Properties of Filament-bound Myosin Light Chain Kinase*

Myosin light chain kinase binds to actin-containing filaments from cells with a greater affinity than to F-actin. However, it is not known if this binding in cells is regulated by Ca2+/calmodulin as it is with F-actin. Therefore, the binding properties of the kinase to stress fibers were examined in smooth muscle-derived A7r5 cells. Full-length myosin light chain kinase or a truncation mutant lacking residues 2–142 was expressed as chimeras containing green fluorescent protein at the C terminus. In intact cells, the full-length kinase bound to stress fibers, whereas the truncated kinase showed diffuse fluorescence in the cytoplasm. After permeabilization with saponin, the fluorescence from the truncated kinase disappeared, whereas the fluorescence of the full-length kinase was retained on stress fibers. Measurements of fluorescence intensities and fluorescence recovery after photobleaching of the full-length myosin light chain kinase in saponin-permeable cells showed that Ca2+/calmodulin did not dissociate the kinase from these filaments. However, the filament-bound kinase was sufficient for Ca2+-dependent phosphorylation of myosin regulatory light chain and contraction of stress fibers. Thus, dissociation of myosin light chain kinase from actin-containing thin filaments is not necessary for phosphorylation of myosin light chain in thick filaments. We note that the distance between the N terminus and the catalytic core of the kinase is sufficient to span the distance between thin and thick filaments.

Myosin light chain kinase binds to actin-containing filaments from cells with a greater affinity than to Factin. However, it is not known if this binding in cells is regulated by Ca 2؉ /calmodulin as it is with F-actin. Therefore, the binding properties of the kinase to stress fibers were examined in smooth muscle-derived A7r5 cells. Full-length myosin light chain kinase or a truncation mutant lacking residues 2-142 was expressed as chimeras containing green fluorescent protein at the C terminus. In intact cells, the full-length kinase bound to stress fibers, whereas the truncated kinase showed diffuse fluorescence in the cytoplasm. After permeabilization with saponin, the fluorescence from the truncated kinase disappeared, whereas the fluorescence of the full-length kinase was retained on stress fibers. Measurements of fluorescence intensities and fluorescence recovery after photobleaching of the full-length myosin light chain kinase in saponin-permeable cells showed that Ca 2؉ /calmodulin did not dissociate the kinase from these filaments. However, the filament-bound kinase was sufficient for Ca 2؉ -dependent phosphorylation of myosin regulatory light chain and contraction of stress fibers. Thus, dissociation of myosin light chain kinase from actin-containing thin filaments is not necessary for phosphorylation of myosin light chain in thick filaments. We note that the distance between the N terminus and the catalytic core of the kinase is sufficient to span the distance between thin and thick filaments.
Phosphorylation of Ser-19 at the N terminus of smooth muscle myosin regulatory light chain (RLC) 1 by Ca 2ϩ /calmodulindependent myosin light chain kinase results in an increase in actin-activated Mg 2ϩ ATPase activity. It is now known that this activity plays a key role in a variety of biological events, including, in part, initiation of smooth muscle contraction (1), potentiation of skeletal and cardiac muscle contraction (2), fibroblast contraction (3,4), endothelial cell retraction (5-7), platelet aggregation and contraction (8,9), growth cone motil-ity in nerve cells (10), and receptor capping in lymphocytes (11). When the intracellular Ca 2ϩ concentration increases upon stimulation of cells by Ca 2ϩ influx through Ca 2ϩ channels in the plasma membrane or through Ca 2ϩ release from intracellular Ca 2ϩ stores, Ca 2ϩ binds to calmodulin. Ca 2ϩ /calmodulin binds to the calmodulin-binding sequence of the kinase, resulting in displacement of the autoinhibitory segment and exposure of the catalytic site for RLC phosphorylation (12,13).
Previous studies in vitro showed that myosin light chain kinase binds to purified F-actin as well as myosin filaments and that binding in both cases is regulated by Ca 2ϩ /calmodulin (14 -18). Biochemical studies by co-sedimentation assays in vitro indicated that the N terminus of myosin light chain kinase was responsible for binding to purified F-actin (14 -16), whereas its C terminus bound to purified myosin (17,18) with binding affinities in the range of 10 Ϫ5 to 10 Ϫ6 M. These binding affinities were decreased in the presence of Ca 2ϩ /calmodulin (18). Additionally, Ca 2ϩ /calmodulin may interact with an N-terminal F-actin-binding site of myosin light chain kinase, distinct from the calmodulin-binding sequence at the C terminus of the catalytic core (15). The F-actin binding site in residues 1-41 had a K a value of approximately 10 5 M Ϫ1 with an IQ-like sequence in residues 26 -41 that bound calmodulin (15).
Studies have also shown that myosin light chain kinase binds to cellular actomyosin-containing filaments. Immunocytochemistry studies found myosin light chain kinase localized to actomyosin-containing stress fibers or myofilaments in nonmuscle and smooth muscle cells, respectively (19,20). Recent investigations in vitro and in vivo showed that the N-terminal half of the kinase (residues 1-655), not the C terminus, was both necessary and sufficient for binding to isolated smooth muscle myofilaments, stress fibers in permeabilized fibroblasts, and myofilaments in intact smooth muscle cells (21). In addition, the apparent binding affinities of myosin light chain kinase to smooth muscle myofilaments and actin-containing thin filaments were greater than to purified smooth muscle F-actin and skeletal muscle myofilaments (21,22). Thus, myosin light chain kinase binds to actomyosin-containing filaments or actin-containing thin filaments in cells with an affinity greater than to F-actin alone, perhaps due to the presence of a distinct anchoring protein or a protein that facilitates binding to F-actin.
