Caldesmon Inhibits Arp2/3-mediated Actin Nucleation*

The Arp2/3 complex greatly accelerates actin polymerization, which is thought to play a major role in cell motility by inducing membrane protrusions including ruffling movements. Membrane ruffles contain a variety of actin-binding proteins, which would modulate Arp2/3-dependent actin polymerization. However, their exact roles in actin polymerization remain to be established. Because caldesmon is present in membrane ruffles, as well as in stress fibers, it may alter Arp2/3-mediated actin polymerization. We have found that caldesmon greatly retards Arp2/3-induced actin polymerization. Kinetic analyses have revealed that caldesmon inhibits the nucleation process, whereas it does not largely reduce elongation. Caldesmon is found to inhibit binding of Arp2/3 to F-actin, which apparently reduces the ability of F-actin as a secondary activator of Arp2/3-mediated nucleation. We also have found that the inhibition of the binding between actin and caldesmon either by Ca2+/calmodulin or by phosphorylation with cdc2 kinase reverses the inhibitory effect of caldesmon on Arp2/3-induced actin polymerization. Our results suggest that caldesmon may be a key protein that modulates membrane ruffling and that this may involve changes in caldesmon phosphorylation and/or intracellular calcium concentrations during signal transduction.

Arp2/3-dependent actin polymerization plays a critical role in controlling motile structures including membrane ruffling (1)(2)(3)(4)(5)(6)(7). Two small G-proteins, Rac and Cdc42, regulate actin polymerization and are responsible for the assembly of membrane ruffles and filopodia, respectively (8). Recent work has elucidated the essential role of the Arp2/3 complex in the Racor Cdc42-mediated actin polymerization (6). Actin alone polymerizes rather slowly because of the rate-limiting process of nucleation. The Arp2/3 complex greatly accelerates the nucleation process in a Rac-or Cdc42-dependent way; Rac or Cdc42 activates the effector proteins of WASP family proteins including N-WASP and WAVE (SCAR). Together they activate the nucleation activity of the Arp2/3 complex and induce rapid actin polymerization. The Arp2/3 complex also binds to the side of actin filaments and initiates branched actin polymerization, resulting in the formation of a dendritic actin network.
The appearance and extent of membrane ruffling vary widely depending on cell types, as well as cellular conditions and external circumstances. For example, fibroblasts generally show more vigorous ruffling than do epithelial cells. Membrane ruffling is regulated by a variety of external signals including serum or growth factors, extracellular matrix, and cell-cell contacts (9 -12). On the other hand, intrinsic signals appear to control constitutive membrane ruffling of motile cells. It is thus conceivable that, in addition to the small G-proteins, other proteins may modulate Arp2/3-dependent actin polymerization. Candidates with such functions are actin-binding proteins present in membrane ruffles. For example, cortactin activates Arp2/3-dependent actin polymerization and stabilizes branched actin polymerization in vitro (13)(14)(15). Tropomyosin, on the other hand, has been shown to reduce Arp2/3-dependent branching and branched nucleation (16). Most recently, coronin has been reported to bind to the Arp2/3/VCA complex and inhibit the de novo nucleation activity of Arp2/3 (17).
Caldesmon is an actin-binding protein that is localized in stress fibers, as well as in membrane ruffles (18,19). In vitro, caldesmon inhibits actin-activated myosin ATPase (20), and caldesmon together with tropomyosin protects actin filaments from severing activities of gelsolin and anneals gelsolin-severed actin filaments (21,22). These two activities are likely to be involved in regulating the actomyosin contractility and stability of stress fibers, respectively (23)(24)(25)(26)(27)(28). It is not clear, however, what functions caldesmon plays in membrane ruffles. We have examined the effects of caldesmon on Arp2/3-dependent actin polymerization and found that caldesmon inhibits the Arp2/3-dependent actin nucleation process.

MATERIALS AND METHODS
Protein Purification-Skeletal muscle G-actin was prepared as described (29), and pyrene dye was conjugated as described (30). The Arp2/3 complex was purified from HeLa cells using the method described in Ref. 31, except that the incubation at 37°C was omitted. The GST 1 -tagged C-terminal region of N-WASP (GST-VCA) was cloned by PCR using the N-WASP-expressing baculovirus (a kind gift from Drs. Takenawa and Miki, Tokyo University) as the template, expressed in bacteria, and purified as described (31). Rat nonmuscle caldesmon was expressed in bacteria and purified as described (32). Smooth muscle caldesmon was purified from chicken gizzard by the method described in Ref. 33 with slight modification as described (34). Cdc2 kinase was prepared as described previously (35). Calmodulin and -phosphatase were purchased from Sigma and New England Biolabs (Beverly, MA), respectively.
