Myosin II thick filament assembly in Dictyostelium is regulated by phosphorylation at three threonines in the tail region of the molecule. Converting these three threonines to aspartates (3×Asp myosin II), which mimics the phosphorylated state, inhibits filament assembly in vitro, and 3×Asp myosin II fails to rescue myosin II–null phenotypes. Here we report a suppressor screen of Dictyostelium myosin II–null cells containing 3×Asp myosin II, which reveals a 21-kD region in the tail that is critical for the phosphorylation control. These data, combined with new structural evidence from electron microscopy and sequence analyses, provide evidence that thick filament assembly control involves the folding of myosin II into a bent monomer, which is unable to incorporate into thick filaments. The data are consistent with a structural model for the bent monomer in which two specific regions of the tail interact to form an antiparallel tetrameric coiled–coil structure.
The spatial and temporal control of the assembly and disassembly of organelles is fundamental to cell and developmental biology. The cytoskeleton is particularly dynamic in vivo, responding to both internal and external signals that induce rapid changes in assembly state. A key component of the actin-based cytoskeleton is myosin II, which has been shown to be essential for cytokinesis of Dictyostelium cells in suspension as well as for efficient chemotaxis and morphogenetic changes in shape during development (De Lozanne and Spudich 1987; Knecht and Loomis 1987; Zang and Spudich 1998). All of these roles require myosin II to be in the form of thick filaments.
A myosin II molecule consists of a pair of heavy chains and two pairs of light chains. Each heavy chain starts with an NH2-terminal globular head, which has ATPase and motor activity, followed by a long α-helical tail. Analysis of the tail sequences of muscle and nonmuscle myosin IIs reveals multiple repeating patterns throughout the entire domain that enable the two tails to form an α-helical coiled–coil rod. The smallest repeat motif consists of seven amino acids occupied at position a–g in the helical turn. Generally, small and hydrophobic residues are found in positions a and d, which form the core and are essential to the formation of the coiled–coil structure (Parry 1981). In all myosin IIs examined, the region of the coiled–coil tail that is required for assembly into thick filaments is in the COOH-terminal half of the tail.
In Dictyostelium, myosin II molecules constantly relocate to multiple locations for participating in various processes. When a cell migrates, myosin II accumulates in the posterior of the cell. During cell division, myosin II accumulates in the cleavage furrow. To accomplish its cellular tasks, myosin II is thought to assemble into bipolar thick filaments and pull together oppositely oriented actin filaments to produce contractile forces. Mutant forms of myosin II that do not assemble into bipolar thick filaments in vitro fail to rescue myosin-null phenotypes, and they do not localize to the furrow during cytokinesis (Fukui et al. 1990; Sabry et al. 1997). Interestingly, neither motor activity nor the existence of the myosin catalytic domain is necessary for myosin II transportation in cells (Yumura and Uyeda 1997; Zang and Spudich 1998).
The mechanism for Dictyostelium myosin filament assembly is thought to proceed in two distinctive stages. The initial slower step occurs by sequential association of myosin II monomers into parallel dimers and antiparallel tetramers. The next step is rapid lateral addition of myosin dimers to bipolar nuclei to form thick filaments (Mahajan and Pardee 1996). A single point mutation in the tail domain has been shown to be sufficient to inhibit Dictyostelium myosin II filament assembly in vitro, apparently by preventing the formation of dimers and antiparallel tetramers (Moores and Spudich 1998).
The initial step of filament assembly for Dictyostelium myosin II is thought to be induced by dephosphorylation of phosphorylated sites in the tail. Target sites for this phosphorylation control have been mapped to three threonines at positions 1823, 1833, and 2029 (denoted TTT) (Vaillancourt et al. 1988; Luck-Vielmetter et al. 1990). These three threonines are located in a regulatory domain that is ∼23% (34 kD) of the length of the tail, starting from the COOH terminus. A short designation for myosins with different amino acids at positions 1823, 1833, and 2029 takes the form of XXX, where X uses the single amino acid code at each of the three positions. NH2-terminal to the regulatory domain, another 34-kD region has been shown to be the smallest fragment of Dictyostelium myosin II that is necessary and sufficient for self-assembly (O'Halloran et al. 1990; Shoffner and De Lozanne 1996). This region is denoted as the assembly domain.
