Phosphorylation of human fascin inhibits its actin binding and bundling activities.

Human fascin is an actin-bundling protein that is thought to be involved in the assembly of actin filament bundles present in microspikes as well as in membrane ruffles and stress fibers. We have found that human fascin is phosphorylated in vivo upon treatment with 12-O-tetradecanoylphorbol-13-acetate, a tumor promoter. The in vivo phosphorylation is gradually increased from 0.13 to 0.30 mol/mol during 2 h of treatment, concomitant with the disappearance of human fascin from stress fibers, membrane ruffles, and microspikes. Human fascin can also be phosphorylated in vitro by protein kinase C at the same sites as observed in vivo as judged by phosphopeptide mapping. The extent of phosphorylation depends on pH: the stoichiometries are 0.05, 0.38, and 0.6 mol of phosphate/mol of protein at pH 7.0, 6.0, and 5.0, respectively. Phosphorylation greatly reduces actin binding of human fascin, while lowering the pH to 6.0 alone does not affect fascin-actin binding. With the incorporation of 0.25 mol of phosphate/mol of protein, the actin binding affinity is reduced from 6.7 × 10 to 1.5 × 10M. The actin bundling activity is also decreased. These results suggest that phosphorylation of fascin plays a role in actin reorganization after treatment with 12-O-tetradecanoylphorbol-13-acetate.

HeLa 55-kDa actin-binding protein and sea urchin fascin are similar in that they both cause aggregation of F-actin into bundles and that their molecular masses are similar. Sea urchin fascin is a 58-kDa protein, first isolated from sea urchin eggs (1). Fascin is thought to act in the formation of microvillar cores at fertilization (2) and is involved in the formation of filopodia by coelomocytes, phagocytotic defense cells of echinoderms (2). We have purified a 55-kDa actin-bundling protein from HeLa cells (3). Immunofluorescent studies have revealed that this human protein is localized in filopodia and membrane ruffles as well as in stress fibers (4), suggesting the involvement of this protein in the organization of these structures.
The molecular cloning of sea urchin fascin by Bryan et al. (5) showed that fascin has 35% identity at the amino acid level to the product of the Drosophila singed gene, a 55-kDa protein of heretofore unknown function. The derived amino acid sequence of sea urchin fascin shows homology to three peptide sequences of HeLa 55-kDa protein, which suggested that HeLa 55-kDa protein is a human homologue of sea urchin fascin. Two groups have independently cloned a human cDNA homologous to Drosophila singed. They have demonstrated, in collaboration with us, that HeLa 55-kDa protein is a homologue of Drosophila singed protein and sea urchin fascin (6,7). Subsequently, mouse and Xenopus fascins have also been cloned (8). The amino acid sequences of these fascins share no apparent homology with other actin-bundling proteins (including villin and fimbrin), indicating that the fascins represent a new family of actin-binding proteins.
It appears that each of these three proteins functions as an actin-bundling protein in microspikes and stress fibers of fibroblasts (4,9), in filopodia of coelomocytes (2), and in bristles and nurse cells of Drosophila (10). Many Drosophila singed mutants exhibit two phenotypes: gnarled bristle development and female sterility (11). A common feature shared between the two phenotypes involves the association of the singed protein with actin fibers (10). In bristles, actin-containing filament bundles that are oriented along the length of the shaft surround microtubules in the center of the bristle. In singed mutants, a decreased number of microfilament bundles are observed, resulting in a short, curved bristle shaft (10,12). In Drosophila oogenesis, each developing oocyte is surrounded by and connected by intercellular bridges to 15 nurse cells (13). Nurse cell cytoplasmic contents flow into the oocyte along actin filaments traversing these cytoplasmic bridges. A singed allele associated with female sterility has been reported to affect the microfilament structure required for this nurse cell cytoplasmic flow (10).
It is not clear what regulates the actin binding and bundling activities of human fascin. This report shows that human fascin is phosphorylated in vivo up to 0.30 mol/mol of protein by TPA 1 treatment. Human fascin can also be phosphorylated in vitro by PKC when the pH is lowered to 6.0. In vitro phosphorylation occurs at the same sites as observed in vivo as judged by phosphopeptide mapping. Phosphorylation greatly reduces the actin binding ability of human fascin, although lowering the pH to 6.0 alone does not affect fascin-actin binding. While it remains to be established whether PKC is indeed responsible for in vivo phosphorylation of human fascin, these results suggest that phosphorylation of human fascin regulates its actin binding and bundling activities, thereby controlling the organization of microspikes as well as stress fibers.

