Dynamic Profiling of Penicillin‐Binding Protein 2a (PBP2a)‐Positive Extracellular Vesicles: Implications for Early Diagnosis and Treatment Monitoring of Methicillin‐Resistant Staphylococcus Aureus Infections

ABSTRACT

Infections caused by methicillin‐resistant staphylococci (MRS), such as methicillin‐resistant Staphylococcus aureus (MRSA), pose significant challenges to public health. The early detection of MRS infections and monitoring of antibiotic resistance profiles are critical for patient management and infection control strategies. Extracellular vesicles (EVs) have emerged as promising biomarkers in infectious disease. By combining single‐particle nano‐flow cytometry and immunoelectron microscopy (immuno‐TEM), we discovered that PBP2a is present on the surface of EVs extracted from MRS. However, whether PBP2a can serve as an EVs‐associated protein marker for diagnosing bacterial infections remains unexplored. Using MRSA as a model strain, mouse models of localized and systemic infections were established, alongside a clinical cohort study, to investigate the dynamics of PBP2a‐positive (PBP2a⁺) EVs in plasma following bacterial infection, infection progression, and in response to antimicrobial therapy. In mouse infection models, PBP2a⁺ EVs were detected in plasma, with variable detection rates observed across different infection models. The study found a progressive correlation between increasing plasma PBP2a⁺ EVs levels and non‐specific inflammation markers (CRP, IL‐6) during infection progression. Antimicrobial therapies, however, inversely affected the ratio of PBP2a⁺ EVs. Furthermore, a clinical cohort study confirmed a direct association between the magnitude of PBP2a⁺ EVs in the plasma of patients with MRSA infection and the severity of infection. The investigation highlights the potential of PBP2a⁺ EVs as plasma biomarkers of MRSA antibiotic resistance, particularly during the early stages of resistant infections. Their persistence in plasma throughout the infectious episode makes them valuable indicators for monitoring disease progression and evaluating the efficacy of antibiotic treatments.

1. Introduction

Bacterial infections, compounded by the rise of antibiotic‐resistant strains, present a formidable challenge to public health (Kadri et al. 2021; Wagenlehner et al. 2024; Wong et al. 2023). Among the diverse bacterial species, MRS, characterized by their unique cell wall composition and declining susceptibility to antibiotics, significantly exacerbates the global burden of infectious diseases (Howden et al. 2023; Severn and Horswill 2022). Staphylococcus aureus, particularly its methicillin‐resistant variant, is of utmost concern due to its widespread prevalence and clinical impact (Tan et al. 2024). Traditional diagnostic approaches, such as bacterial culture and antimicrobial susceptibility testing, suffer from prolonged detection periods and lower sensitivity (Chen et al. 2023). Molecular detection technologies, such as PCR and metagenomic next‐generation sequencing (mNGS), have substantially reduced detection timeframes and improved sensitivity but are still hindered by sample variability and nucleic acid stability issues (Chiu and Miller 2019; Ye et al. 2019; Bashir et al. 2012). Additionally, these methods only provide qualitative evidence of bacterial presence and drug resistance, without reflecting infection severity or treatment effectiveness (Church et al. 2020; Sinha et al. 2018). Therefore, there is an urgent need for new methods that can be used for early diagnosis, therapeutic monitoring, and prognostic assessment of MRS infections.

Bacterial extracellular vesicles, referred to as BEVs (Toyofuku et al. 2023; Juodeikis and Carding 2022; Toyofuku et al. 2019; Wen et al. 2023), have attracted considerable attention due to their pivotal roles in intercellular communication and pathogenesis (Aytar et al. 2022). BEVs encapsulate a cargo that mirrors the phenotypic characteristics of the pathogen, modulate immune responses, and contribute to bacterial survival (Bitto et al. 2021; Briaud and Carroll 2020). The prospect of utilizing BEVs as a non‐invasive reservoir of pathogen‐specific signatures holds significant promise for the development of highly sensitive diagnostic approaches (Xie et al. 2022; Xie, Cools, et al. 2023). BEVs are inherently very small and complex membrane structures, and directly detecting their physical properties is insufficient to confirm their origin or function (Long et al. 2022). The detection of BEVs typically relies on signature proteins, nucleic acids, or other biomarkers located on the vesicle surface or within their contents (Shao et al. 2018). Through the identification of specific markers on the EVs, it becomes feasible to elucidate the informational content and functional roles conveyed by EVs. Here, we focus on the bacterial‐associated markers in the plasma of patients infected with methicillin‐resistant staphylococci (MRS) strains that are widespread in clinical practice. The resistance of these strains is primarily attributed to the acquisition of the mecA gene, which encodes penicillin‐binding protein 2a (PBP2a), conferring resistance to all beta‐lactam antibiotics, including methicillin (Alexander et al. 2023). The detection of PBP2a holds paramount importance in confirming methicillin resistance within bacterial strains (Harrison et al. 2019). By shifting our focus to EVs, we aim to harness this distinctive biological characteristic for the diagnosis of MRS infections. Employing advanced detection methods, such as single‐particle nano‐flow cytometry and super‐resolution immuno‐TEM, we characterized EVs extracted from MRS, with a particular focus on the localization and expression of PBP2a on the surface of EVs. MRSA, as a primary cause of opportunistic infections in clinical settings, carries the mecA gene in the vast majority of its strains. This gene not only profoundly reveals the species specificity of MRSA but also plays a crucial role in encoding the core target protein PBP2a. Studies have conclusively demonstrated the feasibility and high efficiency of rapid PCR detection techniques for the mecA gene in clinical practice to diagnose MRSA infections (Tissari et al. 2010; Kou et al. 2024; Snitser et al. 2020). Additionally, multiple studies have emphasized the significant potential of PBP2a as a marker for MRSA (Neil et al. 2021; Mandal et al. 2014). Therefore, we have selected MRSA as a representative of MRS to explore the practical application value of PBP2a⁺ EVs in the diagnosis of MRSA infections. We constructed mouse models of localized and systemic MRSA infections to closely mimic the natural progression of the infection and to provide a controlled environment for our studies. The primary objective of this research is to investigate the dynamics of PBP2a⁺ EVs in the plasma of mice during MRSA infection and treatment. By monitoring the levels and dynamics of PBP2a⁺ EVs, we aim to determine whether these PBP2a⁺ EVs can serve as reliable biomarkers for the diagnosis and monitoring of MRSA infections. We also advance a clinical cohort study analysing PBP2a⁺ EVs dynamics in human plasma across different infection phases, evaluating their diagnostic value in a clinical setting. Crucially, we conduct a long‐term study of MRSA bloodstream infection patients, meticulously recording PBP2a⁺ EV fluctuations during antibiotic therapy, to validate their effectiveness and sensitivity as biomarkers for therapy outcomes and predicting infection alleviation trends.

