Research ArticleCell Biology

Mechanism of Stretch-Induced Activation of the Mechanotransducer Zyxin in Vascular Cells

+ See all authors and affiliations

Sci. Signal.  11 Dec 2012:
Vol. 5, Issue 254, pp. ra91
DOI: 10.1126/scisignal.2003173


Vascular cells respond to supraphysiological amounts of stretch with a characteristic phenotypic change that results in dysfunctional remodeling of the affected arteries. Although the pathophysiological consequences of stretch-induced signaling are well characterized, the mechanism of mechanotransduction is unclear. We focused on the mechanotransducer zyxin, which translocates to the nucleus to drive gene expression in response to stretch. In cultured human endothelial cells and perfused femoral arteries isolated from wild-type and several knockout mouse strains, we characterized a multistep signaling pathway whereby stretch led to a transient receptor potential channel 3–mediated release of the endothelial vasoconstrictor peptide endothelin-1 (ET-1). ET-1, through autocrine activation of its B-type receptor, elicited the release of pro–atrial natriuretic peptide (ANP), which caused the autocrine activation of the ANP receptor guanylyl cyclase A (GC-A). Activation of GC-A, in turn, led to protein kinase G–mediated phosphorylation of zyxin at serine 142, thereby triggering the translocation of zyxin to the nucleus, where it was required for stretch-induced gene expression. Thus, we have identified a stretch-induced signaling pathway in vascular cells that leads to the activation of zyxin, a cytoskeletal protein specifically involved in transducing mechanical stimuli.


Many cells in the human body experience mechanically active environments. These cells not only sense and adapt to these forces, but they also may require these forces to develop an appropriate phenotype (1). Among the biomechanical forces in the human body, circumferential wall tension or wall stress is the strongest and also the most common. This wall stress is proportional to the product of transmural pressure (for example, blood pressure) and the extension of the organ (for example, the radius of a blood vessel). The resulting force experienced by cells under higher amounts of wall stress is stretch: The higher the stress, the more the cells are stretched. Smooth muscle cells in the bladder are stretched more than once per day; lung epithelial cells are stretched with every breath that we take, and in emphysema, they are chronically overstretched with severe consequences. Arterial, endothelial, and smooth muscle cells are stretched as a result of blood pressure, and they experience a lasting mechanical overload during hypertension (2), again with potentially detrimental consequences for the cellular phenotype and vascular function. Although the physiological and, in the case of supraphysiological amounts of force, pathological effects of stretch-induced phenotype regulation are well known in the cardiovascular system, the specific signal transduction mechanisms at the cellular level are unclear.

We and others have characterized the focal adhesion–associated Lin-11 Isl-1 Mec-3 domain protein zyxin, which is also involved in organizing actin polymerization (3, 4), as a signaling molecule that is specifically activated in response to stretch. In stretched vascular cells, zyxin translocates from focal adhesions to stress fibers and accumulates in the nucleus (57). There, zyxin acts as a transcription factor to mediate stretch-induced changes in the expression of genes whose products are involved in vascular remodeling and disease, such as vascular cell adhesion molecule–1 (VCAM-1), the chemokines interleukin-8 (IL-8) and monocyte chemoattractant protein 1, and the transcription factor hairy/enhancer-of-split related with YRPW motif 1 (Hey-1) in endothelial cells, as well as thrombomodulin and calponin in vascular smooth muscle cells (5, 6). Similar observations were made in cardiomyocytes (8).

As mentioned earlier, the signaling events specifically triggered by constantly increased amounts of stretch are not well characterized. However, endothelial vasoconstrictor peptide endothelin-1 (ET-1), cardiac vasoactive peptide atrial natriuretic peptide (ANP), and transient receptor potential channel C6 (TRPC6) have been characterized as being specifically activated in response to acute changes in wall stress in arteries or the heart. ET-1 is released from endothelial cells almost exclusively in response to stretch, and it causes vasodilation through endothelial B-type ET-1 receptors (ETBRs) (9) and, more dominant in most arteries, a compensatory vasoconstriction through the activation of A-type receptors on smooth muscle cells (9). TRPC6 is a cation channel that is abundant in smooth muscle cells, and it is thought to be crucial for the stretch-induced myogenic vasoconstrictor response that is responsible for maintaining constant perfusion of the kidney and the brain (10, 11). Finally, ANP is a peptide that regulates total fluid volume in the body. This peptide is released from the atria of the heart into the blood in response to supraphysiological filling, that is, stretch of atrial cardiomyocytes. ANP then acts through its A- and B-type guanylyl cyclase receptors (GC-A and GC-B, respectively), cyclic guanosine 3′,5′-monophosphate (cGMP), and cGMP-dependent protein kinase (PKG) on endothelial cells and distal tubule cells of the kidney to reduce blood volume (12, 13). In experiments with zyxin, the only specific vascular mechanotransducer characterized to date, and starting with the stretch-sensitive molecules ET-1, ANP, and the TRPC subfamily as potential mediators of vascular mechanotransduction, we sought to characterize the specific signaling response to stretch in endothelial cells. Understanding this response may provide a new perspective to interfere with stretch-induced vascular remodeling in the early phases of the process.


Stretch-induced zyxin translocation in vascular cells in situ and in vitro

We subjected freshly isolated mouse femoral arteries to perfusion with oxygen-saturated physiological salt solution at pressure gradients of 80 to 60 mmHg at a flow rate of 40 μl/min (low stretch) or 150 to 90 mmHg at a flow rate of 180 μl/min (high stretch). In response to this increased perfusion pressure, zyxin translocated to the nucleus in native arterial endothelial cells (Fig. 1A). Nuclear translocation of zyxin also occurred in native vascular smooth muscle cells, albeit only in very few cells and at a higher perfusion pressure (200 to 120 mmHg; Fig. 1B). A similar observation was made in cultured primary human umbilical vein endothelial cells (Fig. 1C) upon exposure to cyclic stretch (10% elongation at 0.5 Hz), a well-established cell culture model to mimic changes in perfusion pressure or stretch in vitro (5, 6, 14).

