Differential Regulation of Alternatively Spliced Endothelial Cell Myosin Light Chain Kinase Isoforms by p60Src *

The Ca2+/calmodulin-dependent endothelial cell myosin light chain kinase (MLCK) triggers actomyosin contraction essential for vascular barrier regulation and leukocyte diapedesis. Two high molecular weight MLCK splice variants, EC MLCK-1 and EC MLCK-2 (210–214 kDa), in human endothelium are identical except for a deleted single exon in MLCK-2 encoding a 69-amino acid stretch (amino acids 436–505) that contains potentially important consensus sites for phosphorylation by p60Src kinase (Lazar, V., and Garcia, J. G. (1999) Genomics 57, 256–267). We have now found that both recombinant EC MLCK splice variants exhibit comparable enzymatic activities but a 2-fold reduction ofV max, and a 2-fold increase inK 0.5 CaM when compared with the SM MLCK isoform, whereas K m was similar in the three isoforms. However, only EC MLCK-1 is readily phosphorylated by purified p60 Src in vitro, resulting in a 2- to 3-fold increase in EC MLCK-1 enzymatic activity (compared with EC MLCK-2 and SM MLCK). This increased activity of phospho-MLCK-1 was observed over a broad range of submaximal [Ca2+] levels with comparable EC50 [Ca2+] for both phosphorylated and unphosphorylated EC MLCK-1. The sites of tyrosine phosphorylation catalyzed by p60Src are Tyr464 and Tyr471 within the 69-residue stretch deleted in the MLCK-2 splice variant. These results demonstrate for the first time that p60Src-mediated tyrosine phosphorylation represents an important mechanism for splice variant-specific regulation of nonmuscle MLCK and vascular cell function.

The family of myosin light chain kinases (MLCK) 1 expressed in vertebrates are Ca 2ϩ /calmodulin-regulated enzymes that catalyze the transfer of phosphate from Mg 2ϩ -ATP to a serine residue (Ser 19 ) of regulatory myosin light chain (MLC 20 ) (1,2). Members of this MLCK family share structural similarity, with a catalytic core that binds Mg 2ϩ -ATP and MLC 20 and a regulatory segment involved in Ca 2ϩ /calmodulin-dependent activation. Comparisons of primary sequences deduced from cDNA clones demonstrate that skeletal and cardiac muscle MLCK isoforms represent gene products that differ from the gene encoding the smooth muscle and nonmuscle MLCK isoforms localized on human chromosome 3 (3)(4)(5)(6). Furthermore, the physiological role of the gene-specific MLCK isoforms differs significantly in the regulation of actomyosin contraction. In skeletal and cardiac muscles, Ca 2ϩ binds to the regulatory thin filament protein complex containing troponin and tropomyosin and thus allows actin to activate sarcomeric myosin Mg 2ϩ -ATPase. Although MLCK does not initiate muscle contraction in these tissues, MLCK-mediated phosphorylation of MLC 20 may potentiate the rate and extent of force development (7,8). In contrast, Ser 19 phosphorylation of MLC 20 by Ca 2ϩ /calmodulin-dependent enzyme MLCK is essential for the initiation of nonmuscle and smooth muscle contraction (1, 3, 9 -11). In specific nonmuscle tissues, such as the vascular endothelium, this kinase is known to be involved in endothelial cell migration, cell retraction (12), endothelial cell barrier regulation (13), transendothelial migration of neutrophils (14,15), and possibly apoptosis (16). The MLCK isoform abundantly expressed in smooth muscle (SM MLCK) generally exists as a 130-to 150-kDa protein that has been well characterized (for review see Refs. 3 and 11). However, Western blot screening of a variety of embryonic and adult smooth muscle and nonmuscle tissues revealed expression of a high molecular weight MLCK variant with electrophoretic mobility in the range of 208 -214 kDa (17)(18)(19). Garcia and colleagues (20) subsequently sequenced the high molecular weight MLCK isoform cloned from a human endothelial cell cDNA library, revealing an open reading frame, which encodes a protein of 1914 amino acids. Both the low molecular mass (130 -150 kDa) and high molecular mass (208 -214 kDa) MLCK isoforms share essentially identical actin binding, MLC binding, catalytic, and Ca 2ϩ /CaM-regulatory domains. The extreme C-terminal kinase-related protein (KRP) domain, which binds myosin, is contained within both EC MLCK and SM MLCK but can be also expressed as an independent protein capable of stabilizing myofilaments in vitro (3,(21)(22)(23). The C-terminal half of the endothelial MLCK isoform (residues 923-1914) exhibits 99.8% homology to the human low molecular weight MLCK from hippocampus and substantial homology to published SM MLCK sequences from rabbit (94% homology), bovine (95% homology), and chicken (85% homology) (5,6,24,25). However, the exact biological function of the 922-amino acid N-terminal portion, which is unique to the high molecular weight MLCK isoform, is completely unknown. We have previously found that increased levels of endothelial cell protein tyrosine phosphorylation evoked by thrombin or diperoxovanadate are tightly linked to increased MLC 20 phosphorylation, activation of actomyosin contraction, and a dramatic decrease in endothelial cell barrier function (26,27). Further studies demonstrated that the increased kinase activity in MLCK immunoprecipitates strongly correlated with increased EC MLCK phosphorylation on tyrosine residues (26). Endothelial cell MLCK was found to be stably associated with p60 Src kinase after stimulation (26), consistent with potential direct regulation of endothelial MLCK activity by tyrosine phosphorylation. More recently, detailed analysis of MLCK transcripts expressed in human endothelial cells revealed several splice variants of the EC MLCK isoform with predominant expression of the full-length isoform (MLCK-1) and a variant, which is identical to MLCK-1 except for a deleted 69-residue stretch (amino acid residues 437-505) encoded by a single exon (MLCK-2) (28). Within this deleted 69-amino acid stretch is an SH2-binding domain and consensus sites for phosphorylation by the Src family kinases (Tyr 464 , Tyr 485 ). To better understand the role and significance of tyrosine phosphorylation and the function of the novel N terminus in EC MLCK regulation, we have expressed both EC MLCK-1 and EC MLCK-2 isoforms as well as smooth muscle MLCK in the baculovirus system and have characterized the biochemical properties of the purified recombinant proteins. Furthermore, we have assessed the phosphorylation of MLCK-1, MLCK-2, and SM MLCK by p60 Src kinase in vitro, mapped the p60 Src -phosphorylation sites to Tyr 464 and Tyr 471 of MLCK-1, and investigated the effects of MLCK-1 phosphorylation by p60 Src kinase on the MLCK enzymatic activity and regulation by Ca 2ϩ /calmodulin. Our studies indicate the novel up-regulation of high molecular weight endothelial MLCK-1 isoform activity by p60 Src -induced tyrosine phosphorylation and demonstrate isoform-specific endothelial cell MLCK regulation.

Recombinant Donor Plasmids for Baculovirus Expression
Smooth-muscle MLCK-A rabbit uterine smooth muscle MLCK full-length cDNA in a pGEM vector (24) was a generous gift from Dr. Patricia Gallagher (Indiana University). The plasmid was cut at the Eco52I site, followed by blunt-ending with Klenow enzyme. The MLCK insert was then released by digesting plasmid with XbaI, separated from the vector by agarose gel electrophoresis, and purified using a Prep-A-Gene (Bio-Rad, Hercules, CA) kit. The smooth muscle MLCK cDNA was ligated with the pFastBac Hta baculovirus donor plasmid (Bac-To-Bac baculovirus expression system, Life Technologies), which had been digested with StuI and XbaI enzymes.
Nonmuscle MLCK-1 and MLCK-2-MLCK-2 cDNA was obtained from human umbilical vein endothelial cells (HUVEC) by RT-PCR, amplified, and subcloned as described earlier (28), using sense primer 5Ј-ACT GAA TTC ACC ATG GGG GAT GTG AAG CTG and antisense primer 5Ј-GTC AGA ATT CTT GTT TCA CTC TTC TTC CTC TTC C, both containing an EcoRI site. The EC MLCK-2 cDNA was inserted into pFastBac Hta vector at the EcoRI site. To generate MLCK-1 cDNA, an ϳ1.85-kb 5Ј-end fragment of HUVEC MLCK was amplified by RT-PCR using a SuperScript preamplification system (Life Technologies). The first-strand cDNA was synthesized from HUVEC total RNA with oligo-dT primer. The PCR primers were as follows, sense: 5Ј-ACT GCG GCC GCA CCA TGG GGG ATG TGA AGC TG (NotI restriction site followed by HUVEC MLCK-specific 5Ј-end sequence), and antisense: 5Ј-GTA CTC ACT CTT CCT GCT ACT C (ϳ1.85-kb downstream HU-VEC MLCK sequence). The same region (with a 207-bp deletion compared with EC MLCK-1) of the nonmuscle MLCK-2/pFastBac Hta plasmid was excised and replaced with this PCR-amplified fragment encoding the N-terminal portion of MLCK-1. Briefly, the PCR product was digested with NotI, blunt-ended, and cut by BlpI. The respective ϳ1.66-kb fragment from the nonmuscle MLCK-2/pFastBac Hta plasmid was excised with EheI and BlpI, and separated on agarose gel. The PCR-amplified 5Ј-end fragment of MLCK-1 and the 3Ј-end MLCK region previously subcloned into pFastBac Hta vector were ligated to create the recombinant donor plasmid of the whole EC MLCK-1 coding region reported previously (20). All three constructs (SM MLCK, EC MLCK-1, and EC MLCK-2) were verified by restriction analysis, PCR, and complete sequencing.