This study proposes the strategy of utilizing PBP2a⁺ EVs as a marker for MRSA. By detecting the expression of PBP2a on EVs, specific identification and diagnosis of MRSA can be achieved. This approach offers not only a high degree of specificity and sensitivity but also facilitates real‐time monitoring and assessment of disease status through the non‐invasive collection of bodily fluid samples. Therefore, the aim of this research is to explore the feasibility of using PBP2a⁺ EVs as a marker for MRSA, providing new perspectives and methodologies for the early diagnosis and therapeutic efficacy evaluation of MRSA.

2. Materials and Methods

2.1. Bacterial Strains and Culture Conditions

The following bacterial strains were used in this study: MRSA sequence type ST5, ST398, ST1, ST59 and ST188, and methicillin sensitive Staphylococcus aureus (MSSA, ATCC25923). Additionally, methicillin‐resistant Staphylococcus epidermidis (MRSE) and methicillin‐resistant Staphylococcus haemolyticus (MRSH) isolates recovered from patients with bloodstream infections at Shanghai Renji Hospital were included. Escherichia coli (E. coli, ATCC25922) was also included as a control. Bacteria were generally cultured in Luria‐Bertani (LB; Sigma; E. coli) or tryptic soy broth (TSB; Oxoid; all other bacteria) with shaking at 200 rpm at 37°C.

2.2. Isolation of BEVs

The bacterial culture supernatant was subjected to density gradient centrifugation to isolate BEVs, as previously described by Lee et al. (2009). Overnight cultures were diluted at a ratio of 1:100 and incubated in 200 mL of fresh TSB/LB medium at 37°C till OD600 of 0.4–0.5. The supernatant was collected and centrifuged at 4000 × g for 15 min and further filtered through a 0.22 µm filter (Millipore, #SLGPR33RB, USA) to remove suspended bacterial cells. The filtrate was then concentrated to 25 mL via ultrafiltration at 6000 × g for 15 min using a 50 kD Amicon Ultra filter (Millipore, #UFC9100, USA). Subsequently, crude BEVs were obtained by ultracentrifugation at 174,900 × g at 4°C for 3 h using a SW 32Ti rotor (Beckman Coulter). The resulting pellets were resuspended in 4.4 mL of a 50% OptiPrep solution (Sigma, #D1556, Germany) and layered at the bottom of 4 mL of 40% OptiPrep solution and 1.6 mL of 10% OptiPrep solution. The mixture was subjected to centrifugation at 200,000 × g for 15 h at 4°C in an SW 41Ti rotor, and 500 µL of the visible band between the 10% and 40% gradients were collected. The BEVs were stored at −80°C.

2.3. EVs Isolation From Human/Mouse Plasma

Blood plasma samples were collected from healthy individuals and patients with different types of infections at Renji Hospital, Shanghai. There were no statistically significant differences in age or gender among all enrolled participants in the study. Plasma samples from healthy individuals (n = 20) were collected for the control group. The MRSA‐infected patients were sampled from different sites, with 20 cases collected from each specific site, including the respiratory tract, tissue/secretions, and bloodstream. Another 20 patients who suffered from MSSA bloodstream infections were included in the study.

To isolate EVs from plasma, a differential centrifugation method was employed (Dong et al. 2020; Nieuwland and Siljander 2024). Whole blood samples (5 mL from human venous blood/1 mL from mouse venous blood) were collected in EDTA (ethylenediaminetetraacetic acid)‐coated tubes (Becton, Dickinson and Company, #367841, USA). Each blood sample was processed according to the following procedure: The plasma was separated through centrifugation at 1000 × g for 10 min, and two additional centrifugations were conducted at 2500 × g for 15 min to remove cells, large debris, and platelets. Subsequently, the supernatant was centrifuged at 10,000 × g at 4°C for 20 min to eliminate apoptotic bodies, other large EVs, and certain non‐EV components. To further reduce the presence of lipoproteins, which are the predominant plasma proteins, the supernatant was filtered using a 0.45 µm pore size filter (Sigma, #SLHPR33RB, Germany) (Guo et al. 2021). The filtrate was then transferred to a new tube and underwent ultracentrifugation at 120,000 × g at 4°C for 2 h. The resulting pellet was washed with phosphate buffered saline (PBS) and ultracentrifuged again at 120,000 × g at 4°C for 2 h. The EV pellet was resuspended in 50 µL of PBS and stored at −80°C for further studies.

2.4. Capture and Elution of PBP2a⁺ EVs

For the capture of PBP2a⁺ EVs, streptavidin‐coated magnetic beads (Thermo Fisher Scientific, #65001, USA) were first prepared by washing them twice with PBS for 10 min each to remove the storage buffer. Subsequently, biotinylated PBP2a antibody (Thermo Fisher Scientific, #MA5‐46569, USA) was mixed with the streptavidin‐coated magnetic beads at a ratio of 1:10 and incubated at room temperature for 30 min. The magnetic beads were then washed with PBS for 10 min to remove any unbound antibody. Next, the resuspended EVs were mixed with the PBP2a antibody‐conjugated magnetic beads and incubated at room temperature for 1 h. The magnetic beads were separated using a magnetic rack, and the supernatant was discarded. The beads were then washed three times with PBS for 10 min each to remove any unbound EVs.

For the elution of PBP2a⁺ EVs, the magnetic beads were resuspended in 100 µL of elution buffer (0.1 M citrate buffer, pH 3.0) and incubated at room temperature for 5 min with gentle agitation to facilitate the elution of EVs from the magnetic beads. The magnetic beads were separated using a magnetic rack, and the supernatant containing the eluted PBP2a⁺ EVs was collected. To neutralize the low pH of the elution buffer, the eluate was immediately diluted with an equal volume of neutralization buffer (1 M Tris‐HCl, pH 8.0). Finally, the eluted EVs were washed once with PBS by ultracentrifugation at 120,000 × g for 1 h to remove any residual elution buffer and biotin. The purified PBP2a⁺ EVs were collected and used for subsequent analyses.

2.5. Nano‐Flow Cytometry

The EVs were labelled with fluorescent antibodies before performing the nano‐flow cytometry assay. In the experimental group, the EVs were initially incubated with a mouse polyclonal antibody against PBP2a (Life Span Biosciences, #LS‐C343196, USA) at 37°C for 1 h. In the control group, an isotype control antibody IgG (Life Span Biosciences, #LS‐C292295, USA) was used. Following incubation, the EVs were centrifuged at 4°C using the SW55Ti rotor of the Beckman Coulter XE‐90K ultracentrifuge at 120,000 × g for 1 h. This was repeated to remove any unbound antibodies. The EVs were then resuspended in 100 µL of PBS and incubated with Alexa Fluor 488 labelled secondary antibodies (Abcam, #ab150113, UK) at 37°C for 1 h. After incubation, the EVs were washed twice under the same centrifugation conditions to eliminate any free fluorescent antibodies. Finally, the supernatant was discarded, and the EVs were resuspended in 20 µL of PBS in preparation for analysis by nano‐flow cytometry.

In the nano‐flow cytometry assay, the sheath fluid flows at a rate of approximately 40pL/min, while the sample volume flows at approximately 2nL/min. The detection zone volume, defined by the intersection of the laser focal spot (16 µm diameter) and the sample flow (1.4 pm diameter), measures 25 fL. To establish a standard curve, the particle count corresponding to the theoretical concentration of 100 nm polystyrene fluorescent standard beads is determined. The EV samples are analysed using the nano‐flow cytometry assay with the same instrument settings.