Fig. 1

Stretch-induced translocation of zyxin in native and cultured vascular cells. (A) Cross sections of perfused segments of the femoral artery of wild-type (WT, upper panels) and zyxin-deficient (Zyx-KO, lower panels) mice stained for zyxin (Cy3, red) and nuclei (DAPI, blue). Green channel shows the autofluorescence of elastic vessel fibers. Scale bars, 50 μm. Bar graph shows the statistical analysis of zyxin-positive nuclei in C57Bl6 (Bl6) and C57Bl6/129SvJ intercrosses (129S). *P < 0.05 versus static control; n = 9 C57Bl6 mice and n = 6 C57Bl6/129SvJ mice. Data represent means ± SEM. In vessels from zyxin-deficient animals, virtually no Cy3 staining was observed. (B) Enlarged pictures of representative smooth muscle cells in segments perfused at 80/40 mmHg (left) and 200/120 mmHg (right, after 6 hours). Arrowheads depict zyxin-positive regions at the cell borders (left) and in the nucleus (right). Scale bars, 10 μm. (C) Representative confocal immunofluorescence analysis of zyxin localization in static (left) and stretched (right) endothelial cells (10%, 0.5 Hz, for 6 hours). Zyxin (Cy3, red) and the focal adhesion protein paxillin (Cy2, green) were stained together with a nuclear counterstain (DAPI, blue). Scale bars, 50 μm. Data in (B) and (C) are representative of three to six experiments.

Components of the stretch-induced endothelial signaling cascade

For a first analysis of the signaling events that lead to zyxin activation, we tested several inhibitors and agonists that interact with mechanosensitive signal transduction pathways in endothelial cells, smooth muscle cells, or cardiomyocytes by monitoring the nuclear localization of zyxin and zyxin-dependent changes in expression of the gene encoding IL-8 in vitro (Fig. 2). Stretch-induced expression of this gene is strictly zyxin-dependent (6). In some experiments, two other stretch- and zyxin-dependent gene products (VCAM-1 and the transcription factor Hey-1; fig. S1A) were also quantified to ensure that monitoring changes in IL-8 mRNA abundance was a proper readout for stretch-induced gene expression. We also analyzed stretch-induced changes in Hey-1 mRNA abundance in human endothelial cells derived from the aorta and the coronary microcirculation to ensure that experimental findings obtained with venous endothelial cells could be extrapolated to endothelial cells in general (fig. S1B).

Fig. 2

Inhibition of stretch-induced translocation of zyxin in endothelial cells. (A) Left graph: Statistical summary of zyxin localization in stretched endothelial cells (10%, 0.5 Hz, for 6 hours) treated with 100 μM Gd3+ (TRP inhibitor), 1 μM BQ788 (ET-1 antagonist), or 100 μM Rp8 (protein kinase G inhibitor). All three agents were applied 1 hour before the start of the stretch protocol. Nuclei were counted in three independent experiments, with 200 nuclei counted per experiment in two replicates. *P < 0.05 versus static control, #P < 0.05 versus stretched. Right graphs: Analysis of stretch-induced IL-8 mRNA and protein abundance. Endothelial cells were exposed to cyclic stretch as described earlier, and IL-8 mRNA abundance was analyzed by real-time RT-PCR. IL-8 protein release was determined by enzyme-linked immunosorbent assay (ELISA) measurement of conditioned media (10 nM ET-1 for 6 hours and 1 nM ANP for 6 hours). *P < 0.05 versus static control, #P < 0.05 versus stretched; n = 5 experiments. (B) Exemplary immunofluorescence analysis of endothelial cells treated with ET-1 and ANP as described earlier. Zyxin is in red (Cy3), paxillin is in green (Cy2), and nuclei are in blue (DAPI). Scale bars, 50 μm. Data are representative of six experiments. (C) Summary of the statistical analysis of nuclear zyxin translocation in response to ANP and ET-1. *P < 0.05 versus static control, #P < 0.05 versus ET-1– or ANP-treated cells; n = 6 experiments. (D) Real-time RT-PCR analysis of ET-1– or ANP-induced IL-8 abundance. Left: *P < 0.05 versus static control, #P < 0.05 versus ET-1– or ANP-treated; n = 5 experiments. Right: ELISA analysis. *P < 0.05 versus static control; n = 6 experiments.

The general TRP channel inhibitor gadolinium (Gd3+, 100 μM) (15), the ETBR antagonist BQ788 (1 μM), and the PKG inhibitor 8-(4-chlorophenylthio)-guanosine 3′,5′-cyclic monophosphorothioate (Rp8, 100 μM) all prevented both translocation of zyxin to the nucleus (Fig. 2A, left graph) and zyxin-dependent changes in IL-8 mRNA and protein abundance (Fig. 2A, right graphs). That zyxin mediated stretch-induced changes in IL-8 mRNA abundance was confirmed in experiments in which cells were transfected with a small interfering RNA (siRNA) specific for zyxin 72 hours before the cells were subjected to stretch. The efficiency of zyxin knockdown was confirmed by immunofluorescence and Western blotting analysis (fig. S2). Because inhibition of nitric oxide synthesis, another stimulus for PKG activation in vascular cells, had no inhibitory effect on IL-8 mRNA abundance (fig. S1C), we hypothesized that an endogenous (thus, an endothelium-derived) natriuretic peptide acting through GC-A or GC-B was responsible for the stretch-induced activation of PKG in an autocrine manner. The cultured endothelial cells in fact revealed that stretch induced an increase in the abundance of only ANP mRNA among natriuretic peptides (fig. S3). Therefore, we tested the effects of exogenous ANP (1 nM) and ET-1 (10 nM) on zyxin activation. Both peptides mimicked the effects of cyclic stretch on endothelial cells in an Rp8-inhibitable manner, whereas BQ788 counteracted the effects of ET-1, but not those of ANP (Fig. 2, B and C). Moreover, both ET-1 and ANP elicited a similar increase in IL-8 mRNA abundance to that induced by stretch (Fig. 2D).

