Autophosphorylation of type Ibeta cGMP-dependent protein kinase increases basal catalytic activity and enhances allosteric activation by cGMP or cAMP.

Autophosphorylation of purified bovine Iβ isozyme of cGMP-dependent protein kinase (Iβ cGK) in the presence of cGMP or cAMP increased basal kinase activity (−cGMP) as much as 4-fold and reduced the Ka for both cGMP and cAMP; maximum catalytic activity (+cGMP) was not altered. Autophosphorylation proceeded with at least two rate components. The faster rate correlated with phosphorylation of Ser-63. The slower rate, as well as the increase in basal kinase activity and decrease in Ka for cyclic nucleotides, correlated with phosphorylation of Ser-79. Autophosphorylation of either residue was an intramolecular reaction. Autophosphorylation of a proteolytically generated Iβ cGK monomer lacking amino-terminal residues 1-64 increased basal activity (3-fold) and decreased Ka for cAMP (15-fold). This indicated that autophosphorylation of Ser-79 did not require dimeric cGK and that the phosphorylation of Ser-79 in the monomer was sufficient to alter enzymatic characteristics of Iβ cGK. These studies suggested that increases in intracellular cGMP or cAMP could result in autophosphorylation of Iβ cGK, which would increase basal kinase activity as well as the sensitivity of cGK to activation by cGMP or to cross-activation by cAMP. Autophosphorylation could also prolong the increased kinase activity after decline of the second messenger.

Autophosphorylation of purified bovine I␤ isozyme of cGMP-dependent protein kinase (I␤ cGK) in the presence of cGMP or cAMP increased basal kinase activity (؊cGMP) as much as 4-fold and reduced the K a for both cGMP and cAMP; maximum catalytic activity (؉cGMP) was not altered. Autophosphorylation proceeded with at least two rate components. The faster rate correlated with phosphorylation of Ser-63. The slower rate, as well as the increase in basal kinase activity and decrease in K a for cyclic nucleotides, correlated with phosphorylation of Ser-79. Autophosphorylation of either residue was an intramolecular reaction. Autophosphorylation of a proteolytically generated I␤ cGK monomer lacking amino-terminal residues 1-64 increased basal activity (3-fold) and decreased K a for cAMP (15-fold). This indicated that autophosphorylation of Ser-79 did not require dimeric cGK and that the phosphorylation of Ser-79 in the monomer was sufficient to alter enzymatic characteristics of I␤ cGK. These studies suggested that increases in intracellular cGMP or cAMP could result in autophosphorylation of I␤ cGK, which would increase basal kinase activity as well as the sensitivity of cGK to activation by cGMP or to cross-activation by cAMP. Autophosphorylation could also prolong the increased kinase activity after decline of the second messenger.
cGMP-dependent protein kinase (cGK) 1 is localized to few tissues relative to the ubiquitous cAMP-dependent protein kinase (cAK). To date, three mammalian isozymic forms of cGK have been reported, type I␣, type I␤, and type II. I␣ and/or I␤ cGK has been found in smooth muscle cells, cerebellar Purkinje cells, platelets, and neutrophils (1)(2)(3)(4), whereas type II cGK or mRNA corresponding to this isozyme has been reported in intestinal epithelium, brain, and lung (5,6). Several regulatory roles have been ascribed to cGK in mammalian cells (7)(8)(9)(10).
