Inhibition of the Interactions of Cofilin , Destrin , and Deoxyribonuclease I with Actin by Phosphoinositides *

Cofilin is a widely distributed actin-modulating protein that has the ability to bind along the side of Factin and to depolymerize F-actin in a pa-dependent manner. We found that phosphatidylinositol (PI), phosphatidylinositol4-monophosphate (PIP), and phosphatidylinositol4,5-bisphosphate (PIP2) inhibited both actions of cofilin in a dose-dependent manner, while inositol 1,4,5-triphosphate (IPa), 1-oleoyl-2-acetylglycerol (OAG), phosphatidylserine (PS), or phosphatidylcholine (PC) had little or no effect on them. Gel filtration analyses showed that PIP2 bound to cofilin and thereby inhibited the binding of cofilin to G-actin. Destrin is a mammalian, pH-independent actin-depolymerizing protein. The actin-depolymerizing activity of destrin was also inhibited by PI, PIP, and PIP2, but not by IP3, OAG, PS, or PC. In addition, we found further that an actin-depolymerizing activity of bovine pancreas deoxyribonuclease I, a G-actin-sequestering protein, was inhibited by PIP and PIP2, but not by PI, IP3, OAG, PS, or PC. These results together with previous findings (Lassing, I., and Lindberg, U. (1985) Nature 314, 472-474; Janmey, P. A., and Stossel, T. P. (1987) Nature 325, 362-364) suggest that the sensitivity to polyphosphoinositides may be a common feature in vitro among actin-binding proteins that can bind to G-actin and regulate the state of actin polymerization.

Cofilin is a widely distributed actin-modulating protein that has the ability to bind along the side of Factin and to depolymerize F-actin in a pa-dependent manner.
The actin-depolymerizing activity of destrin was also inhibited by PI, PIP, and PIP2, but not by IP3, OAG, PS, or PC. In addition, we found further that an actin-depolymerizing activity of bovine pancreas deoxyribonuclease I, a G-actin-sequestering protein, was inhibited by PIP and PIP2, but not by PI, IP3, OAG, PS, or PC. These results together with previous findings (Lassing, I., and Lindberg, U. (1985) Nature 314, 472-474; Janmey, P. A., and Stossel, T. P. (1987) Nature 325, 362-364) suggest that the sensitivity to polyphosphoinositides may be a common feature in vitro among actin-binding proteins that can bind to G-actin and regulate the state of actin polymerization.
Cofilin is a widely distributed actin-binding protein with an apparent molecular mass of 21 kDa on SDS'-polyacrylamide gel electrophoresis (l-5). It has the ability to bind to both G-and F-actin under physiological conditions. Thus, cofilin not only binds along F-actin with a 1:l molar ratio of cofilin to actin monomer, but also binds to G-actin in the same 1:l molar ratio (1 sequence homologous to that of gelsolin, villin, fragmin, severin, and Acanthamoeba profilin in its C-terminal portion (7). Destrin is a mammalian actin-depolymerizing protein with an apparent molecular mass of 19 kDa on SDS-polyacrylamide gel electrophoresis (8-10). Similar low molecular mass (15-19 kDa) actin-binding proteins have been identified in other organisms: actin-depolymerizing factor in chick embryo brain (11,12), depactin in echinodermatous oocytes and eggs (13), and actophorin in Acanthamoeba (14). These proteins can rapidly depolymerize F-actin in a pH-independent manner by taking actin molecules away from the entire length of F-actin. Recent cDNA cloning studies have shown that the amino acid sequence of destrin, consisting of 165 amino acid residues, is very similar (71% identical) to that of cofilin which consists of 166 amino acid residues (7, 15).
It has been reported that polyphosphoinositides such as phosphatidylinositol4,5-bisphosphate (PIPJ and phosphatidylinositol 4-monophosphate (PIP) cause the dissociation of profilactin by interacting with profilin in uitro (16). The function of Ca'+-sensitive actin-severing proteins, gelsolin and villin, is also modulated by PIP, and PIP in uitro. These polyphosphoinositides can inhibit their actin-severing activities and induce dissociation of gelsolin-actin complex (17)(18)(19). Recently, Hartwig et al. (20) have suggested the possibility of the interaction of gelsolin with polyphosphoinositides in vivo, on the basis of their observation that part of gelsolin localizes with plasma and intracellular membranes. It is interesting to examine whether phosphoinositides affect the interaction of cofilin or destrin with actin, because cofilin and destrin belong to another class of actin-binding proteins with gelsolin, villin, and profilin.
In this paper, we show that phosphoinositides inhibit the binding of cofilin to F-actin, the actin-depolymerizing activity of cofilin and destrin, and the actin-depolymerizing activity of DNase I, a Gactin-sequestering protein. We further demonstrate the direct binding of PIP, to cofilin.

