Structural Model of the Regulatory Domain of Smooth Muscle Heavy Meromyosin*

The goal of this study was to provide structural information about the regulatory domains of double-headed smooth muscle heavy meromyosin, including the N terminus of the regulatory light chain, in both the phosphorylated and unphosphorylated states. We extended our previous photo-cross-linking studies (Wu, X., Clack, B. A., Zhi, G., Stull, J. T., and Cremo, C. R. (1999)J. Biol. Chem. 274, 20328–20335) to determine regions of the regulatory light chain that are cross-linked by a cross-linker attached to Cys108 on the partner regulatory light chain. For this purpose, we have synthesized two new biotinylated sulfhydryl reactive photo-cross-linking reagents, benzophenone, 4-(N-iodoacetamido)-4′-(N-biotinylamido) and benzophenone, 4-(N-maleimido)-4′-(N-biotinylamido). Cross-linked peptides were purified by avidin affinity chromatography and characterized by Edman sequencing and mass spectrometry. Labeled Cys108 from one regulatory light chain cross-linked to71GMMSEAPGPIN81, a loop in the N-terminal half of the regulatory light chain, and to4RAKAKTTKKRPQR16, a region for which there is no atomic resolution data. Both cross-links were to the partner regulatory light chain and occurred in unphosphorylated but not phosphorylated heavy meromyosin. Using these data, data from our previous study, and atomic coordinates from various myosin isoforms, we have constructed a structural model of the regulatory domain in an unphosphorylated double-headed molecule that predicts the general location of the N terminus. The implications for the structural basis of the phosphorylation-mediated regulatory mechanism are discussed.

The actin-activated ATPase activity and motor properties of smooth muscle and nonmuscle myosins are regulated by phosphorylation of the N terminus of the RLC 1 (1)(2)(3)(4). The RLC is a subunit of the two head domains (S1) with each S1 containing one motor domain, ELC and RLC. The two S1 domains are attached to a long ␣-helical coiled-coil domain (tail or rod). The regulatory domain is defined as an RLC and ELC attached to the portion of the heavy chain to which they bind. The unphosphorylated forms of these regulated myosins have low ATPase activity and are unable to move actin filaments, whereas the phosphorylated forms are activated in both respects.
Domain requirements for regulation have been elucidated through studies of various proteolytic and expressed subfragments of SMM. HMM, which lacks the C-terminal two-thirds of the tail, is double-headed and regulated (5), but expressed HMMs with truncated tails failed to form double-headed structures and were found to be unregulated (6 -8) as was S1 (9 -11) and single-headed myosin (12,13). Therefore, two heads are critical for regulation. Two motor domains were found to be required for regulation of a nonmuscle myosin (14). Most mutant constructs that have altered regulation, appear to have not only an increased ATPase activity in the unphosphorylated state but a decreased ATPase activity in the phosphorylated state. This suggests that the phosphorylated state does not reflect a simple case in which the inhibitory mechanism inherent to the native unphosphorylated state has been removed.
The structural basis of the regulatory mechanism is unknown. There are no crystal structures of double-headed, and therefore, regulated constructs. There are no atomic resolution data for the N terminus of the RLC (residues 1-24), which includes the phosphorylated serine at position 19 (smooth muscle isoform). The N terminus of the RLC is highly conserved in myosins that are regulated by phosphorylation and has been shown to be critical to the regulatory properties of the molecule (15).
It is likely that the mechanism whereby phosphorylation controls the motor ATPase activity is common and important to other isoforms that are only modulated by phosphorylation or that accomplish regulation through Ca 2ϩ binding. For example, a class of mutations found in the ␤-myosin isoform from cardiomyopathy patients are found clustered near the N terminus of the RLC (16 -18) and myosin RLC phosphorylation is a key determinant of the stretch activation response in Drosophila muscles (19). Ca 2ϩ binding to the ELC turns on molluscan myosins. The Ca 2ϩ ion mediates interactions between the RLC and the ELC in addition to the heavy chain (20,21). Molluscan myosins, like the smooth and nonmuscle isoforms, require two heads for regulation. It is likely that the Ca 2ϩmediated regulatory mechanism and the phosphorylation-mediated regulatory mechanism will have many structural parallels.
