Tropomyosin and Troponin Regulation of Wild Type and E93K Mutant Actin Filaments from Drosophila Flight Muscle

In the Drosophila flight muscle actin mutant E93K there is a charge reversal on the surface of actin close to the proposed position of tropomyosin when it is in the off state. Using a quantitative in vitro motility assay we have found that the wild type Drosophila ACT88F actin behaved like rabbit skeletal muscle actin when tropomyosin and troponin were added at pCa5 and pCa9. In contrast the effect of tropomyosin upon the E93K mutant actin filament movement was completely different from wild type and resembled the response of wild type with tropomyosin+troponin at pCa9 (i.e. the filaments were switched off). Velocity of E93K actin did not increase, and the fraction of filaments motile was reduced to less than 15% by adding up to 30 nmtropomyosin. When myosin subfragment-1 modified byN-ethylmaleimide was mixed with mutant E93K actin-tropomyosin filaments we observed that it restored motility of the filaments to the level observed with E93K actin alone. We conclude that electrostatic charge on the surface of domain 2 of actin plays a critical role in determining the state of actin-tropomyosin that is a central feature of the steric blocking mechanism of actin filament regulation.

Regulation of vertebrate striated muscle thin filament activity involves a Ca 2ϩ -sensitive change of the state of the thin filament, which affects the interaction between actin molecules and myosin cross-bridges (1). The thin filaments can be described as a two state allosteric and cooperative binding system in which the cooperative unit consists of seven actins and one tropomyosin: troponin I and myosin heads act as negative and positive allosteric effectors respectively (2,3). The equilibrium between the on and off states (K t in the model of Geeves and Halsall (2)) is regulated by Ca 2ϩ through the troponin complex and cross-bridge binding. Troponin-tropomyosin regulation of thin filament activity is found throughout the animal kingdom, including insect striated muscles (4 -6).
Tropomyosin is a dimeric ␣-helical coiled-coil that spans seven monomers along the actin filament (7). Tropomyosin molecules bind end-to-end, thus forming a continuous strand along the actin filament. According to current models, the tropomyosin strand may exist in two or more locations on the surface of the actin (8 -10). In one position, corresponding to the biochemical on state, the negatively charged tropomyosin is located over actin subdomains 3 and 4 in a "trough" of positive charge (11,12). The putative strong and weak myosin binding sites on actin are exposed and the myosin cross-bridges may cycle unimpeded. Recent work using actin mutants supports this model: charge change mutations within the trough of actin such as E311A/R312A in yeast (13) and K238A/E241A/E360H in Dictyostelium (14) result in weaker tropomyosin binding and destabilization of the on state.
In striated muscles from arthropods (Limulus) and vertebrates (Rana) the tropomyosin strand has been shown to be over the junction between actin domain 3 and 1 and over domain 2 in the off state and consequently strong myosin binding sites are blocked (Refs. 8 and 10, Fig. 1). The inhibitory component of troponin, troponin I, is a positively charged protein that binds to actin and switches the tropomyosin to the off state. It is noteworthy that although troponin I controls the actin-tropomyosin state, it does not bind directly to tropomyosin, so it probably acts indirectly through actin (15). Electrostatic charges are likely to play an important part in this interaction.
The amino acids in actin involved in stabilizing the off state have not been positively identified, but there is evidence for troponin I contacts on domains 1 and 2 of actin (16). This is compatible with reports that changes in the shape of domain 2 of actin are associated with regulation (17).
