biofilm-forming strain S. lugdunensis DSM 4804 (Freney et al., 1988) was constructed and studied.
First, no difference in growth kinetics or cell viability was observed between mutant and parental strains, indicating that lytSR is not essential for bacterial growth in S. lugdunensis, as observed for S. epidermidis (Zhu et al., 2010). In order to determine whether lytSR deletion impacted bacterial autolysis, we performed autolysis assays with Triton X-100, a detergent known to remove the inhibition exerted by lipoteichoic acids on the general bacterial autolytic system. The S. lugdunensis ΔlytSR strain tended to exhibit the same rate of Triton X-100induced autolysis as the WT strain, although relatively wide standard deviations were observed at 15-and 30-min time points. This suggests that LytSR was not implicated in induced  Each value is the mean of measurement of eight replicates from two independent preparations. Nematode survival was calculated by the Kaplan-Meier method, and survival differences were tested for significance using the log rank test. The survival kinetics of C. elegans was significantly increased in the presence of ΔlytSR strain (p < 0.0001). (B) Liquid killing assay of C. elegans exposed to the supernatant of WT and ΔlytSR strains. Nematode survival was evaluated after 24, 48, and 120 h of exposure. Each value is the mean ± SEM of measurement of eight replicates from six independent preparations. The nonparametric Mann-Whitney U test was used to compare the means within the same set of experiments. NS, not significant. autolysis for the S. lugdunensis DSM 4804 strain. This result is similar to that obtained for the S. epidermidis 1457 ΔlytSR strain (Zhu et al., 2010). Interestingly, for S. aureus, this characteristic appears to be strain-dependent, since the S. aureus UAMS-1 lytS knock-out mutant did not show increased Triton X-100 induced lysis (Sharma-Kuinkel et al., 2009), whereas the LytSR system plays an important role in the autolysis of the S. aureus RN4220 strain (Brunskill and Bayles, 1996a). Thus, the effects of LytSR on autolysis seem to depend on the genetic backgrounds of staphylococcal strains.
In the same way, our work has shown that LytSR was not involved in death induced by antibiotics targeting cell wall January 2020 | Volume 11 | Article 39 FIGURE 6 | Functional classification according to KEGG pathways of genes significantly upregulated (red) and downregulated (blue) after 6 and 8 h of growth of the ∆lytSR strain compared to the WT strain. Numbers in parentheses represent the total number of genes affected within the genome in each functional class.
Frontiers in Microbiology | www.frontiersin.org synthesis such as penicillin G and glycopeptides. Of note, lytSR deletion did not increase sensitivity to the killing effects of glycopeptides toward the S. lugdunensis DSM 4804 strain, which is tolerant to glycopeptides (Bourgeois et al., 2007). Time-kill kinetic experiments were conducted on cells growing in earlyexponential phase (3 h) as recommended by CLSI (National Committee for Clinical Laboratory Standards, 1999), a time at which lytSR and especially lrgAB operons were poorly expressed. For S. aureus, a lrgAB deletion enhanced penicillin-induced killing of cells approaching stationary growth phase, i.e., when lrgAB operon is maximally expressed, whereas it did not affect penicillin-induced killing of cells growing in early-exponential phase, when lrgAB expression is minimal (Groicher et al., 2000). However, no enhanced killing of penicillin G was observed for S. lugdunensis cells reaching stationary growth phase.
Since biofilm formation is one of the main determinants of S. lugdunensis pathogenicity (Frank et al., 2008;Argemi et al., 2017b), the effect of lytSR deletion on biofilm formation was further investigated. Our results reveal the involvement of LytSR on biofilm production, since the ΔlytSR strain produced significantly less biofilm than the WT strain. Moreover, CLSM analyses showed that the mutation disrupted the normal architecture of the biofilm, with a lack of tower structures. This phenotype was only partially restored by lytSR complementation. Interestingly, a study has shown that lrgAB is specifically expressed within the tower structures during S. aureus biofilm formation (Moormeier et al., 2013). Live/dead staining showed a higher rate of red fluorescence inside the mutant biofilm, which is in favor of a decrease in cell viability. Thus, the impaired biofilm formation of the mutant does not seem related to reduced cell lysis, even if to confirm this fact it would be necessary to quantify eDNA levels inside the biofilms. These results are in contradiction with those obtained for other staphylococcal species. For S. aureus, a lytS mutant and a lrgAB mutant produced more adherent biofilm with increased levels of eDNA in the biofilm matrix Sharma-Kuinkel et al., 2009). For S. epidermidis, a lytSR mutant produced more biofilm but contained a significant decrease in the rate of dead cells inside (Zhu et al., 2010). Thus, LytSR is involved in controlling the formation of staphylococcal biofilms, but the pathways of involvement appear to depend on the species.