The goal of this study was to provide structural information about the regulatory domains of double-headed smooth muscle HMM, including the N terminus of the RLC. In previous work (22) we used a photo-cross-linker, BPIA, to label cysteines in a set of single-cysteine mutants of the RLC, exchange the labeled RLC mutants onto native HMM, and compare the ability of the protein to be photo-cross-linked in two states, unphosphorylated, and thiophosphorylated. For A23C, Q15C and Cys 108 we found cross-linking between the two RLC but only in the unphosphorylated state. This result means that the cysteine sulfur atoms can approach within ϳ8 -9 Å of the RLC of the other head, suggesting that they are oriented toward the partner head. For the other mutants, S59C and T134C, no cross-linking was observed for either state suggesting that these residues are probably not oriented toward the surface of the partner head and cannot approach the other head within 8 -9 Å at any time.
Here we have extended our previous photo-cross-linking studies (22) to determine sites on one RLC that can be crosslinked by a cross-linker attached to Cys 108 of the other RLC. Previously, we used the photo-cross-linker BPIA, but for the present work we have synthesized two new photo-cross-linking reagents, BBPIA and its respective maleimide derivative, which use the same photochemical mechanism as BPIA, but contain a biotin affinity tag to facilitate purification of crosslinked peptides. BBPIA-cross-linked peptides were purified by avidin affinity chromatography and characterized by Edman sequencing and MALDI-MS. BBPIA-labeled Cys 108 was found to cross-link to both 71 GMMSEAPGPIN 81 and to 4 RAKAKTT-KKRPQR 16 of the RLC in smooth u-HMM. Likely targeted residues within these sequences were identified. Both of these cross-links were between the two RLC. Using these data, data from our previous study, and atomic coordinates from various myosin isoforms (21,23,24), we have constructed a model of a regulatory domain in a double-headed molecule. The implications for the structural basis of the phosphorylation-mediated regulatory mechanism are discussed.

MATERIALS AND METHODS
IAA (Sigma) was recrystallized from hexanes before use. Reactions did not proceed smoothly without this precaution. Flash chromatography was performed with silica gel 60 (EM Reagents; 230 -400 mesh). TLC was performed with aluminum-backed TLC plates (5 ϫ 10 cm; 0.2 mm) coated with silica gel 60 F 254 (E. Merck, EM Separations). All reactions were protected from room light.
Synthesis of Compound 1 (see Fig. 1)-The following is a modification of the method of Gilbert and Rando (25). To a solution of biotin (Sigma, free acid, 142 mg, 0.58 mmol), 4,4Ј-diaminobenzophenone (161 mg, 0.76 mmol), diisopropylethylamine (106 mg, 0.82 mmol), and 1-hydroxy-7azabenzotriazole (120 mg, 0.88 mmol) in DMF (5 ml) were added to DCC (134 mg, 0.70 mmol) and stirred for 12 h. The reaction was quenched with H 2 O (50 ml), and the solution was extracted with n-butanol (3 ϫ 25 ml). The combined organic layers were washed with brine (2 ϫ 20 ml), dried using MgSO 4 , filtered, and concentrated to give a tan solid. The material was purified by flash chromatography (CH 2   Synthesis of Compound 2-Syntheses were performed in a standard 2.0-ml plastic microcentrifuge tube. DCC (12.3 mg, 58 mol) was added to a solution of recrystallized IAA (Sigma, 16.2 mg, 86 M) in THF (200 l), and the mixture was stirred for 30 min at 0°C. To this solution was added 1 (10.1 mg, 23 mol) in DMF (400 l) and stirred for 30 min. The progress of the reaction was monitored by analytical TLC (MeOH/CCl 3 (20:80)). The mixture was centrifuged in a microcentrifuge, the supernatant was removed, and the product was precipitated by addition of cold H 2 O (1.5 ml). The precipitated product was dissolved in DMF (400 l) and precipitated again by adding cold H 2 O (1.5 ml). The supernatant was removed to give 2 (11.8 mg, 86%) as