Genomic characterization of Staphylococcus epidermidis Se252 isolated from the rhizosphere of a Brazilian endemic plant

Background

Staphylococcus epidermidis (Se) is commonly regarded as a commensal organism; however, under specific conditions, it may act as an opportunistic pathogen. Here, we report the whole-genome sequencing and comparative genomic analysis of Se strain 252 (Se252), isolated from the rhizosphere of an endemic Brazilian plant.

Results

Se252 exhibits a unique repertoire of genes associated with environmental adaptation and virulence. These include two putative Type VII secretion system (T7SS) effectors and thirteen proteins involved in adhesion, toxin production, and immune evasion—among them, IsaB, which has not been previously reported in Se. Gene family expansions were observed in loci related to phenol-soluble modulins (PSMs), TLpps, LPXTG-motif proteins, nonribosomal peptide synthetases (NRPS), and siderophore biosynthesis (staphylopine, staphyloferrin), as well as quorum-sensing autoinducing peptides. In contrast, Se252 harbors relatively few antibiotic resistance genes.

Conclusions

The genomic profile of Se252 reflects adaptations to a plant-associated environment, yet harbors multiple features potentially enhancing human pathogenicity. These findings highlight the relevance of environmental Se lineages as possible reservoirs of virulence traits with implications for public health.

Graphical Abstract

Supplementary Information

The online version contains supplementary material available at 10.1186/s12864-025-12211-7.

Highlights

• Se252 is the first Staphylococcus epidermidis strain isolated from Brazilian endemic plants.

• The Se252 genome reveals both commensal and opportunistic pathogenic traits.

• New putative T7SS effectors may aid strain adaptation across different ecological niches.

• Se252’s antibiotic resistance genes are products of vertical inheritance.

• The IsaB-like protein was identified in S. epidermidis.

• Se252 shows PSM, TLpp, NRPS, staphyloferrin, and AIP genes.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12864-025-12211-7.

Background

The Brazilian Iron Quadrangle (IQ), located in the south-central region of the state of Minas Gerais, is known for its rich mineral resources, diverse landscapes (such as caves, mountains, and rocky outcrops), and high rates of plant and animal endemism [1]. This region is also considered a biodiversity hotspot [2]. The flora of the IQ has been extensively studied, particularly with regard to its adaptive mechanisms [3–5]. Among the studied plants is Mimosa calodendron Mart. ex Benth. (Fabaceae), a leguminous species endemic to the Rupestrian Fields of Minas Gerais. This plant is typically found in the IQ region, where it grows in association with cangas (ironstone outcrops) at altitudes between 1300 and 1750 m [6–8].

The IQ has been subject to intensive mineral extraction activities since the early 18th century, which continue to this day [9–12]. As a result, M. calodendron is classified as an endangered species. This plant plays a key role in nitrogen fixation in soil [13]. The microbiota associated with M. calodendron has only recently been studied [14]. Among the microbial species isolated, our group was the first to investigate the tolerance and metal-removal capacity of an Alcaligenes faecalis strain from its rhizosphere, demonstrating its potential for rhizoremediation in arsenic-contaminated soils [15]. The genome of this strain has been sequenced and analyzed, providing further insight into the adaptive capabilities of bacteria in metalliferous soils [16].

Another bacterial strain, Staphylococcus epidermidis (Se), was isolated from the same M. calodendron rhizosphere samples. S. epidermidis is predominantly associated with human epithelial tissue, where it is considered part of the natural microbiota. However, it can become pathogenic under certain conditions, particularly when it interacts with other tissues [17, 18]. The genetic repertoire of S. epidermidis enables its persistence in diverse hosts, including humans [19, 20]. Interestingly, S. epidermidis has also been identified in the microbiota of several plant species, including Cannabis sativa seeds [21], sugarcane stems [22], the rhizosphere and phyllosphere of potato plants [23], and rice seeds [24], through 16 S rDNA sequencing. Chaudhry and Patil identified four S. epidermidis species associated with plants of agricultural interest, showing that their genomes, while closely related to animal-associated strains, form a distinct phylogenetic cluster, reflecting their adaptation to plant-associated environments [25].

The occurrence of S. epidermidis in the rhizosphere highlights an underexplored ecological dimension of this species. Plant-associated microbiomes represent complex environments where microbes face nutrient limitation, metal stress, and competition, while also engaging in mutualistic or opportunistic interactions with plants [26]. The detection of S. epidermidis in these niches suggests that the rhizosphere may serve as an ecological reservoir facilitating bacterial persistence outside animal hosts [27]. Such environments can promote genetic diversification, stress tolerance, and the acquisition of traits relevant to host interaction, potentially bridging plant and human ecosystems. Moreover, environmental disturbances such as mining may amplify these dynamics, shaping microbial assemblages and favoring the persistence of opportunistic bacteria like S. epidermidis [28].

The objective of this study was to sequence and analyze the genome of S. epidermidis Se252, isolated from the rhizosphere of M. calodendron, to better understand its genetic adaptations to a non-animal host environment. The findings highlight the potential of S. epidermidis Se252’s gene repertoire to survive and interact with plant environments, shedding light on the possibility of new emerging pathogens arising from environmental degradation [29, 30]. This underscores the potential public health implications of plant-associated S. epidermidis populations in the future.

Methods

Plant sample collection

The Staphylococcus epidermidis (Se) strain 252 (Se252) was isolated from the rhizosphere of wild Mimosa calodendron Mart. (Fabaceae) [14, 16, 31]. The designation as strain 252 refers to its identification as the 252nd isolate within a bacterial collection obtained from endemic plants of the Iron Quadrangle during a preliminary study. Specifically for the isolation of microbiota associated with Mimosa calodendron, plant roots were exposed by excavation to a maximum depth of 5 cm using sterile spatulas [31]. Root fragments approximately 3 cm in length (~ 10 g of tissue) were cut with sterile scalpels, placed, and labeled in 50 ml Falcon tubes previously sterilized by autoclaving, to be processed within hours in the laboratory. All procedures were performed using sterile gloves, which were replaced with new sterile pairs whenever necessary. The bacterial collection was authorized by the Brazilian Ministry of the Environment (MMA) and the Chico Mendes Institute for Biodiversity Conservation (ICMBio), in accordance with the Biodiversity Authorization and Information System (SISBIO) field permit number 54,015. Under this permit, only plant fragments were allowed to be removed, which precluded the creation of a specimen voucher.

Bacterial isolation and conservation

In the laboratory, the roots were washed with distilled water to remove rhizosphere debris. The samples were then immersed in a 2.5% sodium hypochlorite solution for 2 min, followed by a 2-minute rinse in 70% ethanol. Afterward, the samples were rinsed with sterile distilled water and placed on Petri dishes containing Luria Bertani (LB) agar (10 g/L peptone, 10 g/L NaCl, and 5 g/L yeast extract, pH 7.0) supplemented with 0.03 mg/L methyl thiophanate. The plates were incubated at 28 °C for 3–4 days, after which individual colonies were picked using sterile toothpicks and re-plated on fresh LB agar plates. Following isolation, the bacterial cultures were grown in liquid LB medium and supplemented with 30% glycerol for long-term storage at −80 °C.

Bacterial growth

After isolation, Se252 was cultured in 250 mL of LB medium for 2 days at 28 °C with shaking at 220 rpm. Once an optical density of 0.8 was achieved, as measured spectrophotometrically at 600 nm, 0.8 mL aliquots were transferred into sterile 1.5 mL tubes for various assays. For all biological assays conducted in this study Se252 and Staphylococcus aureus ATCC 25,923 strains (obtained from Analítica Labor Belo Horizonte, MG, Brazil) was cultured in either LB medium or Mannitol Agar Salt (MAS) medium (10 g/L peptone, 1 g/L beef extract, 75 g/L NaCl, 10 g/L mannitol, 0.025 g/L phenol red, 15 g/L agar) and incubated at 28–37 °C.