I␣ and I␤ cGK have molecular masses of 76,331 Da and 77,803 Da, respectively (11,12), and they exist as homodimers of subunits that contain a regulatory domain and a catalytic domain in a single polypeptide sequence. The primary structures of the two isoforms differ only in the amino-terminal ϳ100 residues, sharing 36% identity in this region (12). This segment contains the dimerization domain, autophosphorylation sites, and the autoinhibitory domain (9,56). The cyclic nucleotide-binding or catalytic domains of the isozymes have identical sequences (11)(12)(13)(14). However, cyclic nucleotide-binding and activation properties of the isozymes vary, indicating that the amino-terminal domains confer unique properties on the binding sites. The regulatory domain contains two cyclic nucleotide-binding sites, one low affinity site (fast site) and one high affinity site (slow site) (15). For I␣ cGK, these sites exhibit positive cooperativity (15,16). Autophosphorylation is a common property of protein kinases and, in many instances, plays an integral role in the regulation of function of these enzymes (for review see Ref. 17). Both I␣ and I␤ cGK are modified by autophosphorylation. I␣ cGK autophosphorylates a single residue, Thr-58 (18) (0.75 mol/mol of cGK subunit), in 3 h at 30°C when incubated in the presence of cGMP (19). However, in the presence of cAMP, 2.5 mol/mol subunit is incorporated under the same conditions (19) and Ser-50, Ser-72, Thr-84 as well as Thr-58 are modified (18). Autophosphorylation of I␣ cGK decreases cooperativity of cyclic nucleotide binding (20), decreases the K a for cAMP, and decreases the K d of the slow site for cAMP from 3 to ϳ0.5 M (21) without affecting the K d for cGMP (20). I␤ cGK has been studied less extensively than has I␣ cGK. With cGMP-stimulated autophosphorylation of I␤ cGK, ϳ97% of the phosphate is associated with serine residues (22). For relatively short incubations, the major site of cGMP-or cAMPstimulated autophosphorylation of purified bovine aorta I␤ cGK is Ser-63 (56). The present studies provide the first thorough characterization of I␤ cGK autophosphorylation, including proof of a second autophosphorylation site, which is likely responsible for the increase in basal kinase activity and increase in sensitivity to cyclic nucleotides. A preliminary report of some of this work was presented in abstract form (23,24).

EXPERIMENTAL PROCEDURES
Purification of I␣ and I␤ cGK and Kinase Assay-Both I␣ and I␤ cGK were purified to homogeneity from bovine lung and bovine aorta, respectively, according to Francis et al. (25). Purity was determined by SDS-PAGE analysis. The activity of cGK was determined as described previously (25).
Autophosphorylation-Unless otherwise indicated, 30 l of I␣ or I␤ cGK in buffer A (10 mM potassium phosphate, 200 mM NaCl, 2 mM EDTA, and 25 mM 2-mercaptoethanol) was incubated for 2 h at 30°C with 8 l of 5-10 M cGMP or 10 -20 M cAMP and 5 l of a mixture containing 45 mM magnesium acetate, 0.85 mM ATP (ϳ500,000 cpm/l [␥-32 P]ATP). For non-autophosphorylated enzyme, the Mg/ATP was omitted. Assessment of 32 P i incorporation was performed as described * This work was supported by National Institutes of Health Grant DK40029. 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.
[ 3 H]cGMP Dissociation-The [ 3 H]cGMP dissociation behavior of cGK was determined by a modified version of the Millipore filtration assay described by Sugden and Corbin (26). 7 l of enzyme (1 M final concentration) was added to 33 l of reaction mixture containing 10 mM KH 2 PO 4 /K 2 HPO 4 at pH 6.8, 0.2 mM EDTA, 0.5 mg/ml Sigma histone VIII-S, and 20 M [ 3 H]cGMP (specific activity ϳ6000 cpm/pmol). Equilibrium binding reactions were done at 30°C for 30 min. The samples were cooled to 4°C on ice. A 100-fold excess of unlabeled cGMP was added, and at time points indicated, 2 ml of cold saturated aqueous (NH 4 ) 2 SO 4 was added to stop the reaction (27). Samples were filtered through 0.45-m nitrocellulose Millipore filters, washed with 4 ml of the (NH 4 ) 2 SO 4 solution, and then placed into a counting vial containing 1.5 ml of 2% aqueous SDS and mixed in a Vortex shaker. 10 ml of Beckman Readysafe aqueous scintillant was added, and the vials were capped and shaken again before counting. For measurement of total [ 3 H]cGMP binding, 2 ml of the (NH 4 ) 2 SO 4 solution was added to the reaction mixture at time 0 (prior to addition of excess unlabeled cGMP), filtered, and counted as above. Nonspecific [ 3 H]cGMP binding, which was used as blank, was determined by incubation of the enzyme in the presence of 20 M [ 3 H]cGMP and 100-fold excess unlabeled cGMP. Calculation of cGMP dissociation rates from the cyclic nucleotide-binding sites was based on that used for the corresponding cGMP-binding sites in I␣ cGK (15).
