Xenopus Actin Depolymerizing Factor/Cofilin (XAC) Is Responsible for the Turnover of Actin Filaments in Listeria monocytogenes Tails

The COOH-terminal 1,393 bp (464 amino acids) of X. l. gelsolin was cloned by PCR from a λ-YES (Stratagene, La Jolla, CA) Xenopus egg and embryo cDNA library (gift from Jeremy Minshull) (Kinoshita et al., 1995) into a pGEX-2T vector (Pharmacia LKB Biotechnology Inc., Piscataway, NJ). XAC 2 was cloned into a pGEX expression vector as described by Abe et al. (1996). The pGEX expression plasmids were transformed into TG1, and recombinant proteins were expressed and purified on a glutathione column using standard procedures (Smith and Johnson, 1988). For use in the addback experiments or for purification of antibodies, the glutathione S–transferase (GST)-fusion proteins were cleaved with 0.4 mg/ml thrombin (Sigma Chemical Co., St. Louis, MO) in thrombin buffer (100 mM NaCl, 50 mM Tris HCl, pH 7.5, 2.5 mM CaCl₂, 5 mM MgCl₂, 1 mM DTT) at 37°C for 60 min. Thrombin was then removed by passing the cleaved protein over a p-aminobenzamidine Sepharose (Sigma Chemical Co.) column in thrombin buffer and concentrating the flow through a centriprep-10 concentrator (Amicon, Beverly, MA).

Xenopus egg extracts were made as described in Theriot et al. (1994). Briefly, after dejellying meiotically arrested Xenopus laevis eggs in 2% cysteine, pH 7.8, the eggs were washed 4× in 250 ml of XB (100 mM KCl, 10 mM Hepes, pH 7.7, 50 mM sucrose, 5 mM EGTA, 2 mM MgCl₂, 0.1 mM CaCl₂), transferred to 5-ml tubes containing 50 μl 0.5 M EGTA, 5 μl 1 M MgCl₂, 5 μl 1,000× protease inhibitor mix (10 mg/ml leupeptin, pepstatin, and chymostatin in DMSO), and crushed at 10,000 rpm in an HB-4 rotor (Sorvall Instruments, Newtown, CT) at 15°C. The cytoplasmic layer was removed with a 20-gauge needle and syringe and 1/20 volume of energy mix (150 mM creatine phosphate, 20 mM ATP, 2 mM EGTA, pH 7.7, and 20 mM MgCl₂) was added. Aliquots were frozen in liquid nitrogen and stored at −80°C for up to 6 mo.

GST fusion proteins with XAC 2 or gelsolin fragments expressed in Escherichia coli were used for rabbit antibody production (Berkeley Antibody Co., Berkeley, CA). The antibodies were affinity purified on pure XAC or gelsolin cleaved with thrombin from the corresponding GST- fusion proteins expressed in E. coli. Antibodies were affinity purified using published procedures (Harlow and Lane, 1988). Antibodies were eluted from the affinity column with 100 mM glycine, pH 2.5, 150 mM NaCl, neutralized, and dialyzed against 10 mM Hepes, pH 7.7, 100 mM KCl, concentrated using Aquacide II, and redialyzed against 10 mM Hepes, pH 7.7, 100 mM KCl.

XL177 cells were grown on glass coverslips to ∼60% confluency. Cells were infected with the Listeria monocytogenes strain 10403S as described (Theriot et al., 1994) except that ∼10-fold more Listeria were used, and the cells were incubated for 8 h at 23°C (4 h before and 4 h after gentamycin addition) before processing. Coverslips were rinsed in TBS (20 mM Tris, pH 7.4, 150 mM NaCl) before fixation in 4% formaldehyde in TBS for 20 min. The coverslips were rinsed in TBS before cells were permeabilized in TBS + 0.5% Triton X-100 for 10 min. After blocking for 10 min in AbDil (TBS + 2% BSA and 0.5% Na azide), the coverslips were incubated for 30 min with either 2 μg/ml anti-XAC or 6 μg/ml antigelsolin antibodies in AbDil. Texas red–conjugated goat anti–rabbit secondary antibody was used to visualize gelsolin and XAC, and fluorescein-phalloidin (Molecular Probes, Eugene, OR) was used to visualize actin. Coverslips were mounted with FITC-guard (Testog Inc., Chicago, IL).