MATERIALS AND METHODS
Cell Culture-Human neuroblastoma cells (ATCC HTB11 SK-N-SH) were used in this study. Cells were maintained in Dulbecco's modified Eagle's medium containing 10% fetal calf serum in an atmosphere of 5% CO 2 and 95% air at 37°C. 32 P or 35 S Labeling and Immunoprecipitation-For 32 P labeling, cells (one 100-mm dish, 70 -80% confluency) were incubated for 2.5 h in phosphate-free Dulbecco's modified Eagle's medium containing 0.18 mCi/ml [ 32 P]orthophosphoric acid and 10% dialyzed newborn calf serum, and then TPA (800 M dissolved in dimethyl sulfoxide) was added to a final concentration of 240 nM. Cells were continuously labeled until the indicated times. After washing three times with phosphate-free Dulbecco's modified Eagle's medium, cells were lysed in 500 l of hot 2 * This work was supported by Grant CD-40B from the American Cancer Society and Grant R37 CA42742 from NCI. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
ϫ SDS sample buffer (2% SDS, 15% glycerol, 100 mM dithiothreitol, 80 mM Tris-HCl, 0.001% bromphenol blue (pH 6.8)) and stored at Ϫ70°C. Control cells were labeled under the same conditions except for the addition of TPA and were processed in the same way.
For 35 S labeling, cells (one 100-mm dish, 70 -80% confluency) were incubated in methionine-free Dulbecco's modified Eagle's medium containing 55 Ci/ml Tran 35 S-label (ICN) and 10% dialyzed newborn calf serum. Cells were treated with TPA in the same way as described above.
Immunoprecipitation was performed with two monoclonal antibodies (clone 55k-2 and 55k-14), which have been characterized previously (4). After being thawed quickly, the samples lysed in the SDS sample buffer were first diluted 10 times with phosphate-buffered saline (137 mM NaCl, 2.7 mM KCl, 1.5 mM KH 2 PO 4 , 8 mM Na 2 HPO 4 (pH 7.3)) containing 50 mM NaF, homogenized by several passages through a 25-gauge needle, and then centrifuged at 100,000 ϫ g for 30 min. To the supernatants containing the same total radioactivities was added 0.025 volume of ascites fluids of the monoclonal antibodies. After incubation at 4°C for 1.5 h, rabbit antibodies against mouse IgG that had been conjugated to protein A-Sepharose were added and further incubated for 1 h. After washing, one-half of the immune complexes were loaded on one-or two-dimensional gels and analyzed by autoradiography. Human fascin is focused as multiple spots on two-dimensional gels (3). To unambiguously identify fascin spots on two-dimensional gels, the other half of the immunoprecipitate samples were mixed with purified, nonradioactive HeLa fascin and separated on two-dimensional gels, and fascin spots were identified by comparison of the patterns detected by Coomassie Blue staining and by autoradiography. The amounts of radioactivity in 35 S-or 32 P-labeled protein spots separated on one-or two-dimensional gels were quantitated with a PhosphorImager (Molecular Dynamics, Inc., Sunnyvale, CA).
Proteins-HeLa 55-kDa protein (human fascin) was purified from HeLa cells as described previously (3). Skeletal muscle actin was prepared as described (3). PKC was purchased from Life Technologies, Inc., Upstate Biotechnology, Inc., and Promega, all of which gave similar results. Calcineurin was purchased from Upstate Biotechnology, Inc.
In Vitro Phosphorylation of Human Fascin by Protein Kinase C-Human fascin purified from HeLa cells was phosphorylated at 30°C by PKC in a buffer at various pH values containing 30 mM KCl, 3 mM MgCl 2 , 100 ng/ml TPA, 150 g/ml phosphatidylserine, 0.5 mM dithiothreitol, 0.5 mM ATP (0.125 mCi/ml [gamma-32 P]ATP), 1-5 g/ml PKC, and 0.2 mM EGTA or 0.1 mM CaCl 2 . The following buffers were used to phosphorylate human fascin at different pH values: 20 mM Tris buffer at pH 8.0 to 7.5; 20 mM imidazole buffer at pH 7.0; and 20 mM MES buffer at pH 6.5 to 5.0. To determine phosphate incorporation, human fascin was separated on SDS gels, and a 55-kDa band was excised from the gels and counted with a scintillation counter by the Cerenkov method. In some experiments, phosphorylated fascin was dephosphorylated by treatment with calcineurin under the following conditions: 50 mM imidazole HCl (pH 7.0), 0.1 mg/ml bovine serum albumin, 0.025 mg/ml calmodulin, 5 mM MgCl 2 , 0.1 mM CaCl 2 , 1 mM MnCl 2 , 0.5 mM dithiothreitol, 0.4 mM 4-(2-aminoethyl)benzenesulfonyl fluoride hydrochloride, and 10 g/ml calcineurin.