The growth curve in LB medium was determined by measuring the optical density at 600 nm every 8 h for 72 h.

Bacterial DNA extraction

After isolation, Se252 was cultured reaching an optical density of 0.8 (600 nm). Then, 0.8 mL aliquots were taken into sterile 1.5 mL tubes for different assays, including DNA extraction. DNA extraction, sequencing, and genome assembly protocols were adapted from Felestrino et al. 2020 [16]. The DNA was extracted using the Wizard Genomic DNA Purification™ Kit (Promega, USA), following the manufacturer’s instructions. DNA integrity was assessed using a DNA 7500 chip™ via the 2100 Bioanalyzer™ (Agilent Technologies, USA), revealing a predominant presence of fragments >10 kb.

Genome sequencing and assembly

Following DNA extraction, the sequencing library was prepared using the Illumina Nextera DNA Library Preparation™ Kit (Illumina, Inc., USA) with a total input of 40 ng of DNA. Library quantification was performed using the KAPA Library Quantification™ Kit, and the library was sequenced on a MiSeq Reagent™ Kit v2 (500-cycle, paired-end reads - targeting 2 × 250 bp read lengths). Illumina read1 and read2 had an average of >80% and 75% base quality (Q30), respectively. Raw reads were processed using Trimmomatic v0.35 [32] for trimming, and assembly was carried out using SPAdes v3.12.0 [33] and MaSuRCA assembler v3.2.6 [34]. The final genome sequence was produced by comparing SPAdes and MaSuRCA assemblies using cross_match software (http://www.phrap.org), with additional scaffolding and gap-closing using Platanus v1.2.4 [35]. Mapped reads were analyzed using Bowtie2 v2.3.4.1 [36]. A total of 2,841,445 paired-end reads were aligned, yielding a high-quality draft genome with an average coverage of 150x.

Genome comparison

The Se252 genome was compared with other 35 S. epidermidis strains (12142587, APO27, ATCC 12228, CIM28, FRI909, M0026, MC28, NIHLM021, NIHLM031, NIHLM039, NIHLM070, Scl19, Scl31, SE4.6, SE4.7, SE4.8, UC7032, VCU129, SE2.9, SA6, SB7b, SB7c, SA8, s10, NCTC12100, SE90, SE95, UMB1227, RIT611, SESURV_p1_0612, UFMG-H7, Z0118SE0260, Z0118SE0269, 389, and PR246B0), and Staphylococcus aureus (NCTC8325) (Table S1) using OrthologSorter [37], a tool designed for genome comparison based on protein-coding gene content (http://jau.facom.ufms.br/sepidermidis/orthologsorter/). The strain selection was based on the main clades described by Chaudhry and Patil [25], including complete genomes when available.

Phylogenomic analysis

Phylogenomic analysis with Orthologsorter was based on conserved protein families shared across all genomes. Families with a single representative per genome were aligned and concatenated using MUSCLE [38], and non-informative sites were removed with GBlocks [39]. The final alignment was used to infer a maximum likelihood tree in RAxML [40] with the PROTCAT model and 1000 rapid bootstrap replicates.

To complement this analysis, genome similarity at DNA level was also assessed by using the genomic distance metric called MUMi (MUM index). MUMi [41] was calculated by identifying maximal unique matches (MUMs) with a minimal length between bacterial genomes via software package MUMmer3 [42], with default parameters. This genomic distance approach is a fast and reliable estimation of evolutionary divergence and correlates well with ANI based metrics [41].

Pan and core genome analysis

Pan and core genome analyses were performed using OrthologSorter [37]. The analysis was performed using a dedicated web-based tool specifically developed for this purpose (http://jau.facom.ufms.br/sepidermidis/orthologsorter/). Users may define the inclusion or exclusion of genomes for each analysis, thereby enabling the selection of either the core or pan-genome. The core genome was defined as the set of genes conserved across 100% of the analyzed genomes.

Flexible genome diversity analysis

For analysis of the flexible genome, singletons and core genome genes were excluded. Based on the genome selections defined using the Orthologsorter tool (see previous section), a binary matrix of gene presence/absence for the accessory genome was used to generate a heatmap, with hierarchical clustering based on Euclidean distance. The Sa8325 genome served as the reference for this analysis.

Other genomic attributes

The Blast Ring Image Generator (BRIG) [43] was used to visualize the genome on a circular map. Genomic islands were identified using the IslandViewer tool [44], and phage sequences were detected using the PHAST tool [45]. For all analyses, input data consisted of nucleotide (nt), amino acid (aa), or scaffold FASTA files, which are available for download through the interface at http://jau.facom.ufms.br/sepidermidis/download/.

In silico metabolic pathway comparison

Comparative analyses of metabolic pathways were performed using RAST [46] and KEGG [47] platforms. Secondary metabolite gene clusters were analyzed using AntiSMASH 6.0 [48]. For all analyses, input data consisted of nucleotide (nt), amino acid (aa), or scaffold FASTA files, which are available for download through the interface at http://jau.facom.ufms.br/sepidermidis/download/.

TVIISS comparative genomics and effector characterization

Using the OrthologSorter multiple alignment tool (http://jau.facom.ufms.br/sepidermidis/oa/), a comparative analysis of gene presence, absence, and synteny was performed using the S. aureus genome as a reference. Specifically, within the genomic region containing genes associated with the Type VII Secretion System (T7SS), organizational differences were observed among the other genomes analyzed. These variations were tabulated, and the corresponding genes were compared using BLAST. Additionally, selected genes were manually reannotated based on secondary structure analysis and structural prediction, allowing for the inference of putative biological functions.

Virulence factor prediction

All virulence-associated genes from Staphylococcus epidermidis and Staphylococcus aureus available in the VFDB (Virulence Factors of Pathogenic Bacteria) database [49] were retrieved and used as queries in BLAST [50] searches against our sequenced genomes. The comparisons were performed using BLASTp, with a minimum identity threshold of 70%, minimum query coverage of 80%, and an e-value cutoff of 1e⁻⁵. This analysis aimed to identify potential homologs of known virulence factors in the studied strains. The results were tabulated and compared across genomes.

Homology and secondary structure prediction

For selected proteins, multiple protein sequences were aligned using Clustal Omega (version 1.2.4) [51] through the EBI web server. Default settings were used, with the standard progressive alignment method based on a neighbor-joining guide tree and no extra iterations. The resulting guide tree was exported in Newick format and visualized using FigTree v1.4.4 (http://tree.bio.ed.ac.uk/software/figtree/). Secondary structures were predicted using Jpred [52], and signal peptides were identified using SignalP 6.0 [53]. Motif conservation was analyzed using WebLogo [54].

3D protein structure prediction

Structural models were generated using Phyre2 [55] and Robetta [56], with structure superposition using FATCAT [57]. Physicochemical properties of α-helices were analyzed using Heliquest [58]. For this purpose, the FASTA sequences of nucleotide (nt) and amino acid (aa) were used as input.

Arsenic removal and plant growth promotion assays

Both assays were conducted following Felestrino et al. [15], with modifications.

Arsenic removal assay

Se252 and the positive control Alcaligenes faecalis strain Mc250 were cultured in 500 mL flasks containing 250 mL nutrient broth (NB), supplemented with either 500 mM As⁵⁺ or 10 mM As³⁺, or no arsenic (control) [15]. The cultures were incubated at 28 °C and 150 rpm for 3 days. After centrifugation (1200×g for 15 min), the supernatants were analyzed for total arsenic concentration using atomic absorption spectrophotometry.

