Phenotypic and Genotypic Virulence Characterisation of Staphylococcus pettenkoferi Strains Isolated from Human Bloodstream and Diabetic Foot Infections

Staphylococcus pettenkoferi is a recently described coagulase-negative Staphylococcus identified in human diseases, especially in infections of foot ulcers in patients living with diabetes mellitus. To date, its pathogenicity remains underexplored. In this study, whole-genome analysis was performed on a collection of 29 S. pettenkoferi clinical strains isolated from bloodstream and diabetic foot infections with regard to their phylogenetic relationships and comprehensive analysis of their resistome and virulome. Their virulence was explored by their ability to form biofilm, their growth kinetics and in an in vivo zebrafish embryo infection model. Our results identified two distinct clades (I and II) and two subclades (I-a and I-b) with notable genomic differences. All strains had a slow bacterial growth. Three profiles of biofilm formation were noted, with 89.7% of isolates able to produce biofilm and harbouring a high content of biofilm-encoding genes. Two virulence profiles were also observed in the zebrafish model irrespective of the strains’ origin or biofilm profile. Therefore, this study brings new insights in S. pettenkoferi pathogenicity.


Introduction
Diabetes mellitus is considered as a global epidemic with almost 550 million adults affected worldwide, a number predicted to rise to 640 million by 2030 [1]. Diabetic foot ulcers (DFU) represent a significant complication of diabetes mellitus, affecting 34% of patients during their lifetime [2]. Initially superficial, these chronic wounds are responsible for frequent infections (more than 50% of cases) [3] spreading to soft tissues and bone structures in 44 to 68% of cases [2]. Bone infection leads to diabetic foot osteomyelitis (DFOM), a real public health problem responsible for one amputation every 20 s worldwide [4]. DFOM and amputations participate in reducing the life quality of patients, increasing the morbidity and mortality, and represent a costly issue for health care systems [2,[4][5][6].
The majority of these chronic wounds are polymicrobial [7]. Culturomic and metagenomic approaches revealed that DFU and DFOM ecology is established by a predominance of Firmicutes and particularly Staphylococus aureus and coagulase-negative Staphylococcus (CoNS) [7][8][9]. While S. aureus is clearly implicated in the unfavourable course of these wounds [10], the role of CoNS in infected DFUs remains poorly understood, even though their pathogenicity has been reported [11][12][13][14]. Twelve of the fifty CoNS species, including Staphylococcus lugdunensis, Staphylococcus saprophyticus, Staphylococcus schleiferi and Staphylococcus epidermidis, have been identified as true pathogens and are at the origin

Resistome of S. pettenkoferi Strains
The main characteristics and resistance profiles of the 29 S. pettenkoferi clinical isolates are listed in Table 1. Thirteen resistance-associated genes were reported after genomic analysis of the 29 S. pettenkoferi genomes (Table S2). Resistome comparisons between strains isolated from infected DFUs and blood cultures are summarised in Table 2. No difference was noted between the resistome content and the results of susceptibility testing. Moreover, this resistome content was not significantly different between strains isolated from blood culture and infected DFUs. Eleven clinical isolates (41.4%) were resistant to penicillin G (with the presence of the blaZ gene) and only one strain (SP165) carried the mecA gene and was methicillin-resistant. All isolates were susceptible to gentamicin. The only resistance to aminosides was the presence of the ant(9)-Ia gene, encoding an aminoglycoside nucleotidyltransferase, and presenting in three strains isolated from infected DFUs. The SP165 strain was resistant to rifampicin with a mutation in the rpoB gene (p.H481N). NSP009P was resistant to tetracyclin, with the tet(K) gene detected in its genome. The ermA gene, conferring resistance to erythromycin and clindamycin, was only detected in 10.3% (n = 3) of the strains, while the vgaA gene, conferring resistance to streptogramin A and related antibiotics, was identified in 20.7% (n = 6) of the strains. Concerning resistance to quinolones, 20.7% (n = 6) of the genomes harboured a mutation in gyrA (corresponding to p.S84L in gyrase A protein) and 76.7% (n = 23) in the glrA (corresponding to p.S80Y in topoisomerase IV subunit A protein) gene. Strains isolated from infected DFUs more frequently possessed a grlB mutation (corresponding to pE422D in topoisomerase IV subunit B protein) than strains isolated from blood cultures (59% vs. 14.3%, respectively). The fusB gene was detected in 26.7% (n = 8) of the genomes. Among the 14 strains resistant to fosfomycin, 10 harboured the fosB gene, which was found in 37.9% (n = 11) of the studied genomes. Five isolates (17.2%) were multidrug resistant (i.e., resistant to ≥3 antimicrobial groups) (SP165, NSP009P, NSP004P, NSP012P and NSP001H).

