Alternative exon-encoded regions of Drosophila myosin heavy chain modulate ATPase rates and actin sliding velocity.

To investigate the molecular functions of the regions encoded by alternative exons from the single Drosophila myosin heavy chain gene, we made the first kinetic measurements of two muscle myosin isoforms that differ in all alternative regions. Myosin was purified from the indirect flight muscles of wild-type and transgenic flies expressing a major embryonic isoform. The in vitro actin sliding velocity on the flight muscle isoform (6.4 microm x s(-1) at 22 degrees C) is among the fastest reported for a type II myosin and was 9-fold faster than with the embryonic isoform. With smooth muscle tropomyosin bound to actin, the actin sliding velocity on the embryonic isoform increased 6-fold, whereas that on the flight muscle myosin slightly decreased. No difference in the step sizes of Drosophila and rabbit skeletal myosins were found using optical tweezers, suggesting that the slower in vitro velocity with the embryonic isoform is due to altered kinetics. Basal ATPase rates for flight muscle myosin are higher than those of embryonic and rabbit myosin. These differences explain why the embryonic myosin cannot functionally substitute in vivo for the native flight muscle isoform, and demonstrate that one or more of the five myosin heavy chain alternative exons must influence Drosophila myosin kinetics.

To investigate the molecular functions of the regions encoded by alternative exons from the single Drosophila myosin heavy chain gene, we made the first kinetic measurements of two muscle myosin isoforms that differ in all alternative regions. Myosin was purified from the indirect flight muscles of wild-type and transgenic flies expressing a major embryonic isoform. The in vitro actin sliding velocity on the flight muscle isoform (6.4 m⅐s ؊1 at 22°C) is among the fastest reported for a type II myosin and was 9-fold faster than with the embryonic isoform. With smooth muscle tropomyosin bound to actin, the actin sliding velocity on the embryonic isoform increased 6-fold, whereas that on the flight muscle myosin slightly decreased. No difference in the step sizes of Drosophila and rabbit skeletal myosins were found using optical tweezers, suggesting that the slower in vitro velocity with the embryonic isoform is due to altered kinetics. Basal ATPase rates for flight muscle myosin are higher than those of embryonic and rabbit myosin. These differences explain why the embryonic myosin cannot functionally substitute in vivo for the native flight muscle isoform, and demonstrate that one or more of the five myosin heavy chain alternative exons must influence Drosophila myosin kinetics.
The functional properties of muscle myosins have been investigated in many systems. Developmental-and tissue-specific isoforms have distinct ATPase rates and velocities that are the primary determinants of muscle contractile properties (1). However, it is not well understood how functional properties of myosin are determined by variations in myosin heavy chain (MHC) 1 sequence and structure (2,3).
Genetic engineering, sequence comparisons, in vitro motility assays, and crystallographic structures have facilitated recent investigations into MHC structural regions that might influence function properties (3). For example, the two flexible loops (25/50-kDa, loop 1; 50/20-kDa, loop 2) in the chicken skeletal and Dictyostelium myosin heads, unresolved in the atomic structures, have been examined as possible determinants of functional variation between isoforms (4 -6). Substituting different loop 1s with varying charge and length affected ADP release rate, ATPase rate, and in vitro actin sliding velocity (4). However, these changed properties did not correlate well with the relative speed of the native myosin from which the loop was derived, suggesting that additional regions need to be altered in concert to give the unique kinetic and velocity characteristics of a specific isoform.
Investigations using species-specific myosins differing only in one or a few regions result in better insights. In phasic smooth muscle myosin, seven additional amino acids near the nucleotide binding site increase actin-activated ATPase rate and in vitro motility velocities compared with the tonic smooth muscle isoform (7,8). Similarly, scallop catch and phasic muscle isoforms have different motility and ATPase rates (9).
The Drosophila system has the potential to provide even greater insights. Fifteen different MHC isoforms, arising from alternative splicing of mRNA transcripts from the single Mhc gene (10,11), have been found so far in a wide variety of muscle types, ranging from the supercontractile larval muscles to the highly ordered, asynchronous indirect flight muscles (IFMs) (12)(13)(14)(15). Four of the five alternatively spliced exon sets occur in the myosin head coding region and have been proposed to determine functional differences between isoforms ( Fig. 1) (16). The variety of myosins normally expressed and the ease of creating transgenic flies make the Drosophila Mhc gene a powerful system for studying the relationships between myosin sequence and in vivo functional variation (17). However, to take full advantage of this system, it is necessary to develop in vitro assays to determine the different properties of Drosophila myosins. As the IFMs constitute a major part of the adult fly (approximately 20% of body weight), myosin isolation from these muscles following dissection is relatively straightforward compared with attempting recovery from other muscle types.
