Lugdulysin of Staphylococcus lugdunensis, a metalloprotease that inhibits and disrupts protein biolm of Staphylococcus aureus

Staphylococcus lugdunensis is a commensal skin microorganism that, unlike other coagulase-negative staphylococci, presents increasing clinical importance. This species yields a metalloprotease called lugdulysin that may contribute to its higher degree of virulence. This study aimed to determine the biochemical characterization of the lugdulysin produced by S. lugdunensis clinical isolates and investigate its effect on the formation and disruption of biolm of Staphylococcus aureus isolates. The protease was isolated and characterized for its optimal pH and temperature, activity in the presence of inhibitors and enzymatic kinetics. The inuence of metal cofactor supplementation on proteolysis was also evaluated, with and without inhibitors. Finally, the protease capacity to inhibit and disrupt biolms of different S. aureus lineages and biolm matrix was analyzed.

role in the formation and stability of microbial bio lms from many species [21,22]. In S. aureus isolates, these enzymes can limit the growth and detachment of bio lm by sarAand agr-mediated mechanism, respectively [22,23]. Proteases are also necessary to induce Staphylococcus epidermidis bio lm formation by processing the Aap protein [24]. Additionally, Connely et al. [25] have demonstrated that Bacillus subtilis lacking extracellular proteases could not produce bio lms [25]. In all cases, the bio lm matrixes were mainly composed of proteins like occurs in the bio lm of S. lugdunensis isolates [9,26,27].
Despite the initial characterization of lugdulysin carried out by Argemi et al. [7], its chemical and structural characteristics remain unknown. Hence, this study aimed to elucidate biochemical aspects of lugdulysin and investigate its in uence on the formation and disruption of bio lm from S. aureus isolates belonged to different lineages and presenting different bio lm matrices.

Results
Protease puri cation and expression identi cation Bands of 50kDa obtained after electrophoresis were excised from PAGE and MALDI-TOF MS and the search for homologous sequence indicated that it corresponded to putative neutral metalloprotease produced by S. lugdunensis ( Table 1). The subcellular localization prediction by Gpos-mPLoc and PSortB describes the sequence as extracellular (data not shown). The protein has a pre-protein region with yet unknown function [7]. The secondary structure, according to I-TASSER, shows that the protein is composed of 64.73% in loop, 13.27% in sheet and 21.97% in helix, with no transmembrane domains (Fig. 1). The protease sequence was also analyzed for its genetic ontology (Fig. 2). Due to the presence of the HEXXH domain, the enzyme requires a zinc molecule in its catalytic site to be active. The isoelectric point of the enzyme is 4.99. Therefore, the enzyme is possibly active whilst negatively charged. Staphylococcus lugdunensis metalloprotease is highlighted in bold with the respective match, score, and its predicted mass (kDa). Other proteins with more matches are common contaminants inherent to the methodology.