Pasternak et al. 1989 reported evidence from electron microscopy that phosphorylation of the heavy chain promoted a bent conformation of myosin II at approximately two-thirds the length of the tail from the head-neck junction, and these bent myosin II molecules appeared to be excluded from the myosin filament. Further information was obtained from mutant myosin IIs, with the phosphorylatable threonines replaced by either three aspartates or three alanines, expressed in myosin II–null cells (denoted 3×Asp and 3×Ala cells, respectively) (Luck-Vielmetter et al. 1990; Egelhoff et al. 1993). 3×Asp myosin II mimicked the phosphorylated state, did not assemble into thick filaments in vitro, and failed to rescue myosin II–null phenotypes. Opposite to the 3×Asp mutant, 3×Ala myosin II assembled similar to wild-type myosin II in vitro. These data led to the hypothesis that after being phosphorylated, myosin II monomers prefer a bent conformation that is not able to assemble into parallel dimers, the initial unit for filament assembly (Pasternak et al. 1989). In this report we present a structural model for the myosin II bent monomer and provide evidence that two specific alanine-rich regions of the tail interact to form an antiparallel tetrameric coiled–coil structure.
Materials and Methods
Dictyostelium myosin II–null cells (HS1), transformed with pBigMyD to create wild-type myosin II cells (Ruppel et al. 1994) and with pBIG-ASP to create 3×Asp myosin II cells, were grown as described (Egelhoff et al. 1993). The HL-5 growth medium was supplemented with 60 μg/ml streptomycin, 60 U/ml penicillin, and 5 μg/ml G418 (Geneticin; Life Technologies, Inc.). Bacterial lawns were prepared by spreading 2 ml of an overnight Klebsiella aerogenes culture on SM/5 plates (Sussman 1987). Transformations were performed by electroporation (Egelhoff et al. 1991a) and transformed cells were selected by 5 μg/ml G418.
Mutagenesis and Clonal Isolation of Suppressors for 3×Asp Myosin II Cells
Suppressors of 3×Asp myosin II cells were generated by treatment with 4-nitroquinoline-N-oxide (NQNO) or UV irradiation. Conditions for treatment of 3×Asp myosin II cells using chemical mutagen NQNO were as described by Patterson and Spudich 1995. After treatment with NQNO, cells were shaken in suspension for 30 min, washed twice with HL-5, and 1.2 × 106 cells were plated onto bacterial lawns. For suppressors generated by UV irradiation, 1.2 × 106 3×Asp myosin II cells were spotted onto plates containing bacterial lawn and irradiated with UV (0.8–6.4 mJ/cm2).
Plates from either treatment were placed at 22°C and the developmental phenotype was observed 5 d later to screen for suppressors, which recovered the ability to sporulate. Clonal isolation of suppressors was performed as follows: a single sorus from each plaque was picked by a sterile pipet tip, resuspended in 5 ml HL-5, and spread on a new bacterial lawn to obtain single plaques.
Mutation Mapping and Reconstruction
The tail region of each suppressor was sequenced using standard methods to locate each mutation. The mutations were then introduced into the 3×Asp myosin II sequence by PCR overlap extension mutagenesis (Ho et al. 1989), and then subcloned into pLittleMyo (Moores and Spudich 1998). These constructs were subsequently transformed into myosin II–null cells, and the developmental phenotypes were examined.
Electrophoresis and Immunoblotting
Whole-cell lysates were electrophoresed on SDS/7.5% polyacrylamide gels and transferred to nitrocellulose paper. The paper was probed with anti-Dictyostelium myosin II mAbs, My6 (Peltz et al. 1985) or mAb 55 (kindly provided by Dr. Guenther Gerisch, Max Planck Institut für Biochemie, Martinsried, Germany) (Pagh and Gerisch 1986), and then incubated with a HRP-coupled secondary antibody (Bio-Rad). Signals were visualized with an enhanced chemiluminescence system (DuPont).