Actin Binding and Bundling Assays-Human fascin purified from HeLa cells was phosphorylated at pH 6.0 as described above, and as a control, fascin was incubated in the pH 6.0 MES buffer in the same way except for the addition of PKC. Both samples were examined for their actin binding and bundling activities as described (3). Briefly, 12 M F-actin was mixed with varying concentrations of phosphorylated or unphosphorylated human fascin in 20 mM imidazole buffer containing 100 mM KCl, and the pH values of the mixtures were adjusted to pH 7.0. After incubation at room temperature for 1 h, the mixtures were centrifuged in a Beckman Airfuge at 28 p.s.i. for 30 min for the actin binding assay. For the actin bundling assay, the mixtures were centrifuged in an Eppendorf centrifuge for 15 min. Both supernatants and pellets were analyzed by SDS-polyacrylamide gel electrophoresis. The intensities of the 55-kDa bands on stained SDS gels and autoradiographs were analyzed with a densitometer (Chromoscan 3, Joyce-Loebl and Co.) as described (14). Actin bundling activities were confirmed by electron microscopy.
Other Procedures-Two-dimensional phosphopeptide mapping and phosphoamino acid analyses were performed as described previously (21). Two-dimensional gel electrophoresis was performed as described (3). Immunofluorescence was performed as described (4). Protein concentrations were determined using the Bradford assay (15).

RESULTS
In Vivo Phosphorylation of Human Fascin-We investigated, using human neuroblastoma cells, whether human fascin is phosphorylated under a variety of conditions including treatment with TPA (PKC activator), forskolin (protein kinase A activator), or platelet-derived growth factor. We have found that human fascin is phosphorylated to a considerable extent by treatment with TPA, but not with the other agents. As Fig.  1 shows, fascin in the immunoprecipitate is found to be phosphorylated in vivo after a 2-h treatment with TPA, but not before TPA treatment.
We examined the time course of fascin phosphorylation after TPA treatment. As Fig. 2 shows, fascin phosphorylation is not a rapid reaction, gradually increasing over 2 h after TPA treatment. To correlate fascin phosphorylation with changes in the localization of fascin, we have immunolocalized fascin during the time course of fascin phosphorylation (Fig. 3). Human neuroblastoma cells are heterogeneous, consisting of two kinds of cells: spindle-shaped cells with few stress fibers but with long neurite-like extensions (Fig. 3A) and well spread cells with stress fibers and short microspikes (Fig. 3B). In a short time (10 min) after TPA treatment, the neurite-like extensions disappear. However, numerous membrane ruffles that show strong staining with fascin antibody are induced in both types of cells (indicated by arrowheads in Fig. 3, C and D), as reported previously with the other type of cells (16). In addition, short microspikes are often associated with membrane ruffles, which are also stained with fascin antibody. In contrast, stress fibers are more resistant to TPA: fascin staining of stress fibers persists at this time (Fig. 3D), when the phosphorylation level is low. Fascin staining of stress fibers disappears ϳ1-2 h after TPA treatment (Fig. 3F), when the phosphorylation level is reaching a near maximum level. Concomitantly, the fascin staining in membrane ruffles also becomes diffuse, with the appearance of perinuclear cytoplasmic staining of fascin in many cells (Fig. 3E). Cell shape becomes more rounded. These observations suggest that in vivo phosphorylation of fascin appears to be correlated with the disappearance of fascin in stress fibers and membrane ruffles at later stages of TPA treatment.