Other reports support the idea that myosin light chain kinase binds tightly to actomyosin-containing filaments in cells or tissues. For example, smooth muscle tissue strips made permeable with 1% Triton X-100 and stored for several weeks at Ϫ20°C in 50% glycerol were still able to contract in the presence of Mg 2ϩ ATP and Ca 2ϩ /calmodulin (23). Quantitation of myosin light chain kinase showed no significant loss of the kinase after storage. Thus, myosin light chain kinase appears to bind tightly to myofilaments in permeable smooth muscle cells with an apparent affinity greater than values reported for binding to purified F-actin (21). This raises the question of whether myosin light chain kinase must dissociate from actincontaining thin filaments to phosphorylate myosin RLC in the thick filaments (14 -18). Stimulation of smooth muscle tissues can result in a high extent (50 -100%) of RLC phosphorylation, particularly in the presence of a cell-permeable protein phosphatase inhibitor (24 -27). Yet, the concentration of RLC in smooth muscle is estimated as 75-80 mM, a concentration 25 times greater than the concentration of myosin light chain kinase (28,29). Smooth muscle myosin light chain kinase associated with actin-containing filaments may thus dissociate and diffuse to myosin-containing filaments to achieve a high extent of RLC phosphorylation during contraction (14). Thus, Ca 2ϩ /calmodulin may not only activate the kinase but also dissociate it from actin-containing filaments via interaction with a second binding site in the N terminus (15). To test this hypothesis, we transiently expressed myosin light chain kinase with green fluorescent protein and established a system to study its binding to stress fibers in permeable A7r5 cells derived from smooth muscle.

Myosin Light Chain Kinase Binding to Smooth Muscle Myofilaments and F-actin on
Nitrocellulose-coated Coverslips-Extracted smooth muscle myofilaments and purified F-actin filaments (21) were diluted to 1 mg/ml in air-dried BODIPY FL phallacidin (5 units, Molecular Probes, Inc., Eugene, OR) containing 10 mM MOPS at pH 7.0, 50 mM NaCl, 2 mM dithiothreitol, 1 mg/ml bovine serum albumin, and 1 mM MgCl 2 . After adding to nitrocellulose-coated coverslips (advice kindly provided by Dr. Joseph Haeberle, University of Vermont) at room temperature for 30 min, the coverslips were washed with buffer three times to remove nonadhering filaments. Cy3-labeled full-length myosin light chain kinase (0.2 mM) (21) was added to both types of filaments. After incubation at room temperature for 10 min, coverslips were washed with buffer three times to remove unbound kinase. Coverslips were sealed on glass slides for fluorescence imaging.
Construction of MLCK-GFP Plasmids-MLCK-GFP expression vectors were constructed from modified pGREEN LANTERN TM -1 (Life Technologies, Inc.), a cytomegalovirus promoter-driven expression vector containing GFP coding sequences. This GFP coding sequence contains a Ser 65 3 Thr mutation and "humanized" codon usage for increasing its brightness and better expression in mammalian cells (30). Plasmid pGREEN LANTERN TM -1 was modified to delete two nucleotides (GC) between SpeI and NotI cloning sites. This modification shifts an in frame TAG stop codon (within the SpeI restriction enzyme site) at the 5Ј-end of GFP coding sequences to out of frame. A two-nucleotidedeleted SpeI-BclI GFP DNA fragment (1 kilobase pair) was synthesized by polymerase chain reaction using pGREEN LANTERN TM -1 as a template and a primer pair: 5Ј-GCTGACTAGTGCGGCCGCCGCCA-C-3Ј and 5Ј-GGCTGATTATGATCATGAAC-3Ј. This polymerase chain reaction DNA fragment and vector pGREEN LANTERN TM -1 were digested with SpeI and BamHI sites. The 0.54-and 4.5-kilobase-paired DNA fragments from the digested polymerase chain reaction product and the vector, respectively, were ligated and transformed. The modified plasmid (pGFP) was confirmed by DNA sequencing and further used to construct MLCK-GFP fusion protein expression vectors. Fulllength and N-terminal (amino acids 2-142) deleted myosin light chain kinase DNA fragments were synthesized without a stop codon at their 3Ј-ends by polymerase chain reaction using Pat1/J1-J3 (containing fulllength rabbit smooth muscle myosin light chain kinase coding sequences) as a template and primer pairs 5Ј-TCCCCCCGGGATGGATTTCC-GCGCCAAC-3Ј and 5Ј-GACTAGTTGACTCCTCTTCCTCCTCTTCCCC-3Ј, and 5Ј-TCCCCCCGGGATGGAGAGCTCGAAACCTGTGGGC-3Ј and 5Ј-GACTAGTTGACTCCTCTTCCTCCTCTTCCCC-3Ј, respectively. Synthesized myosin light chain kinase DNA fragments (3.6 and 3.0 kilobase pairs) were ligated into pGFP by XmaI and SpeI sites. The resultant plasmids (pFL-GFP and pDN-GFP) were confirmed by restriction enzyme analyses and DNA sequencing.
