Deletion of the Myopathy Loop of Dictyostelium Myosin II and Its Impact on Motor Functions*

One of the putative actin-binding sites ofDictyostelium myosin II is the β-strand-turn-β-strand structure (Ile398-Leu-Ala-Gly-Arg-Asp403-Leu-Val405), the “myopathy loop,” which is located at the distal end of the upper 50-kDa subdomain and next to the conserved arginine (Arg397), whose mutation in human cardiac myosin results in familial hypertrophic cardiomyopathy. The myopathy loop contains the TEDS residue (Asp403), which is a target of the heavy-chain kinase in myosin I. Moreover, the loop contains a cluster of hydrophobic residues (Ile398, Leu399, Leu404, and Val405), whose side chains are fully exposed to the solvent. In our study, the myopathy loop was deleted from Dictyostelium myosin II to investigate its functional roles. The mutation abolished hydrophobic interactions of actin and myosin in the strong binding state during the ATPase cycle. Association of the mutant myosin and actin was maintained only through ionic interactions under these conditions. Without strong hydrophobic interactions, the mutant myosin still exhibited motor functions, although at low levels. It is likely that the observed defects resulted mainly from a loss of the cluster of hydrophobic residues, since replacement of Asp403 or Arg402 with alanine generated a mutant with less severe or no defects compared with those of the deletion mutant.

In the absence of ATP, actin and myosin form a stable "rigor" complex, which is held together mainly by strong hydrophobic interactions. A three-dimensional reconstruction of electron microscopic images of the rigor complex of actin and myosin subfragment 1 (S1) 1 revealed that S1 is in contact with actin at several sites (1,2). One of the putative actin-binding sites of S1 is the ␤-strand-turn-␤-strand structure (Ile 398 -Leu-Ala-Gly-Arg-Asp 403 -Leu-Val 405 , in the case of Dictyostelium myosin II) located at the distal end of the upper 50-kDa subdomain (see Fig. 1) (3,4). Involvement of this ␤-turn-␤ structure in the actin-myosin interaction has been implied by two findings. First, mutation of Arg 403 (equivalent to Arg 397 in Dictyostelium myosin II) to glutamine in human cardiac myosin is the cause of familial hypertrophic cardiomyopathy (5) (thus, this ␤-turn-␤ structure is designated as the myopathy loop). In fact, R403Q mutation in cardiac myosin (6) and R397Q mutation in Dictyostelium myosin II (7) generated mutants with lower affinity to actin, suggesting the involvement of this conserved arginine residue in the actin-myosin interaction. Second, a threonine or serine residue of myosin I, whose location corresponds to a residue at the tip of the myopathy loop (Asp 403 in Dictyostelium myosin II), is phosphorylated by a myosin heavychain kinase (8). This phosphorylation is required for the full activation of the actin-activated ATPase activity of myosin I. In the case of myosin II, however, the residue is glutamate or aspartate as in Dictyostelium myosin II (TEDS rule (9), T or S in myosin I, and E or D in myosin II). In addition to this TEDS residue, the myopathy loop contains a cluster of hydrophobic residues (Ile 398 , Leu 399 , Leu 404 , and Val 405 ) whose bulky side chains are fully exposed to the solvent (see Fig. 1), as if ready to interact with hydrophobic residues on actin.
The second putative actin-binding site is located at the tip of the ␣-helix-loop-␣-helix structure at the distal end of the lower 50-kDa subdomain, and it contains conserved hydrophobic residues ( Fig. 1). Some of the hydrophobic side chains are fully exposed to the solvent and are likely to be in contact with hydrophobic residues located between subdomain 1 and subdomain 3 of actin (2). These two hydrophobic actin-binding sites may be cooperatively involved in the rigor binding of actin and myosin in the absence of ATP as well as in their rigor-like binding in the presence of ADP, i.e. in the strong binding during the ATP hydrolysis cycle.
The third putative site is the 50kDa-20kDa junction, a flexible loop easily recognized in almost all members of the myosin family (10). This loop, rich in basic residues and glycine residues, seems to be in contact with acidic residues of actin to form ionic bonds (1,2,11,12). These ionic bonds are a dominant factor in maintaining the actin-myosin association in the weakbinding state during ATP hydrolysis.
Dictyostelium discoideum cells have only one copy of the heavy-chain gene of myosin II (13). Disruption of the gene generates Dictyostelium myosin-null cells with specific phenotypic defects (14,15). The introduction of a mutant heavy-chain gene into these myosin-null cells generates transformants expressing the mutant myosin in place of the wild type. Using this Dictyostelium expression system developed by Spudich and co-workers (16 -21), we investigated how the myopathy loop of Dictyostelium myosin II is involved in the actin-myosin association and the motor functions of myosin.

