The efflux pump QacA mediates the transition to vancomycin heteroresistance in sequence type 5 methicillin-resistant Staphylococcus aureus via membrane lipid reprogramming

ABSTRACT

The heterogeneous vancomycin-intermediate Staphylococcus aureus (hVISA) phenotype in methicillin-resistant S. aureus (MRSA) is a major driver of vancomycin treatment failure. Within the prevalent ST5-MRSA lineage, the efflux pump gene qacA has emerged as a key factor in the development of hVISA. However, the underlying mechanism remains elusive, as vancomycin is not a typical efflux substrate. In this study, we demonstrated that qacA triggered the vancomycin-susceptible S. aureus (VSSA)-to-hVISA phenotypic conversion independently of direct efflux, cell wall thickening, or biofilm formation. Membrane fluidity assays and fluorescence recovery after photobleaching (FRAP) analysis confirmed that QacA expression significantly increased membrane fluidity and accelerated lateral diffusion rates. Correspondingly, confocal microscopy assays revealed that vancomycin probe binding to cells was reduced in the qacA-carrying strains. Lipidomic profiling further demonstrated that qacA induced membrane lipid reprogramming, characterized by an enrichment of diunsaturated glycerophospholipids. Furthermore, exogenous supplementation with long-chain unsaturated fatty acids exerted potent synergistic bactericidal effects with vancomycin. In conclusion, our study showed that qacA drives the VSSA-to-hVISA transition through a distinct lipid reprogramming. This process forms a hyper-fluid membrane “entropy barrier” that disrupts vancomycin binding to Lipid II, rather than relying on direct drug transport. These findings challenge the traditional understanding of antiseptic resistance determinants and highlight the bacterial membrane lipidome as a critical, yet overlooked, target for potentiating vancomycin activity.

Introduction

Vancomycin, the guideline-recommended first-line treatment for severe methicillin-resistant Staphylococcus aureus (MRSA) infections, exerts its potent bactericidal activity by targeting the D-alanyl-D-alanine moiety in the bacterial cell wall and membrane interface [1]. Nevertheless, its prolonged and extensive clinical use has imposed strong selective pressure, driving the increasing prevalence of vancomycin-intermediate S. aureus (VISA) and heterogeneous VISA (hVISA) [2,3]. Critically, the hVISA phenotype is strongly associated with high rates of vancomycin treatment failure [4,5], posing a significant clinical challenge.

Epidemiological studies have identified sequence type 5 (ST5) as the predominant hVISA clone globally [2,6,7]. Patients infected with ST5 MRSA face a higher risk of microbiological treatment failure compared to those with non-ST5 infections [8], a finding supported by the significantly higher hVISA/VISA detection rate in ST5 isolates versus its variant ST764-MRSA (86.4% vs 26.1%, P < 0.001) [9]. In regions such as East China, the dominance of ST5-MRSA has been linked to the acquisition of a qacA efflux pump determinant [10], with carriage rates approaching 80% in clinical isolates [11]. Our previous study identified qacA as an independent predictor of the hVISA phenotype in ST5-MRSA strains [adjusted odds ratio (aOR), 48.323; 95% confidence interval (CI), 6.004-366.565; P < 0.001] through multivariate analysis [12]. This association has also been observed in other regions; for example, Cho et al. [13] reported that qacA/B-positive MRSA isolates were associated with the hVISA phenotype (OR, 4.09; 95% CI, 1.48-11.34) in South Korea.

QacA, encoded by the qacA gene, a multidrug transmembrane transport protein belonging to the major facilitator superfamily (MFS) family, comprising 14 transmembrane segments [14]. A previous study decoded Cryo-EM structure of QacA, and reported that the negatively charged vestibule of QacA binds to cationic antimicrobial compounds, subsequently driving the efflux of various cationic and lipophilic compounds via a proton motive force-dependent antiport mechanism [15]. While enhanced efflux is a well-established cause of multidrug resistance [16], a critical paradox remains: as vancomycin is not a known pump substrate, the link between qacA and hVISA remains unclear.

Studies show that efflux pump function is tightly coupled to membrane homeostasis [17]. Maintaining membrane fluidity is essential for nutrient transport, morphogenesis, stress adaptation, and cell wall synthesis [18–20]. Bacteria survive environmental challenges by precisely tailoring their fatty acid composition, such as increasing unsaturated or anteiso-branched-chain fatty acids (BCFAs) against cold, and saturated fatty acids against heat or toxicity [21,22]. Notably, S. aureus exploits this lipid remodelling for antibiotic resistance. Vancomycin stress triggers an increase in unsaturated fatty acids to elevate fluidity and preserve cellular integrity [23]. Similarly, in daptomycin resistance, pgsA mutations drive aberrant BCFAs accumulation and the resulting hyper-fluidity synergizes with cell wall thickening to impede drug binding and oligomerization [24]. Conversely, disrupting this remodelled membrane with exogenous medium-chain fatty acids effectively clears persister cells [25]. Ultimately, dynamically altering membrane biophysics via fatty acid modulation is a core molecular mechanism driving bacterial survival and antibiotic resistance.