A cluster of negative charge due to Asp 56 , Glu 57 , and Glu 93 is located at the bottom of domain 2, close to the position occupied by tropomyosin when in the off state ( Fig. 1) (10). We considered the possibility that this negative charge cluster may repulse tropomyosin, thus maintaining the on state and that neutralization of the repulsion by inhibitory proteins may be an essential feature of the inhibitory mechanism. To test this hypothesis, we studied the state of actin-tropomyosin in a mutant of the Drosophila flight muscle-specific actin gene Act88F with a charge reversal mutation of glutamic acid 93 to lysine (E93K actin) (18,19). The E93K mutation produces a flightless phenotype characterized by disordered sarcomeres. However in vitro motility assays have demonstrated that E93K actin filaments can move over myosin and produce force, which is not very different from the wild type (20,21). Using in vitro motility techniques similar to those we used to characterize tropomyosin based regulation in striated and smooth muscles (22,28,32,33), we can demonstrate that charge reversal of amino acid 93 on actin does indeed change the equilibrium of actin-tropomyosin toward the off state.

EXPERIMENTAL PROCEDURES
Actin Preparation-Drosophila WT 1 indirect flight muscle-specific actin was made from a rosy 506 homozygous strain (rosy 506 is an eye color mutation) and E93K mutant actin from a rosy506 Act88F KM88 P-[Act88F E93K ] strain, which has a single copy of the E93K mutant Act88F gene inserted into an Act88F null genome by P-element-mediated transformation (18). Flies were grown and then stored at Ϫ20°C prior to processing.
A myofibrillar preparation was made from 50 g (50,000) of flies according to the method of Saide et al. (23). After a high salt extraction (20 mM KPO 4 , 500 mM KCl, pH 7.0) on ice for 10 min, the myofibril pellet was dehydrated by resuspending first in 6 volumes of ice-cold 50% acetone, then in 6 volumes of ice-cold 100% acetone, and finally allowed to air dry overnight at room temperature.
Actin extraction from the acetone powder was performed essentially after the methods of Bullard (24) and Pardee and Spudich (25) in ACEX (2 mM Tris-Cl, 0.2 mM CaCl 2 , 0.2 mM ATP, 1 mM dithiothreitol, pH 8.0) at 4°C for 60 min. After spinning the extract at 43,000 ϫ g for 15 min at 4°C, a tenth of the extract volume of 10 ϫ polymerization buffer (50 mM Tris-Cl, 500 mM KCl, 20 mM MgCl 2 , 10 mM ATP, pH 8.0) was added and polymerization allowed to proceed for 2 h at 4°C. Solid KCl was added to a final concentration of 800 mM. After incubation for 10 min at 37°C to dissociate the contaminating Drosophila thin filament proteins, the preparation was spun at 90,000 rpm for 20 min at 4°C in a Beckman TLA100.3 rotor. The F-actin pellet was resuspended in 3 ml of ACEX, homogenized, and then dialyzed for 24 h in 3 ϫ 1-liter changes of ACEX. The optical density at A 280 to A 310 , with an extinction coefficient of 0.62 cm 2 /mg, was used to estimate the molarity of the G-actin stock.
Anion exchange chromatography employing a 1-ml Mono Q column (Amersham Pharmacia Inc.) was used to isolate the indirect flight muscle-specific actin isoform from the mixture of the six Drosophila isoforms constituting the G-actin stock. Before application to the column, the actin stock was clarified by first spinning at 90,000 rpm for 20 min in a Beckman TLA100.3 rotor and then filtering through 0.2-mm filters (Millipore). A 28-min segmented gradient from 0 to 500 mM NaCl in 20 mM MOPS, pH 6.5, was then used to resolve the actin mixture into four different isoform peaks (Fig. 2).
The peak B fractions from chromatographic runs of WT (or E93K) actin were pooled and polymerized as before for 2 h at room temperature and then spun at 100,000 rpm for 10 min at 4°C in a Beckman TLA100.3 rotor. Six separate paired preparations of WT and E93K actin were prepared; yield was 30 -100 g per preparation. E93K does not polymerize as readily as WT actin, and in one preparation it was necessary to add heavy meromyosin at 1:500 mol/mol in order to promote polymerization. The F-actin pellets were resuspended in ACEX to a final concentration of about 5 M actin.
