Actin polymerization induced by GTP gamma S in permeabilized neutrophils is induced and maintained by free barbed ends.

To address the mechanisms through which agonists stimulate actin polymerization, we examined the roles of monomer sequestering proteins and free barbed ends on actin polymerization induced by guanosine 5'-3-O-(thio)triphosphate (GTP gamma S) in neutrophils permeabilized with streptolysin O. Addition of profilin (without GTP gamma S) caused a net decrease in F-actin. Thus, merely making profilin available in the cell was not sufficient to induce actin polymerization. On the other hand, addition of profilin hardly affected the polymerization induced by GTP gamma S, while thymosin beta 4 or DNase I decreased this polymerization. These data suggested that GTP gamma S induced polymerization by increasing the availability of barbed ends. In the presence of cytochalasin B, profilin did inhibit polymerization induced by GTP gamma S, demonstrating that GTP gamma S did not inhibit profilin's monomer sequestering ability. The F-actin induced by GTP gamma S was not limited by a time-dependent loss of G-actin or G-proteins from permeabilized cells since, following stimulation with suboptimal concentrations of GTP gamma S, addition of more GTP gamma S induced further polymerization. Barbed ends remained free after F-actin reached plateau since (a) cytochalasin B caused depolymerization of induced F-actin and (b) profilin did not depolymerize induced F-actin unless the cells were first treated with cytochalasin to cap barbed ends. The data indicate that GTP gamma S maintains an increased level of F-actin by keeping at least a few barbed ends available for polymerization.

To address the mechanisms through which agonists stimulate actin polymerization, we examined the roles of monomer sequestering proteins and free barbed ends on actin polymerization induced by guanosine 5-3-O-(thio)triphosphate (GTP␥S) in neutrophils permeabilized with streptolysin O. Addition of profilin (without GTP␥S) caused a net decrease in F-actin. Thus, merely making profilin available in the cell was not sufficient to induce actin polymerization. On the other hand, addition of profilin hardly affected the polymerization induced by GTP␥S, while thymosin ␤ 4 or DNase I decreased this polymerization. These data suggested that GTP␥S induced polymerization by increasing the availability of barbed ends. In the presence of cytochalasin B, profilin did inhibit polymerization induced by GTP␥S, demonstrating that GTP␥S did not inhibit profilin's monomer sequestering ability.
The F-actin induced by GTP␥S was not limited by a time-dependent loss of G-actin or G-proteins from permeabilized cells since, following stimulation with suboptimal concentrations of GTP␥S, addition of more GTP␥S induced further polymerization. Barbed ends remained free after F-actin reached plateau since (a) cytochalasin B caused depolymerization of induced F-actin and (b) profilin did not depolymerize induced F-actin unless the cells were first treated with cytochalasin to cap barbed ends. The data indicate that GTP␥S maintains an increased level of F-actin by keeping at least a few barbed ends available for polymerization.
Neutrophils treated with inflammatory mediators increase their F-actin level by shifting the steady state from G-to F-actin. However, which of the many factors that cause this shift in vitro account for it in vivo are unknown. Possible factors include: (a) a shift in the nucleotide bound to G-actin from ADP to ATP or (b) inhibition of the major monomer sequestering proteins. These factors now seem unlikely (Rosenblatt et al., 1995;Carlier et al., 1993;Redmond et al., 1994;Safer et al., 1990;Cassimeris et al., 1992;Nachmias et al., 1993), so interest is focused on (c) the availability of profilin and (d) the availability of free barbed ends.
Profilin, first identified from its ability to inhibit actin polymerization (Carlsson et al., 1977), is now also known to promote polymerization . When pro-filin is added in vitro to a mixture of thymosin ␤ 4 (T␤4), 1 G-actin, and F-actin, the F-actin decreases or increases depending on whether the filament barbed ends are capped or free Carlier and Pantaloni, 1994). Cells contain T␤4, G-actin, and F-actin. Thus, if in the resting cell some barbed ends are free but profilin is sequestered, an agonist could induce polymerization merely by releasing profilin from a sequestered pool. Profilin might well be sequestered in cells because it is distributed nonhomogeneously in the cell (Buss et al., 1992) and because it binds strongly to inositol bisphosphate (Lassing and Lindberg, 1985). Furthermore, some agonists may release profilin bound to inositol bisphosphate, thus freeing it to promote polymerization (Sohn and Goldschmidt-Clermont, 1994;Machesky and Pollard, 1993). The hypothesis that profilin release induces actin polymerization has not been tested.
