Endothelial Cell Plasma Membrane Biomechanics Mediates Effects of Pro-Inflammatory Factors on Endothelial Mechanosensors: Vicious Circle Formation in Atherogenic Inflammation

Chronic low-grade vascular inflammation and endothelial dysfunction significantly contribute to the pathogenesis of cardiovascular diseases. In endothelial cells (ECs), anti-inflammatory or pro-inflammatory signaling can be induced by different patterns of the fluid shear stress (SS) exerted by blood flow on ECs. Laminar blood flow with high magnitude is anti-inflammatory, while disturbed flow and laminar flow with low magnitude is pro-inflammatory. Endothelial mechanosensors are the key upstream signaling proteins in SS-induced pro- and anti-inflammatory responses. Being transmembrane proteins, mechanosensors, not only experience fluid SS but also become regulated by the biomechanical properties of the lipid bilayer and the cytoskeleton. We review the apparent effects of pro-inflammatory factors (hypoxia, oxidative stress, hypercholesterolemia, and cytokines) on the biomechanics of the lipid bilayer and the cytoskeleton. An analysis of the available data suggests that the formation of a vicious circle may occur, in which pro-inflammatory cytokines enhance and attenuate SS-induced pro-inflammatory and anti-inflammatory signaling, respectively.


Blood Flow Patterns and the Pro-Inflammatory Response of Endothelial Cells
Inflammation is a key mechanism affecting endothelial cells (ECs) and leading to atherosclerosis [1][2][3][4]. ECs are subjected to three main hemodynamic forces: hydrostatic pressure, cyclic stretch, and shear stress (SS). Fluid SS is a frictional force, produced by blood flow, acting tangentially to the surface of the ECs. Depending on the patterns of fluid SS acting on ECs, either anti-or pro-inflammatory mechanisms can be triggered in Two major types of physical effects, integral membrane proteins or ion channels, are expected to draw from the hosting lipid bilayer due to its profiles of electrical charges and mechanical properties [43]. The former one arises due to the consideration that the distribution of the charges of the lipids on either monolayer may or may not show any net charges, but due to their coupling with the integral membrane proteins (MPs) or channel proteins, the charges on both lipids and proteins become redistributed or polarized. As a result, any MP-lipid coupling appears with a distinctive type of physical phenomenon, many charges interactions, as explained using screened Coulomb interactions (SCIs) in ref. [43]. Considering the mechanical properties (bilayer elasticity and lipid intrinsic curvature) of lipid layers, only the bilayer regulation of the integral membrane protein function has long been addressed (see refs. [45][46][47][48][49][50][51][52]). However, in refs. [43,44], it is clearly shown that the SCI model that considers the charge-based interactions among integral MPs and hosting bilayer lipids can correctly address the bilayer regulation of MP or channel functions, and that the charge-based interactions appear to be primary regulators of channel functions. The bilayer mechanical property-based regulation of MP functions also appears in SCI treatment, but only to produce a secondary effect on MP functions. In the elastic bilayer model, the primary effect due to charge-based interactions was totally ignored [45][46][47][48][49][50][51][52]. The charge-based effects have later been consistently found to be appearing as primary molecular mechanisms, especially when using molecular dynamics (MD) simulations, on varieties of membrane-adsorbed peptides and drugs cases (see refs. [53,54]).
Recently, the plasma membrane of immune cells was suggested to integrate multiple biophysical and biochemical stimuli (such as cholesterol content, negatively charged lipids, electrical potential) in order to regulate immune receptor function [55].