EXPERIMENTAL PROCEDURES
Construction and Expression of Mutant Myosin and S1-The myopathy loop (Ile 398 -Leu-Ala-Gly-Arg-Asp 403 -Leu-Val 405 ) was replaced with a dipeptide (AG) by site-directed mutagenesis of the heavy-chain gene of Dictyostelium myosin II (provided by Dr. James A. Spudich, Stanford University). As a result of this mutation, six residues (Ile 398 , Leu 399 , Arg 402 , Asp 403 , Leu 404 , and Val 405 ) were deleted from the heavy chain. Arg 402 or Asp 403 was also changed to alanine by site-directed mutagen-* This work was supported by grants-in-aid for Scientific Research from the Ministry of Education, Science, Sports, and Culture of Japan and a research grant from the Human Frontier Science Program (to K. S.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
¶ To whom correspondence should be addressed. Tel. and Fax: 81-3-5454-6751; E-mail: sutoh@bio.c.u-tokyo.ac.jp. 1 The abbreviations used are: S1, subfragment 1; MOPS, 4-morpholinepropanesulfonic acid. esis. These mutant heavy-chain genes were ligated to the Dictyostelium actin-15 promoter and actin-6 terminator to drive their expression in Dictyostelium cells. They were finally inserted into a multicopy extrachromosomal vector, pBIG. Plasmids carrying these mutant heavychain genes were electroporated into Dictyostelium myosin-null cells in which the myosin-II heavy-chain gene had been knocked out by means of homologous recombination (15). Dictyostelium cells transformed by electroporation were selected in a medium supplemented with 20 g/ml G418 on plastic dishes for a week.
The truncated myosin heavy-chain genes corresponding to wild-type and mutant S1 s were constructed as follows. The heavy chain was truncated at Glu 836 by introducing a stop codon at the corresponding location of the heavy-chain gene. After fusing the actin-15 promoter and actin-6 terminator, the S1 gene was inserted into a multicopy vector, PTIKLOE (provided by Dr. TaroUyeda, National Institute for Advanced Interdisciplinary Research, Japan), which carries the essential and regulatory light-chain genes (22,23). The light-chain genes (provided by Dr. Rex L. Chisholm, Northwestern University) were fused to the actin-15 promoter to achieve high level expression. A histidine tag (His 6 ) was attached at the N terminus of the regulatory light chain for easier purification of S1 by inserting corresponding synthetic oligonucleotides between the start codon and the coding sequence of the light chain. The resulting transformation vector was introduced into Dictyostelium AX2 cells. The transformants were selected as above.
Phenotypes of the Transformed Cells-Growth rates were measured by determining the number of cells cultured in suspension. The incubator was shaken at 150 rpm and 22°C. The development of trans-formed cells was examined on an agar plate covered with a lawn of Escherichia coli cells. Dictyostelium cells (1.2 ϫ 10 4 ) were suspended in 10 mM Tris-Cl, pH 7.5, and then spotted onto the bacterial lawn. When the bacterial lawn had been cleared of bacterial cells by the Dictyostelium cells, the Dictyostelium cells entered the developmental stage.
Protein Purification-Phosphorylated myosin was prepared as described previously (24,25). Wild-type or mutant S1 was prepared as described for myosin with modifications. S1 was extracted from transformed Dictyostelium cells and precipitated as actoS1 after dialysis against a solvent comprising 50 mM NaCl, 10 mM MOPS, pH 7.4, and 2 mM MgCl 2 . S1 was then extracted from the precipitate with a solvent composed of 10 mM MOPS, pH 7.4, 0.25 M NaCl, 7 mM MgCl 2 , and 5 mM ATP. The extract was directly applied to the nickel nitrilotriacetic acid column (Qiagen). After the column was washed with one column volume of the above solvent supplemented with 1 mM ATP, S1 was eluted with a linear gradient of imidazole, pH 7.4, from 10 mM to 0.5 M. The eluted protein was dialyzed against a solvent composed of 50 mM NaCl, 10 mM MOPS, pH 7.4, and 2 mM MgCl 2 . It was centrifuged at 100,000 ϫ g for 30 min (Beckman TL100) before use. Relative concentrations of the proteins were determined by Protein Assay Reagent (Pierce).
ATPase and in Vitro Motility Assays-ATPase activities were measured as described (7,24). In vitro motility assays were performed in the presence of 0.2% or 0.7% methylcellulose as described (12,26,27). Force exerted on an actin filament was measured as described (28).