Indirect plant growth promotion assay

To assess Se252’s potential for plant growth promotion via arsenic removal, a greenhouse experiment with Solanum lycopersicum (Santa Clara 5800) was conducted [15]. Twenty seeds were planted in flasks containing fertilized soil with 2.5 mM As5 + or As3+, with or without Se252 inoculation. After 14 days, plant stem and root growth rates were measured.

Statistical analysis

Statistical significance (p values < 0.05 * and < 0.01 **) for arsenic removal was evaluated using the Kruskal-Wallis nonparametric ANOVA, followed by Dunn’s post hoc test. Analyses were performed with GraphPad Prism 5.

Evaluation of staphyloxanthin production by Se252

Staphyloxanthin production by Se252 was evaluated on LB agar plates with or without potassium bromate (300 mM) and hydrogen peroxide (500 mM), incubated at 37 °C for 36 h. The UV spectra of the bacterial extracts were compared with those of S. aureus strain Sa8325 in the 300–500 nm range for carotenoid peaks.

Results

General and comparative features of the Se252genome

The bacterial strain Se252 was isolated in March 2015 from the rhizosphere of Mimosa calodendron in Nova Lima, Minas Gerais (20°15′83′′S, 43°97′41′′W), within the Iron Quadrangle region [14, 16].DNA was extracted and sequenced, resulting in 10 scaffolds totaling 2,515,570 bp (Fig. S1). No plasmid-related scaffolds were identified.

A phylogenetic analysis, which included Se252, 34 other Se strains isolated from different hosts and niches (Table S1), plus one Staphylococcus aureus(Sa) strain as an outgroup, was performed based on a core genome of 1,797 orthologous protein families (74.2% of the total) (Fig. 1a). This reconstruction revealed three distinct clades (Fig. 1b). Se252 was placed in clade II, alongside Se strains isolated from humans, rodents, goats, and rice plants (Fig. 1b). The composition and diversity of the flexible genomes of the investigated strains are shown in Fig. S2.DNA similarity analysis using a MUMmer matrix was largely consistent with these results (Fig. 1c).

Fig. 1.

Comparative genomic analysis of 36 S. epidermidis genomes. (a) Flower plot showing shared and unique protein-coding genes among strains. The center indicates the core genome (orthologous groups present in all strains); the pink halo represents accessory genes. Numbers near strain codes denote unique genes (black: single-copy; blue: multi-copy). Red numbers indicate gene families shared exclusively with Se252. Icons depict isolation sources, created with BioRender.com (see Table S1). (b) Rooted phylogenomic tree based on core genomes, using S. aureus Sa8325 as outgroup. Clades I–III show strong bootstrap support (red circles). Inverted triangles indicate putative gain events of orthologous families, with protein counts noted. (c) MUMmer matrix showing genomic similarity (red) and dissimilarity (green) among strains (Sa8325 was excluded). (d) Functional classification of strain-specific singletons. WAF: with assigned function; Hyp: hypothetical; Ψ: pseudogene; Tn-ase: transposase. “Plasmid” column shows number of extrachromosomal replicons. Genes linked to virulence/adaptation are in pink squares; shared ones in purple squares. Paralogous genes are marked with copy numbers > 1. “p” indicates plasmid-encoded genes.

Although a small number of singleton genes (ranging from 8 to 143) was identified among the strains analyzed, functional classification revealed that in most genomes at least one gene was associated with virulence or adaptation, potentially conferring a competitive advantage to the strain (Fig. 1d) The complete list of unique genes in the Se252 genome is provided in Table S2. Genes identified exclusively in the outgroup (S. aureus) are presented in Table S3, while those found only in S. epidermidis strains are listed in Table S4.

Virulence genes

Secretion systems and effector comparison

No genes associated with the type II, III, IV, V, VI, or VIII secretion systems were identified in Se genomes. However, a complete cluster of genes encoding the Hly apparatus and the type VII (ESX) secretion system was found. Unlike the type I apparatus, no genes associated with the synthesis of the type VII apparatus were identified in genomes from clade III (Fig. 2a). In clade I genomes, only the structural genes (EsaA, EssA, EssB, EssC) and the helper protein EsaB, along with effectors EsxA and EsaG, were present. In contrast, clade II genomes contain the complete apparatus, along with additional effectors (EsxC, EsxB, EsaE, EsxD), alongside EsxA and EsaG. Notably, Se252, NIHLM031, and FRI909 from clade II exhibited effector profiles similar to those of clade I strains.

Fig. 2.

Synteny and structural analysis of type VII secretion system (T7SS) proteins in S. epidermidis genomes. (a) Presence/absence matrix of structural and effector genes encoding T7SS components. The cladogram reflects the phylogenomic relationships shown in Fig. 1b. T7SS genes are entirely absent in clade III, partially present in clade I, and fully conserved in clade II. Ψ: pseudogene; numbers > 1 indicate paralogs. Numbers in matrix cells represent gene copy number; cell colors correspond to gene/protein labels in panels B and D. TVIIne: two newly identified orthologous protein families (1 and 2) potentially secreted by T7SS. (b) Synteny and gene cluster organization of T7SS in Se252, plant-associated Se (Se2.9 and Se4.7), and S. aureus (ATCC8235) genomes. Background colors indicate conserved or rearranged gene blocks. Modules represent conserved gene clusters encoding core or accessory proteins of the system: Green arrows: putative structural components; red arrows: putative secreted effectors. (c) Evolutionary analysis of TVIIne proteins using EsxA (ESAT-6) and EsxB (ESAT-7) as references; Se252 LXG (colicin IA) serves as outgroup. (d) Predicted 3D structures of selected T7SS-associated proteins The structures identified from 1 to 6 correspond to the respective proteins with the same numbering shown in panel C.

The EsaG effector appears in multiple copies in Se genomes from clades I (3 copies) and II (1 to 5 copies), warranting further investigation. In S. aureus strain Sa8325, 11 copies of this effector were identified. Multiple alignments of these copies revealed high identity and conservation in secondary structures (Fig. S3a), resulting in conserved structural topology (Fig. S3b). Phylogenetic analysis, however, showed that the copies in Se genomes form distinct evolutionary groups (Fig. S3c). In Sa8325, the copies cluster in a single ancestral clade, while Se genomes exhibit four distinct clades, indicating speciation after duplication events.

The structural rearrangements observed in these genes, particularly at the insertion sites in different genomes, further support the speciation hypothesis (Fig. 2b). In Sa8325, genes related to this system are grouped in four modules [59], and Se252 shares similarities with Sa strains, particularly in the organization of modules 1 and 2, while modules 3 and 4 exhibit more variability in plant-associated strains. Structural protein alignments also show high identity across strains, with some notable exceptions (Fig. S3d and S3e).

New candidate effector proteins secreted by the T7SS

From the comparative analysis, two new families of orthologous proteins secreted by the ESX system were identified: TVIIne1 and TVIIne2. TVIIne1 is predominantly found in clade II genomes but also appears in UMB1227 (clade I), RIT611 (clade I), and Sa8325. TVIIne2 is mainly found in clade I genomes, but also appears in Se252, NIHLM031, and FRI909 from clade II (Fig. 2a). Evolutionary analysis indicated speciation of these proteins, with conservation of amino acid sequences in Se strains (Fig. 2c). Three-dimensional structural prediction confirmed that TVIIne1 and TVIIne2 are structurally similar to the EsxA and EsxB heterodimer (PDB 3GWK), suggesting their potential secretion via the ESX apparatus (Fig. 2d).

