Comparative genomic and functional analysis of Staphylococcus caprae clinical isolates: insights into genetic diversity and opportunistic pathogenic potential Authors: Yu-Hee Kim, Junghun Park, Ha Young Yun, Bomi Yoon, Minkyung Shin, Hae-Sun Chung, Jae Hyang Lim, Hee Jung Choi Published: 2026 ABSTRACT Staphylococcus caprae, a coagulase-negative staphylococcus (CoNS), is an opportunistic microorganism increasingly identified in clinical specimens. However, its genomic and functional characterization in human infection have not yet been fully elucidated. In this study, we characterized the genomic and phenotypic diversity of two clinical S. caprae isolates, M89 and S13, to elucidate their opportunistic traits. Whole-genome sequencing was performed to evaluate genomic differences in antimicrobial resistance, virulence determinants, and protein-coding sequences, while their pathogenic potential was assessed using a murine subcutaneous catheter model. Our results revealed that both isolates harbored core resistance genes, such as mecA and blaZ; however, the healthcare-associated strain S13 displayed a broader antimicrobial resistance profile. Although virulence profiling identified conserved factors shared by both strains, including icaABCR, atl, and clfB, a key genomic differentiator was the presence of the accessory Sec (aSec) secretion system in M89, which was absent in S13. Functionally, M89 exhibited significantly greater adherence to endothelial cells and higher colonization efficiency in vivo. Furthermore, M89 infection induced epidermal thickening and elevated proinflammatory cytokine expression in mouse skin tissue. These findings demonstrate significant intra-species heterogeneity in S. caprae regarding host interaction capabilities and underscore the importance of integrating genomic and phenotypic analyses to delineate the pathogenic mechanisms of this emerging opportunistic pathogen. Introduction Clinical isolates of staphylococci are commonly classified into coagulase-producing, highly virulent Staphylococcus aureus and coagulase-negative staphylococci (CoNS), which are generally considered commensals with low pathogenic potential. However, CoNS have increasingly emerged as major opportunistic pathogens in nosocomial setting, particularly in immunocompromised patients and those with indwelling medical devices (Becker et al. 2014). Staphylococcus caprae (S. caprae) is a CoNS species originally associated with animals that commonly invades the skin and mammary glands of goats and can cause mastitis. Despite its initial classification as an animal-associated bacterium, S. caprae has been identified in various human infections, including surgical site infections in patients with orthopedic prostheses and device-related infections such as those involving central venous or urinary catheters (Kanda et al. 1991; Vandenesch et al. 1995; Shuttleworth et al. 1997; Sulaiman et al. 2024). The increasing number of human cases unlinked to livestock exposure, supported by advances in molecular diagnostic tools, suggests that S. caprae may play a more substantial role in healthcare-associated infections than previously recognized (Diez de Los Rios et al. 2025). These observations underscore the need for continued surveillance and molecular characterization of S. caprae clinical isolates to better understand their epidemiology and pathogenic potential. Whole genome sequencing (WGS) has been applied to several CoNS species, leading to the identification of putative virulence factors and antibiotic resistance gene. Watanabe et al. conducted WGS and comparative genomics of methicillin-resistant S. caprae with other CoNS species and reported the presence of virulence factors associated with biofilm formation that are also found in S. epidermidis and S. capitis (Watanabe et al. 2018). In a more comprehensive study, Sun et al. performed genomic and phenotypic characterization of strains within the Epidermidis Cluster Group, further revealed interspecies similarities and differences in virulence-associated gene content (Becker et al. 2014; Sun et al. 2020). Notably, S. caprae strains carry specific virulence genes associated with intercellular adhesion, such as the ica locus, and type VII secretion system components (esaA, essABC). These genetic features have been widely reported in S. aureus (Cramton et al. 1999; Warne et al. 2016) and other CoNS (Chokr et al. 2006). The presence of these conserved machineries in S. caprae may provide significant advantages in immune evasion and persistence within the host environment. While recent genomic studies have provided insights into virulence-associated features of S. caprae, strain-specific genomic features that may contribute to pathogenicity remain poorly characterized. In this study, we performed whole-genome sequencing and functional characterization of two S. caprae clinical isolates, M89 and S13, which were recovered from bloodstream infections under different clinical contexts. Through comparative genomic analysis, virulence gene profiling, and both in vitro and in vivo assays, we investigated strain-specific characteristics potentially associated with adhesion, colonization, and pathogenicity. Materials