Dimerization of cGMP-dependent Protein Kinase Iβ Is Mediated by an Extensive Amino-terminal Leucine Zipper Motif, and Dimerization Modulates Enzyme Function*

All mammalian cGMP-dependent protein kinases (PKGs) are dimeric. Dimerization of PKGs involves sequences located near the amino termini, which contain a conserved, extended leucine zipper motif. In PKG Iβ this includes eight Leu/Ile heptad repeats, and in the present study, deletion and site-directed mutagenesis have been used to systematically delete these repeats or substitute individual Leu/Ile. The enzymatic properties and quaternary structures of these purified PKG mutants have been determined. All had specific enzyme activities comparable to wild type PKG. Simultaneous substitution of alanine at four or more of the Leu/Ile heptad repeats ((L3A/L10A/L17A/I24A), (L31A/I38A/L45A/I52A), (L17A/I24A/L31A/I38A/L45A/I52A), and (L3A/L10A/L45A/I52A)) of the motif produces a monomeric PKG Iβ. Mutation of two Leu/Ile heptad repeats can produce either a dimeric (L3A/L10A) or monomeric (L17A/I24A and L31A/I38A) PKG. Point mutation of Leu-17 or Ile-24 (L17A or I24A) does not disrupt dimerization. These results suggest that all eight Leu/Ile heptad repeats are involved in dimerization of PKG Iβ. Six of the eight repeats are sufficient to mediate dimerization, but substitutions at some positions (Leu-17, Ile-24, Leu-31, and Ile-38) appear to have greater impact than others on dimerization. The Ka of cGMP for activation of monomeric mutants (PKG Iβ (Δ1–52) and PKG Iβ L17A/I24A/L31A/I38A/L45A/I52A) is 2- to 3-fold greater than that for wild type dimeric PKG Iβ, and there is a corresponding 2- to 3-fold increase in cGMP-dissociation rate of the high affinity cGMP-binding site (site A) of these monomers. These results indicate that dimerization increases sensitivity for cGMP activation of the enzyme.

PKG II is a separate gene product, and the amino terminus has little overall similarity to the amino terminus of PKG I isoforms (7,8). PKGs are chimeric proteins that comprise multiple functional domains; these are present in all three isoforms (9). The carboxyl-terminal region of each polypeptide chain contains the catalytic domain that includes the ATP and protein substrate-binding sites, and this domain is the most highly conserved region among the PKGs. This domain catalyzes transfer of the ␥-phosphate from ATP to specific serines or threonines in protein substrates. Immediately amino-terminal to the catalytic domain are two allosteric cGMP-binding sites (site A and site B) that are arranged in tandem. These two sites are homologous to each other, although they differ in cGMPbinding kinetics and cyclic nucleotide analog specificities (7, 9 -12). Binding of cGMP causes a conformational change that is associated with activation of the kinase (13)(14)(15). Immediately amino-terminal to the cGMP-binding sites is the autoinhibitory/autophosphorylation domain; the autophosphorylation sites lie within and near this domain (16 -18). The autoinhibitory domains of each of the PKGs contain a pseudosubstrate sequence that interacts with the catalytic site to block substrate access and thus maintain the kinase in an inactive state (19,20). Autoinhibition is relieved by cGMP binding and/or autophosphorylation in PKG I (11, 17, 20 -22), and cGMP binding activates PKG II (8). Finally, the extreme amino terminus contains the dimerization domain. Proteolysis just carboxylterminal of this domain produces a monomeric PKG with sequence that begins just amino-terminal of the autoinhibitory domain (16,23,24). In vitro, the monomeric PKGs retain the salient properties of dimeric PKGs (autoinhibition, autophosphorylation, cGMP binding, and kinase activity).
In mammalian tissues, cyclic nucleotide-dependent kinases are dimers, and dimerization primarily occurs through interaction at the amino terminus. The regulatory subunits of cyclic AMP-dependent protein kinase are dimerized in an anti-parallel alignment, and the dimerization contacts occur through hydrophobic interactions near the amino terminus of these subunits (25,26). It has been proposed that dimerization of PKG occurs through a leucine zipper motif near the amino terminus, and experimental work with a synthetic peptide based on residues 1-39 of PKG I␣ supports this interpretation (27). However, the role and extent of the leucine zipper motif in dimerization of PKGs has not been experimentally established. All three mammalian PKG isoforms (I␣, I␤, and II) contain a leucine zipper motif in this sequence, although the remainder of the residues in the sequence have limited sequence identity. Other eukaryotic PKGs such as the two Drosophila PKGs (28) and the PKGs found in Caenorhabditis elegans (29), Hydra oligactis, and Apis mellifera (30) also contain a leucine zipper motif in this region (see Fig. 1). The number of heptad repeats in the leucine zipper motif of these proteins varies from five to eight. Lower eukaryotes have monomeric PKGs that lack the leucine zipper motif; PKG from Paramecium has been shown to be a monomer (31), but this kinase retains functional activities of PKG seen in PKGs of higher eukaryotes, although careful studies of cGMP binding and activation of this PKG compared with the mammalian isoforms are lacking.
In addition to being the proposed mechanism of dimerization for PKG, the leucine zipper motif has also been shown to be involved in the interaction of PKG with a number of targeting proteins. PKG I␣ specifically associates with skeletal muscle troponin T (32), GKAP 42 in germ cells (33), and the myosinbinding subunit of myosin phosphatase (34). PKG I␤ specifically binds to skeletal muscle troponin T and inositol 3-phosphate receptor-associated PKG substrate (35). Much of the interaction with the targeting proteins occurs through the region containing the dimerization domain, and mutations within the leucine zipper motif abolish most of these interactions. However, it has not been shown that these substitutions actually disrupt dimerization. PKG has also been shown to phosphorylate these targeting proteins, and localizing PKG to substrates through the leucine zipper motif provides for efficient phosphorylation.