Ability to Form Early Biofilm
The ability of 29 S. pettenkoferi clinical strains to form biofilms was analysed by the Biofilm Ring Test ® (Biofilm Control, Saint-Beauzire, France) and compared against the CIP 107711 reference strain. Biofilm formation index values (BFI) evaluating S. pettenkoferi adhesion ability were distributed in three distinct groups: BFI values = 0 at 4 h of incubation for Group 1; BFI values < 5 at 5 h of incubation for Group 2; and BFI values ≥ 5 at 5 h of incubation for Group 3 ( Figure 2). Interestingly, the majority of S. pettenkoferi strains (n = 17, 58.6%) were able to produce an early biofilm after 4 h of incubation in BHI and were thus classed to Group 1. Group 2 contained nine (31.0%) strains isolated from infected DFU which required a longer incubation time to reach BFI values near 2 at 5 h (a value considered as fixed biofilm in this assay). Finally, three (10.3%) strains constituting Group 3 presented low adherence and potential of biofilm formation. The reference strain CIP 107711 represented our negative for biofilm production with BFI values around 20 during the whole experiment. The three groups showed significant differences in the change in thus classed to Group 1. Group 2 contained nine (31.0%) strains isolated from infect DFU which required a longer incubation time to reach BFI values near 2 at 5 h (a val considered as fixed biofilm in this assay). Finally, three (10.3%) strains constituting Grou 3 presented low adherence and potential of biofilm formation. The reference strain C 107711 represented our negative for biofilm production with BFI values around 20 durin the whole experiment. The three groups showed significant differences in the change BFI values over time (Mann-Whitney test; p < 0.001). Interestingly, all strains isolated fro blood cultures belonged to Group 1, with strong production of early biofilm. Figure 2. Ability of 29 S. pettenkoferi strains to form early biofilm in BHI using BRT ® (Biofilm Co trol). K-means clustering led to three groups: in red, strains with early low biofilm index (BFI) v ues; in green, low adherent strains; in orange, strains with intermediate BFI values; and in blue, t S. pettenkoferi CIP 107711 reference strain with no biofilm production. Dotted horizontal lines: >7, biofilm; <2, fixed biofilm. Means ± standard errors of the mean of BFIs for at least three independe replicates per group are presented. Statistical differences between BFI values of the three groups 3, 4 and 5 h of incubation were obtained using Mann-Whitney tests. ***, p < 0.001. Figure 2. Ability of 29 S. pettenkoferi strains to form early biofilm in BHI using BRT ® (Biofilm Control). K-means clustering led to three groups: in red, strains with early low biofilm index (BFI) values; in green, low adherent strains; in orange, strains with intermediate BFI values; and in blue, the S. pettenkoferi CIP 107711 reference strain with no biofilm production. Dotted horizontal lines: >7, no biofilm; <2, fixed biofilm. Means ± standard errors of the mean of BFIs for at least three independent replicates per group are presented. Statistical differences between BFI values of the three groups at 3, 4 and 5 h of incubation were obtained using Mann-Whitney tests. ***, p < 0.001.

Biofilm-Related Genes within S. pettenkoferi Genomes
Biofilm was further studied by determining the biofilm-encoding genes in all 29 S. pettenkoferi genomes. The coding sequences (CDS) annotated were assigned functions based on Best DIAMOND Hit using the UniProt Reference Clusters (UniRef50) database. The prevalence of these biofilm-associated genes in the 29 studied S. pettenkoferi genomes according to the biofilm production of the strains and their origin (blood cultures or DFUs) are presented in Table 3.
All strains harboured the atlE gene, encoding for autolysin E, a major contributor to primary attachment to abiotic surfaces; the BUY34_0850 gene encoding for phenol soluble modulin beta protein (known for its major role in biofilm detachment); the splF and htrA genes encoding for serine proteases involved in biofilm structuration; and the clpP gene encoding for ClpP protease, implicated in S. epidermidis biofilm development. The poly-Nacetylglucosamine (PNAG) production is dependent on the presence of the intracellular adhesion (ica) operon for which the icaA, icaB and icaC genes were identified in all strains, while icaD and icaR were absent. We also noted a high presence of the most important biofilm regulators genes (agrA and agrC, being part of the Agr quorum sensing system; the rsbUVW-sigB operon; mgrA; sarA; luxS; lytSR; alrRS; cidA; and lgrA). Table 3. Comparison of biofilm-related genes between the S. pettenkoferi blood culture and DFU strains groups and biofilm groups.

