Rapid Nucleotide Exchange Renders Asp-11 Mutant Actins Resistant to Depolymerizing Activity of Cofilin, Leading to Dominant Toxicity in Vivo*

Background: Mutation of Asp-11 is dominant negative in yeast and human actins. Results: Mutant actins exchange bound nucleotides rapidly, cannot bind cofilin, and cofilin-induced depolymerization of mutant and wild type copolymers is slow. Conclusion: Rapid nucleotide exchange with exogenous ATP inhibits cofilin-mediated depolymerization of copolymers, leading to dominant toxicity. Significance: Mechanism of a dominant negative actin mutation is elucidated. Conserved Asp-11 of actin is a part of the nucleotide binding pocket, and its mutation to Gln is dominant lethal in yeast, whereas the mutation to Asn in human α-actin dominantly causes congenital myopathy. To elucidate the molecular mechanism of those dominant negative effects, we prepared Dictyostelium versions of D11N and D11Q mutant actins and characterized them in vitro. D11N and D11Q actins underwent salt-dependent reversible polymerization, although the resultant polymerization products contained small anomalous structures in addition to filaments of normal appearance. Both monomeric and polymeric D11Q actin released bound nucleotides more rapidly than the wild type, and intriguingly, both monomeric and polymeric D11Q actins hardly bound cofilin. The deficiency in cofilin binding can be explained by rapid exchange of bound nucleotide with ATP in solution, because cofilin does not bind ATP-bound actin. Copolymers of D11Q and wild type actins bound cofilin, but cofilin-induced depolymerization of the copolymers was slower than that of wild type filaments, which may presumably be the primary reason why this mutant actin is dominantly toxic in vivo. Purified D11N actin was unstable, which made its quantitative biochemical characterization difficult. However, monomeric D11N actin released nucleotides even faster than D11Q, and we speculate that D11N actin also exerts its toxic effects in vivo through a defective interaction with cofilin. We have recently found that two other dominant negative actin mutants are also defective in cofilin binding, and we propose that the defective cofilin binder is a major class of dominant negative actin mutants.


Conserved Asp-11 of actin is a part of the nucleotide binding pocket, and its mutation to Gln is dominant lethal in yeast,
whereas the mutation to Asn in human ␣-actin dominantly causes congenital myopathy. To elucidate the molecular mechanism of those dominant negative effects, we prepared Dictyostelium versions of D11N and D11Q mutant actins and characterized them in vitro. D11N and D11Q actins underwent salt-dependent reversible polymerization, although the resultant polymerization products contained small anomalous structures in addition to filaments of normal appearance. Both monomeric and polymeric D11Q actin released bound nucleotides more rapidly than the wild type, and intriguingly, both monomeric and polymeric D11Q actins hardly bound cofilin. The deficiency in cofilin binding can be explained by rapid exchange of bound nucleotide with ATP in solution, because cofilin does not bind ATP-bound actin. Copolymers of D11Q and wild type actins bound cofilin, but cofilin-induced depolymerization of the copolymers was slower than that of wild type filaments, which may presumably be the primary reason why this mutant actin is dominantly toxic in vivo. Purified D11N actin was unstable, which made its quantitative biochemical characterization difficult. However, monomeric D11N actin released nucleotides even faster than D11Q, and we speculate that D11N actin also exerts its toxic effects in vivo through a defective interaction with cofilin. We have recently found that two other dominant negative actin mutants are also defective in cofilin binding, and we propose that the defective cofilin binder is a major class of dominant negative actin mutants.
Actin plays a number of important roles in eukaryotic cells, including amoeboid movement, cytokinesis, adhesion, intracellular transport, endocytosis/exocytosis, as well as nuclear roles in transcription regulation. To perform these functions, actin filaments need to be polymerized and depolymerized in a spatially and temporally regulated manner. A number of actinbinding proteins, including those that affect polymerization and depolymerization, have been characterized in vitro, but the physiological functions of each of those actin regulatory proteins in vivo are not fully understood.
Mutant proteins are generally useful tools to elucidate the molecular mechanism of protein functions, and dominant negative mutants often provide unique opportunities in those studies. Because actin is important in many cellular functions, a large number of actin mutations, some of which are dominant negative, have been identified from experimental genetic screens (1)(2)(3)(4) as well as from analyses of human hereditary diseases (5)(6)(7)(8)(9)(10)(11). Unfortunately, however, the expression of recombinant actin requires eukaryotic host cells (12), and the expression of dominantly negative mutant actins would be toxic for the host cells, hampering their biochemical characterization (1). To resolve this problem, we developed a system to express toxic mutant actins in Dictyostelium, which takes advantage of thymosin ␤ that was fused at the C terminus to mutant actin to inhibit copolymerization of the recombinant mutant actin with the endogenous actin of the host cells. The fusion protein was purified and treated with chymotrypsin, which efficiently cleaved the protein immediately after the native final residue of actin, to separate the actin and thymosin moieties (13). This expression system was used to characterize dominant negative actin mutations (14) previously identified in Drosophila indirect flight muscle (2), as well as in our own genetic screens using yeast cells (3). These studies demonstrated the usefulness of dominant negative mutant actins and prompted us to characterize other dominant negative actin mutants.