The leucine zipper motif was first characterized in the transcription factor field (36). It was shown to mediate the heterodimerization of transcription factors, Fos and Jun, and the homodimerization of the yeast transcription factor GCN4 (37). The majority of leucine zipper motifs do not contain more than five heptad repeats. The leucine zipper motif of each of the mammalian PKGs contains several more heptad repeats than does the same motif in transcription factors. Mammalian PKGs contain six to nine heptad repeats, depending on isoform (six heptad repeats in PKG I␣, seven to eight repeats in PKG I␤, and eight to nine repeats in PKG II), whereas transcription factors studied so far contain four to five heptad repeats. Because three to four heptad repeats are sufficient to mediate dimerization of these transcription factors (38 -42), it is not presently possible to predict the contacts required for dimerization of the PKGs. However, other proteins in addition to PKG contain six or more heptad repeats (43)(44)(45)(46)(47)(48)(49); two of these proteins contain more than 20 heptad repeats (43,48). In PKGs and other extended leucine zipper motif proteins, it is not known if the entire leucine zipper motif is required for dimerization or if three to four repeats are sufficient as is the case with transcription factors.
This report examines the role of each of the residues within the mammalian PKG I␤ leucine zipper motif in the mechanism of dimerization of mammalian PKG I␤. In addition, the influence of dimerization on enzyme functions has been determined.  Table I. The point mutations for PKG I␤ L17A/I24A/L31A/   TABLE I  Oligonucleotide pairs utilized in site-directed mutagenesis of PKG I␤  The first primer in each pair is complementary to the sense strand, and the second primer is complementary to the antisense strand. For all  mutants with the exception of L17A and I24A, two Leu/Ile were mutated per oligonucleotide as discussed under "Experimental Procedures." The  mutagenic base pairs are underlined.   PKG I␤ L17A/I24A/L31A/I38A/L45A/I52A  Step 1  5Ј-CTCCAGGAGAAGATCGAGGAGGCGAGGCAGCGGGATGCTCTCGCCGACGAGCTGGAGCTGGAGTTG-3Ј  5Ј-CAACTCCAGCTCCAGCTCGTCGGCGAGAGCATCCCGCTGCCTCGCCTCCTCGATCTTCTCCTGGAG-3Ј  Step 2  5Ј-GCCGACGAGCTGGAGCTGGAGGCGGATCAGAAGGACGAACTGGCCCAGAAGCTGCAGAACGAGCTG-3Ј  5Ј-CAGCTCGTTCTGCAGCTTCTGGCCCAGTTCGTCCTTCTGATCCGCCTCCAGCTCCAGCTCGTCGGC-3Ј  Step 3 5Ј-GCCCAGAAGCTGCAGAACGAGGCGGACAAGTACCGCTCGGTGGCCCGACCAGCCACCCAGCAGGCG-3Ј 5Ј-CGCCTGCTGGGTGGCTGGTCGGGCCACCGAGCGGTACTTGTCCGCCTCGTTCTGCAGCTTCTGGGC-3Ј PKG I␤ L3A/L10A/L17A/I24A Step Step Step I38A/L45A/I52A were made in three steps. The first step mutated Leu-17 and Ile-24. The next step utilized PKG I␤ L17A/I24A as a  template and mutated Leu-31 and Ile-38. PKG I␤ L17A/I24A/L31A/  I38A was used as a template to mutate Leu-45 and Ile-52: The point  mutations for PKG I␤ L3A/L10A/L17A/I24A, PKG I␤ L31A/I38A/L45A/  I52A, and PKG I␤ L3A/L10A/L45A/I52A were made in two steps in the  same manner as discussed for L17A/I24A/L31A/I38A/L45A/I52A. A  580-bp fragment containing the desired mutations was excised from hcGKI␤ using EcoRI/NcoI (New England Biolabs) digestion, and subcloned in EcoRI/NcoI-digested wild type hcGKI␤ clone in the pVL 1392 baculovirus expression vector. Escherichia coli XL1-blue cells were used for transformations with pKSIIϩ, and E. coli DH5␣ were used for transformations with pVL 1392. DNA fragments were purified using a Qiagen gel extraction kit according to the manufacturer's protocol (Qiagen). DNA was purified from large scale vector preparations using a Qiagen Plasmid Midi kit according to the manufacturer's protocol. All DNA segments subjected to mutagenesis and subcloning reactions were sequenced in their entirety to ensure the presence of the desired mutation and proper inframe subcloning.
Mutagenesis of PKG I␤ (⌬1-52)-Full-length PKG I␤ cDNA was ligated into the EcoRI and SmaI unique sites of the baculovirus expression vector pVL 1392. A 1441-bp EcoRI/SacI fragment containing the regulatory domain of human PKG I␤ was ligated into pBluescript IIKS ϩ (Stratagene) for oligonucleotide-directed mutagenesis based on the method of Kunkel et al. (50). The oligonucleotide (Vanderbilt University DNA Core Facility) with the following sequence was constructed to be complementary to the cDNA-coding strand except for the underlined nucleotides that serve to construct the deletion mutation and incorporate a 5Ј NdeI site. ⌬1-52, 5Ј-TGGGTGGCTGGTCGCATATG-GA-ATTCGTACTTGTCCAGCTC-3Ј.
pVL 1392-PKG I␤ (⌬1-52) vector was created by subcloning the EcoRI/SacI fragment for PKG I␤ (⌬1-52) into the large SacI/SmaI fragment of pVL 1392-PKG I␤. pVL 1392 vectors and pBluescript IIKS ϩ were propagated in E. coli (DH5␣ and XL1 Blue, respectively). All plasmids were sequenced on an Automated Biosystems, Inc. DNA Sequencer 373A in the Cancer Center DNA Core facility of Vanderbilt University.
Expression of Wild type and Mutant PKG I␤-Sf9 cells (BD Pharmingen) were cotransfected with BaculoGold linear baculovirus DNA (BD Pharmingen) and one of the mutated hcGKI␤ clones in the pVL 1392 baculovirus expression vector by calcium phosphate method according to the protocol from BD Pharmingen. At 5 days post-infection, the cotransfection supernatant was collected, amplified three times in Sf9 cells, and then used directly as virus stock for expression without additional purification of recombinant viruses. Sf9 cells grown at 27°C in complete Grace's insect medium with 10% fetal bovine serum and 10 g/ml gentamicin (Sigma) in T-175 flasks (Corning) were typically infected with 10 -100 l of viral stock/flask. The optimum volume of viral stock used per T-175 flask was experimentally determined. The Sf9 cell pellet was harvested at 72-96 h post-infection.
