Determination of human myosin III as a motor protein having a protein kinase activity.

The class III myosin is the most divergent member of the myosin superfamily, having a domain with homology to a protein kinase. However, the function of class III myosin at a molecular level is not known at all, and it has been questioned whether it is actually an actin-based motor molecule. Here, we showed that human myosin III has an ATPase activity that is significantly activated by actin (20-fold) with Kactin of 112 microm and Vmax of 0.34 s-1, indicating the mechanoenzymatic activity of myosin III. Furthermore, we found that human myosin III has actin translocating activity (0.11 +/- 0.05 microm/s) using an in vitro actin gliding assay, and it moves toward the plus end of actin filaments. Myosin III containing calmodulin as the light chain subunit showed a protein kinase activity and underwent autophosphorylation. The autophosphorylation was the intramolecular process, and the sites were at the C-terminal end of the motor domain. Autophosphorylation significantly activated the kinase activity, although it did not affect the ATPase activity. The present study is the first report that clearly demonstrates that the class III myosin is an actin-based motor protein having a protein kinase activity.

The class III myosin is the most divergent member of the myosin superfamily, having a domain with homology to a protein kinase. However, the function of class III myosin at a molecular level is not known at all, and it has been questioned whether it is actually an actinbased motor molecule. Here, we showed that human myosin III has an ATPase activity that is significantly activated by actin (20-fold) with K actin of 112 M and V max of 0.34 s ؊1 , indicating the mechanoenzymatic activity of myosin III. Furthermore, we found that human myosin III has actin translocating activity (0.11 ؎ 0.05 m/s) using an in vitro actin gliding assay, and it moves toward the plus end of actin filaments. Myosin III containing calmodulin as the light chain subunit showed a protein kinase activity and underwent autophosphorylation. The autophosphorylation was the intramolecular process, and the sites were at the C-terminal end of the motor domain. Autophosphorylation significantly activated the kinase activity, although it did not affect the ATPase activity. The present study is the first report that clearly demonstrates that the class III myosin is an actin-based motor protein having a protein kinase activity.
Myosin III is a member of the myosin superfamily, which consists of at least 18 classes (1)(2)(3)(4)(5). The class III myosin represents the most divergent member of the myosin superfamily. The motor domain of myosin III shows ϳ24 -27% sequence identity to myosin Is and ϳ22-26% sequence identity to myosin-IIs (6). Of particular interest is that myosin III has an amino terminus domain that resembles to protein kinases. This domain contains the characteristic motifs for protein kinases such as a glycine-rich loop, an invariable Lys residue required for ATP binding, and a catalytic loop. Given its high degree of divergence, if myosin III evolved at the same rate as other myosins, the myosin III lineage would predate the divergence of yeast. Class III myosin was originally found in Drosophila photoreceptor cells and subsequently found in vertebrates including human myosin III (7,8).
The function of Myosin III is best studied in Drosophila photoreceptor cells. Each photoreceptor cell has a specialized organelle consisting of a stack of microvilli known as a rhabdomere. The phototransduction machinery is localized in the rhabdomere (9). Drosophila photoreceptors undergo a pro-longed depolarization afterpotential that persists after cessation of the light stimulus. Prolonged depolarization afterpotential results from the stable conversion of rhodopsin to the light-activated form, metarhodopsin, in response to blue light. During a prolonged depolarization afterpotential, photoreceptor cells become refractory to subsequent prolonged depolarization afterpotential-inducing stimuli and are inactivated. Mutants that are defective for both inactivation and the prolonged depolarization afterpotential are known as neither inactivation nor afterpotential (nina) mutants (10,11), and the myosin III was identified (12) as one of eight nina complementation groups (ninaA-H) that affect a different gene involved in rhodopsin synthesis, ninaC, on the basis of an electroretinogram phenotype, which could be attributed to reduction in the visual pigment rhodopsin (13). The amino acid sequence of the ninaC gene product contains the critical conserved motifs for protein kinases at the amino terminus and myosin motor domain at the C-terminal side of the kinase domain, respectively. There are two spliced variants that were reported, p132 and p174 (12). The p132 protein was primarily localized to the cytoplasm adjacent to the rhabdomeres, and the p174 protein was restricted to the rhabdomeres (14,15). Other than having different subcellular localization, the two ninaC isoforms also exhibit differences in their function. Mutants expressing only the p174 isoform showed little or no effect and displayed the normal phenotype; however, elimination of p174 resulted in a ninaC phenotype as strong as the null allele (15), suggesting that the tail unique to p174 is responsible for localization of p174 to the rhabdomeres.
