2. Mechanotransduction of shear stress: role of the structure proteins

Sensing and transducing shear stress into a biological signal is a process remaining mainly unknown in the peripheral circulation. In addition, shear stress is extremely small, a few dyn per cm2 (compared to thousands for blood pressure). Shear stress stimulates very gently the intraluminal side of the endothelial cells. Thus the first challenge is to understand the sensing process of this force by the glycocalix and/or by or structures in the extracellular space before its transmission to the intracellular machinery through the plasma membrane. A flow sensor is probably made of the interaction between several elements of the extracellular matrix as described in cultured endothelial cells ²² but with a time scale not necessarily compatible with FMD, which occur in the seconds following a change in blood flow. In FMD a possible “flow sensor” might be made of proteoglycans carrying negative charges and binding extracellular sodium cations as described by JA Sevan ¹¹. Indeed, the capacity of proteoglycans to bind sodium is sensitive to shear stress ^{23, 24} and the response of blood vessels to flow involves preferentially sodium influx ²⁵. In addition, shear-induced dilation of small arteries is abolished by removal of sialic acids by neuraminidase ²⁶ and in cultured bovine aortic endothelial cells the partial removal of heparan sulfate proteoglycan with heparinase prevents shear-induced NO release ²⁷. The importance of the glycocalyx is also evidenced in diabetic patients, who have a reduced FMD in association with a lower glycocalyx volume and an enhanced level of circulating glycocalyx elements ²⁸. As discussed in a recent review article describing the importance of a thick glycocalyx for endothelium integrity ²⁹, numerous factors modulate the integrity of the endothelial glycocalyx. Furthermore, endothelial carbohydrate damage during ischemia reperfusion, hypoxia or in an oxidative environment with high oxidized low-density lipoprotein level, contribute to tissue edema and alteration in shear stress sensing ²⁹.

The ultrastructural analyses of the glycocalyx shows that it forms a network, which parallels a cortical web located beneath the membrane and made of filamentous actin ³⁰. Nevertheless, this observation does not show how shear stress is transmitted to this actin network, as shear stress is normally equivalent to zero at the level of the membrane. It is likely that equilibrium between sodium influx as described by JA Sevan ¹¹ and a direct transmission of a deformation to the membrane-anchored complexes and to the cytoskeleton might occur in FMD.

The role of the actin network and of the microtubules in mechanotransduction is now well accepted although the exact pathway involved remains unclear. The use of cytochalasin B to depolymerize and phalloidin to stabilize filamentous actin microfilament and microtubule networks reduced FMD in the rabbit abdominal aorta without affecting acetylcholine-induced dilation ⁴. In addition, the transformation of filamentous into glomerulous (inactive) actin might also play a role in FMD as shear stress activates rapidly this transformation ³¹. Interestingly, in the response of resistance arteries to pressure (myogenic tone), the opposite effect was observed with a rapid formation of filamentous actin after a rise in pressure ³². Microtubule disassembly with nocodazole or colchicine also suppressed FMD in isolated rat gracilis arterioles without affecting the dilation due to acetylcholine or sodium nitroprusside ³³. Thus a selective pathway leading to the production of NO is activated by shear stress in endothelial cells, not by acetylcholine or other endothelium-dependent vasodilator agents. This pathway requires a transmission of the signal from the extracellular matrix to the enzyme eNOS. Several protein complexes are bound to the plasma membrane and usually concentrated in caveolin-rich area, which also link eNOS ³⁴.

Two main complexes have been studied, the dystrophin-dystroglycan complex and the integrins complexes. They have been mainly studied in isolated muscular and endothelial cells respectively. The role of integrins in flow-dependent mechanotransduction has been described in great details in endothelial cells cultured in flow chambers as described in very well documented review articles ^{22, 35–37}. Integrins connects the extracellular matrix to the intracellular machinery. Integrins forming focal adhesion sites activate the focal adhesion kinase FAK and then a cascade of kinases including the MAP kinase pathway leading to cellular remodeling, cytoskeletal rearrangement and elongation of the cell in the direction of flow ³⁵. Although this phenomenon takes hours and involves cells initially cultured without flow these studies provide possible pathways to explore in the understanding of FMD. In cultured isolated blood vessels (aorta and carotid arteries) these pathways have been confirmed and more precisely described in “real” blood vessels^{36, 38}.