Purification of PKG I␤-The Sf9 cell pellet for each T-175 flask (ϳ2 ϫ 10 7 cells) was resuspended in 3 ml of ice-cold 10 mM potassium phosphate, pH 6.8, 1 mM EDTA, and 25 mM ␤-mercaptoethanol (KPEM) containing protease inhibitor mixture tablets (Roche Applied Science) of amount per volume recommended by manufacturer. Cell suspension was homogenized in 10-to 20-ml aliquots by 3 ϫ 10-s bursts in an Ultra Turrex microhomogenizer with a 20-s recovery between bursts. The cell homogenate was centrifuged at 13,000 rpm in a Beckman JA-20 rotor for 30 min at 4°C. The supernatant was loaded onto an 8-aminohexylamino-cAMP-Sepharose (Sigma) column (1 ϫ 1.5 cm) equilibrated with KPEM. The supernatant volume varied depending on number of T-175 flasks infected. The column was washed with 5 ml of KPEM with protease inhibitors followed by 10 ml of 0.5 M NaCl in KPEM with protease inhibitors. 10 mM cAMP in KPEM containing 1 M NaCl and protease inhibitors was added to the column and allowed to soak into the cAMP-Sepharose. The column was incubated for 20 min at 4°C before six 0.3-ml elutions were collected. The column was incubated for 20 additional minutes at 4°C, and six 0.3-ml fractions were collected a second time. Elutions containing kinase activity were pooled and concentrated on Centricon-30 (Amicon). The concentrated sample was chromatographed on a Sephacryl S-200 column (0.9 ϫ 35 cm) equilibrated in KPEM with protease inhibitors. This step removed cAMP from the affinity column elution. 0.5-ml fractions were collected from the Sephacryl S-200 column. For full-length PKG I␤, fractions containing PKG I␤ were applied to a 2-ml DEAE-Sephacel ion exchange column equilibrated in KPEM with protease inhibitors. The column was washed with 20 ml of KPEM with protease inhibitors followed by 20 ml of 0.12 M NaCl in KPEM with protease inhibitors to remove any breakdown products. The column was eluted with 4 ml of 0.3 M NaCl in KPEM with protease inhibitors. All purification steps were done at 4°C, and enzyme was flash-frozen in 0.3 M NaCl and 10% sucrose and stored at Ϫ70°C until use. Four preparations of PKG I␤ (⌬1-52) and two preparations of wild type PKG I␤ and PKG I␤ L17A/I24A/L31A/ I38A/L45A/I52A were expressed, purified, and used in the experiments described in this report.
SDS-PAGE and Western Blot of PKG I␤-PKG was boiled for 4 min in the presence of 10% SDS, 2 M 2-mercaptoethanol, and 0.1% bromphenol blue and subjected to 8% SDS-polyacrylamide gel electrophoresis. Proteins were visualized by Coomassie Brilliant Blue staining. For Western blot, gel was transferred to polyvinylidene difluoride membrane (Millipore). Primary antibody was rabbit polyclonal anti-PKG, and secondary antibody was goat anti-rabbit horseradish peroxidase (BioSource International). The blot was developed using ECL chemiluminescence kit (Amersham Biosciences).
PKG I␤ Kinase Assay-The kinase activity of PKG I␤ was determined by measuring 32 P i incorporation from [␥-32 P]ATP into a synthetic heptapeptide substrate (Arg-Lys-Arg-Ser-Arg-Ala-Glu, Peninsula Laboratories, Inc.) as previously described. Ten l of enzyme was added to 10 l of 100 M cGMP and 50 l of a reaction mixture, which contained 20 mM Tris at pH 7.4, 20 mM magnesium acetate, 200 M ATP, 100 M 3-isobutyl-1-methylxanthine, 136 g/ml heptapeptide, 5,000 -20,000 cpm [␥-32 P]ATP/l, and 0.45 M synthetic peptide inhibitor of the cAMPdependent protein kinase (Peninsula Laboratories, Inc.). The reaction was terminated by applying 50-l aliquots to phosphocellulose papers (Whatman P-81, 1.5 ϫ 2 cm) and washing with four changes of ϳ500 ml 75 mM phosphoric acid and one ethanol change. The papers were dried and counted by Cerenkov method.
cGMP dissociation-PKG I␤ (0.030 -0.050 mg/ml) was incubated for 10 min at 30°C with an equal volume of cGMP-binding mix (25 mM K 2 HPO 4 , 25 mM KH 2 PO 4 , 1 mM EDTA (pH 6.8), 2 M NaCl, 200 M 3-isobutyl-1-methylxanthine, 0.5 mg/ml histone IIAS (Sigma)) and 10 l of [ 3 H]cGMP (Amersham Biosciences) for a final concentration of 3 M [ 3 H]cGMP (ϳ5000 cpm/l). An incubation time of 10 min at 30°C was adequate for saturation of the cGMP-binding sites. After incubation, samples were cooled to 4°C and aliquoted in 10-l portions. The addition of 100-fold excess unlabeled cGMP at time 0 (B o ) initiated the dissociation (exchange) of bound [ 3 H]cGMP. The cGMP exchange was stopped at various time points by the addition of 2 ml of cold aqueous saturated ammonium sulfate. The samples were filtered and washed as described previously (6). The half-life of the bound cGMP was determined by the method of Rannels and Corbin (51).
Determination of Stokes Radius-Purified PKG (ϳ7-10 g) was combined with two internal standards, crystalline catalase (3 mg), and ovalbumin (4 mg), in a volume of 200 l and loaded onto a Sephacryl S-200 gel filtration column (0.9 ϫ 35 cm) equilibrated in KPEM and 150 mM NaCl at 4°C. The column was eluted with the same buffer, and fractions (0.5 ml) were collected and assayed for PKG activity to determine the elution position of the enzyme. Catalase was located by absorbance at 280 nm and/or 400 nm, and ovalbumin was located by absorbance at 280 nm. The column was standardized with protein standards of known Stokes radii: cytochrome c (16.6 Å), ovalbumin (29 Å), bovine serum albumin (35 Å), and catalase (52 Å). The thyroglobulin elution volume was taken as the void volume. Elution positions of the protein standards were used to generate a standard curve of (Ϫlog K av ) 1/2 versus Stokes radius (52) as follows, The Stokes radii of PKGs were determined from the standard curve based on elution volumes.
