Structure and function of the 10 S conformation of smooth muscle myosin.

Smooth myosin regulatory light chain (RLC) was exchanged with RLC labeled with benzophenone-4-iodoacetamide at Cys-108. Irradiation under conditions that favor the folded (10 S) conformation resulted in 10 S cross-linked myosin that could not unfold. Purified 10 S cross-linked myosin was cross-linked between the RLC of one head to light meromyosin between leucine 1554 and glutamate 1583, adjacent to a predicted noncoiled region, approximately 60 nm from the tip of the tail. At high ionic strength without actin, product release from one-half of the heads was slow (like 10 S) whereas the other half were activated. This suggests that tail binding to the RLC carboxyl-terminal domain stabilizes ionic interactions important to slow nucleotide release. With actin, product release from both (un)phosphorylated 10 S cross-linked myosin was from one slow population similar to unphosphorylated filaments. 10 S cross-linked myosin weakly bound actin (dissociation constant > 500 μM) and did not move actin in vitro. Single-headed myosin did not fold or trap nucleotide. These and other data suggest that “trapping” occurs only with both heads and the tail binds to a newly formed site, which includes the RLC carboxyl-terminal domain, once trapping has occurred.

Regulation of actomyosin Mg 2ϩ -ATPases can be myosinlinked through phosphorylation of myosin regulatory light chains (1) or through Ca 2ϩ binding to myosin (2) or actin-linked through Ca 2ϩ binding to the thin filament proteins (3,4). Myosin filaments exhibiting myosin-linked regulation, i.e. from vertebrate smooth muscle (5), striated scallop (6), and vertebrate cytoplasmic systems (7)(8)(9), are unstable under physiological relaxing conditions in the presence of ATP. The filaments are in equilibrium with a soluble monomer in which the tail bends back upon the heads giving rise to a hairpin folded conformation commonly known as folded or 10 S (10). A 6 S monomer with an extended tail is favored at high ionic strengths whereas the 10 S conformation is destabilized. If regulation is actin-linked, as in vertebrate skeletal muscle, myosin filaments are stable in the presence of ATP. The association of the 10 S conformation with myosin-linked regulation suggests that it may be important in the regulatory process. However, unphosphorylated myosin in resting smooth muscles is filamentous (11), and only trace amounts of 10 S myosin is present in both relaxed and contracted states (12), suggesting that additional proteins are required for the stabilization of myosin filaments in the presence of ATP (13,14). Although the physiological role of the 10 S conformation remains to be clarified, it is possible that 10 S molecules are important in the control of filament assembly in developing cells or other cells with changing requirements for contractility (12,15).
The tail interacts with components of the head-tail junction in 10 S myosin. The tail interaction site appears to be at a sharp bend in the tail approximately 100 nm from the head-tail junction (5) which also allows LMM 1 to interact with S2 (5,16). The regulatory light chain is an essential component of the head-tail junction binding site for the tail because light chaindeficient myosin does not fold (17). Experiments with hybrid myosins (17) suggest that there are two different binding sites on the RLC for the tail; one requires residues 1-22 of the RLC and is not present in the skeletal RLC. Deletion mutation studies have shown that Lys-11 through Arg-16 on the positively charged NH 2 -terminal region of the RLC play an important role in the stabilization of the 10 S conformer (18). Folding of the tail inhibits proteolytic cleavage of the heavy chain at the S1-S2 junction (19) and at a site 68 kDa from the NH 2 terminus (20) known to be involved with actin binding.
The structural and functional aspects of the 10 S conformation are interesting for three main reasons. First, the intramolecular interactions in the 10 S monomer may mimic intermolecular interactions in the supramolecular structure of the inactive filament. It is known from microscopy studies that 10 S myosin heads tend to be bent down toward the NH 2 -terminal region of the rod (S2) (5). The tendency toward this bent head attitude was also observed for the heavy meromyosin (HMM) subfragment of myosin that lacks LMM, the COOH-terminal two-thirds of the tail domain (21). The attitude of the heads of 10 S myosin may be correlated to the appearance of relaxed filaments of various regulated myosins (22,23). Second, when ATP is added to filaments to form the 10 S myosin, 1 mol of ADP-P i is "trapped" at each active site (24) to give a molecule with very low ATPase activity that may contain some of the elements of the "off" state of unphosphorylated filaments (25). Third, phosphorylation of the regulatory light chain, which is the regulatory "on" switch for smooth muscle myosins, destabilizes the 10 S conformation and favors filament assembly. A similar coupling between the ability to form a folded 10 S state and myosin-linked regulation exist in molluscan myosin where * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Our interest was to further characterize the interaction of the tail with the head-tail junction in 10 S myosin. A photoactivated probe attached to Cys-108 of the RLC cross-linked to the LMM portion of the tail, effectively "locking" the molecule in the 10 S conformation. In this molecule, Cys-108 of one RLC is cross-linked to a region between Leu-1554 and Glu-1583 of the LMM portion of the tail. The cross-link site on the tail was localized to a region between Leu-1554 and Glu-1583. The relationship between the tail cross-link site and putative weakly coiled-coiled regions of the rod were explored using a sequence analysis approach. A single-turnover kinetic analysis of the 10 S cross-linked myosin showed that the cross-linked head cannot be activated by increasing ionic strength, whereas the uncross-linked head behaved similarly to a control. The 10 S myosin had a very low steady-state actin-activated ATPase activity, and single-turnover experiments suggested that neither head can be activated by actin. 10 S cross-linked myosin did not move actin filaments in an in vitro motility assay and bound weakly to actin. By combining this information with a study of folding in single-headed myosin and the work of others, conclusions were drawn about the specific interactions of the tail with RLC and its relationship to the trapped state. (26) from frozen chicken gizzards (Pel-Freeze) was stored on ice, and experiments were completed within 2 weeks of purification. Myosin concentrations were determined at 0.5 M KCl (6 S monomer) using ⑀ 280 1% ϭ 5.66. Myosins were thiophosphorylated as described previously (27) with myosin light chain kinase (28) and ATP␥S (Boehringer Mannheim) for 1-2 h at room temperature or overnight on ice. After dialysis, the extent of thiophosphorylation was determined from the incorporation of radiolabel from [␥-35 S]ATP. All such indicated samples in the figures were 95-100% thiophosphorylated. Urea gels were used to confirm the extent of thiophosphorylation (29). RLC was isolated from purified smooth muscle myosin as described previously (29) and was stored lyophilized at Ϫ80°C. Concentrations of RLC and labeled RLC was determined by ⑀ 277 1% ϭ 3.37 or by the bicinchoninic acid method (30), respectively. 5,5Ј-Dithiobis(2-nitrobenzoic acid) reagent was used to determine sulfhydryl concentrations (31). Rabbit skeletal actin (32) was stored in the F form on ice and used within 3 weeks of purification. Chicken gizzard tropomyosin (33) was lyophilized and stored at Ϫ80°C.

