Background of the Invention

This invention was made with government support under NIH R01-AR 38910 awarded by the National Institutes of Health. The U.S. Government has certain rights in this invention.

Field of the Invention

The present invention relates to the field of cell biology. Specifically the present invention relates to agents which are capable of binding to the gelsolin binding site on phosphoinositides (PPI) or the PPI binding site on gelsolin and thus inhibiting PPI/gelsolin and PPI/enzyme interactions.

Description of the Related Art

Gelsolin and villin are actin filament severing and capping proteins which are activated by Ca ²⁺ and have profound effects on actin filament organization and assembly. The polyphosphoinositides (PPI) phosphatidylinositol 4-monophosphate (PIP) and phosphatidylinositol 4,5- diphosphate (PIP ₂ ) inactivate the actin filament-severing proteins villin and gelsolin and dissociate them from monomeric and polymeric actin.

Severing of actin filaments and nucleation of actin polymerization are probably essential for the remodeling of cortical actin networks that accompanies nearly all types of cell activation (Stossel, T.P., J. Biol. Chem. 264: 18261-4 (1989)). The family of proteins structurally related to gelsolin is likely to regulate, in part, transformations of the cytoplasmic actin network because these proteins exert powerful effects on actin filament length and initiate filament formation from actin monomers (Matsudaira and Janmey, Cell

54: 139-40 (1988); Yin, H.L. , Bioessays 7: 176-9 (1987)). Differential

activation of severing and nucleating activities in response to changes in the concentrations of Ca ⁺ (Yin, H.L., Nature 287:583-586 (1979)) and polyphosphoinositides (Janmey and Stossel, Nature 325:362-4 (1987)), which are often immediate consequences of cell stimulation, place gelsolin and its homologs directly in the pathway between receptor activation and cytoskeletal remodelling (Stossel, T.P., J. Biol. Chem. 264: 18261-4 (1989)).

In vitro studies have shown that the greatest effect of PPIs on gelsolin (Janmey et al., J. Biol. Chem. 262: 12228-36 (1987); Janmey and Stossel, Nature 525:362-4 (1987)) and the homologous proteins villin (Janmey and Matsudaira, J. Biol. Chem. 265: 16738-43 (1988)) and severin (Yin et al. ,

FEBS Lett. 264:78-80 (1990)) is to inhibit their ability to sever (cleave) actin filaments. At higher concentrations PPIs also inhibit nucleation (Janmey and Stossel, Nature 525:362-4 (1987)) and dissociate complexes of gelsolin and monomeric actin (Janmey et al. , J. Biol. Chem. 262: 12228-36 (1987); Janmey and Stossel, Nature 525:362-4 (1987)). Gelsolin is a multidomain protein with at least three actin-binding sites, and its various effects on actin are thought to result from activation of different sites, alone or in combination. The differential inhibition of severing, nucleation, and actin monomer binding functions of gelsolin by a range of PPI concentrations suggests that there may be multiple PPI-binding sites as well as multiple actin-binding sites, and that the actin-binding site required for severing is the most sensitive to PPIs. Studies of proteolytically derived fragments (Yin et al. , J. Cell Biol. 706:805- 12 (1988)) and of deletion mutants expressed by transfected COS cells (Kwiatkowski et al. , J. Cell. Biol. 70S: 1717-26 (1989)) have shown PPI sensitivity within a fragment derived from the N-terminal 20% of the intact protein (residues 1-160; domain I and the first 10 residues of domain II) (Andre et al. , J. Biol. Chem. 265:722-7 (1988); Bazari et al. , Proc. Natl. Acad. Sci. USA £5:4986-90 (1988); Kwiatkowski et al. , Nature 525:455-8 (1986); Way and Weeds, J. Mol. Biol. 205: 1127-33 (1988)). This fragment also contains the actin filament severing activity of the parent molecule.

Deletion of residues 150-160 eliminates severing activity, but the effect on PPI

binding could not be determined prior to development of the functional assay disclosed herein is not known, since functional assays based on severing were previously unknown (Kwiatkowski et al. , Nature 525:455-8 (1989)).

There is increasing evidence that membrane polyphosphoinositides (PPI), ¹ which are precursors for intracellular second messengers, may also have direct regulatory functions. PIP and PIP ₂ modulate the activity of a number of actin regulatory proteins (Lassing and Lindberg, Nature 574(6010):472-474 (1985); Janmey and Stossel, Nature 325:362-364 (1987); Yin, H.L., et al. , FEBS 264:78-80 (1990); Yu, F.-X., et al. , Science 250: 1413-1415 (1990); Maekawa and Sakai, J. Biol. Chem. 265: 10940-10942

(1990); Yonezawa, N. , et al. , J. Biol. Chem. 265:8382-8386 (1990); Isenberg, G., J. Muscle Res. Cell Motil. 742:4319-4327 (1991)). Among these, the highly specific interactions of PPI with the gelsolin family of proteins (Janmey and Stossel, Nature 525:362-364 (1987); Yu, F.-X., et al. , Science 250: 1413- 1415 (1990); Janmey and Matsudaira, J. Biol. Chem. 265: 16738-16743

(1988); Janmey, P.A., et al. , J. Biol. Chem. 262: 12228-12236 (1987)) and profilin (Lassing and Lindberg, Nature 574(6010):472-474 (1985); Goldschmidt-Clermont, P.J., et al., Science 251: 1575-157 ^' (1990)) have been most well characterized. Unlike the majority of lipid binding actin regulatory proteins (Isenberg, G., J. Muscle Res. Cell Motil. 742:4319-4327 (1991)), gelsolin and profilin bind preferentially to PPI, have much reduced affinity for phosphatidylinositol, and are not inhibited by phosphatidylserine, -ethanolamine or -choline (Lassing and Lindberg, Nature 574(6010): 472-474 (1985); Janmey and Stossel, Nature 525:362-364 (1987); Goldschmidt- Clermont, P.J. , et al. , Science 257: 1231-1233 (1991)). PPI inhibit gelsolin and profilin binding to actin, and these proteins may in turn modulate PIP ₂ metabolism by sequestering/releasing PIP ₂ (Goldschmidt-Clermont, P. J. , et al. , Science 257: 1575-1578 (1990); Goldschmidt-Clermont, P.J., et al. , Science

^J The abbreviations used are: PPI, polyphosphoinositides, PIP ₂ , phosphatidylinositol 4,5-bisphosphate; SDS, sodium dodecyl sulfate; PAGE, polyacrylamide gel electrophoresis.

257: 1231-1233 (1991)). Following agonist stimulation, gelsolin (Hartwig, J.H., et al , J. Cell Biol. 108:467-479 (1989)) and profilin (Hartwig, J.H. , et al , J. Cell Biol 709: 1571-1579 (1989)) translocate towards the plasma membrane and can mediate agonist-induced responses at the membrane- cytoskeletal interface. Both proteins bind multiple PPI cooperatively (Janmey and Stossel, J. Biol Chem. 264:4825-4831 (1989); Machesky, L.M. , et al , Cell Reg. 7:937-950 (1990)) to form high affinity complexes. Although binding is likely to involve electrostatic interactions, the specificity for PPI and not other charged molecules (Lassing and Lindberg, Nature 574(6010): 472-474 (1985); Janmey and Stossel, Nature 525:362-364 (1987); Goldschmidt-

Clermont, P.J., et al , Science 257: 1575-1578 (1990)) suggests that binding requires a particular geometry for the distribution of the phosphoinositide head groups relative to that of basic residues on the proteins. The PPI-binding site(s) of gelsolin is located in the N-terminal half of gelsolin which severs actin filaments in a PPI-regulated manner (Janmey, P. A., et al , J. Biol.

Chem. 262: 12228-12236 (1987); Chaponnier, C, et al , J. Cell Biol. 705: 1473-1481 (1986); Bryan, J. , J. Cell Biol. 706: 1553-1562 (1988); Kwiatkowski, D.P., et al , J. Biol. Chem. 260: 15232-15238 (1985)). This half contains three semi -conserved repeating domains (Kwiatkowski, D.P. , et al. , J. Cell Biol. 108: 1717-1726 (1989)) which appear to act cooperatively to cause severing (Chaponnier, C, et al. , J. Cell Biol. 705: 1473-1481 (1986); Bryan, J., J. Cell Biol. 706: 1553-1562 (1988); Kwiatkowski, D.P., et al , J. Biol. Chem. 260: 15232-15238 (1985); Way, M., et al , J. Cell Biol. 709:593- 605 (1989); Way, M., et al. , EMBOJ. 9:4103-4109 (1990); Yin, H.L., et al, J. Cell Biol. 706:805-812 (1988); Yu, F.-X. , et al , J. Biol. Chem.

266: 19269-19275 (1991)): gelsolin first attaches laterally to actin filaments through an actin binding site located within its domains II-III ((human plasma gelsolin (GS), residues 150-406)) and then breaks the acti actin bond via another actin binding site located in domain I (GS 1-149; this will be referred to as GS149) (Fig. 12A).

Previously, we have found that PPI prevented domains II-III from binding to actin filaments and therefore hypothesized that PPI inhibits severing through the side-binding domain (Yin, H.L., et al , J. Cell Biol 706:805-812 (1988)). This was supported by the findings that a truncated gelsolin (GS160) which contains domain I and 10 amino acids from domain II severed actin filaments in a PPI-regulated manner (Kwiatkowski, D.P., et al , J. Cell Biol. 70S: 1717-1726 (1989)), and synthetic peptides derived from the amino- terminal residues of domain II of gelsolin and villin, a related protein (Matsudaira and Janmey, Cell 54: 139-140 (1988)), bound PPI vesicles and micelles (Janmey, P.A. , et al , J. Biol. Chem. 267(17): 11818-11823 (1992)).

These peptides were more potent than several other unrelated basic peptides with more negatively charged residues, suggesting that in addition to charge, specific aspects of the primary and secondary structure of the gelsolin/villin domain II peptides are important for their interaction with the acidic headgroups of PPIs.

Summary of the Invention

The present study investigated the PPI and F-actin binding properties of synthetic peptides based on the sequences of villin and gelsolin within the domains thought to account for PPI regulation of the parent molecules. Direct binding of peptides and their fluorescently labeled derivatives was assessed by light scattering and fluorescence polarization. Competition between the peptides and native gelsolin for PPIs was measured using functional assays for the actin filament severing activity of gelsolin.

Additionally, we disclose herein that domain I of gelsolin is also inhibited by PPI. Its PPI-binding site is distinct from the actin binding site.

