Endothelial cells under disturbed flow release extracellular vesicles to promote inflammatory polarization of macrophages and accelerate atherosclerosis

Abstract

Background

Extracellular vesicles (EVs) derived from endothelial cells (ECs) are increasingly recognized for their role in the initiation and progression of atherosclerosis. ECs experience varying degrees and types of blood flow depending on their specific arterial locations. In regions of disturbed flow, which are predominant sites for atherosclerotic plaque formation, the impact of disturbed flow on the secretion and function of ECs-derived EVs remains unclear. This study aims to assess the role of disturbed flow in the secretion of EVs from ECs and to evaluate their proatherogenic function.

Results

Our comprehensive experiments revealed that disturbed flow facilitated the secretion of ECs-derived EVs both in vivo and in vitro. Mechanistically, the MAPK pathway transduces mechanical cues from disturbed flow in ECs, leading to increased secretion of EVs. Pharmacological inhibition of the MAPK pathway reduced the secretion of EVs even under disturbed flow conditions. Interestingly, under disturbed flow stimulation, ECs-derived EVs promoted monocyte accumulation and enhanced their invasion of the endothelium. More important, these EVs initiated the inflammatory polarization of macrophages from the M2 to the M1 phenotype. However, the phenotypic switching of vascular smooth muscle cells was not affected by exposure to these EVs.

Conclusions

Taken together, targeting the MAPK signaling pathway holds potential as a novel therapeutic strategy for inhibiting the secretion of EC-derived EVs and mitigating the inflammatory polarization of macrophages, ultimately ameliorating the progression of atherosclerosis.

Graphical Abstract

Supplementary Information

The online version contains supplementary material available at 10.1186/s12915-025-02125-x.

Background

Atherosclerosis is the prevalent pathological character of many vascular diseases, such as acute coronary syndromes, stroke, and peripheral arterial disease [1, 2]. In the last decades, it has remained the dominant cause of death and disability worldly. Endothelial cells (ECs) comprise the inner layer of the normal artery to keep vascular homeostasis by protecting the intact endothelium permeability [3]. Pathological activation and dysfunction of endothelial cells are critical for the initiation and progression of atherosclerosis. After being activated by various pro-inflammatory cytokine and cardiovascular risk factors, including body mass index, systolic blood pressure, non-HDL cholesterol, smoking, and diabetes [4], ECs lose physiological function and recruit monocytes to the endothelial layer followed by chemoattractant cytokines-induced invasion of these monocytes into the intima [2]. Within the intima, monocytes take lipids and differentiate into macrophages [5]. Subsequently, the monocytes-differentiated macrophages became the major contributors to inflammation through their secretion of pro-inflammatory factors and their eventual death (necrosis or apoptosis) [6]. After the initiation of atherosclerotic lesions, vascular smooth muscle cells (VSMCs) subsequently promote the pathophysiological process of atherosclerosis. VSMCs undergo phenotype switching, acquiring increased proliferative and migratory capabilities [7, 8] and release matrix metalloproteinases to break the extracellular matrix (ECM), finally exacerbating the accumulation of lipids and pro-inflammatory cells, such as dying foam cells [8]. Together, the dysfunctional ECs is a trigger for atherosclerosis initiation, which is followed by pathological responses from other vascular cells. Given the critical role of ECs mediating pathological responses from other cells during the initiation of atherosclerosis, directly inhibiting the crosstalk between the dysfunctional ECs and other vascular cells in atherosclerosis holds therapeutic promise, especially for those patients who fail to meet the criteria for stent implantation or exhibit intolerance to current lipid-lowering drugs. However, it raises another interesting question: how do ECs mediate the complex interplay between these pathological responses from other vascular cells?

Emerging evidence has shown that extracellular vehicles (EVs), including exosomes, microvesicles, and apoptotic bodies, mediate cell-to-cell communication by transferring their protein, lipid, and nucleic acid content to target cells [9, 10]. It has been described that the accumulation of EVs is associated with subclinical atherosclerotic lesions [11, 12] and suggested that the EVs derived from ECs are the potential contributor to the crosstalk between dysfunctional ECs and other vascular cells under the proatherogenic stimulus [13]. However, an argument exists for the potential proatherogenic and antiatherogenic effects of ECs-derived EVs. For example, microRNAs miR-145 and miR-143 in ECs-derived EVs maintained VSMCs differentiation, while miR-126 cargo in endothelial EVs activates the phenotypic switching of VSMCs [9]. The debate regarding the effect of ECs-derived EVs is primarily due to the limitation of current research. Most of the previous research stems from EVs isolated from cultured cells without the in vivo pathological condition of atherosclerosis. For instance, composed of the inner layer of the artery, ECs are exposed to a fluid shear stress that varies with magnitude and pattern generated by the flowing blood [14, 15]. The atherosclerotic lesion preferentially develops in branching points or curvatures of the arterial, where ECs sense the disturbed blood flow, not the laminar blood flow [16]. Work in the last decades has identified that disturbed flow (DF) activates various endothelial signaling pathways and downstream events to exacerbate the dysfunction of ECs [17], such as ROS production [18], destabilizing cell–cell junction [19], and inflammation [20]. However, the effect and underlying mechanobiological mechanism of disturbed flow regulating the secretion of ECs-derived EVs remains unknown. On the other hand, the proatherogenic or antiatherogenic function of ECs-derived EVs upon disturbed flow is not investigated clearly.

Results

Increased EVs are accompanied by atherosclerotic lesions and collocated in endothelial cells

Increasing evidence implicates the critical role of EVs in the initiation and progression of atherosclerosis [9]. To validate the relationship between EVs and atherosclerotic arteries, we set up an atherosclerotic mouse model using ApoE^−/− mouse characterized by abundant lipid deposition in the whole artery (Fig. S1A and B). Compared to normal arteries, atherosclerotic arteries showed a noticeably increased number of EVs both in artery tissue and serum (Fig. 1A and B). In cardiovascular disease, EVs originate from diverse donor cells, including endothelial cells, fibroblasts, platelets, smooth muscle cells, leucocytes, monocytes, and macrophages [21]. Among them, endothelial cell-derived EVs are the major sources in patients with atherosclerotic artery disease [22]. It has been shown that EVs derived from activated endothelial cells reflect early vascular dysfunction in those specific patients [23], suggesting that the EVs derived from endothelial cells are a hallmark contributor to the formation of atherosclerosis. Accordingly, we also confirmed the colocalization of endothelial cells marker CD31 and EVs marker Alix, Tsg101. A positive correlation between CD31 expression and EVs release is observed. In the atherosclerotic mouse atherosclerotic artery, Alix showed a higher expression in endothelial cells (Fig. 1C and D). In addition, the normalized fluorescence of Tsg101 in the atherosclerotic artery’s inner layer is increased as well (Fig. 1E and F).

