Modulation of lamellipodial structure and dynamics by NO-dependent phosphorylation of VASP Ser239

Lamellipodial dynamics and behaviour are regulated by proteins of the Ena/VASP family, which are key controllers of actin polymerization in diverse cellular processes (Bear et al., 2000; Coppolino et al., 2001; Jay, 2000; Krause et al., 2000; Laurent et al., 1999; Lawrence and Pryzwansky, 2001; Renfranz and Beckerle, 2002; Vasioukhin and Fuchs, 2001). Lamellipodial protrusion rate is directly correlated with the amount of Ena/VASP proteins within these projections (Rottner et al., 1999). Paradoxically, Ena/VASP proteins appear to retard cell translocation, principally by promoting increased lamellipodial protrusion and turnover (Bear et al., 2000; Bear et al., 2002). The molecular basis of these effects reflects the ability of Ena/VASP proteins to act as anti-capping factors, promoting the barbed-end growth of actin filaments (Barzik et al., 2005). Additionally, these proteins also decrease the branching architecture of actin filaments, but increase actin fibre bundling (Bachmann et al., 1999; Huttelmaier et al., 1999; Laurent et al., 1999).

Cellular regulation of Ena/VASP protein function remains mysterious. These proteins are substrates for both cAMP- and cGMP-dependent protein kinases (PKA and PKG, respectively) (Butt et al., 1994). VASP has three PKA/PKG phosphorylation sites (Krause et al., 2003): Ser157, Ser239 and Thr278 (residue numbers apply to human VASP). Ser157 is the preferred site of phosphorylation by PKA (Butt et al., 1994; Lambrechts et al., 2000) and is conserved within all vertebrate Ena/VASP proteins. Ser239, adjacent to the G-actin-binding site, is preferentially phosphorylated by PKG (Butt et al., 1994). Thr278 is phosphorylated by AMP-activated protein kinase (Blume et al., 2007). The cellular effects of VASP-phosphorylation have remained obscure. Most studies have focused on in vitro phosphorylation of Ena/VASP proteins and how this affects ligand binding and actin polymerization. Ser157 phosphorylation seems to correlate with activation of Ena/VASP proteins, although studies using different technologies have come to opposite conclusions about the effect of PKA on F-actin binding and bundling activity (Harbeck et al., 2000; Laurent et al., 1999). PKA phosphorylation correlates with filopodia formation in neurons (Lebrand et al., 2004) and phosphorylation sites are crucial to Ena/VASP function in cell migration (Loureiro et al., 2002). Analysis of the role of Ser239 phosphorylation within the EVH2 domain suggests that at least in vitro, phosphorylation at this site reduces VASP's actin anti-capping and filament bundling activity (Barzik et al., 2005) and interferes with actin fibre formation (Zhuang et al., 2004). The specific effects of phosphorylation at this site on lamellipodial dynamics remain unknown. Since nitric oxide (NO) has been shown to trigger phosphorylation of Ser239 of VASP (Oelze et al., 2000), we set out to explore the effects of NO-dependent phosphorylation of VASP on its function in human primary renal tubule epithelial cells (PTECs) and embryonic keratinocytes (HEKs).

We examined the effects of NO-dependent phosphorylation of VASP on its function in human primary renal tubule epithelial cells (PTECs). These cells are motile epithelial cells that make abundant NO following cytokine treatment (Glynne et al., 2001). These cells express VASP, which is particularly concentrated in the leading edge of lamellipodia and also in focal adhesions (Fig. 1a), as previously described (Reinhard et al., 1992). To study the dynamic behaviour of VASP within the lamellipodia, we constructed an expression plasmid of VASP fused to the C-terminus of GFP [wild-type (WT) GFP-VASP] and used this to transfect human PTECs. We also constructed a mutant VASP containing a point mutation of serine at position 239 to alanine (S239A) – Ser239 being the major site for cGMP-dependent kinase mediated phosphorylation – and also fused this to GFP in the same manner (S239A GFP-VASP). Both GFP-VASP constructs showed identical localization to that seen in untransfected cells (Fig. 1b,c). There was no difference in the subcellular distribution of WT GFP-VASP compared with S239A GFP-VASP (Fig. 1b,c). Expression levels of the transfected WT and S239A GFP-VASP constructs were virtually identical, as judged both by fluorescence imaging and immunoblotting (Fig. 1d). Control imaging of cells transfected with GFP alone (Fig. 1e) showed homogenous cytoplasmic staining and cells transfected with empty vector showed no fluorescent staining (Fig. 1e), confirming that the observed distribution of GFP-VASP was specific.

