Significant increase in the secretion of extracellular vesicles and antibiotics resistance from methicillin-resistant Staphylococcus aureus induced by ampicillin stress

Extracellular vesicles (EVs) containing specific cargo molecules from the cell of origin are naturally secreted from bacteria. EVs play significant roles in protecting the bacterium, which can contribute to their survival in the presence of antibiotics. Herein, we isolated EVs from methicillin-resistant Staphylococcus aureus (MRSA) in an environment with or without stressor by adding ampicillin at a lower concentration than the minimum inhibitory concentration (MIC). We investigated whether EVs from MRSA under stress condition or normal condition could defend susceptible bacteria in the presence of several β-lactam antibiotics, and directly degrade the antibiotics. A comparative proteomic approach was carried out in both types of EVs to investigate β-lactam resistant determinants. The secretion of EVs from MRSA under antibiotic stressed conditions was increased by 22.4-fold compared with that of EVs without stress. Proteins related to the degradation of β-lactam antibiotics were abundant in EVs released from the stressed condition. Taken together, the present data reveal that EVs from MRSA play a crucial role in the survival of β-lactam susceptible bacteria by acting as the first line of defense against β-lactam antibiotics, and antibiotic stress leads to release EVs with high defense activity.


Results
Physical characterization of EVs. EVs of MRSA ST692 cells were isolated and designated as EV Nor and EV Strs with respect to the culture conditions. Transmission electron microscopy (TEM) analysis showed bi-layered spherical EVs (Fig. 1a,b). Dynamic light scattering (DLS) revealed the average diameter of EV Strs (78.22 ± 0.81 nm) and EV Nor (86.84 ± 0.25 nm) (Fig. 1c,d, and see Supplementary Table S1 online). EV Strs has www.nature.com/scientificreports/ more vesicles of 10-20 nm size than EV Nor , but the size distribution except for 10-20 nm is almost identical with EV Nor . Their polydispersity index (PDI) were measured below 0.3, indicating that the arrangements were monodispersed (Supplementary Table S1). Their zeta potentials were more negative than -30 mV, implying that there were no considerable differences in the cohesion of the vesicles (Fig. 1e,f, and Supplementary Table S1).

Vesiculation enhances upon physiological stress. To understand further how ampicillin stress affects
EVs production, ST692 was exposed to incremental concentrations of ampicillin and each of the respective EVs was purified (Fig. 2). The quantity of secreted EVs increased dose-dependently according to the amount of ampicillin added to the culture of ST692 cells. In particular, the production of EV Strs (64 μg/mL of ampicillin treated) increased by 22.4-fold compared to EV Nor , the untreated control. Even when only 1 μg/mL of ampicillin was added, the yield was 2.5-fold higher than EV Nor .

EVs defend β-lactam susceptible S. aureus cells against β-lactam antibiotics. Minimum inhibi-
tory concentrations (MICs) of the MRSA strain ST692 and the susceptible bacterium S. aureus ATCC29213 were measured to determine whether the different concentrations of EV Strs and EV Nor from ST692 could protect S. aureus ATCC29213 against several antibiotics ( Table 1). The growth kinetics presented in Fig. 3a showed that EVs can protect susceptible bacteria against each antibiotic at a higher concentration than the MIC of S. aureus ATCC29213. For the six β-lactam antibiotics, EV Strs dose-dependently protected S. aureus ATCC29213, allowing it to tolerate antibiotic exposures above the MICs. EV Nor also dose-dependently defended ATCC29213 against ampicillin, and amoxicillin, but did not protect the susceptible bacteria from the other antibiotics tested over the incubation time. Since the equivalent amount of EV Strs protects susceptible bacteria from antibiotics more strongly than that of EV Nor , the susceptible strains grew much faster in the EV Strs group. One microgram per milliliter of EV Strs appears to be more protective than 25 μg/mL of EV Nor . However, neither EV Strs nor EV Nor protected the susceptible bacteria against antibiotics other than the six mentioned above (cefotaxime, imipenem, methicillin, chloramphenicol, gentamicin, kanamycin, streptomycin, and tetracycline) (data not shown).