Determination of Sedimentation Coefficient-Purified PKG (ϳ7-10 g) was combined with two internal standards, crystalline phosphorylase-b (3 mg), and hemoglobin (0.5 mg), in a volume of 200 l in KPEM/ 0.15 M NaCl with protease inhibitors and applied to a 13-ml linear 5-20% sucrose gradient containing KPEM/0.15 M NaCl. The gradients were centrifuged at 37,000 rpm in a Beckman SW 41 rotor for 30 -44 h at 4°C, and fractions (0.5 ml) were collected from the bottom of the tubes. Phosphorylase-b has a s 20,w value of 8.0 S and was located by absorbance at 280 nm. Hemoglobin has a s 20,w value of 3.2 S and was located by absorbance at 280 nm and/or 411 nm. PKG activity was measured to determine the sedimentation position of the enzyme. The sedimentation coefficients of the PKGs were then determined by the distance migrated into the gradients as compared with the standards. The molecular weights were calculated from the Stokes radii and sed-imentation coefficients according to the method of Siegel and Monty (52), using the equation, where N is Avogadro's number; n is the viscosity of medium, assumed to be 1; a is the Stokes radius; s is the sedimentation coefficient; is the density of the medium, assumed to be 1; and is the partial specific volume, assumed to be 0.725 ml/g. Protein Quantification-Protein was determined by the method of Bradford using staining reagent from Bio-Rad and bovine serum albumin fraction V (Sigma) as standard. This method routinely overestimates the amount of PKG protein by 37% (6), and this correction factor was used.
Materials-Reagents for recombinant enzyme preparations were obtained via the Vanderbilt Diabetes Center Molecular Biology Core. The Vanderbilt Diabetes Center Tissue Culture Core facility provided many media and competent cells. Baculovirus expression vector pVL 1392, BaculoGold DNA, and transfection reagents and protocols were obtained from BD Pharmingen. Sf9 log phase cells were purchased from BD Pharmingen. Cytochrome c, ovalbumin, bovine serum albumin, catalase, and phosphorylase-b were from Sigma. Hemoglobin was obtained from Nutritional Biochemicals Corp.

RESULTS AND DISCUSSION
The leucine zipper motif has been proposed to be the mechanism of dimerization of PKG I␤; however, this has not been directly tested for any PKG. Furthermore, the role of dimerization in determining PKG function is poorly understood. This report examines both of these questions.
Requirement of the Leucine Zipper Motif of PKG I␤ for Dimerization-This study first investigated whether the leucine zipper motif is the mechanism of dimerization for PKG I␤. Two types of mutants of human PKG I␤ were created and characterized. First, deletion mutagenesis was used to remove the amino-terminal 52 residues from PKG I␤ (PKG I␤ (⌬1-52)). This segment contains the entire leucine zipper motif (Fig. 1). Second, site-directed mutagenesis of the leucine zipper motif was used to create a mutant in which six of the eight heptad repeats were changed to alanine (L17A/I24A/L31A/I38A/L45A/ I52A). Alanine was chosen, because it is an apolar residue that should maintain the ␣-helical structure of this region, but it should not promote any interactions similar to those in a leucine zipper motif. These proteins were then expressed using the baculovirus system in Sf9 cells and purified as described in "Experimental Procedures." The amount of PKG expressed varied somewhat depending on construct, but each typically displayed at least 10-fold greater cGMP-stimulated kinase activity and 100-fold greater amount of total kinase present over that of mock-infected cells. Expressed PKG had a kinase activity ratio (activity Ϫ cGMP/activity ϩ cGMP) Յ 0.1, which was similar to that of native PKG I␤ purified from bovine aorta (6). These mutant PKGs had similar specific enzyme activities, and affinities for substrates were similar to that of wild type PKG I␤ (Table II) (66), and PKG II (7,67). DG1 and DG2-T1 are the Drosophila PKG isoforms (28). H. oligactis is hydra, and A. mellifera (30) is honeybee. tivity with PKG antibodies in immunoblot experiments (not shown).
The quaternary structures of the two PKG I␤ mutants and of wild type PKG I␤ were obtained according to the method of Siegel and Monty (52). Wild type PKG I␤ was initially characterized, and the measured physical properties of the mutants were compared with the wild type enzyme (Table III). Gel filtration chromatography was used to determine the Stokes radius. As shown in Fig. 2A, the wild type PKG I␤ elution slightly preceded the internal standard catalase, and its Stokes radius was determined to be 55 Å. A small shoulder of kinase activity eluting in later fractions was also observed, suggesting the presence of a proteolytic breakdown product of PKG. On sucrose density gradient centrifugation, wild type PKG I␤ sedimented slightly more slowly than did the internal standard phosphorylase-b (Fig. 3A), and the sedimentation coefficient was determined to be 7.8 S. The calculated molecular mass of wild type PKG I␤ was 180 kDa (Table III). This compares well with the molecular mass of 156 kDa predicted from the amino acid content (5). The Stokes radius, sedimentation coefficient, and calculated molecular weight of wild type PKG I␤ were similar to the corresponding values obtained for native PKG I␤ purified from bovine aorta (6).