Recombinant Regulatory Myosin Light Chain-A plasmid pET3-MLC encoding rabbit vascular smooth muscle regulatory MLC was generously provided by Dr. Patricia Gallagher (Indiana University) and amplified by PCR using primers with unique restriction sites for directional cloning, MLC-sense-EcoRI: 5Ј-TGC TTT GAA TTC ATG TCC AGC AAG CGG GCC AAA GCC AAG-3Ј, and MLC-antisense-XbaI: 5Ј-AAG GAC TCT AGA CTA GTC GTG TTT ATC CTT GGC GCC ATG-3Ј, and then ligated with the pFastBac Hta baculovirus donor plasmid (Bac-To-Bac baculovirus expression system), which had been digested with EcoRI and XbaI enzymes.

Baculovirus Expression of MLC and MLCK Isoforms
Human endothelial cell MLCK-1 and MLCK-2, rabbit smooth muscle MLCK and MLC recombinant baculovirus stocks were prepared using the Bac-To-Bac baculovirus expression system (Life Technologies) according to manufacturer's instructions. The system uses site-specific transposition of foreign genes into a baculovirus shuttle vector, bacmid, propagated in Escherichia coli. Sf9 and Hi5 insect cell suspension cultures at 2 ϫ 10 6 cells/ml were infected with the respective viral stock at a multiplicity of infection range of 0.1-10, and after 1-h incubation, diluted 5-fold with fresh media and grown for 2-4 days at 28°C with continuous shaking. The optimal conditions varied for the three different MLCK isoforms expressed (not shown). For large scale expression and purification, Sf9 cells were infected with baculovirus (multiplicity of infection ϭ 1), and the cells producing SM MLC, rabbit SM MLCK, or human endothelial MLCK-1 and -2 were harvested. The recombinant MLCK and MLC proteins contained a histidine tag at their N-terminal region introduced during subcloning MLCK cDNAs into the pFastBac Hta baculovirus donor vector. Expression of the MLCK isoforms was confirmed by SDS-polyacrylamide gel electrophoresis (29) and Western blot (30) using MLCK-specific D119 antiserum (31) or commercial anti-His tag antibodies.

Purification of Recombinant Proteins
For the isolation of the recombinant MLCK isoforms or MLC, the infected Sf9 cells were harvested by centrifugation at 3000 ϫ g for 5 min and frozen at Ϫ80°C. Frozen insect cells were lysed (1:5 w/v ratio) in ice-cold lysis buffer (50 mM Tris-HCl, pH 8.5, 5 mM 2-mercaptoethanol, 100 mM KCl, 1 mM phenylmethylsulfonyl fluoride, and 1% Nonidet P-40) at 4°C for 2 min. The lysate was centrifuged at 10,000 ϫ g for 10 min, and the supernatant was loaded onto Ni-NTA resin (Qiagen, Santa Clarita, CA). After a wash step with buffer A (20 mM Tris-HCl, pH 8.5, 500 mM KCl, 5 mM 2-mercaptoethanol, 10% glycerol), the expressed MLCK isoforms were eluted with 100 mM imidazole, 20 mM Tris-HCl, pH 8.5, 100 mM KCl, 5 mM 2-mercaptoethanol, 10% glycerol. The protein concentration was determined by Bio-Rad protein assay. The yield of MLCK isoforms ranged from 1.5 to 4 mg of MLCK from a 10-g Sf9 cell pellet. The purified enzymes were aliquoted and stored at Ϫ80°C.

