Subunit Interactions within an Expressed Regulatory Domain of Chicken Skeletal Myosin

The regulatory domain (RD), or neck region of the myosin head, consists of two classes of light chains that stabilize an α-helical segment of the heavy chain. RD from chicken skeletal muscle myosin was prepared in Escherichia coli by coexpression of a 9-kDa heavy chain fragment with the essential light chain. Recombinant regulatory light chain (RLC), wild type or mutant, was added separately to reconstitute the complex. The affinity of RD for divalent cations was determined by measuring the change in fluorescence of a pair of heavy chain tryptophans upon addition of calcium or magnesium. The complex bound divalent cations with high affinity, similar to the association constants determined for native myosin. The intrinsic fluorescence of the tryptophans could be used as a donor to measure the fluorescence resonance energy transfer distance to a single labeled cysteine engineered at position 2 on RLC. Dansylated Cys2 could also serve as a donor by preparing RLC with a second cysteine at position 79 which was labeled with an acceptor probe. These fluorescence resonance energy transfer distances (24–30 Å), together with a previous measurement between Cys2 and Cys155 (Wolff-Long, V. L., Tao, T., and Lowey, S. (1995) J. Biol. Chem. 270, 31111–31118) suggest a location for the NH2 terminus of RLC that appears to preclude a direct interaction between the phosphorylatable serine and specific residues in the COOH-terminal domain.

The nucleotide-and actin-binding sites are located in the globular portion of the myosin head (S1), 1 whereas the light chains are wrapped around a long (ϳ10 nm) ␣-helical region that includes about 70 residues of the COOH terminus of the S1 heavy chain (1) (see Fig. 1). The light chain-binding region is also known as the "regulatory domain" (RD), because regulation of the actin-activated ATPase activity is achieved either by calcium binding to the essential light chain (ELC) of molluscan myosins (2), or by phosphorylation of the regulatory light chain (RLC) of smooth and non-muscle myosins (3). The RLC of vertebrate striated muscle can also be phosphorylated at a homologous serine residue, but here phosphorylation has only a modulatory effect on activity (4). How these signals are transmitted from RD to the active sites remains a major unsolved problem.
A stable fragment containing RLC and ELC bound to a 9-kDa heavy chain peptide was first isolated by proteolytic digestion of scallop myosin (5). This fragment bound calcium with high affinity in the presence of magnesium, mimicking the behavior of intact scallop myosin. Crystallization of RD led to a high resolution structure that revealed the detailed interactions of the three chains in the complex, and their contribution to a calcium-specific binding site in ELC (2). The overall fold of the LCs in scallop RD is similar to that in the neck region of chicken skeletal S1 (1), although there are differences that are probably related to the unique regulatory mechanisms in the two myosins (6).
Our initial goal was to see if a correctly folded RD complex, consisting of two light chains and a heavy chain peptide, could be prepared in an expression system. Not only would this approach eliminate microheterogeneity introduced by proteolysis, but the opportunity would arise to study light chain-heavy chain interactions by site-directed mutagenesis. To date it has not been possible to express sufficient quantities of a striated muscle myosin for biophysical studies. We show here that homogeneous RD, retaining the high affinity nonspecific calcium/magnesium site found in all muscle myosins, can be prepared in good yield in a bacterial expression system. Moreover, by labeling a single engineered cysteine (Cys 2 ) with 1,5-IAE-DANS at the NH 2 terminus of RLC, it was possible to determine the fluorescence resonance energy transfer distance between Cys 2 and a cluster of two tryptophans located at the sharp bend of the heavy chain ␣-helix, which served as a donor to IAEDANS. Dansylated Cys 2 could also be used as a donor by engineering a second cysteine at position 79, which was subsequently labeled with an acceptor molecule. These two FRET distances, together with a previously determined distance between Cys 2 and Cys 155 (7), were used to define an approximate position for the NH 2 terminus of RLC. The first 19 residues of RLC are not visible in the crystal structure of chicken S1, probably due to their flexibility (1). The structure of scallop RD is without 11 NH 2 -terminal RLC residues, since these are removed by proteolysis during the preparation of RD (2). The location of the missing sequence by fluorescence resonance energy transfer, despite the low resolution of this technique, provides a structural framework for considering how the actomyosin interaction can be enhanced by phosphorylation of a single serine on a light chain far removed from the active site. * This work was supported by National Institutes of Health Grant AR17350 and National Science Foundation Grant IBN9119570. 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.

MATERIALS AND METHODS
Expression of Heavy Chain Peptide and ELC-Embryonic chicken skeletal muscle myosin heavy chain cDNA, provided by J. Robbins (8), was used as a template to obtain cDNA encoding residues Leu 774 to Leu 844 by the polymerase chain reaction method. Restriction sites EcoRI and BamHI were created at the 5Ј-and 3Ј-ends, respectively, for cloning into the pT7-7 expression vector. This construct produces 4 extra residues (MARI) at the NH 2 terminus of the expressed protein.
