Actin polymerization in differentiated vascular smooth muscle cells requires vasodilator-stimulated phosphoprotein

Hak Rim Kim, Philip Graceffa, François Ferron, Cynthia Gallant, Malgorzata Boczkowska, Roberto Dominguez, Kathleen G. Morgan


Our group has previously shown that vasoconstrictors increase net actin polymerization in differentiated vascular smooth muscle cells (dVSMC) and that increased actin polymerization is linked to contractility of vascular tissue (Kim et al., Am J Physiol Cell Physiol 295: C768–778, 2008). However, the underlying mechanisms are largely unknown. Here, we evaluated the possible functions of the Ena/vasodilator-stimulated phosphoprotein (VASP) family of actin filament elongation factors in dVSMC. Inhibition of actin filament elongation by cytochalasin D decreases contractility without changing myosin light-chain phosphorylation levels, suggesting that actin filament elongation is necessary for dVSM contraction. VASP is the only Ena/VASP protein highly expressed in aorta tissues, and VASP knockdown decreased smooth muscle contractility. VASP partially colocalizes with α-actinin and vinculin in dVSMC. Profilin, known to associate with G actin and VASP, also colocalizes with α-actinin and vinculin, potentially identifying the dense bodies and the adhesion plaques as hot spots of actin polymerization. The EVH1 domain of Ena/VASP is known to target these proteins to their sites of action. Introduction of an expressed EVH1 domain as a dominant negative inhibits stimulus-induced increases in actin polymerization. VASP phosphorylation, known to inhibit actin polymerization, is decreased during phenylephrine stimulation in dVSMC. We also directly visualized, for the first time, rhodamine-labeled actin incorporation in dVSMC and identified hot spots of actin polymerization in the cell cortex that colocalize with VASP. These results indicate a role for VASP in actin filament assembly, specifically at the cell cortex, that modulates contractility in dVSMC.

  • Ena/VASP
  • EVH1
  • phosphorylation
  • profilin

the dynamic remodeling of the actin cytoskeleton performs many important cellular functions, not only in migrating, proliferating cells but also in nonmigrating, nonproliferating differentiated vascular and airway smooth muscle cells (20, 28), where net changes in actin polymerization regulate vessel and airway diameter, compliance, and contractile force (16, 17, 29). In vascular smooth muscle, remodeling of the nonmuscle actin cytoskeleton containing γ-actin has been proposed to modulate the contractile force generated by interaction of smooth muscle α-actin and myosin in vascular tissue (28, 41). Although several laboratories have reported stretch- and vasoconstrictor-induced changes in net actin polymerization in differentiated vascular smooth muscle (12, 16, 28, 40), the mechanisms involved are not well understood.

Net increases in polymerized actin can occur via decreased depolymerization or increased actin filament nucleation, branching, and elongation (42). A role for the Arp2/3 complex and N-WASP in actin polymerization in airway smooth muscle tissue has been shown (53), and evidence for calcium-dependent and LIM kinase-mediated regulation of cofilin has been reported (46, 55). N-WASP is known to facilitate the formation of actin filament branches (42); however, the degree to which actin filament elongation is regulated in differentiated vascular smooth muscle cells (dVSMC) is presently unknown.

Ena/vasodilator-stimulated phosphoprotein (VASP) proteins are thought to participate in the assembly of focal adhesions and to promote actin polymerization at these sites (11, 34, 47, 51), as well as at the leading edge of lamellipodia and the tips of filopodia (8, 45, 49). Nothing is known about the subcellular localization of Ena/VASP proteins in dVSMC, but actin filament elongation occurs primarily at barbed ends, which in dVSMC exist at both adhesion plaques (also called dense plaques) and dense bodies (which are the sites of attachment of contractile filament bundles). Thus these sites are candidates for actions of the Ena/VASP family.

VASP was initially isolated from human platelets but is also expressed in a wide variety of other cells and tissues (43). VASP belongs to a family of proline-rich proteins that also include mammalian-enabled protein (Mena) and Ena/VASP-like protein (EVL). Ena/VASP proteins have been proposed to support F-actin assembly within cells, both by acting as anti-capping proteins (3) and by directly mediating processive actin filament elongation (8, 15). Single knockout of VASP alone produced no vascular phenotype; however, compensatory overexpression of other family members was cited as a possible, but untested explanation (1). An Ena/VASP triple null mouse model, however, has been reported to lead to increased endothelial permeability, causing fatal vascular leakage and hemorrhaging during development (18). Furthermore, barrier function and force generation of endothelial cells are enhanced by overexpression of VASP, whereas the loss of Ena/VASP activity causes increased endothelial permeability and decreased force generation (33).

Members of the Ena/VASP family of proteins are composed of three highly conserved domains. The NH2-terminal Ena/VASP-homology domain 1 (EVH1) binds the consensus sequence “FPPPP” of target proteins such as vinculin (9), lamellipodin (32), zyxin (44), migfilin (54), and palladin (7) and plays a critical role in the subcellular localization of Eva/VASP. A central proline-rich domain binds signaling proteins and mediates the recruitment of profilin-actin complexes onto the barbed ends of growing filaments (15). Finally, the COOH-terminal Ena/VASP-homology domain 2 (EVH2) contains both G- and F-actin binding sites, as well as a coiled coil tetramerization region (31).

Here, we report that VASP is the sole member of the Ena/VASP family expressed in the aorta tissues studied, that it is targeted to the vicinity of both adhesion plaques and dense bodies, that it is necessary for α-agonist-induced actin polymerization and contraction, and that such polymerization occurs primarily at “hot spots” in plaques at the cell cortex.


Tissue preparation.

