Endothelial mechanosensors: the gatekeepers of vascular homeostasis and adaptation under mechanical stress

Endothelial cells (ECs) not only serve as a barrier between blood and extravascular space to modulate the exchange of fluid, macromolecules and cells, but also play a critical role in regulation of vascular homeostasis and adaptation under mechanical stimulus via intrinsic mechanotransduction. Recently, with the dissection of microdomains responsible for cellular responsiveness to mechanical stimulus, a lot of mechanosensing molecules (mechanosensors) and pathways have been identified in ECs. In addition, there is growing evidence that endothelial mechanosensors not only serve as key vascular gatekeepers, but also contribute to the pathogenesis of various vascular disorders. This review focuses on recent findings in endothelial mechanosensors in subcellular microdomains and their roles in regulation of physiological and pathological functions under mechanical stress.

As a monolayer (endothelium) on the inner surface of vascular wall, endothelial cells (ECs) are constantly exposed to mechanical stimuli including shear stress from blood flow and circumferential stretch from blood pressure. ECs are able to sense mechanical stress, convert it into intracellular biochemical signaling and respond by changing their own behaviors or cross-talk with surrounding cells in order to adapt to the mechanical stimulus. ECs-mediated adaptive signaling contributes to the establishment of structural and functional plasticity of vessels under various mechanical stimuli. It is already known that endothelial adaptive response is crucial to the maintenance of vascular homeostasis and the regulation of vascular adaptation under physiological conditions, while uncontrolled or excessive vascular adaptation has been regarded as one of the earliest events leading to the generation of various pathological conditions [1,2]. Therefore, it has long been one of the main themes in vascular biology to study the cellular and molecular mechanisms underlying endothelial mecha-notransduction, a complex and dynamic process involving extracellular stimuli-sensing, intracellular signaling transduction and regulation of gene expression.
In the past years, significant progress has been made in this filed. A variety of endothelial mechanosensors, which are mostly associated with subcellular microdomains or organelles, have been identified; in addition, their downstream pathways, which may interact with each other at multiple levels, have been revealed [315]. The analysis of genetic and disease models demonstrated that endothelial mechanotransduction not only plays important roles in vascular adaptation or protection against mechanical stress, but also contributes to the pathogenesis of various vascular dis-orders [1620]. In this brief review, we will discuss recent advances in endothelial mechanotransduction with a focus on endothelial mechanosensors and their physiol-ogical and pathological roles. It should be noted that the term of endothelial mechanosensors used in this review is a collection of subcellular microdomains-associated mechanosensing molecules of ECs reported in literatures; it remains to be determined at single molecule level whether they sense physical forces directly or indirectly.

Subcellular microdomains and their associated mechanosensors in ECs
Because ECs are constantly subject to blood flow that has a frictional force and hydrostatic pressure, the search for the mechanosensors has attracted growing attention. Mechanotransduction is mostly initiated from the cell surface due to the deformation and/or fluidity alteration of plasma membrane by mechanical stimulus, transmitted along the cytoskeleton to sites of intercellular junctions and cellmatrix adhesions, and destined to the nucleus for transcriptional regulation [17], suggesting possible roles of these subcellular microdomains in mechanotransduction [21,22]. It has become increasingly clear that a variety of subcellular microdomains function as platforms of mechanosensors to sense and transduce mechanical stimulus. Below are major subcellular microdomains and their associated mechanosensors in ECs.