a tan solid. The solid was dissolved in DMF (20 -30 mM)  Synthesis of the Tritiated Form of Compound 2-All steps were performed in a hood. IAA (17.6 mg, freshly recrystallized) was placed into a tared 3-ml flat-base conical vial equipped with a stir bar. A positive pressure micropipette (Rainin Microman) was used to transfer [ 3 H]IAA (3 mCi (PerkinElmer Life Sciences; ϳ100 -200 mCi mmol Ϫ1 ) from the original vessel to the conical vial with dry THF. THF was removed with a stream of argon. After adding hexanes (250 l), the vial was warmed to 60°C with an aluminum block hot plate until the IAA just dissolved. The vial was cooled to room temperature and then placed on ice for 30 min. The mother liquor was removed with a drawn glass pipette, and the crystals were washed with ice-cold hexanes. Crystals were dried overnight under a stream of argon and quantitated by weight. The coupling reaction was carried out as described above. Protein Preparations-SMM (26) from frozen chicken gizzards was used to prepare regulated HMM (5) as described (9) except that all buffers (except digest) contained fresh diisopropylfluorophosphate (100 M) to prevent post-quench digestion. DIFP converts in time to an unknown compound that causes HMM to lose regulation, giving a fast FTP turnover rate. Ammonium sulfate was added to 60% saturation; higher concentrations precipitated the protease. Native RLC was isolated from SMM (27). Extinction coefficients were: SMM, ⑀ 280 1% ϭ 0.56; HMM, ⑀ 280 1% ϭ 0.65; RLC, ⑀ 277 1% ϭ 0.37. Thiophosphorylation by SMM light chain kinase (5,22) was verified using 10% Tris-glycine gels. Samples (25-40 g) were precipitated with 3 volumes of cold acetone before. Sufficient sample buffer (8 M urea, 33 mM Tris-glycine (pH 8.6), 0.17 mM EDTA, 10 mM fresh DTT, bromphenol blue) was added to give 6 -7 mg/ml protein. These gels gave superior results to urea and/or urea-glycerol gels (27,28 (5,12,22). Unexchanged RLC was removed by gel filtration (5). [ 3 H]BBPIA-labeled u-HMM was dialyzed to 10 mM MOPS (pH 7.0), 0.1 mM EGTA, 0.05 mM DTT, centrifuged (350,000 ϫ g) at 4°C for 10 min, and filtered through a 0.45-micron filter prior to irradiation. Ten percent of the sample was retained, and 90% was irradiated as described (29). Samples were lyophilized, dissolved in 6 M GndHCl, 10 mM MOPS (pH 6.5), 1 mM DTT, and 1 mM EDTA, heated to 50°C for 30 min, and filtered. RLC-RLC dimers (40 kDa) were separated from uncrosslinked RLC and heavy chains by gel filtration in the above buffer (two TSK SW4000 columns (Tosohaas) in series at 0.2 ml/min at 25°C). The RLC-RLC dimer was identified on SDS gels and by scintillation counting.
Four Independent Experiments to Purify and Characterize Photocross-linked Peptides-In experiment 1, RLC-RLC dimer (11.4 nmol) was digested with 6 g of endoproteinase Asp-N (Roche Molecular Biochemicals sequencing grade) for 48 h in 0.1 M GndHCl, 50 mM Tris (pH 8.5) at 34°C, after which another 6 g of endoproteinase Asp-N was added and allowed to react at 4°C for another 48 h. NaCl (0.5 M) was added to the sample prior to loading onto a 1-ml neutravidin column (Pierce) equilibrated in 0.1 M GndHCl, 100 mM Tris (pH 8.5), 0.5 M NaCl, 1 mM DTT, and 1 mM EDTA. The column was washed with the above buffer and 25 mM ammonium bicarbonate, 1 mM DTT. During this wash, 40 -45% of the loaded tritium eluted from the column, as was found for all experiments. Further binding could not be achieved with fresh avidin. Biotinylated peptides were eluted with 70% formic acid as buffers recommended by the manufacturer were ineffective. The sample (2.2 nmol) was lyophilized and applied to a C8 reversed-phase column (Brownlee Aquapore, narrowbore) and eluted with a linear gradient of 0.1% trifluoroacetic acid/H 2 O versus 0.1% trifluoroacetic acid/80% ACN. Radiolabeled peptides (2.4 nmol) eluted over one-third of the gradient. Fractions were lyophilized to near dryness and analyzed by MALDI-MS (Table I).
