Atherosclerotic plaque-derived extracellular vesicles mediate smooth muscle cell phenotypic switching and promote vascular remodeling: EVs promote VSMC phenotypic switching

TL;DR

This study shows that extracellular vesicles from atherosclerotic plaques cause smooth muscle cells to change their function, worsening artery damage.

Contribution

The study identifies a novel role of atherosclerotic plaque-derived EVs in promoting vascular smooth muscle cell phenotypic switching.

Findings

AS-EVs promote VSMC phenotypic switching by downregulating contractile markers.

miR-23a-3p in AS-EVs targets Myl12b to regulate MRTFA and inhibit contractile gene expression.

GW4869 inhibits phenotypic switching in a rat atherosclerosis model.

Abstract

Despite the use of lipid-lowering and anti-inflammatory treatments, the progression of atherosclerosis is relentless in most patients. This suggests the presence of in situ pathological factors that continuously exacerbate lesions. We hypothesized that extracellular vesicles (EVs) within the atherosclerotic microenvironment might act as in situ stimulatory factors on vascular smooth muscle cells (VSMCs), thereby exacerbating atherosclerosis.

A local atherosclerosis model was induced using Ldlr knockout (Ldlr KO) rats fed a high-cholesterol diet and subjected to partial carotid ligation. Immunofluorescence, Western blot (WB), and single-cell sequencing confirmed the phenotypic switching of VSMCs in atherosclerotic plaques from both rats and humans. The phenotypic switching of VSMCs in atherosclerotic rats was characterized by reduced expression of VSMC contraction markers and increased expression of LGALS3, PDGFRB, and SCA1. GW4869 inhibited the phenotypic switching of VSMCs in atherosclerotic plaques in a rat model. EVs were extracted from atherosclerotic carotid tissues using differential centrifugation. Chitosan thermosensitive hydrogels were used for in situ delivery of EVs into the arterial wall of the carotid artery. Immunofluorescence staining revealed that atherosclerotic plaque-derived EVs (AS-EVs) promoted VSMC phenotypic switching and downregulated the expression of VSMC contractile markers in vitro. miRNA analysis of EVs derived from atherosclerotic plaques of rats identified miR-23a-3p and its target gene Myl12b. To investigate the underlying mechanisms, engineered EVs loaded with miR-23a-3p were used to mimic AS-EVs. AS-EVs potentially regulate MRTFA nuclear translocation via miR-23a-3p/Myl12b, inhibit contractile marker expression, promote VSMC phenotypic switching, and further promote carotid artery remodelling.

We conclude that AS-EVs promote VSMC phenotypic switching and exacerbate atherosclerosis.

The online version contains supplementary material available at 10.1007/s00018-026-06094-4.

Figures

F and αSMA shown in Figure 1G, whereas unpaired t tests were applied for all other data analyses. VSMC, vascular smooth muscle cell; IEL: internal elastic lamina; Con: control rats, Ldlr-/- rats fed a normal diet; AS: atherosclerotic rats, Ldlr-/- rats fed a high-fat, high-cholesterol diet and subjected to partial ligation

Full text

Introduction

Atherosclerosis remains a leading cause of morbidity and mortality worldwide [1]. Previous research on atherosclerotic mechanisms has often focused on “extraplaque” factors, such as vascular injury, inflammation, lipid deposition, and blood flow turbulence [2–4]. However, atherosclerotic plaques continue to progress even after external stimuli are removed. Consequently, many patients face a considerable risk of cardiovascular events even when their serum lipid concentration is controlled at very low levels [5, 6]. On the other hand, controlling systemic inflammation did not significantly reduce the risk of cardiovascular events or overall mortality [7, 8]. These findings underscore the necessity of investigating the roles of in situ stimulatory factors in atherosclerotic plaques.

Previous studies have demonstrated that 30% to 70% of cells within plaques are derived from vascular smooth muscle cells (VSMCs) [9, 10]. VSMCs potentially represent the primary cell type stimulated by in situ factors. During the development of atherosclerosis, VSMCs are activated and undergo a phenotypic transition characterized by the downregulation of contractile genes and increased proliferation and protein [11–14] synthesis. Although current techniques have confirmed the varying activation states of VSMCs in atherosclerosis, the precise mechanisms underlying this phenotypic switching remain largely undetermined.

The atherosclerotic microenvironment is known to promote the progression of atherosclerosis through autocrine or paracrine mechanisms [15]. Extracellular vesicles (EVs) have emerged as crucial messengers, transmitting signalling molecules through these mechanisms and contributing to the establishment of the disease microenvironment [16]. Tissue-derived EVs act as key mediators in the tumour microenvironment by interacting with recipient cells, regulating premetastatic niche formation, facilitating immune evasion, and promoting tumour migration and metastasis [17–19]. In nontumour research, EVs mediate disease microenvironment changes. Maitrayee et al. reported that EVs from Alzheimer’s brain tissue transfer neurodegenerative proteins to neurons, increasing neurotoxicity and worsening Alzheimer’s disease pathology [20]. Additionally, EVs can induce pyroptosis through the secretion of activated GSDMD, which induces pyroptosis in neighbouring cells [21]. These findings highlight the role of tissue-derived EVs as carriers of molecular information reflective of the pathological microenvironment [22, 23].

In atherosclerosis, EVs serve as in situ stimulators and may promote plaque development [24–26]. Previous studies have investigated the roles of EVs in mediating VSMC activity, but most have focused on EVs derived from specific cell types or plasma sources. Zhu and colleagues reported that EVs secreted by macrophages after nicotine stimulation promote VSMC migration and proliferation via miR-21-3p [27]. Other researchers have reported that EVs derived from cultured macrophages or VSMCs can serve as sites for mineral aggregation, thereby regulating plaque calcification [28, 29]. Notably, atherosclerotic plaque-derived EVs (AS-EVs) have emerged as a novel research focus. Blaser et al. conducted the first multiomics study on human carotid AS-EVs, revealing their role in cardiovascular diseases. Their findings revealed a 45.9% overlap in protein content between EVs and tissue proteomes, indicating that EVs partially reflect the phenotypic characteristics of the plaque microenvironment. Moreover, the unique signatures detected in EVs highlight the distinct research significance of tissue-derived extracellular vesicles [30]. Recent studies have demonstrated that plaque-derived EVs exacerbate cerebral ischaemic injury through microglial activation [31]. However, whether AS-EVs influence VSMCs within the microenvironment to accelerate plaque progression remains unknown and warrants further exploration.

This study demonstrated that AS-EVs induce VSMC phenotypic switching and vascular remodelling. MiR-23a-3p in EVs was found to regulate actin depolymerization by modulating the expression of the target gene Myl12b, thereby mediating VSMC phenotypic switching. Additionally, in situ administration of AS-EVs promoted the phenotypic switching of VSMCs. In summary, AS-EVs may serve as local stimulatory factors, thereby promoting the progression of atherosclerotic lesions.

Methods

Antibodies

A detailed list of antibodies with catalog numbers and companies is available in (Table S1).

Animals, diet, and treatment

In this study, we utilized male Ldlr knockout (KO) Sprague Dawley (SD) rats, alongside wild-type SD rats. The Ldlr KO rats were sourced from the Institute of Cardiovascular Sciences at Peking University, while the wild-type rats were obtained from the Model Animal Research Institute of Nanjing University (Nanjing, Jiangsu, China) [32].

Carotid atherosclerosis was established in Ldlr KO rats through a combination of a high-fat high-cholesterol (HFHC) diet and partial ligation. Starting at 12 weeks of age, Ldlr KO rats were fed either on an HFHC diet (40% fat and 0.5% cholesterol, TP28502, Trophc Animal Feed High-Tech Co., Nantong, China) or a normal diet as control. At 16 weeks of age, rats on HFHC diet underwent partial ligation of the left carotid artery (LCA), where the left external carotid artery, internal carotid artery, and occipital artery were ligated, while the superior thyroid artery remained intact. During the surgery, the mice were anesthetized with 5% isoflurane and maintained with 2% isoflurane in oxygen (RWD Life Science Co., LTD). Following partial ligation, the rats maintained their assigned diets until they were euthanized at 24 weeks of age.