In order to evaluate the potential involvement of LytSR in the in vivo pathogenicity of S. lugdunensis, the C. elegans nematode was used as a host model organism. This model was chosen because C. elegans has already been successfully used for the study of host-pathogen interactions, e.g., for S. aureus (Sifri et al., 2003) and S. lugdunensis (Gibert et al., 2014). We first used a slow-killing test to observe nematodes' death after bacterial colonization of intestine as biofilm structures. The mutant strain showed significant attenuation of virulence compared to the WT strain. This observation could be related to its lower ability to form biofilm in the nematode's intestine (Tan et al., 1999) and in vitro (this study). On the contrary, no difference was observed in terms of worm death in liquid killing assay or in LDH release by HaCaT cells. This may be in favor of a similar release of toxins or yet unknown cytoplasmic virulence factors by both strains or be related to a lack of sensitivity of this model.
Another part of this study focused on the impact of lytSR on global gene expression by transcriptomic analysis. Two time points were studied, at 6 h of growth equivalent to the mid-exponential phase when the expression of lytSR was maximal and at 8 h equivalent to the late-exponential phase when the expression of lrgAB was maximal. Of note, it was also found that the lrgAB operon transcription of S. aureus appeared maximal when cells entered stationary phase (Groicher et al., 2000). Transcription of lrgAB was constitutively drastically decreased in the ΔlytSR strain, indicating that the operon was activated by LytSR in S. lugdunensis. This is consistent with results for S. aureus and S. epidermidis (Sharma-Kuinkel et al., 2009;Zhu et al., 2010).
In addition, DNA microarray analyses show that the LytSR TCS regulates expression of a wide variety of genes involved in major cellular processes like metabolism of carbohydrates, nucleotides, and amino acids as well as energy metabolism. These results were also observed for other staphylococcal species (Sharma-Kuinkel et al., 2009;Zhu et al., 2010), suggesting that LytSR could affect cell viability and environmental adaptation by regulating the expression of genes controlling bacterial metabolic state. Unlike S. epidermidis, genes involved in pyruvate metabolism like the cidC gene that encodes a pyruvate:menequinone oxidoreductase were downregulated by LytSR. The higher expression level of cidC (2.7) in the S. lugdunensis ΔlytSR strain might partly explain the higher rate of dead cells observed in ΔlytSR biofilm. Indeed, in S. aureus, CidC protein plays a major role in bacterial programmed cell death during the stationary phase and in biofilm by converting intracellular pyruvate to acetate , which leads to cytoplasmic acidification and respiratory inhibition (Thomas et al., 2014). Overflow metabolism and specifically the balance between acetate and acetoin determine the fate of S. aureus cells (van den Esker et al., 2017). LytSR could therefore belong to the regulatory networks implicated in the connection between programmed cell death and metabolism in S. lugdunensis.
