Novel Ca 2 (cid:49) /Calmodulin-dependent Protein Kinase II (cid:103) -Subunit Variants Expressed in Vascular Smooth Muscle, Brain, and Cardiomyocytes*

Ca 2 /calmodulin-dependent protein kinase II (CaM kinase II) (cid:103) -subunits were cloned from a porcine aortic smooth muscle cDNA library resulting in identification of alternatively spliced CaM kinase II (cid:103) B - and (cid:103) C -sub-units and a novel (cid:103) -subunit variant predicted to encode a 60.2-kDa polypeptide, which was designated the (cid:103) G - subunit. A clone predicted to encode a 62.2-kDa (cid:103) -sub-unit, designated as (cid:103) E , was isolated with a variable do- main structure similar to a (cid:103) B -subunit but with a 114-nucleotide insertion in the conserved “association” domain of CaM kinase II subunits. A full-length (cid:103) E -sub-unit construct expressed in COS cells resulted in mul- timeric CaM kinase II holoenzymes (470 kDa) with activation and autoregulatory properties similar to expressed holoenzymes composed of (cid:103) B -, (cid:103) C -, or (cid:103) G -sub-units. Expression of (cid:103) E and related (cid:103) -subunit mRNAs containing the 114-base insertion was documented in porcine tissues by reverse transcriptase-polymerase chain reaction. CaM kinase II subunits containing the 38-amino acid insert were identified by Western analysis of partially purified CaM kinase II from carotid arterial smooth muscle and brain using a sequence-specific anti-peptide antibody. Immunoprecipitations of tissue homo- genates indicated a comparatively high level of medial smooth muscle, and neonatal rat cardiomyocytes using RNAzol (Biotecx). Integrity of the RNA was confirmed by formaldehyde-agarose gel electrophoresis, and concentration was estimated by absorbance at 280/260 nm. RT-PCR amplification reactions were carried out using a commercial kit (Perkin-Elmer Corp.) and Taq DNA polymerase (Fish- er). RT reactions were carried out with oligo(dT) 16 primers and 2 (cid:109) g of total RNA. PCR reactions utilized 500 pmol/ml of each primer. In the “nested” reactions g plasmid or (cid:109) l of the reaction mix from the primary RT-PCR reactions was used as template. PCR amplifications were carried out on a PTC-100 Ther-mocycler (MJ Research, Inc.) using a cycling protocol (35 cycles) recom- mended by the PCR kit manufacturer and the following primers: SPTVPI; near the C terminus of CaM kinase II (cid:100) -subunit) coupled to carrier protein. Miscellaneous Methods— CaM kinase II was partially purified from porcine brain and carotid artery medial smooth muscle by DEAE anion exchange chromatography followed by phosphocellulose cation ex- change chromatography as described previously (26) to specific activities of 60 and 90 nmol/min/ml, respectively. Synthetic oligonucleotides were synthesized on an Applied Biosystems model 394 DNA synthesizer and synthetic peptides on an Applied Biosystems 431A peptide synthesizer in the Core Molecular Biology Lab at the Weis Center for Research. Autoradiograms and ECL exposures were scanned and digi- tized with a Molecular Dynamics personal densitometer and Image-Quant 4.0 software.

Multifunctional Ca 2ϩ /calmodulin-dependent protein kinase II (CaM kinase II) 1 mediates cellular responses induced by increases in second messenger Ca 2ϩ and has been implicated in the control of such essential functions as synaptic transmission (1,2), gene transcription (2,3), and cell growth (4). CaM kinase II specific activity in smooth muscle (5,6) is about 1 ⁄10 of that in brain (7). A significant fraction of the kinase in smooth muscle associates with a myofibrillar fraction (8) and co-purifies with caldesmon, a putative thin filament regulatory protein (9). CaM kinase II is activated in vascular smooth muscle cells over a physiological range of free intracellular [Ca 2ϩ ] (10) and has been shown to be involved in cell migration (11) and modulation of smooth muscle myosin light chain kinase sensitivity to activator [Ca 2ϩ ] (6). CaM kinase II has also been implicated in the control of Ca 2ϩ channels (12) and sarcoplasmic reticulum Ca 2ϩ /ATPase activity in smooth (13) and cardiac muscle (14,15) and as an intermediate in the activation of the mitogenactivated protein kinase signaling cascade induced by Ca 2ϩmobilizing stimuli (16). 2 Endogenous CaM kinase II (7) and recombinant isozymes expressed in Escherichia coli (17) or mammalian cells (18) are large multimeric proteins composed of 8 -10 individual protein kinase subunits, which can be products of four separate genes. The CaM kinase II subunit genes are expressed in a tissuespecific manner in that two of the isoforms (␣ and ␤) appear to be largely restricted to the brain, while the other two isoforms (␦ and ␥) are variably expressed in brain and peripheral tissues (5,19,20). Additional subunit diversity arises through alternative splicing. While the full spectrum and functional significance of subunit heterogeneity has not been established, it has been proposed that it provides a mechanism for producing CaM kinase II isozymes with differing kinetics, substrate specificities, or subcellular localization (21,22).