Recently, a major embryonic isoform (Emb) was transgenically expressed in a myosin null background (18) such that every muscle type expressed only the Emb isoform. Although the transgenic flies were viable, they exhibited a flightless phenotype and were severely impaired in jumping and walking abilities. In addition, the IFMs showed severe myofibrillar degeneration soon after the time at which flight normally commences. Clearly, this embryonic isoform cannot substitute functionally for the IFM isoform. The embryonic isoform differs from that of the IFM at all alternatively spliced regions; therefore, one or more of the alternative exons must be responsible for the functional differences between these isoforms.
We report on the purification and functional characterization of Drosophila IFM and modified embryonic (Emb 18 ) myosins. We present the first optical trap measurements of Drosophila myosin step size, attachment lifetimes, in vitro motility velocities, and isoform-specific ATPase measurements. The results demonstrate that alternatively spliced exon regions of MHC dramatically influence cross-bridge cycle kinetics and set the stage for the evaluation of chimeric MHC constructs to pinpoint functional variations to particular polypeptide sequences encoded by the alternative exons.

EXPERIMENTAL PROCEDURES
Myosin Isolation-Myosin was prepared from the dorsolongitudinal IFMs (DLMs) of 100 -120 wild-type (for muscle genes; the flies carry the yellow body color and white eye mutations) or transgenic Emb 18 flies. Emb 18 myosin (hereafter referred to as Emb 18 ) was expressed transgenically following P element-mediated germline transformation of a modified version of a major embryonic isoform cDNA (17). Emb 18 was modified from a native embryonic myosin to have the shorter adult tailpiece encoded by exon 18. The adult tailpiece enhances myosin accumulation in the adult IFMs (17). By expressing Emb 18 in a Mhc 10 background (null for myosin in flight and jump muscle), the Emb 18 isoform could be isolated from the IFM without contamination by other myosin isoforms. However, it had the IFM-specific essential myosin light chain isoform. Thus, the only differences between the isolated myosin isoforms were located in head regions encoded by exons 3, 7, 9, and 11 ( Fig. 1), and the rod hinge region, encoded by exon 15. DLMs were dissected at 4°C from split thoraces in York Modified Glycerol (YMG: 20 mM potassium phosphate, pH 7.0, 2 mM MgCl 2 , 1 mM EGTA, 8 mM DTT, 2% (v/v) Triton X-100, and 50% (v/v) glycerol) (19) without Triton X-100, but with the addition of a protease inhibitor mixture (Complete, Roche Molecular Biochemicals). A protease inhibitor mixture was also included in all subsequent myosin extraction solutions. IFMs were suspended in YMG (with Triton X-100), incubated for 30 min, centrifuged (8,500 ϫ g), and washed in YMG without Triton X-100 and glycerol. Myosin was extracted into three volumes (55 l) of 1.0 M KCl, 0.15 M potassium phosphate, pH 6.8, 10 mM sodium pyrophosphate, 5 mM MgCl 2 , 0.5 mM EGTA, and 8 mM DTT for 10 min and centrifuged at 8,500 ϫ g for 5 min. The proteins in this initial extraction are shown in Fig. 2A (lane 2). The pellet was sometimes used to make acetone powder for actin isolation (see below). The extracted myosin was precipitated by decreasing KCl to 40 mM and incubating for 16 h (overnight). Following a 15-min centrifugation in a Beckman TL-100.3 rotor at 100,000 ϫ g, the pellet ( Fig Drosophila myosin preparations were quantified by their absorbance at 280 nm using an absorbance coefficient of 0.53 cm Ϫ1 for 1 mg⅐ml Ϫ1 (21). The Bradford assay (Coomassie Plus-200, Pierce) and semiquantitative SDS-PAGE gave similar results when rabbit myosin was used as the protein standard.
Actin Isolation-Actin was isolated from dissected DLMs as described by Razzaq et al. (22). Purified F-actin was centrifuged at 150,000 ϫ g for 1.5 h, and resuspended in actin buffer (25 mM imidazole, pH 7.4, 25 mM KCl, 4 mM MgCl 2, 1 mM EGTA, and 1 mM DTT) (23). Fig.  2C shows the purity of the actin preparation. Actin was quantified by subtracting absorbance at 310 nm from that at 290 nm and dividing by an absorbance coefficient of 0.62 cm Ϫ1 for 1 mg⅐ml Ϫ1 (24). For the motility assay, actin was labeled with rhodamine-phalloidin as described by Kron et al. (23).