Both the composition of the lipid bilayer [22] and the NMMII-generated basal tension of EC [56] can be altered by pro-inflammatory stimuli. Endothelial mechanosensors, Two major types of physical effects, integral membrane proteins or ion channels, are expected to draw from the hosting lipid bilayer due to its profiles of electrical charges and mechanical properties [43]. The former one arises due to the consideration that the distribution of the charges of the lipids on either monolayer may or may not show any net charges, but due to their coupling with the integral membrane proteins (MPs) or channel proteins, the charges on both lipids and proteins become redistributed or polarized. As a result, any MP-lipid coupling appears with a distinctive type of physical phenomenon, many charges interactions, as explained using screened Coulomb interactions (SCIs) in ref. [43]. Considering the mechanical properties (bilayer elasticity and lipid intrinsic curvature) of lipid layers, only the bilayer regulation of the integral membrane protein function has long been addressed (see refs. [45][46][47][48][49][50][51][52]). However, in refs. [43,44], it is clearly shown that the SCI model that considers the charge-based interactions among integral MPs and hosting bilayer lipids can correctly address the bilayer regulation of MP or channel functions, and that the charge-based interactions appear to be primary regulators of channel functions. The bilayer mechanical property-based regulation of MP functions also appears in SCI treatment, but only to produce a secondary effect on MP functions. In the elastic bilayer model, the primary effect due to charge-based interactions was totally ignored [45][46][47][48][49][50][51][52]. The charge-based effects have later been consistently found to be appearing as primary molecular mechanisms, especially when using molecular dynamics (MD) simulations, on varieties of membrane-adsorbed peptides and drugs cases (see refs. [53,54]).
Recently, the plasma membrane of immune cells was suggested to integrate multiple biophysical and biochemical stimuli (such as cholesterol content, negatively charged lipids, electrical potential) in order to regulate immune receptor function [55].
Both the composition of the lipid bilayer [22] and the NMMII-generated basal tension of EC [56] can be altered by pro-inflammatory stimuli. Endothelial mechanosensors, which are embedded in the plasma membrane, not only influence the blood flow, but also regulate the mechanical properties of the lipid bilayer and CSK. Here, we discuss if and how pro-inflammatory factors may change the mechanical properties of the lipid bilayer and CSK. The rigidification of the lipid bilayer and the increase in the cytoskeletal NMMII-generated tension can increase the energy barrier for the activation of endothelial mechanosensors by SS. As a result, high SS can be perceived by ECs as low SS. Since low SS induces the release of pro-inflammatory mediators [2,4] the vicious circle can be formed. It keeps up the low-grade vascular inflammation and promotes the development of endothelial dysfunction and atherosclerosis. Recently, the primary cilia on ECs were suggested to amplify low unidirectional SS signaling, resulting in the activation of NRF2 and the protection of ECs from oxidative damage [6]. We discuss the data suggesting that, under inflammatory conditions, the lipid bilayer and the NMMII-generated tension could dampen high laminar SS. Blood flow exerts extracellular forces, such as hydrodynamic pressure, cyclic stretch, and fluid SS. In addition to these forces, the substrate stiffness is also sensed by ECs [57,58]. The dependence of the mechanical properties of the cortical CSK in HAECs and human umbilical vein ECs (HUVECs) on laminar SS strength was investigated using acoustic force spectroscopy [59]. The exposure of HAECs and HUVECs to laminar SS (6 dyn/cm 2 for up to 48 h) was found to evoke an increase in the membrane cortex stiffness [59].

