Regulation of Actin Polymerization in Cell-free Systems by GTPγS and Cdc42

Rabbit peritoneal exudate PMNs, obtained as described previously, were suspended at 3–6 × 10⁸ cells/ml in saline and incubated with 1 mM di-isopropylfluorophosphate (DFP; 1/1000 dilution of 1 M stock; Sigma Chemical Co., St. Louis, MO), for 5 min on ice. The cells were washed two times with cold saline and resuspended at 3 × 10⁸ cells/ml in intracellular physiological buffer IP: 135 mM KCl, 10 mM NaCl, 2 mM MgCl₂, 2 mM EGTA, 10 mM Hepes, pH 7.1). Protease inhibitors (1 μg/ml leupeptin, 1 μg/ml benzamidine, 10 μg/ml aprotinin, 10 μg/ml Tame) were added, and the cells were incubated in a Parr bomb at 350 lb/in² for 15 min on ice. Upon release of pressure, the lysate was used immediately or stored on ice as described.

Cells were grown in HL-5 based media as previously described (Devreotes et al., 1987). Amoeba were starved at 2 × 10⁷ cells/ml and stimulated at 6-min intervals with 50 nM cAMP for 5 h. Cells were diluted to 1 × 10⁷ cells/ml and shaken at 200 rpm with 3 mM caffeine for 15 min. Cells were spun and resuspended with cold PM (10 mM PO₄ buffer, pH 6.1, and 2 mM MgSO₄) at 5 × 10 ⁷ cells/ml for intact cell stimulations and at 3 × 10⁸ cells/ml for preparation of lysates. At time of lysis, cells were mixed 1:1 with PM containing with 20 mM KCl and 1 mM EGTA and lysed by passage through a 5-μm pore size filter.

D. discoideum mutants were constructed by targeted gene disruption. Mutants lacking the Gα₂ (myc2) and Gβ (LW6 and LW14) subunits have been described previously (Chen et al., 1994; Wu et al., 1995). Controls used in the experiments reported here were the wild-type AX-3 strain.

The low speed supernatant (LSS) was the supernatant of lysate spun at 4°C at 14,000 rpm for 5 min (in microfuge) ∼1.5 × 10⁵ g min. The high speed supernatant (HSS) was made from LSS by spinning at 80,000 rpm for 20 min (∼5.6 × 10⁶ g min) in Beckman TL 100 centrifuge using a 100.1 (D. Discoideum) or a 100.2 or 100.3 rotor (PMN). When the HSS was diluted into TRITC–phalloidin (see below) and pelleted again, the amount of TRITC–phalloidin present in the pellet was 10–20% of that in the lysate. The number of nucleation sites assayed by the rate of pyrenyl actin polymerization (see below) was ∼10% of those present in the lysate. HSS could be frozen at −80°C for at least 6 mo with full retention of GTPγS sensitivity.

F-actin was quantified from TRITC–phalloidin staining of pelleted material, modified slightly from the original description (Howard and Oresajo, 1985). Aliquots of lysates or supernatants were incubated unless stated otherwise at equivalent of 1.5 × 10⁸ cells/ml at 37°C for PMNs or at room temperature for D. discoideum. Reactions were stopped by dilution of aliquots (usually between 30 and 60 μl) into 860 μl IP buffer containing 0.4 μM TRITC–phalloidin (Sigma Chemical Co.) and 0.1% Triton. After staining for 1 h, the samples were spun at 80,000 rpm for 20 min in an ultracentrifuge at high speed (TL100; Beckman Instr., Fullerton, CA) or for 5 min at 14,000 in the microfuge (low speed). The pellets were extracted with 1 ml of methanol, and after ∼20 h the fluorescence (540 ex/575 em) was read. To determine nonsaturable staining, 4 μM unlabeled phallacidin was included.

Pyrenyl actin assays of sites that nucleate polymerization were performed as described previously (Cano et al., 1991).

