The Glycosyltransferases sdgA and sdgB Expression in Staphylococcus Epidermidis Depends On The Conditions of Biolm Formation.

The Staphylococcus aureus’ SdrG protein is glycosylated by SdgA and SdgB for their protection against its degradation by the neutrophil’s cathepsin G. So far, there is not information about the role of Staphylococcus epidermidis’ SdgA nor SdgB in the production of biolm, therefore the main of this work was to determine the distribution and expression of sdrG, sdgA and sdgB genes in S. epidermidis in conditions of biolm. The frequency of the genes sdrG, sdgA and sdgB were evaluated by PCR in a collection of 75 isolates. The isolates were grown in dynamic conditions (in agitation) or static conditions (biolm productor: planktonic or sessile cells). The expression of sdrG, sdgA and sdgB were determined by RT-qPCR in cells grown under dynamic conditions (CGDC), as well as planktonic and sessile cells, and in cells adhered to a catheter (in vivo). The genes sdrG and sdgB were detected in 100% of isolates, meanwhile the gene sdgA was detected in 71% of the samples (p<0.001). The CGDC did not expressed the sdrG, sdgA and sdgB mRNAs. The planktonic and sessile cells expressed sdrG and sdgB, and the same was seen in cells adhered to the catheter. In particular, one isolate, able to induce biolm under cathepsin G treatment, expressed sdrG and sdgB in planktonic, sessile and in cells adhered to the catheter. This suggests that the state of cells adherence is an important factor for the transcription of sdgA, sdgB and sdrG. a at bottom wells plate (Nunc) and adjusted with TSB to a 1:200 dilution. strain inoculated in quadruplicate, human cathepsin G (Gibco) was added only to the strain HS60 at a nal concentration 0.1 µg/ml to induce the biolm formation. The plate was incubated at 37°C for 24 without afterwards, the oating planktonic cells were taken and the were by resuspended in 1 mL 1x PBS. Total RNA was extracted from all the samples and RT-qPCR were performed as

One critical condition in the rst step of bio lm formation, initial adhesion, is the strong adhesion of bacteria to surface of the implanted medical devices, which results in a successful infection. The surface of S. epidermidis is covered with a variety of cell wall proteins engaged to the peptidoglycan (Foster and Höök 1998). The adhesion of the bacteria to the surface of the implanted medical device is driven by the hydrophobicity of the bacterial cell membrane in which certain speci c proteins are involved such as AtlE, Bhp (Heilmann et al. 1997;Tormo et al. 2005) and SSP-1 and SSP-2 (Heilmann et al. 1996;Veenstra et al. 1996; Qin et al. 2007). When medical device is implanted within an organism, the extracellular matrix proteins such as the bronectin, brinogen, vitronectin and collagen from the host quickly cover the surface of device (Baier et al. 1984). Staphylococcus aureus and S. epidermidis possess a type of cell wall proteins called Microbial Surface Recognizing Adhesive Matrix Molecules (MSCRAMMs) that speci cally recognize to extracellular matrix proteins allowing an interaction and bind to them (Bowden et al. 2005; Mack et al. 2006). The Sdr proteins are members of MSCRAMM family, and they have been named as such due to the content of a region of repeated dipeptide serine -aspartate (SD) (McDevitt and Foster 1995). The proteins Sdr can interact speci cally with the extracellular matrix proteins that cover the surface of the implanted medical device (Josefsson et al. 1998).
Three proteins Sdr have been identi ed in S. epidermidis: SdrF that binds to collagen, SdrG that binds to brinogen and SdrH, with a ligand still not elucidated (Nilsson et al. 1998;Lei et al. 1999 (Hazenbos et al. 2013). In addition, the genes that codify for these glycosyltransferases in S. aureus are adjacent to the sdrCDE loci, and this array is completely conserved in different strains of S. aureus.
The lack of glycosylation in the SD region of Sdr proteins by SdgA or SdgB provokes the Sdr proteolysis by the neutrophil cathepsin G protease (Hazenbos et al. 2013). Also, the degradation of the nonglycosylated Sdr proteins has as consequence the loss of the bacterial capability to adhere to human brinogen (Hazenbos et al. 2013). Moreover, it has been demonstrated that the glycosyltransferases SdgA and SdgB are involved in the glycosylation of the plasmin-sensitive protein (Pls) and these Pls glycosyl residues can stimulate bio lm formation in S. aureus (Bleiziffer et al. 2017). The presence of the three genes aggA, aggB (sdgA), and aggC (sdgB) have been detected in S. aureus, which shows that AggA and SdgB contribute to staphylococcal agglutination with brin brils in human plasma (Thomer et al. 2014).
So far, the distribution and expression of S. epidermidis' sdgA and sdgB genes have not been explored in detail, therefore in this work we determined the expression of sdgA and sdgB genes in planktonic and sessile cells (grown in bio lm), in cells grown under dynamic conditions (CGDC), and in cells adhered to a catheter.