Phosphopeptide Generation-4 nmol of I␤ cGK that had been autophosphorylated for 2 h in the presence of 40 (56).
Phosphopeptide Sequence Determination-Sequence analyses on the I␤ cGK phosphopeptides resolved by C-8 reverse-phase chromatography were performed as described by Francis et al. (56).
Determination of Protein Concentration-Protein concentration was determined as described by Francis et al. (56).
Materials-All materials were obtained as described by Francis et al. (56).

RESULTS
Autophosphorylation of I␤ cGK-Purified bovine I␤ cGK was autophosphorylated in the presence of cGMP or cAMP as described under "Experimental Procedures." A subsequent 500fold dilution eliminated carry-over effects of cGMP or cAMP in protein kinase assays as determined by incubations in which ATP was omitted (see "Discussion"). Autophosphorylation in the presence of cGMP or cAMP increased basal kinase activity 3-or 4-fold, respectively ( Fig. 1). There was no change in maximum catalytic activity determined in the presence of saturating concentrations of cyclic nucleotides; the increase in basal activity required both cyclic nucleotide and Mg/ATP in the 2-h incubation; and the enzyme was autophosphorylated but not proteolyzed during incubation, as determined by SDS-PAGE analysis (not shown).
cAMP-stimulated autophosphorylation exhibited at least two rate components and reached ϳ1.0 mol [ 32 P i ]/mol of I␤ cGK subunit in a 90-min incubation. Initial phosphate incorporation was rapid (ϳ0.4 mol [ 32 P i ]/mol in 5 min), followed by a slower incorporation of an additional ϳ0.6 mol [ 32 P i ]/mol over the next 85 min as seen in Fig. 2. While the pattern of autophosphorylation was consistent, total phosphate incorporation varied (from 1 to 2 mol/mol) with different preparations of I␤ cGK and [␥-32 P]ATP. An increase in basal kinase activity occurred during cAMP-stimulated autophosphorylation (Fig. 2) and appeared to correlate best with the slow rate component of phosphate incorporation. cGMP-stimulated autophosphorylation exhibited the same pattern, although with a slightly slower rate of phosphate incorporation (not shown). The same phosphorylation pattern was observed in the absence of cyclic nucleotide (not shown), but the reaction was at least 4-fold slower in the absence than in the presence of cyclic nucleotide (29).
Phosphorylation was not due to contamination by cAMP-dependent protein kinase (cAK) since addition of catalytic subunit of cAK, in the presence or absence of cGMP or cAMP, did not alter total phosphate incorporation into I␤ cGK (not shown). It has been reported that I␣ cGK is not phosphorylated by cAK (30).
The increased basal kinase activity elicited by autophosphorylation was stable for at least 90 min when I␤ cGK was stored at 30°C prior to the kinase assay (not shown) and was linear for at least 2 h at 30°C during the kinase assay (Fig. 3). This activation was not reversed when Ͼ99.5% of cyclic nucleotides were removed by gel filtration chromatography (not shown). These data indicated that the increased basal activity was due to a stable modification of cGK.