With the general goal of understanding the relationship between CaM kinase II structure and cellular localization/function in smooth muscle, studies were carried out to identify ␥-subunit isoforms expressed in mammalian vascular smooth muscle. A novel CaM kinase II ␥-subunit variant (␥ G -subunit) was discovered, and a unique class of ␥-subunit variants with a 38-amino acid insertion in the association domain was documented for the first time in primary neuronal and cardiovascular tissues. When expressed in COS cells, these unique subunits formed multimeric holoenzymes (7-8 subunits) with autoregulatory properties characteristic of known CaM kinase II isozymes. Although the functional significance of the CaM kinase II ␥-subunits with the 38-amino acid insertion is not known, these subunits were variably expressed as proteins in arterial smooth muscle, heart, and brain, where they appeared in heteromultimeric complexes with other more abundant CaM kinase II subunits.

EXPERIMENTAL PROCEDURES
Cloning and Sequencing of CaM Kinase II ␥-Subunit cDNA-A porcine cultured thoracic aorta cell cDNA library (Uni-ZAP XR vector; Stratagene) was screened with a 990-bp DNA probe corresponding to the rat CaM kinase II ␥-subunit catalytic domain. The probe was * This work was supported by National Institutes of Health Grant HL49426 (to H. A. S.) and a fellowship from the American Heart Association, Pennsylvania Affiliate (to S. T. A.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
‡ To whom correspondence should be addressed: Weis Center for Research, Geisinger Clinic, 100 N. Academy Ave., Danville, PA 17822-2612. Tel.: 717-271-8669; Fax: 717-271-6701; E-mail: has @smtp.geisinger.edu. 1 The abbreviations used are: CaM kinase II, Ca 2ϩ /calmodulin-dependent protein kinase II; ECL, enhanced chemiluminescence; HRP, horseradish peroxidase; PCR, polymerase chain reaction; RT, reverse transcriptase; PAGE, polyacrylamide gel electrophoresis; UTR, untranslated region; MOPS, 4-morpholinepropanesulfonic acid; FPLC, fast protein liquid chromatography; CMV, cytomegalovirus. generated by RT-PCR using rat brain RNA and ␥-subunit-specific primers and phosphorylated by nick translation with [ 32 P]dCTP using a kit from Promega. Hybridization was at 50°C followed by four washes in 1-0.1 ϫ SSC, 0.1% SDS at 50°C. Positive clones were categorized by restriction endonuclease mapping and size analysis of PCR-amplified targets using primers that spanned the variable region of the kinase, including regions V 1-3 . By comparison with predicted products, the clones were classified as probable ␥ B -, ␥ C -, or unknown subunits, and representative clones were chosen for sequencing. pBluescript phagemids containing the putative ␥-subunit cDNAs were excised from the Lambda ZAP II vector according to procedures provided by the manufacturer (ExAssist/SOLR System, Stratagene). Both strands of clone 28 were sequenced by the dideoxy chain termination method using [␣-35 S]dATP and Sequenase (U.S. Biochemical Corp.) with overlapping synthetic oligonucleotide primers as depicted in Fig. 1. Other clones were sequenced fully in at least one direction and in both directions in regions corresponding to the variable domains.
Expression Vectors and Construction of Full-length ␥ E -Subunit-pBluescript plasmids containing ␥-subunit inserts (clones 28, 29, and 35) were linearized by digestion with XbaI. ␥-Subunit cDNAs were excised by incomplete digestion with ApaI, which was necessary due to the occurrence of an ApaI site in the conserved 3Ј coding region of the ␥-subunit cDNA. Full-length fragments were gel purified and ligated into pRc/CMV. Clone 6 was incomplete at the 5Ј end but terminated in the conserved catalytic domain at a position corresponding to base ϩ666 in the other ␥-subunits. A full-length sequence was constructed by ligating in a single reaction the following three gel-purified DNA fragments: the 5Ј region of full-length clone 35 (including the 5Ј-untranslated region (UTR)) excised using the XbaI site in the pBluescript multicloning sequence and the NsiI site in the ␥-subunit sequence; the 3Ј fragment of clone 6 (including 3Ј-UTR) produced by digestion with NsiI and ApaI; and linearized pRc/CMV vector digested with XbaI and ApaI. The resulting construct was confirmed by partial sequencing and was shown to be translated into a functional CaM kinase II subunit in transfected COS-7 cells.