ATPase Assays-The Ca-ATPase activity of myosin was measured using a modification of the method described by Pullman (25). Activity was determined in a 0.5-ml reaction mixture containing 10 mM imidazole, pH 6.0, 0.1 M KCl, 10 mM CaCl 2 , and 1 mM [␥-32 P]ATP (500 -800 cpm⅐nmol Ϫ1 ) with 2-5 g of protein. The reaction was initiated by ATP addition. After incubating for 15 min at 22-23°C, the reaction was stopped by addition of 0.1 ml of 1.8 M HClO 4 . Following centrifugation, aliquots of the supernatant were added to 0.5 ml of 5% ammonium molybdate, 2 ml of 1.25 M HClO 4 , and 2.5 ml of isobutanol-benzene (1:1). After vigorous mixing and phase separation, 1 ml of the organic phase containing 32 P i was assayed by Cerenkov counting. We found that the Ca-ATPase activity at pH 6 was approximately twice that obtained at pH 7.4 (data not shown), similar to previous reports for whole thorax myosin (26). Thus, we measured ATPase activity at pH 6.0 to maximize ATPase rates.
For determination of the K m of myosin Ca-ATPase activity, ATP was varied from 5 to 300 M for IFM myosin and from 2 to 40 M for Emb 18 myosin. Specific radioactivity of [ 32 P]ATP was increased to 5,000 -8,000 cpm⅐nmol Ϫ1 . The reaction time was reduced to 3 min to ensure that initial rates were determined.
Mg-ATPase activity was determined in 0.5 ml of solution containing 10 mM imidazole, pH 6.0, 20 mM KCl, 2 mM MgCl 2 , 0.1 mM CaCl 2 , and 1 mM ATP with 2-5 g of protein for 30 min at 22-23°C, and processed as described above for Ca-ATPase.
Sliding Filament Assay-F-actin in vitro motility assays were conducted similarly to Kron et al. (23), except for the following modifications. 1) The nitrocellulose-coated coverslip of the flow cell was blocked by incubating with assay buffer plus BSA (AB/BSA) prior to myosin addition at 0.5 mg⅐ml Ϫ1 . 2) After myosin addition, dead heads (myosin that binds irreversibly to actin) were blocked by adding 5 M phalloidinstabilized IFM actin followed by AB plus ATP, 0.4% methylcellulose, and oxygen scavengers (27). After a 10-min incubation, ATP was washed out of the flow cell by several additions of AB/BSA. Labeled actin was added and the remainder of the protocol was as described by Kron et al. (23). This method was very effective, as continuous movement of actin filaments would last ϳ10 min in unblocked cells before the majority of filaments were arrested by dead heads, while in cells employing dead head blocking, most filaments were still moving after 30 min. Filament movement in unblocked cells was used as an indicator of myosin quality. If continuous movement by the majority of actin filaments did not last for ϳ10 min, the myosin preparation was not used for ATPase measurements.
3) The final activating motility solution contained 0.4% methylcellulose and no KCl.
In some experiments, smooth muscle tropomyosin (smTM) (chicken gizzard, Sigma) was bound to Drosophila IFM actin by mixing at a 1:1 molar ratio. Centrifugation and one-dimensional SDS-PAGE electrophoresis of the pellet and supernatant confirmed the binding of smTM to Drosophila F-actin (data not shown). smTM (100 nM) was included in all solutions added to the flow cell after actin addition to prevent smTM dissociation from actin (28).
The assays were conducted at 22-23°C, and filament movement was recorded onto videotape. Video sequences were captured at 10 frames/s with an Apple PowerMac computer. Actin filament velocity was calculated using the Autotrack Macro (29) from the public domain NIH Image program. A filament's velocity was determined only if it moved smoothly for Ͼ2 s. At least 50 filaments from a flow cell with the best quality of movement were averaged to determine the mean velocity for each myosin preparation. Means for all preparations were then averaged to determine mean myosin isoform velocity. Standard deviations were calculated from the preparation averages. A Student's t test was used for all statistical analysis, with p values Ͻ 0.05 considered statistically significant.