Asp-11 in the budding yeast actin sequence is a part of the nucleotide-binding site ( Fig. 1) and is conserved among all known actins. The D11Q mutant ␣-actin (D13Q in the ␣-actin and D12Q in Dictyostelium actin 15 sequences; the yeast amino acid residue number will be used throughout this paper) translated in vitro was found to be partially defective in binding to DNase I (15). Subsequent overexpression of the corresponding mutant yeast actin in yeast cells demonstrated that it is a dominant lethal mutation (16), although purification of the protein for detailed biochemical characterization was impossible due to the dominant lethality. More recently, the D11N mutation in human ␣-actin was identified to dominantly cause congenital myopathy (7), further pointing to the need to characterize Asp-11-mutant actins in vitro. We thus constructed Dictyostelium versions of the D11N and D11Q mutant actins, purified them using our thymosin-fusion system, and characterized them in a variety of biochemical assays. Our results demonstrated that D11N/D11Q actin polymers are abnormal in a number of ways, including rapid release of bound nucleotides and defective cofilin binding and cofilin-mediated depolymerization, the former presumably causing the latter. Most notably, the copolymer of D11Q and WT actins also exhibited partial resistance to cofilin activity to accelerate depolymerization. Because cofilin is essential for yeast viability (17,18) and for proper muscle development (19) and functions (20,21), defective cofilin-induced depolymerization of WT and D11Q actin copolymers suggested that defective cofilin binding is the primary reason why D11N/D11Q actins are dominantly toxic in yeast and human.

EXPERIMENTAL PROCEDURES
Plasmid Construction-pTIKL ART (13) contains an ART gene, which is the Dictyostelium act15 gene modified to carry four unique restriction sites (the AR gene), followed by a Glybased linker, a synthetic human thymosin ␤ gene, and a His tag. pTIKL GFP-AR carries the GFP-fused AR gene (13). D11N and D11Q mutations were made by a PCR-based method and were subcloned into pTIKL ART and pTIKL GFP-AR after confirmation by DNA sequencing. The mutated sequences are GCTTTAGTTATTAATAACGGTTCTG and GCTTTAGT-TATTCAAAACGGTTCTG for D11N and D11Q, respectively, in which the underlines show the mutated residues.
Cell Culture-pTIKL-based plasmids were electroporated into Dictyostelium discoideum Ax2 or KAX3 cells (22). Transfected cells were selected on plates at 21-22°C in HL5 medium containing 60 g/ml each of penicillin and streptomycin and 12 g/ml G418. For biochemical purification of actin, KAX3 cells expressing either WT or mutant ART were grown on plates in medium containing 40 g/ml G418. Large scale cultures were grown in 25 ϫ 25-cm 2 plastic plates or in 5-liter conical flasks and grown for an additional 24 -36 h with fresh HL5 medium with 10 g/ml G418 on a shaker.
Analyses of GFP Mutant Actin in Dictyostelium Cells-Ax2 cells expressing GFP-actin or its derivative were observed using a confocal laser scanning microscope (13). For Western blotting analysis, washed cells were incubated in 10 mM Hepes, pH 7.4, 150 mM NaCl, 2 mM MgCl 2 , 1 mM EGTA, 1 mM DTT, 0.5% Triton X-100 on ice for 5 min and centrifuged at 15,000 rpm for 5 min at 5°C. Supernatant and pellet fractions were subjected to SDS-PAGE, followed by staining with anti-GFP antibodies raised in rabbits.
Purification of Actin and Cofilin-Recombinant WT and mutant actins were purified as described previously (14). The concentration of actin was estimated using Advanced Protein Assay (Cytoskeleton, Denver, CO) using actin that was calibrated by absorption at 290 nm as the standard.
cDNA of Dictyostelium cofilin (Ddcof1) (23) was isolated from a cDNA library using the primers 5Ј-GGTACCAT-GTCTTCAGGTATTGCT and 5Ј-CAATTGGATTTTGGTA-CATTTTTCAT and, after sequence verification, was inserted at the BamHI and EcoRI sites of pCold I (Takara Bio, Otsu, Japan) or pCold I carrying an mCherry gene inserted at the EcoRI and XbaI sites. These were used to express N-terminally His-tagged cofilin or cofilin-mCherry in Rosetta (DE3) Escherichia coli cells, which were purified using conventional methods.