Synthesis of [ 3 H]Benzophenone-4-iodoacetamide-
The synthesis of this compound has not been reported previously. The reported syntheses of similar compounds (34) are not appropriate for use with readily available and easily handled radioactive starting materials. Here we report an easy synthesis designed for safe handling of radiochemicals. Standard technique was used to carry out the following reactions under an argon atmosphere in the container supplied by DuPont NEN equipped with a stir bar. Room light was avoided. Solid iodoacetic acid (IAA; Sigma; 0.75 eq) was added to 0.25 eq of [ 3 H]IAA (DuPont NEN; 1 mCi at approximately 140 mCi/mmol) to give 1 eq of [ 3 H]IAA (28.5 mol) at a specific activity of approximately 50,000 cpm/nmol. Freshly distilled acetonitrile (ACN; 33 l) was added to dissolve the IAA before 1.25 eq of dicyclohexylcarbodiimide (Sigma) were added (33 l in ACN). The activated ester was allowed to form at 0°C for 10 min before 4 eq of N-hydroxysuccinimide (44 l in ACN) were added. After reacting a further 45 min at 0°C, 1 eq of 4-aminobenzophenone (Sigma) was added dropwise (44 l in ACN) over approximately 5 min. The reaction was then brought slowly, over 1 h, to 40°C. After 48 h at 40°C, residual 4-aminobenzophenone was barely detectable by TLC (100% ethyl acetate; aluminum-backed Silica Gel 60 plates with fluorescent indicator from Merck). Product formation was confirmed using benzophenone-4iodoacetamide (BPIA, Molecular Probes) as a standard (R f ϭ 0.94). The product was clearly resolved from the 4-aminobenzophenone (R f ϭ 0.89) and an unidentified minor product (R f ϭ 0.85). One equivalent of acetic acid was added to quench residual dicyclohexylcarbodiimide, and the reaction was transferred to a 1.5-ml microcentrifuge tube (Eppendorf) with additional ACN (200 l). Dicyclohexylurea was pelleted in a mi-crocentrifuge for 2 min at ambient temperature. The pellet was washed with ACN twice more, and the combined supernatants were evaporated to dryness with a stream of argon to an oil in a 15-ml centrifuge tube. The oil was dissolved in 400 l of MeOH, followed by precipitation of the product and unknown byproduct by adding 400 l of H 2 O. The supernatants obtained after centrifugation contained residual IAA (R f ϭ 0.17), N-hydroxysuccinimide, and 4-aminobenzophenone. These supernatants may be further treated with H 2 O if found to contain appreciable product after TLC analysis. Diethyl ether (500 l) was added to the combined pellets to solubilize the product, leaving the by-product in the pellet. After appropriate washing of the pellets, the combined supernatants were dried with a stream of argon, and the resulting light yellow powder was dissolved in dry distilled dimethyl formamide (Baker) and stored at Ϫ80°C. The product was obtained in 50% overall yield and was verified by comparison of the UV-visible spectrum with the standard (⑀ 298 M ϭ 23,000 M Ϫ1 cm Ϫ1 in MeOH). The final specific radioactivity calculated using this extinction coefficient was within 8% of that calculated from the specific radioactivity of the initial [ 3 H]IAA. The product was a single spot on TLC and was found to be reactive with cysteine at pH 8 (R f ϭ 0 for cysteine adduct in 100% ethyl acetate solvent system).
BPIA Labeling of RLC Cys-108 -To label the single sulfhydryl residue (Cys-108) of RLC, a solution of RLC (1.5 mg/ml) in 50 mM ammonium bicarbonate (pH 7.9), 0.1 mM EGTA, and 0.1 mM EDTA was purged with argon for 30 min on ice. DTT was added to 5 mM, and the sample was purged for an additional 5 min on ice before the reaction was sealed and allowed to reduce for 12 h on ice. After reduction, the protein was dialyzed into 50 mM ammonium bicarbonate (pH 7.9), 0.1 mM EGTA, 0.1 mM EDTA. The total sulfhydryl concentration was monitored over time using 5,5Ј-dithiobis(2-nitrobenzoic acid) (35). A 1.5-fold excess (over total sulfhydryl concentration) of BPIA or [ 3 H]BPIA (dissolved in dimethyl formamide) was added when the remaining DTT concentration was approximately equal to one-half the initial protein concentration. After stirring the reaction at room temperature for 30 -60 min in the dark, the complete loss of reactive sulfhydryls was verified using 5,5Ј-dithiobis(2-nitrobenzoic acid) reagent. Excess free probe was removed using a Sephadex G-25 fine column (Sigma). The use of [ 3 H]BPIA for RLC labeling confirmed a stoichiometry of 1 mol of BPIA to 1 mol of protein, consistent with selective labeling of the single sulfhydryl. The molar extinction coefficient of RLC labeled on Cys-108 with the benzophenone probe (BP-RLC) is ⑀ 302 M ϭ 22,500 M Ϫ1 cm Ϫ1 . Preparation and Photocross-linking of BP-myosin-Nonlabeled RLC on native myosin was replaced by exchange with BP-RLC as described previously (29). Excess free RLC were removed by dialysis into 15 mM Tris-Cl (pH 7.5), 5 mM MgCl 2 , 0.1 mM EGTA, 1 mM DTT followed by pelleting the filaments at 18,000 ϫ g for 20 min. After three additional washes, the final myosin pellet was resuspended in 10 mM sodium phosphate (pH 7.5), 1 mM MgCl 2 , 0.1 mM EGTA, 125 mM NaCl, and 1 mM ATP (10 S conditions) (27). The labeled myosin (BP-myosin) contained 65-90% labeled light chains as assayed by incorporation of [ 3 H]BPIA-labeled RLC. After the final resuspension, the myosin was allowed to equilibrate in 10 S buffer for 30 min on ice, and, immediately thereafter, residual filaments were pelleted. The 10 S myosin supernatant (0.5-8.0 mg/ml) was immediately irradiated as described previously (36) for 10 min. DTT was excluded from the buffer during irradiation because it was found to decrease the photocross-linking efficiency.