The cDNA obtained for chicken skeletal muscle ELC (LC1 isoform) was missing 45 bases from the 5Ј-end that encodes the first 15 residues of LC1 (9). An oligonucleotide containing the additional bases was ligated to the cDNA, and the completed sequence was cloned into an expression vector designated pWY. This vector was engineered from two plasmids, pT7-7 and pGP1-2 (10), for the purpose of having a different origin of replication and antibiotic resistance (kanamycin) from pT7-7 (ampicillin). The light chain expressed from the pWY/LC1 construct has an NH 2 -terminal tag of 4 residues (MARI) from the vector followed by Pro 2 as the first residue of LC1 instead of Ala. Sequencing of the cDNA revealed a six-base deletion (encoding Pro-Ala) in the 5Ј-coding region, which results in six pairs of Pro-Ala instead of the seven pairs found in tandem in the native protein. Comparison of LC1 sequences from different species indicates a high degree of variability in the NH 2 -terminal regions (11), and the deletion was considered unimportant for the experiments presented here.
A construct with both LC1 and wild-type RLC in the same pWY vector was also designed. The pWY/LC1 clone was digested with PstI/ BamHI and the 700-base pair PstI/BglII fragment from pT7-7/LC2 encoding RLC (12) was ligated into this clone. Thus, the cDNA for each light chain was positioned in tandem with the promoter and stop codon. This construct was used to express both light chains together with the pT7-7 vector containing the 9-kDa heavy chain fragment. Protein expression was in BL21(DE3), and bacterial cultures were grown at 37°C for 16 -18 h in the presence of ampicillin and/or kanamycin as described in Wolff-Long et al. (13).
Preparation of RLC-The expression and purification of wild-type and mutant chicken skeletal RLC in the vector pT7-7 has been described (14). Wild-type RLC contains two endogenous Cys residues at positions 126 and 155. In some RLC mutants, e.g. Cys 126 /Cys 155 and Cys 2 , the endogenous tryptophan at position 138 was changed to phenylalanine in order to eliminate the contribution of Trp fluorescence. In other mutants the endogenous Cys residues were selectively removed by mutagenesis to alanine, as in Cys 126 , and new Cys residues were introduced at positions 2 and 79 as in Cys 2 and Cys 2 /Cys 79 . Table I gives a description of all the RLCs, and the location of specific residues is shown in the ribbon diagram of the regulatory domain in Fig. 1.
Purification of the Regulatory Domain-When the 9-kDa heavy chain fragment and ELC are coexpressed, they are found in inclusion bodies, whereas simultaneous expression of heavy chain fragment together with RLC and ELC results in a soluble complex in the cytoplasm. An advantage of the first method is that the soluble bacterial proteins can be readily separated from the inclusion bodies by centrifugation, followed by washing of the pellet with a mild detergent (14). The relatively pure protein in the inclusion bodies can then be used to prepare complexes with a variety of unlabeled or labeled mutant RLCs. Although simultaneous expression of the three chains is in some ways a simpler procedure, a major drawback is the presence of a much higher concentration of contaminating bacterial proteins that are difficult to remove. For this reason, and since we usually wished to incorporate a labeled RLC into the regulatory complex, the first preparative procedure was adopted.
The cell pellet from a 1-liter culture of Escherichia coli BL21(DE3) coexpressing 9-kDa HC and LC1 was lysed, and the inclusion bodies isolated as described (14). The washed pellet was resuspended in 8 M guanidine, 10 mM DTT in PBS (10 mM sodium phosphate, pH 7.2, 150 mM NaCl, 3 mM NaN 3 ), to which RLC (20 -25 mg unchromatographed, or purified and labeled) was added with stirring at room temperature for 1 h. Insoluble cell debris was removed by centrifugation (100,000 ϫ g for 30 min), and the proteins were renatured by dialysis against PBS, 1 mM DTT, 1 mM MgCl 2 at 4°C. Denatured insoluble proteins were removed by centrifugation, and the supernatant was dialyzed against 10 mM sodium phosphate, pH 7.5, 1 mM MgCl 2 , 1 mM DTT, 3 mM NaN 3 for 4 h, clarified, and applied to a hydroxylapatite column (1.5 ϫ 10 cm) equilibrated with the same buffer. The proteins were eluted with a linear gradient (10 -250 mM sodium phosphate in a total volume of 200 ml), and the fractions containing the regulatory complex were pooled and further purified on a DEAE-Sephacel (Pharmacia) column (1 ϫ 10 cm) equilibrated in 10 mM imidazole, pH 7.0, 20 mM NaCl, 2 mM MgCl 2 , 1 mM DTT, 1 mM NaN 3 . After eluting the proteins with a linear gradient to 0.5 M NaCl (total volume 100 ml), the fractions were analyzed by 15% SDS-PAGE and pooled. Trace amounts of remaining impurities were removed by gel filtration chromatography on a Superose 12 column (1 ϫ 30 cm, FPLC, Pharmacia). The yield was 10 -15 mg of complex.
