Genomic driver underlies clonal shift of methicillin-resistant Staphylococcus aureus in the healthcare setting

ABSTRACT

The epidemiology of healthcare-associated methicillin-resistant Staphylococcus aureus (HA-MRSA) has shifted in recent decades, with a notable decline in the lineage ST239 and continued dominance of ST5. The evolutionary mechanisms underlying these contrasting trends remain unclear. We conducted a longitudinal analysis of 3,410 clinical HA-MRSA isolates collected between 2005 and 2020 to assess clonal prevalence and genomic changes. Further phenotypic characteristics were evaluated through in vitro and in vivo experiments, including biofilm formation assays, antibiotic susceptibility testing, polymorphonuclear neutrophil leukocyte (PMN) killing assays, and mouse catheter infection models. The prevalence of ST239 declined markedly from 34.85% in 2005 to 0.52% in 2020, accompanied by reduced biofilm formation and increased susceptibility to vancomycin. Genomic analyses revealed a temporal accumulation of mutations in ST239 lineage, notably a specific tarJ’ point mutation (C665T; Thr222Ile), whose detection frequency increased as ST239 prevalence declined, becoming fixed in 100% of ST239 isolates by 2020. This tarJ’ mutation led to cell wall defects, decreased survival in acidic environments, reduction in biofilm formation, and attenuated persistence in mouse catheter models. In contrast, ST5 exhibited no comparable genomic deterioration and maintained phenotypic stability throughout the same period. The decline of ST239 HA-MRSA is driven by mutation-induced fitness trade-offs, particularly affecting cell wall integrity and persistent infection capacity. These findings highlight how evolutionary constraints and adaptive costs shape the population structure of bacterial pathogens.

Introduction

Methicillin-resistant Staphylococcus aureus (MRSA) remains a leading cause of both healthcare-associated infections (HAIs) and community-acquired infections (CAIs), contributing significantly to global morbidity and mortality [1]. The emergence and spread of MRSA are largely driven by the acquisition of antibiotic resistance genes and the clonal expansion of specific sequence types (STs) [2]. Among these, ST239 and ST5 have emerged as two major healthcare-associated MRSA (HA-MRSA) clones, attracting global attention due to their widespread distribution and public health impact [3–5]. ST239, in particular, has been a predominant HA-MRSA lineage across multiple continents, characterized by extensive multidrug resistance and frequent involvement in hospital outbreaks. Originating in the late 1970s in Australia, ST239 gained prominence with the emergence of the first epidemic MRSA strain in the UK in 1981, known as EMRSA-1 [6]. By the late 1980s, ST239 accounted for a substantial proportion of MRSA cases in England and Wales and became a dominant strain in Asia by the mid-2000s, associated with a majority of HA-MRSA infections in a region home to over 60% of the global population [7,8]. The evolution of ST239 has been characterized by the acquisition of SCCmec type III including a staphylococcal cassette chromosome mec element that confers methicillin resistance and the mobile genetic element-encoded sasX gene, which significantly enhances nasal colonization, lung infection, and immune evasion. These genetic adaptations, coupled with a complex array of resistance determinants and virulence factors, have contributed to the persistence and adaptability of ST239 in nosocomial environments [9–11].

However, recent surveillance data indicate a notable global decline in ST239 prevalence, likely driven by strengthened infection control practices and the emergence of competing MRSA lineages [12,13]. This decline may represent a broader epidemiological shift, enabling other clones to occupy the ecological niches vacated by ST239. For example, the chemotaxis inhibitory protein gene (chp) has been implicated in the replacement of hospital-associated clones like ST239 by community-associated clones such as ST59 in China [14]. Despite this, ST239 and ST59 often occupy distinct ecological niches, limiting the ability of direct genomic comparisons to fully explain lineage displacement under real-world conditions. In contrast, ST5 exhibits remarkable ecological flexibility, circulating in both hospital and community settings. Its evolution is characterized by the acquisition of multiple SCCmec types (I, II, III), indicating a broad adaptability to varied antimicrobial pressures and host environments [15]. This versatility has facilitated the widespread dissemination of ST5 and blurred the traditional boundaries between HA-MRSA and CA-MRSA, thereby complicating infection control efforts [16].

In our previous study, we reported a declining prevalence of ST239 MRSA alongside a stable prevalence of ST5 MRSA within the same healthcare setting by an epidemiologic investigation based on the MRSA isolates recovered from 2008 to 2017 [12]. We hypothesized that ST239 has accumulated deleterious mutations, reducing its fitness and ecological adaptability, whereas ST5 maintains genomic stability, supporting its persistence. To test this, we analysed 3,410 clinical isolates recovered from 2005 to 2020 using whole-genome sequencing (WGS), phenotypic assays, and experimental genetics. These results underscore the importance of ongoing genomic surveillance and a deeper understanding of MRSA evolutionary trajectories to inform strategies for transmission control, treatment optimization, and public health intervention.