Biofilm Steps
Putative

INITIAL ADHESION
To abiotic surfaces bhp Biofilm associated protein 0 (0%) To biotic surfaces fnbpA, fnbpB Fibronectin binding protein A and B 0 (0%) agrA, agrC Accessory genes regulators A and C 7 (100%) agrB, agrD Accessory gene sregulators B and D 0 (0%) LytSR two component regulatory system Interestingly, no significant differences were observed between the presence of biofilmencoding genes and the profile of the biofilm producers, nor between the origin of the strains (DFU and blood culture) and the clades. Finally, a previous study on S. saprophyticus strain MS1146 described the presence of the uafB gene, encoding for an adhesin involved in biofilm, on the pSSAP1 plasmid [28] (Table S3). Notably, this plasmid and the uafB gene were present in strains belonging to the two biofilm-producing groups (9/17, 52.9% in Group 1 and 4/9, 44.4% in Group 2) but absent in bacteria belonging to the Group 3.

Fitness of S. pettenkoferi Strains
Based on our biofilm results, we selected three strains isolated from infected DFUs representing the three groups identified by the BRT ® experiments: NSP003P (clade II, biofilm Group 1, NSP009P (clade I-a, biofilm Group 2) and NSP0023P (clade I-b, biofilm Group 3), as well as one strain isolated from blood cultures, NSP004H (clade I-b, biofilm Group 1). S. aureus reference strain Newman [29] was used as control. Growth profiles of these strains were monitored for over 24 h (Figure 3). . Growth curves of four S. pettenkoferi strains compared to S. aureus Newman strain using the Gompertz equation. A Wilcoxon test was used to compare the slopes of all S. pettenkoferi strains vs. S. aureus Newman; significant difference was at ** p < 0.01. A Dunn test was used to compare growth rates between the S. pettenkoferi strains; significant difference was observed between NSP004H and NSP003P at * p < 0.05.

Virulence Study of S. pettenkoferi Strains Using a Zebrafish Embryos Model
In vivo virulence of the four selected S. pettenkoferi strains (NSP003P, NSP009P, NSP023P and NSP004H) with different biofilm productions and origins was tested using the bath infection model of wounded zebrafish embryos, as previously described for S. pettenkoferi [27]. The viability of zebrafish embryos was monitored for 48 h and we observed that mortalities in injured larvae infected with S. pettenkoferi began at 18 hpi for all strains (Figure 4). Strains showed different virulence profiles: deaths increased up to 95% by 24 hpi for the NSP009P strain, while embryos infected by NSP003P, NSP0023P and NSP004H showed a mortality of around 50% at 36 h (Log-rank Mantel-Cox statistical test, p < 0.001). S. pettenkoferi NSP009P was two times more virulent than the other three strains with less than 5% embryo survival at 24 hpi. Interestingly, NSP009P displayed the same virulence profile as for the S. pettenkoferi SP165 isolate previously described [27]. . Growth curves of four S. pettenkoferi strains compared to S. aureus Newman strain using the Gompertz equation. A Wilcoxon test was used to compare the slopes of all S. pettenkoferi strains vs. S. aureus Newman; significant difference was at ** p < 0.01. A Dunn test was used to compare growth rates between the S. pettenkoferi strains; significant difference was observed between NSP004H and NSP003P at * p < 0.05.
The growth of S. aureus Newman was significantly faster than the S. pettenkoferi strains as illustrated by the maximum absorbances measures in the stationary phase (Ym) (Ym, S. aureus Newman = 1.644, Ym, NSP003P = 1.234, Ym NSP009P = 1.062, Ym NSP023P = 1.086 and Ym, NSP004H = 0.953 respectively, disjoint interval confidence; Table S4). The comparison between each S. pettenkoferi strain showed that NSP0023P presented a significantly higher Ym whereas NSH004H had the lowest. No significant differences between strains were observed for the lag times (K) and inflexion points (1/K) excepted for NSP023P and S. aureus Newman (K NSP023P = 0.246 and K S. aureus Newman = 0.194; 1/K NSP023P = 4.066 and 1/K S. aureus Newman = 5.147, disjoint interval confidence).
Moreover, a significant difference in slopes was observed between all the S. pettenkoferi strains compared to S. aureus Newman (Wilcoxon test; p = 0.009), confirming that bacterial growth was slower for S. pettenkoferi strains. Different growth rates were also observed between NSP004H and NSP003P with a significant slope observed for NSP004H, indicative of a lower growth rate (Dunn test; p = 0.03).

