In the setting of undisturbed laminar flow, nitric oxide (NO) is produced by elevated levels of eNOS in ECs and diffuses across cell membranes to modulate the activity of vascular cells, including SMCs and leukocytes (Pan, 2009). NO-mediated activation of soluble guanylate cyclase and S-nitrosylation has been observed to inhibit SMC proliferation (Ignarro et al., 2001), suppress EC inflammation (Matsushita et al., 2003; Kang-Decker et al., 2007), and regulate vascular tone, blood flow, and oxygen delivery to tissues (Stamler et al., 1997; Haldar and Stamler, 2013). Hydrogen sulfide, a gaseous signaling molecule and upstream regulator of NO, is similarly produced under laminar flow conditions (Huang et al., 2015) and has a critical role in reducing vascular inflammation (Zanardo et al., 2006), promoting antioxidative activity (Muzaffar et al., 2008), and limiting the formation of foam cells (Zhao et al., 2011). Laminar flow conditions have also been found to promote EC alignment (Hahn and Schwartz, 2009), and foster a reducing environment that limits oxidative stress (Berk, 2008), tumor necrosis factor α- and signal transducer and activator of transcription 3-induced inflammation (Ni et al., 2003; Yamawaki et al., 2003), and EC turnover via apoptosis (Pi et al., 2004). The transcription factor Kruppel-like factor 2 (KLF2) has emerged as a critical transcriptional mediator of vascular homeostasis that is induced under laminar shear stress and increases eNOS expression and represses adhesion proteins such as vascular cell adhesion molecule 1 (VCAM-1) and E-selectin (SenBanerjee et al., 2004; Berk, 2008), which are induced by NF-κB in regions of disturbed flow (Hajra et al., 2000; Collins and Cybulsky, 2001). In contrast to regions of laminar flow, altered hemodynamics in atheroprone regions has been associated with increased oxidative stress (Förstermann et al., 2017), reorganization of cytoskeletal and cell-cell junction proteins (Chong et al., 2013; Pfenniger et al., 2013), and enhanced senescence and turnover in ECs (Xu, 2009; Warboys et al., 2014). Curiously, hydrogen sulfide has been observed to impair dilation of coronary arteries by reducing NO production under disturbed flow (Chai et al., 2015), although hydrogen sulfide administration has also been shown to inhibit leukocyte adhesion to the endothelium in regions of shear stress by upregulating Akt/eNOS signaling (Go et al., 2012).

Recent studies continue to elucidate a plethora of EC regulatory pathways that induce a proinflammatory and proatherogenic phenotype in response to disturbed flow (Qu et al., 2020; Zhao et al., 2020; Xia et al., 2021) (Figure 1). The mechanosensitive cation channel Piezo1, though essential for coordinating vascular morphogenesis under physiological shear stress in embryogenesis and adulthood (Li J. et al., 2014), has been previously linked to altered flow-induced inflammation and leukocyte recruitment to the endothelium. Indeed, pathological activation of Piezo1 has been associated with induction of downstream integrin α5 and NF-κB pathways, which contributes to endothelial inflammation and the progression of murine atherosclerosis (Albarrán-Juárez et al., 2018), and elevation of intracellular calcium, which results in actin disruption and increased monocyte adhesion in cultured ECs (Swain and Liddle, 2020). Likewise, integrin α5β1-associated phosphorylation of yes-associated protein (YAP) by c-Abl kinase (Li B. et al., 2019), the coupling of integrin α5 to annexin A2 (Zhang C. et al., 2020), and the guidance receptor plexin D1, upstream of EC integrins (Mehta et al., 2020), have been identified as novel mechanisms for endothelial activation under oscillatory shear stress. In contrast, several EC processes confer in vivo protection against atherosclerosis under disturbed flow, including inhibition of integrin α5 by cartilage oligomeric matrix protein (Lv et al., 2021), increased glycolysis by protein kinase AMP-activated catalytic subunit alpha 1 (Yang et al., 2018), decreased lipid uptake by the transcription factor inhibitor of DNA binding 1 (Zhang K. et al., 2018), and increased EC alignment by Rho GTPase activating protein 18 (Lay et al., 2019). Many of these preclinical findings hold promise for the development of atherosclerosis treatments through pharmacological modulation of the endothelium (e.g., mitigating atherogenic responses or promoting atheroprotective functions in ECs). However, further investigation is required in vitro to determine the generalizability of endothelial responses between vascular cell models (e.g., shared and distinct behaviors of different EC lines under varying flow conditions) (Maurya et al., 2021), as well as in vivo to identify translatability in preclinical models and long-term safety and efficacy in individuals with CVD.

