Direct Photoaffinity Labeling of Gizzard Myosin with [3H]Uridine Diphosphate Places G1uls5 of the Heavy Chain at the Active Site*

The active site of chicken gizzard myosin was labeled by direct photoaffinity labeling with [3H]UDP. [3H] UDP was stably trapped at the active site by addition of vanadate (Vi) and Co2+. The extraordinary stability of the myosin*Co2+*[3H]UDP*Vi complex (tu > 5 days at 0 “C) allowed it to be purified free of extraneous [3H]UDP before irradiation began. Upon UV irradiation, >60% of the trapped [3H]UDP was photoincor- porated into the active site. Only the 200-kDa heavy chain was labeled, confirming earlier results (Maruta, H., and Korn, E. (1981) J. Biol. Chem. 256,499-502) using [3H]UTP. Extensive tryptic digestion of photo- labeled myosin subfragment 1 followed by high per- formance liquid chromatography separations and re-moval of nucleotide phosphates by treatment with al- kaline phosphatase allowed two labeled peptides to be isolated.

Sequencing of the labeled peptides and radioactive counting showed that GIu'~' was the residue labeled.
Since UDP is a "zero-length" cross-linker, Glu"' is located at the purine-binding pocket of the active site of smooth myosin and adjacent to the glytine-rich loop which binds the polyphosphate portion of ATP. This Glu residue is conserved in smooth and nonmuscle myosins and is the same residue identified previously by [3H]UTP photolabeling in Acanthamoeba myosin II (Atkinson, M. A., Robinson, E. A., Appella, E., and Korn, E. D. (1986) J. BioZ. Chem. 261, 1844-1848.
A knowledge of the topology of the active site of myosin is a necessary first step to provide a molecular understanding of the mechanism of ATP hydrolysis and mechanical force generation in muscle. Labeling skeletal Sl' with photoaffinity analogs of ATP has proven to be a useful approach to identify active site amino acids (Cremo et al., 1989;Mahmood et al., 1989;Yount et al., 1987, Okamoto andYount, 1985 moto et al., 1986) has shown that elements of the l7-kDa essential light chains are located at the active site in smooth muscle myosin. In contrast, only residues of the 200-kDa heavy chain in skeletal myosin have been placed at the active site (Cremo et al., 1989;Mahmood et al., 1989;Sutoh, 1987;Sutoh et al., 1986;Okamoto and Yount, 1985;Szilagyi et al., 1979). Thus, even though smooth and skeletal myosins are superficially the same (e.g. molecular mass, subunit composition, and gross morphology), the active site region where ATP is bound and cleaved, appears to be different.
Although ATP photoaffinity analogs have been useful tools to identify amino acids at the active site of myosin, the photoreactive group is often bulky and may identify residues outside the active site. One approach to remedy this problem is to use native photoreactive nucleoside triphosphates as "direct" photoaffinity probes. This type of direct photoaffinity labeling using radioactive nucleotide substrates has been utilized to examine a broad variety of nucleotide binding proteins including aminoacyl-tRNA synthetases (Yue and Schimmel, 1979), DNA polymerases (Pandey et al., 1987;Biswas and Kornberg, 1984;Abraham and Modak, 1984;Hillel and Wu, 1978), ribonucleotide reductase (Kierdaszuk and Eriksson, 1988), recA binding protein (Banks and Sedgwick, 1986), and /3-tubulin (Linse and Mandelkow, 1988;Nath and Himes, 1986).