We next determined the stoichiometry of in vivo phosphorylation in the following way. Because human fascin is known to be focused as multiple spots on two-dimensional gels (3), we first identified, by in vivo 32 P labeling, which spots represent phosphorylated isoforms of human fascin. Then, by labeling in vivo with [ 35 S]methionine followed by two-dimensional gel separation, the level of both phosphorylated and unphosphorylated isoforms of fascin can be determined by PhosphorImager analysis. Fig. 4A shows a Coomassie Blue-stained two-dimensional gel of purified, nonradioactive HeLa fascin for reference, which exhibits two major basic and two minor acidic spots (designated as spots a-d from basic to acidic). Fascin was immunoprecipitated from 32 P-labeled cells after 2 h of TPA treatment. Fig. 4B shows an autoradiograph of a two-dimensional gel of the immunoprecipitates where five spots are observed (spots c-g). Co-electrophoresis of the 32 P-labeled immunoprecipitates with unlabeled human fascin (data not shown) revealed that two (Fig. 4B, spots c and d) of the five phospho- rylated spots comigrate with the two minor acidic forms (Fig.  4A, spots c and d), but not with the two major basic forms (spots a and b), of purified HeLa fascin. The three most acidic spots (Fig. 4B, spots e-g) are too minor to be detected on the Coomassie Blue-stained two-dimensional gel pattern of the purified protein (Fig. 4A). These are, however, isoforms of human fascin because Western blotting using enhanced chemiluminescence (Amersham Corp.) detected these minor spots on twodimensional gels (data not shown).
With the identification of phosphorylated isoforms of human fascin on two-dimensional gels, we determined the stoichiometry of in vivo phosphorylation of human fascin after a 2-h treatment of human neuroblastoma cells with TPA. Cells were labeled with [ 35 S]methionine in the presence or absence of TPA, and fascin was immunoprecipitated and separated on two-dimensional gels. Fig. 4 (C and D) shows the two-dimensional gel patterns of human fascin in the absence and presence of TPA, respectively. Again, the identification of these spots as fascin was confirmed by co-electrophoresis of radioactive spots with unlabeled purified HeLa fascin. Due to the lower sensitivity of 35 S in comparison with 32 P, we cannot see the very minor spots from d to g in the control (Fig. 4C) and from e to g in the TPA-treated sample (Fig. 4D) on these autoradiographs. However, we can detect spots a-d with a PhosphorImager and thus determined the levels of the spots from a to d. We found that the level of phosphorylated fascin (sum of spots c-d divided by sum of spots a-d) is increased from 13 to 30% after a 2-h TPA treatment of human neuroblastoma cells.
In Vitro Phosphorylation of Human Fascin by Protein Kinase C-TPA treatment is known to activate PKC in vivo (17). We thus investigated whether PKC phosphorylates human fascin in vitro. As Fig. 5A (lanes 1 and 3) shows, PKC does phosphorylate human fascin in vitro. HeLa fascin preparation alone has no kinase activity because human fascin is not phosphorylated without the addition of PKC (lanes 2 and 4). Two-dimensional gel analyses revealed that in vitro phosphorylated fascin yields the same phosphoprotein pattern as observed with in vivo phosphorylated fascin. In Fig. 5B, panels 1 and 2 show the Coomassie Blue-stained gel and its corresponding autoradiograph, respectively. Again, five phosphoprotein spots (spots c-g) are identified on the two-dimensional gel (panel 2), of which two (spots c and d) comigrate with acidic minor forms of purified human fascin (panel 1), and two major basic spots (spots a and b) are not phosphorylated. The more acidic spots (spots e-g) are either barely detected or not detected on the Coomassie Blue-stained gel (panel 1). It should be noted here that phosphorylation was performed at pH 7.0, under which condition human fascin is poorly phosphorylated as described below, and thus, an increase in the level of phosphorylated spots (spots c-e) is not evident.
In vitro phosphorylation of fascin by PKC depends greatly on pH (Fig. 6). At pH 7.0 and above, only 5% of fascin is phosphorylated, which initially made us think that fascin is a poor substrate of PKC. However, the stoichiometry of fascin phosphorylation is greatly improved as the pH is lowered, while the activity of PKC itself is decreased. At pH 6.0, ϳ38% of fascin is phosphorylated within 20 min, and 60% of fascin is phosphorylated at pH 5.0. In contrast, phosphorylation of myosin light chain, a good substrate of PKC, is decreased by 30% when the pH is decreased from 7.0 to 5.0. The increase in phosphorylation of fascin at lower pH values is probably not due to the activation of contaminant kinases because a PKC peptide inhibitor (Sigma) inhibits phosphorylation reactions of both fascin and myosin light chain to the same extent at pH 6.0.