CaM Kinase II Assay and Autophosphorylation-CaM kinase II activity was assayed as described previously with autocamtide-2 (KKAL-RRQETVDAL) as a peptide substrate (5). To measure generation of Ca 2ϩ /CaM-independent or "autonomous" activity, COS cell lysate CaM kinase II was autophosphorylated by preincubating for 30 s or 5 min in 50 mM MOPS (pH 7.4), 10 mM magnesium acetate, 3 mM EGTA, 4 mM calcium chloride, 400 nM calmodulin, 0.2 mM ATP, 15 mM 2-mercaptoethanol at 30°C. Controls lacked added Ca 2ϩ /CaM and ATP. Reactions were stopped by removing a 25-l aliquot to tubes (4°C) containing 5 l of 90 mM EDTA. 5-l aliquots were subsequently assayed for "total" and "independent" CaM kinase II activity as described previously (5) but with 6 mM CaCl 2 added to the total tubes to compensate for EDTA in the autophosphorylated lysates. Autonomous activity was defined as [(independent/total activity) ϫ 100].
Size Fractionation of Expressed CaM Kinase II Holoenzymes-Gel filtration chromatography was carried out on a Superose 12 FPLC column (Pharmacia Biotech Inc.) equilibrated and eluted with buffer A (4°C) at a flow rate of 0.25 ml/min, collecting 0.25-ml fractions. The elution of thyroglobulin (669 kDa), catalase (250 kDa), and ovalbumin (67 kDa) was used to calibrate the column. Sucrose gradient centrifugation was carried out as described previously (18,24). Lysates from ␥-subunit-overexpressing COS cells (300 -1000 g of total protein) were layered on 5-20% sucrose gradients (4.5 ml) containing 50 mM MOPS (pH 7.0), 150 mM NaCl, 0.5 mM EDTA, 5% glycerol, 0.5% Nonidet P-40 and centrifuged at 150,000 ϫ g for 16 h at 5°C. 150-l fractions were removed from the top, and aliquots removed for SDS-PAGE (30 l) and CaM kinase II assays (7.5 l). Molecular sizes were estimated relative to the sedimentation of thyroglobulin, catalase, and ovalbumin standards according to the following formula (24): R ϭ (M r (unknown) /M r (std) ) 2/3 , where R is the quotient of the distance traveled by the unknown in the sucrose gradient divided by the distance traveled by the standard. The sizes relative to each standard were averaged to obtain estimated M r values.
Immunoprecipitation and Western Blotting-Anti-peptide antibodies were produced in New Zealand White rabbits as described previously (5,25). A peptide corresponding to amino acids 39 -69 in the catalytic domain of the CaM kinase II ␣-subunit sequence, which is common in all CaM kinase II subunits, was solubilized in phosphate-buffered saline (10 mg/ml), emulsified with Hunter's Titermax adjuvant (CytRx Corp.), and used to produce the antibody designated CK2-CAT. A synthetic peptide (PEGRSSRDRTAPSAGMQPQPSLC) from the V 3 sequence in the ␥ E -subunit was cross-linked to purified protein derivative of tuberculin (Statens Seruminstitut, Copenhagen) with gluteraldehyde (25), emulsified with RIBI adjuvant (RIBI Immunochemical Research Inc.), and used to produce the antibody designated CK2-V3. Antibodies were purified by peptide affinity column chromatography using cyanogen bromide-activated Sepharose 4B (Pharmacia). Fractions were eluted with 0.2 M glycine, dialyzed, and stored in phosphate-buffered saline containing 0.02% sodium azide.
Western blots utilized standard procedures (5). Immunoreactive bands were visualized using horseradish peroxidase (HRP)-coupled 2°a ntibody and the enhanced chemiluminescent (ECL) detection method (Amersham Corp.). In the case of the Western blots carried out on proteins immunoprecipitated from tissue homogenates with CK2-V3, 1°a ntibodies were directly cross-linked to HRP to catalyze the ECL reaction (EZ-link Plus activated peroxidase and Freezyme conjugate purification kit; Pierce). Fresh brain or frozen pulverized heart and carotid artery were homogenized in 5 volumes of buffer A and centrifuged at 100,000 ϫ g for 40 min, and the supernatant was assayed for CaM kinase II and total protein. A volume of brain extract containing 115 nmol/min total CaM kinase II activity and volumes of carotid and heart extracts containing 30 nmol/min of total activity were immunoprecipitated with CK2-V3 (2 g of affinity-purified antibody/mg of extract protein) for 16 h at 4°C. Immune complexes were collected by the addition of protein A-coupled agarose beads (Pierce) and centrifugation at 10,000 ϫ g for 15 s. Immunoprecipitated proteins were washed three times with buffer A and then solubilized in sample buffer and resolved by SDS-PAGE. Control immunoprecipitations were carried out in the presence of 25 M immunizing peptide, 25 M immunizing peptide coupled to carrier protein (purified protein derivative), and 25 M unrelated peptide (IHFHRSG-SPTVPI; near the C terminus of CaM kinase II ␦-subunit) coupled to carrier protein.