In resistance arteries less information is available. A role for integrins in FMD has been shown in rat coronary arterioles. Anti-β₃-integrin antibodies or an antagonist to the matrix-integrin-binding site attenuates FMD ³⁹. This was associated with the inhibition of tyrosine phosphorylation of endothelial cells and with a reduced NO production ³⁹. The same group has also shown that FMD is reduced by the inhibition of the FAK-dependent signaling pathway. This inhibition of FAK prevents flow-induced activation of Akt and eNOS ⁴⁰.

In addition to the β₃ integrin involved in FMD in the rat coronary artery, blockade of the alpha-1 integrin with blocking antibodies severely reduced FMD in mesenteric arteries. Similarly, in mice lacking the alpha-1 integrin, FMD is strongly reduced in resistance arteries, although acetylcholine-induced dilation or contractility to pressure (myogenic tone) or adrenergic stimulation is not affected ⁴¹. Interestingly, activation of the alpha-1 integrin by flow in the mesenteric artery is also associated with Akt activation upstream eNOS phosphorylation ⁴¹. Nevertheless, the role of integrins in FMD remains to be better defined as only 2 vascular beds have been investigated and so far one cannot tell whether β₃ integrins are involved in rat coronary arteries and alpha-1 integrins in mesenteric arteries only or in other resistance arteries. Whether tissue specificity exists and/or whether there is a cross talk between integrins in the different arterial beds is not known.

Besides integrins the dystrophin-dystroglycan complex may also play a role in FMD. Indeed, the lack of a functional dystrophin in mdx mice causes a specific defect in the transduction of shear stress into dilation through the NO-cGMP pathway in endothelial cells being able to dilate normally to other relaxing stimuli⁶. Dystrophin is a main cytoskeletal structure protein ⁴² involved in skeletal and cardiac muscle cells mechanotransduction. Indeed, dystrophin has a key position between membrane structure proteins and the actin cytoskeleton ⁴³. Dystrophin is not only expressed in muscular cells but also in both human and mouse vascular endothelial cells⁶. In mdx mice an incomplete and non-functional dystrophin is produced. In these animals both small and large arteries respond weakly to flow (shear stress) whereas dilation to acetylcholine, which also stimulates NO production by endothelial cells or dilation induced by the NO donor sodium nitroprusside, are normal ⁶. The presence of dystrophin in endothelial cells was initially controversial. Nevertheless, its presence has been demonstrated in endothelial cells in mice, rats and humans ⁶. In addition, it is located at the level of the plasma membrane in both endothelial and smooth muscle cells ⁶. Surprisingly, pressure-induced (myogenic) tone as well as agonist-induced contractions were unaffected although dystrophin is a main structure protein in muscular cells, including smooth muscle cells. The role of dystrophin in FMD was confirmed using mdx mice treated with gentamycin. Gentamycin has the property to induce the recovery of the expression of a full and functional dystrophin in tissues as visualized using antibodies targeting different parts of the protein. In mdx mice treated with gentamycin for 2 weeks FMD was equivalent to control ⁴⁴. Related compounds with a higher efficiency and less toxicity are actually studied in order to find a treatment of some forms of myopathy ⁴⁵. This defect in FMD was also associated with an absence of adaptation of resistance arteries to a chronic rise in blood flow ⁹ (see below, paragraph B-2). Utrophin is structurally close to dystrophin and in many situations can compensate, at least in part for the absence of dystrophin. In human umbilical veins endothelial cells, utrophin forms a complex with beta- and epsilon-sarcoglycan, which associates with caveolin-1 and eNOS ⁴⁶. Nevertheless, the precise role of dystrophin and utrophin in endothelial cells in vivo remains little understood. In endothelial cells FMD has been suggested to involve a pathway different from that of agonist such as acetylcholine or bradykinin ⁴⁷. Thus, dystrophin, which may interact with caveolin-1 and eNOS, might be the first step in a pathway specific for FMD. Nevertheless, no direct link between these proteins has yet been shown in vascular endothelial cells, although in muscular cells the link between caveolin-3 and dystrophin is clear. In dystrophin-related myopathies vascular disorders are not determinant in the evolution of the diseases but they might at least contribute to extent of tissue damage ⁴⁸.