To further understand how TRP channels, ET-1, and ANP were involved in zyxin activation, we performed complementation experiments for zyxin translocation (Fig. 3A) and zyxin-induced IL-8 mRNA expression (Fig. 3B), and we also performed immunofluorescence analysis (fig. S4), which revealed that the effect of TRP channel inhibition could be overcome by the addition of exogenous ET-1 or ANP. This finding was confirmed at the protein level (Fig. 3C). Furthermore, the BQ788-mediated blockade of stretch-induced release of proANP, and thus zyxin activation, was bypassed with exogenous ANP (Fig. 3B). Finally, neither ET-1 nor ANP was able to surmount the Rp8-mediated blockade of PKG, the effector kinase downstream of GC-A– or GC-B–catalyzed formation of cGMP (Fig. 3B). These results suggest that endothelial cells have a hierarchical mechanotransduction cascade in which TRP channel activation led to ET-1 release, which induced ANP release and which was followed by concomitant activation of PKG. This hypothesis was further substantiated by the finding that exogenous ET-1 alone caused the release of proANP from endothelial cells and that both ET-1– and stretch-induced proANP release were blocked by BQ788 and Gd3+ (Fig. 3D). To show that zyxin translocated to the nucleus, we performed Western blotting analyses that verified that ANP-, ET-1–, and stretch-induced zyxin translocation was PKG-dependent (Fig. 3E, left panel) and that the blockade of zyxin translocation by Gd3+ was circumvented by ANP (Fig. 3E, right panel, and, for exemplary Western blotting results, see fig. S5).

Fig. 3

TRP channels, ET-1, and ANP are hierarchically involved in the mechanism of stretch-induced zyxin activation. (A) Summary of the statistical analysis of zyxin localization in cultured endothelial cells pretreated for 1 hour with 100 μM Gd3+, 1 μM BQ788, 100 μM Rp8, 10 nM ET-1, 1 nM ANP, or the indicated combinations before application of the stretch protocol. *P < 0.05 versus static control, #P < 0.05 versus stretched; n = 6 experiments. The immunofluorescence analysis was performed as described in Fig. 1. (B) Real-time RT-PCR analysis of IL-8 mRNA abundance in response to stretch and combinations of Gd3+, BQ788, Rp8, ET-1, or ANP as indicated. *P < 0.05 versus static control, #P < 0.05 versus stretched, P < 0.05 versus Gd3+- and BQ788-treated cells; n = 5 experiments. (C) ELISA analysis of IL-8 protein release from stretched endothelial cells. *P < 0.05 versus static control, #P < 0.05 versus stretched; n = 7 experiments. (D) Release of (left) ET-1 and (right) proANP (as measured by ELISA) from endothelial cells that were stretched, treated with Gd3+ or BQ788, or were stretched followed by treatment with ET-1, or were treated with ET-1 and BQ788. *P < 0.05 versus static control, #P < 0.05 versus stretched or ET-1–treated; n = 6 experiments. (E) Statistical analysis of three independent Western blotting analyses of the nuclear translocation of zyxin in response to stretch, ANP, or ET-1 with or without pretreatment with Rp8 or Gd3+. *P < 0.05 versus static control, #P < 0.05 versus stimulated control (stretch, ET-1, and ANP).

Dependence of zyxin activation on the presence of TRPC3

Although the effects of Gd3+ pointed to the involvement of TRP channels in zyxin translocation, it was unclear whether a single or several TRP channels acting together contributed to zyxin activation and whether the potential candidates were stretch-sensitive. We found that mRNAs encoding TRPC3 through TRPC7, as well as TRPV4, were constitutively expressed in cultured endothelial cells (fig. S6). Therefore, we systematically analyzed the effect of stretch on zyxin translocation in isolated perfused femoral arteries from various TRPC knockout mice. In contrast to femoral arteries from mice triply deficient in TRPC1, TRPC4, and TRPC5 (16, 17), those from mice doubly deficient in TRPC3 and TRPC6 (18, 19) did not show any stretch-inducible nuclear translocation of zyxin (fig. S7), thereby narrowing the candidates down to TRPC3 and TRPC6. Further analysis of femoral arteries derived from mice deficient in either TRPC3 or TRPC6 revealed that native endothelial cells from TRPC3-deficient mice, but not TRPC6-deficient mice, were defective in pressure-induced nuclear translocation of zyxin (Fig. 4A). Moreover, in contrast to aortic smooth muscle cells derived from wild-type mice, cells derived from TRPC3-deficient mice exhibited neither stretch-induced translocation of zyxin to the nucleus (Fig. 4, B and C) nor expression of smooth muscle cell–specific stretch- and zyxin-sensitive genes, such as mRNAs encoding calponin or thrombomodulin (Fig. 4D). These findings were further corroborated by a complete inhibition of stretch-induced IL-8 gene expression in cultured human endothelial cells by the specific TRPC3 inhibitor Pyr3 (10 μM; fig. S8A) (20).

Fig. 4

TRPC3 mediates stretch-induced zyxin activation. (A) Confocal immunofluorescence analysis of cross sections of perfused (with 80/60 or 150/90 mmHg, respectively) femoral artery segments derived from TRPC3 or TRPC6 single-knockout mice stained for zyxin (Cy3, red) and nuclei (DAPI, blue). The green channel depicts the autofluorescence of elastic vessel fibers. (B) Confocal immunofluorescence analysis of zyxin (Cy3, red), α-smooth muscle actin (Cy2, green), and nuclei (DAPI, blue) in stretched (12%, 0.5 Hz, for 6 hours) aortic smooth muscle cells derived from TRPC3-deficient mice (lower images) and littermate WT controls (upper images). For a statistical summary, three independent experiments with smooth muscle cells derived from mice deficient in TRPC1, TRPC4, and TRPC5 (C1,4,5); TRPC3 and TRPC6 (C3,6); TRPC3 (C3); or TRPC6 (C6) were performed and are presented in the bar graph. One hundred randomly selected nuclei were counted from each experiment. *P < 0.05 versus static WT control, #P < 0.05 versus stretched WT; n = 6 experiments. Scale bars, 50 μm. (C) Summary of the statistical analysis of three independent Western blotting analyses of nuclear zyxin translocation in smooth muscle cells derived from WT control or TRPC3-deficient mice in response to stretch (18% elongation for 6 hours at 0.5 Hz), ANP (1 nM), and the TRPC3 inhibitor Pyr3 (10 μM). *P < 0.05 versus static control. (D) Real-time RT-PCR analysis of (left) smooth muscle cell calponin and (right) thrombomodulin mRNA abundances in response to stretch (12%, 0.5 Hz, for 6 hours) in WT cells, cells deficient in both TRPC3 and TRPC6 (C3,6), or cells deficient in either TRPC3 or TRPC6 (C3 and C6, respectively). *P < 0.05 versus static WT control; n = 5 experiments.