Recombinant Rac1 and Cdc42 were expressed in a baculovirus insect cell expression system as described (Heyworth et al., 1993; Xu et al., 1994). The Cdc42 was a glutathione-S-transferase (GST) construct and was isolated on GST beads, as per E. coli proteins, except from a membrane detergent lysate (to obtain the isoprenylated protein only). The amounts of proteins used in the assays were based on activity of the small GTPase proteins as determined by their ability to bind [³⁵S]GTPγS. The original GST–Cdc42 construct was a gift from Dr. R. Cerione (Cornell University, Ithaca, NY). The purified G-proteins were charged with GTPγS by incubating for 10 min at 30°C with 100 mM GTPγS in EDTA/Mg to give a final Mg concentration between 100 and 1,000 nM (Knaus et al., 1992). The Mg concentration was then increased to 2 mM in excess of the EDTA present, and the samples were stored on ice until use. G12V Cdc42 expressed in E. coli was a gift of Dr. J. Meinkoth (University of Pennsylvania Medical School, Philadelphia, PA). The G12V Cdc42 was tested with and without precharging with GTPγS. This protein did induce filopodia when injected into cells (Meinkoth, J., personal communication).

The PAK1 fragment that binds both Rac and Cdc42 is composed of amino acids 65–150 from human PAK1 (Manser et al., 1994). Recombinant Cdc24 and Dbl were expressed in baculovirus and generously provided by Drs. R. Cerione, J. Glaven, and T. Nomanbhoy (Cornell University). For experiments with both the PAK fragment and Cdc24, the supernatants were incubated for 5 min at room temperature with the recombinant proteins before warming to 37°C for 10 min.

The anti-PIP₂ antibody was obtained from PerSeptive Biosystems (Framingham, MA). PIP and PIP₂ were from Boehringer Mannheim (Indianapolis, IN); diacylglycerol (DiC₈) was obtained from Molecular Probes (Eugene, OR); brain extract (Type I: folch fraction I from bovine brain), and all other chemicals, unless noted, were obtained from Sigma Chemical Co.

Liposomes and/or micelles were created by resuspending various dried lipids to 10 mg/ml in 0.1 M Tris, pH 7.5, and sonicating on ice 8 cycles of 5 s on and 10 s off at level 40 on a sonic demembranator (Model 150; Dynatech Labs, Chantilly, VA). Samples were then diluted into IP buffer and mixed with HSS for F-actin assays.

HSS were incubated for various times with buffer or Cdc42-GTPγS before an EM grid was placed on a drop of the solution. The grid was washed on two water drops and then placed on a drop of 1% uranyl acetate, drained, and dried. The samples were viewed at magnifications between 15,000 and 50,000.

Samples were incubated in a tube and then stained with TRITC–phalloidin and observed in the microscope with a 100× objective. Other samples were allowed to polymerize in the presence of phalloidin in a sealed slide coverslip preparation for 5 min at 37°C before observing. Similar results were observed with both methods, although there was more clumping of filaments when they were transferred from tube to slide after polymerization. In most cases, glucose, glucose oxidase, and catalase were included to reduce quenching during observation and photography. Photographs were taken on TMAX 400 and processed with classical darkroom methods.

Samples were prepared as for F-actin determinations but diluted only twofold into TRITC–phalloidin (0.8 μM) and pelleted in the microfuge (14,000 for 5 min) within 10 min. The pellets were suspended in sample buffer and run on 10% polyacrylamide–SDS gels. The Coomassie blue staining band migrating at 43,000 was quantified by scanning in a densitometer and compared to a standard curve of G-actin processed similarly.

PIP₂ synthesis was performed with minor modification as described (Moritz et al., 1992). The PIP/PS micelles to be used as substrate were formed from dried lipid (0.16 μmol of total phospholipid; 80 nmol each), resuspended in 0.6 ml of micelle buffer (50 mM Tris-HCL, pH 7.4, 0.33 M sucrose, 0.133% PEG 20,000, 150 mM NaCl, BSA 0.67 mg/ml), and sonicated for 3 min on ice at 50 W output, 5 s on and 10 sec off.