What
The stiffness of the whole EC, or its plasma membrane with the underlying submembrane actin-based CSK (smACSK), is an integral parameter, evaluating the behavior of the whole cell. At the cellular level, the stiffness of bovine pulmonary arterial ECs (BPAECs) depends on the basal isometric tension, which is determined by NMMII contractility [56]. Thrombin induces a rapid increase in basal isometric tension in BPAECs via the MLCKand RhoA-mediated activation of NMMII [56]. Further studies on the biomechanics of whole ECs should lead to an analysis of the spectrum of intracellularly generated forces that converge on any single transmembrane mechanosensor [60]. Evidence is accumulating that forces generated within the cell regulate mechanical tension across the transmembrane, cytoskeletal, and scaffolding proteins [61][62][63].
Endothelial mechanosensors are subjected to forces generated within the cell, in particular within the lipid bilayer of the plasma membrane, and forces arising from smACSK ( Figure 2). Energy inputs from the lipid bilayer and the smACSK are likely to increase, or decrease, the activation energy required for the stimulation of mechanosensors by fluid SS (Figure 2). Earlier, we proposed that the gradients in hydrostatic pressures across the plasma membrane induced by changes in cell volume are actively probed by cells via the pulling activity of non-muscle myosin II (NMMII) and the pushing activity of smACSK [64]. Considering the forces arising from the lipid bilayer (see Section 1.2), and the forces generated by smACSK, the energy input (E(intracellular)) that is received-in addition to extracellular mechanical forces-by any single mechanosensory can be presented as the sum of the following energies: E(intracellular) = E(lipid bilayer) + E(NMMII) + E(protrusion) + E(resistance) + E(smACSK spring) (1) where E(lipid bilayer) is an energy from the lipid bilayer, E(NMMII) is an energy of NMMIIgenerated pulling (directed into the cell) force, E(protrusion) is an energy of pushing (directed out of the cell) force due to the actin-based assembly of lamellipodia and filopodia, E(resistance) is an energy generated by the lipid bilayer together with smACSK, and E(smACSK spring) is an energy stored by smACSK during its mechanical deformation. and E(smACSK spring) is an energy stored by smACSK during its mechanical deformation.

Figure 2.
Scheme illustrating the actions of intracellular forces on any single endothelial mechanosensor. The transmembrane domain(s) of the mechanosensor is sensing the lipid bilayer lateral pressure. A pulling force from NMMII-generated tension and a pushing force from an actin-based protrusion are also sensed by the mechanosensor, meaning that the energy required for mechanosensor activation can be increased; in this case, the high LSS can be perceived as low LSS, or decreased, in which case the low LSS can be perceived as high LSS. Thus, the interplay of intracellular forces can shift LSS-induced signaling to be either pro-atherogenic or anti-atherogenic. Abbreviations: LSS, laminar shear stress; NMMII, non-muscle myosin II; smACSK, submembrane actinbased cytoskeleton.
Thus, the energies received from within the cell would also influence the activation of an endothelial mechanosensor: E(activation of endothelial mechanosensor) = E(extracellular) + E(intracellular) (2) In this paper, we only pay attention to three intracellular forces: the biomechanics of the lipid bilayer, the pulling force of NMMII, and the pushing force of the lamellipodia. NMMII and smACSK are controlled by many intracellular signaling mechanisms; however, we only consider two functionally antagonistic signaling proteins, RhoA and Rac1. The convergence of several pro-inflammatory stimuli on the two functionally antagonistic small GTPases RhoA and Rac1 is discussed in Section 3.

Biomechanics of the Lipid Bilayer and the Activation Energy of Mechanosensors
There are two mechanisms for the regulation of transmembrane proteins by lipids: a ligand-like mechanism, when the direct high-affinity binding of lipids to proteins occurs, and a solvent-like mechanism, when the addition or removal of a lipid changes the biomechanics of the lipid bilayer [65,66]. For example, many transmembrane proteins contain specific motifs for cholesterol binding: a cholesterol recognition/interaction amino acid consensus (CRAC, R/K-X5-Y-X5-L/V), a reversed CRAC motif named CARC (L/V-X5-Y-X5-R/K), and a cholesterol consensus motif (CCM) [65][66][67][68][69]. In this paper, we mainly pay attention to the solvent-like scenario, which deals with alterations in plasma membrane biomechanics. A pulling force from NMMII-generated tension and a pushing force from an actin-based protrusion are also sensed by the mechanosensor, meaning that the energy required for mechanosensor activation can be increased; in this case, the high LSS can be perceived as low LSS, or decreased, in which case the low LSS can be perceived as high LSS. Thus, the interplay of intracellular forces can shift LSS-induced signaling to be either pro-atherogenic or anti-atherogenic. Abbreviations: LSS, laminar shear stress; NMMII, non-muscle myosin II; smACSK, submembrane actin-based cytoskeleton.
Thus, the energies received from within the cell would also influence the activation of an endothelial mechanosensor: E(activation of endothelial mechanosensor) = E(extracellular) + E(intracellular) (2) In this paper, we only pay attention to three intracellular forces: the biomechanics of the lipid bilayer, the pulling force of NMMII, and the pushing force of the lamellipodia. NMMII and smACSK are controlled by many intracellular signaling mechanisms; however, we only consider two functionally antagonistic signaling proteins, RhoA and Rac1. The convergence of several pro-inflammatory stimuli on the two functionally antagonistic small GTPases RhoA and Rac1 is discussed in Section 3.