MATERIALS AND METHODS
Proteins-Actin was prepared from rabbit skeletal muscle by the method of Spudich and Watt (21) and further purified by Sephadex G-100 gel filtration column equilibrated with 2 mM Hepes, 0.1 mM CaCl*, 0.01% NaNa, 0.2 mM ATP, and 0.05 mM DTT, pH 7.9. Porcine brain cofilin was purified by the previously described method and was indistinguishable from porcine brain cofilin in its interaction with actin (Ref. 15, see below). Porcine brain cofilin and recombinant cofilin bound to F-actin in the same dose-dependent manner at pH 7.0, and the saturation levels of both cofilins were about 0.7-0.&l molar ratio of bound cofilin to actin molecules in F-actin under the assay conditions of Fig. 1. When increasing concentrations of recombinant cofilin were reacted with F-actin in the presence of a fixed concentration of brain cofilin, the molar ratio of recombinant cofilin to brain cofilin in the F-actin pellet was nearly the same as that in the initial reaction mixture (Table I). These results suggest that both cofilins have nearly the same affinity for F-actin. AS shown in Fig. lA, the amount of porcine brain cofilin bound to F-actin was decreased by PIP2 in a dose-dependent manner. In the presence of a 5O:l molar ratio of PIP, to cofilin, cofilin could not bind to F-actin at all. The binding of recombinant cofilin to F-actin was also inhibited by PIP2 in almost the same dose-dependent manner (Fig. 1B). We further investigated the effects of other phospholipids and IPs on the binding of recombinant cofilin to F-actin. PIP had almost the same inhibitory effect as PIPB, and PI had a slightly weaker inhibitory effect than PIP and PIP,. On the other hand, PC, PS, OAG, and IP, did not inhibit the cofilin  binding to F-actin at all (Fig. 1B). It should be noted that none of the phospholipids and IP3 affected the polymerization level of actin in the absence of cofilin.
It is known that an increase in the fluorescence intensity of pyrene labeled to actin accompanies polymerization of actin (25). When F-actin (final concentration 4.0 PM) containing 1% pyrene-labeled actin was mixed with recombinant cofilin (final concentration 4.1 PM) at pH 8.3, the fluorescence intensity of the pyrene was decreased to nearly zero because of the binding of cofilin to actin, as previously reported (1). Prior incubation of cofilin with PIP, (final concentration 360 pM) inhibited this cofilin-induced decrease of fluorescence intensity almost completely (data not shown). This result also suggests that the binding of cofilin to F-actin is inhibited by PIP,. When PIP, (final concentration 850 pM) was added to the mixture of F-actin (final concentration 4.0 pM) and cofilin (final concentration 4.1 ELM) at 37 "C, the fluorescence inten-Inhibition of Cofilin and De&in by Phosphoinositides sity was gradually increased from nearly the zero level to that of F-actin in the absence of cofilin within 30 min (data not shown). This result indicates that dissociation of cofilin from F-actin occurs by the action of PIP2.

Effect of Polyphosphoinositides on the Actin-depolymerizing
Actiuity of Cofilin-Porcine brain cofilin has a pH-dependent actin-depolymerizing activity (6). Recombinant cofdin also depolymerized F-actin in a pH-dependent manner (15). The actin-depolymerizing activity at pH 8.3 and the relatively weak actin-depolymerizing activity at pH 7.0 of both brain cofilin (Fig. W) and recombinant cofilin (Fig. 2, B and C) were reduced to the half level in the presence of about a 2O:l molar ratio of PIP2 to cotilin and were almost completely inhibited in the presence of 50-100 molar ratio of PIP, to cofilin. PIP and PI also inhibited the actin-depolymerizing activity of cofilin (Fig. 2, B and C) in a concentrationdependent manner. But, PC, PS, OAG, and IP3 had little or no inhibitory effect on it (Fig. 2, B and C). It has been shown that incubation of PIP, with divalent cations prior to mixing with gelsolin decreases the ability of PIP, to inhibit the actin-severing function of gelsolin (18). We have then examined the effect of Mg2+ ions on the action of PIP2, PIP, and PI on cotilin. Prior to the mixing with recombinant cofilin, PIPp, PIP, or PI was incubated in the solution containing 2 mM MgCl* at 25 "C for 20 min. In this case, the inhibitory effects of PIPz, PIP, or PI on the cofilin binding to F-actin and on the actin-depolymerizing activity of cofilin were decreased by 80%. On the other hand, addition of Mg" after mixing PIP2 with cofilin caused no decrease in the inhibitory effects of PIP2 (data not shown). These results are consistent with the previous report examining the effect