2.6. Western Blot Analysis

EVs from MRSA, MSSA, MRSE, MRSH, E. coli and plasma were lysed in cold RIPA lysis buffer (Millipore, #20‐188, USA) containing PMSF (Sigma, #78830, Germany) for 20 min on ice. Prior to conducting Western Blot analysis, we measured the concentration of all protein samples using the BCA Protein Assay Kit (Beyotime). The concentration of each sample was determined through a standard curve, and the loading amount was adjusted according to the experimental design to ensure consistent total protein amounts across all lanes. To detect the expression levels of PBP2a protein under different conditions, we adjusted the loading amount of each sample to 20 µg based on the protein concentration. This loading range was optimized in preliminary experiments to ensure clear bands and moderate signal intensity. In the experiment detecting human‐derived markers (TSG101, β‐actin) in PBP2a⁺ EVs, due to the lower amount of EVs in the samples, we loaded a total protein amount of 7.5 µg per lane. EV lysates were denatured with 1× protein loading buffer at 95°C for 10 min. The samples were subjected to sodium dodecyl sulphate‐polyacrylamide gel electrophoresis (SDS‐PAGE) using a 12.5% gel with a vertical electrophoresis device (Bio‐Rad, USA). The dispersed proteins were transferred to a 0.2 µm polyvinylidene fluoride membrane (Millipore, #03010040001, Germany). The membrane was blocked with 5% skim milk (Hangzi Biological Technology Co., Ltd., #HZ102, Shanghai, China) in Tris‐buffered Saline solution containing 0.05% Tween‐20 (blocking solution) for 1 h at 25°C. The membrane was then incubated overnight with a mouse monoclonal antibody against PBP2a (Abnova, #MAB21934, USA, 1:1000) in blocking solution at 4°C. Residual antibodies were removed by washing 3 times, with 10 min each time at 70 rpm. Finally, the membrane was incubated with goat anti‐mouse peroxidase‐conjugated antibody (Abcam, #ab6728, UK, 1:1000) for 1 h at 25°C. Immunoreactive signals were developed using an enhanced chemiluminescence detection kit (Amersham Imager 600, USA).

2.7. Transmission Electron Microscopy (TEM) Imaging

The morphology of EVs was examined using a TEM (Thermo Fisher, TecnaiG2 spirit Biotwin, USA). A 10 µL suspension of EVs was applied onto a carbon‐coated copper grid (Zhongjingkeyi, #BZ110125b, Beijing, China) and incubated with 0.4 µL of 25% glutaraldehyde. Next, the EVs were allowed to settle onto the grid at 25°C for 60 s. After blotting with filter paper, the grid was stained with 10 µL of 2% uranyl acetate for 90 s. The samples were air‐dried and directly observed using a TEM. Images were captured using a Megaview G2 camera and processed using a TEM and Adobe Photoshop software.

2.8. Immunoelectron Microscopy

The specific operations were as follows (experiments were conducted at room temperature unless otherwise specified): (a) A mixture of EVs and 4% paraformaldehyde (PFA) in a 1:1 ratio (total volume of 10–20 µL) was placed onto a plastic film to form droplets, and the positive side of the carbon grid was placed on the droplet and incubated for 20 min. (b) The carbon grid was placed on a droplet of 100 µL PBS and washed twice (3 min/step). (c) The carbon grid was placed on a droplet of 100 µL of 50 mM glycine solution (repeat 3 times; 3 min/cycle). (d) The carbon grid was placed on a droplet of 5% BSA (100 µL) blocking solution and incubated at 4°C overnight. (e) The carbon grid was placed on a droplet of 20 µL of a mouse polyclonal antibody PBP2a (Abnova, #MAB21934, USA, 1:20) for 60 min. (f) The carbon grid was placed on a droplet of wash buffer (100 µL) and washed 6 times (3 min/step). (g) The carbon grid was incubated with the secondary antibody labelled with colloidal gold (Abcam, #ab29619, UK, 1:200) for 60 min. (h) The carbon grid was incubated with a droplet of 0.5% BSA (100 µL) and washed 6 times (3 min/step). (i) The carbon grid was placed on a droplet of PBS (100 µL) and washed 6 times (2 min/step). (j) The carbon grid was placed on a droplet of 1% glutaraldehyde (100 µL) for 2 min. (k) The carbon grid was placed on a droplet of 100 µL double distilled water and washed 6 times (2 min/step). (l) The sample was negatively stained with 10 µL uranyl acetate for 90 s, and the carbon grid was air‐dried before observation.

2.9. Plasma Cytokine Determination

The concentrations of tumour necrosis factor α (TNFα) and interleukin 6 (IL‐6) in plasma samples were determined using the ELISA kit (FineTest Ref.‐EH1560, Wuhan Fine Biotech Co., Ltd., China), following the manufacturer's instructions.

2.10. Bacteraemia Model

The bacteraemia model was established with BALB/C mice (Jiangsu Jicui Pharmaceutical Biotechnology Co., Ltd., China) through the orbital venous plexus injection technique. Following deep anaesthesia, the mice were positioned laterally with their heads secured to expose the eyeball. The injection needle was guided into the eye from the eye socket at an angle of approximately 45°, directed towards the bottom of the eye. With the cutting edge facing downwards, the needle tip was inserted 2–3 mm, allowing for a slow injection of bacteria. Pressure was applied post‐injection to halt bleeding.

The experimental group was administered MRSA (ST398) at a concentration of 10^⁸ CFU/mL, in a volume of 100 µL, while the control group received an equivalent MSSA. Each group consisted of 6 mice. After 24 h post‐injection, 1 mL of blood was collected from each mouse via orbital venous bleeding into EDTA‐coated anticoagulant tubes. The EVs in the plasma were isolated as described above. The obtained EVs were stored at −80°C for future use.

2.11. Pneumonia Model

A precise quantity of MRSA (ST398) bacterial solution was dripped through the left nostril. In the experimental group, the bacterial inoculum concentration was 4×10^⁸ CFU/mL, with a volume of 40 µL, while the control group received an equivalent MSSA. Subsequently, the mouse was maintained in a vertical position for 1 min to ensure optimal absorption. Post‐infection, the mice were categorized into 9 groups based on the infection time: 0, 6, 12, 18, 24, 30, 36, 48, and 72 h, with 6 mice in each group. At the designated time points, 1 mL of blood was collected from each mouse via orbital venous bleeding into anticoagulant tubes coated with EDTA. 500 µL of blood was spread onto blood agar plates and cultured at 37°C for 24 h to ascertain blood culture results. In contrast, another 500 µL of blood was employed for vesicle isolation from the plasma.

For the pneumonia treatment group, a murine pneumonia model was constructed using the same methodology as described above. Following bacterial inoculation, mice were randomly divided into two groups, each consisting of 30 mice: the treatment group and the non‐treatment group.

