Atherosclerosis is a common disease of the cardiovascular system characterized by plaque formation in the artery wall (1). Although some hypothesis based on “inflammation” (2), “lipid” (3) and “immunology” (4) have been proposed to explain the development of atherosclerosis, the pathogenesis of this condition is still not fully understood. Recently, the concept of endothelial to mesenchymal transition (EndMT) has also been put forward to explain the pathophysiological process of atherosclerosis from the perspective of cell trans differentiation (5). EndMT refers to a process in which endothelial cells can be transformed into mesenchymal cells. In this process, endothelial cells acquire the characteristics of mesenchymal cells and feature with loss of cell–cell contact and cell polarity under the condition of biochemical and biomechanical stimulus (6) (Figure 1). The obvious changes involved in EndMT at the molecular level include decreased expression of endothelial cell markers [such as platelet endothelial cell adhesion molecule-1 (CD31), CD34, von Willebrand Factor (vWF), tyrosine kinase with immunoglobulin-like and EGF-like domains 1 (TIE1), TEK receptor tyrosine kinase (TIE2), and vascular endothelial cadherin (CDH5)] and the increased expression of mesenchymal cell markers [such as alpha-smooth muscle actin (α-SMA), ferroptosis suppressor protein 1 (FSP1), Calponin, and smooth muscle 22 alpha (SM22α) (7, 8)]. During EndMT, endothelial cells undergo morphological changes from a cuboidal to a spindle shape. EndMT is a requirement for the formation of endocardial cushions during heart development; cushions resemble the early endothelium (9). EndMT has not only been described in organogenesis, but it has also been implicated in many diseases including pulmonary arterial hypertension (PAH) (10), fibrosis (11), cancer (12), pathological neovascularization (13), and atherosclerosis (14).

In this Review, we focus on the relationship between EndMT and progression of atherosclerosis to help to communicate relevant knowledge for atherosclerosis prevention. We discuss the effects of EndMT and the mechanisms underlying this process in the context of atherosclerosis.

The endothelium plays an important role in withstanding risk factors against atherosclerosis. Various conditions including sustained inflammation, fluid shear stress, ox-LDL, and smoking can facilitate EndMT (15, 16). Sustained inflammation is a typical pathologic feature of atherosclerosis (17). Inflammatory cytokines including interleukin-1beta (IL-1β), tumor necrosis factor alpha (TNF-α), transforming growth factor-beta (TGF-β), and interferon-gamma (IFN-γ) induce endothelial dysfunction and the acquisition of mesenchymal properties, therefore contributing to atherosclerosis (18–20). Long-term exposure to inflammatory cytokines can induce EndMT by altering the morphology of endothelial cells and EndMT-associated markers (21). In addition to the alteration of expression of endothelial/mesenchymal markers, inflammatory cytokines can also activate the TGF-β pathway and the non-TGF-β pathways (19).

Disturbed blood flow is another pathological characteristic of atherosclerosis. Zhou et al. showed that pulsatile shear stress (PS) exerts an atheroprotective role through the maintenance of endothelial cell homeostasis. In contrast, oscillatory shear stress (OS) causes endothelial cell dysfunction (22). Prior literature has also shown that EndMT can be induced by shear stress (23, 24). In addition, data from next-generation sequencing suggested that PS and OS lead to opposite effects in regulating EndMT gene regulation (25). Notably, expression of the twist family basic helix-loop-helix transcription factor (TWIST) expression was observed under conditions of low shear stress in regions of adult arteries (26). (27) demonstrated that low shear stress can result in the depletion of teneleven-translocation 2 (TET2). Overexpression of TET2 alleviated EndMT in atherosclerosis. Other risk factors associated with atherosclerosis can also lead to EndMT. Oxidized low density lipoprotein (ox-LDL) upregulates EndMT transcriptional factor Snail in human aortic endothelial cells (HAECs), in a on ox-LDLreceptor-dependent manner (LOX-1) (28). Moreover, ox-LDL induces accumulation of reactive oxygen species (ROS) accumulation in cells and synergistically promotes radiation-induced EndMT (29, 30). The effect of ox-LDL on EndMT can be inhibited by vaccarin (31) and naringin (32). Smoking is another known essential risk factor of atherosclerosis. Nicotine treated ApoE^−/− mice and human aortic endothelial cells (HAECs) presented with mesenchymal phenotypes, and blocking the nicotine receptor alleviated nicotine-induced EndMT and lesions in mice (33).