All procedures in this study were performed according to protocols approved by the Institutional Care and Use Committee. Ferrets (Marshall Farms, North Rose, NY) were euthanized by an overdose of isoflurane, and the aorta was quickly removed from animals and placed in an oxygenated physiological salt solution (PSS) consisting of (in mM) 120 NaCl, 5.9 KCl, 1.2 NaH2PO4, 25 NaHCO3, 11.5 dextrose, 1 CaCl2, and 1.4 MgCl2 (pH 7.4). Aortas were dissected under a dissecting microscope in oxygenated (95% O2-5% CO2) PSS to remove connective tissue and the endothelium. Tissues were allowed to equilibrate at 37°C for 1 h in oxygenated PSS and then were stimulated for 10 min with 51 mM KCl-PSS (51 mM of NaCl replaced with KCl in PSS) to test their viability and for normalization. The tissues were washed and allowed to relax for 1 h before we began the experiment. At the end of the experiment, the tissues were quick frozen by immersion into dry ice-acetone slurry containing 10 mM DTT for further study.


Quick-frozen tissues were homogenized at 4°C with homogenization buffer containing 50 mM Tris · HCl, pH 7.4, 5 mM EGTA, 50 mM NaCl, 1.0% Nonidet P-40, 1.0% sodium deoxycholate, 3 mM DTT, 67 μM ZnCl2, 29.6 mM β-glycerophosphate, and protease inhibitors and analyzed via immunoblotting as previously published (38). Densitometry was performed with an Odyssey infrared imaging system (LI-COR Biosciences).

Single cell isolation from aorta.

Cells were freshly enzymatically dissociated from aorta tissue by a previously published method (28). Briefly, aorta tissue was cut into small pieces and placed into 7.5 ml of Ca2+- and Mg2+-free HBSS, ∼100 U/ml type II collagenase (Worthington), ∼1.0 U/ml elastase (Roche Diagnostics), and 0.2% BSA (Sigma). The tissue pieces were filtered on a nylon mesh (pore size ∼0.5 mm) and washed with 12 ml Ca2+- and Mg2+-free HBSS and 0.2% BSA. The wash buffer containing the dissociated cells was poured over glass coverslips on ice under an atmosphere of 100% oxygen. One coverslip of each batch was used to confirm that the cells were capable of shortening. All remaining coverslips were put in a PSS containing 300 mM sucrose, which, as previously described (28), increases the tonicity of the solution and prevents stimulus-induced shortening directly at the cross-bridge level, allowing the use of the coverslips for imaging.

Cell staining and microscopy.

All dVSMCs were fixed with 4% paraformaldehyde and washed with 0.1 mM glycine in 1% BSA HBSS. Cells were then permeabilized with 0.1% Triton X-100 and blocked with 10% goat serum in 0.05% Triton X-100 and 1% BSA HBSS. After cells were blocked, they were incubated with the desired primary antibody diluted in 2% goat serum, 0.05% Triton X-100, and 1% BSA HBSS and then incubated with the appropriate secondary antibody and mounted with FluorSave (Calbiochem). Where appropriate, cells were costained with DNase I (300 nM; Alexa Fluor 488) and phalloidin (1:50 dilution; Alexa Fluor 568). Measurements for phalloidin-to-DNase I ratios followed a previously published method (28). Images for measuring VASP intensities for Fig. 5 were taken with a Kr/Ar laser (BioRad Radiance 2000) scanning confocal microscope equipped with a Nikon X-60 (numerical aperture = 1.4) oil immersion objective. Images were recorded and analyzed with Laser Sharp 2000 (BioRad). For deconvolution microscopy, three-dimensional image stacks were acquired with a Nikon Eclipse TE 2000-E inverted microscope equipped with a Nikon Plan Apochromat 60XA (numerical aperture = 1.4) oil immersion objective. Images were recorded by a high-resolution fluorescence charge-coupled device camera (CoolSNAP HQ2, Photometrics) with NIS-Elements Advanced Research (Nikon) software, and out-of-focus fluorescent blur was removed by deconvolution of Z-stacks (Richardson-Lucy algorithm, constrained iterative-maximum likelihood estimation algorithm). In all cases, no primary antibody controls were performed. In addition, for all colabeling experiments, it was confirmed that there was no detectable cross talk between fluorescent labels by exchanging excitation/emission filters on single labeled coverslips.

Labeled actin preparation.

Actin from rabbit skeletal striated or chicken gizzard smooth muscle was prepared by standard procedures (48) and in some cases was further purified by gel filtration chromatography (37) on Sephacryl S-200. Actin lysine(s) was labeled with a rhodamine fluorophore by reacting with 5- (and 6)-carboxyltetramethylrhodamine, succinimidyl ester (Invitrogen) essentially as described (22). Briefly, actin at ∼4.5 mg/ml was reacted at room temperature for 1 h with a 5.6 molar excess of the rhodamine probe in 50 mM KCl, 50 mM PIPES, pH 6.8, 0.2 mM CaCl2, 0.2 mM ATP, and 0.01% NaN3. The reaction was stopped with 5 mM lysine and dialyzed extensively at 4°C in G buffer (2 mM MOPS, 0.2 mM CaCl2, 0.2 ATP 0.2, and 0.01% NaN3, pH 7.5) and finally in G buffer with MgCl2 substituted for CaCl2. Final molar labeling ratios were between 0.8 and 1.2. Results with the rhodamine-labeled actin depended neither on the actin isoform used nor on the use of the gel-filtration purification step.

Cell permeation and introduction of rhodamine actin.

