Locking Regulatory Myosin in the Off-state with Trifluoperazine*

Scallop striated adductor muscle myosin is a regulatory myosin, its activity being controlled directly through calcium binding. Here, we show that millimolar concentrations of trifluoperazine were effective at removal of all regulatory light chains from scallop myosin or myofibrils. More important, 200 m M trifluoperazine, a concentration 10-fold less than that required for light-chain removal, resulted in the reversible elimination of actin-activated and intrinsic ATPase activities. Unlike desensitization induced by metal ion chelation, which leads to an elevation of activity in the absence of calcium concurrent with regulatory light-chain removal, trifluoperazine caused a decline in actin-activated MgATPase activity both in the presence and absence of calcium. Procedures were equally effective with respect to scallop myosin, myofibrils, subfragment-1, or desensitized myofibrils. Increased a -helicity could be induced in the isolated essential light chain through addition of 100–200 m M trifluoperazine. We propose that micromolar concentrations of trifluoperazine disrupt regulation by binding to a single high-affinity site located in the C-terminal domain of the essential light chain, which locks scallop myosin in a conformation resembling the off-state. At millimolar trifluoperazine concentrations, additional binding sites on both light chains would be filled, leading to regulatory light-chain displacement. 2.5 mg/ml dialyzed against a buffer composed of 200 m M KCl, 30 m M MOPS, 15 m M b -mercaptoethanol, 2.0 M MgCl 2 0.6 m M EGTA, and 0.5 m M CaCl 2 (pH 7.0) at 4 °C. In these experiments, the free calcium concentration was calculated to be 2.1 3 10 2 6 M pro- E-LC E-LC TFP 300 m at 222 using shorter scans (225–205 identical conditions. using two separate preparations of E-LC by electroelution.

Scallop striated adductor muscle myosin is a regulatory myosin, its activity being controlled directly through calcium binding. Here, we show that millimolar concentrations of trifluoperazine were effective at removal of all regulatory light chains from scallop myosin or myofibrils. More important, 200 M trifluoperazine, a concentration 10-fold less than that required for lightchain removal, resulted in the reversible elimination of actin-activated and intrinsic ATPase activities. Unlike desensitization induced by metal ion chelation, which leads to an elevation of activity in the absence of calcium concurrent with regulatory light-chain removal, trifluoperazine caused a decline in actin-activated MgATPase activity both in the presence and absence of calcium. Procedures were equally effective with respect to scallop myosin, myofibrils, subfragment-1, or desensitized myofibrils. Increased ␣-helicity could be induced in the isolated essential light chain through addition of 100 -200 M trifluoperazine. We propose that micromolar concentrations of trifluoperazine disrupt regulation by binding to a single high-affinity site located in the Cterminal domain of the essential light chain, which locks scallop myosin in a conformation resembling the offstate. At millimolar trifluoperazine concentrations, additional binding sites on both light chains would be filled, leading to regulatory light-chain displacement.
Trifluoperazine (TFP), 1 a member of the phenothiazine class of drugs, is one of the strongest antagonists of calmodulin action known, capable of binding to calmodulin in the presence of calcium and preventing its stimulatory effects (1)(2)(3). The structure of the TFP⅐calmodulin complex in the presence of calcium has been determined at 2.45-Å resolution (4,5), where it was shown that TFP induces a profound conformational change in calmodulin, converting the elongated dumbbell to a compact globular structure, analogous to the form obtained through the binding of calmodulin to a target peptide (6,7). This effect was accomplished when a single TFP molecule bound to the predominant binding site, a hydrophobic pocket within the C-terminal domain (4). Additional TFP-binding sites are apparent when the TFP concentration is raised (5). Recently, TFP-binding sites on the related calcium-binding protein troponin C have also been identified (8).