In this study, we aimed to delineate the mechanisms associated with qacA-driven vancomycin-susceptible S. aureus (VSSA)-to-hVISA transition by utilizing biophysical and visualization assays in an isogenic S. aureus strain set. We hypothesized that qacA induces a biophysical barrier at the cell wall-membrane interface that impedes vancomycin binding kinetics. By integrating lipidomic profiles and membrane dynamics of these strains, we sought to identify the membrane reprogramming events modulated by qacA that mediate this heteroresistance phenotype.

Materials and methods

Bacterial strains and culture conditions

The representative ST5 VSSA clinical strain S1-2-62 (qacA-negative VSSA), isolated from a pneumonia patient, was used as the primary study strain, with another ST5 strain S1-2-32 used for validation (Table S1). The pOS1-qacA plasmid was constructed as previously described [12]. Unless otherwise stated, strains were cultured in tryptic soy broth (TSB, Oxoid, UK) or on brain heart infusion agar (BHIA, Oxoid, UK). The efflux-deficient plasmid pOS1-qacA-ΔEL (qacAΔI443-T465) was constructed as described by Majumder et al [15]. According to the method described by Hinds et al. [26], we also correlated OD₆₀₀ values (measured with a UV-2102C spectrophotometer, UNICO Instruments, China) with colony-forming unit (CFU) counts for our S. aureus strains (Table S2 and Figure S1).

Susceptibility testing and hVISA detection

Minimum inhibitory concentrations (MIC) were determined by broth microdilution according to Clinical and Laboratory Standards Institute (CLSI) guidelines [27]. The hVISA phenotype was confirmed using a modified population analysis profile-area under the curve (PAP-AUC) method. Briefly, cultures were plated onto BHIA containing 0.5, 1, 2, 2.5, 4, 8 μg/mL vancomycin (National Institutes for Food and Drug Control, China). An AUC ratio (test strain/Mu3) of 0.9-1.3 indicated hVISA [28,29]. S. aureus ATCC 29213 served as the control.

Observation of cell wall morphology via transmission electron microscopy (TEM)

To investigate cell wall thickening, strains were cultured to the exponential phase (OD₆₀₀ = 0.6, approximately 6 × 10⁸ CFU/mL) and treated with vancomycin (2 μg/mL) or saline for 30 min. Cells were harvested, washed, and fixed overnight at 4 °C, followed by post-fixation with 1% osmium tetroxide (Ted Pella Inc, USA). Samples were dehydrated through a graded ethanol series and embedded in Epon 812 resin (SPI, USA). Ultrathin sections (60-80 nm) were obtained, stained with uranyl acetate (SPI, USA) and lead citrate (Sigma, USA), and examined using a TEM (hitachi, Japan). Cell wall thickness was quantified using ImageJ software (v0.6.0) by averaging three measurements per cell from 15 randomly selected cells per group.

Biofilm formation assay

Biofilms were formed in 96-well plates by inoculating TSB supplemented with 0.33% glucose with a bacterial suspension adjusted to an initial OD₆₀₀ of 0.1 (approximately 6 × 10⁷ CFU/mL), followed by incubation at 37 °C for 48 h. After removing planktonic cells, mature biofilms were treated with vancomycin (2 μg/mL) for 24 h. Residual biofilm biomass was quantified by crystal violet staining (1%) (Solarbio, China), and absorbance measurement at 570 nm.

Confocal microscopy of vancomycin binding

Vancomycin binding was visualized using a fluorescent derivative (FITC-vancomycin, Delta, China) as described previously [30]. Cells were pre-stained with 10 μM propidium iodide (PI, Solarbio, China), 10μM 4′,6-diamidino-2-phenylindole (DAPI, MedChemExpress, USA), embedded in agarose pads, and exposed to FITC-vancomycin (10 μg/mL). Time-lapse imaging was performed via confocal laser scanning microscopy (CLSM, Leica SP8 STED 3X, German) using standard FITC, DAPI and PI filters.

Membrane fluidity and potential

Membrane fluidity was assessed by fluorescence polarization using the diphenylhexatriene (DPH) probe (MedChemExpress, USA) [31]. Protoplasts were generated using lysostaphin and DNase I, and the polarization index was calculated. Membrane potential was measured using the 1μM 3,3′-Dipropylthiadicarbocyanine iodide (DiSC3(5), MedChemExpress, USA) fluorescent dye [32]. The fluorescence 620/670 was calculated to monitor potential changes.