PPI binding is localized to between residues 135-149 by deletional mutagenesis and competition with a synthetic peptide. The existence of multiple PPI- binding sites as well as actin binding sites allows stringent and differential regulation of gelsolin activity in response to changes in PPI concentrations.

The present invention is based on the discovery that short peptide sequences from gelsolin are capable of modulating the interaction of PPI with gelsolin as well as modulatating the interconversion of PPI to various isomers and phosphorylation states. Based on these observation, agents are described which are capable of binding to the gelsolin binding site of PPI or the PPI binding site of gelsolin.

Specifically, the present invention discloses peptides with the following amino acid sequences:

CKSGLKYKKGGVASGF (Seq. ID No. 1 corresponding to GS134- 149, hereinafter "SI "), and

KHVVPNEVVVGKLFQVKGRR (Seq. ID No. 2 corresponding to GS150-169, hereinafter "S2")

These peptides, as well as fragments of these sequences, antibodies capable of binding to these peptides, and anti-idiotypic anti-Si or anti-S2 antibodies are capable of inhibiting the interaction of PPI with gelsolin. By inhibiting PPI/gelsolin interaction, the inhibitory effect of PPI on gelsolin mediated actin cleavage can be modulated. The modulation of PPI/gelsolin interaction can be used to further elucidate cellular responses which cause actin turnover. Further, it has been found that the above agents are capable of inhibiting the enzymes which phosphorylate, dephosphorylate, or isomerize PPI by binding to PPI making them unavailable as an enzyme substrate or as a feedback inhibitor of enzymatic activity.

The present invention further discloses methods of inhibiting the interaction of PPI and gelsolin in vivo. Specifically, PPI/gelsolin interactions in vivo can be modulated by supplying to an individual a pharmaceutically acceptable composition containing one of the agents of the present invention. Alternatively, cells can be engineered through recombinant techniques to inducibly or constitutively express the peptide agents of the present invention. By expressing and producing the peptide agent, a cell can be directed to modulate PPI/gelsolin or PPI/enzyme interaction.

The present invention further discloses methods of identifying agents capable of inhibiting PPI/gelsolin interaction. Specifically, an agent is incubated in a sample with gelsolin, PPI, and pyrene-F-actin and the solution diluted to an actin concentration of 300 μM. Agents capable of inhibiting PPI/gelsolin interaction increase the rate at which the fluorescence of such sample decreases as the actin depolymerizes.

Brief Description of the Figures

Figure 1. Fluorescence polarization of pyrene-maleimide-Iabeled peptides.

Upper panel: The polarization of fluorescence from 3.8 μM villin 133-147 (open circles) labeled by pyrene-maleimide at the C-terminal cysteine is shown after addition of increasing amounts of PIP to a solution in TBS with 1 mM

EGTA. The open circles show the results of a similar measurement using pyrene-labeled ABP-1 peptide.

Lower panel: Fluorescence polarization of villin 133-147 caused by increasing concentrations of F-actin in TBS-EGTA.

Figure 2. Maintenance of gelsolin' s F-actin severing activity by peptides in the presence of inhibitory amounts of PIP ₂ . The ability of peptides to compete with intact gelsolin for PIP ₂ was determined by measuring the ability of gelsolin to sever actin filaments in the presence of sufficient PIP ₂ to cause approximately 90 % inhibition of gelsolin in the absence of competitive binding by other ligands. Experimental details are described in the Examples. Open triangles: villin 84-153; closed triangles: gelsolin 150-169; open circles: villin 140-147; closed circles: villin 133-147; large squares: MARCKS 155-173; open squares: arginine trimers; closed squares: lysine pentamers.

Figure 3. Aggregation of PIP ₂ micelles by villin 84-153. The total light scattering intensity (600 nm light, 90°) of 30 μM micellar PIP ₂ in water after addition of increasing amounts of villin peptide.

Figure 4. Effect of Mg ²⁺ on competition between gelsolin and synthetic peptide for PIP ₂ . The ability of gelsolin 150-169 to compete with gelsolin using methods described in the legend to Figure 2 is compared in the presence and absence of 2 mM MgC12 in solutions containing 150 mM KC1, pH 7.4.

Figure 5. Maintenance of gelsolin's F-actin severing activity by peptides in the presence of inhibitory amounts of phosphatidylcholine (PC)/PIP ₂ vesicles. The ability of peptides to compete with intact gelsolin for PIP ₂ in mixed lipid vesicles containing a 10: 1 PC:PIP ₂ molar ratio was determined by measuring the ability of gelsolin to sever actin filaments in the presence of sufficient PIP ₂ to cause approximately 90% inhibition of gelsolin in the absence of competitive binding by other ligands. Experimental details are described in the Examples. Open triangles: profilin; closed triangles: gelsolin 150-169; closed circles: thymosin- -4; large squares: MARCKS 155-173; open squares: arginine trimers; closed squares: lysine pentamers. The large open circles represent the severing activity of 43 nM gelsolin incubated with various amounts of PC:PIP ₂ vesicles in the absence of peptide with effective PIP ₂ concentrations ranging from 0.5 μM to 5 μM.

Figure 6. Average hydrodynamic diameter of PC/PIP ₂ vesicles in various concentrations of gelsolin or gelsolin 150-169. Increasing concentrations of gelsolin (triangles) or the gelsolin 150-169 peptide (circles) were incubated with 50 μg/ml of either 10: 1 PC/PIP, vesicles (solid symbols) or control PC vesicles (open triangles), and the apparent diameter determined by QLS as described in the Examples. The vertical arrow denotes the peptide concentration at which the effect of the PC/PIP ₂ vesicles on gelsolin is reduced by 90%.

Figure 7. Restoration of filament-severing ability of gelsolin inhibited by polyphosphoinositides. Gelsolin/PIP ₂ (closed circles) or gelsolin/PIP complexes (open circles) incapable of severing actin filaments were incubated

for <5 seconds with gelsolin 150-169 at the molar ratios shown, and reversal of the inhibition measured by the filament-severing assay described in the Examples.

Figure 8. Functional characterization of GS149 and GS134. A. SDS- polyacrylamide gel of purified GS149 and GS134 (lanes 3 and 2, respectively).

2 μg proteins were loaded on each lane. Lane 1, MW standards of 45, 31, 21 and 14 kDa. B. Effect of GS149 and GS134 on the extent of actin polymerization. Actin polymerization was determined by the change in fluorescence of pyrene-iodoacetamide labelled actin. Increasing concentrations of gelsolin polypeptides were added to 5.1 μM pyrene actin in a solution containing 0.15 M KC1, 2 mM MgCl ₂ , 0.2 mM CaCl ₂ , 10 mM tris-HCl, pH 7.5, 0.5 mM ATP and 0.5 mM β-mercaptoethanol and the pyrene actin fluorescence was measured after 16 hr. at room temperature. 0% inhibition indicated that the final level of fluorescence was identical to that of actin in the absence of gelsolin. The per cent inhibition was expressed as a function of the molar ratio of gelsolin polypeptide to actin. Each value was the average of duplicate determinations, and results from several experiments were pooled. C. Effect of PIP on interactions with actin. Increasing concentrations of micellar PIP ₂ were added to solutions containing 5.1 μM pyrene-actin and 2.4 μM GS149 or 2.8 μM GS134 and incubated for 16 hr. In the absence of

PIP ₂ , GS149 and GS134 reduced actin polymerization by half. This was defined as 100% actin binding on the Y-axis. Polymerization in the presence of PIP ₂ was expressed as percent of this value, (closed circles) GS149; (squares) GS134.

Figure 9. Binding of gelsolin polypeptides to PIP ₂ detected by gel filtration chromatography. A. Left panel, elution profiles of 25 μM GS149 in the presence of 0, 35 and 212 μM PIP ₂ micelles. GS149 was incubated with PIP ₂ and chromatographed at room temperature on a Superose 12 FPLC column (Pharmacia). Right panel, elution profiles of 26 μM GS134 in the presence

of 0, 53 and 212 μM PIP ₂ . Absorbance at 280 μ was expressed in arbitrary units. B. Coomassie blue stained gels of gelsolin polypeptides in column fractions in the absence and presence of PIP ₂ . GS149 and GS149 (A139) were incubated with 90 μM PIP ₂ and 300 μl of each column fraction were dried down in a SpeedVac and analyzed by SDS-PAGE. Although the starting materials were relatively pure (Fig. IA and Fig. 6A), the column fractions contained more degradation products probably because they were stored at 4°C for several days before processing for SDS-PAGE. GS134 was incubated with 136 μM PIP ₂ and every other fraction was analyzed by SDS-PAGE. C. The amount of bound gelsolin polypeptide, determined from the decrease in absorbance of the free protein peak, was expressed as a function of PIP ₂ concentration. (Filled and open circles) GS149 at 39 and 25 μM, respectively; (squares) 26 μM GS134; (triangles) 38 μM GS150-406. The line connecting GS134 was fitted to the experimental data. The other line was theoretical, corresponding to a stoichiometry of 1 gelsolin polypeptide per 4 PIP ₂ molecules.

Figure 10. Circular dichroism. Spectra for GS 149 (A) and GS 134 (B) were determined in the absence (solid line) or presence of 77 μM PIP ₂ (dashed line). The GS149 and GS134 concentrations, determined by amino acid analysis, were 10 and 17 μM, respectively.

Figure 11. Synthetic peptide competes with gelsolin for PIP . A. P. competes with GS149 binding to PIP ₂ as determined by gel filtration. Mixtures containing 32 μM GS149 and increasing amounts of P, (a synthetic gelsolin peptide, residues 135-149) were gel filtered in the presence of 110 μM micellar PIP ₂ . In the absence of P,, 83% of total GS149 was complexed with PIP,, and this was defined as 100% PIP ₂ binding. The appearance of additional free GS149 in the presence of Pj is defined as a decrease in PIP binding by GS149. B. P, and GS149 competed with gelsolin for PIP ₂ , restoring actin filament severing by gelsolin. The assay was described in

(Janmey and Stossel, J. Biol. Chem. 264:4825-4831 (1989); Janmey, P.A. , et al , J. Biol. Chem. 267(17): 11818-11823 (1992)). Increasing amounts of P], P ₂ , GS149 or pentalysine were mixed with 10: 1 PC/PIP ₂ vesicles at a concentration of 30 μM total PIP ₂ and added to 0.05 μM gelsolin. The peptide/PIP ₂ ratio was calculated assuming that 50% of the PIP ₂ was on the external face of the bilayer and therefore accessible to the peptide. Severing activity by gelsolin was measured by following the decrease in pyrene fluorescence after dilution of pyrene F-actin to 0.3 μM. In the absence of competing peptides, 30 μM PIP ₂ inhibited severing by 90%. None of the competing molecules sever actin filaments, (closed circles) GS149; (open triangles) \ ^~ (closed triangles) P ₂ ; (closed squares) lysine pentamers. The latter two curves were taken from Fig. 5 of Janmey, P. A., et al , J. Biol. Chem. 267(17): 11818-11823 (1992).