Methods

Bacterial strains

A total of 3,410 sequential, non-redundant MRSA isolates were collected from a Chinese tertiary teaching hospital between 2005 and 2020. Among them, 810 ST239 and 1,615 ST5 non-duplicate isolates were selected for minimum inhibitory concentration (MIC) testing to ensure broad temporal representation. Subsequently, 61 ST239 and 65 ST5 isolates (total 126 isolates) from the years 2005, 2010, 2015, and 2020 were randomly chosen from this pool for further investigation (Table S1). In addition, due to the very limited number of ST239 strains available from 2020, two ST239 and five ST5 isolates obtained in 2019 were also included in the growth curve and biofilm formation assays. Species identity of all isolates was confirmed using MALDI-TOF MS (Bruker Daltonics, Bremen, Germany), and all 126 isolates were confirmed to be MRSA strains carrying the mecA gene.

For bacterial culture, planktonic cells were grown in Tryptic Soy Broth (TSB) under shaking conditions (200 rpm), while biofilm formation was assessed in microtiter plates using TSB supplemented with 0.5% glucose (TSBg) under static conditions. For solid media, isolates were cultured on Tryptic Soy Agar (TSA) or sheep blood agar (SBA) plates. All cultures were incubated at 37°C, and isolates were stored at −80°C for long-term preservation.

Antimicrobial susceptibility testing

The antimicrobial susceptibility of all isolates in this study was evaluated using the Biomerieux Vitek 2 system following manufacturer’s instructions. Results were interpreted in accordance with Clinical and Laboratory Standards Institute (CLSI) guidelines. The antimicrobial agents tested included cefoxitin (FOX), gentamicin (GEN), penicillin (PEN), cefazolin (CZO), linezolid (LNZ), levofloxacin (LVX), erythromycin (ERY), Fosfomycin (FOS), rifampin (RIF), vancomycin (VAN) and teicoplanin (TEC). S. aureus ATCC 29213 was used as the quality control strain for susceptibility testing.

The minimum inhibitory concentrations (MICs) of three major antimicrobials – vancomycin (Autobio, Zhengzhou, China), teicoplanin (Bio-Kont, Wenzhou, China), and linezolid (Autobio, Zhengzhou, China) – were determined for representative S. aureus isolate using the E-test method for validation.

Biofilm formation test

A total of 63 ST239 and 70 ST5 MRSA isolates (respectively isolated from 2005, 2010, 2015, 2019 and 2020) were cultured overnight at 37°C, and adjusted to a 0.5 McFarland standard using sterile saline. For each strain, 20 μL of bacterial suspension and 180 μL of TSBg were added to individual wells of a 96-well microplate, with three replicates per strain. Wells containing fresh TSBg without bacteria served as blank controls. The plate was incubated at 37°C for 48 h. After incubation, the supernatant was gently discarded, and the wells were washed twice with sterile phosphate-buffered saline (PBS). Biofilms were fixed with 100 μL of Bouins’ solution for 1 h, stained with 100 μL of 2% crystal violet solution for 1 min at room temperature, and then thoroughly rinsed with distilled water. Biofilm biomass was quantified by measuring the absorbance at 470 or 570 nm using a microplate reader.

Whole-genome sequencing and assembly

Genomic DNA was extracted from overnight cultures of 61 ST239 and 65 ST5 MRSA isolates (respectively isolated from 2005, 2010, 2015, and 2020) grown in TSB using a standard phenol–chloroform extraction method, with enzymatic lysis by lysostaphin. WGS was performed on the Illumina HiSeq 2500 platform (Illumina, San Diego, CA, USA) with 150-bp paired-end reads. Raw sequencing data were subjected to quality control using a Bioanalyzer (Agilent Technologies), and high-quality reads were exported in FASTQ format. Additionally, reads were assembled to contigs de novo using CLC Genomics Workbench 24.0 (Qiagen) with default settings.

Phylogenetic analysis

For single nucleotide polymorphism (SNP)-based phylogeny, high-quality reads were aligned to reference genomes of ST239 and ST5 using CLC Genomics Workbench 24.0 (Qiagen). S. aureus strain TW20 (GenBank accession no. FN433596.1) was used as the reference genome for ST239, and S. aureus strain N315 (GenBank accession no. NC_002745.2) was used for ST5. Maximum likelihood (ML) trees were constructed using the CLC pipeline under the general time-reversible (GTR + G + I) model. Branch support was assessed with 1000 ultrafast bootstrap replicates. The final tree was visualized by ITOL3 (https://itol.embl.de/). The tree scale 0.02 indicates that the expected number of substitutions changes per nucleotide (amino acid for protein sequences) is 0.02.

GO enrichment analyses

Non-synonymous single nucleotide polymorphisms (nsSNPs) identified in the genomes of ST239 and ST5 isolates were annotated to their corresponding genes using the S. aureus reference genome. Gene Ontology (GO) functional annotations were retrieved from the UniProt and Gene Ontology Consortium databases. Enrichment analysis of genes harbouring recurrent or temporally increasing nsSNPs was performed using the topGO R package (Bioconductor) with the “weight01” algorithm and Fisher’s exact test. Significantly enriched GO terms were defined as those with adjusted P-values < 0.05 after Benjamini–Hochberg correction.

Sequence type, spa-type, and SCCmec type analysis

ST, spa-type and SCCmec type were determined based on the assembled contigs from WGS data using the Center for Genomic Epidemiology (CGE) online tools (https://cge.cbs.dtu.dk/services). MLST typing and spa typing were respectively performed using MLST Finder 2.0 and spaTyper 1.0, while SCCmec typing and detection of the mecA gene were carried out using SCCmec Finder 1.2 with default thresholds (≥90% identity, ≥ 60% minimum length) https://cge.cbs.dtu.dk/services/SCCmecFinder/.