Myosin light chains belong to the same large family of calcium-binding proteins as calmodulin, to which they display a similar overall structure (9,10). There are two types of light chain, termed regulatory (R-LC) and essential (E-LC); one member of each type binds to each of the two heads of conventional myosin II. In those regulatory myosins that are activated directly by calcium binding (for review, see Ref. 11), the exact relationship of the heavy chain to the light chains, in the presence of calcium, is now known in detail, the structure of the regulatory domain of scallop myosin having first been established at 2.8-Å resolution (12) and then refined to 2.0-Å resolution (13). Solution studies (14) indicate that TFP can bind to the isolated light chains: measurements of the circular dichroism indicated half-maximal binding of TFP with a dissociation constant in the range of 14 -50 M, the binding of which resulted in a significant change in secondary structure consistent with an increase in ␣-helical content, whereas EPR spectroscopy detected binding at sites of lower affinity with half-maximal effects yielding dissociation constants in the range of 370 -800 M.
The removal of R-LCs from scallop myosin through chelation of metal ions is a well known phenomenon (15,16), complete dissociation being achieved in a reversible manner at elevated temperatures (17,18). However, the same treatment has proven ineffective when applied to smooth muscle and other myosins (11). Recently, TFP was shown to facilitate both R-LC exchange and R-LC dissociation from smooth muscle myosin (19 -21). It was therefore of interest to see whether or not TFP would prove effective at R-LC removal from scallop myosin. Here, we demonstrate that TFP in the millimolar range can remove R-LCs from scallop myosin. However, unlike the ensuing loss of regulation that accompanies metal ion chelation, brought about through an elevation of actin-activated MgAT-Pase in the absence of calcium (17), TFP binding results in a monotonous decline in actin-activated MgATPase both in the presence and absence of calcium. Furthermore, we demonstrate that the loss of regulation brought about by TFP occurs at concentrations an order of magnitude lower than those required to effect R-LC dissociation. These results are discussed in the light of current structural knowledge: we formulate a hypothesis suggesting that TFP may act by interfering with the conformational relay mechanism operating through the interface between the C-terminal lobe of E-LC and regions of the heavy chain, thereby fixing the "off-state" of scallop myosin.

EXPERIMENTAL PROCEDURES
Protein Preparation and EDTA Desensitization-Myofibrils and myosin from scallop striated adductor muscles were prepared as described earlier (22). Desensitization of scallop myofibrils and myosin through metal ion chelation was accomplished by standard procedures (17,23). Ca⅐Mg⅐S-1 was prepared by papain digestion (24), except that the reaction was terminated by the combined addition of N ␣ -p-tosyl-L-lysine chloromethyl ketone (to 10 mM from a 0.5 M stock in 50% ethanol) and leupeptin (to 10 mg/liter from a 0.5 mg/ml stock in buffer), so as to avoid covalent modification of S-1 as compared with the original procedure.
Rabbit F-actin was prepared according to standard protocols (25) from actin acetone powder.
TFP-induced Light-chain Dissociation-Dissociation experiments involving TFP were performed as follows. Myosin (15 mg/ml in a total volume of 60 l) in either high (0.5 M KCl, 20 mM phosphate (pH 7.0), 2 mM MgCl 2 , 3.0 mM NaN 3 , and 10 M CaCl 2 ) or low (0.1 M KCl, 20 mM phosphate (pH 7.0), 2 mM MgCl 2 , 3.0 mM NaN 3 , and 10 M CaCl 2 ) ionic strength buffer was stirred gently for 30 min on ice with TFP added to various final concentrations from a 20 mM stock solution that had been made up in the same buffer. Upon completion of the reaction, 1.4 ml of water were added to each tube, and the material was allowed to stir on ice for 10 min, ensuring myosin filament formation, prior to centrifugation for 15 min at 13,000 rpm and 4°C and lyophilization of the supernatants. Pellets were washed with water (15 min at 4°C) prior to centrifugation, lyophilization, and electrophoretic analysis. Dissociation experiments involving the addition of TFP to scallop myofibrils were similar, except that these were always performed at low ionic strength (40 mM KCl, 20 mM phosphate (pH 7.0), 2 mM MgCl 2 , 3.0 mM NaN 3 , and 10 M CaCl 2 ). TFP was a generous gift from SmithKline Beecham.