Cell Culture and DNA Transfection-A7r5 rat thoracic aorta smooth muscle cells (obtained from ATCC, CRL-1444) were maintained in Dulbecco's modified Eagle's medium containing 4 mM L-glutamine, 1.5 g/liter sodium bicarbonate, 4.5 g/liter glucose, 1 mM sodium pyruvate, 10% fetal bovine serum, 100 units/ml penicillin, and 100 mg/ml streptomycin at 37°C with 5% CO 2 . For DNA transfection, A7r5 cells were seeded onto 40-mm round coverslips (40 Circles-1D; Fisher) in 60-mm Petri dishes at 30 -50% confluence 1 day before transfection. DNA was transfected by a liposome-mediated method according to the manufacturer's instructions. In brief, 12 ml of FuGENE TM -6 (Boehringer Mannheim) was added into 288 ml of serum-free medium. After a 5-min incubation at room temperature, the FuGENE TM -6/medium mixture was added slowly to 3 mg of DNA (15 ml of 0.2 mg/ml DNA). After a 15-min incubation at room temperature, the FuGENE TM -6/DNA mixture was added to cells. Transfected cells were incubated for 2 days before experiments.
Determination of Myosin Light Chain Kinase Activity and Binding in Vitro-Ca 2ϩ /calmodulin-dependent activity of myosin light chain kinase was determined by measuring rates of 32 P incorporation into myosin RLC (31). Transfected or untransfected cells were lysed on ice for 10 min in 50 mM MOPS at pH 7.0, 50 mM MgCl 2 , 0.5 mM EGTA, 1% Nonidet P-40, 10% glycerol, 1 mM dithiothreitol, and protease inhibitors (100 mg/ml phenylmethylsulfonyl fluoride, 20 mg/ml leupeptin, 30 mg/ml aprotenin, 60 mg/ml tosyllysylchlomethyl ketone, and 60 mg/ml tosylphenylalanyl chloromethyl ketone). Cell lysates were clarified by centrifugation at 14,000 ϫ g for 2 min and diluted for kinase activity assays so that 32 P incorporation was linear in respect to time. The minimal kinase activities from untransfected cells were measured as background values. The amounts of expressed MLCK-GFP proteins were quantitated by immunoblotting using either monoclonal antibodies raised to the full-length myosin light chain kinase or polyclonal antibodies raised against the catalytic core of rabbit smooth muscle myosin light chain kinase. Various amounts of rabbit smooth muscle myosin light chain kinase purified from Sf9 cells (21) were used for a standard curve.
For binding analyses in vitro, A7r5 cell lysates containing expressed myosin light chain kinase were further dialyzed against 50 mM MOPS at pH 7.0, 0.5 mM EGTA, 10% glycerol, 1 mM dithiothreitol, and protease inhibitors to remove MgCl 2 . Binding of myosin light chain kinases to smooth muscle myofilaments in vitro was measured by a co-sedimentation procedure (21). The amounts of myosin light chain kinase in the supernatant and pellet fractions were compared by measurements of kinase activity.

Measurement of the Amount of Expressed MLCK-GFP in A7r5
Cells-A7r5 cells transfected with cDNA for MLCK-GFP were sorted by flow cytometry (FACStar Plus; Becton Dickinson, San Jose, CA). After cells were detached by 0.05%/0.53 mM trypsin-EDTA (Life Technologies, Inc.), fetal bovine serum was added to inhibit trypsin activity. Cells were centrifuged at 200 ϫ g for 10 min and washed with phosphatebuffered saline twice before sorting. Cells were sorted based on fluorescence intensity (excitation at 488 nm). About 15% transfected A7r5 cells showed green fluorescence. Positive (fluorescent) and negative (nonfluorescent) cells were collected separately. Numbers of the collected cells after sorting were recorded. Cells were lysed, and the amount of expressed MLCK-GFP was measured by immunoblotting as described previously herein. The average concentration of expressed MLCK-GFP in a single cell was calculated as follows: amount of MLCK-GFP/(number of cells ϫ volume of a single cell). The volume of a single cell was assumed to be 2 picoliters.
Fluorescence Imaging-Fluorescence imaging was performed as described previously (32). Twelve-bit fluorescence images were acquired by a cooled CCD camera (Quantix Photometrics, Tucson, AZ) and Oncor-Image software (Oncor, Gaithersburg, MD). Narrow bandpass interference filters (Omega, Brattleboro, VT) were used to select BODIPY fluorescein or GFP (excitation at 490 nm and emission at 520 nm) and rhodamine or Cy3 (excitation at 550 nm and emission at 575 nm) fluorescence. Cells were kept in an open thermal controlled chamber (Custom Scientific, Dallas, TX) at 37°C during fluorescence imaging. Additional imaging experiments were performed with purified F-actin and detergent-washed gizzard myofilaments with Cy3-labeled myosin light chain kinase (21).