Detection of virulence factors and antimicrobial resistance genes

Virulence factors and antimicrobial resistance genes were identified by mapping assembled genomes against the VFDB 2022 and ResFinder 2020 databases, respectively, using Genomics Workbench 24.0 (Qiagen), with thresholds set at >75% identity and >90% coverage.

In silico prediction of the effects of missense mutations on protein stability

In order to understand the impacts of non-synonymous single nucleotide polymorphisms (nsSNPs) in the genomes on the structure, DUET, a web server for an integrated computational approach for studying missense mutations in proteins is used to study and predict the effects of nsSNPs on protein stability (https://biosig.lab.uq.edu.au/duet/stability).

Allelic gene replacement by homologous recombination and genetic complementation

To construct the C665T point mutation in the tarJ’ gene of ST239 and ST5 MRSA isolates, allelic replacement was performed using the temperature-sensitive plasmid pKOR1 (Table S2), as previously described [17]. DNA fragments harbouring the desired mutation were PCR-amplified from clinical isolates 2005-ST239-1 and 2005-ST5-1 without tarJ’ mutations using primers ST239-tarJ’-A/B/C/D/C2/D2 and ST5-tarJ’-A/B/C/D/C2/D2, respectively (Table S3). The PCR products were cloned into pKOR1 via a Clonase reaction at the attB sites, generating the recombinant plasmid pKOR1-tarJ’ (C665T). This plasmid was first transformed into E. coli Top10 by heat shock, followed by electroporation into S. aureus RN4220, and subsequently introduced into 2005-ST239-1 and 2005-ST5-1. Allelic exchange was performed following the standard protocol, and successful integration was confirmed by PCR and Sanger sequencing using primers pKOR-Up-ApaI-F and pKOR-Dn-KpnI-R. Genomic validation of the point mutation was further confirmed using ST239-tarJ’-F/R and ST5-tarJ’-F/R primer sets. Strain and primer details are listed in Tables S2 and S3, respectively.

Growth assay under normal and acidic conditions

S. aureus strains were cultured in TSB to achieve an OD₆₀₀ of 0.6 (mid-exponential phase). The culture was diluted to OD₆₀₀ = 0.01 and then cultured at 37 °C with agitation at 200 rpm for 24 h. Measurements of OD₆₀₀ were taken every 2 h using a microplate reader (BioTek Synergy H1). Background absorbance from TSB-only wells was subtracted from all measurements.

For acidic stress assays, bacterial suspensions were prepared identically but diluted in TSB acidified to pH 5.5 using sterile 1 M HCl, with pH verified before inoculation (Mettler Toledo pH meter). Growth was monitored at 37°C with 200 rpm agitation for 48 h, with OD₆₀₀ measurements taken at 2-h intervals. A TSB control well (pH 7.4) and an acidified TSB well (pH 5.5) without bacteria served as background controls. Growth curves were plotted as OD₆₀₀ versus time.

Bacterial tolerance to vancomycin killing

For the analysis of biofilm resistance to antibiotic killing, S. aureus biofilms were established in 96-well microtiter plates by incubation for 48 h. The pre-formed biofilms from representative isolates of each lineage were exposed to vancomycin for 3 h or 6 h, which inhibits cell wall synthesis in susceptible bacteria. These exposure times, consistent with previous studies [18], were selected to evaluate antibiotic effects on bacterial cells. Bacterial survival was assessed by determining colony-forming units (CFU). Biofilms at the 0 h time point (prior to vancomycin treatment) served as controls. After gently washing with sterile PBS, the biofilm layers were subjected to 5 min of sonication followed by 15 min of vortexing to disperse the bacteria. CFU counts were determined by serial dilution and plating. The survival fraction was calculated by dividing the CFU after vancomycin treatment by the CFU at 0 h.

Biofilm observation by CLSM

Biofilm formation was quantified as described previously [19]. For biofilm observation by CLSM, biofilm grown in glass-bottom dish (NEST, 801001) was washed with PBS three times, and then sequentially stained with 1 μM of SYTO9 (Invitrogen) for 30 min, 1 μM of propidium iodide (PI) (Yeasen, Shanghai) for 15 min. The stained cells and DNA were visualized by Zeiss LSM 880 upright laser scanning confocal microscope with a 20×objective. Three-dimensional biofilm images were generated by ZEN lite.

Resistance to neutrophil killing

Human neutrophils were isolated from heparinized venous blood of healthy volunteers using the Human Peripheral Blood Neutrophil Isolation Kit (TBD, LZS11131), following the manufacturer’s protocol. For biofilm tolerance assays, S. aureus biofilms were established in TSBg for 48 h in 96-well microtiter plates using equal bacterial inoculation across all strains. Biofilms were then incubated with freshly isolated neutrophils at a multiplicity of infection (MOI) of approximately 10:1. Co-incubation periods of 30, 60, and 90 min were independently assessed. Surviving bacteria were quantified by CFU counting. The survival fraction was calculated by dividing the CFU after neutrophil treatment by the corresponding 0 h CFU. Investigators were blinded to group allocation during the experiment.

