An interface of interaction between photoreceptor cGMP phosphodiesterase catalytic subunits and inhibitory gamma subunits.

Cyclic guanosine 5′-monophosphate (cGMP) phosphodiesterase (PDE) regulates the level of cGMP on transduction of a visual signal in vertebrate photoreceptor cells. Two identical inhibitory PDE γ subunits (Pγs) block catalytic activity of PDE-α and -β subunits (Pαβ) in the dark. The primary regions of Pγ involved in the interaction with Pαβ are a central polycationic region, Pγ-24-45, and a C-terminal region of Pγ. Recently, we have shown that the C-terminal region of Pγ, which is the major Pγ inhibitory domain, blocks PDE activity by binding to the catalytic site of PDE (Artemyev, N. O., Natochin, M., Busman, M., Schey, K. L., and Hamm, H. E. (1996) Proc. Natl. Acad. Sci. U. S. A. 93, 5407-5412). Here, we localize the site on the rod cGMP PDE α subunit that binds to the central polycationic domain of Pγ. This site is located within a region that links a second noncatalytic cGMP binding site with the catalytic domain of PDE. A polypeptide coresponding to this region, Pα-461-553, expressed as a glutathione S-transferase fusion protein in Escherichia coli and isolated after cleavage of the fusion protein with thrombin, blocks inhibition of PDE activity by Pγ. In addition, Pα-461-553 binds to the Pγ-24-45 region (Kd, 7 μM), as measured by a fluorescent increase in a Pγ-24-45Cys peptide labeled with 3-(bromoacetyl)-7-diethylaminocoumarin. The Pα-461-553 region was further characterized by using a set of synthetic peptides. A peptide corresponding to residues 517-541 of Pα (Pα-517-541) effectively suppressed inhibition of PDE activity by Pγ and bound to Pγ-24-45Cys labeled with 3-(bromoacetyl)-7-diethylaminocoumarin (Kd, 22 μM). Pα-517-541 also competes with the activated rod G-protein α-subunit for binding to Pγ labeled with lucifer yellow vinyl sulfone. This suggests that light activation of rod PDE by the G-protein transducin involves competition between transducin α-guanosine 5′-triphosphate and Pα-517-541 for binding to the Pγ-24-45 region. Based on the results, we propose a linear model of interactions between catalytic and inhibitory PDE subunits.

PDE activation leads to the rapid hydrolysis of cytoplasmic cGMP and closure of sodium channels, resulting in hyperpolarization of the rod and cone cells (for review, see Chabre and Deterre (1989) and Stryer (1996)). Rod photoreceptor PDE is composed of two large homologous catalytic ␣ and ␤ subunits (P␣ and P␤) of nearly identical size (M r 99,261 and 98,308, respectively) and two copies of an inhibitory ␥ subunit (P␥, M r 9700) (Ovchinnikov et al., 1986(Ovchinnikov et al., , 1987Deterre et al., 1988;Lipkin et al., 1990). The primary structures of P␣, P␤, and cone-specific P␣Ј subunits  revealed that these PDEs constitute one family, that of PDE6 (Beavo et al., 1994). The photoreceptor PDEs belong to a broader group of cGMPbinding PDEs, which contain two noncatalytic cGMP binding sites located N-terminally to the conserved PDE catalytic domain (Yamazaki et al., 1980;Gillespie and Beavo, 1989;Lipkin et al., 1990;Trong et al., 1990;McAllister-Lucas et al., 1993). Unique features of photoreceptor PDEs are their high k cat /K m parameter and their ability to be inhibited by P␥ and activated by rod and cone G-protein ␣ subunits.
Interactions between PDE catalytic and inhibitory subunits and the mechanism of PDE inhibition have been studied extensively. Two regions of P␥, polycationic region P␥-24 -45 and the C terminus of P␥, have been shown to participate in the interaction with P␣␤ (Lipkin et al., 1988;Brown, 1992;Takemoto et al., 1992). Both of these P␥ domains bind to P␣␤, allowing effective inhibition of PDE activity by the P␥ C terminus. Initial studies indicated that the major sites of P␣ and P␤ interaction with P␥ are different and located in the N-terminal regions (P␣, 16 -30 and 78 -90; P␤, 91-110 and 211-230) in areas with a high level of dissimilarity between catalytic subunits .