100 μg random rabbit IgG (Accurate Chemical and Scientific Corp., Westbury, NY), anti-XAC antibody, or antigelsolin antibody was bound to 30 μl Affiprep protein A (BioRad Labs, Hercules, CA) in TBST for 1 h at 4°C. The pellets were washed with 3× 1 ml XB and then incubated with 50 μl crude cytostatic factor–arrested Xenopus egg extracts for 1 h at 4°C on a rotator. The pellet was removed by centrifuging at 10,000 g in an Eppendorf centrifuge for 20 s. The supernatant was removed and treated as the immunodepleted extract. The pellets were washed 3× 1 ml in TBST, boiled in sample buffer, and analyzed by SDS-PAGE.

Western blots were performed by transferring SDS-PAGE gels electrophoretically to nitrocellulose in 20 mM Tris, 25 mM glycine, 20% methanol. Blots were incubated 1 h in AbDil followed by 1 h of incubation in 2.5 μg/ml antigelsolin or anti-XAC antibody in AbDil at room temperature. Alkaline phosphatase–conjugated anti-rabbit antibody was used as a secondary antibody (Promega Corp., Madison, WI). The amount of gelsolin or XAC depleted from extracts was determined using densitometry of immunoblots by comparing the band intensity of 1 μl of depleted extract to the band intensities of serially diluted undepleted extracts. Immunoblots were digitized using a scanner (model Power Look; UMAX Systems, Hsinchu, Taiwan) and analyzed using Adobe Photoshop (Adobe Systems Inc., Mountain View, CA). Purified XAC and ADF as well as the immunoprecipitates were visualized by staining with 0.25% Coomassie blue R-250 in 45% methanol and 10% acetic acid followed by destaining in 25% methanol and 7% acetic acid.

Listeria monocytogenes strain SLCC-5764 (Leimeister-Wachter and Chakraborty, 1989) was grown overnight at 37°C with constant shaking to stationary phase in brain–heart infusion broth (BHI; Difco Laboratories Inc., Detroit, MI). The Listeria were killed by adding 10 mM iodoacetic acid and incubating for 20 min at room temperature (Theriot et al., 1994). The bacteria were washed once in XB, resuspended in 1/5 original volume in 20% glycerol/XB, and stored at −80°C. Rabbit skeletal muscle actin covalently labeled with N-hydroxysuccinimidyl 5-carboxytetramethyl rhodamine (Molecular Probes Inc.) was made as previously described (Rosenblatt et al., 1995).

Listeria tail morphology and Listeria motility were assayed by mixing 5 μl of depleted or undepleted extract with 0.5 μl each of Listeria and 1 mg/ml rhodamine-labeled actin. In experiments where XAC or ADF proteins were added, XB or proteins were added in a volume of 0.5 μl. 1 μl of this mixture was removed and squashed between a microscope slide and a 22mm² coverslip and allowed to incubate at room temperature for 25 min. Static images or movies of tails were collected using a CCD camera (Princeton Instruments, Trenton, NJ) and fluorescence movies of bacterial motility were acquired using a video camera (model SIT; Dage-MTI, Inc., Wabash, MI), respectively, during a period of 25–60 min after transferring reactions to room temperature. The lengths of tails and total tail fluorescence were quantitated using Winview software (Princeton Instruments, Trenton, NJ). Tail lengths were measured using the program Get Curve (Princeton Instruments), and the length in pixels was converted to microns using a micrometer standard. Total fluorescence in the Listeria tails was measured by multiplying the pixel area by average pixel intensity of a selected area minus the average pixel intensity of a background selected area. The CCD camera responds linearly to fluorescence intensity in the range of 10–3,000 counts/pixel, and we used illumination levels that avoided saturating the signal.

