|Home | About | Journals | Submit | Contact Us | Français|
In regulated myosins, motor and enzymatic activity are toggled between on- and off-states by a switch located on its lever arm, or regulatory domain (RD). This region consists of a long alpha-helical "heavy chain" stabilized by a "regulatory" and an "essential" light chain. The on-state is activated by phosphorylation of the regulatory light chain of smooth muscle RD, or by direct binding of Ca2+ to the essential light chain of molluscan RD. Crystal structures are available only for the molluscan RD. To understand the pathway between the on and off states in more detail, we have now also determined the crystal structure of a molluscan (scallop) RD in the absence of Ca2+. Our results indicate that loss of Ca2+ abolishes most of the interactions between the light chains and may increase flexibility of the RD heavy chain. We propose that disruption of critical links with the C-lobe of the regulatory light chain is the key event initiating the off-state in both smooth muscle and molluscan myosins.
The muscle machinery consists of thick filaments of myosin that translocate along thin filaments of actin (F-actin) powered by ATP hydrolysis. The motor that drives this action is located in the myosin head, which consists of a motor domain and a lever arm domain, here called the regulatory domain (RD). The RD comprises the long heavy chain (HC) α-helix of the lever arm, stabilized by two types of light chain, the essential light chain (ELC) and the regulatory light chain (RLC)1. Each of the light chains, which belong to the calmodulin (CaM) superfamily of proteins2, consists of an N-lobe (helices A, B, C, and D) and a C-lobe (helices E, F, G, and H) that have specific interactions with the HC3; 4. A flexible linker joins helices D and E. The light chains act as regulatory switches in molluscan (e.g., scallop) and vertebrate smooth muscle myosins but not in the myosins of vertebrate striated muscles. In all muscles, contraction is initiated by the release of Ca2+ stored in vesicles. In vertebrate striated muscles, contraction is initiated by the binding of Ca2+ to a troponin-tropomyosin complex on F-actin. Activity of isolated myosin, however, remains locked in an on-state regardless of Ca2+ concentration (for reviews, see refs.5; 6). In regulated myosins, the light chains control myosin activity even in the absence of actin7.
The Ca2+-mediated triggers of molluscan and smooth muscle myosins are different. In molluscan muscle myosin, motor and enzymatic activity are switched "on" when Ca2+ binds directly to the ELC (Figure 1)7; 8; 9. The ELC of molluscan myosins contains a unique sequence of five residues that provide ligands for the binding of Ca2+ even in the presence of excess magnesium. The Ca2+-chelating ability of these residues is stabilized by interactions with a crucial glycine residue of the RLC (Gly117 in scallop)3; 4; 9. The highly homologous vertebrate smooth muscle myosin lacks Ca2+ binding ligands10, and force generation is activated indirectly when Ca2+ is bound by calmodulin, initiating a series of events that include activation of myosin-light chain kinase (MLCK) and finally phosphorylation of the RLC (for review, see ref. 11). Each of these biochemical steps presents a potential entry point for regulatory control of smooth muscle contraction and its rate of activation. As a result, the relatively slow response of smooth muscle to Ca2+ release allows a variety of controls over regulation important for the survival of the organism. The RLC of vertebrate smooth muscle myosin has an extension on the N-terminus that provides a binding site for MLCK, including an arginine critical for kinase activity12. By contrast, the scallop RLC has a shorter extension and lacks this arginine (it is replaced by lysine). As a result, the scallop RLC cannot be phosphorylated by MLCK. Thus, although both light chains are necessary for maintaining the RD structure, regulation of smooth muscle myosin depends only on the RLC, whereas regulation of molluscan myosin depends on both the RLC and ELC. In short, the initial step in the regulatory switch mechanism (i.e. the trigger) of molluscan myosin is clearly different from that of smooth muscle13.