Labeling of RLC-Purified RLC (lyophilized with sucrose) was dissolved in 8 M guanidine, PBS, 10 -20 mM DTT, and stirred at room temperature for 1 h in order to completely reduce the cysteines before dialysis against PBS, 50 M DTT, with at least one change of buffer. The single Cys mutant, Cys 2 (Table I), was reacted with 2.5-5-fold molar excess of 1,5-IAEDANS (Molecular Probes) over total thiols for 2 h on ice in the dark before stopping the reaction by adding DTT to a final concentration of 10 mM. The double Cys mutant, C2/C79, at a concentration of 5-7 mg/ml, was reacted with a 3-fold molar excess of 1,5-IAEDANS for 2 min on ice. The reaction was stopped with 50 mM DTT, and the sample dialyzed against 50 mM sodium phosphate, pH 7.0, 0.2 mM EDTA, 1 mM DTT, and 3 mM NaN 3 . The protein was chromato- and RLC (green, COOH-terminal lobe; and yellow, NH 2 -terminal lobe) are wrapped around an ␣-helical HC peptide (gray) that serves to connect the motor domain to the rod. The NH 2 terminus of RLC (red) starts at Phe 20 since there is no electron density for the first 19 residues in the chicken S1 structure (1). The phosphoserine would be at residue 14 in chicken skeletal RLC. The position of Cys and Trp residues (red) and the Mg 2ϩ -binding site (blue) in RLC, and the Trp pair (red) at the sharp bend of HC are indicated. This structure is of scallop RD (2) which is very similar to chicken RD in its overall fold. Numbering is for the chicken sequence.
graphed on a DEAE-Sephacel (FPLC, Pharmacia) column (1.5 ϫ 10 cm) equilibrated in the above buffer and eluted with a linear gradient to 300 mM sodium phosphate (total volume 150 ml). Fractions containing unreacted, singly labeled (C2 DAN /C79), and doubly labeled RLC were identified by urea-PAGE. C2 DAN /C79 (in PBS, 50 M DTT) was subsequently reacted with a 4-fold molar excess of DDPM (Aldrich) at room temperature for 4 h to obtain C2 DAN /C79 DDP . Iodoacetamide was used to block any unreacted cysteines in control experiments. The light chains were dialyzed exhaustively against PBS to remove excess reagent before being used to prepare the RD complex.
Spectrophotometric analysis of the labeled light chains was used to determine the extent of labeling. Protein concentration was determined by the method of Bradford (15) using wild-type RLC as a standard (absorbance of 0.5 at 280 nm, 1 mg/ml). A molar extinction coefficient of 6200 M Ϫ1 cm Ϫ1 at 340 nm was used for IAEDANS-labeled proteins. The ratio of DANS to light chain was ϳ0.9. The molar extinction coefficient for DDPM-labeled RLC was taken as 2930 M Ϫ1 cm Ϫ1 at 442 nm (7). The ratio of DDP to RLC was approximately 1.
Titration with Divalent Cations-RLC and the regulatory domain (with either wild-type or mutant RLCs) were dialyzed against 0.1 M Hepes, pH 7.2, 0.1 M NaCl, 1 mM NaN 3 for 10 -14 h and clarified at 100,000 ϫ g for 30 min. EGTA was added to a final concentration of 1 mM before titration with calcium ions. Similarly, 1 mM EDTA was used in titrations with magnesium ions. Proteins were diluted to an absorbance of less than 0.05 to reduce the effect of light scattering (and the Raman scattering was removed by subtraction of a buffer blank). Small aliquots (2-20 l) of the stock CaCl 2 or MgCl 2 solution were delivered by a microsyringe (Hamilton Co.) into 2 ml of the protein solution with constant stirring. A SLM 48000 spectrofluorometer was used to measure tryptophan fluorescence (emission maximum at 340 nm) upon excitation at 295 nm. The emission spectra were recorded from 300 to 400 nm at 1-nm intervals with 2-s integration at each wavelength. The sample compartment temperature was maintained at 20°C. Percent decrease in tryptophan fluorescence (after volume correction) was plotted against -log free metal ions (pM) to obtain the concentration (equivalent to 1/K a , the association constant) at which the half-maximal change in fluorescence occurs. The program used to calculate free metal ion concentrations was based on the methodology of Storer and Cornish-Bowden (16).
Energy Transfer Measurements-Steady-state fluorescence energy transfer measurements were performed on an ISS PC1 photon counting spectrofluorometer equipped with a rhodamine B (4 mg/ml in ethanol with a RG 630 filter) quantum counter. Spectra were recorded in ratio mode to correct for lamp intensity fluctuations. The spectral bandwidths for excitation and emission were 8 nm. Emission spectra were recorded at 1-nm intervals with 2-s integration at each wavelength, and were later corrected with software supplied by ISS. A 305 nm cut-off filter was used in the emission path for tryptophan spectra. When IAEDANS was used as a donor, 320 and 470 nm cut-off filters were placed in the excitation and emission paths, respectively.