In experiment 3, RLC-RLC dimer (5 nmol) treated with Asp-N as for experiments 1 and 2. EDTA (1 mM) was added to inhibit Asp-N; Glu-C (Roche Molecular Biochemicals sequencing grade; 1/100 w/w) was added and allowed to digest at 25°C for 3 days. The digest was filtered and loaded onto a 0.5-ml monomeric avidin column (Pierce, Ultralink), and the column was washed as described in experiment 2. The biotinylated peptides were eluted with 70% trifluoroacetic acid. After lyophilization, the sample was dissolved in 200 mM ammonium bicarbonate and clarified by centrifugation. Acylaminoacyl-peptidase (30 g; Roche Molecular Biochemicals sequencing grade; E.C. 3.4.19.1), EDTA (1 mM), and ␤-mercaptoethanol (1 mM) were added, and the digest was allowed to proceed for 24 h at 37°C. The sample was reapplied to a fresh monomeric avidin column and eluted as previously described. After repeated lyophilization the sample was resuspended in 0.1% trifluoroacetic acid/65% ACN to give a precipitate too large to be due to labeled peptides. Therefore the sample was repeatedly centrifuged, and the pellets were washed with 0.1% trifluoroacetic acid/65% ACN. The combined supernatants were lyophilized and treated with Zip-Tips (Millipore; C8) in 0.1% trifluoroacetic acid/10% ACN. Complete Zip-Tip binding required an overnight incubation. Peptides were released from 5 Zip-Tips with 65% ACN (see Fig. 7A). An second sample was prepared by treating the residual from the previous sample with more Zip-Tips in 0.1% trifluoroacetic acid and eluting with 65% ACN (see Fig. 7B). In experiment 4, samples were prepared as in experiment 1, except that proteolysis was with trypsin (1/100 w/w) overnight at 25°C.
Mass Spectral Analysis-Mass spectra were obtained on an Applied Biosystems Voyager DE/RP MALDI instrument in linear positive ion mode. Laser power was 15% above threshold. The matrix was ␣-cyano-4-hydroxy-cinnamic acid prepared as a saturated solution in 50%/50% (v/v) ACN/water with 0.25% trifluoroacetic acid. Data were analyzed using Kaleidograph and MS programs from the Expasy website, Find-Pept Tool and PeptideMass. Observed masses were identified by comparison to calculated masses generated by summing the masses of photo-cross-linker, parent peptides, and predicted target peptides. We subtracted 18 mass units, since 18 mass units can be lost from benzophenone cross-linked peptides treated under similar conditions (30). Most observed masses matched to this in silico data set.
For experiment 1, three spectra from three different fractions from the C8 reversed-phase column were analyzed on at least two different days. Values reported are the average Ϯ standard deviations (Table I). An external calibration was performed with standard peptides (Sequezyme Calmix 2 from Applied Biosystems). For experiment 3, masses [M ϩ H] ϩ were detected in at least 3 of 5 different measurements made on different days on the same sample. Values reported are the average Ϯ standard deviations (Table II). Calibration with internal standards resulted in two problems that prevented simultaneous observation of both unknown masses and standards. First, the sample peaks were suppressed to below detection levels, and second, a sample milieu effect caused the measured masses of peptide standards to shift to higher masses (⌬ ϳ4 -5 Da) than was found for standards in the absence of sample. Therefore an external calibration was performed with standard added to the sample. Then, data were acquired on the sample, alone and the external calibration was applied.

Synthesis of Photo-cross-linkers-We have developed two
new trifunctional photo-cross-linkers (Fig. 1, compounds 2 and  3) to facilitate purification of photo-cross-linked peptides. They are derivatives of the widely used benzophenone chromophore that forms C-C bonds with polypeptides upon irradiation with UV light (31). We have previously described the synthesis of the sulfhydryl-reactive photo-cross-linker, BPIA, including a tritiated form to facilitate isolation of cross-linked peptides (29). The two new probes contain three chemical functionalities; an iodoacetyl (2, BBPIA) or a maleimide (3, BB-maleimide) to selectively react with the cysteine thiolate anion, the photoreactive benzophenone, and the biotin affinity tag to facilitate purification of parent-target covalently cross-linked peptides. We describe easy, economical syntheses for which training in synthetic organic chemistry is not required. The synthesis of 2 provides for safe handling of tritiated compounds with the tritium incorporated on the last step.