To determine whether in vitro phosphorylation of fascin with PKC occurs at the same sites as observed in vivo, two-dimensional phosphopeptide mapping was performed. Fig. 7 (A and  B) shows phosphopeptide maps of in vitro phosphorylated fascin and in vivo phosphorylated fascin, respectively. These two maps appear to be identical, which is confirmed by peptide mapping of a mixture of in vivo and in vitro phosphorylated fascin (Fig. 7C). While the extent of in vitro phosphorylation greatly depends on pH, the patterns of two-dimensional peptide maps are found to be the same (data not shown). We also analyzed the phosphoamino acids in these spots and found that the two major and one minor phosphopeptide spots (indicated by arrowheads 1, 2, and 4) contain phosphoserine, while the other minor spot (indicated by arrowhead 3) contains phospho-threonine (data not shown).
Inhibition of Actin Binding by Phosphorylation-We examined the effects of phosphorylation on fascin-actin binding and found that phosphorylation by PKC inhibits actin binding of fascin. Fascin was first phosphorylated at pH 6.0 with PKC to a molar ratio of 0.25 mol of phosphate/mol of protein, and then actin binding was examined at pH 7.0. As a control, fascin was treated in the same way except that PKC was not added to the phosphorylation reaction mixtures.  3, 4, 7, and 8). Coomassie Blue staining (lanes 1-8) revealed that the levels of free fascin in the supernatants with phosphorylated fascin (lanes 5 and 7) are higher than those with control unphosphorylated fascin (lanes 1 and 3) (compare lane 5 with lane 1 and lane 7 with lane 3), indicating that phosphorylated fascin shows a lower actin affinity. Furthermore, the autoradiographs (lanes 9 -12) corresponding to the Coomassie Blue-stained gels (lanes [5][6][7][8] show that most of the phosphorylated fascin remains in the supernatants (lanes 9  and 11).
We determined the level of radioactivity of fascin in the pellets and supernatants after actin binding. The phosphate incorporation of the original phosphorylated sample is 0.25 mol/mol of protein. After actin binding, the phosphate incorporation of fascin in the supernatant (Fig. 8, lanes 5 and 9) is increased to 0.71 mol of phosphate/mol of protein, while that in the pellet (lanes 6 and 10) is decreased to 0.07 mol/mol of protein. These results indicate that phosphorylated fascin has a greatly reduced actin binding activity.  3 and 4) of unphosphorylated fascin were incubated under the same conditions but without PKC (Coomassie Blue staining; autoradiography is not shown because of lack of phosphate incorporation). Lanes 1, 3, 5, 7, 9, and 11, supernatants; lanes 2, 4, 6, 8, 10, and 12, pellets. Note that more fascin remains in the supernatants (lanes 5 and 7) when phosphorylated fascin is used compared with the control (lanes  1 and 3). Also note that most of the radioactivity remains in the supernatants (lanes 9 and 11), but not in the pellets (lanes 10 and 12). Molecular mass markers are indicated by the arrowheads to the left and are the same as described for Fig. 1.
We used 25% phosphorylated fascin to estimate its actin binding constant. Fig. 9 shows a comparison of actin binding between unphosphorylated and phosphorylated fascin. Phosphorylated fascin shows reduced actin binding. The reduction in actin binding is not due to denaturation caused by pH 6.0 as judged from the following results. First, the actin binding affinity of fascin is not affected by the pH range from 7.5 to 6.0. Second, dephosphorylation of phosphorylated fascin by treatment with calcineurin (protein phosphatase 2B) restores actin binding. Although fascin is not completely dephosphorylated by this treatment, phosphate incorporation is decreased from 0.25 to 0.1 mol of phosphate/mol of protein. As Fig. 9 shows (open square marked with an arrowhead), fascin treated with phosphatase shows a higher affinity than does untreated phosphorylated fascin, indicating that phosphorylation indeed regu-lates actin binding of fascin. Fig. 10 shows Scatchard plots of actin binding, which yield a bell-shaped curve, suggesting that actin binding of either phosphorylated or unphosphorylated human fascin is cooperative. Actin binding constants calculated from 25-75% saturation points give 6.7 ϫ 10 6 M Ϫ1 for control unphosphorylated fascin and 1.5 ϫ 10 6 M Ϫ1 for 25% phosphorylated fascin. These plots also indicate that both control and phosphorylated fascin give a similar saturation of actin binding: one fascin molecule to four to five actin molecules, which is similar to the value reported previously (3).