Mouse experiments

A total of 15 female C57BL/6 mice (Jackson) were used for mouse experiments. All mice were between 6 and 8 weeks of age at the time of use. All animals were euthanized by CO₂ at the end of the studies. Catheter infection model experiments were performed as previously described [20]. Briefly, 1-cm catheter pieces were coated with equal numbers of bacteria and tested in control experiments to be the same range (∼1 × 10³–10⁴). Catheters were then inserted under the dorsal skin of mice. CFU on the catheters and the directly adjacent tissue were determined seven days after the infection when the catheters were picked out. Investigators were blinded as for group allocation when determining CFU.

Statistical analysis

Statistical analyses and picture drawing were performed using GraphPad Prism. For comparisons between two groups, unpaired two-tailed t-tests or Mann–Whitney tests were used, depending on data distribution assessed by the Shapiro–Wilk test. All error bars in graphs represent the standard deviation (SD). A P-value of less than 0.05 was considered statistically significant.

Data availability

The genome sequences of all the isolates have been deposited in the DDBJ/ENA/GenBank under the Bioproject accession number PRJNA1135503.

Ethics approval

All animal work was approved by the Ethics Committee of Renji Hospital, School of Medicine, Shanghai Jiao Tong University, Shanghai, China (Approval Number: RJ2023-088A).

Results

Longitudinal analysis of clonal prevalence, antimicrobial susceptibility, and in vitro biofilm formation in ST239 and ST5 MRSA isolates

A total of 3,410 sequential, non-repetitive MRSA isolates were collected and analysed in this study, including 241 isolates in 2005, 643 in 2008, 349 in 2010, 414 in 2011, 301 in 2012, 340 in 2015, 268 in 2016, 229 in 2017, 188 in 2018, 244 in 2019 and 193 in 2020. Among these, 810 isolates belonged to ST239 and 1,615 to ST5. From 2008 to 2020, we observed a declining trend in the proportion of MRSA among all S. aureus isolates, decreasing from 83.51% to 45.52% (Figure 1(A)). Throughout the study period, ST5 remained consistently prevalent, accounting for over 30% of MRSA isolates each year from 2008 to 2019, although its prevalence dropped to 13.09% in 2020, possibly due to the impact of the COVID-19 pandemic. In contrast, the proportion of ST239 isolates declined markedly, falling from 34.85% in 2005 to 24.35% in 2010, then dramatically to 7.35% in 2015, and further to just 0.52% in 2020 (Figure 1(A)). Dynamical changes in the geometric mean MICs of vancomycin and linezolid for 810 ST239 isolates and 1,615 ST5 isolates are shown in Figure 1(B). For ST239, the geometric mean MIC of vancomycin, a glycopeptide antibiotic, decreased from 1.26 ± 0.26 mg/L in 2005 to 0.50 mg/L in 2020. While the MIC values for ST5 isolates remained relatively stable over the same period, ranging from 0.86 ± 0.36 mg/L to 0.94 ± 0.32 mg/L. Both ST239 and ST5 isolates demonstrated a slight decrease in susceptibility to linezolid over the study period, with the geometric mean MIC increasing from 0.97 ± 0.27 mg/L to 1.00 mg/L for ST239 isolates, and from 0.97 ± 0.47 mg/L to 1.17 ± 0.24 mg/L for ST5 isolates.

Figure 1.

Longitudinal trends in prevalence, antibiotic susceptibility, and biofilm formation of ST239 and ST5 MRSA isolates from 2005 to 2020. A: Temporal dynamics of MRSA clone distribution among S. aureus isolates collected from a Chinese teaching hospital over a 15-year period. Green denotes the overall proportion of MRSA (n = 3,410) among S. aureus isolates; red and blue indicate the relative frequencies of ST239 and ST5 among MRSA, respectively. Dotted lines depict regression trends. B: Average minimum inhibitory concentrations (MICs) of vancomycin and linezolid for ST239 (n = 810) and ST5 (n = 1,615) isolates. Dotted lines indicate temporal trends in mean MIC values; red and blue columns represent ST239 and ST5, respectively. C: Biofilm-forming capacity of ST239 (n = 63) and ST5 (n = 70) isolates, assessed via crystal violet staining and quantified by OD₄₇₀. Red and blue dots represent ST239 and ST5 strains, respectively; grey dots represent sterile medium controls. D: Temporal changes in biofilm formation capacity stratified by year of isolation for both ST239 and ST5 isolates. Precise P-values are provided in the figure, with “ns” indicating no statistically significant difference (P > 0.05).

To investigate the factors underlying the decline in ST239 prevalence and its increased susceptibility to vancomycin, we analysed a total of 133 MRSA isolates. A stratified sampling strategy was employed to ensure temporal representation, comprising 20 isolates of each ST from 2005, 2010, and 2015, and 1 ST239 plus 5 ST5 isolates from 2020. Owing to the limited availability of ST239 strains in 2020, an additional 2 ST239 and 5 ST5 isolates from 2019 were included, yielding a final dataset of 63 ST239 and 70 ST5 isolates. Based on the observation that these ST239 and ST5 isolates exhibited comparable in vitro growth characteristics (Supplementary figure 1), we first evaluated their biofilm formation abilities. Overall, there was no significant difference in biofilm formation between these ST239 and ST5 isolates (Figure 1(C)). However, when stratified by year of isolation, ST239 isolates demonstrated a marked decline in biofilm formation over time, whereas ST5 isolates showed a significant increase (P = 0.037; 95% CI, −0.2693 to −0.0086) (Figure 1(D)). In 2005, ST239 strains exhibited significantly higher biofilm formation compared to ST5 strains (OD₄₇₀: 0.42 ± 0.24 vs. 0.14 ± 0.06, P < 0.001). However, this trend was reversed by 2015 (OD₄₇₀: 0.25 ± 0.17 for ST239 vs. 0.47 ± 0.42 for ST5, P = 0.048) and remained consistent in 2019 and 2020 (OD₄₇₀: 0.17 ± 0.05 for ST239 vs. 0.54 ± 0.33 for ST5, P = 0.049) (Due to the limited number of isolates collected in 2019 and 2020, strains from these two years were combined for statistical analysis).