The cluster of two tryptophans in the heavy chain peptide of RD could be used as an intrinsic energy donor. The acceptor was a single cysteine on RLC labeled with 1,5-IAEDANS. For preparation of RD, RLC mutants were used in which tryptophan residue 138 was replaced with phenylalanine. Two samples of RD had to be independently prepared, one containing Cys 2 blocked with iodoacetamide, and the other containing IAEDANS-labeled Cys 2 . Proteins were dialyzed against Hepes buffer containing 1 mM magnesium chloride and clarified as described above. The tryptophan emission spectra were recorded under identical optical conditions for the unlabeled and labeled complexes diluted into the dialysate or 7 M guanidine hydrochloride. Contributions to the spectra due to Raman scattering were removed by subtraction of the buffer blank. In guanidine, where the complex is fully dissociated and energy transfer is presumably zero, the tryptophan intensities of the two complexes are directly proportional to their concentrations. Therefore, the ratio of the tryptophan fluorescence intensities for the labeled and unlabeled RD complexes in the native state should be the same as in the denatured state, if no energy transfer occurs. After normalizing the concentrations of the two species, any observed decrease in fluorescence of the acceptor-labeled RD relative to the unlabeled sample can be attributed to energy transfer between donor and acceptor. The same procedure was applied to RD specimens in which RLC labeled with IAEDANS was used as the donor and a second cysteine introduced into RLC was labeled with the acceptor DDPM.
The efficiency of energy transfer (E) was calculated according to the following equation, where F da and F d are the measured fluorescence intensities in the presence and absence of the acceptor, respectively. The degree of energy transfer is dependent upon the distance (R) between donor and acceptor and is given by Equation 2, where R 0 , the Forster critical distance at which the energy transfer is 50%, was calculated by experimentally determining the parameters in Equation 3, where Q d is the quantum yield of the donor, n is the refractive index of the medium, taken here as 1.4, J is the overlap integral expressed in mol Ϫ1 cm 3 , and 2 is the orientation factor. The quantum yield, which depends upon the chemical environment of the donor, was calculated for the free light chain and RD by the procedure described by Chen (17) using quinine sulfate (in 0.05 M sulfuric acid) as a standard. The absorbance of proteins at 295 nm and the quantum yield of 0.70 for quinine sulfate (18) were used in the following equation, In order to determine the overlap integral, J, the emission spectrum of the donor (tryptophan excited at 295 nm) and the absorption spectrum of the acceptor (IAEDANS) were recorded at 1-nm intervals from 320 to 380 nm. The value of J was calculated according to the expression, where ⑀ a is the extinction coefficient (mol Ϫ1 cm Ϫ1 ) of the acceptor and F is the fluorescence intensity of the donor at wavelength (cm). Steady-state polarization measurements were performed on RLC and RD in order to estimate the upper and lower limits of the orientation factor 2 (19) essentially as described in Ref. 20. The limiting polarization anisotropies for donor and acceptor were determined using L-format with the emission monochromator set at a fixed wavelength (340 and 480 nm for donor and acceptor, respectively). Sucrose was added to the samples to provide a range of viscosities at 20°C for extrapolation to infinite viscosity in Perrin plots (21).
Other Procedures-SDS-PAGE was performed as described by Laemmli (22). The extent of labeling was monitored by urea-PAGE analysis (23). The extinction coefficients of the recombinant RDs were determined by measuring the absorbance at 280 nm and determining the concentration by the method of Bradford (15). Sedimentation velocity runs were made in a Beckman Model E analytical ultracentrifuge. Myosin light chains were phosphorylated as described previously (24).

RESULTS
Purification and Characterization of Recombinant RD-The 9-kDa heavy chain fragment (HC) could not be obtained in any significant amount when expressed without the light chains, as determined by SDS-PAGE analysis of the crude extracts from overnight cultures (data not shown). Therefore, it was necessary to coexpress HC with both light chains, ELC (LC1 isoform) and RLC (LC2) (Fig. 2A, lane 1) or with ELC alone (lane 2). In the latter method, HC and ELC were segregated into inclusion bodies, which could be readily separated from the bacterial proteins (lane 3). After resuspension in 8 M guanidine, RLC was added to solubilize the complex upon subsequent dialysis into a benign solvent (lane 4). RD was purified by ion-exchange chromatography on hydroxylapatite (lane 5) followed by DEAE-Sephacel (lane 6) to remove any last trace of impurities. The recombinant RLC added to the HC⅐ELC complex was one of several mutants described in Table I. The purified RD remained stable at all concentrations (up to 20 mg/ml) when stored at 4°C.