Live cell imaging allowed us to follow the dynamic changes in cell behaviour and VASP distribution within lamellipodia. Localized concentration of VASP at the edge of the cell correlated with projection of that part of the membrane (Fig. 2a and supplementary material Movie 1). The movie also shows the concentration of VASP within membrane ruffles, which appear to move back from the leading edge of the membrane. Some lamellipodia were a persistent broad front at the leading edge of the cell (Fig. 2a); others were more transient and were extended in directions other than the principal axis of movement (Fig. 2b). Analysis of time-lapse sequences and kymography (Fig. 2c,d) were used to derive the rates of cell movement and lamellipodial protrusion (Fig. 2i), with median values of 0.37 and 0.24 μm min^–1, respectively. Similar rates of cell movement are reported for other cells, e.g. 0.5 μm min^–1 for human epidermal keratinocytes (Hinz et al., 1999), although lamellipodial protrusion velocity in these and other cells was rather higher.

In human PTECs transfected with S239A GFP-VASP, there was no difference in the distribution of VASP in resting cells (Fig. 2e-h and supplementary material Movie 2). Quantification of rates of cell movement and lamellipodial protrusion showed no significant differences between cells transfected with WT or S239A GFP-VASP (Fig. 2i).

Next, we examined the behaviour of the WT and mutant GFP-VASP construct following addition of an NO donor that we expected to phosphorylate WT VASP at the Ser239 site. In resting cells, Ser239 phosphorylation of VASP was undetectable (Fig. 3a), although about 30% of the protein was phosphorylated at Ser157, as judged by the lesser mobility (50 kDa compared with 46 kDa) on SDS PAGE produced by phosphorylation at this site (Butt et al., 1994) and quantified by digital scanning of gel images (Fig. 3b). Following addition of NO donor, there was a rapid phosphorylation of VASP at Ser239, which was readily detectable between 5 and 10 minutes after NO donor addition (Fig. 3a). Nonspecific binding to immunoglobulin is indicated by an asterix above the band of phosphorylated VASP. Since optimal separation from this band did not give very clear separation of the 50 kDa and 46 kDa forms of VASP, we assayed the effects of NO donors on phosphorylation of Ser157 VASP on separate gels. NO donors did not produce any significant difference to Ser157 phosphorylation in these cells, as judged by the lack of any significant change in the proportion of the 46 kDa and 50 kDa forms of VASP (Fig. 3b).

Following addition of NO donor to these cells, we observed profound changes in cell shape and the distribution of WT VASP within lamellipodia (Fig. 3b and supplementary material Movie 3). First, the NO donor produced a dramatic removal of WT VASP from the cell edge, such that after 30 minutes treatment virtually no VASP remained at this site (supplementary material Movie 3 and Fig. 3b). Secondly, NO produced a withdrawal of the edge of the cell, retraction of all cellular processes and cell rounding (supplementary material Movie 3 and Fig. 3b). A kymograph of this retraction is shown in Fig. 3d, which shows that the cell edge retracted with a velocity of 0.99 μm min^–1. In cells transfected with S239A GFP-VASP, the NO donor was without effect, with no significant removal of VASP from the lamellipodial edge (Fig. 3c and supplementary material Movie 4). We quantified these changes, and showed that the amount of WT VASP localized to the lamellipodial edge declined significantly following NO addition (Fig. 3e, P<0.0001, ANOVA), with a half-life of 11.57 minutes (Fig. 3e, insert). In cells transfected with S239A GFP-VASP, there was no significant decline in the amount of VASP localized to the lamellipodial edge (Fig. 3f).

