Title: Phosphorylation of Isolated Human Phosphodiesterase-5 Regulatory Domain Induces an Apparent Conformational Change and Increases cGMP Binding Affinity
Abstract: Substrate binding to the phosphodiesterase-5 (PDE5) catalytic site increases cGMP binding to the regulatory domain (R domain). The latter promotes PDE5 phosphorylation by cyclic nucleotide-dependent protein kinases, which activates catalysis, enhances allosteric cGMP binding, and causes PDE5A1 to apparently elongate. A human PDE5A1 R domain fragment (Val46-Glu539) containing the phosphorylation site (Ser102) and allosteric cGMP-binding sites was studied. The rate, cGMP dependence, and stoichiometry of phosphorylation of the PDE5 R domain by the catalytic subunit of cAMP-dependent protein kinase are comparable with that of the holoenzyme. Migration in native polyacrylamide gels suggests that either cGMP binding or phosphorylation produces distinct conformers of the R domain. Phosphorylation of the R domain increases affinity for cGMP ∼10-fold (K D values 97.8 ± 17 and 10.0 ± 0.5 nm for unphospho- and phospho-R domains, respectively). [3H]cGMP dissociates from the phospho-R domain with a single rate (t12 = 339 ± 30 min) compared with the biphasic pattern of the unphospho-R domain (t12 = 39.0 ± 4.8 and 265 ± 28 min, for the fast and slow components, respectively). Thus, cGMP-directed regulation of PDE5 phosphorylation and the resulting increase in cGMP binding affinity occur largely within the R domain. Conformational change(s) elicited by phosphorylation of the R domain within the PDE5 holoenzyme may also cause or participate in stimulating catalysis. Substrate binding to the phosphodiesterase-5 (PDE5) catalytic site increases cGMP binding to the regulatory domain (R domain). The latter promotes PDE5 phosphorylation by cyclic nucleotide-dependent protein kinases, which activates catalysis, enhances allosteric cGMP binding, and causes PDE5A1 to apparently elongate. A human PDE5A1 R domain fragment (Val46-Glu539) containing the phosphorylation site (Ser102) and allosteric cGMP-binding sites was studied. The rate, cGMP dependence, and stoichiometry of phosphorylation of the PDE5 R domain by the catalytic subunit of cAMP-dependent protein kinase are comparable with that of the holoenzyme. Migration in native polyacrylamide gels suggests that either cGMP binding or phosphorylation produces distinct conformers of the R domain. Phosphorylation of the R domain increases affinity for cGMP ∼10-fold (K D values 97.8 ± 17 and 10.0 ± 0.5 nm for unphospho- and phospho-R domains, respectively). [3H]cGMP dissociates from the phospho-R domain with a single rate (t12 = 339 ± 30 min) compared with the biphasic pattern of the unphospho-R domain (t12 = 39.0 ± 4.8 and 265 ± 28 min, for the fast and slow components, respectively). Thus, cGMP-directed regulation of PDE5 phosphorylation and the resulting increase in cGMP binding affinity occur largely within the R domain. Conformational change(s) elicited by phosphorylation of the R domain within the PDE5 holoenzyme may also cause or participate in stimulating catalysis. The superfamily of cyclic nucleotide phosphodiesterases (PDEs) 1The abbreviations used are: PDE, phosphodiesterase; R domain, regulatory domain; GST, glutathioneS-transferase; PKG, cGMP-dependent protein kinase; PKA, cAMP-dependent protein kinase; IBMX, 3-isobutyl-1-methylxanthine; MOPS, 4-morpholinepropanesulfonic acid is comprised of eleven known families of PDEs that vary in substrate specificity, regulatory properties, and tissue distribution (1Soderling S.H. Beavo J.A. Curr. Opin. Cell Biol. 2000; 12: 174-179Google Scholar, 2Francis S.H. Turko I.V. Corbin J.D. Nucleic Acids Res. Mol. Biol. 2000; 65: 1-52Google Scholar). The regulatory domains of five of the known PDE families (PDEs 2, 5, 6, 10, and 11) contain either one or two sequences known as GAF domains, and these are homologous among the five families. In three of these families (PDEs 2, 5, and 6), at least one of these sequences in each monomer forms an allosteric site for cGMP binding (3Charbonneau H. Beavo J. Houslay M.D. Cyclic Nucleotide Phosphodiesterases: Structure, Regulation and Drug Action. John Wiley & Sons, Inc., New York1990: 267-296Google Scholar, 4Charbonneau H. Prusti R.K. LeTrong H. Sonnenburg W.K. Mullaney P.J. Walsh K.A. Beavo J.A. Proc. Natl. Acad. Sci. U. S. A. 1990; 87: 288-292Google Scholar, 5McAllister-Lucas L.M. Sonnenburg W.K. Kadlecek A. Seger D. LeTrong H. Colbran J.L. Thomas M.K. Walsh K.A. Francis S.H. Corbin J.D. Beavo J.A. J. Biol. Chem. 1993; 268: 22863-22873Google Scholar, 6McAllister-Lucas L.M. Haik T.L. Colbran J.L. Sonnenburg W.K. Seger D. Turko I.V. Beavo J.A. Francis S.H. Corbin J.D. J. Biol. Chem. 1995; 270: 30671-30679Google Scholar, 7Aravind L. Ponting C.P. Trends Biochem. Sci. 1997; 22: 458-459Google Scholar, 8Liu L. Underwood T. Li H. Pamukcu R. Thompson W.J. Cell. Signal. 