Mice in the treatment group received vancomycin (Solarbio, #V8050, China) through subcutaneous injection in the axillary region. The dosage administered was 50 mg/kg per day, and this treatment was continued for 3 consecutive days. In parallel, the non‐treatment group received an equivalent volume of phosphate‐buffered saline (PBS) injected into the same axillary region to serve as a control. At various time points post‐infection, specifically at 12, 24, 36, 48, and 72 h, six mice from both the treatment and non‐treatment groups were selected for sample collection. Blood samples were obtained from each mouse for vesicle extraction. The isolated EVs were then stored at −80°C for future analysis. Simultaneously, myeloperoxidase (MPO) in lung tissue was utilized to quantify the infiltration of neutrophils, a key indicator of the inflammatory response to infection. The assessment also included the evaluation of the distribution and intensity of staining, which aids in determining the severity and spatial characteristics of the infection within the lung tissue.

2.12. Skin‐Invasive Bloodstream Infection Model

In this experiment, a highly virulent MRSA strain (ST398) was selected, and a skin abscess model was established in BALB/c mice through subcutaneous injection. After anaesthesia, the skin at the injection site was gently pinched with the finger of the left hand, and the injection needle was inserted into the skin with the needle hole facing upwards. After injection, the needle was rapidly withdrawn to prevent leakage of the bacterial solution. The bacterial dose used for inoculation is typically 200 µL of 10^¹⁰ CFU/mL bacterial suspension. Mice were randomly divided into 13 groups, with each group consisting of 6 mice. Blood samples were collected from each group every 2 h, and plasma levels of inflammatory cytokines were measured and recorded. Plasma EVs were then extracted and stored at −80°C for further use.

2.13. Meningitis Model

BALB/c mice were anaesthetized to achieve a deep, unresponsive state and then secured on a surgical platform. A precise incision was made in the skin at a location 1–2 mm posterior to the bregma, approximately 1 mm lateral to the midline. Using ophthalmic scissors, a small hole was carefully drilled into the skull to avoid damaging the brain. A needle was gently inserted through this hole using a needle holder, and advanced into the lateral ventricle of the brain. A predetermined volume of 2 µL of bacterial suspension, containing 1×10^⁹ CFU of MRSA or MSSA, was slowly injected into the brain's lateral ventricle, with 6 mice per group. After injection, the needle was meticulously withdrawn to prevent leakage of cerebrospinal fluid or the formation of air bubbles within the cranial cavity. The surgical site was closed by sealing the cranial hole and the skin incision. An antibiotic ointment was applied to the wound to prevent postoperative infection. Twenty‐four hours post‐modelling, blood samples were collected from the mice. EVs were isolated from the blood using differential centrifugation and stored at −80°C for subsequent analysis.

3. Results

3.1. Isolation and Characterization of MRS‐EVs

We isolated and characterized BEVs through a three‐step purification process, which included culture supernatant collection, ultrafiltration through a 50‐kDa cutoff membrane, and an initial ultracentrifugation step (Xie, Cools, et al. 2023). This meticulous methodology was crucial to avoid contamination with cellular debris or large protein aggregates that might be misidentified as BEVs. TEM analysis confirmed the presence of BEVs, exhibiting spherical structures, double membranes, and electron‐dense luminal contents (Figure 1a). Nano‐flow cytometry analysis revealed that the size of MRSA EVs ranged from 60 to 120 nm (Figure 1b), consistent with the typical features of BEVs.

FIGURE 1.

Characterization of MRSA EVs. (a) Transmission electron microscopy (TEM) imaging of MRSA EVs. (b) Nano‐flow cytometry analysis characterizing the size distribution of MRSA EVs. (c) Western blot analysis confirming the expression of PBP2a in MRS‐derived EVs. (d) Immunoelectron microscopy (immuno‐TEM) revealing the distribution of PBP2a. The white arrows indicate the sites of colloidal gold particles which correspond to the positions of the PBP2a. (e) Examination of non‐specific binding of PBP2a to MRSA EVs using isotype control antibodies under immuno‐TEM. (f) Nano‐flow cytometry analysis of PBP2a expression on the surface of EVs from different MRSA clones. The three types of bar charts in the figure represent the experimental group (in pink), the condition control group (in grey), and the blank control group (in yellow), respectively. The markings above the experimental group are used to indicate whether there is a statistical difference compared to the condition control group. The blank control group is used to assess background signals and non‐specific binding. The statistical analysis was performed using an unpaired t‐test. The data were presented as mean values ± standard deviation. In all charts, n.s., p > 0.05; *p < 0.05; **p < 0.01; ***p < 0.001.

Western blot analysis demonstrated a high expression of PBP2a on BEVs derived from clinical isolates of MRS, including MRSA, MRSE, and MRSH (Figure 1c). This finding suggests that PBP2a constitutes a consistent component of MRS‐derived EVs across the population. Importantly, the absence of PBP2a protein in plasma‐derived EVs (Figure 1c), further underscores the specificity of the link between PBP2a and MRS‐derived EVs.

To ascertain the localization of PBP2a on BEVs, we conducted single‐vesicle and submicroscopic analyses. Immuno‐TEM images confirmed the expression of PBP2a protein on the surface of MRSA EVs (Figures 1d,e), MRSE EVs (Figures S1a,b), and MRSH EVs (Figures S1c,d). Further analysis using nano‐flow cytometry revealed that PBP2a⁺ BEVs are present across various MRS species, with the proportion of PBP2a⁺ BEVs exceeding 35% (Figure S1e).

Upon analysing BEVs derived from different MRSA clones, including ST1, ST5, ST398, ST59, and ST188, we found that over 40% of the EVs released by these clones were PBP2a⁺. In contrast, PBP2a⁺ BEVs were barely detectable in other strains (e.g., E. coli and MSSA), as well as plasma (Figure 1f).