The methods for cell permeation and introduction of labeled actin used in this study are previously published (35, 50). Freshly isolated cells on coverslips were transferred to pCa 9 PSS (pCa9-PSS) containing (in mM) 137 KCl, 5.4 NaCl, 0.42 NaH2PO4, 0.44 KH2PO4, 4.17 NaHCO3, 10 HEPES, 0.2 MgCl2, 15 EGTA, 0.2 ATP, and 15 phosphocreatine disodium and then were exposed to pCa9-PSS containing 30 μg/ml saponin for 5 min. Rhodamine actin was diluted to 10 μM in pCa9-HBSS and sonicated for 1 min. The rhodamine actin solution was clarified by centrifugation for 20 min. After permeabilization, the cells were exposed to rhodamine-labeled actin for 20 min and stimulated with phenylephrine (10 μM) for 5 min in the presence of creatine phosphokinase (20 U/ml). The cells were fixed for 10 min in cytoskeleton stabilizing buffer (50 mM NaCl, 30 mM sucrose, 10 mM PIPES, 3 mM MgCl2) containing 0.5% glutaraldehyde and 0.2% Triton X-100.

Differential centrifugation for tissue fractionation.

Tissues were quickly homogenized at 4°C with 0.5 ml of buffer 1 (in mM: 20 Tris · HCl, pH 7.5, 50 NaCl, 250 sucrose, 10 DTT, 3 EGTA, 5 MgCl2, 1 ATP, and protease inhibitors) (19). Homogenized tissue was centrifuged at 100,000 g for 1 h, and the supernatant was collected as the cytosolic fraction. The pellet was resuspended and shaken for 1 h at 4°C in 0.5 ml of buffer 2 (20 mM Tris · HCl, pH 7.5, 250 mM sucrose, 0.5% Triton X-100, 10 mM DTT, 3 mM EGTA, 5 mM MgCl2, 1 mM ATP, and protease inhibitors) and centrifuged at 100,000 g for 1 h. This supernatant was collected as the Triton X-100-soluble or membrane fraction. The pellet was resuspended in 0.5 ml of buffer 3 (20 mM Tris · HCl, pH 7.5, 250 mM sucrose, 0.5% Triton X-100, 1.2% SDS, 10 mM DTT, 3 mM EGTA, 5 mM MgCl2, 1 mM ATP, and protease inhibitors), extracted at 4°C for 1 h, and briefly centrifuged. This final supernatant was collected as the Triton X-100-resistant or cytoskeletal fraction. The cytosolic and cytoskeletal fractions were used for immunoblots of VASP.

Expression and purification of EVH1-TAT, EVH1-TAT mutant, and profilin.

The cDNA encoding for human EVL was purchased from American Type Culture Collection (Manassas, VA). The EVH1 domain was amplified by PCR, and the TAT sequence was added at the COOH terminus using the following primers: forward, 5′-GGTGGTCATATGAGTGAACAGAGTATCTGCCAAGCC-3′; reverse 1, 5′-GCTTTTTACGACCGTAACCGCCTCCTTCTTGGGAATTCATGATGTTCAGG-3′; and reverse 2, 5′-AGAGCCCTCGAGACGGCGACGCTGGCGACGCTTTTTACGACCGTAACCGCCTC-3′. The final product was cloned between NdeI and Xho1 sites of vector pTYB1 (New England BioLabs). This vector comprises a chitin-binding domain (for affinity purification) and an intein (for self-cleavage after purification). The mutant Phe78Ser was generated using the QuickChange XL kit (Stratagene). The constructs were expressed in BL21(DE3) cells (Invitrogen). Cells were grown in “Terrific Broth” medium at 37°C until the optical density at 600 nm reached a value of 1.0–1.2. The expression was induced by addition of 0.5 mM isopropylthio-β-d-galactoside and carried out for 5 h at 20°C. Cells were harvested by centrifugation, resuspended in chitin-affinity-column equilibration buffer (20 mM HEPES, pH 7.5, 500 mM NaCl, 1 mM EDTA, and 100 μM PMSF) and lysed with a microfluidizer device (Microfluidics). After standard purification on a chitin affinity column (New England Biolabs), the proteins were further purified through an ion-exchange MonoS 5/50 GL column (GE Healthcare). Purified proteins were concentrated and dialyzed against PBS buffer. The cDNA encoding for human profilin I (also purchased from American Type Culture Collection) was amplified by PCR and inserted into vector pET29/T7 (Novagen). BL21(DE3) competent cells (Invitrogen) were transformed with the construct and grown in LB medium at 37°C until the optical density at 600 nm reached a value of 0.6. Expression was induced by the addition of 1 mM isopropylthio-β-d-galactoside and carried out for 4 h at 37°C. Cells were harvested by centrifugation, resuspended in 10 mM Tris (pH 7.5), 100 mM glycine, 100 mM NaCl, 1 mM DTT, and lysed with a microfluidizer device. The soluble lysate was purified on an affinity poly-l-proline Sepharose column. Profilin was eluted from the column using 30% (vol/vol) DMSO and dialyzed against PBS buffer.

Introduction of antisense morpholino and organ culture.

Morpholino oligonucleotides were purchased from Gene Tools (Philomath, OR). VASP protein expression was suppressed by using the antisense morpholino oligonucleotide 5′-GGAACAGATGACCGTCTCGCTCATG-3′ complementary to the 5′ part of the VASP gene including its initiation codon. The scrambled morpholino oligonucleotide 5′-GACCGTCGGTTCGACACTCGGAATG-3′ served as control. Intact ferret aortic rings were chemically loaded with the morpholino oligonucleotides at 50 μmol/l using a previously described protocol (30). Briefly, muscles were soaked in a series of four solutions at 4°C. The composition (in mM) of the solutions was as follows: for solution 1, 10 EGTA, 5 Na2ATP, 120 KCl, 2 MgCl2, and 20 TES (30 min); for solution 2, 0.1 EGTA, 5 Na2ATP, 120 KCl, 2 MgCl2, and 20 TES plus 50 μmol/l morpholino (180 min); for solution 3, 0.1 EGTA, 5 Na2ATP, 120 KCl, 10 MgCl2, and 20 TES (30 min); and for solution 4, 120 NaCl, 5.8 KCl, 11 dextrose, 25 NaHCO3, 10 MgCl2, and 1.4 NaH2PO4. After solution 4 was added, the Ca2+ concentration was raised to the normal value of 2.5 mM in gradual steps. Tissues were kept overnight in organ culture at room temperature in a 1:1 mixture of PSS and DMEM in the presence of penicillin (25 U/ml), streptomycin (25 mg/ml), and nystatin (50 U/ml), and the medium was changed everyday. On the fourth day of organ culture after loading was completed, the viability of the preparation and contractile function were tested by measuring the response to 51 mM KCl PSS. At the end of the experiment, the muscle strips were then quick-frozen for biochemical analysis.