Wild-type and 3×Asp myosin IIs were purified from cells grown and harvested as described (Egelhoff et al. 1993), after a modified purification protocol that does not require filament assembly (Moores and Spudich 1998). After being washed with 10 mM Tris-HCl, 1 mM EDTA, pH 7.5, the cell pellets were resuspended in 2 vol/g cells 25 mM Hepes, 50 mM NaCl, 2 mM EDTA, 1 mM DTT, pH 7.4, containing a mixture of protease inhibitors (buffer A) (Moores and Spudich 1998). The cell suspension was frozen by dripping into liquid nitrogen and stored at −80°C. Immediately before preparation, frozen pellets were thawed and lysed in 7 vol/g cells buffer A at 0°C. After sedimentation at 36,000 g for 30 min, the pellets were suspended in 4 vol/g cells 10 mM Hepes, 130 mM NaCl, 1 mM EDTA, 1 mM DTT, pH 7.4. The cell suspension was centrifuged at 36,000 g for 15 min. Extraction of myosin II was achieved by resuspending the pellets in 1.5 vol/g cells 10 mM Hepes, 150 mM NaCl, 5 mM MgCl2, 4 mM ATP, and 1 mM DTT, pH 7.4. After centrifugation at 200,000 g for 30 min, RNase A (5 μg/ml) and PMSF (0.1 mM) were added to the supernatant. The solution was dialyzed against a 12% Aquacide III (Calbiochem) solution in 10 mM Hepes, 200 mM NaCl, 1 mM DTT, pH 7.4, to concentrate the volume ∼20-fold, and further dialyzed in 10 mM Hepes, 200 mM NaCl, 1 mM DTT, pH 7.4 (column buffer). After adding 0.6 M potassium iodide, 20 mM sodium pyrophosphate, and 2 mM ATP, the concentrated dialysate was loaded onto a Biogel A15 gel (Bio-Rad) filtration column (80 ml total volume), which had been equilibrated previously with column buffer and preloaded with 10 ml 0.6 M KI, 20 mM NaPPi. Fractions containing myosin II were pooled and concentrated as mentioned before. Typical final concentrations for myosin IIs were 1–3 mg/ml, as determined by the Bradford method (Bradford 1976), using rabbit skeletal muscle myosin II as the standard.
Rotary Shadowing EM and Analysis
Samples were diluted to 30–50 μg/ml myosin II in 50–70% glycerol, 10 mM Hepes, 200 mM KCl, 1 mM DTT, pH 7.4, and sprayed immediately onto freshly cleaved mica and viewed as described (Flicker et al. 1985). The locations of bends along the tail of myosin II monomers were measured on prints.
Suppressors for 3×Asp Myosin II Cells Were All Intragenic
We used a random mutagenesis approach to identify suppressors responsible for recovery of myosin II function for 3×Asp myosin II cells. 3×Asp myosin II cells are phenotypically identical to myosin II–null cells (Egelhoff et al. 1993). Both fail to complete the Dictyostelium developmental cycle. They arrest at the mound stage (Fig. 1 B). After treatment of cells with the chemical mutagen NQNO or UV irradiation, 3×Asp myosin II cells were spread on bacterial lawns. Any colony that developed past the mound stage was scored as a suppressor. Depending on the extent of suppression, the suppressors were sorted into three groups: limited, medium, and full (Fig. 1, C–E).