Stimulation of zyxin phosphorylation by GC-A through PKG

In a similar approach, we tested whether GC-A was involved in stretch-induced zyxin activation in experiments with mice constitutively deficient in GC-A (21) and mice with an endothelial cell–specific deficiency in GC-A (22). Perfusion of isolated femoral arteries and stretch experiments with cultured aortic smooth muscle cells revealed that GC-A was crucial for the nuclear translocation of zyxin in the native endothelial cells (Fig. 5A) and cultured smooth muscle cells (Fig. 5B). This effect was apparent even though the nuclei of GC-A–deficient cultured smooth muscle cells exhibited a slightly increased basal amount of zyxin compared to those of wild-type mice. Moreover, stretch-sensitive expression of the genes encoding calponin and thrombomodulin was blunted in GC-A–deficient smooth muscle cells and in smooth muscle cells derived from zyxin-deficient mice (Fig. 5C). Analysis of the abundances of mRNAs of the genes encoding GC-A and GC-B by real-time reverse transcription polymerase chain reaction (RT-PCR) assays revealed that GC-A mRNA (1600 ± 190 mRNA copies per nanogram of total RNA, estimated as 40 copies per cell) was more abundant than GC-B mRNA (75 ± 54 copies per nanogram of total RNA, estimated as less than 2 copies per cell), which practically excludes GC-B from playing a role in endothelial cell mechanotransduction.

Fig. 5

The A-type ANP receptor GC-A is critically involved in stretch-induced zyxin activation. (A) Confocal immunofluorescence analyses of cross sections of perfused (with 80/40 or 150/90 mmHg, respectively) femoral artery segments derived from GC-A–deficient mice and WT littermate controls. Staining was performed as described in Fig. 4. Scale bars, 50 μm. (B) Confocal immunofluorescence analysis of zyxin (Cy3, red), α-actin (Cy2, green), and nuclei (DAPI, blue) in stretched (12%, 0.5 Hz, for 6 hours) aortic smooth muscle cells derived from GC-A–deficient mice and WT littermate controls. For a statistical summary, 100 randomly selected nuclei from each of the independent experiments were counted. *P < 0.05 versus static control, #P < 0.05 versus stretched WT; n = 5 experiments. Scale bars, 50 μm. (C) Real-time RT-PCR analysis of calponin (Cal) and thrombomodulin (TM) mRNA abundances in smooth muscle cells derived from age-matched WT or GC-A–deficient animals (GC-A−/−, left panel) or from zyxin-deficient mice (Zyx−/−, right panel). *P < 0.05 versus static controls; n = 4 experiments.

Next, we investigated the hypothesis that zyxin was directly phosphorylated by PKG to be released from focal adhesions. We found that the PKG isoforms 1α and 1β were present in endothelial cells (fig. S8B). Two-dimensional (2D) gel electrophoresis of endothelial cell lysates and nuclear extracts revealed a stretch-induced shift in the isoelectric point of immunoreactive zyxin (from 7.0 to ~6.8), which was lost after exposure to the broad-range protein phosphatase 1 (PP1) (Fig. 6A). Moreover, zyxin was phosphorylated and copurified with nuclear proteins only in extracts of stretched cells, and pretreatment with the PKG inhibitor Rp8 resulted in the inhibition of both the phosphorylation and nuclear translocation of zyxin (Fig. 6, B and C). In that effect, ANP alone also caused a comparable PKG-dependent nuclear accumulation of zyxin (Fig. 6D).

Fig. 6

Stretch-induced nuclear translocation of zyxin requires its phosphorylation at serine 142 by PKG. (A to D) Representative 2D gel electrophoresis and Western blotting analyses of (A and B) total cell lysates and (C and D) nuclear extracts from endothelial cells after stretch exposure (10%, 0.5 Hz, for 6 hours). (A) Zyxin is phosphorylated in response to stretch. PP1 indicates exposure to PP1 after stretch. (B) Phosphorylation of zyxin in response to stretch is PKG-dependent. (C) PKG-mediated phosphorylation of zyxin is necessary for its nuclear accumulation. (D) Treatment with 1 nM ANP alone for 6 hours caused Rp8-sensitive (100 μM) zyxin phosphorylation. (E) Representative confocal immunofluorescence analyses and statistical summary (bar graph) of endothelial cells transfected with plasmids encoding eGFP-tagged expression constructs with a full-length WT zyxin sequence or sequences in which Ser142 was mutated to Gly142 (S142G) or to Glu142 (S142E; green fluorescence). Total zyxin (Cy3, red) was stained together with nuclei (DAPI, blue) to distinguish transfected from nontransfected cells. A minimum of 50 transfected cells were counted for each condition in five independent experiments. *P < 0.05 versus static WT control, #P < 0.05 versus stretched WT; n = 5 experiments. Scale bars, 50 μm.

To identify the amino acid residue in zyxin that was phosphorylated by PKG, we generated enhanced green fluorescent protein (eGFP)–tagged zyxin expression constructs with the wild-type sequence and sequences mutated at potential PKG phosphorylation sites, namely, Ser142, Ser344, and Thr352. These residues were predicted by NetPhosK 1.0 (at (23). Only mutation of Ser142 to glycine resulted in a translocation-incompetent zyxin isoform stably located in focal adhesions (Fig. 6E). Moreover, conversion of Ser142 to glutamate partially mimicked zyxin phosphorylation, which resulted in a distinct stretch-independent accumulation of this mutant protein in the nucleus (Fig. 6E). As expected, exposure of transfected endothelial cells to ANP led to an identical translocation pattern of the mutant proteins as was caused by cyclic stretch (fig. S9).