The following antibodies were used: rabbit polyclonal VASP antibody (1:400), rabbit monoclonal VASP antibody, phospho-VASP (Ser157) antibody, phospho-VASP (Ser239) antibody (all from Cell Signaling), phospho-VASP (Thr278) antibody (from ECM Biosciences), Mena rabbit polyclonal and EVL mouse monoclonal antibodies (gift of F. Gertler, MIT), and rabbit polyclonal profilin antibody (1:400) (from Cytoskeleton). Monoclonal α-smooth muscle actin (clone 1A4; 1:10,000), monoclonal α-tubulin antibody (1:1,000), monoclonal vinculin antibody (1:1,000), monoclonal α-actinin antibody (1:1,000), and monoclonal phospho-myosin light-chain antibody (1:2,000) were all from Sigma. Alexa Fluor 488 goat-anti rabbit IgG, Alexa Fluor 568 goat-anti-rabbit IgG, Alexa Fluor 680 goat anti-mouse IgG, Alexa Fluor 680 goat anti-rabbit IgG (1:2,500) were all from Molecular Probes. IR-Dye 800 goat anti-rabbit IgG and IR-Dye 800 goat anti-mouse IgG were both from Rockland.

Statistical analysis.

All values given in the text are means ± SE. Differences between means were evaluated with a two-tailed Student's t-test. For VASP antisense morpholino experiments, a two-tailed, paired Student's t-test was used. Statistically significant differences were taken at the P < 0.05 levels.


Contractility is decreased by inhibition of actin filament elongation.

In nonmuscle cells, concentrations of cytochalasin D of 500 nM and below are known to specifically inhibit actin filament elongation by capping barbed ends and competing away VASP (3, 13). Thus, as a starting point for determining whether actin filament elongation processes regulate differentiated vascular smooth muscle function, we determined the effect of cytochalasin D in the aorta system. Pretreating aortic rings with 500 nM of cytochalasin D for 1 h significantly inhibited the contractile force induced by a maximally effective (10 μM) concentration of phenylephrine (Fig. 1A).

Fig. 1.

Inhibition of actin barbed-end elongation by cytochalasin D (Cyto-D) decreases contractility in the absence of a change in 20-kDa myosin light-chain (MLC) phosphorylation. A: phenylephrine (PE; 10 μM)-induced force is decreased in Cyto-D (500 nM)-treated vascular smooth muscle tissue. B: immunoblots of phosphorylated MLC (inset) and densitometry of MLC phosphorylation immunoblots normalized to the total MLC levels, with and without Cyto-D treatment (n = 4). C: phalloidin-to-DNase I ratios in freshly isolated differentiated vascular smooth muscle cells (dVSMC) with and without PE and Cyto-D. *P < 0.05 and **P < 0.01 vs. white bar. #P < 0.05 vs. black bar.

The best-studied regulatory mechanism for smooth muscle contractility involves phosphorylation of the 20-kDa myosin light chain (LC20). It has previously been reported that higher concentrations of cytochalasin D (10 μM) can also inhibit contractility in differentiated vascular smooth muscle by decreasing LC20 phosphorylation (52); thus we measured LC20 phosphorylation in our system. As shown in Fig. 1B, under the same conditions in which phenylephrine-induced increases in force were inhibited by cytochalasin D, there was no significant decrease in LC20 phosphorylation. Thus the cytochalasin D-induced inhibition of contractility is not due to alterations in LC20 phosphorylation.

To determine directly whether cytochalasin D treatment inhibits phenylephrine-induced actin polymerization in this system, we utilized a previously described method to assay for changes in F-to-G actin ratios by costaining freshly enzymatically dissociated dVSMCs with fluorescently labeled phalloidin to stain F actin and fluorescently labeled DNase I stain G actin (28). By integrating the total signal from each fluorophore within cellular regions of interest, an index of the F-to-G actin ratio is obtained, even though the actual filaments are below the level of resolution of the microscope. As we have previously reported (28), stimulation with phenylephrine increases phalloidin-to-DNase I ratios significantly compared with that of unstimulated isolated smooth muscle cells (Fig. 1C). However, cytochalasin D pretreatment significantly inhibits phenylephrine-induced increases in the phalloidin-to-DNase I ratios. Cytochalasin D alone causes no significant change in the basal phalloidin-to-DNase I ratio (Fig. 1C).

Expression and distribution of Ena/VASP family members in differentiated vascular smooth muscle.

Ena/VASP proteins are known to function primarily as actin filament elongation factors and may be effectors of the processes inhibited by cytochalasin D. Therefore, we measured the endogenous expression levels of VASP, Mena, and EVL proteins in dVSMC by using specific antibodies in immunoblots of homogenates of aorta and portal vein tissues. Figure 2A shows that VASP is expressed in both aorta and portal vein with greater amounts of VASP detectable in aorta than in portal vein in protein-matched samples. The doublet seen in Fig. 2A is likely due to the known tendency of VASP to show a gel shift when phosphorylated (43). EVL is also expressed in portal vein but was not detectable in aorta samples. Mena (expected molecular mass of 80, 88, and 140 kDa) was not detectable in either aorta or portal vein (Fig. 2B). The fact that only VASP is expressed in aorta simplifies the study of its function in this tissue without the complication of possible functional redundancies from other proteins of the Ena/VASP family.