For each assay of PIP₂ synthesis, a 50 μl sample contained HSS of 3 × 10⁷ cell equivalents/ml in 50 mM Tris-HCl, pH 7.4, 100 mM NaCl, 15 mM MgCl₂, 1 mM EGTA, 80 μM PIP, 80 μM PS, BSA (0.4 mg/ml), 0.25 M sucrose, 0.1% PEG 20,000, 0.04% Triton X-100, 50 μM ATP, with 1–2 Ci/ mmol [³²P]g-ATP (Amersham, Arlington Heights, IL). Samples were incubated for 5 min with no addition or with 100 μM GTPγS or with 100 nM Rac or Cdc42 charged with GTPγS.

The reaction was stopped by adding 188 μl ice-cold mix (CHCl₃/ MeOH/1 N HCl [12/30/3 vol/vol/vol]) and vortexing well. Carrier lipids (5– 10 μg of brain lipids) were added to each tube. Then 25 μl of 1 N HCl and 150 μl of CHCl₃ were added and vortexed. Samples were spun 2 min in microfuge to split phases. The lower CHCl₃ phase was removed to a new polypropylene tube. After addition of 150 μl of CHCl₃ to the aqueous phase, samples were vortexed again, spun, and the lower CHCl₃ phase removed and combined with the previously separated CHCl₃ phase. The combined CHCl₃ phases were then washed once with an equal volume of water. Samples were spun for 2 min, and 200 μl of the CHCl₃ phase was removed, dried in a speed vac, resuspended in 15 μl of CHCl₃/MeOH, and spotted on grooved lanes on a silica gel plate (60A; Whatman, Clifton, NJ). After drying, another 15 μl was added to the tube and respotted on the plate. The TLC plate had been prepared the same day by dipping three times in potassium oxalate solution and then, after drying at room temperature, heating at 85°C for 60 min to dry. Standards of PIP and PIP₂ were also spotted (20 μL of 2 mg/ml in MeOH/CHCl₃) and later detected with I₂. Plates were run in tank with solvent (MeOH/CHCl₃/H₂O/NH₃, 100/70/25/15). Radioactive spots were quantified on screens of a Phosphoimager (Molecular Dynamics, Sunnyvale, CA).

GTPγS addition to lysates of either PMNs or D. discoideum amoeba induced actin polymerization. The F-actin increase was optimal when the lysate concentration was 1.5 × 10⁸ cell equivalents/ml, an ∼10-fold dilution of intact cell cytoplasm (Fig. 1,A). At this concentration, GTPγS typically caused a twofold increase in TRITC–phalloidin staining. This is similar to the increase caused by chemoattractants in intact cells. At lower concentrations, the extent of polymerization is decreased presumably because of dilution of the cytoplasmic components, including the reservoir of sequestered G-actin that is used for polymerization (Tardif et al., 1995). At higher lysate concentrations, the response was limited by a precipitation that occurred during warming of the lysate. The maximal GTPγS-induced F-actin increase was complete in ∼5 min (Fig. 1,B). Maximal F-actin levels were induced with concentrations of GTPγS ⩾30 μM (Fig. 1,C). The response was specific to GTPγS as 100 μM ATPγS induced only a slow increase in F-actin, reaching, after 20 min, about half of the increase induced by 100 μM GTPγS in 3 min. The presence of GTPβS, 100 μM or 1 mM, inhibited the F-actin induced by 100 μM GTPγS by 36 and 58%, respectively. Neither GTP nor chemoattractant plus GTP induced an increase in F-actin. The behavior of amoeba lysates assayed at 22° (not shown) paralleled those of neutrophil lysates at 37°C (Fig. 1).