Regulation of Labeled HMM- Fig. 2 shows that the BBPIAlabeled HMM (sample D) was not as perfectly regulated as untreated HMM (sample A), HMM with RLC labeled with the less bulky BPIA (sample B), or HMM exchanged with unlabeled RLC (sample C). The BBPIA-labeled tp-HMM had a normal activity, but the unphosphorylated form was more active than normal. This suggested that the presence of BBPIA at Cys 108 may alter the functional properties of u-HMM to some extent. However, in this case there remains a sufficient level of regulation to justify continuing with the study. A more complete assessment of regulation of this and other similarly labeled HMMs using the more sensitive single-turnover approach (5,14) is in progress.
Cross-linking Occurs between the Two RLC-Since the regulatory properties of the BBPIA-labeled u-HMM remained largely functional, we expected the photo-cross-linking pattern seen for BBPIA-HMM to be similar to that previously observed for BPIA-HMM (shown to be regulated; (22)). As expected, irradiation caused the formation of RLC-RLC cross-linked dimers but only in the unphosphorylated state (Fig. 3). To determine the site(s) on the RLC into which the BBPIA was photo-inserted (target), the irradiated sample was denatured and passed over an analytical gel filtration column to isolate the RLC-RLC dimers.
Identification of the Target Peptides-We performed four independent experiments to purify and characterize parenttarget cross-linked peptides. In experiment 1, an Asp-N digest of RLC-RLC dimers was applied to an avidin column, and the eluate was separated on a reversed-phase column. Table I shows the peptides in the column fractions identified as targets by MALDI-MS. Peptides 1-4 matched the N terminus. To verify this match, a sample enriched in the N-terminal peptide was treated with acetic anhydride (32) to acetylate lysines. A new set of peaks emerged (data not shown) consistent with 5-6 acetylations (addition of 210 -252 mass units), as predicted for the N terminus. Peptides 5-14 were members of a family of peptides spanning residues 70 -98, and peptides 15-21 spanned residues 49 -51.
In experiment 2, avidin-purified biotinylated peptides were gel-filtered (Fig. 4). For the unirradiated sample, most of the radioactivity eluted in fractions 35-40 that contained parent peptides. For the irradiated sample, the major portion of the radioactivity eluted earlier in the profile, as would be expected for the larger parent-target cross-linked peptides. Edman sequencing was performed for 15 cycles on fractions 6 -8 from Fig. 4 (irradiated sample), and two major peptide sequences were observed (Fig. 5); the parent peptide that was also detected in the MALDI-MS spectra from experiment 1 ((M ϩ H) ϩ ϭ 1634.92), and the target sequence 66 DEYLEGMMSEA-PGPI 80 . . . were found in approximately equal quantities, strongly suggesting that the peptides were cross-linked together. The acetylated N terminus did not sequence. The sequence 49 DKE 51 was not detected in Fig. 5, but it was observed in the MALDI-MS (peptides 16 -21 from Table I). Together these data suggest that, while present, the sequence 49 DKE 51 was in minor amounts relative to the target sequence 66 DEYLEGMMSEAPGPI 80 . . . . Experiment 3 was performed to further define the crosslinked regions identified in experiments 1 and 2. An avidinpurified Asp-N/Glu-C/acetyl aminopeptidase digest of RLC-RLC dimers was treated with Zip-Tips without further purification. Two samples were obtained. Edman sequence analysis of the first sample (Fig. 6A) identified the target sequence starting with Gly 71 that is contained within the sequences identified from experiments 1 (Table I) and 2 (Fig. 5). Table II shows the MALDI-MS data, which is consistent with the sequencing data. Edman sequence analysis for the second sample (Fig. 6B) shows that the major target peptide was the deacetylated N terminus as the sequences starting with Gly 71 was not detected even though they were detected in the MALDI-MS spectra (Table II). Data from experiment 3 are consistent with data from Table I and Fig. 5 and suggest that 71 GMMS 74 , 78 GPIN 81 , and the first 25 residues from the N terminus are targeted. From Fig. 5 it appears that Ser 74 is a cross-linked amino acid, but it is not definitive because the yield for serine is often low. Methionine is often targeted by activated benzophenone, presumably at the methylene carbon adjacent to the sulfur (31). In this case the sequencing data suggest strongly that methionine is not targeted. The targeted residue within the 78 GPIN 81 was not identified, but the sequencing data suggest that it is Asn 81 .