Fig. 2.

Expression of Ena/vasodilator-stimulated phosphoprotein (VASP) family proteins in intact smooth muscle tissues. A: VASP immunoblot of portal vein (PV) and aorta (A) homogenates. B: costain of immunoblot of PV and A homogenates for mammalian-enabled protein (Mena) and Ena/VASP-like protein (EVL).

VASP knockdown decreases phenylephrine-induced contraction in intact aorta.

To specifically analyze the contribution of VASP to contractility of dVSMC tissue, we performed knockdown experiments for VASP in aortic rings with VASP antisense morpholino oligonucleotides. Scrambled morpholino oligonucleotides were used as a control. After oligonucleotide loading, the tissues were maintained in serum-free organ culture for 4 days. As shown in Fig. 3A, phenylephrine-induced contraction of the antisense-treated tissues is significantly decreased compared with tissues treated with the scrambled oligonucleotides. The same tissue strips used in the contractility assay were quick frozen during steady-state contraction and analyzed for protein expression. Western blot analysis after 4 days of organ culture indicates that VASP protein levels were reduced by ∼35% in the antisense tissues strips compared with those treated with scrambled oligonucleotides (Fig. 3, B and C). In contrast, the expression of metavinculin, vinculin, actin, and tubulin, analyzed in parallel, were not changed by treatment with VASP antisense (Fig. 3B). Interestingly, EVL, but not Mena expression, was increased in VASP antisense-treated aortas but not in control aortas (Fig. 3B). These results indicate that VASP is necessary for normal contractility of blood vessels.

Fig. 3.

VASP knockdown decreases PE-induced contraction in intact aorta tissues. A: active force in response to PE (10 μM), normalized by a contraction to KCl (51 mM) at the beginning of the experiment. B: representative immunoblots of VASP, metavinculin, vinculin, actin, α-tubulin, and EVL after treatment with scrambled or VASP antisense morpholinos. C: quantitative analysis of VASP expression levels, normalized by α-tubulin level (n = 9). *P < 0.05.

VASP partially colocalizes with α-actinin and vinculin in dVSMCs.

To begin to elucidate the mechanism of action of VASP in dVSMC, we determined the subcellular distribution of VASP by staining freshly dissociated aorta cells with an anti-VASP antibody in deconvolution microscopy imaging experiments (Fig. 4). The localization of VASP is characteristically in punctate spots or clusters of spots throughout the cell and along the cell periphery.

Fig. 4.

VASP colocalizes with a subset of α-actinin or vinculin-containing structures. A: α-actinin (green). B: VASP (red). C: merged image. D: vinculin (green). E: VASP (red). F: merged image. White arrows indicate spots where vinculin and VASP colocalize. Scale bar = 10 μm.

To determine whether VASP localizes to adhesion plaques and/or dense bodies in dVSMC, we colabeled VASP and vinculin, a marker of adhesion plaques (39), or α-actinin, a marker of dense bodies and adhesion plaques (2). Deconvolution microscopy indicated that VASP was partially colocalized with a subset of the large number of α-actinin staining spots, some of them in the core of the cell and some of them at the cell edge, indicating that some of the VASP localizes to dense bodies as well as adhesion plaques in dVSMCs (Fig. 4, A–C, white arrows). To confirm the adhesion plaque localization of VASP, we costained for vinculin. The distribution of vinculin, being primarily at the cell edge (Fig. 4D), is distinctly different from that of α-actinin (Fig. 4A). Vinculin and VASP are present within the same general areas of the plasma membrane in plaque-like structures (white arrows). Adhesion plaques are large structures, on the order of a micrometer or more (26), and, as shown in movie S1 (supplemental materials for this article are available online at the Website), vinculin staining spans several optical sections. The colocalization with VASP occurs only in a subset of the area stained in each vinculin patch, as would be expected if VASP were located at the barbed ends of actin filaments, where they intersect with the inner edge of the adhesion plaque.

Responses of VASP after contractile stimulations in dVSMCs.

To investigate the possibility that VASP might undergo a change in targeting upon stimulation, we quantified the content of VASP in the cytosol and the cytoskeletal fraction from differential centrifugation of homogenates of dVSMC, with and without phenylephrine or the phorbol ester 12-deoxyphorbol 13-isobutyrate 20-acetate (DPBA) (Fig. 5A). Stimulation of differentiated vascular smooth muscle tissues with DPBA induces a significant redistribution of VASP from the cytosolic fraction to the cytoskeletal fraction (Triton X-100 insoluble). Stimulation with phenylephrine showed a similar trend, but the difference was not significant (Fig. 5A).

Fig. 5.

Stimulus-induced changes in VASP subcellular distribution. A: densitometric analysis of VASP in cytosolic fractions (gray bars) and Triton X-100-insoluble fractions (cytoskeletal, black bars), normalized as a fraction of the total VASP, from unstimulated or stimulated tissues (n = 5). B: confocal fluorescent microscopy images showing raw intensity of VASP staining in the absence (a) or the presence of stimuli [PE (b) or 12-deoxyphorbol 13-isobutyrate 20-acetate (DPBA; c)]. White boxes represent regions of interest (ROIs) used for quantitative analysis. Scale bar = 10 μm. C: box-and-whisker plots of integrated intensities of VASP staining. The middle line of the box indicates median, the top of the box indicates 75th quartile, the bottom of the box indicates 25th quartile, and whiskers extend to the highest and lowest values (n = 18–27). *P < 0.05; **P < 0.01. ns, Not significant.