Protease biochemical characterization
The proteolytic enzyme activity from 546s S. lugdunensis isolate identi ed by mass spectrometry was measured by a colorimetric biochemical assay using azocasein as substrate. The EDTA caused a signi cant reduction in proteolytic activity, with a statistically signi cant decrease (p < 0.01) compared to the activity without inhibition (Fig. 3A). There was no recovery in activity with any metals supplemented after EDTA inhibition (Fig. 3B). The gure shows activity with 5 mM EDTA and no metals added as a control. Then, 1 mM calcium, magnesium, zinc, and manganese were added in the other bars showing no signi cant change between themselves and control.
Metal saturation in the medium was not able to recover the enzymatic activity of lugdulysin. Not only was there no recovery of activity after EDTA inhibition by metal ion supplementation, as there was no signi cant change in proteolytic activity with different metal supplementation without inhibition. In Fig. 3C, the rst bar shows unsupplemented protease activity, while the other bars show the same protease supplemented with 1 mM calcium, magnesium, zinc, and manganese. There is no signi cant difference between unsupplemented and supplemented protease activity. A negligible activity level was observed from pH 2.0 to 4.0, with a slight increase in activity from pH 6.0 and reaching peak activity at pH 7.0, reducing at pH 8.0 (Fig. 3D).
Regarding the in uence of temperature (Fig. 3E), it is notable that the highest proteolytic activity are shown between 37°C and 40°C, with temperatures above or below these values exhibiting considerable decrease.
The proteolytic activity begins immediately, with absorbance above control in 10 minutes of reaction Disruption and inhibition of bio lm production by proteases Trypsin can determine the bio lm composition between protein and non-protein bio lm, as it degrades speci cally the proteins present in it. Figure 4A shows the effect of trypsin or lugdulysin on bio lm formation in six staphylococcal isolates. Trypsin inhibited the formation of more than 70% of the S. lugdunensis protein bio lm and between 60% and 70% of the S. aureus protein bio lm. Lugdulysin metalloprotease showed a reduction above 50% for S. aureus protein bio lm and between 20% and 40% for S. lugdunensis bio lm. Less than 10% inhibition in S. aureus polysaccharide bio lm was observed.
Finally, the protease disruptive potential in pre-formed bio lm was evaluated. Trypsin disrupted between 40% and 50% of S. aureus protein bio lm and more than 70% of S. lugdunensis bio lm. Lugdulysin promoted a reduction from 40-50% in S. aureus protein bio lm, similar to trypsin, but no bio lm reduction was observed for S. lugdunensis. Contrastingly, the impact of lugdulysin in S. lugdunensis bio lm was shallow compared to trypsin (p < 0.0001). Non-protein matrix bio lm isolates did not shown any reduction with the addition of trypsin or lugdulysin (Fig. 4B).

Discussion
Lugdulysin could be involved in S. lugdunensis pathogenicity according to previous ndings. However, its fundamental biochemical properties had not yet been determined. Based on the genomic annotation of S. lugdunensis, only one metalloprotease presents a molecular weight of 37 kDa produced by this species, as brie y proposed by Argemi et al. [7]. The present study supports this nding, where the extracellular proteins were concentrated, and the proteolytic activity was con rmed by degradation of azocasein. This assay was also used to investigate the in uence of EDTA, a metalloprotease inhibitor that extinguished the enzyme's proteolytic activity.
Furthermore, the similarity of the protease to the previously characterized enzyme (MEROPS Accession number MER0001182) was con rmed by mass-spectrometry peptide sequencing. As there was no previous characterization of this enzyme, various functional properties, such as its activity lifespan, optimal pH and temperature were investigated, the in uence of inhibitors and supplementation with metallic ions. We further analyzed different functional properties and through in silico analysis, providing information regarding cell location, isoelectric point, putative cleavage preference and the 3D structure of lugdulysin. Zinc-ion binding sites were identi ed, which was already expected by the presence of the HEXXH domain. Predict Protein server identi ed putative metalloendopeptidase activity, cleaving L-amino acid-rich peptides at inner parts of peptides instead of amino or carboxy-terminal [28]. Given the presence of a signal peptide, the protease is likely secreted, as previously suggested by Argemi [7].
In the present study, EDTA signi cantly reduced the proteolytic activity, con rming the lugdulysin as a metalloprotease [29,30]. In our attempt to recover its activity after inhibition, there was no observable recovery after adding Ca 2+ , Mg 2+ , Zn 2+ or Mn 2+ . It is possible that its structure is affected by the metal removal, which could prevent the recovery of the proteolytic activity [31]. It is known that metalloproteases respond diversely to different metallic ions. The response of the lugdulysin to ions is similar to aureolysin, another staphylococcal metalloprotease produced by the S. aureus [32]. For both metalloproteases, after inactivation with EDTA, ions were also unable to restore the protease activity. On the other hand, hyicolysin, a metalloprotease of S. hyicus, has its activity entirely recovered by adding zinc and cobalt, and partially restored with Mg 2+ , Ca 2+ and Cu 2+ [17]. These data indicate a close relationship with S. aureus, as already shown among other features presented by both species, such as adhesins and hemolysins [9].
The biochemical characterization indicated that, despite its structural similarity with hyicolysin [7], lugdulysin shares more biochemical similarities with aureolysin. For instance, the optimum pH and temperature of hyicolysin are 7.4 to 7.9 and 55 ºC, respectively [17], while for lugdulysin and aureolysin is 7.0 and 37 ºC. Indeed, whilst the clinical relevance of hyicolysin remains unclear, it is known that aureolysin play an essential role in the pathogenesis of S. aureus infections. This metalloprotease can modulate the pathogenesis of osteomyelitis by triggering alterations in bone turnover [16], promotes the escape of immune response, inhibiting the complement system by degradation of C3 component [33] and cleaves staphylococcal surface-associated proteins allowing the transition from an adherent to an invasive phenotype [34]. Thus, the biochemical properties of lugdulysin, like aureolysin, could be related to the emergence of S. lugdunensis as pathogenic bacteria implicated in severe infections, especially endocarditis and osteoarticular infections [18].
The effect of proteases against bio lms of S. aureus is already well established in the literature [35,36]. Here, lugdulysin signi cantly reduced the S. aureus pre-formed protein bio lm, similar to trypsin, inhibiting the production of protein bio lm by S. aureus. Diversely, the same outcome did not occur in S. lugdunensis isolates.
To this species, lugdulysin did not affect the pre-formed bio lm, causing only a slight inhibition in bio lm production. These results are surprising because not even the bio lm of S. aureus is so resistant to own metalloprotease as the bio lm of S. lugdunensis appears to be resistant to lugdulysin. In vitro studies demonstrated that aureolysin signi cantly disrupt the S. aureus pre-formed protein bio lm [22,36]. Even more, Abraham & Jeferson [36] found that the inactivation of the aureolysin gene augments S. aureus bio lm production. Nonetheless, some species produce proteases that can protect and modulate the bio lm formation whereas attack bio lms of bacteria from other species [37,38]. However, to our knowledge, there are no reports to date of any staphylococcal protease that simultaneously acts to protect the bio lm of the species that produces it and causes damage to the bio lm of another species. The resistance of the S. lugdunensis protein bio lm to its protease could demonstrate a possible new mechanism that may generate a competitive advantage among staphylococci.
Our study presents some limitations. First, biochemical studies are needed for a better comprehension of lugdulysin characteristics, such as substrate speci city. Second, we have not established the mechanism by which lugdulysin affects S. aureus bio lm. It remains unclear whether it degrades the protein matrix or acts on proteins involved in bio lm gene expression. Also, the effect of lugdulisin on S. aureus was only evaluated in vitro, which limits the understanding of the role of this metalloprotease as a virulence factor.