Only p174 is required for wild-type electrophysiology and to prevent retinal degeneration. However, it is not known whether the electroretinogram phenotype and retinal degeneration are obligatorily coupled. Analysis of mutants with mutations in the kinase homologous domain demonstrated that the ninaC kinase domain is required for normal electroretinogram, suggesting that the ninaC kinase domain may be required for phototransduction. Mutants that have the entire myosin homologous domain deleted displayed a phenotype that is indistinguishable from the null mutant, suggesting that the myosin homologous domain is required for normal phototransduction and to prevent retinal degeneration (16). Immunolocalization performed on the kinase domain-and myosin domain-deleted mutants shows that the kinase domain is not required for localization of p174 to the rhabdomeres, but myosin domaindeleted mutants no longer specifically localize to the rhabdomeres and were also detected in the cell body, suggesting that the myosin domain is required to concentrate p174 in the rhabdomeres (16).
The Drosophila ninaC proteins have also been demonstrated to bind calmodulin (17). The calmodulin localization seems to be dependent on the ninaC proteins, since the mutant flies lacking the p174 protein did not concentrate calmodulin in the rhabdomeres, and mutant flies lacking p132 did not show any detectable amount of calmodulin in the cytoplasm. A defect in vision was also detected when calmodulin was not concentrated in the rhabdomeres, suggesting a role for calmodulin in phototransduction. Recently, it was shown that the ninaC protein directly binds to the PDZ domain protein INAD that is thought to play a role in assembling the phototransduction signaling complex (18). Whereas INAD mutants failed to change ninaC localization, disruption of the ninaC/INAD interaction delayed termination of the photoreceptor response, suggesting the role of the complex in rapid deactivation of the photoresponse. These findings have suggested that myosin III plays a significant role in the phototransduction process.
Of interest are the recent findings of the involvement of the myosin III gene in human hereditary disorders. Dose and Burnside (19) reported that MYO3B maps a region that overlaps the locus for a Bardet-Biedl syndrome. Walsh et al. (20) also reported that MYO3A is responsible for nonsyndromic recessive progressive hearing loss, DFNB30. However, a key problem is that whereas the class III myosin contains the domain homologous to the motor domain of the other myosin superfamily, there is no hard evidence that the class III myosin is actually the motor protein. Because the sequence homology of myosin III at the "myosin" domain is low among the myosin superfamily members, it has been suggested that myosin III may not be a motor protein but rather a signaling molecule. The objective of this research is to clarify whether myosin III is an actin-based motor protein. In the present paper, we show for the first time that human myosin III has an actin-activated ATPase activity and an actin translocating activity, clearly indicating that myosin III is an actin-based motor molecule with protein kinase function.

EXPERIMENTAL PROCEDURES
Materials-Restriction enzymes and modifying enzymes were purchased from New England Biolabs (Beverly, MA). Actin was prepared from rabbit skeletal muscle according to Spudich and Watt (21). Skeletal muscle myosin S1 fragment was prepared as described previously (22).
Generation of the Human Myosin IIIA Construct-Total RNA was prepared from human retinal pigment epithelia cell line ARPE-19 (ATCC, Manassas, VA) using an RNeasy minikit (Qiagen, Hilden, Germany). Poly(A) ϩ RNA was isolated using an Oligotex mRNA minikit (Qiagen). Myosin IIIA cDNA was generated by reverse transcription (Superscript reverse transcriptase II; Invitrogen) with specific primers. The cDNAs (nucleotides 1-2200 and 2026 -3873) were amplified using sense primer 5Ј-GCTAGCTTGGCGTCTGAGCATTCTTATGC-3Ј and antisense primer 5Ј-GTCGACTGCTCAGCTCATCACCATTC-3Ј and using sense primer 5Ј-GCTAGCTGGTCACTAGAGGAGAAAC-3Ј and antisense primer 5Ј-GTCGACCTGGGTAGCTTGGCCTTTCTG-3Ј, respectively. The amplified cDNAs were digested with NheI/NcoI and NcoI/SalI, respectively. The digested cDNAs were ligated into pFast-Bac1 baculovirus transfer vector (Invitrogen), containing c-myc and an octahistidine tag. This construct (huM3AIQ2) contains the kinase domain, motor domain, neck domain containing two IQ motifs, and short tail domain with a c-myc tag and an octahistidine tag at the C-terminal end. The recombinant baculovirus expressing huM3AIQ2 protein was produced according to the manufacturer's protocol.