Fig. 1.

The atherosclerotic plaque is associated with increased ECs-derived EVs. A, B Transmission electron microscopy (TEM) and nanoparticle tracking analysis (NTA) were performed to determine the amount and morphology of EVs harvested from atherosclerotic mouse artery tissue and serum, respectively. Scale bar = 200 nm, n = 6. C–F Representative immunofluorescence images and quantitation analysis showing the location and expression of EVs marker Alix or Tsg101 and endothelial cells marker (CD31) in mouse atherosclerotic artery. ApoE^−/− mice were fed a western diet for 3 weeks. Alix (red) or Tsg101 (red) and CD31 (green). The nucleus was stained with DAPI (blue). The white arrows indicate the colocalization of EVs and endothelial cells. M mouse, L lumen. Scale bar = 20 µm, n = 3. ^***P < 0.001. G–J Representative immunofluorescence images and quantitation analysis showing the location and expression of EVs marker Alix or Tsg101 and endothelial cells marker (CD31) in the human atherosclerotic samples. Alix (red) or Tsg101 (red) and CD31 (green). The nucleus was stained with DAPI (blue). The white arrows indicate the colocalization of EVs and endothelial cells. H human, L lumen. Scale bar = 20 µm, n = 3. ^***P < 0.001

Similarly, human atherosclerotic arteries also revealed increased colocalization of CD31 and EVs markers (Fig. 1G–H and I–J). The HE and immunohistochemistry staining were used to validate the human atherosclerotic artery. The disorganized smooth muscle layer and lipid accumulation show the pathological characteristic of atherosclerosis (Fig. S1C). In line with this, an enhanced inflammatory response was indicated by the higher expression of CD 68, the marker of pro-inflammatory macrophages [24] (Fig. S1D).

Disturbed flow facilitates the secretion of extracellular vesicles derived from endothelial cells

Vascular heterogeneity and homeostasis were regulated by different fluid shear stress due to constant exposure to different blood flows with various velocities and patterns [15]. Interestingly, atherosclerosis preferentially occurs in the bifurcation, branching, and bending of blood vessels exposed to disturbed flow [16]. The endothelial cell layer senses disturbed flow, leading to dysfunction and inflammation. However, the effect of disturbed flow regulating the secretion of EVs derived from ECs remains unknown. Accordingly, we loaded disturbed flow on endothelial cells in vivo and in vitro to unravel this unclear effect. Firstly, the partial left carotid artery (LCA) ligation was used to load disturbed flow in vivo [25], and the surgery diagram is shown in Fig. S2A–B. After ligating, ultrasound shows flow velocity profiles and reveals that partial ligation induces flow reversal (Fig. S2C) in LCA. After 3 weeks of a high-fat diet post-ligation, the lipid lesion was determined by oil red O staining. It shows strong evidence that the disturbed flow in partial ligated LCA and artery arch (AA), where the endothelial cells sense the intrinsic disturbed flow, exacerbates atherosclerosis (Fig. S2D). It was supported by the HE staining results. The lumen loss and accumulation of inflammation cells were observed in the disturbed flow region (Fig. S2E and F). Together, we successfully set up the disturbed flow loading system in vivo. Next, we examined the secretion of EVs from endothelial cells under disturbed flow.

Immunostaining revealed that Tsg101 was upregulated in endothelial cells in the ligated LCA, compared with those in normal LCA, where the endothelial cells sense the laminar blood flow. In contrast, endothelial cells in the unlighted right carotid artery (RCA) showed a stable Tsg101 expression in the ligated as well as the control group mice (Fig. 2A and C). In line with this, the Alix expression also shows an increased tendency in the ligated artery while keeping stable in the laminar flow (LF) region (Fig. 2B and D). Similarly, Tsg101 and Alix exhibited increased endothelial expression in the AA, another disturbed flow region (Fig. S3A–D). To identify the pro-secretion role of the disturbed flow on EVs, we expose the endothelial cells to disturbed flow in vitro and determine the amount of ECs-derived EVs. Distinctly, the elevated amount of EVs from endothelial cells induced by disturbed flow was observed (Fig. 2E–G). Furthermore, the western blot analysis supported that with increased expression of Rab27a, a promoter for EVs secretion [26], and decreased expression of Rab7, an inhibitor for EVs secretion [27] (Fig. 2H–J). Collectively, it gives solid evidence that the disturbed flow promotes the release of EVs derived from endothelial cells.

Fig. 2.

Disturbed flow is the promoter of secreting ECs-derived EVs in proatherogenic area. A–D Immunofluorescence and quantitation analysis of Tsg101 or Alix in LCA or RCA with or without partial ligation treatment. Green: CD31; red: Tsg101 or Alix; blue: nucleus. L lumen. Scale bar = 20 μm, n = 3. ^*P < 0.05, ^**P < 0.01. E TEM assay was performed to identify the EVs harvested from HUVECs subjected to laminar flow (LF) or disturbed flow (DF). F–G Concentration and size range of EVs were determined by NTA. ^**P < 0.01, n = 3. H–J Western blot and quantification of Rab27a and Rab7 in HUVECs subjected to LF or DF. ^**P < 0.01, n = 3

MAPK signaling initiates the disturbed flow-based mechanical clue to release endothelial EVs

