Sites of Interaction between Kinase-related Protein and Smooth Muscle Myosin*

Kinase-related protein, also known as KRP or telokin, is an independently expressed protein product derived from a gene within the gene for myosin light chain kinase (MLCK). KRP binds to unphosphorylated smooth muscle myosin filaments and stabilizes them against ATP-induced depolymerization in vitro. KRP competes with MLCK for binding to myosin, suggesting that both proteins bind to myosin by the KRP domain (Shirinsky, V. P., Vorotnikov, A. V., Birukov, K. G., Nanaev, A. K., Collinge, M., Lukas, T. J., Sellers, J. R., and Watterson, D. M. (1993)J. Biol. Chem. 268, 16578–16583). In this study, we investigated which regions of myosin and KRP interact in vitro. Using cosedimentation assays, we determined that KRP binds to unphosphorylated myosin with a stoichiometry of 1 mol of KRP/1 mol of myosin and an affinity of 5.5 μm. KRP slows the rate of proteolytic cleavage of the head-tail junction of heavy meromyosin by papain and chymotrypsin, suggesting it is binding to this region of myosin. In addition, competition experiments, using soluble headless fragments of nonmuscle myosin, confirmed that KRP interacts with the regulatory light chain binding region of myosin. The regions important for KRP’s binding to myosin were investigated using bacterially expressed KRP truncation mutants. We determined that the acid-rich sequence between Gly138 and Asp151 of KRP is required for high affinity myosin binding, and that the amino terminus and β-barrel regions weakly interact with myosin. All KRP truncations, at concentrations comparable to theirK D values, exhibited some stabilization of myosin filaments against ATP depolymerization in vitro, suggesting that KRP’s ability to stabilize myosin filaments is commensurate with its myosin binding affinity. KRP weakened the K m but not the V max of phosphorylation of myosin by MLCK, demonstrating that bound KRP does not prevent MLCK from activating myosin.

In resting smooth muscle, myosin exists in a predominantly dephosphorylated and inactive state (1) and contraction is initiated by a calcium-calmodulin-dependent phosphorylation of the regulatory light chain (RLC) 1 of myosin by myosin light chain kinase (MLCK) (2). In vitro, phosphorylation of myosin activates its MgATPase activity in the presence of actin and also allows myosin to translocate actin filaments (3). Phosphorylation also affects the equilibrium between filamentous and monomeric myosins in vitro (4). Filaments formed from unphosphorylated myosin are dissociated by MgATP into a monomeric form of myosin in which the tail forms a hairpin fold back onto the neck region of the myosin head. This folded monomeric myosin has been termed "10 S" myosin based on its sedimentation coefficient in the analytical ultracentrifuge, while phosphorylated monomeric myosin exists primarily in an extended conformation, called "6 S" (5,6). Phosphorylation of the RLC greatly stabilizes filaments against MgATP dissociation (4,7).
In vivo, smooth muscle myosin filaments appear to be stable even under conditions where myosin is predominantly unphosphorylated (1,8,9). The discrepancy between myosin's filamentous state in vivo and in vitro may be explained by the presence of regulatory proteins, which stabilize unphosphorylated smooth muscle myosin filaments in vivo. One candidate for such a protein is an abundant myosin-binding protein, termed kinase-related protein (KRP) and also referred to as telokin. KRP binds to unphosphorylated myosin filaments in vitro and stabilizes them against MgATP-dependent depolymerization (10). KRP binds tightly to the heavy meromyosin (HMM), but not to the subfragment-1 (S1) or rod domains of myosin, suggesting that KRP's binding site is in the S1-S2 region of myosin. In addition, KRP may prevent myosin from adopting the 10 S conformation, by competing for the region on myosin where the tail folds over, which is thought to include the RLC binding site (11).
The gene encoding KRP is nested within the gene for MLCK, and, in fact, the 157-residue sequence of KRP is completely identical to the sequence of the carboxyl terminus of MLCK (12,13). Western analyses estimate that in avian gizzard smooth muscle, KRP is abundant compared with MLCK, and is approximately stoichiometric with myosin (10,14). It is likely that MLCK interacts with myosin primarily by its KRP domain, since KRP and MLCK compete for myosin binding and a trun-cated MLCK that lacks the KRP domain cannot bind myosin as well as intact MLCK (10).