Bacterial protection percentage of EV Strs and EV Nor from antibiotics.
To determine the degree of bacterial protection of each EV against antibiotics, we performed quantitative plate assays based on growth kinetics (Fig. 3b). Notably, the viability of the susceptible cells inoculated with EV Strs at 25 μg/mL in the presence of respective antibiotics showed no loss in viability compared to the culture of susceptible cells without antibiotics. In contrast, the viability of the susceptible cells with EV Nor at 25 μg/mL in the presence of cefoperazone, cefazolin, cefalexin, and cloxacillin showed no viability, which can be explained by the rapid killing by the respective antibiotic concentrations.
Specific molecules of EVs are important for the protection of bacteria. After the growth curve experiment, all samples were plated on TSA agar with or without respective antibiotics in the same concentration as was used in the growth curve experiment (see Supplementary Fig. S1 online). All samples that grew in the above experiment were grown in TSA agar but not in TSA agar with respective antibiotics. These results sug- Figure 2. Induction of EVs production by treatments with stressors. (a) EVs were purified and quantified from cultures of ST692 cells treated with 64, 16, 4, 1, or 0 μg/mL ampicillin. EVs yields were averaged and normalized to untreated controls to adjust fold change. One way ANOVA was used for analyses and data were presented as mean ± standard deviations (SD). *P < 0.05, ****P < 0.0001. www.nature.com/scientificreports/ gested that the enhanced survival rates were not due to a mutation of the β-lactam resistant genetic materials of EVs transferred to susceptible S. aureus, but rather from the molecules owned by EVs that protected the bacteria from the β-lactam antibiotic environment. In addition, colonies grown in TSA agar were identified as S. aureus at the species level using MALDI-TOF MS (data not shown).

EVs protect different genera of Gram-negative bacteria against ampicillin.
To determine whether the EVs isolated from the Gram-positive MRSA can protect Gram-negative bacteria, which belong to different genera, growth curve profiles in the presence of growth-inhibiting concentrations of ampicillin and qualitative plate assays were carried out. The MICs of ampicillin against Escherichia coli (RC85), Edwardsiella tarda (ED45), and Salmonella spp. (Sal26B) were 8 μg/mL, 4 μg/mL, and 8 μg/mL, respectively 9 . Both EV Strs and EV Nor defended RC85, ED45, and Sal26B against ampicillin (Fig. 4a). ED45 cells treated with 5 μg/mL EV Nor exhibited growth at 12 h, whereas both RC85 and Sal26B cells supplemented with 5 μg/mL exhibited growth at 2 h. When EV Nor 1 μg/mL was added to each strain, RC85 exhibited growth at 2 h, ED45 at 24 h, and Sal26B at 4 h. Since the MIC of ED45 was lower than that of RC85 or Sal26B, the protective effect caused by EVs molecules appears to be slower. Quantitative plate assays were investigated by counting the CFU based on growth kinetics (Fig. 4b). Both EV Strs and EV Nor dose-dependently protected each susceptible strains against the bactericidal effect of ampicillin.