Atherosclerosis studies (Oil red O, HE, EVG)

The carotid artery tissue (n = 4 rats per group) was longitudinally sectioned and stained with Oil Red O. Serial cross-sections were stained with hematoxylin and eosin (HE) and Oil Red O to assess atherosclerotic lesions. Additionally, serial cross-sections were stained with Elastica van Gieson (EVG) to measure the extent of neointimal hyperplasia. Intimal area and intimal/medial area ratio were calculated using Image J software. The medial area was defined as the area between the external elastic lamina and the internal elastic lamina. Image J software was used for quantification.

EV isolation and identification

The methodology for extracting tissue EVs was detailed in our previous publication [32]. In brief, EVs were isolated from the extracellular matrix of fresh tissue using a gentle digestion process. A segment of left common carotid artery (LCA) tissue, approximately 1 cm in length, was incubated in 100 µL of collagenase (1 mg/mL). The tissue was ground, minced, and digested at 37 °C for 20 min. After digestion, 300 µL of protease inhibitor (final concentration 1 µL/mL) was added. The sample was centrifuged at 300 g for 5 min, 2000 g for 10 min, 10,000 g for 30 min, and ultracentrifuged at 100,000 g for 30 min to collect the precipitate. The pellet was resuspended in PBS and stored at −80 °C for future use. EVs extracted from atherosclerotic carotid artery segments, approximately 1 cm in length, were resuspended in 100 µL of PBS, yielding an average EV concentration of 8 × 10^10/mL. EVs were validated by western blot for TSG101, CD63, and CD81 and by nanoscale flow cytometry (NanoFCM) for size distribution analysis.

Preparation of chitosan hydrogels and in situ simulation of EV release

The methods were described previously [27]. Briefly, the stock solutions were prepared as follows: (1) 2% chitosan stock solution (Aladdin, C105799) was prepared using 0.1 M acetic acid and kept in an ice bath. (2) 50% β-glycerophosphate (β-GP, Aladdin, D106347) solution was prepared using double-distilled water. Both stock solutions were sterilized using 0.22 μm filters. The chitosan stock solution, 50% β-glycerophosphate solution, and extracellular vesicle solution were thoroughly mixed in a 7:3:5 ratio and injected near the carotid artery(n = 4 rats per group)using a syringe. Injections were administered every 10 days, with each dose containing 4*10^9 EVs particles, for a total of three injections.

EV uptake

EVs were labeled with PKH26 (Sigma Aldrich, USA) according to the instructions of manufacturer. VSMCs were cultured on slides and co-incubated with EVs for 24 h. Immunofluorescence staining was performed on VSMCs, with DAPI (D9542, Sigma Aldrich) used for nuclear staining. Engineered EVs loaded with cy3-miR-23a-3p were inherently fluorescent. For in vivo evaluation of EV uptake, PKH26-labeled EVs were released in situ using chitosan hydrogel. After 24 h, the carotid artery was embedded in OCT (4583, Sakura Finetek, USA) and cryosections were prepared. Tissue sections were stained with αSMA (1:300, Abcam) and DAPI. Images of the final slides or sections were acquired using an Olympus BX51 microscope (Olympus, Japan).

In vivo inhibition of EV formation

As previously mentioned [32], the neutral sphingomyelinase inhibitor GW4869 was used to inhibit EV formation. Following partial ligation of the carotid artery, rats (n = 4) were intraperitoneally injected with GW4869 (S7609, Selleck, China) at a dose of 0.7 mg/kg, dissolved in a solution containing 5% DMSO in saline. Control rats (n = 4) received an equivalent volume of 5% DMSO in saline. Injections were administered twice weekly for two months. Subsequently, the carotid artery was embedded in OCT, and frozen sections were obtained for histological analysis.

VSMC culture

Primary VSMCs were cultured as previously reported [33, 34] with minor modifications. After euthanizing 3–4 week Sprague-Dawley rats via cervical dislocation, the thoracic aorta was harvested. Primary vascular smooth muscle cells (VSMCs) were then isolated through collagenase digestion (2 mg/ml, Collagenase Type II, Sigma-Aldrich, USA). In brief, after initial digestion with collagenase, the intima and adventitia of the thoracic aorta were separated. The media was then further digested with collagenase. The digestion was subsequently terminated, followed by resuspension and centrifugation, and finally, the cells were plated in culture dishes that had been precoated with collagen (Collagen from rat tail, C7661, Sigma-Aldrich). The isolated VSMCs were cultured in DMEM/F12 medium (Dulbecco’s Modified Eagle Medium/Nutrient Mixture F12) (Gibco, USA) supplemented with 10% fetal bovine serum (FBS), 100 U/ml penicillin, and 100 µg/ml streptomycin. Using the first three generations of cells for subsequent experiments.

Immunofluorescence and cytoskeletal staining

Immunofluorescence staining was performed on 15 μm-thick vascular frozen sections and cell coverslips. Briefly, the sections and coverslips were fixed with 4% paraformaldehyde (PFA) at room temperature for 10 min. Permeabilization and blocking were carried out using 5% goat serum, 1% BSA, and 0.1% Triton X-100 for 1 h. Subsequently, the samples were incubated with primary antibodies overnight at 4 °C.The sections and coverslips were incubated with appropriate secondary antibodies and DAPI in the following day. Images were obtained using an Olympus BX51 microscope (Olympus, Japan) and subsequently processed with Adobe Photoshop (version 25.0.0). Quantification of positive signals was performed using ImageJ software (version 1.53c).

Combined FISH and immunofluorescence staining were employed to detect miR-23a-3p. A Cy5-labeled miR-23a-3p probe was used to detect miR-23a-3p in rat carotid artery tissue, and a Cy3-labeled probe was used for miR-23a-3p detection in human carotid intima from endarterectomy specimens. The probes were designed by RiboBio. Briefly, the sections were fixed in 4% PFA, permeabilized, and blocked, followed by overnight incubation with the probe hybridization solution. In the next day, immunofluorescence staining was performed, and nuclei were stained with DAPI.

The cytoskeletal staining was performed according to previous studies [35]. Adherent cells were cultured on a coverslip. The cells were washed three times with PBS and then fixed with 4% paraformaldehyde (PFA) for 10–15 min. (Note: Avoid using fixatives containing methanol, as methanol can disrupt the cytoskeletal structure or interfere with dye binding, leading to loss of filamentous staining.) The cells were washed three times with PBS and permeabilized with PBS containing 0.1% Triton X-100 for 5 min. Rehydration was performed in PBS for 5–10 min, followed by three PBS washes. A staining solution containing 9 µg/mL (0.3 µM) fluorescent DNase I (Alexa Fluor 488 DNase I conjugate, Invitrogen) and 0.165 µM fluorescent phalloidin (Alexa Fluor 594 phalloidin, Invitrogen) was added to the coverslip (100 µL). The cells were incubated for 15–20 min for staining, then washed three times with PBS. Nuclei were counterstained with DAPI for 10 min, and images were captured using an Olympus BX51 microscope (Olympus, Japan).

RNA extraction and qPCR analysis

Total RNA was extracted from cells and tissues using TRIzol reagent (Invitrogen, USA) and reverse transcribed to cDNA using the RevertAid First Strand cDNA Synthesis Kit (Thermo Scientific, USA). Real-time PCR was performed using SYBR Green in a 25 µL reaction on a Stratagene Mx3000P QPCR system (Agilent Technologies, USA). Additionally, miRNA-specific reverse transcription primers were used for miRNA reverse transcription. miRNA qPCR was conducted using miRNA-specific primers, with U6 serving as the internal control. The reverse transcription and qPCR primers were provided by RiboBio. Primers for reverse transcription were provided in Table S2. The primer pairs used for real-time PCR were listed in Table S3.

Immunoblotting

Samples were lysed using RIPA lysis buffer (Cell Signaling technology, USA) supplemented with 1% PMSF. Protein concentration was determined using the BCA assay. After denaturation with loading buffer, the samples were used for immunoblotting. Equal amounts of protein were separated by SDS-PAGE and transferred onto PVDF membranes. The membranes were blocked with non-fat milk and incubated with primary antibodies overnight. The following day, the membranes were incubated with secondary antibodies for 1 h. Specific protein signals were detected using the Immobilon Western Chemiluminescent HRP Substrate (Millipore, USA).