Various methods were used to establish the composition of RD. The co-elution of the light and heavy chains from the ion-exchange columns during purification suggested that they form a complex. Upon gel filtration chromatography, RD eluted between the chymotryptic S1 (ϳ100 kDa) and the free light chains (ϳ20 kDa) consistent with a molecular mass of ϳ50 kDa. The sedimentation velocity pattern for purified RD showed a single, homogeneous peak with a sedimentation coefficient of 3.3 S (Fig. 2B). The extinction coefficients of RDs made with wild-type and Cys 126 /Cys 155 were 0.5 and 0.4, respectively, which agrees well with the theoretical values based on the number of tryptophans and tyrosines in a 1:1:1 complex (25). Although the ratio of the two LCs and HC as determined by densitometry of the Coomassie Blue-stained SDS gels was 1:1:2, it is probable that the large number of charged residues in the heavy chain peptide caused a disproportionate absorption of dye.
Intrinsic Fluorescence of Heavy Chain Tryptophans-There are three tryptophans in wild-type RD, two in the heavy chain peptide (Trp 829 -Pro 830 -Trp 831 ) and one in wild-type RLC (Trp 138 ) (Fig. 1). A mutant RLC was prepared in which Trp 138 was replaced by Phe so that Trp fluorescence from the heavy chain could be used as a single reporter group or donor site in FRET experiments. Tryptophan emission spectra for RD containing wild-type or mutant RLC were compared with free wild-type RLC (Fig. 3). RD(mut) containing two tryptophans from the heavy chain showed far less fluorescence than wildtype RLC which has only a single tryptophan. The diminished intensity of fluorescence suggests intramolecular quenching by the heavy chain tryptophans due to their close proximity in the structure (WPW). In contrast, the single Trp 138 in wild-type RLC appears to increase its fluorescence in the more hydrophobic environment of the heavy chain complex, as judged by the enhanced fluorescence of RD(wt) compared with the sum of the fluorescence intensity of wild-type RLC and RD(mut). In a denaturing solvent such as guanidine hydrochloride, the fluorescence intensity is more proportional to the tryptophan content of the protein: i.e. the fluorescence of RD(wt) is equal to the sum of the fluorescence of RD(mut) and wild-type RLC (Fig. 3). The emission maxima in guanidine hydrochloride showed a red shift (353 nm) compared with the emission maxima at 343 and 340 nm for WT and RDs, respectively, in a benign solvent. This can be explained by the increased exposure of the Trps to solvent in an unfolded, denatured state.
Binding of Divalent Metal Ions to RLC and RD-Titrations with divalent cations were carried out to determine if expressed RD would bind calcium and magnesium with affinities similar to those found for native myosin. First, the intrinsic fluorescence of Trp 138 in RLC was measured as a function of calcium (or magnesium) to determine the binding of these ions to free RLC (data shown only for Ca 2ϩ binding) (Fig. 4, A and  B). The total decrease in fluorescence intensity was ϳ18% for Ca 2ϩ and ϳ12% for Mg 2ϩ ions. The association constants of 10 5 M Ϫ1 for Ca 2ϩ and 10 3 M Ϫ1 for Mg 2ϩ obtained here agree well with those reported by Alexis and Gratzer (26) for rabbit skeletal RLC. Despite the considerable distance between the cation binding site in the NH 2 -terminal domain and the single Trp in the COOH terminus of RLC, the conformational change associated with metal binding must be transmitted to the Trp, as evidenced by a change in signal.
Similarly, the decrease in fluorescence for RD(wt) or RD-(mut) as a function of calcium (or magnesium) was used to determine the affinity of RLC for metal ions in the heavy chain bound state. Association constants of 10 7 -10 8 M Ϫ1 for Ca 2ϩ , and 10 6 for Mg 2ϩ were obtained (Fig. 4, C and D). The smaller fluorescence change with RD(wt) compared with RD(mut) can be attributed to the high level of background fluorescence contributed by Trp 138 in wild-type RLC. The large fluorescence change in RD(mut) (ϳ32%), despite the low absolute intensity of the signal, suggests that the heavy chain tryptophans are perturbed to a greater degree by metal binding than the single Trp in RLC. This may be related to the hydrophobic interactions of residues in the NH 2 -terminal domain of RLC with the FIG. 3. The corrected tryptophan fluorescence emission spectra of RD and RLC. The excitation wavelength is 295 nm. Molar protein concentration is constant for all species. RD(wt) (solid line) contains two heavy chain Trp(WPW) and a wild-type RLC that has a single Trp, see Table I. RD(mut) (dotted line) contains the two heavy chain Trp, and the mutant RLC (C126/C155) that has no Trp. The low fluorescence intensity of RD-(mut) relative to wild-type RLC(WT) (broken line) in a benign solvent (PBS) suggests intramolecular quenching of the heavy chain Trps. In a denaturing solvent, such as guanidine hydrochloride, the fluorescence intensity more closely approximates the tryptophan content of the protein: i.e. RD(wt) Ͼ RD(mut) Ͼ WT.   Table I. B, sedimentation velocity pattern for purified RD showed a single, homogeneous peak. tryptophans in the COOH terminus of the heavy chain (2). The binding constants obtained here are in good agreement with earlier values (3 ϫ 10 7 M Ϫ1 for Ca 2ϩ ) determined by more direct binding methods (27).