More surprisingly, our results have shown that LytSR was involved in the regulation of genes encoding known or putative virulence and colonization factors (Heilbronner et al., 2011). LytSR seems to downregulate genes encoding surface-anchored proteins including S. lugdunensis surface proteins (SlsB and SlsC; Heilbronner et al., 2011) and Fbl, the main fibrinogenbinding protein of S. lugdunensis (Marlinghaus et al., 2012). Moreover, the genes sstABD encoding putative iron siderophore uptake proteins were shown to be downregulated by LytSR. SstABCD ABC transporter is essential to use catechols and catecholamines as an iron source and therefore to staphylococcal survival and virulence (Beasley et al., 2011). On the contrary, among genes upregulated in a lytSRdependent manner, are genes encoding the bifunctional autolysin AtlL implicated in S. lugdunensis pathogenesis (Gibert et al., 2014;Hussain et al., 2015) as well as SLUG_11580 encoding a putative autolysin. However, our study did not show any involvement of LytSR in autolysis, suggesting that other systems might be involved in the control of autolysins in S. lugdunensis. For example, in S. epidermidis, atlE transcription is regulated by agr (Vuong et al., 2003), a system also described in S. lugdunensis (Vandenesch et al., 1993) but not yet functionally characterized.
Interestingly, genes encoding the putative type VII secretion system (essB, esaB, and essA) of S. lugdunensis were also upregulated by LytSR. These genes have been described by Warne et al. (2016) as homologous to those of S. aureus. This secretion system is implicated in the excretion of extracellular proteins across S. aureus cell membranes and has been associated with virulence, especially in abscess development (Warne et al., 2016). Several other genes encoding putative virulence factors have been shown to be upregulated by LytSR, like the CAAX amino terminal protease family proteins described as putative membrane-bound metalloproteases (Pei and Grishin, 2001), secretory antigen precursor homologous to the extracellular immune dominant protein SsaA postulated to be involved in S. epidermidis biofilm-associated infections (Lang et al., 2000), or the peptide methionine sulfoxide reductase MsrA, which is implicated in oxidative stress tolerance and virulence in S. aureus (Singh et al., 2015(Singh et al., , 2018. Of note, apart from atlL, no gene involved in S. lugdunensis biofilm formation, such as isdC (Missineo et al., 2014) or comEB (Rajendran et al., 2015), was upregulated by LytSR. The decreased capacity of the ΔlytSR strain to form biofilm could therefore be related to the downregulation of atllL (Gibert et al., 2014) and ssaA (Lang et al., 2000;Resch et al., 2005) or be indirectly linked to the altered metabolic state of the cells.
Lastly, LytSR was shown to upregulate (SarA and MarR families) and downregulate (MerR and GntR families) the transcription of several other transcriptional regulators controlling a wide diversity of metabolic processes and virulence (Rodionov, 2007;Jenul and Horswill, 2018). Further studies are necessary to determine whether the involvement of LytSR in the transcription of genes described here is direct or indirect through interaction with other regulators. Especially, electrophoretic gel shift assays could be conducted to determine whether the LytR response regulator binds directly to the promoter regions of its target genes.

CONCLUSION
This study is the first to characterize a two-component regulatory system in S. lugdunensis, a particularly virulent CoNS. Our results demonstrate that LytSR plays a significant role in the biofilm formation of S. lugdunensis, probably in connection with cell death and metabolism. Transcriptional analyses have shed new light on the putative virulence factors of S. lugdunensis that could be regulated by LytSR. Furthermore, LytSR appears to play a role in the virulence of S. lugdunensis, as evidenced by the in vivo model of C. elegans infection, but this warrants confirmation in a rat endocarditis model for example. Overall, LytSR appears as a major TCS implicated in S. lugdunensis pathogenesis. Further studies of this regulatory system are crucial to investigate how LytSR activation fits into the regulatory networks modulating S. lugdunensis virulence.

DATA AVAILABILITY STATEMENT
Data are available in the Gene Expression Omnibus (GEO), accession number GSE139909.

AUTHOR CONTRIBUTIONS
SD, VO, and MP-C designed the study. J-MF, VO, and YZ performed the microarray analyses. TC and OL performed the cytotoxicity assays and C. elegans virulence assays. ED, JD, LD, MN, SD, SL, and VO performed all other experiments. SD, VO, and MP-C analyzed the data and wrote the manuscript. All authors have read and approved the final version of the manuscript.

FUNDING
The Caenorhabditis elegans WT Bristol strain N2 was provided by the CGC, which was funded by NIH Office of Research Infrastructure Programs (P40 OD010440).