Polymerization Assay-Actin polymerization was measured by pyrene fluorescence using a Perkin Elmer spectrofluorometer, LS-50B. Mg-ATP G-actin (2 M 5% pyrene-labeled actin) was mixed with various proteins in polymerization buffer (final condition: 10 mM imidazole buffer, pH 7, 50 mM KCl, 1 mM MgCl 2 , 1 mM EGTA, 10 mM DTT) at time 0. When we examined the effects of Ca 2ϩ /calmodulin, 0.4 mM CaCl 2 was added instead of EGTA. The high concentration of DTT was included to prevent the dimerization of caldesmon (36).
Phosphorylation of Caldesmon with cdc2 Kinase-Smooth muscle caldesmon was phosphorylated by cdc2 kinase as described (35). The level of phosphorylation (determined by 32 P incorporation) was found to be 4.5 Ϯ 0.5 mol/mol. In some experiments, phosphorylated caldesmon was dephosphorylated by the incubation with -phosphatase at 30°C for 30 min, which resulted in the decrease of the incorporation to 0.8 Ϯ 0.4 mol/mol. -Phosphatase was denatured by heat treatment, and dephosphorylated caldesmon was recovered in a heat-stable fraction.
Actin Binding Assay-F-actin was polymerized from Mg-ATP Gactin (final concentration, 2 M) in the polymerization buffer (10 mM imidazole buffer, pH 7, 50 mM KCl, 1 mM MgCl 2 , 1 mM EGTA, 10 mM DTT, 4 M phalloidin) for 40 min. After polymerization, caldesmon (final concentration, 0.5 M) was added and incubated for 10 min, and then varying concentrations (5-20 nM) of Arp2/3 and GST-VCA (50 nM) were added. The samples were incubated for 20 min at room temperature and centrifuged in a Beckman Airfuge (26 p.s.i. ϫ 20 min). Both pellets and supernatants were suspended in SDS sample buffer and analyzed by the Western blot method using mouse monoclonal antibody against Arp3 (BD Biosciences Pharmingen, San Diego, CA). A standard curve was made by immunoblotting known concentrations of Arp2/3 complex with the same antibody and used to quantitatively determine the levels of Arp3 using Kodak one-dimensional Image Analysis Software (Eastman Kodak Co., Rochester, NY).
Electron Microscopy-The average length of actin filaments at the equilibrium state (after pyrene fluorescence reached a plateau) was measured by electron microscopy using negative staining technique (36). More than 1000 actin filaments were measured for each polymerization experiment.

Caldesmon Effectively Inhibits Arp2/3-dependent Actin
Polymerization-We first examined effects of caldesmon on Arp2/ 3-mediated actin polymerization. Actin (2 M), Arp2/3 (15 nM), and VCA (50 nM) were mixed with varying concentrations (0 -2 M) of either smooth muscle or nonmuscle caldesmon, and polymerization was initiated by the addition of salt at time 0. As Fig. 1 (A and B) shows, the higher the concentration of caldesmon, the longer the delay in actin polymerization. To quantitate the effects of caldesmon, we measured the delay times required to reach one-fifth of the final fluorescence, as well as slopes at the half-maximal fluorescence (proportional to the number of barbed ends, assuming that caldesmon did not change the elongation rate). As Fig. 2 shows, the delay time is increased from 100 to 200 s for smooth muscle caldesmon and to 210 s for nonmuscle caldesmon. The delay is saturated at 0.25 M for smooth muscle caldesmon and at 0.5-1 M for nonmuscle caldesmon. The inhibition of the polymerization rates is saturated at approximately 1 M of either type of caldesmon. It is noteworthy that caldesmon at a very low concentration (50 nM) has more effect on the delay time than on the slope of polymerization. The inhibition by caldesmon is at least as effective as that reported for tropomyosin (16). Without Arp2/3, the effect of caldesmon on actin polymerization was minimal. The delay time was changed from 165 to 180 s when 0.5 M of caldesmon was added in the absence of Arp2/3 (Fig. 1C). The difference was only 9%. In contrast, the same concentration (0.5 M) of caldesmon delayed Arp2/3-dependent polymerization to a much greater extent (from 99.4 to 193.2 s, corresponding to a 94% increase; see Fig. 2B). The polymerization rate (slopes at the half-maximal fluorescence) also showed little change in the absence of Arp2/3; caldesmon decreased the rate by merely 5%. In contrast, the same concentration of caldesmon decreased the rate by 62% in the presence of Arp2/3 (Fig. 2B). These results suggest that caldesmon affects Arp2/3-dependent processes of actin polymerization.