The formation of actin filaments in response to GTPγS was also observed in a LSS of the lysate. Since most of the F-actin in the lysate is removed at low speed, the increases in F-actin levels relative to basal levels were greater in the LSS; however, the absolute change in TRITC–phalloidin staining was similar, indicating that a similar amount of actin polymerized. The large fractional increase allowed us to examine the increase in F-actin morphologically. The filaments, when stained with TRITC–phalloidin, could be observed in the fluorescent microscope. In the presence of GTPγS there were many filaments which often clustered into bundles and meshworks; without GTPγS, few filaments were seen (Fig. 1, D and E).

D. discoideum mutants lacking heterotrimeric G-protein subunits were tested in vivo and in vitro for actin polymerization responses. Amoebae lacking the α subunit, Gα₂, do not exhibit chemotaxis to the chemoattractant cAMP, and mutants lacking the β subunit fail completely to move towards any chemoattractant. These defects parallel deficiencies in actin polymerization (Fig. 2,a, and other data not shown). In the Gα₂ null cells, cAMP does not induce actin polymerization; in the Gβ null cells, no chemoattractants trigger a response. Nevertheless, in lysates of either mutant, GTPγS stimulated actin polymerization (Fig. 2 b). The response was similar to that in wild-type cells. These observations suggested that the target for GTPγS in lysates is not a heterotrimeric G-protein but might be a downstream small G-protein.

Addition of purified GTPγS-activated baculovirus-expressed human Cdc42 induced actin polymerization in lysates of both PMNs and D. discoideum amoeba. Further studies with PMN supernatants indicated that the rate of polymerization depended on the Cdc42 concentration (Fig. 3,A), while the final level of F-actin achieved was similar for concentrations ⩾50 nM (50–200 nM; Fig. 3,B). The amount of GTPγS carried over after charging the Cdc42 caused little or no polymerization (300 nM GTPγS was present per 100 nM Cdc42 [Fig. 1 C]).

In vivo, agonists that increase the levels of F-actin usually also increase the filament number and the availability of free barbed ends (Cano et al., 1991). Thus it was important to determine if the increase in polymerized actin observed in this in vitro assay also correlated with an increase in free barbed ends. Lysates were warmed with buffer, Cdc42, or GTPγS, and then diluted into pyrene- labeled actin; and the rate of pyrenyl actin polymerization was followed by the increase in pyrene fluorescence. Both GTPγS and Cdc42 induced an approximately twofold increase in the rate of polyme365368rization (Fig. 3 C). This increase in rate of polymerization was inhibited by 2 μM cytochalasin b (not shown), indicating that the elongation was occurring at the barbed ends of filaments.

The responses in this assay required appropriately processed Cdc42. Other members of the Rho family of small GTPases, Rac1 or Rho, expressed in baculovirus and charged with GTPγS did not stimulate actin polymerization in PMN lysates. The activated Rac was shown to be effective in assays of PIP₂ synthesis (see below). Cdc42 expressed in E. coli did not increase F-actin levels, suggesting that a specific modification such as geranylgeranylation was important for activity (Fig. 4 A; Heyworth et al., 1993). The activity of Cdc42 may also be enhanced by some aggregation. In preliminary experiments, centrifugation of the activated Cdc42 preparation at 80,000 rpm for 20 min, pelleted between 15 and 50% of the immunoreactive Cdc42 (as detected by Western blots) and ∼75% of its ability to induce actin polymerization (data not shown).

The effects of Cdc42 were specific. The activity required activation by GTPγS; Cdc42 without nucleotide or Cdc42 bound with GDP was ineffective. The effects of activated Cdc42 were blocked by a fragment of the enzyme PAK1, which contains the binding site for Cdc42 (Manser et al., 1995). Inhibition by the PAK fragment was most pronounced at early times; upon continued incubation, slow polymerization resulted in F-actin accumulation (Fig. 4 B). The GTPγS requirement and the inhibition by the PAK fragment indicates that the activity is not due to a contaminant in the Cdc42 preparation.