Free barbed ends are of interest because in vitro increasing the fraction of filaments with free (uncapped) barbed ends induces polymerization (Yin and Stossel, 1979;Pollard, 1986). However, in cells, evidence that free barbed ends regulate the actin steady state is equivocal. On one hand, cells lysed after stimulation with agonist have an increased number of sites that nucleate barbed end elongation of exogenous actin (Carson et al., 1986;Condeelis et al., 1988;Hall et al., 1989;Hartwig, 1992;Nachmias et al., 1993). This increase in nucleation sites occurs even when net polymerization is blocked by cytochalasin, suggesting these sites may be the cause rather than the effect of polymerization (Hartwig, 1992). 2 On the other hand, most of the filaments in the lysate of resting neutrophils appear to have free barbed ends, and the increase in number of barbed ends upon stimulation is matched by an equal increase in pointed ends (Cano et al., 1991;Carson, et al., 1986). Merely doubling the number of filaments should not shift the critical concentration. Furthermore, it is unclear if these barbed ends are available in the intact cell or freed only upon lysis and dilution. Indeed, at high cell concentrations, lysates contain sufficient capping activity to cap all the barbed ends present in both control and stimulated cells (Cassimeris et al., 1992;Southwick and DiNubile, 1986). Thus, it is unclear what fraction of filaments are free in the intact cell and whether this fraction is altered by stimulation. Attempts to determine whether free barbed ends stimulate actin polymerization in intact cells by injection of free barbed ends (small actin filaments) gives negative or ambiguous results (Sanders and Wang, 1990;Handel et al., 1990).
Also equivocal is evidence that F-actin is regulated by freeing barbed ends that is based on experiments with cytochalasin, a barbed end capper. Cytochalasin inhibits agonist-induced polymerization in neutrophils, suggesting that polymerization occurs primarily at barbed ends (MacLean-Fletcher and Pollard, 1980;White et al., 1983). However, this does not indicate that the freeing of barbed ends is the regulated event; a similar inhibition by cytochalasin would be expected were agonist releasing profilin to facilitate polymerization on existing free ends. Furthermore, cytochalasin has many effects such as stabilizing actin dimers and increasing the rate of ATP hydrolysis (Sampath and Pollard, 1991). Thus, cytochalasin could exert its effects in cells by shifting the nucleotide bound to G-actin from ATP to ADP. So overall, the evidence supporting regulation of F-actin by freeing of barbed ends is weak. One cannot study how an agonist causes actin polymerization in vitro because after cell lysis both natural agonists and GTP␥S cease to stimulate an increase in F-actin. However, permeabilized neutrophils, like intact cells, double their Factin level upon addition of appropriate agonist (Downey et al., 1989;Therrien and Naccache, 1989;Bengtsson et al., 1990;Redmond et al., 1994). Neutrophils permeabilized with streptolysin O (SLO) have large pores, allowing entry of exogenous proteins up to 120,000 Da (Redmond et al., 1994;Bhakdi et al., 1993). This permitted us to modulate the cytoplasmic G-actin by means of exogenous monomer binding proteins. We utilized the different properties of monomeric actin binding proteins to dissect the changes that lead to actin polymerization. In particular, we utilized the unusual properties of G-actin profilin complex, which can contribute G-actin to the barbed but not pointed end of an actin filament (Tilney et al., 1983;Pollard and Cooper, 1984;Pring et al., 1992;Pantaloni and Carlier, 1993;Giuliano and Taylor, 1994). The results put on firmer ground the idea that increase in F-actin by chemoattractants is mediated through an increase in the availability of free barbed ends.

EXPERIMENTAL PROCEDURES
Materials-GTP␥S, TRITC-phalloidin, phallacidin, and DNase I were obtained from Sigma. The G-actin binding by DNase I was calibrated by ability of purified G-actin to inhibit DNase I activity; this was 50% of the DNase I present based on protein concentration.
Streptolysin O Permeabilization-One part of the cell suspension at 2 or 5 ϫ 10 7 cells/ml was mixed with three parts of streptococcal cytotoxin streptolysin O (Murex, Norcross, GA) in phosphate-buffered saline at 2.0 international units (IU)/ml, and incubated on ice for 3 min. Thus, cells were permeabilized with 2 or 0.9 ϫ 10 Ϫ7 IU of SLO/cell (the higher ratio of SLO/cell resulted in faster exchange of proteins between the cell and the medium). The excess of SLO was removed by pelleting in a microcentrifuge for 15 s at 15,000 rpm. The cell pellet was resuspended at 4 ϫ 10 6 cells/ml in intracellular physiological (IP) buffer (135 mM KCl, 10 mM NaCl, 2 mM MgCl 2 , 2 mM EGTA, 10 mM Hepes, pH 7.0) at room temperature. Cells were incubated at room temperature for 2 min (warming time) before stimulus was added. The pores induced by streptolysin O do not form in the cold, so the cells only become permeable after warming. The time courses plotted in the figures begin after warming at room temperature for 2 min.