Single Molecule Mechanical Experiments-The single actomyosin cross-bridge mechanical experiments were carried out using an optical tweezers transducer in the "three-bead" configuration (30 -32). Two independently trapped latex beads, attached to the ends of an actin filament, held the filament close to a glass bead coated with a low surface density of Drosophila myosin molecules (0.1 g of protein applied to a 22 ϫ 22-mm 2 coverslip) allowing interactions between single myosin heads and the actin filament. Beam steering of the traps and calibrations were performed as described previously (31,32). Actomyosin interactions were measured at trap stiffnesses between 0.02 and 0.04 pN⅐nm Ϫ1 at 22-23°C. The optical trap assay buffer was 25 mM KCl, 4 mM MgCl 2 , 1 mM EGTA, 25 mM imidazolium chloride, pH 7.4, 20 mM DTT and glucose/glucose oxidase/catalase (0.2 mg⅐ml Ϫ1 glucose oxidase, 0.05 mg⅐ml Ϫ1 catalase, and 3 mg⅐ml Ϫ1 glucose). It included an ATP regeneration system (2 mM creatine phosphate, 0.1 mg⅐ml Ϫ1 creatine phosphokinase) and 5 g⅐ml Ϫ1 rhodamine-phalloidin-labeled actin (32). The ATP concentration was 3 M for both step size and attached lifetime measurements. The latter estimates the second order rate constant of A⅐M ϩ ATP 3 A ϩ M⅐ATP and was determined by fitting a least squares regression to a plot of frequency against the logarithm of binned event durations (31).
The size of the working strokes of the two Drosophila myosins inter-acting with rabbit or Drosophila IFM actin were determined by analyzing hundreds of displacement events (for details, see Refs. 31 and 32). Histograms of the displacement events were fitted with a Gaussian distribution whose midpoint was shifted from zero, reflecting the size of the working stroke, and whose width was determined by the trap stiffness (31).

RESULTS
Protein Isolation-The Drosophila myosin isolation method yielded an enzymatically active myosin capable of moving actin filaments in a smooth and continuous manner. The IFM myosin yield was 300 -400 g (2.5-3.3 g/fly) and was at least 90% pure, as judged from the proportion of heavy chain (200 kDa) and light chains on SDS-PAGE gels (Fig. 2B). Two light chains associate with Drosophila MHC: the essential light chain and the regulatory light chain, which appears as two bands due to multiple phosphorylated versions (ϳ14 isoelectric variants; Refs. 33 and 34). A negligible amount of actin was present. The Emb 18 myosin yield was 200 g (Ϸ2 g/fly), and its purity was similar to the IFM isoform (Fig. 2B). The purified Drosophila myosins were kept in storage buffer at 5-10 mg⅐ml Ϫ1 , and were used the same day they were isolated, as ATPase activity declines to 50 -75% of the original activity within 2 days. The decline in activity is similar to that reported for other myosin types, including rat cardiac myosin (35).   The yield of IFM actin was ϳ0.5 g/fly, similar to that reported by Razzaq et al. (22).
ATPase Activities of IFM and Emb 18 Myosins-Drosophila IFM myosin exhibited a high Ca-ATPase activity, of 7.5-9 s Ϫ1 (Table I). These values are the same as for phosphorylated myosin from whole Drosophila thoraces (36). The Ca-ATPase assay conditions, selected to maximize ATPase rate, were similar to those used by Takahashi et al. (36), except that we used 10 mM calcium rather than 3 mM. The Emb 18 Ca-ATPase activity is similar to that reported for whole larval myosin (26), which was 25% that of their adult myosin mixture (whole thorax). The K m for ATP of the Emb 18 myosin Ca-ATPase activity was approximately one-fifth that of the IFM myosin (Fig. 3).
Mg-ATPase activity of the IFM isoform was only 1-2% of the Ca-ATPase activity. The Ca-ATPase and Mg-ATPase activities of transgenic Emb 18 myosin were 20 -37% of those obtained with IFM myosin (Table I, Fig. 3). Drosophila myosin K-ATPase (EDTA) activity at 0.5 M KCl was slightly higher than the Mg-ATPase rates (data not shown), and about 50% of the K-ATPase activity reported by Takahashi and Maruyama (26) for whole thorax myosin.