Virulence Study of S. pettenkoferi Strains Using a Zebrafish Embryos Model
In vivo virulence of the four selected S. pettenkoferi strains (NSP003P, NSP009P, NSP023P and NSP004H) with different biofilm productions and origins was tested using the bath infection model of wounded zebrafish embryos, as previously described for S. pettenkoferi [27]. The viability of zebrafish embryos was monitored for 48 h and we observed that mortalities in injured larvae infected with S. pettenkoferi began at 18 hpi for all strains (Figure 4). Strains showed different virulence profiles: deaths increased up to 95% by 24 hpi for the NSP009P strain, while embryos infected by NSP003P, NSP0023P and NSP004H showed a mortality of around 50% at 36 h (Log-rank Mantel-Cox statistical test, p < 0.001). S. pettenkoferi NSP009P was two times more virulent than the other three strains with less than 5% embryo survival at 24 hpi. Interestingly, NSP009P displayed the same virulence profile as for the S. pettenkoferi SP165 isolate previously described [27].
vs. S. aureus Newman; significant difference was at ** p < 0.01. A Dunn test was used to growth rates between the S. pettenkoferi strains; significant difference was observed NSP004H and NSP003P at * p < 0.05.

Virulence Study of S. pettenkoferi Strains Using a Zebrafish Embryos Model
In vivo virulence of the four selected S. pettenkoferi strains (NSP003P, NS NSP023P and NSP004H) with different biofilm productions and origins was teste the bath infection model of wounded zebrafish embryos, as previously describe pettenkoferi [27]. The viability of zebrafish embryos was monitored for 48 h and served that mortalities in injured larvae infected with S. pettenkoferi began at 18 hp strains (Figure 4). Strains showed different virulence profiles: deaths increased up by 24 hpi for the NSP009P strain, while embryos infected by NSP003P, NSP002 NSP004H showed a mortality of around 50% at 36 h (Log-rank Mantel-Cox statisti p < 0.001). S. pettenkoferi NSP009P was two times more virulent than the other three with less than 5% embryo survival at 24 hpi. Interestingly, NSP009P displayed th virulence profile as for the S. pettenkoferi SP165 isolate previously described [27].

Genomic Analysis of Putative Virulence Factors of the S. pettenkoferi Strains
Due to the evidence that the S. pettenkoferi strains are virulent and presented two profiles, we investigated their virulome content. A total of 44 putative homologues were identified by genomics analyses (Table 4). Most detected genes were related to immune system evasion and invasiveness (e.g., capA, eno, katA, tuf, esXA and five genes encoding for the T7SSb apparatus), whereas most of the toxins-encoding genes were absent (e.g., hly, hlb, hlg, sea, sel, seg, tst-1, etA, etB, etD and lukDE).

Discussion
Chronic non-healing wounds are a major complication of diabetes mellitus, associated with severe outcomes such as high morbidity and health care costs [4,30]. Polymicrobial colonisation, biofilm formation, and collaboration between pathogenic and colonising bacteria are hypothesised to impair healing of DFUs, thus contributing to severe complications such as osteomyelitis and amputation [2,3]. . Sequences comparison of S. pettenkoferi genomes. The NSP009P strain was compared against SP165 (a virulent strain) and the NSP003P, NSP023P and NSP004H genomes. The inner light red ring represents the NSP009P genome; the middle red one shows the SP165 genome after BLASTn match; while the outer green, light blue and dark blue rings correspond to NSP023P, NSP003P and NSP004H, respectively. Missing regions identified by the BLAST analysis on the CGView server's PROKSEE software are shown as 'gaps' on each of the circular genomes. Arrows represent regions shared between SP165 and NSP009P but not present in the NSP023P, NSP003P and NSP004H genomes. The inner ring corresponds to the GC skew of S. pettenkoferi NSP009P isolate. Light grey profile indicates overabundance of GC nucleotides, whereas dark grey shows the opposite. The inner black ring represents the GC%.