Cell membrane lipidomics

To investigate the mechanism by which qacA enhances membrane fluidity, we performed a lipidomic analysis of the cell membrane, following a previously described procedure [33]. Briefly, overnight bacterial cultures were diluted 1:100 into 500 mL of TSB and grown to an OD₆₀₀ of 0.6 (approximately 6 × 10⁸ CFU/mL). Following a 30-min vancomycin (2 μg/mL) treatment, cells were harvested, resuspended in Tris-sucrose solution, and treated with lysostaphin (30 μg/mL) and phenylmethylsulfonyl fluoride (PMSF, 2 mM) to lyse cell walls and inhibit proteases. Cell lysis was performed using an AVESTIN Emulsiflex (Beckman Coulter, USA), followed by ultracentrifugation (38,000 rpm, 4 °C, 45 min) to collect total membrane fractions. Membranes were washed, re-pelleted, resuspended in 25% sucrose, and flash-frozen. Lipids were extracted using a methyl tert-butyl ether (MTBE): methanol (3:1, v/v) system containing internal standards. The supernatant was dried under nitrogen, reconstituted in acetonitrile: isopropanol (1:1, v/v), and analyzed using an ultra performance liquid chromatography-tandem mass spectrometry (UPLC-MS/MS) system (ExionLC™ AD coupled with QTRAP® 6500+, SCIEX, USA).

Chromatographic separation was performed on a Thermo Accucore™ C30 column (2.6 μm, 2.1 mm × 100 mm) at 45 °C. The mobile phases consisted of (A) acetonitrile/water (60:40, v/v) and (B) acetonitrile/isopropanol (10:90, v/v), both containing 0.1% formic acid and 10 mM ammonium formate. A gradient elution programme was applied at a flow rate of 0.35 mL/min: 0–2 min, 80%–70% A; 2–4 min, 70%–40% A; 4–9 min, 40%–15% A; 9–14 min, 15%–10% A; 14–15.5 min, 10%–5% A; maintained at 5% A until 17.3 min; and re-equilibrated to 80% A from 17.5 to 20 min. Mass spectrometry was conducted using an electrospray ionization (ESI) source (500 °C) with optimized ion source gas and curtain gas settings. Lipid quantification was performed in multiple reaction monitoring (MRM) mode. Data were processed using the metware database and normalized to protein concentration via bicinchoninic acid (BCA) assay (ThermoFisher Scientific, USA). Lipid concentrations (pmol/mg protein) were calculated using a semi-absolute quantification method [34]. Briefly, the peak area ratio (R) of each analyte to its corresponding class-specific internal standard was multiplied by the internal standard concentration (c), the final extract volume (V), and a class-specific correction factor (F) to obtain the absolute amount (pmol) of the analyte. This amount was then normalized to the total protein content (mg) derived from the sample volume (m) and protein concentration (n1). The final lipid concentration was calculated using the following formula:

$$\mathrm{X}={\displaystyle \frac{\mathrm{R}\times \mathrm{c}\times \mathrm{F}\times \mathrm{V}}{\mathrm{m}\times \mathrm{n}1}}$$

Identified metabolites were annotated using Kyoto Encyclopedia of Genes and Genomes (KEGG) Compound database (http://www. kegg.jp/ kegg/compound/), annotated metabolites were then mapped to KEGG Pathway database (http://www.kegg.jp/kegg/pathway.html). Differentially abundant lipids were identified based on Variable Importance in Projection (VIP) scores >1 from the Orthogonal Partial Least Squares Discriminant Analysis (OPLS-DA) model and P-values < 0.05. In order to avoid overfitting, a permutation test (200 permutations) was performed.

Reverse transcription quantitative polymerase chain reaction (RT-qPCR)

Total RNA was extracted from exponential-phase cultures (RNA extraction kit, Accurate Biotechnology, China) and reverse-transcribed to cDNA (cDNA Reverse Transcription Kit, Vazyme, China). Gene expression was quantified by RT-qPCR using SYBR Green (Vazyme, China). The samples were evaluated on the QuantStudio 1 Real-Time PCR System (Thermo Fisher Scientific, USA) using approved cycling protocols. The relative expression values were calculated using 2^(-ΔΔCt) method, with gyrB served as the internal reference.

Synergy assays

Synergy between vancomycin and other agents was evaluated. γ-linolenic acid (HY-N7140), α-linolenic acid (HY-N0728), arachidonic acid (HY-109590), oleic acid (HY-N1446), decanoic acid (HY-W015309), octanoic acid (HY-41417R), and lauric acid (HY-Y0366R) were obtained from MedChemExpress (USA). Quercetin was obtained from Macklin (China). Time-kill assays were performed by incubating cultures with vancomycin (2 μg/mL), alone or in combination with other agents. CFUs were enumerated at 0 and 6 h.