Interaction of PIP, with Cofilin-We
have examined whether PIP2 binds to cofilin or actin by gel filtration chromatography. Sephadex G-100 gel filtration column chromatographies were carried out in the presence of 100 mM NaCl. PIP2 forms small micelles of molecular mass 93,000 Da (26) and was eluted near the void volume (Fig. 3A, -).
When a low concentration of G-actin with (0) or without (0) PIP2 was passed through the column, actin was eluted at the position corresponding to a monomer in both cases (Fig. 3A), indicating that PIP, does not bind to actin under the conditions used. In contrast, the elution position of cofilin changed when cofilin mixed with PIP, was passed through the column. Almost all the cofilin was co-eluted with PIP2 near the void volume (Fig. 3B, A). This clearly indicates that PIP2 binds to cofilin.
Next, a mixture of cofilin and actin (2:l molar ratio) without (Fig. 3C) or with (Fig. 30) PIP, was passed through the column. In Fig. 3C, elution profiles of actin (0)  the Sepbadex G-100 column and the elution profiles of cofilin (A) and actin (0) were determined.
We confirmed by actin polymerization assays with UV measurements as well as fluorescence measurements of pyrene-labeled actin that under the same conditions used here no sign of polymerization of actin (-1 pM) was detected irrespective of the presence or absence of cofilin or PIP* within 1 h. Each run of the gel filtration column chromatography was also completed within 1 b. De&in (final concentration 2.9 PM) was incubated with various concentrations of each lipid for 10 min at 25 "C. The molar ratio of each lipid to destrin is located in the figure (absci.ssa). F-actin (final concentration 2.5 pM) was added to the mixtures. After the incubation for 30 min at 25 "C, the mixtures were centrifuged. The amounts of nonsedimented actin were determined as described under "Materials and Methods." Buffer conditions were 20 mM Hepes, 4 mM Pipes, 70 mM KCl, 0.01 mM CaC12, and 0.02 mM ATP, pH 7.0. The amount of nonsedimented actin in the absence of phospholipids or IP, was -1.5 PM, and this was regarded as 100%. 0, 0, X, n , 0, A, and A indicate PIP,, PIP, PI, IP,, OAG, PC, and PS, respectively.
25% of the cofilin was eluted at the position of actin-cofilin complex. Thus, the molar ratio of cofilin to actin at the position of the actin-cofilin complex was about 0.6. This relatively low ratio may result from dissociation of cofilin from actin during the elution because of the low concentration of actin and cofilin. When a mixture of cofilin and actin with PIP? (5O:l molar ratio to cofilin) was passed through the column, cofilin was co-eluted with PIPs near the void volume (Fig. 30, A) and almost all actin eluted at the position corresponding to a monomer (Fig. 30, O).' These results clearly indicate that PIP2 binds to cofilin and thereby inhibits the interaction of cofilin with actin. Effect of Phosphoinositides on the Actin-depolymerizing Activity of Destrin-Destrin is a mammalian, pH-and Ca*+independent actin depolymerizing protein (8,9), whose amino acid sequence has been revealed to be very similar to that of cofilin (7, 15). Therefore, we have asked whether the activity of destrin is also regulated by phosphoinositides. Fig. 4 shows that PIP2 (0) and PIP (0) inhibited the ability of destrin to depolymerize F-actin in nearly the same dose-dependent manner. The half-maximal inhibitory effect was seen at about a 1O:l molar ratio of PIP or PIP2 to destrin. Therefore, their inhibitory effect on the activity of destrin is slightly stronger than that on cofilin's activity. PI showed a slightly weaker inhibitory effect on the activity of destrin than PIP and PIP*, but PS, PC, IP.?, and OAG had little or no inhibitory effects (Fig. 4).
Effect of Polyphosphoinositides on the Actin-depolymerizing Activity of DNase I-We found that the nuclease activity of ' The peak position of actin eluted in the experiment shown in Fig.  30 (0) was slightly shifted from that of actin monomer shown in Fig.  3A (O), indicating some weak interactions of actin with cofilin and/ or PIP, under the conditions used. Part of cofilin might be complexed with actin under the conditions, because the actin-depolymerizing activity of cofilin was not inhibited completely in the presence of about a 5O:l molar ratio of PIP2 to cotilin, as shown in Fig. 2. Moreover, our preliminary experiment showed that in the absence of NaCl, i.e. under the conditions of low ionic strength, actin was coeluted with PIP, near the void volume. Therefore, even in the presence of 100 mM NaCl a weak interaction might exist between actin and PIP?. This may partly account for a very slight shift of elution of actin in the presence of PIP?, as was observed in Fig. 3A (0 and 0) and Fig. 30 (0).