ATP or AMPPNP actin filaments were made by diluting rhodamine- labeled actin in G-buffer (5 mM Tris HCl, pH 8.0, 0.2 mM CaCl₂, 0.2 mM DTT) containing either 0.2 mM ATP or 0.2 mM AMPPNP to a final concentration of 12.8 μM in 100 μl. The mixtures were either passed by gravity or spun through 1 ml G-25 (Pharmacia LKB Biotechnology, Inc.) columns preequilibrated in G-buffer plus the 0.2 mM of the appropriate nucleotide for 1 min in a clinical centrifuge at mid-speed into tubes containing 25 μl 0.25 M KCl, 0.25 M Tris, pH 8.0, and 1 μl 0.1 M of the appropriate nucleotide and incubated for 30 min at room temperature. Filamentous actin was recovered by centrifuging the actin for 15 min at 436,000 g in a centrifuge (model TLA100; Beckman Instrs., Palo Alto, CA) and resuspending the pellet in 100 μl F-buffer (50 mM KCl, 50 mM Tris, pH 8.0, 0.2 mM DTT, 0.5 mM ATP or AMPPNP).

Nucleotide incorporation was analyzed by centrifuging the various F-actin preparations through a 600 μl 40% glycerol F-buffer (50 mM KCl, 50 mM Tris, pH 8.0, 0.2 mM DTT) cushion at 436,000 g in a table top centrifuge (model TLA100; Beckman Instrs.) for 60 min. The pellet was resuspended in 50 μl 8 M urea for 15 min at room temperature and then diluted with 100 μl of H₂O and spun through a 10-kD cut-off filter (Millipore Corp., Bedford, MA). The nucleotides in the filtrate were then analyzed by HPLC on a 1 ml Mono Q column (Pharmacia LKB Biotechnology, Inc.) equilibrated in 100 mM NH₄HCO₃ and eluted in a 100–500 mM NH₄ HCO₃ gradient over 30 min at a flow rate of 1 ml/min. Peak areas were analyzed and recorded at OD₂₅₄ using Gilson software (Worthington, OH).

The ATP- or AMPPNP-containing F-actin was then mixed 1:1 with F-buffer, 0.1 mg/ml recombinant XAC, or crude cytostatic factor–arrested Xenopus egg extracts. Remaining filaments from the above mixtures were visualized on the microscope and quantitated by fluorimetry. Images of the above reactions were recorded by squashing 1 μl of each reaction between a microscope slide and a 22-mm² coverslip using a microscope (Nikon, Inc., Melville, NY), a CCD camera, and Winview software. After incubating the F-actins with buffer, XAC, or extract for 10 min at room temperature, the remaining F-actin in the mixture was pelleted at 436,000 g for 15 min at 4°C in a centrifuge (model TLA100; Beckman Instrs.). The pellets were resuspended in 0.1% SDS and the fluoresence was measured on a fluorimeter (model Aminco; SLM Instruments, Inc., Urbana, IL). Percent remaining F-actin was calculated as the fluorescence of XAC- or extract-treated pellet/fluorescence of buffer-treated pellet.

Since we suspected that severing proteins might accelerate actin turnover in Listeria tails, we examined the localization of two candidate actin-severing proteins, gelsolin and XAC, in Listeria-infected Xenopus tissue culture cells (XL 177). Antibodies were raised to the COOH-terminal half of Xenopus laevis gelsolin (Ankenbauer et al., 1988) and to the full-length XAC 2 protein (Abe et al., 1996) and affinity purified. Since XAC 1 and 2 differ by only four amino acids, we made only antibodies to XAC 2, which should also recognize XAC 1. Immunoblots of XL 177 lysate show that antibodies to gelsolin and XAC recognize a single band of the expected molecular mass in both cases (Fig. 1,A, lanes 2 and 3). Fluorescein-labeled phalloidin and affinity-purified antibodies to XAC or gelsolin were used to visualize the intracellular distributions of F-actin (Fig. 1, B and D), gelsolin (Fig. 1,C), and XAC (Fig. 1 E) in XL 177 cells infected with Listeria monocytogenes. Both XAC and gelsolin colocalize with the F-actin staining in the Listeria tails. Both gelsolin and XAC antibodies also give punctate staining throughout the rest of the cell. In addition, gelsolin typically stained coincidentally with F-actin in the stress fibers of the cell, whereas XAC does not.