Purification of Photocross-linked Myosin-The 10 S cross-linked myosin was purified based upon its inability to form filaments. The irradiated sample was dialyzed into 15 mM Tris-Cl (pH 7.5), 5 mM MgCl 2 , 0.1 mM EGTA, 1 mM DTT to promote filament formation of the uncrosslinked myosin. After pelleting the filaments by centrifugation at 100,000 ϫ g for 10 min, the cross-linked myosin in the supernatant was tested for purity by gel filtration (see below). The purity of the resulting cross-linked species was typically Ͼ90% (see also "Results," Fig. 1B). The overall yield of cross-linked myosin was typically 1 mg per 10 mg of myosin used for light chain exchange.
HPLC Gel Filtration of Photocross-linked Myosin-Gel filtration of myosins (50 -200 g) was performed on a TSK G4000SW column (Toso-Haas, 7.5 mm ϫ 60 cm, 13-m particle size, with a 75 ϫ 7.5 mm guard column). The column was run at room temperature in 10 mM HEPES (pH 7.3), 1 mM MgCl 2 , 0.1 mM EGTA, 25 M ATP. The KCl concentration was as indicated in the figure legends. The flow (0.75 ml/min) was controlled by a Beckman HPLC, and peaks were monitored at 280 nm (0.1 absorbance unit full scale) with a Beckman 165 variable wavelength detector.
Gel Electrophoresis-Polyacrylamide gels used for scanning (BioImage Visage 60 -110) contained 12% glycerol in the separating gel (pH 9.5) (37). Western blots were performed using a rabbit anti-RLC antibody gifted to us from the laboratory of Dr. James Stull. Donkey anti-rabbit antibody labeled with horseradish peroxidase (Amersham) was used as secondary antibody. Transfer of proteins from gels to nitrocellulose and subsequent Western blot was performed according to standard procedures (38). Renaissance (DuPont NEN) chemiluminescence kit was used to visualize the antibody cross-reaction with RLC.
Isolation of Cross-linked Peptides-10 S cross-linked myosin (6.2 mg prepared from myosin exchanged with [ 3 H]BP-RLC; see "Materials and Methods") was lyophilized, brought to 1 mg/ml in 6 M guanidine hydrochloride, 25 mM HEPES (pH 7.4), 10 mM CaCl 2 and equilibrated at 35°C for 1 h to unfold the protein prior to addition of 1/50 (w/w) Staphylococcus aureus strain V8 endoproteinase Glu-C (V8 protease, Sigma). After 2 h at 35°C, the mixture was diluted to a final guanidine hydrochloride concentration of 4 M, and the treatment with protease was repeated. After a final dilution to 2 M guanidine hydrochloride and a third treatment with protease, the mixture was incubated for 14 h at 35°C. The peptide mixture was adjusted to 6 M guanidine hydrochloride, the pH was brought to 8.3 with Tris base, and the solution was flushed with argon prior to addition of 2 mM DTT. After 1 h at 35°C, another aliquot of DTT was added (4 mM final) and allowed to react for 15 min before iodoacetamide was added to 10 mM. After 1 h at 35°C, the reaction was quenched with the addition of DTT to 20 mM. Water was added to reduce the guanidine hydrochloride concentration to 4 M, and the solution was filtered through a 0.4-m filter to prepare the solution for HPLC. One-half of the peptide mixture (7 nmol) was applied by repetitive injections onto a Brownlee Aquapore C8 RP 300 HPLC column (220 ϫ 4.6 mm) equilibrated in aqueous 5 mM potassium phosphate (pH 6.9) at room temperature (solvent A). After 10 min at 100% solvent A, a linear gradient to 33% solvent B (B ϭ 65% acetonitrile, 35% H 2 O) was applied in 42 min, followed by a gradient to 55% B in 100 min. The flow rate was 1 ml min Ϫ1 , and 1-ml fractions were collected. Approximately 50 -60% of the applied radioactivity eluted from the column. Fifteen percent of the eluted radioactivity was found in the void volume and was not further analyzed. The remainder of the radioactivity eluted as a broad peak between fractions 90 and 140 (data not shown). A second identical separation was performed, and the radioactive fractions were combined with the first separation and divided into 4 pools. Each pool was separately injected onto the above column equilibrated in aqueous 0.115% trifluoroacetic acid (solvent C). After 10 min at 100% solvent C, a linear gradient to 40% D (D ϭ 0.1% trifluoroacetic acid in 65% acetonitrile) in 40 min was followed by a gradient to 100% D in 120 min at 1 ml min Ϫ1 . All radioactivity eluted between fractions 100 and 130 (between 65% and 85% D) in 6 major peaks, each of which was differentially enriched in each of the four applied pools (data not shown). It was not practical to sequence all the radioactive peak fractions from all four HPLC separations. Therefore, absorbance spectral data were collected between 215 and 350 nm (Waters 990 diode array detector), and fractions were selected that exhibited an absorbance band at 300 -310 nm (due to the irradiated photocross-linker) with minimal absorbance at 280 nm. The LMM region of the myosin tail contains only 11 aromatic residues, none of which are tryptophans, and the RLC contains no tryptophans. After spectral analysis of all four HPLC chromatograms, representative fractions from all of the six major radioactive peaks were submitted for sequencing ( Fig. 4) and represent approximately 30% of the radioactivity eluted in the four separations.