Biomechanics of the Lipid Bilayer and the Activation Energy of Mechanosensors
There are two mechanisms for the regulation of transmembrane proteins by lipids: a ligand-like mechanism, when the direct high-affinity binding of lipids to proteins occurs, and a solvent-like mechanism, when the addition or removal of a lipid changes the biomechanics of the lipid bilayer [65,66]. For example, many transmembrane proteins contain specific motifs for cholesterol binding: a cholesterol recognition/interaction amino acid consensus (CRAC, R/K-X 5 -Y-X 5 -L/V), a reversed CRAC motif named CARC (L/V-X 5 -Y-X 5 -R/K), and a cholesterol consensus motif (CCM) [65][66][67][68][69]. In this paper, we mainly pay attention to the solvent-like scenario, which deals with alterations in plasma membrane biomechanics.
There is convincing evidence that the lipid composition, which determines the lipid bilayer's fluidity, influences both basal and SS-induced GTPase activities of G αq and G αi3 subunits of heterotrimeric G proteins in phospholipid vesicles [70]. The incorporation of lysophosphatidylcholine into liposomes increases the fluidity of the lipid bilayer and elevates the basal activity of G αq and G αi3 proteins from 0.47 to 1.35 pmol/min per µg of protein [70]. The incorporation of benzyl alcohol, another fluidizing agent, increases the basal activity of G proteins from 0.47 to 2.37 pmol/min per µg of proteins. On the other hand, the incorporation of cholesterol, which decreases bilayer fluidity, diminishes the basal activity of Gαq and Gαi3 proteins from 0.47 to 0.113 pmol/min per µg of protein and reduces the activation of G proteins by SS [70].
In HUVECs, fluid SS (from 0.7 to 33 dyn/cm 2 ) induces an increase in membrane fluidity [71]. The addition of benzyl alcohol also increases the membrane fluidity [71]. The exposure of BAECs to fluid SS increases the plasma membrane fluidity [72]. Additionally, BAECs' membrane fluidity is increased and decreased by benzyl alcohol (a fluidizing agent) and cholesterol (a rigidifying agent), respectively [72]. In HPAECs, fluid SS increases membrane fluidity [30,31]. Furthermore, the plasma membranes in HPAECs discriminate between cyclic stretch and fluid SS, in that cyclic stretch increases the lipid bilayer order and decreases fluidity, while fluid SS decreases the lipid bilayer order and increases fluidity [73].
Due to all the changes in the membrane composition, which have been explained above, a membrane's two major physical properties, namely, the charge profiles and mechanical properties, may especially become altered. Consequently, as explained earlier, the membrane regulation of integral MP functions also changes [43,[45][46][47][48][49][50][51][52]. However, the exact energy (generally refereed as the 'free energy of bilayer-integral protein coupling') that plays important roles in such membrane regulation of integral protein functions has been correctly calculated using the SCI models that consider charge-based interactions [43,53,54]. The SCI model explains all the parameters behind calculating the free energy of bilayerintegral protein coupling, and that this energy consists of both components drawn from the charge properties and mechanical properties of the bilayer and MPs. Thus, it appears to be a universal mechanism which also raises some universal probability functions related to any bilayer-MP coupling energetics (see details in ref. [54]). MD simulations on lipid-drug pair interactions in the bilayer environment have especially demonstrated these universal probability functions to be primarily relying on two major types of charge-based lipid-drug interactions, namely the electrostatic and van der Waals interactions.
The small GTPase, RhoA, and its effector, Rho-associated coiled-coil-containing kinase (ROCK), activate NMMII, both via the phosphorylation of RLC at Ser19/Thr18 [96] and the phosphorylation and inhibition of MLCP [97]. ROCK, the main downstream effector of RhoA, has two isoforms: ROCK1 and ROCK2 [98]. Rac1 and its effector, p21activated kinase (PAK), inhibit NMMII via the phosphorylation of MLCK by PAK [99]. Rac1 itself can be regulated by cell-generated tension: in rat aortic smooth muscle cells, the inhibition of myosin contractility via the inhibition of Rho-kinase with Y-27632 or MLCK with ML-7 increased Rac1 activity [61].
Seemingly, the RhoA-dependent activation of NMMII and the generation of the centripetal force, as well as Rac1-dependent protrusive actin polymerization, would affect the endothelial mechanosensors. As pro-inflammatory agents, via the activation of RhoA and/or Rac1, they not only regulate the endothelial permeability, but also may tune mechanosensors because of the inducing pulling (RhoA-mediated NMMII contractility) or pushing (Rac1-mediated actin-based protrusions) of intracellular forces. The small GTPase, RhoA, and its effector, Rho-associated coiled-coil-containing kinase (ROCK), activate NMMII, both via the phosphorylation of RLC at Ser19/Thr18 [96] and the phosphorylation and inhibition of MLCP [97]. ROCK, the main downstream effector of RhoA, has two isoforms: ROCK1 and ROCK2 [98]. Rac1 and its effector, p21-activated kinase (PAK), inhibit NMMII via the phosphorylation of MLCK by PAK [99]. Rac1 itself can be regulated by cell-generated tension: in rat aortic smooth muscle cells, the inhibition of myosin contractility via the inhibition of Rho-kinase with Y-27632 or MLCK with ML-7 increased Rac1 activity [61].
Seemingly, the RhoA-dependent activation of NMMII and the generation of the centripetal force, as well as Rac1-dependent protrusive actin polymerization, would affect the endothelial mechanosensors. As pro-inflammatory agents, via the activation of RhoA and/or Rac1, they not only regulate the endothelial permeability, but also may tune mechanosensors because of the inducing pulling (RhoA-mediated NMMII contractility) or pushing (Rac1-mediated actin-based protrusions) of intracellular forces.