and methods Collection of bacterial isolates and clinical data S. caprae clinical isolates were collected from February to September 2023 at the clinical microbiology laboratories of Ewha Womans University Mokdong Hospital and Ewha Womans University Seoul Hospital. This study was approved by the Institutional Review Board of the Ewha Womans University Mokdong Hospital (IRB File No. EUMC 2022-11-038-001) and Ewha Womans University Seoul Hospital (IRB File No. SEUMC 2023-01-015). Strain M89 was isolated from a single peripheral blood culture set of an 80-year-old woman admitted with COVID-19. Strain S13 was isolated from a 77-year-old woman with diabetic nephropathy. The patient experienced cardiac arrest on hospital day 13 and was transferred to the intensive care unit, where a femoral central venous catheter was inserted. S. caprae was isolated from catheter-drawn blood cultures eight days after catheter placement. Antimicrobial susceptibility test Antimicrobial susceptibility test was performed using the VITEK2 system with a 601 panel (bioMérieux, Marcy-l’Étoile, France). The results were interpreted according to the Clinical and Laboratory Standards Institute (CLSI) M100 guidelines for Coagulase-Negative Staphylococci (CoNS). Whole genome sequencing Genomic DNA was isolated using the NucleoSpin Microbial DNA mini kit (Macherey-Nagel, Nordrhein-Westfalen, Germany) and DNA concentrations were measured with the DropSense 96 spectrophotometer (Trinean, Gentbrugge, Belgium). Fragmentation of 500 ng of gDNA was performed using the Covaris S220 ultrasonicator (Woburn, MA, USA) under the following parameters: 175 W peak incident power, 5% duty factor, 200 cycles per burst, and 35 s of treatment time. Sequencing libraries were prepared with the TruSeq Nano DNA sample prep kit (Illumina, San Diego, CA, USA) in accordance with the manufacturer’s protocol. Library quality and fragment sizes were evaluated using an Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA), and quantification was performed by qPCR using the CFX96 Real-Time System (Bio-Rad, Hercules, CA, USA). Libraries were normalized and sequenced on the Illumina MiSeq platform using a 600-cycle MiSeq Reagent Kit v3, generating 2 × 300 bp paired-end reads. The raw FASTQ files were uploaded to the TrueBac ID system (version 1.92; CJ Bioscience, Seoul, South Korea), a whole-genome-based bacterial identification platform supported by a curated reference genome database and a specialized bioinformatics pipeline (Ha et al. 2019). Species-level identification was performed based on average nucleotide identity (ANI ≥ 95%) and alignment coverage thresholds (≥ 70%) against the TrueBac reference database. Genomic assemblies were generated using SPAdes 3.13.0 (Bankevich et al. 2012) and protein-coding sequences (CDSs) were predicted using Prodigal 2.6.2 (Hyatt et al. 2010). Phylogenetic analysis The phylogenetic tree was constructed using the Type Strain Genome Server (TYGS, http://tygs.dsmz.de) (Meier-Kolthoff and Goker 2019; Freese et al. 2026). In addition to M89 and S13, the whole-genome sequences of four previously reported S. caprae strains isolated from human clinical specimens (Watanabe et al. 2018; Sun et al. 2020) were retrieved from the NCBI database and included in the analysis. Resistance and virulence gene prediction The assembled genome sequences were analyzed to predict antibiotic resistance genes and virulence factors. Resfinder (https://genepi.dk/resfinder) was used to identify acquired antimicrobial resistance genes, while virulence factors were predicted using VFanalyzer, available through the Virulence Factor Database (VFDB, https://www.mgc.ac.cn/cgi-bin/VFs/v5/main.cgi). All analyses were performed using default parameters. Visualization of genome comparisons To visualize the genotypic differences between the S. caprae reference strain NCTC 12196 (GenBank accession: GCA_900458385.1) and the two clinical isolates (M89 and S13), the BLAST Ring Image Generator (BRIG, http://sourceforge.net/projects/brig/) was used (Alikhan et al. 2011). The reference genome was set as the central ring, with the clinical genomes aligned in outer rings for comparison. GC content and GC skew were visualized in the innermost rings to illustrate compositional variation across the reference genome. All analyses were performed using default parameters. Cell wall protein fractionation and western blot Overnight cultures of M89 and S13 strains were centrifuged at 12,000 × g for 5 min to pellet the cells. The cell pellets were washed three times with PBS and lysed in 6 M urea (Thermo Fisher Scientific) supplemented with a protease inhibitor cocktail (Roche, Basel, Switzerland). The suspension was incubated at room temperature (RT) for 20 min on an orbital shaker. Following lysis, the samples were centrifuged at 16,000 × g for 10 min at 4 °C. The resulting supernatants, representing the cell wall-associated protein fraction, were collected. Protein concentrations were determined using a BCA Protein Assay kit (Thermo Fisher Scientific), and equal amounts of protein (25 µg) were resolved by SDS-PAGE and transferred onto a PVDF membrane (EMD Millipore, Darmstadt, Germany). Membranes were blocked in