S. epidermidis isolates
We worked with 29 isolates from ocular infection (OI) with capacity to generate bio lm (13 isolates; 44.8%) or not (16 isolates; 55.2%), and 46 isolates from healthy skin (HS), also with capacity to generate bio lm (2 isolates; 4.3%) or not (44 isolates; 95.7%). The HS60 isolate was used as non-bio lm producer control; in this last one, the bio lm was induced with cathepsin G protease treatment.
For the expression assays, the strain type RP62A was used as a bio lm producer control.
The isolates were cultured in 3 ml of Tryptone soy-broth (TSB; Sigma-Aldrich, Estado de Mexico, Mexico) and incubated overnight at 37°C. The cells were obtained by centrifugation at 12000 rpm for 3 min, then 200 µl of Whinston solution (2% Triton X-100, 1% SDS, 10 nM NaCl, 10 mM Tris base at pH 8.0 and 1 mM EDTA) were added. A mechanical disruption was conducted for 1 min at 3000 rpm followed by incubation in ice for 30 s (this was done in 5 cycles), then another 200 µl of the Whinston solution were added. The DNA was extracted with phenol-chloroform-isoamylic alcohol (25:24:1) and precipitated with one volume of isopropanol. The genes sdrG, sdgA and sdgB were ampli ed using the oligonucleotides shown in Table 1. The PCR ampli cations were carried out with MyTaq™ DNA polymerase (Invitrogen, MA, USA) according to the manufacturer´s instructions.  Table 1). Gene products were aligned with BLAST genes to obtain homology. The SdgA (SAUSA300_0549) and SdgB (SAUSA300_0550) gene products from Staphylococcus aureus USA300_FPR3757 were used as reference. To determine the genomic arrangement the azoreductase gene (SAUSA300_0545) was used as a reference point, this gene is orthologous in all genomes. The phylogenetic tree was obtained with the PhyML program (www.atgc-montpellier.fr/PhyML) using the concatenation of ve proteins (RecA, RpoB, AtpD, GyrB and ClpB) with the LG evolutionary model and a bootstrap of 100.

Bio lm formation
Isolates were inoculated in TSB (Sigma-Aldrich) and incubated for 24 h at 37°C. Then, in 6 -well tissue culture plates (Nunc, Thermo Fisher Scienti c, MA, USA) the cells were diluted 1:200 in TSB medium. In the case of HS60 isolate, TSB medium was supplemented with 0.1 µg/mL of cathepsin G (Gibco, Thermo Fisher Scienti c). Plates were incubated for 24 hr at 37°C without stirring (static conditions). In the dynamic growing conditions assays, the plates were incubated for 24 hr at 37°C and kept shaking at 2.6 Hz, as Stepanovic et al. (2001) described. The bio lm formation was determined as described by Christensen et al. (1985).
Expression by RT-qPCR evaluated under dynamic conditions (non-bio lm condition) The strain S. epidermidis RP62A and the bio lm producers clinical and commensal isolates were inoculated in 6 well plates (Nunc, Thermo Fisher Scienti c) containing 3 ml of TSB (Sigma-Aldrich) and were incubated in dynamic conditions for 13 h at 37°C shaking at 2.6 Hz, as described by Stepanović et al. (2001); under these conditions the formation of bio lm was also tested as indicated above. The cell culture was transferred to a sterile tube and the no adherent cells were recovered by centrifugation at 12000 rpm and washed with DEPC treated water. The RNA puri cation and RT-qPCR were performed as previously described by Martínez-García et al. (2019). The expression of 16S rRNA was used as a control, and sdrG, sdgA and sdgB were determined using the primers listed in Table 1. Relative expression was determined by the 2 −ΔΔCt method.