The crystal structure of KRP has been solved and shows that the protein is a seven-stranded C2 immunoglobulin-like ␤barrel with trailing arms at either end (15). The amino-terminal 35 amino acids and the carboxyl-terminal 19 amino acids were not resolved in the crystal structure, suggesting that they are in an unstructured or flexible conformation. The amino terminus contains two putative sites for phosphorylation by cyclic AMP-dependent kinase (PKA) and mitogen-activated protein kinase (MAPK) (16), while the carboxyl terminus is distinguished by a high content of acidic residues. To determine the sites of interaction between KRP and myosin, we have quantified KRP binding to myosin and examined its interaction with various proteolytic and recombinant fragments of myosin. We show that KRP binds at the RLC region of the myosin heavy chain, probably where the two heads join to form the rod region of myosin. We have expressed amino-and carboxylterminal truncation mutants of KRP and studied their binding to myosin. Furthermore, we demonstrate that the acidic carboxyl terminus of KRP is important for high affinity myosin binding and subsequent stabilization of myosin filaments. Finally, we show that KRP affects the K m , but not the V max for phosphorylation of myosin by MLCK. Some of these data have been presented in preliminary form (17).
Bacterial extracts were incubated at 30°C for 15 min in binding buffer containing 0.1% Triton X-100, 0.10 mM leupeptin, 0.1 mM PMSF, 10 mM ␤-mercaptoethanol, and 0.01% of each of the following: chymostatin, pepstatin, N ␣ -p-tosyl-L-lysine chloromethyl ketone hydrochloride, and N-tosyl-L-phenylalanine chloromethyl ketone. Following brief sonication and centrifugation at 20,000 ϫ g for 15 min at 4°C, the supernatant was filtered through a 0.45-m filter. The supernatant (10 ml) was applied to a 3.0-ml bed volume TALON metal affinity column (CLONTECH, Palo Alto, CA), which has a high affinity for recombinant 6-His-tagged proteins. After extensive washing with 10 volumes of binding buffer, the protein was eluted using 2 volumes of elution buffer (binding buffer containing a final concentration of 60 mM imidazole). The column was stripped of remaining protein using binding buffer containing 1 M imidazole and was reused. Protein fractions were analyzed electrophoretically, pooled, concentrated using Aquacide (Calbiochem, San Diego, CA), and dialyzed against Buffer A (50 mM NaCl, 10 mM MOPS (pH 7.0), 1 mM MgCl 2 , 0.1 mM EGTA, 3 mM NaN 3 , and 3 mM DTT).
Protein Concentration Determination-Protein concentrations were measured spectrophotometerically (Beckman, Columbia, MD). For KRP, the following extinction coefficients were calculated according to Gill For bacterially expressed KRP truncations, nonradioactive samples were incubated and cosedimented as above, with myosin and 0.1 mg/ml bovine serum albumin (BSA), which was used as a volume marker for the pellet. Binding was analyzed by gel electrophoresis on 12.5% SDSpolyacrylamide gels. The supernatants (40 l) were removed and added to 10 l of 5 ϫ Laemmli sample buffer (24) and DTT, boiled, and electrophoresed. The pellets were rinsed in buffer A, resuspended in 25 l of 1 ϫ sample buffer, boiled, and electrophoresed. Gels were stained with Coomassie Blue R-250 dye, destained, and scanned on a densitometer (Molecular Dynamics, Sunnyvale, CA). After correcting for the H 2 O volume of the pellets by measurement of BSA, the mol of KRP bound/ mol of myosin was quantitated by densitometry and the data were fit to the equation described above.
For competition experiments, soluble nonmuscle myosin fragments (ranging from 1-6 molar excess) were added to smooth muscle myosin and recombinant KRP (1:3 molar ratio), in 50 l of buffer A for 15 min at 25°C. Cosedimentation and densitometry, as described above, were utilized to identify a depletion of KRP from the pellet. Percent competition was calculated by the proportion of KRP bound to myosin in the presence of soluble fragments versus KRP bound to myosin in their absence.