ATPase Assay-The actin-activated ATPase activity was measured in 30 mM imidazole (pH 7.5), 25 mM KCl, 2 mM MgCl 2 , 1 mM EGTA, 1 mM dithiothreitol with or without F-actin at 25°C. All assays were initiated by adding [␥-32 P]ATP to the reaction mixture. The liberated 32 P was measured as described previously (25). The actin-activated ATPase activity was also measured in the presence of an ATP-regenerating system (20 units/ml pyruvate kinase, 2.5 mM phosphoenol pyruvate). The liberated pyruvate was determined as described (26).
Actin Co-sedimentation Assay-The huM3AIQ2 was incubated in buffer containing 30 mM imidazole (pH 7.5), 25 mM KCl, 2 mM MgCl 2 , 1 mM EGTA, 1 mM dithiothreitol, and 20 M F-actin at 25°C for 10 min. The sample was ultracentrifuged at 100,000 ϫ g for 10 min, and then the supernatant and the pellets were analyzed by SDS-PAGE. The amount of the co-sedimented huM3AIQ2 heavy chain was determined by densitometry as described previously (27).
Gel Filtration Chromatography-huM3AIQ2 was dialyzed against a solution containing 20 mM Tris-HCl (pH 7.5), 0.15 M KCl, 5 mM MgCl 2 , 5 mM ATP, 5 mM 2-mercaptoethanol, and 1 mM EGTA and then applied to a Sephacryl S-300HR column (1.0 ϫ 46 cm). The protein was eluted with a dialyzed buffer at a flow rate of 0.16 ml/min. Fractions were analyzed by immunoblotting using anti-hexahistidine antibodies as described above. K av was calculated using the equation, where V e , V o , and V t are the elution volume, the void volume, and the total volume, respectively.
In Vitro Motility Assay-The actin gliding velocity was measured by in vitro actin gliding assay. A coverslip was first coated with nitrocellulose and huM3AIQ2 was then applied to the coverslip. The movement of actin filaments was observed in 25 mM imidazole (pH 7.4), 25 mM KCl, 5 mM MgCl 2 , 1 mM EGTA, 1 mM dithiothreitol, 18 g/ml catalase, 0.1 g/ml glucose oxidase, 3.0 mg/ml glucose, 4 mM ATP, and 0.5% methylcellulose at 25°C. Glucose oxidase and catalase were added in the assay solution to reduce the fluorescence quenching as described previously (28). Dual labeled F-actin was prepared by the method of Homma et al. (29). Actin filament velocity was calculated from the movement distance and the elapsed time in successive snapshots. Student's t test was used for statistical comparison of mean values. A value of p Ͻ 0.01 was considered to be significant.

Expression and Isolation of the Recombinant Human Myosin
III-The human myosin III construct was produced (see "Experimental Procedures") and expressed in sf9 insect cells. The construct (huM3AIQ2) contains the entire kinase domain, the myosin homology domain, and the two IQ domains with C-terminal histidine tag and c-myc tag to aid in purification (Fig.  1A). The histidine tagging at the C-terminal end of the molecule has been performed with conventional (30) as well as unconventional myosin (31,32), and no influence on motor function has been observed. The cells were co-infected with an appropriate ratio of human myosin III-expressing virus and calmodulin-expressing virus, since it was reported that calmodulin copurified with myosin III heavy chain (33), suggesting that calmodulin plays a role as the light chain subunits. Fig. 1B shows SDS-polyacrylamide gel electrophoresis of the purification steps of huM3AIQ2. The purified huM3AIQ2 construct was composed of a high molecular mass band and a low molecular mass band free from 200-kDa sf9 conventional myosin and actin. The high molecular mass band (150 kDa) was consistent with the calculated molecular mass of huM3AIQ2 and was recognized by anti-myc antibodies (data not shown) (Fig. 1B), indicating that the high molecular mass band is the expressed myosin III heavy chain. The small subunits showed a mobility shift with a change in Ca 2ϩ that is characteristic of calmodulin, suggesting that the small subunit is indeed calmodulin (Fig. 1B). To examine whether myosin III has oneheaded or two-headed structure, the purified myosin III was subjected to gel filtration chromatography analysis using Sephacryl S-300HR (Fig. 2). The molecular mass of the purified huM3AIQ2 was estimated to be 172 kDa based upon the standard curve with known molecular mass standard proteins. The value was consistent with the calculated molecular mass of the construct with two bound calmodulins of the single-headed form (183,926 Da).