There are, however, similarities between regulation in molluscan and smooth muscle myosins (see, for example, ref. 14). There is evidence that in both myosins one or more of the last several carboxy-terminal residues of the RLC are required for regulation15; 16. Of these residues, scallop Lys149 (gizzard smooth Lys163), is conserved in all regulated myosins but is absent from all unregulated myosin II's2. In both isoforms, regulation requires a myosin dimer containing two myosin heads and a portion of the coiled-coil rod which connects them14; 17. Heavy meromyosin (HMM) is an example of a regulated two-headed myosin fragment. In a hybrid of HMM containing scallop heavy chains, scallop ELCs, and smooth muscle RLCs, the MgATPase and motor activity can be regulated by either phosphorylation or by Ca2+-binding18; 19. This finding indicates that the smooth muscle RLC contains all the components required by both trigger mechanisms, providing that the appropriate ELC is present. If we assume that both triggers can induce the same off-state conformation, then the hybrid results suggest that both phosphorylation (the smooth muscle trigger) and Ca2+-binding (the molluscan trigger) are two separate pathways activating the same underlying regulatory machinery. Cryo-EM studies indicate that the asymmetric off-state conformation of smooth muscle (involving quaternary interactions between two myosin heads)20; 21; 22 closely resembles that of scallop23, as well as tarantula myosin thick filaments (which are regulated by phosphorylation as in smooth muscle myosin)24. In summary, although the initial trigger for regulation differs in scallop and smooth muscle myosins, experimental results suggest that the off-state conformation - and at least some components of the regulatory machinery - may be similar for both types of myosin.
We do not yet know how dissociation of Ca2+ from scallop RD sets in motion a conformational change leading to the off-state. In scallop myosin, regulation is abolished if the last eleven carboxy-terminal residues of its RLC are substituted with different amino acids16, but specific binding of Ca2+ remains intact. This finding demonstrates that the RLC has distinct functional regions involved in switch activation. This view is also supported by a D39A point mutation in scallop RLC, which weakens RLC binding and results in loss of regulation without loss of Ca2+ binding16.
In order to investigate Ca2+-mediated regulation of scallop myosin, 2.8 Å and 2.0 Å crystal structures of scallop Ca2+ -bound RD have been determined in this laboratory and were interpreted as corresponding to the switch as it appears in the on-state3; 4. It was proposed that the absence of Ca2+ from the its ELC binding site might result in electrostatic repulsion between the coordinating oxygen atoms, leading to a local conformational change that would weaken ELC/RLC interactions and allow more flexibility of the heavy chain. This flexibility would allow formation of off-state interactions between two myosin heads that would affect ATPase activity and contractility3. For example, head-head contacts between the motor domains could trap them in conformations that prevent catalysis and mechanical motion, such as indicated by the cryo-EM studies cited above20; 21; 22; 24. A structure of Ca2+ -free scallop myosin has not yet been reported, however, to confirm this hypothesis. Moreover, as a result of proteolysis, a number of N-terminal residues of the RLC were missing from previously published scallop RD crystal structures, including a serine (Ser6) at a position corresponding to the regulatory phosphorylation site in homologous smooth muscle RLC. To date, the smooth muscle RD has not been crystallized. To extend the investigation of how the molluscan regulatory switch toggles myosin between the on- and off-states and in an attempt to gain some insight into the phosphorylation machinery of smooth muscle myosin, we have determined two scallop RD crystal structures, reconstituted with complete light chains. One is a 2.6 Å resolution structure of an RD containing an ELC D19A point mutation (D19A-RD), which prevents Ca2+ from binding to the regulatory trigger site (see below under "The Ca2+-RD Binding Site" for preliminary work involving wild-type RD in the presence of EGTA). The other structure is of Ca2+-RD at 2.1 Å resolution, to be used as a control for the conformation of the RLC N-terminus as well as the Ca2+ binding site. Our results indicate a weakening of ELC/RLC interactions as a result of local conformational changes at the Ca2+-binding site and lead us to a simple hypothesis that accounts for the on-off switch in regulated myosins.