Miscellaneous Methods-CaM kinase II was partially purified from porcine brain and carotid artery medial smooth muscle by DEAE anion exchange chromatography followed by phosphocellulose cation exchange chromatography as described previously (26) to specific activities of 60 and 90 nmol/min/ml, respectively. Synthetic oligonucleotides were synthesized on an Applied Biosystems model 394 DNA synthesizer and synthetic peptides on an Applied Biosystems 431A peptide synthesizer in the Core Molecular Biology Lab at the Weis Center for Research. Autoradiograms and ECL exposures were scanned and digitized with a Molecular Dynamics personal densitometer and Image-Quant 4.0 software.

RESULTS
Cloning and Identification of CaM Kinase II ␥-Subunit Variants-Four candidate CaM kinase II ␥-subunit clones were isolated from a cultured porcine thoracic aortic cell cDNA li-brary and sequenced as shown in Fig. 1. 3 A comparison of clone 28 sequence with published rat and human ␥-subunit sequences indicated that it was a novel CaM kinase II ␥-subunit variant. It was designated the ␥ G -subunit, consistent with published nomenclature for alternatively spliced ␥-subunit variants (18,27,28). This ␥-subunit contained two sequences encoding 21 (V 1 region) and 23 (V 2 region) amino acids, each of which has been shown to be individually expressed in some CaM kinase II ␥-subunit isoforms (18,27) (Fig. 2). Clone 35 was found to lack the V 1 insertion identifying it as a ␥ B -subunit cDNA, while clone 29 lacked inserts in V 1 and V 2 , identifying it as a ␥ C -subunit. Nucleic acid sequences of the porcine cDNAs were 95% identical to the human sequence in the coding region and 3Ј-UTR but only 29% identical in an overlapping region of 21 nucleotides in the 5Ј-UTR. Predicted amino acid sequences of the porcine ␥ B and ␥ C clones were 100% identical to published human ␥ B -and ␥ C -subunit sequences.
Clone 6 was incomplete, lacking the 5Ј-UTR and first 665 bases of the ␥-subunit coding sequence. Its sequence was otherwise identical to a ␥ B -subunit but with an additional 114base insert at a position corresponding to bases 1216 and 1217 of the ␥ G sequence (Fig. 3). The insertion resulted in the substitution of an alanine for a valine followed by 38 amino acids in the translated sequence in a region considered to be part of the conserved association domain (Figs. 2 and 3). Three ␥-subunit variants containing an identical 114-base insert (referred to here as V 3 ) and either no V 2 insert, the 69-base V 2 insert, or a truncated 27-base V 2 insert were recently identified by RT-PCR in cultured human biliary tumor cells and designated as putative ␥ F -, ␥ E -, and ␥ D -subunits (28). Based on this, we designated clone 6 as a partial ␥ E -subunit cDNA.
To investigate the possibility that the V 3 sequence was CaM kinase II subunit-and/or species-specific, RT-PCR was used to amplify products spanning this domain from neonatal rat cardiomyocyte RNA using primers specific for the rat ␥-subunit sequence. Two major products were amplified consistent in size 3 Complete cDNA sequences and deduced amino acid sequences for the four porcine clones shown in Fig. 1 have been entered in Gen-Bank TM and assigned the following accession numbers: ␥ B (clone 35), U72970; ␥ C (clone 28), U72071; ␥ E (clone 6), U72972; ␥ G (clone 29), U72973. The sequence of the rat ␥-subunit RT-PCR product spanning the V 3 domain shown in Fig. 3 has been assigned accession number U73503. with predicted ␥-subunit targets containing or lacking the 114base V 3 insert. Sequencing the RT-PCR products from rat mRNA confirmed a homologous 114-base V 3 insert, nearly identical to the porcine and human sequences except for two nucleotides, which resulted in two amino acid substitutions (Fig. 3). Similar experiments using ␦-subunit-specific PCR primers and swine brain, carotid artery, and cardiomyocyte mRNA amplified only a single product of the size predicted for a ␦-subunit lacking a V 3 insertion (not shown).
Properties of ␥-Subunit Variants Expressed in COS-7 Cells-To assess the biochemical properties of the novel CaM kinase II ␥ G -and ␥ E -subunits, the cDNAs were subcloned into a mammalian expression vector (pRc/CMV) and expressed in transiently transfected COS-7 cells. Because the putative ␥ Esubunit clone was incomplete (clone 6), and additional fulllength clones were not isolated from the library, a full-length sequence was constructed (see "Experimental Procedures"). Transfection of the cDNAs in COS cells resulted in 6 -27-fold increases in Ca 2ϩ /CaM-stimulated autocamtide-2 kinase activity in cell lysates (Fig. 4A). Western analysis identified CaM kinase II subunits in each of the overexpressing cell lines consistent in size with those predicted based on amino acid or presence (open symbols) of activators to regulate autophosphorylation and conversion to an activator-independent or "autonomous" form. The autonomous activities in these samples were subsequently assayed using exogenous substrate in the absence of free Ca 2ϩ . Total activities in the autophosphorylated samples were not different from their respective nonautophosphoryated controls (n ϭ 2-3).

