Sample collection

The bacterial isolates used in this study were collected as part of an environmental investigation, as described previously [17,18]. In brief, a total of 32 surface swab samples were collected (October 2013) from four fitness centers located in the Memphis metropolitan area, Tennessee, using cotton-tipped swabs (Sanicult™; Thermo Remel/Starplex Scientific Inc., Etobicoke, ON, Canada). Trained volunteers swabbed skin-contact surfaces of exercise equipment like Nautilus machines, power striders, stationary bikes, treadmills, elliptical machines, leg presses, dumbbells, stair handrails, and toilet handles. Swabs were immediately placed into tubes containing sterile diluent solution and transported in a refrigerated container to the laboratory for further analysis. Pure cultures of bacteria were isolated by repeated streaking on tryptic soy agar (TSA) medium.

Bacterial culture and DNA extraction

Bacteria were cultured in tryptic soy broth (Sigma-Aldrich, USA) for 18 h at 37 °C, followed by DNA extraction using an Omega BIO-TEK bacterial DNA extraction kit (USA, Cat. No. D3350-02). The quality and quantity of the extracted DNA were assessed using a NanoDrop spectrophotometer (Thermo Scientific, USA) and 0.8% agarose gel, respectively.

Library preparation, sequencing, and read quality assessment

The shotgun genomic library was prepared using an Illumina DNA preparation kit (Illumina cat no 20060060). The libraries were pooled and quantified in qPCR, followed by sequencing on a NovaSeq 6000 platform using V1.5 sequencing kit in paired-end mode (2 × 150 bp configuration). Fastq files, generated from the sequencing platform were used for downstream analysis. We used FastQC v0.11.9 to assess raw reads quality and fastp v0.23.4 [19] with parameters -f 14, -t 10, -F 14, -T 10, -q 25 for trimming.

Genome assembly construction and de novo assembly statistics

We used a k-mer based de novo approach to assemble the genome fragments with SPAdes v3.15.5 [20] utilizing de Bruijn graph algorithms and the “--careful” option to generate longer contigs from short read data (2 × 150 bp). Assembly statistics were computed using QUAST v5.2.0 [21] with default parameters.

Genome integrity evaluation and annotation prediction

The genome annotation was performed byProkka v1.14.6 [22] with specific settings: --kingdom bacteria, --addgenes, --evalue 1e-9, --rfam, -gcode 11, and –force. Following annotation, the completeness and contamination of the generated genomes were assessed using CheckM v1.2.2 [23].

WGS-based prediction of putative antibiotic resistance genes (ARGs)

Putative antibiotic resistance genes in the two bacterial genomes were assayed in the RGI main server (https://card.mcmaster.ca/analyze/rgi) using the CARD (Comprehensive Antibiotic Resistance Database) database v3.2.1 [24] considering perfect (100% identity) and strict (>95% identity) parameters.

WGS-based prediction of associated virulence factors

The genes associated with virulence determinants were predicted in abricate v1.0.1 (https://github.com/tseemann/abricate) using VFDB database [25]. We used the contig file generated from the assembled construction with 70% identity and 80% coverage parameters.

Multilocus sequence typing (MLST) using WGS data

The sequence types (ST) of the two isolates were identified through multilocus sequence typing (MLST) analysis based on sequence variations in specific housekeeping genes. For this purpose, we used the mlst package v2.23.0, which facilitates MLST analysis, and accessed typing schemes from the PubMLST database [26].

WGS-based phylogenetic tree construction

The phylogenetic relationship between the genomes of our isolates and other similar sequence type clones (E. faecalis ST40 and S. haemolyticus ST52) available in the NCBI Genome Database (https://www.ncbi.nlm.nih.gov/genome/) was analyzed using an online tool named REALPHY (https://realphy.unibas.ch/realphy/), which integrates Bowtie2 for mapping-based alignment and PhyML for inferring phylogenetic trees. For this purpose, five genomes of E. faecalis ST40 and five genomes of S. haemolyticus ST52 were downloaded to construct phylogenetic trees, allowing a comparative assessment of genetic relatedness. Detailed information about these isolates, including accession numbers and metadata, is provided in S1 Table. This analysis helped contextualize the genomic characteristics of our isolates within the broader phylogenetic landscape of these sequence types.

Comparative proteomics and gene family expansion/contraction analysis

To elucidate the evolutionary relationships between S. haemolyticus S5, E. faecalis S3, and other closely related strains within their respective species, we utilized the OrthoFinder algorithm as implemented in the OrthoVenn 3 package [27]. This analysis was conducted with an e-value threshold of 1e − 5 and an inflation value set to 1.5, enabling accurate identification and clustering of orthologous gene families. To further explore the evolutionary dynamics, we performed gene family expansion and contraction analysis using CAFÉ v5.0 [28]. This program employs a birth-and-death process model to estimate rates of gene gain and loss across a given phylogenetic tree. By analyzing the distribution of family sizes under this model, we assessed the significance of observed differences in gene family sizes among taxa, providing valuable insights into the adaptive evolution of these strains.