MiRNA transfection and dual-luciferase reporter assay

VSMCs were transfected with 50 nM miR-23a-3p negative control (N.C.)/mimic (Ribo Bio, Co., Ltd, China) utilizing Lipofectamine 3000 reagent (Thermo Fisher Scientific, USA). The Myl12b 3’UTR dual-luciferase reporter vector plasmid (wild-type WT/mutant MUT) was constructed by Ribo Bio. 293 T cells were plated in 96-well plates and co-transfected with 1 µg of a vector harboring the 3′UTR of rat Myl12b and 100 nM of miR-23a-3p negative control (N.C.)/mimic (Ribo Bio, Co., Ltd, China) using Lipofectamine 3000 reagent (Thermo Fisher Scientific, USA). Luciferase activity was quantified 24h after transfection. The Dual-Luciferase^®^ Reporter Assay System (Cat# E1910, Promega, USA) was employed for Luciferase activity, in combination with the GloMax^®^-Multi + Detection System (Promega, USA).

Preparation of engineered EVs loaded with miR-23a-3p

ExoLoad Small RNA Loading Kit (Enze Kangtai, Cat#ELSR-06, Beijing) was used to prepare engineered EVs loaded with miR-23a-3p [36]. The protocol involved mixing 300 pmol cy3-miR-23a-3p (RiboBio, Guangzhou), 330 µl EVs (Enze Kangtai, ContrExo 293 F EV reference, Cat#CTE-061), 60 µl EV-Transit Peptide (ETP), and 33 µl reaction buffer. The mixture was incubated at 37 °C in the dark for 2 h with shaking at 150 rpm. Following incubation, the sample was transferred to a 100 kD ultrafiltration tube (Millipore), and wash buffer was added to reach a volume of 4 ml. Centrifugation was carried out at 4,000 g until the volume was reduced to 100 µl. A second wash with wash buffer was performed, centrifuging to a final volume of 200 µl. The EVs were gently pipetted from the membrane and transferred to an EP tube, yielding the loaded EVs. The dosage of miR-23a-3p was measured via fluorescence microplate reader.

Cell transfection and lentiviral transduction

MiR-23a-3p mimic (micrON™ miRNA mimic), inhibitor (micrOFF™ miRNA inhibitor), and their respective negative controls were purchased from RiboBio (Guangzhou, China). Cell transfection was performed using Lipofectamine™ 3000 Transfection Reagent (Invitrogen) according to the manufacturer’s instructions. Subsequent experiments were conducted 48 h after transfection. To investigate the function of the Myl12b gene, an Myl12b overexpression lentivirus was constructed by GeneChem (GeneChem, Shanghai, China). The Myl12b overexpression lentivirus was produced using the lentiviral vector GV341 (Ubi-MCS-3FLAG-SV40-puromycin). Both the overexpression lentivirus and the negative control (NC) lentivirus were provided by GeneChem (Shanghai, China). Viral particles were added to the cultured cells at a multiplicity of infection (MOI) of 10, with the addition of 1× HitransG A (GeneChem, Shanghai, China) to enhance transfection efficiency. Fresh culture medium was replaced 24 h after transduction, and the expression level of the target gene was analyzed by Western blot. Successfully transduced cells were used for subsequent functional experiments and analyses.

VSMC proliferation and migration assay

The proliferation ability of VSMCs was assessed using a commercial kit (Cat#C10310-1, RiboBio). VSMCs were stimulated with PBS or AS-EVs for 24 h, followed by incubation with 50 µM EdU for 2 h. Staining procedures and result analysis were performed according to the manufacturer’s instructions. DAPI staining was performed to identify nuclei. Immunofluorescence sections were captured using Olympus BX51 microscope (Olympus, Japan), and the quantification of positively labeled cells was performed using Image J software (version 1.53c).

For the scratch assay, VSMCs were seeded in 6-well plates and allowed to adhere overnight to reach 90%–95% confluence. After 48 h of serum starvation, a scratch was made in the cell monolayer using a sterile 200 µL pipette tip, and initial photomicrographs were taken using a Nikon Eclipse Ti-U phase-contrast microscope. Cells were then stimulated with Con-EVs or AS-EVs, and a second set of images was taken after 24 h to measure SMC migration over the scratched area.

Carotid dissociation and single cell preparation

After euthanizing the rats, left ventricular perfusion was performed. Carotid artery tissues were then harvested and immediately placed in cold PBS. Three samples from both the control and atherosclerosis groups were collected, with each sample comprising pooled tissues from five carotid arteries. Single-cell suspensions of carotid artery cells were prepared following established protocols from the literature [37]. Briefly, after removing the perivascular adipose tissue, carotid arteries from both groups were cut into 1 mm pieces and digested with an enzyme solution containing 2 mg/mL type I collagenase (Worthington Biochemical Corp., Lakewood, NJ), 1 mg/mL elastase, and 10 µg/mL DNase I (Sigma, USA) for 20 min at 37 °C. The cell suspension was filtered through a 70 μm strainer and washed twice with PBS. The cells were then resuspended in PBS containing 0.04% bovine serum albumin, achieving a viability of over 80%. These resuspended cells were subsequently used for sequencing. Additionally, during the single-cell dissociation process, the supernatant from each centrifugation step was collected for the extraction of EVs.

Single-cell RNA sequencing

Qualified carotid artery cell suspension was loaded onto the Chromium single cell controller (10x Genomics) to generate single-cell gel beads in the emulsion according to the manufacturer’s protocol using single cell 3 ‘Library and Gel Bead Kit V3.1(10x Genomics, 1000121) and Chromium Single Cell G Chip Kit (10x Genomics, 1000120). In brief, single-cell suspension, reagents, gel beads, and partitioning oil were loaded onto 10x Chromium Chip B. Single-cell RNA was barcoded via reverse transcription in individual GEMs. Next, all cDNAs were pooled to construct a general library, which was then sent for sequencing. The libraries were finally sequenced using an Illumina Novaseq6000 sequencer with a sequencing depth of at least 100,000 reads per cell with pair-end 150 bp (PE150) reading strategy (performed by CapitalBio Technology, Beijing).

Single-cell RNA data analysis

Single-cell expression data derived from FASTQ files were processed using Cell Ranger software (7.2.0, 10X Genomics). Following batch effect removal across samples via the R package “harmony” according to the pipeline (http://satijalab.org/seurat/), quality control was performed to exclude cells with < 200 genes or mitochondrial gene per centages > 25%. The dataset was then normalized using the NormalizeData function and scaled with the ScaleData function in Seurat (v4.3.0). Dimensionality reduction was performed using PCA, and visualization was realized by t-SNE and UMAP. Cluster-specific gene markers were identified using the FindAllMarkers function. For cell annotation, signature genes of each cluster were mapped against well-established vascular cell-type markers from previous studies [33, 34]. Gene expression was visualized with violin plots, dot plots, heatmaps, and t-SNE plots created with the Suerat functions VlnPlot, DotPlot, DoHeatmap, and FeaturePlot. GO and KEGG enrichment analyses of cluster markers were performed using KOBAS software with Benjamini-Hochberg multiple testing adjustment, using the top 20 marker genes of each cluster. The contractile and synthetic gene sets were scored using the AddModuleScore function in Seurat (v4.0.4).

EV RNA sequencing and scRNA-seq combined analysis

As previously mentioned, all supernatants from the single-cell dissociation process were collected for the extraction of EVs. RNA was extracted using the Exosome RNA Purification Kit (Simgen), and library preparation was performed with the SMARTer Stranded Total RNA-Seq Kit V2 (Takara). Sequencing was carried out by Echo Biotech Co., Ltd., Beiing. China. The target genes of the upregulated miRNAs in AS-EVs (compared to control EVs) were overlapped with the downregulated gene sets of VSMCs in the atherosclerosis rats (compared to the control rats), followed by KEGG pathway enrichment analysis.