TBS-T containing 1% BSA for 1 h at RT. For glycan detection, membranes were incubated overnight at 4 °C with biotin-conjugated succinylated wheat germ agglutinin (biotin-sWGA; Vector Laboratories, Newark, CA, USA) diluted in TBS-T containing 1% BSA. After washing with TBS-T, membranes were incubated with streptavidin-HRP (Abcam, Cambridge, UK) for 30 min at RT. Membranes were washed again, developed using ECL substrate (Nippon Genetics Europe, Düren, Germany), and visualized with a ChemiDoc imaging system (Bio-Rad, Hercules, CA, USA). Microplate biofilm formation assay In vitro biofilm-forming ability was assessed using a microplate biofilm formation assay as previously described (O’Toole 2011). Staphylococcus epidermidis strains RP62A and M7 were used as positive and negative controls, respectively. Bacteria were grown to the mid-logarithmic phase at 37℃ in Muller Hinton II Broth (MHB; BD Biosciences, Franklin Lakes, NJ, USA) and diluted 1:100 in fresh MHB to achieve a normalized initial OD600 of approximately 0.05. Then, 200 µl of each bacterial suspension was inoculated in quadruplicate into sterile 96-well tissue culture microplate and centrifuged at 1,000 rpm for 2 min. The plates were incubated for 2 days at 37℃, and the absorbance was measured at 600 nm. Planktonic cells were removed, and the wells were washed twice with PBS (Welgene, Gyeongsan, South Korea). The bacterial biofilms were fixed with 100 µl of methanol for 5 min, followed by staining with 100 µl of 0.1% crystal violet for 10 min at RT. The wells were then washed 3–4 times with DPBS and air-dried for 1 h at RT. Subsequently, 200 µl of absolute ethanol was added to each well and incubated at 4℃ for 30 min. The absorbance was measured at 540 nm. The extent of biofilm formation was calculated using the following formula (Poursina et al. 2018): Specific Biofilm formation (SBF) Index = (AB – CW)/G, where AB is the OD540 nm of the stained bacteria, CW is the OD540 nm of the control wells, and G is the OD600 nm of planktonic bacteria in the medium. Endothelial cell attachment assay The human endothelial cell line EA.hy926 (CRL-2922, ATCC, Manassas, VA, USA) was cultivated in DMEM (Welgene) supplemented with 10% FBS (Welgene) and 1% antibiotics-antimycotics (Welgene). For the endothelial cell adherence assay, cells were seeded in 12-well tissue culture plates (SPL) and cultured for 2 days until reaching confluence. The confluent monolayers were washed with PBS and each bacterial strain was added at a multiplicity of infection (MOI) of 10 in antibiotics-antimycotics free medium. The plates were incubated for 30 min at 37℃. Non-adherent bacteria were removed by washing with PBS, and the adherent bacteria was quantified by lysing the host cells with sterile distilled water, followed by serial dilution and plating onto nutrient agar plates. Bacterial adherence to endothelial cells was expressed as the percentage of the initial inoculum that remained attached to the endothelial cell monolayers (Li et al. 2019). in vivo subcutaneous catheter colonization model A subcutaneous catheter graft model was established using 10-week-old male C57BL/6 mice (Orient Bio, Sungnam, South Korea). Mice were anesthetized with intraperitoneal (IP) injection of Alfaxan (40 mg/kg) and xylazine (10 mg/kg). The dorsal region was shaved, followed by hair removal with a depilatory cream and disinfection with povidone-iodine and an alcohol swab. A small incision was made to create a subcutaneous pocket, into which a 1 cm segment of a 22-gauge intravenous (IV) catheter pre-filled with bacterial suspension (1ⅹ106 CFU) was implanted. Mice were assigned to a control group (n = 3) or infection groups (n = 4 per strain). Mice were sacrificed 3 days post-implantation by IP injection of avertin (500 mg/kg), and the catheters were aseptically harvested. Each catheter was placed in a 5 ml round-bottom tube containing 2 ml of PBS, followed by sonication in an ultrasonic bath (60 Hz/450 W, BSFS-10 S) and vigorous vortexing. The resulting bacterial suspensions were serially diluted and plated onto nutrient agar plates to quantify CFUs. Skin specimens were collected from the catheter insertion sites using 8 mm biopsy punch (BP-80 F, KAI Industries, Gifu, Japan) for further histological and PCR analyses. All animal experiments were approved by the Animal Ethics Committee of Ewha Womans University College of Medicine (approval number MEDIACUC23-020-t) and performed in accordance with relevant institutional guidelines and regulations. Scanning electron microscopy (SEM) Catheter samples were initially fixed in 2.5% glutaraldehyde prepared in 0.1 M phosphate buffer (pH 7.4), followed by three washes with PBS for 10 min each. Post-fixation was performed in 1% osmium tetroxide prepared in 0.1 M phosphate buffer (pH 7.4). The samples were then washed with the same buffer, dehydrated through a graded ethanol series (60%, 70%, 80%, 90%, 95%, and 100%), and substituted with isoamyl acetate. Critical point drying was carried out using a CPD300 (Leica, Vienna, Austria). The dried specimens were mounted on stubs with carbon tape, sputter-coated with gold using a Q150RS (Quorum, Laughton, UK), and examined using a field-emission scanning