FIG. 3. CaM kinase II ␥-subunit V 3 domain.
Nucleotide and deduced amino acid sequences of the V 3 domain and some flanking conserved sequence are from porcine ␥ G -and ␥ E -subunit clones and a rat cardiomyocyte RT-PCR product amplified using rat ␥-subunit specific primers, which flanked this region. The porcine sequence differs from human ␥-subunit sequences (18) by a single nucleotide (t in porcine, c in human) shown in boldface type and marked with an asterisk. Differences in nucleotide and deduced amino acid sequence between porcine and rat sequence in this region are in boldface type. sequences (Fig. 4B). A distinguishing feature of CaM kinase II is that upon activation it undergoes a transition to a Ca 2ϩ / calmodulin-independent or "autonomous" kinase activity, a reaction that is facilitated by the multimeric structure of the holoenzyme (29). Preincubation in the absence of exogenous substrate resulted in the rapid and extensive conversion of each expressed ␥-subunit CaM kinase II isozyme to its autonomous form (Fig. 4C).
The position of the V 3 insertion within the conserved association domain raised the possibility that multimeric complexes composed of ␥ E -subunits might be disrupted or modified in size. However, gel filtration analysis of both CaM kinase II ␥ E -and ␥ B -subunits expressed in COS cells (Fig. 5A) and Western analysis of the column fractions with CK2-CAT (Fig. 5B) indicated that most of the kinase nearly co-eluted with a thyroglobulin (669 kDa) standard. Sucrose gradient centrifugation confirmed expression of ␥ B,E,G -subunits as multimers in COS cells with estimated sizes as follows: ␥ B , 448 kDa; ␥ E , 471 kDa; ␥ G , 425 kDa. These data suggest that the ␥-subunits form multimers of 7 or 8 subunits, a finding consistent with previous results describing the human ␥ B -subunit (18). A variable fraction of Ca 2ϩ /CaM-dependent kinase activity, consistent in size with subunit monomers, was also resolved by both gel filtration (Fig. 5A) and sucrose density centrifugation. However, only weakly immunoreactive 35-40-kDa bands could be detected in these fractions (Fig. 5B), suggesting that the source of kinase activity was proteolytic fragments of the expressed subunits.
Structural Characterization of Expressed CaM Kinase II ␥-Subunits Containing the V 3 Insert-To better define the ␥-subunit structures that contain the V 3 insertion in vivo, RT-PCR was carried out using mRNA from porcine brain, ca-rotid artery, and cardiomyocytes such that only ␥-subunitderived targets that contained the 114-base V 3 insert and spanned the V 1 domain were amplified (Fig. 6). A second PCR amplification was carried out using templates amplified from the primary RT-PCR reactions and nested primers spanning V 1 and V 2 . The predominant products obtained from this reaction were identical in size to the product amplified from the control ␥ B -subunit clone. This experiment confirmed that ␥-subunits containing the V 3 insert are in fact expressed at the mRNA level in primary porcine tissues and indicated that the insert occurred mainly in combination with the 69-base V 2 insert (consistent with the sequence of clone 6 or the predicted ␥ Esubunit) or possibly a 63-base V 1 insert, which would represent yet another ␥-subunit variant. Other minor products were amplified with the nested primers, some of which were similar in size to the products amplified from control ␥ C and ␥ G subunit clones, suggesting low level in vivo expression of alternative ␥-subunit isoforms containing the V 3 insert.
Relative Expression of ␥-Subunit Variants in Vivo-To provide an estimate of the relative expression of ␥-subunits containing the V 3 insert compared with other ␥-subunits, RT-PCR experiments were carried out using primers that spanned all three variable domains. Products amplified from porcine brain, arterial smooth muscle, and cardiomyocyte mRNA and control plasmid DNA (Fig. 7A) were Southern blotted with 32 P-labeled oligonucleotide probes that hybridized specifically to the 63base V 1 , 69-base V 2 , or 114-base V 3 sequence or to a sequence common to all ␥-subunits (Fig. 7B). Hybridization with the common probe produced an autoradiogram with a pattern essentially identical to the ethidium bromide-stained gel (Fig.  7A). In smooth muscle, the most abundant PCR product hybridized with the ␥ B -specific V 2 probe and was consistent in size with that amplified from the control ␥ B -subunit cDNA. Less abundant were products that were consistent in size with that expected from the ␥ C -subunit and hybridized only with the common probe (not shown). Minor amounts of a 781-base product that hybridized with both the V 1 -and V 2 -specific probes were amplified from smooth muscle mRNA, consistent with low level expression of the ␥ G -subunit. Hybridization of the arterial smooth muscle PCR products with the V 3 probe was weak, suggesting that ␥-subunit transcripts containing the 114-base V 3 insert must be a minor fraction of total ␥-subunit mRNA in this tissue.