Saturable Staining with TRITC-Phalloidin as Assay of F-actin Levels-F-actin levels were determined by phalloidin staining as described by Howard and Oresajo (1985) modified by Cano et al. (1992a). Changes in the state of actin were stopped by addition of 1% glutaraldehyde and TRITC-phalloidin (final concentration, 2 ϫ 10 Ϫ7 M). To determine the nonsaturable staining, unlabeled phalloidin (final concentration, 2 ϫ 10 Ϫ6 M) was added with TRITC-phalloidin to some samples. Samples were stained for at least 1 h, pelleted for 7 min at 15,000 rpm in a microcentrifuge and drained. From each pellet, corresponding to 2 ϫ 10 6 cells, the TRITC-phalloidin was extracted with 0.8 ml of methanol for at least 48 h before the rhodamine fluorescence (540 ex /575 em ; Slits: 3, 20) was read in a Perkin Elmer LS5 fluorimeter. TRITC-phalloidin staining values reported are the saturable staining: the total fluorescence extracted minus the nonsaturable fluorescence, i.e. the fluorescence extracted from cells stained in the presence of unlabeled phalloidin.
Nucleation Sites Assay-Permeabilized cells with or without stimulus were processed as described above. The nucleation sites assay was performed as described previously using 4 ϫ 10 5 cells/ml in assay buffer (Cano et al., 1991).
Profilin Purification-Profilin was isolated from calf spleen according to the method of Kaiser et al. (1989) and Janmey (1991). The actin and profilin were eluted with extraction buffer containing 3 M and 7 M urea, respectively. The profilin peak, identified by A 280 nm , was immediately dialyzed against 20 mM Tris-HCl, pH 7.4, containing 0.1% sodium azide. Profilin sample was run on 12.5% SDS gel to check for purity. In one profilin preparation contaminated with actin, we repeated the polyproline column to obtain a clean profilin sample. Yield was approximately 50 mg of profilin/300 g of spleen.
Thymosin ␤4 (T␤4) was isolated from calf spleen by the method of Cassimeris et al. (1992) with minor modifications.
Modeling of Data-To determine if the amount of polymerization induced by GTP␥S could be caused by (a) a shift in the affinity of the filaments from 0.5 M to 0.1 M (as might happen if barbed ends became available), we calculated the amount of G-actin that would be released from a T␤4-bound pool after such a shift. The amount released depended on the amount of T␤4-actin complexes present in the SLOpermeabilized cell and changes in this amount caused by the presence of exogenous monomer binding proteins (Cassimeris et al., 1992;Fechheimer and Zigmond, 1993;Nachmias et al., 1993;Pantaloni and Carlier, 1993). For these calculations, we assumed (a) that PMNs contain about 250 M T␤4 and about 115 M G-actin (Cassimeris et al., 1992) (for simplicity, the contribution of profilin was ignored); (b) that the K d of the binding of cellular G-actin to DNase I was 1 nM and to T␤4 is 0.6 M ; (c) that the rate of equilibration of G-actin with T␤4 was very rapid, thus the DNase I and exogenous T␤4 in the permeabilized cells were considered to be at steady state with the endogenous G-actin/T␤4 .
The amount of G-actin, T␤4, and profilin in a permeabilized cell is not constant because endogenous proteins are continually leaving. Analysis of the G-actin released into the medium (assayed by DNase I inhibition) indicated that at the time of stimulation with GTP␥S (i.e. 2 min) about half of the G-actin pool has left the cell (Redmond et al., 1994). Since T␤4 is in rapid equilibrium with the G-actin, we assumed that half of the T␤4 has also left the cell. Indeed, slightly more than 50% of the profilin released by Triton lysis was released into the medium after warming SLO-permeabilized cells for 2 min. Profilin released from permeabilized cells or after Triton lysis was collected on a polyproline column, eluted with urea, and quantified by comparison to profilin standards on Coomassie Blue-stained SDS gels (Kaiser et al., 1989).
Thus after 2 min, with about 125 M T␤4 and 57.5 M G-actin remaining in the cell, about 57 M T␤4 would be bound to G-actin and the free G-actin concentration would be ϳ0.5 M (in equilibrium with the free pointed end). We then calculated the amount of G-actin that would be released from T␤4 if the affinity of F-actin were decreased to 0.1 M, allowing actin to polymerize until the free G-actin decreased to 0.1 M. This would allow G-actin bound to T␤4 to be released until the complex was in equilibrium with 0.1 M G-actin. The amount of G-actin that would polymerize in the absence of exogenous monomer binding proteins (ϳ40 M) was set at 100%.
Since the concentration of free G-actin in the medium outside the SLO-permeabilized cells was extremely low, it was unlikely that Gactin binding proteins in the external medium would increase the rate of G-actin exit from permeabilized cells. Thus, the effects must be mediated by the proteins that entered the cell. To estimate the effects of exogenous monomer binders on polymerization induced by GTP␥S stimulation, we assumed the concentration of monomer binder inside the cell (bound or free) during the polymerization (i.e. between 3 and 5 min) was equal to the concentration added to the medium. While this is an overestimate, the actual value would be greater than half of this concentration.