Given the involvement of the disturbed flow in regulating endothelial EVs secreting and their function, our subsequent objective was to investigate the underlying biomechanical mechanisms responsible for the release of endothelial EVs under disturbed flow. We performed transcriptome sequencing of endothelial cells exposed to disturbed flow. In comparison to endothelial cells under laminar flow, the volcano plot showed 158 upregulated and 327 downregulated genes in endothelial cells upon disturbed flow stimulation (Fig. 3A). Gene ontology enrichment analysis also showed that the differently expressed genes after disturbed flow stimulation could be enriched in the functions related to EVs (Fig. 3B). Next, we conducted a KEGG pathway analysis based on the transcriptome sequencing results. Remarkably, the MAPK signaling pathway was found and emerged as the most potent transducing clue mediating disturbed flow in endothelial cells (Fig. 3C). To further validate this, we examined the expression of several essential components of the classical MAPK signaling pathway in endothelial cells exposed to disturbed flow. After exposure to disturbed flow, the upstream components of the MAPK signaling pathway Ras and Raf1 significantly increased in endothelial cells. On the other hand, the downstream components ERK (extracellular signal-regulated kinase) showed a declined expression tendency accompanied by enhanced phosphorylated ERK1/2 (p-ERK), which results in an increased ratio of p-ERK/ERK (Fig. 3D–G). In addition, we performed immunostaining to identify the expression of Ras, Raf1, ERK1/2, and p-ERK1/2 in endothelial cells after different flow loading. It confirmed the same expression tendency in the western blot assay and showed the nuclear translocation of p-ERK1/2 (Fig. 3H–K). Furthermore, several genes related to MAPK signaling pathway also showed significantly different expression (Fig. S4A–D). The above results suggested that the MAPK signaling pathway was activated in endothelial cells after subjecting to disturbed flow.

Fig. 3.

MAPK signaling pathway response to disturbed flow in endothelial cells. A Transcriptome sequencing analysis of HUVECs subjected to different flow stimulation. Red: upregulated genes, blue: downregulated genes; gray: non-significant genes. B Gene ontology (GO) enrichment for differentially expressed genes in HUVECs after different flow stimulation. C KEGG pathway analysis for differently expressed genes in HUVECs after exposure to disturbed flow. D–G Western blot assay and quantitation analysis of Ras, Raf1, ERK1/2, and p-ERK1/2 in HUVECs upon disturbed flow stimulation. LF laminar flow, DF disturbed flow. β-actin was used as the internal control. ^*P < 0.05, ^**P < 0.01, n = 3. H–K Representative immunofluorescence images and quantitation analysis of Ras, Raf1, ERK1/2, and p-ERK1/2 expressions in HUVECs upon disturbed flow stimulation. Red: Ras, Raf1, ERK1/2, and p-ERK1/2 proteins, green: F-actin, blue: DAPI. Scale bar = 20 μm. ^*P < 0.05, ^***P < 0.001, n = 3

To further investigate the effect of the MAPK signaling pathway on the secretion of endothelial EVs under disturbed flow. Initially, the PD98059 inhibitor was employed to suppress the activation of the MAPK signaling pathway under disturbed flow stimulation. After exposure to PD98059, the expression of Ras and Raf1 does not increase even under disturbed flow stimulation. Meanwhile, the unchanged ratio of p-ERK/ERK showed inactivation of the MAPK signaling pathway (Fig. 4A–B). Accordingly, inhibition of the MAPK signaling pathway with the PD98059 inhibitor resulted in declined Rab27a and Rab7 expression (Fig. 4C–E). This expression tendency of Rab27a and Rab7 was further confirmed by immunostaining (Fig. 4F–H). Subsequently, NTA (nanoparticle tracking analysis) and BCA (bicinchoninic acid assay) assays were conducted to examine the number and protein concentration of endothelial EVs after inhibiting the MAPK signaling pathway. The decreased amount and protein concentration of endothelial EVs were observed even under disturbed flow stimulation (Fig. 4I–L). These results provide solid evidence that the MAPK signaling pathway is central for the disturbed flow-induced endothelial EVs. However, another interesting question arises: what is the function of EVs derived from endothelial cells upon disturbed flow stimulation? Our subsequent objective was to detect the effect of endothelial EVs under disturbed flow in the initiation and progression of atherosclerosis.

Fig. 4.

Inhibition of MAPK signaling pathway blocked the secretion of ECs-derived EVs under disturbed flow. A–B Western blot assay and quantification analysis of key factors in MAPK pathway in HUVECs pre-treated with or without PD98059 upon the disturbed flow stimulation, β-actin was used as the internal control. ^**P < 0.01, ^***P < 0.001, n = 3. C–E Western blot assay and quantification analysis of Rab27a and Rab7 in HUVECs with or without PD98059 treatment upon disturbed flow stimulation, β-actin was used as the internal control. ^*P < 0.05, n = 3. F–H Representative immunofluorescence images and quantitative fluorescence intensity analysis of Rab27a and Rab7 in HUVECs pre-treated with or without PD98059 upon disturbed flow stimulation. Red: Rab27a, Rab7; green: F-actin; blue: DAPI. Scale bar = 20 μm.^*P < 0.05, ^**P < 0.01, ^***P < 0.001, n = 3. I–J The amount and size range of ECs-derived EVs were determined by NTA assay. ^***P < 0.001, n = 3. K–L The protein concentration in ECs-derived EVs was detected by the BCA assay. ^***P < 0.001, n = 3

Endothelial EVs under disturbed flow accelerates the progression of atherosclerosis

Monocytes recruited to the endothelium is an initial hallmark of atherosclerosis [28]. The circulating monocytes in the bloodstream preferentially bind to adhesion molecules expressed by activated endothelial cells located in the disturbed flow region [29]. Here, we validated the tendency of monocytes towards recruitment in disturbed flow regions. In the in vitro disturbed flow loading system, more monocytes adhered to the endothelial layer was observed (Fig. 5A and B). However, the reason why monocytes likely accumulated in the endothelium under disturbed flow is not apparent. Subsequently, we subjected monocytes to endothelial EVs with different flow exposures to reveal the role of EVs in mediating cellular communication between monocytes and endothelial cells. After staining with DiI, a lipophilic cell membrane dye, the endothelial EVs under disturbed flow were less uptaken by monocytes (Fig. 5C and D). The flow cytometry assay further verified the results (Fig. 5E and F). Interestingly, the endothelial EVs under disturbed flow promote the trans-endothelium migration of monocytes (Fig. 5H and I). Taken together, it suggested that endothelial EVs under disturbed flow accelerated the recruitment of monocytes on the endothelium, followed by trans-endothelium migration.