In Vitro Motility Assay-The in vitro motility assay was performed as described by Fraser and Marston (22) using 100 g/ml skeletal muscle HMM on cover glasses coated with silicone by soaking in 0.2% dichloromethylsilane in chloroform. F-actin was labeled with rhodaminephalloidin () as described by Kron et al. (26). F-actin⅐-tropomyosin and actin⅐-tropomyosin-troponin complexes were formed at 10 ϫ assay concentration: 100 nM actin⅐ , 150 nM tropomyosin, and 0 -600 nM troponin were mixed in 50 mM KCl, 25 mM imidazole-HCl, 4 mM MgCl 2 , 1 mM EDTA, 5 mM dithiothreitol, pH 7.4 (buffer A), and incubated for 30 -60 min. The complexes were diluted 10-fold immediately prior to actin, ACT88-E93K actin; , rhodamine-phalloidin; HMM, heavy meromyosin; NEM-S-1, myosin subfragment-1 modified by N-ethylmaleimide. A and B in the wild type profile are positively identified as as ACT88F actin, since they are absent in the KM88 ACT88F null mutant chromatogram. Peak A is arthrin, and peak B is ACT88F-modϩ (posttranslationally altered products of the Act88F gene; the nature of modϩ is unknown (45)). In the E93K profile peaks A and B have moved as a result of the amino acid substitution. The positions of peaks C and D are not altered and represent products of other actin genes. Peak B represents pure WT or E93K flight muscle-specific actin. C, wild type actin 88F on a two-dimensional gel. It shows one major spot and a minor one of the same mass. There is no arthrin.
infusion into the motility cell. We have demonstrated that tropomyosin does not dissociate during the experiment under these conditions and that additional tropomyosin in buffers C and D is not necessary (22).
A flow cell was prepared from a freshly siliconized coverslip and a microscope slide as described by Kron et al. (26). Assay components and buffers were infused into the flow cell at 30 -60-s intervals. Two 50-l aliquots of HMM at 100 g/ml were infused in buffer A (50 mM KCl, 25 mM imidazole-HCl, 4 mM MgCl 2 , 1 mM EDTA, 5 mM dithiothreitol, pH 7.4) to provide a coating of immobilized HMM on the coverslip. This was followed by 2 ϫ 50 l of buffer B (buffer A ϩ 0.5 mg/ml bovine serum albumin) then 2 ϫ 50 l of 10 nM actin⅐ Ϯ associated tropomyosintroponin in buffer A. 50 l of buffer C (buffer B ϩ 0.1 mg/ml glucose oxidase, 0.02 mg/ml catalase, 3 mg/ml glucose, 0.5% methylcellulose, Ϯ troponin at assay concentration) and 50 l of buffer D (buffer C ϩ 1 mM ATP) were then infused. Ca 2ϩ concentration was varied by incorporating Ca 2ϩ ⅐EGTA buffers in the final assay buffers C and D.
The movement of actin⅐-tropomyosin filaments over the immobilized skeletal muscle HMM was observed under a Zeiss epifluorescence microscope (63ϫ/1.4 objective) with a DAGE-SIT-68 camera and recorded on video tape. Videos were digitized, and the movement was analyzed to determine fraction of filaments moving and velocity of motile filaments using either the automatic tracking program described by Marston et al. (27) or the manual tracking procedure in the cases where filaments were not easily detected (22,28).

Comparison of Rabbit Skeletal and Wild Type Drosophila
Flight Muscle Actin Filament Motility-Actin filament movement over immobilized heavy meromyosin was studied in an in vitro motility assay. With this assay it is possible to measure a number of parameters in place of the single parameter of ATPase. These are fraction of actin filaments moving, velocity of the filaments that are moving, and the number of filaments attached to myosin per unit area. It has been shown that Ca 2ϩ can regulate all three of these parameters independently through troponin-tropomyosin (22,28,29).