Polymerization Assay-Monomeric WT or mutant actin was diluted in G-buffer (2 mM Hepes, pH 7.4, 0.2 mM CaCl 2 , 0.1 mM ATP, 7 mM ␤-mercaptoethanol, 0.05% NaN 3 ) and incubated on ice for 10 min. Polymerization was induced by the addition of concentrated F-buffer, and the increase in light scatter was monitored at 360 nm at 22°C, using a 100-l cuvette and a fluorescence spectrophotometer. The final concentration of each component was 5 M actin, 2 mM Hepes, pH 7.4, 100 mM KCl, 2.5 mM MgCl 2 , 0.5 mM EGTA, 0.2 mM ATP, and 0.2 mM DTT.
Depolymerization Assays-WT, mutant, or a 1:1 mixture of WT and mutant actins were allowed to polymerize in F-buffer at RT for 2 h. The concentration of total actin was 5 M. Latrunculin A (Wako, Osaka, Japan) or DMSO control was added at a final concentration of 60 M or 0.3%. Following a 10-min incu- bation, the mixtures were centrifuged at 250,000 ϫ g for 10 min at 20°C, and the supernatant and pellet fractions were subjected to SDS-PAGE. Alternatively, WT actin labeled at Cys-374 with pyrene (24) was used, and the decrease in pyrene fluorescence was monitored using a fluorescence spectrophotometer with excitation and emission wavelengths of 365 and 407 nm, respectively.
For direct observation of depolymerization of individual actin filaments, actin filaments (10 M) were labeled with 200 M Alexa-Fluor 488 succinimidyl ester (Invitrogen, Tokyo, Japan) in a buffer consisting of 2 mM Hepes, pH 7.4, 50 mM KCl, 2.5 mM MgCl 2 , 0.5 mM EGTA, and 0.2 mM ATP on ice overnight. The reaction was stopped by the addition of 0.1 M Tris-Cl, pH 7. 4. After removing unbound dye by ion exchange resin (Dowex, 1 ϫ 80, 100 -200 mesh), the labeled actin was dialyzed against G-buffer. The resultant labeled monomeric actin was mixed with either unlabeled WT or D11Q actin in G-buffer at a 1:1 ratio, polymerized in F-buffer at RT for 2 h as above and, after dilution in assay buffer (10 mM Hepes, pH 7.4, 25 mM KCl, 4 mM MgCl 2 , 10 mM DTT, and 0.5% bovine serum albumin), introduced into flow cells coated with 25 g/ml skeletal heavy meromyosin, followed by incubation for 2 min. The flow cells were then perfused with a copious amount of F-buffer containing 10 mM DTT and the oxygen scavenger system, and actin filaments were observed with a fluorescence microscope (BX60, Olympus, Tokyo, Japan) equipped with an EB-CCD camera (C7190, Hamamatsu Photonics, Hamamatsu, Japan) at 25°C.
Phosphate Release Assay-The time course of P i release from polymerizing actin was measured by using an EnzChek phosphate assay kit (Invitrogen). Actin was polymerized at 10 M, as described under "Polymerization Assay," in the presence of 2-amino-6-mercapto-7-methylpurine riboside and 1 unit/ml purine nucleoside phosphorylase, and the absorbance at 360 nm was monitored.
Stopped Flow Analyses-The rates of 1,N 6 -ethenoadenosine 5Ј-triphosphate (⑀-ATP) 5 release from monomeric actin was measured at 25°C using a stopped flow system (SX18MV: Applied Photophysics, Leatherhead, UK). Actin filaments were dialyzed against G-buffer for 24 h, followed by second dialysis against G-buffer that contained 0.2 mM ⑀-ATP (Invitrogen) in place of ATP for 24 h (WT) or 48 h (mutants). This solution was rapidly mixed with an equal volume of G-buffer that contained 1 mM CaATP, and fluorescence excited by 360 nm light and passed through a 395-nm cutoff filter was monitored.
Cofilin Binding Assay-WT, mutant, or a 1:1 mixture of WT and mutant actins were allowed to polymerize in 20 mM Pipes, pH 6.5, 50 mM KCl, 2.5 mM MgCl 2 , 0.5 mM EGTA, 0.2 mM ATP, and 0.2 mM DTT at RT for 2 h. Cofilin was added at a final concentration of 2.5 M, and 5 min later, the mixture was centrifuged at 250,000 ϫ g at 20°C for 10 min. The supernatant and pellet fractions were subjected to SDS-PAGE.
For microscopic observation of cofilin binding to actin filaments, 1 M Alexa-Fluor 488-labeled actin filaments were mixed with 2 M cofilin-mCherry in buffer consisting of 10 mM Pipes, pH 6.5, 50 mM KCl, 1 mM MgCl 2 , 5 mM DTT, and 1 mM ATP. After incubation for 1 min at 25°C, labeled actin filaments were diluted 10-fold in the same buffer and observed with a fluorescence microscope.