Fluorescence recovery after photobleaching (FRAP)

To investigate the lateral diffusion rate of membrane components, the image-based FRAP protocol was performed as described by Oubekka et al. [35]. The overnight culture was diluted 1:100, grown to OD₆₀₀= 0.5 (approximately 3 × 10⁸ CFU/mL), washed with phosphate-buffered saline (PBS), and incubated with DAPI and FITC-vancomycin probes at 37 °C for 30 min. After washing with PBS, the cells were immobilized under an agarose pad. Briefly, fluorescence intensity images were acquired at a fixed resolution of 512 × 512 pixels with a scanning frequency of 1000 Hz. Bidirectional scanning and the FRAP Booster module were enabled, and the pinhole size was set to 2.0. After delineating the regions of interest (ROIs), each assay was initiated by capturing a single pre-bleach image at 3% laser intensity. Subsequently, a photobleaching pulse was applied to the ROI at 100% laser intensity for 796 ms. The laser power was then attenuated back to 3%, and a series of 150 post-bleach images were captured at 796-ms intervals. Finally, 20 additional images were acquired at 1-s intervals using standard laser intensity.

Statistical analysis

Data were analyzed using GraphPad Prism 11.0.1. Differences were assessed using Student’s t-test (two groups) or ANOVA (multiple groups). Data are presented as the mean ± standard deviation (SD) from at least three biological replicates per group. Significance was set at P < 0.05.

Results

QacA does not mediate the transition from VSSA to hVISA via cell wall thickening or efflux mechanisms

Transformation of the qacA gene into the clinical strain S1-2-62 (MIC = 1 μg/mL) resulted in a phenotypic shift from VSSA to hVISA (Figure 1(A and B) and Table S3), whereas the vancomycin MIC remained unchanged. In the absence of antibiotics, strains exhibited identical growth kinetics, indicating that qacA acquisition does not impair bacterial growth. Under vancomycin pressure (2 and 4 μg/mL), colony counts of S1-2-62 (pOS1-qacA) were higher than those of S1-2-62 (wild type, WT) and S1-2-62 (pOS1-empty) at 24 h (Figure 1(C)). We selected 2 µg/mL as the optimal and representative selective pressure to investigate the subsequent adaptive mechanism.

Figure 1.

Acquisition of qacA promotes vancomycin hetero-resistance. (A) The vancomycin resistance levels were evaluated by population analysis profiles and the colonies were counted after incubation at 37 °C for 48 h. The well-characterized strains Mu3 and ATCC 29213 were included as the standard hVISA and VSSA reference strains, respectively. The dashed line represents the limit of detection. (B) The area under the curve of population analysis profile was calculated. (C) Time-killing curves of S1-2-62 (WT), S1-2-62 (pOS1-empty) and S1-2-62 (pOS1-qacA) with different concentrations of vancomycin (0, 1, 2 and 4 μg/mL). (D) Determination of membrane potential of strains after treatment of saline, and vancomycin (2 μg/mL). (E) Determination of membrane fluidity of strains after treatment of saline, vancomycin (2 μg/mL) and CCCP (2 μg/mL). Membrane fluidity is inversely proportional to fluorescence polarization. Data are presented as the mean ± SD from three biological replicates per group (n = 3). * P < 0.05, ** P < 0.01, **** P < 0.0001. CFU: colony-forming units. WT: wild type; VAN: vancomycin; CCCP: carbonyl cyanide m-chlorophenylhydrazone.

Since QacA is an efflux pump protein, we investigated whether it mediates heterogeneous resistance by actively extruding vancomycin. Chlorhexidine, a known substrate, served as a positive control. We assessed MIC changes in S1-2-62 (pOS1-qacA) when treated with vancomycin or chlorhexidine in combination with the efflux pump inhibitor carbonyl cyanide m-chlorophenylhydrazone (CCCP). In the presence of CCCP, the MIC of chlorhexidine decreased by 4-fold, whereas the vancomycin MIC remained unchanged, suggesting that QacA-mediated hVISA formation is likely not dependent on vancomycin efflux (Figure S2).

Given that vancomycin targets Lipid II in the cell wall, we compared the cell wall morphology of S1-2-62 (pOS1-qacA) and S1-2-62 (pOS1-empty) treated with 2 μg/mL vancomycin using TEM. No significant differences in cell wall thickness were observed between the two strains (Figure S3(A–C)). Biofilm assays revealed that while vancomycin treatment reduced biofilm biomass in both strains, there was no significant difference between them (Figure S3(D)).

Collectively, these findings imply that QacA does not mediate resistance via direct efflux of vancomycin, cell wall thickening or biofilm formation, but rather through alternative mechanisms.

QacA mediates the VSSA-to-hVISA transition by enhancing membrane fluidity and impeding vancomycin-target binding

The membrane potential (Δψ) dynamics of the strains were evaluated using the DiSC3(5) fluorescent probe. We observed that the membrane potential of the wild-type and empty vector strains remained largely unchanged in either the presence or absence of vancomycin. In contrast, both S1-2-62 (pOS1-qacA) strain and the efflux-deficient S1-2-62 (pOS1-qacA-ΔEL) strain exhibited a decrease in fluorescence intensity upon vancomycin exposure (Figure 1(D)). These observations suggest that the qacA gene alters the stability of the bacterial electrochemical gradient under stress, independent of its efflux function, potentially reflecting underlying modifications in the physical state and fluidity of the cytoplasmic membrane.