Uracil contains a photoreactive 5,6 double bond and will form photoadducts with nearby amino acids when exposed to UV radiation (Smith, 1969). This property makes UTP a useful photoaffinity analog of ATP and allows the structure of the purine binding pocket of the active site of myosin and other nucleotide binding proteins to be investigated. ["HIUTP has been used effectively to label myosin from turkey gizzard and Acanthamoeba myosin II to show that elements of the heavy chain provide part of the active site (Maruta and Korn, 1981). Subsequent work with Acanthamoeba myosin II (Atkinson et al., 1986) has identified Glu"' of the heavy chain as a residue at the active site. An inherent problem with these studies was that the high energy UV light necessary for activating UTP (or other nucleotides) rapidly inactivates myosin so that it was not possible to correlate inactivation with the covalent photolabeling of the enzyme. These conditions make it difficult to rule out possible nonspecific labeling. One method to alleviate this possible complication is to "trap" a nucleoside diphosphate molecule at the active site with orthovanadate (Vi) and divalent metals (Goodno, 1979). Replacing Mg'+ with Co'+ in the trapped complex will suppress the known photooxidation of active site residues by vanadate Grammer et al., 1988). The vanadate trapping approach allows the myosin . Co'+. NDP . Vi complex to be isolated before irradiation begins, thus insuring specific photolabeling of only active site residues. In this work we report that [3H]UDP, when trapped at the active site with vanadate and Co'+ and irradiated with UV light, appears to react exclusively with G~u'*~ of the gizzard myosin heavy chain. The results place Gluls5 in the adenine binding pocket of the active site, adjacent to the glycine-rich loop which is thought to bind the polyphosphate portion of ATP. This particular Glu residue is conserved in a variety of smooth and nonmuscle myosins, yet is replaced by Val in skeletal myosin. These results, in combination with other photoaffinity labeling studies of the active site of myosin, provide further evidence for differences in the structure and composition of the active site in smooth and skeletal myosins.  Goodno (1982). All buffers contained 0.025% sodium azide (Sigma) as a preservative. Enzyme Preparations-Dephosphorylated smooth muscle myosin was isolated from fresh chicken gizzards as described by Ebashi (1976) and stored in 50% glycerol at -20 "C. K'EDTA ATPase activities were quantified as previously described by Wells et al. (1979) (Maruta and Korn, 1981).3 Co'+ was used because it quenches the photoreaction of vanadate itself with active site residues which occurs in the presence of Mg*+ ). Fig. 1  of bound [3H]UDP (Fig. 2, inset). Gizzard myosin.Co'+.
[3H] UDP. Vi complexes have half-lives greater than 5 days at 0 "C, indicating that only a small amount (~1%) of bound [3H] UDP leaks from the active site in the short time necessary to purify and irradiate the complex.
Analysis of the Location of pH]UDP-The photolabeled myosin was analyzed by SDS gel electrophoresis to identify the labeled subunits (Fig. 3). All the radioactivity was associated with the 200-kDa heavy chains. Neither the 20-kDa regulatory light chains nor the l7-kDa essential light chains were labeled. These results indicated that the photoreactive 56 double bond of the pyrimidine ring in [3H]UDP is located in close proximity to a portion of the heavy chain in the active site.
To further localize the photoincorporated [3H]UDP within the 200-kDa heavy chain, photolabeled Sl (see below) was partially digested with trypsin and analyzed on a 12% SDS polyacrylamide gel (Fig. 4) digested Sl was then analyzed by electrophoresis in a 12% SDS-polyacrylamide gel with a 5% stacking gel. Gel lanes were sliced, solubilized, and counted as described in Fig. 3. tic fragment, a region previously identified to contain the ATP binding site (Maruta and Korn, 1981).
Production and Isolation of pH]UDP-labeled Peptides (Scheme I,-Photolabeled gizzard myosin was digested with papain, and photolabeled Sl was isolated (see "Experimental Procedures").
The Sl was extensively digested with trypsin and peptides separated on a Brownlee RP-300 semipreparative column (see "Experimental Procedures" and Fig. 5). Two radioactive peaks (I and II) were independently pooled for further purification.

Isolation of Peptide I-Peptide
I was very hydrophilic, as it eluted in the void volume on the reverse phase semipreparative column (Fig. 5). Peptide I did not bind to an ODS (C,,) column at pH 2 or 6 and was irreversibly retained on an anion exchange column (AX-lo, Varian; 5 mM KHIPOI, pH 5.0, and a gradient of O-O.5 M KCl). In order to make peptide I more hydrophobic, the phosphate groups of the covalently bound [3H]UDP were removed with bacterial alkaline phosphatase (see "Experimental Procedures") and the digest was purified on an ODS column at pH 2. Radioactive peptides eluting in the void volume (Fig. 6) were treated with bacterial alkaline phosphatase again and repurified. The radioactive peptides eluting at 46 min (Fig. 6) were collected and further purified at pH 6.0 (Fig. 7) followed by a desalting step (Fig. 8) to prepare the peptide for sequencing.