To evaluate the effects of exogenous monomer binders on the pool of T␤4-actin present inside the cell and responding to the change in free G-actin concentration, we assumed that essentially all of the DNase I, which has an affinity for G-actin of about 1 nM , that entered the cell would bind G-actin. This would decrease the pool of T␤4-actin complex present, lower the concentration of free G-actin, and cause F-actin depolymerization, as observed ( Fig. 1). We assumed that a fraction of the T␤4 that entered the cell would bind G-actin (the fraction determined by its affinity, which was assumed to be the same as endogenous T␤4, 0.6 M). This would increase the amount of T␤4actin complex present and cause depolymerization. While some of the G-actin released by depolymerization would leave the cell, some would bind to the monomer binders present; for this calculation, the contribution of G-actin released by depolymerization was ignored.
We then calculated for each concentration of monomer binder in the cell the change in the amount of G-actin that would be bound to T␤4 (endogenous and exogenous) if stimulation changed the critical concentration as described above. This change was expressed as a percentage of the change occurring in the absence of exogenous monomer-binding proteins.
Calculation of Critical Concentration-The apparent actin critical concentration A (the free [G-actin] at which no net filament elongation occurs) was calculated as a function of the fraction of filaments whose barbed ends are not blocked by capping proteins other than profilin, f u , and at various total profilin concentrations, [P] t , as follows. Following Pantaloni and Carlier (1993), we assumed that profilin binding to G-actin and barbed filament ends was at equilibrium, yielding where [P] is free profilin, [PA] is the profilin-G-actin complex, f p is the fraction of barbed filament ends with profilin bound, and f f is the fraction free to elongate. K P and KЈ P are the affinity constants of profilin for free G-actin and barbed filament ends, respectively. A is then obtained by setting the net filament elongation rate to zero where k ϩ , k pϩ , and kЈ ϩ are the rate constants of filament elongation at the barbed and pointed ends by G-actin, and at the barbed end by the profilin-G-actin complex respectively; and k Ϫ , k pϪ and kЈ Ϫ are the corresponding depolymerization rate constants. This equation, which yields a cubic equation in A, was conveniently solved numerically by binary search with an initial range from zero to the pointed end critical concentration.

RESULTS
Exogenous G-actin Binding Proteins, Including Profilin, Lower Basal F-actin Levels-Inclusion of G-actin sequestering proteins, DNase I or T␤4, in the buffer at the time of permeabilization (warming) caused a decrease in basal F-actin of SLO-permeabilized cells as indicated by TRITC-phalloidin staining (lower curves in Fig. 1, A and B). The magnitude of the decrease depended on the concentration of sequesting protein added. It was unlikely that these effects were mediated by sequestering G-actin that had left the permeabilized cell, since the cell concentration was such that concentration of G-actin in the medium was always exceedingly low (see "Experimental Procedures"). Rather, the proteins presumably acted after entering the permeabilized cell and sequestering the free G-actin there. In order to affect the high concentrations of buffered G-actin present in the cytoplasm, high concentrations of the monomer binding protein were needed. With DNase, a 50% decrease in basal F-actin staining was seen after warming for 4 min in 15 M and a maximal decrease of about 60% in 30 M. With T␤4, a 50% decrease was seen with about 50 M. As expected, the molar concentrations of T␤4 required to cause a given amount of depolymerization was greater than that of DNase I since the affinity of T␤4 for cellular G-actin is much lower than that of DNase I (K d ϭ 0.6 M for T␤4 versus ϳ1 nM for DNase I; Weber et al., 1992).
Inclusion of profilin in the buffer at the time of permeabilization also decreased the basal F-actin level (lower curve in Fig.  1C). The effects of profilin were similar, although often not as pronounced as those of T␤4. After warming for 4 min in 40 M profilin, the basal F-actin decreased by 40 Ϯ 10%. The fact that addition of profilin decreased the F-actin in unstimulated cells indicated that profilin had entered the cells and was able to sequester G-actin. The fact that the addition of profilin in the cell caused a decrease, not an increase, in F-actin ruled out the possibility that merely freeing profilin from a sequestered pool could account for the increase in cellular F-actin upon stimulation.

Comparison of Profilin with Other G-actin Sequestering Proteins Indicates That GTP␥S Increases the Fraction of Filaments
with Free Barbed Ends-Addition of GTP␥S to permeabilized neutrophils stimulated actin polymerization, assayed by saturable TRITC-phalloidin staining (Redmond et al., 1994). Addition of DNase I or T␤4 caused dose-dependent decrease in the F-actin induced by GTP␥S (Fig. 1, A and B). GTP␥S could still increase the F-actin level in the presence of DNase I or T␤4 at concentrations that decreased the basal level. This suggested that GTP␥S must either (a) free a pool of G-actin not available to the basal F-actin and/or (b) increase the affinity of the filaments for G-actin such that the free G-actin concentration was below the critical concentration of the basal F-actin but above that of stimulated F-actin. This might occur if barbed ends were made available by GTP␥S stimulation. Interestingly, profilin at concentrations up to 100 M caused little decrease (9% Ϯ 8%) in the level of F-actin achieved after GTP␥S addition (Fig. 1C). The presence of profilin had little effect on the rate of polymerization; rather the increase in the magnitude of the change in F-actin seen in the presence of profilin resulted from polymerization continuing until it reached the same level (or occasionally greater) in the presence as in the absence of exogenous profilin ( Fig. 2A). In contrast, in the presence of exogenous T␤4, polymerization never reached the level achieved in its absence (Fig. 2B). The ability of profilin to decrease basal F-actin without significantly inhibiting GTP␥S induced polymerization is consistent with the hypothesis that barbed ends were capped in the resting cell and freed upon stimulation.