EndMT is a hallmark of endothelial plasticity. A detailed Review by William C. Aird previously discussed the phenotypic heterogeneity of the endothelium. In the embryonic period, endothelial cell differentiation plays an important role in organ development (34). Endothelial cells differentiate to meet the needs of organ development. For example, multiple lineage hematopoietic progenitors are derived from hemogenic endothelial cells via process endothelial-to-hematopoietic transition (EHT) (35). EndMT is essential for heart development. Heart endothelial cells give rise to cardiac fibroblasts and smooth muscle cells, contributing to the formation of the endocardial cushion formation (36). In the adult, disease can occur when homeostasis of endothelial cells is disrupted and these cells remain in an EndMT state.

In atherosclerosis, endothelial cells are exposed to various biochemical and physical stimuli derived from the circulating blood. Low-density lipoprotein (LDL), cholesterol, and wall shear stress may induce EndMT and disrupt endothelial cell homeostasis, leading to endothelial dysfunction and thereby contributing to the development of atherosclerosis. In addition, it has been well documented that inflammatory factors such as IL-1β, TNF-α, TGF-β, and endotoxins can induce endothelial dysfunction via EndMT. In inflammation conditions, activation of endothelial cells occurs, leading to phenotypic and molecular changes (16, 37–39). Endothelial cells are heterogeneous and show a high degree of plasticity in both normal and atheromatous conditions. (40) defined eight endothelial cell clusters in the heart and aorta of patients with diabetic atherosclerosis at single cell level, three of which expressed mesenchymal markers, indicating EndMT markers. Further analysis suggested that the proportion of EndMT-derived fibroblast like cells was higher in atherosclerosis group compared to the normal group, with alterations in extracellular-matrix organization, adhesion, and apoptosis.

The presence of endothelial cell-derived mesenchymal like cells in plaques provides robust evidence of the involvement of EndMT in atherosclerosis. Previous studies demonstrated that atherosclerotic plaques contain mesenchymal cells (including fibroblasts and smooth muscle cells), which regulate inflammation, extra-cellular matrix and collagen production, and plaque structural integrity and play a key functional role in atherosclerosis (41, 42). To investigate the origins of atherosclerosis-associated fibroblasts, (43) examined and confirmed the presence of EndMT-derived fibroblast-like cells present in atherosclerotic lesions through lineage-tracking, suggesting a role of EndMT atherosclerosis development. Matrix metalloproteinases (MMPs) are associated with unstable atherosclerotic lesions (44, 45), and EndMT-derived fibroblast-like cells express higher levels of MMP1, MMP9, and MMP10 compared with normal fibroblasts. In addition, TGF-β signaling, oxidative stress, and hypoxia facilitate endothelial cells conversion to mesenchymal cells and are all hallmarks of atherosclerosis. During EndMT process, active endothelial cells express adhesion molecules, such as intercellular adhesion molecule-1 (ICAM-1) and vascular cell adhesion molecule-1 (VCAM-1), which enhance monocyte, leukocyte, and macrophage recruitment and infiltration (46). Interestingly, (47) found evidence for a crosstalk between macrophages and EndMT: macrophages in atherosclerotic lesions in vivo upregulate the expression of mesothelial markers, promoting EndMT, and, conversely, EndMT cells impact the function, such as the capacity of lipid uptake, and phenotypes of macrophages. In addition to in vivo and in vitro studies, single cell sequencing technology has been helpful to further our understanding of the landscape and pathophysiology of human atherosclerotic plaques. Recently, a study identified 14 cell populations including endothelial cells, smooth muscle cells, mast cells, B cells, myeloid cells, and T cells and identified multiple cellular activation states in plaques (48). One of the identified subclasses of endothelial cells expressed the smooth muscle cell markers such as alpha-actin 2 (ACTA2), notch receptor 3 (NOTCH3), and myosin heavy chain 11 (MYH11), suggesting that this subtype was undergoing EndMT and providing additional evidence on plasticity of endothelial cell plasticity. Altogether, these findings confirm that EndMT is closely associated with plaque initiation and development.

A substantial body of research pinpoints to several features of atherosclerosis that can lead to EndMT via several signaling pathways (49, 50). The TGF-β signaling pathway is a canonical pathway modulating EndMT, which has been shown to crosstalk with other pathways including the fibroblast growth factor (FGF), Notch, and bone morphogenetic protein (BMP) pathways (Figure 2). In addition, non-coding RNAs also paly an essential role in EndMT. We next discuss the mechanisms underlying EndMT in the context of atherosclerosis.