Verification and Characterization of MLCK Constructs
Human endothelial MLCK-1 and MLCK-2 splice variants as well as the rabbit smooth muscle MLCK isoform cDNAs were subcloned into baculovirus vector pFastBac Hta as described above. The alternatively spliced regions in MLCK-1 and MLCK-2 mRNAs were verified by PCR amplification (not shown) using primers located upstream and downstream of nucleotides 1428 -1634 as well as by complete sequencing of the MLCK-1 and MLCK-2 cDNA inserts. Translation of completely sequenced cDNA inserts encoding EC MLCK splice variants revealed four amino acid residue differences between the unique N-terminal sequence of the EC MLCK-1 and MLCK-2 (Phe 629 , Cys 681 , Gly 714 , and Leu 806 ) and the previously reported EC MLCK cDNA sequence (Gen-Bank accession number U48959). These sequence differences, which may represent polymorphisms within the human MLCK gene, did not involve either the consensus sequence for potential tyrosine phosphorylation catalyzed by p60 Src (Tyr 464 , Tyr 471 , Tyr 485 ), the putative SH2binding sites (Tyr 59 , Tyr 464 ), the SH3 domains (Pro 314 -Arg 318 , Arg 373 -Pro 379 ) previously proposed for EC MLCK (26,28), and did not affect MLCK enzymatic properties (shown below). Rigorous analysis of the cDNA sequences encoding the C terminus of EC MLCK common to both high and low molecular weight MLCK isoforms, revealed three variances (Phe 925 /Leu, Ala 1179 /Val, and Lys 1233 /Glu) in the baculovirusexpressed recombinant MLCK-1 and MLCK-2 isoforms that do not correspond to GenBank accession number 48959 or to homologous regions of the reported human, rabbit, and bovine SM MLCK variants, respectively (6,24,25). Leu 925 resides within the putative actin-binding domain spanning residues 910 -1036 of MLCK-1, whereas Val 1179 and Glu 1233 do not lie within functional domains described for smooth muscle MLCK (32).

Phosphorylation of MLCK by p60 Src in Vitro
Purified MLCKs were dialyzed and brought to 0.1 mg/ml concentration using reaction buffer containing 25 mM Tris-HCl, pH 7.5, 20 mM KCl, 5 mM Mg 2ϩ acetate, 0.5 mM leupeptin. Phosphorylation of MLCK diluted in reaction buffer was started by adding 0.2 mM ATP, 10 Ci/ml [␥-32 P]ATP and 75 units/ml recombinant p60 src kinase (Upstate Biotechnology, Inc., Lake Placid, NY; final concentrations). Synthetic MLCK-1 peptides were used for p60 src phosphorylation assays at 0.1 mg/ml final concentration. In certain experiments, a putative specific p60 src inhibitor, PP-2 (Calbiochem-Novabiochem Corp., La Jolla, CA), was added to reaction tubes at 500 nM final concentration. The phosphorylation reaction was performed at 22°C, and 10-l aliquots of reaction mixture were applied onto cellulose phosphate filters P81 (Whatman, UK) at specified periods of time. The filters were washed to remove unincorporated label, and specific incorporation of 32 P into MLCK was determined by scintillation counting. The p60 Src phosphorylation of the MLCK synthetic peptides as substrates was performed during 30 min at 22°C under the same conditions. For scintillation counting, synthetic MLCK peptides were spotted onto nitrocellulose, and unbound radioactive label was washed out with solution containing 20% methanol and 2% disodium pyrophosphate. After completion of the phosphorylation reaction, phospho-MLCKs were aliquoted and stored at Ϫ80°C until use in kinase assays.