The pattern of PCR products amplified from cardiomyocyte RNA was similar to that in smooth muscle, but in this case products consistent in size with the ␥ E -subunits and hybridizing with the V 3 -specific probe were more abundant. In contrast, the largest and most abundant ␥-subunit products amplified from porcine brain RNA included the 114-base V 3 insertion. Probes specific for both the V 2 and V 1 inserts also hybridized with these targets, implying the expression of both the ␥ Esubunit and another isoform predicted to have the 63-base V 1 insert in combination with the 114-base V 3 insert. Only minor amounts of a 718-base product, which would reflect expression of the ␥ B -subunit, or a 649-base product, consistent with the ␥ C -subunit, were amplified from brain RNA. A minor product slightly smaller than the ␥ G -subunit and hybridizing with the V 1 -specific primer was amplified from brain RNA, probably reflecting expression of the ␥ A -subunit, which was the original ␥-subunit isoform cloned from brain (27).
Identification of ␥-Subunits with the V 3 Insert in Brain and Arterial Smooth Muscle-To confirm expression of CaM kinase II ␥-subunits containing the 114-base V 3 insert at the protein level in primary tissues and to gain insight into its assembly into multimeric CaM kinase II isozymes in vivo, an anti-peptide antibody was produced (CK2-V3) against a portion of the predicted 38-amino acid insert. The specificity of CK2-V3 was established by Western analysis of lysates from COS cells overexpressing recombinant ␥-subunits. In a mixture of such lysates a 62-kDa band was identified, consistent with the predicted size of the ␥ E -subunit (Fig. 8A). A 52-kDa band in the lysates also cross-reacted with CK2-V3 but not with CK2-CAT (Fig. 8B), indicating a protein unrelated to CaM kinase II. A consistent observation was a relatively weaker immunoreactivity of the expressed ␥ C -and ␥ E -subunits with CK2-CAT compared with ␥ B -and ␥ G -subunits, which is apparent in mixtures of COS cell lysates containing approximately equal activities of each of the expressed subunits (Fig. 8B, lane 1).
Analysis of partially purified fractions of CaM kinase II from carotid artery smooth muscle with CK2-CAT indicated CaM kinase II subunits with apparent sizes in the range of 54 -58 kDa (Fig. 8B). These subunits were not cross-reactive with CaM kinase II ␣or ␤-subunit-specific antibodies (not shown) and therefore probably represent mixtures of ␥and ␦-subunits. Similar fractions from porcine brain contained primarily 50and 58 -60-kDa subunits, which can be attributed to expression of CaM kinase II ␣and ␤-subunits as previously established for this tissue (7). Bands consistent in size with the ␥ E -subunit (62 kDa) were barely detectable in either brain or carotid fractions immunoblotted with CK2-CAT. When the same membrane was probed with CK2-V3, a 62-kDa band that co-migrated with recombinant ␥ E -subunit standards was detected in the partially purified CaM kinase II fractions from both tissues, indicating expression of the ␥ E -subunit or a similarly sized CaM kinase II subunit containing the V 3 insert (Fig. 8A). Smooth muscle CaM kinase II preparations also contained CK2-V3 cross-reactive bands, which were estimated to be 56 and 64 kDa, consistent in size with ␥ C -and ␥ G -like subunits containing the 38-amino acid V 3 insert. The preparation from brain also contained minor amounts of unidentified faster and slower migrating subunits cross-reactive with the CK2-V3 antibody. To determine the composition of native CaM kinase II isozymes containing ␥ G -and related subunits, aliquots of porcine brain, carotid artery medial smooth muscle, and cardiac tissue homogenates were immunoprecipitated with CK2-V3. The immunoprecipitated proteins were resolved by SDS-PAGE and immunoblotted with CK2-V3 (Fig. 9A) and CK2-CAT (Fig. 9B), which had been directly conjugated to FIG. 7. Analysis of ␥-subunits expressed in brain, carotid artery vascular smooth muscle, and heart. Total RNA from the tissues was analyzed by RT-PCR and Southern blotting. The PCR primers spanned all three variable regions, providing an analysis of the entire pool of ␥-subunit variant mRNAs. A, ethidium bromide-stained agarose gel of size-fractionated RT-PCR products compared with controls amplified from a mixture of plasmid DNAs containing ␥-subunit sequences. B, Southern analysis of the same gel using end-labeled 32 Poligonucleotides (Var1, Var2, and Var3) complementary to sequences specific for each of the variable domain sequences (V 1 , V 2 , and V 3 , respectively). The same blot was sequentially hybridized and stripped of each probe. HRP. This avoided the need for secondary HRP-coupled antirabbit IgG antibodies in the detection protocol that would have also detected the IgG subunits in the immunoprecipitates and obscured signals in the 50-kDa range. CK2-V3 immunoprecipitates from all three primary tissues contained a primary 62-kDa band, which was immunoblotted with the same antibody and co-migrated with the ␥ E -subunit immunoprecipitated from overexpressing COS cells (Fig. 9A). The addition of a 180-fold molar excess of the immunizing peptide (relative to estimated antibody binding sites), either free or coupled to the carrier protein, prevented immunoprecipitation of ␥ E - (Fig. 9A) and co-immunoprecipitating subunits (Fig. 9B). An unrelated peptide had no effect, establishing the specificity of the immunoprecipitations. Co-immunoprecipitation of other CaM kinase II subunits from all three primary tissues, as detected with CK2-CAT (Fig. 9B), indicated the formation of CaM kinase II heteromultimers in vivo. Compared with carotid or heart extracts, brain extracts contained higher concentrations of ␥-subunits immunoprecipitable with CK2-V3, consistent with the relatively high level of mRNA expression for these subunits (Fig. 7B).