Autophosphorylation of I␣ cGK in the presence of cAMP was reported to increase basal kinase activity and decrease the K a for cAMP by ϳ6-fold (21). Our results confirmed the increase in basal activity of I␣ cGK by cAMP and indicated that cGMPstimulated autophosphorylation also increased the basal activity of I␣ cGK (not shown). In addition to the effect on basal activity, cAMP-or cGMP-stimulated autophosphorylation of I␤ cGK decreased the K a for either cyclic nucleotide ϳ2-fold (0.83-0.48 M cGMP, 7.6 -2.8 M cAMP) (Fig. 4). Fig. 4 illustrates the impact of autophosphorylation on the sensitivity of I␤ cGK to physiological concentrations of cGMP or cAMP (shown in dashed boxes). Porcine coronary artery smooth muscle is rich in cGK, and intracellular concentrations of cGMP (0.09 -0.27 M) and cAMP (0.42-1.26 M) have been determined in that tissue (31,32). As seen in Fig. 4, kinase activity of non-autophosphorylated I␤ cGK varied from 16 to 30% of maximum at the lower and upper end of the physiological range of cGMP concentration (0.09 -0.27 M), respectively, an increase of 1.9-fold. In contrast, the activity of autophosphorylated I␤ cGK was 56% of maximum at 0.27 M cGMP and represented a 3.5-fold increase over that of non-autophosphorylated enzyme at 0.09 M cGMP (56 versus 16%). Similarly, the activity of non-autophosphorylated I␤ cGK at 1.26 M cAMP, which is at the upper end of the physiological range, was 1.5-fold greater than that exhibited at 0.42 M cAMP. The activity of autophosphorylated I␤ cGK was 3.3-fold greater at 1.26 M cAMP compared with that of non-autophosphorylated enzyme at 0.42 M cAMP. These data demonstrated that the phosphorylation state of I␤ cGK, as well as the level of cyclic nucleotides, is a significant determinant of I␤ cGK activity.
Autophosphorylation of I␤ cGK was also studied at concentrations of enzyme as well as concentrations of cyclic nucleotides that approximate intracellular levels in porcine coronary artery smooth muscle (31,32). Treatment of this tissue with either 10 M sodium nitroprusside or 100 M isoproterenol causes a ϳ3-fold elevation of cGMP and cAMP, respectively (see above) (32 H]cGMP in the presence of 100-fold excess unlabeled cGMP was measured as described under "Experimental Procedures." Dissociation was rapid under these conditions (ϳ80% dissociated in 4 s), and no difference was observed between autophosphorylated and non-autophosphorylated I␤ cGK (not shown). The presence of histone VIII-S (0.5 mg/ml, final concentration) in the incubation significantly decreased the rate of dissociation, but no difference in dissociation rates between autophosphorylated and non-autophosphorylated I␤ cGK could be detected (Fig. 5).
[ 3 H]cGMP dissociated from the fast and slow sites of I␤ cGK with a t 1 ⁄2 of 3.7 s and 8.1 min, respectively (not shown). Under the same conditions, dissociation rates from the fast and slow sites of I␣ cGK yielded a t 1 ⁄2 of 6.2 s and 18.2 min, respectively (not shown). These rates are in accordance with published values (15,22).
Comparison of Effect of cGMP on Autophosphorylation of I␣ and I␤ cGK-The stoichiometry of phosphate incorporation into purified I␣ cGK has been reported to be higher in the presence of cAMP than in the presence of cGMP (19). This was also generally observed with I␤ cGK in the present studies and was probably due to the slower rate of phosphate incorporation in the presence of cGMP compared with that in the presence of cAMP. However, there was a difference in the pattern of cGMPstimulated autophosphorylation of I␣ and I␤ cGK (Fig. 6). With I␣ cGK, autophosphorylation was stimulated (33) by cGMP concentrations that saturate only one cyclic nucleotide-binding site (15). At cGMP concentrations sufficient to partially saturate the second cyclic nucleotide-binding site, autophosphorylation was inhibited (33) (Fig. 6). Cyclic nucleotide analog studies (not shown) suggested that the extent of autophosphorylation of I␣ cGK was inversely related to the affinity of the cyclic nucleotide for both binding sites. This supports the original observation that autophosphorylation was stimulated by cyclic nucleotide bound in the slow site alone, or perhaps in the fast site alone, but was inhibited by saturation of both cyclic nucleotide-binding sites (33). However, defining the contribution of the binding sites based on the use of cyclic nucleotide analogs is problematic since those that are relatively fast siteselective are still likely to bind to the high affinity site at lower concentrations than is required for their association with the fast site. Therefore, partial saturation of only the fast site without prior saturation of the slow site is unlikely. Thus, fast site-selective cyclic nucleotide analogs may stimulate autophosphorylation by binding to the slow site. This introduces the possibility that cyclic nucleotide binding to the slow site stimulates autophosphorylation of I␣ cGK, whereas binding to the fast site inhibits the reaction.