Stretch-induced zyxin activation in other cell types

Finally, to test whether the signaling cascade that we described was specific for vascular cells or was common to other cell types that experience enhanced amounts of stretch in vivo, we analyzed the human lung epithelial cell line A549 and primary smooth muscle cells isolated from the bladders of wild-type and zyxin knockout mice (fig. S10). Both cell types responded to cyclic stretch with zyxin translocation to the nucleus and stretch-induced gene expression. At least in case of smooth muscle cells from the bladder, stretch-induced gene expression was zyxin-dependent, suggesting that this response was stimulus-specific and conserved in different types of mechanosensitive cells.


Zyxin is a protein with dual functions. Attached to the cytoskeleton, the protein is part of the machinery that organizes the function, dynamics, and maintenance of actin fibers (3, 4). However, in vascular cells, zyxin translocates from the cytoskeleton to the nucleus and initiates gene expression in response to mechanical overload (5, 6). Here, we identified the mechanism by which zyxin becomes capable of entering the nucleus of endothelial cells and other mechanosensitive cells. This finding may open a new perspective on vascular mechanotransduction in physiological and pathophysiological settings.

In vivo, two relevant hemodynamic forces act on endothelial cells. Whereas circumferential wall stress basically is the product of vessel radius and blood pressure, fluid shear stress can be derived from the flow rate and the viscosity of the fluid, but it is inversely proportional to the third power of the vessel radius. The question arising from this concept is whether the two models used in this work can specifically address wall stress and whether fluid shear stress may be disregarded as a factor for zyxin activation. When comparing perfusion of femoral artery segments with different pressure gradients in our model, the perfused vessel will be distended, resulting in a parallel increase in flow with pressure (up to four- to fivefold). However, regarding fluid shear stress, this increase in flow will be fully compensated by the concomitant increase in vessel diameter (160 to 180% of the nonperfused state); thus, accounting for the third power dependency of radius, fluid shear stress is stable or even decreased at high perfusion pressure. In the case of cultured endothelial cells, there is no flow. Thus, whereas stretch is exactly adjustable (5, 6, 14), fluid shear stress is absent. A potential drawback for this mechanically well-defined model is the fact that because of the lack of fluid shear stress, endothelial cells rapidly change their phenotype even after one passage in culture. Therefore, we used endothelial cells from umbilical cords because these cells are available in sufficient amounts to perform experiments without any passaging. However, because the umbilical vein is an embryonic vessel, endothelial cells from this source need to be carefully characterized regarding their response to cyclic stretch. Therefore, we also performed exemplary experiments with arterial endothelial cells [human coronary and aortic endothelial cells (6)]. These experiments showed that our cultured human umbilical vein endothelial cells responded to stretch similarly to human arterial endothelial cells.

We showed that TRPC3 was crucial for stretch-induced zyxin activation in vascular cells and thus was a prime candidate for being part of the elusive stretch-sensing complex in vascular cells. This result was not anticipated because it seemed attractive to think of TRPC channels as a family of somewhat redundant molecules that might compensate for each other to a certain degree (24). Moreover, TRPC3 plays some role in hearing (24) and the myogenic response (25), and therefore, it did not seem to be very specific at a functional level. However, reports showing that TRPC3 is involved in the development (26) and promotion (27, 28) of hypertension support our finding that TRPC3 is central in the response to stretch in vascular cells. In addition to TRPC-3 and zyxin, ET-1 and ANP play a central role in the endothelial response to stretch. Both peptides are released under conditions of mechanical overload, but in a rather different context. Although stretch-induced release of ET-1 from endothelial cells contributes to vasoconstriction (29, 30), its newly characterized autocrine role as a mechanotransducing factor is surprising. In view of the pattern of zyxin-induced gene products (6), however, this finding may well explain observations that ET-1 acts as a mitogen in arteries (31, 32). The systemic function of ANP, which was not previously recognized to be released from endothelial cells in appreciable amounts, is principally stretch-related because the peptide counteracts cardiac (blood) volume overload by promoting salt and water secretion by the kidney. Other endothelial autacoids released in response to stretch include angiotensin II and bradykinin, which have been implicated in stretch-induced signaling in endothelial cells (33) and other cells (34). At this stage, we cannot exclude that these or other factors may alternatively activate zyxin in response to mechanical overload. However, although the pathway that we describe may be considered somewhat more redundant in the future, the central principle of a stretch-induced activation of the mechanotransducer protein zyxin will still be a specific response of vascular cells to mechanical overload. In this context, kinases other than PKG may be able to phosphorylate zyxin at Ser142 in endothelial cells (35). Among these, adenosine 5′-monophosphate–activated kinase may be a promising candidate because it can be activated by PKG in another cell model (36). Again, although such sidetracks may exist, Ser142 is a classical substrate of PKG.

Although given the complex pattern of gene products controlled by zyxin in response to stretch (6), zyxin may be the most prominent target of future research, and the characterization of other stretch-induced targets of PKG is of considerable interest. Among these may be TRPC3, because PKG inhibits the activity of this channel in other cell types (37). In the case of TRPC3, this mechanism may constitute a mode of feedback inhibition to prevent prolonged or exaggerated activation of the endothelial response to stretch.

Finally, the question remains as to how the signaling pathway that we described relates to other factors that are activated by cyclic stretch in vascular cells. A prototypic example is the transcription factor activating protein 1 (AP1) (38, 39); however, AP1 also mediates inflammatory processes induced by a multitude of stimuli in various cell types and thus seems to be a general stress response protein rather than a protein specifically induced by stretch. Therefore, we believe that apart from several stress-induced signaling pathways that may or may not have pathophysiological importance, the stretch-induced signaling cascade that we described is specific and pivotal for early vascular remodeling processes in response to mechanical overload.