The freeing of barbed ends, by shifting the affinity of F-actin (for G-actin), could account for the concentration dependence of both DNase I and thymosin ␤ 4 on the amount of GTP␥Sinduced actin polymerization. The data from these experiments were compared with those predicted from a simple model (see "Experimental Procedures") in which the F-actin increase equalled the amount of G-actin released from a T␤4 complex and polymerizing when the affinity of the F-actin shifted from 0.5 to 0.1 M. When we assumed that the concentration of monomer binder in the cell by the time of fixation (5 min) equalled the concentration added to the medium, the predicted increases in F-actin were similar to those observed (Fig. 3). The fact that the model fits both the DNase I and T␤4 data suggested that stimulation was not selectively inactivating T␤4, since this would have made T␤4 a less effective inhibitor than DNase I. The model did not replicate the slight increase in the magnitude of the F-actin induced by GTP␥S in the presence of low concentrations of T␤4. An increase in the magnitude of F-actin induced by GTP␥S was also seen with profilin at all concentrations but not with DNase I. This increase was probably due to polymerization of the profilin and T␤4-actin complexes that arose as a result of the depolymerization of basal F-actin. These complexes were ignored in the model, as noted under "Experimental Procedures," but could contribute to the increase in F-actin after stimulation. However, the DNase Iactin complexes are of too high affinity to contribute.
The hypothesis that GTP␥S acts by making free barbed ends available is supported by experiments with cytochalasin. Cytochalasin B inhibited much of the GTP␥S-induced polymerization. Maximal inhibition was achieved with concentrations of cytochalasin B between 2 and 10 M. At these concentrations, the polymerization induced by 100 M GTP␥S was inhibited by 64 Ϯ 3% (Fig. 4A). Since in intact cells cytochalasin B completely inhibits the polymerization induced by chemoattractant, it is not clear why the inhibition in SLO-permeabilized cells was not complete. In the presence of cytochalasin, the magnitude of the decrease in F-actin caused by either profilin (Fig. 4B) or 40 M T␤4 (not shown) was similar before and after GTP␥S stimulation. This ruled out the possibility that the failure of profilin to decrease the GTP␥S-induced F-actin in the absence of cytochalasin was due to GTP␥S inhibiting its sequestering ability. The presence of cytochalasin B had little effect on basal F-actin level or on the ability of 40 M profilin to lower the basal F-actin levels (Fig. 4B). This is consistent with the hypothesis that the steady state was already determined primarily by the pointed ends.
The Rate and Level of Polymerization Depended on the Concentration of GTP␥S Added-If GTP␥S induces actin polymerization by making barbed ends available for polymerization, one might expect that the rate of polymerization would depend on the number of barbed ends free and thus on the GTP␥S concentration. Indeed, the rate of polymerization was a function of the concentration of GTP␥S added. A plateau level of staining was achieved about 45 s after the addition of 100 M GTP␥S, about 1 min after addition of 1 M GTP␥S and about 4 min after the addition of 100 nM GTP␥S (Fig. 5).
To determine if the cessation of polymerization at the plateau was due to depletion of GTP␥S, we examined whether GTP␥S was still available at the time the plateau level of F-actin was reached. Supernatants of permeabilized cells that had been incubated 15 min with or without a suboptimal concentration of GTP␥S (300 nM) were ultrafiltered (cut off 10,000 daltons) to remove proteins such as G-actin and monomer binding proteins and then tested for their ability to stimulate polymerization in freshly permeabilized cells. The supernatant from control cells had no effect on F-actin levels, but the supernatant from stimulated cells induced polymerization to approximately the same level as that observed with the original cells (data not shown). Thus, the termination of net polymerization at plateau was not limited by depletion of GTP␥S.