Cofilin binding to monomeric actin was detected by crosslinking 7 M actin and 14 M cofilin in G-buffer with 40 mM 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide (Sigma) for 5 min at 25°C. The reaction was stopped by the addition of SDS sample buffer, and the samples were analyzed by SDS-PAGE.
Cofilin-induced Depolymerization Assay-WT, D11Q, and 1:1 mixtures of Alexa-Fluor 488-labeled WT actin and unlabeled WT or D11Q actin were polymerized in F-buffer at RT for 2 h or at 5°C overnight. These were diluted to 5 M in F-buffer that contained 2 mM Hepes, pH 7. 4. Hepes, pH 8.35, and cofilin were added to the final concentrations of 10 mM and 10 M, respectively, and 15 min later, the mixture was centrifuged at 250,000 ϫ g for 10 min at 25°C. The supernatant and pellet fractions were subjected to SDS-PAGE and stained with Coomassie Blue. The gel was also viewed on a Molecular Dynamics Typhoon 8600 Imager (GE Healthcare) with 532 nm excitation light.
For microscopic observation of cofilin-induced severing/depolymerization of actin filaments, copolymers of Alexa-Fluor 488-labeled WT actin and either unlabeled WT or D11Q actin were diluted to 40 nM in Hepes buffer (10 mM Hepes, pH 8.35, 30 mM KCl, 1 mM MgCl 2 , 0.2 mM ATP, and 5 mM DTT), including 2 M cofilin. After incubation for 5 min at 25°C, the mixture was observed with a fluorescence microscope.
DNase I Inhibition Assay-DNase I activity was measured at 22°C by the change in A 260 when DNA (50 g/ml) (Sigma) was added to a mixture of 0.7 nM DNase I (Sigma) and 10 nM G-actin in G-buffer.

RESULTS
Purification of D11N/D11Q Actins-D11N and D11Q mutant actins were expressed and purified as fusion proteins with thymosin ␤ and a polyhistidine tag. After separating actin and thymosin-His tag moieties by chymotryptic digestion, we were able to further purify mutant actins by Q-Sepharose column chromatography, followed by a cycle of polymerization, pelleting by ultracentrifugation, and dialysis against G-buffer, in a manner similar to WT actin.
Polymerization of D11N and D11Q Actins-When monomeric D11Q actin in G-buffer was induced to polymerize by the addition of salts, light scatter increased with a time course slower than the WT actin ( Fig. 2A). Filaments were visualized by fluorescence microscopy following rhodamine-phalloidin staining of the D11Q polymers (Fig. 2B), suggesting that D11Q actin filaments were normal. Salt-induced polymerization of different batches of D11N actin yielded variable patterns of increase in light scattering (data not shown). However, as an example in Fig. 2A, they had a common tendency to steadily rise well above the steady state levels achieved by the same concentration of WT actin (arrow), suggestive of slow aggregate formation. When partially purified D11N actin from a Q-Sepharose column was allowed to polymerize and stained with rhodamine-phalloidin, a large number of normal filaments were visualized. However, when those filaments were dialyzed against G-buffer and ultracentrifuged and the supernatant fraction following ultracentrifugation was allowed to polymerize and stained with rhodamine-phalloidin, relatively few very short filaments, many of which appeared as dots, were observed (Fig. 2B).
We next observed negatively stained D11N/D11Q actins under polymerization conditions by electron microscopy (Fig.  3). To our surprise, D11N/D11Q polymers contained small oligomeric structures without a noticeable double-helical appearance ( Fig. 3, C and E), and numerous ring-like structures were observed in D11N polymers (Fig. 3, E and F). There were relatively few filaments of normal appearance, particularly in D11N polymers. The addition of phalloidin did not noticeably increase the filamentous fractions of D11Q polymers (Fig. 3D). These results and the fact that those mutant actins were purified normally by a cycle of polymerization and depolymerization together indicate that D11N/D11Q actins are able to undergo salt-dependent reversible polymerization, but the polymerized products contain normal filaments and abnormal oligomeric structures. Each D11N ring appeared to consist of five subunits (Fig. 3F). Similar rings had been observed with Ca 2ϩ -WT actin of fission yeast under polymerization conditions (25), although the fission yeast actin rings were somewhat smaller than those of Asp-11-mutant actins.