Protein Kinase Activity of Human Myosin III and Its Autophosphorylation-The protein kinase activity of the purified human myosin III was examined. Among the protein substrates tested, human myosin III phosphorylated myosin regulatory light chain, calponin, actin, and myelin basic protein  ( Fig. 3A). The result clearly demonstrated that it has protein kinase activity. It should be mentioned that the phosphorylation of these substrate proteins was not observed without adding myosin III, indicating that the phosphorylation is catalyzed by myosin III but not contaminated kinases in the substrate protein samples. The best substrate among the substrates tested was myelin basic protein, and the observed rates in the assay conditions were 0.27 nmol/min/mg, 0.13 nmol/min/mg, 0.016 nmol/min/mg, and 9.3 ϫ 10 Ϫ4 nmol/min/mg, for myelin basic protein, calponin, myosin II regulatory light chain, and actin, respectively. Of interest is that human myosin III also showed the autophosphorylation activity (Fig. 3B, a) and the maximum extent of phosphorylation of 1.8 mol of P i /mol was obtained (Fig. 3B, b). Similar autophosphorylation activity was observed with the construct containing the third IQ motif (Met 1 -Lys 1408 ) (not shown). The result suggests that there are at least two autophosphorylation sites in the huM3AIQ2 construct. The phosphoamino acid analysis of the autophosphorylated myosin III revealed that both serine and threonine are phosphorylated, although threonine phosphorylation was predominant (Fig. 3B, c). To detect tyrosine phosphorylation, thin layer electrophoresis was also performed at higher pH, where phosphotyrosine is well separated from phosphoserine and phosphothreonine (34). There was no phosphotyrosine detected (not shown). To examine whether the autophosphorylation is the intermolecular process or intramolecular process, the rate of autophosphorylation was examined as a function of huM3AIQ2 concentration. The rate of autophosphorylation was constant regardless of myosin III concentration (Fig. 3B, d).
The result suggests that the autophosphorylation is an intramolecular process rather than intermolecular reaction. The result further indicates that the phosphorylation is not due to the potentially contaminated kinases, because if so, the rate of myosin III heavy chain phosphorylation increases with the increase in myosin III concentration in the reaction mixture. We studied whether autophosphorylation affects the protein kinase activity of myosin III. huM3AIQ2 was preincubated with cold ATP to complete autophosphorylation, and then [␥-32 P]ATP and the protein substrate were added simultaneously, and the incorporation of 32 P into the substrate was monitored by autoradiography at various times (Fig. 4A). The result clearly indicated that the autophosphorylation significantly activated the protein kinase activity of myosin III. The radioactivity of the phosphorylated substrate was quantita- tively analyzed by using a phosphor imager (Fujifilm FLA-5000). The result indicates that the autophosphorylation activated the kinase activity 3.3-fold (Fig. 4B).
To further determine the domain of myosin III autophospho-rylation, the autophosphorylated huM3AIQ2 was subjected to limited proteolysis with chymotrypsin (Fig. 5). The proteolysis first produced the phosphorylated peptide of 32 kDa, which was further digested to 24-kDa and then 20-kDa peptide. The autoradiography suggests that all of the phosphorylation sites are retained in the 20-kDa peptide. To determine the location of the 20-kDa domain in human myosin III, we determined the Nterminal sequence of the 20-kDa peptide. The peptide had a free N terminus, and the sequence was Ile-Asn-Leu-Ala-Lys. This sequence corresponds to Ile 901 -Lys 905 in human myosin III sequence, thus indicating that the 20-kDa peptide is the C-terminal end of the motor domain.