Single-turnover Assays-Single-turnover assays (39) were performed at 0.2 mg/ml myosin in 15 mM Tris-Cl (pH 7.5), 5 mM MgCl 2 , 0.1 mM EGTA, and 2 mM DTT at varying ionic strengths as indicated in the figure legends. Actin-activated single-turnover assays were performed in the same buffer with 70 mM KCl in the presence or absence of 5 M actin and 1.25 M tropomyosin. Each myosin solution was equilibrated to 25°C, and the fluorescence was set arbitrarily at 1000 before adding formycin triphosphate (FTP; synthesized as described (40)) to 4 times the myosin concentration (2 times the number of sites). The fluorescence increased as the FTP bound to myosin. Nucleotide binding was allowed to reach a maximum before ATP was added to a final concentration of 100 M. The release of FDP was measured by monitoring the decrease in fluorescence as a function of time. All assays were performed on an SLM 4800 fluorometer in a 3-ml stirred cuvette at a final volume of 1.0 ml at 25°C. The excitation wavelength was 313 nm (8 nm band pass), and the emission was detected through a 335 nm Schott low cut-off filter.
The fluorescence decay curves were analyzed by fitting each to a single or double exponential decay equation using Sigma Plot Scientific Graph System from Jandel Scientific that uses a weighted leastsquares fit. In the single exponential equation f(x) ϭ aexp Ϫbx ϩ c, a equals the total amplitude of the fluorescence decay that decreases to a final value c with a time constant of 1/b. Constraints limit b to values Ͼ0 to prevent exponential rise. Samples that were composed of more than one population were fit to the double exponential equation f(x) ϭ aexp Ϫbx ϩ cexp Ϫdx ϩ e. The total amplitude of the fluorescence decay is the sum of a ϩ c and a is the amplitude of the decay due to the relatively fast population and c is the amplitude of the decay due to the slow population. The exponential starts at a ϩ c and decreases to a final value e with increasing time (x) when b, d Ͼ 0. The rate of decrease was determined by two time constants, 1/b corresponds to the fast rate and 1/d corresponds to the slower rate. The mole fractions of the fast and slow sites were given by a and c, respectively.

Preparation and Purification of 10 S Cross-linked Myosin-
Initial studies of myosin exchanged with BP-RLC were aimed at identifying interactions of the COOH-terminal domain of RLC with the rest of the myosin molecule. Irradiation of BPmyosin in the 10 S conformation (Fig. 1, lane 3) results in a covalent linkage of the BP-RLC with the myosin heavy chain (HC) as evidenced by a new band migrating above the HC. Light chain dimers were not detected. The presence of the RLC in the band that migrates above the free HC on PAGE was confirmed by Western blot of the gel (data not shown). This band could not be detected after irradiation in the 6 S conformation (data not shown). The 10 S cross-linked myosin was isolated from the mixture based on its inability to form filaments. After irradiation, the sample was dialyzed to remove ATP, and the myosin that is competent to form filaments is removed from the sample by sedimentation. The resulting pellet was enriched in the uncross-linked HC (Fig. 1, lane 4), the supernatant (Fig. 1, lane 5) was enriched in the cross-linked species. The supernatant is also enriched in an intramolecularly cross-linked RLC visible between the RLC and the ELC. The species can be removed by gel filtration and is therefore not bound to myosin (see below). Gel filtration under conditions that favor the 6 S conformation (Fig. 1B) was used to determine the purity of the sample from lane 5 (Fig. 1A). The majority of the 10 S cross-linked myosin sample eluted at 18.75 min which is the elution time of control myosin under 10 S conditions (see Fig. 2). A small peak (Fig. 1B) corresponding to noncross-linked myosin eluted at the native 6 S elution time of 15 min (see Fig.  2). The purity of cross-linked myosin in lane 5 ( Fig. 1A) was estimated to be 92% from the areas under the two peaks (Fig.  5B). For subsequent analysis, 10 S cross-linked myosin samples were not purified by gel filtration and are therefore similar to the sample shown in lane 5 (Fig. 1A). This level of purification will be referred to as 10 S cross-linked myosin.
SDS-PAGE analysis of the major peak in Fig. 1A (lane 6) suggested that 10 S cross-linked myosin contained only one HC-RLC cross-link, whereas the second HC and RLC migrated at the free HC and free RLC positions. The relative amount of cross-linked and free HC was confirmed to be approximately 1:1 by scanning of a 12% glycerol SDS gel for a sample with 90% exchange (see "Materials and Methods," data not shown). Therefore, 10 S cross-linked myosin appears to contain a single HC-RLC cross-link; however, the presence of a small amount of myosin containing two cross-links cannot be ruled out. Assuming an exchange efficiency of 80% and a cross-linking efficiency of 5%, by calculation only 4% of the purified 10 S cross-linked could potentially contain two cross-links, assuming an equal probability of cross-linking for each head. Therefore, under our conditions, it is not surprising that we found only one head cross-linked to the tail.