DISCUSSION
Compared with CaM kinase II isozymes from brain, which are composed of ␣and/or ␤-subunits, relatively little work has been carried out on the properties and function of CaM kinase II isozymes expressed in peripheral tissues, which are composed primarily of ␦and/or ␥-subunits. Porcine models have been widely used in both in vivo cardiovascular studies (30) and in vitro, where biophysical studies on isolated arterial tissues and biochemical studies of purified proteins have contributed significantly to our current knowledge concerning the regulation of smooth muscle contractility (31). Our preliminary biochemical characterization of porcine carotid artery CaM kinase II (8) provided a rationale for the present studies, which were initiated to gain a better understanding of smooth muscle CaM kinase II structure.
Attention was focused on CaM kinase II ␥-subunits, which appear to be the principal isoform expressed in this differentiated tissue (32). 4 Analysis of ␥-subunits cloned from the cultured porcine aortic vascular smooth muscle library resulted in the identification of two novel variants designated ␥ G and ␥ E , bringing the total number of reported ␥-subunits to seven (18,28). The 21-amino acid V 1 sequence in the ␥ G -subunit is common only to the original ␥-subunit (␥ A ) cloned from brain and is homologous to a similar insert in the CaM kinase II ␤-subunit (33). The ␥ G -subunit was shown in the present study to be expressed in carotid and heart, but it appears to be a relatively small fraction of the total ␥-subunit pool. The 114-base ␥-subunit V 3 insert, which we identified in clone 6, was also identified by RT-PCR in human biliary tumor cells and reported while the current study was in progress (28). Although that analysis did not include the V 1 region of the ␥-subunit shown in Fig. 2, a partial sequence containing both the 69-base V 2 insert and 114-base V 3 insert was designated as a ␥ E -subunit. To avoid confusion, we chose to retain this nomenclature and designated clone 6 as a partial ␥ E -subunit. As shown here, this 114-base V 3 insert is well conserved across mammalian species with only a few differences in nucleotide and primary sequence between the rat and porcine or human homologues.
This study provides the first direct evidence that ␥ E -subunit and related subunit mRNAs that contain the 38-amino acid V 3 insert are expressed and translated in vivo in primary tissues. CaM kinase II subunits containing this sequence were found to be variably expressed and were most abundant in brain, where they were the principal ␥-subunit variant. While clearly detectable as protein in partially purified fractions of CaM kinase II from carotid artery smooth muscle, based on the RT-PCR analysis of ␥-subunit RNA and immunoprecipitation experiments with the V 3 insert specific antibody, ␥ E -like subunits appear to represent a relatively small fraction of the total ␥-subunit pool in this tissue and also in heart. It should be noted that the ␥ E -subunit was cloned from a cDNA library from cultured aortic smooth muscle cells, which are known to revert to a dedifferentiated phenotype similar to smooth muscle cells found in the developing animal (34). In the present study, neonatal animals were the source of the porcine brain and hearts, while carotid arteries were from mature swine. It is possible that ␥ E -and related subunits may be developmentally regulated and expressed at higher levels in tissues from immature animals.