3.2. The Origin of PBP2a⁺ EV in Plasma Following MRSA Infection

PBP2a⁺EVs are, in fact, composed of two components: the PBP2a protein and the vesicle carrier. The bacterial origin of the PBP2a protein is beyond doubt, as normal animal hosts lack the genetic material to encode this protein and thus do not possess the genetic basis for its expression. However, the origin of the vesicle carrier for the PBP2a molecule is subject to multiple uncertainties. After bacterial infection of the host, various EV formation mechanisms may be activated within the body, such as secretion by MRSA bacteria, bacterial lysis, processing and presentation of host EVs by immune cells, or detachment and adhesion of PBP2a to other EVs. To trace the origin of these vesicle carriers, we conducted meticulous detection and analysis of human‐derived markers (such as CD9, CD11, TSG101, and β‐actin) on the target EVs. Initially, we successfully captured PBP2a⁺ EVs from plasma EVs in patients with MRSA bloodstream infections using streptavidin‐coated magnetic beads combined with biotinylated PBP2a antibodies. Figure 2a depicts the detailed process of capturing PBP2a⁺ EVs from the plasma of patients with MRSA bloodstream infections using streptavidin‐coated magnetic beads in conjunction with biotinylated PBP2a antibodies. This method allows for the efficient separation and enrichment of the target EVs, laying the groundwork for subsequent experimental analysis. TEM images clearly display the morphology of the EVs enriched by the magnetic beads, as shown in Figure 2b. Further analysis using a nano‐flow cytometer revealed that the proportion of PBP2a⁺ EVs in the enriched sample was over 90%, as shown in Figure 2c. Subsequently, we employed nano‐flow cytometry to further detect the presence of host‐related markers CD9, CD81, CD63, and CD11 on PBP2a⁺ EVs. Results showed that the positivity rates for CD63, CD81, CD9, and CD11 in healthy controls ranged from 7% to 21%, while 3%–5% of EVs in PBP2a⁺ EVs expressed human CD markers. Additionally, immunoblot analysis confirmed the presence of trace amounts of TSG101 in PBP2a⁺ EVs (Figure 2e). Considering the limited number of PBP2a⁺ EVs extracted after enrichment and purification, the low content of marker proteins may have resulted in weak bands (the protein loading amount was 7.5 micrograms per lane). Both nano‐flow cytometry and immunoblot analysis showed that human markers were present in the target EVs. Therefore, our study supports the presence of a human‐derived EV subpopulation within PBP2a⁺ EVs but does not rule out the possibility of direct bacterial secretion. This dual nature of PBP2a⁺ EVs precisely reflects the intricate interplay between the host and the pathogen during MRSA infection. We use the term  PBP2a⁺ EVs’ to refer to EVs that carry this marker, regardless of whether they originate from bacteria or the host. This approach helps us better understand the roles of these EVs in infection processes.

FIGURE 2.

Investigation of the Origin of PBP2a⁺EVs in the Plasma of Patients with Bloodstream Infections. (a) Schematic of PBP2a⁺ EVs captured from plasma. (b) TEM image of PBP2a⁺ EVs captured by magnetic beads. (c) Nano‐flow cytometry analysis of the proportion of PBP2a⁺ EVs captured by magnetic beads. (d) Nano‐flow cytometry detection of host exosome markers CD9, CD63, CD81 and CD11 on PBP2a⁺ EVs in plasma. (e) Western blot detection of the host exosome marker TSG101 and β‐actin on PBP2a⁺ EVs. Lane 1: protein marker; Lane 2: magnetic bead‐captured PBP2a⁺ EVs; Lane 3: exosomes derived from EVs of human plasma (positive control); Lane 4: BEVs from MRSA (negative control).

3.3. Dynamics of PBP2a⁺ EVs in Plasma Following MRSA Infection

To further investigate the dynamics of PBP2a⁺ EVs in plasma post‐MRSA infection, we established pneumonia, meningitis, skin abscess, and bloodstream infection models in BALB/c mice using MRSA (ST398), with MSSA serving as controls. Twelve hours post‐MRSA infection, mice blood samples were collected, bacterial cultures were performed, and EVs were extracted from the plasma. The bacterial cultures indicated that the bacteria had not invaded the bloodstream 12 h post‐infection (Figure S2a). Histological examination (stained with H&E) clearly confirmed the presence of infection at each site (Figures S2b,c,d). We then analysed the extracted plasma EVs using nano‐flow cytometry. Notably, the levels of PBP2a⁺ EVs in the plasma of MRSA‐infected mice were significantly higher than those in MSSA‐infected controls (Figure 3a). The results of the experiments on non‐specific antibody binding can be found in Figures S3a and S3b. These results suggest that the increased quantity of PBP2a⁺ EVs may be predictive of infection occurrence. To further substantiate that the PBP2a⁺ particles we detected are indeed of vesicular origin, we treated the plasma samples with a 1% nonionic surfactant Triton X‐100 for 1 h (Tian et al. 2019). The Triton X‐100 molecules, which possess a hydrophilic head and a hydrophobic tail, can penetrate the lipid bilayer, disrupting its structure. This action compromises the integrity of the lipid bilayer, resulting in the disintegration of vesicle structures, thereby providing the necessary conditions for subsequent analysis. Following this treatment, the proportion of PBP2a⁺ particles decreased to approximately 3% of the levels observed prior to treatment, as shown in Figure S4. These findings indicate that the majority of PBP2a⁺ particles are derived from vesicular structures rather than lipoproteins. We then investigated the dynamic changes in PBP2a⁺ EVs within plasma after MRSA infection and their correlation with the host's inflammatory response. Using the ST398 strain, we successfully established a murine pneumonia model and continuously collected blood samples from mice at various time points. To investigate disease progression, we employed two primary analytical methods: bacterial cultivation and the extraction and analysis of EVs in plasma. A schematic diagram of this process is presented in Figure 3b.

FIGURE 3.

Analysis of PBP2a⁺EVs and inflammatory markers in the blood of mice post‐MRSA infection. (a) Nano‐flow cytometry analysis of PBP2a⁺ EVs proportions in plasma from mice with different infection models. (b) Schematic diagram illustrating the process of constructing a murine pneumonia model. (c) Immunohistochemical analysis of lung tissues from mice infected with different bacteria. (d) Immuno‐TEM image showed the presence of PBP2a⁺ EVs in the plasma. The white arrows denote the presence of PBP2a⁺ EVs in the plasma. (e) Immuno‐TEM image of non‐specific binding control for plasma EVs in mice with pulmonary infection. (f) Dynamics of PBP2a⁺ EVs in the plasma of mice. The statistical analysis was conducted using an independent samples t‐test. (g), (h) Variation in C‐reactive protein (CRP)/interleukin‐6 (IL‐6) levels in the blood of MRSA‐infected mice with pulmonary infection. The data above were presented as mean values ± standard deviation. In all charts, n.s., p > 0.05; *p < 0.05; **p < 0.01; ***p < 0.001.

Bacterial cultures of blood samples collected at various time points were all negative, as shown in Figure S5. However, we conducted an immunohistochemical analysis of lung tissues from MRSA‐ and MSSA‐infected mice. The analysis showed the presence of brown spots in Figure 3c, indicating a significant upregulation of Ly6G expression, a marker for neutrophils. This finding directly indicates substantial neutrophil infiltration, suggesting a pronounced inflammatory response in the lungs and further confirming the success of our murine pneumonia model.

For the analysis of plasma EVs, we employed advanced immuno‐TEM. This allowed for the detection of PBP2a⁺ EVs in the blood of infected mice. As demonstrated in Figures 3d and 3e, showing the presence of PBP2a⁺ EVs in the blood of lung‐infected mice. Nano‐flow cytometry analysis of PBP2a⁺ EVs in the blood of lung‐infected mice revealed a progressive increase in the proportion of PBP2a⁺ EVs during the early stages of infection, peaking at approximately 30 h post‐infection, followed by a gradual decline (Figure 3f). The results of the negative control experiments can be found in Figures S6a and S6b. The temporal patterns of CRP (C‐reactive protein) and IL‐6 (interleukin‐6) levels in mouse blood, as shown in Figures 3g and 3h, indicate that these inflammatory markers increased over time before gradually decreasing. Notably, PBP2a⁺ EVs in plasma were detectable as early as 6 h post‐infection, reflecting the early pattern of non‐specific inflammatory cytokine changes, indicating the potential of PBP2a⁺ EVs as an early marker for MRSA infection. The similar patterns of change observed among PBP2a⁺ EVs, CRP, and IL‐6 levels suggest that PBP2a⁺ EVs may serve as an indicator of infection severity to a certain extent.