Figure 12. Existence of basic amino acid motifs among PPI binding proteins. A. The structure of the gelsolin N-terminal half (human plasma gelsolin residues 1-406), which contains three semiconserved repeating domains. It severs actin filaments, and severing requires the cooperative interaction between an actin filament side binding site (located in GSII-III) and another actin monomer binding site located in GSI. In this paper, we demonstrated that Pi, which is at the C-terminus of GSI, bound PIP ₂ . P ₂ , the contiguous sequence also bound PIP ₂ (Janmey, P.A., et al , J. Biol. Chem. 267(17): 11818-11823 (1992)), suggesting that gelsolin has at least two PIP ₂ binding sites. B. Comparison of the putative gelsolin domains I and II PPI binding sites with sequences from other PPI-binding actin regulatory proteins. Numbers indicate amino acid residues in human plasma gelsolin (Kwiatkowski ,

D.P., et al , Nature 525:455-458 (1986)), mouse gCap39 (Yu, F.-X., et al , Science 250: 1413-1415 (1990)), chicken villin (Bazari, W.L., et al , Proc. Natl. Acad. Sci. USA 85:4986-4990 (1988)), porcine cofilin (Matsuzaki, F. , et al , J. Biol Chem. 265: 11564-11568 (1988)) and human profilin (Ampe, C. , et al. , FEBS 228: 17-21 (1988)). The number of amino acids (x) between

the first and second basic residues range from 3 to 6. C. Similarity between gelsolin domain I PPI binding motif with a sequence in the conserved "X box" of the rat phospholipase C (PLC) family (Rhee, S.G., et al , Science 244:546- 550 (1989)).

Figure 13. Expression of gelsolin domain I mutants in which a single lysine was substituted by alanine. A. SDS polyacrylamide gel. 3 μg of GS149 (A141), GS149 (A139) and GS149 (lanes 2-4, respectively) were analyzed. Lane 1, M.W. standards of 66, 45, 31, 21 and 14 kDa. B. Comparison of gelsolin domain I interactions with actin and PIP ₂ . 2.7 μM pyrene actin was polymerized in the presence of 1.3 μM gelsolin polypeptide as described in

Fig. 1. a, no PIP ₂ ; b, 14 μM PIP ₂ ; c, 28 μM PIP ₂ .

Brief Description of the Preferred Embodiments

The present invention is based on the novel observation that peptides whose amino acid sequence are depicted in Seq. ID Nos 1 and 2, as well as fragments thereof, are capable of 1) modulating (inhibiting) the interactions of

PPI with gelsolin, and 2) modulating (inhibiting or stimulating) PPI dependent kinases, phosphatases, and isomerases which convert PPI to various isomers and phosphorylation states. Based on these observations, it is now possible to 1) modulate actin severing by modulating gelsolin/PPI interaction, and 2) modulated the conversion of PPI to other forms of PPI by modulating

PPI/enzyme binding.

In one embodiment of the present invention, agents are described which are capable of modulating PPI/gelsolin interaction. By incubating one of the agents of the present invention with a sample containing PPI and gelsolin, PPI/gelsclin interaction can be inhibited. Further, if the sample additionally contains actin, the agents of the present invention will modulate PPI's inhibition of gelsolin mediated actin cleavage.

In one aspect of this embodiment, the agents are peptides whose sequences are depicted in Seq. ID Nos. 1 and 2, or fragments of these sequences. The peptide agents of the present invention are capable of binding to the gelsolin binding site on a PPI.

5. One skilled in the art can readily generate such peptide agents using protein synthesis techniques known in the art. Alternatively, a host can be transformed with DNA sequences capable of constitutively or inducibly expressing the desired peptide in order to produce large quantities of the desired peptide. One skilled in the art can readily adapt, without undue 0 experimentation, any host/vector system and protein purification techniques currently available to express and isolate the peptide agents of the present invention.

Fragments of Seq. ID Nos. 1 and 2 which possess the ability to bind to the gelsolin binding site on PPI can be readily generated using the above- 5 described procedures. One skilled in the art can utilize the actin severing assay disclosed herein as a means of testing and identifying, without undue experimentation, fragments of Seq. ID Nos. 1 and 2 which possess the desired binding characteristics.

In another aspect of this embodiment, the modulating agents are 0 antibodies or peptides (anti-peptide peptides) which are capable of binding to a peptide whose amino acid sequence is depicted in Seq. ID Nos. 1 and 2. The antibody and anti-peptide agents of the present invention are capable of binding to the PPI binding site on gelsolin.

One skilled in the art can readily adapt currently available procedures 5 to generate both monoclonal and polyclonal antibodies capable of binding to a specific peptide sequence in order to generate the antibody agents of the present invention, for example see Harlow et al, " Antibodies," Cold Spring Harbor Press, (1988), and Campbell, A.M., "Monoclonal Antibody Technology: Laboratory Techniques in Biochemistry and Molecular Biology, " 0 Elsevier Science Publishers, Amsterdam, The Netherlands (1984).

One skilled in the art can readily adapt currently available procedures to generate peptides capable of binding to a specific peptide sequence in order to generate the antipeptide peptides of the present invention, for example see Hurby et al., Application of Synthetic Peptides: Antisense Peptides", In Synthetic Peptides, A User's Guide, W.H. Freeman, NY, pp. 289-307 (1992), and Kaspczak et al., Biochemistry 28:9230-8 (1989).

The anti-peptide peptides of the present invention can be generated in one of two fashions. First, the anti-peptide can be generated by replacing the basic amino acid residues found in Seq. ID Nos. 1 and 2 with acid residues, while maintaining hydrophobic and uncharged polar groups. For example, the lysine, arginine, and/or histidine residues found in Seq. ID Nos. 1 or 2 are replaced with aspartic acid or glutamic acid and glutamic acid residues in Seq. ID No. 2 is replaced by lysine, arginine or histidine.

Alternatively, the anti-peptides of the present invention can be generated by synthesizing or expressing the peptides encoded by the antisense strand of DNA which encodes peptides of Seq. ID Nos. 1 or 2. Peptides produced in this fashion are, in general, similar to those discribed above since codons complementary to those coding for basic residues generally code for acidic residues. In another aspect of this embodiment, the agents are anti-idiotypic antibodies capable of binding to the antigen binding sites of the anti-Si or anti- S2 antibodies. The anti-idiotypic antibody agents of the present invention are capable of binding the gelsolin binding site on a PPI.

Anti-idiotypic antibodies can be generated by any of the methods described above using one of the antibodies of the present invention as an immunogen. One skilled in the art can readily adapt known methods in order to generate the anti-idiotypic antibodies of the present invention.

In another embodiment of the present invention, methods are described for 1) modulating the interaction of PPI with gelsolin and 2) modulating the phosphorylation, dephosphorylation and isomerization of PPI.

In detail, PPI/gelsolin interactions can be inhibited by supplying to a sample (comprising PPI and gelsolin) one of the agents of the present invention which is capable of binding to either the gelsolin binding site on a PPI or the PPI binding site on gelsolin. By supplying such an agent, the PPI mediated inhibition of actin cleavage by gelsolin can be modulate by inhibiting

PPI/gelsolin binding.

Further, it has been found that agents capable of binding the gelsolin binding site on PPI modulate, the interconversion of PPI to various isomers and phosphorylation states. Some of the enzymes found to be modulated by the agents of the present invention include, but are not limited to, PI 4-kinase,

PI 3-kinase, PI(4)P 5-kinase, and phospholipase C-γ.

Other enzymes which have been found to be modulated by the agents of the present invention include, but are not limited to other PPI specific phosphatases and other PPI specific kinases. Other reactions which have been found to be modulated by the agents of the present invention include, but are not limited to PI -- > PI(4)P or PI(3)P PI(4)P - > PI(4,5)P ₂ PI(4,5)P ₂ - > PI(3,4,5)P ₃ PI(3)P - > PI(3,4)P ₂

PI(3,4)P ₂ - > PI(3,4,5)P ₃ PIP -- > DAG (diacyl glycerol) + IP ₂ PIP ₂ - > DAG + IP ₃ PIP ₃ --> DAG -I- IP ₄ For a review of PPI nomenclature see Vance, "Phospholipid

Metabolism In Eukaryotes" In Biochemistry And Lipid Membranes, Ed. Vance et al. Benjamin/Cummings, Menlo Park, p. 242-270 (1985).

As used herein, "modulate" is defined as the ability of an agent to stimulate or inhibit an enzymatic reaction or a naturally occuring non- enzymatic interaction. For example, PPI inhibition of gelsolin mediated actin cleavage can be modulated. The agents of the present invention modulated

th is inhibition by decreasing PPI/gelsolin interaction. This decrease in PPI/gelsolin interaction leads to the stimulation of gelsolin mediated actin cleavage.

The type of modulation exerted by the agents of the present invention varies between enzymes which use a PPI as a substrate. For example, an agent can bind PPI, removing it as a substrate and thus inhibiting enzymes which use the particular PPI as a substrate. Alternatively, the agent can bind to PPI and thus stimulate an enzyme which is feedback inhibited by the presence of the PPI. The samples in which the various enzymatic activities can be modulated can either be comprised of living cells or tissues, or can be cell free, comprising components which are either synthesized or derived from living cells.

In samples which contain living cells, the cells themself can be modified to produce the protein agents of the present invention. Specifically, one skilled in the art can readily modify a cell using recombinant techniques such that one of the agents of the present invention is either constitutively or inducibly expressed.

In detail, a cell is transformed with an expression vector capable of directing the expression of an agent of the present invention. Methods for generating such an expression vector and for transforming cells are well known in the art (Sambrook et at., "Molecular Cloning, " Cold Spring Harbor Press (1989). By utilizing an inducible promoter, the expression of the agent can be controlled. Whether or not the sample comprises living cells or tissue, modulation of the various enzymes can be used to further elucidate biochemical pathways as well as serve as a basis of developing treatments for pathological conditions and disease states which are associated with deficiencies or overexpression of the various modulated enzymes. Such condition include, but are not limited to condition which lead to cytoskeletal damage and an inflammatory reaction.