To confirm this, we further measured membrane fluidity using a fluorescence polarization assay. Treatment with 2 μg/mL vancomycin significantly enhanced membrane fluidity in S1-2-62 (pOS1-qacA), whereas no such changes were observed in the other strains. The efflux pump inhibitor CCCP had no effect on membrane fluidity (Figure 1(E)). Consistently, introduction of the pOS1-qacA plasmid into another ST5 clinical strain (S1-2-32) also resulted in increased membrane fluidity and vancomycin heteroresistance (Figure S4). Additionally, FRAP analysis revealed that S1-2-62 (pOS1-qacA) strain exhibited an accelerated fluorescence recovery at 150s post-bleaching compared to the S1-2-62 (pOS1-empty) strain. This rapid restoration of fluorescence suggests an increased molecular diffusion rate driven by the modified membrane environment (Figure S5).

We further compared differences in affinities of binding to vancomycin between two strains. After 24 h, the free vancomycin in broth medium in the S1-2-62 (pOS1-qacA) group was slightly higher, indicating that more vancomycin molecules bind to cells of S1-2-62 (pOS1-empty) (Figure S6). Given that this bulk concentration assay provides only a static snapshot, we further conducted dynamic visualization. Confocal laser scanning microscopy (CLSM) further revealed that within the same time interval, the empty-vector strain exhibits higher vancomycin probe binding compared to the qacA-carrying strain (Figure 2). This suggests that qacA-induced changes effectively reduce the overall binding capacity of the bacteria for vancomycin, which correlates with resistance.

Figure 2.

Dynamic intracellular accumulation of FITC-vancomycin. (A, C) Representative time-lapse confocal images showing the distribution of FITC-vancomycin (green) in S1-2-62 (pOS1-empty) (A) and S1-2-62 (pOS1-qacA) (C) over time. (B, D) Linear fluorescence intensity profiles analyzed across representative bacterial cells (indicated by red lines) at the 40-min time point for the empty vector control (B) and the qacA-positive strain (D). Time points and scale bars are displayed in the top-right and bottom-left corners, respectively.

QacA enhances membrane fluidity through lipid reprogramming

Based on the above findings, we compared the lipidomic profiles of S1-2-62 (pOS1-qacA) and S1-2-62 (pOS1-empty) under vancomycin stress to elucidate the mechanism driving the phenotypic switch. A total of 500 lipid species were detected. Vancomycin treatment reduced the total lipid content in both strains (Figure 3(A)).

Figure 3.

Summary lipidomic analysis of S1-2-62 (pOS1-empty) and S1-2-62 (pOS1-qacA) strains in the presence or absence of vancomycin. (A) Total lipid content; (B) Stacked bar chart of unsaturated fatty acids categorized by degree of unsaturation in glycerophospholipids; (C) Stacked bar chart of unsaturated fatty acids categorized by carbon chain length in glycerophospholipids. Unsaturated fatty acids contain ≥1 double bond. Data are presented as the mean ± standard deviation from four biological replicates per group (n = 4). VAN: vancomycin.

Literature suggests that increased levels of short-chain, unsaturated, and branched-chain fatty acids influence membrane fluidity [18]. We analyzed the structural distribution of glycerophospholipids, the primary constituents of the bacterial membrane. The S1-2-62 (pOS1-qacA) strain under vancomycin stress showed a trend toward higher proportion of diunsaturated glycerophospholipids (Figure 3(B)). Additionally, the acyl chain length distribution shifted, with an increase in the proportion of species featuring 31 and 33 carbons (Figure 3(C)). These refined structural changes in glycerophospholipids likely drive the alterations in membrane biophysics.

OPLS-DA analysis showed distinct separation among the four groups (Figure S7). Comparing pre- and post-vancomycin treatment, S1-2-62 (pOS1-empty) exhibited 12 differential lipids (10 downregulated, 2 upregulated). S1-2-62 (pOS1-qacA) also showed 12 differential lipids (6 downregulated, 6 upregulated). However, phospholipid products showed divergent trends. Phosphatidylglycerol (PG 18:2_18:3), phosphatidylcholine (PC 16:0_18:2) and phosphatidylinositol (PI 18:2_16:0) were upregulated in S1-2-62 (pOS1-qacA) but PC (10:0_18:0) and PI (18:0_17:1) were downregulated in S1-2-62 (pOS1-empty) (Figure 4(A and B)). KEGG enrichment analysis indicated that differential lipids in both strains were primarily enriched in the glycerophospholipid metabolism pathway (Figure 4(C and D)). We further measured the expression of key genes in glycerophospholipid pathway (Figure 4(E)). Although downregulated by vancomycin, mprF expression in S1-2-62 (pOS1-qacA) was relatively higher compared to the pOS1-empty strain, possibly driven by the accumulation of its substrate, phosphatidylglycerol (Figure 4(F)). The expression changes of other genes in this pathway are presented in Figure S8. Additionally, we quantified the relative abundance of lysyl-phosphatidylglycerol (lysyl-PG) (Table S4). We found that the S1-2-62 (pOS1-qacA) strain contained a significantly higher proportion of lysyl-PG (32:0), a critical lipid responsible for modulating the membrane surface charge (Figures S9 and S10).