Actin-activated ATPase Activity-Whereas the class III myosin contains the myosin head homology domain, it is less conserved than other myosin superfamily members, and one of the most critical questions for myosin III is whether myosin III is an actin-based motor protein. To address this question, we examined the actin-activated ATPase activity of the purified human myosin III. We found that human myosin III exhibited the Mg 2ϩ -ATPase activity that was significantly activated by actin. Fig. 6 shows the actin-activated ATPase activity as a function of actin concentration. The actin concentration required for the saturation of the activation was significantly higher than that of myosin V (35, 36) but comparable with that of conventional myosin, and K actin of 112 M and V max of 0.34 s Ϫ1 were obtained in 25 mM KCl at 25°C (Fig. 6). It should be noted that the autophosphorylation did not significantly affect the actin-activated ATPase activity (not shown). Fig. 7 shows the actin-activated ATPase activity of huM3AIQ2 as a function of ATP concentration in the presence of the ATP regeneration system. A solid line is the calculated line based upon the equation According to the analysis, V max and K ATP were estimated to be 0.11 s Ϫ1 and 6.2 M, respectively. Whereas we measured the ATPase activity of huM3AIQ2, we found that the rate of P i liberation significantly decreased with time (Fig. 8). Approximately 50 M ATP of the initial 200 M ATP was consumed at 120 min. Because K ATP was 6.2 M, it is unlikely that the decrease in the activity is due to the depletion of ATP. This time-dependent inhibition of the P i liberation was abolished when the ATP regeneration system was included in the reaction mixture (Fig. 8). In contrast, the time-dependent inhibition of the ATPase activity was not observed for skeletal S1 subfragment (Fig. 8, inset). Similar time-dependent inhibition of the ATPase activity has been reported for myosin V. The result FIG. 4. Effect of autophosphorylation of myosin III on its kinase activity. 20 g/ml huM3AIQ2 was incubated in the presence or absence of 0.2 mM ATP for 90 min as described under "Experimental Procedures," to complete autophosphorylation of myosin III, and then 0.1 mg/ml MBP and 0.2 mM [␥-32 P]ATP were added to the reaction mixture. The specific radioactivity was adjusted for the phosphorylation of myelin basic protein by phosphorylated and unphosphorylated myosin III. The reaction mixture was incubated for a time as indicated and subjected to SDS-PAGE. The rate of phosphorylation was determined using an FLA-5000 phosphor imager (Fujifilm). A, autoradiogram. B, time course of phosphorylation of myelin basic protein determined by the phosphor imager. Open circles, phosphorylation of myelin basic protein by autophosphorylated huM3AIQ2; closed circles, phosphorylation of myelin basic protein by nonautophosphorylated huM3AIQ2 .   FIG. 5. Limited proteolysis of the autophosphorylated huM3AIQ2. huM3AIQ2 was autophosphorylated as described in the legend to Fig.  3. 0.1 mg/ml autophosphorylated huM3AIQ2 was incubated with 2.5 g/ml chymotrypsin for various times, and then the reaction was stopped by adding 5% trichloroacetic acid. After the protein was precipitated by centrifugation, the pellets were dissolved by adding 0.1 M NaHCO 3 and subjected to SDS-PAGE. A, Coomassie Brilliant Blue staining. B, autoradiogram.
suggested that ADP produced during the ATPase reaction inhibits the ATPase activity, presumably due to relatively high affinity for ADP. Based upon the equation , and K ATP ϭ 6.2 M, K ADP was estimated from the initial phase of the fitting curve to be 1.5 M. To further confirm this notion, the effect of ADP on the actin-activated ATPase activity of huM3AIQ2 was examined (Fig. 9). The addition of ADP to the reaction mixture significantly inhibited the actin-activated ATPase activity of huM3AIQ2 (Fig. 9). On the other hand, the activity of the skeletal S1 fragment was not inhibited (Fig. 9, inset). Binding of Myosin III to Actin-It is known that conventional myosin dissociates from actin upon ATP binding per each cross-bridge cycle. On the other hand, processive myosins such as myosin V and myosin VI have apparent high affinity for actin and do not readily dissociate from actin upon ATP binding. Fig. 10 shows the binding to and dissociation from actin of huM3AIQ2. huM3AIQ2 was mixed with actin in the presence and absence of Mg 2ϩ -ATP, and the fractions of myosin III bound to and dissociated from actin were determined by sedimentation analysis. huM3AIQ2 bound to actin in the absence of ATP, but the majority of the bound huM3AIQ2 was dissociated from actin upon the addition of ATP. The results suggest that human myosin III readily dissociates from actin upon ATP binding, suggesting that human myosin III spends the majority of time in the actin-dissociated form during the cross-bridge cycle.