2002; 14: 45-51Google Scholar, 9Yamazaki A. Sen I. Bitensky M.W. Casnellie J.E. Greengard P. J. Biol. Chem. 1980; 255: 11619-11624Google Scholar, 10Stroop S.D. Beavo J.A. J. Biol. Chem. 1991; 266: 23802-23809Google Scholar). These latter PDEs belong to a subgroup of PDE families known as cGMP binding PDEs. In PDE2, cGMP binding to the cGMP binding allosteric sites increases catalytic activity of the enzyme toward cAMP and cGMP by an unknown mechanism (11Beavo J.A. Hardman J.G. Sutherland E.W. J. Biol. Chem. 1971; 246: 3841-3846Google Scholar, 12Manganiello V.C. Tanaka T. Murashima S. Beavo J. Houslay M.D. Cyclic Nucleotide Phosphodiesterases: Stucture, Regulation and Drug Action. John Wiley & Sons, Inc., New York1990: 61-85Google Scholar). In PDE5, the effect of allosteric cGMP binding is still not clear, but cGMP binding to the PDE5 R domain controls phosphorylation of a specific serine that is near the amino terminus of the R domain (13Thomas M.K. Francis S.H. Corbin J.D. J. Biol. Chem. 1990; 265: 14971-14978Google Scholar, 14Turko I.V. Francis S.H. Corbin J.D. Biochem. J. 1998; 329: 505-510Google Scholar). Phosphorylation at this serine activates both PDE5 catalytic and allosteric cGMP-binding activities (15Corbin J.D. Turko I.V. Beasley A. Francis S.H. Eur. J. Biochem. 2000; 267: 2760-2767Google Scholar). Phosphorylation of PDE5 occurs in intact vascular smooth muscle cells after cGMP elevation, and this phosphorylation is associated with increased catalytic activity of this enzyme (16Wyatt T.A. Naftilan A.J. Francis S.H. Corbin J.D. Am. J. Physiol. 1998; 274: 448-455Google Scholar, 17Mullershausen F. Russwurm M. Thompson W.J. Liu L. Koesling D. Friebe A. J. Cell Biol. 2001; 155: 271-278Google Scholar, 18Rybalkin S.D. Rybalkina I.G. Feil R. Hofmann F. Beavo J.A. J. Biol. Chem. 2002; 277: 3310-3317Google Scholar). Mechanism(s) by which cGMP binding and phosphorylation of the R domain of PDE5 alter enzyme properties are not understood. PDE5 exists in at least two conformational states. In absence of cGMP, the enzyme assumes an apparently more compact structure (19Francis S.H. Chu D.M. Thomas M.K. Beasley A. Grimes K. Busch J.L. Turko I.V. Haik T.L. Corbin J.D. Methods. 1998; 14: 81-92Google Scholar). Occupation of the PDE5 catalytic site by either cGMP or inhibitors such as 3-isobuty-1-methylxanthine, zaprinast, or sildenafil stimulates cGMP binding to the allosteric sites, and when both catalytic and binding sites are occupied PDE5 undergoes an apparent elongation (19Francis S.H. Chu D.M. Thomas M.K. Beasley A. Grimes K. Busch J.L. Turko I.V. Haik T.L. Corbin J.D. Methods. 1998; 14: 81-92Google Scholar). However, little is known about specific structural events that occur in PDEs upon interaction with regulatory agents or after post-translational modification by phosphorylation. Regulation of PDE5 involves cGMP interaction with the R domain, which is required for efficient post-translational modification through phosphorylation (13Thomas M.K. Francis S.H. Corbin J.D. J. Biol. Chem. 1990; 265: 14971-14978Google Scholar, 15Corbin J.D. Turko I.V. Beasley A. Francis S.H. Eur. J. Biochem. 2000; 267: 2760-2767Google Scholar). Models of allosteric regulation of many enzymes include modulation of the equilibrium of distribution of proteins between two conformers,i.e. a relaxed (R) state that is active and has higher affinity for substrate and a taut (T) state that is less active and has lower affinity for substrate (20Monod J. Wyman J. Changeaux J.P. J. Mol. Biol. 1965; 12: 88-118Google Scholar, 21Johnson L.N. Barford D. Owen D.J. Noble M.E.M. Garman E.F. Adv. Second Messenger Phosphoprotein Res. 1997; 31: 11-28Google Scholar). Phosphorylase is an example of this model wherein either binding of an allosteric activator, 5′AMP, or phosphorylation by phosphorylase kinase alters equilibrium between the two states to produce a more active enzyme (21Johnson L.N. Barford D. Owen D.J. Noble M.E.M. Garman E.F. Adv. Second Messenger Phosphoprotein Res. 1997; 31: 11-28Google Scholar). PDE5 may conform to that model, because cGMP binding to allosteric binding sites induces an apparent conformational change in PDE5, perhaps converting it from a less active state to a more active state (15Corbin J.D. Turko I.V. Beasley A. Francis S.H. Eur. J. Biochem. 2000; 267: 2760-2767Google Scholar, 19Francis S.H. Chu D.M. Thomas M.K. Beasley A. Grimes K. Busch J.L. Turko I.V. Haik T.L. Corbin J.D. Methods. 1998; 14: 81-92Google Scholar). However, phosphorylation of Ser92 in bovine PDE5 and activation of PDE5 occur efficiently only when cGMP is elevated, and the enzyme is already in the cGMP-bound (R) conformation. Therefore, it is possible that this phosphorylation effectively sequesters more PDE5 in the R state or induces yet a third conformation (R*) that differs from the fraction of unphospho-PDE5 fluctuating between cGMP-bound (R) and free (T) states (T ↔ R ↔ R*). If so, it would represent an additional overlay on the allosteric regulation of PDE5. Protein phosphatase action would be required to dephosphorylate phospho-PDE5 for this portion of cellular PDE5 to once again re-enter the equilibrium between cGMP-bound (R) and cGMP-free (T) PDE5. To examine functional and conformational changes associated with phosphorylation and activation of PDE5 further, we have studied intact PDE5, as well as the isolated PDE5 R domain, using a recombinant protein containing the phosphorylation site and the allosteric cGMP-binding sites. These studies reveal that regulation of PDE5 phosphorylation by cGMP and enhanced cGMP binding induced by PDE5 phosphorylation is associated strongly with changes in conformation and in function within the R domain of the enzyme. A truncated R domain of PDE5 was expressed as a GST fusion protein inEscherichia coli by slight modification of the method reported previously (8Liu L. Underwood T. Li H. Pamukcu R. Thompson W.J. Cell. Signal. 