Statistical analysis

Each experiment was repeated at least three times independently with similar results. Statistical analyses were performed using GraphPad Prism9.5.1. Parameters are expressed as the mean ± SD. Data normality was evaluated using the Shapiro–Wilk test. For comparisons between two groups, if the data conformed to a normal distribution, a two-tailed Student’s t-test was applied; for non-normally distributed data, the nonparametric Mann-Whitney test was utilized. When comparing more than two groups, if the data were normally distributed and satisfied the homogeneity of variance assumption, a one-way ANOVA combined with Tukey’s post hoc test was used to compare the means across groups. For non-normally distributed data, the nonparametric Kruskal-Wallis test with Dunn’s multiple comparison test was used instead. Detailed methods for each analysis are described in the figure legends. Statistical significance was set at P < 0.05.

Results

An atherosclerosis model demonstrated neointimal hyperplasia and VSMC phenotypic switching

To establish a carotid atherosclerosis model, Ldlr-KO rats were fed a high-fat, high-cholesterol diet combined with partial carotid artery ligation. Ldlr-KO rats fed a regular chow diet served as controls (Figure S1A). Gross specimens of the carotid artery stained with oil red confirmed successful model construction (Figure S1B). Oil red (Figure S1C) and HE staining (Figure S1D) of cross-sectional slices of the carotid artery revealed significant intimal thickening and vascular lipid infiltration in atherosclerotic rats. Immunofluorescence staining revealed more pronounced macrophage infiltration in atherosclerotic rats (Figure S1E).

Next, the phenotypic switching of VSMCs and neointimal hyperplasia in atherosclerotic rats were investigated. EVG staining (Fig. 1A) and αSMA staining (Fig. 1B) revealed significant neointimal hyperplasia and VSMC proliferation in atherosclerotic rats. The neointimal area and intima/media area ratio were increased in atherosclerotic rats (Fig. 1C, P < 0.0001 vs. control rats). BODIPY fluorescence staining revealed stronger signals in the VSMCs of atherosclerotic rats, suggesting significant lipid accumulation within the lesioned tissues because of the net effect of lipid influx and efflux (Fig. 1D). VSMC phenotypic switching was further evidenced by the expression of switching markers PDGFRB, SCA1, and LGALS3 (Fig. 1E-F, P < 0.0001 for LGALS3; P = 0.0022 for PDGFRB; P < 0.0001 for SCA1 vs. control). Human carotid artery specimens obtained from carotid endarterectomy also showed intracellular cholesterol infiltration in VSMCs and expression of the markers PDGFRB, SCA1, and LGALS3 (Fig. 1J). Additionally, the expression of the contractile markers αSMA, MYH11, and TAGLN was significantly lower in the vascular tissues of atherosclerotic rats than in those of control rats at both the mRNA level (Fig. 1G; P = 0.0056 for Myh11; P = 0.0018 for αSma; P = 0.0029 for Tagln) and the protein level **(**Fig. 1H-I; P = 0.0003 for MYH11; P = 0.0286 for αSMA; P = 0.0052 for TAGLN), whereas the expression of the phenotypic switching marker OPN was significantly upregulated (Fig. 1H-I; P = 0.0074 for OPN). These results indicated that neointimal formation and VSMC phenotypic switching occurred in the atherosclerosis model.Fig. 1VSMC phenotypic switching and vascular remodelling in atherosclerotic rats. A, Representative photomicrographs of Verhoeff’s/van Gieson-stained carotid artery sections from control rats (Ldlr-/- rats fed a normal diet) and atherosclerotic rats (Ldlr-/- rats fed a high-fat high-cholesterol diet and subjected to partial ligation) (n = 3 rats per group). Scale bars: 50 μm. B, Representative immunofluorescence staining of αSMA (red) and DAPI (blue) in carotid arteries from control rats and atherosclerotic rats (n = 3 rats per group). Scale bars: 50 μm. C, Quantification of the intimal area and the ratio of the intimal area to the medial area. Each dot represents the intimal area of one carotid artery section (left); Each dot represents the intima/media area ratio of one carotid artery section (right) (n=4 rats per group, 2 sections per rat). D, Representative fluorescence image of intracellular lipid accumulation in VSMCs (n = 3 rats per group). Low magnification scale bar: 100 μm; high-magnification scale bar: 50 μm. E-F, Representative images (E) and quantification (F) of αSMA (red), PDGFRB (green), SCA1 (green), and LGALS3 (green) and their colocalization in carotid artery sections (n = 3 rats per group; two sections per rat). Scale bar: 50 μm. G, αSma, Myh11, and Tagln mRNA expression levels in carotid arteries were detected using real-time PCR (n = 4 rats per group). H-I, Representative immunoblots and densitometric analysis of MYH11, αSMA, TAGLN, and Osteopontin in the lysates of carotid arteries (n = 4 rats per group). J, Representative images of αSMA (red), BODIPY lipid (green), PDGFRB (green), SCA1 (green), or LGALS3 (green) and their colocalization (yellow) in human carotid artery sections (n = 3 carotid samples in each group). Scale bar: 50 μm. All the data are presented as the mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001, **** P < 0.0001 vs. the control group. Statistical analysis: The Mann‒Whitney test was used for the analysis of PDGFRβ shown in Fig. 1F and αSMA shown in Figure 1G, whereas unpaired t tests were applied for all other data analyses. VSMC, vascular smooth muscle cell; IEL: internal elastic lamina; Con: control rats, Ldlr-/- rats fed a normal diet; AS: atherosclerotic rats, Ldlr-/- rats fed a high-fat, high-cholesterol diet and subjected to partial ligation

ScRNA sequencing revealed VSMC phenotypic switching

To further demonstrate the phenotypic switching in the atherosclerosis model, single-cell transcriptome capture (3 samples per group, with each sample being a mix of 5 animals) was performed. We present the results of our investigation of VSMCs. Four distinct SMC subpopulations were identified: SMC0, SMC1, SMC2, and SMC3 (Fig. 2A). SMC0, an indicator of contractile SMCs, was characterized by high expression of contractile markers such as Cnn1 and Itga8. SMC1 was characterized by high expression of Vcam1 and Eng. SMC2 was characterized by high expression of Dcn, Gsn, and Col14a1. SMC3 was characterized by high expression of Sncg, Fabp4, Rgs5 and Lgfbp4 (Fig. 2C). The functional enrichment of KEGG pathways and GO results validated the annotation of these SMC subpopulations (Fig. 2E, S2). Compared with those in the control group, the quantity and proportion of the SMC1, SMC2, and SMC3 subsets in the AS group were greater (Fig. 2B, D). The expression of the VSMC phenotypic switching markers Lgals3, Pdgfrb, and Spp1 was significantly greater in the AS group compared with the control group (Fig. 2F). To further illustrate phenotypic switching in the AS group of VSMCs, we considered SMC1, SMC2, and SMC3 as a whole and constructed a heatmap of the expression of contractile and phenotypic switching markers (Fig. 2G). Additionally, we defined sets of contractile and phenotypic switching genes and used cell scores to evaluate the extent to which individual cells expressed these defined gene sets. The results indicated that SMC0 cells exhibited high expression of contractile genes and low expression of genes related to phenotypic switching, whereas SMC1, SMC2, and SMC3 cells exhibited high expression of genes related to phenotypic switching and relatively low expression of contractile genes (Fig. 2H). The contractile gene set showed a trend of being lower compared to control VSMCs, whereas phenotypic switching gene set showed the opposite trend (Fig. S3). These findings suggest that VSMCs indeed undergo phenotypic switching in atherosclerosis.Fig. 2ScRNA-seq detects phenotypic switching in VSMCs. A, t-SNE visualization of VSMCs with colours indicating identified subtypes. Accepted manuscript Each dot corresponds to one single cell, and each cell cluster is labelled according to cell type and three typical markers B, t-SNE plots showing VSMC subtypes examined in the carotid arteries of control and AS rats. C, Heatmap displaying marker genes used for identifying different cell subsets. D, Numbers (left) and proportions (right) of the four VSMC subsets in the two groups. E, KEGG enrichment analysis of differentially expressed genes in four VSMC subtypes from control and AS rats. F, Compared with those in the control group, Lgals3, Pdgfrb, and Spp1 were significantly increased in the AS group. The adjusted p value (p_val_adj) was 0 for Lgals3, 8.54274E-18 for Pdgfrb, and 3.5545E-174 for Spp1. *** p_val_adj <0.001. G, Heatmap illustrating the expression levels of genes related to SMC contraction and the synthetic phenotype in VSMCs, comparing the control and AS groups. H, t-SNE plots displaying scores for contraction-related and phenotype switching gene sets with colour-coded visualization. Each point represents a single cell. I, Ordering of contractile smooth muscle cells (SMCs, represented by SMC0) and phenotypic switching SMC clusters (SMC1, SMC2, SMC3) along pseudotime in a Monocle2 defined 2D state space. Cell ordering was inferred via Monocle’s differentialGeneTest function, based on expression profiles of the top 500 differentially expressed genes (DEGs) between contractile SMCs and phenotypic switching SMC clusters. Each point corresponds to an individual SMC. J, Monocle components correlated with functional features of contractile and phenotypic switching clusters, including contraction/phenotypic switching scores (calculated via mean expression of cell status-related gene sets). Solid lines: locally weighted smoothing (LOESS) fitting of score-Monocle component relationships. Violin plots (top/bottom corner): functional score distribution across clusters. P-values (Pearson correlation): P<2.2×10⁻¹⁶. VSMC: vascular smooth muscle cell; Con: control rats, Ldlr-/- rats fed a normal diet; AS: atherosclerotic rats, Ldlr-/- rats fed a high-fat high-cholesterol diet and subjected to partial ligation