Caveolae and lipid rafts
Plasma membrane of cells has been organized by lipid rafts, which are enriched in cholesterol and sphingolipids, into various discrete microdomains called caveolae, omegashaped pits in the plasma membrane with a diameter of 6080 nm [2325]. Accumulated evidence suggests that caveolae act as platforms for conducting a variety of cellular functions including mechanotransduction [2630]. Caveolin-1 (Cav-1) protein is one of major components in caveolae. A recent study showed that after exposure to shear stress for 24 h, Cav-1/caveolae were enhanced and concentrated across the apical aspects of ECs [26]. In contrast, after stretching vascular caveolae became flattened, which may not only serve as a buffering system to decrease membrane tension and prevent cell damage or lysis, but also trigger cell protective signaling [27,28]. Altered blood pressure and flow changed the gene expression and protein phosphorylation levels of resident caveolae proteins including caveolins and cavins [29]. Direct in vivo evidence for a role of caveolae in mechanotransduction was obtained from Cav-1 knockout mice [30]. The Cav-1 deficient mice displayed defects in both acute flow-dependent dilation and chronic flow-dependent remodeling. Because both effects were rescued by Cav-1 re-expression in the endothelium, Cav-1 appears to be a critical mechanosensing molecule in ECs [30,31]. Accumulated data suggest that caveolae and its associated molecules play multiple roles in vascular paradigms via mechanotransduction [32,33].

Membrane proteins: ion channels and receptors
Several types of membrane proteins, such as some ion channels, VEGFR2 and G protein-coupled receptors (GPCRs), are known to sense and/or transduce mechanical stimuli in ECs. Although some of these membrane proteins are located at or associated with caveolae [3235], their mechanotransducing functions are not necessarily related to subcellular microdomains. We thus use this section to discuss these proteins, focusing on ion channels and GPCRs.
More than two decades ago, it was proposed that ion channels are endothelial mechanotransducers [3]. Consistent with this, shear stress was shown to activate K + current in ECs [36], which was later found to mediate the secretion of vasodilators [37] and gene expression [38]. Following these findings, shear stress was also shown to be able to activate calcium transients [39]; stretch was found to activate non-selective cation channels in ECs [40]. It has already been known that multiple mechanosensing ion channels, including stretch-activated (TRPC1 and TRPV2), shear stress-activated (TRPV4 and TRPP1/2), and pressure-activated (TRPM4 and TRPC6) channels are expressed in ECs [41]. It is worth noting that mechanical stimuli may activate other signaling molecules that indirectly activate ion channels, as clearly shown in the case of TRPV4 activation by cell swelling [42].
An early study showed that mechanical stimulus activated specific GTP binding proteins (G proteins) in ECs within 1 s of flow onset, representing one of the earliest mechanochemical signal transduction events reported to date in shear-stimulated endothelium [5]. A recent study showed that G alpha (q/11), the heterotrimeric G protein subunits, form a protein complex with PECAM-1, another mechanosensing molecule in cellular junction involved in rapid mechanosensing response of ECs [43]. Interestingly, G alpha (q/11) was absent from junctions in atheroprone areas as well as in all arterial sections of PECAM-1 knockout mice, suggesting that the G alpha (q/11)-PECAM-1 complex is a critical mediator of vascular diseases [43]. The activation of G proteins may be directly [44] or indirectly triggered by GPCRs under flow stimulus [45]. Recently, the chemokine receptors CXCR1 and CXCR2, known as G protein-coupled receptors required for migratory response of ECs toward the shear stress-dependent CXCL8 (Interleukin-8), are potentially novel mechanosensors in ECs for hemodynamic forces [46]. The detailed role of these receptors in mechanotransduction awaits further investigation.

Primary cilia
A primary cilium, which connects to the cytoskeletal microtubules of the cells doublets via the basal body of the cell, protrudes from the cell into the lumen of the vessel with the length of proximately 15 μm. Primary cilium is increasingly regarded as a mechanosensing organelle in several cellular systems including ECs [47]. Primary cilia in ECs exposed to unidirectional fluid flow bend in the direction of flow. In addition, ECs with cilia displayed a stronger response to shear stress in vitro than those without cilia [48]. It is known that mechanosensing molecules, such as polycystins and integrins are present in cilia of ECs. Consistently, ECs without primary cilia failed to convert mechanical stimulus into intracellular calcium and nitric oxide signaling in ECs [49]. Interestingly, fluid flow has been shown to cause loss of primary cilia from the endothelial surface [50], suggesting a feedback between mechanical stimulus and ciliary structure [51]. A recent study indicated that loss of primary cilia primed shear-induced endothelial-to-mesenchymal transition, indicating a functional link between primary cilia and flow-related endothelial performance [52].