EVs enable degradation of β-lactam antibiotics.
To explain the reason why EVs were capable of protecting bacteria in the β-lactam antibiotic environment, LC-ESI-QQQ analysis was carried out to measure concentrations of antibiotics after treatment of EVs in the presence of antibiotics in a cell-free system (Fig. 5). The concentration of six antibiotics was dramatically decreased in samples treated with 5 μg/mL of EV Strs for 3 h (five antibiotics except for cefalexin) or 24 h (cefalexin) compared with respective antibiotics without EVs. Samples treated with 1 μg/mL of EV Strs could hydrolyze six antibiotics, but the capacity appears to be lower than that of 5 μg/mL. Moreover, 5 μg/mL of EV Nor also completely decomposed ampicillin and amoxicillin in 3 h, but the activity against cefoperazone and cefazolin was very weak, and against cefalexin and cloxacillin was not detected. The extent of the capacity of EV Strs and EV Nor to hydrolyze ampicillin for 6 h was compared by increasing the dose of ampicillin after EVs concentrations were fixed to a certain amount (Fig. 6). EV Strs degraded 640 μg/mL of ampicillin in 3 h, but EV Nor could not degrade the equivalent concentration of antibiotics for 6 h.

Proteomic characterization of EV Strs and EV Nor .
We compared the protein constituents of EV Strs and EV Nor using LC-MS/MS analysis because delineation of the biological role of the protein components of EVs is important in understanding their relevance to antibiotic resistance. A total of 204 proteins were identified using the uni_bacteria database (Fig. 7a, See Dataset 1 online). EV Strs alone has 159 proteins, of which 67 proteins overlapped with EV Nor . All proteins possessed by EV Strs and EV Nor were classified according to these categories: relevant biological processes, cellular components, and molecular functions ( Fig. 7b-d). Each of the proteins of EV Strs and EV Nor was also classified with the same categories ( Fig. 7e-g). The introduction of the LC-MS/MS data into the Staphylococcus aureus USA300 strain database revealed that there are several different proteins identified which are not included in the uni_bacteria database. So we re-examined the proteins under the categories, biological processes, cellular components, and molecular functions of both EV Strs and EV Nor , separately and combined (see Supplementary Fig. S2 and Dataset 2 online). The results obtained here implied that the majority of the protein composition of EVs was altered and some of the proteins are more abundant which might be the result of the reaction of the host bacteria to the exposure of sub-MICs of ampicillin. www.nature.com/scientificreports/ Comparative analysis of EVs protein concerned with resistance to a β-lactam antibiotic. Among the EVs proteins analyzed from the two databases (Dataset 1 and Dataset 2), we examined the candidate proteins related to the consumption of β-lactam antibiotics ( Table 2). The β-lactamase proteins encoded by blaZ and SA1529 and native PBPs such as PBP1, -2, -3 were upregulated when compared to the protein compositions of EV Strs versus EV Nor .

β-lactamase content in EVs was regulated by ampicillin stress.
To further elaborate on the relevance of ampicillin stress in the total production of β-lactamase in EVs of the host bacteria, we then analyzed the β-lactamase activity associated with the degradation of various β-lactam antibiotics. All samples of EVs from cultures treated with increasing concentrations of ampicillin showed higher absorbance than the positive control, other samples had higher absorbance than negative control but lower than the positive control (Fig. 8a).
To quantify the β-lactamase activity of EVs was presented as milliunit per milligram of protein (Fig. 8b). The activity of EVs isolated from bacteria exposed to increasing concentration of ampicillin in sub-MICs increased gradually, wherein, EVs treated with 1 μg/mL exhibited more β-lactamase activity than corresponding EV Nor , by 6.5-fold. When comparing the samples with β-lactamase activities per mg, the EV Strs (treated with 64 μg/mL of ampicillin) was the highest, followed by the supernatant from stress condition (Fig. 8c). All of the samples subjected to ampicillin stress showed significantly increased levels as opposed to those in the stress-free environment, specifically in the supernatant by 15.8-fold, whole-cell lysate by 4.0-fold, and EVs by 19.1-fold.   www.nature.com/scientificreports/