Tryptic Cleavage and MALDI TOF Analysis of MLCK-1 Phosphopeptides
A 200-l aliquot of MLCK-1 (0.3 mg/ml) phosphorylated by p60 src in vitro was partially digested by incubation with trypsin (1:100 w/w) for 5 min at room temperature in buffer containing 50 mM Tris-HCl, pH 7.5, 5 mM EGTA, 5 mM MgCl 2 . The reaction was terminated by adding phenylmethylsulfonyl fluoride at 2 mM final concentration. The reaction mixture was lysed in SDS-sample buffer, peptides were separated on SDS-polyacrylamide gel electrophoresis, and phosphotyrosine peptides were identified by autoradiography and Western blot with anti-phosphotyrosine antibody. A portion of the polyacrylamide gel containing a major 55-kDa MLCK-1 tryptic peptide, which incorporated 32 P and cross-reacted with anti-phosphotyrosine antibody, was excised and further processed. After 3 ϫ 30 min washes in distilled water, the gel piece was trimmed and the phosphopeptide incorporated into gel was subjected to destaining and complete trypsinolysis. Gel destaining required the addition of 100 l of 1:1 (v/v) acetonitrile:25 mM ammonium bicarbonate for ϳ30 min. The gel was then dried down completely using a Speed Vac concentrator (Savant), and further trypsinolysis was performed. The gel was incubated overnight at 37°C with 1 l of trypsin solution (0.1 g/l in 1% acetic acid) and 25 l of 25 mM ammonium bicarbonate, pH 7.8. After trypsinolysis, the MLCK peptides were eluted from the gel with a 1:1 (v/v) acetonitrile:water, 5% trifluoroacetic acid solution and concentrated to several microliters by Speed Vac. This peptide mixture was then analyzed by mass spectrometry. Mass spectra were acquired on a Kratos Axima CFR (Manchester, United Kingdom) time-of-flight mass spectrometer. Briefly, an aliquot of the peptide digest (0.3 l) was placed on the sample plate followed by the addition of 0.3 l of saturated ammonium sulfate and 0.3 l of matrix solution (saturated solution of ␣-cyano-4-hydroxycinnamic acid in 1:1 (v/v) ethanol:water). The mixture was air-dried (ϳ10 min), and the sample plate was inserted into the mass spectrometer. deletion of a 69-residue stretch encoded by a single exon in MLCK-2 (28). The baculovirus-expressed and -purified MLCK isoforms (MLCK-1, MLCK-2, and SM MLCK) were analyzed by gel electrophoresis and revealed protein bands with expected sizes 214, 206, and 150 kDa, respectively (Fig. 1B), which reacted on the Western blot with MLCK-specific D119 antiserum (Fig. 1C). After purification and rigorous sequencing analysis, the three recombinant MLCK isoforms were examined for their intrinsic enzymatic properties. Both EC MLCK-1 and EC MLCK-2 exhibited comparable V max values (11.9 Ϯ 3.2 and 10.9 Ϯ 1.8 mol/min/mg, respectively), which were slightly reduced compared with the rabbit SM MLCK isoform (17.0 Ϯ 2.5 mol/min/mg) ( Table I). The K m values reflecting substrate affinity were not significantly different among the three recombinant MLCK preparations, whereas the K 0.5 calmodulin was higher for the EC MLCK isoforms (0.49 and 0.42 nM) compared with the rabbit SM MLCK isoform (0.21 nM) (Table I). Overall, the K m , V max , K 0.5 calmodulin values for the baculovirus-expressed endothelial MLCK-1 and MLCK-2 splice variants are in good agreement with published values for MLCK activity derived from other tissues (24,(35)(36)(37). Furthermore, all three recombinant MLCK isoforms possess identical substrate specificity and preferentially phosphorylate Ser 19 and Thr 18 of regulatory MLC 20 (Table I), because the substitution of Ala for either Ser 19 alone or both Ser 19 and Thr 18 in the MLC mutants resulted in dramatic reduction in MLC 20 phosphorylation. These results are in complete agreement with the previously reported preferred sequential phosphorylation of Ser 19 followed by phosphorylation at the MLC second site (Thr 18 ), which is phosphorylated more slowly than Ser 19 and requires relatively high concentrations of myosin light chain kinase (2).

Kinetic Characteristics of Recombinant Endothelial
In Vitro Phosphorylation of MLCK Isoforms by p60 Src -Augmentation of tyrosine protein phosphorylation in endothelial cells in vivo correlates with increased phosphotyrosine content in MLCK immunoprecipitates, increased MLCK activity, and increased MLC phosphorylation (26,38). More recently, EC MLCK has been shown to be stably associated with p60 Src and a well recognized p60 Src substrate, the actin-binding protein cortactin (26). To further characterize the role of tyrosine phosphorylation in EC MLCK regulation, the three recombinant MLCK isoforms were used as substrates for in vitro phosphorylation catalyzed by p60 Src . EC MLCK-1 exhibited substantial time-dependent p60 src -catalyzed incorporation of radioactive phosphate (Fig. 2), whereas p60 src -mediated 32 P incorporation did not occur in either MLCK-2, the EC MLCK splice variant, lacking the 69-amino acid stretch containing the p60 src consensus site, nor in SM MLCK, which completely lacks the novel N terminus (Fig. 2). The 32 P incorporation into MLCK-1 was essentially abolished by the specific p60 src kinase inhibitor PP-2 (250 nM). However, all three MLCK isoforms exhibited low level of 32 P incorporation even in the absence of p60 src (0.63 Ϯ 0.21 mol of PO 4 /mol of protein) consistent with the MLCK autophosphorylation previously described for smooth muscle MLCK (39,40). This was confirmed by heat treatment of the MLCK preparations (70°C for 5 min), which completely abolished incorporation of 32 P into MLCK in the absence of p60 src (data not shown). The preferential p60 src -catalyzed phosphorylation of MLCK-1 compared with other MLCK isoforms strongly suggested that the specific site of tyrosine phosphorylation of endothelial MLCK by p60 src resides within the 69amino acid residue stretch in the N-terminal part of the MLCK molecule encoded by a single exon (28), which is not expressed in MLCK-2. However, the stoichiometry of phosphate incorporation into MLCK-1 catalyzed by p60 src (2.32 Ϯ 0.32 mol of PO 4 /mol of protein) strongly suggested the potential presence of a secondary tyrosine residue within MLCK-1, which may also be phosphorylated by p60 src .