The current structures reveal the first comparison between Ca2+-bound and Ca2+ -free scallop RD. The overall conformation of the Ca2+-free D19A-RD structure is similar to that observed in both the current and previous scallop Ca2+-RD and Ca2+-S1 crystal structures (Figure 1a)3; 4. By contrast to the previous crystallographic analyses of partially proteolyzed RD, however, the current experiments include the entire N-terminus of the RLC. The crystal structures show that the N-terminus of the RLC adopts extended, partially disordered, conformations which are different in the Ca2+ -bound and Ca2+ -free structures. It is difficult, however, to relate these differences simply to the state of the Ca2+ binding site. Although the overall conformations of the Ca2+ -RD and Ca2+ -free D19A-RD structures are similar, we observe a local change at the Ca2+ binding site (Figure 1a,c), which appears to weaken electrostatic interactions between the two light chains and suggests greater flexibility when Ca2+ is not bound.
In the Ca2+-RD structure, the Ca2+ ion is coordinated by seven oxygen ligands, in agreement with previous scallop RD crystal structures3; 4. These ligands include three main-chain carbonyls (ELC residues Asp19, Gly23, and Ala27), three side-chain carboxyl oxygen atoms (residues Asp19, Asp22, and Asp25), as well as one water molecule (Figure 1b). In the absence of Ca2+ (Figure 1c), there is no electron density for the water ligand. In addition, the Asp22 side-chain is turned away from the Ca2+ binding site. Unlike the Ca2+-RD structure, the electron density for ELC residue Gly23 in D19A-RD is difficult to interpret, but the carbonyl group clearly does not interact with the Ca2+ site and appears to have rotated away from it. Neighboring residue Arg24 is partially disordered. The side-chain of residue Asp25 now occupies two alternative conformations, one pointing toward the Ca2+ site and one pointing away from it. In preliminary studies, we crystallized wild-type RD in the presence of EGTA to remove Ca2+. Electron density could still be seen, however, at the Ca2+-binding site (results not shown), and we could not determine whether that density was the result of Ca2+ or water. The D19A-RD construct was designed to ensure that no cation is bound at this site (see Materials and Methods). Difference Fourier (Fo−Fc) electron density is still visible in the Ca2+ binding site of D19A-RD, although there are too few coordinating oxygen ligands to justify a Ca2+ atom at that position. An atom at that location can readily form hydrogen bonds with the carboxyl groups of Asp19 and Ala27. This Fo−Fc density vanishes if a water molecule is modeled into the site. Modeling Mg2+ or Ca2+, however, results in a negative density. Thus, the site is probably occupied by a structural water, not a cation. This finding agrees with the observation that D19A-RD has a pKca of about 4, compared to a value of about 6 for other scallop myosin S1 and RD constructs25; 26.
Compared to the Ca2+-RD coordinates, the D19A-RD coordinates suggest that removing the Ca2+ ion may increase mobility and weaken interactions at the ELC/RLC interface. The main-chain B-factors of ELC residues Asp22 to Gly26, comprising most of the Ca2+ binding loop, rise sharply compared to nearby residues (Figure 2) and spike at Arg24, an indication of high mobility. No such B-factor rise is seen in the equivalent residues of the Ca2+-RD structure.
Critical electrostatic interactions between the two light chains have been lessened in the Ca2+-free structure, particularly in the supporting "meshwork" near Gly23 and Asp19 which is vital for coordination of the Ca2+ ion3; 4 (Figure 3b,c). The main-chain hydrogen bond between ELC residue Arg24 and RLC residue Gly117 has broken, as have two hydrogen bonds between the side-chains of ELC residue Arg24 and RLC residue Asp118. Some interactions between the RLC and HC by the ELC/RLC interface have become stretched to distances of 3.1 Å or greater, and a hydrogen bond between RLC residue Leu113 and HC residue Gln812 has been broken (Figure 3b,c). The overall interactions between this ELC loop and the HC have remained intact, however, including a 3.5 Å hydrophobic contact between the Phe20 benzene ring and the side-chain of HC residue Arg805. Adjacent HC residues, Lys796 through Gln804, have no interactions with either light chain, suggesting that this region may become more flexible than surrounding regions of the HC in the absence of strong interactions between the ELC, RLC, and HC, as previously proposed3. Indeed, these residues have elevated B-factors only in the D19A-RD structure, but not in the Ca2+ -RD coordinates (Figure 2). This region is, in fact, precisely the location of a slight bend in the HC that brings the light chains close enough to interact. It is possible that the full range of HC flexibility cannot be visualized in the D19A-RD structure because of the stabilizing effects of crystal contacts, which are very similar to those of the Ca2+-RD structure. To summarize, in the Ca2+-free structure, interactions between the ELC, RLC, and HC are weak near the ion's binding site, and, as a result, the adjacent solvent-exposed region of the HC is likely to be significantly more flexible than in Ca2+-bound structures.