ATPase Measurements-Actin-activated MgATPase measurements were performed either on a pH-stat (Radiometer America, Inc.) or through colorimetric analysis, both procedures being described previously (17,26).
pH-stat measurements of actin-activated MgATPase of myosin were performed in the presence of 1.0 mM ATP, 2 mM MgCl 2 , and either 0.1 mM EGTA or 0.1 mM EGTA plus 0.2 mM CaCl 2 (17); F-actin was present at a 10ϫ molar ratio to myosin. When measuring rates in the presence of calcium, TFP was added from a 50 mM stock (dissolved in 10 mM phosphate, 1.0 mM MgCl 2 , and 40 mM NaCl (pH 7.0)) to 5.0-or 10-ml aliquots of myosin (0.1 mg/ml) using a Hamilton syringe. Measurements of rates in the presence of TFP but in the absence of calcium were performed in a similar manner, except that aliquots of myosin (1.0 mg/ml) were treated with TFP released from a freshly prepared 0.5 M stock solution using a Hamilton syringe. In the case of colorimetric measurements, MgATPase rates were measured in separate solutions containing 20 mM NaCl, 20 mM Tris-HCl, 3.0 mM MgCl 2 , 2.0 mM ATP, and 0.2 mM EGTA (pH 7.5) with (0.1 mg/ml myosin) or without (1.0 mg/ml myosin) the addition of CaCl 2 to 0.25 mM (26). For these determinations, the required concentrations of TFP were incorporated into the separate buffers from a 20 mM stock solution.
Measurements of the actin-activated MgATPase rates of intact or desensitized myofibrils were made in a similar way to the colorimetric determination of myosin rates, except that no additional actin was added. The buffer was 20 mM NaCl, 20 mM Tris-HCl, 3.0 mM MgCl 2 , 2.0 mM ATP, and 0.2 mM EGTA (pH 7.5) with or without the addition of CaCl 2 to 0.25 mM. Actin-activated Ca⅐Mg⅐S-1 MgATPase rates (Ϯcalcium) were measured in a buffer similar to that described for the colorimetric determination of myosin rates, except that F-actin was present at a 20ϫ molar ratio to S-1, the latter being present at a concentration of 50 g/ml. The intrinsic "K ϩ EDTA"-ATPase activity of myosin, Ca⅐Mg⅐S-1, and intact or desensitized myofibrils was monitored in the presence 2 mM EDTA, 0.5 M KCl, 1 mM ATP, and 20 mM Tris (pH 7.5) (17).
Calcium Binding-Calcium binding was performed using a 45 Ca-EGTA buffering system as described previously (17), except that the pellet volume was estimated from the pellet weight (27) rather than by a double isotope method and corrected accordingly. Briefly, 10 -15-mg aliquots of myosin or myofibrils were washed twice in radioactive buffers comprising 40 mM KCl, 30 mM MOPS, 15 mM ␤-mercaptoethanol, 2.0 mM MgCl 2 , 0.6 mM EGTA, and 0.5 mM CaCl 2 (pH 7.0) at 4°C prior to dissolution of the weighed pellets in 0.5 ml of 0.1 M NaOH, followed by scintillation counting. Calcium binding to S-1 was determined by equilibrium microdialysis, the protein being assayed in 60-l aliquots at a Circular Dichroism-To determine if TFP directly elicited a change in the secondary structure of E-LC, far-UV CD spectra were obtained using a Jobin-Yvon JY-CD6 spectropolarimeter. E-LC (19 M), in the presence or absence of 200 M TFP, was incubated in 30 mM phosphate buffer (pH 7.0) at room temperature for 15-30 min. CD spectra were obtained at 20°C using the following instrumental settings: range, 250 -185 nm; scan rate, 5 nm/min; time constant, 2 s; and wavelength increment, 0.1 nm. Spectra of the buffer alone (ϮTFP), taken under identical conditions, were first subtracted from the corresponding protein spectra prior to obtaining a difference spectrum by subtracting the corrected E-LC spectrum (ϪTFP) from the corrected E-LC spectrum (ϩTFP). In addition, induction of the ␣-helix was monitored as a function of increasing TFP concentrations (0, 40, 50, 80, 100, 120, 150, 160, 200, 250, and 300 M) at 208 and 222 nm using shorter scans (225-205 nm) run under identical conditions. Reproducible results were obtained from measurements taken using two separate preparations of E-LC purified by electroelution.