Expression by RT-qPCR in bio lm cells (static growing conditions)
The strain S. epidermidis RP62A (bio lm producer control), the clinical and commensal isolates with bio lm phenotype, and the strain S. epidermidis HS60 were inoculated in TSB (sigma-Aldrich) and incubated at 37°C and shaking overnight. From those cultures, an inoculum was taken and transferred to a 6 at bottom wells plate (Nunc) and adjusted with TSB to a 1:200 dilution. Each strain was inoculated in quadruplicate, human cathepsin G (Gibco) was added only to the strain HS60 at a nal concentration of 0.1 µg/ml to induce the bio lm formation. The plate was incubated at 37°C for 24 h, without stirring, afterwards, the oating planktonic cells were taken and the sessile cells were obtained by scratching and resuspended in 1 mL of 1x PBS. Total RNA was extracted from all the samples and RT-qPCR were performed as previously described by Martínez-García et al. (2019).

Expression by RT-qPCR in a model of catheter (in vivo)
Female Balb/c mice were used in a model of subcutaneous implanted device-related infection according to Sander et al. (2012). This study was carried out following the recommendations of the bioethics review board of the "Escuela Nacional de Ciencias Biológicas-IPN." The mice were weighed and anesthetized by intraperitoneal injection with locain (100 mg/g of weight). The hair was removed from the back using electric hair clippers, and a small incision was made. Then, 1 cm of a sterile 14-gauge te on intravenous catheter (Exel International, FL, USA) was inserted, and the incision was sutured. Five mice were used as control (no bacterial inoculum) and the rest of the mice were injected through the catheter with a bacterial inoculum of 1.5x10 8 CFU in 20 µL of sterile PBS. The distribution of the strains utilized as inoculum for the catheters was as follows: ve mice were inoculated with the strain type RP62A, ve mice with the strain HS60 and ve mice with the strain HS60 supplemented with cathepsin G (0.1 µg/ml), ve mice with bio lm producer IO7 isolate and ve mice with bio lm producer HS6 isolate. After 7 days post-infection, the animals were sacri ced. The catheters were removed, and each one was put in 1 mL of 1x PBS and sonicated at 200 Hz for 5 min twice. On one hand, the CFU/mL for each catheter was determined by decimal dilutions, and on the other hand, a bacterial pellet was obtained by centrifugation in order to extract the total RNA and carry out the RT-qPCR as described previously.

Statistical analysis
To determine the proportion analysis, the accurate Fisher test was conducted. In order to analyze the expression levels, a two way ANOVA and Tuckey tests were conducted. These analyses were carried out with the software GraphPad Prism version 7.0.

Results And Discussion
Distribution of sdrG, sdgA, sdgB genes and genotypes A total of 75 isolates were tested and they were grouped in bio lm producers and non-bio lm producers. The sdrG and sdgB genes were detected indistinctly in all the isolates tested (100%), except for the gene sdgA, which was only detected in 71% (Table 2), showing a signi cant statistical difference (p<0.001) in relation to sdrG and sdgB genes, however there was not a difference between bio lm producers and nonbio lm producers. The gene sdrG was found in all isolates analyzed, which is in agreement with the former ndings, since the presence of the gene sdrG is reported between 78-91% of S. epidermidis strains isolated from orthopedic infections and from central venous catheter-associated infections (Arciola et al. Detection of the genes was carried out by PCR as described in materials and methods.
The genotypes identi ed in the isolates were two; 1) sdrG + , sdgA + , sdgB + and 2) sdrG + , sdgA − , sdgB + , where the genotype 1 was predominant (71%) over the genotype 2 (29%) considering all the isolates (p<0.001). Analyzing the frequency of genotypes within each source of isolation (IO and HS) genotype 1 is also predominant in both cases (OI p<0.001 and HS p<0.05) ( Table 3). Genomic array of the genes sdrG, sdgA and sdgB in S. epidermidis Several Staphylococcus genomes were analyzed in order to determine the diversity in genomic organization and conservation of the mentioned genes. It was found that there was some diversity in the presence and organization of these genes. Analyzing the Staphylococcus species, it was observed that one or more homologues of sdrG were present and the glycosylases sdgAB have been disorganized. In the gure 1 it can be observed that, in other species, the sdgAB genes are immediately contiguous, in the opposite direction, but in the S. epidermidis strains this organization was lost and only sdgB conserved its position with respect to sdrG (Fig. 1). The other glycosylase, sdgA, appeared at a considerable distance, between 200 and 365 kb (as is the case with ATCC12228). In other strains, it happened that sdrG is found at more distant positions (191 and 660 kb), as in 13T0028, JMUB898 and NCTC13838, that do not appear in the gure. sdgB is normally contiguous with the sdrG gene in all S. epidermidis strains analyzed; these two genes are in a opposite direction. Strains ATCC14990, CDC121, FDAARGOS_153 and FDAARGOS_161 presented another gene homologous to sdrG, annotated as sesJ (belonging to MSCRAMM family, cell wall-anchored protein), in chromosome positions unrelated to this cluster. The S. aureus strain USA300_FPR3757 showed similar organization as the MSHR1132 strain, three contiguous sdrCDE homologues. The phylogenetic tree showed representative Staphylococcus species and it was useful to associate phylogeny with the genetic conservation and diversity of these genes.
From this analysis we can infer that there was a translocation event in the glycosylase sdgA segment in S. epidermidis. On the other hand, there was conservation of the sdrG and sdgB positions, showing a greater functional or regulatory commitment. Given the diversity of these components and the loss of organization, it is possible that these are factors that in uence the pathogenicity of the strains.