The function of EVs mixed with a β-lactamase inhibitor.
To determine whether β-lactamase is involved in the antibiotic degradation capacity of EVs, we observed the growth kinetics after the treatment of EVs and β-lactamase inhibitors (sulbactam) in ATCC29213 cells in the respective antibiotic environment (Fig. 8d). EV Strs degraded each β-lactam antibiotics more than EV Nor , allowing susceptible cells to grow in an antibiotic environment, but no susceptible cells grew within 48 h in the respective EVs added with sulbactam.
Ampicillin stress affects β-lactam antibiotic tolerance of bacteria. The MICs of each antibiotic against stressed and normal ST692 cells were compared ( Table 3). Treatment of ampicillin lower than MICs (64 μg/mL) increased the MICs from 2 to 4 times for β-lactam antibiotics except for cefotaxime, imipenem, and  www.nature.com/scientificreports/  www.nature.com/scientificreports/ methicillin, but when the stressed ST692 cells were sub-cultured in antibiotic-free agar they reverted to their normal susceptibility to a β-lactam antibiotic. The MICs of stressed ST692 cells showed no change in all other antibiotics except for β-lactam antibiotics.

Discussion
In this study, the potential function in β-lactam resistance of EVs released by bacteria in stressed conditions was examined. EVs from MRSA cultured with or without sub-MICs of ampicillin were purified, and their capability to degrade several β-lactam antibiotics investigated, and their proteomics compared to identified β-lactam antibiotic resistance-related protein constituents. It was found that the production of MRSA EVs increased dose-dependently by ampicillin treatment, and EVs from cultures treated with ampicillin became more potent in degrading β-lactam antibiotics. Several factors that trigger vesicle production including genetic background of the strains, and factors such as temperature, iron and oxygen availability, media composition, and growth phase 10,22 . In addition, bacterial vesiculation could be triggered more intensely when bacteria were challenged with certain antibiotics at sublethal concentrations. A previous study showed that the treatment of β-lactam antibiotics creates holes in the peptidoglycan layer of Gram-positive bacteria which cause protrusion of cytoplasmic membrane material into the extracellular area and more EVs are generated 10,23-25 . There is no significant difference in the size of EVs isolated from MRSA treated w/ and w/o ampicillin by the analysis of TEM and DLS ( Fig. 1 and Supplementary Table S1 in manuscript). However, the total protein amount of respective EVs was different in BCA protein assay. In addition, we have confirmed that the production of EVs augmented when ampicillin or gentamicin was treated with different MRSA strains named ST541 in sub-inhibitory concentration (data not shown).
Several types of research have shown that vesicles from bacteria can affect various antibiotics or antimicrobial peptides by hydrolysis/sequestration and acting as decoys. For instance, the vesicles can protect the bacteria for some antibiotics 16,24,26 such as the case of colistin 18 , melittin 18 , polymyxin B 26 , and daptomycin 24 thereby assisting the survival of the bacterium both in vitro and ex vivo 24 . It has been demonstrated in the past that vesicles can defense bacteria by degrading some β-lactam antibiotics, like amoxicillin, ampicillin, cefoperazone, and cefotaxime through their own proteins which are related to β-lactam resistance 7,9,19 . Packed into vesicles, substances related to antibiotic resistance can be delivered over long distances, unharmed by dilution and degradation 14 ; such packaging poses a great benefit for the antibiotic resistance mechanism of the bacteria.
β-lactams are one of the most widely utilized groups of antibiotics available for the treatment of several bacterial infections 27 . β-lactam antibiotics attach to transpeptidase enzymes, also referred to penicillin-binding proteins (PBPs), interfering them from forming a peptidoglycan layer that produces the cell wall. Due to differences in cell wall organization, Gram-positive bacteria are bounded by a single membrane making them more sensitive to the bactericidal activity of β-lactam antibiotics than Gram-negative which are surrounded by an outer membrane that can protect exposure to antibiotics 28 . These bacteria have been exposed to naturally occurring antibiotic compounds for at least one billion years, and the evolutionary advantages they have adapted have led to the rapid development of resistance mechanisms for survival against antibiotics which are currently being used for medical applications 29 . The major mechanisms by which Gram-positive bacteria have developed to avoid the inhibitory effect of β-lactam antibiotics are as follows 30 : (1) β-lactamases are secreted into the extracellular space to degrade β-lactam antibiotics before they reach the cell wall; (2) Expression of mutated PBPs that are still capable of synthesizing the