electron microscope (Sigma 300, Carl Zeiss, Oberkochen, Germany). All electron microscopy procedures were performed at the Nano-Imaging Laboratory, Institute of Medical Science, Ewha Womans University. Histological analysis Skin biopsy samples were fixed overnight in 10% neutral-buffered formalin and paraffin-embedded. Tissue sections (4 μm thickness) were stained with hematoxylin and eosin (H&E), and images were captured using a slide scanner (VS200, Olympus, Tokyo, Japan). Epidermal thickness was measured in 12 randomly selected fields per sample. Real-time PCR analysis Skin specimens were finely minced using a razor blade, and total RNA was extracted using the Tri RNA reagent (Farvogen) according to the manufacturer’s instructions. Complementary DNA (cDNA) was synthesized from total RNA using the AccuPower RocketScript Cycle RT Premix (Bioneer, Daejeon, South Korea). The relative mRNA expression levels were determined by quantitative real-time PCR using the Kapa SYBR FAST qPCR Mater Mix (KAPA Biosystems, Wilmington, MA, USA) on a StepOnePlus Real-Time PCR System (Applied Biosystems, Foster City, CA, USA). Primers were designed using the Primer3 web tool (version 4.1.0), and the primer sequences are listed as follows: 5’-CAACCAACAAGTGATATTCTCCATG-3’ and 5’- GATCCACACTCTCCAGCTGCA-3’ for IL-1β; 5’-CATCTTCTCAAAATTCGAGTGACAA-3’ and 5’- TGGGAGTAGACAAGGTACAACCC-3’ for TNF-α; 5’-GCTCCAGAAGGCCCTCAGA-3’ and 5’-AGCTTTCCCTCCGCATTGA-3’ for IL-17; 5’-ACAGGAGAAGGGACGCCAT-3’ and 5’-GAAGCCCTACAGACGAGCTCA-3’ for IL-4; 5’-AGAGGGAAATCGTGCGTGAC-3’ and 5’-CAATAGTGATGACCTGGCCGT-3’ for GAPDH as a housekeeping gene. Amplification was performed for 40 cycles, consisting of a denaturation step at 95 °C for 15 s and an amplification and signal acquisition step at 60 °C for 1 min. Cycle threshold (Ct) values were obtained automatically, and the relative expression levels of the target genes were calculated suing the 2^(∆Ct) methods, where ∆Ct = Ct(GAPDH) – Ct(target gene). Statistical analysis Statistical analyses were conducted using GraphPad Prism (version 10.6.1; GraphPad Software, La Jolla, CA, USA). Welch’s t-test was used for comparisons between two groups, and one-way ANOVA followed by Tukey’s post-hoc test was used for comparisons among three or more groups. Two-way ANOVA was applied to analyze time-course changes in body weight. Data are presented as mean ± standard error of the mean (SEM) , and p-values < 0.05 were considered statistically significant. Results General features of S. caprae genomes In order to identify biological characteristics of the isolated strains, we performed whole-genome sequencing. The assembled genome of isolate M89 comprises 2,724,872 base pairs with a GC content of 33.4%, encoding 2,617 predicted protein-coding sequences (CDSs), 58 tRNAs, and 13 rRNAs. Of these, 1,943 genes CDSs were assigned to clusters of orthologous groups (COG) functional categories. The genome of isolate S13 was slightly smaller, spanning 2,650,044 base pairs with the same GC content, and contained 2,507 predicted CDSs, 59 tRNAs, and 10 rRNAs, of which 1,872 were classified into COG categories. Phylogenetic analysis using the TYGS web server was performed to determine the genomic placement of isolates M89 and S13 among known staphylococcal species. The analysis confirmed that both M89 and S13 clustered tightly within the S. caprae lineage, forming distinct species clade closely related to species such as S. capitis and S. epidermidis (Fig. 1). Furthermore, M89 and S13 segregated into two distinct sub-clusters alongside previously reported human clinical isolates (Zheng et al. 2015; Watanabe et al. 2018) demonstrating their genomic alignment with pathogenic S. caprae strains. Fig. 1Phylogenomic analysis of S. caprae clinical isolates M89 and S13. Whole-genome-based phylogenetic tree generated using the Type Strain Genome Server (TYGS) showing the relationships of S. caprae isolates M89 (GenBank accession: JBUBAQ000000000) and S13 (GenBank accession: JBUBAP000000000) alongside reference human clinical strains (GCA_000931485.1, GCA_003966625.1, GCA_003966605.1, and GCA_003966585.1) and the type strain NCTC 12196. Numbers at the node branches represent percentage bootstrap support values Phylogenomic analysis of S. caprae clinical isolates M89 and S13. Whole-genome-based phylogenetic tree generated using the Type Strain Genome Server (TYGS) showing the relationships of S. caprae isolates M89 (GenBank accession: JBUBAQ000000000) and S13 (GenBank accession: JBUBAP000000000) alongside reference human clinical strains (GCA_000931485.1, GCA_003966625.1, GCA_003966605.1, and GCA_003966585.1) and the type strain NCTC 12196. Numbers at the node branches represent percentage bootstrap support values Identification of antimicrobial resistance genes and their phenotypic correlation First, antimicrobial resistance genes were identified at the genomic level to compare the differences in antimicrobial resistance between the two isolates. A total of four antibiotic resistance genes associated with three classes of antibiotics were identified in the S. caprae M89 genome. These included the aminoglycosides resistance gene aacA-aphD, the β-lactam resistance genes blaZ and