When a 1:1 mixture of D11Q actin labeled with Alexa-Fluor 594 and WT actin labeled with Alexa-Fluor 488 were allowed to polymerize and observed under a fluorescence microscope, the same filaments were visualized by both fluorophores (Fig. 4A). Interestingly, fluorescence intensities of both Alexa-Fluor 488 and Alexa-Fluor 594 were not homogeneous along the length of copolymers, suggesting the possibility that WT and D11Q actins tend to segregate from each other and form clusters. In addition, very bright fluorescent dots of Alexa-Fluor 594 were

Asp-11 Mutant Actin
observed along the length (Fig. 4A, arrowheads). This may represent oligomeric structures of D11Q actin associated along the sides of filaments, as seen in electron micrographs (arrowheads in Fig. 3, C and D). When GFP-fused D11Q actin was expressed in Dictyostelium cells, the fluorescence was localized along cell edges, thin projections, and macropinocytotic cups (Fig. 4B). This is similar to the distribution of GFP-WT actin and is distinct from nonpolymerizable GFP-WT actin fused C-terminally with thymosin ␤ (Fig. 4B). The cytoplasmic fluorescence of GFP-D11Q actin, which was derived from monomeric or oligomeric GFP-actin, was low and was comparable with that of GFP-WT actin, suggesting that both GFP-D11Q and GFP-WT actins copolymerized with endogenous actin with a similar efficiency in vivo. Western blotting analysis (Fig. 4C) showed that 53 Ϯ 15% of GFP-WT actin was recovered in the Triton-insoluble fraction, whereas 30 Ϯ 2.3% of GFP-D11Q actin was in the insoluble fraction (n ϭ 3; the insoluble fraction of GFP-D11Q actin fused with thymosin ␤ was 7.0 Ϯ 0.8%).
Depolymerization of D11Q Actins-Next, depolymerization of D11Q polymer was analyzed by three different methods. In the first experiment, depolymerization was induced by the addition of latrunculin A to sequester monomeric actin from the solution (26), and after incubation for 10 min, the polymeric and depolymerized fractions were separated by ultracentrifugation followed by SDS-PAGE (Fig. 5A). Most of the WT actin was recovered in the supernatant fraction after the latrunculin treatment. In contrast, the majority of D11Q actin molecules were pelleted after the latrunculin treatment. Intriguingly, a 1:1 mixture of WT and D11Q copolymer was also more resistant to the latrunculin treatment than WT filaments. D11N actin behaved similarly to D11Q actin in this depolymerization assay (Fig. 5B). In the second experiment, we copolymerized pyrenyl WT actin and D11Q actin, and after the addition of latrunculin A, the decrease in pyrene fluorescence was followed to monitor the depolymerization of WT subunits in the copolymers (Fig.  5C). This experiment demonstrated that WT subunits that copolymerized with D11Q actin were indeed significantly slower to depolymerize than those in WT homopolymer filaments.
These results suggest that Asp-11-mutant actins are not only slow to depolymerize by themselves, but they also slowly depolymerize WT actin that copolymerized with them. However, we were unable to rule out the possibility that D11Q actin did not bind latrunculin A. Additionally, we were unable to rule out the  possibility that the fluorescence signal observed in Fig. 5C was derived from pyrenyl WT actin trapped in nonfilamentous oligomeric structures. Thus, in the third set of experiments, copolymers of Alexa-Fluor 488 WT actin and either unlabeled WT or D11Q actin were immobilized on glass surfaces coated with skeletal heavy meromyosin, and after washing out free monomeric actin, depolymerization of individual filaments was followed by fluorescence microscopy. After 10 min of incubation, originally longer WT filaments became shorter than WT/D11Q copolymers (Fig. 5D). Filaments often fragmented, presumably due to strong excitation light, and very short fragments diffused away because the density of heavy meromyosin molecules on the surface was low. We identified fragmentation events based on sequential images taken at 5-min intervals and analyzed the changes in length of unfragmented filaments. WT homopolymer filaments shortened at a rate of 0.16 Ϯ 0.083 m/min (average Ϯ S.D., n ϭ 55), which is roughly consistent with the estimate of skeletal Mg 2ϩ -actin (0.1 m/s) under slightly different buffer conditions (27). In contrast, Alexa-Fluor 488-WT/D11Q copolymer shortened at a 2-fold slower rate (0.072 Ϯ 0.049 m/min; n ϭ 56). This result, statistically significant by Student's t test (p Ͻ 10 Ϫ8 ), qualitatively confirmed those of the latrunculin experiments and demonstrated that copolymerization with D11Q actin slows depolymerization of WT actin in filaments.