Identification of p60 src Phosphorylation Sites within MLCK-1-To search for tyrosine phosphorylation site(s) within MLCK-1, we applied mass spectroscopy analysis of tryptic fragments obtained from MLCK-1 phosphorylated by p60 src . As an initial step, a limited trypsinolysis of radiolabeled phospho-MLCK-1 was performed, and tyrosine phosphorylation site(s) mapped to the tryptic fragment with an approximate molecular weight of 55 kDa based on autoradiography data and Western blot analysis with anti-phosphotyrosine antibody (inset, Fig.  3A). Mass spectrometric analysis of MLCK-1 peptides obtained after complete tryptic digestion of a 55-kDa MLCK-1 fragment  revealed a group of peptides corresponding to the N-terminal portion of MLCK-1 (Fig. 3A). Furthermore, we identified a single characteristic peak (Fig. 3A) not seen in the digests of MLCK-1 preincubated with ATP without p60 src (data not shown). This peak has m/z ratio 2486 corresponding to the theoretical mass of the tryptic peptide 457-476 of MLCK-1 ( 457 QEGSIEVYEDAGSHYLCLLK 476 ) with two incorporated phosphate groups (underlined). MALDI TOF analysis allowed isotopic resolution of the m/z 2486 peak (isotopic variants with m/z ratios 2484.9, 2485.9, and 2486.9, respectively) further supporting mapping of the peak to the diphospho-(457-476)-MLCK-1 fragment. To further prove phosphorylation of sites Y 464 and Y 471 by p60 src , we used synthetic peptides 460 SIEVYEDAGSHYLCLL 475 and 478 RTRDSGTYSCTASNA 492 corresponding to amino acid residues 460 -475 and 478 -492 of full-length MLCK-1 protein, respectively, for in vitro p60 src phosphorylation assay (Fig. 3B). Only the peptide containing Y 464 and Y 471 was readily phosphorylated in the presence of p60 src , whereas another candidate peptide containing the potential p60 src consensus phosphorylation site, Y 485 , as well as irrelevant peptide PEKVPPPKPATPDFRSVL (residues 968 -985 of MLCK-1), which lacks tyrosine residues, were not phosphorylated. The relatively low stoichiometry of p60 src -mediated phosphate incorporation into peptide 460 -475 (ϳ0.015 mol of PO 4 /mol of peptide, 10-min reaction) suggests that other MLCK-1 epitope(s) in proximity to amino acid residues 460 -475 may be required for the optimal p60 src activity. As recently demonstrated, the interaction of the p60 src SH3 domain with the ligand sequence of the substrate is important for activation of the catalytic domain and autophosphorylation (41), and the addition of an SH3 domain ligand to a substrate peptide increases its phosphorylation 10-fold via lowering of the K m value of the substrate and kinase activation (42). Thus, putative SH3-and SH2-binding domains present in MLCK-1 N-terminal portion (Fig. 1A) may be important for p60 src -catalyzed MLCK-1 phosphorylation, and further studies are underway to address this question. Finally, mass spectrometry analysis of the fragments obtained after cyanogen bromide cleavage of the MLCK-1 suggested that the two phosphate groups are contained within the Lys 1721 -Met 1761 EC MLCK fragment (data not shown) consistent with the presence of the EC MLCK-1 autophosphorylation sites Thr 1748 and Ser 1760 , which are homologous to the previously described SM MLCK autophosphorylation sites Thr 803 and Ser 815 (39). Differential Activation of MLCK-1 and MLCK-2 by p60 srcmediated Phosphorylation-To explore whether phosphorylation by p60 src alters endothelial MLCK-1 enzymatic properties, MLCK-1 samples were preincubated with either ATP and p60 src ("phospho-MLCK") or ATP alone ("dephospho-EC MLCK"), followed by assessment of in vitro kinase activity. Phosphorylation of EC MLCK-1 by p60 src increased EC MLCK-1 kinase activity 2-fold (Fig. 4), whereas the enzymatic activities of EC MLCK-2 and SM MLCK were not affected by p60 src and were comparable to that measured for EC MLCK-1 in the absence of p60 src . These results are again consistent with the inability of p60 src to phosphorylate EC MLCK-2 (as shown FIG. 3. Identification of the tyrosine phosphorylation