Conclusions
This report identi ed optimal conditions of function and stability of the lugdulysin produced by S. lugdunensis isolates from human clinical specimens. Moreover, we showed the striking effect of this protease on the formation and dispersion of bio lm from different S. aureus isolates, a pathogen of signi cant clinical relevance which bio lm has been associated with antimicrobial resistance and invasive medical devices related infections. These preliminary ndings show that lugdulysin may be a new mechanism of competition and/or modulation of the staphylococcal bio lm.

Bacterial isolates
Previous published bacterial isolates of S. lugdunensis (541s and 546s) and S. aureus (63a, 1636a, 1176a and 1348a) [9,39,40] were selected for the study. These isolates belonged to the Laboratory of Hospital Infection collection at the Federal University of Rio de Janeiro. They presented distinct genotypic pro les determined by the PFGE technique. In addition, the bio lm characteristics and biochemical composition were also determined previously, according to Ferreira et al. [41] ( Table 2).  SA Frankfurt., Germany). The recovered supernatant was then subjected to a tangential ow ltration procedure (Millipore Lab Scale TFF System, Millipore, Mass., USA) using an initial volume of 1L and reducing to approximately 30 mL for approximately 4 hours. The supernatant was concentrated using a 50 kDa ultra ltration membrane (Pellicon XL 50 -Millipore), and lyophilized. The dry weight was determined and the material was resolubilized in Tris-HCl 20 mM pH 7.0.