Human PTECs do not express constitutive NO synthases, but do express inducible NO synthase (iNOS) following exposure to pro-inflammatory cytokines (Glynne et al., 2001). In order to test the effect of endogenous NO production within these cells, we stimulated human PTECs transfected with either WT GFP-VASP or S239A GFP-VASP for 16 hours with cytokines and then measured the percentage of lamellipodial area occupied by VASP (Fig. 3g). Following this stimulation, very little WT GFP-VASP was detected in lamellipodia but, by contrast, the median amount of S239A GFP-VASP within lamellipodia remained at a significantly higher level (Fig. 3g, P=0.0022, Mann-Whitney test). Thus, endogenous NO production has an effect on VASP similar to that of an NO donor, and this effect was abrogated in the S239A GFP-VASP mutant.

NO donors in these cells do not produce detectable cytotoxicity in this time frame (Glynne et al., 2001) (data not shown), although they do induce cell shedding after 24 hours (Glynne et al., 2001). Additionally, the lack of effect of NO donor on cells transfected with S239 GFP-VASP stresses the crucial importance of Ser239 phosphorylation as the basis for the observed changes.

The removal of VASP from lamallipodia following treatment with NO donor suggested this should produce alterations in cell movement, because most cells move by extension of the leading edge lamellipodium. We measured cell speeds in human PTECs transfected with WT or S239 GFP-VASP. Following addition of NO donor, there was a ∼30% reduction in average cell speed in cells transfected with WT GFP-VASP, a significant reduction (Fig. 3h). By contrast, in cells transfected with S239A GFP-VASP, there was only a small, not statistically significant, reduction in cell speed following NO addition (Fig. 3h). Thus, loss of VASP from the leading edge of the cell after NO addition leads to a reduction in cell speed.

The retraction of the cell edge following treatment with NO donor led to considerable changes in cell shape which we quantified by measuring changes in total cell area. In our time-lapse imaging we also recorded the phase-contrast view of each cell at the same time as the signal from the WT GFP-VASP (Fig. 4a). This showed the edge of the cell (arrow heads) correlating with the presence of the WT GFP-VASP. In cells expressing WT GFP-VASP, 30 minutes after NO addition, the total cell area was reduced to just under 60% of the value observed immediately before NO addition (Fig. 4b), a significant reduction (P<0.01, two-tailed t-test). The aggregation of VASP after NO addition noticed here was not a consistent feature following NO donor addition. By contrast, cells transfected with the S239A GFP-VASP mutant showed no significant change in cell shape over the same time period (Fig. 4b). Thus, S239A GFP-VASP following NO treatment remains constitutively active compared with WT GFP-VASP.

The NO donor produced a steady decline in lamellipodial VASP over a 30-minute period, associated with persistent cell rounding (Figs 3 and 4). We were not able to follow cells reliably after this time to study potential recovery but, on those occasions when we were able to image cells for prolonged periods, there was no recovery of VASP distribution or cell shape in WT GFP-VASP after a 60-minute period and cells transfected with S239A GFP-VASP retained their normal appearance (data not shown).

We repeated our studies of the behaviour of WT GFP-VASP and S239A GFP-VASP in a different primary human cell, embryonic keratinocytes (HEKs). Following addition of NO, very similar changes in amount of VASP within lamellipodia and reduction of cell area were seen in cells transfected with WT GFP-VASP (Fig. 5a). These changes were absent in cells transfected with S239A GFP-VASP (Fig. 5b). We quantified these changes as decribed for PTECs and found a significant decline in the amount of WT GFP-VASP localized to the retracting cell edge following NO addition, with a half-life of 18.6 minutes (Fig. 5c). No significant decline in the amount of VASP at the lamellipodial edge was seen in cells transfected with S239A GFP-VASP (Fig. 5d). In addition, there was a significant reduction in cell area following NO treatment in cells transfected with WT GFP-VASP that was not seen with the S239A mutant (Fig. 5e). Thus, the critical dependence of VASP Ser239 phosphorylation on the disappearance of VASP from lamellipodia and reduction in cell area in HEKs following NO treatment was comparable with that found in PTECs.