Caveolin-1 is a structure protein with a key position in caveolae and it is associated to many proteins including eNOS. In endothelial cells chronically exposed to shear stress both caveolin and eNOS expression increase and cells become more sensitive to subsequent stimulation by flow ⁴⁹. In addition, FMD is improved in mice lacking caveolin-1, which link eNOS in an inactive state ³⁴. Thus, an attractive hypothesis is that the actors of the response to flow would be concentrated in caveolae, from integrins, dystroglycans and focal adhesion sites on one side to dystrophin, eNOS, HSPs, Akt on the other side. Among the proteins potentially associated to caveolae, PECAM-1 might also play a role in mechanotransduction. In gracilis muscle arterioles isolated from PECAM-1-deficient mice, FMD is strongly reduced ⁵⁰. PECAM-1 is a scaffold protein able to modulate eNOS activity ⁵¹. Indeed, earlier studies of shear stress distribution in endothelial cell have demonstrated that tension is higher at the level of cell–cell junctions ⁵². In addition, in endothelial cells exposed to increasing shear stress, proteins associated with cell–cell junction undergo a rapid reorganization ⁵³. Interestingly, polymorphisms in the gene encoding for PECAM-1 are associated to a higher risk of coronary artery disease⁵⁴.

Vimentin is probably another key element in the chain linking shear stress to vasodilatation although its position in the sequence is not yet known. Indeed, in mice lacking vimentin FMD is selectively reduced in resistance arteries ⁵. In cultured cells, vimentin has been shown to be sensitive to mechanical factor ⁵⁵ including shear stress ⁵⁶. The vimentin network is important in the stabilization of the cell. Indeed, in vimentin knockout mice, which have no obvious phenotype ⁵⁷, regional defects at the cellular level may require an very precise mechanical organization as FMD ^{5, 58} and cell migration ⁵⁹ are reduced in mice lacking vimentin. Thus, interactions between vimentin and other cytoskeletal elements may be critical in the vascular response to shear stress. Nevertheless, a direct observation and measurement of intermediate filaments position during flow has shown that short-term changes in flow do not induce assembly or disassembly of vimentin network segments ⁶⁰. On the other hand, vimentin rearrangement depends on PI3kinase stimulation, which has a key role in FMD as well as in the actin network reorganization. Vimentin has also been shown to regulate focal contact size when stimulated by shear stress ⁶¹. This helps stabilizing cell–matrix adhesions in endothelial cells ⁶¹. More generally, intermediate filaments are involved in the targeting of proteins to specific cellular locations, in the transport and distribution of organelles, and in the regulation of cell–cell and cell–matrix adhesion as recently described in a review article ⁶². We have investigated the role of another main intermediate filament, desmin, in vascular tone. Nevertheless, desmin has no selective effect on FMD as its absence lead to a non-selective decrease in contractility and dilatory capacity of arteries ⁸. Indeed, desmin is most likely involved in the contractility of smooth and skeletal muscle. Defects in the gene encoding for desmin induce rare but severe cardiomyopathies ⁴⁴.

Thus, in the response of arteries to changes in blood flow the role of cytoskeletal or structure proteins is well accepted although it remains to be better defined and it remains many empty boxes. Nevertheless, changes in shear stress are likely activating extracellular elements within the glycocalyx transmitting the signal to membrane-anchored complexes. These latter might be different depending on cell type, species or environment. So far PECAM, integrins and the dystrophin-dystroglycan complex have been identified as the best candidates in this flow-dependent mechanotransduction. Nevertheless, their precise role and their interaction are not yet understood. The signal is then transmitted to eNOS through activation of FAK and PI3kinase-Akt. The location of vimentin in this pathway, although essential, remains to be determined.

2. Role of structure proteins in the adaptation of the microcirculation to chronic changes in blood flow

Considering our initial study showing that FMD was selectively impaired in vimentin-null mice ⁵, we submitted these animals to carotid artery ligation ⁷. Whereas vimentin is essential in FMD (its absence selectively reduced FMD, see above), in response to a chronic increase in blood flow the absence of vimentin might oppose remodeling ⁷. Indeed, intermediate filaments determine, together with actin and microtubules, cells architecture and shape. In addition, Intermediate filaments are generally considered the most rigid elements involved in the overall cell shape ⁷⁹. Thus intermediate filaments, which play a selective role in FMD (vimentin) or in vascular contractility (desmin), are probably regulating the remodeling process following a chronic change in blood flow.