The Actin Translocating Activity of Myosin III-The above results clearly indicate that human myosin III has the signif- ). According to the analysis, V max and K ATP were 0.11 s Ϫ1 and 6.2 M, respectively. Note that there was some ATPase activity without the addition of ATP, and this is due to the presence of residual nucleotide in the actin preparation.  Fig. 6. Approximately 50 M ATP was consumed at 120 min. The inset shows that the actin-activated ATPase activity of skeletal S1 was not affected by the presence of the ATP regeneration system. 1 mM ATP was used for skeletal S1. ). According to the analysis, a K ADP of huM3AIQ2 was 8.5 M. The inset shows that the actin-activated ATPase activity of skeletal S1 was not affected by the presence of ADP.
icant actin-activated ATPase activity that is thought to be coupled with the myosin motor function. To directly demonstrate the motor function of human myosin III, we performed an in vitro actin gliding assay (37). As shown in Fig. 11A, huM3AIQ2 translocated the actin filaments. The mean actin translocating velocity was 0.11 Ϯ 0.05 m/s that was significantly slower than other myosin superfamily members (31, 35, 38 -42). Interestingly, high Ca 2ϩ concentration did not abolish the motility activity of human myosin III in contrast to other calmodulin binding myosins such as myosin V (31,35), myosin I (27,43), and myosin IX (41).
Direction of the Movement-Whereas it was thought that myosin moves toward the plus end of actin filament, recent studies have revealed that there are myosins that move in the opposite direction on actin filaments (38). It was originally hypothesized that the unique large insertion of myosin VI between the motor domain and the neck domain was responsible for the reverse directionality of motility, thus making the class VI myosin the only minus-directed myosin. However, recent studies have revealed that this is not the case and that there are additional members in the myosin superfamily that move in the minus direction (29,41). To determine the direction of movement of human myosin III, we utilized F-actin filaments in the in vitro motility assay that were labeled throughout with fluorescein and labeled with a rhodamine cap at the filament's pointed end (see "Experimental Procedures"). The dual fluorescence-labeled F-actin filaments were visualized under the fluorescence microscope moving on coverslips coated with huM3AIQ2. As shown in Fig. 12A, huM3AIQ2 moved the dual fluorescence-labeled F-actin with the pointed end at the front of the movement. This means that myosin III moves toward the barbed end, as is known for the conventional myosins. Fig. 12B shows a histogram of the velocities of polarity-marked actin filaments on huM3AIQ2-coated coverslips. Whereas some variation of the sliding velocity was observed, all actin filaments moved in the same direction. The result shows that human myosin III is a plus-directed motor.

DISCUSSION
At least 18 classes of myosins are present (1,2), and these myosins are thought to play a key role in diverse actin-based cellular motile processes. For the class III myosin, most of the studies have been made with Drosophila ninaC gene product and the importance of ninaC myosin III in the phototransduction process has been indicated (15,16). Recent genetic studies have also revealed the involvement of the class III myosin gene in human sensory disorders (19). A key piece of information lacking in understanding the function of myosin III is the functional study of myosin III at a molecular level, and there is no evidence indicating whether myosin III is actually the actinbased motor protein. The present study showed for the first time that class III myosin is an actin-based motor protein having a protein kinase activity.
In the present study, we used the recombinant myosin III rather than attempting to purify it from tissues. Obviously, it would be quite difficult to obtain enough myosin III from tissues for the biochemical experiments, and the recombinant technique overcomes this problem. Furthermore, it is known that various types of myosins are present in the same tissue, and it would be difficult to completely eliminate the contamination of other myosins when myosin III is purified from the tissues. The produced human myosin III construct contains the protein kinase domain, the motor domain, and the two IQ domains adjacent to the motor domain that cover all biochemical functional domains of human myosin III.