Discussion
Chronic non-healing wounds are a major complication of diabetes mellitus, associated with severe outcomes such as high morbidity and health care costs [4,30]. Polymicrobial colonisation, biofilm formation, and collaboration between pathogenic and colonising bacteria are hypothesised to impair healing of DFUs, thus contributing to severe complications such as osteomyelitis and amputation [2,3].
The microbial community present on foot ulcers in patients living with diabetes is highly complex and diverse, with CoNS frequently isolated from DFOM specimens [8,9]. Whereas CoNS have long been considered as colonising bacteria belonging to cutaneous commensal microbiota, some species such as S. lugdunensis have been identified as true pathogens [14]. However, the role of each CoNS species in infected DFUs remains unclear to date [14,31]. Among them, S. pettenkoferi was reported to cause severe infections [13,[16][17][18][19][20][21][22][23][24], predominantly in immunocompromised patients. Recently, we determined the pathogenicity of one S. pettenkoferi clinical isolate from DFOM, and demonstrated its in vivo virulence associated with a large production of biofilm [27]. Here, we confirm that S. pettenkoferi is a pathogen by studying a collection of 29 S. pettenkoferi clinical isolates from blood cultures and infected DFU.
Biofilm formation is an important pathophysiological hallmark in DFU, involving most of the bacteria present in chronic wounds, and inducing prolonged inflammation and chronicity of the lesions, as well as the development of antibiotic resistance resulting in difficulties treating these wounds [32]. It has been established that CoNS participate in biofilm formation that could contribute to the chronicity of DFU. However, biofilm-related studies tend to only focus on S. epidermidis, as this species is the most frequently isolated [33]. Few studies have evaluated the biofilm formation of S. pettenkoferi and, surprisingly, the previous investigations on clinical, animal or environmental isolates reported no production of biofilm [34,35]. We corroborated that the reference strain CIP 107711 was unable to form biofilm as previously published [35]. However, our results showed that clinical S. pettenkoferi strains were able to form biofilm, as previously published for one isolate, SP165 [27]. We could classify the strains in three different groups with distinct biofilm behaviours: a large majority of strains (n = 17 and the SP165 isolate) corresponding to fast and high producers, a second group of strains (n = 9) with an intermediate profile, and three strains with a low potential for biofilm formation. This classification confirmed that most clinical S. pettenkoferi strains (n = 26, 89.7%) were able to form biofilm. Notably, all strains isolated from blood cultures belonged to the faster biofilm producers group, which is consistent with previous studies showing that 52-100% of CoNS species isolated from bloodstream formed a biofilm [36][37][38].
To corroborate our results, S. pettenkoferi genome analysis identified genes known to participate in biofilm formation. All S. pettenkoferi genomes harboured the atlE gene encoding for the homologous autolytic enzyme AtlE, described in S. epidermidis [39]. This protein is a major contributor to primary attachment to abiotic surfaces and a critical enzyme required for the release of eDNA, a component of the staphylococcal in vitro biofilm matrix [40]. Moreover, the NSP012P and NSP023P strains possessed the gene encoding for the Bap homologue protein (Bhp), described in S. epidermidis [41]. In S. aureus, this protein has been identified as the main determinant for successful surface adherence, as well as intercellular accumulation during biofilm development [42,43]. In addition, the PNAG, a polysaccharide synthesized by products of the icaADBC locus, is a well-known mechanism of intercellular adhesion in staphylococcal biofilms. Some genes of this locus (the icaA, icaB and icaC homologue genes) were identified in all studied strains, although the icaD and icaR genes were absent. Interestingly, a similar analysis was reported by Dutta et al. (2018) in S. pettenkoferi isolated from a cat peritonitis case [34]. The icaADBC locus organisation may differ between staphylococci, as S. lugdunensis also does not contain the icaR gene [44]. Moreover, the ica operon expression can be activated by sarA and partially by the sigmaB and srrAB two-component systems [45], and such homologues of major staphylococcal biofilm-regulator-encoding genes were detected in the S. pettenkoferi genomes. This included positive regulators such as sigmaB (sigB), expressed from and regulated by the rsbUVW-sigB operon also present in S. pettenkoferi genomes; sarA, critical for biofilm formation mainly through repression of proteases; and mgrA and luxS, two negative regulators. However, the major regulatory system of staphylococcal biofilm formation is the accessory gene regulator (Agr), known to regulate the production of proteases and phenol-soluble modulins (PSMs), main contributors to biofilm maturation and disassembly in S. aureus and S. epidermidis [46,47]. Phenol-soluble modulin beta (PSMβ) protein and ClpP protease homologues were also annotated in all S. pettenkoferi genomes. PSMβ is a key effector in biofilm structuring and dispersion, both in vitro and in vivo [48]. A role in biofilm development has also been identified for the S. epidermidis protease ClpP and its expression is regulated by the agr system [49]. Finally, the pSSAP1 plasmid described in the S. saprophyticus strain MS1146 and containing the adhesin-encoding gene uafB was also detected in all the strains. The UafB adhesin is known to contribute to cell surface hydrophobicity and to mediate adhesion to fibronectin and fibrinogen [28]. Therefore, this adhesin could play a role in biofilm formation in S. pettenkoferi strains, as it is present in 50% of the biofilm producers' genomes and absent in the non-producer group.
In accordance with Ahmad-Mansour et al. [27], we confirmed that S. pettenkoferi strains showed relatively slow growth compared to S. aureus. One hypothesis predicts a positive association between the evolution of host immunity and pathogen virulence involving an increase in replication rates in response to host resistance [50]. Remarkably, a specific set of putative virulence factors (e.g., capA, katA and nucH) involved in immune evasion were identified in all S. pettenkoferi strains. These immune evasion determinants and the slow growth replication of S. pettenkoferi may be involved in its persistence during host infection.
Our study confirmed that S. pettenkoferi is a virulent pathogen in the zebrafish model of infection, which represents an important animal model to study host-pathogen interactions in vivo [51]. Zebrafish embryos possess a fully functional innate immune system and have been used for their ability to closely mimic human diseases [52]. Bacterial infections in zebrafish embryos are most frequently performed by microinjection. As S. pettenkoferi was isolated from chronic wounds, we used an adapted model where a bath infection model was chosen with larvae fin excision to mimic "sterile" wounding damage and inflammation [53]. Immersion following the larvae injury model was recently validated to study S. pettenkoferi virulence [27]. The studied strains with different origins and biofilm production ability showed virulence in this model. However, two clear virulence profiles were observed with one strain (NSP009P) being highly virulent, as previously published for the SP165 isolate [27], while the other three studied strains displayed a less virulent profile. However, no difference was observed in virulome content nor in biofilm behaviour. In addition, our genome phylogenetic analysis obtained by WGS data confirmed the presence of distinct clusters and subclusters, each comprising closely related S. pettenkoferi isolates from the two different clinical origins. A previous phylogenetic study between 25 S. pettenkoferi strains showed that clinical isolates were closely related to environmental isolates, irrespective of their geographic origin [35]. Interestingly, despite the single geographical origin of the strains analysed in this study, a similar repartition between their strains and our strains was observed except for clade II, absent in this previous study. This clade corresponds to a new group of S. pettenkoferi, genetically distant from the other strains, and will require phylogenetic investigations on a larger scale. More importantly, no links could be highlighted between biofilm production groups nor with strain origin and cluster membership. However, the two highly virulent strains SP165 and NSP009P are closely related in the same clade, I-a, with seven conserved regions, though more investigation is needed to determine if these regions could correspond to pathogenicity islands.