Cells Stimulated with Suboptimal Concentrations of GTP␥S Were Able to Increase F-actin Level upon Addition of More GTP␥S-It was interesting that the plateau level of polymerization depended on the concentration of GTP␥S added. If a significant number of barbed ends were freed and remained free upon stimulation, then even at low concentrations of GTP␥S, polymerization might continue until the free G-actin was decreased nearly to the critical concentration of the barbed end (Young et al., 1990). It was possible that the plateau level of polymerization achieved depended on the rate of polymerization and was limited eventually by depletion of the cellular concentration of G-actin, which was continually being lost from the permeabilized cells. However, cells stimulated with a suboptimal dose of GTP␥S were able to further polymerize actin when stimulated again with a higher concentration of GTP␥S, indicating that after the F-actin had reached a plateau, a reservoir of G-actin and G-proteins remained and was capable of responding (Fig. 6). Cells incubated for 4 min with 0.1 M GTP␥S increased their F-actin level further when 100 M GTP␥S was added. The amount of polymerization achieved after the sequential additions of GTP␥S was greater than that induced in cells warmed for 6 min before addition of 100 M GTP␥S. Cell responsiveness does decrease with the duration of warming, probably because G-actin and other components are leaving the cell (Redmond et al., 1994); the higher level of F-actin achieved after sequential additions of GTP␥S may be due to the fact that the initial polymerization decreased the amount of G-actin that left the cell during this period. 3 Barbed Ends Remain Free after the F-actin Has Reached Its Plateau Level-The sustained level of polymerization seen following GTP␥S stimulation may indicate that GTP␥S inhibited depolymerization or that the F-actin in permeabilized cells, like lysed cells, was inhibited from depolymerizing (Cano et al., 1992b). However, addition of various concentrations of T␤4 (10, 20, or 40 M) at 6 min after permeabilization (4 min after GTP␥S stimulation) caused equal decreases in stimulated and basal F-actin (not shown). Alternatively, the sustained F-actin level may indicate that a fraction of filaments still have their barbed ends free. This possibility is supported by the fact that addition of cytochalasin lowered the plateau level of F-actin. 3 min after addition of cytochalasin B, there was a loss of 35 Ϯ 7%, n ϭ 3, of the F-actin induced by GTP␥S (Fig. 7). The addition of cytochalasin B to unstimulated cells had little effect on basal F-actin.
This conclusion that barbed ends remained free was further supported by the observation that addition of profilin to stimulated cells did not decrease the plateau level of F-actin. The TRITC-phalloidin staining remained unchanged or was increased slightly (Fig. 8A). Profilin did cause depolymerization of basal F-actin in unstimulated cells that had been permeabilized for the same period of time, indicating that prolonged permeabilization did not on its own make cells insensitive to profilin (Fig. 8B). However, when profilin was added to cells that had been stimulated with GTP␥S and then treated with cytochalasin B to block all barbed ends, it did cause depolymerization (Fig. 8A). Thus, when the barbed ends were capped with cytochalasin, profilin did cause depolymerization even in the presence of GTP␥S. This indicated that the differential effects of profilin on basal and stimulated cells were due to differences in the availability of free barbed ends and reinforced the conclusion that GTP␥S did not inhibit the G-actin sequestering ability of profilin.
The Number of Free Barbed Ends Available at Plateau Appears to be Low-The effects of profilin, cytochalasin, and the differential effects of DNase I and T␤4 on basal and stimulated F-actin all supported the conclusion that GTP␥S induced actin polymerization by increasing the availability of free barbed ends. However, the fraction of filaments in the permeabilized cells having free barbed ends was not known. The presence of barbed ends would be expected to increase the rate of depolymerization when the free G-actin concentration was below the critical concentration. We compared the rate of depolymerization of resting and stimulated cells after addition of DNase I to decrease the free G-actin concentration. Addition of DNase I (25 M) decreased the GTP␥S-induced and basal F-actin levels (Fig. 9). The initial rate of depolymerization of GTP␥S-induced F-actin was only slightly greater (about 2-fold) than that of basal F-actin. With time, DNase I reduced the F-actin in both control and stimulated cells to Յ20% of the original basal level. Thus, the stimulated F-actin did depolymerize, and the failure to detect differences in the rates of depolymerization between control and stimulated cells could not be due to the fact that only basal F-actin was depolymerizing.
To determine if this slight increase in rate was due to barbed-end depolymerization, we compared the rate of depolymerization induced by DNase I in the presence and absence of cytochalasin, which blocks barbed end depolymerization. The presence of cytochalasin did not detectably decrease the rate of depolymerization over that seen with DNase alone (data not shown). Nor did the presence of cytochalasin decrease the rate of depolymerization observed when (a) the DNase concentration was increased to 100 M to insure that the G-actin concentration was below the critical concentration of the barbed end or (b) the DNase was added soon after GTP␥S when the F-actin level was still rising.