Genomic evolution of ST239 and ST5 MRSA isolates

To investigate the underlying mechanism of this phenomenon, we analysed the genomic information of these 126 bacterial strains. Comprehensive information on these isolates is provided in Table S1, including ST, spa type, SCCmec type, source of the original samples, and antimicrobial susceptibility profiles for multiple antibiotics. Phylogenetic analysis was performed on the 126 MRSA strains, and their associated different virulence and antimicrobial resistance genes were identified (Figure 2). ST239 isolates were predominantly associated with spa types t030and t037, whereas ST5 isolates exhibited greater diversity, including spa types t601, t2460, t002, t311, and others. ST239 strains isolated in 2005 were mainly t037, which was largely replaced by t030 by 2010. In contrast, the spa types of ST5 strains showed no clear temporal association. ST239 could be resolved into two phylogenetic sub-lineages which we term for reference ST239-A, ST239-B. Comparing these to more traditional spa typing points to a rough agreement between sub-lineages and dominant spa type (Clade ST239-A: t030, ST239-B: t037). Similarly, ST5 had four major sub-lineages (Clade ST5-A: other spa types, ST5-B: t601 and t2460, ST5-C: t002 and t311, ST5-D: t002 and t601), though with some within sub-lineage diversity in spa type assignment (Figure 2).

Figure 2.

Phylogenetic analysis of ST239 and ST5 MRSA isolates. Maximum-likelihood phylogenetic trees were constructed based on whole-genome SNPs from 61 ST239 and 65 ST5 MRSA isolates. For each tree, the outer rings annotate spa types, year of isolation, and presence or absence of representative virulence and antimicrobial resistance genes that differ between the two lineages.

Genomic analysis revealed distinct patterns of virulence and antimicrobial resistance genes among different STs. Overall, ST239 strains harboured a higher prevalence of antimicrobial resistance genes, while ST5 strains carried a greater number of virulence genes. A total of 73 virulence genes and 17 resistance genes were identified in ST239, while 80 virulence genes and 16 resistance genes were found in ST5. These virulence factors were associated with bacterial adherence, effector delivery systems, exoenzymes, exotoxins, and immune modulation. Genes related to adherence and immune modulation were more frequently identified in ST239 strains than in ST5. For example, the carriage rates of clfA and clfB in ST239 were 60.66% (37/61) and 90.16% (55/61), respectively, compared to only 1.54% (1/65) and 61.54% (40/65) in ST5. Additionally, eap/map genes were present in ST239 but absent in ST5. Regarding immune modulation genes, ST239 carried the cap8H-K genes, which was not found in ST5, whereas chp was absent in ST239 but present in 49.23% (32/65) of ST5 isolates. In contrast, ST5 carried more effector delivery system genes, including essC, esxC, esxB, esaE, esxD, and esaD, which were not found in ST239. Differences were also observed in exoenzyme and exotoxin gene profiles: ST239 carried vWbp but lacked aur, while ST5 showed the opposite pattern. ST239 did not carry seb, sec, or sell, but had higher carriage rates of selk (81.97%, 50/61) and selq (85.25%, 52/61). The tsst-1 gene was absent in ST239, while present in 87.69% (57/65) of ST5 isolates. As for resistance genes, aadD was more frequently detected in ST5 (41.54%, 27/65) than in ST239 (1.64%, 1/61). However, with the exception of lnu(G), lnu(A), and bleO, other resistance genes were more commonly carried by ST239 strains.

Different sub-lineages within the same ST also exhibited distinct patterns of virulence and resistance genes. For instance, only 4.76% (1/21) of ST239-A isolates carried the clfA gene, whereas 90% (27/30) of ST239-B isolates were clfA-positive (clfA+). Moreover, ST239-A strains did not harbour the resistance genes, including ant(6)-Ia, aph(3’)-III, tet(K) and dfrG. Within the ST5 lineage, clfB was predominantly detected in ST5-B and ST5-D sub-lineages, while the absence of sdrC-E genes was mostly observed in ST5-D. Besides, ST5-A strains carried fewer resistance genes compared to other ST5 sub-lineages.