Sliding Filament Assay-The in vitro motility assay required a few modifications from the standard rabbit heavy meromyosin (HMM) assay (23) to obtain smooth and continuous movement of actin on Drosophila myosin. Actin filaments diffused off the myosin-coated surface after ATP addition unless 0.4% methylcellulose was added to the AB/BSA/glucose/glucose oxidase/catalase and ATP solutions. Actin interactions with myosin were enhanced when the overall ionic strength was lowered to about 25 mM (no added KCl). Higher ionic strengths decreased the number of filaments moving, and some filaments would leave the myosin surface even in the presence of methylcellulose. The best actin movement (greatest number of filaments moving smoothly and continuously) occurred at ϳ0.5 mg⅐ml Ϫ1 of myosin, which is at the high end of values reported for other myosin types (0.1-0.5 mg⅐ml Ϫ1 ; Ref. 28). Movement was not achieved below 0.3 mg/ml. Mean actin velocity on IFM myosin, 6.4 m⅐s Ϫ1 , was faster than that typically found with rabbit myosin (4 -5 m⅐s Ϫ1 ; Refs. 23 and 28). Mean actin velocity on IFM myosin was 9-fold faster than with Emb 18 myosin (Table II). Although it was much slower, the embryonic myosin still required methylcellulose to prevent actin diffusion away from the myosin-coated surface, and moved actin best at a myosin concentration of 0.5 mg⅐ml Ϫ1 .
Surprisingly, the binding of smTM to actin increased filament velocity over Emb 18 myosin almost 6-fold (Table II, Fig.  4A). In contrast, smTM had a slight inhibitory effect on IFM myosin driven movement, reducing the velocity (Fig. 4B) and decreasing the number of filaments that moved smoothly and uniformly. Reduced movement quality was not observed with Emb 18 myosin. Thus, with smTM bound to the Drosophila IFM actin, the difference in sliding velocity between IFM and Emb 18 myosins was only 1.4-fold.
Myosin Displacement Size-The displacements produced by the two Drosophila myosins in single molecule interactions with an F-actin in the optical trap are shown in Fig. 5. A short section of signal recording the position of a bead (one of a bead pair attached to F-actin) is shown (Fig. 5A), which is similar to those presented previously for other myosins (32,37). There are periods of reduced noise due to the increased stiffness when myosin, on the third bead, binds to the actin filament. Signal noise is due to thermal fluctuation of the bead held in the optical trap. The amplitude of the thermal fluctuations was estimated from a running variance (lower trace), which is used to objectively and automatically identify individual myosin at-tachment events. A histogram (Fig. 5B, gray) shows the distribution of bead positions during periods when no myosin is attached. This noise can be fitted with a Gaussian distribution whose S.D. is defined by the combined stiffness of the optical tweezers ( ϭ ͌(kT/ trap )). A distribution of individual mean displacements produced by Drosophila IFM myosin attachments is shown (Fig. 5B, black). These data are well fit by a Gaussian distribution of the thermal noise of the unattached bead-actin-bead assembly, but stiffened in the x axis by the size of the myosin working stroke. These are representative data from all the interactions of an individual F-actin bead pair. The shift in position of this distribution from the zero position on the x axis estimates the average size of the myosin displacements (37). For the Drosophila IFM myosin this produces a displacement of 3.91 Ϯ 2.36 nm (mean of 3547 attachment events obtained from 43 F-actin filaments on a total of 13 IFM myosin preparations; Ϯ S.D. of 43 means). For Emb 18 myosin we obtained a mean displacement of 4.38 Ϯ 2.27 nm (1011 events involving 8 F-actin filaments and 4 myosin preparations; Ϯ S.D. of 8 means). These estimates are not significantly different (p ϭ 0.605, Student's t test), but are rather smaller than previously measured with rabbit HMM (5.5 nm) (37) and S-1 (32). Fig. 5C shows the distribution of displacement means (binned into 2-nm bins) for each actin filament interacting with either Emb 18 (gray) or IFM (black). We found that data collected from different pedestals (third bead) with the same Factin filament had a large variation in mean displacement. Because of this large variability between different estimates, we performed additional control experiments. 1) Only Drosophila myosin preparations that demonstrated good quality movement in the in vitro motility assay were used for step size measurements. 2) Step size of rabbit full-length myosin was also measured under the same conditions (Fig. 5C). This gave a step size of 4.20 Ϯ 0.93 nm (mean Ϯ S.D. of 8 filament means, each with Ͼ25 events), which is not significantly different from the values for the Drosophila isoforms (p ϭ 0.832, one-way analysis of variance). 3) Neither using rabbit actin instead of IFM actin nor adding tropomyosin affected the step size of either Drosophila myosin.