The sequence deleted in PKG I␤ (⌬1-52) contained the entire leucine zipper motif and was characterized first. Previous work had shown that a proteolytically derived fragment of PKG I␤ beginning at Gln-62 is monomeric and retains the salient features of wild type PKG I␤ (24). However, it was not known if deletion of only the first 52 residues, which include the leucine zipper motif, would also produce a dimer or monomer. Amino acid sequencing of PKG I␤ (⌬1-52) verified that the amino terminus of the construct began at Arg-53 as predicted (15). Sephacryl S-200 gel filtration chromatography was used to determine a Stokes radius for PKG I␤ (⌬1-52). As shown in Table III, the Stokes radius was 37 Å. Analysis of the proteins by sucrose density gradient centrifugation determined that the sedimentation coefficient was 5.0 S (Table III). The values for the Stokes radii and sedimentation coefficients were used to calculate a molecular weight using the Siegel and Monty equation (52) as described under "Experimental Procedures." The calculated molecular mass of PKG I␤ (⌬1-52) was 76 kDa, consistent with it being a monomer; this value is approximately half the molecular weight of the dimeric wild type PKG (5). Although this mutant established that the first 52 residues of PKG I␤ are important for dimerization, it did not specifically address the role of the leucine zipper motif as the sole mediator of dimerization. To address this question, six of the eight Leu/ Ile heptad repeats were mutated to alanine, beginning at Leu-17 and extending through Ile-52. Because a minimum of 3-4 Leu/Ile heptad repeats are required for dimerization in transcription factors (38 -42), these substitutions would be predicted to disrupt dimerization of PKG I␤ if it was mediated by a leucine zipper.
To determine if the leucine zipper motif is directly involved in dimerization, the quaternary structure of PKG I␤ L17A/ I24A/L31A/I38A/L45A/I52A (Leu Zip Mut) was examined. On gel filtration chromatography, PKG I␤ L17A/I24A/L31A/I38A/ L45A/I52A eluted between the two internal standards, catalase and ovalbumin (Fig. 2B), and significantly later than did wild type PKG I␤ (compare Fig. 2, A and B). PKG I␤ (⌬1-52) and PKG I␤ L17A/I24A/L31A/I38A/L45A/I52A had Stokes radii of 37 and 44 Å, respectively (Table III). The Stokes radii for PKG I␤ (⌬1-52) and L17A/I24A/L31A/I38A/L45A/I52A differed slightly, which was in keeping with the ϳ6000-Da difference in their predicted molecular masses. Because the Stokes radius reflects shape as well as size, it is also possible that the L17A/ I24A/L31A/I38A/L45A/I52A mutant exists in a more extended conformation compared with PKG I␤ (⌬1-52). On sucrose density gradient centrifugation, PKG I␤ L17A/I24A/L31A/I38A/ L45A/I52A sedimented between the two internal standards (Fig. 3B). This contrasts with the pattern of wild type PKG I␤, which sedimented slightly more slowly than did the internal standard phosphorylase-b (compare Fig. 3, A to B). The sedimentation coefficient of PKG I␤ L17A/I24A/L31A/I38A/L45A/ I52A was 5.4 S as compared with 7.8 S for wild type PKG I␤ (Table III). The values for the Stokes radius and sedimentation coefficient were used to calculate a molecular mass of 98 kDa for PKG I␤ L17A/I24A/L31A/I38A/L45A/I52A (Table III) using the Siegel and Monty equation (52) as described under "Experimental Procedures." The calculated molecular weight of PKG I␤ L17A/I24A/L31A/I38A/L45A/I52A suggests that it is a monomer and that the leucine zipper motif is required for dimerization of PKG I␤.
Identifying residues of the leucine zipper of PKG I␤ that contribute to dimerization-The leucine zipper motif of PKG I␤ is composed of eight heptad repeats as compared with four to five repeats in the leucine zipper of transcription factors studied so far. Therefore, we investigated the contribution of the eight heptad repeats to dimerization of PKG I␤. Three separate quadruple mutants were created, and a calculated molecular weight was determined for each using the method described above (Table III). All mutants were expressed in the baculovirus/Sf9 cell system and purified to near homogeneity as described under "Experimental Procedures." All of these mutants migrated as ϳ78-kDa proteins on SDS-PAGE and retained the salient enzymatic features of wild type PKG I␤ (data not shown). PKG I␤ L3A/L10A/L17A/I24A contained mutations of the amino-terminal four Leu/Ile heptad repeats but the four carboxyl-terminal Leu/Ile heptad repeats remained intact. Conversely, in PKG I␤ L31A/I38A/L45A/I52A the carboxylterminal 4 Leu/Ile heptad repeats were replaced with alanine, which left the amino-terminal four repeats unaltered. PKG I␤ L3A/L10A/L45A/I52A substituted two Leu/Ile heptad repeats at either end of the motif, leaving the middle four Leu/Ile heptad repeats intact. Physical properties of these mutants were determined as described above. As shown in Table III, the calculated molecular weights of these mutants indicated that all three mutants are monomeric. These results indicated that four Leu/Ile heptad repeats within an expanse of 28 amino acids are not sufficient to mediate dimerization in PKG I␤. In this respect, the number and arrangement of Leu/Ile heptad repeats required for dimerization of PKG I␤ differs markedly from that reported for previously studied leucine zipper motifs of transcription factors (38 -42). Because four heptad repeats are not sufficient for dimerization of PKG I␤, three additional mutations were created; in each, only two Leu/Ile repeats of the leucine zipper motif were replaced with alanine. These mutants were L3A/L10A, L17A/ I24A, and L31A/I38A (Table III). The calculated molecular weight of the L3A/L10A PKG indicated that it is a dimeric enzyme, suggesting that the carboxyl-terminal six Leu/Ile repeats were sufficient to mediate dimerization. However, the L17A/I24A and L31A/I38A mutants displayed physical properties that were more complex in that the calculated molecular mass was intermediate between a monomer and dimer. The calculated molecular masses of these mutants were 120 and  140 kDa, respectively. To further clarify the properties of these two mutants, additional analyses were performed.