ST239 MRSA have a temporal trend of increased genetic mutations

To further investigate genomic changes in the MRSA strains, we conducted the comparative genomics and analysed the number of single nucleotide variants (SNVs), including SNVs differing from the reference genome (SNV_noref) and single nucleotide polymorphisms (SNPs). In ST239 isolates, the average number of SNVs and SNPs per genome increased significantly from 188.43 and 90.68 in 2005 to 549.29 and 447.59 in 2010, coinciding with the initial marked decline of ST239. These values remained relatively stable in 2015, at 498.00 and 405.80, respectively. In contrast, the corresponding numbers in ST5 isolates remained relatively stable at approximately 280 throughout the same period (Figure 3(A)). Further analysis of SNPs revealed a similar trend in both synonymous and nonsynonymous SNPs (Figure 3(B)). The heatmap illustrated SNP differences among the isolates over time (Figure 3(C)), showing that ST239 strains harboured a significantly higher number of SNPs compared to ST5, particularly in year 2010 and 2015.

Figure 3.

Temporal genomic evolution of ST239 and ST5 MRSA isolates. A: Temporal comparison of single nucleotide variants (SNVs) in ST239 and ST5 isolates from 2005, 2010, and 2015, including total SNVs (SNV_noref) and SNPs relative to the reference genome. B: Comparison of synonymous (purple) and non-synonymous (yellow) mutations in ST239 and ST5 isolates across the same timepoints. C: Heatmap illustrating the temporal trend of total SNPs per genome in individual ST239 and ST5 strains, with SNP counts increasing from blue to red; isolation years are indicated by colour-coded bars. D: Gene Ontology (GO) enrichment analysis of mutated genes in 61 ST239 genomes from 2005, 2010, and 2015, showing pathway-level functional enrichment across time. E: Detection frequency of the tarJ’ (C665T) mutation in ST239 isolates over time (blue line), plotted alongside the overall ST239 isolation rate (red line); a dotted line indicates the trend.

Further Gene Ontology (GO) analysis identified 22 genes with increasing non-synonymous mutation frequencies from 2005 to 2010 in ST239 isolates including tarJ’, dltA, femA, isdE, ezrA, atpC, ctaB, hssS, and prsA. These genes were enriched in pathways associated with the cell wall and cell membrane (n = 9), cytoplasm (n = 6), chromosome (n = 5), and ribosome (n = 2) (Figure 3(D)). Since most of the mutated genes were associated with the cell wall and cell membrane, we hypothesized that such mutations might be detrimental to the evolution of ST239. To investigate whether similar changes occurred in ST5, we analysed the mutation frequencies of genes involved in cell wall and membrane function. Unlike ST239, no polarity-altering nonsynonymous mutations in tarJ’ were detected in ST5 isolates. Then, we analysed the mutated gene of tarJ’ in 810 non-repetitive ST239 MRSA isolates collected in this study. The proportion of tarJ’ mutations in ST239 showed a dramatic increase coinciding with the clonal decline, rising from 2.38% in 2005 to 61.18% in 2010. This proportion remained relatively stable at approximately 40% during the subsequent years (2016–2018), and eventually reached 100% in 2020 (Figure 3(E)). Besides, we found that ST239 strains harbouring the tarJ’ mutation formed a distinct Clade (ST239-A) and appeared to have replaced those lacking the mutation (ST239-B) based on the phylogenetic analysis (Supplementary Figure 2). Therefore, we next focused on the role of the tarJ’ mutation in this study.

Identification and functional validation of tarJ’ mutation as a driver of evolution

The tarJ’ gene encodes ribulose-5-phosphate reductase 2, a homologous gene of tarJ that catalyses the NADPH-dependent reduction of D-ribulose-5-phosphate to D-ribitol-5-phosphate, representing a key step in wall teichoic acid (WTA) biosynthesis [21–23] (Supplementary figure 3A). WTAs are critical for bacterial pathogenesis and physiological functions such as cell division and cell wall homeostasis. The tarJ’ mutation identified in ST239 involves a cytosine-to-thymine substitution at nucleotide position 665 (C665T), resulting in a threonine (Thr) to isoleucine (Ile) change at amino acid 222 (T222I). Structural analysis using DUET predicted this mutation to be destabilizing (ΔΔG = －0.214 kcal/mol), likely due to disruption of hydrogen bonding by threonine and altered local interactions caused by the more hydrophobic isoleucine (Supplementary figure 3B-3D). To assess the functional impact of this mutation, isogenic ST239 tarJ’ (C665T) and ST5 tarJ’ (C665T) mutants were constructed using 2005-ST239-1 and 2005-ST5-1 strains, both of which harbour the wild-type tarJ’ allele.

Given the critical role of tarJ’ in WTA biosynthesis, we first examined the cell wall morphology of S. aureus isolates using transmission electron microscopy (TEM). Compared to the ST239 wild-type (ST239 WT), the tarJ’ mutant (C665T) exhibited disrupted and lysed cell walls with rough surfaces and numerous protrusions (Figure 4(A)). Similar structural alterations were observed in the ST5 mutant strain (Supplementary figure 4). While the mutation did not affect bacterial growth under normal conditions, both ST239 and ST5 tarJ’ (C665T) mutants showed significantly reduced growth under acidic conditions at 8 h (P = 0.013, 95% CI: −0.9449 to −0.2045 for ST239; P = 0.004, 95% CI: −0.8983 to −0.3171 for ST5) (Figure 4(B)). Under high-concentration vancomycin stress (50×MIC for 3 h), survival rates of the mutant strains were markedly lower than those of their respective wild-type counterparts (P = 0.013, 95% CI: −0.2683 to −0.05571 for ST239; P = 0.047, 95% CI: −0.1535 to −0.001862 for ST5) (Figure 4(C)). MIC assays further demonstrated increased susceptibility of the tarJ’ mutants to vancomycin and teicoplanin (vancomycin MIC reduced from 1 to 0.5 mg/L; teicoplanin from 0.5 to 0.25 mg/L), while linezolid susceptibility remained unchanged (Figure 4(D)). Notably, biofilm formation was significantly impaired in the mutant strains, as quantified by crystal violet staining (P < 0.0001, 95% CI: −0.8636 to −0.4339 for ST239; P = 0.001, 95% CI: −0.7160 to −0.3065 for ST5) and further confirmed by confocal microscopy (Figure 4(E)).