Isolation of Peptide II-Peptide II, which eluted at 25 min on the reverse phase semipreparative column (Fig. 5), was further purified first at pH 6.0 (Fig. 9), then at pH 2 (Fig. 10). The desalting step (Fig. 10) was necessary to remove inorganic phosphate, which is known to be an inhibitor of alkaline phosphatase (Reid and Wilson, 1971). Radioactive peptide II was then treated with bacterial alkaline phosphatase (see "Experimental Procedures") and purified at pH 2 ( Fig. 11) to prepare the peptide for sequencing.
Sequence Analysis of Peptides I and IZ-The amino acid sequence of peptide I was found to be T(X)NTK (Fig. 24). The residue at cycle 2 corresponds to G1u's5 in the sequence of the gizzard myosin heavy chain (Yanagisawa et al., 1987). Radioactivity was first released at cycle 2, further indicating that G1ulR5 contained the [3H]uridine label. The sequence of peptide II was TGESGAGKT(X)NTK (Fig. 12B), where the residue at cycle 10 corresponds to Gl@ in the amino acid sequence of the gizzard myosin heavy chain (Yanagisawa et al., 1987). The radioactivity detected after cycle 10 was most likely due to the continued partial extraction of the [3H]uridine-glutamic acid-phenylthiohydantoin adduct from the polybrene coated membrane during each subsequent cycle.

DISCUSSION
The goal of this work was to identify and locate active site amino acid residues of gizzard myosin which are photolabeled by [3H]UDP. Vanadate trapping methodology (Goodno, 1979(Goodno, , 1982 was employed to stably bind [3H]UDP at the active site and allow the myosin.Co'+. L3H]UDP.Vi complex to be purified away from free [3H]UDP. The addition of Co*+ in the trapped complex is known to prevent the photooxidation of active site residues by vanadate Cole and Yount, 1989). Subsequent irradiation of the purified complex resulted in approximately 60% incorporation of trapped [3H]UDP into the heavy chain (Fig. 2). Following extensive trypsin digestion of the labeled Sl, two major [3H] UDP-labeled peptide pools were each dephosphorylated with bacterial alkaline phosphatase and purified further by reverse phase HPLC (Scheme I). This procedure allowed the effective separation of labeled peptides from contaminating peptides in one or two additional HPLC steps. In addition, the absence of phosphate groups on the photoincorporated [3H]uridine allowed the labeled residue, G1uls5, to be positively identified by its radioactivity during the appropriate cycle of peptide sequencing.
The phenylthiohydantoins of phosphorylated amino acids, e.g. phosphoserine (Murakami et al., 1990) are known to be poorly extracted by the organic solvents used in automated peptide sequencing. Hence, phenylthiohydantoins of amino acids photolabeled with nucleotides or nucleotide derivatives will likely never be identified because of the strong affinity of the phosphates for the positively charged polybrene coating on the filters. Peptide I had the sequence Thr-(X)-Asn-Thr-Lys which corresponds to residues 184-188 of the gizzard myosin heavy chain (Yanagisawa et al., 1987). Peptide II was similar to peptide I except that it contained an eight residue NH?-terminal extension giving a peptide corresponding to residues 176-188. In sequencing both peptides, the cycle corresponding to GUI? gave no identifiable phenylthiohydantoin-derivative and contained the [3H]uridine label (Fig. 12 is located. These results are unexpected based on inspection of the amino acid composition around G1ula5. This residue is conserved in all smooth and non-muscle myosins and is surrounded by a predominance of hydrophilic residues (Table I). The nature of this peptide region suggests it is located at the surface of the myosin head. In contrast, previous photoaffinity labeling studies with skeletal myosin (Okamoto and Yount, 1985;Sutoh, 1987) have implicated the hydrophobic region around Trp13" as providing the adenine binding site. It may be that the methylene groups of the side chain of G~u"~ lie next to the adenine ring with the y-carboxyl group hydrogen-bonded elsewhere.