GTP␥S Increased the Number of Sites Able to Nucleate Actin Polymerization after Cell Lysis-To determine if permeablized cells, like intact cells, did increase the number of nucleation sites available after lysis, we examined the ability of permeabilized cells lysed with Triton to nucleate barbed-end elongation of exogenous pyrenyl actin (Cano et al., 1991). GTP␥S stimulation of SLO-permeabilized cells increased the number of filament ends that were free following cell lysis. The time course of increase in nucleation sites following stimulation with 100 M GTP␥S paralleled the F-actin polymerization; a maximum increase occurred at about 45 s and then was maintained for at least 10 min (Fig. 10A). The number of sites induced within 3 min of stimulation with GTP␥S depended on the concentration of GTP␥S added. With 100 M GTP␥S, the number was 2.04 Ϯ 0.43 times that of unstimulated cells (n ϭ 5); with 1 M GTP␥S, the number was 1.52 Ϯ 0.30 times the basal level (n ϭ 5) (Fig. 10B). It was not possible to detect an increase with concentrations less than 0.3 M GTP␥S.
It is important to note that unstimulated cells also nucleated barbed-end elongation after lysis. The presence of 2 M cytochalasin reduced the rate of polymerization of 1.5 M pyrenylactin more than 80%, suggesting that after lysis, greater than 20% of the filaments had free barbed ends (Korn et al., 1987). However, as noted above, the effects of adding cytochalasin and profilin to unstimulated cells suggested that most barbed ends were capped. Thus, sites that nucleate barbed end polymerization after lysis may not have been free in the permeabilized cell before lysis.
In the Permeabilized Cell, the Fraction of Filaments with Free Barbed Ends May Be Small-The similar rate of DNase I induced depolymerization of resting and stimulated F-actin and the lack of effect by cytochalasin suggested that in stimulated cells a relatively small fraction of the filaments had free barbed ends. This is compatible with the fact that freeing a small fraction of the total number of barbed ends is sufficient to shift the actin critical concentration (Young et al., 1990). With pure actin, using parameters described under "Experimental Procedures," the transition from all barbed ends capped to all barbed ends uncapped would shift the critical concentration from 0.5 to 0.115 M; uncapping only 10% of the filaments could decrease the critical concentration from 0.5 to 0.2 M. Since profilin has been shown to alter the level of polymerization, we asked whether profilin would affect this change in critical concentration as a function of free barbed ends. Indeed, the presence of profilin exaggerates the effects of uncapping. Using rate constants estimated by Pantaloni and Carlier (1993) in the presence of 5 M profilin and sufficient G-actin, freeing 10% of the barbed end would decrease the critical concentration to 0.046 M; freeing only 1% of the ends would decrease the critical concentration to 0.226 M (Fig. 11; Pring et al., 1992). In the presence of 30 M profilin, the critical concentrations with 10 and 1% of the barbed ends free would be 0.025 and 0.14 M, respectively. We estimated from preliminary data (see "Experimental Procedures") that about 10 M profilin would be present in the cells 2 min after permeabilization. Thus, the freeing of less than 10% barbed ends could significantly shift the critical concentration and induce polymerization.

DISCUSSION
Permeabilized Cells Are Useful for Studies of the Regulation of Actin Polymerization-Since permeabilized neutrophils respond to agonists by polymerizing actin, they allow investigation, not possible after cell lysis, of the mechanisms regulating the actin steady state. Permeabilized neutrophils mimic the behavior of intact neutrophils in that they can be stimulated through chemoattractant receptor or downstream of the receptor by GTP␥S (Redmond et al., 1994). The basal F-actin level as well as that induced by stimulation are similar to those in intact cells. Finally, as demonstrated here, like intact cells, the increase in F-actin is associated with an increase in the number of free barbed ends available after lysis to nucleate actin elongation of exogenous pyrenyl actin.
For the studies described here we have chosen to stimulate polymerization with GTP␥S. GTP␥S presumably acts, at least in part, through the pertussis toxin-sensitive trimeric G-protein through which chemoattractants act, although it may also activate downstream small G-proteins (Redmond et al., 1994). . At various times, 100 l of cells were withdrawn and mixed with 900 l of an assay buffer that contained 1 M pyrenyl G-actin. The initial rate of increase in pyrenyl G-actin polymerization (pyrenyl fluorescence) was plotted as a function of stimulation time before lysis. The data from five different experiments were pooled by setting as 100% the rate of pyrenyl G-actin polymerization induced by unstimulated permeabilized cells. Error bars indicate the S.D. B, dose response. SLO-permeabilized cells were stimulated with various concentrations of GTP␥S. 3 min later, the cells were lysed, and the initial rate of pyrenyl G-actin polymerization was measured as described above. The data from four different experiments were pooled as described above.
Stimulation with GTP␥S has the advantages that (a) it induces a large and stable polymerization of F-actin and (b) the response is not limited by receptor dynamics, e.g. phosphorylation and internalization, or by GTP hydrolysis. Thus the magnitude of the F-actin increase may closely reflect the properties of actin steady state.
While the cytoplasm of a permeabilized cell is changing over time as soluble components diffuse out, it is possible to systematically investigate the effects of exogenous factors within a limited time window. Because the experiments are very rapid, one can alter the composition of the cytoplasm without the secondary effects that can result from chronic alterations, for example, following transfection of intact cells.