Protein pro le analysis
The concentrated supernatant was analyzed using SDS-PAGE (Sodium Dodecyl Sulphate Polyacrylamide Gel Electrophoresis) with 15% acrylamide according to Laemli [42] and stained according to Neuhoff [43]. The coomassie-stained gel was then destained and submitted to in-gel digestion using trypsin 20µg / mL (Sigma-Aldrich), adapted from Shevchenko [44]. The trypsinized peptides were analyzed in a MALDI-TOF/TOF type mass spectrometer (Micro ex LT, Bruker, USA) [44]. Mascot Distiller software was used to analyze the processed bands for MALDI-TOF/TOF analysis and raw data (Matrix Science, version 2.2.1.0). Protein identi cation was performed by searching for homologous sequences in online Mascot software ( Table 2). In silico analysis of the metalloprotease was performed with the sequence deposited by previous characterization [7], available at UniProt under the accession number A0A133QCC8. Proteases domain and other biochemical characteristics were proposed using the automated Predict Protein (https://predictprotein.org/), using a database of up-to-date public sequences, performing alignments, and predicting protein function and structure [45]. Prediction of cell location was performed using automated domain service, while analysis of other biochemical characteristics was performed using the automated service to predict subcellular localization of Cell-PLoc proteins, Gpos-mPLoc, for prediction speci cally in Gram-Positive bacteria [46]. Those data correlated to data obtained in PSortB, which performed the same analysis [47]. The putative structure was obtained via homology modeling using the protein prediction webserver I-TASSER [48].

Protease biochemical characterization
To determine proteolytic activity, a colorimetric method determined by Lei and coworkers [29] was carried out, with adaptations. As a positive control, trypsin (Sigma-Aldrich) (initial concentration 20 µg/mL) was used, and, for the negative control, the protease received immediate addition of 10% trichloroacetic acid (TCA). The activity units (U) that were determined as the variation of 440 nm absorbance per mg of protein per hour. This protocol was used to evaluate the effect of inhibitors on protease, with the addition of 5mM EDTA in the reaction [47]. Various divalent metals were supplemented in the reaction medium to evaluate the metal in uence on proteolytic activity, following proteolytic activity method [29] and adding 1µL of 1M metallic salt. In this study, Calcium Chloride (VETEC, Brazil), Zinc Chloride (Isofar, Brazil), Manganese Chloride (Carlo Erba, Italy) and Magnesium Chloride (Sigma-Aldrich, USA) were used. To assess the recovery of proteolytic activity, the azocasein method [29] was carried out with EDTA-inhibited enzyme. Various metallic salts were supplemented in the reaction medium to overcome the EDTA chelating potential following the abovementioned method. The optimum pH was determined with various buffers with different pH were used. The selected buffers were pH 2.0 KCl-HCl buffer, pH 4.0 acetic acid/sodium acetate buffer, pH 6.0 and pH 8.0 phosphate buffer, 20 mM. The method was conducted as per the previously described method [29] with 37°C incubation. To evaluate optimum enzyme operating temperature, the azocasein method [29] was performed at pHk9l 7.0 in different temperatures: 10°C, 20°C, 30°C, 40°C, 50°C, 60°C, 90°C. The rate of substrate cleavage was determined with a substrate serial dilution curve following, for 15 minutes. The concentrations of azocasein used were: 0.04, 0.08, 0.16, 0.32, 0.640 and 1.28 mg/mL. Michaelis-Menten and Lineweaver-Burk curves were plotted based on these data to determine Km and V max for azocasein.
Protease in uence on bio lm formation Metalloprotease in uence on bio lm formation was analyzed by the quantitative micromethod to evaluate bio lm production described by Stepanovic [49] with modi cations. Twenty microliters of a bacterial suspension in sterile distilled water corresponding to the 0.5 McFarland standard were added in triplicate to the wells containing 180 µl of tryptic soy broth, TSB (Becton, Dickinson and Company; Sparks, MD, USA), supplemented with 1% glucose (Isofar; Duque de Caxias, RJ, Brazil). Finally, 15 µL of freeze-dried protease diluted in PBS (20 µg/mL) were added in triplicate for each. The plates were incubated at 37ºC for 24h. Then, wells were washed twice with sterile PBS buffer (pH 7.2), dried for 1 h at 60°C, and stained with 200 µL of a0.1% safranin solution (w/v, in water) for 15 min. After, wells were washed twice again, and 200 µL of a 95% ethanol solution was added. Absorbance (OD492nm) was read after 30 min of incubation at room temperature. The culture medium without bacterial inoculum was used as a negative control and trypsin (20 µg/mL) was used as the positive control. All experiments were performed in triplicate at 3 independent times.
Protease in uence on pre-formed bio lm To observe the protease in uence in pre-formed bio lm, the bio lm formation was initially performed on 96-well plate as previously described [49] with modi cations. After the bacterial suspension is added to the wells and the microplate were incubated at 37ºC for 24h. The medium was discarded, and three washing steps with 200 µL sterile PBS buffer, pH 7.2 (Laborclin, Brazil) were performed to remove unbound cells. Then, 135 µL of PBS 100 mM pH 7.5 and 15 µL of freeze-dried protease (20 µg/mL) were added in triplicate for each isolate, and the 96-well plate was submitted to a second incubation at 37°C for 1 = h. After, the medium was discarded, another wash was performed with 200µL PBS to remove unbound cells, followed by incubation at 60°C for 1 hour, and wells were stained with 200 µL of a 0.1% safranin solution (w/v, in water) for 15 min. After, wells were washed twice again, and 200 µL of a 95% ethanol solution was added. Absorbance (OD492nm) was read after 30 min of incubation at room temperature. The culture medium without bacterial inoculum was used as a negative control, and trypsin (20 µg/mL) was used as the positive control. All experiments were also performed in triplicate at three independent times.  Figure 1 Predicted 3D structure of the Staphylococcus lugdunensis neutral metalloprotease by I-TASSER homology modelling. TM-score = 0.44 ± 0.14; RMSD = 12.5 ± 4.3 Å. HEXXH (HEYQH) domain highlighted in red (1A).