Human myosin III showed the actin-translocating velocity of 0.11 Ϯ 0.05 m/s at 25°C. The slow velocity is consistent with the slow cycling rate, i.e. the low actin-activated ATPase activity (0.34 s Ϫ1 ). One of the critical issues to understanding the physiological function of myosin motors is the processivity. It has been known that myosin dissociates from actin upon ATP binding and spends a majority of the time during the crossbridge cycle in the actin-dissociated form. In other words, each cross-bridge cycle involves the dissociation and reassociation of myosin motor domain to actin. However, recent findings that myosin V (44 -47), myosin VI (48,49), and myosin IX (41) move along the actin filament for multiple steps without dissociating from actin thus processive have fundamentally expanded the physiological roles of myosins in various biological contractile and motile activities. Whereas the question of whether myosin III is a processive or nonprocessive motor needs to be answered by further studies, the present results suggest that myosin III is a nonprocessive motor. The K actin value was much higher than the myosin V and myosin VI, which are known to be processive motors, and similar to myosin II, a nonprocessive myosin. Furthermore, the predominant fraction of myosin III was dissociated from actin in the presence of ATP, which suggests that myosin III spends the majority of time in weak actin binding states during the cross-bridge cycling. These properties are consistent with the notion that myosin III is a nonprocessive motor.
Directionality of myosin movement is another important issue to understand the physiological function of myosin III. The present results clearly indicated that myosin III is a plus-ended motor unlike myosin VI (38). Together, the present results suggest that myosin III is a single-headed plus directed nonprocessive motor. It is known that myosin III is present in a stack of microvilli in photoreceptor cells known as a rhabdomere, which resembles the brush border microvilli, and the rhabdomeric microvilli are structurally similar and consist of highly ordered microvilli composed of actin filaments connected to the surrounding plasma membrane by radial links (50). In terms of the properties as a motor protein, myosin III is quite similar to brush border myosin I (i.e. slow actin-translocating velocity (43), nonprocessive nature, and plus directionality). Therefore, it is plausible that myosin III, which is localized to the rhabdomeric microvilli, may have a similar function as the brush border myosin I, linking the actin filament bundle that forms the structural core of the microvilli with the microvillar plasma membrane and functioning as a mechanoenzyme (51). This view is consistent with the fact that the deletion of the motor domain of ninaC myosin III abolishes the rhabdomere localization of myosin III (16).
However, a major difference between these two actin-based motors is the presence of protein kinase activity in myosin III. Previously, we found that the kinase domain of Drosophila ninaC myosin III actually exhibits the protein kinase activity (52). The present result is consistent with the earlier study and, furthermore, showed that the kinase activity is retained with the presence of the motor domain of myosin III, indicating that the myosin III molecule has both the motor function and the protein kinase function. Interestingly, the substrate specificity of human myosin III was different from that of the Drosophila ninaC gene product previously reported (52). Whereas both class III myosins carry the N-terminal kinase domain, the sequence similarity of the kinase domain of this myosin III is not high, showing only 49% identity. The different substrate specificity of these myosin IIIs is consistent with this notion.
We found that myosin III undergoes autophosphorylation with an intramolecular process. The autophosphorylated domain was assigned to be the C-terminal end of the motor domain. Because the autophosphorylation is the intramolecular reaction, it is likely that the kinase domain interacts with the C-terminal end of the motor domain of its own molecule. In other words, the kinase domain lies in close proximity to the C-terminal end of the motor domain.
Quite interestingly, the autophosphorylation significantly activated the protein kinase activity of human myosin III. The results raise the following scenario for the autophosphorylation-induced activation of the kinase activity of myosin III. The kinase-active site is in close contact with the autophosphorylation sites at the C-terminal side of the motor domain that lies near the autoinhibitory region of the kinase. The phosphorylation attenuates the inhibitory activity of the autoinhibitory region, thus activating the protein kinase activity of myosin III. It is known that calmodulin-dependent protein kinase II is significantly activated by autophosphorylation at Thr 287 , which lies near the autoinhibitory region of the kinase (53)(54)(55). The autophosphorylation-induced activation of the protein kinase activity is also known for myosin light chain kinase, in which the autophosphorylation site is found to be in close proximity to the autoinhibitory region (56). Because autophosphorylation is the intramolecular process, the rate of autophosphorylation is independent to myosin III concentration, and therefore the observed rate in the present study probably represents the rate in the cell environment. It is reasonable to assume that myosin III continuously undergoes autophosphorylation in the cell, where plenty of ATP is present. Because the rate of autophosphorylation is quite slow, it is plausible that the phosphorylation state of myosin III is regulated via regulation of protein phosphatase activity. The determination of the physiological role of the protein kinase function of myosin III requires further studies.