3.4. Monitoring of PBP2a⁺ EVs During Progressive Infection

We constructed a mouse model that simulates the deterioration from skin infection to bloodstream infection using the MRSA ST398 strain. As the experiment was conducted, from the onset of skin abscesses to the full development of bacteraemia, we systematically monitored the dynamics of PBP2a⁺ EVs in the plasma of the mice. Concurrently, we measured the levels of C‐reactive protein (CRP) and interleukin‐6 (IL‐6) to fully assess the status of the inflammatory response. Additionally, we conducted bacterial cultures on the blood samples collected at various time points during the progression from localized skin abscess to bacteraemia. The aim was to confirm the specific timing and extent of bacterial invasion into the bloodstream, providing key information for understanding the infection process and the potential mechanisms of action of PBP2a⁺ EVs.

Fourteen hours after local skin infection in mice, blood cultures became positive, and the number of bacterial colonies increased with a prolonged infection period (Figure 4c). Immuno‐TEM confirmed the presence of PBP2a⁺ EVs in the blood (Figures 4a,b). Nano‐flow cytometry analysis revealed that the proportion of PBP2a⁺ EVs in plasma started to increase between 2 and 4 h post‐infection and continued to rise as the infection progressed (Figure 4d; results of the negative control experiments are shown in Figures S7a and S7b). Concurrently, dynamic monitoring of inflammatory factors in the blood of mice revealed that CRP and IL‐6 gradually increased over the course of the infection (Figures 4e and 4f).

FIGURE 4.

Dynamics of PBP2a⁺EVs and inflammatory markers in mouse blood during infection progression. (a) Immuno‐TEM image depicting the presence of PBP2a⁺ EVs in plasma of mice with bacteraemia. The white arrows indicate the location of colloidal gold particles. (b) Immuno‐TEM image of non‐specific binding control for plasma EVs in mouse blood. (c) Bacterial culture results indicating the progression from localized skin infection to bloodstream infection in mice. Note: The watermarks in blood agar plates are visible as black shadows. (d) Changes in the levels of PBP2a⁺ EVs in the blood of mice as the infection progresses from local to systemic. (e), (f) Variation in C‐reactive protein (CRP)/interleukin‐6 (IL‐6) levels in the blood of mice during the progression from local to bloodstream infection. (g) Changes in the levels of PBP2a⁺ EVs in the blood of mice treated with vancomycin. (h), (i) Variation in CRP/IL‐6 levels in the blood of vancomycin‐treated mice. (j) Immunohistochemical staining of lung tissues from MRSA‐infected mice before and after treatment with vancomycin. The brown spots are Ly6G positive, indicating neutrophil infiltration. The data above were presented as mean values ± standard deviation.

The results indicate that PBP2a⁺ EVs are detectable in plasma prior to the entry of MRSA into the bloodstream. As the local infection progresses to bacteraemia, the changes in PBP2a⁺ EVs are more pronounced compared to non‐specific inflammatory markers such as CRP and IL‐6. Therefore, PBP2a⁺ EVs hold potential as a biomarker for early detection and monitoring the progression of MRSA infection.

3.5. Changes in PBP2a⁺ EVs During MRSA Treatment

We further investigated the changes in PBP2a⁺ EVs in blood during the antibiotic treatment against MRSA. Utilizing a BALB/c mouse pneumonia model, we detected the proportion of PBP2a⁺ EVs in the plasma at 12, 24, 36, 48 and 72 h post‐MRSA infection. Our findings indicated that in the vancomycin treatment group, the proportion of PBP2a⁺ EVs in the plasma gradually decreased, whereas it increased over time in the non‐intervention group (Figure 4g; results of the negative control experiments are shown in Figures S8a,b,c,d). Concurrently, we observed a gradual reduction in the levels of inflammatory factors in the plasma of the treatment group (Figures 4h,i). Additionally, immunohistochemical analysis of Ly6G in lung tissues from the treatment group demonstrated a progressive alleviation of pulmonary inflammation (Figure 4j). Collectively, these results suggest that the proportion of PBP2a⁺ EVs holds potential as an indicator for assessing the therapeutic efficacy in the treatment of MRSA infection.

3.6. Clinical Relevance of PBP2a⁺ EVs in Human MRSA Infections

Given the correlation between plasma PBP2a⁺ EVs and MRSA infection in localized and bacteraemia mouse infection models, we aimed to determine the clinical relevance of PBP2a⁺ EVs in human infections. We conducted a comprehensive study involving the collection of blood samples from two distinct cohorts: healthy individuals and patients diagnosed with MRSA infections. These infections were identified across various anatomical sites, encompassing the respiratory tract, soft tissues/secretions, and the bloodstream. The experimental workflow is depicted in Figure 5a. Immuno‐TEM observations revealed the presence of PBP2a⁺ EVs in the plasma of patients with bacteraemia (Figures 5b,c; Figure S9). Nano‐flow cytometry identified PBP2a⁺ EVs in the blood samples of patients with MRSA strains isolated from respiratory, tissue, and blood.

FIGURE 5.

Diagnostic value of PBP2a⁺EVs in clinical cases of MRSA infection. (a) Schematic illustration of the human blood sample processing procedure. (b) Immuno‐TEM image showing the presence of PBP2a⁺ EVs in human blood. (c) Control immuno‐TEM image for non‐specific binding of plasma EVs in human blood. (d) Nano‐flow cytometry analysis depicting the statistical distribution of PBP2a⁺ EVs proportions in plasma samples from different patients. The statistical analysis was performed using a one‐way ANOVA. (e) Graphical representation of the variation in C‐reactive protein (CRP) levels in plasma samples from different patients. The statistical analysis was performed using a one‐way ANOVA. (f) Longitudinal monitoring over 10 days of PBP2a⁺ EVs, CRP levels, white blood cell (WBC) counts, and body temperature in patients with MRSA bloodstream infection. Body temperature is plotted on the right axis, while all other parameters are plotted on the left axis. The proportion of PBP2a⁺ EVs is expressed as a percentage (%). The CRP levels are measured in milligrams per millilitre (mg/mL), and the white blood cell counts are measured in billions per litre (10^⁹/L). The data were presented as mean values ± standard deviation. (g) Receiver operating characteristic (ROC) curves for the diagnostic evaluation of PBP2a⁺ EVs in MRSA infections across various anatomical sites. In all charts, n.s., p > 0.05; *p < 0.05; **p < 0.01; ***p < 0.001.

The identification of PBP2a⁺ EVs was statistically significant when compared with the plasma of healthy controls (Figure 5d; results of the negative control experiments are shown in Figures S10a and S10b), highlighting the potential of PBP2a⁺ EVs as a diagnostic biomarker for MRSA infections. A pronounced elevation in cytokine levels was observed in patients with bloodstream infections, significantly exceeding the levels found in patients with localized infections (Figure 5e). This increase corresponded with the proportionate changes in PBP2a⁺ EVs, with the highest plasma levels of PBP2a⁺ EVs detected in patients with bacteraemia.