Gel Filtration-The elution time of a smooth muscle myosin control from a gel filtration column (Fig. 2) decreases with increasing ionic strength as the equilibrium distribution shifts from the folded 10 S conformation to the 6 S extended form (17,41). BP-myosin (before irradiation) showed similar behavior (data not shown). In contrast, 10 S cross-linked myosin eluted near the 10 S position at all ionic strengths tested (Fig. 2) indicating that it was not capable of adopting the extended conformation (Fig. 2). Further studies are required to determine whether the small decrease in elution time with increasing ionic strength observed here may be related to changes in the attitudes of the heads as suggested by previous studies with HMM (21).
Localization of the Cross-link-Identification of the region of the HC that was cross-linked to the RLC in the 10 S crosslinked myosin was accomplished by limited proteolysis with V8 protease followed by Western blot analysis of the resulting digest (Fig. 3). The fragments produced by digestion of control myosin under the conditions used have been identified previously (19) and are labeled accordingly. The fragments produced by treatment of the 10 S cross-linked myosin with V8 protease (Fig. 3, lanes 6 and 7) were identical to those produced by treatment of control myosin (Fig. 3, lane 4) except for a single additional band visible slightly above HMM. This new band migrated at an apparent molecular mass of 130 kDa, consistent with the RLC (20 kDa) covalently linked to a heavy chain fragment of 95 kDa. The ratio of the S1/LMM bands was markedly lower in the control myosin digest (Fig. 3, lane 4) compared to the 10 S cross-linked digest (Fig. 3, lanes 6 and 7). The decrease in amount of LMM present in the 10 S crosslinked along with the appearance of the unique 130-kDa band implicates the LMM region of the HC in the cross-link to the RLC. This is supported by the lack of new bands corresponding to HMM-RLC (130 ϩ 20 kDa) (Fig. 3, lanes 5-7). The RLC did not cross-link to S2 since a new band was not apparent at the predicted 42-ϩ 20-kDa position. Since overlap of bands could mask new bands in the digest of 10 S cross-linked myosin, Western blot analysis using anti-RLC antibody was performed to visualize the RLC (Fig. 3, right side). The anti-RLC antibody cross-reacts only with the 130-kDa band that is unique to the proteolyzed 10 S cross-linked myosin (Fig. 3, lanes 10 and 11) and the free RLC. The undigested HC-RLC cross-link band cross-reacts with this antibody in the undigested 10 S crosslinked myosin as expected (lane 9) and showed a decreased intensity as the sample is proteolyzed (Fig. 3, lanes 10 and 11). A minor product that migrated below the HC (Fig. 3, lane 9) increased with the age of the myosin. These data are consistent with photocross-linking of the RLC to the LMM portion of the myosin tail.
To identify the LMM region involved in the RLC cross-link, denatured 10 S cross-linked myosin containing a [ 3 H]benzophenone-labeled RLC was exhaustively digested with V8 protease. The resulting radioactive cross-linked peptides eluted later than most other peptides in the mixture under both neutral and acidic conditions on a reversed-phase HPLC column, providing sufficient purification after only two separations (data not shown). The cross-linked peptides eluted late because they were composed of a long RLC sequence beginning at Gly-71 and including the labeled Cys-108, attached to a family of shorter LMM sequences containing residues 1554 to 1583 (Fig. 4). All of the 80 to 250 pmol (3,800 to 12,000 cpm) of tritium submitted for sequencing remained bound to the filter (after up to 24 cycles). The peptides in Fig. 4 are thought to represent the cross-linking site from three lines of reasoning: 1) sequences are within LMM as dictated by the data in Fig. 3, 2) sequences in this region were found in all samples, 3) amino acids corresponding to these sequences were found in yields consistent with the known amount of radioactivity submitted for sequencing.
In 10 S cross-linked myosin, one RLC is cross-linked to the LMM portion of the tail whereas the other RLC on the other head remains uncross-linked. To attempt to resolve populations of differentially activated heads, the rate of product release from the 10 S cross-linked myosin in the absence of actin was measured using the fluorescent ATP derivative FTP during a single turnover. The FTP turnover of unmodified myosin was slow at low ionic strength (0.15 M KCl) as expected for the 10 S conformation (Fig. 6A). The turnover reached a maximal rate about 16-fold higher at approximately 0.35 M KCl as expected for the transition to the extended 6 S monomer. These fluorescence decay curves fit well to a single-exponential model. Fit to a double-exponential model gave two rates of nearly equal magnitude and did not improve the fit. Rates of turnover by myosin exchanged with unmodified RLC were nearly identical to the control for each ionic strength tested (data not shown), which confirmed that the exchange procedure did not notably alter the myosin activity, as previously shown (29,42,43). At low ionic strength, the 10 S cross-linked sample spanning the radioactivity profile from the four acidic HPLC separations (see "Materials and Methods") were submitted for sequencing as described previously (29). The detected sequences from the LMM portion of the tail (number notation from Ref. 46) are underlined. All samples contained the parent peptide from the RLC, G 71 MMSEAPGPINFTMFLTMFGEK . . . (peptide was not sequenced beyond 22 cycles). In addition, sample a contained the RLC sequence A 36 FNMIDQN; sample b contained V 1065 QLQD LQSKYSDGERVR-TELNE from the HC; samples e and f contained the RLC sequence A 36 FNMIDQNRAGFID. Some cycles contained small quantities of amino acids that could not be attributed to the indicated peptides.