cell wall but are incapable of binding to β-lactam antibiotics. Therefore, EVs that www.nature.com/scientificreports/ have the ability to resist the bactericidal effect of β-lactam antibiotics as newly observed in Gram-positive strains are more important than those from the less susceptible Gram-negative strains. The β-lactamase enzyme 7 and metallo β-lactamase superfamily protein 31 hydrolytically destroy β-lactam antibiotics and both proteins were significantly upregulated in EV Strs compared with EV Nor (Table 2 and see Dataset  1 and Dataset 2 online). It is assumed that the β-lactamase may be bound to the membrane of EVs 32 , therefore the EVs secreted from Gram-positive S. aureus exhibit β-lactamase activity (Fig. 8) and degrade antibiotics in the extracellular environment (Fig. 5). The number of PBPs located on the extracellular surface of the cytoplasmic membrane varies between bacterial species 33,34 . Cell wall synthesis in S. aureus is intrinsically controlled by four native PBPs, PBP1 to 4, and β-lactam resistance of MRSA is determined by the production of one non-native PBP, PBP2a 33 . Once β-lactam covalently bound to native PBPs, this stable covalent adduct could not be removed by neutral buffers, acids or detergents 35,36 , therefore we assumed that the native PBPs which are more abundant in EV Strs than EV Nor ( Table 2) are capable of holding β-lactam antibiotics. This study hypothesized that EV Strs are more efficient in degrading β-lactam antibiotics than EV Nor , and this capability can be due to the proteins produced in abundance since bacteria have a mechanism to selectively package-specific proteins and concentrate them into the vesicles to resist stressors 37,38 .
We confirmed that EVs did not affect the growth of strains in the presence of the other tested antibiotics (data not shown) which include imipenem, cefotaxime, and methicillin. These three antibiotics belong in the group of β-lactam antibiotics but were not degraded by the EVs containing β-lactamases due to their resistance to β-lactamase [39][40][41] . Therefore, although the activity of β-lactamase was increased fourfold by ampicillin stress in ST692 cells (Fig. 8c), these three mentioned antibiotics were not influenced in terms of their MICs by the stressed ST692 (Table 3).
Ampicillin below concentrations of MICs value intensifies the production of chromosomal β-lactamase 42,43 and this phenomenon was confirmed in Fig. 8C and Table 3. These findings suggested that MRSA temporarily facilitates the intrinsic ability of bacterial defense mechanism to over-produce β-lactamase as a reaction after exposure to high levels of ampicillin drug, thereby causing an increased hyposensitivity to β-lactam antibiotics. Thus, for an appropriate antibiotic challenge, it might require an increased dose of antibiotics to treat bacteria.
Actually, S. aureus has been identified as one of the microbial infections in many polymicrobial infections 44 . Studies proving 'cooperative interaction' between different strains of bacteria had been discussed in the past. For example, Candida albicans and S. aureus colonized human mucosal surfaces with commensals and cause enhanced disease severity during biofilm-related coinfections [44][45][46] . Haemophilus influenzae and S. aureus showed cooperative interactions and both colonized in nasopharynx, instances, and genital tract 44,47 . S. aureus co-colonized together with Enterococcus faecalis in the intestinal tracts 44 , and conjugation between the two strains causes a horizontal transfer of the vanA gene, resulting in multidrug-resistant staphylococci 48,49 . Schaar et al. also indicated that vesicles including β-lactamase help to protect producer bacteria as well as co-occurring organisms in the human 19,50 . The results of Fig. 4 further imply that the EVs secreted from MRSA can indeed protect co-existing bacterial communities against β-lactam antibiotics.
To summarize, global protein modulation of EVs by ampicillin stress response of MRSA involves processes that are directly related to β-lactam antibiotic resistance. EVs naturally secreted from MRSA possess β-lactamresistant proteins, which can help bacteria to survive in the antibiotic environment by hydrolyzing the capacity of antibiotics. Inducing stress on MRSA with a sub-lethal dose of ampicillin could stimulate production of EVs enhancing the ability to consume antibiotic compared with EVs released in no-stress condition. Therefore, proper drug treatment is necessary to impede the progeny of MRSA because indiscriminate abuse of β-lactam antibiotics may create an opportunity for bacteria to be more resistant to β-lactam antibiotics. Besides, this is a novel mechanism not related to PBP2a-based, the major resistance mechanism of β-lactam antibiotics of MRSA strains known up to this point. This information equips us with a new perspective on how to lessen (if not eradicate) the impact of multi-drug resistant bacteria which impose a grave threat to global public health.