In previous work it has been shown that addition of arthropod (Limulus) tropomyosin to actin increases filament velocity (4), but this is not observed with vertebrate striated muscle tropomyosin. We therefore decided to test the Drosophila actin with vertebrate smooth muscle tropomyosin (from sheep aorta) as well as with vertebrate striated muscle tropomyosin (from rabbit). Smooth muscle tropomyosin tends to activate actomyosin ATPase, while skeletal muscle tropomyosin is usually inhibitory (30,31). In the in vitro motility assay smooth muscle tropomyosin gives an increase in filament velocity similar to Limulus tropomyosin (4, 32, 33). The increased velocity has been suggested as indicating that actin-tropomyosin is predominantly in the on state, since smooth muscle tropomyosin has a higher K t and size of cooperative unit compared with skeletal muscle tropomyosin (28,30,31) The regulation of Drosophila flight muscle actin filament movement by troponin and tropomyosin has not been studied before, and it is possible that Drosophila actin could have regulatory characteristics quite different from vertebrate actin. Table I compares Drosophila actin and the well studied rabbit striated muscle actin filaments (22,28,33). The two types of actin filaments moved at the same velocities with a high proportion of the filaments motile. Smooth muscle tropomyosin increased the velocity of both types of actin filaments by around 25%, while skeletal muscle tropomyosin had no effect on velocity, although cosedimentation experiments showed it did bind under the conditions of the motility assay (22). The fraction of filaments motile was not affected by either tropomyosin.
In previous experiments we have demonstrated in our system that addition of up to 16 nM troponin at Ca9 results in a marked decrease in the fraction of filaments moving accompanied by a modest decrease in the velocity of the filaments that remain motile (22,28). Addition of skeletal muscle troponin to wild type 88F actin-skeletal muscle tropomyosin filaments produced the same decrease in the fraction of filaments motile (half-maximal effect at about 4 nM troponin) and the same small decrease in velocity of motile filaments (16% at 16 nM troponin) as was observed previously with skeletal muscle actin filaments (compare Fig. 3 with Fig. 3 in Ref. 22). This behavior was observed irrespective of the type of tropomyosin used and is characteristic of actin-tropomyosin filaments in the off state (28). Different technical approaches to measuring troponin control of filament motility have produced differing patterns of results in some experimental circumstances (these are discussed in Ref. 28); however, all the published analyses agree that reduction in the proportion of filaments that are motile is a characteristic feature of actin-tropomyosin filaments switched off by troponin or other inhibitory proteins (22,32,46,47). At pCa5 troponin did not alter the fraction of filaments motile or the velocity.
We conclude that the motility parameters of actin from rabbit and Drosophila are indistinguishable, and they are regulated by tropomyosin and troponin in the same way. This justifies the use of mammalian regulatory proteins in our investigation of mutant Drosophila actin.
Comparison of Wild Type and E93K Mutant Actin-Although the E93K mutation has a flightless phenotype E93K actin, filaments have been reported to be capable of movement in the in vitro motility assay and of exerting force (20,21). We found that wild type and E93K actin moved at almost the same velocity at 30°C. At 28°C wild type moved 26% faster than E93K actin, and it has been reported that at 23°C wild type moved 50% faster than E93K actin (20). Thus our results are in broad agreement with previous experiments using E93K actin and confirm that the mutation does not seriously compromise the actin-myosin interaction. Some differences are to be expected, since we have used a different surface (siliconized glass), and we performed experiments at higher ionic strength than previously and used 0.5% methyl cellulose (20,21). We have found that silica surfaces seem to provide less resistance to movement than nitrocellulose and also that the addition of methylcellulose to the medium enables weakly bound filaments to remain attached to the myosin surface (28). Thus the conditions we have adopted would minimize differences in motility Percentage change due to 10 nM skeletal muscle tropomyosin Percentage change due to 20 nM smooth muscle tropomyosin performance between wild type and mutant actin. When smooth or skeletal muscle tropomyosin were added to wild type actin filaments, the fraction of filaments that were motile did not change. In contrast, when either smooth or striated muscle tropomyosin was added to the mutant E93K actin filaments, the fraction of filaments motile was greatly reduced (Fig. 4). Experiments were done with six paired preparations of wild type and E93K actin. The fraction of wild type actin filaments motile was 79 Ϯ 3.6% (S.E., n ϭ 7) and of E93K actin 71 Ϯ 2% (S.E., n ϭ 7). Tropomyosin progressively reduced the fraction of E93K actin filaments motile reaching 14 Ϯ 2% (n ϭ 5) with 16 nM skeletal muscle tropomyosin and 17 Ϯ 2% (n ϭ 3) with 16 nM smooth muscle tropomyosin; in contrast, tropomyosin did not affect WT actin filament motility (Fig. 4).