Bacterial Strains
All S. pettenkoferi strains belonged to the collection of the Department of Microbiology at Nîmes University Hospital (France). Twenty-two S. pettenkoferi strains were isolated from infected DFUs or DFOM and seven from blood cultures between July 2018 and June 2022. Bacterial identification was obtained by mass spectrometry (Vitek-MS ® , Biomérieux, Marcy-l'Étoile, France). Antimicrobial susceptibility testing was performed by the disc diffusion test according to the European Committee for Antimicrobial Susceptibility Testing (EUCAST) recommendations [54]. Vancomycin, teicoplanin and daptomycin MICs were determined using broth microdilution procedures (UMIC) (Bruker, Champs-sur-Marne, France). The S. pettenkoferi reference strain CIP 107711 (Culture Collection University of Gothenburg, Department of Clinical Bacteriology, University of Gothenburg, Sweden) was used as control in biofilm assays. The S. aureus reference strain Newman [29] was used as control in growth curves tests.

Whole-Genome Sequencing and Bioinformatics Analysis
S. pettenkoferi strains were cultivated aerobically at 37 • C for 48 h on Columbia sheep blood agar plates (5%) (Biomérieux). Following the manufacturer's instructions, genomic DNA was extracted using the DNeasy UltraClean Microbial Kit (Qiagen, Aarhus, Denmark). Whole-genome sequencing (WGS) was carried out using Illumina MiSeq sequencing equipment (Illumina, San Diego, CA, USA) using paired-end (PE) read libraries (PE250) made with an Illumina DNA Library Prep Kit (Illumina) according to the manufacturer's instructions. To examine data quality, raw readings were processed using FastQC (v.0.11.9). Sequencing reads were quality-trimmed and assembled de novo using CLC Genomics Workbench 22.0.2 (Qiagen) with default parameters to include only contigs > 1000 nt. Each S. pettenkoferi contig sequence was then annotated with Find Prokaryotic Genes, Annotate CDS with Best DIAMOND Hit and Annotate CDS with Pfam Domains functions within the Functional Analysis tool of the Microbial Genomics Module on the CLC Genomics Workbench with settings as previously described [55]. Virulence and biofilm genes were identified from each annotated S. pettenkoferi genome. ResFinder (2022-06-22), PointFinder for S. aureus (2019-08), CARD and the NCBI Bacterial Antimicrobial Resistance Reference Gene (2021-08) databases were also used to identify antimicrobial resistance genes [56][57][58]. Sequences obtained were aligned by using the multisequence alignment tool in CLC Genomics Workbench 22.0 with default parameters. Phylogenetic relationships were inferred from this alignment using a distance matrix generated under the assumptions of the Jukes and Cantor model [59]. The phylogenetic tree was constructed by using the neighbour-joining method implemented in the software [60]. WGS was subjected to other investigations, such as circular genome representation using the Proksee programme with CGView.js as its genome drawing engine [61]. Whole-genome bioinformatic data are summarised in Table S1. All the identified genomics sequences have been deposited on the GenBank website accession bioproject: PRJNA870410.

Kinetics of Early Biofilm Formation
To evaluate the capacity of S. pettenkoferi to form biofilm, we used the Biofilm Ring Test ® technique (Biofilm Control, Saint-Beauzire, France) following the manufacturer's recommendations [62]. This assay is based on the measurement of the mobility of superparamagnetic microbeads embedded in biofilms subjected to a magnetic field. Standardised bacterial cultures were deposited on 96-well microtiter plates (Falcon 96 Flat Bottom Transparent, Coming, USA) containing magnetic beads (TON004). The plates were incubated without shaking at 37 • C. After incubation, 100 µL of liquid contrast (LIC001) solution was added on the top of each well, and plates were inserted for 1 min onto a magnetic block and then into a reader (an Epson Scanner modified for microplate reading). Images of each well were acquired at 3 h, 4 h and 5 h. Data were analysed using the BFC Elements 3.0 software (Biofilm Control), giving a result in the form of a Biofilm Formation Index (BFI). Each experiment was performed in triplicate. A negative control was systematically included in each experiment corresponding to Brain Heart Infusion (BHI) medium and beads without bacterial suspension. BFI values were calculated for each well, ranging from 0 (total bead immobilisation, i.e., strongly adherent cells/strong biofilm formation) to 16 of 20 20 (no bead aggregation, i.e., non-adherent cells/no biofilm formation in the experiment conditions). A high BFI value (>7) corresponds to high mobility of the beads under magnetic action (no biofilm), while a low value (<2) corresponds to full immobilisation of the beads due to the sessile cells. Three experiments with two repeats were performed per strain and per incubation time.