Effect of Asp-11 Mutations on Nucleotide Exchange and Phosphate Release
Rates-The strategic position of Asp-11 in the nucleotide binding pocket and the aberrant polymerization/ depolymerization properties of the Asp-11-mutant actins suggested that those mutant actins have altered nucleotide binding properties. Thus, we first examined the release rates of bound nucleotides by measuring the decrease in the fluorescence of ⑀-ATP when bound ⑀-ATP was released in the presence of excess ATP (Fig. 6A). ⑀-ATP that was bound to WT actin was released at a rate of 0.012 Ϯ 0.0029 s Ϫ1 , which is consistent with previous measurements using skeletal actin (28,29). In contrast, ⑀-ATP bound to monomeric D11Q actin was released at an ϳ40-fold faster rate of 0.42 Ϯ 0.098 s Ϫ1 . Release from D11N actin was even 10-fold faster, at 4.0 Ϯ 0.16 s Ϫ1 . We have not directly measured the affinities of WT or mutant actins for ATP, but the extremely rapid dissociation of ⑀-ATP from the mutant actins suggested much lower affinity of monomeric mutant actins, especially of D11N actin, for ATP. This speculation and the fact that purified D11N actin loses competence to polymerize normal filaments (Fig. 2), as well as large batch-tobatch variations among different D11N preparations (data not shown), suggested that D11N actin quickly denatured during and/or after dialysis against G-buffer containing 0.1 mM ATP. This precluded D11N actin from further quantitative biochemical characterizations, and we focused our subsequent analyses on D11Q actin.
We next compared the rates of nucleotide release from WT and D11Q filaments (Fig. 6B). Consistent with previous measurements that nucleotide release is very slow from skeletal actin filaments (30 -32), the increase in fluorescence of ⑀-ATP was very small and slow, if at all, when phalloidin-stabilized WT filaments that were dialyzed against F-buffer containing 0.1 mM ATP and then treated with the Dowex resin to remove free ATP were mixed with 0.1 mM ⑀-ATP. In contrast, a large increase in fluorescence intensity was observed with D11Q filaments over the 3-h time course. The time course did not fit a single exponential curve well, suggesting the presence of two or more different populations of D11Q subunits, such as the ATP-bound and ADP-bound forms or the normal filaments and oligomeric structures. In any case, it was clearly demonstrated that D11Q actin subunits in filaments release bound nucleotides much more rapidly than the WT subunits in filaments and rebind ATP.

Asp-11 Mutant Actin
A nearly stoichiometric amount of phosphate was released from polymerizing WT actin with a relatively short lag following an increase in light scattering ( Fig. 2A). Phosphate was released from polymerizing D11Q actin as well, although the lag between polymerization and phosphate release was significantly larger with D11Q actin, suggesting either slower ATP hydrolysis or higher affinity for P i after hydrolysis on D11Q actin subunits in filaments. Nonetheless, despite the rapid release of bound ATP, D11Q actin clearly retains the activity to hydrolyze ATP in a polymerization-dependent manner.
Effect of Asp-11 Mutations on Cofilin Binding and Depolymerization by Cofilin-Cofilin is the major actin depolymerizing factor in vivo (33,34), with activities to sever filaments (35), depolymerize actin filaments (36,37) by enhancing subunit dissociation from pointed ends of the filaments (38), and bind to actin monomers (36,37). Thus, we next examined the effects of D11Q mutation on interactions with cofilin.
Cosedimentation of cofilin with actin polymers at pH 6.5, a condition under which cofilin binds to actin filaments without significantly depolymerizing them (39 -41), showed that, although WT filaments and copolymers of WT and D11Q actins bound cofilin with similar affinities, D11Q homopolymers hardly bound cofilin (Fig. 7A). However, it was not possible to determine unequivocally that D11Q subunits within normal homopolymer filaments bound cofilin, because unknown fractions of the mutant actin molecules were sequestered in oligomeric structures. Thus, we added mCherry-fused cofilin to Alexa-Fluor 488-labeled D11Q or WT filaments, and we found under a fluorescence microscope that cofilin-mCherry hardly bound the D11Q filaments, although it bound and disassembled WT filaments (Fig. 7B).
Binding of cofilin to monomeric D11Q actin was next assayed by cross-linking in G-buffer. Although G-buffer has a very low concentration of salts and would enhance actin-cofilin binding more than under physiological conditions, we found that monomeric D11Q actin cross-linked to cofilin at a significantly slower rate than WT actin (Fig. 7C). Taken together, we concluded that D11Q actin has lower affinities for cofilin in both filamentous and monomeric forms, as well as in the small oligomeric structures.
We next assayed the activities of cofilin against D11Q actin at pH 8.3, the condition under which cofilin efficiently depolymerizes actin filaments (39 -41), using two different assays. In the first set of experiments, cofilin-induced depolymerization was assayed by ultracentrifugation followed by SDS-PAGE. Incubation with 10 M cofilin for 15 min released more than half of the subunits to the supernatant fraction from the WT filaments. In contrast, copolymer filaments of Alexa-Fluor 488-WT actin and unlabeled D11Q actin were significantly resistant to depolymerization by cofilin (Fig. 8A). Observation of gel fluorescence demonstrated that Alexa-Fluor-labeled WT actin was also protected from cofilin activity when copolymerized with D11Q actin (Fig. 8A). Densitometric scanning of the gels showed that cofilin-induced depolymerization of Alexa-Fluor 488-WT actin in copolymers with D11Q actin was 38 Ϯ 9% (n ϭ 3) of that in homopolymers.