Glycocalyx
Glycocalyx, a mesh-like structure of hydrated proteoglycans and glycosaminoglycans, covers the luminal side of ECs in both arteries and veins. Glycocalyx thickness increases with vessel diameter, ranging from 2 to 4.5 µm in arteries [53,54]. In addition to be a barrier that determines the vascular permeability and restricts the molecules and leukocytes from reaching the endothelium, glycocalyx is also involved in endothelial mechanotransduction. Flow stimulus modulated the production and distribution of glycocalyx [55]. After exposure to shear stress for 24 h, the glycocalyx components, e.g., heparan sulfate, chondroitin sulfate, glypican-1 and syndecan-1, were enhanced and associated with the changes in membrane rafts and the actin [26]. In contrast, the removal of glycocalyx by heparitinase significantly reduced flow-induced endothelial responses [8,56,57]. The redistribution of the glycocalyx also appears to act as a cell-adaptive mechanism by reducing the shear gradients that the cell surface experiences, which play a central role in mediating fluid shear stress-induced cell motility and proliferative response [26,55]. Further in vivo study found that the glycocalyx is present as soon as blood flow is initiated, which contributes to normal vascular development [58]. Evidence has been provided to show the glycocalyx is the first line of defence against atheror-ombotic disease and specifically its dysfunction is thought to be the first step during atherothrombosis [59].

Cellular junction and adhesion molecules
Endothelial cellular junctions include tight, gap and ad-herens junctions that are mediated by occludin/claudins, connexins and VE-cadherin, respectively [60]. These junction molecules have been proposed as putative mechanotransducers, which are able to sense blood flow [61]. A mechanosensory complex of proteins at cellular junctions, which include PECAM-1, VE-cadherin and VEGFR2, has been well studied in flow and stretch-mediated mechanotransduction in ECs [2,6265]. Cellular adhesion to the surrounding extracellular environment is essential to numerous aspects of cellular physiology including mechanobiology. Focal adhesions constitute discrete contact sites where integrin receptors connect the extracellular filamentous meshwork to the intracellular cytoskeleton, mediating bidirectional mechanical transduction through endothelial cell [66]. The integrins in ECs are known to be mechanotransducers critically involved in force-sensing processes by exerting their action between the extracellular matrix and the contractile actomyosin cytoskeleton [67]. Notably, integrin ligation by matrix proteins at focal adhesions in ECs is possibly required to initiate the signaling pathway leading to shear stress-induced vasodilation and blood pressure regulation [6,68,69]. As a key focal adhesion component and intracellular signaling transducer, the role of focal adhesion kinase (FAK) in endothelial mechanotransduction has been mostly studied. Mechanical stretch rapidly increased tyrosine phosphorylation of FAK [70]. Up to now, the studies have showed that FAK is involved in endothelial responses to various patterns of mechanical stimuli including cyclic stretch and shear stress, regulating vascular cell proliferation, remodeling and vascular inflammation [71].

Other subcellular microdomains
In order to effectively and coordinately sense, transmit and respond to mechanical stimuli, other parts/compartments, such as cytoskeleton and nuclei, of cells also play important roles [7274]. Cytoskeletal components determine and maintain cell shape and integrity. Mechanical stimuli, such as shear stress and stretch, cause to some extent the changes of cellular morphology, which exerts effects on interconnected network of cytoskeleton and triggers cellular response. A delicate study with a laser trap showed that the mechanical force even as small as 5.5 pN applied to actin stress fibers was strong enough to trigger an influx of calcium ions, presumably owing to the activation of mechanosensitive ion channels in the plasma membrane [75]. Such a stimulus-response process may be very fast, sometimes at the speed of the order of hundreds of milliseconds [76]. Shear stress activates eNOS and endothelin-1 gene expression through a pathway involving the intermediate filament vimentin, the microtubule network and actin [77,78]. Interestingly, blood flow-induced vascular remodeling is facilitated by the genetic absence of the intermediate filaments, vimentin and desmin suggesting that these elements oppose the process [79,80]. Recent studies suggested that the nu-cleus is an important contributor to the overall mechanics of the cell [81]. The nucleus can serve as an intracellular "mechanostat", a structure that is able to sense and respond to changes in the mechanical properties of the cellular environment by changing its own stiffness [81,82]. A very recent study proposed that endothelial nucleus is a sensor of blood flow direction and strength via the hydrodynamic drag applied to their nuclei [15]. The detailed mechanism in nucleus-mediated mechanotransduction and its significance in vascular biology and in mechanical stress-related diseases need to be further explored.