Emerging evidence has also implicated disturbed blood flow in the dysregulation of EC proliferation and turnover during early atherosclerosis. Under physiological conditions, ECs that become senescent or apoptotic are replaced by the replication of neighboring cells, and more significant disruptions to the endothelium (e.g., during injury) are mitigated by circulating endothelial progenitor cells (Mannarino and Pirro, 2008). Atherosclerosis is characterized by cellular processes that promote EC turnover and induce proliferation of the activated endothelium (Xu, 2007). However, whether this proliferation is protective or detrimental has not been fully resolved. The mechanosensitive transcription factor zinc finger and BTB domain containing 46, for instance, inhibits proliferation in quiescent ECs and is downregulated by disturbed flow in vitro (Wang et al., 2019a). Hemodynamic signaling also impacts epigenetic pathways, and we have reviewed the impact of epigenetics on atherosclerosis elsewhere (Khyzha et al., 2017). For example, altered flow has been shown to induce DNA methyltransferase 1-mediated hypermethylation of the endothelium and its subsequent inhibition limits atherosclerosis via the cell cycle regulator cyclin A (Zhang et al., 2017). Furthermore, laminar flow-induced autophagy and expression of the deacetylase sirtuin 1 (SIRT1) together inhibit Hippo/YAP signaling to attenuate atherosclerotic plaque formation (Yuan P. et al., 2020), and hypermethylation of eNOS promoter elements under chronic disturbed flow has been shown to contribute to repressed eNOS expression in wildtype mice (Ku et al., 2021).

Flow-sensitive microRNAs have also been identified as key modulators of endothelial proliferation and turnover in atherosclerosis (Kumar et al., 2019), including miR-126-5p, which has been shown to contribute to proliferative reserve in ECs and prevents plaque development through upregulation of Notch signaling (Schober et al., 2014). Curiously, increased expression of miR-126 has been found in atheroprone regions of the endothelium (Zhou et al., 2013). Furthermore, physiological flow conditions have also been observed to promote antiproliferative microRNA activity within the endothelium, whereby KLF2-induced miR-23b represses cyclin H to reduce activity of the cyclin-dependent kinase–activating kinase complex and limit cell cycle progression (Wang et al., 2014). The regulatory interplay of antagonistic, flow-sensitive microRNAs may serve as a mechanism for fine-tuning EC proliferation and warrants further characterization in both quiescent and activated endothelial states.

A robust characterization of the transcriptional and epigenetic regulatory elements that contribute to endothelial dysfunction has remained an important challenge for atherosclerosis research and has been recently addressed with next-generation approaches for global EC profiling. Of note, the integration of chromatin immunoprecipitation, chromatin accessibility, and RNA sequencing has enabled identification of diverse DNA regulatory elements associated with disturbed flow and proinflammatory activation (e.g., NF-κB and hypoxia inducible factor 1α, as well as ETS, zinc finger, and activator protein 1 transcription factor families) (Hogan et al., 2017; Bondareva et al., 2019; Alizada et al., 2021). Moreover, combined single-cell RNA sequencing and genome-wide chromatin accessibility assays have been used to profile the genome- and epigenome-wide changes associated with proatherogenic ECs under oscillatory shear stress (Andueza et al., 2020). Likewise, these technologies have allowed for characterization of EC heterogeneity in human atherosclerotic plaque (Depuydt et al., 2020) and identification of coronary artery disease-associated genetic variants in the open chromatin regions of activated ECs (Örd et al., 2021). The future integration of global EC profiling with functional assays will aid in the validation of putative modulators of atherogenesis, which may serve as prospective therapeutic targets for CVD treatment. Notably, real-time monitoring of the vascular endothelium has been enabled with organ-on-a-chip technologies (Sei et al., 2017), allowing for specific and quantifiable testing to elucidate the impact of environmental stimuli on endothelial function. Nevertheless, the complex cellular milieu of the atherosclerotic plaque remains difficult to fully model.