Under the same conditions, autophosphorylation of I␤ cGK was not inhibited by saturating cGMP concentrations (Fig. 6). This lack of inhibition was apparently not due to the lower cGMP affinity of I␤ cGK relative to I␣ cGK, since autophosphorylation of I␤ cGK was also stimulated at low concentrations and not inhibited at high concentrations of the cyclic nucleotide analog 8-Br-␤-phenyl-1,N 2 -etheno-cGMP (not shown), which has a ϳ28-fold higher affinity for I␤ cGK than does cGMP (34).
Identification of Phosphoacceptor Sites-The two apparent time components of autophosphorylation (Fig. 2) were examined for multiple-site autophosphorylation. I␤ cGK was autophosphorylated in the presence of cAMP and [␥-32 P]ATP for 7 min or 2 h and then proteolyzed with trypsin. The resulting 32 P-peptides were resolved by high performance liquid chromatography C8 reverse phase chromatography (Fig. 7). When I␤ cGK was incubated for 7 min (0.8 mol [ 32 P i ]/mol I␤ cGK subunit), the only significant 32 P-peptide peak eluted at ϳ15% acetonitrile (Fig. 7A), confirming the primary phosphoacceptor site as Ser-63 (56). The basal kinase activity and K a for cAMP of cGK autophosphorylated for 7 min was unchanged compared with that of the non-autophosphorylated cGK (not shown). The profile derived from I␤ cGK autophosphorylated for 2 h (1.8 mol [ 32 P i ]/mol I␤ cGK subunit) in the presence of cAMP also had a major radiolabeled peak that eluted with ϳ15% acetonitrile, as well as three other peaks that eluted with ϳ30% acetonitrile (Fig. 7B). The basal activity and K a for cAMP of cGK autophosphorylated for 2 h was 5-fold greater and 2.6-fold lower, respectively, than that of the non-autophosphorylated enzyme (not shown). All three peaks eluting with ϳ30% acetonitrile were determined to be derived from the same phosphopeptide, 75 RQAIS * AEPTAFDIQDLSHVTLPFYPK . . . , but slight differences in the peptides accounted for the different elution positions. Peak 1 contained the peptide as written above. The amino terminus of peak 2 was Gln-76 rather than Arg-75. Upon sequential Edman degradation of these peptides, the 32 P was released at Ser-79. The amino terminus of peak 3 was blocked; however, tandem mass spectrometric analysis determined that the mass of the peak 3 peptide was consistent with the peak 2 peptide containing one phosphoserine and having Gln-76 cyclized. The slower rate of autophosphorylation correlated with modification of Ser-79 since phosphorylation of Ser-63 was essentially complete within 7 min (Fig. 7).
Studies of the Requirement of Quaternary Structure for Autophosphorylation of I␤ cGK-The order of reaction of I␤ cGK autophosphorylation was examined. The initial rate (Ser-63) of phosphate incorporation into I␤ cGK was constant over a 40fold range of enzyme concentration (13-513 nM). A Van't Hoff plot would yield a slope of 1 for an intramolecular reaction or a slope of 2 for an intermolecular reaction (35). As seen in Fig. 8, autophosphorylation of Ser-63 was first-order (slope ϭ 0.98) and, therefore, independent of enzyme concentration, which represents an intramolecular reaction. The reaction was carried out for only 2 min due to the lability of I␤ cGK at low concentrations. Because of this lability, examination of the slow-rate component of autophosphorylation (Ser-79) in native cGK proved difficult. This problem was circumvented by utilizing a monomeric form of I␤ cGK. During extended storage, purified dimeric I␤ cGK is proteolyzed carboxyl-terminal to the dimerization domain, thus producing cGMP-dependent monomer (22,36). An aged I␤ cGK preparation was determined to contain approximately 20% full-length and 80% monomeric enzyme. SDS-PAGE revealed that proteolysis generated two different monomers with molecular masses of ϳ70 and ϳ68 kDa. Sequential Edman degradation of the monomers determined that the amino terminus of the 70-kDa monomer was Ser-65 and that of the 68-kDa monomer was Ser-79. Thus, neither contained Ser-63, the primary phosphoacceptor. This preparation was used to examine the order of reaction for Ser-79 modification.