To further investigate the clinical diagnostic value of PBP2a⁺ EVs, we collected blood samples from three patients diagnosed with MRSA bloodstream infection. The proportion of PBP2a⁺ EVs in the plasma was dynamically monitored over 10 consecutive days. As shown in Figure 5f, the levels of PBP2a⁺ EVs initially increased and subsequently decreased, correlating with body temperature, CRP levels, and white blood cell counts. These findings suggest that PBP2a⁺ EVs could serve as a diagnostic marker for MRSA infection and may also be capable of monitoring the progression of infection. To further explore the diagnostic potential of PBP2a⁺ EVs in various infectious conditions, we constructed ROC curves for respiratory infections, tissue infections, and bloodstream infections. The AUCs for respiratory infections, tissue infections, and bloodstream infections were calculated to be 0.8025, 0.8513, and 0.9350, respectively (Figure 5g). These values indicate the relative performance of PBP2a⁺ EVs in distinguishing between infected and non‐infected subjects across the different infection categories.

4. Discussion

In our study, we investigated the distribution characteristics of PBP2a on the surface of MRS‐derived EVs from both single‐vesicle and multi‐vesicle dimensions. PBP2a, encoded by the mecA gene, is a unique penicillin‐binding protein that confers resistance to β‐lactam antibiotics, including methicillin (Maillard and Pascoe 2024). Previous studies have identified PBP2a as an important membrane protein component in MRSA (Wang et al. 2018). Our findings that over 35% of EVs secreted by common clinical strains of MRSA, MRSE, and MRSH express the PBP2a represent a pivotal advancement in clinical microbiology. Furthermore, the consistent expression of PBP2a on the surface of MRSA EVs across different sequence types (STs) suggests its potential as a marker for isolating and identifying EVs of MRS origin from biological samples. The identification of PBP2a, a non‐human cellular component, in clinical samples signifies a major breakthrough in infectious disease diagnostics and management.

The potential of BEVs to serve as non‐invasive diagnostic biomarkers for bacterial infections has garnered considerable interest, with numerous studies exploring their clinical implications (Cruz‐Samperio et al. 2023; Xie, Haesebrouck, et al. 2023). Recent research has highlighted the association between increased levels of LPS⁺ EVs in plasma and compromised barrier integrity in patients with inflammatory bowel disease (IBD), HIV, and cancer therapy‐induced mucositis (Tulkens et al. 2020; Ou et al. 2023). Despite these promising findings, the diagnostic value of other subtypes of BEVs in clinical practice remains to be fully validated. Our study focused on the MRSA model strain to evaluate the efficacy of PBP2a⁺ EVs as non‐invasive diagnostic biomarkers for MRSA infections. In this study, our decision to use PBP2a as the sole marker is not arbitrary. On one hand, it is based on its pivotal role in the expression of MRSA drug resistance. As one of the primary resistance mechanisms in MRSA, the expression of PBP2a is closely linked to the bacterium's resistance profile, making it a highly specific diagnostic target (Alexander et al. 2023). On the other hand, our study employs a nano‐flow cytometer to detect target EVs, a method primarily based on the principle of membrane protein marking. It is worth noting that, apart from PBP2a, no other membrane molecules have been reported as representative markers for MRSA. Given the compelling evidence supporting PBP2a as a marker, we have decided to use PBP2a as the marker for target EVs without introducing any additional markers.

It is particularly noteworthy that the origin of PBP2a⁺ EVs in plasma likely differs from that of BEVs obtained from pure cell cultures, given the diverse vesicle formation mechanisms that can occur within living organisms. To trace the origin of these target vesicular carriers, we conducted detailed detection and analysis by enriching PBP2a⁺ EVs using magnetic beads. Our results showed that approximately 3%–5% of the target EVs in the experimental group contained human‐derived markers, with faint positive bands detected in immunoblot analysis. Although the proportion of human markers detected in the PBP2a⁺ EVs is relatively low, this does not exclude the possibility of human‐derived vesicular carriers being present. Normal human hosts do not possess the genetic coding for the PBP2a protein, lacking the genetic basis for its expression. The precise origin of PBP2a⁺ EVs in human plasma remains an open question and will require further research and exploration in the future. However, it is noteworthy that the PBP2a protein on these EVs is closely associated with MRSA. The diagnostic value of PBP2a⁺ EVs does not rely on their complete bacterial origin but is based on their strong correlation with MRSA infection.

We observed elevated levels of PBP2a⁺ EVs in the blood of MRSA‐infected mice compared to those infected with MSSA. This distinction between MRSA and MSSA infections underscores the potential of PBP2a⁺ EVs as biomarkers for differentiating methicillin‐resistant and methicillin‐sensitive strains and highlights a significant diagnostic opportunity. The stark differences in PBP2a⁺ EVs levels in the blood of mice with localized infections in the lungs and skin, as well as systemic bloodstream infections, suggest that these EVs can act as systemic markers for ongoing local infections. This observation is particularly noteworthy as it implies that PBP2a⁺ EVs could provide a diagnostic signal even when the infection site is not immediately apparent. In clinical practice, this could facilitate the prompt initiation of targeted antimicrobial therapy, potentially improving patient outcomes.

CRP and IL‐6 are indispensable inflammatory markers, widely used to assess acute‐phase responses due to their significance in diagnosing, monitoring, and managing inflammatory conditions (Liu et al. 2021; Singh and Newman 2011). Our study investigates the dynamics of PBP2a⁺ EVs in a murine pneumonia model, correlating their levels with those of CRP and IL‐6. Our findings reveal a consistent trend between the plasma levels of PBP2a⁺ EVs and the initial rise of CRP and IL‐6 during the early stages of infection. This synchrony provides valuable insights into the complex interplay between PBP2a⁺ EVs and the host's inflammatory response. The detection of PBP2a⁺ EVs in our murine pneumonia model offers a unique perspective on the early events of MRSA infection. PBP2a⁺ EVs could serve as an early indicator of MRSA infection, potentially preceding traditional markers of inflammation.

Previous research has firmly established the role of BEVs in the pathogenesis and progression of numerous diseases (Xie, Cools, et al. 2023; Hong et al. 2023; Fizanne et al. 2023). These EVs, shed from various cell types including bacteria, play a crucial role in intercellular communication by carrying bioactive molecules that can modulate the host's response and contribute to disease processes (Toyofuku et al. 2023). Building upon these findings, our study investigated the presence and dynamics of PBP2a⁺ EVs in the context of microbial infections. We focused on a murine model of infection, where the disease progressed from a localized state to a systemic one, closely mimicking the clinical scenario of bacterial spread from a focal point of infection to systemic bacteraemia. Our observations revealed a significant increase in the proportion of PBP2a⁺ EVs as the infection transitioned from localized to systemic. This increase parallels the known role of EVs in disease progression, suggesting that these EVs may play a similar role in the pathogenesis of bacterial infections. The presence of PBP2a⁺ EVs could potentially serve as a marker of disease severity or a mechanism by which the infection disseminates and establishes a systemic foothold.