Figure 4.

In vitro functional consequences of the tarJ’ (C665T) mutation in ST239 and ST5 MRSA isolates. A: Transmission electron microscopy (TEM) images showing altered cell wall morphology in tarJ’-mutant strains compared to their wild-type counterparts in ST239 backgrounds. B: Growth curves of wild-type and tarJ’-mutant ST239 and ST5 strains under normal and low pH conditions. C: Survival rates of wild-type and mutant strains following 0, 3, and 6 h of high-concentration (50×MIC) vancomycin exposure. D: Minimum inhibitory concentrations (MICs) of linezolid, teicoplanin, and vancomycin against wild-type and tarJ’-mutant ST239 and ST5 strains. E: Quantification of biofilm formation by crystal violet staining in wild-type and mutant isolates. Confocal laser scanning microscopy (CLSM) images show biofilm architecture after staining with propidium iodide (PI) and SYTO9. Images were captured at 40× magnification. Precise P-values are provided in the figure, with “ns” indicating no statistically significant difference (P > 0.05).

After assessing the impact of the tarJ’ mutation on S. aureus isolates under in vitro conditions, we next examined its effects under in vivo conditions. In the polymorphonuclear neutrophil leukocyte (PMN) killing assay, the tarJ’ (C665T) mutant strains exhibited a trend towards reduced resistance to neutrophil-mediated killing at 60 min, with a statistically significant decrease observed at 90 min (P = 0.02 for ST239; P = 0.01 for ST5) (Figure 5(A)). To further investigate pathogenicity in vivo, we employed a murine catheter-associated infection model (Figure 5(B)). Mice infected with tarJ’ (C665T) mutant isolates displayed attenuated pathogenic features compared to those infected with wild-type strains, including decreased dermal and submucosal thickening and reduced inflammatory cell infiltration, as assessed by hematoxylin and eosin (H&E) staining (Figure 5(C)). At 7 days post-infection, colony-forming units (CFU) from both catheter surfaces and surrounding skin tissues (SST) were quantified. Mutant strains exhibited significantly diminished adherence to catheters (P = 0.002 for ST239; P = 0.047 for ST5) and reduced bacterial loads in SST (P = 0.037 for ST239; P = 0.049 for ST5) (Figure 5(D)). These findings suggest that the tarJ’ (C665T) mutation impairs S. aureus fitness in host environments, potentially contributing to the epidemiological decline of the ST239 lineage.

Figure 5.

Impact of the tarJ’ (C665T) mutation on bacterial fitness in host environments. A: Resistance of biofilms formed by wild-type and tarJ’-mutant ST239 and ST5 strains to polymorphonuclear neutrophil (PMN)-mediated killing, assessed by CFU quantification following 0, 30, 60 and 90-minute co-incubation. B: Schematic illustration of the mouse subcutaneous catheter infection model. C: Hematoxylin and eosin (H&E) staining of skin tissues collected from mice infected with wild-type or tarJ’-mutant ST239 and ST5 isolates. D: Bacterial burden in catheters and adjacent skin tissue (SST) measured by CFU enumeration 7 days post-infection. Precise P-values are provided in the figure, with “ns” indicating no statistically significant difference (P > 0.05).

Discussion

A significant global decline in ST239 MRSA prevalence has been widely documented. Previous hypotheses have attributed this decline to competitive displacement by CA-MRSA lineages such as ST59 harbouring chp in China [14] or dual HA- and CA-MRSA lineages such as ST22 through enhanced desiccation tolerance and faster growth rates across from Singapore and Australia to Malaysia [24]. While the chp-mediated niche competition explains CA-MRSA expansion, it fails to elucidate the intra-hospital displacement between HA-MRSA clones such as ST239 and ST5 in China.

Consistent with previous reports, our longitudinal genomic and phenotypic analyses of ST239 and ST5 isolates reveal that clinical selection pressures have driven bacterial evolution, with fitness loss contributing to the failure of ST239 to maintain its ecological niche. This supports the idea that fitness costs played an important role in the demise of this once-dominant pathogen lineage [25]. First, the differential patterns between ST239 sub-lineages illuminate this mechanism. Clade ST239-B, characterized by a higher burden of resistance genes – including near-universal carriage of clfA and multiple resistance operons – underwent more pronounced genomic decay than Clade ST239-A, in which only 4.8% of isolates carried clfA. This observation aligns with evolutionary models suggesting that excessive accessory gene load in genetically static lineages accelerates fitness decline. In contrast, ST5 maintained a “virulence-adapted” genomic profile, harbouring significantly more virulence genes (80 in ST5 vs. 73 in ST239), including effector delivery system components (essC, esxC) that facilitate immune evasion, widespread exotoxins (tsst-1 in 87.7% of isolates), and conserved adherence factors such as clfB. This balanced genomic architecture enabled ST5 sub-lineages – particularly Clade ST5-D, which retained both clfB and tsst-1 – to persist under clinical selective pressures without accumulating deleterious mutations in core functional pathways.