Protease Protection Assays-For papain digestions, 5 M HMM was incubated in the presence or absence of 20 M KRP in buffer A at 25°C for 5 min. The reaction was initiated with the addition of 200 units of papain/mg of myosin (Worthington, Freehold, NJ) in papain activating solution (0.1 M NaCl, 10 mM NaPO 4 , 0.2 mM EDTA, pH 7.0) Time points were collected every 30 s for 5 min, and the reaction was stopped with 20 mM iodoacetic acid. Samples were added to 2 ϫ sample buffer, boiled, and electrophoresed on 12.5% SDS-polyacrylamide gels. Coomassie Blue-stained gels were quantitated by densitometry, and the digestion was monitored by the relative amount of S2 produced at each time point. Data were generated from four experiments.
For chymotrypsin digestions, 3.7 M HMM was incubated with 22 M KRP in buffer A (without ATP) at 25°C for 5 min. The reaction was initiated with the addition of ␣-chymotrypsin (Sigma) to a 1:20 ratio (w/w) of protease to HMM, and time points were collected at 5, 10, 20, 40, and 80 min. The digestion was terminated with 1 mM PMSF. Samples were electrophoresed and quantitated by densitometry (NIH Image, Bethesda, MD) of the amount of RLC digested, with normalization to undigested essential light chain (ELC). Data were generated from three experiments.
ATPase Assays-The protection of myosin filaments against ATPdepolymerization by KRP truncations was monitored by measurement of the (NH 4 ) ϩ EDTA-ATPase activities of soluble myosin. Myosin purified from chicken gizzards was used for these experiments, as this source of myosin was consistently soluble upon addition of 2 mM MgATP (Ͼ95%). Myosin filaments were prepared by dialysis against 150 mM NaCl, 10 mM MOPS (pH 7.0), 1 mM MgCl 2 , 0.1 mM EGTA, 3 mM NaN 3 , and 3 mM DTT. 3 M myosin filaments and 15 M KRP truncations were incubated in buffer A brought to 100 mM NaCl in the presence of 2 mM MgATP, and cosedimented as above. Following centrifugation, 5 l of the supernatants was added to the following ATPase mixture: 0.4 M NH 4 Cl 2 , 35 mM EDTA, 30 mM NaCl, 25 mM HEPES (pH 8.0), 2 mM MOPS, 0.2 mM MgCl 2 , 0.2 mM EGTA, 5 mM [␥-32 P]ATP, 0.5 mg/ml BSA. The rate of phosphate release was measured at 25°C over 15 min according to Sellers et al. (20). Samples were measured in duplicate, and the percent of soluble myosin was determined by normalization to the myosin control without KRP.

RESULTS
KRP Binding Affinity and Stoichiometry-Studies of KRP binding to myosin were performed by a cosedimentation assay, using conditions in which KRP is soluble and myosin is filamentous and sedimentable. KRP has two sites for phosphorylation by PKA and MAPK, respectively. PKA-phosphorylated KRP has the same binding capacity as unphosphorylated KRP (16). Therefore, [ 32 P]KRP was used in binding studies to determine KRP's binding affinity and stoichiometry of binding to myosin. The K D was 5.5 M, and a stoichiometry of 1.10 mol of KRP/mol of myosin molecule was determined (Fig. 1).
KRP Binds to the S1-S2 Domain of Myosin-Previous experiments showed that KRP binds well to unphosphorylated HMM, but not to the S1 or rod domains of myosin (10). In addition, the ability of KRP to protect myosin against ATPinduced depolymerization in vitro suggested that KRP may prevent myosin from adopting the 10 S (folded) conformation, which is thought to occur by binding of the rod of myosin to its S1-S2 junction (6,11). We examined whether KRP binds to this site by using proteolytic protection assays. Papain will proteolytically digest HMM into S1 and S2 fragments (25). Fig. 2A depicts a time course of this reaction in the presence and absence of KRP. Inclusion of KRP retarded the rate of papain digestion of the heavy chain of HMM and also inhibited the rate of proteolysis of the RLC. In Fig. 2B, this effect is quantified as the proportion of S2 produced over time. KRP also afforded protection of the RLC of HMM against chymotrypsin digestion (Fig. 2C). In both experiments, KRP itself was not digested, indicating that it was not merely acting as a competitor for proteolysis by the enzymes.