Fast Kinetics-Time courses of mant-ATP binding or mant-ADP release to and from wild-type or mutant S1 were followed as described (29) by a stopped-flow apparatus with a fluorescence detector (Applied Photophysics SX18MV).
Actin-S1 Binding in the Absence of Nucleotide or in the Presence of ADP-Varying amounts of wild-type or mutant S1 were mixed with a fixed amount of pyrene-labeled F-actin (0.25 M) (30) in 50 mM NaCl, 10 mM MOPS, pH 7.4, and 2 mM MgCl 2 or in this solvent supplemented with 1 mM ADP. Fluorescence measurements were performed at 25°C with a Perkin-Elmer LS50B luminescence spectrophotometer with excitation at 365 nm and emission at 410 nm.
For pelleting experiments, varying amounts of F-actin were mixed with a fixed amount of wild-type or mutant S1 (final concentration, 1.5 M) in either 50 mM NaCl, 10 mM MOPS, pH 7.4, and 2 mM MgCl 2 or in 0.5 M NaCl, 10 mM MOPS, pH 7.4, and 2 mM MgCl 2 . The mixture was centrifuged at 15 min at 4°C (Beckman TL100). The supernatant and precipitate were separated and analyzed by SDS gel electrophoresis to determine the relative amount of S1 coprecipitated with F-actin.

Construction of Mutant Myosins and Their Subfragments-A
short stretch of the heavy chain of Dictyostelium myosin II spanning residues 398 -405 (ILAGRDLV) was deleted, and the gap was connected by a dipeptide (AG). This mutant was designated as ⌬ mutant. Since the residues at the gap are in contact with each other in the three-dimensional structure of the motor domain of Dictyostelium myosin II (4, 31), they can be connected with the dipeptide without distorting the conformation around the deleted site, as shown below. The sequence ILAGRDLV was replaced with AG, and therefore, the mutation actually removed 6 amino acid residues (Ile 398 , Leu 399 , Arg 402 , Asp 403 , Leu 404 , and Val 405 ). Two point mutations, R402A and D403A, were also introduced into Dictyostelium myosin II to discriminate the effects of the removal of charged side chains from those of hydrophobic side chains. These mutants were expressed as full-length myosins or their soluble, single-headed subfragments (subfragment 1 or S1) in Dictyostelium cells.
Disruption of the Rigor Binding of Actin and Myosin-Effects of mutations on the actin-myosin interaction in the absence of ATP, i.e. under the rigor conditions, were examined by using S1 and pyrene-labeled F-actin (30). As shown in Fig. 2A, the pyrene fluorescence was quantitatively quenched upon the addition of wild-type S1, an indication that the rigor complex was formed. When ⌬ S1 was mixed with pyrene-labeled actin, however, no quenching was observed at the concentrations examined ( Fig. 2A). To examine the possibility that ⌬ S1 bound to actin without quenching the pyrene fluorescence, the binding of actin and S1 was directly determined by pelleting experiments (Fig. 3). In the absence of ATP, wild-type S1 was coprecipitated with F-actin quantitatively by ultracentrifugation. Unlike the wild type, ⌬ S1 remained in the supernatant until a large excess of actin was added, indicating that it actually lost tight binding with actin, as expected from the fluorescence measurements. From the pelleting experiments, the dissociation constant of ⌬ S1 and actin was calculated to be ϳ5 M, whereas that of wild-type S1 and actin was approximately 10 nM (32).
This weak interaction of ⌬ S1 and actin in the absence of ATP was completely abolished when 0.5 M NaCl was included in the solvent (Fig. 3), indicating that this binding was dominated by ionic interactions even in the absence of ATP. Contrary to the case of ⌬ S1, the tight binding of wild-type S1 and actin was not affected in a high ionic strength solvent, as previously shown. It is very likely that the loss of hydrophobic interactions between ⌬ S1 and actin resulted from a loss of hydrophobic side chains in the myopathy loop, not from a loss of charged side chains of Asp 403 and Arg 402 , since R402A or D403A S1 bound to actin as tightly as the wild type ( Fig. 2A). These results suggest that the hydrophobic residues in the myopathy loop are involved in the interface to actin in the rigor state.