Mutagenesis was performed on a strain of Dictyostelium that had its endogenous mhcA gene deleted (HS1) and contains an extrachromosomal plasmid expressing mhcA–3×Asp myosin II (pBIG-ASP). To check whether the suppressor mutations were intragenic or extragenic, the plasmid from each suppressor was rescued and retransformed into unmutagenized myosin II–null cells, and the transformed cells were spread on bacterial lawns. The phenotypes of all the suppressors were reproduced, verifying that all 28 suppressor mutations were intragenic. The characteristics of the suppressors are shown in Table. Typically the expression level of myosins from the suppressors was similar to that from the wild-type and 3×Asp myosin II cells, but the size of the myosins varied (Fig. 2). 7 of 28 of the suppressors were full-length myosin II, 9 were small internal deletions of 1–7 residues, and 12 were truncations from the COOH terminus. There was no correlation between the means of mutagenesis (NQNO or UV) and the sizes of myosin IIs expressed from the suppressors. The sizes of the 12 ΔCOOH terminus suppressor myosin IIs were the same or larger than a previously studied mutant myosin II called ΔC34. ΔC34–myosin II, a truncated Dictyostelium myosin II lacking the 34-kD COOH terminus of the tail (the regulatory domain, residues 1819–2116; see Fig. 3), constitutively assembles into thick filaments, and ΔC34–myosin cells are able to complete the Dictyostelium developmental cycle and form fruiting bodies (O'Halloran and Spudich 1990). Our ΔCOOH terminus myosin II suppressors are likely to be longer variants of the ΔC34–myosin II constitutive assembly phenotype, and we therefore focused on the remaining 16 suppressors.
The Myosin II Suppressor Mutations All Mapped to a Specific Region of the Tail
Sequencing results for all of the 16 full-length or near full-length suppressors revealed mutations that strikingly all lie in a 21-kD region (∼182 amino acids) towards the COOH terminus of the tail (Fig. 3). All of the seven suppressors with single residue mutations resulted from changes of a single nucleotide base pair. The strongest suppressor was strain D1823Y, which had an aspartate to tyrosine mutation at position 1823 (denoted YDD). This position is one of the three targets for myosin II heavy chain kinase (Egelhoff et al. 1993). This residue corresponds to position d in the heptad repeat of the myosin II tail. The rest of the suppressors with single residue changes resulted in the introduction of a proline, which does not exist in the tail of wild-type Dictyostelium myosin II. Interestingly, three independent suppressors recovered from our screen affected Arg 1880, and two affected Arg 1926. Such multiple hits imply that these positions may play critical roles in regulating filament assembly. The nine small internal deletion group of suppressors had deletions of one to seven amino acids in this region. As shown in Fig. 3 B, the locations of the deletion and the single-residue mutation groups did not mix. The deletions all mapped beyond position 1926.
To be certain that the mapped mutations were responsible for the suppression phenotype and not mutations elsewhere in the myosin II that we may have missed, these mutations were recreated using PCR overlap extension mutagenesis of a 3×Asp myosin II gene contained within an extrachromosomal Dictyostelium expression vector. The plasmids were transformed into myosin II–null cells and the development ability on bacterial lawns was tested. The sporulation phenotypes were identical to the original suppressor strains (e.g., see Fig. 1, F–H).
Electron Microscopy Reveals that 3×Asp Myosin II Is Mainly in a Bent Monomer Conformation
Dictyostelium 3×Asp myosin II molecules were monomeric at high ionic strength. Rotary shadowed 3×Asp myosin II exhibited primarily two conformations under this condition: straight and bent monomers (Fig. 4 A). Various forms of the bent monomers were observed. In 20% of the bent monomer images, the COOH terminus of the tail folded back tightly and resulted in an apparently shorter tail (Fig. 4 A, lower panel). 77% of the 3×Asp myosin II molecules were found to be in the bent conformation (n = 400). On the other hand, only 23% of the wild-type myosin II molecules were found to be bent (n = 280). The percentage of the bent wild-type molecules is consistent with the previous finding that freshly purified wild-type myosin IIs are 20–30% phosphorylated in the heavy chain (Kuczmarski and Spudich 1980).
The majority of the 3×Asp myosin II molecules bend at ∼1,200 Å, located at approximately two-thirds the length of the tail from the head–neck junction (Fig. 4 B). This bent position is similar to that measured for wild-type myosin II monomers (Pasternak et al. 1989). However, we also observed a previously unfound, minor population of bends at ∼1,000 Å (Fig. 4 B). These values are interesting, in that they fit well with the structural motifs described below for the myosin II tail. The relative proportion of bends at 1,000 and 1,200 Å appeared to be the same in 3×Asp and wild-type myosins (∼35%).