Skeletal and smooth muscle tropomyosin had different ef-fects upon filament velocity. The skeletal muscle tropomyosin had no effect on velocity of the wild type actin as observed previously with rabbit actin (Table I) and slightly decreased velocity of E93K actin filaments (Fig. 4). The decrease was greater at 25°C (up to 37%, Fig. 4) than at 27°C (up to 19%).
Smooth muscle tropomyosin enhanced the velocity of wild type actin filaments by 25% at 30°C (Fig. 4, Table I), but this was not observed with E93K actin filaments, instead velocity declined in the same way as it did with skeletal muscle tropomyosin (Fig. 4).
It will be noted that the the large decrease in the fraction of E93K actin filaments motile and the small decrease in velocity when tropomyosin was added (Fig. 4) resembles the effect of adding troponin at pCa9 to wild type actin-tropomyosin filaments (Fig. 3). As has been discussed, this effect is characteristic of actin-tropomyosin filaments in the off state; thus it appears that the E93K mutation in actin causes tropomyosin to bind to actin only in the off state (32).
To confirm that the "switch off" of E93K actin filament movement was due to the tropomyosin-mediated mechanism, we added NEM-S-1 to E93K actin-tropomyosin. NEM-S-1 forms strong bonds to actin, even in the presence of ATP, thus it switches actin-tropomyosin to the on state independently of regulatory proteins (29,34). We have shown previously that adding NEM-S-1 to actin-tropomyosin resulted in an increase in filament motility (28). When NEM-S-1 was mixed with E93K actin-smooth muscle tropomyosin filaments, we observed that it restored both the fraction of filaments motile and their velocity to the level observed with E93K actin when tropomyosin was absent (Fig. 5). This reversal of inhibition required less than 0.4 NEM-S-1 per actin, thus confirming that it was acting cooperatively upon tropomyosin in the same way as it does with wild type striated muscle actin-tropomyosin (28,34). E93K actin-tropomyosin filaments were also reactivated by adding troponin in the presence of Ca 2ϩ (Fig. 5). This indicates that the effect of the E93K mutation is to shift the on/off equilbrium of the thin filament toward the off state to a certain degree, but the Ca 2ϩ -activated troponin complex is capable of shifting the equilibrium back to the on state. In contrast a deletion mutation in tropomyosin has been shown to produce thin filaments that cannot be activated by Ca 2ϩ and troponin (36), although they can be switched on by rigor S-1.
Molecular Interpretation of the Effect of the E93K Mutation-Recent structural work has indicated that tropomyosin is loosly associated with actin and that its location may be dependent upon the charge distribution on the surface of actin. The neg-  (14)) lead to a reduction in tropomyosin binding affinity and modification of regulatory behavior. In addition mutations in tropomyosin can also modulate regulation (33,36) The likely position of tropomyosin when in the off state has only recently been visualized by electron microscopy and threedimensional reconstructions. In both vertebrates and arthropods tropomyosin is seen to have moved across the face of the actin monomer so that it is over the junction between domains 1 and 3 and over domain 2 at the top of the actin monomer, covering the putative myosin binding site (8, 10) (Fig. 1). Although this description provides an explanation of how the steric blocking model of regulation could work, it provides no clues as to why tropomyosin is located in this position in the presence of inhibitory proteins.
