Probing Domain Functions of Chimeric PDE6α′/PDE5 cGMP-Phosphodiesterase*

Chimeric cGMP phosphodiesterases (PDEs) have been constructed using components of the cGMP-binding PDE (PDE5) and cone photoreceptor phosphodiesterase (PDE6α′) in order to study structure and function of the photoreceptor enzyme. A fully functional chimeric PDE6α′/PDE5 enzyme containing the PDE6α′ noncatalytic cGMP-binding sites, and the PDE5 catalytic domain has been efficiently expressed in the baculovirus/High Five cell system. The catalytic properties of this chimera were practically indistinguishable from those of PDE5, whereas the noncatalytic cGMP binding was similar to that of native purified PDE6α′. The inhibitory γ subunit of PDE6 (Pγ) enhanced the affinity of cGMP binding at noncatalytic sites of native PDE6α′ by ∼6-fold. The polycationic region of Pγ, Pγ-24–45, was mainly responsible for this effect, while the inhibitory domain of Pγ, Pγ-63–87, was ineffective. On the contrary, Pγ failed to inhibit catalytic activity of the chimeric PDE6α′/PDE5 or to modulate its noncatalytic cGMP binding. Substitutions of Ala residues for the conserved Asn, Asn193 or Asn402, in the two N(K/R)XD-like motifs of the chimeric PDE noncatalytic cGMP-binding sites, each led to a loss of the noncatalytic cGMP binding. Our data suggest that both putative noncatalytic sites of PDE6α′ are important for binding of cGMP, and that the two binding sites are coupled. Furthermore, mutation Asn402 → Ala resulted in an approximately 10-fold increase of theK m value for cGMP, indicating that occupation of the noncatalytic cGMP- binding sites of PDE6α′ may regulate catalytic properties of the enzyme.

Photoreceptor phosphodiesterases (PDEs) 1 serve as effector enzymes in the G protein-mediated visual transduction cascade (1)(2)(3). During transduction of the visual signal in vertebrate photoreceptor rod and cone cells, the activated G protein (transducin) ␣ subunit stimulates PDE catalytic activity by relieving the inhibitory constraint imposed by two identical inhibitory P␥ subunits. A recently adopted classification of cyclic nucleotide PDEs recognizes seven different families based on primary sequence and regulation (4). PDEs within each of the families have 60% or more homology while similarities between different families are 40% or less. According to this nomenclature, photoreceptor rod and cone PDEs comprise the PDE6 family (4). Rod photoreceptor PDE is composed of two large homologous catalytic ␣ and ␤ subunits of nearly identical size (molecular masses of 99.2 and 98.3 kDa) and two copies of an inhibitory ␥ subunit (molecular mass 9.7 kDa) (5)(6)(7)(8). Cone PDE is composed of two identical ␣Ј subunits (molecular masses of 98.7 kDa) (9,10), which share Ͼ60% homology with PDE6␣ and PDE6␤ (11). An inhibitory cone P␥ subunit that is highly homologous to rod P␥ and specific for a subset of cone photoreceptors has been identified (12). Recently, a rod-specific ␦ subunit has been cloned and shown to solubilize the membrane-bound PDE (13). The role of the PDE ␦ subunit in phototransduction is not well understood.
The photoreceptor PDEs contain two noncatalytic cGMPbinding sites located N-terminally to the conserved PDE catalytic domain (11). Originally identified in frog rod PDE (14), noncatalytic cGMP-binding sites were later shown for bovine rod and cone PDEs (9,15). Despite a similar binding stoichiometry (ϳ0.9 mol of cGMP/mol of subunit), the affinities for cGMP binding by bovine cone and rod PDEs are different. Cone PDE binds cGMP with K d ϭ 10 nM, and rod PDE binds cGMP with even higher affinity (9,15). Two other PDE families have similar noncatalytic cGMP-binding sites, cGMP-stimulated PDEs (PDE2) and cGMP binding, cGMP-specific PDEs (PDE5) (16,17). Noncatalytic cGMP binding by photoreceptor PDEs may play an important role in phototransduction and light adaptation. Yamazaki et al. (18) have shown that activated PDE in frog rods has reduced affinity for cGMP at the noncatalytic sites, and addition of P␥ stimulates cGMP binding. Conversely, occupation of the cGMP-binding sites may affect the P␥ affinity for PDE catalytic subunits (19).