Relative expression of sdgA and sdgB genes under dynamic growth conditions (non-bio lm condition)
From all the isolates grown under dynamic conditions no one formed bio lm (data no shown). The cells grown under dynamic conditions (CGDC) expressed only the 16S rRNA, and the expression of sdrG, sdgA and sdgB was not detected in the CGDC coming from the clinical and commensal isolates bio lm producers (data no shown), nor in the CGDC coming from the non-bio lm producers isolates (data no shown).
The sdrG expression in S. epidermidis has been scarcely studied. The expression of the sdrG gene do not occur in vitro nor under certain conditions such as: iron depletion, bacterial growth in the presence of 5% of CO 2 , conditions stimulated with whole human blood or with 70% of human serum (Sellman et al. 2008). The expression of sdrG gene has been only detected under in vivo conditions through the systemic infection of S. epidermidis in mice (Sellman et al. 2008). Moreover, it has been demonstrated that the abundance of the gene sdrG on the cell surface of S. epidermidis dramatically improves their ability to bind to brinogen-coated implanted medical devices (Vanzieleghem et al. 2015) implicating a role of SdrG in the in vivo adhesion conditions. The assays for the expression of sdrG in CGDC did not demonstrate that the induction of this gene can be achieved, as occur in sdgA y sdgB genes under these conditions. Our results con rm that there is a correlation between the expression of the gene sdrG and the expression of the genes sdgA and sdgB.
Relative expression of the sdrG, sdgA and sdgB in bio lm growth conditions (static growth) The expression of the genes of interest was assayed in S. epidermidis RP62A strain (bio lm producer). The bio lm was produced in static growth conditions, and there was absence of bio lm in the dynamic growth conditions (Fig. 2a). In the bio lm conditions the expression of sdrG and sdgB were observed in the planktonic and in sessile cells, but sdgA was not expressed (Fig. 2b). In the dynamic growth conditions, the CGDC did not expressed sdrG, sdgB nor sdgA (Fig. 2b). Also we did not nd statistical difference among the samples tested (p>0.05).
Afterwards, the relative expression of the genes of interests was tested in the clinical (13 isolates) and commensal (2 isolates) S. epidermidis isolates with bio lm producers phenotype. In the static growth condition, all the isolates produced bio lm (Fig. 3a), and both planktonic (Fig. 3b) and sessile cells (Fig.  3c) expressed sdrG and sdgB, and in some isolates the levels of expression of sdrG were higher than sdgB (p<0.05). Concerning the gene sdgA, both planktonic and sessile cells coming from three clinical isolates expressed sdgA ( Fig. 3b and c).
The in silico analysis of 20 genomes of S. epidermidis demonstrated differences in the genomic arrays reported for S. aureus (Hazenbos et al. 2013). In the analysis, 13 genomes of S. epidermidis the sequence of the gene sdgA was not found. This con rms the reports on S. aureus, in which is stated that the gene sdgA is not essential for the protection of the Sdr proteins against the proteolysis caused by the cathepsin G (Hazenbos et al. 2013; Thomer et al. 2014). In the present work, we suggest that in the case of S. epidermidis, the gene sdgA may not be essential because its expression was absent in planktonic and sessile cells in the bio lm producers OI isolates, in which case only 3 out of 13 expressed the gene sdgA.
We found that the in vitro growth conditions for the production of bio lm promote the expression of sdrG and glycosyltransferases sdgA and sdgB in planktonic cells as well as in sessile cells indicating a correlation between the expression of sdrG gene and glycosyltransferases. In a recent report it was demonstrated that umbelliferone, a natural product of the coumarin family, exerts an anti-bio lm effect on S. epidermidis by turning down the initial adhesion and cellular accumulation, due to the reduced expression of adhesion encoding genes (icaD, atlE, aap, bhp, ebh, sdrG, and sdrF) (Swetha et al. 2019). To the best of our knowledge, there are no reports on the expression of the genes sdgA and sdgB of S. epidermidis under bio lm conditions.