Protein concentrations were determined either spectrophotometrically (for myosin, E 1 cm 1% ϭ 5.3 (22); for S-1, E 1 cm 1% ϭ 8.0 (24)) or by the Bio-Rad protein assay (28). Free calcium concentrations were computed using an iterative procedure (17). (25 M) for 30 min at 0°C in high ionic strength buffer with increasing concentrations of TFP (up to 3.3 mM) resulted in the loss of R-LCs, as monitored by urea gel analysis of the resulting pellets and supernatants ( Fig. 1, a and b), performed as described under "Experimental Procedures." Densitometry of lanes from urea gels on which pellets from TFP-treated myosin had been loaded indicated that R-LC dissociation was Ͼ60% complete following 2.0 mM TFP treatment (TFP/myosin molar ratio ϭ 80) and complete at 3.0 mM TFP (Fig. 1c). No E-LC dissociation was observed during this process (Fig. 1b). Similar results were obtained upon exposure of myosin to TFP at low ionic strength (Fig. 1c). EDTA-desensitized scallop myofibrils, from which all R-LCs had been removed, were also treated with increasing concentrations of TFP (data not shown), yet only traces of E-LC were displaced (Ͻ5%), the vast majority remaining tightly associated with the heavy chain. Results were invariant of the presence or absence of calcium. Taken together, these results indicate that millimolar concentrations of TFP are effective in the specific removal of R-LCs from scallop adductor muscle myosin, as has been found for smooth muscle myosin (19 -21).

Effect of Trifluoperazine on Regulatory Light-chain Dissociation from Scallop Myosin-Incubation of scallop myosin
Effect of Trifluoperazine on the Rate of ATP Hydrolysis by Scallop Myosin-Increasing concentrations of TFP gave rise to a decline in actin-activated MgATPase of scallop myosin both in the presence and absence of calcium (Fig. 2). Actin-activated MgATPase rates were halved by the addition of ϳ0.1 mM TFP ( Fig. 2), whereas R-LC removal was half complete at ϳ1.5 mM TFP (Fig. 1). It may be noted that the relationship of TFP inactivation to R-LC removal is completely different from the situation resulting from EDTA treatment of scallop myosin. Whereas EDTA treatment leads to a decline in actin-activated MgATPase in the presence of calcium and a rise in actinactivated MgATPase in the absence of calcium (17), TFP treatment brings about a decline in actin-activated ATPase rates irrespective of calcium levels. Furthermore, EDTA-induced dissociation of R-LCs is causative with respect to desensitization (15,17), whereas the effect of TFP on actin-activated MgAT-Pase is independent of and precedes any effect on R-LC loss (compare results shown in Figs. 1-3).
A decline in the actin-activated MgATPase rate of scallop myofibrils in the presence or absence of calcium could also be demonstrated (Fig. 3). Significantly, increasing TFP concentrations also resulted in the attenuation of the actin-activated MgATPase activity of EDTA-desensitized scallop myofibrils (Fig. 3), demonstrating that the effect of TFP action on ATP hydrolysis does not require the presence of R-LC for its mediation.
Reversibility of the effect of TFP on ATPase activity was demonstrated through exhaustive dialysis of scallop myofibrils treated with 200 M TFP (Table I), a concentration that leads to a significant reduction of actin-activated MgATPase, yet does not displace R-LC (Fig. 1). Measurement of the actin-activated MgATPase of these TFP-treated myofibrils, before and after dialysis, demonstrated reversibility; rates of activity were restored to intact levels both in the presence and absence of calcium (Table I). By contrast, exhaustive dialysis of myofibrils treated with either 2 or 4 mM TFP could not restore any activity even when incubated with a 12ϫ molar ratio of R-LC to myosin throughout the dialysis period (data not shown).
Increasing concentrations of TFP gave rise to a gradual decline in the intrinsic rate of K ϩ EDTA-ATP hydrolysis, an indicator of active-site functionality, as seen using preparations of myosin and intact or EDTA-desensitized myofibrils (Fig. 4). These rates were diminished by ϳ70% at 250 M TFP, a concentration where inhibition of actin-activated MgATPase was complete (compare results shown in Figs. 2 and 4).