The C terminus of caldesmon is indispensable for the inhibitory effect. Although the N terminus of caldesmon contains a major myosin-binding site, the C terminus has the actin-binding domains in addition to the domains that bind to tropomyosin or calmodulin (37). We found that the C-terminal half (Glu 235 -Val 531 of rat nonmuscle caldesmon) was able to show inhibitory effects on Arp2/3-dependent actin polymerization similar to the full-length molecule, whereas the N terminus showed no effects on actin polymerization (results not shown). This result suggests that actin binding ability is important for the inhibition.
Caldesmon Does Not Largely Alter Elongation Rates-The retardation of Arp2/3-dependent actin polymerization could be caused by the reduction of the elongation rate from barbed ends, by the inhibition of actin nucleation, or by both. We first examined whether caldesmon slows the elongation from barbed ends of Arp2/3-capped actin seeds. To make seeds, actin (2 M, without pyrene label) was polymerized in the presence of the Arp2/3 complex (20 nM) and VCA (50 nM). The seeds were then diluted 10-fold and incubated with 0.3 M pyrene-labeled Gactin with or without caldesmon to monitor barbed end polymerization. We conclude from the following reasons that this assay reasonably tests the effects of caldesmon on elongation from barbed ends. First, the concentration of G-actin is below the critical concentration of pointed ends so that barbed end polymerization was monitored (final concentrations: 0.2 M F-actin seeds, 0.3 M G-actin, with or without 0.5 M caldesmon). Second, caldesmon did not increase the annealing of Arp2/3-capped actin filaments in this condition, and thus the number of barbed ends was not changed with or without caldes-mon. When Arp2/3-capped actin filaments with two different colors were mixed with or without caldesmon, the percentage of annealed actin filaments (visualized by combined actin filaments with two colors) was the same, indicating that caldesmon did not dissociate Arp2/3 complexes from pointed ends. Third, no nucleation occurred under this condition. It was possible that Arp2/3 and VCA used for making seeds could induce de novo nucleation. However, no actin polymerization was observed within 200 s (Fig. 3A, filled circles) when 0.5 M G-actin was mixed with 2 nM Arp2/3 and 5 nM VCA (the same concentrations used in this assay). Fourth, caldesmon binding is faster than actin elongation at the barbed ends. Because the binding of caldesmon to F-actin is rather a fast reaction (k ϩ ϭ 1 ϫ 10 7 M Ϫ1 s Ϫ1 ) (38), most actin filaments would be saturated with caldesmon before any significant actin polymerization could occur at the free barbed ends; half of caldesmon binds to actin in 0.1 s in this condition, whereas only 0.2% of actin seems to polymerize at the ends at the same time.
As Fig. 3B shows, caldesmon inhibits the elongation rates to open circles, actin plus nonmuscle caldesmon; closed circles, control (actin polymerization without seeds; G-actin of 0.5 M was incubated with 2 nM Arp2/3 and 5 nM VCA). Virtually no actin polymerization occurred. The buffer conditions were the same as those of Fig. 1. B, the initial elongation rates were calculated from the slopes in A. The values were normalized as "actin alone" to 100%. a small extent. The rates were decreased 20% by smooth muscle caldesmon and 10% by nonmuscle caldesmon. This small extent of inhibition, however, does not account for the large decreases in the rates of actin polymerization by caldesmon as shown in Figs. 1 and 2 in which the slope at half-maximum fluorescence was decreased by 62% by 0.5 M caldesmon.