TABLE I In vitro motility assay of phosphorylated myosins
In vitro motility assays were performed as previously described (62) except the motility buffer contained 60 mM KCl, 20 mM MOPS (pH 7.2), 5 mM MgCl 2 , 50 mM DTT, 0.1 mM EGTA, 200 nM gizzard tropomyosin, 2.5 mg/ml glucose, 0.1 mg/ml glucose oxidase, 2 g/ml catalase, 0.7% methylcellulose, and the assay temperature was 30°C. All samples were phosphorylated during the motility assays on the coverslip under conditions which are known to give Ͼ95% phosphorylation of control myosin (62). 10 S cross-linked myosin was found to thiophosphorylate to a similar extent in control experiments using [␥- 35 The mean reported is the velocity of actin filaments that were consistently moving (i.e. where the ratios of the standard deviation to the mean is less than 0.3) (62). It has been shown previously that irradiation does not affect the ATPase and folding of smooth muscle myosin (64).
b Exchange condition myosin is myosin that has been exchanged with unlabeled RLC. c BP-RLC myosin was exchanged with BP-RLC but was not irradiated. d 10 S cross-linked myosin was exchanged with unlabeled RLC. The extent of RLC exchange was not quantified due to limited protein. However, typical exchange efficiency was between 65 and 90%. (Fig. 6B) behaved as a single population with a turnover rate identical to the control (Fig. 6A). As the ionic strength was increased, the best fit to the data revealed that one population retained the initial slow rate characteristic of 10 S myosin, while a second population was partially activated. This activated population reached a maximal rate approximately 12fold higher than the value at 0.15 M KCl. The population within the 10 S cross-linked sample that remained slow at all ionic strengths tested represented between 40 and 45% of the myosin (Fig. 6C, filled bars). Eight percent of the myosin was noncross-linked (see above) and thus the turnover from both of these heads should contribute to the fast population. The 10 S cross-linked myosin contains only one cross-linked head. Therefore, the data in Fig. 6, B and C, suggest that the crosslinked head retains the slow turnover rate characteristic of 10 S myosin, while the "free" head can be activated at high ionic strength to about 70% of the value of unmodified myosin.
To examine the phosphorylation-dependent regulation of 10 S cross-linked myosin, the effect of phosphorylation upon the actin-activated ATPase was studied by two methods. Steadystate activity of control myosin (Fig. 7) was regulated by phosphorylation and showed a similar dependence on actin concentration as found previously (25). In contrast, unphosphorylated and thiophosphorylated 10 S cross-linked myosin (98% pure) showed minimal actin activation of hydrolysis suggesting a very high K m or a reduced V max . Single-turnover assays (Fig. 8) were performed in an attempt to detect a regulated population of heads in the 10 S sample. Both phosphorylated (D) and unphosphorylated (C) 10 S cross-linked myosin (0.002/s) showed decay curves similar to unphosphorylated control myosin (A; 0.001/s). Cross-linked samples did not appear to contain an appreciable amplitude corresponding to the fast population that was found in thiophosphorylated control myosin (B; 0.052/s).
The movement of actin filaments by phosphorylated myosin was measured by in vitro motility assays (Table I). Movement was not detected in unphosphorylated samples (data not shown) indicating well-regulated myosin. Unmodified myosin and myosin exchanged with unmodified RLC moved actin filaments at 1.1-1.2 m/s as previously measured (44). In contrast, phosphorylated 10 S cross-linked myosin did not significantly move actin filaments. To demonstrate that cross-linked myosin was not irreversibly inactivated, a second exchange procedure, using native unlabeled RLC, was performed to test for recovery of motility. This second exchange resulted in some of the myosin possessing one unmodified RLC per head with a third RLC cross-linked to the tail but no longer bound to the head. The rate of actin motion was returned to approximately 50% of the control value. removed by dialysis) to actin was obtained by mixing 10 S cross-linked myosin (0.86 M) with actin (25 M) at 4°C in the same buffer used for ATPase assays. Radioactivity of supernatant and pellet were measured after centrifugation (265,000 ϫ g for 10 min). The minimum value of the equilibrium dissociation constant was calculated to be approximately 500 M, as no binding could be detected within 5%.
Single-headed Smooth Muscle Myosin-It was of interest to determine if single-headed smooth muscle myosin could fold to the "10 S" conformation. This point has been previously addressed (17), but the single-headed myosin used lacked an NH 2 -terminal portion of RLC that has since been shown to stabilize the 10 S state (18). Single-headed smooth muscle myosin (44) with full-length native RLC was used in the experiment described here. The ionic strength dependence of myosin conformation was monitored by gel filtration (Fig. 9). All unphosphorylated double-headed myosin was folded at or below 0.25 M KCl. In contrast, folding for unphosphorylated singleheaded myosin was observed only at the lowest ionic strength tested (0.15 M KCl). Here, only 30% of the sample was folded as the remaining 70% of the protein was filamentous and did not elute from the column (see legend, Fig. 9). Thiophosphorylated single-headed myosin did not fold at any ionic strength tested. Therefore, stable binding of the tail to the remainder of the molecule requires two heads. Under identical conditions, but at a higher MgCl 2 concentration (5 mM), the steady-state ATPase activity of both thiophosphorylated and unphosphorylated single-headed myosin was high (0.03/s) and independent of ionic strength between 0.15 M KCl and 0.4 M KCl (data not shown). This suggests that the trapped state (0.001/s; Fig. 6A) also requires two heads. DISCUSSION Benzophenone-labeled RLC (Cys-108) photocross-linked to the LMM portion of the smooth muscle myosin tail after irradiation in the 10 S conformation (Fig. 3). This is the first direct evidence that the tail interacts with the COOH-terminal domain of the RLC. Photocross-linked sequences from LMM were found to be between Leu-1554 and Glu-1583 (Fig. 4), which is about midway along the length of LMM. The EELE portion of the sequence spans the most negatively charged cluster of amino acids within a 28-residue repeat zone (45,46), as determined by the averaged charge over 38 complete zones.