Sephacryl S-200 gel filtration profile of PKG I␤ L31A/I38A is shown in Fig. 4A. Approximately 10 g of PKG protein was applied to the gel filtration column. A major peak of protein kinase activity (peaking at fraction 34) was observed, but there was a significant trailing shoulder (fractions 36 -40). The molar concentration of PKG I␤ in the peak fraction (fraction 34) was calculated to be 51 nM using a specific enzyme activity of 2.5 mol/min/mg for PKG. To further evaluate the characteristics of this mutant, an early fraction (fraction 33) containing 0.56 g of PKG was rechromatographed on the Sephacryl S-200 gel filtration column. Following this rechromatography, PKG I␤ L31A/I38A eluted in a single peak between the two internal standards, catalase and ovalbumin (Fig. 4B), and had a Stokes radius of 41 Å, consistent with it being a monomer (Table III). The molar concentration of PKG I␤ in the peak fraction (fraction 36) following rechromatography was calculated to be 0.95 nM, which was significantly lower than the concentration of 51 nM obtained after the first chromatography. Western blot analysis of the PKG peak in the S-200 fractions from the second gel filtration (Fig. 4B) displayed a 78-kDa band as well as a minor 70-kDa band that was most likely a proteolytic breakdown product (data not shown). Thus, at the higher concentration (51 nM), PKG I␤ L31A/I38A appeared to be in equilibrium between a dimer and a monomer. However, at a 50-fold lower concentration (0.95 nM), it was exclusively a monomer. When 0.16 g of PKG I␤ L31A/I38A from fraction #33 in Fig. 4A was subjected to sucrose density gradient centrifugation, it sedimented between the two standards, phosphorylase-b and hemoglobin, also consistent with it being a monomer (data not shown). Its sedimentation coefficient was 5.2 S and the calculated molecular mass was 88 kDa (Table III). This calculated molecular mass was significantly lower than the 140-kDa value ( Table   III) that was obtained after the first gel filtration, in which the enzyme was present at relatively high concentration.
Because the calculated molecular weight for the double mutant PKG I␤ L17A/I24A was also intermediate between a dimer and a monomer, it was also analyzed for a concentration effect on the predicted quaternary structure. PKG I␤ L17A/I24A (0.43 g) was subjected to gel filtration and sucrose density gradient centrifugation to determine if it would also behave as a monomer at low concentrations. Using both gel filtration and sucrose density gradient centrifugation, the enzyme exhibited a single peak in a manner consistent with it being a monomer (data not shown). The Stokes radius was 41 Å, and the sedimentation coefficient was 5.1 S, which resulted in a calculated molecular mass of 86 kDa (Table III). To test whether this monomerization occurred exclusively from dilution or from weakening of dimer contacts, both wild type PKG I␤ and the double mutant L3A/L10A were subjected to Sephacryl S-200 gel filtration at similarly low enzyme concentration (0.47 g of wild type and L3A/L10A applied to column). Both proteins eluted in a manner consistent with a dimeric quaternary structure. The Stokes radii for the wild type and L3A/L10A PKGs were 55 and 52 Å, respectively. Thus, neither wild type PKG I␤ nor PKG I␤ L3A/L10A were monomerized after extensive dilution. It is possible that the Leu/Ile heptad repeats in the central portion of the leucine zipper motif contribute more importantly to dimerization than do the flanking Leu/Ile heptad repeats of the motif. Thus, mutation of two of these central Leu/Ile heptad repeats weakened dimerization to the point that a partial monomerization became complete after ϳ50-fold dilution. It can be estimated that the K D for dimerization for both L17A/I24A and L31A/I38A is between 0.95 and 51 nM. Because the wild type and PKG I␤ L3A/L10A forms remained dimers at ϳ1 nM, it can be stated that the K D for dimerization of each of these forms is Ͻ1 nM.
Single residue mutations of Leu-17 and Ile-24 to alanine were also characterized for quaternary structure. Both PKG I␤ L17A and PKG I␤ I24A were dimers based on their calculated molecular mass (Table III), suggesting that mutating one of the heptad repeats is not sufficient to disrupt dimerization. Based on these results, it is concluded that six heptad repeats are sufficient to mediate dimerization of PKG I␤. The Leu/Ile residues in the central portion of the zipper motif are more critical, albeit not required, for dimerization than are those at the amino-terminal end. To our knowledge, this is the first evidence of a leucine zipper motif that requires six heptad repeats to mediate dimerization. Other leucine zipper motifs contain six or more heptad repeats (43)(44)(45)(46)(47)(48)(49). These include non-transcription factor proteins (zipper protein, enterophilin, luman, and myocilin) as well as some transcription factors (LZIP-1, LZIP-2, ATB2, and hMAF). Zipper protein and enterophilin each contain more than 20 heptad repeats (43,48); luman, LZIP-1, LZIP-2, ATB2, hMAF, and myocilin vary from 7-13 heptad repeats in their leucine zippers (44 -47, 49). With the exception of myocilin, it has not been determined how many heptad repeats are needed for dimerization of these proteins. It appears that the amino-terminal four heptad repeats are sufficient for dimerization of myocilin (53). However, the heptad repeats were mutated to glycine. This residue will often destabilize ␣-helices, and it is possible that it disrupted the secondary structure of myocilin in the region of the leucine zipper motif.
Comparison of the PKG Leucine Zipper Motif to Those in Transcription Factors-Because the majority of the work on the mechanism of dimerization by leucine zipper motifs has been in the transcription factor field, the features of the PKG leucine zipper motif can be compared with that of transcription factors. The structure of leucine zipper motifs in proteins is similar to a coiled-coil (54,55). This motif consists of a heptad repeat of leucines/isoleucines (Leu/Ile) that occurs within an ␣-helical structure. The Leu/Ile heptad repeats align on one side of the ␣-helix, and two right-handed helices coil around each other to form a slight left-handed superhelical twist. The helical repeat is reduced from 3.6 (classic ␣-helix) to 3.5 residues per turn, which allows for the alignment of the leucines on one face. The sequences of leucine zippers are often described in terms of a heptad pattern (abcdefg) n , with the critical leucine at position d, but other residues within this motif contribute importantly to stabilization of the dimer.