The role of TGF-β signaling pathway on EndMT has been extensively studied. In brief, the TGF-β family ligands bind to type I receptors and type II receptors, phosphorylating and thereby activating the transducer small mother against decapentaplegic (SMAD). Nuclear import of active SMAD transmits regulates gene transcription (51). In addition, TGF-β receptors can also activate other signaling pathways. Specifically, TGF-β signaling can be divided into SMAD-dependent and non-SMAD-dependent pathway (52). Growing evidence demonstrated the role of TGF-β signaling pathway in regulating cell proliferation, differentiation, adhesion, migration, and apoptosis not only in both embryonic development and the pathology of human disease (53). (14) showed that risk factors of atherosclerosis such as oscillatory shear stress and inflammation-induced loss of fibroblast growth factor receptor 1 (FGFR1) expression can activate TGF-β signaling and contribute to EndMT. FGFR1 depletion induces EndMT by upregulation of smooth muscle markers and mesenchymal markers which are also targets of TGF-β (54). Although the crucial role of TGF-β in EndMT has been well established, the different roles of the various isoforms of TGF-β are less known. To this aim, Sabbineni et al. (55) compared the effect of three isoforms of TGF-β (TGF-β1–3) on EndMT and found that TGFβ2 was the one most associated with EndMT. In addition, TGFβ2 is required for epithelial-to-mesenchymal cell transformation during endocardial cushion (56). Experiments treating human microvascular endothelial cells (HMECs) with different TGFβ isoforms indicated that only TGF-β2 substantially increased Smad2/3, p38 mitogen-activated protein kinase (MAPK), and mesenchymal transcription factors Snail and forkhead box protein C2 (Foxc2). Furthermore, TGF-β2 and IL-1β co-stimulate EndMT to activate nuclear factor-kappaB (NF-κB) (19).

Given the critical role of TGF-β in regulating EndMT, TGF-β inhibition may be an approach to reverse EndMT. Knockout of endothelium-specific TGF-β receptor relieved EndMT and kidney fibrosis (57). However, TGF-β exhibits potent regulatory functions, and inhibition of this pathway might also result in unwanted side effects.

Altogether, this evidence suggests that TGF-β may also be a promising therapeutic target for atherosclerosis, and further studies to confirm this hypothesis and assess safety are needed.

The studies by Chen et al. (14) and Evrard et al. (43) provide strong evidence for the involvement of EndMT in atherosclerosis. EndMT drives the initiation of atherosclerosis by accelerating plaque growth and instability of plaque. Previous research has shown that EndMT is regulated in a number of signaling pathways, in particular the canonical TGF-β signaling, and several atherosclerosis risk factors, e.g., inflammation, shear stress, ox-LDL, and smoking. Conversely, FGF signaling plays a protective role on EndMT. Remarkably, several non-coding RNAs were reported to modulating EndMT, offering therapeutic application to treat atherosclerosis (74). Epigenetic mechanisms involved in EndMT regulation are poorly understood. Histone deacetylases (HDACs) play a role in EndMT, in particular HDAC3 and HDAC9, and promoted atherosclerosis progression (98). Overall, the molecular mechanisms underlying EndMT remain largely unknown, and additional research is needed to discover new targets that can be explored in reverse atherosclerosis.

EndMT is associated with the formation of atherosclerotic lesions. Therefore, reversing or inhibiting EndMT might help to prevent the development or progression of atherosclerosis. In vivo, suppression of EndMT shows promising effects in alleviating atherosclerosis. However, animal and cell models of atherosclerosis present many limitations, and the study and detection of EndMT in humans offer great challenges. More research is needed to understand the role of EndMT in atherosclerosis to ultimately offer new insights for the treatment of atherosclerosis.

QH, HW, and ZZ contributed to the conception of the study, performed the literature search, and wrote the manuscript. QH, YG, and ZZ edited the manuscript. ZY and YG assisted in the literature search and critically revised the article for important intellectual content. All authors have read and approved the final manuscript.

This study was supported by the Guangdong Provincial Key Laboratory of Precision Medicine and Clinical Translation Research of Hakka Population under Grant 2018B030322003; the Science and Technology Program of Meizhou under Grant 2019B0202001.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.