Aged I␤ cGK, containing a mixture of full-length and monomeric enzyme, was diluted over a 64-fold range (20 -1300 nM) and autophosphorylated in the presence of cAMP. The reaction was halted at 7 min by adding SDS-reducing buffer and boiling. Samples were subjected to SDS-PAGE to resolve the full-length protein and the two monomers. Monomer bands were excised, and the radioactivity was measured. The 70-kDa I␤ cGK monomer was readily autophosphorylated and was used to calculate the order of reaction for Ser-79 modification. As with phosphorylation of Ser-63 in intact I␤ cGK, the Van't Hoff plot generated from the autophosphorylation rate of Ser-79 indicated a first-order reaction (slope ϭ 1.1) (Fig. 8).
Thus, autophosphorylation of Ser-63 is an intramolecular event within the native dimeric enzyme, but it could not be ascertained whether this is an intrasubunit or intersubunit process. Using the monomeric enzyme, autophosphorylation of Ser-79 has been determined to occur via an intrasubunit process.
Phosphate incorporation into the 68-kDa I␤ cGK monomer was only 15% of that incorporated into the 70-kDa I␤ cGK monomer (not shown). This suggested that Ser-79 is not readily modified without the presence of additional residues amino-terminal to this residue. It also suggested that there are no major phosphoacceptors carboxyl-terminal to Ser-79.
The availability of the 70-kDa I␤ cGK monomer provided an opportunity to examine the effects of Ser-79 autophosphorylation in the absence of Ser-63. The 70-kDa I␤ cGK monomer was separated from both the full-length enzyme and the 68-kDa I␤ cGK monomer by DEAE-Sephacel chromatography as described previously (25). The 70-kDa I␤ cGK monomer was then subjected to autophosphorylation. Basal kinase activity and cAMP K a for the autophosphorylated 70-kDa I␤ cGK monomer was compared with that of the non-autophosphorylated 70-kDa I␤ cGK monomer (Fig. 9). Autophosphorylation increased basal kinase activity from 17 to 52% of maximum and decreased the K a of cAMP 15-fold. These results confirmed that modification of Ser-79 is sufficient to produce the functional effects observed with autophosphorylation of full-length I␤ cGK. In addition, the rate of Ser-79 modification in the absence of Ser-63 was similar (ϳ2-fold greater) to that of the native dimer (not shown). This suggested that in the native dimeric cGK, autophosphorylation of these sites in I␤ cGK is not hierarchical.

DISCUSSION
Kinase activity of I␤ cGK is increased by cGMP or cAMP in three ways. First, cyclic nucleotide association with the binding sites of the enzyme relieves autoinhibition of the catalytic site. Second, cyclic nucleotide-stimulated autophosphorylation of the autoinhibitory domain impedes interaction of the pseudosubstrate site with the catalytic center. Finally, autophosphorylation increases sensitivity of the enzyme to activation by cyclic nucleotides. At subsaturating levels of cyclic nucleotide, these effects in combination would produce a significantly greater stimulation of kinase activity than would occur by either process alone. Thus, cyclic nucleotides cause the enzyme to autophosphorylate and concomitantly autoactivate.