Caldesmon Reduces Nucleation Rates-We then examined how caldesmon affects the apparent nucleation rate, k*, and the apparent elongation rate, k ϩ . According to the theory of actin polymerization devised by Oosawa and Asakura (39), the nucleation rate k* is proportional to the inverse of ϽiϾ*t, where ϽiϾ is the average length of actin filaments at the equilibrium state, and t is the time when a certain amount of monomers has polymerized. On the other hand, the elongation rate k ϩ is proportional to ϽiϾ/t. Fig. 4 shows the time courses of actin polymerization in the absence (panel A) or presence (panel B) of caldesmon when the concentrations of the Arp2/3 complex were changed from 0 to 20 nM. At the equilibrium, the lengths of more than 1000 actin filaments were measured by electron microscopy, and the length distribution is shown as the histogram in Fig. 5. It is clear that filament length becomes longer when caldesmon is present. Table I shows the average lengths, ϽiϾ, of actin filaments, as well as times (t1 ⁄5 ) at which 20% of F-actin was polymerized relative to the final level of F-actin. Both average lengths and t1 ⁄5 become longer in the presence of caldesmon than in its absence. We have then calculated relative changes in k* and k ϩ by normalizing the rates, k* and k ϩ , of polymer-ization of actin alone to 1.0. Fig. 6 shows the plots of relative values of k* and k ϩ versus Arp2/3 concentrations.
These plots indicate that caldesmon greatly decreases the nucleation rate, k* (Fig. 6A), but not the elongation rate, k ϩ (Fig. 6B). As expected, k* increased as the concentration of Arp2/3 increases (Fig. 6A). Importantly, k* in the presence of caldesmon (open squares) is decreased to one-third of that in the absence of caldesmon (filled squares). On the other hand, k ϩ (Fig. 6B) seems to be the same whether caldesmon is present (open squares) or absent (filled squares), although k ϩ decreases when the Arp2/3 concentration becomes high. The reason for the decrease is not clear at this time. It is interesting to note that the Arp2/3-dependent increase in k* is proportional to the second power of Arp2/3 concentrations (Fig. 6A, inset). This relationship may be reflected by the fact that Arp2/3 nucleation activity is greatly increased when the Arp2/3 complex binds to the side of an actin filament (29,40).  Fig. 4, the lengths of actin filaments were determined by electron microscopy, and distribution is shown as histograms. Open bars, without caldesmon; filled bars, with caldesmon. Note that the filament lengths become longer in the presence of caldesmon.
Caldesmon Does Not Interact with the Arp2/3 Complex-The kinetic analyses shown above (Figs. 4 -6) strongly suggest that caldesmon inhibits the nucleation activity of Arp2/3 complexes. One possibility was that caldesmon might directly associate with Arp2/3 complex or VCA and interfere the nucleation activity. However, we were unable to detect the binding of caldes-mon to the Arp2/3 complex, VCA, or a VCA-Arp2/3 complex when the association was examined by a pull-down assay using affinity column chromatography or by the surface plasmon resonance method (data not shown). This suggests that caldesmon may indirectly inhibit the nucleation activity of Arp2/3 complex.

Caldesmon Inhibits Branched Nucleation of Arp2/3 by Reducing Actin Binding of Arp2/3-How does caldesmon indi-
rectly reduce the nucleation activity of Arp2/3? Arp2/3 is able to nucleate actin in two ways. One is de novo nucleation (without polymerized actin), and the other is branched nucleation via binding of Arp2/3 to the side of F-actin filaments. Branched nucleation has been reported to be far more effective than de novo nucleation (29,40). Because caldesmon binds to the side of actin filaments, it is possible that caldesmon inhibits the latter nucleation by inhibiting Arp2/3-F-actin binding.
To test this possibility, we performed the following two experiments. We first asked whether caldesmon reduces the ability of F-actin seeds as the secondary activator of Arp2/3 nucleation. F-actin seeds (final concentration, 0.3 M) were first mixed with or without 0.5 M caldesmon, and then Arp2/3, VCA, and pyrene-labeled G-actin were added to initiate polymerization. As Fig. 7 shows, actin polymerization in the presence of Arp2/3 occurred very rapidly with almost no lag time when F-actin seeds without caldesmon were added (Fig. 7, filled squares; compare the actin polymerization without Factin seeds shown in Fig. 1). The addition of caldesmon (Fig.  7A, nonmuscle caldesmon; Fig. 7B, smooth muscle caldesmon) to F-actin seeds, however, greatly reduces the ability of F-actin seeds (compare open squares with filled squares). The effect is specific to Arp2/3-dependent actin polymerization because, in the absence of Arp2/3, the addition of either nonmuscle (Fig.  7A) or smooth muscle (Fig. 7B) caldesmon showed only small (for nonmuscle caldesmon) or no (for smooth muscle caldesmon) effects on actin polymerization (compare filled and open circles). These results suggest that caldesmon inhibits branched nucleation mediated by the binding of Arp2/3 to F-actin seeds. However, because there was an excess amount of caldesmon in these assays, it is possible that free caldesmon may also inhibit de novo nucleation of Arp2/3. We next examined whether caldesmon inhibits binding of Arp2/3 to F-actin, because actin binding of Arp2/3 is required for branched nucleation. Using co-sedimentation assays, we examined whether caldesmon affects actin binding of Arp2/3. F-actin (2 M) with or without caldesmon (0.5 M) was incubated with varying concentrations of Arp2/3 (5-20 nM), and the levels of Arp2/3 in F-actin pellets were determined by quantitative Western blot method using the monoclonal antibody against Arp3. As Fig. 8 shows, the levels of bound Arp2/3 are reduced to 30 -60% of the controls, when actin was preincubated with either nonmuscle or smooth muscle caldesmon. These results indicate that caldesmon inhibits Arp2/3 binding to F-actin.