Given the observed effects of NO on VASP distribution that are dependent on Ser239, we predicted that a phosphomimetic mutation at this site (S239D) should also produce differences in VASP localization within lamellipodia. We transfected human PTECs with a GFP-tagged VASP construct bearing this mutation (S239D GFP-VASP) and analysed the distribution of this mutant VASP (Fig. 6). In contrast to the uniform presence of WT VASP within lamellipodia (Fig. 6a), the S239D GFP-VASP mutant was only found within about 12% of lamellipodia (Fig. 6b). Cells expressing the mutant did not, however, show any rounding (Fig. 6a), suggesting that endogenous VASP maintained normal cell architecture.

We assayed whether inhibition of cGMP-dependent protein kinases or interruption of production of cGMP from guanylate cyclase has any effects on the NO-triggered changes of human PTEC lamellipodia and cell shape. We used small interfering RNA (siRNA) to downregulate the expression of PKG type I and type II (PKG I and PKG II, respectively) in human PTECs (Pfeifer et al., 1996; Pfeifer et al., 1998). Downregulation of PKG I was without effect on the decline in WT GFP-VASP localization and changes in cell shape following addition of NO donor (Fig. 7a and supplementary material Movie 5). Kymographic analysis showed clearly the retraction of the cell edge following addition of NO donor (Fig. 7b). Analysis of data from a number of cells showed that WT GFP-VASP was removed from the lamellipodial edge with a half-life of 10.13 minutes following addition of NO donor in cells treated with siRNA targeting PKG I (Fig. 7e,h), virtually identical kinetics to those seen in cells in which PKG I was not downregulated (Fig. 7h and compare Fig. 3e). Similarly, reduction of PKG I levels was without effect on the reduction in cell area following addition of NO donor to human PTECs (Fig. 6g).

By contrast, we found that reduction in PKG II levels prevented any effect of NO on the cells (Fig. 7c,d). Thus, in cells transfected with WT GFP-VASP, downregulation of PKG II prevented both the reduction in VASP localization to lamellipodia (Fig. 7f and Movie 6) and the reduction in cell area (Fig. 7g). Thus, we concluded that type II PKG was required to mediate the effects of NO on VASP.

Following addition of NO donor, cGMP levels are elevated by the action of the NO-sensitive guanylate cyclase (Hofmann et al., 2000). This is specifically inhibited by the drug 1H-[1,2,4]oxadiazole [4,3-a]quinoxalin-1-one (ODQ) (Lee et al., 2000). Following ODQ treatment for 30 minutes, there was a small but not significant reduction in the amount of VASP localized to the lamellipodial edge (Fig. 7i). Addition of NO donor produced no further decrease in VASP localization to the lamellipodial edge, nor did it alter cell shape following addition of NO donor (Fig. 7j). Thus, the effects of NO on VASP and cell shape are inhibited by ODQ, suggesting that the generation of cGMP through guanylate cyclase is required.

Staining of cells for filamentous actin with fluorescent phalloidin showed that following treatment with NO donor the changes in VASP distribution were paralleled by alterations in actin distribution. Compared with untreated control cells (Fig. 8a), filamentous actin within lamelliopodia was virtually abolished (Fig. 8b), although occasionally cells showed an intact cortical actin cytoskeleton. As noticed in live cell imaging, following treatment with NO donor, cells tended to round up and appear smaller (Fig. 8b). However, there was little change in VASP within focal adhesions following addition of NO donor (Fig. 8c), in agreement with previous reports, which found no effect of phosphorylation of VASP on its localization to focal adhesions unless all three possible phosphorylation sites were mutated (Smolenski et al., 2000).