Ideally in peptide sequence analysis, labeled amino acids can be identified by a low yield for the respective sequencing cycle if the data are not complicated by concurrent sequences. Conversely, labeling of certain residues may be ruled out by noting a consistently high yield. The exact amino acids covalently linked to Cys-108 were not resolved from the data used to create Fig. 4. It is interesting however that none of the methionines appeared to be labeled even though photoactivated benzophenone (47) is known to be relatively selective for methionine (48). These residues may be protected from labeling because they are in the a or d sequence positions forming the hydrophobic interface between the two ␣-helices. From yield analysis, Arg-1570 and Ser-1580 best fit the criteria for potentially labeled residues, although other residues cannot be excluded. At least three amino acids are labeled spanning a 30-residue region corresponding to approximately 4.5 nm of the tail. Multiple amino acid labeling could be explained by various mechanisms. For example, it might be consistent with a bend in the tail that interacts with both heads, where Cys-108 of the two RLC could be exposed to different regions of the tail. Alternatively, it may be due to a rapidly reversible interaction of the tail with the heads during the photolabeling event, giving the activated probe access to the tail in partially bound states.
It was of interest to merge this new information about the position of the photocross-linked site with other structural details of the myosin tail region. To assist in this process we modeled the tail structure using a pairwise residue correlation program (49) which predicts the likelihood for ␣-helical coiledcoil. The results for the chicken gizzard myosin tail (Fig. 5, top) are compared with other myosins. The photocross-linked region is shown hatched on the gizzard myosin sequence. Skip residues are labeled with arrows (see legend). For gizzard myosin, four main regions (Fig. 5, labeled A, B, C, and D) are predicted to have a lower probability of ␣-helical coiled-coil structure than the majority of the tail sequence. Region A correlates with the position of the HMM-LMM junction and to the position of a bend in the rod observed in 10 S myosin approximately 50 nm from the head-tail junction (5). Region B is a previously unidentified weakly coiled region. It was strongly predicted for all myosins regulated by phosphorylation with the exception of Drosophila nonmuscle myosin. It was not predicted for scallop myosin that is known to fold. Site C is adjacent to, but not overlapping, the photocross-linked region and appears to correlate well with the sharp bend in the tail near the head-tail junction (5,16) in 10 S smooth muscle myosin. Measurements of shadowed 10 S molecules place this bend 60 nm from the COOH terminus of the 158 nm tail (5). Arg-1570 (see Fig. 4) is 60.6 nm from the COOH terminus of the tail (Glu-1978) assuming a helical rise of 0.1485 nm/residue for an ␣-helical coiledcoil (45) including the "nonhelical" tailpiece ( Fig. 5, site D). Therefore, both the Paircoil analysis and the electron microscopy evidence are consistent with a bending in the myosin tail near the area of the photocross-linking site. Vertebrate skeletal myosins are not known to bend in this region (50), but scallop myosin has been shown to bend (51). As many myosins have not been studied for their folding ability, it is difficult to generalize about the importance of site C in stabilizing a folded conformation.
10 S cross-linked myosin contained a cross-link from only one of the two heads to the tail. Single-turnover kinetics was used to enzymatically distinguish between the two heads. In the absence of actin, the single-turnover data for control unphosphorylated myosin was best to fit to a single decay rate (data not shown). A population of unregulated myosins, as was observed by others (52), was not detected. As the ionic strength was varied between 10 S and 6 S conditions (Fig. 6A), a 17-fold increase in turnover rate was observed, similar to previous steady-state measurements (19). The single-turnover rate at 0.15 M KCl was 0.001/s, similar to previously reported values under similar conditions (25). This rate was similar to the faster of the two components (0.004/s) observed in other experiments (24) under comparable conditions. Whereas one head of 10 S cross-linked myosin remained inactive at all ionic strengths tested, the second head was activated to 70% of the value (Fig. 6B) obtained for unphosphorylated myosin (Fig.  6A). Motility experiments showed that this stabilized trapped state is not from irreversible inactivation of the head (Table I). Therefore, even in the presence of one inactive head, the other head can be partially activated (to 70% of control) in the absence of actin at high ionic strength. A stabilized interaction of the tail with a region near Cys-108 on a RLC is sufficient for a slow product release in that head. Perhaps tail binding to the head stabilizes an ionic interaction that is important to slow P i release.
It was of interest to determine the extent of phosphorylationdependent regulation of the actin-activated Mg 2ϩ -ATPase activity in 10 S cross-linked myosin. 10 S cross-linked myosin had a low actin-activated Mg-ATPase similar to unphosphorylated control myosin (Fig. 7). Although the 10 S cross-linked myosin was fully thiophosphorylatable (1.9 -2 mol of thiophosphate per mol of myosin), thiophosphorylation did not appear to influence FIG. 7. Effect of actin on steady-state ATPase activity. Activity was determined from the rate of hydrolysis of [␥-32 P]ATP as described (61), except charcoal was removed from the quenched reaction by pelleting using a microcentrifuge. The assay conditions were 0.2 mg/ml myosin, 30 mM KCl, 15 mM Tris-Cl (pH 7.5), 5 mM MgCl 2 , 2 mM DTT, 0.1 mM EGTA, 1 mM ATP at 25°C. Rabbit skeletal actin (32) was dialyzed extensively against assay buffer to remove ATP. Chicken gizzard tropomyosin was added to actin at a ratio of one tropomyosin per four actin and allowed to equilibrate on ice for 2 h prior to addition to myosin. At time ϭ 0, [␥-32 P]ATP (1 mM final) was added to the actomyosin solution. Aliquots were quenched at indicated times for thiophosphorylated control myosin (open circles), nonphosphorylated control myosin (closed circles), thiophosphorylated 10 S cross-linked myosin (open squares), and nonphosphorylated 10 S cross-linked myosin (closed squares). All data points represent n Ն 2, where n ϭ the number of determinations. For noncross-linked samples, most data points are the average where n Ն 3. Due to limited amounts of cross-linked sample, the majority of data points are averages of duplicates. For n Ն 3, error bars show the S.D.; for n ϭ 2, error bars show the range. Lines were not drawn through the data to allow error bars to be seen. this steady-state activity. This result was confirmed using a single-turnover assay (Fig. 8). At the highest concentration of actin appropriate for this protocol (5 M), both unphosphorylated and phosphorylated 10 S myosin had a slow rate of product release similar to unphosphorylated control myosin. Both heads of 10 S cross-linked myosin appear to have a low actinactivated Mg 2ϩ -ATPase that is not phosphorylation-dependent. Therefore, based upon the conditions tested in this assay, the presence of one inactive head (see above) appears to prevent phosphorylation-dependent activation of the other head in a molecule with a folded tail.