To determine the function of gelsolin and XAC in Listeria tails, we compared Listeria tails in mock-depleted Xenopus egg extracts to those in extracts depleted of XAC or gelsolin. Using our antibodies complexed to protein A beads, we were able to remove the bulk of either protein from Xenopus egg extracts. Quantitative immunoblots show that ∼75% of XAC (Fig. 2,B, lane 2) and >95% of gelsolin (Fig. 2,A, lane 3) were depleted from extracts compared to random rabbit IgG–depleted extracts (Fig. 2, A and B, lane 1). The XAC antibody precipitated XAC (19 kD) and an unknown band with an approximate molecular mass of 28 kD (Fig. 2,C, lane 2). The gelsolin antibody precipitated only gelsolin (∼93 kD) (Fig. 2 C, lane 3) when compared to the rabbit IgG control pellet (lane 1).

The IgG control depletion (Fig. 3,A) produced Listeria tails of the same length as untreated extract (15 ± 2.5 and 18 ± 1.5 μm, respectively). The tails in the gelsolin- depleted extracts (Fig. 3,B) were the same length and contained the same polymer mass as those of the IgG- depleted extracts (Fig. 3,A). However, the tails formed in the XAC-depleted extracts (Fig. 3,C) were on the average four to six times longer and displayed 13–21-fold more total fluorescence than those of the control extracts. Addback of pure recombinant XAC (Fig. 3,D) or chicken ADF to approximately endogenous concentrations (2.7 μM) rescued the long tail phenotype (12.3 ± 1.4 and 11.4 ± 1.6 μm, respectively). Addition of the same amount of a recombinant mutant version of ADF (S3E ADF) (Fig. 3,E), which behaves like constitutively phosphorylated ADF and has ∼10% of wild-type ADF activity on pure actin (data not shown), does not rescue the long tail phenotype. The purity of the proteins used in the addback experiment (Fig. 2,D, lanes 1 and 2) and the inability of the S3E ADF to rescue the XAC depletion phenotype strongly suggest that the observed long tail phenotype is due to the removal of XAC. The graph in Fig. 3 F shows the quantitation of the tail lengths and the amounts of tail fluorescence of the different phenotypes and provides quantitative support for our conclusion that increased tail length is due to XAC depletion, and not depletion of some other protein. These experiments demonstrate that XAC is required for the rapid depolymerization of actin filaments in Listeria tails in Xenopus extracts. While gelsolin is also concentrated within these tails, it does not appear to be essential for the depolymerization of Listeria tail actin.

Quantitative immunoblots of Xenopus egg extracts using XAC antibodies and bacterially expressed XAC as a standard revealed that XAC is present at ∼2.1 μM in extracts (data not shown). Addition of recombinant XAC to a final concentration of 7.1 μM in the extract (Fig. 4,B) produced tails that were ∼0.33 times the length and had ninefold less total fluorescence than tails in a control extract (Fig. 4,D). Doubling the amount of XAC added to the Listeria assay to give a final concentration of 12.1 μM did not result in any further decrease in tail length (Fig. 4, C and D). When XAC was added to concentrations above 12.1 μM final concentration, few tails formed. Instead, rodlike structures containing rhodamine actin could be seen throughout the extract (data not shown). These rods appear analogous to those seen when actin and cofilin are concentrated in the nucleus upon heat shock or DMSO treatment to cells (Nishida et al., 1987; Ono et al., 1993) and probably represent a nonfilamentous coaggregate of XAC and actin. It is likely that few tails can form in such high concentrations of XAC since the actin required for Listeria tail formation may be sequestered in these XAC/actin rodlike aggregates. Thus, a Listeria tail segment of ∼5 μm is resistant to XAC depolymerization even when XAC is added up to nearly saturating concentrations.