Future study is warranted to address several outstanding research areas in EndMT and atherosclerosis, including the characterization of EndMT as a discrete versus continuous process, as well as the potential beneficial versus detrimental role of EndMT in plaque vulnerability. First, the proatherogenic endothelium has recently been shown to exist in a “metastable or partial” state of EndMT, where perturbed ECs perform both endothelial and mesenchymal functions (Helmke et al., 2019; Fledderus et al., 2021). The metastable nature of EndMT may be further elucidated with next-generation technologies that profile the atherosclerotic plaque and neighboring regions at the single-cell level, as recently used in studies of disturbed flow and endothelial reprogramming (Andueza et al., 2020) and diabetic atherogenesis (Zhao et al., 2021), as well as through the incorporation of computational models that predict endothelial activation and EndMT in various genetic and environmental conditions (Weinstein et al., 2020). Second, although augmentation of EndMT has previously been associated with increased plaque progression (Chen et al., 2015), emerging studies have highlighted a potential protective role of EndMT by maintaining plaque stability in the absence of SMC-derived myofibroblast-like cells (Evrard et al., 2016; Newman et al., 2021). Evidently, a more nuanced understanding of EndMT in the context of plaque development and rupture is required before therapeutic inhibition of EndMT is considered for the treatment of atherosclerotic CVD.

In recent years, vascular cell-cell communication mediated by the secretion of extracellular vesicles (i.e., EVs; nano-sized packages of proteins, mRNAs, noncoding RNAs, and lipids) has emerged as an important research area for the pathogenesis and treatment of atherosclerosis (Charla et al., 2020; Chen Y.-T. et al., 2020; Wang H. et al., 2020). Three broad categories of EVs – exosomes (30–150 nm in diameter), microparticles (100 nm–1 μm in diameter), and apoptotic bodies (1–5 μm in diameter) – have been shown to modulate vascular function in atherogenesis via intercellular signaling among ECs, as well as between ECs and other immune and non-immune cell types (He et al., 2018; Li M. et al., 2018; Charla et al., 2020; Peng et al., 2020) (Figure 1). Within the vascular endothelium, EC-derived EVs with distinct microRNA content have been observed to both induce (Jansen et al., 2013; Arderiu et al., 2015) and inhibit angiogenesis (Ou et al., 2011; Liang et al., 2017, 2) via paracrine regulation of nearby ECs, whereby EVs fuse to recipient cells, and the release of EV-derived microRNAs allows for silencing of complementary mRNA targets. Moreover, the uptake of EC-secreted microparticles by neighboring ECs was shown to promote ICAM-1 expression and monocyte adhesion via miR-222 in diabetic conditions (Jansen et al., 2015), and EV-derived apoptotic bodies containing miR-126 were found to reduce macrophage content in plaque and limit atherosclerotic burden (Zernecke et al., 2009). The specific role of the proatherogenic environment in governing EV loading and function remains of interest for future study, and characterization of this environment has begun with exposure of donor ECs to proatherogenic stimuli (i.e., oxLDL and IL-6), which were observed to decrease thrombospondin 1 and increase angiogenesis in recipient ECs via EV-derived miR-92a-3p (Liu Y. et al., 2019).