Fluorescence Recovery after Photobleaching (FRAP)-FRAP was performed on the saponin-permeable A7r5 cells as described previously (23) with the following modification. An argon ion laser (Spectra-Physics 2017) operated at 1 watt on the 488 nm line was directed into the Filament-bound Myosin Light Chain Kinase microscope (Zeiss Axiovert 35) with the beam radius 4.43 mm at the specimen plane. The proportion of the cell photobleached was approximately 3%. Photobleaching time was 2-30 ms, resulting in a 20 -70% decrease in the fluorescence intensity. At least three half-lives of data were recorded for each recovery. The recovery time for some experiments was monitored for 15 min. Cells were maintained at 37°C in a sealed chamber. A control experiment using A7r5 cells microinjected with 167-kDa fluorescein-labeled dextran (Sigma; 4 mg/ml in 10 mM MOPS, 30 mM magnesium acetate, 100 mM NaCl at pH 7.1) was also included in FRAP measurements under similar conditions, except the cells were permeabilized with 30 mg/ml ␤-escin at 37°C for 10 min (23).
Measurement of RLC Phosphorylation by Urea-Glycerol PAGE-A7r5 cells were treated with 0.02% saponin in Ca 2ϩ -free buffer at 37°C for 10 min and washed with Ca 2ϩ -free buffer three times. After washing, permeable cells were treated with various buffers at room temperature for 3 or 30 min: Ca 2ϩ -free buffer, Ca 2ϩ buffer alone, Ca 2ϩ buffer plus 10 mM wortmannin (Sigma), and Ca 2ϩ buffer plus 4 mM concentraetion of a peptide (KKRAARATSNVFS-amide, synthesized by Genosys Biotechnologies, Inc.) containing the sequence around the phosphorylatable serine in the smooth muscle RLC (33), respectively. The Ca 2ϩ buffer solution included 10 mM Ca 2ϩ and 100 nM calmodulin with or without 3 mM okadaic acid, a protein phosphatase type 1 and 2A inhibitor (Calbiochem). In addition to other chemicals described above, after treatments the solutions were aspirated completely. Ice-cold 10% trichloroacetic acid and 10 mM dithiothreitol were added, and cells were frozen by putting culture dishes into liquid nitrogen immediately. Cell protein was collected and incubated on ice for 20 min, and insoluble proteins were collected by centrifugation at 4,700 ϫ g for 1 min. Pellets were washed with ethyl ether (Fisher) three times and resuspended in 8 M urea sample buffer as described previously (34). The samples were analyzed by urea-glycerol PAGE and immunoblotting using a monoclonal antibody against smooth muscle RLC (34). The relative amounts of nonphosphorylated, monophosphorylated, and diphosphorylated RLCs were measured by quantitative immunoblots.

Myosin Light Chain Kinase Binding to Smooth Muscle Myofilaments and F-actin on Nitrocellulose-coated Coverslips-Ex-
tracted smooth muscle myofilaments and purified smooth muscle F-actin mixed with BODIPY FL phallacidin were bound to nitrocellulose-coated coverslips. Cy3-labeled myosin light chain kinase was added to the filaments to test its binding ability. Fig. 1 shows phallacidin binding to actin in both smooth muscle myofilaments and smooth muscle F-actin. The Cy3-labeled myosin light chain kinase strongly bound only to smooth muscle myofilaments (Fig. 1). These results are similar to previous results obtained with a cosedimentation assay where myosin light chain kinase bound to gizzard smooth muscle myofila-ments with a greater affinity compared with purified smooth muscle F-actin (21).

Expression and Biochemical Properties of MLCK-GFP in Intact A7r5
Smooth Muscle Cells-To study the physiological effects and binding properties of myosin light chain kinase in cells, MLCK-GFP fusion protein expression vectors were constructed. The cDNA encoding GFP was fused at the C terminus of full-length (FL) or N-terminal deleted (DN) myosin light chain kinase ( Fig. 2A). The N-terminal 2-142 amino acids were deleted in DN MLCK-GFP. This region is within the N-terminal half of the kinase, which is responsible for high affinity myofilament binding in vitro and in vivo (21) and contains the actin binding segment (14 -16). GFP, FL MLCK-GFP, and DN MLCK-GFP were transiently expressed in A7r5 smooth muscle cells (Fig. 2B). As expected, fluorescence from the 27-kDa GFP was diffuse and was found in both the nucleus and cytoplasm in transfected cells (Fig. 2B, left part). The 153-kDa FL MLCK-GFP was localized primarily to filaments with some diffuse cytoplasmic fluorescence (Fig. 2B, middle part). The amount of diffuse cytoplasmic fluorescence was greater in cells expressing larger amounts of the kinase. The 137-kDa DN MLCK-GFP showed only diffuse fluorescence in the cytoplasm without apparent binding to filaments (Fig. 2B, right part). Both FL MLCK-GFP and DN MLCK-GFP were excluded from nuclei, most likely due to their masses. MLCK-GFP with deletion of the kinase C terminus (amino acids 1004 -1147) was also expressed and found primarily on filaments similar to FL MLCK-GFP (data not shown). These results show that the site for high affinity binding resides in residues 2-142 and that the GFP tag at the C terminus of full-length myosin light chain kinase did not affect kinase binding to stress fibers in cells. Similar results were also obtained in transfected NIH 3T3 and CV1 fibroblast cells (data not shown).