On metastatic potential, for example, overexpression of gelsolin in fibroblasts

is positively correlated with their rate of directed locomotion in response to extracellular signals (Cunningham et al, Science 257: 1233-6 (1991). Since part of the signalling pathway involves PPI's, modulation of gelsolin/PPI interactions in these cells are expected to alter their ability to respond to stimuli and move.

For example, the mechanisms involved in gelsolin/PPI/actin interactions are not well known by using the methods described herein, and one skilled in the art can begin to identify key interactions which are responsible for cellular response to cell damage. By identifying key interactions, drugs can ultimately be designed for various medical treatments which require modulating these interactions.

The present invention further provides methods of treating pathological conditions which are associated with the increased or decrease of PPI/gelsolin or PPI/enzyme interactions. In detail, the agents of the present invention can be formulated using known procedures to obtain pharmaceutically acceptable compositions. In providing a patient with an agent of the present invention, the dosage of administered agent will vary depending upon such factors as the patient's age, weight, height, sex, general medical condition, previous medical history, etc. In general, when the agent is an antibody, a dosage of antibody which is in the range of from about 1 pg/kg to 10 mg/kg (body weight of patient) is preferred. When the agent is a peptide, it is preferable to administer such molecules in a dosage which also ranges from about 1 pg/kg to 10 mg/kg (body weight of patient) although a lower or higher dosage may also be administered.

The agents of the present invention may be administered to patients intravenously, intramuscularly, subcutaneously, enterally, topically or parenterally. When administering the agent by injection, the administration may be by continuous injections, or by single or multiple boluses. The agents of the present invention are intended to be provided to recipient subjects in an amount sufficient to "physiologically effective. " An

amount is said to be physiologically effective if the dosage, route of administration, etc. of the agent are sufficient to modulate gelsolin/PPI interaction or to bind the particular PPI. For example, one of the agents of the present invention is provided to a patient for the intention of modulating gelsolin/PPI interaction is said to be physiologically effective if it is provided in sufficient dosage to suppress PPI inhibition of gelsolin mediated actin cleavage.

Additionally, the agents of the present invention can be administered either alone or in combination with one or more additional agents disclosed herein.

The administration of the agent of the present invention may be for either a "prophylactic" or "therapeutic" purpose. When provided prophylacti- cally, the agents are provided in advance of the onset of the pathological condition, response, or onset of symptoms. The prophylactic administration of the compound(s) serves to prevent or attenuate any subsequent event.

When provided therapeutically, the agent is provided at (or shortly after) the onset of the appearance of a symptom or the diagnosis of a pathological condition. The therapeutic administration of the agent serves to attenuate any actual symptom. A composition is said to be "pharmacologically acceptable" if its administration can be tolerated by a recipient patient. Such an agent is said to be administered in a "therapeutically effective amount" if the amount administered is physiologically significant. An agent is physiologically significant if its presence results in a detectable change in the physiology of a recipient patient.

As described previously, the agents of the present invention can be formulated according to known methods of preparing pharmaceutically useful compositions, whereby these materials, or their functional derivatives, are combined with a pharmaceutically acceptable carrier vehicle. Suitable vehicles and their formulation, inclusive of other human proteins, e.g., human serum albumin, are described, for example, in Remington's Pharmaceutical Sciences

(16th ed. , Osol, A., Ed., Mack, Easton PA (1980)). In order to form a pharmaceutically acceptable composition suitable for effective administration, such compositions will contain an effective amount of an agent of the present invention together with a suitable amount of carrier. Additionally, the antibodies of the present invention may be humanized, through chimerization or CDR grafting, to become more "pharmacologically acceptable" to a patient.

Additional pharmaceutical methods may be employed to control the duration of action. Control release preparations may be achieved through the use of polymers to complex or absorb the agents of the present invention. The rate and duration of the controlled delivery may be regulated to a certain extent by selecting an appropriate macromolecule matrix, by varying the concentration of macromolecules incorporated, as well as the methods of incorporation. Another possible method to control the duration of action by controlled release preparations is to incorporate the agents of the present invention into particles of a polymeric material, such as polyesters, polyamino acids, hydrogels, poly(lactic acid) or ethylene vinyl acetate copolymers. Alternatively, instead of incorporating these agents into polymeric particles, it is possible to entrap these materials in microcapsules prepared, for example, by coacervation techniques or by interfacial polymerization, by gelatine or poly(methylmethacylate) microcapsulation, or in colloidal drug delivery systems, for example, liposomes, albumin microspheres, microemulsions, nanoparticles, and nanocapsules or in macroemulsions.

The invention further includes a pharmaceutical composition comprising one or more agents selected from the group consisting of; (a) a peptide whose sequence is selected from the group consisting of Seq. ID Nos. 1 and 2, or a fragment thereof, (b) an antibody capable of binding to a peptide defined in Seq. ID Nos. 1 or 2, and (c) anti-idiotypic antibodies capable of binding to the antigen binding site of the above antibodies.

The present invention further discloses methods of identifying agents capable of inhibiting PPI/gelsolin interaction. In detail, agents are identified

as being capable of inhibiting PPI/gelsolin interaction by their ability to remove PPI inhibition of gelsolin mediated actin cleavage.

Specifically, an agent is incubated with a sample comprising gelsolin, pyrene-F-actin (preferably diluted to below its critical monomer concentration), and a PPI capable of binding to gelsolin thus inhibiting gelsolin mediated actin cleavage. The sample is then incubated under condition which would allow gelsolin to cleave actin if the reaction was not inhibited by PPI. During incubation, the rate of fluorescence is monitored. The rate of fluorescence change of the sample is compared to the rate of fluorescence of a sample not containing the agent (negative control), and either a sample not containing the PPI or a sample containing one of the peptide agents of the present invention (positive control) . Agents capable of inhibiting PPI/gelsolin interaction lead to a more rapid decrease in fluorescence intensity of the sample and the rate of decrease is proportional to the degree of inhibition. The concentration of the components of the above assay will vary depending on the type and nature of detection employed. In one application of this embodiment, from about 200 to 300 nM pyrene-F actin, from about 33 to 50 nM gelsolin, and from about 10 to 20 μM PIP^ micelles or mixed lipid vesicles containing various amounts of PIP ₂ is employed and fluorescence is measure using a Greg ISS instrument in an L configuration (see example section for details).

One skilled in the art can readily adapt actin labeled in other fluorescent fashions as well as other PIP compositions and means of determining fluorescence for utilization in the above described assay. Utilizing such an assay, one skilled in the art can now identify novel agents, such as peptides and carbohydrates, which may be ultimately useful in treating numerous pathological conditions.

Having now generally described the invention, the agents and methods of obtaining same will be more readily understood through reference to the following examples which are provided by way of illustration, and are not intended to be limiting of the present invention, unless specified.

Examples

Materials and Methods

Proteins and Peptides. Actin (Spudich and Watt, J. Biol. Chem. 246:4866-4871 (1971)) and gelsolin (Chaponnier et al , J. Cell. Biol. 705: 1473-1481 (1986)) were purified by published methods, and actin was labeled with either pyrene-iodoacetamine (Kouyama and Mihashi, Eur. J. Biochem. 774:33-38 (1981)) or pyrene-maleimide (Kawasaki et al, Biochim. Biophys. Ada 446:166-178 (1976)) as previously described. G-actin was prepared by dialysis into solutions containing 2 mM Tris, pH 7.4, 0.2 mM CaCl ₂ , 0.2 mM dithiothreitol, 0.5 mM ATP (buffer A) and polymerized by addition of 2 mM MgCl ₂ and 150 mM KC1 (to form buffer B). Gelsolin was stored frozen in 10 mM Tris, 100 M KC1, 1 mM EGTA, ² pH 7.4 (TBS- EGTA). Thymosin b-4 (Safer et al , J. Biol. Chem. 266:4029-4032 (1991)) was the generous gift of Daniel Safer and Vivianne Nachmias, University of Pennsylvania. Peptides based on the sequences of yeast ABP-1 (Drubin et al. ,

Nature 545:288-90 (1990)) and MARCKS (Graff et al , J. Biol. Chem. 264:21818-23 (1989)) were provided by David King and David Drubin, University of California, and Marcus Thelen and Alan Aderem, Rockefeller University, respectively. A peptide containing residues 150-169 of gelsolin was provided by Blake Gepinsky, Biogen, Cambridge, MA. Peptides based on the corresponding domains of villin were synthesized using t-boc chemistry on an Applied Biosystems model 430A peptide synthesizer. For villin 134-147 a cysteine residue was included at the C terminus. For the remaining villin peptides, a glycine-cysteine sequence was included, with the glycine serving

² The abbreviations used are: EGTA, [ethylenebis(oxyethylene- nitrilo)]tetraacetic acid; PPI, polyphosphoinositide; PIP, phosphatidylinositol 4-monophosphate; PIP ₂ , phosphatidylinositol 4, 5 -bis-phosphate; PC, phosphatidylcholine; MARCKS, myristoylated, alanine-rich kinase C substrate; DLS, dynamic light scattering.

as a spacer residue. The peptides were labeled with pyrene-maleimide for the polarization experiments or with iodoacetamide to prevent disulfide exchange. All peptides were dissolved in TBS-EGTA or water, and their sequences are summarized in Table 1.

_ Phospholipids and Vesicle Preparation. Dioleoyl-L-α- phosphatidylcholine (PC), phosphatidylinositol 4-monophosphate (PIP), and phosphatidylinositol 4,5-bisphosphate (PIP ₂ ) were obtained from Sigma. PIP and PIP ₂ were dissolved by sonication in water as described elsewhere (Janmey and Stossel, J. Biol. Chem. 264:4825-31 (1989)). Vesicles containing PC or PC and PIP ₂ at a 10: 1 molar ratio were prepared by sonication. 1 mg of lyophilized PC was added to 0.1 mg of PIP ₂ in 1 ml of H ₂ O and sonicated for 1 min at maximum power as previously described (Janmey and Stossel, J. Biol. Chem. 264:4825-31 (1989)).

Dynamic Light Scattering. The hydrodynamic diameter of phospholipid vesicles was calculated from the intensity autocorrelation functions measured by dynamic light scattering (DLS) using a Brookhaven Instruments BI30AT apparatus, PC or PC/PIP ₂ (10: 1) were diluted to a concentration of 56 μg/ml in buffer B and centrifuged at 16,000 x g for 15 min to remove dust and aggregates. DLS measurements were made on 1-ml samples in cylindrical scattering tubes at angles from 60 to 120 degrees. The diffusion constant D was calculated from the slope of a plot of the average decay constant, measured by a second order cumulant fit, versus the inverse scattering vector, and the diameter d calculated from the expression

d . = kT

3πηD

where k is Boltzmann's constant, T is the absolute temperature (298 K), and h is the solvent viscosity (0.89 centipose) (Schmitz, K., An introduction to

dynamic light scattering by macromolecules, Academic Press, Boston, p. 449 (1990)).