Fig. 5.

ECs-derived EVs induce the endothelium adhesion of THP-1. A–B Representative fluorescence images and quantitation analysis of the adherent THP-1 to HUVECs with LF or DF stimulation. Red: mCHerry-HUVECs; green: GFP-THP-1. Scale bar = 100 μm. ^**P < 0.01, n = 3. C–D Representative fluorescence images and quantitative analysis of endothelial EVs labeled with DiI are uptaken by THP-1. Red: EVs, green: F-actin; blue: nucleus. Scale bar = 20 μm.^*P < 0.05, n = 3. E–F Flow cytometry showed that the ECs-derived EVs were uptaken by THP-1 after LF or DF stimulation. ^***P < 0.001, n = 3. G Schematic diagram of transwell model for detecting the trans-membrane ability of THP-1 subjected to different endothelial EVs. H–I Crystal violet staining and analysis of THP-1 after penetrating HUVEC monolayer. Blank: only HUVEC cells seeded on the insert of the transwell. Scale bar = 200 μm. ^**P < 0.01, ^***P < 0.001, ns denotes not significant, n = 3

Macrophages, the major immune cells in atherosclerotic plaques, were differentiated from monocytes after their penetration into artery intima and play a key role in promoting atherosclerosis [30]. The internalization of lipids in macrophages reflects their polarization and eventually drives macrophages into foam cells, finally resulting in a necrotic core [31]. Generally, in the ex vivo study, macrophages were divided into three main subpopulations, including M0, M1, and M2 type. The M0 type is also called naive macrophages. The M1 type is referred to as pro-inflammatory macrophages, while the M2 type is known as anti-inflammatory macrophages [32]. Various stimuli, such as cytokines, lipids, and iron can influence inflammatory polarization of macrophage (M2 to M1) contributing to atherosclerotic plaque formation. Nonetheless, the impact of endothelial EVs on macrophages inflammatory polarization has not been identified. Therefore, we next examined the inflammatory polarization of macrophages derived from monocytes after PMA incubation under different endothelial EVs stimulation. Oil red O staining showed that lipid droplets accumulated in M1 macrophages with LPS induction and DF-EVs. In contrast, neither the M2 macrophages induced by IL-4 nor macrophages incubated with LF-EVs exhibit lipid accumulation (Fig. 6A). Immunofluorescence staining revealed that M2 macrophages exposed to endothelial EVs under disturbed flow conditions exhibited increased expression of CD86 and CD80, markers indicative of M1 macrophages. Conversely, the expression of CD206, a marker associated with M2 macrophages, was reduced (Fig. 6B). These findings were further corroborated by flow cytometry analysis. Incubation with DF-EVs suppressed the anti-inflammatory M2 phenotype and enhanced the pro-inflammatory M1 phenotype (Fig. 6C–F). Additionally, we subsequently measured the effect of DF-EVs on the M0 and M1 macrophages. It suggested that DF-EVs exposure had no discernible effect on M1 polarization in M0 or M1 macrophages (Fig. S5A–F). We further validated the role of DF-EVs in promoting pro-inflammatory polarization of M2 macrophages by detecting the expression of inflammation-related genes. Accordingly, all the pro-inflammatory genes, as M1 macrophages markers, exhibited higher expression after incubating with DF-EVs (Fig. S5G–H). On the contrary, anti-inflammatory genes, as M2 macrophages markers, were downregulated (Fig. S5I–J). The above results showed that endothelial EVs subjected to disturbed flow promoted the inflammatory polarization of macrophage and the formation of foam cells, which together facilitate atherosclerosis progression.

Fig. 6.

ECs-derived EVs accelerate the inflammatory polarization of M2 macrophages. A Oil red O staining to detect the formation of foam cells differentiated from M0 macrophages. Scale bar = 200 μm. B Representative immunofluorescence images of phenotypic switching of macrophages with different endothelial EVs stimulation. M2 (CD206⁺, green) to M1 (CD86⁺ and CD80⁺, red), blue: nucleus. Scale bar = 20 μm. C–F Flow cytometry analysis for validating phenotypic switching of macrophages after different endothelial EVs stimulation. M2 (CD206⁺), M1 (CD86⁺ and CD80⁺). Mean fluorescence intensity (MFI) values were employed to statistically evaluate the different marker expression. ^*P < 0.05, ^***P < 0.001, ns denotes not significant, n = 3

On the other hand, “contractile to synthetic” phenotype transition of vascular smooth muscle cells (VSMCs) is a critical factor in the development of atherosclerosis [33]. Depending on the phenotypic switching, VSMCs acquired enhanced proliferative and migratory capabilities that exacerbate plaque formation [8]. The contractile phenotype is characterized by markers such as smooth muscle α-actin (α-SMA), smooth muscle protein 22-α (SM22α), calponin, and myosin heavy chain 11. In contrast, the synthetic phenotype is indicated by markers like matrix metalloproteinase-2 (MMP2) [7]. Our previous study demonstrated that increased interstitial flow initiated the phenotypic switching of VSMCs, and the phenotypic switching of VSMCs is the upstream inducer for the secretion of pro-calcified EVs from VSMCs [34]. Notably, the donor cell and recipient cell are both VSMCs, and we wonder how VSMCs respond to the EVs from other cell origins. Next, we sought to determine whether endothelial EVs under different flow conditions can alter the phenotypic switching of VSMCs in vitro. Flow cytometry assay showed that HASMCs were able to phagocytose more DF-EVs (Fig. S6A and B), which is confirmed by the fluorescence staining (Fig. S6C and D). Interestingly, HASMCs did not undergo phenotypic switching after subjecting to DF-EVs. Western blot and qRT-PCR both confirmed that the markers indicating contractile and synthetic phenotype keep stable expression (Fig. S6E–O).