To visually illustrate the differentiation process of SMCs, we applied the unsupervised inference method Monocle to construct the potential developmental trajectory of SMC populations [35] (Fig. 2I). As shown in the corresponding trajectory visualization, SMC0 and the other three SMC subsets were localized at opposite ends of the trajectory, respectively. Notably, within the trajectory space, SMC1 exhibited closer spatial proximity to contractile SMC0, suggesting that contractile SMCs have the capacity to differentiate into the other three SMC states, with SMC1 potentially acting as an early transitional subtype during this differentiation. To further characterize the trajectory’s dynamic features, we defined contraction and phenotypic switching scores. Integrating these functional scores into trajectory analysis revealed that Component 1 was negatively associated with the contraction score but positively correlated with the phenotypic switching score (Fig. 2J). Collectively, these results indicate that contractile SMCs tend to differentiate into distinct cell populations.

Atherosclerotic plaque-derived EVs mediate VSMC phenotypic switching in vivo

We hypothesized that EVs derived from atherosclerotic plaques contributed to plaque progression via localized stimulation. The expression of proteins associated with EV secretion, namely, RAB27A and VAMP8, was investigated in control and atherosclerotic rats. Immunofluorescence analysis revealed increased RAB27A and VAMP8 expression in atherosclerotic rats than in control rats (Fig. 3A, B). Furthermore, analysis of the GSE28829 dataset revealed elevated VAMP8 expression in late atherosclerotic plaques compared with early atherosclerotic plaques of the human carotid artery (Fig. 3C; P < 0.0001).Fig. 3Effects of EVs on VSMC phenotypic switching in atherosclerosis. A-B, Representative immunofluorescence staining of RAB27A (green), VAMP8 (green) and DAPI (blue) in carotid arteries from control and AS rats (n = 3 rats per group). Scale bar: 50 μm. C. Gene expression of Vamp8 was reanalysed using publicly available RNA-seq results from GSE28829 from the Gene Expression Omnibus (GEO) database, including early plaques (n = 13) and advanced plaques (n = 16). D, Representative immunofluorescence staining of αSMA (red), elastic lamina (green) and DAPI (blue) in carotid artery atherosclerotic rats treated with or without GW4869 (n = 4 rats per group). Low-magnification scale bar: 100 μm; high-magnification scale bar: 50 μm. E, Quantification of the intimal area and the ratio of the intimal area to the medial area. Each dot represents the intimal area of one carotid artery section (left); Each dot represents the intima/media area ratio of one carotid artery section (right) (n = 4 rats per group, 2 sections per rat). F, Representative immunofluorescence staining of αSMA (red), CD68 (green) and DAPI (blue) in carotid arteries (n = 4 rats per group). Low-magnification scale bar: 100 μm; high-magnification scale bar: 50 μm. G, Representative fluorescence images showing intracellular lipid accumulation in carotid arteries (n = 4 rats per group). Low-magnification scale bar: 100 μm; high-magnification scale bar: 50 μm. H-I, Representative images (H) and quantification (I) of αSMA (red), PDGFRB (green), SCA1 (green), and LGALS3 (green) and their colocalization in carotid artery sections (n = 4 rats per group). The data are presented as the mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001, **** P < 0.0001 vs. the AS group. Statistical analysis: Unpaired t test. Con, Ldlr-/- rats fed a normal diet; AS, atherosclerotic rats, Ldlr-/- rats fed a high fat, high-cholesterol diet and subjected to partial ligation

To further investigate the role of EVs in atherosclerotic plaque progression, we used the EV release inhibitor GW4869 to prevent the release of EVs from plaques. The outcomes revealed an amelioration of atherosclerosis in atherosclerotic rats treated with GW4869, as evidenced by a notable reduction in neointima formation (Fig. 3D), a significant decrease in both the neointima area and the intima-to-media area ratio (Fig. 3E; P < 0.0001 for the intima area, P < 0.0001 for the intima/media area ratio compared with atherosclerotic rats without GW4869), and a decrease in macrophage infiltration (Fig. 3F). These findings suggest that EVs are potentially involved in plaque formation.

Compared with those in the model group, the degree of lipid infiltration in the GW4869 group was significantly greater. (Fig. 3G). Additionally, the expression of the VSMC phenotypic markers PDGFRB, SCA1, and LGALS3 was significantly reduced (Fig. 3H, I; P < 0.0001 for PDGFRB; P = 0.0002 for SCA1; P = 0.0286 for LGALS3). These results suggest that EVs are involved in plaque formation and that inhibiting their release could mitigate the phenotypic transition of VSMCs.

We isolated EVs from carotid tissues using differential centrifugation to investigate their characteristics. Transmission electron microscopy confirmed the presence of typical cup-shaped EV structures (Fig. 4A). Western blot analysis revealed markers characteristic of EVs, such as CD63, CD81, and TSG101, but not GAPDH (Fig. 4B). Nanoflow cytometry revealed that the size distribution of the EVs ranged between 40 and 160 nm (Fig. 4C). Consistent with the results of the scRNA sequencing of VSMCs in atherosclerotic carotids, AS-EVs expressed low levels of contractile markers such as Cnn1,* Actg2*,* Eln*, and Speg but high levels of Lgals3,* Spp1*, and Sncg (Fig. 4D). Analysis of EV mRNA and cell cluster-specific transcriptomes preliminarily suggested that macrophages may serve as the primary source of extracellular vesicles (Figure S4).Fig. 4EVs delivered by chitosan hydrogels induce in vivo phenotypic switching of VSMCs. A, Representative TEM images of EVs derived from carotid artery tissues (n = 3 rats). Scale bar: 50 nm. B, Representative immunoblots of EV markers (CD63, CD81 and TSG101) and GAPDH in carotid artery tissue-derived EVs. C, Nanoflow cytometric analyses were conducted on EVs isolated from carotid artery tissues (n = 4 rats per group). The concentration depicted in the figure represents the number of EVs extracted from approximately 1 cm of fresh left carotid artery Accepted manuscript tissue and subsequently resuspended in 100 μl of PBS. D. The quadrant plot displays the expression of characteristic genes in VSMCs and EVs. E, Optical images of the chitosan solution (4 °C) and hydrogel (37 °C). F, Representative fluorescence images of a rat carotid artery treated with a thermosensitive chitosan hydrogel that incorporates PKH26-labelled EVs. Scale bar: 50 μm. G-H. EVs were administered to the vasculature at 4×10^9 particles three times every 10 days. Representative images (G) and quantification (H) of αSMA (red PDGFRB (green), SCA1 (green), and LGALS3 (green) expression and their colocalization in carotid artery sections (n = 4 rats per group). (I) Comparison of the carotid lumen cross-sectional area. Each dot represents the vascular wall area of one carotid artery section (n = 4 rats per group). The data are presented as the mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 vs. the Con-EV group. Statistical analysis: Unpaired t test (H, LGALS3&PDGFRB), or Mann‒Whitney test (H, SCA1; I). VSMC, vascular smooth muscle cell; Con, Ldlr-/- rats fed a normal diet; AS, atherosclerotic rats, Ldlr-/- rats fed a high-fat, high-cholesterol diet and subjected to partial ligation; EVs, extracellular vesicles; Con-EVs, control EVs; AS-EVs, atherosclerotic plaque derived EVs