Beyond the endothelium, EC-EVs regulate both immune and non-immune vascular cell populations involved in the development of atherosclerosis (Li M. et al., 2018; He et al., 2018) and circulating EVs may have systemic effects at distant sites (Bär et al., 2019; Liu et al., 2021). Exosomal microRNAs mediate crosstalk between ECs and macrophages by suppressing (e.g., miR-10a) (Njock et al., 2015) or promoting monocyte activation (e.g., miR-155) (He et al., 2018), reducing macrophage infiltration, and delaying plaque progression (e.g., miR-126, miR-210, and miR-216) (Wang et al., 2019b). In turn, monocyte- and macrophage-derived EVs have been shown to induce EC apoptosis and promote murine atherosclerosis through elevated blood lipids, oxidative stress, and inflammation (Aharon et al., 2008; Li K. et al., 2021), as well as by reducing proliferation and angiogenesis in human coronary artery ECs via miR-503-5p (Wang et al., 2021). Reciprocal communication between cells of the vasculature has been similarly observed in ECs and SMCs, where EC-secreted EVs induce both atheroprone (Boyer et al., 2020; Yuan X. et al., 2020; Zhang Z. et al., 2020) and atheroprotective SMC phenotypes (Hergenreider et al., 2012; Xiang et al., 2021), while SMC-derived EVs govern endothelial migration under stimulation by platelet-derived growth factor (Heo et al., 2020) and can increase endothelial permeability and potentiate atherosclerosis via miR-155 (Zheng et al., 2017).

EC-EVs that are secreted into systemic circulation may serve as biomarkers for CVD and have the potential to govern cell behaviour at distant areas of the vasculature, although definitive evidence for this regulation is lacking (Bär et al., 2019; Liu et al., 2021). Preliminary animal studies have demonstrated that microRNAs secreted by circulating blood cells, and presumably contained within EVs, can modulate SMC activity in atherogenesis (Shan et al., 2015) and have proapoptotic and antiproliferative effects on ECs (Chu et al., 2017). Further investigation is warranted to determine whether circulating EVs are present in sufficient quantities to elicit systemic effects in patients and if their regulatory functions extend to other CVDs (e.g., cardiac fibrosis and ischemic heart disease), as well as to identify the precise mechanisms by which circulating EVs promote vascular dysfunction in recipient cells. Notably, future study is also required to determine whether EVs undergo transcytosis or pass through intercellular or intracellular gaps to the subendothelial space and elicit direct effects on medial cells, or are endocytosed by ECs, which then serve as indirect mediators of altered vascular function. The continued integration of EV profiling in patients with atherosclerosis (i.e., local EVs in plaque; circulating EVs in plasma) with experimental models of EV activity will shed light on the causal role of EVs and EV-contents in the setting of atherosclerotic CVD. Such models will require the transition from in vitro co-culture towards in vivo tracking approaches to allow for a robust, physiological characterization of EV activity, which will be made possible with advances in the visualization of EV release and uptake (Durak-Kozica et al., 2018; Oesterreicher et al., 2020). With respect to the diagnostic potential of vascular EVs, the application of machine learning strategies to predict CVD from circulating EV biomarkers has garnered recent interest (Burrello et al., 2020; Castellani et al., 2020) and will likely play an important role in the development of precision cardiovascular medicine over the coming decades.

The NLRP3 inflammasome is now recognized as a key player in the coordination of vascular inflammation and onset of atherogenesis (Duewell et al., 2010; Jin and Fu, 2019). NLRP3-mediated induction of atherosclerosis has been associated with a host of causal factors including hypoxia (Folco et al., 2014), cholesterol crystals and oxLDL (Duewell et al., 2010), and disturbed flow (Xiao et al., 2013). Activation of NLRP3 subsequently leads to the maturation of proinflammatory cytokines (e.g., IL-1β and IL-18), increases the migration and lipid loading of macrophages (Li X. et al., 2014), and promotes pyroptosis, a proinflammatory form of programmed cell death that contributes to the release of additional inflammatory mediators (Bergsbaken et al., 2009). Therapeutic agents that directly inhibit the NLRP3 inflammasome, including the natural compound arglabin (Abderrazak et al., 2015), colchicine (Fernando et al., 2017), and the small molecule inhibitor MCC950 (van der Heijden et al., 2017), have been shown to ameliorate endothelial inflammation and atherosclerosis in preclinical studies. Moreover, adjunct colchicine therapy has entered clinical trials for repressing vascular inflammation in acute coronary syndrome (Bouabdallaoui et al., 2020) and coronary artery disease (Nidorf et al., 2013; Kajikawa et al., 2019; Hays et al., 2021). Previous trials have employed several outcomes related to improved cardiovascular health, including CVD endpoints, serum inflammatory markers, and proxy measures for endothelial dysfunction including flow-mediated vasodilation and exercise-induced coronary blood flow. Although the latter two of these measures remain the gold standard for clinical assessment of the perturbed endothelium (Flammer et al., 2012; Xu et al., 2021), they rely on endothelial responsiveness to vasodilatory or vasoconstrictive manipulation and do not adequately capture changes in EC phenotype during atherogenesis (e.g., permeability, metabolism, EndMT, or intercellular communication). Emerging studies have begun to address this limitation by applying machine learning techniques to predict CVD events from clinical data, CT imaging, and circulating biomarkers, including matrix metalloproteinase 9 and polymeric immunoglobulin receptor (Tamarappoo et al., 2021), very low-density lipoprotein and leucine (Coelewij et al., 2021), and serum EVs (Burrello et al., 2020; Castellani et al., 2020). Alongside functional assessment of the endothelium, robust models for CVD prediction have the potential to identify subclinical disease (Coelewij et al., 2021), aid in risk stratification, and inform treatment and prevention strategies.