In experiment 4, a tryptic digest was performed to identify targeted residues from the N terminus by MALDI-MS (Fig. 7). Other previously identified targeted regions were not observed in this experiment because the peptides were too large. Five different peptides were matched to a parent mass plus an arginine residue. Arg 4 is the only single arginine predicted from a tryptic digest and is therefore a targeted residue. A lysine residue was also targeted and is most likely Lys 12 , but Lys 150 was also possible. The dipeptide AK could be residues 5-6 and/or 7-8. By combining the information from Figs. 6B and 7, we can be reasonably sure that Ser 1 , Ser 2 , Ala 5 , Ala 7 , Thr 9 , and Gln 15 are not targeted. Lys 3 , Lys 8 , and Thr 10 are potentially targeted, and Arg 4 , Lys 6 , Lys 11 , Lys 12 , Arg 13 , Pro 14 , and Arg 16 are probably targeted. Therefore, the region 4 RAKAKTTKKRPQR 16 can approach within cross-linking distance of benzophenone. Products of Protein Photo-cross-linking with Benzophenone Derivatives-Our study has revealed new mechanistic information about benzophenone photochemistry. It is known that the product of benzophenone cross-linked to the ␣-carbon of glycine can dehydrate (31) and that peptides containing a targeted methionine exist in a form that is 18 mass units lighter than expected (30). Most of our cross-linked peptides were 18 mass units lighter than predicted (31). For example, peptide 1-25 was found in the dehydrated form (Table I). This peptide does not contain glycine and we have no evidence for cross-linking of methionine. We identified Lys and Arg as targeted amino acids in this region. This suggests that these amino acids are also prone to dehydration as may be Ser and Asn (Tables I and II). We do not know if this dehydration occurs during the photolysis   (Fig. 2). In contrast, labeling of Q15C and A23C with BBPIA significantly disrupted regulation (data not shown), whereas regulation was intact with BPIA (22). This suggests that these residues are positioned in critical locations important for regulation.

Summary of Target Peptides and Residues-By
Motor Domains Are Not Required for the Phosphorylationdependent Structural Changes-We have previously shown (22) that the HMM cross-linking pattern was not altered by ADP or ATP. And we showed that RLC-RLC cross-linking occurred in an unphosphorylated construct lacking motor domains and that phosphorylation abolished the cross-linking. Therefore, it appears that our experiments are sensing a phosphorylation-induced conformational change in the RLC that does not require motor domains. Similarly, Rosenfeld et al. (33) showed that RLC rotational motion in a construct lacking the motor domains is increased by RLC phosphorylation. None of the interactions in Table III were observed in tp-HMM in any nucleotide state.
Structural Model of u-HMM Regulatory Domain Structure- Table III summarizes data we considered to develop a computational model of the u-HMM regulatory domain. The general features of the model are shown in Fig. 8, the details of which will be published elsewhere. It describes the relative orientation of the regulatory domains during the cross-linking event.
To build this symmetrical model both benzophenone moieties (one from each RLC Cys 108 ) were locked within 1.4 Å of Gly 78 of the RLC from the other head, and the structure was adjusted to avoid Van der Waal's overlap and to agree with Table III other symmetrical models were consistent with the data in Table III. The two RLC are side by side in an antiparallel manner. Fig. 8 is not meant to indicate specific RLC interactions, but merely suggests their relative orientation and separation during the cross-linking event. The two Phe 25 (the most N-terminal residues for which we have atomic resolution data) are close together at the interface between the two RLCs in the center of Fig. 8A and in the lower portion of the RLCs in Fig.  8C.
To test our model, we performed an additional cross-linking experiment. Irradiation of BPIA-labeled T83C HMM formed RLC-RLC dimers, only in the unphosphorylated state, and nucleotide did not appear to affect the result (data not shown). This result is consistent (Fig. 8) as Thr 83 is positioned at the top of the groove between the two RLC within cross-linking distance to the partner RLC.