Gene cluster prediction

To identify potential gene clusters within the genomes of S. haemolyticus S5 and E. faecalis S3, we utilized the bacterial version of the antiSMASH v6.0 tool [29]. This powerful bioinformatics tool specializes in the prediction of secondary metabolite biosynthetic gene clusters. The nucleotide sequences and corresponding GenBank files generated during the genome annotation process were used as inputs. Gene cluster prediction was performed using the ‘strict’ detection parameter to ensure high-confidence results. This approach allowed for the identification of genomic regions potentially involved in the production of bioactive compounds, providing further insights into the functional and adaptive capabilities of these isolates.

Raw sequence read, quality check, and trimming

The sequencing platform generated 1,063,199 and 901,900 paired-end reads for E. faecalis S3 and S. haemolyticus S5, respectively (S2 Table). The initial quality assessment revealed Q30 bases of 87.86% and 86.96% for E. faecalis S3 and S. haemolyticus S5, respectively, prior to filtering. Subsequently, theraw reads were trimmed to improve read quality for downstream analysis, resulting in increased Q30 bases to 88.28% for E. faecalis S3 and 87.29% for S. haemolyticus S5 (S2 Table). The high percentage of Q30 bases indicates the overall good quality of the sequencing data.

Genome assembly from fragmented DNA sequence and annotation

The trimmed reads were used for genome assembly, resulting in 39 contigs for E. faecalis S3, with the largest contig spanning 590,494 bp and an N50 value of 183,877 bp. For S. haemolyticus S5, the assembly produced 58 contigs, with the largest contig measuring 249,943 bp and an N50 value of 93,523 bp. Assembling complete genomes from fragmented data is challenging and critical in whole-genome sequence (WGS) analysis. Various assemblers are available, but we opted for the k-mer based de novo assembler SPAdes [30] based on the standard recommendations [31], followed by annotation using Prokka. Annotation revealed GC contents of 37.41% for E. faecalis S3 and 32.77% for S. haemolyticus S5. E. faecalis S3 had 2826 CDS, 5 rRNA, 78 misc_RNA, and 52 tRNA genes, while S. haemolyticus S5 had 2385 CDS, 12 rRNA, 78 misc_RNA, and 64 tRNA genes. Completeness and contamination assessments showed high completeness scores of 99.63% for E. faecalis S3 and 99.62% for S. haemolyticus S5, with no detectable contamination, indicating that the genomes are nearly complete and suitable for further genomic analysis.

WGS-based antimicrobial resistance gene prediction

The putative antimicrobial resistance (AMR) genes were predicted and are presented in Table 1. E. faecalis S3 harbored multiple AMR genes, including dfr(E), tet(M), efrA, efrB, IsaA, IreK, and emeA, conferring resistance to various antibiotic classes such as diaminopyrimidine, tetracycline, macrolide, streptogramin, and pleuromutilin (Table 1). Additionally, the emeA gene in E. faecalis encodes an efflux pump that expels biocides and ammonium-containing compounds commonly used in disinfectants [32]. A previous study reported a high prevalence of emeA gene (42.8%) among clinical isolates of E. faecalis [32]. The presence of the emeA gene in E. faecalis S3 is prevalent, and therefore, cleaning surfaces with ammonium-based disinfectants alone may not effectively eliminate this pathogen. However, phenotypic resistance was not assessed in this study. Previous studies have also identified several AMR genes, such as lsa(A), RlmA(II), LiaR, LiaS, dfr(E), and tet(M), in E. faecalis isolated from diseased fish [22] and resistance to chloramphenicol, linezolid, and teicoplanin in dental plaque isolates [33]. In contrast, S. haemolyticus S5 exhibited resistance to macrolide (mphC), aminoglycoside (ANT(4’)-Ia), tetracycline (tetK), and fluoroquinolone (qacA) antibiotics (Table 1). A study of 365 S. haemolyticus isolates found 91.3% and 85.4% resistance to cefoxitin and erythromycin, respectively, and 57.3%, 52.8%, and 3.7% resistance to co-trimoxazole, clindamycin, and linezolid, respectively [34]. The presence of multidrug-resistant (MDR) bacteria on environmental surfaces, such as those found in healthcare facilities [35], public restrooms [36], supermarket shopping carts [37], and fitness facilities [38], is a growing concern for community health. Surveillance of these environments is a key component of the One Health approach. This approach recognizes that human health is closely linked to the surrounding environment, making regular surveillance essential for effective public health management [39]. Through source tracking, monitoring, prevention strategies, and intervention policies, the One Health framework aims to control the spread of MDR bacteria and protect community health. In April 2019, the Interagency Coordination Group on Antimicrobial Resistance, comprising the WHO, OIE, and FAO, presented a report titled “We Can’t Wait: Securing the Future Against Drug-Resistant Infections” to the UN Secretary-General. To combat this situation, all agencies agreed to raise awareness of the AMR problem among general people, healthcare professionals, and policymakers to control the spread of antibiotic use and AMR bacteria [40].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Putative antibiotic resistance genes present in E. faecalis S3 and S. haemolyticus S5.