More recently, using a cross-linking approach we have demonstrated that the C terminus of P␥ interacts with region P␣-751-763 located within the PDE catalytic domain (Artemyev et al., 1996). In this study, to identify a P␣ region for binding to P␥-24 -45, we have expressed several large domains of P␣ as GST fusion proteins in Escherichia coli. One of these fusion proteins contained a region unique for photoreceptor PDEs that links a second noncatalytic cGMP binding site with the catalytic domain. This protein was used to obtain a polypeptide, P␣-461-553, which blocks inhibition of PDE activity by P␥ and binds to P␥-24 -45. Characterization of the P␣-461-553 region using a set of synthetic peptides revealed that the P␣-517-541 site is primarily responsible for P␣ binding to the polycationic region of P␥.

EXPERIMENTAL PROCEDURES
Materials-cGMP, GTP␥S, and T4 DNA ligase were obtained from Boehringer Mannheim. Restriction enzymes were from New England Biolabs. 3-(Bromoacetyl)-7-diethylaminocoumarin was purchased from Molecular Probes, Inc. Trypsin and soybean trypsin inhibitor were from Worthington. All other reagents were purchased from Sigma.
Preparation of G t␣ GTP␥S, Trypsin-activated PDE, and P␥-Bovine rod outer segment membranes were prepared by the method of Papermaster and Dreyer (1974). The G t␣ GTP␥S was extracted from membranes with GTP␥S and purified using chromatography on a Blue-Sepharose CL-6B column as described by Kleuss et al. (1987). PDE was extracted from rod outer segment membranes as described by Baehr et al. (1979). PDE was purified, and trypsin-activated PDE (taPDE) was prepared as described earlier . The purified proteins were kept in 40% glycerol at Ϫ20°C. The P␥ subunit was expressed in E. coli and purified on a SP-Sepharose fast flow column and on a C-4 HPLC column (Microsorb-MW, Rainin) as described by Skiba et al. (1995).
Construction and Expression of P␣ GST Fusion Proteins-A vector, containing cDNA coding for the full-length P␣ subunit (Ovchinnikov et al., 1987), was kindly provided by Dr. N. Skiba (University of Illinois, Chicago, IL). Fragments of P␣ cDNA corresponding to sequences 2-553, 2-854, 92-553, 92-854, 461-854, and 461-553 were amplified by polymerase chain reaction with appropriate upstream primers containing the XbaI site and downstream primers containing the SalI site. These fragments were digested with XbaI and SalI and ligated into the pGEX-KG expression vector (Guan and Dixon, 1991), which was cut with the same enzymes. The GST-P␣-461-553 cDNA sequence was verified by automated DNA sequencing at the University of Iowa DNA Core Facility using the 5Ј-pGEX sequencing primer (Pharmacia Biotech Inc.). A single silent nucleotide C 3 T substitution was found in AGC codon for Ser-549. All DNA manipulations were performed using standard techniques (Maniatis et al., 1989). Expression of P␣ GST fusion proteins even at a low temperature (25°C) and a low isopropyl-1-thio-␤-D-galactopyranoside concentration (20 M) resulted in the accumulation of recombinant protein in inclusion bodies. We attempted to refold the GST fusion proteins after dissolving the inclusion bodies in 6 M urea. The GST-P␣-461-553 polypeptide was relatively stable in solution after removing urea. Typically, E. coli cells, after a 5-h induction with 0.5 mM isopropyl-1-thio-␤-D-galactopyranoside, were spun down and resuspended in PBS at a ratio of 1:20. The cells were disrupted by ultrasonication, and pellets were collected after the cell lysate had been centrifuged for 30 min at 50,000 ϫ g. Insoluble inclusion bodies were separated from cells and membranes by five consecutive washes in PBS buffer containing 0.5% Triton X-100 and five washes in PBS with 1% Nonidet P-40. After the final wash with PBS, inclusion bodies were dissolved in 6 M urea, and ␤-mercaptoethanol was added to a final concentration of 10 mM. The urea was removed on a Sephadex G-25 column equilibrated with PBS. GST-P␣-461-553 fusion protein was then digested with thrombin (1 NIH unit/2 mg of protein) overnight at room temperature. Proteins were then precipitated with 10% trichloroacetic acid. The pellet was washed three times with 20 mM Tris-HCl (pH 7.4) and dissolved in 20 mM Tris-HCl buffer (pH 7.4) containing 6 M urea and 100 mM NaCl. The P␣-461-553 polypeptide was purified on a Superose 12 (Pharmacia) column equilibrated with the same buffer. Fractions containing P␣-461-553 were desalted on a PD-10 column (Pharmacia), equilibrated with double distilled H 2 O, lyophilized on a SpeedVac concentrator (Savant), and dissolved in buffer A (10 mM HEPES, pH 7.8, 100 mM NaCl, and 1 mM MgSO 4 ).