When lung‐infected mice were treated with vancomycin, we observed a progressive decline in plasma PBP2a⁺ EV levels. This decrease is particularly noteworthy, as it suggests that PBP2a⁺ EVs could be a potential biomarker for monitoring therapeutic response in MRSA infections. Traditionally, the assessment of treatment efficacy has relied on clinical symptoms and inflammatory markers such as CRP and IL‐6, which may not always accurately reflect the extent of infection resolution (Ridker et al. 2021). The introduction of PBP2a⁺ EVs as a biomarker may provide a more direct and sensitive measure of the host‐pathogen interaction and the impact of antibiotic treatment.

The diagnostic efficacy of PBP2a⁺ EVs was rigorously evaluated across a spectrum of infectious conditions, including respiratory, tissue, and bloodstream infections. To quantitatively assess their diagnostic potential, we constructed Receiver Operating Characteristic (ROC) curves for each category, a standard approach in diagnostic test evaluation.

The area under the ROC curve (AUC) is a robust indicator of a test's discriminatory power, with higher values suggesting superior diagnostic accuracy. Our analysis yielded AUCs of 0.8025 for respiratory infections, 0.8513 for tissue infections, and an impressive 0.9350 for bloodstream infections. These results underscore the potential of PBP2a⁺ EVs as a diagnostic biomarker, particularly in the context of bloodstream infections where the AUC approaches the ideal value of 1, indicative of excellent sensitivity and specificity.

The variation in AUCs across different infection types may reflect the distinct pathophysiological processes and immune responses elicited by these infections. Notably, the higher AUC for bloodstream infections suggests that PBP2a⁺ EVs could be particularly useful in the rapid and accurate diagnosis of sepsis, a condition with significant clinical implications and a pressing need for effective biomarkers.

While the AUC for respiratory infections is slightly lower, it still falls within the range that is considered to provide good diagnostic accuracy. This may be attributed to the complexity of respiratory infections, which often involve a variety of pathogens and host responses that could influence the release and detection of PBP2a⁺ EVs.

Our findings are in line with the growing body of evidence that supports the role of EVs as promising diagnostic agents in infectious diseases. However, the generation of PBP2a⁺ EVs in vivo is a highly complex process. It may involve a diverse array of EVs types, including free MRS‐derived EVs, hybrids of mammalian cells/MRS EVs, mammalian cell EVs that have picked up PBP2a, and potentially lipoprotein complexes also containing PBP2a that have come through the 0.45‐micron filter. Since the current study mainly serves as a proof‐of‐concept investigation to demonstrate the feasibility of using PBP2a⁺ EVs as biomarkers for MRSA infection, it is necessary to conduct future in‐depth research focusing on clarifying the specific origins of PBP2a⁺ EVs in the context of in vivo infections. Additionally, it is important to note that the clinical application of PBP2a⁺ EVs as a diagnostic tool will require further validation in larger cohorts and across different populations to ensure the generalizability of our results.

In summary, our findings highlight the potential of PBP2a⁺ EVs, particularly those derived from MRSA, as reliable biomarkers for the early diagnosis and monitoring of bacterial infections. The detection of PBP2a on EVs could have significant implications for the diagnosis and treatment of MRS infections. As a key determinant of methicillin resistance, the presence of PBP2a on EVs could serve as a novel biomarker for rapidly identifying MRS strains, thereby accelerating the initiation of targeted antibiotic therapy and improving patient outcomes. Future research should focus on optimizing the detection methods for PBP2a⁺ EVs and integrating them into clinical diagnostic algorithms to enhance the early identification and management of infectious diseases. The potential clinical application of PBP2a⁺ EVs could significantly impact the management of MRSA infections in the global healthcare system.

Author Contributions

Qianqian Gao engaged in the conceptual framework, development of methodology, execution of the investigation, and management of data curation. She was also responsible for the initial drafting of the manuscript, revising it critically for intellectual content, creating visualizations, and confirming the accuracy of the underlying data. Wenwu Zhou participated in conceptualization and methodology development, conducted formal analysis, managed data curation, and verified the underlying data. He contributed to the original draft preparation, manuscript revision, visualization. Zhen Shen participated in conceptualization and methodology development, conducted formal analysis, managed data curation, and verified the underlying data. Tianchi Chen was involved in conceptualization and methodology development, performed formal analysis, and participated in data curation. Cong Hu assisted with the methodology and conducted formal analysis and was also involved in the critical review and editing of the manuscript. Liang Dong played a role in the conceptualization of the study and was instrumental in drafting the original manuscript as well as revising it critically for important intellectual content. Da Han actively engaged in conceptualization and methodology development, secured resources, and was a key contributor to the original draft and subsequent revisions of the manuscript. He also provided supervision and administered the project. Min Li was responsible for the conceptualization, acquisition of resources, original drafting, and review and editing of the manuscript. She provided supervision, managed the project, and secured funding for the research. All authors have reviewed and given their approval to the final manuscript version. Qianqian Gao and Wenwu Zhou have verified the underlying data.

Ethics Statement

The animal study was reviewed and approved by the Ethics Committee of Renji Hospital, School of Medicine, Shanghai Jiaotong University, Shanghai, China. (Ethics number: IACUC‐2022‐Mi‐205). In this study, the collection of human blood samples was conducted in strict accordance with ethical standards and relevant legal regulations. All samples were derived from surplus blood remaining after routine medical testing, ensuring that the original medical use and purpose of the samples were not compromised. Utilizing the guidelines set forth by the Ethics Review Committee, the requirement for obtaining separate informed consent has been explicitly waived for the scientific use of such discarded, non‐identifiable residual blood samples that do not contain any personally identifiable information.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting Figures: jev270111‐sup‐0001‐FigureS1.tif

Supporting Figures: jev270111‐sup‐0002‐FigureS2.tif

Supporting Figures: jev270111‐sup‐0003‐FigureS3.tif

Supporting Figures: jev270111‐sup‐0004‐FigureS4.tif

Supporting Figures: jev270111‐sup‐0005‐FigureS5.tif

Supporting Figures: jev270111‐sup‐0006‐FigureS6.tif

Supporting Figures: jev270111‐sup‐0007‐FigureS7.tif

Supporting Figures: jev270111‐sup‐0007‐FigureS8.tif

Supporting Figures: jev270111‐sup‐0010‐FigureS10.tif

Supporting Figures: jev270111‐sup‐0011‐SuppMat.docx

Data Availability Statement

This study did not generate any new datasets or analyse previously unpublished data. All information and data referenced in this manuscript are derived from publicly available sources and have been appropriately cited within the text. Any additional information or clarification required regarding the data and materials used in this study can be addressed by contacting the corresponding author. Should there be any specific data or materials from this study that are available upon request, they will be shared in accordance with the journal's policies. However, at this time, there are no supplementary data or materials to deposit in a data repository.