To determine whether the addition or depletion of XAC or gelsolin had an effect on the rate of actin polymerization, we measured the rate of Listeria movement in extracts either depleted of XAC or gelsolin, or containing additional XAC (Fig. 5). We used Listeria movement as an assay since the movement rate reflects the actin polymerization rate at the bacterial surface (Sanger et al., 1992; Theriot et al., 1992). The rate of Listeria movement in the XAC depleted extracts did not vary greatly from the mock-depleted extracts. The lack of effect that XAC-depletion had on actin assembly rates may indicate that XAC is not involved in actin polymerization. However, we cannot rule out such an involvement since only 75% of the XAC could be removed from the extracts with our reagents. Depletion of gelsolin increased the rate of Listeria movement by ∼1 μm/min (∼20%). The manipulations required for depletion slowed Listeria movement by ∼1 μm/min (compare IgG-depleted to addition of buffer). This may be due to a decrease of ATP, dilution of actin, or other factors during the depletion procedure (2–3 h at 4°C). Addition of XAC to 5.0 μM seemed to increase the rate of polymerization by ∼1 μm/min compared to when buffer alone is added, despite the shorter tails produced by this concentration (Fig. 4 B). These slight differences cannot account for the large changes in tail length and total fluorescence upon XAC depletion. Thus, the effects of XAC on actin depolymerization greatly outweigh those upon polymerization.

Upon polymerization, actin hydrolyzes its bound ATP and the terminal phosphate is slowly released. It has been suggested that the loss of the terminal phosphate could serve as a clock that regulates the lifetime of a filament by controlling the activity of ADF/cofilin proteins (Maciver et al., 1991; Moon and Drubin, 1995). To test this hypothesis, we made rhodamine-labeled actin filaments containing AMPPNP, a slowly hydrolyzable ATP analog, and tested their resistance to the depolymerizing activities of Xenopus egg extracts and purified XAC. Rhodamine ATP-actin filaments depolymerize rapidly in the presence of XAC (Fig. 6,C) and Xenopus egg extract (Fig. 6,E) compared to buffer (Fig. 6,A). In contrast, rhodamine-labeled AMPPNP filaments depolymerized to a limited extent in XAC (Fig. 6,D) or Xenopus egg extract (Fig. 6,F) compared to in buffer alone (Fig. 6,B). Thus, AMPPNP filaments were more stable to depolymerization in XAC or extract (Fig 6, D and F) compared to ATP filaments under the same conditions (Fig. 6, C and E). To quantitate actin filament depolymerization, rhodamine-labeled filaments were added to Xenopus extracts or purified XAC and total polymerized actin was recovered by sedimentation. The fraction of the added actin left in polymer was determined by resuspending the pellets in SDS and quantitating the fluorescence with a fluorimeter (Fig. 6,G). AMPPNP caused twice as much of the added F-actin to be recovered in the pellet fraction. Comparing Fig. 6, E and F, with G, we should note that the visual and sedimentation assays are measuring different parameters. If the labeled actin repolymerized into new filaments in the presence of extract, the filaments will not be visible to the CCD camera because they are diluted with endogenous actin (Fig. 6, E and F). However the diluted actin will still sediment, giving rise to the high fraction of label in the pellet fraction in the presence of extract (Fig. 6 G, Xenopus extract). Both types of assay demonstrate that the AMPPNP filaments are relatively resistant to depolymerization by pure XAC and total extract. Two other slowly hydrolyzable ATP analogs, ATPγS and AMPPCP, showed similar stabilizing effects (data not shown), suggesting that the actin stability is due to the state of the bound nucleotide rather than nonspecific effects on actin structure. The results from both the microscopy and the fluorimetry assays suggest that the nucleotide content of actin filaments regulates XAC.

Incorporation of AMPPNP into actin filaments in these experiments was modest; ∼12% of the ATP-binding sites in actin filaments incorporated AMPPNP, and the remainder contained ADP. By contrast, ATP actin filaments contained approximately only 2% ATP and 98% ADP. Despite low incorporation, ∼1.5 and 2 times as much AMPPNP-containing actin filaments as ATP-containing actin filaments pelleted in the presence of Xenopus extract or pure XAC, respectively.