Discussion

Until 2011, “extracellular vesicles” (EVs) was used as a generic term to define all lipid bilayer-enclosed extracellular structures [35]. EVs could transfer cargos, including functional proteins, mRNAs, and microRNAs, from donor to acceptor cells, thereby inducing a response in the recipient cells to biological signals contained in EVs from the donor cells. The critical role in cell–cell communication has been ascribed to EVs in many pathophysiological situations, including cancer, immune responses, cardiovascular diseases, regeneration, and stem cell-based therapy [36, 37]. Among them, cardiovascular diseases (CVDs) are the leading cause of death [38], with most due to atherosclerotic cardiovascular disease [39]. The contribution of EVs is observed during the initiation and development of atherosclerosis. In atherosclerotic plaque, EVs can originate from different donor cells, including leukocytes, erythrocytes, smooth muscle cells, and endothelial cells [40]. In the last decades, dysfunctional endothelial cells have been recognized as the early marker for atherosclerosis, which can be detected before apparent structural changes to the vessel wall, such as plaque formation or vascular remodeling, are diagnosed by angiography or ultrasound [41]. On the other hand, some literature has identified that endothelial cells are the major source of EVs in patients with coronary artery disease, and the increased packaging of miR-92a-3p in EVs from endothelial cells promotes migration and proliferation in the recipient cell [42]. It suggested that EVs derived from the dysfunctional endothelial cells were key mediator in promoting atherosclerosis.

The dysfunctional endothelial cells need to sort proper cargo into EVs to communicate the right message in a regulated and context-specific manner. In recent years, a variety of lipidomic, proteomic, and RNA sequencing studies have revealed that EV cargo composition differs depending upon the pathological microenvironment where the donor cell is located [43]. However, no detailed data are available on the mechanisms regulating endothelial EVs formation and their respective function in the atherosclerotic artery. Except for some biological risk factors, including inflammatory cytokines and high levels of non-HDL cholesterol, dysfunctional endothelial cells can also be induced by biomechanical stimuli. Due to the specific location of the artery, endothelial cells directly contact with blood flow, and they are acted upon by the fluid shear stress that varies with the magnitude and pattern generated by the flowing blood [14, 15, 44]. Interestingly, atherosclerosis does not arise uniformly on arterial surfaces throughout the vasculature. It tends to be localized to specific sites, such as branches and curvatures where blood flow deviates from normal straight vessel patterns [16]. In the atherosclerotic lesion, endothelial cells sense the disturbed blood flow, not the laminar blood flow in straight vessels. Work in the last decades has identified that disturbed flow activates various endothelial signaling pathways and downstream events to exacerbate the dysfunction of endothelial cells [17]. However, the effect and mechanobiological mechanism of disturbed flow regulating the secretion of EVs derived from ECs remains unknown. Furthermore, the proatherogenic or antiatherogenic function of ECs-derived EVs upon disturbed flow is mainly unexplored.

Here, we established a disturbed flow loading system in vivo and in vitro. Notably, the number of EVs derived from endothelial cells subjected to the disturbed flow significantly increased. Then, we conducted transcriptome sequencing to investigate the biomechanical mechanisms responsible for regulating the secretion of endothelial EVs. By employing comprehensive molecular experiments and pharmaceutical inhibition, the MAPK signaling pathway was validated to be the mediator in the disturbed flow-based mechanical clue. In addition, it has been verified that the MAPK signaling pathway responds to steady flow (12 dyn/cm²) to control the transcriptional activities of c-Jun and Elk-1 in endothelial cells [45]. On the other hand, the MAPK signaling pathway could be activated by other stimuli, such as inflammatory cytokines and oxidative stress. The elevated TNF-α and IL-1β significantly increased the ratio of p-ERK/ERK, indicating the activation of the MAPK signaling pathway in endothelial cells exposed to high ox-LDL [46]. Together, the MAPK signaling pathway is implicated in various pathological behaviors of endothelial cells during atherosclerosis. Therefore, pharmaceutical blocking MAPK-ERK signaling has been proposed as an alternative therapy for atherosclerosis instead of current lowering lipid drugs. However, since the MAPK signaling pathway plays a crucial role in various cellular processes, its prolonged MAPK inhibition could potentially lead to adverse effects. It suggested that a FDA-approved MAPK inhibitor used for tumor therapy may cause cutaneous side effects [47]. Therefore, except for pharmaceutical inhibition, PROteolysis TArgeting Chimeras (PROTACs) have attracted significant attention for finding the more proper methods to overcome the side effects and drug resistance caused by small molecular drugs [48]. The Heightman group designed representative PROTACs of ERK1 and ERK2. After 16 h post-ERK-CLIPTAC was added to the cell, the degradation of ERK1 or ERK2 was observed [49]. Unfortunately, the therapeutic effect of this ERK-CLIPTAC was not investigated in atherosclerotic endothelial cells.

Furthermore, we examined the exact impact of ECs-derived EVs on the pathological responses from other vascular cells and those EVs promoted the accumulation of monocytes and facilitated their endothelium invasion (Fig. 5). The pro-inflammatory effect of ECs-derived EVs with the disturbed flow on macrophages was detected. We found that the disturbed flow-induced ECs-derived EVs accelerated the inflammatory polarization of macrophages (Fig. 6). Generally, the pathological response from VSMCs is implicated in advanced atherosclerosis. In advancing atherosclerosis, VSMCs undergo complex structural and functional changes, giving rise to a broad spectrum of phenotypes. Classically, VSMCs transited from a contractile phenotype, characterized by markers such as smooth muscle α-actin (α-SMA), smooth muscle protein 22-α (SM22α), calponin, and myosin heavy chain 11, to a synthetic phenotype. This synthetic phenotype is indicated by markers like matrix metalloproteinase-2 (MMP2), which enhance the proliferative and migratory capabilities of VSMCs and promote the intimal invasion of VSMCs, finally accelerating the formation of plaque [7, 8]. Our previous work has demonstrated that EVs derived from VSMCs could be released into the extracellular space and provide nucleation sites for vascular calcification [34], characterized by osteogenic differentiation of VSMCs, another type of phenotypic switching. On the contrary, in this study, we found that the phenotypic switching of VSMCs did not occur even with the EVs derived from ECs upon disturbed flow stimulation (Fig. S6). We postulate that the underlying mechanism may be that the EVs derived from endothelial cells cannot penetrate the vascular media layer, being obstructed by the internal elastic membrane. Based on the literature and our research data, we proposed that endothelial EVs under disturbed flow are primarily involved in the early stage rather than the late stage of atherosclerosis. However, we are not yet clear about the specific contents within the endothelial EVs inducing the invasion of monocytes and the polarization of macrophages. Our previous study showed that miR-34c-5p enriched in ECs under laminar flow could induce the conversion of M1 macrophages into M2 macrophages [50]. Additionally, some literature indicated that tumor-derived EVs containing miRNAs 29, 222, and 940, CD63, gp130, ICAM-1, and IL-6Rb alternate activation of macrophages [51]. On the contrary, the EVs derived from macrophages also can affect various functions of the donor cells. Macrophage-derived EVs enriched with lncRNA GAS5 enhance the apoptosis of ECs and promote VSMCs migration [52]. A variety of mechanosensitive molecules located on the membrane of endothelial cells could initiate the disturbed flow-driven mechanical clue. The matricellular protein CCN1 mediated the nuclear accumulation of transcription factor nuclear factor-κB to exhibit higher oxidative stress, expression of endothelin-1 and monocyte chemoattractant protein-1, and monocyte homing in endothelial cells subjected to disturbed flow [53]. Furthermore, Toll-like receptors 4 (TRL4), plexin D1 (PLXND1), and integrin α5 also have been involved in the disturbed flow-induced dysfunctional endothelial cells [17, 54, 55]. However, it remains unclear whether these mechanosensitive molecules mediate the MAPK signaling pathway activated by disturbed flow. Accordingly, our future research should focus on revealing the mechanobiological signaling propagated from the exterior of the endothelial cells to the interior under disturbed flow.