To further investigate the effects of AS-EVs on the carotid artery in situ, a thermosensitive chitosan hydrogel was used to locally release EVs around the carotid artery. As a paracrine effect, we propose that EVs may migrate from the perivascular space into the intima through mechanisms such as diffusion across the extracellular matrix and uptake by adventitial vessels. This temperature-sensitive hydrogel remained in solution on ice and transitioned into a gel state at 37 °C (Fig. 4E). Immunofluorescence analysis demonstrated the potential of the carotid artery to uptake EVs encapsulated in chitosan hydrogels (Fig. 4F). EVs derived from the carotid artery of control rats were used as control EVs. Compared with those with encapsulated control EVs, those with encapsulated AS-EVs exhibited more pronounced upregulation of VSMC phenotypic switching markers (Fig. 4G-H; P < 0.0001 for PDGFRB; P = 0.0286 for SCA1; P < 0.0001 for LGALS3) and more significant thickening of the vascular walls (Fig. 4I; P = 0.0286).

AS-EVs mediate VSMC phenotypic switching in vitro

VSMCs were exposed to EVs at various concentrations (1 × 10^7, 1 × 10^8, and 1 × 10^9 particles/mL) to assess their effects on the mRNA expression levels of VSMC contractile markers (Fig. 5 A). Immunofluorescence assays confirmed the uptake of EVs by the VSMCs (Fig. 5B). PCR analysis revealed that a concentration of 1 × 10^9 particles/mL significantly suppressed the expression of VSMC contractile markers, including αSMA, MYH11, and TAGLN (P = 0.0039 for Myh11; P = 0.0053 for αSma; P = 0.0160 for Tagln) (Fig. 5 C). Compared with control cells, VSMCs treated with AS-EVs exhibited a morphological shift from an elongated, spindle-like form to a flattened, spreading appearance, accompanied by a significant reduction in the cell length-to-width ratio (Fig. 5D-E; P < 0.0001) and increased expression of the phenotypic transition marker osteopontin (OPN) (Fig. 5D). Furthermore, compared with those in control cells, the protein levels of VSMC contractile markers in VSMCs treated with AS-EVs were markedly lower, as determined by western blot analysis (Fig. 5F-G; P = 0.0286 for MYH11, P = 0.0360 for αSMA, P = 0.0412 for TAGLN). Furthermore, the results of the EdU proliferation and scratch assays indicated that compared with control cells, VSMCs treated with AS-EVs demonstrated increased proliferative capacity (Fig. 5H-J; P = 0.0004) and migratory potential (Fig. 5I-K; P = 0.0059). Collectively, these findings suggested that EVs derived from atherosclerotic plaques promoted the phenotypic switching of VSMCs in vitro.Fig. 5EVs derived from atherosclerotic plaques promote the phenotypic switching of VSMCs in vitro. A, Schematic diagram illustrating the experimental design is provided. VSMCs were serum-starved for 24 h and then treated with Con-EVs or AS-EVs to evaluate their effects. B, Representative immunofluorescence images showing Accepted manuscript the uptake of PKH26-labelled EVs by VSMCs (green) after incubation for 24 h. Scale bar: 20 μm. C, Relative expression levels of contractile genes in VSMCs (n = 3 experiments). D, Representative immunostaining images for contractile markers (MYH11, green; and αSMA, red) and synthetic markers (osteopontin, magenta). Scale bar: 20 μm. E, Statistical analysis of the length‒width ratio of the VSMCs (n = 4 biological replicates per group, and 3 cells were randomly selected from each group for statistics). F-G. Representative immunoblots (F) and densitometric analysis (G) of MYH11, αSMA, and TAGLN in VSMCs (n = 4 experiments). H, Representative images of SMC proliferation. Scale bar: 50 μm. I, Representative phase-contrast images of SMC migration in the scratch assay. Scale bar: 50 μm. J, Quantification of the ratio of the number of EdU-positive cells to the total number of cell. Each dot represents the proportion of EDU positive cells in one field of view (n = 4 biological replicates per group). K, Quantification of the migrated area. Each dot represents the wound healing rate of one field of view (n = 4 biological replicates per group). All the data are presented as the mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001 vs. the control group. Statistical analysis: Unpaired t test (J, K), Mann-Whitney test (E, G), one-way ANOVA with Tukey’s post hoc test (C). VSMC, vascular smooth muscle cell; Con, Ldlr-/- rats fed a normal diet; AS, atherosclerotic rats, Ldlr-/- rats fed a high-fat, high-cholesterol diet and subjected to partial ligation; EVs, extracellular vesicles; Con-EVs, control EVs; AS-EVs, atherosclerotic plaque derived EVs

EVs packed with miR-23a-3p promote VSMC phenotypic switching

A prominent function of EVs is their ability to deliver miRNAs to target cells. We sequenced miRNAs in EVs from control and atherosclerotic rats and validated the differentially expressed miRNAs [32]. KEGG and GO enrichment analyses revealed that genes targeted by upregulated miRNAs were enriched in smooth muscle cell contraction pathways, whereas those targeted by downregulated miRNAs were enriched in pathways related to cell proliferation, migration, adhesion, and extracellular matrix synthesis (Fig. S5). We analysed both the EV miRNA and the scRNA-seq data and compared the target genes of the upregulated miRNAs in AS-EVs with those of the downregulated genes in the VSMCs of atherosclerotic rats. The intersecting genes were enriched primarily in the KEGG pathways associated with VSMC phenotypic switching (Fig. 6A). The results of sequencing analysis suggested that miRNAs transported by EVs might play a role in mediating the phenotypic switching of VSMCs.Fig. 6miR-23a-3p is a key factor in VSMC phenotypic switching induced by atherosclerotic plaque-derived EVs. A, KEGG pathway enrichment of intersecting genes between upregulated miRNAs in AS-EVs and downregulated genes in the VSMCs of atherosclerotic rats. B, The table presents the genes associated with the three pathways (the cytoskeleton in muscle cells, vascular smooth muscle contraction, and focal adhesion) and their corresponding miRNAs, which are upregulated in AS-EVs. Representative immunoblots and densitometric analysis of proteins corresponding to target genes in the lysates of carotid arteries (C, D) and VSMCs (C, E) (n = 3 biological replicates per group). F, In situ hybridization for miR 23a-3p in Con rats, AS rats and human carotid vessels (n = 3 per group). Scale bar: 50 μm. G, After serum starvation for 24 h, miR-23a-3p expression levels in VSMCs stimulated with Con-EVs and AS-EVs for 24 h were detected. Each dot represents the relative expression level of miR-23a-3p in each biological replicate (n = 6 biological replicates per group). H, Representative immunofluorescence images of MYH11 (green) and DAPI (blue) in the different groups. Scale bar: 20 μm. I-J, Representative immunoblots and densitometric analysis of MYH11, αSMA, TAGLN and MYL12B expression in different groups (n = 5 experiments). All the data are presented as the mean ± SD. *P < 0.05, **P <0.01 vs. the control group. Statistical analysis: Unpaired t test (D, E), Mann Whitney test(G), and one-way ANOVA with Tukey’s post hoc test (J). ns, not statistically significant. VSMC, vascular smooth muscle cell; Con, Ldlr-/- rats fed a normal diet; AS, atherosclerotic rats, Ldlr-/- rats fed a high-fat high-cholesterol diet and underwent partial ligation; EVs, extracellular vesicles; Con-EVs, control EVs; AS-EVs, atherosclerotic plaque-derived EVs; NC-i, negative control inhibitor; 23a-3p-i, miR-23a-3p inhibitor

We focused specifically on pathways related to phenotypic switching, including the cytoskeleton in muscle cells, vascular smooth muscle contraction, and focal adhesion. The target genes enriched in these three pathways and their corresponding miRNAs are detailed in Fig. 6B. We subsequently conducted both in vivo and in vitro experiments to identify which gene was truly affected at the protein level. Notably, PPP1R12A and MYL12B expression levels in the carotid tissue of atherosclerotic rats markedly decreased compared with those in the control rats (Fig. 6C, D; P = 0.0431 for PPP1R12A and P = 0.0038 for MYL12B). Furthermore, the MYL12B expression level was significantly lower in the VSMCs treated with AS-EVs than in the control (Fig. 6C, E; P = 0.0256). The upregulated miRNA corresponding to Myl12b in AS-EVs is miR-23a-3p. Thus, AS-EVs potentially facilitated the phenotypic switching of VSMCs via miR-23a-3p and its target gene, Myl12b.