Rac1 in the Regulation of Actin Polymerization Pushing Force in Lamellipodia
ECs are known to form lamellipodia, which are protrusive actin-based structures [102]. Lamellipodia assembly is mainly governed by Rac1 [87,88]. Lamellipodia push the plasma membrane out from the cell interior, and this pressure can increase or decrease the energy required for the activation of any endothelial mechanosensory by SS. For example, VE-cadherin, an element of the mechanosensory PECAM-1/VE-cadherin/VEGFR2 complex [8], is a component in lamellipodia in HUVECs [102]. There is an interesting interplay between pushing, which is generated by protrusive actin polymerization, and NMMIIdependent pulling forces in the formation of VE-cadherin mediated adherens junctions in HUVECs [102]. It should be noted that NMMII contractility is required for lamellipodia formation [102,103].

Pro-Inflammatory Stimuli and the Lipid Bilayer Biomechanics
Hypoxia itself can be a mechanical signal for ECs, as a decrease in the number of dioxygen molecules dissolved in the lipid bilayer of the plasma membrane is a mechanical stimulus that can influence the mechanosensitive transmembrane proteins and, thus, participate in the hypoxia response [104].
However, hypoxia can also lead to the increased production of reactive oxygen species (ROS) [105], which may increase membrane lipid peroxidation, with effects on the mechanical properties of the lipid bilayer. Oxysterols, products of cholesterol oxidation, promote the development of atherosclerosis [106]. Connections between dyslipidemia and the mechanical properties of ECs were reviewed elsewhere [22]. Hypercholesterolemia leads to the accumulation of cholesterol in ECs, and promotes the development of inflammation and atherosclerosis [22,107,108]. An increase in low-density lipoprotein (LDL) cholesterol levels and a decrease in high-density lipoprotein (HDL) cholesterol levels in blood plasma are among the key risk factors for atherogenesis [107,109].