This study is limited by the lack of primary human monocytes and endothelial cells, due to the inherent difficulty in acquiring human blood and vascular samples. Consequently, it is challenging to obtain experimental results that accurately reflect real conditions in vivo. Although we have shown that endothelial EVs under disturbed flow promote monocyte infiltration and inflammatory polarization of macrophage (M2 to M1), the specific EV cargo mediating these effects remains unidentified. Identifying the precise components of EVs poses a significant challenge, as it is unclear whether proteins, nucleic acids, or lipids play a primary role. Additionally, it is uncertain whether these components act independently, synergistically, or antagonistically, necessitating further investigation.

Conclusions

Collectively, we confirmed that disturbed flow promotes the ECs-derived EVs both in vivo and in vitro. Through RNA sequencing and other comprehensive experiments, we identified the central role of MAPK signaling in the mechanosensory transduction machinery driving disturbed flow-induced atherosclerotic ECs-derived EVs. Notably, the EVs derived from ECs subjected to disturbed flow significantly promoted the accumulation of monocytes and facilitated their invasion of the endothelium. Additionally, we found that these EVs accelerated the inflammatory polarization of macrophages. Interestingly, phenotypic switching of VSMCs was not observed even with EVs derived from ECs exposed to disturbed flow. Our results provide valuable insights into potential therapeutic targets for inhibiting the secretion of ECs-derived EVs in the treatment of atherosclerosis, offering a novel avenue for intervention in this disease.

Methods

Cell culture

Cells used in this study included human umbilical vein endothelial cells (HUVECs, Jiangsu Blood Research Institute), human aortic smooth muscle cells (HASMCs, ScienCell), and Tohoku Hospital Pediatrics-1 (THP-1, Warner Bio-Logy). HUVECs and HASMCs were maintained in EV-free Dulbecco’s modified Eagle medium (DMEM, Gibco) with 10% fetal bovine serum (FBS, Gibco) and 1% penicillin/streptomycin (PS, Hyclone). THP-1 cell line was cultured with EVs-free PRIM-1640 (Gibco) supplemented with 10% FBS and 1% PS. All cells were cultured in humidified 5% CO₂ at 37 °C. The extracellular vesicles in FBS were depleted by overnight ultracentrifugation at 110,000 g, followed by filtration through a 0.22-μm filter unit (Millipore).

The macrophage used in this study is derived from THP-1 [56]. Briefly, THP-1 cells were expanded in suspension flasks at 2.5 × 10⁵/mL in RPMI 1640 medium supplemented with 10% FBS and 1% PS solution. Following expansion, THP-1 cells were transferred to ultralow attachment culture flasks and stimulated with phorbol 12-myristate-13-acetate (PMA, 100 ng/mL final concentration) for 48 h at 37 °C and 5% CO₂, to induce differentiation to M0 macrophages. Following PMA treatment, M0 macrophages were washed in phosphate-buffered saline (PBS). For differentiation into the M1 and M2 phenotypes, the M0 macrophages were incubated for an additional 48 h in media supplemented with LPS (100 ng/ml) for M1 activation or IL-4 (20 ng/ml) for M2 activation [57].

Patient samples

Samples of human atherosclerotic plaques were obtained from patients after heart transplant recipients, while normal arteries were obtained after heart transplant donors. The sample-collecting process strictly followed the approved guidelines and obtained the informed consent of all participants. The study received ethical approval from the Institutional Medical Ethics Committee, ensuring compliance with ethical standards. The collected tissues were directly stored in liquid nitrogen or 4% paraformaldehyde for further use.

Animals

The animal experiments protocol followed the Medical Ethics Committee of Sichuan University. C57BL/6 mice and apolipoprotein E knockout (ApoE^−/−) mice were used in this study. The mice were purchased from Chengdu Dashuo Biotechnology Co., LTD. All mice were kept under specific pathogen-free conditions at a constant temperature of 25 °C. The mice were free of food and water. ApoE^−/− mice were used to establish the atherosclerotic mice by feeding a high-fat western diet (5 g/mouse/day).

Partial carotid ligation

Anesthesia was induced with 3% isoflurane and maintained with 1% isoflurane throughout the surgical procedure. The mice were placed on a heating pad set to 37 °C until the surgery was completed and monitored during recovery. The in vivo disturbed flow model was set up by partial ligation of the left carotid artery [58, 59]. Briefly, the left carotid artery (LCA) and its branches, including the internal carotid artery (ICA), occipital artery (OA), and external carotid artery (ECA), were ligated, leaving the superior thyroid artery (STA) intact. The ligation model could induce disturbed flow (DF) in LCA, while the right carotid artery (RCA) still senses laminar flow (LF), which was set up as the control group. After 3 weeks of ligation, the mice were euthanized by CO₂ inhalation at a flow rate of 3.5 L/min for 4 min, followed by confirmation of death.