Reversal of the Inhibition of Caldesmon by Ca 2ϩ /Calmodulin or by Phosphorylation with cdc2 Kinase-The interaction of caldesmon with actin is regulated by Ca 2ϩ /calmodulin or by phosphorylation of caldesmon with a variety of kinases including cdc2 kinase (35,41). Because actin binding of caldesmon is important for its effects on Arp2/3, we examined whether these agents regulate the effects of caldesmon on Arp2/3 nucleation. As Fig. 9 shows, Ca 2ϩ /calmodulin negated the effects of the inhibition of caldesmon. Although caldesmon alone (filled triangles) retarded Arp2/3-dependent actin polymerization greatly (compare with filled squares), the addition of Ca 2ϩ / calmodulin (open triangles) to caldesmon almost completely reversed the inhibitory effect of caldesmon. As an additional  Fig. 4.   FIG. 6. Caldesmon reduces nucleation rates of Arp2/3-dependent actin polymerization but not the elongation rates. A, the relative nucleation rate constants (k*) are calculated from Table I according to the theory of Oosawa and Asakura on protein polymerization (39) and plotted versus Arp2/3 concentrations. The rate constants are normalized as actin alone to 1. Inset, the Arp2/3-dependent increases in the nucleation rates (k* Ϫ 1) are proportional to the square of Arp2/3 concentrations. Closed squares, without caldesmon; open squares, with caldesmon. B, the relative elongation rate constants (k ϩ ) calculated from Table I are  We also found that phosphorylation of caldesmon with cdc2 kinase completely negates the inhibition of caldesmon (Fig. 10). Although unphosphorylated caldesmon (Fig. 10, ϫ) greatly inhibited Arp2/3-dependent actin polymerization, caldesmon phosphorylated by cdc2 kinase to 4 -5 mol/mol showed no inhibition on Arp2/3-dependent actin polymerization. The time course without caldesmon (filled squares) was identical to that with phosphorylated caldesmon (open circles). The release of inhibition was phosphorylation-specific; when caldesmon was dephosphorylated with -phosphatase, caldesmon regained its inhibitory effect on Arp2/3-dependent actin polymerization (filled triangles). Dephosphorylation was not complete (phosphorylation level was 0.8 Ϯ 0.4 mol/mol), which appears to explain the fact that dephosphorylated caldesmon showed less inhibition than unphosphorylated caldesmon. DISCUSSION Our data indicate that caldesmon reduces the nucleation rate of the Arp2/3-dependent actin polymerization by inhibit-ing actin binding of Arp2/3. Three lines of evidence support this conclusion. First, kinetic analyses showed that caldesmon reduced the nucleation rate, whereas it did not alter the elongation rate largely (Fig. 6). Second, F-actin seeds bound with caldesmon were much less effective in activating nucleation by Arp2/3 than were F-actin seeds without caldesmon (Fig. 7). Finally, caldesmon inhibited actin binding of Arp2/3 (Fig. 8). These results strongly suggest that caldesmon reduces the nucleation rate by inhibiting branched nucleation, which is similar to the way of the inhibition reported for tropomyosin (16). The inhibition of branched nucleation is probably related to the way of actin binding of these two proteins; both tropomyosin and caldesmon bind to the side of filaments. Although it has not been shown that tropomyosin inhibits actin binding of Arp2/3, side binding of tropomyosin is likely to reduce the actin binding affinity of Arp2/3 as did caldesmon. Another possibility for the inhibition of branching would be that caldesmon stimulates the conversion of ATP-actin to ADP-actin. However, we did not detect any stimulation of actin ATPase by caldesmon (data not shown).