The in situ hybridization results revealed elevated expression of miR-23a-3p in atherosclerotic carotids of rats, which was corroborated by findings in human carotid plaques (Fig. 6F). In vitro experiments demonstrated a significant increase in miR-23a-3p levels in VSMCs stimulated with AS-EVs (Fig. 6G; P = 0.0108). The miR-23a-3p mimic suppressed the expression of contractile VSMC markers (Figure S6; P = 0.0159 for Myh11; P < 0.0001 for αSma; P = 0.0147 for Tagln). Using inhibitors to target miR-23a-3p, the morphological changes in VSMCs caused by AS-EVs were ameliorated, thereby restoring their characteristic elongated spindle shape (Fig. 6H). Additionally, the inhibition of miR-23a-3p expression significantly mitigated the downregulation of the expression of contractile markers in VSMCs induced by AS-EVs (Fig. 6I-J; P = 0.0017 for MYH11, P = 0.0054 for αSMA, and P = 0.0140 for TAGLN). Collectively, these findings suggested that AS-EVs facilitated the phenotypic transition of VSMCs via the function of miR-23a-3p.

miR-23a-3p promotes the VSMC phenotypic transition by targeting Myl12b

As previously mentioned, Myl12b was identified as a target gene of miR-23a-3p. The 3’ untranslated region (3’UTR) of Myl12b contains a binding site for miR-23a-3p (Fig. 7A). The direct interaction between miR-23a-3p and Myl12b was confirmed using a dual-luciferase reporter assay, which demonstrated that mutation of the binding site abolished miRNA-mediated repression (Fig. 7B, P = 0.0032).Fig. 7Impact of the target gene Myl12b on miR-23a-3p-mediated regulation of VSMC phenotypic switching. A, Putative binding sites between MYL12B mRNA and miR-23a-3p. The Accepted manuscript nucleotides highlighted in red represent mutant binding sites and were designed for use in the luciferase reporter assay. B, Statistical results of the dual luciferase reporter assay (n = 4 biological replicates). C, Schematic diagram of the procedure for engineering EVs loaded with miR-23a-3p. D, Representative immunofluorescence images showing the uptake of engineered EVs loaded with cy3-miR-23a-3p by VSMCs (green) after 24 h of incubation. Scale bar: 20 μm E, Immunoblotting images and analyses of MYH11, αSMA, TAGLN, and MYL12B expression in the VSMCs of different groups (n = 4 biological replicates). F, Representative phalloidin staining of G-actin and F-actin in the VSMCs of the different groups. Scale bars, 50 μm. G, Representative immunofluorescence images of nuclear MRTFA in the VSMCs of different groups. Scale bars, 25 μm. The data are presented as the mean ± SD. *P < 0.05, **P < 0.01. Statistical analysis: Unpaired t test (B) or one-way ANOVA with Tukey’s post hoc test (E). VSMC, vascular smooth muscle cell; EVs, extracellular vesicles

Engineered EVs were obtained through coincubation with EV standards and miR-23a-3p mimics, followed by ultrafiltration and resuspension (Fig. 7C). First, an immunofluorescence assay verified the uptake of these engineered EVs by the VSMCs (Fig. 7D). The engineered EVs loaded with miR-23a-3p were found to inhibit MYL12B expression (Fig. 7E, P = 0.0010,). Lentiviral transfection was utilized to overexpress Myl12b in VSMCs. Myl12b overexpression significantly ameliorated the downregulation of VSMC contractile markers induced by engineered EVs loaded with miR-23a-3p (Fig. 7E; P = 0.0144 for MYH11, P = 0.0088 for αSMA, P = 0.0368 for TAGLN), demonstrating that Myl12b is pivotal for the miR-23a-3p-mediated regulation of VSMC phenotypic transition.

MYL12B modulates the mutual conversion between F-actin and G-actin [36]. The G-actin concentration in the cytoplasm regulates the nuclear translocation of MRTFA, which promotes the transcription of contractile genes in VSMCs [37]. We found that engineered EVs loaded with miR-23a-3p promoted F-actin depolymerization and concurrent MRTFA nuclear egress (Fig. 7F–G), whereas Myl12b overexpression restored F-actin polymerization and facilitated MRTFA nuclear entry (Fig. 7F-G).

Discussion

This study revealed that AS-EVs suppressed the target gene Myl12b by transferring miR-23a-3p to VSMCs. This mechanism promoted actin depolymerization, leading to the nuclear translocation of MRTFA and, consequently, facilitating the phenotypic switching of VSMCs (Fig. 8). During this process, AS-EVs stimulate VSMCs locally through a paracrine mechanism, exacerbating atherosclerosis progression. A chitosan-based thermosensitive hydrogel was employed to deliver AS-EVs to the carotid arteries, mimicking the localized release of EVs. These findings further demonstrated that AS-EVs could induce the phenotypic switching of VSMCs in situ.Fig. 8**Schematic diagram illustrating how AS-EVs stimulate vascular remodelling in the microenvironment. **Atherosclerotic plaque cells secrete atherogenic EVs containing miR-23a-3p into the interstitial space. EVs are absorbed by VSMCs within the plaque microenvironment. MiR-23a-3p is then released into VSMCs, where it targets Myl12b, promoting actin depolymerization. This results in the translocation of MRTFA from the nucleus to the cytoplasm, thereby inhibiting the expression of contractile genes. Consequently, this leads to vascular remodelling and ultimately exacerbates the progression of the lesion. AS-EVs, atherosclerotic plaque-derived EVs

In this study, we developed a carotid atherosclerosis model in Ldlr^−^/^−^ rats using a combination of a high-cholesterol diet and partial ligation of the carotid artery. Rats fed only the high-cholesterol diet exhibited slower atherosclerosis progression and delayed plaque formation in the carotid artery. The ligation method was chosen to promote localized plaque formation, enhancing experimental reproducibility and facilitating subsequent analyses of the plaque microenvironment. Rats were selected for this study because of their abundant carotid artery tissue, which is conducive to isolating EVs from the carotid artery for functional characterization. In this preliminary mechanistic study, we exclusively utilized male rats to reduce variability in this preliminary mechanistic investigation, acknowledging that female rodents experience hormonal fluctuations throughout the oestrous cycle that could introduce additional variability to the results. We recognize this as a limitation, as it constrains the applicability of our findings to female physiology. However, considering emerging evidence of sex-specific differences in atherosclerotic progression, future research should include female rats to examine the influence of sex on carotid plaque pathophysiology [38–40].

VSMC phenotypic switching was observed in our models of carotid atherosclerosis. In recent years, the significance of VSMC research has been markedly enhanced by the widespread implementation of lineage tracing and single-cell sequencing technologies [10, 41]. Consistent with previous studies, our research confirmed that the expression of contractile markers was downregulated in atherosclerosis models [42]. Notably, the markers that are indicative of phenotypic switching are heterogeneous and vary with the stage of atherosclerosis [1]. The markers can be classified into three categories: (1) markers that are upregulated following phenotypic switching but do not indicate the direction of cellular differentiation, such as Klf4, Oct4, and Pdgfrb [43–45]; (2) markers that are upregulated and represent an intermediate state, such as Lgals3 and Sca1, and intermediate-state VSMCs can further transform into more differentiated cells under pathological conditions [46–48]; and (3) markers that are upregulated indicate the direction of phenotypic switching. For example, Opn, Fn1, and Dcn are markers of a phenotype associated with extracellular matrix synthesis, whereas Cd68 and Mac3 are markers of a macrophage-like phenotype [1, 34]. This study focused primarily on three markers, Pdgfrb, Lgals3, and Sca1, to underscore the phenotypic switching of VSMCs without delving into specific differentiation pathways. Further exploration could investigate various differentiation endpoints and their implications for plaque vulnerability. The results of single-cell analysis not only suggested the presence of phenotypic switching in VSMCs during atherosclerosis but also indicated that these cells were in various stages of transdifferentiation (Fig. 2E, Fig. S2). These findings were consistent with results from other atherosclerosis models and human atherosclerotic tissues reported in the literature [33, 49].