Peptide Synthesis-Peptides corresponding to P␣ residues 21-31, 467-491, 492-516, and 517-541 (Ovchinnikov et al., 1987) and P␥ residues 24 -45Cys (Ovchinnikov et al., 1986) were synthesized by the solid-phase Merrifield method on an Applied Biosystems automated peptide synthesizer. The extra cysteine was added to the C terminus of the P␥-24 -45 sequence as a site for the introduction of the environmentally sensitive fluorescent probe 3-(bromoacetyl)-7-diethylaminocoumarin (BC). Each peptide was purified by reverse-phase HPLC on a preparative Aquapore Octyl column (25 ϫ 1 cm; Applied Biosystems). The purity and chemical formula of each peptide were confirmed by fast atom bombardment mass spectrometry and analytical reverse-phase HPLC.
Preparation of P␥-24 -45BC and P␥LY-The P␥-24 -45Cys peptide (0.5 mg) was dissolved in 0.3 ml of 10 mM HEPES (pH 7.8). A 2-fold molar excess of 3-(bromoacetyl)-7-diethylaminocoumarin in acetonitrile was added to the peptide solution, and the mixture was incubated for 15 min at room temperature. The P␥-24 -45BC was purified by reversephase HPLC on a Microsorb-MW C-4 column (Rainin). P␥ was labeled at its single cysteine (Cys-68) with lucifer yellow vinyl sulfone (P␥LY) and purified as described by .
Fluorescent Assays-Fluorescent assays were performed on a F-2000 fluorescence spectrophotometer (Hitachi) in 1 ml of buffer A. The fluorescence of P␥-24 -45BC was monitored with excitation at 445 nm and emission at 500 nm. The concentration of P␥-24 -45BC was determined using ⑀ 445 ϭ 53,000. The assay of interaction between P␥LY and G t␣ GTP␥S was carried out essentially as described by .
Analytical Methods-The PDE activity was measured using the proton-evolution assay of Liebman and Evanczuk (1982). The assay was performed at room temperature in 200 l of buffer A. The reaction was initiated by addition of cGMP (3 mM final concentration). The pH was monitored with a pH microelectrode (Microelectrode, Inc.). Protein concentrations were determined by the method of Bradford (1976) using IgG as a standard or using calculated extinction coefficients at 280 nm. Sodium dodecyl sulfate-polyacrylamide gel electrophoresis was performed by the method of Laemmli (1970) in 14% acrylamide gels. Fitting of the experimental data was performed with nonlinear least squares criteria using GraphPad Prizm software.

RESULTS
Expression and Purification of P␣-461-553-P␣ sequences corresponding to residues 2-553, 2-854, 92-553, 92-854, 461-854, and 461-553 were cloned into the pGEX-KG vector and overexpressed as GST fusion proteins in E. coli. The recombinant proteins were found primarily in the inclusion bodies. Solubilization of inclusion bodies in 6 M urea, followed by removal of urea and cleavage of GST fusion proteins with thrombin, resulted in the formation of recombinant P␣ polypeptides with their expected molecular masses. However, all the P␣ polypeptides, except P␣-461-553, were unstable and aggregated into complexes that migrated with high molecular mass on gel-filtration columns. Formation of inclusion bodies with GST-P␣-461-553 led to a very high yield of this protein. After washing, a fraction of the inclusion bodies (ϳ1 g/liter of culture) contained ϳ90% GST-P␣-461-553 (Fig. 1, lane 1). Complete cleavage by thrombin resulted in the formation of three major protein bands at ϳ27, 25, and 13 kDa. (Fig. 1, lane 2). Formation of the 25-kDa band may reflect additional nonspecific cleavage of the proteolytically sensitive poly-Gly loop connecting GST and P␣-461-553. The P␣-461-553 (13 kDa) protein was purified on a Superose 12 gel-filtration column (Fig. 1,  lane 3).