Depending on vessel size, the magnitude of wall stress under normotensive conditions is ~4 to 6 N/cm2, but easily rises to 10 N/cm2 during hypertension. This increase impairs the ability of the blood vessel to locally control blood flow, and it triggers a compensatory hypertrophic or hyperplastic remodeling response to regain this ability. The downside of this initially meaningful response, however, is a permanent change in the phenotype of both endothelial and smooth muscle cells. Although much is known about the later stages of hypertension-induced arterial remodeling, which are characterized by chronic activation of multiple unspecific growth-promoting and proinflammatory pathways (40), the initial signaling events elicited by the rise in wall stress and their impact on the subsequent remodeling process are still ill-defined. With the stretch-induced, TRPC3–ET-1–ANP–mediated activation of zyxin, we have described a specific response of vascular cells to this biomechanical stimulus.

In our cell culture and in situ perfusion models, endothelial cells were by far more sensitive to stretch than were smooth muscle cells. This holds true for the threshold of the response as well as for the amount of zyxin that translocated to the nucleus. This observation may reflect the fact that in hypertension as well as in atherosclerosis, another vascular disease strongly promoted by mechanical overload, endothelial dysfunction typically precedes the maladaptive response of the smooth muscle cells in the affected arteries. The complex set of genes and pathways regulated by zyxin (6) is highly suggestive of a remodeling process associated with mechanical overload. Thus, zyxin may be a gatekeeper at the crossroads of maintenance and remodeling of vascular structure and function. In this regard, it is noteworthy that the translocation of zyxin to the nucleus in perfused arteries is observed only when the extent of wall stress is comparable to that observed in hypertensive individuals.

Not only is wall stress an important molding factor in the development of several organs, such as the lung or the urogenital tract, but it also plays a critical role in maintaining their function in the adult organism. In the case of chronically enhanced wall stress, (mal)adaptive remodeling processes may also ensue in these organs (41). ANP and ET-1 are in fact capable of regulating gene expression in epithelial cells (42) and nonvascular smooth muscle cells (43). Accordingly, the organ with the most pronounced dynamics in wall stress, the lung, contains the largest amount of zyxin (4). Although the mechanotransduction cascade that we described may differ in some molecular details, our data suggest that stretch induces the nuclear translocation and transcriptional activity of zyxin in nonvascular cells too. Therefore, zyxin appears to act as a general mechanotransducer that is not restricted to vascular cells but specifically responds to increased stretch.

Collectively, our data define an unsuspected, but highly specific, stretch-induced signaling cascade that is likely to exist in cells sensitive to this biomechanical force (Fig. 7). Because zyxin-induced gene expression seems to be a hallmark of early stretch-induced changes in the phenotype of endothelial cells (6), further characterization of this signaling pathway, especially of previously uncharacterized PKG target proteins, may be relevant for several stretch-induced maladaptive remodeling processes such as emphysema in the lung or hypertension-induced arterial stiffening.

Fig. 7

Stretch-induced activation of zyxin in endothelial cells. Inhibitors and knockout animals used in this work are depicted in blue. Stretch, presumably indirectly through an increase in diacylglycerol, causes activation of TRPC3, with the subsequent influx of extracellular calcium enhancing the release of (preformed) ET-1. ET-1, by binding to the ETB receptor, increases the intracellular calcium concentration beyond the threshold required to trigger ANP release. Through binding to GC-A, ANP evokes an increase in the activity of PKG, which in turn phosphorylates zyxin at Ser142, thereby causing its translocation to the nucleus. Zyxin, through interacting with a specific PyPu box enhancer element, alters the expression of various mechanosensitive genes.

Materials and Methods


Unless noted otherwise, all reagents and peptides were from Sigma.

Vessel perfusion and immunohistochemistry

All animal studies were performed with permission from the Regional Council of Karlsruhe and conformed to the Guide for the Care and Use of Laboratory Animals (National Institutes of Health publication no. 85-23, revised 1996). TRPC3 and TRPC6 double-knockout mice were obtained by crossing TRPC3−/− mice (18) with TRPC6−/− mice (19). TRPC1,4,5 triple-knockout mice were generated with TRPC1−/− mice (16), TRPC4−/− mice (17), and mice in which exon 4 of Trpc5 was deleted (P. Weißgerber, S. Stolz, V. Flockerzi, and M. Freichel; Department of Experimental Pharmacology, University of Homburg/Saar, Germany). All TRPC-deficient mice were on a mixed C57Bl6/129SvJ genetic background. Wild-type C57Bl6 mice served as controls for zyxin-deficient and GC-A–deficient mice. First-generation offspring of C57Bl6 and 129SvJ intercrosses served as controls for the TRP-deficient mice. All mice [with the indicated gene defect(s) zyxin knockout (4); the two GC-A knockouts (21, 22); TRPC1,4,5 triple-knockout; and TRPC3,6 double- or single-knockout] were euthanized in CO2 chambers. For the isolation of femoral arteries, the hindlimb was excised, and the femoral artery, which was submerged under perfusion buffer [119.0 mM NaCl, 1.25 mM CaCl2, 4.70 mM KCl, 1.17 mM MgSO4, 2.10 mM NaHCO3, 1.18 mM KH2PO4; partial pressure of oxygen (Po2; 160 mmHg)/partial pressure of CO2 (Pco2; 37 mmHg) that was constantly maintained at pH 7.4], was separated from the accompanying vein and dissected from connective tissue. Segments of the femoral artery (0.5 to 1 cm, 100 to 120 μm in diameter) were cut and mounted onto glass capillaries (diameter, 120 μm) fitted for the use in the Pressure Myograph System 110P (Danish Myo Technology). After equilibration (at 10 to 30 mmHg for 1 hour), femoral arteries were subjected to defined perfusion conditions: 37°C, flow of 20 to 230 μl/min, depending on perfusion pressures varying from 20 to 200 mmHg. The vessel chamber was continuously refilled with prewarmed and equilibrated perfusion buffer. The use of a pressure transducer together with the MyoVIEW system and software enabled continuous control of temperature, pressure, and vessel diameter. For all mouse experiments, experimental and analytical procedures were performed in a blinded fashion regarding the genotype of the mice.