Pro-Inflammatory Cytokines in RhoA and Rac1 Regulation in ECs
As discussed above, RhoA-ROCK activation via the induction of actomyosin contractility may increase the tension experienced by EC mechanosensors, leading to an increase in the pro-inflammatory response in ECs. Pro-inflammatory mediators unbalance RhoA-Rac1 activities and homeostasis, resulting in changes to the intracellularly generated tension and mechanosensors' activation threshold.
Ang II can induce vascular inflammation and remodeling [10]. Ang II acts via two types of Ang receptors, type 1 (AT 1 R) and type 2 receptor (AT 2 R), which significantly differ in their physiological effects [119]. Signaling through AT 1 R leads to vasoconstriction, oxidative stress, and inflammation, while signaling through AT 2 R mediates anti-inflammatory effects [120,121] and prevents the development of hypertension in animal models of hypertension [119]. In BAECs, the stimulation of AT 1 R leads to the sequential activation of G α12/13 and RhoA [122]. In BAECs, Ang II acting via AT 1 R activates Rac1, and elevates focal adhesion complexes and actin fiber formation [118]. In contrast, the activation of AT 2 R has been linked with the negative regulation of RhoA activity in vascular smooth muscle cells [123]. In BAECs, C-reactive protein (CRP) activates the RhoA-ROCK pathway to induce the expression of plasminogen activator inhibitor-1 [124].

Piezo1
Piezo1 is a transmembrane cation channel [131][132][133] that is gated by membrane tension and SS [134,135]. Full-length Piezo1 was cloned from the mouse neuroblastoma N2A cell line and expressed in several other cell lines [136]. Piezo1 exhibits activation by stretching [136,137]. Fluorescent Piezo1 constructs were expressed in HEK293 cells [134]. Blebs formed in these transfected cells were deficient in the cytoskeletal proteins, and the basal Piezo1 activity in the bleb-attached patches was higher than in whole-cell-attached patches, suggesting that the membrane tension is a main driver of mechanosensitive gating of Piezo1, whereas the CSK has a mechanoprotective role [134].
Piezo1 performs multiple roles in the cardiovascular system, and links mechanical stimuli to the triggering of both pro-and anti-atherogenic responses in ECs [138]. Piezo1 is expressed in many cell types participating in the development of atherosclerosis, including ECs, vascular smooth muscles cells, T and B cells, and monocytes, which undergo sequential transition into macrophages and lipid-engorged foam cells [138]. The endothelial-specific deletion of Piezo1 in mice impairs SS-mediated vascular development [131], sprouting angiogenesis, and vascular lumen formation [139]. Moreover, EC Piezo1 can sense disturbed blood flow and is linked to inflammatory signaling [140]. In human umbilical arterial ECs (HUAECs), the fluid SS-induced activation of Piezo1 leads to ATP release and autocrine stimulation of the purinergic P2Y 2 receptor and its downstream effectors, G αq and G α11 , which triggers the activation of eNOS, NO release, and vasodilatation [141]. Depending on blood flow patterns, Piezo1 activation, together with the activation of purinergic P2Y 2 receptor and G αq/11 , may lead to atheroprotective signaling or atherogenic signaling in response to laminar SS or disturbed SS, respectively [140]. In the case of disturbed flow, the induction of atherogenic signaling proceeds via the activation of integrin by SS sensors Piezo1 and P2Y 2 -G αq/11 [140].