The Cdc42-induced filaments were collected on EM grids, negatively stained, and observed in the electron microscope. An HSS was used for these experiments. The filaments formed upon incubation with 100 nM Cdc42 for 5 min were present primarily as individual filaments of varying lengths up to at least 8 μm (Fig. 5,A); occasionally filament bundles were observed. Searches of grids made from HSS warmed for 5 min without Cdc42 revealed only an occasional filament (not shown). Similarly, TRITC–phalloidin-stained filaments were rarely seen in the HSS warmed without Cdc42 (Fig. 5,B). When HSS was warmed on the slide in the presence 100 nM Cdc42 and TRITC–phalloidin, many small filaments were observed (Fig. 5,C). These increases in filament number were consistent with the increase in barbed ends that nucleated polymerization, described above (Fig. 3,C). When warmed in a tube and then transferred to the slide, filament bundles and clusters were observed (the clusters presumably form during mixing; Fig. 5 D).

Interestingly, the GST–Cdc42 fusion protein still attached to glutathione bead could induce polymerization at the surface of the bead (Fig. 5, E and G). The ability of Cdc42-loaded beads to induce polymerization required activation by GTPγS (Fig. 5,F) and was blocked by the PAK GTPase binding fragment (Fig. 5 H). Like actin polymerization induced in vivo and in permeabilized cells, the GTPγS and Cdc42-induced polymerization was blocked by cytochalasin (data not shown).

Both activated Cdc42 and GTPγS were able to induce actin polymerization in LSS (1.5 × 10⁵ g min). However, only activated Cdc42 could induce polymerization in HSS (5.6 × 10⁶ g min). The magnitude of the Cdc42-induced response in a lysate and an HSS made from that lysate was similar (Fig. 6,A). The ability of GTPγS to induce polymerization in HSS could be restored by addition of a small amount of lysate or resuspended pellet (Fig. 6 B). Addition of lysate equal to only 3% of the volume of the HSS was sufficient to allow polymerization. Increasing the amount of lysate increased the rate of polymerization.

Since GTPγS bound Cdc42-induced polymerization in the HSS, all components downstream of Cdc42 that are needed for actin polymerization must be present. The failure of GTPγS to act suggested that HSS lacks either a G-protein or a GEF needed for GTPγS to bind to a G-protein. Treatment of the pelleted material with GTPγS was not sufficient to induce polymerization upon dilution into the HSS. Rather, GTPγS had to be present after mixing the HSS with the particulate fraction (data not shown), suggesting that the G-protein is present in the supernatant and that the particulate fraction stimulates nucleotide exchange.

In PMN lysates, small G-proteins of the Rho family are found in the supernatant, while the GEFs are in a particulate fractions (Bokoch et al., 1994; Phillips et al., 1995). Thus, the possibility that the HSS was lacking guanine nucleotide exchange activity seemed likely. To test this idea we sought to supply exchange activity in various ways. First, we sought to release the G-protein from possible inhibition by guanine nucleotide dissociation inhibitory factor (GDI) by addition of phospholipids (Chuang et al., 1993; Abo et al., 1994). Addition of lamellar or micellar lipids derived from PMNs by CHCl₃/MeOH extraction (not shown) or commercially available extracts from brain allowed GTPγS to stimulate an increase in F-actin (Fig. 7 A). Liposomes of pure anionic phospholipids were also effective. The rank order effectiveness of various lipids in allowing GTPγS to induce actin polymerization was similar to that effective in displacing GDI (Chaung et al., 1993). Phosphatidylinositol (PI) the most active pure lipid tested, allowed GTPγS to induce actin polymerization when present at 15 μM. Commercially available phosphatidic acid (PA), PIP, and PIP₂ also allowed response to GTPγS when present between 50 and 200 μM. Diacylglycerol (1-oleoyl-2-acetyl glycerol) was not effective even at concentrations up to 450 μM. At the concentrations tested, most of the lipids had little effect on the F-actin level in the absence of GTPγS. However, concentrations of PI >50 μM and brain lipids and phosphatidic acid >100 μM did increase F-actin levels in the absence of GTPγS.