We also immunostained cells for VASP in the presence or absence of stimuli to see whether a detectable translocation occurred. No obvious translocation was observed; however, we did consistently notice that stimulation of freshly dissociated aortic cells, with either phenylephrine or DPBA, led to a subtle, but measurable, increase in the absolute intensity of VASP staining with the polyclonal antibody (Cell Signaling) (Fig. 5B). To quantify these results, we integrated the signal intensity for VASP staining within a defined region of interest spanning the cell width using confocal microscopy (Fig. 5C). As shown in Fig. 5C, stimulation of cells with either phenylephrine or DPBA increases the signal intensity of VASP in a statistically significant manner, compared with that of unstimulated cells. The mechanism of change in signal intensity is unknown at this time but could be the result of a conformational change in the molecule that increases the exposure of epitopes to the VASP antibody and that could also have functional consequences for protein-protein interactions.

VASP localizes at newly incorporated actin filaments in dVSMC.

Although changes in F-to-G actin ratios have been previously reported, G-actin incorporation into filaments has never been directly visualized in any differentiated smooth muscle system. To directly visualize G-actin incorporation in dVSMCs and to see where in the cell G-actin incorporation occurs, freshly dissociated aorta cells were permeabilized in pCa9-HBSS containing 30 μg/ml saponin for 2 min and then 10 μM of rhodamine-labeled G actin was introduced by the method of Symons and Mitchison (50). After 2 min, the cells were stimulated with phenylephrine for 5 min and then fixed and costained for VASP or for F actin with phalloidin. As shown in Fig. 6A, rhodamine-labeled smooth muscle G actin (Rh-SM-actin) incorporates predominantly at foci along the cell membrane. The newly formed actin structures are also stained with Alexa 488-labeled phalloidin, indicating that Rh-SM-actin is incorporated into filamentous structures (Fig. 6C). To confirm that these structures truly represent actin incorporation into filaments or whether they might result from Rh-SM-actin aggregation, we repeated the same procedures using Rh-SM-actin samples for which ATP (necessary for polymerization) had been depleted with hexokinase (50); however, no detectable incorporated actin was seen (data not shown). To investigate whether VASP is involved in actin incorporation at the cell cortex, we costained for VASP and found that it is colocalized in these newly formed actin structures (Fig. 6F, white arrows). These data strongly suggest that VASP is actively involved in actin filament polymerization in structures in the cell cortex in dVSMC, presumably associated with adhesion plaques.

Fig. 6.

Incorporated rhodamine-labeled actin (Rh-actin) colocalizes with VASP in dVSMCs. A–C: comparison of incorporated Rh-actin (A; red) with phalloidin (B; green) distribution. C: merged image. D–F: colocalization of incorporated Rh-actin (D; red) with endogenous VASP (E; green). F: merged image. White arrows indicate examples of colocalized spots. Scale bars = 10 μm.

The EVH1 domain inhibits phenylephrine-induced increases in actin polymerization.

The above findings suggest that VASP is involved in actin dynamics in dVSMC. To further test this hypothesis, we also assayed for changes in F-to-G actin ratios. It was not technically feasible to measure F-to-G actin ratios in the morpholino knockdown strips because of the small sample sizes and prolonged protocol of those experiments; therefore, we devised an alternative, dominant-negative approach (Fig. 7A), which could be used at the single cell level, within the lifespan of the freshly dissociated cells (4–6 h). The EVH1 domains of Ena/VASP proteins are necessary for targeting and function of these proteins. They share high sequence identity, particularly within the FPPPP-binding site. Thus we expressed and purified the EVH1 domain of human EVL, which is 62% identical to that of VASP, as a fusion protein with the protein transduction HIV-1 TAT sequence at the COOH-terminal end (Fig. 7A) to allow entry into the intact dVSMC. This EVH1 was selected because of the availability of a high-resolution crystal structure in complex with a target peptide that could be used as a template for mutagenesis design (14). The EVH1-TAT protein was introduced into smooth muscle cells to act as a decoy, or dominant negative, for VASP localization. On the basis of the crystal structure, an EVH1-TAT mutant was designed in which residue Phe78 on the target-binding interface was replaced by Ser, which is predicted to interfere with target binding (Fig. 7A). Indeed, a similar mutation of the EVH1 domain of Homer (Phe to Ala) abolished target binding (4). Thus the EVH1-TAT Phe78Ser mutant was expressed and introduced into cells as a negative control. If our hypothesis that VASP regulates actin dynamics in dVSMC is correct, the wild-type EVH1 domain should prevent phenylephrine-induced actin dynamics but the point mutant should be ineffective.

Fig. 7.

Introduction of a recombinant Ena/VASP-homology domain 1 (EVH1) inhibits stimulus-induced increases of actin polymerization in intact cells. A: structural model and sequence of the EVH1-TAT hybrid construct used in this study. The EVH1 portion of the model (blue) is from the crystal structure of the EVH1 domain of mouse EVL bound to an FPPPP-containing target peptide (14). COOH-terminal TAT structure is from the NMR structure of HIV-1 TAT (Protein Data Bank accession code 1TBC). The mutant construct Phe78Ser, deficient in target binding (4), was also expressed and used in control experiments. B: phalloidin/DNase I assay as an index of actin polymerization. ROIs stained with fluorescent phalloidin and DNase I were integrated in freshly isolated cells exposed to the indicated conditions. **P < 0.01 vs. white bar. #P < 0.05 vs. black bar.

Thus, to determine, in a cause-and-effect manner, whether the introduction of the EVH1 domain of Ena/VASP proteins can inhibit actin polymerization in dVSMC, we measured phalloidin-to-DNase I staining ratios. As shown in Fig. 7B, the addition of phenylephrine significantly increased the phalloidin-to-DNase I ratio. In contrast, the introduction of EVH1 in freshly dissociated dVSMC significantly inhibited subsequent phenylephrine-induced increases of the phalloidin-to-DNase I ratio. However, when the EVH1 Phe78Ser mutant was instead introduced before the application of phenylephrine, there was no significant inhibition of phenylephrine-induced increase in the phalloidin-to-DNase I ratio (Fig. 7B). This result points to an essential role for VASP in α-adrenergic agonist-induced actin polymerization in dVSMCs.