Mechanosensory PECAM-1/VE-Cadherin/VEGFR2 Complex
VE-cadherin is both a key player in the regulation of the endothelial barrier function [94,95] and an element in the PECAM-1/VE-cadherin/VEGFR2 mechanosensory complex [8]. Interestingly, a direct association between PECAM-1 and G αq/11 was reported in HUVECs [142,143]. In HUVECs, the dissociation of G αq/11 from PECAM-1 is dependent on G αq/11 activation by fluid SS, and on the patterns of fluid SS. Here, impulse and oscillatory, but not ramped-transient flow induced G αq/11 dissociation from PECAM-1 [142]. VEGFR2 is localized in lipid rafts, and interference in the lipid raft structure may impair its activation by SS [144]. The transmembrane domains of VE-cadherin and VEGFR2 and VEGFR3 can interact with each other, and this interaction is important for SS signal transduction onto and the activation of VEGFR2/3 [145]. The linking of VE-cadherin to the actin CSK seems to be mediated by βand α-catenins, as it takes place in epithelial cells containing epithelial (E)-cadherin [146]. E-cadherin binds to β-catenin, β-catenin interacts with α-catenin, and α-catenin binds to actin filaments [146].
The SS response of ECs includes re-arrangement in the EC cytoskeleton and the redistribution of the intracellular forces acting on PECAM-1 and VE-cadherin [63]. The onset of the flow elicited an increase in the tension on PECAM-1 and a decrease in the tension on VE-cadherin [63]. Vimentin, an intermediate filament, appears to transmit NMMII tension onto PECAM-1 [63]. In our hypothesis, RhoA-and Rac1-dependent intracellular forces can tune the mechanosensitivity of the PECAM-1/VE-cadherin/VEGFR2 complex ( Figure 5). Indeed, in HUVECs, the VE-cadherin located in the lamellipodia undergoes exhibits both pushing and pulling forces [102]. S1P, via the S1P 1 receptor, activates Rac1 and suppresses pro-inflammatory elevation in endothelial permeability [126,127] (Figure 5). Rac1 inhibits the RhoA-ROCK pathway that reduces NMMII-generated tension on VE-cadherin, resulting in the stabilization of VE-cadherin [101]. On the other hand, NMMII contractility also contributes to lamellipodia formation [102,103].

Heterotrimeric G Proteins and GPCRs
There is experimental evidence for the activation of GPCRs and heterotrimeric G proteins by fluid SS in the absence of GPCR agonists. The exposure of HUVECs to fluid SS (10 dyne/cm 2 ) induces the rapid (within 1 s) activation of Gαq and Gαi3 [147]. The reconstitution of purified Gαq and Gαi3 in phospholipid liposomes reveals that the lipid composition significantly influences the activation of G proteins in response to fluid SS (0-30 dynes/cm 2 ) [70]. In human coronary artery ECs (HCAECs), Gαq/11 proteins are activated by fluid SS independently of upstream GPCRs [148]. In addition, in HCAECs, the stimulation of Gαq/11 proteins by fluid SS is independent of the Piezo1 channel [143].
The human bradykinin type 2 receptor (B2R), with an inserted yellow fluorescent protein and fused to a cyan fluorescent protein, was expressed in BAECs [149]. The re-