On the other hand, we also analyzed if the correlation between the expression of sdrG gene and glycosyltransferases also happen in isolates with non-bio lm forming phenotype. The isolate S. epidermidis HS60 has a non-bio lm forming phenotype, and to promote its bio lm formation it is necessary the presence of the cathepsin G. In this isolate, the presence of cathepsin G induced the bio lm formation (Fig. 4a) and the genes sdrG, sdgB, and sdgA were expressed (Fig. 4b)  The catheters inoculated with different strains of S. epidermidis (RP62A, HS60, IO7, HS6) were implanted in the back of Balb/c mice. Seven days post implantation the mice were sacri ced and the catheters were recovered. For each catheter the CFU/mL was quanti ed and the bacterial pellet was used for RT-qPCR. In the case of S. epidermidis RP62A, which is the model strain for bio lm production, 1.4x10 7 CFU/mL were obtained, and the same number was for IO7 isolate, and HS6 isolate was 1.0x10 7 CFU/mL. In the case of S. epidermidis HS60, which is not a bio lm producer, we got 2.5x10 6 CFU/mL, however in this same strain supplemented with cathepsin G reached a count of 9.3x10 6 CFU/mL (Fig. 5a). There was a signi cant difference between the strain S. epidermidis HS60 that was not supplemented with cathepsin G and the strain that was supplemented with it (p<0.05).
From the bacteria recovered from the catheters, the expression of sdrG, sdgA and sdgB were evaluated. In the S. epidermidis RP62A, OI7 and HS6 isolates there was the expression of the sdgB and sdrG mRNAs. In the S. epidermidis HS60 isolate sdgB and sdrG mRNAs were expressed, either treated or not treated with cathepsin G, meanwhile sdgA was only expressed in the isolate that was not treated with cathepsin G (Fig. 5b). In both isolates there was no signi cant difference among the different treatments (p>0.05).
Regarding S. epidermidis HS60 (non-bio lm producer phenotype), we observed that in the absence of cathepsin G treatment (bio lm inducer in this strain) HS60 adhered to the catheter and expressed the sdrG gene and the glycosyltransferases sdgA and sdgB genes, suggesting that the adherence of the bacteria to the catheter is enough stimuli for the expression of the genes. This is related to the fact that SdrG provides a strong binding force and slow dissociation with the brinogen and clustering of SdrG, and these biophysical features provide a molecular foundation for the ability of S. epidermidis to colonize to implanted biomaterials and to withstand physiological shear forces (Herman et al. 2014).

Conclusion
It was demonstrated that sdrG is expressed under bio lm conditions (in-vitro and in-vivo), at the same time, the genes sdgB and sdgA are also expressed indicating an association between sdrG and the glycosyltransferases. Moreover, the sdgA gene was expressed at a low proportion among the isolates, which suggests that it is not an essential gene for the SdrG glycosylation. So, strains with non-bio lm forming phenotype can be potentially bio lm producers since in their genome are the sdrG and sdgB genes and their expression can contribute to the adhesion to the catheter.  Figure 1 Genomic arrays found in Staphylococcus spp. The analysis was performed with 30 fully sequenced Staphylococcus genomes. The SdgA (SAUSA300_0549) and SdgB (SAUSA300_0550) gene products from Staphylococcus aureus USA300_FPR3757 were used as reference. To determine the genomic arrangement the azoreductase gene (SAUSA300_0545) was used as a reference point, this gene is orthologous in all genomes. The phylogenetic tree was obtained with the PhyML program with the LG evolutionary model and a bootstrap of 100.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. Table1supplementary.docx