The tail domain of Dictyostelium myosin II consists of 1,298 residues, and has no proline interruptions. The coiled–coil prediction algorithm Coils (Lupas et al. 1991) predicts small, distinct regions in the Dictyostelium myosin II tail that have low probabilities to form a coiled–coil structure (Fig. 5). The two most unfavorable regions for coiled–coil structure locate at ∼1,000 and 1,200 Å from the head–neck junction. Consistent with this prediction, to optimize the pattern of charged and uncharged residues for the periodicity of stable coiled–coil structure in the tail, two skips of two amino acids each were necessary to be inserted into the tail sequence lineup at these two regions (Warrick et al. 1986). Similar correlation between the bends observed from EM versus skips in the tail has been shown in smooth and skeletal muscle myosins (Offer 1990), and in Acanthamoeba myosin (Hammer et al. 1987). These results indicate that these two positions at ∼1,000 and 1,200 Å from the head–neck junction are hinge regions in the Dictyostelium tail domain, as seen in other myosin IIs (Tan et al. 1992).
In another region of the tail, closer to the myosin head (∼400 Å from the head–neck junction), there is a small area that has a somewhat lower probability to be in a coiled–coil than the majority of the tail (Fig. 5). In C. elegans, a similar domain has been described as the prehinge (Hoppe and Waterston 1996). A similar region (∼440 Å from the head–neck junction) predicted from muscle myosin IIs has been proposed to serve as a hinge, which allows the head domain to swing away from the thick filament, thus allowing the myosin heads greater freedom to interact with the actin filaments (Offer 1990).
Alanine-rich Core Domains in the Tail Are Consistent with an Antiparallel Tetrameric Coiled–Coil Structure
An analysis of the Dictyostelium myosin II tail sequence revealed two regions (denoted A and B) in the tail that are unusually rich in alanine residues at the core (a or d) positions of the heptad repeats (Fig. 6). Region A spans 18 heptad repeats (between residue numbers 1383 and 1508). Region B contains 23 heptad repeats (residue 1806–1966). More than 75% of all of the alanines at core a and d positions along the tail are found in regions A and B, although these regions account for only 22% of the total tail. The 300–amino acid gap between regions A and B contains the previously identified assembly domain (O'Halloran et al. 1990), which contains only two alanines. The midpoint of the gap is at ∼1,200 Å from the head–neck junction. Core a and d positions rich in alanine residues have been shown to be important for the proper packing of four α-helices into an antiparallel α-helical coiled–coil structure (Monera et al. 1996). Moreover, X-ray crystal structure determination indicates that the ColE1 Rop protein forms a highly regular four stranded α-helix bundle with alanine residues populated in the hydrophobic core (Banner et al. 1987).
The results presented here, together with earlier results (Pasternak et al. 1989; Tan et al. 1992), suggest the following structural model of phosphorylation control of myosin II thick filament assembly (Fig. 7). Phosphorylation by myosin II heavy chain kinase produces charges on the outside of the coiled–coil tail that help stabilize the bent form of the myosin. Bent myosin II molecules cannot associate with other molecules to form parallel dimers, and therefore no antiparallel tetramers appear for the next phase of filament formation. Myosin II heavy chain phosphatase removes phosphates from the bent monomers, and the molecules return to their straight conformation. We propose that the threonine pair 1823/1833 could act as a nucleation site, which when phosphorylated, initiates the bent monomer conformation by orienting the two strands of dimeric coiled–coils. Once nucleated, regions A and B may zip up into an antiparallel four-stranded structure, possibly similar to that observed for the ColE1 Rop protein (Banner et al. 1987). Formation of such a structure results in a major bend at ∼1,200 Å from the head–neck junction. Moreover, the previously identified assembly domain (Fig. 7, blue) (O'Halloran and Spudich 1990) is sequestered as a loop between regions A and B. This conformation of the assembly domain prevents intermolecular interactions that lead to formation of thick filaments. The equilibrium between the bent versus straight conformations is delicately poised, and can be easily disturbed by mutations at multiple sites in the tail (e.g., the suppressor mutations described here; see also Moores and Spudich 1998).