Fluorescence Energy Transfer between Heavy Chain Tryptophans and IAEDANS-labeled RLC-
The intrinsic fluorescence of the two heavy chain tryptophans can also be used as a donor in energy transfer experiments. By introducing a single IAE-DANS acceptor probe into the Cys 2 residue of RLC, we could measure the energy transfer distance between the cluster of Trp at the bend of the heavy chain helix and the NH 2 terminus of RLC. The tryptophan emission spectra of RD (excitation at 295 nm) prepared either with or without the IAEDANS-labeled RLC are shown in Fig. 5A. The tryptophan emission is reduced by about 9% when Cys 2 is labeled with the IAEDANS acceptor (RD(C2 DAN )). Although donor quenching is small, the large enhancement in IAEDANS fluorescence at 500 nm compared with the free labeled light chain (C2 DAN ) is strong qualitative evidence for energy transfer from the heavy chain tryptophans to the dansyl probe on RLC.
We also examined the excitation spectra for several constructs that contained either no tryptophan or had tryptophans located at different distances from the acceptor probe, IAE-DANS (Fig. 5B). The appearance of a tryptophan component between 280 and 300 nm in the excitation spectrum of IAE-DANS (emission measured at 500 nm) is strong evidence that energy transfer has occurred. C2 DAN has no tryptophan and therefore shows little fluorescence when excited in the region of 290 nm. C126 DAN shows strong fluorescence when the single tryptophan at position 138 is excited due to the close proximity of the labeled Cys 126 . Similarly, the tryptophan cluster (WPW) on the heavy chain shows energy transfer, although much weaker, to the dansylated Cys 2 in RD(C2 DAN ).
The efficiency of energy transfer (E) between the heavy chain Trp and the IAEDANS probe on RLC, as determined from emission spectra, is given in Table II. In order to convert E into a distance, the quantum yield (Q d ) and the overlap integral (J) had to be determined for RD(mut) (see "Materials and Methods"). The low value of 0.02 obtained for Q d is not surprising since the tryptophan emission is unusually low for this complex (see Fig. 3). The overlap integral for the Trp/IAEDANS pair was calculated to be 6.1 ϫ 10 Ϫ15 M Ϫ1 cm 3 . The orientation factor, 2 , used in the calculations was the isotropically aver- to 0.16, indicating considerable randomization of the probe orientation (21). A value of 2/3 for the orientation factor has generally been used in distance calculations, with no major errors in distance resulting from this assumption. Based on the above parameters, R 0 , the distance at which E is 50%, was 16 Å which yields a distance of 23-24 Å between -WPW-and Cys 2 in RD. This distance did not change significantly upon phosphorylation of RLC, or in the presence or absence of divalent cations.
As a control, the distance between Trp 138 and Cys 126 was determined in the RLC mutant, Cys 126 (see Table I) using IAEDANS as the acceptor. Between 80 and 85% quenching was observed in the presence or absence of metal ions. A Q d of 0.14 was obtained for the single Trp in RLC, which is close to the value reported for free tryptophan (17). The value for the overlap integral was assumed to be the same as for the above, particularly, since a similar J value was reported for a Trp cluster and single IAEDANS-labeled Cys mutant in a membrane protein (20). The range of anisotropy values measured for Trp 138 /IAEDANS in RLC was also similar to RD, and a value of 2/3 was taken for the orientation factor. These parameters led to a value of 22 Å for R 0 , from which a distance of 16 -17 Å was calculated (Table II). This FRET distance agrees well with the distance of 18 Å for the corresponding residues in the crystal structure of scallop RD (2), and with an earlier distance determination by fluorescence spectroscopy (28).