An in depth investigation of the structure and function of photoreceptor PDEs in general, and the noncatalytic cGMPbinding sites in particular, requires the development of an effective system for functional expression of PDE6. Functional expression of rod PDE6␣ and PDE6␤ was reported in human kidney cells (20) and in the baculovirus/Sf9 system (21). However, the yields and activity of recombinant PDE6 in both expression systems were very low, perhaps explaining the lack of subsequent systematic studies on recombinant PDE6. A recent study has demonstrated high yield functional expression of PDE5 in the baculovirus system (22). In this study, we report the efficient functional expression of PDE6␣Ј/PDE5 chimeric enzyme and mutational analysis of PDE6␣Ј noncatalytic cGMP-binding sites.

EXPERIMENTAL PROCEDURES
Materials-cGMP was obtained from Boehringer Mannheim.
[H 3 ]cGMP (8.5 Ci/mmol) was a product of Amersham Pharmacia Biotech. All restriction enzymes were purchased from New England Biolabs. 3-(Bromoacetyl)-7-diethylaminocoumarin (BC) was purchased from Molecular Probes, Inc. Trypsin and soybean trypsin inhibitor were from Worthington. Zaprinast and all other reagents were purchased from Sigma.
Generation of Constructs for Expression of PDE6␣Ј, PDE5, and Chimeric PDE6␣Ј/PDE5-A vector containing bovine PDE6␣Ј cDNA (pBlueScriptPDE6␣Ј) was provided by Dr. M. Applebury (Harvard University) (10). Originally, vectors for PDE6␣Ј and four PDE6␣Ј/PDE5 chimeric PDEs were constructed for expression using the BacPAK baculovirus expression system (CLONTECH): Chi1, PDE5-1-509/ PDE6␣Ј441-855; Chi2, PDE5-1-566/PDE6␣Ј522-855; Chi3, PDE6␣Ј1-522/PDE5-567-863; and Chi4, PDE6␣Ј1-440/PDE5-510-863. The pBacPAK9-PDE6␣Ј construct was obtained with ligation of the SmaI/ XhoI fragment of pBlueScript PDE6␣Ј into similarly digested pBacPak9 (CLONTECH) and used as a template for polymerase chain reaction amplification. Appropriate polymerase chain reaction fragments were amplified using primers containing chosen restriction sites. After digestion with restriction enzymes (Chi1, BalI, PacI; Chi2, BspEI, PacI; Chi3, BamHI, BspEI; and Chi4, BamHI, BalI) the fragments were ligated into the pBacPak9PDE5 (22) vector digested with the same enzymes. However, due to persistent irregularities in shipments of the BacPAK6 viral DNA (Bsu36I digest) required for production of a recombinant virus, PDE6␣Ј, PDE5 and the chimeric PDEs were subcloned into the pFastBacHTb vector for expression as His 6 -tagged proteins using the Bac-to-Bac system (Life Technologies, Inc.). cDNA coding for PDE6␣Ј was isolated after digestion of pBlueScriptPDE6␣Ј with SalI and partial digestion with NcoI, and then ligated into the SalI/NcoIdigested vector pFastBacHTb (Life Technologies, Inc.) to produce the pFastBacHTbPDE6␣Ј vector for expression of PDE6␣Ј. The pBlue-ScriptPDE5 plasmid (17,26) was cut with NaeI and SalI, and the PDE5 cDNA fragment lacking codons for the first 3 amino acid residues was subcloned into the pFastBacHTb vector digested with StuI and SalI to produce the pFastBacHTb-PDE5 vector for expression of PDE5. To obtain pFastBacHTbChi1 and pFastBacHTbChi2, the pBacPak9Chi1 and pBacPak9Chi2 were cut with NaeI and SalI and ligated into pFast-BacHTb digested with StuI and XhoI. As a result of this subcloning, Chi1 and Chi2 lacked 3 N-terminal amino acid residues from the PDE5 sequence. The NcoI (partial)/ScaI digested pBacPak9Chi3 and pBacPak9Chi4 were ligated into pFastBacHTb cut with NcoI and StuI to yield pFastBacHTbChi3 and pFastBacHTbChi4, respectively. The sequences of all chimeric proteins were confirmed by automated DNA sequencing at the University of Iowa DNA core facility. Viral amplifications were carried out using Sf21 cells. The High Five insect cells (Invitrogen) were used for protein expression. Generation of the recombinant bacmids, transfection of Sf21 and High Five cells, viral amplifications and infections of insect cells for protein expression were carried out according to the manufacturer's recommendations (Life Technologies, Inc.).