MSCRAMMs and immune system evasion proteins

Thirteen virulence factors, including proteins associated with adhesion (Atl, EbpS, SdrG, SdrH) and immune evasion (Ebh, VraX), were identified in all Se strains (Fig. 3a). Notably, the ebh gene, found in all genomes, showed a high number of pseudogenes, with variation not following any clear phylogenetic pattern. Additionally, proteins associated with lipoteichoic acid synthesis did not vary across the Se strains.

Fig. 3.

Virulence factors in S. epidermidis strains. (a) Presence/absence matrix of genes associated with adherence, exotoxin and exoenzyme production, biofilm formation, and immune evasion. Comparative strains from S. haemolyticus JSJC1435, S. saprophyticus ATCC 15,305, and S. epidermidis RP62A (highlighted in blue) were included based on VFDB data. Color codes are indicated in the figure legend. The cladogram corresponds to the phylogenomic clustering shown in Fig. 1b. Ψ: pseudogene; numbers > 1 indicate paralogs. (b) Pairwise alignment of IsaB from Sa8325 and IsaB-like from Se252. Sp: signal peptide; Id: sequence identity. Below is an analysis of confidence, conservation, and quality of the amino acid residues composing IsaB-like proteins in S. epidermidis genomes. It can be observed that regions with the highest degree of conservation correspond to positions forming α-helices (red) and β-sheets (green) identified in the structure. (c) Predicted 3D structures of IsaB (Sa8325) and IsaB-like (Se252) proteins using Robetta. (d) Error estimation (Å) of structural predictions shown in panel C. (e) Structural superposition of IsaB (green) and IsaB-like (orange) proteins from panel C using Dali. (f) Syntenic organization of genomic regions flanking isaB (Sa8325) and isaB-like (Se252). Gene abbreviations: aur – aureolysin; clfB – clumping factor B; PM – phenol-soluble modulins; IS6 – transposase family IS6; arcCDBRargR – arginine deiminase pathway. Background shading indicates homologous genes.

A new IsaB-like protein in S. epidermidis 

A new orthologous protein family was identified in all Se strains, annotated as a homolog of S. aureus IsaB (immunodominant staphylococcal antigen B). Despite low sequence identity (34% when compared to Sa8325), the secondary structure of the Se IsaB-like protein was highly conserved (98% identity across Se strains), suggesting its potential role in virulence (Fig. 3b). Structural analysis revealed conserved motifs and folding patterns, indicating functional significance despite low sequence identity (Fig. 3c-e and Fig. S4).

Toxin‒antitoxin systems

A diverse array of toxin-antitoxin systems was identified in Se genomes (Fig. S5a), with multiple copies of type I toxin-antitoxin systems not present in S. aureus.

Phenol-soluble modulins

The genes encoding phenol-soluble modulins (pmtABCD) were identified in all Se genomes (Fig. S5b). In contrast to S. aureus strain Sa8325, which contained only one β-phenol and one δ-phenol modulin, Se genomes showed additional copies (α, ɛ, and β). These proteins exhibited a range of structural forms, primarily consisting of amphipathic alpha helices or hairpin structures (Fig. S5c-d).

Tandem lipoproteins and LPXTG cell wall anchor proteins

A total of 174 genes encoding tandem lipoproteins (LPPs) were identified across the studied genomes, with a notable expansion in clade II genomes and fewer representatives in clade III genomes (Fig. S6a). In some cases, these duplications reflect serial genome duplications, as seen in the UFMGH7 genome. Notably, many of these tandem Lpp genes are inserted in the region corresponding to genomic island 4 in the Se252 strain. Global alignment of the protein sequences revealed a conserved signal peptide (1–20) and high structural conservation (Fig. S6b), despite differing evolutionary histories (Fig. S6c).

Regarding LPXTG proteins, three families were present in all strains, suggesting vertical inheritance, with one family absent in Sa8325 (Fig. S7a-b). Additionally, 13 families of LPXTG proteins were identified in specific strain sets, with clade II strains showing gene amplification, while clade I strains isolated from rice seeds exhibited a high number of pseudogenes in these families.

Secondary metabolite biosynthesis-related gene clusters

The analysis of biosynthesis-related gene clusters (BGCs) revealed four clusters present in all investigated genomes, indicative of vertical inheritance. These included one cluster for non-ribosomal peptide synthesis (NRPs) of unknown function, two clusters for staphylopine and staphyloferrin metallophore synthesis, and a fourth cluster for cyclic-lactone autoinducer (AIP) synthesis (Fig. 4a). Clades I and III contained a cluster for PKS synthesis of unknown function (Fig. S8), while some clade II genomes harbored a cluster for sactipeptide synthesis (hycABCDEF), linked to antimicrobial potential. Analysis of domain composition and conservation of Hyc sequences, empirically validated in species such as S. delphini, S. felsis, and Bacillus subtilis, revealed a high degree of conservation (Fig. 4b). Furthermore, the full staphyloxanthin synthesis cluster was identified in clade II genomes, inserted in genomic island 1 of Se252, with ssbA genes in tandem with crtOPQMN genes and two repetitive regions (Fig. S1a- region 4). Metabolic analysis identified the synthesis route for staphyloxanthin and 4,4’-diapolycopemedial, both linked to photon free radical scavenging (Fig. 4c). However, Se252 did not produce staphyloxanthin under standard synthesis conditions, in contrast to a pigment-producing S. aureus strain (Fig. 4d).

Fig. 4.

Genomic regions associated with secondary metabolite biosynthesis in S. epidermidis. (a) Presence/absence matrix of biosynthetic gene clusters (BGCs) across S. epidermidis strains, identified using antiSMASH. The cladogram reflects the phylogenomic clustering from Fig. 1b. Conserved BGCs across all strains are marked in red. The number inside each square indicates the total number of paralogous copies. (b) Structural and functional characterization of the sactipeptide BGC. Left: protein domain architecture of genes in the hycABCDEF cluster. TM – transmembrane domain; C – cytoplasmic domain; SAM – S-adenosylmethionine domain; ABCt – ABC transporter domain; M16 – M16 domain. Right: amino acid sequence of the subtilosin (bacteriocin) encoded by hycS, with multiple alignment and consensus comparison to homologs in Staphylococcus and Bacillus genomes. (c) Gene cluster (Crt) organization involved in staphyloxanthin biosynthesis. Each gene present in the cluster participates in a reaction involved in the biosynthesis of staphyloxanthin or 4,4′-diapolycopenedial. (d) Spectrophotometric detection of staphyloxanthin in S. aureus and Se252, with absorbance peaks at 455 and 475 nm indicating pigment presence. (e) Classification of autoinducing peptides (AIPs) based on agrD sequences, with clade-specific color coding as in the AgrD phylogeny. TI and TV indicate the classification of the identified AIPs with their respective amino acid variants, along with the model organisms from which the S. epidermidis strains were isolated. (f) Phylogenetic analysis of the identified AIPs. Note that clades I, II, and III are more closely related to each other than to types IV and V.

Quorum sensing

The argBDCA genes associated with quorum sensing were present in all strains. ArgB and ArgA showed high identity (95–98%) between orthologs, while AgrC and AgrD exhibited lower identity (73–87%). Notably, variation in the AgrD protein, responsible for the autoinducing peptide (AIP), revealed four distinct structural configurations, differing from S. aureus AIP type V (Fig. S9). Type II was the most prevalent AIP configuration in the Se strains, irrespective of the host source (Fig. 4e). Phylogenetic analysis showed that peptides of types I, II, and III are more closely related to each other than to types IV and V (Fig. 4f).