Figure 4.

Vancomycin-induced membrane lipid alterations in S1-2-62 (pOS1-empty) and S1-2-62 (pOS1-qacA) strains. (A,C) Analysis of the S1-2-62 (pOS1-empty) strain between presence and absence of vancomycin: Volcano plot of differential lipids (A) and KEGG pathway enrichment analysis (C). (B,D) Analysis of the S1-2-62 (pOS1-qacA) strain between presence and absence of vancomycin: Volcano plot of differential lipids (B) and KEGG pathway enrichment analysis (D). (E) Classic glycerophospholipid metabolism pathway map. (F) Transcriptional expression levels of mprF gene across groups. Data are presented as the mean ± SD from three biological replicates per group (n = 3). * P < 0.05, **** P < 0.0001. MG: monoglyceride; PI: phosphatidylinositol; FFA: free fatty acid; PE: phosphatidylethanolamine; LPE: lysophosphatidylethanolamine; PC: phosphatidylcholine; LPC: lysophosphatidylcholine; PA: Phosphatidic acid; LPA: lysophosphatidic acid; DG: diglyceride; Cer: ceramide; PG: phosphatidylglycerol; LPG: lysophosphatidylglycerol; TG: triglyceride; VAN: vancomycin; VIP: variable importance in projection.

Further comparison between the two strains revealed that in the absence of vancomycin, only 5 lipids differed (all downregulated). Under vancomycin pressure, 3 lipids differed (2 downregulated, 1 upregulated) (Figure 5(A and B)). KEGG analysis enriched these differences in glycerolipid and metabolic pathways (Figure 5(C and D)). Notably, a specific free fatty acid (FFA 19:1) enriched in the fatty acid degradation pathway was downregulated in S1-2-62 (pOS1-qacA) under vancomycin stress (Figure 5(F)).

Figure 5.

Lipidomic differences between S1-2-62 (pOS1-empty) and S1-2-62 (pOS1-qacA) in the absence or presence of vancomycin. (A, C) Volcano plot of differential lipids (A) and KEGG pathway enrichment analysis (C) between the two strains in the absence of vancomycin. (B, D) Volcano plot of differential lipids (B) and KEGG pathway enrichment analysis (D) between the two strains in the presence of vancomycin. (E) Venn diagram displaying the overlap of differential lipids between the two comparison groups. (F) Comparison of FFA (19:1) levels across all groups. Cer: ceramide; TG: triglyceride; FFA: free fatty acid; VAN: vancomycin; VIP: variable importance in projection.

Unsaturated fatty acids synergize with vancomycin by disrupting membrane adaptation

Given that qacA mediates tolerance via altered lipid distribution, we validated this mechanism by combining vancomycin with exogenous fatty acids. γ-linolenic acid (GLA), α-linolenic acid (ALA), arachidonic acid (AA), and oleic acid (OLA) are long-chain unsaturated fatty acids. In the absence of vancomycin, GLA, ALA, and AA alone reduced S1-2-62 (pOS1-qacA) colony counts. When combined with vancomycin, GLA, ALA, AA, and OLA exhibited synergistic bactericidal activity, with colony counts significantly lower than in the vancomycin-only group. In contrast, saturated fatty acids (decanoic acid, octanoic acid, and lauric acid) antagonized vancomycin activity, promoting bacterial growth. This identical phenomenon was also observed in the S1-2-62 (pOS1-empty) strain. Notably, ALA (an isomer of GLA) failed to suppress growth in the empty-vector strain when combined with vancomycin, yet it significantly reduced colony counts in the qacA-carrying strain, indicating that the potentiating effects of some unsaturated fatty acids are isomer-specific (Figure 6). Additionally, quercetin, a traditional Chinese medicine component targeting membrane fatty acids, also showed synergy with vancomycin (Figure S11).

Figure 6.

Bactericidal activity of vancomycin in combination with fatty acids against S1-2-62 (pOS1-empty) (A) and S1-2-62 (pOS1-qacA)(B). Cultures were grown to the exponential phase and treated with fatty acids and vancomycin (2 μg/mL). Colony-forming units (CFUs) were enumerated after 6 h. All fatty acids were applied at 1-fold MIC: GLA (32 μg/mL), ALA (64 μg/mL), AA (32 μg/mL), OLA (32 μg/mL), DA (128 μg/mL), OCA (128 μg/mL), and LA (32 μg/mL). VAN: vancomycin; GLA: Gamma-Linolenic acid (18:3); ALA: α-Linolenic acid (18:3); AA: Arachidonic acid (20:4); OLA: Oleic acid (18:1); DA: Decanoic acid (10:0); OCA: Octanoic acid (8:0); LA: Lauric acid (12:0). Data are presented as the mean ± SD from three biological replicates per group (n = 3). *** P < 0.001, **** P < 0.0001, ns: no significance.