We have used the ability to reconstitute Listeria motility in Xenopus egg extracts to test directly the role of the gelsolin and ADF/cofilin proteins in the promotion of actin filament depolymerization. We have raised specific antibodies to gelsolin and to XAC, the major ADF/cofilin protein known to be present in Xenopus egg extracts. Using immunodepletions and adding back purified proteins, our studies demonstrate that ADF/cofilin is the major factor responsible for the rapid turnover of actin filaments in Listeria tails. Removal of 75% of the ADF/cofilin from extracts considerably lengthened Listeria tails and greatly increased the total actin polymer mass present in the tails. Both tail length and actin polymer mass in tails were restored to control levels by adding back pure XAC at physiological concentrations to XAC-depleted extracts. In addition, adding excess XAC shortens Listeria tails and reduces the polymer mass in the tail, and AMPPNP actin is resistant to depolymerization by both XAC and whole extracts. Taken together, these data strongly implicate XAC as a central component of the machinery responsible for rapid actin filament turnover in extracts. Although we have no direct evidence, we predict that ADF/cofilin proteins are required for the dynamic organization of populations of actin filaments within living cells that turn over rapidly, such as the filaments within lamellipodia. Three arguments support our prediction: Xenopus egg extracts can support the actin polymerization and depolymerization in Listeria tails at rates comparable to those in intact cells and suggest they reflect the actin dynamics within cell cytoplasm. XAC is required to maintain the turnover of actin filaments in these tails in extracts. Because ADF/cofilin proteins are essential in every species in which they have been found, we may infer that they are required for essential processes like rapid actin turnover in the cell. Finally, since XAC is concentrated in Listeria tails and the leading edge of cells, we suspect it is also required for rapid filament turnover in vivo.

In contrast to XAC, gelsolin depletions had no significant effect on Listeria tail length or tail polymer mass, leaving the functional role of gelsolin in the regulation of actin dynamics an open question. Since gelsolin does not affect the depolymerization of Listeria tails, its concentration in these tails is curious. Perhaps gelsolin is concentrated in these tails strictly because of its actin binding activity, or by another of its activities such as actin capping. Although the Xenopus egg extracts used in our assays contain 5 mM EGTA, it is difficult to analyze any local concentrations of Ca²⁺ that may be due to vesicle release. Since Ca²⁺ is required for gelsolin activity, it is possible that the conditions in our cell-free egg extracts are not able to support the activity of gelsolin. Therefore, we cannot rule out a role for gelsolin in tail dynamics in vivo. However, the presence of gelsolin in Listeria tails does not greatly affect actin assembly or disassembly in Listeria tails in Xenopus extracts.

In our addition of varying amounts of excess XAC to extracts until all the actin was driven into abnormal rod structures, Listeria tails shrank to ∼5 μm but no further. The persistence of a resistant tail segment in up to sixfold the normal concentrations of XAC suggests that an additional factor may control the extent of actin depolymerization by XAC. Several investigators have suggested that the energy of ATP hydrolysis could be used to regulate the lifetime of a filament (Pollard, 1986; Carlier, 1988; Maciver et al., 1991; Moon and Drubin, 1995). In vitro studies of actin filament assembly have shown that upon polymerization of ATP-actin, the bound ATP is hydrolyzed and the terminal phosphate is slowly released (Carlier, 1987). This slow rate of phosphate loss relative to the polymerization rate should then produce filaments that contain a segment of ADP + inorganic phosphate (ADP.Pi). Since ATP- and ADP.Pi-bound actin are known to make stronger intersubunit bonds than ADP-actin (Carlier, 1991; Carlier et al., 1985), phosphate release could play a role in regulating actin filament stability. However, since in vitro filaments depolymerize very slowly, phosphate release alone is not sufficient to direct rapid disassembly of actin filaments in the cell. In vivo, perhaps the role of slow phosphate release in actin is to regulate either the binding or activity of actin-severing proteins. Severing proteins such as ADF and cofilin may preferentially sever the ADP subunits of actin filaments while the ATP and ADP.Pi subunits are resistant to severing (Fig. 7). In support of this, the depolymerizing activity of actophorin, an Acanthamoeba member of the ADF/cofilin family, can be inhibited by addition of 25 mM phosphate, which presumably mimics ADP.Pi filaments (Maciver et al., 1991). In addition, actophorin has been shown to bind tightly to ADPcontaining G-actin and weakly to ATP-actin (Maciver and Weeds, 1994). However, other studies show that chick ADF has a higher affinity for ATP-actin than for ADP-actin (Hayden et al., 1993) and have left the role of the bound nucleotide in regulating depolymerization in other species an open question. Our results showing the resistance of AMPPNP-containing actin filaments to the depolymerizing activities of both XAC and concentrated Xenopus egg extracts lend support to the idea that the nucleotide content of an actin filament controls filament lifetime by regulating the activity of ADF/cofilin family members.