Regulation of phosphorylation of VASP by phenylephrine in dVSMC.

VASP is a known substrate of serine/threonine kinases, and the regulation of its phosphorylation has been shown by others to regulate the function of VASP. VASP has been shown to be phosphorylated at three sites (Ser157, Ser239, and Thr278) by PKA, PKG and AMP kinase (6, 10, 56), and the phosphorylation of VASP has been shown to correlate with the targeting and interactions with actin of VASP (5, 21). Recently, Benz et al. (5) further defined the mechanisms by which the differential phosphorylation of these three sites controls VASP-driven actin filament formation using phosphomimetic VASP mutants.

Here, we determined the regulation of VASP phosphorylation by the α-adrenergic receptor agonist phenylephrine in differentiated vascular smooth muscle. For these experiments, we stimulated tissue strips with phenylephrine (10 μM) at different time points. As shown in Fig. 8A, active force reached steady state by 5 min. By systemic analysis of VASP phosphorylation with phospho-site-specific antibodies, we found that the phosphorylation of VASP at Ser157 and Ser239 was decreased significantly below resting levels at 1 min after stimulation (Fig. 8B). Also, the phosphorylation of VASP at Thr278 was decreased significantly at 2 and 5 min after stimulation (Fig. 8B). These results are consistent with the model, proposed by Harbeck et al. (21) and Benz et al. (5), and further support an involvement of VASP in actin polymerization induced by the α-adrenergic receptor agonist phenylephrine in dVSMCs.

Fig. 8.

Phenylephrine-induced phosphorylation of VASP at serine 157 (Ser157), serine 239 (Ser239), and threonine 278 (Thr278) in dVSMCs. A: time-dependent increases in active force in response to PE (10 μM). B: quantitative analysis of immnuoblots of phosphorylation of VASP (Ser157, Ser239, and Thr278) after PE stimulation, normalized to unstimulated level (n = 4). *P < 0.05.

Profilin distribution in dVSMCs.

Profilin, a G-actin binding protein, is known to play a critical role in the regulation of actin polymerization dynamics by Ena/VASP proteins (25). To determine the subcellular distribution of profilin in dVSMCs as a further test of whether VASP is associated with actin elongation complexes, we stained freshly dissociated aorta cells with an anti-profilin antibody (Fig. 9A). Profilin staining is distributed in punctate spots throughout the cell and along the margin of the cell. Additionally, the patterns of distribution of profilin and G actin are not qualitatively changed by phenylephrine (Fig. 9A). Because profilin is a G-actin binding protein, we compared, in the same cells, the distribution of G actin, stained by Alexa 488-labeled DNase I (Fig. 9A). In contrast to the punctate nature of profilin staining, G actin is, interestingly, distributed broadly throughout the cells. These results suggest that, in dVSMC, the fraction of actin bound to profilin is a small subset of total actin. For this reason, we quantitated actin and profilin in calibrated immunoblots using purified actin and recombinant profilin as standards (see supplemental Fig. S1; supplemental materials for this article are available online at the Website). This approach indicated that, in the aorta, profilin is ∼0.095% and actin is ∼14.7% of total protein. When we take into account that G actin is 19.2% of total actin in this tissue when it is unstimulated (28) and that profilin forms a 1:1 complex with G actin, we find that only ∼10% of total G actin would be bound to profilin in aorta of differentiated vascular smooth muscle. Although higher profiling-to-actin ratios have been reported for other cell types (23), it is of interest that platelets also have low profiling-to-actin ratios (36).

Fig. 9.

Profilin distribution in dVSMC. A: deconvolution fluorescent microscopy of profilin (red) and DNase I (green) in unstimulated or PE-stimulated cells. B: differential centrifugation of aorta homogenates. Immunoblots of profilin in cytosol (CSO), membrane (MEM), and cytoskeleton (CSK) fractions of unstimulated- or PE-stimulated tissue are shown. C: deconvolution microscopy of profilin and α-actinin colabeled dVSMCs: a, α-actinin (green); b, profilin (red); c, merged image. Yellow arrows indicate examples of colocalized spots. Scale bar = 10 μm.

For comparison to the cell staining results, we used differential ultracentrifugation of unstimulated and phenylephrine-stimulated differentiated vascular smooth muscle tissue to separate a cytosolic fraction (proteins soluble in the absence of detergent), a membrane fraction (proteins soluble in 0.5% Triton X-100), and a cytoskeletal fraction (0.5% Triton X-100 insoluble). In both unstimulated and phenylephrine-stimulated tissues, profilin remains in the cytosolic fraction (Fig. 9B). However, as a positive control, other actin-binding proteins, metavinculin and vinculin, can be detected in all three fractions. These results can be reconciled if the punctate spots of profilin staining in cells represent a labile attachment of profilin to punctate cellular structures, likely sites of potential actin polymerization at filament barbed ends.

To determine whether these structures correspond to dense bodies or adhesion plaques, we colabeled cells for profilin and α-actinin (Fig. 9C). Fluorescent deconvolution microscopy indicated that a subset of profilin is colocalized with α-actinin staining, both in the core of the cell and at the cell edges (yellow arrows), indicating that some of the profilin localizes in dense bodies in the core of the dVSMCs and that some localizes in adhesion plaques in the cell cortex. Together with our results on fluorescent G-actin incorporation into structures at the cell periphery (Fig. 6), these findings indicate that only the population of profilin associated with adhesion plaques at the cell edge is actively involved in actin polymerization.