Model Predicts Location of RLC N Terminus-Our model allows us to predict the position of the first 24 RLC residues, a region for which there is no atomic resolution data. This area is of particular interest because it contains the critical regulatory phosphorylation site at Ser 19 . First, we generated an independent model of these first 24 residues by using secondary structure and disorder prediction tools and analysis of kinase structures with bound substrate peptides. Our model predicts that Ser 1 to Thr 10 or Lys 11 forms a helix followed by a disordered region that cannot be assigned secondary structure. This latter portion could maximally extend ϳ32 Å from Lys 11 to Met 24 . Fig. 8B shows the independently modeled N-terminal 24 residues (gray) from the lower RLC (blue) placed onto the regulatory domain model in a manner consistent with our data. We have shown that residues within 4 RAKAKTTKKRPQR 16 can approach within 8 -9 Å of the Cys 108 sulfur of the partner RLC (Fig. 6B, Tables I and II, and Fig. 7). All the probable and potentially targeted residues found within residues 1-11 (Fig.  7), a region that we have predicted to be a helix, are located on one face of such a helix. This suggests that residues 1-11 may be folded into a helix. However because so many residues within this region were targeted, the helix and the benzophenone are not highly restricted in space relative to one another. Much of the N terminus including Ser 19 (red) lies in a groove between the two RLC. This placement may explain the fact that we observed altered regulation in A23C and Q15C mutants labeled with a bulky group. Previous experiments with the less bulky BPIA-labeled Q15C and A23C on u-HMM (22) showed that RLC-RLC dimers were formed upon irradiation, but the site of labeling was not identified. This model places these two residues, flanking the Ser 19 (pink), within crosslinking distance of the partner RLC.
As seen in Fig. 8B, the N terminus interacts with the Cterminal domain of the partner RLC. Several studies have shown that elements of the C-terminal domain are crucial to proper regulatory properties (34 -36). It is tantalizing to suggest that the N terminus is positioned to control the conformation of the heavy chain helix to which the two RLC bind. It has been previously suggested that Ca 2ϩ binding in up-regulated scallop myosin may play a role in stabilizing the regulatory domain through tightened interactions between the RLC, ELC, and heavy chain (21). The N terminus may also be strategically placed to control the interface between the two RLC and the attitude and flexibility of the linker connecting the two domains of the RLC. This latter region is known to be important to the regulatory mechanism (37) and has been noted to be different between the skeletal and scallop structures (21).

Model Predicts Phosphorylation-induced Motion of RLC N Terminus-Phosphorylation must result in movement of
Cys 108 out of cross-linking distance to the 71 GMMSEAPGPIN 81 and 4 RAKAKTTKKRPQR 16 targets of the partner RLC as photo-cross-linked RLC-RLC dimers were not observed in tp-HMM. It may be that movements of the two targets are consequences of one another. Our model shows that cross-linking could occur from Gln 15 and Ala 23 to the partner RLC, in agreement with Table III. However, upon phosphorylation, neither Gln 15 or Ala 23 can cross-link to the partner RLC suggesting a phosphorylation-induced movement of a significant portion of the N terminus. This could occur if the phospho-serine folded back the N terminus upon itself through coordination with its or other basic residues. In our model, the phosphorylated serine is in a strategic position to dramatically alter the interactions of the N terminus with such a coordination. It is interesting that phosphorylation sites are often in regions of known disorder (38), and we predict that the region around Ser 19 is disordered. Disorder is also common in protein-protein interactions that are part of regulatory switches (38). The role of phosphorylation may be to transition this region of the N terminus a Observed masses from two Zip-tip treated samples from experiment #3. All peptides were found in both samples ( Fig. 6A and 6B), except the first two which were found only in the sample from Fig. 6B Table I. Parent sequences were the same as in Table I, except that VIRNAFAC(bz)F ϭ 1519.83 was also included because it was observed in the spectra.
b Hydrated form of the peptide.

FIG. 7. Peptides in N-terminal region identified by MALDI-MS (experiment 4).
The first 16 residues of the RLC sequence are shown. Each line spanning the residues represents an independent measurement of a mass matching that region. Darker lines represent more than one observation. Dotted lines mean that the match could be in more than one place; either of two AK positions, respectively. The position of the lines above the sequence is not important.
from an extended (shown in our model) to a folded structure.
Does Phosphorylation Dissociate Regulatory Domains?-Our model is consistent with previous work showing that isolated u-SMM regulatory domains dimerize (33) and further suggests that interactions between the two RLC provide the stabilizing forces. However, phosphorylation has little effect upon the stability of the dimers (33). Our results using zero-length crosslinking were consistent with these findings. 2 Rosenfeld et al. (33) showed that the rotational motion of a fluorophore on Cys 108 in a construct lacking motor domains is increased by RLC phosphorylation. However, the rotational motion was slower than expected for mobile regulatory domains moving independently of the rods. In light of these findings, the transition we observe from cross-linking in the u-HMM to no crosslinking in the tp-HMM may be the result of local motions that do not destabilize RLC interactions. Our data, which shows that the N terminus moves out of cross-linking distance to Cys 108 , may be sufficient to explain the changes in rotational motion observed by Rosenfeld et al. (33).