Adaptive repertoire genes

Antibiotic resistance

A comprehensive analysis of antibiotic resistance genes in the genomes revealed 18 genes present in all Se and Sa8325 strains, including mecA (β-lactams), mepB and lmrS (multidrug), norAB and alrRS (quinolones), fosB (fosfomycin), folA (trimethoprim), rsmA (various), liaRS (Bacitracin/vancomycin), graF (glycopeptide), and vrgGF-graSRX (bacitracin) (Fig. 5a). Resistance genes such as blaZ, blaI, and blaR1 (β-lactams), mphC (macrolide), and msrA (erythromycin, streptogramin) were identified in some strains. Other resistance-related genes (e.g., mecR1, tetK, dfrG, aph(2), and others) were found sporadically across strains (Table S5).

Fig. 5.

Distribution and genomic context of antibiotic resistance genes in S. epidermidis genomes. (a) Presence/absence matrix of antibiotic resistance genes. Ψ: pseudogene; numbers > 1 indicate paralogs; p: gene located on a plasmid; /p: paralogs split between plasmid and chromosome; x/yΨ: x intact and y frameshifted paralogs. The purple color was used solely to indicate the reference genes employed in the search for orthologous copies, while the red color highlights Sa8325. Icons were obtained from BioRender. Arrows above the matrix denote tandem gene arrangements. The vrgGFgraSRX cluster is associated with bacitracin resistance. Genes conserved across all strains (putative vertically inherited) are marked in black; genes likely acquired via horizontal gene transfer (HGT) are marked in green. (b) Genomic context of reference resistance genes (purple in A), showing co-localization with flanking genes (± 10 kbp). Colored backgrounds indicate functional categories of neighboring genes (see legend). (c) Synteny and organization of resistance loci potentially acquired via HGT, with gene functions summarized in the figure legend. The general function of the genes (indicated by different colors) is shown in the figure legend.

Strains with high antibiotic resistance potential exhibited plasmid-borne resistance genes, except for SURVp10812, which had minimal resistance genes and no plasmid-associated ones. Strains from farmed animals (e.g., dogs, cattle, pigs) showed elevated resistance gene presence, even without plasmids in some cases. The Se252 strain had only vertically inherited resistance genes. Genes associated with antibiotic resistance were often found near integrases, recombinases, and other mobile genetic elements, supporting the hypothesis of horizontal gene transfer (HGT) for some genes, including tetK/tet(38) and fosB/fosD (Fig. 5b-c). Notably, virulence-associated genes like staphostatin and staphopain were identified in these regions, suggesting potential pathogenicity islands.

Carbohydrate uptake

All Se genomes possessed genes for glucose, N-acetylglucosamine, maltose, sucrose, α-glucoside, mannose, lactose, and fructose transporters (Fig. S10a). In contrast, Sa8325 had additional transporters for trehalose, mannitol, galactitol, and ascorbate. The mannitol transporter, linked to S. aureus differentiation, was identified as a potential genomic island in Sa8325 (Fig. S10b). Experimental growth analysis of Se252 revealed no mannitol fermentation capability but better growth at 28 °C compared to other strains, consistent with its plant rhizosphere origin (Fig. S10c-d).

Iron, nitrogen, phosphorous and acetoin metabolism

Genes essential for iron, nitrogen, and phosphorus acquisition and metabolism were universally present across the strains (Fig. S11), suggesting their role as core genetic traits of Staphylococcus species. These included systems for hemin uptake (Hrt), Fe²⁺ internalization (Feo), and sulfur and nitrogen assimilation pathways. Additionally, genes related to acetoin and butanediol synthesis, known to promote plant growth, were identified, further supporting the environmental adaptability of Se252.

Resistance and metabolism of metals

Genes involved in metal detoxification and metabolism, including those for zinc, copper, sulfur, and nickel, were present in all genomes, indicating a shared molecular signature (Fig. 6a and Fig. S12). Specific variations were noted for genes related to potassium, mercury, cadmium, and arsenic metabolism. For cadmium and mercury, resistance genes showed evidence of HGT in specific genomes, often linked to genomic islands. Se252 harbors a broad repertoire of genes associated with arsenic metabolism (Fig. 6b), demonstrated capacity for removal of arsenic, eliminating 30% of As³⁺ and 13% of As⁵⁺ from solution (Fig. 6c), and was associated with a trend toward enhanced plant growth in arsenic-contaminated soil (Fig. 6d), although this increase was not statistically significant compared to plants grown in contaminated soil without bacterial treatment.

Fig. 6.

Metal resistance and metabolism in S. epidermidis genomes. (a) Overview of metal-related metabolic pathways in Se genomes. Presence/absence analysis of each gene is detailed in Fig. S12. Genes likely acquired via horizontal gene transfer (HGT) are indicated with a white background, including those in the kdp (potassium uptake) and mer/met (mercury metabolism) systems. Ni – nickel; K – potassium; Fe – iron; S – sulfur; Cu – copper; Cd – cadmium; Hg – mercury; Zn – zinc; P – phosphorus; N – nitrogen. Two gene clusters are highlighted: the mer cluster, associated with Hg transport in the ATCC12228 genome, and the kdp and lar clusters, associated with K and Ni transport in the M0026 genome. Both are located in putative HGT islands flanked by transposable elements. (b) Arsenic metabolism in strain Se252. (c) In vitro arsenic removal assay. Strain Mc250 (Alcaligenes faecalis) was used as a removal efficiency control. (d) Indirect plant growth promotion assay in the presence of arsenite (As³⁺) and arsenate (As⁵⁺), with or without Se252 inoculation. Bar graphs show mean plant elongation. *p ≤ 0.05; **p ≤ 0.01. # Previously published data on growth promotion by other strains.

Discussion

Staphylococcus is among the most studied bacterial genera, particularly S. aureus, which is associated with various human and animal pathologies [60]. In contrast, S. epidermidis, a coagulase-negative bacterium naturally found in the epithelial and nosocomial microflora, is less significant in human disease but of considerable biological interest. It is linked to opportunistic infections [61–64] and plays a key role in controlling the propagation of other pathogenic bacteria, such as S. aureus, in the nasal cavity [65]. Additionally, S. epidermidis is important for protecting against skin neoplasms [66–70], inflammation induced by Propionibacterium acnes [71], and tissue damage from water loss [64]. However, as a pathogen, it is the leading cause of implant-associated infections [18] and has been linked to skin conditions such as atopic dermatitis [72] and Netherton syndrome [73]. Bloodstream infections, particularly in neonates, are challenging to treat and may result in sepsis, septic shock, or infective endocarditis [74, 75]. Furthermore, S. epidermidis can serve as a reservoir for virulence and antibiotic resistance genes that may be transferred to other pathogens, including S. aureus [19, 76].

While most studies focus on human interactions with S. epidermidis, research also highlights its relationships with animals of economic importance [77]. Notably, humans can act as reservoirs for strains that colonize animals [78–82], emphasizing the role of environmental factors in bacterial evolution and pathogenesis. Although S. epidermidis has been found in plant microbiotas (Table S6), its association with rice seeds (Oryza sativa) as potential endophytes was only reported in 2016 [24, 25], opening new research avenues for understanding its adaptability across environments [83]. However, studies on plant-associated strains remain limited.