Comparison of the sequences in the leucine zipper motifs of PKGs and transcription factors reveals notable differences. In B-ZIP transcription factors, leucine occupies 84% of the d positions (56). In contrast, leucine occupies only 53% of the d positions in the PKG isoforms (Fig. 1), and isoleucine is present in 36% of the isoforms with valine and cysteine comprising the remaining 11% of the d positions. Leucine has been shown to be the optimal residue to occupy the d position in dimeric structures, because the C␣-C␤ bond of the knob side chain of leucine points directly at the adjacent helix (perpendicular packing). However, ␤-branched residues in the d position (isoleucine and valine) are less advantageous for dimerization, because they can cause steric clashes in this configuration (57). The a position is often stabilized by hydrophobic amino acids or by asparagines (57)(58)(59). There are 51 a positions in the PKG isoforms listed in Fig. 1. A non-polar residue or an asparagine is present in only four of these a positions. Another feature that stabilizes dimerization is an interhelical salt bridge between the e and g positions (60 -62). The side chain of the g position interacts with the other helix at the e position five amino acids closer to the carboxyl terminus. There are only two possible e-g interactions among all the PKG isoforms; PKG I␤ contains Arg-20 at a g position and Glu-25 at an e position five amino acids closer to the carboxyl terminus, and the H. oligactis isoform contains Lys-40 at a g position and Glu-45 at the next e position closer to the carboxyl terminus. In the absence of these stabilizing contacts, PKG may require a more extensive interface of leucine/isoleucine contacts to stabilize the enzyme as a dimer.
The characteristics of the extended leucine zipper motifs of luman, hMAF, LZIP-1, LZIP-2, ATB2, and myocilin were compared with those of transcription factors containing four to five heptad repeats. The number of heptad repeats in the extended motifs varied from 7 to 13 (44 -47, 49). Although two proteins (enterophilin and zipper protein) contained motifs with greater than 20 residues (43, 48), these will not be considered, because the sizes of their motifs were much greater than that of PKG or any of the other proteins. The d positions of luman, hMAF, LZIP-1, LZIP-2, ATB2, and myocilin are more leucine rich than those of PKG but less leucine-rich than those of transcription factors containing four to five heptad repeats; they are comprised of 75% leucine and 25% non-leucine (Ser, Tyr, Ile, Val, Met, and Thr). The a positions of luman, hMAF, LZIP-1, LZIP-2, ATB2, and myocilin resemble those of transcription factors with four to five heptad repeats. They contain 41% hydrophobic residues (Val, Leu, and Ile), and one or two asparagines per protein in the a position. There are no extensive e-g interactions within either luman, hMAF, LZIP-1, LZIP-2, ATB2, or myocilin, but more possible interactions exist in these proteins than seen in the PKGs. Each protein contains one or two possible e-g salt bridges, and myocilin also has two possible e-g electrostatic repulsions. The extended leucine zipper motifs of luman, hMAF, LZIP-1, LZIP-2, ATB2, and myocilin share more similarities with transcription factors containing four to five heptad repeats than do the PKGs. Given this information, it appears unlikely that luman, hMAF, LZIP-1, LZIP-2, ATB2, and myocilin would require their entire extended leucine zipper motif for dimerization, but that still remains to be determined.
The features of the leucine zipper motif of PKG differ from those of other proteins with extended leucine zipper motifs and also from the classic four-to five-heptad leucine zipper motif characterized in transcription factors. To our knowledge, no proteins have been shown to require six or more heptad repeats to mediate dimerization through a leucine zipper. Dimerization of PKGs could possibly occur through a different type of leucine zipper with features that are dissimilar to transcription factors. Another possibility is that dimerization occurs through a novel interaction not related to leucine zippers. Structural investigation is needed to determine whether PKG utilizes an extensive leucine zipper mechanism or if a novel interaction involving the heptad repeats is responsible for dimerization of this enzyme.
Mutagenesis of Leucine Zipper Motifs in Other Proteins-As noted above, in proteins known to employ a leucine zipper for dimerization, leucine is the most common residue at the d position. Single replacement by conservative residues such as valine or isoleucine typically has minimal effect on dimerization. The disadvantage of having these latter residues in the d position is the potential steric clashes from their ␤-branched side chains. If all of the d-position leucines of dimeric GCN4, which has four heptad repeats, were mutated to isoleucine or valine, trimers or tetramers resulted because the interactions formed by the ␤-branched side chains of these residues favor formation of these oligomers rather than dimeric structures (57). Substitution of a single leucine in this position had minimal effects on dimerization of GCN4 and other proteins (38 -42, 63, 64). Two studies examined the effect on structural stability, or dimerization, upon mutagenesis of a single leucine in the d position. The energetic contribution of a single amino acid substitution was determined by measuring thermal stability (63) or resistance to chemical denaturation (65). The leucine zipper domain of dimeric vitellogenin binding protein (chicken homolog of the mammalian protein TEF) contains five heptad repeats. The fifth d position was mutated to each of seven different amino acids (Leu, Met, Ile, Val, Cys, Ala, and Ser), and dimerization still occurred in each mutant. Leucine was found to be the most stabilizing followed by Met, Ile, Val, Cys, Ala, and Ser (63). A similar study using a synthetic 38amino acid peptide found leucine to be the most stabilizing residue in the d position followed by Met, Ile, Tyr, Phe, Val, Gln, Ala, Trp, Asn, His, Thr, Lys, Ser, Asp, Glu, Arg, and Gly (65).
Extensive mutagenesis studies have been performed on the leucine zipper motifs of the transcription factors Fos, Jun, and GCN4. Disruption of dimerization varied depending on the particular protein and the numbers of heptad repeats that were mutated. GCN4 contains four d-position leucines in its leucine zipper motif. Substitution of a single leucine in GCN4 (38) did not disrupt dimerization, regardless of which of the four dposition leucines was changed, and a wide variety of amino acids were tolerated in these substitutions. When two of any of the four d-position leucines were mutated to a variety of residues, dimerization was typically disrupted. Dimerization was preserved only in a few instances when leucine was conservatively replaced with valine, phenylalanine, and/or isoleucine. These results suggest that a minimum of three d position zipper residues are critical for dimerization of GCN4. Similar results were seen in mutagenesis studies of the dimerization requirements for the heterodimer of Fos and Jun (39 -42). Both Fos and Jun contain five d-position leucines in their leucine zipper motifs. When any one d-position leucine was mutated, heterodimerization was retained. Replacing any two of the five d-position leucines in Jun or any one of the five d-position leucines in Fos did not disrupt heterodimerization. In addition, if both Fos and Jun were mutated, heterodimerization still occurred in most cases providing that Jun contained three d-position leucines and Fos contained four d position leucines. There were some exceptions depending on which combination of d-position leucines was mutated. Therefore, to achieve a stable heterodimer between Jun and Fos, Jun requires a minimum of three d-position leucines, and Fos requires a minimum of four d-position leucines. The heterodimerization of Jun and Fos also tolerated a variety of residues in the d positions. The only residue that routinely abolished dimerization was proline, most likely by disrupting the secondary structure of the ␣-helix.