Site-directed Mutagenesis of Chi4 -Site-directed mutagenesis of Chi4 was performed using QuikChange TM kit (Stratagene). Pairs of complementary oligonucleotides coding for substitutions of PDE6␣Ј Asn 193 3 Ala and PDE6␣Ј Asn 402 3 Ala were synthesized. The pFas-tHTbChi4 plasmid was used as a template for polymerase chain reaction using Pfu DNA polymerase. The polymerase chain reaction products were treated with DpnI specific for methylated and hemymethylated DNA and then transformed into E. coli DH5␣. The sequences of mutants were confirmed by automated DNA sequencing.
Purification of Recombinant PDE6␣Ј, PDE5, and Chimeric PDEs-High Five cells were harvested at 60 h after infection, washed with 20 mM Tris-HCl buffer, pH 7.8, containing 50 mM NaCl, and then resuspended in the same buffer containing 10 g/ml pepstatin, 5 g/ml leupeptin, and 0.2 mM phenylmethylsulfonyl fluoride. After 30-s pulse sonication for a total time of 3 min using a 550 Sonic Dismembrator (Fisher) the supernatant (100,000 ϫ g, 45 min) was loaded onto Ni 2ϩ resin (Novagen) equilibrated with 20 mM Tris-HCl buffer, pH 7.8, containing 10 mM imidazole. The resin was washed with the same buffer containing 500 mM NaCl. Proteins were eluted with the buffer containing 250 mM imidazole, and 2 mM ␤-mercaptoethanol was added to the eluate. Purified proteins were dialyzed against 40% glycerol and stored at Ϫ20°C.
cGMP Binding Assay-The cGMP binding assay was performed in a total volume of 100 l of 10 mM sodium phosphate buffer (pH 7.0), containing 2 mM EDTA, 5 mM ␤-mercaptoethanol, 2 pmol of PDE, [H 3 ]cGMP (50,000 cpm), and varying concentrations of unlabeled cGMP. The binding reactions were incubated for 2 h at 4°C and then applied onto 0.45-m MultiScreen HATF filters (Millipore). The filters were washed with 250 l of 10 mM K 2 HPO 4 (pH 6.8) containing 1 mM EDTA, dried, and punched into the scintillation vials using Multiple Punch assembly (Millipore). No cGMP hydrolysis was detected under the conditions of the cGMP binding assay.
Preparation of Anti-PDE6␣Ј Antiserum-The SspI fragment of pGEX-KG (27) containing a glutathione S-transferase protein and polylinker was ligated into pET-11a (Novagen) digested with BamHI and NdeI (treated with mung bean nuclease). The resulting vector was used for subcloning of the NcoI (partial)-XhoI fragment of pBlueScript-PDE6␣Ј, containing the coding region for PDE6␣Ј to produce pET-11aglutathione S-transferase-PDE6␣Ј. The PDE6␣Ј polypeptide was expressed as a glutathione S-transferase fusion protein in E. coli similarly as described in Natochin and Artemyev (28). Glutathione S-transferase-PDE6␣Ј was found mainly in inclusion bodies. The inclusion bodies were dissolved in 6 M urea. The fusion protein was cleaved with thrombin after removal of urea (28). The band corresponding to PDE6␣Ј was cut from the SDS-gels, and the gel slices were used to immunize sheep to obtain a specific polyclonal antiserum against PDE6␣Ј. Immunizations were carried out by Elmira Biologicals (Iowa City).
Miscellaneous-PDE activity was measured using [ 3 H]cGMP as described previously (29). Less than 15% of cGMP was hydrolyzed during these reactions. Fluorescent assays were performed on an F-2000 fluorescence spectrophotometer (Hitachi) as described previously (28,30). Peptides corresponding to P␥ residues 24 -45 and 63-87 were synthesized and purified as described in Natochin and Artemyev (28). Protein concentrations were determined by the method of Bradford (31) using IgG as a standard or using calculated extinction coefficients at 280 nm. SDS-polyacrylamide gel electrophoresis was performed by the method of Laemmli (32) in 10 -12% acrylamide gels. For Western immunoblotting, proteins were transferred to nitrocellulose (0.1 m, Schleicher & Schuell) (33) and analyzed using a sheep anti-PDE6␣Ј antiserum with a 1:3000 dilution and a rabbit anti-PDE5 antiserum with a 1:2000 dilution (26). The antibody-antigen complexes were detected using antigoat/sheep or anti-rabbit IgG conjugated to horseradish peroxidase and ECL reagent (Amersham Pharmacia Biotech). Fitting of the experimental data was performed with nonlinear least squares criteria using GraphPad Prizm Software (30). The K i , K d , and IC 50 values are expressed as mean Ϯ S.E. for three independent measurements.