Our work identifies for the first time an important functional consequence of PKG phosphorylation of VASP Ser239 on lamellipodia. Phosphorylation at this amino acid residue results in the rapid loss of VASP from lamellipodia, which are withdrawn into the cell leading to cell rounding. We propose that NO and other mediators of changes in cGMP levels are, thus, regulators of VASP function and can produce alterations in actin-rich cellular protrusions, such as the lamellipodia studied here. This provides a powerful means of controlling lamellipodial dynamics and, hence, potentially altering directed cell movement. Three sets of observations reported here support this hypothesis. First, treatment of primary human epithelial cells and embryonic keratinocytes with a NO donor results in loss of VASP from the lamellipodial edge, retraction of the lamellipodia and cell rounding. These changes were all prevented in cells transfected with a mutant VASP in which the Ser239 phosphorylation site had been mutated to Ala239 and mimicked by cells expressing the S239D phosphomimetic VASP mutant S239D GFP-VASP. Second, we demonstrate that the NO-induced alterations are dependent on PKG II and guanylate cyclase, consistent with the generation of cGMP following NO donor treatment and the subsequent activation of PKG following phosphorylation of VASP Ser239. Third, in cells treated with cytokines to induce endogenous NO production, loss of VASP from the lamellipodial edge was again entirely dependent on the presence of Ser239. Taken together, these data strongly support an important role for phosphorylation of VASP Ser239 in controlling lamellipodial structure and function.

Understanding how actin polymerization is controlled is of central importance in many cellular functions. Control of this polymerization is essential in changing cell behaviour, such as from a motile to non-motile phenotype, or changing cell direction during chemotaxis (Bretschneider et al., 2004; Diez et al., 2005). Ena/VASP family proteins are crucial in directing actin polymerization, but since these proteins are constitutively present within cells, a means must exist to control their activity. Phosphorylation is an attractive method of control that is rapid and can be tightly regulated.

All Ena/VASP members share a common tripartite structure with an N-terminal Ena/VASP homology 1 (EVH1) domain, a central proline-rich region and a C-terminal EVH2 domain. The EVH1 domain binds to proteins with a poly-proline II helix, such as vinculin and zyxin, that directs Ena/VASP proteins to focal complexes and adhesions. The central proline-rich core binds to the G-actin-binding protein profilin, as well as to a number of proteins containing SH3 and WW domains. The EVH2 domain binds both G- and F-actin, as well as containing a coiled-coil motif that is essential for Ena/VASP oligomerization. The Ser239 studied here is located immediately adjacent to the G-actin-binding site and biochemical data suggest that the anti-capping and filament-bundling activities of VASP are inhibited by phosphorylation at this site (Barzik et al., 2005). However, Ena/VASP proteins are also essential for filopodial production (Mejillano et al., 2004), which may account for the lack of these processes in cells treated with NO donor. Ser239 phosphorylation also inhibits actin-fibre formation (Zhuang et al., 2004). The exact physicochemical effects of Ser239 phosphorylation, however, requires further analysis as not all studies have found that VASP antagonises actin-capping proteins (Samarin et al., 2003; Schirenbeck et al., 2006).

Loss of VASP localization at lamellipodia after its phosphorylation at Ser239 might also indicate a crucial importance of this residue in the interaction of VASP with scaffolding proteins, such as lamellipodin, that appear to direct VASP to the lamellipodial edge (Krause et al., 2004). Such interactions might be important in linking environmental cues to changes in cell behaviour controlled by VASP and actin polymerization. NO can play an important role in such changes in cell locomotion, for example, in osteoclasts (Yaroslavskiy et al., 2005) and endothelial cells (Smolenski et al., 2000). NO and cGMP changes are also important in controlling neuronal development (Bicker, 2005; Haase and Bicker, 2003). The effects of NO on cell motility are complex: in endothelial cells PKG I activation inhibits cell motility (Smolenski et al., 2000), although another study found that NO supports endothelial cell migration (Murohara et al., 1999). Movement requires a coordinated cycle of adhesion and detachment, and endogenous NO effects within living cells might be subject to rapid changes that allow cycles of lamellipodial extension and retraction. We found that NO-mediated phosphorylation of VASP at Ser239 was associated with a reduction in cell speed (Fig. 3h), as might be expected given the retraction of lamellipodia and cell rounding (Fig. 4). Since phosphorylation at Ser239 prevents the anti-capping effects of VASP (Barzik et al., 2005), this suggests that this anti-capping activity of VASP is crucial for its ability to produce lamellipodial spreading. However, given that complete lack of VASP – paradoxically – increases cell speed (Bear et al., 2000), functions other than its anti-capping activity must be important in regulating cell motility – for example by interacting with the substratum. NO does not appear to influence VASP distribution in focal adhesions (Fig. 8) (Smolenski et al., 2000). Loss of lamellipodia with retention of cell adhesion might account for the reduction in cell speed following NO addition.