A limited number of determinations were made to ascertain the effect of increasing concentrations of TFP on the actin-  (Table II). These data demonstrate that increasing TFP concentrations led to a decline in the actin-activated MgATPase activity of myosin S-1. However, the effect of TFP was not as pronounced as that observed with myosin; whereas a concentration of 250 M TFP resulted in complete elimination of myosin actin-activated MgATPase (Fig. 2), at least 30% activity remained in the case of S-1 (Table II).
Effect of Trifluoperazine on Calcium-specific Binding to Scallop Myosin and Ca⅐Mg⅐S-1-Calcium-specific binding (2.1 M free calcium in the presence of 1 mM MgCl 2 ) (17) was performed on scallop myosin over a range of TFP concentrations (Fig. 5). At 200 M TFP, a concentration where the elimination of actinactivated MgATPase was Ͼ80% complete (Fig. 2) yet R-LC losses remained minimal (Fig. 1), ϳ70% of the normal calciumspecific binding capacity was retained following TFP treatment (Fig. 5). Full restoration of normal calcium binding values was observed (data not shown) in sister aliquots treated with 200 M TFP followed by overnight dialysis to remove TFP and to facilitate R-LC rebinding.
Treatment of scallop Ca⅐Mg⅐S-1 with TFP did not perturb calcium binding to the calcium-specific binding site (Fig. 5), which remains intact on this proteolytic fragment despite cleavage of 11 residues from the N terminus of R-LC (12,24). Under conditions of 2.1 M free calcium in the presence of 1 mM free magnesium, calcium bound to S-1 at a ratio of ϳ1 mol/mol, and this level of binding remained unchanged throughout TFP titration (Fig. 5), even when the range was extended to include 1.5 mM TFP (data not shown).
Effect of Trifluoperazine on the Circular Dichroic Spectra of the Scallop Essential Light Chain-Far-UV CD spectra were obtained to determine if TFP could elicit, directly, a change in the secondary structure of scallop E-LC. A difference spectrum (Fig. 6) obtained through subtraction of the corrected E-LC spectrum (ϪTFP) from the corrected E-LC spectrum (ϩTFP) indicates the induction of the ␣-helix by the addition of 200 M TFP. This change was also monitored (data not shown) at the characteristic wavelengths associated with far-UV CD ␣-helical extrema, 208 and 222 nm, as a function of increasing TFP concentration and shown to be maximal following the addition of 100 M TFP. DISCUSSION The selective and total dissociation of R-LC from scallop myosin by TFP treatment (Fig. 1) is reminiscent of R-LC removal from scallop myosin as a consequence of 10 mM EDTA treatment at elevated temperatures (17). Furthermore, increasing concentrations of TFP gave rise to a decline in actinactivated MgATPase in the presence of calcium (Figs. 2 and 3), superficially also similar to the effect of R-LC loss through metal ion chelation (17). However, the effects of TFP and EDTA on scallop myosin are not the same and differ from each other in several ways, as described below.
Of fundamental importance are our observations of the effect of TFP on actin-activated MgATPase in the absence of calcium: the inhibited state of intact scallop myosin. In the case of R-LC removal through divalent cation chelation, a biphasic rise in this ATPase activity as a function of R-LC loss was observed (17,18), providing insight into the cooperative nature of regulatory myosins. By contrast, TFP treatment of scallop myosin or myofibrils led to a monotonic decline in actin-activated MgATPase (ϪCa 2ϩ ) (Figs. 2 and 3). Furthermore, the sensitivity of actin-activated MgATPase activity to TFP took place over Whereas calcium-specific binding to Ca⅐Mg⅐S-1 remained unaffected by TFP treatment over the concentration range examined, Ͻ60% calciumspecific binding remained at 500 M TFP in the case of scallop myosin. Calcium-specific binding to scallop myosin was ϳ75% of intact values at 150 M TFP, a concentration where actin-activated MgATPase was inactivated by Ͼ50% (Fig. 2). Details are given under "Experimental Procedures." a concentration range an order of magnitude lower than that required to elicit R-LC removal (compare Figs. 1c and 2). In both the presence and absence of calcium, 50% inactivation of actin-activated myosin MgATPase occurred at 100 -150 M TFP, and full inactivation was achieved at 300 M TFP (Fig. 2), whereas R-LC removal began only at TFP concentrations Ͼ500 M, with full dissociation being achieved at 3.0 mM TFP (Fig. 1). These effects were reversible at 200 M TFP (Table I).