Features of mechanosensing and transducing system
The localization of the mechanosensors in microdomains mentioned above is not absolute; in addition, their behaviors are dynamic. For example, caveolae harbor VEGFR2 [33] and PECAM-1 [34], which are known to be mechanosensors in cellular junctions [2,62,63]. Consistently, VEGFR2 activation was impaired in endothelial cells from Cav-1 KO mice [33]. Furthermore, the mechanosensing molecules frequently interact with each other. Recent studies showed that G alpha (q/11), the heterotrimeric G protein subunits, form a protein complex with PECAM-1 to rapidly respond to mechanical stimuli [43,83]. Such dynamic and interactive feature allows the mechanosensors to finely and synergistically sense and respond to environmental stimuli.
Although mechanosensors transmit different signaling pathways, their downstream pathways are often integrated to activate some key enzymes or to induce the expression of a set of genes to perform certain functions. For example, the shear stress or acute stretch induced activation of eNOS, which is crucial in maintaining vascular homeostasis by dilating vessel and inhibiting inflammatory responses, can be finely-tuned by several upstream kinases [84,85]. The production of reactive oxygen species (ROS), a key molecule initiating many pro-atherogenic events, is tightly controlled by a variety of stimulatory and inhibitory pathways under various flow patterns and hypertensive stretch [8688]. Similarly, the activity of NF-B, a critical pro-inflammatory nucleus factor, is modulated by several signaling pathways [89]. So far, however, little is known about the signaling adaptors that mediate signal transduction between mechanosensors and downstream pathways. The scaffolding molecules, i.e., Shc and Grb2-associated binding protein 1 may be the candidates as they are mechanosensitive and able to coordinate multiple signaling pathways from various receptors [6,9094].

Physiological and pathological roles of mechanosensors
In the in vivo setting, endothelial mechanosensing system operates in the context of cell-cell contact and under various patterns of flow in blood vessels. Therefore, the mechanosensors in different ECs in different parts or organs appear to be heterogenous. It is known that in response to mechanical stress, ECs interact with vascular smooth muscle cells (VSMCs) [95,96]. Since VSMCs, which are also heterogenous throughout the vascular system, are able to control vessel diameter by contraction or relaxation via putative mechanosensors, i.e., DEG/ENaC proteins [97], the physical contact between ECs and VSMCs generates the heterogeneity of endothelial mechanosensors. Studies suggested that the expression of some mechanosensors is dependent on the location of the vessel and the local flow pattern. For example, the layer of glycocalyx in the internal carotid sinus region is thinner than that in the common carotid artery [16], whereas primary cilium is present in areas of low and disturbed flows in the adult aortic [98]. Accumulated results indicate that several types of hemodynamic forces are important in maintaining normal functions of the EC under physiological condition, while other types may lead to endothelial dysfunction under physiological status, contributing to the development of vascular diseases.

Requirement of endothelial mechanosensors for vascular adaptation and remodeling
Mechanical stress-induced immediate responses include biochemical changes, such as the activation of mechanosensors and secondary signaling pathways, increase in calcium concentration and nitric oxide release; cellular responses, such as membrane deformation, cytoskeletal remodeling and cell secretion; vessel changes, such as flow-mediated dilatation and pressure-mediated vascular tone. These responses serve to maintain vascular homeostasis through locally changing hemodynamics and prevent vascular cells from damage by mechanical injury [65,85,99]. Mechanical stress-induced sustained responses are relatively slow, adaptive and eventually lead to structural-wall remodeling, involving gene expression, cell differentiation and growth. Previous studies found that the vessel diameter is determined by an endothelium-dependent interplay between shear stress and the local pressure profile [100102], suggesting the importance of endothelial mechanotransduction in sensing and transducing vascular mechanical stress in vascular remodeling. Consistently, lack of caveolae-mediated mechanotransduction impaired acute flow-dependent dilation and chronic flow-dependent remodeling [30].