mecA, and the fusB gene conferring resistance to fusidic acid, a steroid antibacterial agent. In the S. caprae S13 genome, in addition to aacA-aphD, blaZ and mecA, four additional resistance genes were identified. These included ant9, which confers resistance to the aminoglycosides; fosD, involved in fosfomycin resistance; ermA, associated with macrolide-lincosamide-streptogramin B resistance; and mupA, which mediates resistance to mupirocin. To determine whether these genomic features manifested functionally, we evaluated the phenotypic antimicrobial susceptibility profiles of the two strains (Table 1). The presence of blaZ and mecA in both strains aligned with their phenotypic resistance to penicillin and oxacillin. Regarding aminoglycoside resistance, the shared aacA-aphD gene correlated with reduced susceptibility to gentamicin. In addition, the ermA gene, identified only in S13, is associated with phenotypic resistance to erythromycin and clindamycin. This broader resistance profile observed in S13 may be a consequence of prolonged exposure to antibiotics commonly used in hospital settings, which resulted in the survival of strains carrying multiple resistance genes. Table 1 Antibiotics susceptibility profiles of S. caprae M89 and S13AntibioticsM89S13Ciprofloxacin≤ 0.5 (S)≥ 8.0 (R)Clindamycin≤ 0.12 (S)≤ 0.25 (R), inducibleErythromycin≤ 0.25 (S)≥ 8.0 (R)Gentamicin8 (I)≥ 16.0 (R)Linezolid1 (S)2 (S)Oxacillin MIC2 (R)≥ 4.0 (R)Penicillin G≥ 0.5 (R)≥ 0.5 (R)Rifampin≤ 0.03 (S)≤ 0.5 (S)Teicoplanin1 (S)≤ 0.5 (S)Tetracyclin≤ 1.0 (S)≤ 1.0 (S)Trimethoprim/sulfamethoxazole≤ 10.0 (S)≤ 10.0 (S)Vancomycin1 (S)1 (S)Values represent the Minimum Inhibitory Concentrations (MICs); Susceptibility categories (S, Susceptible; I, Intermediate; R, Resistant) were interpreted based on CLSI M100 guidelines Table 1 Antibiotics susceptibility profiles of S. caprae M89 and S13AntibioticsM89S13Ciprofloxacin≤ 0.5 (S)≥ 8.0 (R)Clindamycin≤ 0.12 (S)≤ 0.25 (R), inducibleErythromycin≤ 0.25 (S)≥ 8.0 (R)Gentamicin8 (I)≥ 16.0 (R)Linezolid1 (S)2 (S)Oxacillin MIC2 (R)≥ 4.0 (R)Penicillin G≥ 0.5 (R)≥ 0.5 (R)Rifampin≤ 0.03 (S)≤ 0.5 (S)Teicoplanin1 (S)≤ 0.5 (S)Tetracyclin≤ 1.0 (S)≤ 1.0 (S)Trimethoprim/sulfamethoxazole≤ 10.0 (S)≤ 10.0 (S)Vancomycin1 (S)1 (S)Values represent the Minimum Inhibitory Concentrations (MICs); Susceptibility categories (S, Susceptible; I, Intermediate; R, Resistant) were interpreted based on CLSI M100 guidelines Antibiotics susceptibility profiles of S. caprae M89 and S13 Trimethoprim/ sulfamethoxazole Values represent the Minimum Inhibitory Concentrations (MICs); Susceptibility categories (S, Susceptible; I, Intermediate; R, Resistant) were interpreted based on CLSI M100 guidelines Virulence genes in S. caprae M89 and S13 To identify the pathogenic potential of the isolated strains, we performed bacterial virulence factor prediction using the VFDB (Table 2). Adherence-related genes—including icaABCR (intercellular adhesion), ebh (fibronectin binding protein), ebp (elastin binding protein), atl (autolysin), and clfB (clumping factor B)—were present in both isolates. Both also encoded components of the type VII secretion system (esaAB, essABC, and esxA), as well as toxin-related genes (hlb, β-hemolysin; cylR2, cytolysin regulator), enzymes (geh, lip, nuc), and the capsule-associated gene capB. Strain S13 carried additional virulence determinants absent from M89. These include serine V8 protease gene sspA, involved in extracellular matrix degradation, and capC, another capsule-related gene implicated in immune evasion. These findings suggest that while the two strains share a core set of virulence factors, the presence of strain-specific genes in S13 may reflect adaptation to different clinical environments. Table 2TPredicted virulence factors present in S. caprae M89 and S13CategoryVirulence factorsGeneM89S13Adherence & BiofilmAutolysinatlorf01470orf00812Fibronectin binding proteinebhorf00734orf00358Clumping factor BclfBorf00363orf01761Elastin binding proteinebporf00697orf00321Intercellular adhesinicaAorf01490orf02239icaBorf01492orf02237icaCorf01493orf02236icaRorf01489orf02240Invasion & ExoenzymesLipasegehorf01798orf01932liporf01494; orf01542orf02190; orf02235Serine V8 proteasesspA–orf01793Thermonucleasenucorf00852orf00476Beta hemolysinhlborf01854orf02004Immune evasionCapsule related genes–orf00333; orf00334; orf00345; orf00346; orf00347; orf00348; orf00338orf01731; orf01732; orf01743; orf01744; orf01745; orf01746; orf01736Polyglutamic acid capsulecapBorf00197orf01593capCorf00196–Other capsule-relatedugeorf00339orf01737CytolysincylR2orf00024orf01418LPSwbtPorf00336orf01734Secretion systemType VII secretion systemesaAorf01113orf00936esaBorf01111orf00934essAorf01112orf00935essBorf01110orf00933essCorf01109orf00932esxAorf01114orf00937 Table 2TPredicted virulence factors present in S. caprae M89 and S13CategoryVirulence factorsGeneM89S13Adherence & BiofilmAutolysinatlorf01470orf00812Fibronectin binding proteinebhorf00734orf00358Clumping factor BclfBorf00363orf01761Elastin binding proteinebporf00697orf00321Intercellular adhesinicaAorf01490orf02239icaBorf01492orf02237icaCorf01493orf02236icaRorf01489orf02240Invasion & ExoenzymesLipasegehorf01798orf01932liporf01494; orf01542orf02190; orf02235Serine