TFP-induced removal of R-LC was specific, with no E-LC being lost during treatment (Fig. 1). When scallop R-LC was removed totally by EDTA treatment prior to TFP addition, E-LC removal remained refractory to TFP treatment, although some small losses (Ͻ5%) did occur (data not shown). Such losses are consistent with data suggesting that prior R-LC removal destabilizes E-LC attachment to the myosin heavy chain (21,29). For smooth muscle myosin, a combined treatment of 5-10 mM TFP and 4.5 M ammonium chloride facilitates E-LC exchange and even complete removal of both R-LCs and E-LCs (21,30). E-LC exchange, in the absence of R-LC removal, has been shown to occur through the action of 1 mM TFP on permeabilized smooth muscle cells (31).
It is significant that the incremental addition of TFP to EDTA-desensitized myofibrils was just as effective at attenuating actin-activated MgATPase as the addition of TFP to intact myofibrils (Fig. 3). TFP also inactivated actin-activated S-1 MgATPase (Table II), analogous to its effect on myosin (Fig.  2) and myofibrils (Fig. 3). Whereas complete removal of R-LCs by EDTA treatment does not impair the K ϩ EDTA-ATPase of scallop myosin (17), TFP treatment does cause a significant decline in this indicator of active-site function, with Ͻ35% activity remaining at 250 M TFP (Fig. 4), a concentration where inhibition of actin-activated MgATPase was complete. It is not clear as to why these profound effects of TFP on activity and regulation were not apparent in earlier studies (32) measuring tension development in chemically skinned scallop muscle fibers in the presence of TFP.
Although calcium binding was impaired somewhat by TFP treatment (Fig. 5), Ͼ75% calcium-specific binding remained at 150 M TFP, a concentration where actin-activated MgATPase was Ͼ50% inactivated and where no loss of R-LC had occurred. Although this relationship between calcium binding and actin activation is expected for heavy meromyosin, myosin should exhibit nearly 100% activity at this level of calcium binding (26); that it does not do so is indicative of an inhibitory effect by TFP above and beyond its effect on calcium binding. At 500 M TFP, only 40 -50% calcium-specific binding remained in the case of scallop myosin, whereas calcium-specific binding to Ca⅐Mg⅐S-1 remained unimpaired (Fig. 5). This result emphasizes the functional dislocation of regulation from activity in scallop Ca⅐Mg⅐S-1, a molecule in which regulatory capability is completely impaired (15,33). Scallop Ca⅐Mg⅐S-1 is also known to be more refractory to EDTA-induced dissociation of R-LC as compared with the two-headed myosin structure (24).
From the above results, it is clear that the effects of TFP binding are manifest over two distinct concentration ranges, indicative of at least two different sets of binding sites. Whereas 300 M TFP is required to eliminate actin activation and regulation (50% inactivation at 100 -150 M TFP), 3000 M TFP is required to remove R-LCs (50% inactivation at 1300 -1600 M TFP). Spectroscopic studies on TFP binding to skeletal muscle myosin light chains in solution have demonstrated the presence of two types of TFP-binding site, differing in affinity by at least an order of magnitude. Measurements of the far-UV CD spectra indicated that TFP elicited increases in helical content with half-maximal effects at ϳ20 -50 M TFP for LC1 and LC3 and ϳ14 M for LC2 (14). By contrast, paramagnetic spin probes attached to the same light chains exhibited a change in rotational correlation time upon TFP addition, the effect being half-maximal at ϳ370 -809 M TFP, an order of magnitude higher than the CD results (14). Here, through far-UV CD analysis (Fig. 6), we have confirmed that micromolar concentrations of TFP have a direct effect in inducing an increase in ␣-helicity in isolated scallop E-LC; the effect is maximal by ϳ100 M. The higher affinities of the two sets of TFP-dependent transitions determined from spectroscopic data on isolated light chains as compared with the two ranges of TFP concentration required for the two different functional results obtained here on scallop myosin may reflect the fact that functional effects can be assessed on only the multichain structure.