Fluorescence Energy Transfer between Donor/Acceptor Probes on RLC Bound to the Heavy Chain-To fix the position of Cys 2 in RD, a minimum of three distances needs to be determined. In an earlier study, the FRET distance between Cys 2 and Cys 155 in the heavy chain-bound RLC was measured (7). For a third distance, Cys 2 and Cys 79 in RLC were labeled with IAEDANS and DDPM as donor and acceptor probes, respectively, and reconstituted with heavy chain peptide in RD. The greater reactivity of Cys 2 compared with Cys 79 again enabled us to prepare the RLC with donor and acceptor probes in a 1:1 stoichiometry (7). The emission spectra of RD containing mutant RLC labeled with donor, IAEDANS, at position 2 (C2 DAN /C79) and RD containing RLC with donor and acceptor (C2 DAN /C79 DDP ) is shown in Fig. 6. The fluorescence of IAE-DANS is strongly quenched by the non-fluorescent acceptor, DDPM, see Table II. In the presence of guanidine hydrochloride no energy transfer occurs, and the fluorescence intensity of the light chains with or without acceptor was the same. A distance of 29 -30 Å was calculated between Cys 2 and Cys 79 , using param-eters from the literature for these probes (7). As with the other sites, phosphorylation had no effect on the distance. DISCUSSION The structure and function of myosin have been studied historically by a reductionist approach: first, myosin was cleaved by proteolytic enzymes into heavy meromyosin and light meromyosin, which allowed a separation of the enzymatic from the assembly properties of this macromolecule. Next, the active globular head region, subfragment-1 (S1), was cleaved from the long ␣-helical rod (29). Although proteolytic procedures could be used to reduce S1 even further (5, 30), it became far simpler to prepare the "motor domain" by molecular biolog-  ical methods (31). Here we describe the first preparation of the "regulatory domain" or neck region of chicken skeletal myosin S1 by recombinant methods. The regulatory domain appropriately derived its name from the structure of the invertebrate, scallop myosin, a molecule that is regulated by a Ca 2ϩ -specific site in the neck region of the myosin head (32). The RD consists of an essential (ELC) and a regulatory (RLC) light chain, which together stabilize the S1 heavy chain in an ␣-helical conformation as it joins the ␣-helical coiled-coil rod (1). The crystal structure of scallop RD (2) first showed that this multisubunit interaction creates a Ca 2ϩ -binding site on the NH 2 terminus of the ELC, a result which could not be predicted from the sequence, since it fails to correspond to a typical EF-hand (2). The only other myosins regulated by Ca 2ϩ are the "unconventional" myosins which contain primarily calmodulin in the neck region of the head (33).
Unlike the invertebrate and unconventional myosins, the vertebrate myosin IIs are regulated to varying degrees by a phosphorylated serine in the NH 2 terminus of the RLC (34).
Smooth and nonmuscle myosins are completely inactive in the dephosphorylated state, whereas the activity of skeletal myosin is only modulated by the state of phosphorylation (4). This raises the intriguing question of which properties of myosin dictate whether a single phosphorylation site in the regulatory domain can accelerate product release in the motor domain? In a recent study, it was suggested that the RLC must possess certain arginine residues in its COOH-terminal domain which can bond to the phosphoserine in the NH 2 terminus (35). Only thick filament-regulated myosins have conserved arginine residues in the E to H-helices (nomenclature refers to calmodulin) of RLC that can serve as possible candidates for such an interaction. This might explain why skeletal RLC, which lacks these conserved Arg residues, cannot replace smooth RLC in the regulation of smooth muscle myosin (36).
Location of the NH 2 Terminus of RLC-The two FRET distances determined here, together with an earlier determination of the distance between Cys 2 and Cys 155 in RLC exchanged into myosin (7), define a locus for Cys 2 , which marks the beginning of the sequence for RLC (Fig. 7). The 19 residues in RLC, that are not visible in the crystal structure of chicken S1, appear to lie at an angle to the first A-helix of the EF-hand that forms the Mg 2ϩ -binding site. The region of closest approach between the phosphoserine and the remainder of the RLC molecule would be in the vicinity of the D/E linker segment that joins the NH 2and COOH-terminal lobes. Although, theoretically, Cys 2 could lie 180°away from this location, this orientation is preferred insofar as it maximizes side chain interactions between the NH 2 -terminal peptide and the linker region. Interestingly, the NH 2 terminus of troponin C is located in approximately the same position relative to the A-helix and the central linker helix (37). An Arg residue in the NH 2 terminus of TnC has been shown to form a H-bond with a Glu residue in the D-helix of the NH 2 -terminal domain (37). Assuming that skeletal and smooth RLC share a similar structure, the localization studies described here preclude a direct interaction between the phosphoserine and specific arginine residues in the COOH-terminal domain. However, before discussing alternative mechanisms for the activation of myosin's activity, one should question: 1) how faithfully the recombinant regulatory domain mimics the neck region of the native myosin molecule, and 2) how reliable are resonance energy transfer measurements in determining distances?
The structural determination of the regulatory domain of scallop myosin was carried out on a complex produced by a series of proteolytic digestions of the native protein (2, 5).  Table II, together with the FRET distance determined in myosin between Cys 2 and Cys 155 (30 Å) (7), a location for the NH 2 terminus of RLC (Cys 2 ) is defined.