What is the mechanism of actin depolymerization by XAC? XAC could depolymerize by end-wise removal of subunits, by severing, or by both mechanisms. We interpret our results with AMPPNP actin as favoring the endwise mechanism. We reason that a severing protein would not be greatly affected by an actin filament with only 12% of its subunits substituted with AMPPNP. By contrast, an end-wise depolymerizing protein would be blocked whenever AMPPNP subunits were present at the filament ends. Thus, the large inhibition of depolymerization by AMPPNP filaments may be accounted for if XAC primarily depolymerized using an end-wise mechanism. The most direct way of distinguishing severing and end-wise mechanisms will come from imaging depolymerization by ADF/ cofilin proteins.

On the basis of our findings, we can postulate a model for how XAC recycles actin subunits in the Listeria tail (Fig. 7). A complex of proteins at the back of Listeria (Welch et al., 1997) induces polymerization of ATP-bound actin. Once actin polymerizes, the bound ATP is hydrolyzed and the terminal phosphate is slowly released. The resulting ADP-containing actin subunits interact more weakly within the filament than the ATP subunits and allow binding of XAC. XAC either depolymerizes single subunits or short fragments of filament. XAC is then released from the depolymerized actin subunit and recycled for another round of depolymerization. The ADP in the depolymerized actin is then exchanged for ATP by bulk mass or by catalysis from profilin, and this subunit is now available for another round of polymerization or remains unpolymerized by binding thymosin β₄ or other sequestering proteins.

This model provides a framework for understanding how an actin subunit is recycled from one round of polymerization to the next in regions of the cell that rapidly turn over actin filaments. Clearly, many questions remain regarding the details of this model. Future work will need to address whether XAC works primarily by severing or end-wise mechanisms in the cell. We are currently examining how XAC is recycled for another round of depolymerization after it is bound to the actin subunit and the role that XAC phosphorylation may play in its recycling. Other studies will need to focus on how the actin subunit is recycled for repolymerization. Analysis of the nucleotide content of actin has revealed that ∼90% of unpolymerized actin in the cell is bound to ATP (Rosenblatt et al., 1995), suggesting that the actin nucleotide is exchanged early in the pathway. Information about whether the ATP- or ADP-bound actin is used for polymerization is still lacking. The 10% of actin that is ADP bound could be bound to XAC. This population of actin could undergo nucleotide exchange and be used directly for another round of polymerization, leaving the remaining 90% of actin sequestered from polymerization. However, the lack of an effect that XAC depletion has on the rate of polymerization at early time points would suggest that this is not the case. Future studies will need to address what population of actin is used for polymerization.

We thank the entire Mitchison lab for providing a stimulating and supportive environment and many great dinners. In particular, we thank Aneil Mallavarapu, Arshad Desai, and Jason Swedlow for help with computers and microscopes, Matt Welch for performing Listeria infections, Paul Peluso (University of California at San Francisco, San Francisco, CA) for help with the fluorimeter, and Claire Walczak for advice on cloning and bacterial protein expression. We are indebted to Matt Welch and especially Arshad Desai and Michael Redd (University of California at San Francisco, San Francisco, CA) for critically reading this manuscript and for stimulating conversations. We are grateful to Jeremy Minshull (Maxygen Corp., Santa Clara, CA) for giving us a Xenopus egg cDNA library and to Dr. Werner W. Franke (German Cancer Research Center, Heidelberg, Germany) for giving us X.l. gelsolin antibodies used for preliminary experiments. We also thank Tom Pollard (The Salk Institute, La Jolla, CA) who early on suggested depleting ADF/cofilin from Xenopus extracts.

This work was supported by a National Science Foundation predoctoral fellowship to J. Rosenblatt and National Institutes of Health grants (GM35126) to J.R. Bamburg and (GM48027) to T.J. Mitchison.

ADF

actin depolymerizing factor

ADP.Pi

ADP + inorganic phosphate

AMPPNP

5′adenylylamido-diphosphate

F-actin

filamentous actin

XAC

Xenopus ADF/Cofilin