The duration of myosin attachments to F-actin in the optical tweezers traces provides a measure of the second order rate constant of A.M ϩ ATP 3 A ϩ M⅐ATP. Since the rate of detachment is so fast once ATP binds, the rate constant esti- mates the ATP affinity of nucleotide free actomyosin. At 3 M ATP, we obtained estimates for the two Drosophila myosins (Table II), which are essentially the same and similar to that obtained (31) for the heavy meromyosin fragment of rabbit myosin (0.8 ϫ 10 6 M Ϫ1 ⅐s Ϫ1 ). DISCUSSION We have demonstrated that myosin can be isolated and purified from Drosophila IFMs in sufficient quantities for optical trapping, in vitro motility, and ATPase assays. More significantly, we can isolate other myosin isoforms by expressing them transgenically in the IFMs of mutant flies, Mhc 10 , which do not express the native IFM myosin. Myosin isoform differences do not appear to influence assembly properties (17,18). Even though embryonic isoforms cause myofibrillar breakdown and a reduction in myosin content as the flies age (17,18), yields of Emb 18 from young flies (1-2 days old) were no less than 50% compared with wild type.
Actin Velocity in Vitro and in Vivo-Both Drosophila myosins move actin in vitro only at high surface myosin concentrations, in the presence of methylcellulose and at low ionic strength. All these factors suggest that these myosins have much lower actin affinities than rabbit skeletal muscle myosin. Methylcellulose enhances solution viscosity, reducing diffusion of actin filaments away from the surface and increasing the probability of further interactions with myosin heads. It thus maintains movement at very low myosin surface densities (27) or when actomyosin affinity is reduced, either by various actin or myosin mutants (6,38) or by higher ionic strengths (39,40). Indirect flight muscle myosin from Lethocerus (waterbug) has a K m for actin, measured from actin-activated ATPase studies, which is 6 -15 times greater than vertebrate fast muscle (41). Thus, it seems likely that IFM myosins may generally have low actin affinities. Since Emb 18 has similar in vitro motility requirements, the low actin affinity likely extends also to the embryonic isoform, normally expressed in the supercontractile muscles of the larva.
Our results show that Drosophila IFM myosin produces one of the fastest in vitro actin sliding velocities reported for a skeletal muscle isoform. It is faster than rabbit skeletal myosin (4 -5 m⅐s Ϫ1 ) under similar conditions and temperature (27,42). The high in vitro velocities produced by the IFM myosin are consistent with its role in powering one of the fastest contracting Drosophila skeletal muscle types. Drosophila IFMs produce oscillatory contractions of ϳ200 Hz at 22°C (43) and shorten by about 3.4% (44). The thick and thin filaments of each set of IFMs, the opposing dorso-longitudinal and dorsoventral muscles, slide past each other for up to 2.5 ms during a contraction. Starting with a sarcomere length at rest (45) of 3.4 m, the length change for each half sarcomere will be 58 nm and contraction velocity will be around ϳ23 m⅐s Ϫ1 , almost 4 times the speed measured in the motility assay. In comparison, rabbit psoas muscle shortens about 7 m⅐s Ϫ1 per half sarcomere at 22°C (46,47) or about 3-fold slower than Drosophila IFMs.
It is clear from the Drosophila IFM myosin data, as previously observed with rabbit skeletal myosin, that in vitro velocity tends to underestimate the values of V max achieved in vivo (47). This is partly due to the random orientation of crossbridge heads (48) and because the filaments are not fully unloaded (e.g. binding of other myosin heads) in vitro. Further, in vitro velocities are usually higher with a full complement of thin filament accessory proteins (28).
The mean velocity produced by the Emb 18 myosin is 9-fold slower than the IFM isoform and correlates with the slower contraction velocities of the larval body wall muscles. Emb myosin is normally expressed in the supercontractile dorsal oblique muscles that have long sarcomeres and a high degree of shortening (49). The mechanical properties of these muscles are unknown, but they are used to power crawling at 1 mm⅐s Ϫ1 , 2 with individual contractions visible to the naked eye, and must have much slower contraction velocities than IFMs. The actin velocity on Emb 18 myosin that we have measured more closely resembles those described for phasic smooth muscle isoforms than those of vertebrate slow skeletal muscle isoforms (4,7,8).