In the second set of experiments, fluorescence microscopy was used to monitor the disappearance of WT homopolymer and WT/D11Q copolymer filaments. When 40 nM Alexa-Fluor 488-WT actin filaments were incubated with 2 M cofilin, filaments disappeared almost completely within 6 min (Fig. 8B). In contrast, many filaments remained when 40 nM 1:1 copolymer filaments of Alexa-Fluor 488-WT and unlabeled D11Q actins were treated with 2 M cofilin for 6 min. Some filaments remained even after 26 min (data not shown), further indicating that copolymer filaments of WT and D11Q actins were significantly resistant to the depolymerizing activity of cofilin.
Strikingly, in F-buffer that contained 1 mM ADP in addition to 0.1 mM ATP, cofilin was able to depolymerize D11Q filaments fairly efficiently, although not as efficiently as WT actin in the presence of ADP (Fig. 9). This result suggested that ADP in the buffer was incorporated into D11Q subunits in the filaments, due to the very rapid exchange of bound nucleotides, Arrow shows the position of the cross-linked actin-cofilin. Average of three independent measurements indicated that the cross-linking of D11Q actin was 47 Ϯ 15% slower than WT actin, and this difference is statistically significant with p Ͻ 0.05 by Student's t test. Higher molecular weight ladders formed in D11Q-cofilin cross-link reactions were formed even when D11N or D11Q actin, but not WT actin, was treated with 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide in G-buffer in the absence of cofilin. and made them susceptible to cofilin activity, whereas in the standard F-buffer that only contained ATP, most of the D11Q subunits carried bound ATP even if bound ATP was hydrolyzed to ADP, which conferred resistance to cofilin activity. Under a more physiological condition, i.e. in the presence of 1 mM ATP and 50 M ADP (42), D11Q actin was resistant to cofilin activity, suggesting that D11Q actin subunits were resistant to cofilin activity in vivo.
Allosteric Effect of the Asp-11 Mutations on the Structure of DNase Loop-Finally, we examined the effects of the Asp-11 mutations on the structure of the DNase loop. In the first experiment, the structure of the DNase loop was monitored by its susceptibility to cleavage by subtilisin (43). SDS-PAGE analysis showed that monomeric WT actin was cleaved almost completely by 1 g/ml subtilisin in 5 min under our experimental conditions. In contrast, monomeric D11Q actin was cleaved at a much slower rate under the same conditions (Fig. 10A).
The DNase loop is involved in binding DNase I and inhibiting its enzymatic activity (44,45). Again, D11Q actin was less efficient than WT actin to bind and inhibit DNase I (Fig. 10B).

DISCUSSION
The Asp-11 mutation has been shown to be dominantly negative both in yeast actin (16) and human ␣-actin, the latter leading to congenital myopathy (7). Our in vitro characterizations revealed that both D11N and D11Q mutant actins undergo salt-dependent reversible polymerization, and the resultant filaments appear normal when observed by low reso-lution electron microscopy. However, relatively few normal filaments were formed with purified D11N actin, presumably due to denaturation during overnight dialysis against G-buffer containing 0.1 mM ATP, which forced us to focus our detailed biochemical analyses on D11Q actin. D11Q filaments moved more or less normally on surfaces coated with skeletal heavy meromyosin (supplemental data),

Asp-11 Mutant Actin
and interaction of monomeric D11Q with profilin and thymosin ␤ appeared normal as well in the context of fusion proteins in vivo (supplemental data). Nonetheless, D11Q actin showed a number of biochemical defects. For instance, D11Q actin filaments depolymerized more slowly than WT filaments, as did copolymer filaments of WT and D11Q actins (Fig. 5). Furthermore, both monomer and filament forms of D11Q rapidly exchanged bound nucleotides with free nucleotides in solution (Fig. 6) and failed to interact properly with cofilin (Figs. 7-9). ADP concentration is much lower than ATP in cells (42) as well as in standard G-and F-buffers, so that the rapid exchange of bound nucleotides would allow most of the D11Q actin molecules to carry ATP, even if the hydrolysis activity is normal, both in vivo and in vitro. ATP-bound skeletal actin is slower to depolymerize than ADP-bound actin (46), and this may be at least one of the reasons why D11Q actin filaments are slower to depolymerize. In copolymer filaments of D11Q and WT, slow dissociation of ATP-bound D11Q subunits from depolymerizing ends would cause pauses, leading to slower average depolymerization rates of copolymers. Although our analyses on D11N actin were limited, we believe the same explanation is applicable to D11N actin because D11N actin monomers released bound nucleotides even more rapidly.