The low Mg-ATPase activities (Figs. 7 and 8) and lack of ability to move actin filaments (Table I) can be explained by a weak binding of 10 S cross-linked myosin with actin; the minimum value of the equilibrium dissociation constant was approximately 500 M in the absence of nucleotide (data not shown). It appears that one head-tail cross-link confers weak binding to both heads. The constant for the dissociation of the acto-10 S myosin-AMPPNP complex is approximately 42 M (20). However, this value may be an underestimate of the dissociation constant, because actin binding can induce 10 S myosin to assemble into filaments (53,54). The measurement described here is not similarly affected, since 10 S cross-linked myosin cannot form filaments. 10 S myosin has a slower rate of proteolysis than 6 S myosin at an actin binding site between the central 50-kDa and the NH 2 -terminal 20-kDa region of the HC (20). Weak actin interaction observed for the 10 S crosslinked myosin may be explained by a stabilized conformation of the HC at this actin binding site.
Recent kinetic experiments with smooth muscle myosin show that a trapped ADP-P i -6 S state isomerizes to the 10 S state (54). This means that head-tail binding will be observed in myosins with long-lived ADP-P i states as has been proposed previously (24). However, head-tail binding is not required for slow product release (19,21,55,56). The structure of the tail binding site in 10 S myosin reveals information about the myosin-ADP-P i state. The tail binds to the COOH-terminal domain of the RLC (this work) which is approximately 7.5 nm from the nucleotide binding site (57). From the study of purified single-headed myosin, the tail binding site appears to include both heads (Fig. 9), and both heads are required for trapping (this work and Ref. 44). Two heads are also required for phosphorylation-dependent regulation of smooth muscle myosin (44). The region of the tail found to be photocross-linked (Fig. 4) in 10 S cross-linked myosin is consistent with binding of the tail to both heads (Fig. 9). The sequence from Lys-11 to Arg-16 in the NH 2 -terminal region of the gizzard RLC is important for the stabilization of the 10 S conformation (18). If these residues are deleted, the myosin behaves like the phosphorylated form, i.e. the Mg 2ϩ -ATPase is high and independent of ionic strength (18). Therefore, this deletion mutant is another example of a molecule that cannot trap nucleotide and therefore does not The single-turnover of FTP was monitored by the decay in fluorescence under the assay conditions described under "Materials and Methods" except that 70 mM KCl was included in the buffer. At time ϭ 0, the initial fluorescence of each myosin solution was arbitrarily set to 1000. All additions to the cuvette are indicated by a drop in the signal to 0. FTP was added (to 2 times the number of active sites), and the fluorescence was allowed to reach a maximum before the addition of actin. After the fluorescence was again allowed to reach a maximum, ATP was added and the decay in fluorescence due to the turnover of FTP was monitored. The fluorescence obtained after addition of ATP was always about 10% lower than the fluorescence reached prior to the ATP addition. This drop in fluorescence may be due to a change in the viscosity and turbidity of the actomyosin after ATP addition and may mask a small amplitude due to unregulated myosin. This rapid drop in intensity was not observed in the absence of actin (i.e. as in the data used to generate Fig. 6). See legend of Fig. 6 for appropriate further details of the method. Unphosphorylated control myosin (A), thiophosphorylated control myosin (B), unphosphorylated 10 S cross-linked myosin (C), and thiophosphorylated 10 S cross-linked myosin (D). stably fold. The data for these two modified myosins, singleheaded and the above-mentioned deletion mutant, suggest that Lys-11 to Arg-16 of the RLC are important for interaction between the heads that is required for trapping and subsequent formation of a tail binding site.
In conclusion, the 10 S conformation of smooth muscle myosin involves an interaction between the COOH-terminal domain of the RLC (near Cys-108) with residues Leu-1554 to Glu-1583 of the LMM portion of the myosin tail. If this interaction is stabilized by cross-linking, a head conformation with a slow rate of product release is stabilized even under high ionic strength conditions. Within the same molecule, a head that is not cross-linked to the tail remains largely unaffected by the presence of the folded tail at high ionic strength. It is significant that the COOH-terminal domain of the RLC has been shown to be important for phosphorylation-dependent regulation in smooth muscle myosin (42,58) and for regulation by Ca 2ϩ binding in scallop myosin (59). Two heads are required to form the 10 S conformation (this work), and two heads are also required for phosphorylation-dependent regulation in smooth muscle myosin (44). In the stabilized 10 S conformation studied here, it appears that both heads bind very weakly to actin, are not activated by actin or phosphorylation, and do not move actin in vitro. Additional studies with smooth muscle myosin filaments are underway to further examine the structural aspects of the phosphorylation-dependent interactions of the RLC.  (50 -200 g) was performed on a TSK G4000SW column (TosoHaas, 7.5 mm ϫ 60 cm, 13-m particle size, with a 75 ϫ 7.5 mm guard column). The column was run at room temperature in 10 mM HEPES (pH 7.3), 1 mM MgCl 2 , 0.1 mM EGTA, 25 M ATP at the indicated KCl concentration. The flow (0.75 ml/min) was controlled by a Beckman HPLC, and peaks were monitored at 280 nm (0.1 absorbance unit full scale) with a Beckman 165 variable wavelength detector. The uncertainty in the measurements was approximately Ϯ 0.2 min. The percent of applied protein which eluted was determined by the area under the peak divided by the area under the peak at 0.5 M KCl. Protein that was filamentous did not elute from the column.