Implications of the Model for S2 Structure-The distance between the C termini of the heavy chains at the head-rod junction is 47 Å in our model, suggesting that the rod cannot adopt a coiled-coil up to the head-rod junction. We were unable to find any satisfactory models with such a coiled rod conformation. Our model is consistent with the idea that optimal mechanical performance of SMM may require the rod to uncoil near the heads (39).
Comparison to Other Models-The data from Table III do not fit a model developed from a three-dimensional reconstruction of frozen hydrated expressed smooth u-HMM on a lipid bilayer (40). Neither can RLC Cys 108 reach RLC residues 71-81 from either head, nor are inter-head Cys 108 to N termini interactions likely without large rearrangements. Since the rod interacts with the lipid bilayer and the heads are arranged on top of the rod, it may be such an interaction stabilizes a structure not found in solution.
We previously reported (22) that our cross-linking data did not fit a computed model (24) of scallop myosin regulatory domains (atomic resolution data from Xie et al. (20) in the presence of Ca 2ϩ ) attached to a model of the C-terminal portion of the ␣-helical coiled-coil S2 (tail) domain. The data presented here reinforce that conclusion. To generate our model from the scallop model, the coiled-coil must be unwound to allow the opposite faces of the RLC to interact. We are unaware of any structural data in the scallop system that supports the computed model. SMM 10 S Studies Support the Model-The 10 S or folded conformation of u-SMM, like u-HMM, is kinetically inactive. Several studies have shown that the N terminus of the RLC is required to form a 10 S SMM structure (41,42) with the tail folded onto the heads. However, myosin containing an N-smooth/C-skeletal RLC chimera, which contains a complete N terminus, also fails to fold to 10 S (34). This suggests that an interaction of the N terminus with the C-terminal domain may be critical to generate a binding site on the heads for the tail in 10 S. Since single-headed myosin fails to form the 10 S (29, 41) 2 D. Pouchnick and C. R. Cremo, unpublished results. this is another regulatory domain. A, the C termini of the heavy chain helices (residue Leu 837 ) are extending out of the page toward you. The distance between these two terminal residues is 47 Å. Benzophenone is shown in yellow space-filling attached to Cys 108 . The benzophenone on the right is attached to the red RLC. The RLC sequence 71 GMMSEA-PGPIN 81 is shown in green ribbon. The ␣-carbon of Gly 78 (space-filling) is Ͻ1 Å from the carbonyl carbon of the benzophenone. Ser 59 is shown in white ribbon. T134C is also in white ribbon, but is mostly obscured. Phe 25 of each RLC is found in the center of the structure, near the linker (residues Gly 95 -Pro 98 ) between the RLC N-terminal and the C-terminal domains. B, same orientation as A and illustrates the predicted orientation of the blue RLC N terminus (residues 1-24; shown in gray). The N terminus has been positioned manually. The equivalent region of the red RLC is not shown for clarity. Ser 19 is pink and 10 TKKRP 14 is shown in cyan. Cys 108 is shown in yellow ribbon. C, B rotated 90°toward you about the x-axis. Heavy chains now extend downward. Thr 134 is visible as white ribbon in the upper portion of both RLC. Ser 59 is most clearly seen in the red RLC. All modeling was performed on a Silicon Graphics, Inc. (Mountain View, CA) Octane work station using Insight II (Version 2000; Accelrys). it appears that elements from both heads are required to generate this binding site. We have previously shown (29) that BPIA-labeled Cys 108 photo-cross-links to the tail in the 10 S conformation. All these data point to a structure where the RLC N terminus interacts with the RLC C-terminal domain on the partner head and that this interaction is required for tail interaction near Cys 108 . These data taken together are highly consistent with our model, which places Cys 108 in the RLC C-terminal domain close to the N terminus of the partner RLC. We propose that an interaction similar to that in Fig. 8 is important for down-regulation of SMM.