Fluorescence Polarization. The fluorescence polarization of pyrene- maleimide-labeled peptides was measured using a Greg (Urbana, IL) ISS instrument in an L configuration. A sample containing peptide in the absence of either F-actin or PIP was first measured using buffer as a blank correction.

Either F-actin or freshly sonicated PIP (Janmey and Stossel, /. Biol. Chem.

264:4825-31 (1989)) was added in increments to both the fluorescent sample and the buffer blank. Light scattering from F-actin was a significant contribution (up to 50 % ) to the uncorrected fluorescent polarization of peptides at the highest actin concentrations. Each reported polarization is the average of three to five measurements.

Functional Assay of Gelsolin Severing Activity The ability of gelsolin to sever actin filaments was measured by a fluorescence assay employing pyrene-iodoacetamide-labeled actin. The basis of this assay, described in detail elsewhere (Janmey and Stossel, J. Muscle Res. Cell Motil. 7:446-454 (1986); Kouyama and Mihashi, Eur. J. Biochem. 774:33-38 (1981)), is that the fluorescence of pyrene actin is much greater for polymeric than monomeric actin. A sample of F-actin is diluted below its critical monomer concentration into solutions containing gelsolin with or without various combinations of PPIs and peptides. Since actin filaments depolymerize only from their ends the rate of fluorescence decrease, proportional to the depolymerization rate, depends on the number of ends, and therefore on the number of cuts introduced by gelsolin. Solutions contained between 200 and 300 nM pyrene-F actin, 33-50 nM gelsolin, and 10-20 μM PIP ₂ micelles or mixed lipid vesicles containing various amounts of PIP ₂ . To calculate peptide/PIP ₂ ratios for experiments using PC/PIP ₂ bilayer vesicles, the effective PIP ₂ concentration is 50% of total PIP ₂ , assuming that half of the PIP ₂ is in the inner leaflet of the bilayer and therefore inaccessible to the

peptide. In cases where peptides were also added, the order of addition of the various solutes was buffer B, PIP or PIP ₂ peptide, gelsolin, and F-actin. Control experiments showed that neither PPIs nor any of the peptides by themselves affected the rate of fluorescence decrease using these conditions.

Generation of Truncated Gelsolin Expression Constructs. pLcIIFXGS, a human plasma gelsolin expression vector (kindly provided by Drs. M. Way and A. Weeds (Way, M. , et al , J. Cell Biol. 709:593-605 (1989))) was linearized at the Hindlll site, blunted and ligated with BamHI linkers. The insert was released by BamHI digestion and subcloned into M13mpl8. Termination codons were inserted at specific sites by oligonucleotide-directed mutagenesis (Amersham version 2 system). The cDNA constructs were sequenced to confirm the mutation. The cDNA was cloned into the BamHI site of pet3a, the T7 RNA polymerase directed expression vector of Sturdier et al. (Studier, F.W., et al , Methods Enzymol. 785:60-89 (1990)). We switched from pLcII to pet vector because the latter gave superior expression. The resultant proteins contained a fusion peptide (MASMTGGQQMGRGSIEGRA (Seq. ID No. 3) at their N-terminus. A gelsolin cDNA encoding domains II-III (GS 150-406) was generated by restriction enzyme digestion and site-directed mutagenesis.

Expression and Purification of Truncated Gelsolin. BL21 (DE3) pLysS cells transfected with expression constructs were grown in LB medium containing 100-200 μg/ml ampicillin at 37°C to an OD^ nm of 0.6, and induced by 0.4 mM isopropyl-1-thio-jS-galactopyranoside as described previously (Yu, F.-X., et al , J. Biol. Chem. 266: 19269-19275 (1991)). Gelsolin polypeptides were obtained by solubilizing inclusion bodies with 8 M urea (no detergents), and purified by sequential anion and cation exchange chromatography (Yu, F.-X., et al , J. Biol. Chem. 266: 19269-19275 (1991)). Protein concentration was determined by the method of Bradford (Bradford, M.M., Ann. Rev. Biochem. 72:248-254 (1976)), and protein purity was

assessed by electrophoresis on 5-20% acrylamide gradient gels in the presence of SDS.

Functional Characterization of Gelsolin Polypeptides

Actin Monomer Binding. 5.1 μM pyrene-iodoacetamide labeled actin (Kouyama and Mihashi, Eur. J. Biochem. 774:33-38 (1981)) was polymerized for 16 hr. at room temperature in a solution containing 0.15 M KC1, 2 mM

MgCl ₂ , 0.2 mM CaCl ₂ , 10 mM Tris-HCl, pH 7.5, 0.5 mM ATP and 0.5 mM β-mercaptoethanol in the presence of increasing amounts of GS149 or GS134.

GS149 prevented actin monomers from polymerizing, resulting in a decrease in pyrene-actin fluorescence intensity compared with actin control (Way, M., et al , EMBO J. 9:4103-4109 (1990); Yu, F.-X. , et al , J. Biol. Chem.

266:19269-19275 (1991)). To study the effect of PIP ₂ on actin monomer binding, 2.4 μM GS149 or 2.8 μM GSI 34 was incubated with PIP ₂ before addition of pyrene actin. PIP (Sigma) micelles were prepared by dissolving PIP ₂ in water to a final concentration of 1 mg/ml, and sonicated for 30s. at maximum power (model W185; Heat Systems Ultrasonics, Inc., Farmingdale,

NY) as described (Janmey and Stossel, J. Biol. Chem. 264:4825-4831 (1989)).

Gel Filtration Assay for PIP ₂ Binding. The assay is similar to that described for studying PIP ₂ binding of profilin (Goldschmidt-Clermont, P.J., et al , Science 257: 1231-1233 (1991); Machesky, L.M., et al , Cell Reg.

7:937-950 (1990)) and CapZ (Heiss and Cooper, Biochemistry 50:8753-8758 (1991)). It is based on the fact that PIP ₂ micelles are large (90 kDa) compared to the gelsolin polypeptides (15 and 28 kDa), and polypeptide/PIP ₂ complexes will elute earlier than the unbound polypeptide. Gelsolin polypeptides were incubated with PIP ₂ micelles for 5 min. at room temperature, and 100 μl of the mixture was chromatographed at room temperature on a Superose 12 HR 10/30 column (FPLC system, Pharmacia) equilibrated with a buffer containing 5 mM Tris-HCl, pH 7.5, 75 mM KC1 and 0.1 mM NaN ₃ . PIP ₂ was not

included in the elution buffer. The elution was at 0.5 ml/min., and 0.5 ml fractions were collected. The elution profile was monitored by absorbance at 280 μm. 300 μl of selected fractions were dried down in a Speedvac, and analyzed by SDS-polyacrylamide gel electrophoresis. The amount of protein bound to PIP was determined from the decrease in the protein absorbance peak, taking into account the contribution of contaminating proteins in the starting material.

Rescue of gelsolin severing activity by competition for PIP ₂ . The basis of this assay was described in detail elsewhere (Janmey, P. A., et al , J. Biol. Chem. 267(17): 11818-11823 (1992)). Actin filament severing by gelsolin was measured by a fluorescence assay using pyrene-iodoacetamide- labeled actin (Janmey and Stossel, Nature 525:362-364 (1987)). F-actin was diluted to below 0.3 μM into solutions containing 0.05 μM gelsolin with or without PIP ₂ /PC vesicles. PIP ₂ inhibits actin severing. Proteins/peptides which bind PIP ₂ will compete with gelsolin for PIP ₂ , restoring severing (Janmey,

P.A., et al , J. Biol. Chem. 267(17): 11818-11823 (1992)). Control experiments showed that none of the proteins/peptides severed actin filaments in the absence of gelsolin. Lipid vesicles of approximately one hundred nm diameter containing PIP ₂ and PC at a 1 : 10 molar ratio were formed by sonication as described (Janmey, P. A., et al , J. Biol. Chem. 267(17): 11818-

11823 (1992)). Gelsolin was prepared from human plasma (Chaponnier, C , et al., 3. Cell Biol. 705: 1473-1481 (1986)).

Synthetic Peptide. P,, a peptide encompassing GSI 35- 149

(CKSGLKYKKGGVASGF, (Seq. ID No. 1, N-terminal cysteine added) was synthesized by standard solid phase methods and purified by g-HPLC using an acetonitrile gradient in 0.1 % trifluoroacetic acid. Its concentration was determined by amino acid analysis.

Circular Dichroism. CD spectra of GS149 and GS134 in the presence or absence of PIP ₂ micelles were measured using an Aviv model 60DS spectrometer at 25 °C in 1 mm pathlength cells. Spectra were scanned at 1 nm intervals for 3 s, and three scans were averaged. The proteins were dialyzed against 5 mM Tris-HCl, pH 7.5, centrifuged and filtered before assay.

Protein concentrations were determined by amino acid analysis.