To address this gap, our team isolated a novel S. epidermidis strain (Se252) from the rhizosphere of Mimosa calodendron, an endemic plant in Brazil’s Iron Quadrangle region [14, 31], and was found to tolerate high arsenic concentrations, suggesting potential for bioremediation applications [15]. Given the paucity of genomic studies on plant-associated strains, we sequenced the high-quality draft genome of Se252 and compared it with 34 other S. epidermidis strains, both methicillin-resistant (MRSE) and non-MRSE, from diverse sources, including animals and plants, alongside a S. aureus strain. This study aimed to investigate molecular and genetic responses relevant to key biological and medical questions. One key question is whether Se252, or other plant-associated strains, may act as commensal/opportunistic pathogens or reservoirs of virulence genes transferable to other pathogenic species.

Structurally, Se252’s genome is similar to other S. epidermidis strains, but without plasmids, as previously described for strains ATCC12228, SE90, SE95, SESURVp10612, Z0118SE0260, Z0118SE0269 and Z0118SE0269. Phylogenomic analysis grouped Se252 in clade II, alongside strains from rice, rodents, humans, and goats, but distinct from other rice-seed isolates [25]. Although the number of strain-specific genes was low, at least one gene associated with virulence or adaptation was identified in 50% of the strains, with Se252 harboring 11 (~ 0.48%) such genes. Notably, 21% of the genes across all strains were identified as part of the flexible genome, supporting earlier findings that the S. epidermidis pangenome remains open and subject to variation that influences adaptation, virulence, and antibiotic resistance [83, 84]. The deep analysis of Se252 genome highlights several intriguing findings related to its genomic architecture and potential for pathogenicity, including genes involved in adhesion, colonization, immune evasion, pathogenesis, and competition within the core genome suggests a strong potential for vertical inheritance and adaptability. Key genes such as those encoding microbial surface components recognizing adhesive matrix molecules (MSCRAMMs), including SdrG, SdrH, and EbpS [85–87], were found to be conserved across strains, including Se252. These genes play pivotal roles in adhesion to host tissues, which could be critical in both commensal and pathogenic interactions.

Interestingly, some genes associated with virulence, such as exoenzymes and exotoxins (e.g., Hlb, SspA, SspB, and Nuc2 [64, 73, 88–91], were universally present across the strains investigated. This suggests that these virulence factors may be vertically inherited within the species and might contribute to the pathogenic potential of S. epidermidis, including in Se252. The Nuc2, in particular, has been linked to biofilm formation [91], a hallmark of chronic infections, further supporting the strain’s pathogenic potential.

The study also uncovered variability in the presence of certain virulence genes, such as enterotoxins (Sec, SelL, Seb) and Hld (δ-hemolysin), which were restricted to some strains, especially those isolated from humans [92, 93]. This heterogeneity indicates a complex evolutionary pattern, with some virulence factors likely acquired through HGT. The presence of tandem-lipoproteins (TLpp) and LPXTG proteins also exhibited high variability across the strains, supporting the notion of HGT as a key mechanism in shaping the virulence profile of S. epidermidis [94, 95].

A noteworthy finding from the comparative genomic analyses concerns the diversity and abundance of toxin-antitoxin (TA) systems identified across the genomes analyzed. A striking variability was observed both among the S. epidermidis strains and in comparison to S. aureus, suggesting a possible correlation with the ecological origin and environmental pressures associated with each strain’s isolation niche. TA systems are known to contribute to bacterial adaptation by promoting survival under stressful conditions, facilitating persister cell formation, and stabilizing mobile genetic elements that carry adaptive traits [96]. Beyond their role in stress tolerance, TA systems also function as regulatory modules supporting phenotypic plasticity and may serve as defense mechanisms against phage infection [97]. The observed heterogeneity in TA repertoires likely reflects niche-specific selective pressures and may play a key role in long-term persistence and ecological specialization of S. epidermidis strains.

Additionally, regarding other proteins potentially associated with virulence, we highlight those whose genes are located in phage regions and proteins that function as phenol-soluble modulins (Table S7). Phage region 2, using Se252 as a reference, shows signatures corresponding to a SaPI (Staphylococcus aureus pathogenicity island) of approximately 18.2 Kbp, containing at least seven genes previously described as broadly distributed virulence-associated factors [98]. These are capable of replicating as an unstable plasmid [99] and are regulated in a cascade manner [100]. The SaPI was also identified in other genomes investigated in this study, but with heterogeneity in the presence and synteny of the genes. However, the presence of genes encoding superantigens was particularly notable, with the gene encoding enterotoxin B (SEB, also known as toxic shock syndrome toxin-1, TSST-1) [101] being restricted to a few strains isolated from humans (UMB1227, Se95, Se90, FRI909, NCTC12100, and UC7032). Although characterized as part of the flexible genome of these strains, the maintenance of copies of genes encoding these potential superantigens in Se252, as well as in other species associated with plant hosts, represents another indication of an additional virulence factor in these strains when in contact with each other and animal hosts. This contributes to the diversification of the already described repertoire, supporting the argument that the flexible genome contributes to adaptation to different hosts, as discussed below. It is important to highlight that for Se252, isolated from the rhizosphere (a highly dynamic environment characterized by high microbial density and root exudates) this setting promotes the exchange of plasmids, transposons, and phages carrying virulence and resistance genes, facilitating the rapid dissemination and diversification of these genes and contributing to bacterial adaptation in plant-associated environments [102]. Therefore, HGT events are fundamental to the variability of virulence genes in natural environments, especially within the rhizosphere.

Regarding proteins associated with immune system evasion, only VraX [103] and Ebh [104] were identified in all S. epidermidis strains investigated, which contrasts with the presence of capsule-synthesizing proteins, which were absent in all strains of this study but have been described for some specific strains [105]. The proteins IsaB [106] and Aur [107], though not included in the VFDB (virulence factor database), were also categorized here due to the role they play. Notably, the presence and function of IsaB in S. epidermidis (IsaB-like) have so far been overlooked in functional studies as a virulence factor, unlike in S. aureus, where this has been described [108]. We believe that part of this oversight is due to the fact that the gene encoding this protein in S. epidermidis is so distinct from its ortholog in S. aureus that standard sequence comparison and alignment tools, which use default parameters for rapid nucleotide sequences, fail to identify it in S. epidermidis genomes. This identification can only be obtained from translated sequences, revealing a low degree of identity, around 34%. However, structural overlap analysis surprisingly showed that the topology of IsaB-like is identical to that of IsaB, suggesting its homologous function. Another interesting aspect of this gene is that isaB in S. aureus is in tandem with aur, a pair not observed in S. epidermidis genomes. Although isaB is located in a different genomic region, the region where aur is inserted remains conserved, with high synteny in flanking genes, similar to the region in S. aureus. It is not possible to determine, based on the analysis, whether this positional modification occurred due to rearrangement followed by speciation or successive events of gene loss and gain. The fact remains that Aur shows a high degree of identity among orthologous copies, while IsaB/IsaB-like does not, suggesting that the latter may be undergoing selective pressure, which makes future functional studies on this protein even more intriguing.

Finally, but not least in the virulence and adaptation process, we highlight the phenol-soluble modulins (PSMs) [109], which, in S. epidermidis strains, have a greater number of copies and greater diversity than in S. aureus strains. Six distinct copies of genes encoding β-PSM (with only one orthologous copy in the reference S. aureus genome), two copies of α-PSM, and two copies of ɛ-PSM (absent in S. aureus) were identified. Additionally, one copy of δ-PSM, which corresponds to the C-terminal portion of the Hld protein in S. aureus, shows vertical inheritance in the strains analyzed, indicating its functional importance in the biological system. However, in some cases, it is located in unstable genomic regions, possibly islands of HGT [110], making the discussion even more interesting from an evolutionary perspective. Although they can vary in size from 21 to 50 residues, forming one or two alpha-helices, in all cases, the maintenance of an amphipathic structure typical of these proteins was observed [111]. Initially described as a pro-inflammatory complex [112], PSMs are linked to different physiological processes, including immune evasion and aggressive virulence, due to their strong neutrophil lysis ability in humans [113].