V8 proteasesspA–orf01793Thermonucleasenucorf00852orf00476Beta hemolysinhlborf01854orf02004Immune evasionCapsule related genes–orf00333; orf00334; orf00345; orf00346; orf00347; orf00348; orf00338orf01731; orf01732; orf01743; orf01744; orf01745; orf01746; orf01736Polyglutamic acid capsulecapBorf00197orf01593capCorf00196–Other capsule-relatedugeorf00339orf01737CytolysincylR2orf00024orf01418LPSwbtPorf00336orf01734Secretion systemType VII secretion systemesaAorf01113orf00936esaBorf01111orf00934essAorf01112orf00935essBorf01110orf00933essCorf01109orf00932esxAorf01114orf00937 TPredicted virulence factors present in S. caprae M89 and S13 Loss of the aSec system distinguishes S. caprae S13 from M89 and the reference strain To compare the overall genomic similarities and differences of the isolated strains with a reference genome, the genomes of M89 and S13 were aligned to the reference genome NCTC 12196 using the BLAST Ring Image Generator (BRIG) (Fig. 2A). Both clinical isolates exhibited high overall sequence similarity to the reference genome. Notably, at around 210 kbp, S13 lack a region that is present in the reference strain and M89. The missing region spans approximately 12 kb and this encodes components of the accessory Sec (aSec) system, a specialized transport system which mediates the export of heavily glycosylated cell wall-anchored proteins, known as serine-rich adhesin for platelets (SraP) (Braunstein et al. 2019). Genetic organization of the aSec system identified in M89 is shown in Fig. 2B. The gene cluster contains eight coding DNA sequences (CDSs), all arranged in reverse orientation, that include core components of the aSec system: SecA2/SecY2, Asp1/2/3, the related glycosyltranferases Gtf2 and tagE, and the putative platelet binding protein GspB. Although S13 lacks the aSec system, it retains a CDS encoding SraP, suggesting that the absence of the secretion machinery may impair proper surface localization of the glycosylated proteins. To examine this, we extracted cell wall proteins from M89 and S13 and analyzed for carbohydrate contents using a glycan detection assay. Results showed that M89 exhibited a distinct glycoprotein band in the high molecular weight region, whereas S13 did not show this specific band but displayed a smear-like signals (Fig. 2C). In addition, different band patterns were observed in different molecular weight ranges, such as a decrease in band around 250 kDa in M89 and around 90 kDa in S13. This demonstrates that there are differences in the glycosylation and localization of cell wall proteins among the strains. Fig. 2Comparative genomic analysis and loss of aSec system in S. caprae S13. A Whole-genome alignment of M89 and S13 against the reference strain S. caprae NCTC 12196 using BRIG. S13 lacks a 12 kb region (~ 210 kbp) corresponding to the accessory Sec (aSec) secretion system present in M89 and the reference genome. B Genetic organization of the aSec locus in M89, comprising secA2/secY2, asp1-3, gtf2, tagE, and gspB. C Glycan detection assay of cell-wall proteins showing a high-molecular-weight glycoprotein band in M89 but not in S13 Comparative genomic analysis and loss of aSec system in S. caprae S13. A Whole-genome alignment of M89 and S13 against the reference strain S. caprae NCTC 12196 using BRIG. S13 lacks a 12 kb region (~ 210 kbp) corresponding to the accessory Sec (aSec) secretion system present in M89 and the reference genome. B Genetic organization of the aSec locus in M89, comprising secA2/secY2, asp1-3, gtf2, tagE, and gspB. C Glycan detection assay of cell-wall proteins showing a high-molecular-weight glycoprotein band in M89 but not in S13 Differential pathogenic potential between S. caprae M89 and S13 To evaluate the pathogenic potential of S. caprae strains M89 and S13, a series of functional assays were performed. In an in vitro biofilm formation assay, S13 exhibited a significantly greater capacity for biofilm formation than M89 (Fig. 3A). In contrast, in the ex vivo endothelial cell attachment assay, M89 showed significantly stronger adherence to host cells compared with S13 (Fig. 3B). To mimic catheter-related infection, an in vivo subcutaneous catheter colonization model was employed. Catheters preincubated with bacterial suspensions were implanted subcutaneously into mice, and colonization on the catheters was examined using scanning electron microscopy three days post-implantation. Results showed that the M89-infected catheters were covered with compact bacterial aggregates embedded in a dense extracellular matrix, whereas the S13-infected catheters showed bacterial cells distributed as small aggregates (Fig. 3C). Furthermore, bacteria collected from the catheters were cultured on nutrient agar plates to quantify viable bacteria. Consistently, M89-infected catheters showed a significantly higher bacterial load than those infected with S13 (Fig. 3D). A transient reduction in body weight was observed one day after catheter implantation, which recovered in the control group on the following day. However, mice implanted with catheters containing bacterial suspensions showed limited recovery, and specifically, the M89-infected group exhibited