Immunohistochemistry was performed as described previously (44). Briefly, vessel segments were fixed in zinc fixative for 24 hours, which was followed by a standard dehydration process and paraffin embedding at 60°C overnight. For staining, paraffin sections (5 μm) were deparaffinized, permeabilized, and blocked in 0.1% Triton X-100 in Hanks’ balanced salt solution (HBSS) for 10 min and in 0.25% casein, 0.1% bovine serum albumin in 50 mM tris (pH 7.6) for 1 hour, incubated with the indicated primary antibodies (1:100 dilution) at 4°C for 12 hours, and stained by incubation with Cy2- or Cy3-labeled secondary antibodies (diluted 1:100 in blocking buffer; Jackson Laboratories) for 2 hours. Finally, samples were counterstained with 4′,6-diamidino-2-phenylindole (DAPI; 1 μg/ml) for 10 min and mounted in ProLong (Invitrogen). For confocal microscopy analysis, an IX81 microscope equipped with an IX-DSU disk unit and the MT20 multiwavelength illumination system was used in combination with the Cell^R Software package (Olympus). For a demonstration of the specificity of zyxin staining, vessels from zyxin-deficient mice were stained (see Fig. 1).

Isolation and growth of human umbilical vein endothelial cells

Endothelial cells were isolated from fresh human umbilical cord veins as described previously (45) and were directly cultured on BioFlex collagen I elastomers (Flexcell International Corporation). All procedures were approved by the local ethics committee. Typically, >98% cells stained positive for von Willebrand factor (vWF), whereas α-smooth muscle actin was not detectable. Upon reaching 80% confluence, endothelial cells were used for experiments under static or cyclic stretch conditions (elongation of 10% at 0.5 Hz with a sinusoidal profile) in a Flexcell FX-5000 Tension System (30). A549 lung epithelial cells were cultured under standard conditions as recommended by the American Type Culture Collection. Human coronary and aortic endothelial cells were obtained from PromoCell and were cultured as described earlier for umbilical vein endothelial cells.

Isolation and growth of mouse cells

Aortae isolated from wild-type and knockout mice were cut into small rings (~1 mm) and rinsed in HBSS until all of the blood was removed. The tissue was treated overnight at 37°C in 5% CO2 in 1.4 ml of Dulbecco’s modified Eagle’s medium (DMEM) containing smooth muscle cell growth medium 2 (1:1; PromoCell) supplemented with 5% fetal bovine serum (FBS) and containing 250 μl of collagenase solution (1%; Sigma). The resulting cell suspension was centrifuged for 5 min at 200g. The pellet was resuspended in 2 ml of smooth muscle cell growth medium [a 1:1 mixture of DMEM and smooth muscle cell growth medium containing 5% FBS, penicillin (50 U/ml), streptomycin (50 μg/ml), and fungizone (0.25 μg/ml)] and was seeded into a 6-cm culture dish. After the first passage, smooth muscle cells were propagated with culture medium (DMEM with 15% FBS and antibiotics as defined earlier). Typically, cells were >90% positive for α-smooth muscle actin and were essentially negative for the endothelial cell marker vWF. Bladder smooth muscle cells were isolated from the medial layer after mechanically dissecting the urothelium and outer connective tissue layer. The remaining tissue was cut into small pieces (~1 mm) and essentially treated as described for the aorta. Upon reaching about 80% confluency, primary cells were directly transferred onto BioFlex membranes for experiments. About 70% of all cells were positive for α-smooth muscle actin.

Gene expression analysis

Isolation of RNA and real-time RT-PCR analysis were performed as described previously (6) with a LightCycler 1.2 (Roche). For a list of all synthetic nucleic acids, see table S1. The annealing temperature for all primers was 58°C, and elongation was for 30 s. The amount of total RNA per reaction and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) mRNA abundance were used as internal loading standards. Copy numbers of mRNA were defined as copies per nanogram of total RNA. For some gene products, relative changes in mRNA abundance were monitored by classical RT-PCR, as described previously (30). Briefly, reactions were stopped in the exponential phase and subjected to agarose electrophoresis, and band intensities were analyzed with a Gel Doc XR unit and Quantity One software package version 4.06 (Bio-Rad) and normalized to the respective intensities of bands of GAPDH mRNA. In these cases, mRNA abundance was relatively quantified as percentage of control.

Analysis of protein abundance and localization

Western blotting analysis by 1D SDS–polyacrylamide gel electrophoresis (SDS-PAGE) was performed according to standard procedures, with 7 to 15 μg of protein loaded in each lane, as described previously (6). The primary antibodies used were rabbit antibody against zyxin (B71, provided by M. Beckerle, Huntsman Cancer Institute, Salt Lake City, UT), and monoclonal mouse antibody against β-actin (Sigma) was used to control for loading and transfer. The secondary antibodies used were against mouse and rabbit immunoglobulins and were peroxidase-coupled (Sigma). ECL Plus reagent (GE Healthcare) was used for the visualization of protein bands. Band intensities were analyzed with a ChemiDoc XRS unit and Quantity One software and were normalized to those corresponding to β-actin.

Preparation of nuclei and nuclear extracts

To enrich the nuclear protein fraction from cultured cells for 2D gel electrophoresis analysis, we used the ReadyPrep Protein Extraction Kit (Bio-Rad) according to the manufacturer’s protocol. The nuclear pellet was solubilized in 2D gel rehydration buffer [7 M urea, 2 M thiourea, 4% (w/v) CHAPS, 0.2% ampholytes (pH 3 to 10; Bio-Rad), 15 mM dithiothreitol (DTT), 0.01% bromphenol blue] and used for 2D gel electrophoresis. For Western blotting analysis, the pellet and cytosolic fractions were directly solubilized with loading buffer and subjected to SDS-PAGE as described earlier.