P␣-461-553 Blocks Inhibition of taPDE by P␥-To elucidate whether the P␣-461-553 region contains a site for P␥ interaction, we assayed whether it could compete with P␣␤ by measuring the effects of purified P␣-461-553 on inhibition of taPDE by P␥. Limited proteolysis of PDE with trypsin removes intrinsic P␥ subunits and small farnesylated and geranyl-geranylated C-terminal fragments of P␣ and P␤, respectively (Hurley and Stryer, 1982;Catty and Deterre, 1991), leading to significantly increased activity compared with holoenzyme. Subsequent addition of P␥ reinhibits PDE activity. The activity of taPDE was determined following addition of increasing concentrations of P␥ in the absence or presence of the P␣-461-553 peptide. At 10 M P␣-461-553, the EC 50 of taPDE inhibition by P␥ shifts from 0.22 to 6 nM (Fig. 2).
P␣-461-553 Binds to P␥-24 -45-As was recently demonstrated, the C terminus of P␥ interacts with the catalytic domains of P␣ and, likely, P␤ (Artemyev et al., 1996). To determine whether P␣-461-553 binds to another known region of the P␣␤-P␥ interface, namely that of P␥-24 -45, we developed a fluorescent assay to monitor interaction between P␥-24 -45 and P␣␤. A peptide, P␥-24 -45Cys, was synthesized and labeled with the environmentally sensitive fluorescent probe 3-(bromoacetyl)-7-diethylaminocoumarin. Addition of taPDE to P␥-24 -45BC significantly increased the fluorescence of the probe in a dose-dependent manner (Fig. 3A). The binding curve shows a single class of binding sites with a K d of 37 nM. Addition of excess P␥ reversed the fluorescent increase, indicating that the assay is specific (data not shown). Similarly, when P␣-461-553 was added to P␥-24 -45BC, there was a dramatic increase in probe fluorescence (maximum f/f 0 , 7.2) (Fig. 3B). The affinity of P␥-24 -45BC binding to P␣-461-553 (K d , 7.3 M) is lower than that of P␣-24 -45BC/P␣␤ interaction (K d , 37 nM). This suggests that the polycationic region of P␥ may have an additional binding site on P␣␤. Alternatively, P␣-461-553 could have a tendency to self-association, which would explain some cooperativity seen in its binding to P␥-24 -45BC (Fig. 3B).
Effects of Synthetic Peptides from the P␣-461-553 Region on Inhibition of taPDE Activity by P␥-Peptides corresponding to sequences P␣-467-491, P␣-492-516, and P␣-517-541 were synthesized and tested for their ability to suppress the taPDE inhibition by P␥. In the presence of 1, 3, or 10 M P␣-517-541, the EC 50 of taPDE inhibition by P␥ shifted in a linear fashion from ϳ0.22 to 4, 15, or 41 nM, respectively (Fig. 4A). Incomplete folding of the larger P␣-461-553 polypeptide may explain in part why the shorter P␣-517-541 peptide is more effective in blocking inhibition of taPDE activity by P␥.
The P␣-467-491 peptide was effective only at relatively high concentrations. When present at 100 M, it shifts the EC 50 for taPDE inhibition by P␥ to ϳ5 nM (not shown). This result indicates that P␣-467-491 might be a part of the P␥ binding site, however, the P␣-467-491 sequence contains a patch of acidic residues, ECEEEE (Fig. 6B), which most likely interacts with the positively charged polycationic region of P␥. Therefore, it is not clear whether the effects of P␣-467-491 at high concentrations are functionally significant. Peptides P␣-492-516 and P␣-21-31 (control) had no effect at concentrations up to 100 M. Peptide P␣-21-31 was used in control experiments, as it is a part of the proposed N-terminal P␥ binding domain on P␣ .
Peptide P␣-517-541 Binds to P␥-24 -45-Next we examined binding of peptides P␣-467-491, P␣-492-516, P␣-517-541, and P␣-21-31 to the polycationic region of P␥ using the fluorescent assay. Addition of P␣-517-541 to P␥-24 -45BC caused a significant increase (maximum f/f 0 , 7.4) in the probe fluorescence, similar to the effects of taPDE and P␣-461-553. P␣-517-541 bound to P␥-24 -45BC with a K d of 22 M (Fig. 4B). Unlabeled peptide P␥-24 -45Cys reversed the fluorescent increase of P␥-24 -45BC caused by binding of P␣-517-541 (not shown). Other peptides at concentrations up to 100 M had no effect on P␥24 -45BC fluorescence. The affinity of the P␣-517-541/P␥-24 -45BC interaction is somewhat lower than what might have been expected based on the potency of this peptide to block taPDE inhibition by P␥. P␣-517-541 may have additional interactions with P␥ outside the P␥-24 -45 region, or the fluorescent probe may interfere to some extent with the interaction.