Example 1

PPI Binding Sites on Gelsolin

Residues 133-147 YNVQRLLHVKGKKNVC (Seq ID No. 4) of villin (Bazari et al. , Proc. Natl. Acad. Sci. USA 85:4986-90 (1988)) (corresponding to 157-172 in gelsolin) which are implicated in binding the sides of F-actin prior to severing (Kwiatkowski et al , J. Cell. Biol. 108: 1717-26 (1989)) also bind to polyphosphoinositides. Figure 1 shows the increase in fluorescence polarization of the pyrene-maleimide-labeled peptide caused by either PIP vesicles (Fig. la) or F-actin (Fig. lb). The saturable increase in polarization caused by either ligand suggests a specific interaction of the peptide with both PIP and actin filaments. Half-maximal polarization of 3.8 μ6M peptide is observed at a PIP concentration of 10 μ6M, approximately equal to that required for half-maximal inhibition of villin or gelsolin' s severing activity (Janmey et al , J. Biol. Chem. 262: 12228-36 (1987); Janmey and Matsudaira,

J. Biol. Chem. 265:16738-43 (1988)). Since the PPI-binding site of gelsolin probably binds to multiple PPI molecules (Janmey and Stossel, Nature 525:362-4 (1987); Janmey and Stossel, J. Biol. Chem. 264:4825-31 (1989)) and since sonicated aqueous PIP preparations contain a mixture of micelles and vesicles, in which for stearic reasons not all PIP headgroups are accessible to ligands in solution, the affinity of the peptides for PIP cannot be accurately estimated. However, assuming that all the PIP headgroups are accessible to the peptide, the apparent stoichiometry is approximately 1 peptide, 5 PIP, and

the dissociation constant is less than micromolar. The specificity of the binding is shown by the lack of polarization seen in another labeled peptide, derived from the actin-binding protein yeast ABP-1 (Drubin et al , Nature 545:288-90 (1990)). PIP also had no effect on the fluorescence polarization of either free pyrene-maleimide-SH or pyrene-labeled G-actin (data not shown). The binding of the labeled peptide to F-actin is relatively weak and appears to be highly dynamic, since this peptide does not cosediment with F- actin during high speed centrifugation unless the peptide is dimerized (M. deArruda and P. Matsudaira, unpublished experiments). Figure 2 shows the ability of various peptides to compete with gelsolin for binding to PPIs as determined by their ability to prevent inhibition of gelsolin' s severing activity by PIP ₂ . In these experiments, PIP ₂ micelles were incubated with peptides for 5 seconds prior to addition of gelsolin and the commencement of the severing assay by adding pyrene-labeled F-actin. The largest peptides encompassing residues 84-153 of villin (open triangles) or residues 150-169 of gelsolin (closed triangles) are the most effective competitors, on a molar basis. The smaller villin peptide of residues 133-147 is slightly less efficient, but still binds with approximately the same molar ratio to P1P2 as does intact gelsolin or villin. The highly basic sequence 140-147 which is not essential for PIP ₂ sensitivity in the homoLogous domain (164-173) of gelsolin, has much less ability to compete with gelsolin for PIP2, although at concentrations nearly equimolar to PIP ₂ , this peptide, like penta-lysine and neomycin, which have similar numbers of free amino groups, prevents the inhibition of gelsolin by PIP ₂ . Similarly, the calmodulin-binding domain of MARCKS, which contains 7 basic residues in a 25 residue span is at least an order of magnitude less effective than the larger gelsolin/villin peptides, even though it has more net positive charge than gelsolin 150-169.

The ability of the larger peptides to prevent PIP ₂ -binding to gelsolin at peptide:PIP ₂ ratios as low as 1: 100 is likely to result partly from the ability of these peptides to aggregate PIP ₂ . Figure 3 shows the effect of the villin 84-

153 peptide on the light scattering intensity of a solution of PIP ₂ micelles. A

1: 15 peptide:PIP ₂ ratio causes an appreciable increase in scattering and at nearly equimolar ratios, which presumably cause a nearly total charge neutralization, results in the formation of insoluble aggregates. The inability of PIP ₂ to inhibit gelsolin in the presence of low molar amount of peptides therefore is likely the result of a perturbation of the micellar structure of PIP ₂ by the peptide, in addition to direct competition between the peptide and gelsolin for individual PIP ₂ headgroups. This effect on the supramolecular arrangement of PPI's is further enhanced by the presence of divalent cations which promote a transformation of PIP ₂ micelles into a presumed hexagonally packed structure (Goldschmidt, C.P.J., et al, Science 247: 1575-8 (1990);

Janmey, P. A., et al , J. Biol. Chem. 262: 12228-36 (1987); Janmey and Stossel, J. Biol. Chem. 264:4825-31 (1989)). This effect of divalent cations is sufficiently slow that it does not interfere with the filament severing assay, but the effect of the poly-basic peptides may be in part due to an acceleration of the rearrangement of PIP ₂ micelles. Consistent with this expectation,

Figure 4 shows that the gelsolin peptide is more effective in preventing gelsolin inhibition by PIP ₂ in the presence than in the absence of Mg ² + . Similarly, more peptide is required to prevent inhibition of gelsolin by PIP, which does not aggregate in Mg ²⁺ , although both PIP and PIP ₂ micelles inhibit gelsolin at approximately equal molar ratios (Janmey, P. A., et al , J. Biol.

Chem. 262:12228-36 (1987)) (data not shown).

In order to distinguish the effects of direct binding of the peptides to PPI's from the effects of PIP ₂ aggregation, small mixed lipid vesicles containing PIP ₂ in PC were prepared by sonication as described in Materials and Methods. Such PC/PIP ₂ vesicles inhibit gelsolin' s severing function as avidly as PIP ₂ micelles (Janmey and Stossel, J. Biol. Chem. 264:4825-31 (1989)), and as Figure 5 shows, the gelsolin peptide was able to prevent inhibition of gelsolin by these PC/PIP ₂ vesicles. Half maximal effect of peptide on PC/PIP ₂ vesicles occurred at a molar ratio of 10 PIP ₂ monomers on the outer leaflet of the bilayer to 1 gelsolin 150-169 peptide. The effect of the gelsolin 150-169 peptide on inhibition of gelsolin by PC/PIP ₂ vesicles is

as great as that of profilin, and at least one order of magnitude greater than that of the MARCKS peptide, thymosin- _/ 3-4, or other basic peptides. Similar results were seen using mixed micelles of PIP ₂ and Triton X-100 (1:2 molar ratio), confirming that the effects of peptide on PPI's do not require pure PPI micelles (data not shown). A comparison between the affinity of the gelsolin

150-169 peptide and intact gelsolin can be made by comparing the effect of the peptide on severing by a constant amount of gelsolin in the presence of PIP ₂ (closed triangles) to the severing activity of a constant amount of gelsolin incubated in various concentrations of PIP ₂ (open circles). The molar ratio of gelsolin to PIP ₂ at which half of the severing activity is lost is equal to the ratio of peptide to PIP ₂ at which half of the severing activity is restored to maximally inhibited gelsolin. This result suggests that the affinity of the gelsolin 150-169 peptide for PIP ₂ is very similar to that of native gelsolin. Moreover, when the bacterially expressed 150-389 fragment (domains II-III) of gelsolin is examined in this competition assay, its apparent affinity for PIP ₂ is also indistinguishable from that of the 150-169 peptide (J. Lamb, F.X. Yu and H.L. Yin, unpublished experiments).

DLS measurements of average vesicle size confirmed that the gelsolin 150-169 peptide did not induce aggregation or other changes in vesicle size under conditions at which it nearly totally prevented their ability to inhibit gelsolin. Molar ratios as high as 1 peptide to 5 PIP, monomers in mixed vesicles had no effect on vesicle size, in contrast to gelsolin itself, which induced changes in apparent vesicle size at a molar ratio of 1 gelsolin to 25 PLP ₂ monomers (Figure 6). This result suggests that the ability of the peptide to prevent the inhibition of gelsolin by PIP ₂ in mixed vesicles is due only to the binding of the peptide to PIP ₂ headgroups and not to aggregation of lipid particles.

The binding of gelsolin to PPI's has previously been shown to be highly cooperative and critically dependent on the physical state of the PPI's. With the exception of high concentrations of detergent (Janmey and Stossel,

Nature 525:362-4 (1987)), all of the agents previously reported to inhibit

gelsolin-PPI binding when they are added first to PPI's (Janmey, P. A., et al. , J. Biol. Chem. 262: 12228-36 (1987); Chaponnier, C, et al, J. Cell. Biol. 705: 1473-1481 (1986)), do not dissociate gelsolin-PPI complexes if these are formed prior to addition of the competitor. In contrast to these PPI-binding

5. molecules, which include neomycin and several different lipids, Figure 7 shows that the gelsolin-derived peptide completely reverses the inhibition of gelsolin by PIP and PIP ₂ micelles. Higher concentrations of the peptide are required to reverse gelsolin-PPI complexes than are required to prevent their formation, consistent with the highly cooperative nature of these protein- 0 phospholipid complexes. The ability of the peptides to restore severing activity to gelsolin that has been complexed to PPIs suggests that such a regulation in vivo would be readily reversed by displacement of the lipid by the appropriate competitive ligand. In no case was any of the peptides themselves found to sever actin filaments or to inhibit their severing by 5 gelsolin.

Based on the above, it has now been demonstrated that the PPI- regulated F-actin severing site resides in a poly-basic region between domains I and II. This has been verified by showing that synthetic peptides based on the sequences of these proteins in this region bind to PPI's and prevent them 0 from inhibiting the actin filament-severing activity of intact gelsolin. These peptides also bind to F-actin, suggesting that the regulation of severing activity by PPI's results from competitive binding of PPI's and F-actin to the same or very closely spaced sites on the severing proteins. The F-actin binding of these peptides, and perhaps of the corresponding sites in the intact proteins, 5 is of only moderate affinity and sufficiently labile so that they do not co- sediment with F-actin, nor inhibit the formation of very high affinity bonds consequent to severing and capping of actin filaments by the multiple actin- binding sites of intact gelsolin or villin.

Although it is likely that basic amino acids are important for the 0 interaction between actin severing proteins and PPI's or F-actin, this interaction is not simply electrostatic, since the effects of the specific peptides

are not mimicked by oligomers of lysine or arginine, or by the MARCKS peptide. Even the actin-binding protein thymosin-/3-4 which contains numerous positive charge clusters is 100 times less effective than gelsolin 150- 169 in competing with gelsolin for PIP ₂ in phospholipid bilayers. Interestingly, large polymers of arginine, lysine or histidine cannot be used to assess the restoration of gelsolin' s severing activity in the presence of PPI's because these poly-basic amino acids are themselves potent inhibitors of either severing or actin depolymerization, perhaps because they either block the site on F-actin to which gelsolin must bind or they form actin bundles which are resistant to severing (unpublished data). The specific villin/gelsolin peptides cause neither inhibition of severing nor actin bundling at the low concentrations needed to inhibit gelsolin-PPI interactions. These properties of basic peptides based of villin and gelsolin contrast with those of the peptide based on the calmodulin and C-kinase substrate site of MARCKS protein, which has also recently been shown to bind F-actin (Hartwig, J., et al ,

Nature 556:618-622 (1992)). The MARCKS peptide is a potent bundler of actin filaments, and MARCKS activity has not been demonstrated to be affected by PPI's.

Several proteins outside the gelsolin group have been identified which also bind both F-actin and PPI's. The list to date includes myosin 1 (Adams and Pollard, Nature 540:565-8 (1989)), calpactin, destrin (Yonezawa, N. , et al , J. Biol. Chem. 265:8382-8386 (1990)), and cofilin (Yonezawa, N., et al , J. Biol. Chem. 265:8382-8386 (1990)). All four of these proteins contain a poly basic domain implicated in their ability to bind to F-actin. The actin monomer binding protein profilin also binds PPI's (Goldschmidt, C.P.J. , et al , Science 247: 1575-8 (1990); Lassing and Lindberg, Nature 574(6010):472-474 (1985); Lassing and Lindberg, J. CellBiochem. 57:255-67 (1988)) and is a highly basic protein with an isoelectric point of 9.3. The importance of basic domains in binding to F-actin is likely the result of the net negative charge on an actin filament due to the exposure of the highly acidic

N-terminus of actin on the surface of the filament (DasGupta, G., et al ,

Biochemistry 29:3319-24 (1990)). An actin- and PIP ₂ -binding 12 amino acid peptide based on the sequence of cofilin has recently been described (Yonezawa, N. , et al , J. Biol. Chem. 266: 17218-21 (1991)). This peptide contains 2 lysine residues but otherwise bears no obvious similarity to gelsolin 150-169. This difference is consistent with the large functional differences between gelsolin and cofilin.