First, using a model of carotid artery ligation we found that the contralateral artery submitted to high flow remodeled excessively in vimentin-null mice ⁷. Similarly, in mice lacking desmin, mesenteric artery remodeling is higher than in control mice with a larger diameter increase ⁸. These findings are in agreement with the concept that the intermediate filaments desmin and vimentin stabilize the arterial wall structure and thus oppose diameter expansion when chronically stimulated by shear stress. This is in accordance with the observation that vimentin content increases after exposition to high flow or pressure, thus strengthening cellular architecture ⁸⁰. In addition, a proteomic analysis performed in mesenteric resistance submitted chronically to high blood flow for one week shows a significant increase in vimentin content⁸¹. Interestingly, in mature rats (8–10 month-old) vimentin content was lower than in young animals and no diameter expansion occurred after a chronicle increase in blood flow ⁸¹. In high flow resistance arteries several heat shock proteins (HSP) expression level increased as well, which is a characteristic of a response to a stress ⁸¹. HSPs may be involved in the stabilization of cytoskeletal element after diameter expansion ⁸². High flow-induced remodeling is also associated with eNOS overexpression and depends on NO production in resistance arteries ^{74, 81}. In addition, in large blood vessels ⁷⁶ and most probably in resistance arteries (preliminary result: ⁸³), high flow-induced remodeling requires a rise in reactive oxygen species (ROS) production. Indeed, in these studies ROS scavenging ⁸³ or NADPH oxidase p47 subunit knockout⁷⁶ prevents remodeling due to high flow. NO and ROS might form peroxynitrites, which would then activate MMPs ⁷⁶. In cultured endothelial cells shear stress activates ROS production ⁸⁴, which stimulates tyrosine kinases ⁸⁵. Furthermore, in the same bovine aortic endothelial cells, pulsatile shear stress increases superoxides formation and upregulates NADPH oxidase (p22phox) gene expression ⁸⁶. Superoxides and NO are also able to activate HSPs production ⁸⁷ and both are produced by endothelial cells in response to shear stress ⁴⁷. Enolase, which has properties similar to those of HSPs, is also increased in mesenteric resistance arteries submitted to high flow in young and mature rats ⁸¹. In young rats diameter and medial mass increased whereas in mature rats medial hypertrophy occurred without increase in diameter⁸¹.

Thus, flow-activated NO and ROS production would preferentially stimulate MMPs and diameter enlargement whereas HSPs would take part in wall hypertrophy. Nevertheless, basal ROS level might determine the type of response obtained after stimulation of endothelial cells by a chronic rise in blood flow. Indeed, in hypertensive rats (SHRs), mesenteric arteries submitted to high flow do not remodel (no increase in diameter) but the treatment of SHRs with the antioxidant tempol restores a significant remodeling ⁸⁸. Thus basal ROS level might determine the type of response obtained after a chronic increase in blood flow in rat mesenteric arteries. Nevertheless, this issue requires further investigations, especially in other diseases such as diabetes or obesity also characterized by high oxidative stress and altered endothelial function. It should be noted that in obese Zucker rats, basal ROS production is higher than in control lean rats and that increasing chronically blood flow in mesenteric arteries remain able to increase arterial diameter. This latter was accompanied by a rise in ROS production in both lean and obese rats⁷³.

By contrast with mice lacking vimentin or desmin, mice lacking a functional dystrophin (mdx mice) do not remodel in response to a chronic increase in flow ⁹ although FMD is reduced in mdx mice, as it is in mice lacking vimentin. Thus, both vimentin and dystrophin are activated by an acute rise in shear stress in a sequence starting outside the cells and being transmitted across the plasma membrane through a pathway that might involve dystrophin first and then intermediate filaments. The activation of this pathway leads to a rapid vasodilatation FMD) not requiring a change in cell architecture. In the vascular response to a chronic rise in blood flow vimentin and desmin oppose the structural changes required for the diameter expansion. This is probably due to their role in the maintenance of cell architecture. On the other hand, dystrophin, located exclusively beneath the plasma membrane, has a role in the transmission of the signal (rise in shear stress) to the cellular machinery (eNOS, NADPH-oxidase) needed for NO and ROS production, essential for high flow-induced diameter increase in resistance arteries.