It is likely that other ␥-subunit variants containing both the 23-amino acid V 1 and 38-amino acid V 3 inserts exist. Indirect evidence for this included the hybridization of both V 1 -and V 3 -specific probes to ␥-subunit RT-PCR products amplified from brain RNA (Fig. 7) and multiple bands in the Western blots using the V 3 sequence-specific antibody (Fig. 8). A novel CaM kinase II ␤-subunit cDNA, designated ␤ 3 , cloned from neonatal rat islet cells (35) contains a longer nonhomologous sequence (258 bases encoding 86 amino acids) inserted at the same position in the association domain as the ␥-subunit V 3 insert. In the ␤-subunit gene, this is at the boundary between exons IV and V (33). The Drosophila CaM kinase II gene is also alternatively spliced at this site, producing four CaM kinase II subunit variants (36) indicating evolutionary conservation of variants with modifications in this region. However, this class of splice variants may be subunit-specific, since similar ␦-subunit variants were not identified in the cDNA library or by RT-PCR using brain, smooth muscle, or heart mRNA-and ␦-subunit-specific primers. 4 H. A. Singer, unpublished observations. FIG. 9. Immunoprecipitation of CaM kinase II holoenzymes with CK2-V3. Brain, carotid artery, heart extracts, and control COS cell lysates from cells expressing the ␥ B -and ␥ E -subunits were immunoprecipitated using the antibody that specifically recognizes CaM kinase II subunits containing the unique 38-amino acid V 3 insert (IP: CK2-V3). The precipitated proteins were then resolved by SDS-PAGE and immunoblotted using the same antibody (CK2-V3*) that had been directly conjugated to HRP to provide a catalyst for chemiluminescent detection (panel A). The blots were then stripped and reprobed with HRP-conjugated CK2-CAT (CK2-CAT*) (panel B), a subunit-nonselective antibody, to detect other CaM kinase II subunits that co-precipitated in multimeric complexes with ␥ E -like subunits. The lanes containing carotid and heart samples are shown at an exposure 3 times longer than the control and brain samples to partially compensate for differences in the relative signal strengths. Brain immunoprecipitation (IP) and control lanes are as follows. BLAST searches of the peptide sequence data bases failed to detect peptide sequences strongly homologous to the 38-amino acid ␥-subunit that might provide clues as to the possible function of the insert. Expression of this sequence in the ␥-subunits does not grossly alter holoenzyme structure or autoregulatory properties, as evidenced by the close similarity between ␥ B -and ␥ E -subunits expressed in COS cells with respect to holoenzyme size and kinetics of autophosphorylation-dependent generation of autonomous kinase activity (Fig. 4). At this point we can only speculate that the V 3 sequence may be important in targeting CaM kinase II isozymes to specific subcellular compartments as has been shown for a specific V 2 sequence found in other CaM kinase II subunits, which targets isozymes to the nucleus (21,22).
In general, not much is known specifically about the subunit composition of CaM kinase II holoenzymes. Therefore, it is significant that multiple CaM kinase II subunits were found to co-immunoprecipitate from tissue homogenates with subunits containing the 38-amino acid V 3 sequence using the V 3 sequence-specific antibody. While the co-immunoprecipitating subunits have not been identified, their sizes in brain are consistent with ␣-(50 kDa) and ␤-subunits (58 -60 kDa). In the case of smooth muscle, which contains primarily 54-and 58-kDa CaM kinase II subunits, the 62-kDa ␥ E -subunits appear to differentially co-precipitate (co-assemble) with the 58-kDa subunits. In heart, the primary co-precipitating subunit(s) has an apparent size of 50 -52 kDa and may represent a ␦ 2 -and/or ␦ 3 -subunit that has been described in this tissue (5,20). This experiment indicates that in vivo ␥ E -subunits form heteromultimeric holoenzymes with other more abundant CaM kinase II subunits. Similar evidence of heteromultimeric CaM kinase II holoenzymes in brain and carotid smooth muscle was obtained by immunoprecipitation with an antibody specific for the unique C terminus of the ␦ 2 -subunit and related subunits (8).
A smooth muscle CaM kinase II activity has been purified from chicken gizzard and reported to have a tetrameric structure, slow autophosphorylation and autoactivation responses, and unique autophosphorylation sites (32). These properties were suggested to be due to a smooth muscle ␥ B -subunit. However, in the present and previous studies (18), CaM kinase II holoenzymes composed of ␥ B -subunits expressed in COS cells were found to have properties typical of most other reported CaM kinase II holoenzymes, including a multimeric structure of 6 -9 subunits and rapid kinetics of autophosphorylation under optimal conditions (Fig. 4). It is possible that the specific V 2 sequence in the avian ␥ B -subunit (32) accounts for the unusual properties of the kinase. With respect to our original goal of defining CaM kinase II structure in mammalian vascular smooth muscle, we conclude that at least four ␥-subunit variants (␥ B , ␥ C , ␥ E , and ␥ G ) are expressed in carotid arterial smooth muscle. Of these, the ␥ B -and ␥ C -subunit mRNAs, which are translated into proteins of 58 and 56 kDa, respectively, are most abundant. Recombinant ␥ B -and ␥ C -subunits comigrate with, and could account for, the 58-and 54-kDa CaM kinase II subunits purified from carotid artery (Fig. 8). Additional studies aimed at defining the expression and properties of CaM kinase II holoenzymes containing ␦-subunit variants are required to complete the characterization of the arterial smooth muscle CaM kinase II.