Consistent with the phenotypic switching dynamics observed in our carotid atherosclerosis models, VSMC phenotypic plasticity, as a conserved pathological hallmark, is also present in other aortic remodelling disorders, such as aortic dissection [50]. Indeed, VSMC phenotypic switching in both atherosclerosis and aortic dissection shares core features, reflecting a conserved adaptive response of VSMCs to pathological stimuli [51]. This phenotypic transition disrupts vascular homeostasis. In atherosclerosis, phenotypically switched VSMCs can transform into fibromyocytes, fibrochondrocytes, and proinflammatory/macrophage-like cells, promoting intimal thickening and plaque formation [34]. In aortic dissection, these cells transdifferentiate into fibro-like and lipo-like cells, which drive medial layer degradation by increasing the secretion of proteolytic enzymes (such as matrix metalloproteinases) that breakdown elastic fibres and collagen, thereby advancing the degenerative remodelling of the arterial wall [35]. Notably, although VSMC phenotypic plasticity is present in both processes, the specific subtypes of transformed cells and their functional priorities vary with disease stage. Further investigations are needed to elucidate how the types and proportions of these phenotypically switched cells evolve across distinct stages of disease progression, which could provide insights into disease-specific therapeutic targets.

Our study revealed that AS-EVs promote the phenotypic switching of VSMCs in situ and in vitro, influencing cell contractility and contributing to vascular remodelling. We provide the first evidence that in situ microenvironmental alterations are mediated by AS-EVs. Unlike previous studies in which EVs were typically derived from in vitro cultured cell lines, we identified limitations of such EVs because of their lack of representation of the true pathological microenvironment [27]. Instead, our research focused on AS-EVs. We found that GW4869, a pharmacological inhibitor of EV release, alleviated the development of atherosclerosis and suppressed VSMC phenotypic switching (Fig. 4G-I). However, it is important to acknowledge that GW4869 was administered via intraperitoneal injection, a route that may systemically inhibit the secretion of EVs across multiple cell types, including potentially beneficial EVs. Notably, current technological constraints present challenges in achieving precise inhibition of EVs specifically within atherosclerotic plaques [52]. Although the intraperitoneal administration of GW4869 effectively reduces EV release, it lacks the necessary specificity. Furthermore, as demonstrated by Lallemand et al., it is plausible that the beneficial effects of GW4869 on carotid atherosclerosis may also be mediated through the suppression of proinflammatory cytokines, independent of its effects on EVs [53]. Taken together, these considerations highlight the need to exercise caution in interpreting the effects of GW4869 as being exclusively or primarily due to its inhibition of EVs. We found that proteins associated with EV secretion, such as RAB27A and VAMP8, were highly expressed in atherosclerotic rats (Fig. 3A and B). This evidence suggested that AS-EVs might be involved in the progression of atherosclerosis.

Numerous studies have emphasized the importance of identifying the origin of EVs [54]. In this study, we hypothesized that macrophages might serve as a potential cellular source within atherosclerotic plaques responsible for pathological EV secretion. Concordance analysis between differentially upregulated mRNAs in AS-derived EVs (compared with those in control EVs) and highly expressed genes across cell type-specific clusters revealed increased macrophage activity and a greater propensity for EV secretion (Fig. S4A, S5A). Furthermore, an analysis of differentially expressed mRNAs in EVs and cell type-specific differential genes further confirmed that macrophages are more likely to secrete EVs (Figure S4B, S5B). Overall, macrophages appear to be more active and are likely the source cells of EVs. To experimentally validate this prediction, we detected the macrophage-specific marker F4/80 on isolated AS-EVs using immunoblotting (Figure S4C), providing direct evidence that at least a subset of AS-EVs are of macrophage origin. This observation aligns with our previous report demonstrating that AS-EVs express macrophage marker [32]. Collectively, these data reinforce the notion that macrophages are likely the source cells of AS EVs. Although these results support the conclusion that macrophages are likely the major source of pathological EVs in atherosclerosis, future studies employing a macrophage-specific Cre-LoxP lineage-tracing system will be necessary to definitively quantify the contribution of macrophages to AS-EVs.

MiRNAs are key components found in EVs and contribute significantly to their pathogenesis mechanisms [30, 55]. Our study demonstrated that AS-EVs influenced the nuclear localization of MRTFA through the miR-23a-3p–Myl12b axis, ultimately affecting VSMC phenotypic switching. The contractile phenotype of VSMCs is sustained through the interaction between serum response factor (SRF) and its cofactor, myocardin-related transcription factor A (MRTFA), with the CArG box [56]. MRTFA, a member of the myocardin family, functions as a coactivator of SRF. MRTFA activity is modulated by cytoskeletal dynamics. Unpolymerized G-actin inhibits MRTFA activity. When G-actin polymerizes into F-actin, the cytoplasmic concentration of G-actin decreases, allowing MRTFA to translocate into the nucleus. In the nucleus, MRTFA binds to SRF, thereby initiating the transcription of contractile genes [36, 57, 58]. Myl12b, a target of miR-23a-3p, belongs to the myosin regulatory subunit family and is associated with actin polymerization [59].

A key strength of this study is its elucidation of how AS-EVs from the atherosclerotic microenvironment stimulate lesion progression. AS-EVs may be secreted by macrophages, with miR-23a-3p within these EVs regulating actin depolymerization via the target gene Myl12b, thereby inducing VSMC phenotypic switching. MiR-23a-3p is a microRNA that plays diverse roles in pathological processes, including angiogenesis, tumour cell metastasis, and drug resistance in cancer [60–62]. Additionally, miR-23a-3p promotes the development of atherosclerotic lesions by enhancing inflammatory responses [32, 63]. In this study, we revealed that AS-EVs carrying miR-23a-3p can promote VSMC phenotypic switching, indicating that EVs and miR-23a-3p might serve as potential therapeutic targets for preventing and treating atherosclerosis. However, current research shows that there are no drugs available to modulate the secretion of EVs in atherosclerosis. Moreover, if drugs are used to inhibit the secretion of EVs, beneficial EVs may also be suppressed. Future research should focus on the heterogeneity of EVs, with the aim of targeting the secretion of specific types of EVs. Moreover, miR-23a-3p shows promise as both a biomarker for atherosclerosis and a therapeutic target, warranting further investigation into its role in disease progression and treatment.

However, this study has several limitations. First, the absence of lineage tracing technology limits the ability to definitively track VSMC phenotypic switching. Although experiments and single-cell sequencing provide indirect evidence, lineage tracing remains the gold standard for tracking smooth muscle cells that have lost their markers. Second, although it is hypothesized that AS-EVs are secreted by macrophages, further experimental validation is needed. Third, current approaches cannot locally suppress EV release within plaques, and this issue requires further investigation.

Conclusion

Phenotypic switching of VSMCs and vascular remodelling occur in atherosclerosis, as determined by single-cell sequencing, confirming that VSMCs can transdifferentiate into various cell types. Chitosan hydrogels loaded with EVs locally promoted the phenotypic switching of carotid VSMCs and thickening of the vascular wall. Mechanistically, in carotid atherosclerosis, EVs containing miR-23a-3p, which targeted Myl12b, were released, thereby promoting nuclear translocation of MRTFA and facilitating VSMC phenotypic switching. The inhibition of miR-23a-3p eliminated this effect, whereas the overexpression of Myl12b alleviated miR-23a-3p-induced VSMC phenotypic switching. This study elucidated a novel mechanism underlying the self-aggravation of atherosclerosis. EVs and their miRNA contents represent potential novel therapeutic targets for the prevention and treatment of atherosclerotic diseases.

Supplementary Information

Below is the link to the electronic supplementary material.Supplementary Material 1 (DOCX 2.21 MB)