P␣-517-541 Competes with G t␣ GTP␥S for the Interaction with P␥-Evidence suggests that the polycationic region of P␥ contains binding sites for both P␣␤ and G t␣ GTP Brown, 1992 al., 1992). To determine whether P␣-517-541 can compete with G t␣ GTP␥S for binding to P␥-24 -45, we used a fluorescent assay of the interaction between G t␣ GTP␥S and P␥ labeled with lucifer yellow vinyl sulfone . P␣-517-541 had no effect on the basal fluorescence of P␥LY but was able to fully reverse the fluorescent increase of P␥LY that resulted from addition of G t␣ GTP␥S (Fig. 5). A K d of 3 M for the P␣-517-541⅐P␥LY complex was calculated from the competition curve based on the K d of 36 nM for the G t␣ GTP␥S⅐P␥LY complex . Peptides P␣-467-491, P␣-492-516, and P␣-21-31 at concentrations up to 100 M did not compete with G t␣ GTP␥S for the interaction with P␥LY.

DISCUSSION
In rod photoreceptor cells, GTP-bound transducin ␣ subunits interact with a PDE complex of two catalytic ␣␤ subunits and two inhibitory ␥ subunits, leading to displacement of the inhibitory subunits and subsequent PDE activation. The interfaces of critical interactions between P␥ and P␣␤ or P␥ and G t␣ GTP have been analyzed in a number of studies (Lipkin et al., 1988;Artemyev et al., , 1993Brown, 1992;Takemoto et al., 1992;Skiba et al., 1995Skiba et al., , 1996. Although functional sites of P␥ have been elucidated, the interactive surfaces on G t␣ GTP and P␣␤ are not well defined. The central polycationic region of P␥, residues 24 -45, together with the C terminus, is involved in the interaction between P␥ and the PDE catalytic subunits. The primary role of the P␥-24 -45 region is to enhance the affinity of the P␥-P␣␤ interaction. To map the P␥-24 -45 binding site on the P␣ subunit, we have expressed a region linking a second noncatalytic cGMP-binding site with the catalytic domain of P␣ as a GST fusion protein in E. coli. This selection was made for the following reasons; this region: (a) appears to be unique for photoreceptor PDEs; (b) contains several patches of acidic amino acid residues (Fig. 6B), and (c) has unknown function. The choice is consistent with the unique ability of photoreceptor PDE to be inhibited by P␥ and with the polycationic nature of the P␥-24 -45 region. The P␣-461-553 polypeptide, purified after cleavage of the GST fusion protein, blocks inhibition of PDE activity by P␥. In addition, P␣-461-553 binds to fluorescently labeled P␥-24 -45BC and induces a maximal increase in fluorescence, similar to that of taPDE binding with P␥-24 -45BC.
Synthetic peptides were used to localize the P␥-24 -45 binding to residues 517-541 of P␣. Peptide P␣-517-541 effectively suppressed inhibition of PDE activity by P␥ and bound to P␥-24 -45BC in the fluorescent assay. Despite the fact that P␣-517-541 contains 4 acidic and 3 basic residues and P␣-467-491 contains 8 acidic and 3 basic residues, the latter was able to interfere with the P␥ inhibition only at high concentrations and did not show binding to P␥-24 -45BC. This implies that interactions other than electrostatic ones are very important for the P␥-24 -45 binding to P␣␤. It also eliminates a concern one might have regarding the specificity of the initially observed interaction between the overall negatively charged region P␣-461-553 and the polycationic region P␥-24 -45. Nevertheless, P␣-467-491 may represent an additional, weaker interaction site between P␣ and P␥-24 -45. Results of the study by  indicated that a weak interaction between region P␣-453-563 and P␥ may complement major interactions that involve the N-terminal regions of P␣, P␣-16 -30, and P␣-78 -90. Our data suggest that the P␣-517-541 domain is the major site of P␣-P␥ interaction that binds to P␥-24 -45. Different methods used in this study and the study by  could in part account for the different conclusions. Also, peptides P␣-16 -30 and P␣-78 -90 are basic and may at high concentrations bind to P␣␤ and compete with P␥ by a mechanism similar to known nonspecific effects of histones and protamines on PDE (Miki et al., 1975). The conclusion that highly dissimilar regions of P␣ and P␤ interact with P␥  appear to be inconsistent with the evidence that both P␣ and P␤ interact with identical sites on P␥ with high affinity (Wensel and Stryer, 1990;Brown, 1992).