Competition Experiments Reveal Binding to the RLC Binding Region-KRP protection of myosin against proteolysis suggested that it bound to the S1-S2 junction of myosin. Since KRP

FIG. 2. KRP protects the myosin heavy chain and the RLC against proteolysis at the S1-S2 junction.
HMM was digested by papain and chymotrypsin with and without KRP present. A, two 12.5% SDS-polyacrylamide gels, which represent samples from 11-min papain digestions. B, quantification of the rate of papain digestion of the myosin heavy chain, depicted by the amounts of S2 produced over time. C, quantification of the rate of chymotryptic digestion of the RLC. bound similarly to nonmuscle and smooth muscle myosins (data not shown), we used nonmuscle myosin fragments containing this region in competition binding experiments, to more precisely localize KRP's binding site. Soluble headless fragments of nonmuscle myosin IIB corresponding to S2, S2 and bound RLC, and S2 and bound RLC and ELC were prepared by baculovirus expression (Fig. 3A). The soluble fragments were included in cosedimentation assays with smooth muscle myosin and KRP, and the depletion of KRP from the pellet indicated competitive binding to the soluble fragment. The two constructs containing the RLC binding site competed strongly with myosin for KRP binding, while S2 alone did not compete as well, indicating a weaker interaction with KRP (Fig. 3B).
Expression of KRP Truncations in Bacteria-Truncation mutants of KRP were made by PCR of a nonmuscle MLCK cDNA template. The truncations were designed based upon the crystal structure previously determined (15) and the primary structure, as shown in Fig. 4 (A and B). The three-dimensional structure of KRP consists of a core seven-stranded ␤-barrel with a C2-type IgG motif. The amino-and carboxyl-terminal tails, containing amino acids 1-35 and 138 -157, respectively, were not resolved in the crystal structure, presumably due to their flexibility. The consensus sites for phosphorylation by PKA and MAPK are found within the amino terminus, and an acid-rich domain characterizes the carboxyl terminus (13 of 17 residues are acidic). Truncations were created to understand the significance of these regions for myosin binding and stabilization properties in vitro. Fig. 5A is a schematic representation of the expressed truncations, and their migration on a 20% SDS-polyacrylamide gel is shown in Fig. 5B. Since preparations of truncations 5, 6, and 7 consistently migrated as more than one band, the integrity of these truncations was confirmed by mass spectroscopy. This method revealed the expected sequences with only a methionine missing in some cases, suggesting that the extra bands may be due to intramolecular bonds (data not shown).

The COOH Terminus of KRP Is Significant for Myosin Binding and Filament Stabilization in Vitro-
The KRP truncations were assayed for their myosin binding capacities using cosedimentation assays. The binding affinities (K D , dissociation constant) of the KRP truncations are indicated in Table I. Truncation of the first 23 residues at the amino terminus, K(12-157) and K(23-157), did not affect myosin binding, while truncation of the first 36 residues, K(36 -157), weakened the binding slightly. Truncation of 7 residues at the carboxyl terminus, K(1-150), also did not affect binding affinity, whereas removal of 19 residues at the carboxyl terminus, K(1-138), weakened KRP binding by about 5-fold. The smallest truncations, representing the amino acids resolved in the crystal structure, K (36 -138), and the core ␤-barrel, K(43-136), also had weaker binding affinities.