Filament Barbed Ends Are Mostly Capped in Unstimulated Cells but Are Available following GTP␥S Addition-The effect of exogenous profilin on permeabilized cells has provided strong evidence that (a) the presence of profilin alone was not sufficient to stimulate polymerization and (b) GTP␥S-induced polymerization and maintenance of actin polymerization occurs through a shift in the actin steady state toward that of the barbed ends. Thus, while addition of exogenous DNase I and T␤4 caused depolymerization of basal F-actin and inhibited the increase in F-actin induced by GTP␥S, addition of profilin decreased the basal F-actin level but had little or no effect on the GTP␥S-induced increase in F-actin. Since the profilin-Gactin complex, unlike complexes with the other G-actin binding proteins, does not block polymerization at the filament barbed end, these data suggested that in the resting cell most barbed ends were capped but after stimulation at least some were free. This conclusion was supported by the observation that profilin decreased the F-actin after stimulation only if cytochalasin was also present. This observation ruled out the possibility that GTP␥S inactivated profilin. While these studies do not rule out the possibility that stimulation regulates other factors, the regulation of the availability of free barbed ends may account for many other agonist-induced changes in steady state.
A 5-fold decrease in the critical concentration caused by freeing barbed ends could account for the greater inhibition of basal than stimulated F-actin by DNase I and T␤4. The fact that both DNase and T␤4 were equally well fit by a simple model suggests that the properties of T␤4 were not modified, even transiently, by stimulation with GTP␥S. Furthermore, in the presence of cytochalasin, both profilin and T␤4 caused similar decreases in F-actin in the presence or absence of GTP␥S, indicating that neither protein was inhibited by GTP␥S. This extends previous studies that had shown that there was no stable modification of T␤4 following stimulation (Safer et al., 1990;Cassimeris et al., 1992;Nachmias et al., 1993), but had left open the possibility that it might be transiently modified at the time of stimulation. Local modification of T␤4 or profilin is not ruled out by the studies presented here.
The fraction of filaments that have free barbed ends before and after polymerization in the permeabilized cell is not known. In the resting cell, the fraction appears to be low as profilin acts as an effective monomer-sequestering protein.
However, the fact that profilin did not decrease resting F-actin levels as well as T␤4 may indicate that in the resting cell some barbed ends are free. Indeed, in some experiments cytochalasin did decrease the basal F-actin slightly, and profilin had a slightly greater effect on basal F-actin when cytochalasin was present (see Fig. 4b). Although cell lysates made after stimulation showed a large increase in the number of free barbed ends, in the permeabilized cell, it was not possible to detect the presence of free barbed ends by studies of the rate of depolymerization. The 2-fold increase in rate of depolymerization induced by DNase I in stimulated versus control cells may result from a 2-fold increase in filament number associated with the increase in F-actin, i.e. 2 times more pointed ends (Cano et al., 1991). Furthermore, the presence of cytochalasin did not detectably slow the DNase-induced depolymerization of stimulated F-actin.
One can imagine various factors including the presence of cross-linking proteins that might limit our ability to detect barbed ends from the kinetics of depolymerization. We therefore also examine the rate of polymerization induced by phalloidin. Addition of phalloidin to permeabilized cells stimulated actin polymerization in both resting and stimulated cells. 3 The phalloidin presumably acts by inhibiting depolymerization while allowing addition of monomer at either filament end. However, the rate of polymerization was no greater in stimulated than in control cells, even when the cells were stimulated with suboptimal concentrations of GTP␥S to insure that a reservoir of G-actin remained. Nor could we detect an effect of cytochalasin on this rate, even though in vitro cytochalasin effectively inhibited polymerization at the barbed end in the presence of phalloidin.
The discrepancy between the data showing that free barbed ends determine the stimulated F-actin steady state and the inability to actually measure free barbed ends in stimulated permeabilized cells is reconciled by a calculation showing that in the presence of profilin, a very small fraction of filaments with free barbed ends can shift the critical concentration. Thus, the number available at any time may be well below current detection limits. These studies shed no light on the mechanism through which barbed ends become available upon GTP␥S stimulation. They may arise from uncapping, cutting of filaments or de novo nucleation of new filaments. These mechanisms remain to be defined.
Conclusion-Studies using neutrophils permeable to profilin and other G-actin sequestering proteins provide strong evidence that GTP␥S increases and maintains elevated F-actin levels by increasing the fraction of actin filaments with barbed ends available for polymerization. The number of free barbed ends appears to be a small fraction of the total filaments present.  (uncapped). Profilin binding to free G-actin and to the barbed filament ends was assumed to be at equilibrium, after Pantaloni and Carlier (1993), with K d values of 0.5 and 7 M, respectively. The on and off rate constants of both G-actin and profilin-G-actin at the barbed ends were 10 M Ϫ1 s Ϫ1 and 1 s Ϫ1 , and for G-actin at the pointed ends 0.4 M Ϫ1 s Ϫ1 and 0.2 s Ϫ1 .