Discussion

This study unveils a previously unrecognized metabolic linkage between qacA and vancomycin heteroresistance. By demonstrating that qacA induces a specific lipid reprogramming, characterized by an enrichment of diunsaturated glycerophospholipids, alongside the accumulation of PG, PC, and PI, which fundamentally alters membrane physical properties, we challenge the conventional efflux-centric view of qacA. This remodelling induces a hyper-fluid state that functions as a biophysical barrier, effectively disrupting vancomycin’s target engagement at the cell wall-membrane interface. Furthermore, we observed that supplementation with long-chain unsaturated fatty acids restores vancomycin susceptibility in qacA-carrying hVISA strains.

QacA is classically characterized as a multidrug efflux pump transporting divalent cationic compounds like chlorhexidine [15]. However, given that vancomycin has a large molecular weight (1449 Da) and targets the Lipid II in cell wall-membrane-interface, it is unlikely to be a direct substrate for QacA-mediated expulsion. Bayer et al. [36] previously demonstrated that qacA confers resistance to tPMP-1 via membrane rigidification independent of efflux activity. Consistent with this “non-efflux” paradigm, our study confirms that qacA transformation drives the VSSA-to-hVISA transition without altering cell wall thickness, biofilm formation or exhibiting direct vancomycin efflux. Instead, we observed significantly enhanced membrane fluidity in the qacA-carrying strain, whereas no such alteration was observed in the efflux-deficient qacA-ΔEL strain. This indicates that the qacA-mediated enhancement of membrane fluidity is not due to a substantial metabolic burden or a non-specific membrane stress response, but rather represents a specific evolutionary function of QacA.

Previous studies on hVISA/VISA resistance mechanisms have largely focused on cell wall thickening driven by cell wall stress regulatory systems [29,37]. However, the role of membrane lipid remodelling in antibiotic resistance is gaining increasing attention; for instance, mutations in mprF and pgsA have been shown to reduce drug susceptibility by altering membrane surface charge or phospholipid composition [24,38–41]. Meanwhile, Lee et al. [42] recently reported that overexpression of AcrAB-TolC efflux pump in Salmonella Typhimurium is coupled with extensive membrane lipid remodelling, which alters envelope membrane lipid and protein compositions to restrict the accumulation of non-substrate compounds. Building on these, we employed lipidomics to investigate the mechanistic role of qacA. In the pOS1-empty strain, differential lipids in the presence versus absence of vancomycin exhibited an overall downregulation, reflecting an amplified global stress response. Conversely, the lipidomic profile of the pOS1-qacA strain deviated starkly from that of the control, pointing to an active reprogramming of the lipidome rather than a simple magnification of stress signals. Under vancomycin stress, qacA drove a shift in the membrane fatty acid profile toward diunsaturated glycerophospholipids. Notably, Mirani et al. [23] have reported analogous membrane fatty acid profiles in VISA/VRSA, lending support to our finding that these alterations result in significantly enhanced membrane fluidity, facilitating survival under vancomycin stress. This alteration in physical properties established a biophysical barrier that impeded vancomycin-target engagement [43,44]; indeed, confocal microscopy revealed that vancomycin binding to cells was reduced in the qacA-carrying strains compared with empty-vector strains, indicating that the extent of vancomycin binding correlates with resistance (Figure 2). We hypothesize that this hyper-fluid membrane environment may create an entropic barrier, potentially isolating the target Lipid II by transiently masking its D-Ala-D-Ala terminus or by enabling its lateral diffusion rate to exceed the kinetic capture rate of vancomycin, thereby limiting its interaction with the drug. This concept is now biophysically corroborated by our FRAP analysis (Figure S5), which demonstrates a significantly accelerated diffusion of fluorescent vancomycin in the qacA strain.

Furthermore, qacA may also operate via an electrostatic repulsion mechanism. We observed that qacA mediated the accumulation of the substrate PG and maintained elevated mprF expression levels even under vancomycin pressure. This facilitates the conversion of PG into positively charged lysyl-PG, thereby increasing the net positive surface charge of the membrane [45]. This electrostatic shift further hinders the binding and insertion of the cationic vancomycin molecule. Collectively, qacA orchestrates the transition of VSSA to hVISA through the dual mechanisms of a membrane biophysical barrier and surface charge modulation (Figure 7).

Figure 7.