Growth Curves Evaluation
Growth profiles were performed in Trypticase Soy Broth (TSB). Then, bacterial suspensions were calibrated to obtain an optical density of around 0.1 at 600 nm. Each bacterial suspension was added to a 96-well plate incubated at 37 • C for 48 h. The absorbance at 600 nm of each well was determined by the Infinite M Mano automatic absorbance reader (Tecan, Männedorf, Switzerland). Each experiment was performed in triplicate. The nonlinear regression model of Gompertz [63] was used with GraphPadPrism 9.2 software (San Diego, CA, USA) to acquire the equations of each growth curve, as well as various notable points (Ym, Y0, K and 1/K). The Gompertz equation can be written as follows: where Y(t) corresponds to the absorbance at a time t, Ym corresponds to the maximum absorbance (stationary phase), Y0 corresponds to the absorbance at t = 0 h, K determines the lag time (1/h), and 1/K (h) is the inflection point of the exponential phase. Inflection points enabled us to determine the beginning of the exponential phase and to calculate the slope of the linear trend line of this phase.

Infection of Danio Rerio Embryos
Following the previous protocol [27], S. pettenkoferi strains were cultured in TSB overnight at 37 • C. The cultures were diluted 1:20 and allowed to grow in TSB medium until they reached the mid-exponential growth phase (OD 600 = 0.7-0.9). Then, the bacteria were centrifuged for 10 min at 4000 rpm, and the pellets were resuspended in fish water with 60 µg/mL of sea salt (Instant Ocean) in distilled water with 4 × 10 −4 N sodium hydroxide at a concentration of 8 × 10 7 bacteria/mL. After diluting the inoculum in phosphatebuffered saline (PBS), the samples were plated onto TSB agar plates to determine the number of bacteria present. Experiments were conducted with the GAB zebrafish line in fish water at a temperature of 28 • C. The wounded embryos were placed on a Petri plate, given 0.28 mg/mL tricaine as anaesthesia, and a small tail fin injury was caused prior to infection using a 26-gauge needle under a stereomicroscope (Motic). After tail transection, groups of 24 damaged embryos were placed in Petri dishes with the appropriate bacterial solution (or fish water as a control), and then they were distributed individually into 96-well Falcon plates containing 200 µL of the bacterial suspension (or fish water). Throughout the incubation procedure, the plates were maintained at a constant temperature of 28 • C (bacteria were kept throughout the experiment in fish water, which does not support S. pettenkoferi growth). The number of dead embryos could be counted visually based on the absence of a heartbeat.

Ethical Statement
The animal experiments were carried out at the University of Montpellier in accordance with European Union recommendations for the care and use of laboratory animals (https://ec.europa.eu/environment/chemicals/lab_animals/index_en.htm (accessed on 15 March 2021)) and were authorised by the Direction Sanitaire et Vétérinaire de l'Hérault and Comité d'Ethique pour l'Expérimentation Animale under reference CEEALR-B4-172-37 and APAFIS#5737-2016061511212601 v3. Adult zebrafish were not killed for this study, and breeding of adult fish followed the international norms set out by the EU Animal Protection Directive 2010/63/EU. According to the EU Animal Protection Directive 2010/63/EU, all studies were conducted prior to the embryos' free feeding stage and did not constitute animal experimentation. The cardiac rhythm was used as a clinical criterion for survival graphs. At the end of the survival monitoring, the plates were parafilmed and quickly frozen, stored at −20 • C for 48 h to ensure embryos' death, and then autoclaved with the bacterially contaminated waste.

Statistical Analysis
Statistical analyses were performed using GraphPad Prism version 9.2 and R software version 4.1.0. Tests used for the p-value determination are mentioned in each table and figure legend. A p-value < 0.05 was considered to reflect a statistically significant difference.

Conclusions
This study highlights that S. pettenkoferi possesses an important virulence potential as demonstrated in vivo with the zebrafish model of infection, is able to form biofilm and contains a large panel of virulence determinants. Therefore, the genomic analysis of S. pettenkoferi constitutes a valuable resource to study molecular mechanisms involved in its pathogenicity.  Informed Consent Statement: Informed consent was obtained from all subjects involved in the study.
Data Availability Statement: Data supporting reported results can be found in the file "S. pettenkoferi project" contained in the computer of the VBIC unit.