The specific activity of FL MLCK-GFP from transfected cell lysates was compared with full-length myosin light chain kinase purified from Sf9 cells (Fig. 3A). The kinase activity of FL MLCK-GFP was Ca 2ϩ /calmodulin-dependent, and the specific activity was 1280 pmol/min/pmol, comparable with that of purified full-length myosin light chain kinase without GFP (1980 pmol/min/pmol). The ability of FL MLCK-GFP to bind to smooth muscle myofilaments in vitro was also measured by the co-sedimentation assay. Cell lysates containing 5 nM FL MLCK-GFP were used to compare the binding ability with FIG. 1. Myosin light chain kinase binding to smooth muscle myofilaments and F-actin on nitrocellulosecoated coverslips. Extracted smooth muscle myofilaments and purified F-actin filaments (1 mg/ml) were mixed with BODIPY FL phallacidin (5 units; Molecular Probes) in a buffer containing 10 mM MOPS at pH 7.0, 50 mM NaCl, 2 mM dithiothreitol, 1 mg/ml bovine serum albumin, and 1 mM MgCl 2 to identify actincontaining filaments. The respective filamentous solutions were adhered to nitrocellulose-coated coverslips, and Cy3-labeled full-length myosin light chain kinase was added. After incubation for 10 min, coverslips were washed three times to remove the unbound kinase. Filaments were double imaged under a fluorescence microscope to detect BODIPY fluorescein and Cy3 fluorescence (for details, see "Experimental Procedures"). Typical fluorescence micrographs are presented for three experiments.
purified full-length myosin light chain kinase to detergentwashed myofilaments from gizzard smooth muscle (Fig. 3B). Myosin light chain kinase with or without GFP bound smooth muscle myofilaments similarly under these conditions. Thus, the introduction of green fluorescent protein at the C terminus of myosin light chain kinase had no significant effect on kinase catalytic or binding properties. These results indicate that the FL MLCK-GFP may be used to characterize the cellular binding properties of myosin light chain kinase in cells.
The concentration of expressed FL MLCK-GFP in a single A7r5 cell was estimated by flow cytometry and immunoblotting (data not shown). The estimated FL MLCK-GFP concentration was 1.1 Ϯ 0.3 mM, which is in a range similar for the endogenous myosin light chain kinase in smooth muscle cells in tissues (3-4 mM), although it is at least 10 times greater than the amounts found in nonmuscle cells (35).

Myosin Light Chain Kinase Binding in Permeable A7r5
Cells-To examine Ca 2ϩ regulation of myosin light chain kinase binding to cellular actomyosin filaments, transfected A7r5 cells were made selectively permeable by saponin. A7r5 cells were transfected with DN MLCK-GFP or FL MLCK-GFP and intact cells were imaged (Fig. 4, top panels). After treatment with saponin for 10 min, the diffuse cytoplasmic fluorescence due to DN MLCK-GFP was completely removed, indicating release of the kinase from cells (Fig. 4, middle left panel). This was not due to loss of actin-containing stress fibers as shown by staining with rhodamine-labeled phalloidin (Fig. 4, bottom left  panel). In the FL MLCK-GFP-transfected A7r5 cells, the diffuse fluorescence was released from the cells by saponin treatment, while the filament-associated FL MLCK-GFP remained (Fig. 4, middle right panel).
Effect of Ca 2ϩ /Calmodulin on Myosin Light Chain Kinase

Binding to Filaments in Permeable Cells-To test whether
Ca 2ϩ /calmodulin affects myosin light chain kinase binding in cells, fluorescence intensities of FL MLCK-GFP bound to filaments in saponin-permeable cells were compared in the presence of either 4 mM EGTA or 10 mM Ca 2ϩ plus 1 mM calmodulin and 10 mM wortmannin (Fig. 5A). Although calmodulin is not completely removed from cellular filaments by EGTA (23), calmodulin was added to the buffer containing Ca 2ϩ to assure a sufficient amount for binding to the kinase. Wortmannin was added to the Ca 2ϩ -containing buffer to prevent the robust contraction of cell filaments for fluorescence measurements. Cells were either treated with EGTA buffer first and then washed with the Ca 2ϩ /calmodulin/wortmannin buffer or treated with Ca 2ϩ /calmodulin/wortmannin buffer first and then washed with the EGTA buffer. The fluorescence intensities of FL MLCK-GFP remained unchanged in both circumstances (Fig.  5A). Similar results were obtained by comparison with EGTA buffer and 1 mM Ca 2ϩ /100 nM calmodulin buffer (data not shown). Thus, myosin light chain kinase binding to actin-containing stress fibers is not released by Ca 2ϩ /calmodulin.
To test whether myosin light chain kinase may diffuse laterally along filaments, the mobility of FL MLCK-GFP was examined in the saponin-permeable A7r5 cells by FRAP (Fig.  5B, closed symbols). After photobleaching, there was no recovery of fluorescence for up to 15 min, indicating that FL MLCK-GFP bound to filaments with no evidence of mobility. There was no fluorescence recovery in either EGTA buffer or 10 mM Ca 2ϩ , 1 mM calmodulin, 10 mM wortmannin after photobleaching. Similar results were obtained with the addition of 1 mM Ca 2ϩ , 100 nM calmodulin buffer for 80 s (data not shown). Additionally, similar results for measurements of fluorescence intensities and FRAP were obtained by perfusion of Cy3-labeled myosin light chain kinase in Triton X-100-permeable stress fibers in Swiss 3T3 cells (data not shown). In contrast, the fluorescence of 167-kDa fluorescein-labeled dextran in ␤-escin-permeable cells was recovered 100% in 1 min after photobleaching in either EGTA or Ca 2ϩ /calmodulin (Fig. 5B, open  symbols). These data collectively indicate that myosin light chain kinase binds to actin-containing filaments with high affinity and does not dissociate or diffuse along the filaments in the presence of Ca 2ϩ /calmodulin.