Recent studies have also highlighted the potential of endogenous specialized proresolving mediators (SPMs) as anti-inflammatory therapies for limiting endothelial inflammation in CVD (Fredman and Tabas, 2017; Xu et al., 2021). Resolvin D1 (RvD1) belongs to a family of SPMs formed from metabolic processing of polyunsaturated docosahexaenoic acid and eicosapentaenoic acid, and is synthesized by ECs under proinflammatory conditions including stimulation with oxLDL (Dufour et al., 2018). Treatment of ECs with RvD1 has been observed to repress lipopolysaccharide- and cholesterol crystal-stimulated EC-monocyte interactions in vitro by downregulating NFκB-ICAM-1/VCAM-1 signaling (Chattopadhyay et al., 2018; Pichavaram et al., 2019), and attenuates leukocyte trafficking to the endothelium in mice fed a cholesterol rich diet (Pichavaram et al., 2019). Beyond the resolvin family, the SPMs maresin 1 and lipoxin A4 have likewise been observed to disrupt leukocyte adhesion to activated ECs (Filep et al., 1999; Chatterjee et al., 2014), suggesting that endogenous production or exogenous administration of diverse SPM metabolites may have therapeutic benefits for limiting endothelial activation in the setting of atherosclerosis.

Notably, the potential therapeutic impact of promoting inflammation resolution through efferocytosis (i.e., the clearing of apoptotic cells) is being established with preclinical studies, wherein atherosclerotic plaque progression has been attenuated by proefferocytic cluster of differentiation 47 (CD47)-blocking therapies (Kojima et al., 2016; Flores et al., 2020) or activation of the macrophage cell surface receptor MER proto-oncogene, tyrosine kinase (Thorp et al., 2008; Cai et al., 2018). While these effects appear to be mediated by lesional macrophages, ECs could facilitate the transport of inflammation resolving agents to cells within the developing plaque. Indeed, a proefferocytic and anti-atherosclerotic nanoparticle therapy has been previously hypothesized to function by uptake within circulating monocytes or delivery to macrophages within the intima (Flores et al., 2020), the latter of which would require the monocyte-independent transport of nanoparticles across the endothelium. In this role, ECs may contribute to therapeutic delivery by modulating vascular permeability and trafficking nanoparticles to macrophages via EC-secreted EVs. Clinically, the promise of inflammation resolving strategies has been highlighted in a subset of patients enrolled in a phase 1b-2 trial of the proefferocytic anti-CD47 antibody magrolimab for treating B-cell non-Hodgkin’s lymphoma, where vascular inflammation was significantly reduced in the carotid arteries of patients with CVD after 9 weeks of treatment (Jarr et al., 2021).