sites within the N-terminal 55-kDa MLCK-1 tryptic fragment. The p60 src -catalyzed phosphorylation of the 55-kDa tryptic MLCK-1 fragment was detected by autoradiography and immunoreactivity with anti-phosphotyrosine antibody (upper inset). After excision from the gel, the phosphoprotein was subjected to complete trypsinolysis as described under "Materials and Methods," and the peptide digest was analyzed using mass spectrometry. A, high mass spectrum resolution of the peak with an average m/z ratio 2486, which mathematically corresponds to tryptic fragment 457-476 of MLCK-1 with two incorporated phosphate groups. Shown are peaks with m/z ratios 2484.9, 2485.9, and 2486. in Fig. 2) and indicate a significant enhancement of MLCK-1 kinase activity by p60 src -mediated phosphorylation. Enzymatic activity of all three isoforms was Ca 2ϩ /CaM-dependent, because chelation of free Ca 2ϩ with 2 mM EGTA (Fig. 4) or removal of calmodulin from the kinase reaction mixture (data not shown) completely abolished MLC phosphorylation catalyzed by either phospho-or dephospho-MLCK-1 preparations by MLCK-2 and SM MLCK. An inhibitor of smooth muscle MLCK activity, ML-7 (5 ϫ 10 Ϫ6 M) also abolished the enzymatic activity of EC MLCK-1 (phospho-and dephospho-), EC MLCK-2, and SM MLCK (Fig. 4). Finally, phosphorylation of MLCK-1 by p60 src did not alter Ca 2ϩ /CaM-dependent regulation, because the values for half-maximal activation of phospho-and dephospho-MLCK-1 determined over a range of free Ca 2ϩ concentrations (10 Ϫ8 to 10 Ϫ5 M) were comparable (pCa 6.56 versus pCa 6.50, respectively) despite an increase of ϳ2fold in enzymatic activity toward MLC 20 in the phospho-MLCK-1 preparation (Fig. 5). These data suggest that, although similar Ca 2ϩ concentrations are required for MLCK-1 activity, tyrosine phosphorylation promotes increased MLC phosphorylation at lower Ca 2ϩ concentrations within the cells. DISCUSSION In contrast to smooth muscle, only the high molecular weight MLCK isoform (208 -214 kDa) is expressed in endothelium (18 -20, 22). Molecular cloning of MLCK from human endothelial cells (20) revealed a high molecular weight MLCK variant containing a unique 922-residue N-terminal domain not expressed in the low molecular weight MLCK isoform, which is abundantly expressed in smooth muscle. Comparison of the cDNA encoding human high and low molecular weight MLCKs, when combined with results of chromosome mapping of human MLCK to single locus with chromosomal localization to 3qcen-q21 (6), suggests that mammalian MLCK genomic organization is highly similar to the "gene within a gene" organization of the avian smooth muscle/nonmuscle MLCK gene expressing two size class MLCK variants and one nonkinase protein (KRP), which are encoded by exons 1-31, 15-31, and 29A-31, respectively (17,43). The complexity of the human MLCK genomic organization was recently further emphasized by the detection of five splice variants of high molecular weight MLCK in nonmuscle and smooth muscle tissues using RT-PCR approaches (28). These data, which elucidated the considerable expression of EC MLCK-2, were strongly consistent with the potential functional diversity of the expressed smooth muscle and non-muscle MLCK proteins. Among endothelial cell MLCK splice variants, the MLCK-1 and MLCK-2 appear to be preferentially expressed (28), although all five have been identified in tissues. Using purified recombinant MLCK-1 and MLCK-2 expressed in a baculovirus system, we have now characterized for the first time the kinetic parameters of the high molecular weight MLCK isoforms from human endothelial cells. Comparisons of the V max and K m of these high molecular weight isoforms to recombinant rabbit uterine smooth muscle MLCK reveal very similar enzymatic properties of the three MLCK isoforms. However, a 2-fold increase in K 0.5 calmodulin observed in endothelial MLCK splice variants, may suggest a lower sensitivity of the intracellular EC MLCK for regulation by Ca 2ϩ /calmodulin as compared with SM MLCK.