The interaction between mechanosensors and mechanical stresses is an important determinant of vascular diseases
The geometric distribution of mechanical stress is closely associated with site-specific susceptibility to pathological changes [1,2,18,103], such as atherosclerosis and inflamma-tion in arteries, suggesting the effects of mechanical stress on the phenotypic changes of ECs and vascular wall. Clinical studies suggested that the sites of atherosclerosis are not random, but preferentially located at arterial branches and curvatures where the local flow is disturbed. The effects of disturbed flow on vascular homeostasis, the signaling pathways and the involvements in pathological conditions have been well studied. For the details, please refer to recent excellent reviews [1,2,89,104]. On the other hand, recent studies showed that the increased density and expression of mechanosensors could initiate the mal-adaptation of vessels [105,106]. The pressure-activated cation channel upregulation in intact endothelium of aorta was suggested to contribute to severe genetic hypertension in a rat model [106]. Interestingly, PECAM-1-knockout mice did not activate NF-B and downstream inflammatory genes in regions of disturbed flow, suggesting that this mechanosensing pathway is required for the earliest-known events in atherogenesis [63]. Therefore, the interaction between endothelial mechanosensors and mechanical stresses determines the site-specific susceptibility of vascular pathogenesis. So far, except for the atherosclerosis, which is well known to be associated with endothelial mechanotransduction [1,2], abnormal cilia function (ciliopathy) in ECs has been linked to renal diseases [107,108], while endothelial glycocalyx dysfunction has been regarded as a key factor in the develop-ment of microvascular hyperpermeability-related diseases [109,110]. It can be expected that in the near future, more and more mechanosensor-related diseases will be revealed.
The main feature of endothelial dysfunction is unable to sense the changes in hemodynamic forces and blood-borne signals, and to respond suitably. Besides, conventional risk factors were shown to damage mechanosensors [111]. In this regard, the suitable density and normal function of mechanosensors of ECs are the gatekeepers of cardiovascular functions. However, many vascular pathological changes are mediated through endothelial mechanosensing and transducing system [1]. Such opposite roles imply mechanosensors are double-edged swords, whose functions depend on the complex interactions among geometric mechanical stimuli, the expression and functional status of endothelial mechanosensors, and conventional risk factors.

Perspectives
In the last decades, with assembling of pieces of our understanding about mechanosensing and mechanotransducing process, a preliminary network of mechanosensing mechanisms and biological functions of mechanotransduction in ECs has gradually emerged ( Figure 1). However, there are two major challenges in the field. The first challenge lies in the basic research aspect. A single endothelial mechanosensor unlikely exists; rather, mechanosensing and transducing process occurs at multiple subcellular microdomains. How do those multiple mechanosensors work together to effectively sense the mechanical stimulus and transmit into biochemical signaling pathways? How do ECs integrate multiple signaling pathways to exert functionally unified responses to mechanical stresses? How do ECs transmit mechanical signaling into surrounding cells, especially smooth muscle cells, to coordinate various responses to maintain vascular homeostasis and adapt to mechanical stress? To answer these questions, we need to use approaches from biophysics, molecular cell biology, physiology, developmental biology, bioengineering and bioinformatics to carry out comprehensive studies.
Another challenge comes from translational medicine research. With better understanding of endothelial mechanotransduction, we are prompted to develop new diagnostic, therapeutic and pharmacological approaches for treatment of cardiovascular diseases. Since current cardiovascular drugs are mostly aimed at lowering the risk factors (e.g., diabetes, hyperlipidemia, and hypertension), targeting at endothelial mechanosensors and/or intracellular regulators together could be more beneficial. Because those mechanosensors and signaling regulators are required for other fundamental cellular processes, caution should be rendered when translating the bench data into clinical applications.