Oil red O staining

The aorta was perfused with 4% paraformaldehyde for 48 h. The aorta segment was transferred to the oil red O dye solution and incubated at room temperature for 15 min. After staining, wash with 60% isopropyl alcohol and distilled water. The quantitative analysis of atherosclerotic plaque areas was completed using ImageJ software. On the other hand, for lipid staining in macrophages, the cells were kept in the presence of PMA and oxidized low-density lipoprotein ox-LDL (80 μg/mL final concentration). After incubating with LF-EVs or DF-EVs for 24 h, the harvested cells were fixed with 4% paraformaldehyde at 37 °C for 30 min. Intracellular lipid droplets were observed by staining with oil red O.

In vitro flow loading experiment

The laminar flow loading system (parallel-plate flow chamber) was applied to HUVECs as described previously [59–61]. HUVECs were seeded on 25 × 75 mm glass slides with the seeding density 2 × 10⁴ cells/cm². When the cells arrived at 100% confluence, they were subjected to LF of 15 dyn/cm² for 8 h. The shear stress (τ) was determined according to the formula:

$$\tau =\frac{6\mu Q}{W{H}^2}$$

where μ represents the viscosity of the circulating buffer, W is the flow field width (95 mm), and H represents height (0.3 mm). Additionally, Q represents the flow rate (155 mL/min).

The disturbed flow was applied to HUVECs as described previously [62, 63]. HUVECs were seeded on a square glass dish with a 10 cm length and the seeding density is the same as that used in LF loading. Subsequently, the disturbed flow was achieved by placing the glass dish on a platform shaker set at a frequency of 0.5 Hz, ± 7° rotating angle for 8 h. Depending on these parameters, the disturbed shear stress used in this study was about 4.5 dyn/cm². Inhibitor PD98059 (APExBIO) was used to inhibit the MAPK pathway. The inhibitor (50 μM) was applied 48 h on the HUVECs before the different flow loading and retained throughout the fluid shear stress loading process [64].

Preparation of EVs and nanoparticle tracking analysis (NTA)

Whole blood was collected from the model mice and stored at room temperature for 30 min. To isolate serum, the blood was centrifuged at 3000 g for 15 min at 4 °C. The supernatant was collected and centrifuged again at 3000 g for 20 min at 4 °C. Subsequently, 0.2 mL of EVs purification solution (EX010) was added to 1 mL of serum according to the manufacture’s instruction, mixed gently and incubated overnight at 4 °C. The following day, the mixture was centrifuged at 1500 g for 30 min at 4 °C, and the supernatant was discarded. To remove any residual liquid, the pellet was centrifuged at 1500 g for 5 min at 4 °C. The resulting pellet contained EVs, which were subsequently used for further experiments.

For the isolation of EVs from aortic tissue, aortic samples were dissected to remove fat and connective tissue. The tissue was then washed with sterile PBS (without calcium and magnesium) and cut into 5 mm³ pieces. The pieces were incubated in DMEM medium containing collagenase I (1 mg/mL) at 37 °C for 4 h, with shaking every 30 min. The mixture was filtered with a 40-μm strainer followed by centrifugation at 500 g for 5 min, further centrifuged at 10,000 g for 10 min at 4 °C. Then, 0.2 mL of EVs purification solution (EX010) was added to every 1 mL of supernatant, mix well, and left overnight at 4 °C. The next day, the EVs in the tissues were obtained by centrifugation at 1500 g for 30 min at 4 °C, and the supernatant was discarded, another centrifugation at 1500 g for 5 min to completely remove the residual liquid. The resulting pellet contained the EVs from vascular tissue, which were prepared for subsequent experiments. The conditioned medium (EVs-free DMEM medium) was obtained after culturing HUVECs subjected to LF or DF. The EVs (LF-EVs or DF-EVs) were collected from the conditioned medium by ultracentrifugation and resuspended in PBS [65]. Then, the size and number of EVs were immediately examined by NTA using the ZetaView instrument (PMX120, Particle Metrix). Data analysis was performed using Zetaview Analytical Software.

Transmission electron microscopy (TEM)

For observation of the morphology of EVs, the resuspended EVs in PBS were dropped onto a copper mesh for 3 min of adsorption. Subsequently, the solution was removed using filter papers, and 2% aqueous uranyl acetate was applied to the copper grid for 3 min at room temperature to perform negative staining. After the sample was dried in darkness for 20 min, the EVs were visualized using TEM (HT7800, Hitachi High-Tec Co., Ltd.) at 80 kV.

EVs uptake assays

To assess the uptake of EVs by recipient cell, we labeled equal amounts of purified EVs (3 × 10¹⁰) with DiI (C1036, Beyotime) according to the manufacturer’s instructions. Then, the labeled EVs were added to THP-1, macrophages, and VSMCs (5 × 10⁴), respectively. After 30 min coculturing at 37 °C, the cells were centrifuged at 1500 rpm for 5 min and analyzed by flow cytometry (BD Biosciences), and the data was handled with FlowJo 10.0 software. For immunofluorescence, 4% paraformaldehyde was added to the various cells for 15 min. The nuclei were stained with DAPI (1:1000, Beyotime), and the cytoskeleton was stained with FITC-Phalloidin. Finally, PBS was added to the cells, and the cells were photographed by confocal microscope (LSM710, Zeiss).

Transwell assay

HUVECs (3 × 10⁴) were seeded in the apical chamber of transwell inserts (8-μm pore size and 24-mm diameter, Corning®) in 200 μL of DMEM and allowed to attach overnight to form a dense inner layer at 37 °C and 5% CO₂. THP-1 cells (3 × 10⁴) were added to the apical chamber, and 1 × 10¹⁰ LF-EVs or DF-EVs were added, respectively. HUVECs cultured in transwell inserts without THP-1 and EVs stimulation served as a blank control. In addition, HUVECs cultured with equal amounts of THP-1 cells but without stimulation of EVs were used to measure the intrinsic transmembrane capability of THP-1. After 24 h, the cells settle down in the bottom of plate well were collected by centrifugation and counted. Meanwhile, the transwell insert was cleaned with PBS and fixed with 4% formaldehyde for 30 min, cleaned with PBS again, and stained with 1% crystal violet (Beyotime) for 15 min. Cells located on the upper membrane were carefully scraped off. Three randomly selected fields on the underside membrane were photographed with a microscope, and the number of cells was counted using ImageJ software. The total number of THP-1 cells crossing the HUVECs monolayer was the THP-1 cells in the lower compartment, plus these cells on the lower membrane. The cell number of all the groups was normalized to the only HUVECs group.