In cells, however, spontaneous depolymerization of actin filaments is unlikely to play any important roles, as the concentration of monomeric actin is above the critical concentration for polymerization, and it is generally believed that cofilin-mediated depolymerization from the pointed ends of filaments is physiologically relevant (38). D11Q actin monomers and homopolymers do not bind cofilin (Figs. 7 and 8) and render copolymer filaments with WT actin partially resistant to the depolymerizing activity of cofilin (Fig. 8). Considering very rapid turnover of actin subunits in dynamic structures within nonmuscle cells (47)(48)(49), this slow depolymerization of copolymer filaments of WT and D11Q mutant actins may well be deleterious for nonmuscle cells, including yeast. Again, the rapid exchange of bound nucleotides would explain why D11Q filaments do not bind cofilin, because the cellular concentration of ATP is much higher than that of ADP (42), and cofilin is unable to bind ATP-bound actin filaments (38). This view is consistent with the fact that D11Q filaments were efficiently depolymerized by cofilin in F-buffer containing 1 mM ADP in place of ATP (Fig. 9).
The inability of monomeric D11Q actin molecules to bind cofilin would cause additional problems. The cellular concentration of total actin is well above the critical concentration for polymerization, and a number of actin-binding proteins are present to maintain a polymerization-competent monomeric actin pool. Although there is evidence against the simple idea that cofilin sequesters monomeric ADP-actin from polymerization (50), differential high affinity of cofilin for ADP-actin monomer over ATP-actin monomer (38,51) suggests a role of cofilin in this complex process, which would not work with Asp-11-mutant actins in the cells.
In light of the traditional notion that the turnover of sarcomeric actin is slow in muscle cells (52), it is not intuitively obvious if the same cofilin-related explanations are applicable to the dominant negative effect of D11N mutation in the ␣-actin gene leading to myopathy. However, it is now established that actin subunits turnover rapidly in both developing and mature muscle cells (53). Furthermore, cofilin is expressed in muscle cells (54), and recent studies provided evidence that mutation in cofilin causes nemaline myopathy (21) and that cofilin is required for muscle development (19) and maintenance (20). In the mutant muscle cells, D11N actin is probably present at the same concentration as WT. This would lead to a modest retardation of cofilin-mediated depolymerization of copolymer actin filaments and disturb the turnover of the actin monomer, so that the mutant muscle cells become sick but do not die. Dictyostelium cells expressing GFP-D11Q actin did not show noticeable defects in growth or cell morphology (data not shown). This is presumably because the relative content of GFP-D11Q actin was much less than that of endogenous actin, as was the case with other GFP-mutant actins (14).
G146V is another dominant lethal actin mutation in yeast, which also inhibits cofilin binding (3). The K336I mutation, which in human ␣-actin causes congenital myopathy (7), also makes Dictyostelium actin incapable of binding cofilin. 6 Furthermore, the P332A mutation in ␥-actin, which causes autosomal dominant nonsyndromic progressive deafness, was resistant to depolymerization by cofilin (55). Taken together, we propose that a significant fraction of polymerization-competent dominant negative mutant actins exerts toxic effects by dominantly disturbing cofilin-mediated dynamic regulation of the actin cytoskeleton. It was recently found that N12D mutation in the ␤-actin gene causes Baraitser-Winter syndrome (11), and it will be interesting to investigate if this mutation, which occurred right next to D11N in the opposite direction, increases or decreases the sensitivity to the cofilin activity.
A recent high resolution structural study demonstrated that the side chain of Asp-11 indirectly interacts with the ␤-phosphate of ADP through a water molecule (56). It thus makes sense that mutating Asp-11 changes the affinity for nucleotides, although it is not intuitively obvious why changing to Asn causes a more severe phenotype than to Gln. We speculate that this modification of nucleotide binding affinity can explain much of the defective interaction of D11Q actin with cofilin. However, cofilin was unable to depolymerize D11Q actin filaments as efficiently as WT filaments even in the presence of 1 mM ADP. This is difficult to explain by rapid nucleotide exchange, and it may be due to slower ATP hydrolysis or P i release from D11Q actin in filaments. The slower polymerization of Asp-11-mutant actins is also difficult to explain by rapid nucleotide exchange. Furthermore, D11Q mutation allosterically affected the conformation of subdomain 2 of monomeric actin in G-buffer, and this too is difficult to explain by rapid nucleotide exchange, together suggesting additional mechanisms by which Asp-11 mutations affect the interaction with cofilin and/or impair cellular function of the mutant actin. Allosteric interactions between subdomain 2 and the nucleotide binding cleft (57,58), including those involving cofilin (59), have been reported. Thus, this allosteric effect of Asp-11 mutations on subdomain 2 may impair the interaction with cofilin and other actin subunits during polymerization, because subdomain 2 is a major binding site for cofilin (60,61) as well as for the adjacent actin subunit within the same protofilament (56,(62)(63)(64). Detailed structural analyses of Asp-11-mutant actins should shed light on these biologically important intramolecular communications.