That no remodeling was seen in mdx mice might be of importance in dystrophin-related myopathies such as Duchenne’s muscular dystrophy. Indeed, in situations such as ischemia, physical effort or wound healing, an inappropriate vascular wall adaptation to an increasing need in blood flow might be deleterious and increase tissue damages. So far it is considered that no important vascular disorder occurs in dystrophin-associated myopathies ⁴⁸. Nevertheless remodeling of the vasculature might be necessary in the process of skeletal and/or cardiac muscle regeneration using different approaches (review: ⁸⁹). In a selective muscular therapy aiming to restore a functional dystrophin in patients the microcirculation would need a proper adaptation with diameter expansion of preexisting vessels in addition of angiogenesis. In this situation arteries lacking this capacity to adapt would reduce the efficacy of such a therapy.

Chronic changes in blood flow may also involve the stimulation of the glycocalyx and the transmission of the signal from the glycocalyx to the F-actin network located beneath the plasma membrane as described above for acute responses (FMD). Stimulation by shear stress is followed by a change in caveolar density, thus rendering the signal more selective for the NO-pathway ⁹⁰. This adaptation takes place between 1 and 3 days after a chronic change in flow. Nevertheless, no study has yet investigated the role of the glycocalyx specifically in high flow-induced remodeling. Similarly no study has yet investigated the role of the transmembrane complexes, especially as regard to the potential role of the integrins-FAK and Src pathway, which is likely to play a role in the process.

Activation of the structure proteins in response to a chronic rise in blood flow might also involve the RhoA/Rho kinase system. A first study has been performed in the model described in figure 1 using a cDNA microarray analysis (14 000 genes were analyzed). This analysis, confirmed by an Rt-PCR analysis showed an overexpression of RhoA and Rho kinase at day 4 after ligature followed a down-regulation at day 16 and 32 ⁹¹. This is consistent with a role of the RhoA-Rho kinase system in high flow-induced remodeling. This is in agreement with previous studies showing that the RhoA/Rho-kinase pathway is involved in many cellular activities besides the cytoskeleton including vesicle trafficking, endocytosis, cellular differentiation and gene transcription (review: ⁹²). Although the RhoA-Rho kinase system has a central role in cell function including pathways activated by shear stress, a treatment with the Rho-kinase inhibitor Fasudil failed to prevent or to reduce high flow-remodeling ⁹¹, although the relatively low dose of fasudil used might not be high enough to block sufficiently the activity of the enzyme. In a model of chronic increase in flow in the pulmonary artery, a 10-fold higher dose of fasudil prevented arterial wall hypertrophy ⁹³. Nevertheless, in this study the rise in flow was much higher than in the ligated mesenteric arteries (4-fold increase versus 60%) and blood pressure probably increased substantially in the pulmonary circulation as the model is based on a carotid shunt⁹³. Thus, the hypertrophy due to a rise in blood flow and pressure most probably involves the RhoA pathway. This is in accordance with previous studies demonstrating that fasudil is efficient against hypertension and the associated hypertrophy. In addition, RhoA is involved in the signaling pathway activated by trophic hormones involved in hypertrophy⁹⁴.

Thus, A chronic increase in blood flow in resistance arteries induces a rise in diameter associated with a medial hypertrophy. So far the pathway involved have been mainly investigated in first order mesenteric arteries. This is a limitation, although the mesenteric circulation has a key role in vascular homeostasis. Nevertheless, in the mesenteric arterial bed the existence of arcading arteries allows ligating alternatively second order arteries and thus to create low and high blood flow areas. In summary, the increase in blood flow induces overexpression of eNOS and NADPH-oxidase, which produce NO and ROS, both important for the activation of MMPs and diameter enlargement. Growth and migration of vascular cells probably results of the stimulation by shear stress of the local renin-angiotensin system and RhoA inducing activation of MAP kinases needed for the process.