Direct Photoaffinity Labeling of Gizzard Myosin with UDP in that in skeletal myosin a Val replaces Glulss (Table I), and Val may simply be less reactive towards photoactivated UDP than Glu. Alternatively, it may be the composition of the active sites of skeletal and smooth muscle myosins are different, reflecting their different modes of regulation.
Photoaffinity labeling studies with the photoprobe NANDP provide additional evidence that suggest the active site topologies of smooth and skeletal myosins are different. NANDP contains a photoreactive nitroaryl-azido group which binds in the adenine subsite of the active site. The azido group of NANDP is located on the end opposite the diphosphate and should label residues deep within the adenine binding pocket. NANDP, when trapped and photoincorporated at the active site of skeletal Sl, labeled Tq? of the heavy chain (Okamoto and Yount, 1985). No labeling of the light chains was observed. Gizzard myosin, however, is labeled by NANDP in both the heavy chain and the 17-kDa essential light chain (Okamoto et al., 1986). This finding demonstrates further differences in active site topology between smooth and skeletal myosins.
In contrast to the above results, photoaffinity labeling studies with the photoprobe Bz*ATP indicate that the heavy chain conformation near the ribose binding site for ATP in both smooth and skeletal myosins are essentially identical. Bz*ATP contains a photoreactive benzophenone group esterified to the 2' or 3' hydroxyls of the ribose ring of ATP and will label residues near the ribose binding subsite of the active site. BzzATP has been shown to label Se? of the heavy chain in skeletal myosin (Mahmood et al., 1989) and Pro324 of the heavy chain in gizzard myosin (Cole and Yount, 1990). Both of these residues are located within the central 50-kDa tryptic fragment of their respective Sl heavy chains and suggest that the gross morphology and folding of this part of the heavy chain around the ATP binding site is similar in these two myosins.
The photochemistry of nucleotide bases with proteins remains largely unknown.
However, we expect that the $6 double bond of uracil is the photoreactive center of UDP. The 5,6 double bond of pyrimidine bases is known to undergo a ?r -+ K* transition when exposed of UV light and form covalent photoadducts with nearby molecules (Wang, 1976;Smith, 1969). In a few cases, the photoproducts of direct photoaffinity labeling have been isolated and characterized.
For example, photolabeling studies of diptheria toxin and Pseudomonas exotoxin with ['%]NAD have identified glutamic acid at the active sites of these proteins (Carroll and Collier, 1987;Carroll et al., 1985). It is also not known if amino acids add preferentially at the C" or C6 position of the pyrimidine ring. Kierdaszuk and Eriksson (1988) have isolated and characterized the major photoadduct of ribonucleotide reductase and dTTP and have shown it to contain a covalent bond between the thiol of CysZg2 and the 5-methyl group of thymidine.
[cr-"'PJTTP is only one-sixth as efficient as UDP in photolabeling gizzard myosin," a result which suggests that the C" position of uracil may be the major site of reaction with Gl@". The conformation of ["H]UDP bound at the active site may explain why a Glu residue next to the comparable glycine rich loop of gizzard myosin becomes labeled. The orientation of pyrimidine nucleotides about the glycosyl bond is predominantly anti (Saenger, 1984). This conformation exhibits minimal steric hindrance between the ribose and uracil rings and places the photoreactive 5,6 double bond over the ribose, closer to the diphosphates.
In contrast to NANDP, UDP should label an area in the adenine pocket that is closer to the phosphate binding region. Hence, even though G1uls5 is directly adjacent to the GESGAGKT sequence known to be at the triphosphate binding site (Cremo et al., 1989), it could still be close enough to add to the 5,6 double bond of the trapped UDP. Whether G~u"~ also plays an important role in binding to the adenine ring of ATP awaits the solution of the crystal structure of the gizzard Sl.ADP complex at atomic resolution.