To examine if the myopathy loop is also involved in the interface to actin in the presence of ADP, the pyrene-actin was titrated with wild-type and mutant S1s (Fig. 2B). When wildtype S1 was added to pyrene-actin in the presence of 1 mM MgADP, the rigor-like tight association of actin and S1 was observed, although the affinity was lower than that in the absence of nucleotide. The addition of R402A or D403A S1 also resulted in association to actin with slightly lower affinity than that of the wild type, indicating some contribution of these charged residues to the binding with actin in the presence of ADP. The addition of ⌬ S1, however, resulted in only slight quenching of the pyrene fluorescence at the concentrations examined here, showing that ⌬ S1 lost the rigor-like association with actin in the presence of ADP. Thus, deletion of the myopathy loop, especially deletion of the hydrophobic residues in the loop, abolished both the rigor binding in the absence of nucleotide and the rigor-like binding in the presence of ADP. The results indicate that the myopathy loop contributes to the interface with actin both in the absence of nucleotide and in the presence of ADP.
Effect of Mutations on the Enzymatic Properties of Myosin in the Absence of Actin-We examined the effects of mutations on the enzymatic properties of myosin to know if structural changes induced by the mutations were confined to the mutation site. First, a steady-state rate of hydrolysis of MgATP by wild-type or mutant S1 was determined by observing intrinsic tryptophan fluorescence (33). As shown in Table I, the basal MgATPase activities of ⌬, R402A, and D403A S1s were very similar to those of the wild type, indicating that these mutations did not disturb the ATPase site. To further support this notion, we measured the rate of binding of mant-ATP to wildtype or mutant S1 by means of the enhancement of fluorescence of the mant moiety upon binding (29). The rate of release of mant-ADP from wild-type or mutant S1 was also determined by the decrease in fluorescence of the mant moiety upon addition of excess ATP (29). All of these measurements, summarized in Table I, showed that no significant difference was detected between wild-type and mutant S1s. Thus, it is likely that conformational changes induced by the ⌬, R402A, and D403A mutations were confined to the mutation site.
Effect of the Deletion of the Myopathy Loop on the in Vitro Functions of Myosin-Although the ⌬ mutation did not affect the enzymatic properties of myosin in the absence of actin, as mentioned above, a remarkable difference was observed between wild-type and mutant full-length myosins when their FIG. 2. Binding of S1 to pyrene-labeled actin. A, varying amounts of the wild type (q), D403A (OE), R402A (), and ⌬ S1 (f) were mixed with a fixed amount of pyrene-labeled F-actin (final concentration, 0.25 M) in 50 mM NaCl, 10 mM MOPS, pH 7.4, and 2 mM MgCl 2 at 25°C. B, similar measurements were performed in the above solvent supplemented with 1 mM ADP. Note that higher concentrations of S1 than in A were used here. Fluorescence measurements were performed with excitation at 365 nm and emission at 410 nm.

FIG. 3.
Binding of S1 to actin as measured by pelleting experiments. Varying amounts of F-actin were mixed with a fixed amount of wild-type S1 (OE, ϫ) or mutant S1 (ࡗ,f) (1.  All of these full-length myosins showed a normal level of basal MgATPase activities, consistent with the S1 experiments as above. It seems that hydrophobic residues in the myopathy loop are required for the full actin-activation of the MgATPase activity of myosin. The negative charge of Asp 403 also contributed to the full activation, which is consistent with the observation that phosphorylation of the TEDS residue of myosin I by the heavy-chain kinase is required for the full activation of its actin-activated ATPase activity (8). The positive charge of Arg 402 did not play a role in the activation. The motor functions of the mutants were further investigated by means of in vitro motility assays. Surprisingly, ⌬ myosin could drive the sliding of actin filaments, although slowly. Under the standard motility assay conditions in the presence of 0.7% methylcellulose, continuous, one-directional sliding of actin filaments was observed. The velocity of sliding driven by ⌬ myosin was 0.2 m/s, whereas that by the wild type was 1.2 m/sec (Table II). In the presence of 0.2% methylcellulose, one-directional sliding of actin filaments was observed for a while. They then detached from the glass surface, showing that the actin-myosin interaction was rather weak. In the presence of a higher concentration of salt (75 mM KCl) and 0.7% methyl cellulose, some of the filaments started one-dimensional Brownian motion; they moved back and forth. This behavior was not observed for sliding driven by the wild type. Thus, the association of ⌬ myosin and actin was more dependent on the ionic strength of the solvent than that of wild-type myosin and actin. Under the standard motility conditions, ⌬ myosin could also exert a low level of force on actin filaments: 4.4 piconewtons/m of actin filament and 32 piconewtons/m of actin filament for mutant and wild-type myosin, respectively (Table II).
D403A myosin drove actin filaments at a velocity of 0.5 m/s. This result is consistent with the observation that the V max value of actin-activated ATPase activity of D403A myosin was lower than that of the wild type (Table II). As expected from its actin-activated ATPase activity, R402A myosin drove the sliding of actin filaments at a similar velocity to that of the wild type. It seems that a loss of hydrophobic residues in the myopathy loop was mainly responsible for the slow sliding of actin filaments driven by ⌬ myosin. Loss of a negative charge at Asp 403 partially contributed to the defects.