Endothelium adhesion assay

Firstly, HUVECs were transfected with lentivirus-mCherry, and THP-1 was infected with Calcein AM according to the manufacturer’s instructions. Then, HUVECs were cultured on the slides and dish for different flow loading, and the THP-1 was added to the circulating buffer (EVs-free DMEM medium). After 8 h of flow loading, the HUVECs were washed 3 times with cold PBS. The THP-1 adhered and penetrated the endothelium layer was observed by Zeiss confocal microscope.

Western blot

Collected cells or EVs were lysed with RIPA buffer (Beyotime Biotechnology Co.) with 1% protease inhibitor (SAB), 1% phosphatase inhibitor (SAB), and 1% phenylmethylsulfonyl fluoride (SAB) for 30 min on ice. The proteins were extracted by centrifugation at 12,000 g for 10 min at 4 °C, and the concentration was examined by the Enhanced BCA Protein Assay Kit (Beyotime Biotechnology Co.). The proteins were isolated with 4–12% SDS-PAGE gels loaded in the exact amounts per well (30 μg) and transferred to a polyvinylidene difluoride (PVDF) membrane. After 2 h of blocking with 5% skim milk at room temperature, the membrane was incubated overnight with specific primary antibodies (diluted in 5% skim milk with TBST, the detailed information of primary antibodies was listed in Table S1) at 4 °C and treated with the corresponding HRP-conjugated secondary antibodies (1:3000) for 2 h at room temperature. The bands were finally visualized and analyzed in a Molecular Image® Chemi-DocTM XRS + system with Image LabTM Software using enhanced chemiluminescence (ECL, 4ABIO). β-actin served as the internal control.

Immunofluorescence staining

For the immunofluorescence of cells, 1 × 10⁵ cells of each group were seeded onto a coverslip placed in 24-well plates and subjecting to different treatments. The cells were then rinsed with cold PBS and fixed with 4% paraformaldehyde for 15 min. After blocking with 5% normal goat serum (Solarbio) for 30 min at room temperature, the cells were incubated with primary antibodies, which are listed in Table S1, overnight at 4 °C. The corresponding fluorochrome-labeled secondary antibodies (1:1000, diluted in 5% BSA) were applied to the cells for 1 h after washing with PBS three times. The nuclei were subsequently stained with DAPI (1:1000, Beyotime) for 10 min and rinsed with PBS. On the other hand, the paraffin sections of vessel tissues were deparaffinized in xylene and gradient ethanol, followed by immersion in boiled sodium citrate buffer for antigen retrieval. Subsequently, the samples were blocked, incubated with antibodies, and stained as described above. The images were captured using the Zeiss confocal microscope (LSM710, Zeiss). Full details regarding antibodies are presented in Table S1.

Flow cytometry

The M0, M1, and M2 macrophages were subjected to LF-EVs or DF-EVs for 24 h, respectively. After that, the cells were collected and resuspended with flow buffer (0.5% BSA in PBS). Then, the cells were incubated with APC/Fire™ 750 anti-human CD206 (MMR) Antibody (BioLgend), PE anti-human CD86 Antibody (BioLgend) or PE anti-human CD80 Recombinant Antibody (BioLgend) for 30 min at room temperature. The cells were washed with flow buffer three times and harvested for flow cytometric analysis (BD Biosciences). Results were analyzed using FlowJo 10.0 software.

RNA sequencing

Total RNA was extracted from the flow-loaded cells using Trizol (Invitrogen Life Technologies) according to the manufacturer’s instructions. The quality control and integrity of total RNA were assessed using the NanoDrop ND-1000 (Thermo Scientific). RNA sequencing was subsequently performed by Illumina NovaSeq 6000. The abundance of transcripts and high-quality sequences for analysis were estimated using Agilent 2100 Bioanalyzer. The FPKM value (≥ 0.5) of genes or transcripts was analyzed by the R package Ballgown.

Quantitative real-time PCR (qRT-PCR)

The total mRNA was extracted by the total RNA Isolation Kit (RE-03113; Forgene) according to the manufacturer’s instruction and quantified through NanoDrop. The first strand cDNA was synthesized by the Evo M-MLV RT Mix Kit with gDNA Clean for qPCR Ver.2 (No. AG11728, Accurate Biotechnology). The primer sequences are shown in Table S2. Subsequently, qRT-PCR amplification was performed using the SYBR Green Premix Pro Taq HS qPCR kit (No. AG11701; Accurate Biotechnology) and the real-time PCR detection system (Bio-Rad; CFX Connect). β-actin served the internal control to normalize expression values by the delta delta-CT method.

Statistical analysis

The results are presented as the mean ± sd. All data were analyzed with GraphPad Prism 9 software. Data with n ≥ 6 were tested for normality using the Anderson–Darling, D’Agostino-Pearson, Kolmogorov–Smirnov, or Shapiro–Wilk test. For data following a normal distribution and exhibiting homogeneous variance, differences between treatment groups were analyzed using an unpaired t-test for comparisons between two groups or one-way/two-way ANOVA for multiple groups. Post hoc analysis, specifically Tukey’s honestly significant difference test, was conducted to assess statistical significance among multiple groups. For data with n < 6, nonparametric tests were applied. Values of P < 0.05 were considered statistically significant.

Authors’ contributions

Conceptualization, XL and HY; Methodology, investigation and data curation, LD, ZH, FF, YZ and TZ; Writing-Original draft preparation, ZH and HY; Writing-Reviewing and Editing, GL and MM; Funding acquisition; XL, HY and YZ. All authors read and approved the final manuscript.

Funding

This study was supported by the National Natural Science Foundation of China (12372315, 32371369, 32301089) and the Natural Science Foundation of Tibet (XZ202301ZR0012G).

Data availability

Data is provided within the manuscript files.

Declarations

Ethics approval and consent to participate

This study was conducted in accordance with the guidelines set by the China Council on Animal Care and was approved by the Medical Ethics Committee of Sichuan University (No. K2021015).

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.