A second way to increase exchange of nucleotides on small G-proteins is through magnesium chelation. We chelated magnesium in the HSS with EDTA and incubated at room temperature in the presence of GTPγS. When Mg was restored to 2 mM and the supernatant warmed to 37°C, a large increase in actin polymerization occurred (Fig. 7,B). Neither EDTA treatment without GTPγS nor EDTA treatment with GDP induced polymerization. Finally, we tested the ability of recombinant GEFs to restore the ability of GTPγS to stimulate actin polymerization in HSS. Addition of 500 nM recombinant Cdc24, a GEF from S. cerevesiae specific for Cdc42 (Cerione and Zheng, 1996) had no effect on its own but allowed GTPγS to induce actin polymerization in the HSS; the F-actin increase varied between 0.5- and 3-fold in three different experiments (Fig. 7,C). Addition of Dbl, a GEF for Cdc42 and Rho, but not Rac, also allowed GTPγS to induce up to a sixfold increase in F-actin in the HSS. Activity was seen with concentrations as low as 62 nM (Fig. 7 D). Note that since the basal F-actin in an HSS, i.e., after dilution and staining with TRITC–phalloidin and pelleting again, is only ⩽20% of the TRITC–phalloidin in the lysate, the increase in F-actin in the HSS is greater than in the lysate.

Various polyphosphatidylinositol remodeling enzymes are potential downstream targets for Cdc42. A known target of Cdc42 is the 85-kD regulatory subunit of PI-3 kinase (Zheng et al., 1994; Tolias et al., 1995). However, addition of 100 nM wortmannin, an inhibitor of PI-3 kinase, did not inhibit the induction of actin polymerization by either Cdc42 or GTPγS (not shown). A second potential downstream target is PIP₂ synthesis, which is simulated by Rac in permeabilized platelets and has been proposed to regulate actin polymerization (Hartwig et al., 1995). We investigated whether PIP₂ synthesis was responsible for actin polymerization in PMN HSSs. In the PMN HSS, it was not possible to detect incorporation of ³²P from γ-labeled ATP into bands comigrating on thin layer chromatography with PIP or PIP₂. However, upon addition of substrate lipids, micelles of equimolar PS with PIP, incorporation of ³²P into PIP₂ was detected. Recombinant Rac stimulated synthesis of PIP₂ from PIP nearly tenfold (Fig. 8, A and B). Cdc42 stimulated PIP₂ synthesis about threefold; however GTPγS stimulated little or no synthesis of PIP₂ even when brain lipids were included to stimulate nucleotide exchange (Fig. 8, A and B). Parallel experiments run on the same day confirmed that each of these agents acted on actin polymerization as described above: Rac did not induce actin polymerization, while both Cdc42 and GTPγS (plus brain lipids) did. Thus, PIP₂ synthesis did not correlate with actin polymerization. Furthermore, as noted above, addition of PIP₂ liposomes did not stimulate actin polymerization. Finally, we tested antibodies to PIP₂ that have been effective in blocking functions of PIP₂ in some systems (Fukami et al., 1988; Gilmore and Burridges, 1996). The antibody, at concentrations effective in injection experiments, did not inhibit Cdc42-induced actin polymerization in a HSS (Fig. 8 C). This again suggested that PIP₂ is not needed downstream of Cdc42.

We have developed a cell-free system that, upon stimulation with intermediates of the signal transduction pathway, exhibits an increase in F-actin. Lysates of PMN and D. discoideum, cells widely separated in evolution, both respond under similar conditions to GTPγS by increasing actin polymerization. The HSS is stable for several hours on ice or can be frozen and stored at −80° for several months without loss of activity. Furthermore, the lysate is amenable to fractionation and to addition of exogenous materials.