In comparing the current scallop RD structures, we find that the heavy chain is slightly more bent in Ca2+-RD than in the Ca2+-free D19A-RD, but it is not clear whether this is a result of the Ca2+ occupancy at its ELC binding site or a sampling of the normal plasticity of the RD HC. We find, however, that if we allow a modest straightening of the RD heavy chain by the ELC/RLC interface, we can improve the molecular fit to the recent HMM off-state cryo-EM map20; 21 (Himmel, unpublished result). Crystallographic experiments are in progress with a molluscan HMM off-state conformation to confirm this observation.
One outstanding question is why the RLC carboxy-terminal residues of helix H are required for regulation in both molluscan and smooth muscle myosins, although those residues are not required for either Ca2+ or phosphate binding. As mentioned above, one of these residues, scallop Lys149 (gizzard smooth Lys163), is conserved only in regulated myosins2. In the Ca2+-RD structure, Lys149 forms a hydrogen bond with the side chain of RLC residue Thr83 (located on helix E), whereas in D19A-RD this hydrogen bond is broken. This observation led us to consider whether all regulated myosins can modulate the rigidity of the lever arm heavy chain by maintaining links only in the on-state between (1) the RLC N- and C-lobes and (2) between the RLC C-lobe and the ELC. This possibility would be consistent with the finding from a cross-linking study that the RLC N-lobe must rotate away from the ELC to achieve the off-state in regulated myosin27. It would appear that a true on-state in smooth muscle myosin, and possibly even in molluscan myosin, might involve salt bridge and hydrogen bond contacts between RLC helices A, E, and H that would be further stabilized by a phosphate binding site near helix A. Breaking of these links - either indirectly by increasing mobility of the RLC C-lobe and ELC N-lobe (in molluscan muscle), or directly by dephosphorylation (in smooth muscle) -would permit myosin to fold into an off-state conformation. The helix A N-terminal residues that would be expected to interact with helix E are in close proximity to scallop RLC residue Ser6, whose equivalent in smooth muscle serves as the phosphorylation switch (Figure 4). Indeed, examination of the structure of the lever arm of chicken skeletal myosin28 has previously led Trybus et al. to propose that phosphate might itself form one or more salt bridges between RLC helices A and H during the on-state of smooth muscle myosin29. With the exception of the Lys149-Thr83 salt bridge, however, we do not consistently see links between helices A, E, and H in molluscan myosin structures to support these models. A structure of smooth muscle RD with a phosphorylated RLC would therefore be an important contribution to our understanding of myosin regulation. An important step, nevertheless, can be taken toward understanding the regulatory role of RLC helix H by noting that disruption of the Lys149-Thr83 salt bridge increases the degrees of freedom of movement between the RLC C- and N-lobes.
This notion allows us to posit a simple picture for the switch mechanism, derived directly from the crystallographic observation that Ca2+ dissociation weakens links with the RLC C-lobe in molluscan myosin. It is possible that the key to initiating an off-state is destabilization of the RLC C-lobe. In scallop myosin, this step would be achieved by breaking Ca2+-dependent links to the ELC. The N-terminal segment of the RLC, which contains the phosphate binding site in smooth muscle myosin, is adjacent to the C-lobe of the RLC, including helices E and H (Figure 4). Although a smooth muscle RD structure has not yet been reported, it might well resemble that of squid S130 and chicken skeletal S128.In these structures, the RLC N-lobe is rotated in such a way that links between helices A, E, and H may be possible if the RLC N-terminus is phosphorylated (Figure 5). If so, then these links would be broken by dephosphorylation in smooth muscle myosin. As a result of sequence differences in the ELC N-lobe, we can assume that ELC/RLC contacts are likely to be different in smooth muscle than in molluscan myosins. If those contacts are not extensive, then it would appear that regulation in both types of myosin depends critically on the ability to modulate the stability of key interactions with the RLC C-lobe.