2002; 14: 45-51Google Scholar). The cDNA fragment coding for the regulatory domain of PDE5 was amplified by PCR from human fetal lung cDNA (Clontech) using specific primers 5′-AAAAGAATTCTGTTAGAAAAGCCACCAGAGAAATG-3′ and 5′-AAAACTCGAGCTCTCTTGTTTCTTCCTCTGCTG-3′ (Val46-Glu539 of human PDE5A1). The amplified fragment was subcloned into the EcoRI andXhoI sites of pGEM-5X3, resulting in the expression plasmid coding for GST-fused human PDE5 regulatory domain. GST-fused PDE5 R domain was phosphorylated by PKA primarily at Ser102, but there was also a low level of phosphate incorporation into Thr50 (-Arg-Lys-Ala-Thr50-), which is near the junction with the GST tag. In contrast, native PDE5 is phosphorylated by PKG or PKA catalytic subunit (C subunit) only at Ser92 of the bovine enzyme or Ser102 of human PDE5 (8Liu L. Underwood T. Li H. Pamukcu R. Thompson W.J. Cell. Signal. 2002; 14: 45-51Google Scholar, 13Thomas M.K. Francis S.H. Corbin J.D. J. Biol. Chem. 1990; 265: 14971-14978Google Scholar). Therefore, Thr50 was mutagenized to alanine using QuikChange site-directed mutagenesis kit (Stratagene) and primer sets 5′-GAAAAGCCGCCAGAGAAATGGTCAATG-3′ and 5′-CATTGACCATTTCTGGCGGCTTTTC-3′. The resulting pGEX-PDE5 R domain-T50A, which is called GST-fused PDE5 regulatory domain or PDE5 R domain herein, was then expressed in E. coli (BL21 DE3 stain) by 100 μmisopropyl-1-thio-β-d-galactopyranoside at 22 °C for 18 h. Cells were disrupted in PBS by a sonicator and then centrifuged at 48,000 × g. The supernatant was loaded onto a glutathione-Sepharose 4B affinity column, and the GST-fused PDE5 regulatory domain was eluted with 10 mm reduced glutathione in 50 mm Tris-HCl, pH 8.0. Eluate was stored in the same buffer containing 10% sucrose and 0.15 m NaCl at −70 °C. The protein was incubated with fresh 5 mmdithiothreitol overnight at 4 °C before use. Purity of the protein was assessed using 10% SDS-PAGE followed by Coomassie Blue staining. Characteristics of the effects of phosphorylation on the properties of the original PDE5 R domain construct and of the PDE5 R domain (T50A) were the same. PDE5 R domain was phosphorylated routinely using free C subunit of PKA that had been purified to apparent homogeneity from bovine heart by a method published previously (22Flockhart D.A. Corbin J.D. Marangos P.J. Campbell I.C. Cohen R.M. Brain Receptor Methodologies. Academic Press, Orlando, FL1984: 209-215Google Scholar). For experiments to determine rate and cGMP dependence of PDE5 and PDE5 R domain phosphorylation, the reaction was conducted in a mixture containing isolated PDE5 R domain (1.4 μm) or purified His-tag bovine PDE5 (1.4 μm), purified bovine C subunit (100 nm), 143 μm sildenafil, 20 mm magnesium acetate, 0.1 mm ATP (with addition of trace [32P]ATP), 10 mm potassium phosphate, pH 6.8, 0.1 m NaCl, and in absence or presence of 50 μm cGMP at 30 °C. At varying times, aliquots of the reaction mixture were spotted onto phosphocellulose paper (P81), washed four times in 0.5% phosphoric acid, dried, and counted. Alternatively, aliquots were used for SDS-PAGE analysis as described below and followed by radioautography. For experiments to prepare quantitatively phosphorylated PDE5 R domain, the phosphorylation was conducted as follows. PDE5 R domain (2 nmol, 7.5 μm) was incubated with 200 μm ATP and 2 mm magnesium acetate in the presence of 10 μm[3H]cGMP and 3 μm C subunit. In some instances, [32P]ATP was included in trace amounts to quantitate stoichiometry of phosphorylation. After incubation for 2 h at 30 °C, cIMP at a final concentration of 3.3 mm was added to exchange with cGMP bound to PDE5 R domain and incubated for an additional 1 h. Tubes were placed on ice and incubated for 1 h. Cyclic IMP has ∼10-fold lower affinity for PDE5 allosteric binding sites than does cGMP and therefore can be removed more readily from the protein using gel filtration. Phospho-PDE5 R domain was concentrated using ULTRAFREE-0.5 (Millipore) to reduce concentrations of free cIMP and cGMP. The concentrated sample was loaded onto Sephacryl S200 (0.9 × 50 cm) equilibrated with 10 mm potassium phosphate, pH 6.8, 2 mm EDTA, 5 mm dithiothreitol, 0.15 m NaCl to remove free ATP, [3H]cGMP, and cIMP. Eluted samples were pooled and concentrated to ∼500 μl using an Amicon filtration cell equipped with a PM-30 membrane. Stoichiometry of phosphate incorporation into PDE5 R domain was assessed by [32P] incorporation and by the slowed migration of phospho-PDE5 R domain in native polyacrylamide gel electrophoresis. More than 90% of total protein was phosphorylated, and phosphorylation of the PDE5 R domain (T50A) was exclusively at serine. The PDE5 R domain mutant lacking the phosphorylation site at Ser102 (S102A) was not phosphorylated under these conditions (data not shown). According to radioactivity of [3H]cGMP of the purified phosphorylated PDE5 R domain, the cGMP content of the protein was less than 1% of total cGMP-binding sites. Unphospho-R domain was treated using the same conditions but in the absence of ATP, magnesium acetate, and C subunit. Both the unphospho- and phospho-R domains eluted as sharp peaks from Sephacryl S200 at a position consistent with a dimeric structure of this size. The