Subsequent investigations revealed that the decline of ST239 is driven by the accumulation of mutations that impose fitness costs, particularly within cell wall biosynthesis pathways. A pivotal discovery was the identification of a tarJ’ (C665T; Thr222Ile) point mutation, whose prevalence increased dramatically from 2.4% in 2005 to 100% in 2020 among ST239 isolates – mirroring the clone’s epidemiological collapse. Phylogenetic analysis further showed that ST239 strains harbouring the tarJ’ mutation formed a distinct Clade (ST239-A), which appeared to have replaced the ancestral Clade (ST239-B). In contrast, ST5 maintained stable prevalence, likely due to the absence of comparable mutations in WTA or peptidoglycan biosynthesis genes, thereby preserving core cell wall functions. We further analysed publicly available ST239 genome sequences deposited in NCBI from other countries and regions of China (Supplementary Table 4). We found that the tarJ’ C665T mutation was absent in the limited isolates from Singapore, but was detected in isolates collected after 2010 from several regions in China (23/33, 69.70%) (e.g. Beijing, Chongqing, Henan, Guangdong). These findings suggest potential geographic heterogeneity in the occurrence of this mutation. Future large-scale comparative genomic studies incorporating international collections will be essential to establish whether the tarJ’ C665T mutation contributed more broadly to the global decline of ST239 or represents a regionally restricted evolutionary event.

The Thr222Ile substitution in TarJ’ was predicted to destabilize the protein (ΔΔG = –0.214 kcal/mol), likely impairing its enzymatic function in converting D-ribulose-5-phosphate to D-ribitol-5-phosphate, a key precursor in WTA biosynthesis [21]. Functionally, this mutation reproduced hallmark phenotypes of ST239 decline, including significantly reduced biofilm formation, attenuated viability in acidic conditions, increased vancomycin susceptibility, and compromised cell wall structure, as observed via transmission electron microscopy. Moreover, tarJ’-mutant strains displayed attenuated pathogenesis in vivo, with significantly reduced catheter adhesion and impaired resistance to neutrophil-mediated killing. These results demonstrate that the tarJ’ mutation disrupts WTA biosynthesis and compromises cell wall integrity, biofilm development, and antibiotic resistance, which are traits essential for persistence in healthcare environments. In contrast, the continued success of ST5 is underpinned by genomic stability and the absence of deleterious mutations in critical physiological pathways.

Our study highlights the delicate balance between resistance acquisition and genomic resilience in MRSA clones. The historical dominance of ST239 was likely driven by the acquisition of SCCmec type III and a streamlined genome that enabled rapid clonal expansion [26]. The paradoxical fixation of the tarJ’ C665T mutation, despite its apparent fitness cost, may be attributed to genomic decay in the recombination-limited ST239 lineage, stochastic processes such as genetic drift or bottlenecks, and transient advantages under specific hospital conditions [10, 16, 22, 25]. In contrast, ST5’s larger and more dynamic genome, incorporating diverse SCCmec types and accessory genes, provided the flexibility to buffer fitness costs while maintaining resistance, thereby supporting long-term persistence [27]. Once fitter clones such as ST5 emerged, ecological competition likely hastened the decline of ST239 [14, 25]. This contrast underscores a broader evolutionary principle: lineages with limited genomic plasticity are prone to erosion, whereas those with greater adaptability sustain clonal success. Clinically, tarJ’ mutations may serve as biomarkers of clonal vulnerability, and inhibition of wall teichoic acid biosynthesis [28]. Moreover, the reduced biofilm formation observed in tarJ’-mutant strains suggests increased susceptibility to environmental disinfection, presenting an exploitable trade-off for infection control.

In this study, the functional role of the tarJ’ mutation was demonstrated only in ST239. Further investigations are required to determine whether this mutation also contributes to the decline of other MRSA clones and to assess the potential presence of compensatory mutations that may transiently restore bacterial fitness. In addition, the impact of host microenvironmental factors – including pH, osmotic pressure, and immune stress – on exacerbating tarJ’-associated defects remains to be clarified. Finally, the other eight genes associated with the cell wall and cell membrane identified in our study were not fully characterized and thus represent promising targets for future investigation. Incorporating genomic decay metrics into antimicrobial resistance surveillance could enhance our ability to predict clonal trajectories and inform precision interventions, particularly against genomically stable lineages like ST5. Taken together, our findings establish tarJ’ mutation as a critical driver of fitness loss in ST239, providing a mechanistic basis for its epidemiological decline and underscoring the divergent evolutionary strategies that shape clonal persistence versus collapse.

Funding Statement

This study was supported by National Natural Science Foundation of China [grant number 82472336 (to L.H), 82272395 (to LH), 81974311 (to LH), and W2411070 (to ML)], the Ningbo Natural Science Foundation Ningbo [grant number 2024J419 (to LS)]. The funders had no role in the study design, collection, analysis and interpretation of data, writing of the report, or in the decision to submit the article for publication.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

All data generated or analysed during this study are included in this published article or in the supplementary information files.

Supplemental Material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/22221751.2025.2576582.