Cell Contraction and RLC Phosphorylation-To determine if filament-bound myosin light chain kinase could phosphorylate myosin RLC in thick filaments in cells, cell contractility was first measured in the presence of Ca 2ϩ . Fig. 6A (top to bottom  parts) shows intact cells transfected with FL MLCK-GFP, followed by saponin-permeabilization in Ca 2ϩ -free buffer containing Mg 2ϩ ATP. The subsequent addition of Ca 2ϩ in the presence of Mg 2ϩ ATP results in contraction of filaments containing bound-myosin light chain kinase. This Ca 2ϩ -dependent contraction was inhibited in the presence of 10 mM wortmannin, a myosin light chain kinase inhibitor (Fig. 6B), or 4 mM RLC peptide, a competitive substrate (Fig. 6C).
To determine whether RLC phosphorylation mediates this Ca 2ϩ -dependent contraction by the filament-bound kinase, RLC phosphorylation was measured in the saponin-permeabilized cells (Fig. 7). Nonphosphorylated, monophosphorylated, and diphosphorylated RLCs were separated by urea-glycerol PAGE and measured by immunoblotting (Fig. 7A). As expected, RLC phosphorylation in saponin-permeable cells in Ca 2ϩ -free buffer alone for 3 min was low (Fig. 7, A and B). When a protein phosphatase inhibitor, okadaic acid, was added to the Ca 2ϩfree buffer, RLC phosphorylation increased from 0.14 to 0.43 mol of phosphate/mol of RLC. When wortmannin was added with okadaic acid to the Ca 2ϩ -free buffer for 3 min, RLC phosphorylation was similar (0.41 mol of phosphate/mol of RLC) to the extent of phosphorylation obtained with okadaic acid alone (Fig. 7, A and B). Thus, there was apparently some myosin light chain kinase-independent phosphorylation of RLC. However, inclusion of 10 mM Ca 2ϩ and 100 nM calmodulin with okadaic acid resulted in a greater extent of RLC phosphorylation, 0.87 mol of phosphate/mol of RLC (Fig. 7, A and B). This Ca 2ϩ / calmodulin-dependent RLC phosphorylation was abolished by wortmannin or RLC peptide (0.40 and 0.25 mol of phosphate/ mol of RLC, respectively) (Fig. 7, A and B). When the RLC peptide substrate concentration was 50-fold in excess of RLC in a typical kinase assay, the myosin light chain kinase activity toward RLC was inhibited greater than 90% (data not shown).
In the absence of Ca 2ϩ , RLC phosphorylation was 0.17 mol of phosphate/mol of RLC at 30 min after saponin treatment, results similar to those obtained at 3 min (Fig. 7, A and B). In the presence of Ca 2ϩ /calmodulin but in the absence of okadaic acid, the extent of RLC phosphorylation increased to 0.60 mol of phosphate/mol of RLC at 30 min. This value increased to 1.60 mol of phosphate/mol of RLC when okadaic acid was added (Fig. 7, A and B). These results suggest that filament-bound myosin light chain kinase phosphorylates myosin RLC to initiate cell contraction. DISCUSSION Previous investigations showed that myosin light chain kinase bound to F-actin and F-actin-containing filaments in cells (14 -20). Although it was assumed the binding in cells was to F-actin, a recent report indicated a significant difference in FIG. 4. Myosin light chain kinase binding in permeable A7r5 cells. DN MLCK-GFP (left panels) and FL MLCK-GFP (right panels) in transfected A7r5 cells were imaged before permeabilization (Intact, top panels). After imaging the intact cells, cells were treated with 0.02% saponin in Ca 2ϩ -free buffer (20 mM PIPES at pH 6.8, 4 mM EGTA, 5 mM MgSO 4 , 90 mM K ϩ -gluconate, 5.3 mM Na 2 ATP, 0.1% bovine serum albumin, 0.1 mM ionomycin, 1.5 mM thapsigargin, 0.1 mM phenylmethylsulfonyl fluoride, and 10 mg/ml leupeptin) at 37°C for 10 min. After permeabilization, cells were washed three times with the Ca 2ϩ -free buffer without saponin, and the same cells were imaged (Saponin, middle panels). Rhodamine-labeled phalloidin was added to the Ca 2ϩfree buffer for 2 min and then washed with the Ca 2ϩ -free buffer three times following by imaging (phalloidin, bottom panels). apparent affinities for the two types of filaments (21). Using a cosedimentation assay, Lin et al. (21) showed that the kinase binding affinity was greater for detergent-washed smooth muscle myofilaments compared with purified F-actin or detergentwashed myofilaments from skeletal muscle. This observation was extended herein with fluorescence imaging of Cy3-labeled kinase binding to purified F-actin or gizzard myofilaments. Myosin light chain kinase bound to gizzard myofilaments with no significant binding to F-actin. These results are consistent with high affinity binding of the kinase to myofilaments, perhaps due to an anchoring or accessory protein. They also raise a question as to whether this binding is regulated by Ca 2ϩ / calmodulin like the low affinity binding of kinase to purified F-actin.