Methods
Bacterial strains. Methicillin-resistant Staphylococcus aureus C-S03-S237 strain of ST692 51 isolated from the chicken was provided from Animal and Plant Quarantine Agency, Korea and β-lactam-sensitive S. aureus ATCC29213 strain was purchased from ATCC. Luria-Bertani (LB; Oxoid) broth or tryptone soya agar (TSA; Oxoid, United Kingdom) were used to grow both cells at 37 °C. Ampicillin-sensitive bacteria, Escherichia coli RC85 9 , and Edwardsiella tarda ED45 52 were cultured on TSA and Salmonella spp. Sal26B 53 was incubated on brain heart infusion (BHI; Oxoid) agar at 37 °C.

Determination of minimum inhibitory concentrations.
Nine β-lactam antibiotics known to confer bactericidal effects by inhibiting cell wall biosynthesis, namely ampicillin, amoxicillin, cefalexin, cefazolin, cefoperazone, cefotaxime, cloxacillin, imipenem, and methicillin (Sigma-Aldrich, USA) and five other class antibiotics, such as chloramphenicol, gentamicin, kanamycin, streptomycin, and tetracycline (Sigma-Aldrich) were selected. The minimum inhibitory concentration (MIC) of each antimicrobial agent was determined in ST692 and ATCC29213 cells using the broth-dilution method in 96-well plates 9,54 according to Clinical and Laboratory Standards Institute (CLSI) guidelines, except that cation-adjusted Muller Hinton broth was substituted with LB. The MIC values were measured from three independent experiments.
EVs isolation and characterization. EVs from ST692 cells were purified from bacterial culture supernatant as described previously 7 , with some modifications. Briefly, when isolating EV Nor , the strain was cultured in nutrient broth (NB; Difco) without any antibiotic addition. To determine the change in the production of EVs, the ampicillin dose was treated at 1, 4, 16, or 64 μg/mL (EV Strs means EVs treated with 64 μg/mL of ampicillin www.nature.com/scientificreports/ when bacteria are cultured). Each culture medium was centrifuged at 6,000×g for 15 min and concentrated by QuixStand Benchtop system (GE Healthcare, Sweden). Each supernatant was centrifuged at 150,000×g at 4 °C for 3 h. Further purification was performed by a continuous sucrose density gradient followed by ultracentrifugation. The final EV pellet was resuspended in 10 mM Tris-HCl (pH 8.0) (Biopure, Korea) and filtered through a 0.2 μm filter (Thermo Fisher Scientific, IL). The protein yields of EVs samples were measured using a Pierce BCA protein assay kit (Thermo Fisher Scientific, USA). Transmission electron microscopy (TEM) of EVs was performed as previously described 9 using a Tecnai G2 Spirit Twin TEM system (FEI, USA). Dynamic light scattering (DLS) of EVs for particle size distribution and measurement of zeta potential was performed as described previously 9 using a Nano ZS instrument (Malvern Instruments, Malvern, UK) and the Zetasizer software (version 7.11; Malvern Instruments).
Proteome analysis by liquid chromatography combined with tandem mass spectrometry (LC-MS/MS). Each of ST692 EVs was mixed with sample buffer and separated by Sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) according to Laemmli's method 55 and in-gel digestion was performed as previously described 9  Quantitative protein profiling, statistics and database searching. Peptide peaks were detected with an average peak width of 1 min and matched with a mass accuracy of at least 0.6 Da. Differentially expressed proteins were defined to exhibit more than twofold or greater increase/decrease in comparable intensity or the complete appearance/disappearance of the spot. After alignment of the retention times of the chromatogram, normalization was carried out with the measured intensity distribution and the proteome was quantified with the peak intensity ratio. In silicon analysis of functional associations. Gene ontology (GO) terms such as biological process, cellular component, and molecular function are derived from differentially expressed proteins obtained using a software tool for researching annotations of proteins (STRAP) version 1.5 (Boston University School of Medicine, USA).