Proposed mechanism of QacA-mediated vancomycin heteroresistance in S. aureus. The efflux pump protein QacA does not mediate heteroresistance via the direct efflux of vancomycin. Instead, it drives lipid reprogramming by consuming free fatty acids as an energy source. This metabolic shift increases the proportions of diunsaturated glycerophospholipids, as well as the abundance of specific lipids including PG, PI, and PC. These alterations enhance membrane fluidity and lateral diffusion rates, thereby establishing a dynamic biophysical barrier that impeding vancomycin from binding to its lethal target, Lipid II. Concurrently, the accumulation of PG induces a relative upregulation of mprF, promoting the synthesis of positively charged Lysyl-PG. This increase in surface positive charge generates electrostatic repulsion against the cationic vancomycin molecule, further blocking its entry. Finally, exogenous unsaturated fatty acids can synergize with vancomycin by disrupting this specific membrane fatty acid distribution. FFA, free fatty acid; PG, phosphatidylglycerol; Lysyl-PG, lysyl-phosphatidylglycerol; PI, phosphatidylinositol; PC, phosphatidylcholine.

Specific fatty acid signatures have even been established as reliable biomarkers for distinguishing between quinolone-resistant and methicillin-resistant strains [46,47]. Additionally, the specific downregulation of FFA (19:1) in qacA-positive strains suggests its potential as a lipidomic biomarker for this hVISA phenotype. However, it should be noted that this lipid failed to meet the threshold for significance after false discovery rate correction by the Benjamini-Hochberg method, indicating that its role as a robust biomarker remains preliminary and requires further verification (Table S9).

The reliance of qacA-mediated hVISA on membrane plasticity offers a therapeutic vulnerability. Previous studies suggest that exogenous unsaturated fatty acids can synergize with vancomycin by disrupting membrane organization [25,48–53]. We found that long-chain unsaturated fatty acids are also capable of reversing the resistance phenotype conferred by qacA. Importantly, the potentiation of vancomycin by long-chain unsaturated fatty acids was not restricted to the qacA-carrying strain but was equally profound in the qacA-negative background. It suggests that the application of unsaturated fatty acids represents a broader therapeutic strategy to resensitize S. aureus to vancomycin. Notably, Quercetin, which inhibits fatty acid synthesis (FabG) and modulates membrane composition [54], demonstrated significant synergy with vancomycin, reducing bacterial load by 1-log₁₀ CFU/mL. This validates the strategy of targeting membrane lipid metabolism to disarm qacA-mediated protection.

This study has limitations. First, our investigation was restricted exclusively to ST5 clinical isolates; thus, the observed mechanisms of lipid reprogramming may be lineage-specific and geographically limited, warranting further validation across diverse sequence types. Second, although we co-expressed the native repressor qacR to mitigate overexpression from our high-copy plasmid, absolute QacA protein levels may still exceed those in natural isolates. The precise quantitative impact of this lipidomic reprogramming may vary in clinical strains. Third, the direct molecular interaction between the QacA protein and specific lipid synthesis enzymes remains to be crystallized. Finally, the in vivo efficacy of the proposed combination therapies requires validation in animal models.

Ultimately, our findings contribute to a better understanding of the ongoing MRSA crisis by revealing that the VSSA-to-hVISA transition is facilitated by a biophysical entropy barrier – a qacA-mediated lipid reprogramming that serves as an evolutionary stepping stone toward high-level resistance. Because qacA is plasmid-borne and confers resistance to common hospital disinfectants, current biocide usage may inadvertently co-select for these vancomycin-preadapted strains. Recognizing this biophysical mechanism informs antimicrobial stewardship by highlighting the risks of vancomycin monotherapy, while providing a therapeutic strategy to deploy membrane-modulating adjuvants that dismantle this barrier and extend the clinical lifespan of last-resort antibiotics.

Conclusion

In conclusion, our study establishes that qacA drives the VSSA-to-hVISA transition in ST5-MRSA through membrane lipid reprogramming rather than direct drug efflux. By promoting the accumulation of diunsaturated glycerophospholipids, along with PG, PC and PI, qacA enhances membrane fluidity and lateral diffusion rate, impeding vancomycin from binding to its lethal target, Lipid II. Importantly, disrupting this adaptive fatty acid profile, using either long-chain unsaturated fatty acids or quercetin, demonstrates synergistic bactericidal activity with vancomycin. These findings highlight membrane lipid metabolism as a critical therapeutic target for reversing qacA-mediated heterogeneous resistance.

Funding Statement

The study was funded by the Pujiang Talent Project (24PJD007) and National Natural Science Foundation of China (82504870).

Disclosure statement

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

Data sharing statement

After publication, any data included in the main text or Supplementary Materials that demonstrate the study’s conclusions will be accessible upon reasonable request to the corresponding author Jing Zhang (zhangj61@fudan.edu.cn). Requests for reanalysis of the data contained in this article for the purposes of replicating results will be granted. Requests for analyses beyond the scope of this publication will be evaluated by the corresponding author to evaluate the scientific and ethical appropriateness of the proposed data use.

Supplemental Material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/22221751.2026.2698241.