a significant decrease in body weight three days post-implantation compared with the control group (Fig. 3E). Inflammatory gene expression analysis in mouse skin revealed increased expression of TNF-α and IL-1β only in the M89-infected group, whereas expression levels of IL-4 and IL-17 were not altered (Fig. 3F). Furthermore, histopathological analysis was performed on mouse skin at the catheter implantation site. H&E-stained sections showed thickening of the stratum corneum and stratum spinosum, accompanied by parakeratosis, only in the M89-infected group (Fig. 3G). Consistently, measurement of epidermal thickness confirmed a significant increase in the M89-infected mice compared with the S13-infected group (Fig. 3H). Together, these findings demonstrate that M89 exhibits enhanced host tissue adherence and colonization in both ex vivo and in vivo models compared with strain S13. Fig. 3Differential pathogenic potential between M89 and S13. A Biofilm formation assay showing stronger biofilm formation by S13 than M89. B Endothelial cell attachment assay demonstrating greater adherence of M89 to host cells. C SEM images of subcutaneous catheters. M89 forms dense aggregates within an extracellular matrix, whereas S13 shows sparse clusters. D Quantification of viable bacteria recovered from catheters confirming higher colonization by M89. E Body weight changes showing delayed recovery in M89-infected mice. F Expression of inflammatory cytokines showing increased TNF-α and IL-1β levels only in M89-infected skin. G, H Histological analysis revealing epidermal thickening and parakeratosis in M89-infected tissue. Data are presented as mean ± SEM and were analyzed using Student’s t-tests or one-way ANOVA (*p < 0.05,**p < 0.01, ***p < 0.001). Control group (CON), n = 3; M89 and S13 groups, n = 4 per group Differential pathogenic potential between M89 and S13. A Biofilm formation assay showing stronger biofilm formation by S13 than M89. B Endothelial cell attachment assay demonstrating greater adherence of M89 to host cells. C SEM images of subcutaneous catheters. M89 forms dense aggregates within an extracellular matrix, whereas S13 shows sparse clusters. D Quantification of viable bacteria recovered from catheters confirming higher colonization by M89. E Body weight changes showing delayed recovery in M89-infected mice. F Expression of inflammatory cytokines showing increased TNF-α and IL-1β levels only in M89-infected skin. G, H Histological analysis revealing epidermal thickening and parakeratosis in M89-infected tissue. Data are presented as mean ± SEM and were analyzed using Student’s t-tests or one-way ANOVA (*p < 0.05,**p < 0.01, ***p < 0.001). Control group (CON), n = 3; M89 and S13 groups, n = 4 per group Discussion In this study, we performed a genomic and phenotypic characterization of two S. caprae clinical isolates. Both isolates shared core genomic features and an antibiotic resistance profile, including resistance to β-lactams mediated by the presence of blaZ and mecA. However, the S13 strain, isolated from a healthcare-associated setting, exhibited a broader resistance spectrum, including resistance to macrolides, lincosamides, and aminoglycosides. Comparative genomic analysis revealed that M89, but not S13, harbored an aSec system, which facilitate the secretion of surface-anchored glycoprotein. Consistent with this, M89 demonstrated enhanced adherence to endothelial cells and greater colonization of implanted catheters in vivo compared to S13. Originally regarded as an animal-associated CoNS species, S. caprae has increasingly been implicated in human infections and the clinical cases reported previously show that S. caprae infections can be diverse. For instance, some reports have described bone and joint infections caused by S. caprae in the absence of orthopedic devices, such as a case of vertebral osteomyelitis in an immunocompromised patient (Gowda et al. 2018) or subacute osteomyelitis in a healthy adolescent following minor trauma (Vazquez et al. 2023). This clinical adaptability of S. caprae in human hosts is further supported by comparative genomic analyses of human clinical isolates. For example, Watanabe et al. characterized three methicillin-resistant S. caprae strains isolated from distinct clinical scenarios in a hospital setting: JMUB145 from a febrile patient undergoing chemotherapy, JMUB898 from a vascular graft infection, and JMUB590 from a nasal screening swab (Watanabe et al. 2018). Beyond antibiotic resistance, these human specimens possessed shared biofilm-forming machinery similar to S. epidermidis and S. capitis. In our phylogenomic analysis, these strains were clustered with the hospital-acquired multidrug-resistant strain S13. In another instance, S. caprae strain was isolated from human cerebrospinal fluid (Zheng et al. 2015), and this invasive isolate was closely clustered with our bloodstream strain M89. This suggests a possibility that these strains share a core lineage specialized for invasive human infections. These cases, together with our data, suggest that the traditional distinction between commensal CoNS and primary pathogens like S. aureus is becoming increasingly blurred as CoNS species