https://doi.org/10.1371/journal.pone.0324894.t001

Public gymnasium centers are high-contact environments where individuals frequently share fitness equipment and floor surface, often under conditions of limited or inconsistent disinfection. The combination of perspiration-induced warmth and moisture, along with frequent skin-to-skin contact, creates an optimal environment for bacterial persistence and transmission [41]. The detection of MDR bacteria in such settings is critical for understanding potential non-clinical transmission routes and for guiding effective public health interventions.

WGS-based virulence factor prediction

The prediction of virulence factors is crucial for assessing the pathogenic potential of a pathogen. In our analysis, E. faecalis S3 harbored 19 virulence genes (identity 92–99%, coverage 86–100%), including biofilm-associated pilus genes (EbpA, EbpB, EbpC), gelatinase (gelE), and serine protease (sprE), known for enhancing pathogenicity (Table 2). Previous studies have highlighted the importance of the ebpABC-bps gene cluster in attachment, a critical step in biofilm formation initiation [42]. Furthermore, gelatinase and serine protease encoded by gelE and sprE are also reported as major virulence factors in E. faecalis [43]. In comparison, S. haemolyticus S5 showed only two virulence genes, ClpP and ClpC, with ClpP known for enhancing infectivity in Staphylococcus species [44]. Thus, the presence of virulence genes alongside antibiotic resistance properties suggests that these isolates may pose a higher risk of pathogenicity.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Predicted virulence factors in E. faecalis S3 and S. haemolyticus S5.

https://doi.org/10.1371/journal.pone.0324894.t002

Multilocus sequence typing (MLST) of the isolated strains using WGS

S. haemolyticus S5 was identified as ST52 based on allelic variations in arcC(2), SH_1200(1), hemH(6), leuB(1), SH1431(2), cfxE(1), and Ribose_ABC(4). The prevalence of sequence type (ST) 52 is relatively uncommon in nature. However, a previous study reported the emergence of ST52 strains of S. haemolyticus isolated from hospitals and hospital-associated environments in China [14]. Interestingly, all these isolates exhibited resistance to a broad spectrum of antibiotics. In contrast, E. faecalis S3 was assigned ST40 based on allelic variations in gdh(3), gyd(6), pstS(23), gki(12), aroE(9), xpt(10), and yqiL(7). ST40 is a major clone reported across diverse habitats, including humans, animals, and environmental sources [45] and is associated with infections [46]. MLST is a standardized, sequence-based method that characterizes bacterial strains by analyzing the allelic profiles of several housekeeping genes [47]. This makes it a valuable tool for epidemiological tracking, antimicrobial resistance monitoring, and understanding the population structure and evolutionary history of pathogens [48]. A comparative genomics analysis revealed that the distribution of virulence-associated genes was not significantly different between E. faecalis ST40 strains isolated from animal and human sources, nor between those from commensal and clinical backgrounds [45]. In contrast, S. haemolyticus ST52 is an uncommon strain type and has primarily been reported from home environments [49]. Its detection in a public gymnasium setting may indicate potential environmental dissemination and warrants further investigation into its reservoirs, transmission pathways, and public health implications. Furthermore, pathogenic strains with an MLST designation can be compared with the same ST isolates in the global database to understand genome-relatedness and antibiotic resistance patterns. Thus, MLST can guide antibiotic selection against AMR bacteria, though correlations between MLST and resistance can vary by species, as seen in recent studies with Staphylococcus epidermidis [50].

WGS-based phylogenetic tree construction of the sequenced strains

A whole-genome sequencing (WGS)-based phylogenetic tree was constructed using five strains of S. haemolyticus ST52 and five strains of E. faecalis ST40 (S1 Table). The analysis revealed that S. haemolyticus S5 exhibited a close genetic relationship with S. haemolyticus C86FS1, a strain isolated from hospital-associated feces in China (Fig. 1A). Notably, like S. haemolyticus S5, the S. haemolyticus C86FS1 strain was resistant to multiple antibiotics, including ciprofloxacin, gentamicin, and linezolid [14]. Similarly, E. faecalis S3 showed a close relationship with E. faecalis 1MPA3, a strain isolated from a hospital-associated environment in South Africa (Fig. 1B). This genetic proximity highlights the global dissemination of specific antibiotic-resistant clones in healthcare-associated environments.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Core genome-based phylogenetic tree of the selected strains.