In addition to providing a likely site for binding to acidic phospholipid surfaces, a polybasic domain in the primary structure of a protein appears to promote conformational changes in the rest of the protein when it binds phospholipid. For example, polylysine transforms from a random coil in aqueous solution to a jS-sheet or an α-helix when it binds phosphatidylserine or phosphatidic acid respectively (Fukushima, K., et al , Biophys. Chem. 54:83-90 (1989)). Such a transformation in a region linking two actin binding domains in gelsolin could alter the spatial relationship of these domains sufficiently to perturb their cooperative effect on actin without any of the actin-binding sites being completely eliminated. Indeed, secondary structure prediction algorithms predict a random structure for most of the gelsolin sequence bridging domains I and II, suggesting both that the basic residues in this region have the flexibility to conform to the acidic headgroups in a PPI cluster, and that such binding may strongly perturb the conformation of the protein. Preliminary circular dichroism measurements of the peptide confirm that, as predicted from secondary structure algorithms, its conformation is that of a random coil in aqueous solution at pH 7.0. However, in the presence of PIP ₂ the peptide adopts an almost completely α-helical conformation (Xian, W. , et al. , FASEB J. 6: A87 (1992)). In contrast, the calmodulin-binding site of MARCKS is strongly predicted to be α-helical in the absence of ligand, and the resulting constraints placed on the spacing of the basic residues may decrease their ability to form multiple tight contacts with acidic phospholipids.

The present studies implicate small regions of actin severing proteins as being likely to participate in binding the proteins to polyphosphoinositides.

Synthetic peptides with as little as 20 amino acids bind as avidly as intact

gelsolin or villin to PIP and PIP ₂ . These peptides are therefore useful for determining the structure of phosphoinositide binding sites in other proteins and can serve as potent pharmacologic inhibitors of PPI turnover, due to their ability to sequester PPI's from proteins that potentially bind them in vivo.

Example 2

Gelsolin Domain I is Inhibited by PIP ₂ While a C-terminal Truncated

Mutant is Not

Fig. 8A shows the SDS-polyacrylamide gel recombinant gelsolin domain 1 (GS 1-149, referred to as GS149, lane 3) and a further truncated protein GS134 (GS1-134, lane 2). As shown previously (Kwiatkowski, D.P. , et al , J. Biol. Chem. 260: 15232-15238 (1985); Way, M., et al , EMBO J. 9:4103-4109 (1990); Yu, F.-X., et al , J. Biol. Chem. 266: 19269-19275 (1991); Kwiatkowski, D.P., et al , J. Cell Biol. 708: 1717-1726 (1989)), GS149 inhibited actin polymerization by binding actin monomers stoichiometrically, so maximal inhibition was observed at a 1 : 1 mutant/actin molar ratio (Fig. 8B). The presence of a 17 amino acid fusion peptide at the N-terminus of gelsolin domain I did not interfere with its actin binding activity. GS134 was equally effective, confirming previous reports that gelsolin domain I is functional after truncation to residue 126 (Way, M., et al , EMBO J. 9:4103-4109 (1990); Kwiatkowski, D.P., et al , J. Cell Biol. 108: 1717-

1726 (1989); Way, M., et al , J. Cell. Biol. 776: 1135-1143 (1992)). Fig. 8C shows that PIP ₂ inhibited GS149 binding to actin and half-maximal inhibition of 2.4 μM GS149 was observed at 12 μM PIP ₂ . No inhibition was observed with 100 μM phosphatidylcholine or phosphatidylserine (data not shown), suggesting that the effect of PIP, was specific, as has been shown previously for gelsolin (Janmey and Stossel, Nature 525:362-364 (1987); Janmey and Stossel, J. Biol. Chem. 264:4825-4831 (1989)) and its homolog, gCap39 (Yu, F.-X., et al , Science 250: 1413-1415 (1990)). The dose response is similar

to that required for inhibition of severing (Janmey and Stossel, Nature 325:362-364 (1987); Yin, H.L., et al , J. Cell Biol 706:805-812 (1988)) and actin filament side binding (Yin, H.L., et al , J. Cell Biol 706:805-812 (1988)), suggesting that PIP, regulation of gelsolin severing may be mediated through the domain I site as well.

- In contrast, GS134 was minimally inhibited by 12 μM PIP ₂ . Therefore, the deleted residues 135-149 (KSGLKYKKGGVASGF (Seq. ID No. 1 minus terminal C residue) are required for optimal PIP regulation. Loss of regulation was unlikely to be due to global denaturation, since GSI 34 inhibited actin polymerization as well as GS149. This was supported by the finding by others that an even shorter fragment (GS126) binds actin monomer (Way, M., et al , EMBO J. 9:4103-4109 (1990); Way, M., et al , J. Cell. Biol. 776: 1135-1143 (1992)).

Example 3

Gel Filtration Analysis of Binding to PIP ₂

The abilities of GS149 and GS134 to bind PIP ₂ micelles were compared by gel filtration analyses. This method had been used to demonstrate high affinity PIP binding by profilin (Goldschmidt-Clermont, P.J., et al , Science 257: 1231-1233 (1991); Machesky, L.M., et al , Cell Reg. 7:937-950 (1990)) and CapZ (Heiss and Cooper, Biochemistry 50:8753-8758 (1991)). Fig. 9A, left panel shows that in the absence of PIP,, GS149 eluted with a V _e (elution volume) of 13.9 ml, and the peak was asymmetric. Addition of PIP, changed the protein elution profile. At 35 μM PIP ₂ , the peak height was reduced and the original peak was resolved from a slower migrating peak with of 14.7 ml. At 212 μM PIP ₂ , the largest peak eluted at V _e of 11.3 ml; the original V _e 13.9 ml peak disappeared while the V _e 14.7 ml peak remained unchanged. SDS- polyacrylamide gel analysis (Fig. 9B) shows that after gel filtration, the GS149 fractions contained GS149 and two smaller bands. These bands were most

likely GS149 degradation products; freshly prepared GS149 always contained variable amounts of lower molecular weight bands (Fig. 8A & Fig. 13A) which became more prominent on storage. It is clear from Fig. 9B that PIP, shifted the elution of GS149 but did not alter the elution of the smaller fragments. This suggested that GS149 bound PIP ₂ , while the smaller fragments did not. The lack of protein trailing behind the GS149:micelle complexes was consistent with high PIP, binding affinity, further supporting the conclusion based on efficient inhibition of function by PIP ₂ (Fig. 8C).

The amount of GS149 bound to PIP, was calculated from the decrease in the original GS149 absorbance peak and plotted against PIP ₂ concentration in Fig. 9C. It increased linearly with PIP, concentration until all of the protein was complexed (open circle at 140 μM PIP ₂ ). No difference in the binding of 39 and 25 μM GS149 was observed, as long as saturation was not reached. The slope of the plot gave a ratio of 1 GS149 to 4 PIP ₂ , which is similar to the binding stoichiometry reported for human platelet profilin (139 amino acids and binds 5 PIP ₂ (Machesky, L.M., et al , Cell Reg. 7:937-950 (1990))). A similar curve was also obtained with GS150-406, the actin filament side binding domains II-III of gelsolin (GS 150-406). Although this fragment has been previously shown to be inhibited by PPI (Yin, H.L. , et al , J. Cell Biol 706:805-812 (1988)), our result presented here is the first direct measurement of its binding to PIP,, and establishes that the two PPI-binding domains of gelsolin bound PIP ₂ with comparably high affinity.

In contrast, no GSI 34 binding was detected below 53 μM PIP, (Fig. 9A). At 212 μM PIP ₂ , a faster migrating protein peak (V _e 11.3 ml) appeared, suggesting that some GS134 became associated with PIP ₂ , although to a much lesser extent than GS149. The GS134 preparation contained a contaminant (V _e of 12.6 ml in Fig. 9A) which did not bind PIP,. Fig. 9C shows that 7 μM GS134 bound to 140 μM PIP, whereas equivalent binding of GS149 was observed at 5 times lower PIP, concentration. Thus, deletion of residues 135- 149 decreased PIP binding affinity and reduced PIP, inhibition of actin binding, establishing that these residues are critically important for the high

affinity interaction of gelsolin domain I with PIP,. They are however not required for actin monomer binding, suggesting that the PPI and actin binding sites are located on different residues.

Example 4

PIP ₂ Induces a Conformational Change in GS149 but not GS134

Fig. 10 shows the circular dichroism profiles of GS149 in the absence and presence of PIP,. Computer curve fitting indicated that in the absence of PIP ₂ , GS149 contained about 55 % jS-sheets, 35 % random coil and 10% α- helix. In the presence of PIP,, the profile was consistent with 25 % random coil, 65 % 3-sheet and 10% α-helix. In contrast, PIP, did not induce significant conformational change in GSI 34, confirming that the truncated polypeptide did not bind PIP,. In the absence of PIP ₂ , GSI 34 had a different conformation than GS149. Surprisingly, GS134 had a similar jS-sheet and random coil content as GS149 complexed with PIP,. This may be coincidental, and does not necessarily indicate that GS134 and GS149:PIP, complexes had identical secondary structures. Nonetheless, since the former binds actin while the latter does not. it is not possible to conclude at present that the conformational change in GS149 after PIP, binding accounts for inhibition of actin binding.

Example 5

Domain I Synthetic Peptide Interferes with PIP ₂ Binding

We tested the ability of a synthetic peptide (_ _* ) containing the deleted sequence to compete with gelsolin polypeptides for PIP ₂ . Fig. 11 A shows that

P) inhibited GS149 binding to PIP, micelles in the gel filtration assay. 50% inhibition of binding was observed at a P, peptide/GS149 molar ratio of 0.48,

suggesting that P, bound PIP, with an affinity and stoichiometry comparable to that of the parent GS149. This was confirmed in a functional assay which is based on the ability of PIP, to prevent gelsolin from severing actin filaments and competing peptides to rescue gelsolin from PIP, inhibition (Janmey and Stossel, J. Biol. Chem. 264:4825-4831 (1989); Janmey, P. A., et al , J. Biol.