Other Functions of the Extreme Amino-terminal Domain of PKGs-In addition to mediating dimerization, the region of PKG containing the leucine zipper motif has been reported to be responsible for associating with targeting proteins. In most instances, mutation of selected leucines or isoleucines within the leucine zipper motif of the PKGs abolishes the interaction with binding proteins, but no one has previously assessed the effect of these mutations on PKG dimerization. The double mutants L12A/I19A and I33A/L40A in the leucine zipper motif of PKG I␣ displayed little or no binding to myosin-binding subunit of myosin phosphatases (34), and mutation of I19A/ L40A abolished interaction with skeletal muscle troponin T (32). These authors interpreted these effects as being due to disruption of the dimerization of PKG I␣. If so, the results of these studies suggest that only four heptad repeats are not sufficient to sustain dimerization in PKG I␣. As in PKG I␤, the leucine zipper motif of PKG I␣ is dissimilar to that seen in transcription factors. However, the effect of these mutations on the dimerization of PKG I␣ is currently unknown and must be experimentally established. Fos, Jun, and GCN4 only require three to four heptad repeats to mediate dimerization (38 -42). The L12A/I19A mutant of PKG I␣ leaves the carboxyl-terminal four heptad repeats intact (see Fig. 1), and the I33A/L40A mutant leaves the amino-terminal three heptad repeats and the last carboxyl-terminal repeat intact. If the leucine zipper motif of PKG I␣ was similar to that of transcription factors, these mutations should not disrupt dimerization. However, if mutating two of the six heptad repeats in PKG I␣ disrupts dimerization, this would indicate that PKG I␣ may require an extended leucine zipper motif to mediate dimerization, similar to the case for PKG I␤. Because the leucine zipper motif of PKG is quite extensive, it is possible that a portion of the heptad repeats is involved in associating with targeting proteins and that several heptad repeats are necessary for the dual roles of this region, i.e. mediating dimerization and association with targeting proteins.
Activation of Wild Type and Monomeric Mutants by cGMP-Although all mammalian PKGs are dimeric, monomeric PKGs produced by limited proteolysis (23,24) or truncation mutagenesis retain many of the salient features of the dimeric enzyme. As discussed above, the region of PKGs containing the leucine zipper motif has been shown to play a critical role in selective interaction of PKG isoforms with other proteins (32)(33)(34)(35), and in some instances, this interaction is required for phosphorylation of a substrate by PKG (32)(33)(34). However, the contribution of dimerization to the enzymatic functions of PKGs has been unclear. The effect of dimerization on the potency of cGMP activation of PKG I␤ was determined by measuring kinase activity of the dimeric wild type PKG I␤ and monomeric mutants PKG I␤ (⌬1-52) and PKG I␤ L17A/I24A/L31A/I38A/ L45A/I52A in the presence of increasing concentrations of cGMP (Fig. 5). Kinase activation constants (K a ) were determined for all three PKG constructs using GraphPad Prism graphics. Wild type PKG was more sensitive to activation by cGMP than were the monomeric enzymes. There was a 2-to 3-fold increase in K a of the two monomeric mutants compared with wild type PKG I␤ (Table II). The K a increase was seen with both monomeric PKGs in the present investigation along with two proteolytically derived monomers of PKG I␤ (16,24), which provides strong argument that the change in K a is caused by dimerization and not by a conformational change induced by mutagenesis or truncation of the amino-terminal region.
Cyclic GMP Dissociation/Exchange from Wild Type and Monomeric PKG I␤ Mutants-The difference in K a of cGMP for dimeric and monomeric PKG I␤ could be due to a change in affinity for cGMP, or it could be attributed to a change in relief of the potency of autoinhibition by cGMP binding. To determine if cGMP-binding affinity differed between the dimeric and monomeric PKG I␤ (wild type, ⌬1-52, and L17A/I24A/L31A/I38A/ L45A/I52A), the K D value for cGMP binding was determined for each. There was no detectable difference in this value between wild type PKG and the monomeric PKGs (data not shown). Because PKG I␤ contains two cGMP-binding sites per subunit, the K D represents an average of the affinities of these two sites. If the affinity of only one of the binding sites is altered, then this method might not be sufficiently sensitive to measure a change in K D . To determine if one or both of the cGMPbinding sites is affected by dimerization, [ 3 H]cGMP dissociation/exchange from the enzymes was examined. The pattern of [ 3 H]cGMP dissociation from all three PKGs was biphasic, consistent with the two kinetically distinct cGMP-binding sites that were previously reported (Fig. 6) (11). Because dissociation of the fast cGMP-binding site for all constructs is complete at the first time point (3 s), [ 3 H]cGMP dissociation rate of the dimeric wild type and monomeric PKGs could not be determined for this site. The dissociation rate of [ 3 H]cGMP from the slow cGMP-binding site was two to three times more rapid in the two monomeric PKGs than in wild type PKG I␤ as reflected in the t1 ⁄2 values (Table II). This finding was consistent with the 2-to 3-fold increase in K a in the two monomeric PKGs. The results suggested that dimerization increases the affinity of the slow cGMP-binding site for cGMP, resulting in improved affinity of the enzyme for cGMP.
To summarize studies of enzyme functions, catalytic activities of the wild type and the two monomeric mutant PKGs were comparable. They displayed similar specific enzyme activities and K m values for both ATP and a heptapeptide substrate (RKRSRAE). However, dimerization enhanced cGMP activation of the enzyme, which was associated with increased affinity of the slow cGMP binding site (site A).
Concluding Remarks-The results indicate that the leucine zipper motif of PKG I␤ is responsible for dimerization. Furthermore, six of the eight heptad repeats are required to mediate and stabilize this interaction. To date, this is the most extensive leucine zipper motif that has been shown to be required for dimerization. In addition to binding targeting proteins, the advantage of PKG I␤ dimerization is increased sensitivity to cGMP activation of kinase activity, which results at least in part from increased affinity of the slow cGMP binding site (site A).