For several reasons, the increased basal activity of both I␣ and I␤ cGK following autophosphorylation cannot be explained by trace amounts of cyclic nucleotides carried over from the autophosphorylation incubation. First, dilution of non-autophosphorylated I␤ cGK that had been incubated in the absence or presence of cGMP or cAMP yields identical activities. Second, the characteristics of cGMP dissociation from autophosphorylated I␣ (19,20,37) or I␤ cGK do not differ from those of the unmodified enzymes. Third, the increased basal activity is For I␤ cGK, the reaction was terminated after 2 min by spotting aliquots onto phosphocellulose papers (Whatman P81), and [ 32 P i ] incorporation was determined (slope ϭ 0.98). For the I␤ cGK monomer, the reaction was terminated after 7 min by addition of SDS-reducing buffer and boiling. 32 P i incorporation was determined by measuring the radioactivity associated with the monomer band isolated from SDS-PAGE gels stained with Coomassie Blue. (Slope ϭ 1.1.) not reversed upon removal of substrates and ligands by gel filtration chromatography. Thus, the increase in basal kinase activity and the increase in affinity for cyclic nucleotides result from a stable modification of the enzyme that closely correlates with autophosphorylation.
Autophosphorylation of I␤ cGK under these conditions modifies two residues, Ser-63 and Ser-79. Phosphorylation of Ser-63 is rapid and saturates in ϳ7 min at 30°C but has minimal, if any, effect on kinase activity or cyclic nucleotide sensitivity. Phosphorylation of Ser-79 proceeds at a rate ϳ10fold slower than that for phosphorylation of Ser-63 and correlates with an increased basal kinase activity and decreased K a for cyclic nucleotides. It is not certain if autophosphorylation of Ser-79 in dimeric I␤ cGK is sufficient to increase basal activity or if modification of both Ser-63 and Ser-79 is required. However, modification of Ser-79 in a monomeric I␤ cGK lacking Ser-63 is sufficient to elicit these changes in activity. It is possible that Ser-63 alters an unknown activity of the enzyme, such as stability to proteolysis, intracellular localization, or substrate recognition.
Van't Hoff analysis reveals that the autophosphorylation of I␤ cGK is intramolecular. This is confirmed by the recent observation in this laboratory that the rate of I␤ cGK autophosphorylation remains constant in the presence or absence of a 5-fold excess of pre-autophosphorylated I␤ cGK (Francis et al. (56)).
I␣ and I␤ cGK contain substrate sites (autophosphorylation sites) and putative pseudosubstrate sites within their autoinhibitory domains. The sequence of a pseudosubstrate site resembles that of a substrate and is consequently thought to confer inhibition of catalysis by competition with the active site of the enzyme (17, 38 -40). However, by definition the pseudosubstrate site lacks a phosphoacceptor site. The pseudosubstrate sites of I␤ cGK (56) and of I␣ cGK (11,12,41) are 74 KRQA * ISAEP and 59 RAQG * ISAEP, respectively. The residue preceding the asterisk (*) is thought to occupy the phosphate acceptor site (P o ) because it corresponds to RRXS * X of the consensus phosphorylation sequence in type II regulatory subunit of cAK. The major sites of autophosphorylation in I␣ cGK (Ser-50, Thr-58, Ser-72, Thr-84) (42) and I␤ cGK (Ser-63 and Ser-79) flank the pseudosubstrate site. In the case of I␤ cGK, Ser-79 is located in the P ϩ2 position. Phosphorylation of this residue is proposed to cause the activity change in I␤ cGK, but phosphorylation of the homologous residue in I␣ cGK (Ser-64) presumably cannot explain the activity change in this enzyme since this site is reported to be only a minor phosphoacceptor (18).