(A) and (B) indicate the phylogenetic relationship between S. haemolyticus S5 and E. faecalis S3 with other strains isolated from different surfaces. Macrococcoides caseolyticum FDAARGOS 868 and Vagococcus intermedius STAA11 were taken as outgroups for the construction of a phylogenetic tree for S. haemolyticus S5 and E. faecalis S3, respectively.

https://doi.org/10.1371/journal.pone.0324894.g001

Protein ortholog and gene family expansion and contraction

We compared protein orthologs between S.haemolyticus S5 and five other strains belonging to the ST52 clone (S1 Table). The analysis identified 2,359 proteins in 2,302 clusters and 49 singletons in S. haemolyticus S5 (Fig 2A). Among these, 2,153 clusters were shared with other strains, while S. haemolyticus S5 exhibited one unique cluster (Fig 2B). Gene family expansion and contraction analysis revealed that S. haemolyticus S5 had one expanded gene family (GO:0032196), while 29 gene families, primarily associated with metabolic processes (S3 Table), were contracted in this strain (Fig 2C). The expanded gene family encodes a protein, the putative transposon Tn552 DNA-invertase Bin3, which is a type of mobile genetic element. This protein plays a critical role in the horizontal transfer of antibiotic resistance genes, contributing to the dissemination of resistance within microbial communities [51]. The expansion of this gene family in S. haemolyticus S5 underscores a potential emerging threat, as it enhances the strain’s ability to spread antibiotic resistance within the community.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. The protein cluster and protein ortholog comparison.

(A) An overview of the protein clusters and singletons in E. faecalis S3 and other closely related strains, (B) The distribution of shared and unique protein clusters among different strains of E. faecalis, (C) The expansion and contraction of gene families in E. faecalis S3 in comparison with other strains. (D) The protein clusters in S. haemolyticus S5 and its closely related strains, (E) The number of shared and unique protein clusters in S. haemolyticus S3 in comparison to other strains. (F) The evolutionary tree indicates the number of gene family expansions and contractions in S. haemolyticus S5 over the evolutionary period.

https://doi.org/10.1371/journal.pone.0324894.g002

The protein orthology analysis among E. faecalis strains (S1 Table) revealed that E. faecalis S3 possesses 2,807 proteins organized into 2,757 clusters and 42 singletons (Fig 2D). Comparative orthology indicated that E. faecalis S3 contains a unique cluster not found in the other strains (Fig 2E). Evolutionary analysis of E. faecalis S3 identified 77 contracted gene families, but no gene family expansions were observed (Fig 2F). These contracted gene families are associated with diverse biological functions, including metabolic processes, carbohydrate transport, responses to stimuli, and other critical pathways (S4 Table). The lack of gene family expansion in E. faecalis S3, coupled with significant contractions in key functional categories, may influence its adaptability and pathogenic potential in specific environments. These findings provide valuable insights into the evolutionary dynamics of E. faecalis S3 and its unique functional attributes.

WGS-based gene cluster prediction

The gene cluster prediction analysis is depicted in Fig. 3. It reveals the presence of a cyclic lactone autoinducer gene cluster in E. faecalis S3, consisting of a biosynthetic gene (agrA), a regulatory gene (hypothetical protein), and a transporter gene (stp) (Fig. 3A). However, a protein BLAST search unveiled that the regulatory hypothetical protein is encoded by a transcriptional family gene lysR, exhibiting 100% similarity with E. faecalis strain ADL-337. The agr system, involving agrB, regulates quorum-sensing responses and virulence in E. faecalis and other Gram-positive pathogens [52]. In S. haemolyticus S5, two gene clusters were identified: cyclic lactone autoinducer and NI-Siderophores (Fig 3B). The lactone-containing autoinducer peptide activates the agr system, a global regulator in Staphylococci that influences virulence-associated protein production [53]. The NI-Siderophores cluster includes SnbF, which is involved in siderophore production, which is crucial for iron acquisition and enhances virulence in Staphylococcus [54]. These gene clusters highlight the pathogenic potential of E. faecalis S3 and S. haemolyticus S5, underscoring their ability to regulate virulence and adapt to host environments.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. The predicted gene cluster in both E. faecalis S3 and S. haemolyticus S5.

(A) The cyclic lactone autoinducer gene cluster in E. faecalis S3. The core biosynthetic gene agrB encodes regulator protein B, (B) and (C) represent cyclic lactone autoinducer and NI-siderophore gene clusters predicted in S. haemolyticus S5, respectively. Both gene clusters are involved in pathogenicity.

https://doi.org/10.1371/journal.pone.0324894.g003