Although the functional measures of Emb 18 correlate with the mechanical properties of its native muscles, it should be noted that due to transgenic expression the Emb 18 isoform has the IFM-specific essential light chain. The Drosophila genome contains single copies of both the regulatory and essential muscle myosin light chain genes. The IFMs express a unique essential light chain isoform that differs in 14 amino acids of the C terminus due to alternative splicing of its transcript (50). As light chain isoforms influence velocity (51), the properties of the endogenous embryonic isoform may differ slightly from those measured for Emb 18 4. In vitro actin sliding velocities. A, actin velocity with IFM myosin was slightly decreased when smTM was bound to actin. B, actin velocity was much slower with Emb 18 myosin, but was dramatically increased with smTM bound to actin.
transgenically expressed Emb 18 contain the same essential light chain, any functional differences we observed must derive from the myosin heavy chain sequence.
The binding of smooth muscle tropomyosin dramatically increases actin velocity on Emb 18 myosin but slightly decreased the sliding rate on IFM myosin. The molecular basis for this potentiation, which has also been found for actin activation of myosin ATPase (52,53) remains generally unexplained. The effect varies according to myosin type and isoform. Smooth muscle tropomyosin increases actin sliding velocity ϳ10-fold with phosphorylated Limulus striated muscle myosin (54) and 2-4-fold on smooth muscle myosin (55), but only slightly increases velocities on rabbit skeletal HMM (56). On this basis the Emb 18 isoform acts like a smooth muscle myosin, while the IFM myosin behaves more like rabbit skeletal myosin.
IFM Myosin ATPase Rates-The fast nature of the IFM myosin isoform was supported by ATPase measurements. The Ca-ATPase rate of the IFM isoform was ϳ2-4-fold higher than rabbit skeletal muscle and Lethocerus flight muscle myosins measured under similar conditions (36,57). The significance of the K m difference for ATP of the Ca-ATPase activation for the two isoforms is not immediately apparent, since ATP concentrations in vivo are vastly in excess of 1 M. However, it will be interesting to determine which alternative exon influences this property, even if it does not impact the fly physiologically.
White et al. (41) measured Mg-ATPase values of 0.01-0.05 s Ϫ1 for Lethocerus myosin subfragment 1, S-1, at 20°C, pH 7; this is about the same as our Drosophila IFM myosin value (0.1 s Ϫ1 ). Drosophila IFM ATPase activity correlates with IFMs being classified as fast muscles. However, Lethocerus flight muscles, which are physiologically similar, only contract at 30 Hz, yet have nearly the same myosin ATPase activities. Clearly, myosin ATPase activity does not scale with contraction velocity within this muscle type. This observation is very similar to the results of Molloy et al. (58), who showed that IFM fiber ATPase activity is constant across a wide range of insect wing beat frequencies, with the exception of the wasp.
Single Molecule Mechanics-There were no significant differences in the overall mean displacements measured for rabbit, IFM, and Emb 18 full-length myosins. However, there was large scatter in the individual estimates of mean step size obtained for each myosin isoform. Separate estimates of the mean step size were obtained from each third bead tested (and therefore each individual myosin molecule). The individual binding events used to obtain each estimate had an amplitude distribution that was Gaussian with S.D. similar to that of the free bead movement (as described by Molloy et al. (31)). The new feature in the data that we report here is that the mean estimates (some obtained using the same actin filament but testing a different myosin molecule of the same isoform) have a greater standard error of the mean than expected on the basis of sampling statistics. In other words, trials on the same myosin type gave results that were significantly variable between myosin molecules tested. This discovery implies that the individual myosin molecules being tested behave differently in this assay. There are three possible interpretations. 1) The myosin molecules are biochemically heterogeneous as a result of differential post-translational modifications; 2) the Drosophila myosins occupy a wider range of functionally active molecular orientations in the way they are deposited on the third bead substrate, some of which restrict the production of the full displacement; or 3) the myosin step size depends upon the geometry of myosin binding to actin and the degrees of freedom of movement within the myosin molecule. Any single myosin molecule would give a movement proportional to the cosine of the angle that it makes to the actin filament (as suggested by Molloy et al. (Ref. 31) and as shown by Tanaka et al. (Ref. 48)). Since in these assays the myosin head orientation is random with respect to the F-actin axis, different flexibilities within the myosins might produce step size distributions with different mean variance. On the basis of our data set, it is not possible to FIG. 5. Step size determination for IFM and Emb 18 myosin. A, a raw data recording from an optical trapping experiment is shown. A single actin filament was held suspended between two 1-m diameter plastic microspheres close to a third, surface-fixed bead on which was deposited a low surface density of IFM myosin (see text for details). The position of one of the optically