KRP truncations (15 M) exhibited a similar pattern in their abilities to protect against the depolymerization of myosin (3 M) by MgATP in vitro. All three amino-terminal KRP truncations, K(12-157), K(23-157), and K(36 -157), stabilized myosin filaments like wild type KRP. However, all of the carboxylterminal truncations, K(1-138), K(36 -138), and K(43-136) which lack the acidic region, failed to stabilize myosin filaments (Table I) 4. Crystal structure and primary sequence of KRP. A, the KRP core is a seven-sheet ␤-barrel of a C2 immunoglobulin-like motif (15). The structure is depicted here with the amino terminus at the right and the carboxyl terminus at the left. The amino-terminal 35 residues and the carboxyl-terminal 19 residues were unresolved in the crystal structure. B, the KRP primary sequence is 157 amino acids. Those residues resolved in the crystal structure are indicated here in bold, and the phosphorylation sites for PKA and MAPK are underlined.
binding truncation mutants to levels where binding was comparable to K D led to further stabilization. For example, inclusion of 60 M of K(1-138) and K(36 -138) left 45% and 18% myosin soluble, respectively. The relative abilities of KRP truncations to stabilize myosin filaments in vitro appears to be proportional to their binding affinities. Altogether the data confirm that KRP binds myosin with high affinity by its acidic tail and weakly interacts with myosin by its amino terminus and core ␤-barrel.
Effect of KRP on Phosphorylation of HMM by MLCK-The evidence presented above suggests that KRP binds to the S1-S2 junction, probably in the vicinity of the phosphorylatable residue on the RLC and that this region is likely the same site to which MLCK binds. This raises the possibility that KRP might act as a competitive inhibitor of RLC phosphorylation. To directly examine this prospect, the steady state kinetics of phosphorylation of HMM was measured in the absence and presence of a near saturating concentration of KRP (50 M). The data show that KRP increases the K m for phosphorylation approximately 10-fold, but does not dramatically affect the V max (Fig. 6). DISCUSSION A knowledge of the binding sites between KRP and myosin is necessary for understanding how KRP stabilizes myosin filaments and elucidating the structural and functional relationships between MLCK and KRP. The question of how KRP binds myosin was especially intriguing to address because of the shared structural motifs in KRP and MLCK. The predominant structure of KRP is a seven-stranded C2-type immunoglobulin (IgG)-like motif (15). The amino-and carboxyl termini of KRP, which were not resolved in the crystal structure, are also characterized by interesting sequences, including phosphorylation sites at the amino terminus of KRP (16), and a highly acidic carboxyl terminus (13 of 17 residues).
MLCK has two other IgG repeats, in addition to the KRP IgG repeat, which are located amino-terminal to the catalytic core (26). None of these IgG repeats are essential for MLCK catalysis (27). IgG repeats are found in other myosin-binding pro- teins. MyBP-C, also called C-protein, is a skeletal muscle thick filament protein, which contains seven copies of the IgG motif (28) and can bind to LMM and S2 (29,30). The carboxylterminal IgG motif is responsible for high affinity binding to myosin (31,32). Moreover, Okagawi et al. (31) compared the primary sequences of the myosin binding IgG motif in MyBP-C and the carboxyl-terminal IgG motif of MLCK (the KRP domain), and predicted 15 residues within the KRP ␤-barrel that might be important for binding to myosin. Another myosinbinding protein is vertebrate titin, a 3-MDa protein that spans the skeletal sarcomere and is implicated in filament positioning and stabilization (33,34). Titin contains 122 IgG motifs, located mostly throughout the I band domain, but uses several IgG motifs within the A band to interact with LMM (35,36). A titin homologue in C. elegans, twitchin, contains 26 IgG motifs (37). Interestingly, the carboxyl terminus of twitchin is organized into domains that are similar in protein sequence and arrangement to MLCK (38). Recently, Kobe et al. (38) used the crystal structure of the twitchin carboxyl-terminal IgG motif to model the corresponding structure in MLCK (the KRP domain). Based on this model and the myosin binding sequences that were predicted by Okagawi et al. (31), the authors specified 6 residues within the KRP ␤-barrel, Phe 44 , Asp 50 , Val 82 , Val 118 , Glu 123 , Ala 124 , which might participate in binding to myosin. In addition, since titin uses one of its IgG motifs to bind to skeletal LMM (35), Kobe et al. (38) suggested that MLCK might interact with the homologous residues, Leu 1824 -Arg 1840 , in the LMM portion of smooth muscle myosin.