We report here that the actin filament elongation factor VASP is necessary for smooth muscle contraction. Our group (28) and others (20) have reported that agonist-stimulated smooth muscle contraction involves actin filament turnover. However, actin incorporation had not been directly visualized in contractile smooth muscle cells. In this study, we directly visualized actin incorporation in dVSMCs with the use of rhodamine-labeled actin and identified sites of actin polymerization near the cell membrane that colocalize with VASP, suggesting that clusters of actin filaments connecting with adhesion plaques (dense plaques) are the specific site for actin polymerization in dVSMCs. We confirmed the involvement of adhesion plaques by showing the adjacency of VASP to vinculin, an adhesion plaque marker.

Remodeling of the actin cytoskeleton is known to be involved in many cellular events, including cell migration, axon guidance, and pathogen motility (31, 42, 51). Actin cytoskeleton plasticity in dVSMC is only beginning to be investigated, and the mechanisms involved are less well understood [reviewed in Kim et al. (27)]. We have recently reported that γ-actin in dVSMC is specifically involved in filament remodeling (28), whereas the smooth muscle α-actin, which makes up the contractile filament bundles (41), is less dynamic. However, the exact location for actin remodeling has not been previously determined and the mechanism(s) for filament elongation has not been elucidated.

In the present study, we have shown that inhibition of actin filament elongation by cytochalasin D, a pharmacological barbed-end capper (13), significantly decreased the contractile force induced by the α-agonist phenylephrine. The best-studied mechanism of regulation of differentiated vascular smooth muscle contractility is via phosphorylation of myosin LC20 (24), and it is known that phenylephrine increases LC20 phosphorylation levels in this tissue type. In contrast, we observed a net inhibition of actin polymerization but unchanged LC20 phosphorylation during treatment with cytochalasin D. These results indicate that elongation of actin filaments is necessary for efficient smooth muscle contraction.

Obvious candidates in the regulation of actin filament elongation are the Ena/VASP proteins (11, 34, 47, 51). Three Ena/VASP family members have been detected at the message level in blood vessels of mice (1, 18). However, we found that only VASP is expressed as a protein at detectable levels, pointing to a role of VASP in the regulation of actin filament elongation in the aorta tissues used here. This assumption was confirmed by our knockdown of VASP, which showed reduced contractility. Interestingly, in VASP knockdown tissues, EVL expression was increased, which is consistent with previous observations in VASP knockout mice (1). However, a contractility deficit persisted in the present study, indicating either insufficient EVL expression to compensate for the loss of VASP or a different role for EVL compared with VASP in differentiated vascular smooth muscle.

We further tested the function of VASP in actin filament elongation in dVSMCs by using a recombinant EVH1 Ena/VASP domain as a decoy/dominant-negative construct. The EVH1 domain is known to target Ena/VASP proteins to FPPPP-containing targets at its sites of action, including vinculin (9), lamellipodin (32), zyxin (44), migfilin (54), and palladin (7). As a negative control, we compared the effect of the EVH1 mutant Phe78Ser, defective in targeting binding (4). Whereas the mutant EVH1 had no effect on phenylephrine-induced increases in actin polymerization, the native domain significantly inhibited phenylephrine-induced increases in actin polymerization. This result is consistent with a role for VASP in actin polymerization in dVSMCs and further points to Phe78 in the EVH1 domain as a critical residue for target binding.

It is known that VASP has three phosphorylation sites (Ser157, Ser239, and Thr278) and that phosphorylation has been suggested to regulate the function of VASP (21). Ser157 is involved in the targeting of VASP to its possible functional sites, whereas Ser239 and Thr278 are involved in the regulation of actin polymerization driven by VASP (5). As shown in Fig. 8, the decreased phosphorylation of VASP after phenylephrine stimulation is consistent with this model. At the present time, the mechanism by which VASP phosphorylation is decreased after phenylephrine stimulation is unknown. It could be due to either decreases in kinase activity or increases in phosphatase activity.

The concentration of actin monomers in cells is typically higher than the critical concentration for spontaneous actin polymerization, and it has been suggested that G-actin binding proteins such as profilin or thymosin-β4 regulate the pool of cellular G actin (42). Profilin, in particular, promotes the exchange of ADP for ATP on actin (23). When bound to profilin, actin monomers are preferentially added onto the barbed ends of actin filaments (15). The pattern of profilin distribution in dVSMCs was somewhat unexpected; profilin was distributed in dot-like structures, whereas G actin, stained by labeled DNase I, was distributed homogenously through the cell. However, our results indicate that only ∼10% of the total G-actin pool is bound by profilin in dVSMC. This result indicates that other G-actin-binding proteins, possibly thymosin-β4, contribute to maintaining the reservoir of actin monomers in these cells and that the fraction of G actin bound to profilin is found at sites of potential active filament assembly such as dense plaques.

In summary, the results presented here indicate that the elongation of actin filaments, mediated by VASP in dVSMCs, is necessary for normal vascular contractility and that sites of active actin polymerization in dVSMCs are most active at foci in the cell cortex.


This work was supported by National Heart, Lung, and Blood Institute Grant P01 HL-86655.


No conflicts of interest are declared by the authors.


We thank Dr. Frank B. Gertler (MIT, Cambridge, MA) for the generous gift of Mena and EVL antibodies.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
  31. 31.
  32. 32.
  33. 33.
  34. 34.
  35. 35.
  36. 36.
  37. 37.
  38. 38.
  39. 39.
  40. 40.
  41. 41.
  42. 42.
  43. 43.
  44. 44.
  45. 45.
  46. 46.
  47. 47.
  48. 48.
  49. 49.
  50. 50.
  51. 51.
  52. 52.
  53. 53.
  54. 54.
  55. 55.
  56. 56.
View Abstract