trapped beads was monitored with a four-quadrant photodetector. The upper record shows how the position of the optically trapped bead varies with time. Intervals of high noise are interspersed with intervals of reduced noise. Noise amplitude reflects the system stiffness and periods of low noise correspond to myosin binding events. The lower trace is the calculated S.D. of the data record. Events are defined as periods during which the S.D. remained below the threshold level (dotted line). The amplitude of separate events is measured from the displacement from local mean position. Panel B shows (gray) the distribution of background thermal vibration of the bead (in the absence of myosin attachments). The shape of this histogram is consistent with the principle of equipartition of energy in which the Gaussian distribution of x positions is given by A ϭ A 0 exp(Ϫx 2 /kT) (where is the combined optical trap stiffness of 0.038 pN⅐nm Ϫ1 , see text). The black histogram shows the distribution in amplitude of myosin binding events measured from a much longer data recording than shown in panel A (335 events). The mean position in this data set is shifted by about 5 nm from zero. Panel C shows how the estimates of mean position are distributed between experimental runs. Individual estimates are highly variable between runs, and even within experiments performed using the same actin filament. This implies that behavior of the individual myosin molecules is heterogeneous. The overall means determined from many experimental runs measured for IFM (black), Emb 18 (gray), and rabbit myosins (white) are not significantly different from each other (see text for details). discriminate among these models. However, it does appear that, in these experiments, the insect myosin step sizes give a broader distribution than rabbit myosin.
The Functional Role of MHC Alternative Exons-At least one of the five MHC regions encoded by alternative exons must be responsible for the observed differences in actomyosin kinetics between the IFM and Emb 18 isoforms. The remainder of the molecule is identical between these two isoforms as it arises from constitutive exons of the same gene, and because we engineered the adult tailpiece onto Emb 18 . Based on mapping the four motor domain alternative exons (exons 3, 7, 9, and 11) onto the three-dimensional chicken skeletal myosin head (16), we currently believe that exon 7 may be responsible for the measured differences in ATPase rates as it encodes part of the lip of the nucleotide site and is part of "switch one." Exons 11 and/or 3 could contribute to the differences in actin sliding velocity, as they are located near the pivot point of the lever arm ( Fig. 1) and have been proposed previously to affect step size (17). However, no difference in step size was found in this study; therefore, neither exons 3a and 3b nor exons 11e and 11c regulate step size. We cannot exclude the possibility that exon 11a, 11b, or 11d expressed in one or more of the other myosin isoforms may affect step size. Exon 9 encodes the "rigid relay loop" that may be involved in propagating signals from the nucleotide site to the actin binding site and lever arm region (59). The final alternative exon, 15, encodes the rod hinge region. Although it has been shown that myosin S-1 head fragment is sufficient for in vitro motility (31,60), exon 15 could be involved in modulating function in vivo (61).
None of the four alternative exon regions in the myosin head corresponds to the loop 1 and loop 2 regions that have been investigated as possible sources of functional variation between myosins from different muscle types (4 -9). The seven-amino acid insert, which determines differences in ATPase rate and in vitro motility velocity between the tonic and phasic smooth muscle isoforms, is not within a homologous region encoded by a Drosophila alternative exon (7,8). Thus, multiple mechanisms appear to have evolved to modify myosin function.
In summary, we have observed dramatic differences in the in vitro motilities, the effects of smooth muscle tropomyosin on motility, Ca-and Mg-ATPases, and K m for ATP between the IFM and Emb 18 myosin isoforms. These differences are certainly sufficient to explain why the Emb 18 isoform cannot functionally substitute for the native isoforms in the flight muscle. Most obviously, the Emb isoform is not fast enough to power flight. Although the details of how these functional differences arise (e.g. different ADP release rates) remain to be investigated, we have shown that functional myosins can be expressed and purified from the IFMs in sufficient quantities for biochemical experiments. This will enable us to develop more sophisticated assays to determine which steps in the cross-bridge cycle are different between these two isoforms.
By exchanging alternative exons between the IFM and Emb 18 isoforms, we can relate the functional differences described here to specific structural regions of the myosin heavy chain. These alternative regions can then be examined in more detail using site-directed mutagenesis to determine specific amino acids or groups of residues that are responsible for particular functional modifications. The advantage of the Drosophila Mhc system for this analysis of myosin is that we can, in transgenic organisms, observe the effects from the level of myosin molecular properties, through muscle mechanics (20,33) to whole animal locomotory performance (17) to gain a fully integrative understanding of the role of the alternative exon regions.