Two-dimensional gel electrophoresis

Analysis of zyxin phosphorylation was performed by 2D gel electrophoresis according to O’Farrell with minor modifications (45). Protein load for the first dimension [60.000 V×h (pH 3 to 10) at 20°C in a Protean IEF Cell apparatus (Bio-Rad)] was 150 μg. For the second dimension, individual strips were equilibrated in 2% DTT and subjected to 10% SDS-PAGE. Western blotting analysis of these gels was performed as described earlier with a Trans-Blot Cell (Bio-Rad).

Enzyme-linked immunosorbent assay

ELISA kits were used for quantitative determination of human IL-8 (R&D Systems), ET-1 (R&D Systems), and proANP (Biomedica) in cell culture supernatants according to the manufacturers’ protocols.

Confocal immunofluorescence analysis of cultured cells

Confocal immunofluorescence analysis was performed as described previously (6, 44). Briefly, after fixation, permeabilization, and blocking, BioFlex membranes were incubated cell-side down with 15 μl of primary antibodies diluted in blocking buffer for 2 hours at room temperature. The antibodies used were zyxin rabbit antiserum B71 at a 1:250 dilution and paxillin monoclonal mouse antibody (BD Transduction Laboratories) at a 1:100 dilution. Thereafter, membranes were incubated cell-side down with either or both Cy2 (green)– and Cy3 (red)–conjugated secondary antibodies for 1 hour at ambient temperature. The secondary antibodies used were donkey-derived anti-mouse and anti-rabbit antibodies (Jackson Laboratories). Finally, to provide a nuclear counterstain, we incubated the membranes with DAPI (blue; 1 μg/ml) for 10 min at room temperature. Membranes were then mounted onto a large coverslip with ProLong (Invitrogen). Confocal microscopy was performed as outlined earlier.

Plasmid cloning, mutagenesis, and cell transfections

Zyxin expression plasmids were constructed by subcloning a full-length PCR fragment including the first stop codon (positions 143 to 1895) derived from endothelial complementary DNA (cDNA) into the cDNA 6.2/N-EmGFP TOPO 5.9-kb vector, with the TOPO cloning reaction used according to the manufacturer’s recommendations (TOPO Mammalian Expression Vector Kit, Invitrogen). Site-directed mutagenesis was performed to generate glycine and alanine (S142G/S344A/T352A) or glutamate and aspartate (S142E/S344E/T352D) mutants with the QuickChange II site-directed mutagenesis kit (Stratagene), according to the manufacturer’s protocol, with the mutagenesis primers listed in table S1. All clones were sequenced and shown to have no unexpected mutations. Transient transfection of endothelial cells with the plasmids encoding wild-type or mutant eGFP-zyxin constructs was performed with polyethylenimine (PEI). For each well of cells to be transfected, plasmid (2 μg) was incubated with PEI in Opti-MEM I (Invitrogen) at 0.32 g/liter in a final volume of 200 μl at ambient temperature for 30 min and then was layered dropwise onto the endothelial cells and incubated for 6 hours in medium without antibiotics and FBS at 37°C and in 5% CO2. After termination of the procedure by adding normal endothelial cell growth medium, transfection efficiency was found to be 35%. Twenty-four hours after transfection, endothelial cells were transferred to a BioFlex membrane for stretch experiments. The amount of transfected cells on the BioFlex plates typically dropped to about 10%.

Transfection of endothelial cells with siRNA

Briefly, for each well of a six-well plate, endothelial cells were transfected with 3 μg of zyxin-specific siRNA (Qiagen). The Hs_ZYX_1_HP validated target sequence is 5′-AAGGTGAGCAGTATTGATTTG-3′. Transfections were performed by magnet-assisted transfection (IBA) as described previously (6). Gene silencing was optimal 72 hours after transfection (fig. S2).

Statistical analysis

All quantitative data are presented as means ± SEM of the indicated number of independent experiments with cells or samples obtained from individual umbilical cords or femoral arteries and aortae from different animals, respectively. Repeated-measure analysis of variance followed by a Tukey-Kramer post hoc test was performed as appropriate with the InStat software package version 3.06 (GraphPad Software), and P < 0.05 was considered to be statistically significant.

Supplementary Materials

Fig. S1. Stretch-induced, zyxin-dependent gene expression.

Fig. S2. Technical controls and standards.

Fig. S3. Expression of natriuretic peptides by human umbilical vein endothelial cells.

Fig. S4. Representative immunofluorescence analyses of samples used for quantitative representations.

Fig. S5. Western blotting analysis of stretch-induced zyxin translocation from focal adhesions (cytosolic fraction) to the nucleus (nuclear fraction) in cells after subcellular fractionation.

Fig. S6. Expression of TRP channels in endothelial cells.

Fig. S7. TRPC3 or TRPC6 mediates stretch-induced zyxin activation.

Fig. S8. Characterization of the effect of TRPC3 inhibition on stretch-induced responses.

Fig. S9. Phosphorylation of the stretch-sensitive residue Ser142 is necessary for ANP-induced nuclear translocation of zyxin.

Fig. S10. Wall tension–induced zyxin signaling in lung epithelial and bladder smooth muscle cells.

Table S1. Synthetic oligonucleotides used for cloning and for real-time RT-PCR analysis.

References and Notes

Acknowledgments: This work is dedicated to M. Cattaruzza, a respected colleague, supervisor, and friend, who unexpectedly passed away before its completion. We thank D. Heide and R. Cattaruzza for expert technical assistance and L. Hoffman and M. Beckerle for providing zyxin-specific antibodies and zyxin-deficient mice. Funding: This work was supported in part by the German Research Foundation (grant CA 262/1-3 to M.C.), the European Commission (Marie Curie Initial Training Network SmArt to M.H.), and the Intramural Research Program of the NIH (Z01-ES101684 to L.B.). Author contributions: S.S.B., A.W., and M.C. performed and analyzed the data; M.F. and L.B. provided knockout mice, designed experiments, and wrote the manuscript; and M.C. and M.H. designed experiments and wrote the manuscript. Competing interests: The authors declare that they have no competing interests.
View Abstract

Related Content

Navigate This Article