Initiation of actin polymerization in cell lysates appears to occur at the level of small G-proteins of the Rho family. Thus, addition of Cdc42 charged with GTPγS induced polymerization. Indeed, coupling between the chemoattractant receptor and actin polymerization appears lost upon lysis. The insensitivity of lysates does not reflect an uncoupling of receptor and heterotrimeric G-protein, since GTPγS regulates chemoattractant binding in lysates of both cell types (Devreotes and Zigmond, 1988). In lysates of D. discoideum amoeba, GTPγS stimulates adenylate cyclase, a response that depends critically on the β-subunit. Yet, in the same lysates, actin polymerization is completely independent of the presence of the β-subunit of the heterotrimeric G-protein. It seems most likely that coupling between the trimeric G-protein and downstream effectors of actin polymerization are lost upon cell lysis. It is possible that lysis itself activates guanine nucleotide exchange activity toward the downstream Rho GTPases, making regulation by trimeric G-protein redundant.

In vitro actin polymerization occurs in concentrated lysates (2- to 3-fold dilution of packed cell pellet) of various cells and oocytes (Kane, 1986; Fukami et al., 1988; Whitehead et al., 1995a,b). It is not clear what initiates this polymerization, which exceeds that in the cell at the time of lysis (Kane 1986). It is possible that in these concentrated extracts, activation of GEF activity upon lysis allows endogenous GTP to activate small G-proteins, which then induce actin polymerization. In the more dilute extracts used in this study, GTP is ineffective, presumably because it is hydrolyzed before it has time to act.

In the assays described here, downstream of Cdc42, all components required for actin polymerization are present in the HSS. Thus, in the HSS of both neutrophils and D. discoideum cells, GTPγS-charged Cdc42 could induce polymerization. However, the HSS appeared to lack the guanine nucleotide exchange activity required for GTPγS to induce polymerization. Activity in PMN HSS was restored by addition of recombinant GEFs (Cdc24 and Dbl). Our data are consistent with observations in hematopoetic cells that most of the Rac and Cdc42 are present in the supernatant (Bokoch et al., 1994; Phillips et al., 1995), while the GEFs, or factors that activate GEFs, are present in a particulate fraction (Bokoch et al., 1994; Whitehead et al., 1995a,b; Chardin et al., 1996).

Acidic lipids restored the ability of GTPγS to induce polymerization in the HSS, probably acting by displacement of GDI (Chuang et al., 1993). Lipids are known to facilitate guanine nucleotide exchange in several ways: (a) acidic lipids can displace GDI (Chuang et al., 1993). (b) Polyphosphatidylinositols can activate GEFs by binding to their PH domain (Chardin et al., 1996). (c) Polyphosphatidylinositols can also directly stimulate release of GDP from small G-proteins, including Cdc42, although since GTPγS can not bind this complex, PIP₂ alone can not activate Cdc42 (Terui et al., 1994; Zhang et al., 1996). In displacement of GDI, phosphatidic acid and PI are more effective than PIs; in stimulating GEFs and releasing GDP, PIP₂ is much more effective than PI. The fact that PI was more effective than PIP or PIP₂ in allowing GTPγS to induce actin polymerization suggested that, in this system, the lipids functioned primarily through GDI displacement.

PIP₂-mediated uncapping of actin filaments has been proposed to be the downstream effector of Rac, because Rac stimulates PIP₂ synthesis (Hartwig et al., 1995; Barkalow et al., 1996; Schafer et al., 1996). However, in our system there was no correlation between PIP₂ synthesis and actin polymerization. Recombinant Rac stimulated PIP₂ synthesis, but the increase was not accompanied by increased actin polymerization. In addition, although GTPγS induced polymerization, it did not stimulate PIP₂ synthesis. That GTPγS did not stimulate PIP₂ synthesis in the HSS suggested that it did not activate endogenous Rac. This failure may be ascribed to the absence of cellular membranes, since the ability of GTPγS to activate Rac in PMN lysates depends on GEFs that pellet with the cellular membranes (Bokoch et al., 1994). Further excluding PIP₂ as a downstream mediator of actin polymerization, addition of exogenous PIP₂ did not increase F-actin level, and antibodies against PIP₂ did not block Cdc42-induced actin polymerization.