In addition to this arsenal of virulence and immune evasion factors, all S. epidermidis strains exhibit the same quorum sensing system encoded by the agrBDCA genes, considered the most important virulence regulatory system in S. aureus [114]. Responsible for mediating cell-to-cell communication via the secretion of autoinducing peptides (AIP) in coagulase-negative staphylococci species, it has been reported that it can block the function of the analogous system in S. aureus, suggesting a new avenue for probiotic therapies [115]. Structurally, the system is composed of a membrane protein (AgrB) involved in the secretion of the pre-processed peptide (AgrD), a membrane sensor protein (AgrC) responsible for recognizing the active peptide in the extracellular space, and transducing signals to AgrA, which in turn acts as a transcriptional regulator of virulence gene expression, including hemolysins, PSMs, among others [116, 117]. AgrB and AgrA are reported to have high identity among orthologous copies [118]. However, greater variation in the composition of AgrC and AgrD proteins has been reported. This occurs because variations in AIP structure leads to recognition by specific sensor proteins, demonstrating that non-native AIP induces biofilm dispersal gene expression in cell cultures and represents new tools to study the role of quorum sensing in S. epidermidis infections [119].

Based on this premise, we conducted an analysis of AIP sequences, which revealed at least four different types in the genomes investigated, in comparison to the AIP from S. aureus used as a reference. Three of these were also differentiated from other AIPs previously described for other S. epidermidis species [119]. Interestingly, the so-called AIP-II in this study was the most widely distributed among the strains investigated, representing the greatest diversity of host isolation (human, animal, rice seeds, and Se252 isolated from cacti). This finding reinforces the hypothesis widely supported here that S. epidermidis isolated from plants has high potential for colonization and induction of virulence in animal hosts (Fig. 7). These findings not only demonstrate the genetic and structural diversity of quorum sensing signals among S. epidermidis strains but also suggest practical implications. The ability of S. epidermidis AIPs to inhibit S. aureus agr-mediated virulence expression has already been proposed as a promising strategy for controlling infections [120]. Strains such as Se252, isolated from non-clinical environments and carrying divergent AIP profiles, may serve as valuable sources for the identification of natural quorum sensing inhibitors. This opens up opportunities for the development of novel probiotic therapies aimed at attenuating S. aureus pathogenicity via interspecies quorum sensing interference, a strategy that bypasses the selective pressure associated with conventional antibiotics and may reduce the emergence of resistance.

Fig. 7.

Metabolic profile of S. epidermidisSe252in soil, plant-associated niches, and as a potential opportunistic human pathogen

At the center, strain Se252 is highlighted in the proposed model of interaction and physiological relationship with human tissue and, simultaneously, with plant root and rhizosphere tissues, demonstrating its potential for adaptation and interaction across different environments. All genes and proteins depicted as interacting with human tissue were compiled based on previously reported literature data. Details regarding the biological function of each component are provided throughout the manuscript. Proteins are represented either by oval symbols or structural models, all clearly labeled with their respective protein names. For the plant-related model, however, these gene and protein activities remain hypothetical and speculative, requiring further functional investigation. Arrows indicate activation, secretion, or induced effects, while blocked lines represent inhibition or antagonistic effects. The colors of arrows and lines were varied solely to emphasize flow or action. White boxes highlight complex biological functions. AMF – antimicrobial factors; NRPS – nonribosomal peptide synthetases; PKS – polyketide synthases; RS – reactive species; CW – cell wall.

Another important issue, directly related to the virulence factors discussed above, is whether there are significant differences in the repertoire of genes associated with antibiotic resistance or metal tolerance between strains isolated from animal and plant hosts. Bacteria resistant to antimicrobials have become serious problems, as they impact both public health and the economy [121]. A systematic analysis of this resistance profile revealed that pathogens associated with the genus Staphylococcus are among those presenting the greatest risk of causing difficult-to-treat pathologies [122]. Multidrug-resistant S. epidermidis strains have been identified globally [123] and are occasionally linked to major outbreaks [124]. To identify the in-silico resistance profile of the S. epidermidis strains investigated in this study, a thorough comparative analysis of the presence and absence of genes associated with this function was conducted. Approximately 53% of all genes identified as having resistance functions correspond to the resistome, which presents an evolutionary profile of vertical inheritance. This suggests that such resistance is a natural response of bacteria in this species to antibiotics from different classes. Notable genes include those encoding the proteins MecA (beta-lactam), MepB (tigecycline), NorA and NorB, AlrS and AlrR (fluoroquinolone), LmrS (lincomycin), FosB/FosD (fosfomycin), FolA/DrfC (trimethoprim), RsmA (various antibiotics), LiaR and LiaS (bacitracin or vancomycin), GraF (glycopeptide), and proteins that are components of the multidrug export system VrgGF-GraSRX (antimicrobial peptides). In contrast, 16 other genes were identified in specific strains, associated with horizontal inheritance and clear signatures of transfer, such as phage insertions or transposition elements [125]. These include genes encoding proteins such as MepA and MepR (tigecycline), Tet(38)/Tetk (tetracycline), Blaz (beta-lactam), DfrG (trimethoprim), BlaI/MecI, BlaR1, and MecR1 (beta-lactam), MphC (macrolide), MsrA (erythromycin and streptogramin), Aph(s) and Ant(4) (aminoglycoside), BleO (bleomycin), ErmC (streptogramin, macrolide, and lincosamide), and LnuA (lincosamide), with an additional copy of FosB/FosD.

A closer examination of resistance profiles based on host origin reveals some interesting insights. In Se252, no gene originating from horizontal transfer was identified, possibly due to the peculiar characteristics and restricted human access to the environment from which it was isolated. This limited exposure to microbiota and exogenous hosts may have restricted the strain’s resistome, allowing it to survive in this specific niche [126]. This can be explained by the fact that, in this specialized and isolated niche, selective pressures from antibiotics and horizontal gene transfer are minimal, resulting in a narrower repertoire of resistance genes compared to strains from more dynamic environments. This streamlined resistome reflects an evolutionary adaptation that conserves energy by avoiding maintenance of unnecessary resistance determinants, enabling the strain to efficiently survive and persist. Additionally, reduced gene flow limits acquisition of new resistance traits, stabilizing the genome for local conditions.