regulatory domain of PDE5 has been shown previously to account for dimerization of the enzyme (2Francis S.H. Turko I.V. Corbin J.D. Nucleic Acids Res. Mol. Biol. 2000; 65: 1-52Google Scholar, 23Thomas M.K. Francis S.H. Corbin J.D. J. Biol. Chem. 1990; 265: 14964-14970Google Scholar). Cyclic GMP-binding assays were conducted in a reaction mixture that contained 40 mm MOPS, pH 7.5, 0.55 mm EGTA, 15 mm magnesium acetate, 0.17 mg/ml bovine serum albumin, and various concentrations of [3H]cGMP as indicated. The binding reaction was initiated by addition of isolated PDE5 R domain. After incubation for various periods at 30 °C, 1.5 ml of cold KP buffer (10 mmpotassium phosphate, pH 6.8) was added to each sample. Samples were filtered immediately onto pre-moistened Millipore HAWP filters (pore size 0.45 μm), which were then washed with 1.5 ml of cold KP buffer three times. Filters were then dried and counted using non-aqueous scintillant, and radioactivity was then counted. Blanks containing no protein were run for each concentration of [3H]cGMP where cGMP was varied. Otherwise, nonspecific binding was determined by incubation of PDE5 R domain in the presence of 1 mmunlabeled cGMP. Unphospho-PDE5 R domain (16 nm) and phospho-PDE5 R domain (8.3 nm) were incubated for 25 min at 30 °C in a [3H]cGMP-binding mixture containing 1 μm [3H]cGMP (see above) to allow saturation of cGMP-binding sites with [3H]cGMP. Cyclic GMP binding stoichiometry was the same at 1 and 3 μm[3H]cGMP. Reaction mixtures were transferred to 4 °C for 25 min and then aliquots were removed for determination of total bound [3H]cGMP (Bo). A 100-fold excess of unlabeled cGMP was then added, and aliquots were removed at indicated times (t) for determination of [3H]cGMP remaining bound (Bt) using the Millipore filtration technique described above. Results were corrected by subtraction of a parallel control performed with no enzyme and plotted as ln Bt/Bo versus time, where Bt/Bo represents the fraction of [3H]cGMP remaining bound at any time. For SDS-PAGE, aliquots (15 μl) of protein samples were boiled in 10% SDS/2 m2-mercaptoethanol for 4 min, combined with 2 μl of bromphenol blue (1 mg/ml) and subjected to SDS-PAGE (10% gels) as described previously (5McAllister-Lucas L.M. Sonnenburg W.K. Kadlecek A. Seger D. LeTrong H. Colbran J.L. Thomas M.K. Walsh K.A. Francis S.H. Corbin J.D. Beavo J.A. J. Biol. Chem. 1993; 268: 22863-22873Google Scholar). In experiments utilizing [32P]ATP for phosphorylation of PDE5 R domain, the dye front was run into the lower buffer to remove [32P]ATP prior to radioautography. For native gel electrophoresis, two protocols were used. For studies of PDE5 R domain, protein samples were combined with sample buffer and loaded into wells of pre-cast 7.5% Tris-HCl acrylamide gels (Bio-Rad). Electrophoresis used 25 mm Tris, 192 mm glycine running buffer, pH 8.3, and electrophoresis proceeded for 40 min at constant voltage (250 V) at room temperature according to the manufacturer's instructions. In some instances, cGMP (1 mm) was added to the running buffer. For studies of recombinant PDE5 holoenzyme, native gels (9.5%) were prepared daily as described previously (19Francis S.H. Chu D.M. Thomas M.K. Beasley A. Grimes K. Busch J.L. Turko I.V. Haik T.L. Corbin J.D. Methods. 1998; 14: 81-92Google Scholar), and electrophoresis was conducted at constant voltage (50 V) at 4 °C for 24 h. In all instances, bovine serum albumin standard for native gels (Sigma) was run alongside to serve as markers. Proteins were visualized with Coomassie Blue or silver stain. Protein concentration of GST-fused PDE5 R domain was determined by the Bradford method and by amino acid analysis. The value determined by the Bradford assay using bovine serum albumin as standard was in accordance with that determined by amino acid analysis. [γ-32P]ATP was purchased from PerkinElmer Life Sciences. [8-3H]cGMP was fromAmersham Biosciences. Acryliquid 40 and Instabis were purchased from Eastman Kodak Co. Preformed native gels were purchased from Bio-Rad. Low molecular weight SDS-polyacrylamide gel electrophoresis standards were purchased from Amersham Biosciences. cGMP, ATP, magnesium acetate, bovine serum albumin, and IBMX were from Sigma. Our laboratory has reported that both native and recombinant PDE5 holoenzyme can be phosphorylated in vitro by PKG or C subunit of PKA at a specific serine in the regulatory domain of PDE5 (13Thomas M.K. Francis S.H. Corbin J.D. J. Biol. Chem. 1990; 265: 14971-14978Google Scholar, 14Turko I.V. Francis S.H. Corbin J.D. Biochem. J. 1998; 329: 505-510Google Scholar). Phosphorylation of this serine (Ser92 and Ser102 in bovine and human PDE5A1, respectively) is stimulated potently by cGMP binding to the allosteric cGMP binding sites in the R domain, i.e. a substrate-modulated phosphorylation. Phosphorylation of PDE5 holoenzyme has been shown recently to stimulate catalytic activity and to increase cGMP binding affinity at the allosteric sites (15Corbin J.D. Turko I.V. Beasley A. Francis S.H. Eur. J. Biochem. 