To understand the basis of the on-off triggers in regulated myosins, we have determined structures for Ca2+-bound and Ca2+-free scallop RD reconstituted with complete (non-proteolyzed) light chains. These molluscan structures indicate that, when Ca2+ dissociates from the ELC of scallop myosin, a small conformational change near the Ca2+ binding site leads to a loss of interactions between the ELC N-lobe and RLC C-lobe. This finding, together with elevated B-factors in nearby HC residues in the absence of Ca2+ binding, supports a previous hypothesis3 that the HC of Ca2+-free molluscan myosin becomes significantly more flexible than when Ca2+ is bound. Although we have not been able to draw definitive conclusions about the phosphorylation trigger in smooth muscle, it is possible that in both molluscan and smooth muscle myosins the trigger modulates the intrinsic flexibility of the RD and thereby controls the number of conformations available to the myosin head. We propose here that disruption of key links to the RLC C-lobe may be the basis for initiating the off-state in all regulated myosins.
The scallop RD heavy chain fragment was prepared from papain-digested myosin and reconstituted with intact RLC and ELC as previously described26. The D19A ELC mutant was obtained by subcloning into the Nde 1 and EcoR1 sites of pMW172 as described9. Purified RD was stored on ice in 5 mM HEPES pH 7, 20 mM NaCl, 2 mM MgCl2, 1 mM EGTA, 3 mM NaN3, 0.5 µg/ml leupeptin, 0.7 µg/ml pepstatin, and, for the Ca2+-RD, 0.2 mM CaCl2. The D19A-RD preparation was dialyzed extensively against large volumes of 1 mM EGTA overnight to remove Ca2+ ions.
Crystals were grown at 4°C in microseeded hanging drops consisting of equal volumes of 16.9 mg/ml protein and precipitant solution (for Ca2+-RD: 150 mM HEPES pH 7.0, 50 mM (NH4)2SO4, 0.5 mM CaCl2, 5 mM MgCl2, 5 mM NaN3, 16.5% (wt/vol) PEG 4000; for D19A-RD: 100 mM bicine pH 8.3, 45 mM (NH4)2SO4, 5 mM MgCl2, 2.5 mM EGTA, 3.0 mM NaN3, 15% (wt/vol) PEG 4000, 4% (vol/vol) PEG 200, 6% (wt/vol) 2,3-butanediol). Crystals were transferred to mother liquor containing 25% (vol/vol) glycerol as a cryoprotectant. They were subsequently flash-cooled in liquid propane, stored as frozen popsicles in liquid N2. Crystals for both Ca2+-RD and D19A-RD were triclinic. Data were collected at a temperature of 100 Kelvin at the Brookhaven National Laboratory National Synchrotron Light Source and the Cornell High Energy Synchrotron Source and were processed using HKL-DENZO-SCALEPACK31.
Phases were solved by molecular replacement and rigid body refinement with the program AMoRe32 using a proteolytically-derived scallop Ca2+-RD structure (PDB accession number 1WDC)3 as an initial search model. Stepwise model building and refinement were conducted using the "O" graphics package33, the Coot graphics package34, XPLOR35, and CNS36 with a bulk solvent correction. Water molecules were added manually in the final stages of refinement.
We dedicate this paper to the memory of S. Mui, a gifted colleague who will be missed. We are grateful to A. Houdusse for preliminary work on this project, J. H. Brown for helpful discussion and advice, as well as to the staffs of the Cornell High Energy Synchrotron Source and the Brookhaven National Laboratory for assistance with data collection. Figures were generated in Molscript, PovRay, and Pymol. This work was supported by grants to C.C. from the National Institutes of Health (AR17346) and the Muscular Dystrophy Association, as well as to C.C. and A. G. S. from the National Institutes of Health (AR41808).
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Abbreviations: CaM, ELC, EM, HC, HMM, MLCK, RD, RLC
Coordinates have been deposited with the Protein Data Bank with accession no. XXXX for Ca2+-RD and 3JTD for D19A-RD.