TFP binding to the hydrophobic pocket within the C-terminal lobe of E-LC would appear to be the most likely cause of ATPase inactivation. There is a monotonic decline in actinactivated MgATPase (ϮCa 2ϩ ) that is complete at 300 M TFP, and prior removal of R-LC by EDTA still permits TFP inactivation of the desensitized myosin and exhibits the same concentration dependence (Fig. 3), suggesting that the heavy chain⅐E-LC complex can be induced to attain this extreme  -1 (d). a, location of the TFP-binding site within the C-terminal half of bovine brain calmodulin (Ser 81 -Ala 147 ) (4); b, location of the putative TFP-binding site within the C-terminal half of scallop E-LC (Phe 84 -Gly 152 ) as found within the regulatory domain structure (12), as seen from the same perspective as for calmodulin (a); c, relationship of the putative TFP-binding site within E-LC as found within the smooth muscle myosin motor domain⅐E-LC ADP⅐AlF 4 Ϫ complex (35), illustrating its proximity to heavy-chain amino acid residues at the entrance to the nucleotide-binding pocket; d, relationship of the putative TFP-binding site within E-LC as found within scallop adductor myosin MgADP⅐S-1 (36), illustrating its proximity to heavy-chain amino acid residues within the SH3 domain. In c and d, perspectives were chosen to illustrate the proximity of the E-LC C-terminal TFP-binding lobe to key heavy-chain surface features; the view of the converter region, which maintains its relationship to E-LC in both structures (35), is approximately the same in both cases. Diagrams were drawn using RasMol (Version 2.4); coordinates were downloaded from the Protein Data Bank, Research Collaboratory for Structural Bioinformatics, Rutgers University. Codes used were as follows: 1ctr (TFP⅐calmodulin complex) (4), 1scm (scallop E-LC from the regulatory domain complex) (12), 1br1 (chicken smooth muscle myosin motor domain⅐E-LC complex in the presence of AlF 4 Ϫ (MD E-LC) (35), and 1b7t (scallop adductor myosin S-1 in the presence of MgADP) (36). Illustrations are rendered as wire-frame structures on which the ␣-carbon backbone has been highlighted. Residues shaded green indicate amino acids known (a) or inferred (b-d) to be involved in TFP binding. Equivalent residues were determined through CLUSTAL analysis of the amino acid sequences of calmodulin (50), scallop E-LC (27), and chicken gizzard E-LC inhibited state. Also, TFP can induce, directly, a conformational change in isolated E-LC (Fig. 6). In the stoichiometric TFP⅐calmodulin complex, the hydrophobic tricyclic ring of TFP localizes to a hydrophobic pocket within the C-terminal lobe of calmodulin, entrapping TFP⅐calmodulin in the same conformation as that seen when calmodulin binds to a target peptide substrate (4,6,7). Although the C-terminal lobes of scallop R-LC and E-LC exhibit semi-open conformations, in contrast to the C-terminal lobe of calmodulin, which displays a fully open conformation (10), the required hydrophobic pocket remains available, albeit less deeply placed (Fig. 7, a and b). Of the 12 residues directly involved in TFP binding within the C-terminal lobe of calmodulin (4), 7 are identical to sequences found within either Argopecten irradians (27) or Pecten maximus (34) E-LC, and another two represent conservative replacements; furthermore, the spatial configuration of these residues appears to be conserved (Fig. 7, a and b). This places TFP within the hydrophobic pocket located between the two domains of the C-terminal lobe of E-LC (Fig. 7b).