2000; 267: 2760-2767Google Scholar). Phosphorylation of PDE5 holoenzyme is tightly regulated, because cGMP binding to the R domain is minimal unless cGMP, a cGMP analog, or an inhibitor such as IBMX or sildenafil is bound at the catalytic site. Biochemical evidence suggests that exposure of the phosphorylation site in PDE5 is a stepwise process in which substrate/inhibitor binds to the catalytic site and causes a conformational change that increases cGMP binding to the R domain (13Thomas M.K. Francis S.H. Corbin J.D. J. Biol. Chem. 1990; 265: 14971-14978Google Scholar, 23Thomas M.K. Francis S.H. Corbin J.D. J. Biol. Chem. 1990; 265: 14964-14970Google Scholar, 24Francis S.H. Lincoln T.M. Corbin J.D. J. Biol. Chem. 1980; 255: 620-626Google Scholar). Upon cGMP occupation of the allosteric binding sites, the phosphorylation site is exposed and can then be phosphorylated by either PKG or C subunit of PKA. PKG is a ten times better catalyst than PKA in vitro for this phosphorylation and would be activated under cellular conditions when cGMP is elevated. However, it cannot be ruled out that under some conditions PKA could also act as a catalyst in vivo. Whether the molecular events associated with substrate-directed phosphorylation of PDE5 and the phosphorylation-induced increase in cGMP-binding affinity reside largely in the PDE5 R domain or require the entire molecule is the focus of this report. Native gel electrophoresis of recombinant bovine PDE5A1 commonly reveals a doublet of protein bands in which the more rapidly migrating species almost always predominates. For the studies shown in Fig. 1 (control lane), we used a PDE5 preparation that was comprised largely of the more rapidly migrating band to present a clear demonstration of the electrophoretic gel shift that occurs in response to ligand binding or phosphorylation. Prior incubation of the enzyme with the PDE5 inhibitor zaprinast, which interacts with the catalytic site of the enzyme, does not alter the migration of the protein band significantly. However, pre-incubation of PDE5 with zaprinast, in combination with cGMP, causes a major redistribution of the PDE5 protein bands so that a more slowly migrating form is prominently visible (Fig. 1). This result is consistent with the interpretation that cGMP binding to allosteric sites in PDE5 causes an apparent elongation of the protein and demonstrates that simultaneous occupation of both catalytic and allosteric cGMP binding sites induces a significant change in overall structure of the protein. Addition of high concentrations of cGMP alone can also cause a quantitative electrophoretic gel shift of PDE5 into the more slowly migrating band (data not shown). This effect is consistent with simultaneous cGMP occupation of both the catalytic and allosteric cGMP binding sites. However, ongoing hydrolysis of cGMP by the PDE5 catalytic site during the preincubation and the rapid migration of cGMP toward the anode progressively reduces the concentration of the ligand and its effect. Introduction of additional negative charges into the PDE5 R domain either by binding cGMP or by covalent incorporation of phosphate would be predicted to accelerate mobility because of an overall increase in electronegativity. It was established earlier that phosphorylation of PDE5 by PKG or C subunit of PKA requires occupation of the allosteric cGMP binding sites of PDE5 (13Thomas M.K. Francis S.H. Corbin J.D. J. Biol. Chem. 1990; 265: 14971-14978Google Scholar). Results in Fig. 1 also demonstrate for the first time that when PDE5 is phosphorylated under these conditions,32P incorporation occurs exclusively in the more slowly migrating PDE5 band and does not redistribute into the lower band. Phosphorylated PDE5 appears to co-migrate with the more slowly migrating band observed in the unphospho-PDE5 in the presence of cGMP and zaprinast (lane 2). Whether these two forms differ could not be determined. This question has now been pursued further using the isolated PDE5 R domain to determine whether this apparent elongation upon phosphorylation occurs in absence of the catalytic domain. A GST fusion protein, PDE5 R domain, containing glutathione S-transferase fused to a segment of the human PDE5A1 R domain extending from Val46 through Glu539 (8Liu L. Underwood T. Li H. Pamukcu R. Thompson W.J. Cell. Signal. 2002; 14: 45-51Google Scholar) with a predicted monomer molecular weight of ∼85,000, has been used to examine effects of cGMP binding to or phosphorylation of the PDE5 R domain. The amino acid sequence of PDE5 R domain includes the specific serine (Ser102) in human PDE5A1 that is homologous to Ser92 in bovine PDE5 and is phosphorylated by either PKG or C subunit of PKA (8Liu L. Underwood T. Li H. Pamukcu R. Thompson W.J. Cell. Signal. 2002; 14: 45-51Google Scholar, 13Thomas M.K. Francis S.H. Corbin J.D. J. Biol. Chem. 1990; 265: 14971-14978Google Scholar). The sequence terminates in the region predicted to conclude the b GAF domain (5McAllister-Lucas L.M. Sonnenburg W.K. Kadlecek A. Seger D. LeTrong H. Colbran J.L. Thomas M.K. Walsh K.A. Francis S.H. Corbin J.D. Beavo J.A. J. Biol. Chem. 1993; 268: 22863-22873Google Scholar). The protein used for the present studies is highly purified (>95%) as assessed using SDS-PAGE followed by Coomassie or silver staining. Kinetic characteristics of [3H]cGMP binding by the isolated R domain mimic closely those of PDE5 holoenzyme determined in the presence of a PDE inhibitor, such as sildenafil or IBMX, that occupies the catalytic site and relieves autoinhibition of cGMP binding at the allosteric sites (data not shown). The K D values for cGMP binding are 97.8 ± 17 and 106 ± 8 nm, mean ± S.E. for the isolated R domain and PDE5 holoenzyme, respectively. These similarities suggest that the isolated R domain is in the active, i.e. the deinhibited state, and affirm the usefulness of this protein to study PDE5 function. As shown in Fig. 2,A and C, cGMP stimulates phosphorylation of the isolated R domain by the C subunit of PKA. This is presumed to occur by cGMP binding to PDE5 R domain, because cGMP does not bind to the C subunit. Phosphorylation of the isolated R domain is time-dependent and cGMP-stimulated under conditions used for these studies (Fig. 2, A andB). This pattern was obtained using four separate purified preparations of the PDE5 R domain. The rate and cGMP dependence of phosphorylation of the isolated R domain by C subunit are comparable with that of PDE5 holoenzyme (Fig. 2, B and C). The stoichiometry of phosphorylation of PDE5 R domain by PKA is 0.91 ± 0.15 mol of phosphate per monomer of PDE5 R domain when a higher concentration of C subunit is used (see “Experimental Procedures”) and radiolabel is associated with serine phosphate. The cGMP stimulation of phosphorylation of the R domain demonstrates for the first time a significant communication between two functional subdomains within the isolated R domain of PDE5, i.e. the allosteric cGMP-binding subdomain and the phosphorylation subdomain. The energy provided by cGMP binding to the allosteric cGMP binding sites alone is sufficient to effect exposure of the phosphorylation site. Although occupation of the catalytic site of intact PDE5 by cGMP, cGMP analogs, or PDE inhibitors is established as an important initial step in the mechanism of phosphorylation of PDE5 holoenzyme, its main effect is apparently to enhance cGMP binding to the allosteric sites by relieving autoinhibition of this function. However, the presence of PDE5 catalytic domain is not required to effect conformational changes that occur in the R domain to bring about exposure of the phosphorylation site and its availability for modification by the kinases. PKG can also be used as the catalyst for the phosphorylation reaction (13Thomas M.K. Francis S.H. Corbin J.D. J. Biol. Chem. 1990; 265: 14971-14978Google Scholar). With PKG as catalyst, 8-bromo-cGMP (4 μm) was included to activate PKGIα fully, but cGMP (10 μm) was still required to achieve significant phosphorylation of PDE5 R domain (data not shown), because cGMP is required to occupy the cGMP binding sites on PDE5 R domain. PKG is an ∼10-fold better catalyst than PKA for phosphorylation of PDE5 or isolated PDE5 R domain, but it has not been used for studies in this report for three reasons (13Thomas M.K. Francis S.H. Corbin J.D. J. Biol. Chem. 1990; 265: 14971-14978Google Scholar). First, when PKG is used to phosphorylate the isolated PDE5 R domain, interpretation of the results is complicated by the fact that cGMP, or a cGMP analog, is required for full PKG catalytic activity, and cGMP is also required for the substrate-directed stimulation of phosphorylation of PDE5. Second, under these conditions, PKG undergoes autophosphorylation, and after electrophoresis, the PKG and PDE5 R domain protein bands are closely juxtaposed. Third, the presence of PKG in the reaction mixture complicates subsequent interpretation of cGMP-binding studies, because both the PKG and the PDE5 R domain bind cGMP, and because of similarity in size, they are difficult to separate using conventional techniques. Therefore, the free C subunit of PKA has been used for the studies described herein. PKA C subunit phosphorylates the same site in PDE5 as does PKG, and its activity is cyclic nucleotide-independent. Therefore, the effects produced by addition of cGMP can be determined directly to be substrate-directed. To assess whether phosphorylation in the isolated PDE5 R domain results in an apparent elongation of the protein as occurs in PDE5 holoenzyme, the mobility of unphospho- and phospho-R domains on native polyacrylamide gels was examined. Unphospho- and phospho-R domains were prepared using non-radiolabeled ATP according to the protocol for quantitative phosphorylation as described under “Experimental Procedures.” Following incubation to generate unphospho- and phospho-R domains, aliquots of the respective proteins were analyzed by native PAGE (Bio-Rad) in the absence and presence of cGMP (Fig.3). In Fig. 3 (lanes 2–5), unphospho- and phospho-R domains were pre-incubated at 4 °C overnight either in absence or presence of 20 mm cGMP prior to electrophoresis for 40 min at room temperature at constant voltage (250 V). A high concentration of cGMP is used in the preincubation, because cGMP, as a small negatively charged molecule, migrates rapidly toward the anode and is removed quickly from the environs of the protein. At lower concentrations of cGMP, the mobility shift in the protein diminishes progressively (data not shown). Bovine serum albumin standards for native gels were run alongside (lanes 1 and8) and are shown as controls for relative migrations of the various forms of PDE5 R domain. In Fig. 3, lane 2, two species of the PDE5 R domain are visible, suggesting that the protein can exist in two conformational states, but the basis of the distinction between these two forms is not known. The more rapidly migrating species is by far the major form in most of our