A number of protein kinases, including cAMP-dependent protein kinase A, protein kinase C, Ca 2ϩ /CaM-dependent protein kinase II, and p21-activated kinase have been demonstrated to phosphorylate the smooth muscle MLCK isoform in vitro and in vivo (44 -47). Serine/threonine phosphorylation within MLCK calmodulin-binding domain results in a 10-fold increase in K CaM reflecting a 3.5-fold decrease in the association rate and a 6-fold increase in the dissociation rate between MLCK and Ca 2ϩ /CaM (3,45,46,48,49) and thus reduced MLCK enzymatic activity. In turn, phosphorylation by p21activated kinase decreases MLCK-1 catalytic activity by ϳ50% via decrease in maximum velocity (V max ) without affecting K CaM (47). In addition, Thr 803 , Ser 815 , and Ser 823 of the smooth muscle MLCK isoform undergo autophosphorylation in vitro also resulting in decreased MLCK affinity to Ca 2ϩ /calmodulin (39).
In contrast to serine/threonine phosphorylation of SM MLCK and EC MLCK, which attenuates MLCK activity (3,20,31,46,48), information is limited regarding phosphorylation sites within the MLCK molecule, which serve to enhance its enzymatic activity. Phosphorylation of smooth muscle MLCK by mitogen-activated protein kinase in vitro has been reported to stimulate smooth muscle MLCK activity (50), although we have not yet found mitogen-activated protein kinase to affect EC MLCK activity in this manner. However, recent studies have defined the involvement of tyrosine phosphorylation in EC MLCK regulation (26,27,51). Augmentation of protein tyrosine phosphorylation increased MLC 20 phosphorylation and cell contraction in endothelial cells, which strongly correlated with an increase in MLCK phosphotyrosine content, enhanced MLCK enzymatic activity, and the stable association of EC MLCK with activated p60 src (26). Our present results appear to be consistent with the hypothesized novel role of the unique N terminus in EC MLCK regulation via tyrosine phosphorylation. We now demonstrate for the first time the in vitro phosphorylation of the full-length EC MLCK-1 by p60 Src kinase on Tyr 464 and Tyr 471 , post-translational modifications not observed in the EC MLCK-2 splice variant lacking the 69-residue stretch (amino acids 436 -505) in the N terminus, which is encoded by a single exon deleted in the EC MLCK-2 isoform (28). Our future studies using site-directed mutagenesis approach are aimed at the determination of the sequence of phosphorylation events and the role of each tyrosine phosphorylation site in the regulation of MLCK-1.
In summary, we have characterized the kinetic properties of endothelial MLCK splice variants and demonstrated a novel mechanism of MLCK-1 regulation by p60 src phosphorylation. The phosphorylation sites (Tyr 464 and Tyr 471 ) are located within unique N-terminal domain (436 -505) of endothelial MLCK-1 isoform not expressed in smooth muscle MLCK or in the alternatively spliced endothelial isoform MLCK-2. Consistent with this finding, only MLCK-1 activity is regulated by p60 src -catalyzed phosphorylation. These data demonstrate the importance of the novel N-terminal domain in the specific regulation of the MLCK isoform present in nonmuscle cells. As we have previously demonstrated, the tyrosine phosphorylation of EC MLCK increases its association with both p60 src kinase, as well as with the actin-binding protein and the p60 src substrate cortactin (26), we speculate that MLCK-1 tyrosine phosphorylation may be involved in contractile complex scaffolding and contribute to Ca 2ϩ sensitization of the endothelial contractile apparatus. Based on our data, we speculate that p60 src -catalyzed tyrosine phosphorylation contributes to the local and selective activation of endothelial cell MLCK-1 under submaximal Ca 2ϩ concentrations providing a mechanism that may tightly orchestrate critical cytoskeletal rearrangements and ultimately the cellular contraction, which is critical for endothelial cell-dependent biological processes, such as vascular barrier regulation, transendothelial leukocyte diapedesis, and angiogenesis.