Further work will be required to determine which properties of Cdc42 are critical for its induction of actin polymerization and why Rac was ineffective. Cdc42 activity required GTPγS and some modification, probably geranylgeranylation that occurred during expression in Sf9 cells but not in E. coli. Activity may also be enhanced by some macromolecular organization since the Cdc42 pelleted at high speed had disproportionately high activity. Cdc42 activity in solution and attached to a bead was inhibited by the GTPase-binding fragment of PAK, which presumably acts by blocking interaction of Cdc42 with its targets.

Parallel preparations of recombinant Rac were inactive in stimulating polymerization in PMN lysate and HSS. This was surprising since Rac stimulates actin polymerization in permeabilized platelets (Hartwig et al., 1995). Perhaps, in the PMN, Rac function has been diverted to regulate the NADPH oxidase (Bokoch, 1994). Alternatively perhaps, lysis disrupts coupling of Rac to actin polymerization as it disrupts coupling of heterotrimeric G-proteins to actin polymerization.

Further work will also be required to determine whether GTPγS acts through endogenous Cdc42. Addition of the GTPase binding fragment of PAK to PMN lysates did not block GTPγS-induced actin polymerization (Joyce, M., and S.H. Zigmond, unpublished result) suggesting Cdc42 might not be the endogenous target. However, supporting endogenous Cdc42, recombinant exchange factors Cdc24 and Dbl, specific, respectively, for Cdc42 and for Cdc42 and Rho, allowed GTPγS to function in the HSS (Cerione and Zheng, 1996). The concentration of endogenous Cdc42 in the HSS of 1.5 × 10⁸ cells/ml was estimated by Western blots to be ∼4.5 ± 2 nM (mean ±SD; Lartique, J., H. Sun, and S.H. Zigmond, unpublished observation), ∼10-fold lower than the concentration of recombinant Cdc42 required to induce actin polymerization. Thus, the endogenous Cdc42 may be more active than the recombinant protein. Increased activity might result from complexes that associate endogenous Cdc42 with its downstream target. If so, the failure of the PAK fragment to inhibit the GTPγS stimulation might also be explained: this fragment might bind too slowly to compete under these conditions.

In summary, the ability to induce actin polymerization in vitro allows dissection of the pathways involved in this important process. Using this assay, we have begun this investigation. Future studies will aim at restoring upstream coupling to agonist and at defining the endogenous downstream target for GTPγS and subsequent elements that mediate actin polymerization.

We are grateful for excellent technical assistance from Alexandra Kudrjavcev-DeMilner, Hai Sun, and Julien Lartique. We are most grateful to Drs. K. Zhou and U. Knaus (The Scripps Research Institute) for purified PAK fragment and BV Cdc42, respectively; Dr. J. Meinkoth (University of Pennsylvania Medical School) for E. coli-expressed Cdc42; Drs. R. Cerione, T. Nomanbhoy, and J. Glaven (Cornell University) for recombinant Cdc24 and Dbl expressed in baculovirus; Dr. L. Tilney, and Pat Connelly (University of Pennsylvania) for help with EM; and Drs. S. Rittenhouse (Jefferson University Medical School), H. Goldfine, and B. Wolf (University of Pennsylvania Medical School) for help with the phospholipid synthesis.

F-actin

filamentous actin

G-actin; globular actin; GDI

guanine nucleotide dissociation inhibitory factor

GEF

guanine nucleotide exchange factor

GST

glutathione-S-transferase

HSS

high speed supernatant

LSS

low speed supernatant

PI

phosphatidylinositol

PIP₂

phosphatidylinositol bisphosphate

PMN

polymorphonuclear leukocytes