In contrast, the three strains with the greatest resistome diversity were isolated from nonhuman animal hosts. Among PR246B0 strains isolated from pigs, Z0118SE0269 strains from dogs, and UFMGH7 strains from cow, 28 and 25 of the 34 resistance genes evaluated were identified. Though not the primary focus of this discussion, this result raises significant questions. Did these organisms acquire a greater resistance repertoire due to increased and indiscriminate antibiotic exposure while treating related pathologies? Or is this increased resistance a secondary consequence of water, soil, and food contamination in these animals? The World Health Organization has long advocated for reduced antibiotic use in both humans and farm animals, citing studies showing resistance transfer between different organisms [127]. Eco-evo-devo approaches in drug development and interventions represent a new avenue of research that may help mitigate antibiotic resistance while maintaining a healthy interaction between humans and animals [128], a perspective that could be applied to Staphylococcus and its associated pathologies [129].In addition to resistance, understanding the repertoire of genes associated with metabolism and metal tolerance in Staphylococcus is also critical. There are two main reasons for this relationship. Evolutionarily, the diversity of the resistome often correlates with the diversity of metal resistance genes, with co-resistance frequently mediated by the same mobile genetic elements [130]. This phenomenon has been described in species of the genus Staphylococcus [131, 132]. In this study, several strains exhibited resistance to metals such as Hg, Cd, As, and unusually, K. Another aspect, more physiological in nature, involves nutritional immunity [133]. This process, marked by strong selective pressure on metal bioavailability, drives an evolutionary arms race between host and pathogen, influencing strategies for metal sequestration [134, 135]. The presence of metal resistance genes in S. epidermidis strains isolated from environmental sources may reflect adaptive responses to selective pressures imposed by contaminated habitats. In metalliferous environments, prolonged exposure to elevated concentrations of heavy metals such as arsenic, mercury, and cadmium can drive the enrichment and retention of genes involved in metal detoxification, efflux, and sequestration [136]. Furthermore, environments such as the rhizosphere are hotspots for genetic exchange, where microbial interactions, root-derived compounds, and fluctuating physicochemical conditions promote the maintenance of adaptive traits [102]. This ecological and evolutionary context supports the hypothesis that heavy metal-rich environments act as strong selective filters, shaping the resistome and metal tolerance repertoire of S. epidermidis populations outside the clinical setting.

Additionally, the host’s diet can greatly impact the success or failure of bacterial colonization [137], as seen in S. aureus in diabetic patients’ epithelial tissue [138]. As a pathogen, Staphylococcus aims to counteract nutritional immunity by substituting ions, secreting metal chelators (staphyloferritin and staphylopine) [139], and internalizing chelators produced by the host (heme) [68], thereby increasing metal uptake. This discussion was extended to include the identification of all genes related to these strategies in the strains investigated, contributing to the indirect resolution of the questions posed earlier.

For Se252, the repertoire of metal tolerance genes was notably different from what we anticipated before sequencing. Given that Se252 was isolated from one of the world’s largest contamination hotspots for metals and metalloids [140], we expected a higher representation of metal resistance genes than in other genomes from the region [16, 141]. Surprisingly, for most metals (Ni, Zn, Cd, Cu, S, and Fe), the genes involved in acquisition, metabolism, and homeostasis were the same and highly conserved, suggesting vertical inheritance within S. epidermidis. This implies that, if functional, these genes may assist in the evasion of nutritional immunity, much like orthologous genes in human isolates [142], supporting the previously presented virulence discussion.

Flexible genomes are often crucial for bacteria’s adaptation to environmental conditions and pathogenesis [143, 144]. In S. epidermidis, this topic has also been debated [145, 146]. Based on these premises, we expected to find a greater repertoire of flexible genes in plant-associated strains. However, our findings refute this hypothesis, revealing no significant differences in flexible genomes between plant- and animal-associated strains. This contradicts earlier studies that suggested genes exclusive to plant-associated S. epidermidis [25], which could explain their adaptation to plant hosts. Moreover, our analysis of these genomes revealed strong divergence, with only a single family of orthologous proteins shared among them.

Interestingly, a set of genes present in the flexible genome and potentially contributing to Se plant host adaptation were identified in the type VII ESX secretion apparatus. First described in mycobacteria, this apparatus is involved in adaptation and virulence [147]. In S. aureus, it plays a role in inducing apoptosis and evading antimicrobial host fatty acids [148, 149]. Although little is known about its function in S. epidermidis, we found that strains in clade III (isolated from animals) lack the genes for the structural apparatus or effector proteins. In contrast, strains in clade II (isolated from both plants and animals) possess these genes, including a putative novel secreted protein (TVIIne2) in Se252. This suggests that the ESX system could be important for plant adaptation, though its role in animal-associated strains requires further exploration.

Thus, despite 21% of genes being accessory to each genome, no significant repertoire was found to validate the hypothesis of distinct adaptation strategies based on host origin. These findings support the idea that plant-associated strains could act as commensal or pathogenic organisms in animal hosts, and vice versa. This perspective aligns with the notion that plants may provide a beneficial habitat for human pathogens [150]. Additionally, research on animal pathogens’ survival and growth promotion in plants presents new opportunities for studying commensal organisms like Staphylococcus. Functional studies using a One Health approach are critical for understanding the ecological and pathophysiological aspects of emerging diseases, including those caused by S. epidermidis [151]. Furthermore, the conservation of natural environments, such as the Brazilian iron quadrangle, remains essential to prevent the emergence and spread of potential new pathogens [152, 153].

Conclusions

This study highlights the complex genomic landscape of Staphylococcus epidermidis, revealing a mosaic of conserved and flexible elements that challenge the notion of strict niche specialization. Despite being isolated from a plant-associated environment, strain Se252 shares core features with strains from diverse hosts, including genes implicated in adhesion, immune evasion, antibiotic resistance, and potential virulence. The identification of unique and horizontally acquired genes in strains from plants and animals, particularly those associated with metal metabolism and resistance, emphasizes the dynamic nature of the S. epidermidis genome and its capacity for functional expansion. These findings reinforce the idea that environmental and commensal strains may serve as reservoirs of adaptive traits with pathogenic potential. While genomic content alone cannot confirm the transition between ecological roles, it provides a valuable framework for targeted functional studies. In this context, Se252 emerges as a compelling model for exploring the evolutionary and functional continuum between environmental persistence and opportunistic pathogenicity.

Abbreviations

Se

Staphylococcus epidermidis

Sa

Staphylococcus aureus

Se252

S. epidermidis strain 252

Sa8325

S. aureus strain NCTC8325

PSM

Phenol–soluble modulins

PTS

Phosphotransferase system

PKS

Polyketide synthase

NRPs

Nonribosomal proteins

TLpp

Tandem lipoproteins

LXPTG

Surface adhesion protein with an LXPTG domain (X = any aa)

ABC

ATP binding cassette transporters

HGT

Horizontal gene transfer

BGC

Biosynthesis–related gene cluster

T7SS

Type 7 (VII) secretion system

MRSE

Methicillin–resistant Staphylococcus epidermidis

MSCRAMMs

Microbial surface components recognizing adhesive matrix molecules

Authors’ contributions

Conceptualization, LMM, EBF, WLC, and CCMG; methodology, ABS, EBF, IFC, WLC, CGCL, CHPF, AKS, DFR, RCM, JPM, LCMR, MRAD, CCMG, JGS, NFA, and AMV; software, LMM, ABS, AMV, JCS, and NFA.; validation, LMM, ABS, AMV, NFA, JCS, and CCMG; formal analysis and investigation, ABS, EBF, IFC, WLC, CGCL, CHPF, AKS, DFR, RCM, JPM, LCMR, and MRAD; resources, LMM.; data curation, LMM, ABS, AMV, NFA, JCS, and CCMG; writing—original draft preparation, ABS and LMM; writing—review, editing and visualization, ABS, LMM, EBF, IFC, WLC, CGCL, CHPF, AKS, DFR, RCM, JPM, LCMR, MRAD, CCMG, JGS, NFA, and AMV.; supervision, LMM; project administration, ABS and LMM; funding acquisition, LMM. All the authors have read and agreed to the published version of the manuscript.

Funding

This study was partly financed by the Foundation of Protection of Research of the State of Minas Gerais – FAPEMIG (process APQ-02357-17) and by UFOP grants. LMM, NFA, AMV, and JCS received a research fellowship from CNPq. The funders had no role in the study design, data collection, analysis, publication decision, or manuscript preparation.

Data availability

The sequence of the Se252 genome is available at GenBank under accession numbers RQJI00000000.1, Bioproject PRJNA505119, and Biosample SAMN10411012.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.