Several lines of evidence point to a possible involvement of domain 2 of actin in transmitting a signal from regulatory protein to tropomyosin. Actin monomers within actin filaments are flexible and conformational changes within the monomer are mainly in the region of domain 2 (38,39). Furthermore detailed modeling of x-ray diffraction patterns from active and relaxed muscles have been interpreted in terms of a movement of domain 2 (17). Interestingly in this model the distance between domain 2 and tropomyosin is halved when the muscle is switched off, but there are only minor changes in the distance between tropomyosin and the other three domains.
Since it is proposed that tropomyosin is stabilized in the on position by actin surface charges in domains 3 and 4, it seems reasonable to propose that charges on the surface of domains 1 and 2 contribute to stabilizing tropomyosin in the off position. It has long been appreciated that in inhibitory proteins such as troponin I and caldesmon inhibitory function is related to concentrations of positive charge (the inhibitory peptide of troponin I, residues 104 -115, contains 4 arginines and 2 lysines (40,41)). There are complementary clusters of negative charged amino acids on the surface of actin, the most notable of which is at the bottom of domain 2 (Asp 56 , Glu 57 , Glu 93 ). This negatively charged patch is close to tropomyosin in the off state (10) as illustrated in Fig. 1. The critical nature of this region to muscle function is highlighted by the fact that the E93K mu-tation produces a flightless phenotype, and the E93A/R95A mutation in yeast is lethal (37). Mutations in this region affect sarcomere assembly (18), and this may be a consequence of the poor polymerizability of E93K actin or due to an altered interaction with tropomyosin. The motility parameters of E93K actin are not greatly different from WT, but the interaction of E93K with tropomyosin is altered as shown by these experiments (Fig. 4). The observation that a charge reversal mutation of glutamic acid 93 to lysine results in tropomyosin binding to actin in the off state, as defined by the motility parameters (low fraction of filaments motile with little change in velocity) suggests one aspect of actin structure that may be important for regulation. Alterations of other amino acids in this charged cluster (D56A/E57A) seem to have similar effects upon actintropomyosin state (48), but charge changes at the nearby surface helix (E99A/E100A) or in the middle of domain 1 (D24A/ D25A) have no effect upon regulation (49), this indicating a specific role for this site.
We propose that electrostatic charge on the surface of domain 2 of actin plays a critical role in determining the equilibrium between states of actin-tropomyosin that is a central feature of the steric blocking mechanism of actin filament regulation (8,42). Concentrated negative charge in this region, as found in wild type actin, promotes the on state, and the absence or reversal of such charge, as found with the E93K actin mutant and actin-troponin I complex, permits tropomyosin to move toward domain 2 and thus promotes the off state. It is possible that the ability of structurally unrelated proteins such as troponin and caldesmon to inhibit actin-tropomyosin interaction with myosin may simply be due to neutralizing negative charge on the surface of actin, thus permitting tropomyosin to move to the off position on actin. It would also be compatible with the general observation that inhibitory proteins bind to actin rather than to tropomyosin. FIG. 5. Restoration of the motility of E93K actin⅐-tropomyosin filaments by the addition of NEM-S-1 or troponin at pCa5. The motility of 10 nM E93K actin⅐ ϩ 20 nM smooth muscle tropomyosin was measured at 25°C. Left, NEM-S-1 was prepared (44) and added after the E93K actin⅐-smooth muscle tropomyosin in buffers C and D. Right, troponin was mixed with E93K actin⅐-smooth muscle tropomyosin in buffer A and applied to the motility cell. Troponin and pCa5 Ca/ EGTA buffer was present in buffers C and D. Filament movement in buffer D was analyzed as described in the legend to