Chem. 267(17): 11818-11823 (1992)) (Fig. 11B). P, which had no severing activity restored severing by gelsolin in the presence of PIP,/PC vesicles as effectively as GS149, suggesting that P _: and GS149 bound PIP, with comparably high affinities. The severing recovery curves for P] and GS149 were similar to that shown previously for a domain II N-terminal peptide P,

(GS150-169, KHVVPNEVVVQRLFQVKGRR (Seq. ID No. 2), (Janmey, P. A., et al , J. Biol. Chem. 267(17): 11818-11823 (1992))) indicating that the newly identified domain I site bound PPI with as high affinity as domain II. We have shown previously by dynamic light scattering that P ₂ did not aggregate the PIP,:PC mixed vesicles, suggesting that it prevented inhibition of gelsolin by PIP ₂ by binding to PIP,, and not nonspecifically by aggregating them. This was supported by the finding that pentalysine which causes aggregation of acidic lipids was nonetheless much less effective and did not completely restore severing activity even at very high concentrations (Janmey, P.A., et al . J. Biol. Chem. 267(17): 11818-11823 (1992)).

Comparison of P, and P, sequences shows that although they have little overall homology, each has a high concentration of basic residues (Fig. 12A). To determine if all of the basic amino acids in the region defined by P, are important for PPI binding, GS149 mutants in which a single lysine (K) was substituted with alanine (A) were generated by site-directed mutagenesis. Fig.

13A shows an SDS-polyacrylamide gel of GS149 and two mutants, GS149(A141) (the mutated residue is designated in parentheses) and GS149(A139). The mutants inhibited actin polymerization and were regulated by PIP, to a similar extent as the wild type GS149 (Fig. 13B). Therefore, the replacement of a single lysine with alanine had little effect on the ability of

this domain to bind actin or PIP,. That these mutants bound PIP, was confirmed by gel filtration (data not shown).

Discussion

We have identified a new polyphosphoinositide (PPI) binding site on gelsolin. We have shown that gelsolin domain I (residues 1-149) bind PIP, and its interaction with actin is inhibited by PIP, at molar ratios similar to that required to inhibit actin filament severing by gelsolin. Circular dichroism studies showed that PIP, binding reduced the random coil content of gelsolin domain I and increased the proportion of β-sheet. C-terminal truncation studies showed that residues 135-149 were important for GS149 interaction with PIP, because GS134 which is missing this sequence was no longer PIP, sensitive and did not bind PIP,. P,, a synthetic peptide containing these 15 residues, inhibited binding of GS149 to PIP, and prevented PIP, from inhibiting actin filament severing by gelsolin more effectively than pentalysine. These results strongly suggest that P _! contains the PIP binding site of GS149.

Gelsolin domain I contains an actin binding site which is necessary for severing actin-actin bonds after gelsolin attaches laterally to actin filaments via domains II-III. The interactions of domain I with actin have been characterized extensively, and residues important for act in binding have been identified by deletional and site-directed mutagenesis (Bryan, J., J. Cell Biol.

706: 1553-1562 (1988); Kwiatkowski, D.P., et al , J. Biol. Chem. 260: 15232- 15238 (1985); Way, M., et al , J. Cell Biol. 709:593-605 (1989); Way, M. , et al , EMBO J. 9:4103-4109 (1990); Yu, F.-X., et al, J. Biol. Chem. 266: 19269-19275 (1991); Kwiatkowski, D.P., et al , J. Cell Biol 708: 1717- 1726 (1989); Way, M., et al , J. Cell. Biol 776: 1135-1143 (1992); Pope, B., et al. , FEBS 280:70-74 (1991); Bubb, M.R., et al. , J. Biol. Chem. 266:3820- 3826 (1991)). This is however the first report that actin binding by domain I is regulated by PPI. Our results show that domain I bound PPI as well as domains II-III, which we have previously shown to be inhibited by PIP, (Yin,

H.L., et al , J. Cell Biol. 706:805-812 (1988)). Therefore, it is likely that PPI inhibit both actin binding steps in the severing process. The existence of multiple PPI-binding sites as well as multiple actin binding sites in gelsolin may explain why severing is particularly sensitive to PPI inhibition. It may also explain why actin filament end capping and nucleation by gelsolin, which require only one of the two N-terminal PPI-regulated domains, are also inhibited by PPI but to a lesser extent (Janmey and Stossel, Nature 525:362- 364 (1987)).

The region important for PPI binding in domain I was located to within the C-terminal 15 residues of domain I. Deletion of this sequence resulted in substantial loss of PPI binding and regulation of actin binding, whereas a synthetic peptide containing this sequence competed effectively with domain I and gelsolin for PIP,. The PPI binding site is distinct from the actin binding site, because the C-terminal truncated domain I polypeptide bound actin as well as full length domain I. Likewise, we showed recently that a synthetic peptide (P,) from the N-terminus of domain II (GS150-169) competed effectively with gelsolin polypeptides for PIP ₂ , suggesting that it also contains a PPI-binding site (Janmey, P.A., et al. , J. Biol. Chem. 267(17): 11818-11823 (1992)). This sequence is contiguous with the putative domain I PPI-binding site identified here. Secondary structure analyses according to Chou-Fasman and Garnier-Osguthorpe-Robson routines (MAX-VECTOR PROGRAM) predict that the sequence spanning P, and P, contains mostly random coils, which will allow the basic residues in this region to have the flexibility to conform to the acidic headgroups in a PPI cluster. PPI binding may induce a conformational change in the PPI binding domain and in the rest of the molecule as well to inhibit actin binding and severing. Circular dichroism studies with GS149 suggest that the con formation change in domain I involves a transition from random coil to /3-sheet. In addition, there is preliminary evidence that P, which has a random coil conformation in solution adopts a predominantly α helical conformation on binding to PIP ₂ (Xian, W. , et al ,

FASEB 6:A87 (1992)). Inhibition of actin binding may result from an induced

change in the conformation of the actin binding sites or indirectly through changes in other parts of the molecule to limit access to actin and/or cooperative interactions between different actin binding domains.

Each of the putative PPI binding sites on gelsolin contains a stretch of basic residues spaced almost identically: KxxxKxKK in P, and KxxxxKxRR in P, (Fig. 5B). Site directed mutagenesis studies of gelsolin domain I showed that replacement of two of the lysines individually with alanine did not affect PPI binding, suggesting that these lysines are not essential for interaction with PPI, providing that the other residues are still present. This is not surprising because PPI binding is likely to be mediated by multiple charge-charge interactions as well as hydrogen bonding (Janmey and Stossel, J. Biol. Chem. 264:4825-4831 (1989)), so neutralization of a single charge may not significantly affect binding. Additional experiments will be required to determine the effects of replacing other lysines singly and in combination, and of deleting the P, sequence in domain II, to assess its importance for PPI binding directly. Furthermore, although our results suggest that the region spanning domains I and II is important for PPI regulation, it does not rule out that other sites on gelsolin may also be involved. These sites may account for the residual PPI binding and regulation of GSI 34. Although we have not yet identified the residues in the P, and P, motifs which are required for high affinity PPI binding, it is interesting to note that this motif is found in two other members of the gelsolin family, villin (Bazari, W.L., et al , Proc. Natl. Acad. Sci. USA 85: :4986-4990 (1988)) and gCap39 (Yu, F.-X. , et al. , Science 250: 1413-1415 (1990)), and a number of unrelated PPI regulated actin binding proteins (Fig. 12B & C). Among the gelsolin family, gelsolin is the only member with two putative PIP, binding sites, while villin and gCap39 have one in their domains II. Severin does not have the spacing of the basic motif even though it is inhibited by PIP, (Yin, H.L., et al , FEBS 264:78-80 (1990)). It will be important to compare the PPI binding affinity and specificity of the gelsolin family of proteins, to sort out the functional significance of having different number of negatively charged

residues and different primary structures for the binding sites. There are indications that villin is less sensitive to PPI than gelsolin (Janmey and Matsudaira, J. Biol. Chem. 265: 16738-16743 (1988)). Subtle differences in PPI regulation among these proteins would not be surprising, since considerable differences in their Ca ²⁺ -regulation have already been demonstrated (Yu, F.-X., et al , Science 250: 1413-1415 (1990); Janmey and Matsudaira, J. Biol. Chem. 265: 16738-16743 (1988); Yu, F.-X., et al , J. Biol. Chem. 266: 19269-19275 (1991)). Human profilin (Ampe, C , et al , FEBS 228: 17-21 (1988)) has a modified motif, with a 6 amino acid spacing between the first two basic residues, and substitution of K with H. Since H is partly charged at neutral pH, human profilin may have lower PIP, binding affinity than gelsolin. Acanthamoeba profilin II which binds PIP ₂ but with at least 10 fold lower affinity than human profilin (Machesky, L.M. , et al , Cell Reg. 7:937-950 (1990)) does not have the putative motif (Ampe, C, et al , FEBS 228: 17-21 (1988)). Cofilin has a similar motif in residues 13-22.

Although an unrelated peptide derived from cofilin 104-115 was recently shown to bind PPI- and actin (Yonezawa, N., et al, J. Biol. Chem. 265:8382- 8386 (1991)), it remains possible that cofilin residues 13-22 contain another PPI binding site. The PIP, binding motif is not restricted to actin regulatory proteins. Several inositol-specific phospholipase C isozymes (Studier, F.W. , et al , Methods Enzymol 785:60-89 (1990)) have a similar motif (Fig. 5C) at the extreme C-terminus of a conserved domain, referred to as the "X box" (Rhee, S.G., et al. , Science 244:546-550 (1989)). The existence of a common motif among a large variety of PPI-binding proteins is consistent with its importance for high affinity polyphosphoinositide-specific binding.

Net Charge Seq. ID No.

Villin 111-153 KQGGVASGMKHVETNTYNVQRLLHVKGKKNWAAEVEM + 6.0 5 + ++ - + + + ++

Villin 133-147 YNVQRLLHVKGKKNVC +4.5 6 + + + + +

Villin 140-147 HVKGKKNVC +3.5 7 + + + +

Gelsolin 150-169 KHWPNEVWQRLFQVKGRR +4.5 2 ++ + + ++

MARCKS 155-173 KRFSFKKSFKLSGFSFKKN + 7.0 8 ++ ++ + ++

Table 1. Primary structures and net charges of synthetic peptides used for competition with gelsolin for binding polyphosphoinositides. Charges are denoted below each residue symbol.