preparations. Phosphorylation of the isolated R domain shifts the protein almost quantitatively into a more slowly migrating species (lane 3) that clearly differs from either band seen in lane 1 containing the unphospho-R domain. This suggests that phosphorylation of the isolated R domain induces a conformational change that is distinct from either species observed in the unphospho-R domain preparation. To better determine the relationship of the cGMP-bound form of the isolated PDE5 R domain to the protein bands observed in phospho-PDE5 R domain, the same unphospho- and phospho-forms of the R domain were pre-incubated with 20 mm cGMP prior to native gel electrophoresis. Because cGMP is diluted when sample is combined with loading buffer, and cGMP in the sample migrates rapidly toward the anode, very high concentrations of cGMP were used in an attempt to maximize the likelihood that cGMP would remain bound to PDE5 R domain during electrophoresis. As seen in lane 4, preincubation of the unphospho-R domain with cGMP causes a redistribution of protein between the two protein bands observed in lane 2. Cyclic GMP binding to the unphospho-R- domain causes a major shift of protein into a more slowly migrating band. This shift is consistent with the interpretation that cGMP binding to R domain of PDE5 holoenzyme induces a significant conformational change in this region to expose the phosphorylatable serine to PKG or PKA action. Furthermore, the combined effect of cGMP binding and phosphorylation (lane 5) clearly differs from the effect elicited by cGMP binding alone (lane 4). Whether migration of the phospho-PDE5 R domain in absence of cGMP (lane 3) differs from the migration of the phospho-PDE5 R domain in presence of cGMP (lane 5) could not be ascertained with confidence. The results could be explained by the phosphorylation-induced increase in cGMP binding affinity of the PDE5 R domain. That is, the phospho-R domain might retain cGMP with greater efficiency throughout electrophoresis. To examine this possibility, the same unphospho- and phospho-PDE5 R domain samples were preincubated with 20 mmcGMP prior to loading into the native gel wells. Electrophoresis was conducted using the standard Tris-glycine buffer containing 1 mm cGMP (lanes 6–8). Because cGMP is negatively charged, cGMP in the running buffer migrates progressively toward the anode throughout the 40-min electrophoresis, thereby maintaining a high cGMP concentration in the microenvironment of the migrating R domain. Unphospho-PDE5 R domain under conditions designed to saturate cGMP binding (lane 6) migrates as a single band that clearly differs from the cGMP-bound phospho-R domain. Therefore, the cGMP-bound form of the phospho-PDE5 R domain is conformationally distinct from that of the cGMP-bound form of unphospho-PDE5 R domain (comparelanes 4 and 5 or lanes 6 and7). This conformational difference is likely to relate to functional differences between these two forms of the protein. When the catalytic site of PDE5 holoenzyme is not occupied by a cyclic nucleotide analog such as IBMX or sildenafil, cGMP binding to the enzyme is minimal. However, the isolated R domain of PDE5 has cGMP-binding characteristics similar to that of the PDE5 holoenzyme whose catalytic site is occupied by cyclic nucleotide analogs such as IBMX or sildenafil. Considering the similarity of theK D values of the two proteins (106 ± 9 for PDE5 holoenzyme, and 97.8 ± 17 nm, mean ± S.E. for the isolated R domain), this suggests that the isolated R domain is in the active form for cGMP binding and is similar kinetically to the holoenzyme when inhibitors or cGMP are bound to the catalytic site. Phosphorylation of the PDE5 holoenzyme causes increased cGMP-binding affinity at its allosteric sites. To determine the molecular basis for the increase in affinity of allosteric cGMP binding following phosphorylation more precisely, the isolated PDE5 R domain was stoichiometrically phosphorylated as described under “Experimental Procedures.” As seen in Fig. 4 A, the phospho-R domain was shifted quantitatively on native gel electrophoresis. As shown in Fig. 4, phosphorylation of the isolated PDE5 R domain improves cGMP-binding affinity ∼10-fold with no effect on total cGMP binding (Fig.4 B). Stoichiometry of cGMP-binding to unphospho- and phospho-PDE5 R domain is 0.50 ± 0.03 and 0.69 ± 0.02 mol/mol, respectively (n = 14). TheK d value for cGMP binding to the unphospho-PDE5 R domain is 97.8 ± 1 nm compared with 10.0 ± 0.5 nm, mean ± S.E. for the phospho-PDE5 R domain. These values agree well with those reported previously by this laboratory (15Corbin J.D. Turko I.V. Beasley A. Francis S.H. Eur. J. Biochem. 