Effect of EVs on the growth of bacteria in the presence of β-lactam antibiotics. The effects of
EV Nor and EV Strs on the cytotoxicity of β-lactam antibiotics were monitored by assessing the growth curves of EVs-treated Staphylococcus aureus (ATCC29213) cells as previously described with slight modifications 9 . The following antibiotics were used at the growth-inhibiting concentrations: ampicillin, 40 9 . The count obtained in the absence of antibiotics was taken as 100%, and the corresponding counts in the presence of different concentrations EV Nor or EV Strs plus the respective β-lactam antibiotics were calculated. Colonies from each cultured sample (n = 5, colonies per sample) were randomly selected and identified by matrix-assisted laser desorption ionization-time of flight mass spectrometry (MALDI-TOF MS) 56 to check for contamination.

Measurement of antibiotic concentrations.
To evaluate whether EVs could directly degrade β-lactam antibiotics, the effects of EV Nor and EV Strs on the concentrations of six antibiotics in a cell-free system were analyzed by liquid chromatography/electrospray ionization mass spectrometry (LC-ESI-QQQ-MS/MS; 6420 Triple Quad LC/MS; Agilent) as previously described with slight modifications 9 . One microgram per milliliter or 5 μg/mL of EV Strs or 5 μg/mL of EV Nor in PBS were mixed with ampicillin (40 μg/mL), cefoperazone (8 μg/ mL), cefazolin (1.25 μg/mL), amoxicillin (40 μg/mL), cefalexin (4 μg/mL) or cloxacillin (1.25 μg/mL). Filtered PBS containing the respective antibiotics without EVs was used as a positive control. In addition, to compare the degradability of EV Nor and EV Strs against ampicillin, 5 μg/mL of each EVs was added to various concentrations

Quantification of β-lactamase activity.
To test for differences in β-lactamase activity between wholecell lysates, supernatants, and EVs from the stressed condition and normal condition, a colorimetric β-lactamase activity assay kit (BioVision, Canada) was used according to the manufacturer's instructions. The assay involves the hydrolysis of nitrocefin which produces a colored product that is measured by spectrophotometry (OD 490 ). A Bradford assay kit (Thermo Fisher Scientific) was used to determine the protein concentrations of samples. Equivalent concentrations of each sample were dispensed to the wells of a clear flat-bottomed 96-well, and the provided nitrocefin and buffer were added. The OD 490 was immediately measured in kinetic mode.

Statistical analysis.
Statistical analyses were carried out using Graphpad Prism, version 8.1.1. (GraphPad, CA, USA). Significant differences were determined by Student's t-test, One-way Analysis of Variance (ANOVA), Two-way ANOVA, or Turkey's multiple comparison test. All data were expressed as means ± standard deviations (SD). Differences were considered statistically significant at P < 0.05.

Data availability
All data generated or analyzed during this study are included in this published article and its supplemental material files.