A fusion protein containing myosin light chain kinase and GFP (MLCK-GFP) was expressed to monitor kinase binding to actin-containing filaments in cells. It has similar catalytic and binding properties as purified smooth muscle myosin light chain kinase and therefore may be used to characterize binding of myosin light chain kinase to actin-containing stress fibers in cells. FL MLCK-GFP, although primarily localized to the filaments in intact cells, showed some diffuse cytoplasmic fluorescence that depended on the extent of expression. This cytoplasmic fluorescence may be due to saturation of kinase binding sites on stress fibers in A7r5 cells, which is supported by the observations that it diffused out when cells were treated with saponin. Smooth muscle cells in tissues have a greater capacity for binding myosin light chain kinase, because previous reports showed that the kinase bound completely to skinned fibers (23) or detergent-washed myofilaments (36). The FL MLCK-GFP on actin-containing filaments remains bound after saponin permeabilization. These results obtained from MLCK-GFP-transfected A7r5 cells are similar to those in Triton X-100-permeable fibroblasts and intact smooth muscle cells perfused or microinjected with Cy3-labeled myosin light chain kinases (herein, and see Ref. 21).
Binding of myosin light chain kinase to actin-containing filaments in permeable cells is not affected by Ca 2ϩ /calmodulin based on the comparison of fluorescence intensities in the presence or absence of Ca 2ϩ /calmodulin. In addition, FRAP measurements in permeable cells indicate that myosin light chain kinase does not diffuse laterally along the filaments. Previous studies in vitro showed the binding affinity of myosin light chain kinase for purified F-actin or purified myosin was decreased in the presence of Ca 2ϩ /calmodulin (15,18). Additionally, Ca 2ϩ /calmodulin also decreased the binding affinity of the N-terminal fragment (1-114 residues) of the kinase for purified F-actin and another potential Ca 2ϩ /calmodulin binding site was localized within the N-terminal actin binding sequence (residues 26 -41) (15). However, the binding affinities (10 5 to 10 6 M Ϫ1 ) of myosin light chain kinase and its actin binding sequence for F-actin alone are low (14,15,18) relative to the almost irreversible binding of the kinase in skinned fibers (23) and detergent-washed myofilaments (21, 36) from smooth muscles. Deletion of the N-terminal 142 residues of MLCK-GFP prevents binding to actin-containing filaments in cells consistent with previous results (21) when the N-terminal half of myosin light chain kinase was deleted. These results indicate that a recently identified Ca 2ϩ /calmodulin-insensitive binding site in myosin light chain kinase (residues 319 -721 preceding the catalytic core; Ref. 15) is not responsible for binding to filaments in cells. These results also show that the anchoring motif is in the N terminus of myosin light chain kinase. Importantly, this high affinity binding of myosin light chain kinase to actin-containing filaments in saponin-permeable A7r5 cells and in Triton X-100-permeable Swiss 3T3 fibroblasts was not affected by Ca 2ϩ /calmodulin. Although Ca 2ϩ /calmodulin is required for activation of catalysis, it does not significantly dissociate myosin light chain kinase from cellular filaments.
Filament-bound myosin light chain kinase in saponin-permeabilized cells is sufficient for RLC phosphorylation and cell contraction in the presence of Ca 2ϩ /calmodulin. Wortmannin and a competitive peptide substrate inhibited Ca 2ϩ -dependent RLC phosphorylation and smooth muscle cell contraction in saponin-permeable cells. Previous studies have shown that wortmannin is a potent and selective inhibitor for myosin light chain kinase and phosphatidylinositol 3-kinase, which is involved in formation of inositol 1,4,5-trisphosphate and Ca 2ϩ signaling (37,38). However, in experiments reported here, inhibition of phosphatidylinositol 3-kinase is not an important consideration, since this kinase does not phosphorylate myosin RLC, and a permeable cell system was used so that filamentbound myosin light chain kinase could be activated directly by the addition of Ca 2ϩ and calmodulin. Wortmannin has little or no effect on the activities of other kinases including protein kinase C, cAMP-dependent protein kinase, cGMP-dependent protein kinase, calmodulin-dependent protein kinase II, tyrosine kinases, and phosphatidylinositol 4-kinase (37,38). In saponin-permeabilized cells, both wortmannin and RLC peptide inhibited Ca 2ϩ -dependent RLC phosphorylation and cell contraction, which is consistent with the idea that these coupled events (3-7) are due to filament-bound myosin light chain kinase.
In our studies, three forms (non-, mono-, and diphosphorylated) of RLC were observed in saponin-permeable cells in the presence of Ca 2ϩ /calmodulin and/or okadaic acid. Phosphorylation at Ser-19 of smooth muscle myosin RLC by myosin light chain kinase has been identified both in vitro and in smooth muscle tissues as the major monophosphorylation site important for actin-activated MgATPase activity and myosin motility (25,39). However, Thr-18 is also phosphorylated but at a slower rate by myosin light chain kinase in vitro, leading to diphosphorylated RLC (40,41). In contracting smooth muscle tissues, monophosphorylated and diphosphorylated forms of myosin RLC were also observed (34). However, because phosphorylation at Thr-18 has no effect on the velocity of myosin movement in motility assays in vitro (42), the physiological importance of diphosphorylated RLC needs clarification by further investigations.