are frequently reported to harbor a diverse array of virulence and resistance determinants (Argemi et al. 2019; Grazul et al. 2023). Compared to previously reported S. caprae strains (Argemi et al. 2019), virulence gene profiling revealed that our isolates carry genes encoding capsule proteins, adhesins, and enzymes, whereas phenol-soluble modulins (PSMs) were not detected. PSMs are cytolytic and proinflammatory virulence factors in S. aureus and S. epidermidis (Peschel and Otto 2013), and their absence in M89 and S13 may reflect the relatively lower virulence potential of these clinical isolates. However, both strains harbored β-hemolysin (hlb) and the cytolysin regulator (cylR2), which were not reported in previous S. caprae studies. These genes may contribute to membrane disruption or tissue damage during infection (Coburn and Gilmore 2003; Singh et al. 2014; Jia et al. 2024) potentially compensating for the lack of PSM-mediated cytotoxicity. Further studies are required to validate this compensatory mechanism experimentally. At the genomic level, we hypothesize that the difference in pathogenicity between these two strains is related to the presence or absence of specific systems, such as the aSec system. This specialized secretion machinery is responsible for exporting glycosylated surface proteins such as SraP (Braunstein et al. 2019). Although SraP CDS is present in both strains, the absence of aSec in S13 likely limits proper glycosylation and surface localization. Supporting this, a high-molecular-weight glycoprotein in the cell wall fraction was not detected in S13, and functional assays showed its reduced adhesion and colonization capacities compared to M89. However, since only two strains were compared and gene deletion experiments were not performed, a causal relationship cannot be definitively proven. Nevertheless, based on this data, we can hypothesize that the aSec system may be involved in interactions with the host. Interestingly, the bloodstream isolate M89 showed higher ex vivo endothelial attachment and in vivo catheter colonization while the catheter isolate S13 had superior in vitro biofilm formation ability. This may seem counterintuitive, but it suggests that the clinical origin of an isolate does not always determine the phenotypic potential of a strain. In addition, S13 showed lower colonization ability but had a much broader antibiotic resistance profile. This indicates that S13 survives in clinical settings mainly by enduring antibiotic pressure and using non-specific binding rather than high virulence. This observation may reflect a potential trade-off between antibiotic resistance and virulence traits (Beceiro et al. 2013; Melnyk et al. 2015). It has been proposed that the acquisition of resistance determinants can impose a fitness cost, potentially attenuating bacterial adhesion, growth, or invasiveness under certain conditions. While the exact mechanisms remain to be clarified, the inverse association observed in our study may align with the resistance-virulence balance hypothesis shaping the pathogenicity of S. caprae in different clinical contexts. Despite providing novel insights into the strain-specific genomic and phenotypic features of S. caprae, this study has several limitations. First, the analysis was based on only two clinical isolates, M89 and S13, which may not fully capture the genetic and functional diversity of the species. In particular, the difficulty in obtaining and reliably distinguishing between ‘infectious’ and ‘commensal’ S. caprae strains remains a challenge in elucidating clear pathogenic characteristics. Therefore, generalizing our findings to S. caprae as a whole should be approached with caution. Second, although comparative genomics revealed the presence or absence of key virulence-associated genes such as the aSec system and capC, functional validation through genetic manipulation (e.g., gene knockout or complementation) was not performed. Therefore, the direct contribution of these genes to pathogenesis remains to be established. Lastly, while phenol-soluble modulins (PSMs)—major cytolytic toxins in S. aureus and S. epidermidis—were absent in both strains, the impact of this genetic deficiency on cytotoxic potential was not experimentally assessed. In conclusion, our integrative genomic and phenotypic analysis of two S. caprae clinical isolates highlights strain-specific differences in antimicrobial resistance and virulence, providing insights into the pathogenic potential of this emerging human pathogen. The presence of the aSec system in M89 may confer enhanced host interaction capabilities, while the broader resistance repertoire of S13 reflects adaptation to healthcare environments with high antibiotic pressure. These findings highlight the necessity for increased surveillance of S. caprae in clinical settings and warrant further investigations using larger strain collections and genetic tools to delineate the molecular mechanisms governing its pathogenicity. As S. caprae infection continues to be reported in hospital settings, a deeper understanding of its strain-specific characteristics will be important in designing diagnostic, therapeutic, and preventive strategies.