Effect of the Disruption of Hydrophobic Interactions on the in Vivo Functions of Myosin-The mutant myosins were expressed in Dictyostelium myosin-null cells to examine if they were functional in vivo. Transformants expressing ⌬ myosin exhibited phenotypes very similar to those of myosin-null cells; they neither grew in suspension nor formed fruiting bodies upon starvation. Unlike ⌬ myosin, R402A or D403A myosin completely complemented the phenotypic defects of myosinnull cells (data not shown). These results suggest that hydrophobic residues in the myopathy loop are essential for the in vivo functions of Dictyostelium myosin II, whereas the charged residues, including the TEDS residue, are not.

DISCUSSION
It has been proposed that the ␣-loop-␣ structure at the distal end of the lower 50-kDa subdomain interacts with the hydrophobic pocket formed by subdomain 1 and subdomain 3 of actin (2). It has also been proposed that the ␤-turn-␤ structure at the distal end of the upper 50-kDa subdomain, the myopathy loop, is in contact with the upper surface of the ATP binding cleft of actin (2). Thus, in the rigor binding, two distal ends of the upper and lower 50-kDa domains are likely to associate with actin through hydrophobic interactions and hold subdomain 1 of actin like two fingers holding a ball. Since deletion of one of the hydrophobic sites, i.e. the hydrophobic residues in the myopathy loop, abolished the rigor binding as shown here, cooperation between these two sites seems to be essential for the tight, hydrophobic association of actin and myosin. Consistent with this notion is the fact that replacement of one of the conserved hydrophobic residues in the ␣-loop-␣ structure (Phe 535 in Dictyostelium myosin II) with alanine dramatically reduced the affinity of actin and myosin. 2 It must be mentioned here that deletion of the loop abolished not only the rigor binding but also the rigor-like binding of actin and S1 in the presence of ADP, whereas replacement of Arg 402 or Asp 403 with alanine resulted in no or only slight destabilization of the complex under these conditions. The results show that the hydrophobic residues in this loop contribute to the hydrophobic interface to actin in the absence of nucleotide as well as in the presence of ADP, i.e. in the strong-binding state during ATPase cycle.
Deletion of the loop did not completely abolish the motor functions of myosin. The basal MgATPase activity of ⌬ myosin was activated to some extent upon the addition of actin. Moreover, the mutant could support the slow sliding of actin filaments and exert a low level of force on them. The continuous, 2 N. Sasaki and K. Sutoh, unpublished result. one-directional sliding of actin filaments driven by ⌬ myosin was disrupted upon increasing the concentration of KCl in the assay solvent from 25 to 75 mM. These results suggest that the hydrophobic actin-myosin binding in the strong-binding state is crucial for effective energy transduction for sliding and force generation but not essential for these motor functions. As far as weak ionic interactions between ⌬ myosin and actin being maintained in a low ionic strength solvent, ⌬ myosin could support the sliding of actin filaments and exert force on them, although at low levels. In view of the fact that the actin-⌬ myosin interaction was highly dependent on the ionic strength of the solvent, this weak ionic association of ⌬ myosin and actin must have been disrupted in the living cells, consistent with the observation that ⌬ myosin was not functional in Dictyostelium myosin-null cells even though it retained a low level of in vitro motor functions.
Removal of a negative charge from the TEDS residue by D403A mutation, which might mimic the dephosphorylation reaction, generated less severe defects than ⌬ mutation. Consistent with the result, the mutant myosin could completely complement the defects of myosin-null cells. Thus, the negative charge of the TEDS residue is not essential for in vivo and in vitro motor functions, although it is important for the full activation of ATPase activity of myosin and for efficient sliding of actin filaments. Removal of the positive charge in Arg 402 rarely affected motor functions. It seems that the loss of the hydrophobic side chains was the main cause of defects observed for ⌬ myosin and ⌬ S1. Loss of the negative charge at Asp 403 only partially contributed to the defects.
In crystal structures of the motor domain of Dictyostelium myosin II (4,31), the side chains of the hydrophobic residues in the myopathy loop are fully exposed to the solvent (Fig. 1). Chicken S1 has a similar cluster of exposed hydrophobic side chains at the corresponding location (3). Furthermore, almost all classes of myosins have a similar cluster of hydrophobic residues at a similar location along their sequences, although those amino acid residues are not conserved (10). This cluster of hydrophobic residues must be crucial for effective actinactivation of ATPase activity, sliding of actin filaments, and force generation.