Position 1823 appears to be a particularly critical position for switching between the bent and open conformations. YDD is the only suppressor with a full-length myosin II that completely restores wild-type development. With the aspartate to tyrosine mutation at position 1823, YDD revives myosin II functions possibly by removing the negatively charged aspartate at position 1823 in 3×Asp myosin II that participates in an electrostatic interaction needed for the formation of the bent monomer. Unlike the other two myosin heavy chain kinase sites, residue 1823 corresponds to position d of the heptad repeat of the myosin tail. Position d is at the core of the coiled–coil, which most commonly consists of hydrophobic residues. To tolerate a negatively charged residue at this position in the bent conformation, it is conceivable that a positively charged residue(s) from region A contributes in the core of the antiparallel tetrameric coiled–coil structure. Two possible candidates are lysine 1481 or 1526, at positions a and d, respectively. An interruption of the heptad repeat by polar residues has been observed in the SNARE complex, which is also a highly regulated tetrameric coiled–coil structure (Sutton et al. 1998).
The relative importance of the three threonine targets for myosin heavy chain kinase and phosphatase have been explored by constructing all possible combinations of threonines and aspartates and examining the myosin II functions (Nock, S., W. Liang, H. Warrick, and J.A. Spudich, unpublished data). Consistent with the current report, threonine 1823 appears to be the most critical phosphorylation site for myosin II functions. Simply replacing aspartate at position 1823 with a threonine in 3×Asp myosin II (denoted TDD) is enough to reverse the null phenotypes into wild-type. On the other hand, DTD partially recovers myosin II function, which indicates that position 1833 does play some role. Position 2029 does not appear to be required, because DDT is identical to myosin-null cells. These arguments assume, of course, that an aspartate fully mimics a phosphorylated threonine.
Because TDD has a wild-type phenotype, it is interesting that we did not get a simple reversion from aspartate to threonine in our suppressor screen. This could be due to the fact that this mutation would require two nucleotide changes, which is expected to occur in lower probability. It has been reported that kinases that specifically phosphorylate the three threonines also accept serines (Luck-Vielmetter et al. 1990). The mutation from an aspartate to a serine would also require two nucleotide changes.
The interaction within the alanine-rich core regions A and B appears to be highly sensitive to even single amino acid changes in the tail. All of the suppressors with single point mutations other than YDD, are located in region B, and except for YDD, they all involve the replacement of an amino acid residue with proline. That a single proline at any of three positions appears to be sufficient to destabilize the antiparallel tetrameric coiled–coil domain is possibly explained by its well-known disruptive effect on α-helical structure. The locations of suppressors with proline substitutions (Fig. 7, orange) and deletion mutations (Fig. 7, green) are distinctly clustered. This suggests that the interactions that lead to suppression are different in the two sections of domain B. In one section, prolines may destabilize the bent monomer by disrupting the tetrameric coiled–coil structure. In the other section, deletions may shift the alignment of the coiled–coil regions in the bent monomer, resulting in disruption of critical contacts such as salt bridges important for maintaining the bent monomer. It is important to note that the model described in Fig. 7 only deals with the first step of filament assembly. It is possible that the interactions identified by the suppressors could be between different molecules in higher ordered structures. These structures would be the subsequent steps of the filament assembly pathway. Thus, the proline mutations may locally disrupt the thick filament substructure such that the aspartate residues could be better accommodated. This could shift the equilibrium to favor filament formation. Similarly, the shift of alignment by the deletion mutations could disrupt critical contacts important for the thick filament structure.
It is interesting that no suppressors were found in region A. One possibility is that any mutation in this region affects another step in the pathway for thick filament assembly, which is therefore unable to survive our screening process. In fact, the second half of region A has been strongly implicated in formation of parallel dimers (Pasternak et al. 1989), which are the building blocks for thick filaments.