2000; 267: 2760-2767Google Scholar) for the effect of phosphorylation on cGMP-binding affinity of PDE5 holoenzyme. For PDE5 holoenzyme, phosphorylation increases the allosteric cGMP binding affinity ∼4-fold. Because cGMP binding to the allosteric cGMP-binding sites of the isolated PDE5 R domain or PDE5 holoenzyme is required to expose the serine for phosphorylation by PKA or PKG, it would be predicted by the principle of reciprocity that phosphorylation should affect cGMP binding (26Weber G. Adv. Protein Chem. 1975; 29: 1-83Google Scholar). The results are consistent with that prediction. The similar effects of phosphorylation to increase cGMP-binding affinity in both the PDE5 holoenzyme and its isolated R domain suggest that the interactions that provide for enhanced function of these sites reside solely within this segment of human PDE5A1 sequence, i.e. Val46through Glu539. Allosteric cGMP binding to PDE5 and to isolated PDE5 R domain has been studied previously by measuring the dissociation of [3H]cGMP from the enzyme in the presence of saturating levels of unlabeled cGMP (6McAllister-Lucas L.M. Haik T.L. Colbran J.L. Sonnenburg W.K. Seger D. Turko I.V. Beavo J.A. Francis S.H. Corbin J.D. J. Biol. Chem. 1995; 270: 30671-30679Google Scholar, 23Thomas M.K. Francis S.H. Corbin J.D. J. Biol. Chem. 1990; 265: 14964-14970Google Scholar). Both the PDE5 holoenzyme and the isolated PDE5 R domain exhibit curvilinear dissociation exchange kinetics consistent with the presence of two cGMP-binding sites with different affinities. The effect of phosphorylation of the PDE5 R domain to increase cGMP binding affinity was investigated further by measuring the rate of cGMP dissociation from the phospho- and unphospho-PDE5 R domains (Fig. 5). Unphospho- and phospho-PDE5 R domains have similar stoichiometries of cGMP binding at zero time, i.e. 0.50 and 0.69 mol of cGMP/mol PDE5 R domain, respectively, and this is similar to the stoichiometry of cGMP binding per PDE5 monomer in the intact holoenzyme (≤1 mol/mol). The same stoichiometry of cGMP binding by the isolated PDE5 R domain is achieved using either 1 or 3 μm[3H]cGMP. The unphospho-PDE5 R domain shows the typical curvilinear dissociation pattern for [3H]cGMP, and the t12values compare with those reported previously for the recombinant and native PDE5 (Fig. 5). Data for unphospho-R domain were analyzed using GraphPad Prism and Origin graphing programs using a biexponential decay model. Results indicated that the lower affinity and higher affinity cGMP binding sites in unphospho-R domain account for 52.4 ± 4.9 and 47.8 ± 4.8 of the cGMP binding sites, respectively. Thet12 value for the lower affinity site was determined to be 39.0 ± 4.8 min compared with at12 of 265 ± 28 min for the higher affinity site; correlation coefficient = 0.99. In contrast, the dissociation of [3H]cGMP from the phospho-PDE5 R domain occurs with a single rate component (t12 = 339 ± 30 min, S.E. correlation coefficient = 0.94) that is significantly slower than dissociation of [3H]cGMP from the unphospho-R domain. Thus, phosphorylation converts the curvilinear pattern of cGMP dissociation from the unphospho-PDE5-R domain into a single high affinity kinetic component. This suggests, but does not prove, that the sites are identical structurally. Phosphorylation may relieve kinetic asymmetry between two sites. The functional asymmetry in cGMP binding to allosteric sites in PDE5 could be produced in several ways. First, there could be two entirely separate cGMP binding sites with different kinetic properties,e.g. cGMP could bind to sites in GAF a domain and in GAF b domain. Current evidence argues against, but does not disprove, this possibility. Second, a single type of allosteric cGMP-binding site e.g. the two GAF a sites, with identical primary structure could have inherently different kinetic properties because of structural constraints imposed by contacts within the dimeric PDE5 R-domain. Third, cGMP saturation of the allosteric binding sites in the unphospho-R domain of PDE5 could impose conformational changes that induce kinetic differences on sites that have identical primary structures, e.g. the two GAFa domains. However, it is clear that the effect of phosphorylation relieves this kinetic asymmetry by increasing the affinity of the fast site of cGMP binding, and this effect is likely to be relevant physiologically. Results presented herein demonstrate that the isolated PDE5 R domain can exist in at least three states depending on the state of phosphorylation of a specific serine in this region and the extent of cGMP binding to the allosteric site(s) in the R domain. Phosphorylation of the isolated PDE5 R domain is stimulated strongly by cGMP occupation of the allosteric cGMP binding sites, and the resulting phosphorylation induces a distinct conformational change that is associated in turn with a 10-fold increase in binding affinity for cGMP. In addition, phosphorylation converts the curvilinear pattern of [3H]cGMP dissociation of the unphospho-PDE5 R domain to a single high affinity component in the phospho-PDE5 R domain. Cyclic GMP binding to the isolated R domain causes an apparent elongation, and phosphorylation of the R domain produces an even more pronounced effect as reflected by significantly slower mobility in native PAGE. This communication between the phosphorylation subdomain and the allosteric cGMP-binding subdomain occurs within the context of the PDE5 R domain alone and does not require the presence of the PDE5 catalytic domain. However, these characteristics of the PDE5 R domain are associated with and may provide for activation of PDE5 by phosphorylation. Results of this study provide important insights into molecular processes that contribute to functional changes induced by phosphorylation of selected domains within the chimeric structure of PDE5. We thank Dr. Charles Cobb, Dept. of Molecular Physiology and Biophysics, Vanderbilt University for assistance in analyzing kinetic data.