Phosphorylation of VASP at other amino acid residues also regulates its activity. Phosphorylation of Ser157 produces a substantial conformational change, resulting in altered mobility in SDS-PAGE. Phosphorylation at this site in vivo is largely as a result of PKA activation (Butt et al., 1994). In a study using site-specific mutations of Mena, which has phosphorylation sites homologous to Ser157 and Ser239 of VASP, the equivalent of Ser157 was shown to be essential for VASP function in a motility assay (Loureiro et al., 2002). However, Loureiro and colleagues did not examine the effects of potential activators, such as NO, of phosphorylation equivalent to that of Ser239. Phosphorylation at Thr278 occurs through cAMP-activated protein kinase, and impairs actin-stress-fibre formation and alters cell morphology (Blume et al., 2007). NO had only small effects on Ser157 phosphorylation (Fig. 3a); its effect on Thr278 is not known. Phosphorylation at these sites might also contribute to alterations in VASP distribution and cell motility. However, we have clearly demonstrated here the important effect that NO-dependent phosphorylation of Ser239 has on distribution of VASP, cell shape and motility. The human PTECs used in this study also express, in addition to VASP, Mena and EVL (data not shown), but as our studies were performed using transfected VASP constructs fused to GFP, we cannot make any conclusions as to the possible control of Mena or EVL function by phosphorylation.

We have tested the hypothesis that Ser239 phosphorylation of VASP affects lamellipodial structure and function. We have demonstrated that NO-induced phosphorylation at this site removes VASP from the lamellipodial edge, and leads to retraction of lamellipodia and cell rounding. This is an important means of controlling VASP activity and, hence, potentially regulating actin-based processes, such as directed cell motility.

The NO donor spermine NONOate [N-(2-aminoethyl)-N-(2-hydroxy-2-nitrosohydrazinol-1-2-ethylenediamine)] and the guanylate cyclase inhibitor ODQ (1H-[1,2,4]oxadiazolo [4,3-a]quinoxalin-1-one) were from Calbiochem. Rabbit anti-VASP (M4) was from Alexis Biochemicals. Mouse anti-phosphorylated VASP (Ser239) antibody, clone 16C2, was from Chemicon. Rabbit anti-PKG I (sc-25229) and anti-PKG II (sc-25430) antibodies were from Santa Cruz Biotechnology. Anti-actin antibody was from Abcam. Biotinylated anti-rabbit and anti-mouse antibodies were from Vector Laboratories. Alexa-Fluor-488-conjugated phalloidin and Texas-Red-X-conjugated phalloidin were from Molecular Probes.

Human PTECs were harvested from surgical specimens of human kidneys, obtained under Institutional and National ethical procedures. Isolation and culture of these cells was as described (Glynne et al., 2002; Glynne et al., 2001). Stimulation with cytokines was performed with interleukin-1, tumor necrosis factor and interferon γ exactly as described (Glynne et al., 2001). Human embryonic keratinocytes were obtained from Lonza Biologics (Slough, UK) and grown in media as supplied by the manufacturer.

cDNA of WT VASP, and of sequences of mutated S239A VASP and S239D VASP were cloned in-frame to the C-terminus of GFP in the vector pQBI25f (QBiogene), using the restriction sites of BamHI and EcoRI (resulting in constructs WT GFP-VASP, and S239A GFP-VASP and S239D GFP-VASP, respectively). Human PTECs were transfected using the AMAXA 2 system (AMAXA, Germany), with the basic epithelial transfection kit and programme T-13 and HEKs with the kit specific for these cells. Typically, 1×10⁶ cells were transfected with 1 μg of plasmid and then used for imaging ∼16 hours later.

For live cell imaging, transfected cells were plated in 2-well glass microscope-chamber slides, maintained in medium containing HEPES. Cells were imaged on a warmed (37°C) stage using a Zeiss Axiovert Inverted Fluorescent Microscope, with ×63 PlanFluor objective and a digital CCD camera (Hamamatsu C4742-95-12ER).