We attempted to observe directly, by fluorescent microscopy, whether caldesmon inhibits branched F-actin formations. However, we found that this assay had the following problems, which made results unreliable. First, the binding of caldesmon made F-actin filaments more rigid and resistant to shearing force than F-actin without caldesmon. During microscopic ob- servation, we observed that branched F-actin without caldesmon was frequently broken at its branch sites by shearing force, whereas branched F-actin with caldesmon was preserved better under the same conditions. Second, caldesmon tends to form a dimer in the absence of reducing agents, and dimer caldesmon can cross-link actin filaments (36). Although a high concentration of DTT reduced dimer formation of most but not all caldesmon, residual caldesmon dimers appear to generate loosely cross-linked actin filaments. Such loosely cross-linked actin filaments were quite similar to branched filaments under a microscope, again making accurate quantitation of branching quite difficult.
The inhibition of Arp2/3-induced actin polymerization by caldesmon is likely to have physiological significance. First, caldesmon and Arp2/3 are co-localized in membrane ruffles. Second, the intracellular concentration of caldesmon appears high enough to affect Arp2/3-dependent actin polymerization. For example, fibroblasts contain ϳ2-8 M caldesmon (32). Because total actin concentration is approximately 100 -200 M, the molar ratio of caldesmon to actin could be between 1:100 and 1:12, in which range caldesmon was shown to affect actin polymerization in this work (Figs. 1 and 2). Fourth, changes in caldesmon concentrations appear to alter the motile activity of the peripheral membranes. For example, caldesmon has been reported to be down-regulated in many transformed cells (19,26). This down-regulation is well correlated to the increases of membrane ruffling upon cell transformation. Furthermore, in our preliminary experiments, we found that microinjection of caldesmon into normal rat kidney cells reduced membrane ruffling movement considerably (data not shown).
Caldesmon may be a key molecule that could confer phosphorylation-dependent or Ca 2ϩ /calmodulin-dependent regulation of Arp2/3-mediated actin polymerization. We have demonstrated that the inhibition of caldesmon of Arp2/3-dependent actin polymerization is released by phosphorylation with cdc2 kinase (Fig. 10) and by Ca 2ϩ /calmodulin (Fig. 9). This release is likely to be caused by the inhibition of actin binding of caldesmon by these agents. Although tropomyosin has been reported to inhibit Arp2/3-dependent actin polymerization (16), it is mainly localized in stress fibers, and the actin binding of tropomyosin is not regulated like caldesmon. Because other kinases including MAPK (43) and protein kinase C (44) were also reported to regulate the actin binding of caldesmon, caldesmon may play a role in motile phenomena in a variety of signal transduction pathways.
For example, growth factors like PDGF are known to induce membrane ruffles. PDGF also causes caldesmon phosphorylation by MAPK. Because the phosphorylation sites of caldesmon by MAPK partially overlap with those by cdc2 kinase, phosphorylation with MAPK is likely to reverse the inhibition of caldesmon on Arp2/3 nucleation. Indeed, our preliminary data showed that caldesmon phosphorylated by MAPK lost its inhibition of Arp2/3-dependent actin polymerization (data not shown). Thus, although growth factors like PDGF activate small G-proteins such as Cdc42 and Rac, which in turn initiate membrane ruffles, phosphorylation of caldesmon by MAPK would enhance Arp2/3-induced actin polymerization, leading to more vigorous membrane ruffling and cell movement during growth factor treatment (45,46). This notion appeared to be supported by a report showing that caldesmon phosphorylation by MAPK is involved in PDGF-stimulated cell migration of smooth muscle cells (46).
Caldesmon phosphorylation by cdc2 kinase may also have physiological significance. It has been recently reported that Arp2/3 plays an important role in the completion of cytokinesis in yeast (47) as well as Drosophila (42). We have shown here that phosphorylation of caldesmon by cdc2 kinase completely abolished the inhibition of caldesmon on Arp2/3 (Fig. 10). Because our previous study has shown that caldesmon is phosphorylated by cdc2 kinase during mitosis (35,41), it is possible that the release of the inhibition of caldesmon may be involved in Arp2/3-mediated assembly of contractile rings. Future studies should be conducted to define how the regulation of actincaldesmon binding controls actin polymerization and cell motility during signal transduction.