This proposed model for KRP binding to myosin is inconsistent with the results obtained here. Our in vitro assays with bacterially expressed KRP truncation mutants indicated that the acid-rich carboxyl terminus of KRP, corresponding to the sequence Gly 138 -Glu 150 , is the primary determinant of KRP's binding affinity for myosin. The Kobe et al. (38) model postulates that important residues for myosin binding are contained within the KRP ␤-barrel, but we found that truncation mutants representing amino acids resolved in the crystal structure, K(36 -138), and the core ␤-barrel, K(43-136), bound only weakly to myosin. However, it is important to note that the carboxyl terminus truncations could at least partially stabilize myosin filaments in the presence of ATP, if sufficiently high concentrations were used to obtain binding. The amino terminus is not critical for interaction with myosin, as indicated by our binding assays with truncations and by the lack of effects of phosphorylation of KRP on its binding to myosin (16). These assays suggest that the KRP ␤-barrel and the amino-terminal extension do not determine the high affinity binding to myosin, but that the ␤-barrel does weakly bind to myosin.
In addition, we have shown that KRP clearly does not bind to the region of smooth muscle myosin postulated by Kobe et al. (38). It should be noted that there are numerous examples of IgG motifs that do not bind LMM, such as the two aminoterminal IgG motifs of MLCK, which do not appear to bind myosin (27).
Evidence from Sellers and Pato (19), Shirinsky et al. (10), and the present study indicates that KRP and MLCK share a primary myosin docking site at the S1-S2 junction determined by the KRP domain that is independent of MLCK's catalytic interaction with the RLC. The idea that MLCK possesses both a catalytic binding site and a "docking" site that is shared with KRP is supported by the steady state kinetic analysis of HMM phosphorylation, which clearly demonstrates that KRP binding to myosin only affects the K m and not the V max of phosphorylation of myosin by MLCK. Since the KRP binding site is not required for catalysis, MLCK could phosphorylate myosin light chains even if KRP is bound, but MLCK now can only bind via its catalytic site. However, because KRP is more abundant in a smooth muscle cell, and evidence suggests that MLCK predominantly associates with F-actin in vivo (39,40), it is likely that more myosin molecules have KRP bound than MLCK. Nevertheless, MLCK should be able to efficiently phosphorylate myosin.
KRP binding to the S1-S2 junction of myosin is consistent with its apparent ability to shift the myosin equilibrium from the 10 S to the 6 S conformation in vitro. This region has been implicated in the formation of myosin in the folded 10 S conformation. A peptide corresponding to 12 residues, Leu 835 -Lys 846 , from the head/neck junction of the myosin heavy chain, inhibits the formation of the 10 S conformation (41). This is the location where we propose KRP binds. KRP's ability to protect myosin against papain proteolysis at the S1-S2 junction is also strikingly similar to the resistance of both the heavy chains and light chains to digestion when myosin is in the 10 S conformation (42,43). In addition, during the preparation of this manuscript, a paper was published reporting the crosslinking of KRP and smooth muscle myosin (44). KRP was shown to cross-link to both the RLC and the heavy chain and was localized to the neck region of myosin by electron microscopy. In addition, the cross-linked myosin was unable to form the 10 S conformation.
Recently Olney et al. (11) showed that myosin can be crosslinked in the 10 S conformation, using a probe on Cys 108 of the RLC, to a site approximately one-third from the end of the tail. The cross-linked region on LMM contains one of the negatively charged regions within the 28-residue repeats, which characterize the myosin tail. It is interesting that the KRP sequence that we have identified to be most important for binding to myosin is also very acidic. We propose that the negatively charged carboxyl terminus of KRP may bind electrostatically to the same region of myosin to which the acidic tail interacts during 10 S formation and effectively compete with this tail to prevent 10 S formation.
ln summary, our studies have shown that KRP binds tightly via its acidic tail to the site on myosin where S1 and S2 join. This binding site is logical based on KRP's ability to affect the 6 S-10 S myosin equilibrium, and lends credence to KRP's supposed in vivo function. Furthermore, implications that the 10 S conformation is related to ionic interactions between LMM and the S1-S2 junction correlate well with our finding that KRP binds to myosin via acidic residues. Further studies are currently under way to elucidate the specific residues of myosin and KRP that interact.