Once a suitable field of view had been chosen, cells were imaged for at least 30 minutes prior to any experimental additions. Where indicated, spermine NONOate was added to a final concentration of 200 μM, following which cells were imaged for at least another 30 minutes. Decomposed NO donor was without effect.

Time-lapse images were acquired using computer-driven shutters and Openlab 5.0 software (Improvision); images were acquired every 30 seconds. Image analysis was performed using Image J (Rasband, W.S., ImageJ, U.S. National Institutes of Health, Bethesda, MA, http://rsb.info.nih.gov/ij/, 1997-2006). Cell speeds were calculated by using the cell outliner Plugin of Image J and measuring the rate of displacement of the cell centroid. To quantify changes following NO donor addition, mean cell speeds 30 minutes prior to donor addition were compared with mean speeds 20 minutes afterwards and expressed as a percentage change. Kymographic analysis of cell movement (Hinz et al., 1999) was performed using plug-ins provided by J. Rietdorf and A. Seitz, EMBL Heidelberg (http://www.embl.de/eamnet/html/kymograph.html). The area of the lamellipodial edge occupied by GFP-VASP was performed using Image J. This was calculated by measuring the area of the edge of all lamellipodia that contained a signal from GFP-VASP (more than five times of background intensity) at various time points. Results were standardized by expressing the data as percent of the value calculated for a given cell immediately prior to NO donor addition. This allowed effects on different cells with different sizes of lamellipodia to be compared directly. To quantify the effects of the S239D mutation on VASP distribution, we scored a cell as having VASP localized to the lamellipodium if more than 10% of the lamellipodial edge had a signal from VASP that was of more than five times of background intensity. Images were prepared for publication using Adobe Photoshop.

For immunoprecipitation experiments, cells were washed once with cold PBS and lysed in Phosphosafe extraction buffer (Novogen) for 5 minutes at room temperature. Lysates were cleared by centrifugation and incubated with anti-VASP (1:125) for 60 minutes, rotating at 4°C. After incubation, 50 μl of protein G sepharose was added and the mixtures were rotated for a further 60 minutes at 4°C. The beads were then washed four times in Phosphosafe extraction buffer and finally resuspended in gel loading buffer. The bound proteins were resolved by SDS–PAGE and immunoblotting was performed as described below. The amount of phosphorylated VASP Ser157 at various times after NO addition was quantified as a percentage of total VASP by scanning gels immunoblotted with antibody to total VASP. VASP phosphorylated at Ser157 runs with a lower mobility on SDS PAGE with a molecular mass of 50 kDa compared with 46 kDa for VASP not phosphorylated at Ser157.

Protein extracts were separated by SDS-PAGE and transferred to Hybond-P membrane (Amersham). After blotting, the membranes were stained with Ponceau S solution (Sigma). Membranes were blocked with 5% skimmed milk in PBS-T (1×PBS with 0.1% Tween-20) and processed for enhanced chemiluminescence by standard methods.

The knockdown of PKG I and PKG II were performed using Custom Stealth™ Duplex RNAi (Invitrogen) directed against PKG I nucleotides 1100-1124 and PKG II nucleotides 1344-1368. The RNAi oligonucleotides were transiently transfected in human PTECs using the AMAXA 2 system, as previously described. The relative knockdown efficiency was determined using antibodies against both PKG I and PKG II. A non-specific oligonucleotide was used as a negative control (Ambion).

Comparison between two groups was made using Student's t-test, two-tailed, unequal variances, with data having a normal distribution by Shapiro-Wilk test. Changes in cell speed following NO donor addition were tested for a difference from zero change using a one-sample t-test. For comparisons between larger numbers of groups, analysis of variance was used. For clearly skewed data, comparisons were made using the Mann-Whitney test. The differences in the proportion of lamellipodia containing WT GFP-VASP or S239A GFP-VASP were tested by Fisher's exact test. In all cases, a result was considered significant if P<0.05. Statistical testing and curve fitting was performed using the computer program Prism 4.0 (GraphPad).

The work was supported by the Dr Hadwen Trust for Humane Research and the Wellcome Trust (to T.J.E.) and the Deutsche Forschungsgemeinschaft SFB688 (to T.R.).