Several myosin II tail mutants have been constructed with COOH-terminal and internal deletions. COOH-terminal deletion mutants are functional to different extents as long as the assembly domain (Fig. 7, blue) is intact, allowing filaments to form (Egelhoff et al. 1991b; Lee et al. 1994). Mutants that remove parts of the proposed tetrameric coiled–coil structure (region B; Fig. 7, green and orange stripes) give rise to intermediate in vivo phenotypes, indicating that regulation is impaired (Lee et al. 1994). Deletions that remove fragments between the head and region A (Fig. 7, gray stripes) appear to be functional, but if part of region A is removed an intermediate phenotype is observed (Kubalak et al. 1992; Shu et al. 1999). These observations are consistent with the proposal that the tetrameric coiled–coil structure is important for efficient regulation.
Phosphorylation of myosin II heavy chain has been found to occur in a variety of nonmuscle cells as well as in the catch muscle of mollusks, and may be a general mechanism of regulating myosin II function (Tan et al. 1992). It is possible that the hinges in the tail domains of these other myosins are designed for regulation of myosin IIs through a mechanism similar to that proposed here. The bending position of Acanthamoeba myosin II locates at a proportionally similar position as the Dictyostelium region (Hammer et al. 1987). Furthermore, the Acanthamoeba myosin II is phosphorylated at three serines located at the end of the tail (Collins et al. 1982). However, it is controversial whether phosphorylation regulates assembly of the Acanthamoeba myosin II (Collins et al. 1982; Sinard and Pollard 1989).
Myosin II from a molluscan catch muscle is phosphorylated at two serines in the tail domain (Castellani et al. 1988). After phosphorylation, myosin II solubility is enhanced and the molecule folds (Castellani and Cohen 1987), reminiscent of the behavior of Dictyostelium myosin II. Recently, heavy chain phosphorylation of vertebrate nonmuscle myosin IIs and even of smooth muscle myosin II has been reported (Korn and Hammer 1988; Kelley and Adelstein 1990; Fukui and Morita 1996). It remains to be determined whether the model proposed in this study is universal to myosin IIs that are regulated by heavy chain phosphorylation.
Smooth muscle myosin II has two hinge regions located at approximately one-third and two-thirds the length of the tail from the head–neck junction domain (Tan et al. 1992). Although little is known regarding the effects of heavy chain phosphorylation for smooth muscle myosin, in vitro regulation of conformational changes in this myosin, and control of assembly has been reported to be mainly by light chain phosphorylation (Craig et al. 1983; Trybus 1989). However, there is controversy about the state or the extent of light chain phosphorylation change in the filamentous state of smooth muscle myosin II in vivo (Post et al. 1995; Somlyo et al. 1981).
Cells have evolved intricate mechanisms for the control of macromolecular assemblies. The myosin II thick filament is just one example where a delicately balanced equilibrium between monomer and filament forms is used to control cellular function. The model suggested here proposes that phosphorylation on a single threonine residue on a >200-kD protein results in stabilization of a bent monomer form of the molecule. This structural modification shifts the equilibrium to make filament assembly less favorable. Specific kinases and phosphatases can be activated, repressed, or spatially distributed to provide the appropriate regulation signals. A related model has been proposed for the regulation of the activity of heat shock factor in Drosophilia (Rabindran et al. 1993). There, coiled–coil regions of the protein interact within the protein as a monomer or between proteins as the active trimer form. The regulation of macromolecular assemblies via coiled–coil interactions is likely a widely applicable, highly dynamic, important cellular mechanism.
We are grateful to Dr. Guenther Gerisch for his generous gift of the monoclonal antibody mAb 55. We thank Lynne Mercer for help with the electron microscopy and taking the electron microscopy pictures.
W. Liang is supported by a fellowship from the American Cancer Society. This work was supported by National Institutes of Health grant 46551 to J.A. Spudich.
Abbreviation used in this paper: NQNO, 4-nitroquinoline-N-oxide.