Heterotrimeric G Proteins and GPCRs
There is experimental evidence for the activation of GPCRs and heterotrimeric G proteins by fluid SS in the absence of GPCR agonists. The exposure of HUVECs to fluid SS (10 dyne/cm 2 ) induces the rapid (within 1 s) activation of G αq and G αi3 [147]. The reconstitution of purified G αq and G αi3 in phospholipid liposomes reveals that the lipid composition significantly influences the activation of G proteins in response to fluid SS (0-30 dynes/cm 2 ) [70]. In human coronary artery ECs (HCAECs), G αq/11 proteins are activated by fluid SS independently of upstream GPCRs [148]. In addition, in HCAECs, the stimulation of G αq/11 proteins by fluid SS is independent of the Piezo1 channel [143].
The human bradykinin type 2 receptor (B 2 R), with an inserted yellow fluorescent protein and fused to a cyan fluorescent protein, was expressed in BAECs [149]. The recombinant B 2 R was activated by fluid SS, hypotonic stress, and benzyl alcohol in the absence of a B 2 R agonist [149]. Several long-chain polyunsaturated fatty acids (eicosapentaenoic acid, docosahexaenoic acid, docosadienoic acid, and dihomo-γ-linoleic acid) activated B 2 R in BAECs and in B 2 R-transfected HEK293 cells in a ligand-independent manner [150]. The search for a structural motif(s) responsible for the mechanosensitivity of the human histamine type 1 (H 1 ) receptor in HUVECs revealed that helix 8, located in the cytoplasmic C-terminus of the H 1 receptor, is essential [151]. Whether intracellular forces, such as NMMII-dependent contractions, influence the activities of heterotrimeric G proteins and GPCRs remains poorly understood.

Integrins
Integrins, heterodimeric adhesion proteins linking extracellular matrix proteins to the cytoskeleton, comprise 18 αand 8 β-subunits, which form 24 heterodimers [152]. Integrins sense various mechanical stresses and convert a mechanical force into biochemical signaling within the cell [58]. The multiple roles of integrins-expressed on ECs, leukocytes, monocytes/macrophages, vascular smooth muscle cells, and platelets-in the pathogenesis of atherosclerosis are reviewed in depth elsewhere [153,154].
Integrins are sensitive to the lipid bilayer ordering and thickness [155][156][157][158]. Moreover, in PAECs and mouse embryonic fibroblasts, integrins themselves seem to increase the lipid bilayer order [155]. The exposure of HUVECs to pro-inflammatory oscillatory SS (0.5 ± 4 dyn/cm 2 ) or anti-inflammatory pulsatile SS (12 ± 4 dyn/cm 2 ) induced the opposite effects on α5β1 integrin localization in lipid rafts [156]. Oscillatory SS and pulsatile SS increase or decrease the levels of α5β1 integrins in lipid raft regions, respectively [156]. The exposure of HAECs to three non-lipid amphiphiles (vitamin E, Triton X-100, and benzyl alcohol) was used for the study of β1 integrin dependence on lipid bilayer order and domain thickness [157]. Only benzyl alcohol partitioned into the liquid-disordered domains and thinned these domains, enhancing β1 integrin affinity and valency, and inducing β1 integrin clustering [157]. Lietha and Izard suggested that mechanical stress-induced β integrin activation is mediated by membrane thinning [158].
Integrins are linked to the actin-based cytoskeleton via numerous scaffolding proteins, such as talin, vinculin, filamin A, and zyxin [58,152]. Integrins themselves regulate NMMII [153]. NMMII-generated contractility is required for the maintenance of integrin adhesion complexes [159]. In the stationary membranes of BAECs, the tension across vinculin was generated by NMMII and its upstream regulator RhoA, and was about 2.5 pN (0.25 µdyn) [62].

Conclusions
The experimental data considered here suggest that internal cellular forces-mainly NM-MII contractility and actin-based lamellipodial protrusions, together with pro-inflammatory cholesterol-and oxidative stress-induced changes in the biomechanics of the lipid bilayertune the mechanosensitivity of the endothelial mechanosensors. Since RhoA and Rac1, key regulators of NMMII contractility and actin filaments assembly, are themselves critically regulated by pro-inflammatory agents, the biomechanics of the plasma membrane of ECs can be softened or stiffened upon inflammation, with subsequent alteration in ECs response to fluid SS. Fluid SS, depending on its patterns, is among the main controllers of endothelial inflammation. We propose that the altered-by first exposure of ECs to pro-inflammatory agents-mechanosensitivity of mechanosensors leads to their aberrant activation, and to the formation of a vicious cycle where pro-inflammatory signaling is sustained. Low-grade vascular inflammation then promotes atherogenesis.