An interesting result was observed in saponin-permeable cells in Ca 2ϩ -free buffer. RLC phosphorylation was increased in Ca 2ϩ -free buffer containing okadaic acid, a known phosphatase type 1 and 2A inhibitor (26). In addition, wortmannin did not inhibit this RLC phosphorylation. These results suggest that some low kinase activity toward RLC is independent of the Ca 2ϩ /calmodulin-dependent myosin light chain kinase pathway. Ca 2ϩ -independent RLC phosphorylation in the presence of a protein phosphatase inhibitor such as okadaic acid has also been reported for smooth muscle tissues (26). Recent studies also showed in vitro and in vivo that protein kinase C and Rho-associated kinase (Rho-kinase) may phosphorylate RLC FIG. 7. RLC phosphorylation by filament-bound myosin light chain kinase in permeable A7r5 cells. FL MLCK-GFP-transfected A7r5 cells were permeabilized with 0.02% saponin in Ca 2ϩ -free buffer at 37°C for 10 min and washed with Ca 2ϩ -free buffer three times. After washing, cells were treated at room temperature for 3 or 30 min with Ca 2ϩ -free buffer or 10 mM Ca 2ϩ buffer containing the following components: 3 mM okadaic acid, 10 mM wortmannin, and 4 mM RLC peptide. A, representative immunoblots of RLC phosphorylation after ureaglycerol PAGE. Migrations of the nonphosphorylated (non-P), monophosphorylated (mono-P), and diphosphorylated (di-P) forms of RLC are indicated. Lanes 1-9 correspond to the conditions in B from left to right. B, quantitation of phosphorylated RLC. The data represent the means Ϯ S.E. for three or four experiments. (43)(44)(45). Although protein kinase C phosphorylated Thr-9, Ser-1, and/or Ser-2 of myosin RLC in vitro, it did not increase actin-activated Mg 2ϩ ATPase activity or myosin motility (42,44). Smooth muscle contraction still occurred at low Ca 2ϩ concentrations, suggesting that a Ca 2ϩ -independent isoform of protein kinase C⑀ could conceivably phosphorylate RLC (46). On the other hand, Rho-kinase activated by GTP-bound RhoA may increase the extent of RLC phosphorylation and induce smooth muscle contraction in the absence of Ca 2ϩ (45). Increasing RLC phosphorylation by Rho-kinase appears to be primarily indirect through the inactivation of myosin phosphatase by phosphorylation of the myosin-binding subunit of the phosphatase (47). However, very little Rho-kinase was present in Triton X-100-permeable fibers (45), and both inactive and active Rhokinases did not appear to be localized to actomyosin-containing fibers based on immunofluorescence staining (48). Additional investigations are needed to identify specifically this Ca 2ϩindependent kinase and its potential physiological relevance.
In summary, our results suggest that myosin light chain kinase bound to actin-containing filaments with a high affinity at its N terminus is sufficient for RLC phosphorylation and cell contraction when activated by Ca 2ϩ /calmodulin. This raises an interesting question as to how the kinase phosphorylates myosin RLC in thick filaments if it is bound to actin-containing thin filaments. Structural studies by sedimentation velocity show that axial ratios of full-length myosin light chain kinase are 18.0 and 16.8 in the absence and presence of Ca 2ϩ /calmodulin, respectively (49). Based on sedimentation velocity data, the calculated length of the recombinant smooth muscle myosin light chain kinase is 540 Å. 2 Thus, the shape of a full-length myosin light chain kinase molecule is elongated either in the presence or absence of Ca 2ϩ /calmodulin. Myosin light chain kinase may reach and phosphorylate RLC on myosin thick filaments while it remains attached to actin-containing thin filaments. In smooth muscle, the surface-to-surface distance from thick and thin filaments is approximately 150 Å (50). It has been shown by small angle x-ray and neutron scattering that myosin light chain kinase containing the catalytic core and regulatory segment forms an elongated ellipsoid with a maximum linear dimension of 95 and 78 Å in the absence and presence of Ca 2ϩ /calmodulin, respectively (51). The catalytic cleft is in the middle of these ellipsoids. Smooth muscle myosin light chain kinase contains structural motifs between the N terminus and catalytic core, including the tandem repeat region and two immunoglobulin-like (Ig) and one fibronectin-like (Fn) motifs ( Fig. 2A). The three-dimensional structures of individual Ig and Fn motifs are known (52,53), and calculations show that three of these motifs would extend the catalytic core at least 130 Å. The three-dimensional structures of the tandem repeat region, the intervening sequences between the immunoglobulin-like and fibronectin-like motifs, and the N terminus of myosin light chain kinase are not known, but clearly the intervening distance between the N-terminal binding segment and the substrate-binding cleft in the catalytic core seems more than sufficient to span the 150 Å between thin and thick filaments. Although a plausible hypothesis, additional investigations are needed to understand how myosin light chain kinase bound to actin-containing filaments in cells phosphorylates myosin light chain in thick filaments.