Analysis of the sequences from the regions of P␤ and cone P␣Ј that correspond to P␣-517-541 indicates that P␣ and P␤ are more than 80% identical in this region, whereas P␣Ј has only ϳ40% homology to P␣ or P␤ (Fig. 6A). Most likely, P␤-515-539 binds P␥-24 -45, as is seen with P␣-517-541. The lower homology between rod PDE and cone PDE in this region may explain why rod P␥ inhibits rod PDE more effectively (K i , 80 pM) than cone PDE (K i , 600 pM) (Hamilton et al., 1993). The high level of homology between the P␣ and P␤ sites that bind the P␥ C terminus (Artemyev et al., 1996) and P␥-24 -45 favors a model with similar, although not necessarily identical, affinities for the P␥-P␣ and P␥-P␤ binding sites (Wensel and Stryer, 1990). Fig. 6C shows a linear model for the P␥ interactions with P␣ and P␤ based on our results.
Previous studies have demonstrated that the P␥-24 -45 site is involved in the interaction with G t␣ GTP (Lipkin et al., 1988;Takemoto et al., 1992). Here we show that P␣-517-541 effectively competes with G t␣ GTP␥S for the binding to P␥LY. The question still to be answered is whether this competition is functionally relevant. A very tight binding of P␥ to P␣␤ with a K d Ͻ 50 pM (Wensel and Stryer, 1990) does not allow for P␥ dissociation from P␣␤ during the time of photoresponse (0.2 s). However, noncompetitive binding of G t␣ GTP to the P␥-63-76 region near the major P␥ inhibitory domain P␥-77-87 (Skiba et al., 1995) may induce a conformational change of P␥, resulting in the availability of the P␥-24 -45 region for the interaction with G t␣ GTP. This would increase the affinity of G t␣ GTP-P␥ interaction in the active complex G t␣ GTP-P␥-P␣␤-P␥-G t␣ GTP.
Current classification of cyclic nucleotide phosphodiesterases separates all known mammalian PDEs into seven families (Beavo et al., 1994). All PDEs contain a highly conserved catalytic domain within the C-terminal part of the enzyme. Photoreceptor PDEs have similar general domain organization with two PDE families: cGMP-stimulated PDE and cGMPbinding, cGMP-specific PDE (cGB-PDE). These PDEs have two internally homologous repeats for noncatalytic cGMP binding Lipkin et al., 1990;Trong et al., 1990;McAllister-Lucas et al., 1993. Direct binding studies have shown that only two molecules of cGMP are bound to the dimeric PDE molecule, which has four cGMP binding segments (Gillespie and Beavo, 1989;Thomas et al., 1990;Stroop and Beavo, 1991). Perhaps, each cGMP binding site is formed by two analogous motifs from both catalytic PDE subunits (McAllister-Lucas et al., 1995). A segment of approximately 60 -70 amino acid residues connects the second noncatalytic cGMP binding site with the PDE catalytic domain. The functional role of this domain for different cGMP-binding PDEs is not known. We performed a local alignment search (BLAST) to compare sequences from bovine photoreceptor PDEs corresponding to this "linker" region against protein sequence data bases. This search revealed 65-70% homology between the linker regions of P␣ and P␤ from different species and 45-50% homology between P␣ or P␤ and cone P␣Ј. Interestingly, the cGB-PDE also has a significant level of homology to photoreceptor PDEs within this region. The sequence corresponding to residues 524 -583 of bovine cGB-PDE is 35-37% identical (51-55% similar) to the P␣-481-540 and P␤-479 -538 regions (Fig.  6B). Thus far, no protein modulators of activity of cGB-and cGMP-stimulated PDEs have been identified. The universal Ca 2ϩ -binding protein modulator calmodulin stimulates activity of calmodulin-dependent PDEs through binding to two sites located N-terminally to the catalytic domain . It is tempting to speculate that the linker regions in PDEs with noncatalytic cGMP binding sites could represent potential sites for PDE regulation by other proteins.