Despite the absence of the motor domain, the structure of scallop RD was remarkably similar in its overall features to that of chicken S1 (1), implying that RD is a stable folding unit, and that the structure of RD in all myosin IIs is highly conserved. The remarkable stability of RD is evident from its ability to fold spontaneously upon replacement of a denaturing solvent with a benign solvent (5). The recombinant chicken RD used in this study was also prepared by reconstitution out of a denaturant, although the same complex could be prepared by simultaneous expression of the two LCs and the heavy chain peptide, which avoids the use of denaturants. The only functional property that can readily be measured is the binding affinity of RD for divalent cations. The good agreement between the binding constants obtained here and those reported for myosin, provides strong evidence that RD retains its native properties and is a good model system. An important advantage in using RD for structural studies compared with S1 is that the endogenous tryptophans on the heavy chain can be used as a donor group without introducing potentially disruptive extrinsic probes. Thr 79 was chosen for mutagenesis to Cys, because its position on the D-helix is solvent-exposed, and a probe at this site would be expected to have a minimal effect on RLC/heavy chain interactions.
The accuracy of distance measurements by resonance energy transfer is thought to be limited by assumptions about the orientation factor, 2 (see "Discussion" in Wolff-Long et al. (7) and references therein). Even with this uncertainty, usually less than 20%, FRET distance measurements are in a range that is sufficiently narrow to determine whether an interaction is likely to occur between the NH 2 and COOH termini of RLC. An unexpected experimental problem encountered here was the low quantum yield of the two neighboring tryptophans in the heavy chain peptide. This led to a low value for R 0 (16 Å), and therefore very little quenching of the Trp fluorescence by the acceptor probe (Ͻ10%). To perform these measurements, it was necessary to prepare RD with a high degree of purity since any contamination by fluorescent impurities would seriously compromise the data. The reproducibility of the results, and the demonstration of acceptor excitation, enhanced our confidence in the validity of these distance determinations.
Regulation by Phosphorylation-Protein phosphorylation stabilizes different conformational states of regulated molecules in order to enhance or repress biological activity. A classic example of such a mechanism is glycogen phosphorylase, whose crystal structure has been solved in both the active and inactive form (38). It was shown that phosphorylation of the NH 2 -terminal peptide results in a structural change from a disordered state to a distorted 3 10 helix, which is stabilized at the subunit interface of the glycogen phosphorylase dimer by a series of hydrogen bonds and van der Waals interactions. These hydrophobic interactions disrupt COOH-and NH 2 -terminal regions and new contacts are made which promote an active state.
It is difficult to envision how this type of allosteric mechanism can be readily applied to the myosin molecule. The two subunits of myosin are relatively independent entities in the head or S1 regions, and do not become joined until the junction with the ␣-helical coiled-coil rod. Computer modeling of the head/rod junction has suggested that the principal contact between the S1s occurs between the RLCs of neighboring heads (39). Experimentally, it was shown that mutant RLCs containing cysteines in the NH 2 -terminal domain formed intermolecular disulfide bonds between the heads (13). It is quite conceivable, therefore, that phosphorylation of the NH 2 terminus might induce a structural change that could disrupt the RLC interface interactions and free the S1 heads from their con-strained configuration. As interhead interactions are weakened, the intrasubunit interactions between the RLC and the heavy chain may be strengthened, thereby propagating a structural change to the active site that enhances the release of the products of ATP hydrolysis.
What is the evidence in support of such a mechanism? It is well established that single-headed smooth muscle myosin is active independent of phosphorylation (40,41). This observation is consistent with the hypothesis that the "off" state at a minimum involves interactions between the RLCs of the two heads, but it does not tell us about the affinities of these interactions. It has been shown that mutations in the COOHterminal region of smooth RLC can markedly reduce the affinity of RLC for the heavy chain, in parallel with a loss of myosin's motor activity (34,42). We have found that a single point mutation in skeletal RLC, which lowers its affinity to the heavy chain, also causes a reduction in the in vitro motility of skeletal myosin, both in the single-and double-headed state. 2 The general principal that emerges from these studies is that strong binding of RLC to the heavy chain is a prerequisite for the "on" state. This conclusion can also be applied to singleheaded myosin I, where removal of calmodulin results in a loss of motor activity (43). Even recent physiological experiments in skeletal muscle fibers have shown that phosphorylation increases Ca 2ϩ binding to RLC, an effect that would be expected to increase the RLC/heavy chain interaction, and may therefore be responsible for the increase in force production at submaximal levels of Ca 2ϩ (44).
This brings us back to the question of why mutating six residues in the COOH-terminal domain of skeletal RLC to those present in smooth RLC (of which four were Arg) had the effect of restoring regulation to a smooth myosin heavy chain, which was locked permanently in the off state with wild-type skeletal RLC (36). Instead of proposing a direct coordination between the phosphoserine and Arg residues (35), we suggest that the mutated skeletal RLC interacts more strongly with smooth heavy chain than wild-type skeletal RLC, with the consequence that phosphorylation can now activate this chimeric myosin. This alternative explanation pictures myosin regulation to be based on a complex series of molecular interactions between the RLC and the heavy chain that can be perturbed by phosphorylation, and transmitted to the active site through the regulatory domain.