Finally, we examined the effects on actin bundling activities. As Fig. 11 shows, the actin bundling activity of phosphorylated fascin is considerably lower than that of the control. The concentrations required for the half-maximal bundling activity are 1.1 M for unphosphorylated fascin and 1.6 M for phosphorylated fascin (0.25 mol of phosphate/mol of protein). DISCUSSION We have shown in this paper that 1) fascin is phosphorylated in vivo up to 30% after TPA treatment; 2) PKC is able to phosphorylate fascin in vitro at the same sites as observed in vivo; and 3) in vitro phosphorylation greatly reduces fascinactin binding. TPA treatment is known to induce movement of membrane ruffles and to cause disassembly of stress fibers in many cells (16). Human neuroblastoma cells also show similar but complicated effects upon TPA treatment (Fig. 3). In a short time (10 min) after TPA treatment, when the phosphorylation level is still low, long neurite-like extensions are rapidly retracted; instead, numerous membrane ruffles are induced, which can be strongly stained with fascin antibody. Stress fibers are disassembled gradually over 2 h of incubation, which apparently causes rounding of most cells. Concomitantly, fascin staining of stress fibers as well as membrane ruffles disappears, resulting in more diffuse staining in the cytoplasm. The maximum level of phosphorylation roughly corresponds to these alterations of fascin distribution. These observations suggest that fascin phosphorylation may play a role in the dissociation of fascin from stress fibers and membrane ruffles, but does not appear to be involved in the initial rapid retraction of neurite-like extension upon TPA treatment.
In vivo phosphorylation of fascin is substoichiometric (ϳ30%). However, the actin binding experiments with partially phosphorylated fascin (Fig. 9) have shown that the actin binding affinity is reduced to one-fourth of its normal level, suggesting that phosphorylation at this level can have a considerable effect on fascin-actin interactions. Furthermore, there have been some reports that substoichiometric phosphorylation produces biological effects. For example, phosphorylation of only 5 or 20% (depending on the source of smooth muscle) of myosin light chain is sufficient for tension development in smooth muscle (18,19). We have also reported (22) that 15-30% of myosin light chain is phosphorylated at Ser-1/Ser-2 during mitosis, which is suggested to inhibit myosin from premature activation before cytokinesis.
It is interesting to note that phosphorylation of human fascin with PKC depends greatly on pH. The optimal pH of PKC is known to be pH 7.5, and indeed, myosin light chain is better phosphorylated by PKC at pH 7.0 or above. Thus, the pH dependence of fascin phosphorylation should not be attributed to the pH-dependent activity of PKC itself. Rather, it is likely that fascin phosphorylation sites become more accessible to PKC at a lower pH. Lowering the pH to 6.0 alone does not affect the actin binding ability, suggesting that fascin is not denatured at this pH. The observation that actin binding ability is recovered by dephosphorylation with calcineurin (protein phosphatase 2B) treatment confirms that phosphorylation actually regulates actin binding of fascin.
At present, we cannot conclude that PKC is responsible for phosphorylation of fascin in vivo. It is possible that local changes in pH may occur where both fascin and PKC are present. Indeed, some isoforms of PKC are known to be localized in microfilament structures (20). However, it is not known whether pH drops upon TPA treatment. Furthermore, the time course of PKC activation upon TPA treatment does not appear to coincide with the time course of fascin phosphorylation. Further studies are required to identify whether PKC or another kinase (or kinases) is involved in fascin phosphorylation in vivo.
Phosphorylation appears to have a drastic effect on actin binding of fascin. The actin binding assay (Fig. 11) shows that the actin binding constant was decreased 4-fold with the incorporation of 0.25 mol of phosphate/mol of human fascin. However, this actin binding constant was obtained with a mixture of 75% unphosphorylated and 25% phosphorylated human fascin. Densitometry of the Coomassie Blue-stained gel shown in lanes 5 and 6 of Fig. 8 revealed that ϳ72% of fascin binds to actin. On the other hand, densitometry of the autoradiograph of the same lanes showed that only 20% of radioactive fascin binds to actin. Thus, if we were able to use 100% phosphorylated fascin, we would find a much reduced actin binding constant, perhaps Ͼ10 times lower than that of unphosphorylated fascin.
We have recently identified one of the sites of human fascin phosphorylation as Ser-39. 2 This site is well conserved among many fascins, including human, mouse, Xenopus, and Drosophila (sea urchin fascin has a Thr residue instead). Furthermore, the sequence around this site is one of the most conserved domains of these fascins, suggesting its importance in the functions of fascin. We are in the process of generating fascin mutants with the Ser residue replaced by Ala or Asp. It will be very interesting to see whether the expression of such mutants changes the organization of filopodia, membrane ruffles, and stress fibers. Such studies will facilitate the understanding of the physiological significance of phosphorylation of fascin in microfilament organization and cell motility.