Currently, crystal structures of myosin heads from two regulatory myosins are available: the chicken smooth muscle myosin motor domain⅐E-LC complex, crystallized in the presence of MgADP⅐AlF 4 Ϫ or MgADP⅐BeF x to 3.5-and 3.6-Å resolution, respectively (35); and scallop (A. irradians) S-1, crystallized in the presence of MgADP to 2.5-Å resolution (36). These are very different structures (Fig. 7, c and d). With respect to the chicken smooth muscle myosin structure (35), the G-helix within the TFP-binding lobe of E-LC abuts the motor domain close to the 25/50-kDa loop (Ala 198 -Lys 204 ) (Fig. 7c), a region of the heavy chain implicated as a major determinant in the rate of ADP release (37)(38)(39)(40)(41). Additionally, other surface loop structures (Met 140 -His 147 , Ser 163 -Asp 170 , and Phe 256 -Tyr 261 in the chicken smooth muscle myosin structure (35)) are capable of interaction with this TFP-binding lobe of E-LC. By contrast, within the scallop MgADP⅐S-1 structure (36), the TFP-binding lobe of E-LC abuts the motor domain close to the N terminus (especially Ala 45 -Lys 60 ) (Fig. 7d) within the so-called SH3 domain, which has been suggested to limit the potential swing of the lever arm (35). This region of contact is close to the heavychain region (Glu 21 -Asn 29 ) found in the nucleotide-free skeletal myosin structure (42), where it abuts the F-helix within the C-terminal domain of E-LC (as described (35)). Although crystallized in the presence of MgADP, the point is well made (36) that scallop MgADP⅐S-1 may be considered to be in an "ATPlike" or pre-power stroke state; consequently, this structure is classified as being in the weak binding state. Here, too, the F-helix of E-LC makes contact with heavy-chain residues, located within an ␣-helix close to the N terminus (Fig. 7d) (36). If the conformational off-state locked by TFP corresponds to either of the above structures (Fig. 7, c and d), it can be seen that TFP will bind at a key location along alternative sequential conformational relays that carry information from the triggering site within the N-terminal domain of E-LC, via the Cterminal lobe of E-LC, to the ATP-binding pocket. We speculate that the interface between the 25/50-kDa loop and E-LC is most likely to participate in the off-state (as seen in Fig. 7c) because the 25/50-kDa loop is known to be a major determinant of the rate of ADP release (37)(38)(39)(40)(41), and the rate-limiting state in the absence of calcium is likely to be a weak binding ADP⅐P i intermediate (45,46), not an ATP-like structure. Further work is needed to test this hypothesis through both structural and functional means. However, it is unlikely that either structure, alone, can account for all features of regulation, two heads being required (26,43,44).
Our results may also suggest that different mechanisms for the control of the off-state exist in regulatory myosins controlled either by direct calcium binding or through phosphorylation. Although the triggers are very different (calcium binds to scallop myosin E-LC (12,47) whereas phosphorylation occurs on smooth muscle myosin R-LC (48)), conventional wisdom has been that the conformational sequelae of the two triggering mechanisms rapidly converge toward a common pathway in these conserved structures to facilitate full activation at the active site. However, smooth muscle heavy meromyosin lacking E-LC (LC17), produced through overexpression of recombinant baculovirus, exhibited a 75% reduction in the rate of displacement of actin filaments, yet this movement remained phosphorylation-dependent (49). Furthermore, smooth muscle myosin lacking LC17, produced through R-LC (LC20) readdition to light chain-depleted myosin, had reduced rates of superprecipitation and actin-activated MgATPase, yet these activities remained phosphorylation-dependent (21). If, as our results suggest, E-LC is vital for maintenance of the off-state, it is unlikely that regulated activity could remain in the absence of E-LC. Consequently, in addition to distinct triggering mechanisms, these two forms of myosin-linked regulation may also exhibit different conformational relays between their respective triggering sites and the active site.
In summary, we have shown that TFP has a dual effect on scallop myosin. Application of millimolar concentrations of TFP leads to dissociation of R-LC from scallop myosin. However, concentrations of TFP an order of magnitude lower than those required to displace R-LCs can eliminate the actin-activated MgATPase activity of scallop myosin, irrespective of the presence and absence of calcium. This occurs in a manner that is independent of the presence or absence of R-LC. Furthermore, TFP can directly affect the conformation of E-LC; comparison with calmodulin indicates that the putative binding site is conserved. The exact manner by which this inhibited state is maintained remains to be determined, but we hypothesize that this structure closely mimics the off-state of a regulatory myosin, which has, so far, eluded structural analysis.