The mechanism by which autophosphorylation or cyclic nucleotide binding activates cGK is unknown; however, these processes could be quite similar. Because of the presence of autophosphorylation sites in the vicinity of the autoinhibitory domain, activation is likely due to displacement of the pseudosubstrate site from the catalytic site. In order for phospho-transfer to either exogenous substrates or autophosphorylation sites to occur, the tight interaction between the catalytic domain and the autoinhibitory domain must be disrupted. Such a disruption could be brought about by a conformational change due to electrostatic repulsion between the incorporated phosphate, which carries a negative charge, and a negative charge(s) in the catalytic domain. The structure of the cAK catalytic subunit (C-subunit) co-crystallized with a peptide derived from the heat-stable protein kinase inhibitor (43) confirms results generated from biochemical studies (38,44,45) indicating that electrostatic interactions are crucial to C-subunit-substrate association. Two Arg residues in the consensus substrate recognition sequence (RRXS * X) are found in the protein kinase inhibitor as well as in the autoinhibitory region of the cAK regulatory subunits and are thought to interact with four Glu residues of the C-subunit (43). Incorporation of an electronegative phosphate group near the two Arg residues in the consensus recognition sequence might repel the Glu residues in the active site of the C-subunit. Indeed, autophosphorylation of the Ser residue in the autoinhibitory site (RRVS * V) of the type II regulatory subunit reduces its affinity for Csubunit by 10-fold (46), and phosphorylation of the Ser-99 residue in the pseudosubstrate site ( 94 RRGAIS * ) of type I regulatory subunit by cGK greatly reduces its affinity for C-subunit (47,48). The position of Ser-99 in type I regulatory subunit is homologous to that of the phosphoacceptor Ser-79 in I␤ cGK (P 2ϩ ). The catalytic domain of type I cGK shares 41% identity to the cAK C-subunit (determined by using Ref. 49) and includes the four Glu residues discussed above. Therefore, it is feasible that electrostatic repulsion between the phosphate group of the autophosphorylated Ser-79 (I␤ cGK) and the Glu residues of the catalytic domain could elicit a conformational change that activates the enzyme. However, evidence indicates that the catalytic and regulatory domains of cAK and cGK interact at positions outside the pseudosubstrate site (50 -52, 56). Thus, autophosphorylation must also overcome these additional sites of interaction. Using synthetic peptides, it was recently demonstrated that phosphorylation alters the charge of basic residues that flank the phosphoacceptor (53). This effect, in combination with repulsion of the catalytic domain or other components of the regulatory domain, and a reduction in the hydrophobicity of the pseudosubstrate region could provide a potent force to induce conformational changes resulting in activation of the enzyme.
Purified bovine I␣ cGK contains endogenous covalent phosphate, indicating that phosphorylation of this isoform occurs in vivo (19), although the sites of in vivo modification have not been determined. The following additional observations suggest that autophosphorylation of both I␣ and I␤ cGK has physiological relevance and plays a role in activation. 1) Autoactivation by autophosphorylation of sites in the autoinhibitory domain occurs with both isoforms even though the amino acid sequences of the autophosphorylated regions for the two isozymes are very different. 2) Autophosphorylation reduces the K a for cAMP in both isozymes, thus facilitating cross- FIG. 9. Basal kinase activity and cAMP K a of non-autophosphorylated and autophosphorylated 70-kDa I␤ cGK monomer. 2.0 M 70-kDa I␤ cGK monomer was autophosphorylated in the presence of 40 M cAMP as described under "Experimental Procedures." The reaction mixtures were then diluted 500-fold in cold buffer A containing 0.5 mg/ml bovine serum albumin and assayed for protein kinase activity in presence of the indicated cAMP concentrations. activation by cAMP. 3) Autoactivation of I␤ cGK by autophosphorylation occurs at presumed physiological concentrations of enzyme and cyclic nucleotides. 4) Autophosphorylation of I␣ (54) or I␤ cGK (24) is unaffected by the presence of substrate, suggesting that autophosphorylation/autoactivation occurs in preference to phosphorylation of exogenous substrates. Furthermore, in vitro autophosphorylation of I␤ cGK proceeds as rapidly as phosphorylation of exogenous substrates such as the heptapeptide RLRSRAE and the in vivo substrate vimentin (24). The in vitro autophosphorylation rate of the secondary site is relatively slow, but cGK could be maintained in a partially autophosphorylated state in vivo due to the presence of significant steady-state levels of cyclic nucleotides even in the presence of low levels of stimulants in most tissues (55). Again, this would depend on the relative rates of autophosphorylation and dephosphorylation.
In vitro autophosphorylation of I␣ or I␤ cGK is an intramolecular reaction that activates these enzymes and increases the sensitivities to activation by cGMP or cAMP. When cGMP is elevated, autophosphorylation may account for as much of the increased cGK activity as that contributed by cGMP binding to the enzyme. Finally, autophosphorylation could protract the effects of extracellular signals that induce transient elevations of cyclic nucleotide levels.