DNase I was inhibited
by PIP*. The inhibition by PIP2 was dose-dependent and about an 8O:l molar ratio of PIP2 to DNase I was required for the half-maximal inhibition (data not shown). It has been shown that DNase I can induce a relatively slow depolymerization of F-actin by sequestering monomeric actins which are in equilibrium with F-actin (27-29). Then, we have examined the effect of phosphoinositides on the actin-depolymerizing activity of DNase I. PIP2 and PIP inhibited the actin-depolymerizing activity of DNase I in a dose-dependent manner and the half-maximal inhibitions were obtained at about 25:l and 5O:l molar ratios of PIP2 and PIP to DNase I, respectively (Fig. 5). These inhibitory effects of PIP, and PIP on DNase I were slightly weaker than those on cofilin and destrin. Although PI inhibited the actions of cofilin and destrin (Figs. 1, 2, and 4), it did not inhibit the actin-depolymerizing activity of DNase I (Fig. 5, X). PS, PC, IP3, and OAG had little or no effect on the action of DNase I (Fig. 5) as in the case of cofilin or destrin. Thus, the sensitivity of DNase I to phosphoinositides is somewhat different from that of cofilin or destrin.

DISCUSSION
It has already been reported that polyphosphoinositides affect the functions of a G-actin-sequestering protein, profilin (16), and Ca*'-sensitive actin-severing/capping proteins, gelsolin (17-19) and villin (19) in uitro. In this study, we have found that besides these actin-regulatory proteins, an F-actinside-binding and -depolymerizing protein, cofilin, and actindepolymerizing/severing protein, destrin, and a G-actin sequestering protein, DNase I, are sensitive to polyphosphoinositides.
Gel filtration column chromatography on Sephadex G-100 revealed that PIP, binds to cotilin but not to actin. It has also been reported that PIP2 binds to profilin (16) and gelsolin (18) in uitro. As PIP2 inhibits the functions of destrin, DNase I and villin, PIP2 may bind to these proteins, too. Thus, at least these six kinds of actin-binding proteins have the ability to bind to PIP,. It is interesting that the cross-linking site in the actin sequence for DNase I, the segment of 48-82 amino acid residues (31), is different from the sites for gelsolin and cofilin, which have been determined to be the N-terminal and/or C-terminal portions of actin (3,30). Furthermore, these actin-binding proteins are clearly different from one another in their mode of interaction with actin as described above. Therefore, despite the common feature of the sensitivity to PIPz, the molecular mechanism underlying the interaction with actin may not be common among these proteins.
Previous studies using gelsolin fragments generated by limited proteolysis (32) or by gene truncation of plasma gelsolin cDNA (33) revealed that the PIP2 binding site on the gelsolin sequence is on an 11-amino acid sequence (150-160 amino acid residues). It is also suggested that this 11-amino acid sequence may be the F-actin-side-binding site (33). Neither cofilin nor destrin has a sequence similar to the 11-amino acid sequence. Therefore, we do not know at present whether a consensus motif for the phosphoinositide-binding sites of these actin-binding proteins exists. Both PIP2 and PIP inhibit the function of gelsolin, villin, profilin, cofilin, destrin, and DNase I in uitro. PI has an inhibitory effect on cofilin (Figs. 1 and 2) and destrin ( Fig. 4) but has little or no effect on gelsolin (18), profilin (IS), and DNase I (Fig. 5). The effect of PI on the function of villin has not yet been reported. Thus, there is a slight but clear difference in the sensitivity to different phosphoinositides between these actin-binding proteins. Cofilin and destrin have distinct but similar actin-modulating functions in vitro and their amino acid sequences are highly homologous (15). This study has shown that, in addition to these similarities, cofilin and destrin share the common feature of the sensitivity to PIP*, PIP, and PI.
It has not been determined whether the function of other actin-binding proteins, such as a-actinin, filamin, and tropomyosin, is inhibited by phosphoinositides, although it was reported that a-actinin forms a complex with diacylglycerol and palmitic acid (34). It can be concluded, however, that a sensitivity to polyphosphoinositides in vitro is a common feature at least among the actin-binding proteins that bind to G-actin and regulate the polymerization state of actin.