Figures
Residues HIS241, HIS245 (highlighted in yellow) and GLU242 (highlighted in beige) are the catalytic triad, zoomed in Figure 1B. GLU268 was also highlighted as a yellow stick as it is a known catalytic residue conserved among M30 proteases. The catalytic zinc ion is represented by a grey sphere pointed by an arrow. The structure is C-α trace coloured according to sequence from dark blue (N terminus) to red (C terminus). Distances measured from the catalytic zinc to the catalytic amino acids were 2.8, 3.1 and 4.0 Å for the residues HIS241, HIS245 and GLU268, respectively. Figures were drawn using PyMOL (www.pymol.org). Figure 2 database. The protease has two main domains, the HEXXH responsible for zinc (and other divalent ions) binding and the catalytic triad responsible for the peptidase activity. As per the highlighted boxes, the peptidase domain has aminopeptidase and endopeptidase activities, which are metal dependent. Moreover, the protease has also peptide binding capacity, which may be correlated with its activity.

Figure 3
Aspects associated with the lugdulysin activity puri ed from Staphylococcus lugdunensis supernatant using azocasein as substrate. A) Inhibition of metalloprotease activity with EDTA 5 mM (control = without EDTA) (p = 0.0045). B) Proteolytic activity recovery test with metallic ions supplementationCa2+, Mg2+, Zn2+or Mn2+. No Page 16/17 signi cant difference was found in relation to EDTA inhibition (p = 0.9803, 0.9530, 0.9763 and 0.9772 respectively). C) Effect in the enzymatic reaction by the addition of Ca2+, Mg2+, Zn2+orMn2+. Supplementation with metallic ions did not present signi cant difference when compared to no supplementation as control (p = 0.9930, 0.8772, 0.9950 and 0.1715 respectively). D) Optimal pH determination. The X-axis represents the pH values tested, while y-axis shows the enzymatic activity in U. E) Determination of the optimal temperature conditions. X-axis represents temperature range used; y-axis shows the enzymatic activity in U. F) Enzymatic activity vs time. X-axis represents the timespan used; y-axis shows enzymatic activity in units of absorbance. G) Lugdulysin Michaelis-Menten constant determination. H) Double reciprocal Lineweaver-Burk graph.

Figure 4
In uence of the lugdulysin or trypsin on S. aureus and S. lugdunensis bio lms. A) Effect of trypsin or lugdulysin non bio lm formation. Trypsin inhibited more than 70% of the S. lugdunensis protein bio lm (isolates 541s and