Epigenetic interventions allow for broad targeting of the diverse mechanisms underlying endothelial dysfunction and have shown recent promise as anti-atherosclerotic therapies (Fledderus et al., 2021). The methyltransferase enhancer of zeste homologue 2 (EZH2) and histone deacetylase SIRT1 respectively induce and limit endothelial dysfunction in murine models of CVD (Zhou et al., 2011; Lv et al., 2016) and represent novel avenues for epigenetic treatment. EZH2 antagonism and SIRT1 agonism similarly increase NO production (Gracia-Sancho et al., 2010; Kumar et al., 2013) and decrease EC activation (Zhou et al., 2011; Wang J. et al., 2020) and EndMT (Maleszewska et al., 2016; Liu Z.-H. et al., 2019) in experimental models of CVD (Figure 2). More recently, natural antioxidants including epigallocatechin and naringenin have been shown to reduce oxidative damage in ECs via SIRT1 (Li H. et al., 2021; Pai et al., 2021), and an intermittent fasting regimen was found to ameliorate vascular dysfunction in a murine model of diabetes by activating the SIRT1 pathway (Hammer et al., 2021). Small molecule agents that target EZH2 and SIRT1 are currently available or under clinical study in several disease areas (e.g., oncology, dermatology, and nephrology) (Ganesan et al., 2019; Fledderus et al., 2021), and may be investigated as potential anti-atherosclerotic agents by incorporating CVD endpoints into ongoing clinical trials. Beyond their desired therapeutic potential, the unintended effects of genome-wide modification via epigenetic agents also warrant consideration. Future studies may improve upon the specificity of these agents by making use of targeted CRISPR/Cas9 epigenome-editing techniques for epigenetic modifications associated with CVD risk loci (Cano-Rodriguez et al., 2016; Cano-Rodriguez and Rots, 2016; Xu et al., 2018). Alternatively, targeted EC therapies, as discussed below, may be considered for the resolution of vascular inflammation at the site of the activated endothelium in atherosclerosis. Prospective agents for targeted resolution include ICAM-1- or VCAM-1-targeted nanocarriers, which have previously been employed to limit neurovascular (Lutton et al., 2017; Marcos-Contreras et al., 2020) and pulmonary inflammation (Li S. et al., 2018; Park et al., 2021) in experimental models of disease.

Advances in the development of nanoparticles (on a scale of <0.1 μm) have likewise afforded new opportunities for anti-atherosclerotic therapies that resolve endothelial dysfunction (Flores et al., 2019; Marchio et al., 2019; Chen L. et al., 2020). Vascular nanotherapies have been broadly classified as those which resolve inflammation and dysfunction in efferocytosis, limit plaque neovascularization and neointimal growth, modulate lipid metabolism, and decrease thrombosis (Flores et al., 2019). Several of these therapies have been shown to target the perturbed endothelium in experimental models of CVD, including the encapsulation of five cell adhesion molecule small interfering RNA, which reduced leukocyte migration to plaque and suppressed post-myocardial infarction inflammation in murine models (Sager et al., 2016). Integrin-targeted nanoparticles containing the antiangiogenic compound fumagillin have likewise been shown to limit neovascularization in mice and demonstrate prolonged activity when combined with statin treatment (Winter et al., 2006; Winter et al., 2008) (Figure 2). More recently, engineered endothelial adrenoreceptor- (Ul Ain et al., 2017), VCAM-1- (Distasio et al., 2021), and P-selectin-targeted nanoparticles (Mocanu et al., 2021), which respectively carried the genes eNOS and IL-10 and receptor for advanced glycation end products-silencing RNA, have been found to localize to ECs and resolve inflammation within the murine atherosclerotic plaque. Nano-selenium particles have also been observed to improve endothelial dysfunction in murine atherosclerosis via Na⁺/H⁺ exchanger 1 inhibition (Zhu et al., 2019). Despite the therapeutic advantages offered by nanoparticles (e.g., reduced off-target effects and increased efficacy per dose), several methodological and biological challenges remain for clinical translation (Flores et al., 2019). Importantly, robust protocols are required to determine the composition and purity of nanoparticles derived using various formulations (e.g., lipid or polymeric). Furthermore, the long-term stability of nanoparticles, as well as their safety and selectivity for the activated endothelium in patients with atherosclerosis, must be assessed before these agents are clinically adopted. Nevertheless, vascular nanotherapies remain a promising avenue for the targeted resolution of endothelial dysfunction and reduction of plaque progression in atherosclerotic CVD, with encouraging evidence being generated for the systemic administration of nanoparticle-enveloped small interfering RNA in nonhuman primates to inhibit EC gene expression in multiple organs (Khan et al., 2018).