Expression of PDE6␣Ј and PDE6␣Ј/PDE5 Chimeras in the baculovirus/High Five Cell System-
The native PDE6␣Ј is known to form a catalytic homodimer, so it was thought that the baculovirus-expressed protein might be more stable and active than would the rod PDE6␣ and ␤ co-expressed in the same system (21). A typical total yield for recombinant PDE6␣Ј was ϳ3 mg/100 ml of culture, but only 10 -15% of this was soluble. Soluble PDE6␣Ј purified using affinity chromatography on nickel-nitrilotriacetic acid resin showed no cGMP hydrolytic activity, as well as no detectable binding of cGMP to the noncatalytic sites (not shown). PDE5 is similar to PDE6␣Ј with respect to formation of a homodimer, and the PDE5 has been successfully expressed in the baculovirus/insect cell system (22). In an attempt to achieve functional expression, four chimeric PDEs between PDE6␣Ј and PDE5 have been con-structed (Fig. 1A). Total yields for the chimeric PDEs and PDE5 were comparable, 3-5 mg/100 ml of culture. The solubility of chimeras (20 -40%) was notably better than the solubility of recombinant PDE6␣Ј, but somewhat lower than that of PDE5 (ϳ50%). Western blot analysis showed that polyclonal anti-PDE6␣Ј and anti-PDE5 antisera did not cross-react with PDE5 and PDE6␣Ј, respectively. Both antisera, however, recognized all of the chimeric PDEs (Fig. 1B). All recombinant PDEs were tested for PDE activity and noncatalytic cGMP binding. Chi1, Chi2, and Chi3 were catalytically inactive. PDE5 and Chi4 were fully functional.
Catalytic Properties of PDE6␣Ј/PDE5 Chi4 -PDE5 and Chi4 had a specific activity of 9 -10 mol of cGMP⅐mol PDE Ϫ1 ⅐s Ϫ1 , which was similar to that of native purified PDE5 reported earlier (34). Chi4 hydrolyzed cGMP with a K m value of 1.5 M, which is comparable to the K m value of 3.0 M for recombinant PDE5 (Fig. 2A) and a K m value of 5.6 M for purified native PDE5 (34). Zaprinast, a specific competitive inhibitor of PDE5 and PDE6, inhibited cGMP hydrolysis by Chi4 with an IC 50 value of 0.65 M. A similar IC 50 value (0.75 M) was seen for inhibition of recombinant PDE5 by zaprinast (Fig. 2B) and has been obtained previously for native PDE5 (0.3-0.8 M) (22,34). Both PDE5 and Chi4 migrated on a Sephacryl S-200 HR column as catalytic dimers with apparent molecular mass ϳ180 kDa (not shown). Overall, the catalytic properties of Chi4 were practically indistinguishable from those of recombinant or native PDE5.
A recent study has suggested that PDE5 may have a protein modulator(s) with homology to the polycationic region of P␥, P␥-24 -45 (35). We have tested the possibility that P␥ interacts with PDE5 or Chi4. First, we determined whether recombinant PDE5 or Chi4 preparations might contain bound P␥-like proteins using limited proteolysis with trypsin. Limited proteolysis of native PDE6 with trypsin removes P␥ and activates the enzyme (6). We found that limited trypsinization of PDE5 or Chi4 does not alter their electrophoretic mobility or catalytic properties. In addition, results of cGMP binding experiments and fluorescence binding assays using P␥83BC were not affected as well (not shown), suggesting that there is no P␥-like protein modulator bound to PDE5 or Chi4 expressed in High Five cells. P␥ efficiently inhibited PDE6␣Ј(K i 0.17 nM). However, it had no effect on the catalytic activity of Chi4 (Fig. 3A) and PDE5 (not shown) even at an inhibitor concentration as high as 5 M. The peptide, P␥-24 -45, did not affect the activity of any of the tested enzymes, PDE6␣Ј, PDE5, or Chi4 (not shown).
Recently, we have developed a novel assay that uses a fluorescent signal to report binding of the P␥ C terminus to rod PDE6␣␤ (30). The assay utilizes a P␥ mutant, P␥83Cys labeled with the fluorescent probe BC at the cysteine residue (P␥83BC). HoloPDE6␣Ј was trypsinized (tPDE6␣Ј) to remove intrinsic inhibitory P␥ subunits. Using this fluorescence assay, addition of tPDE6␣Ј to P␥83BC resulted in an ϳ8-fold maximal fluorescence increase of P␥83BC fluorescence with a K d value  of 40 nM (Fig. 3B), which was readily reversible by subsequent addition of excess P␥ (not shown). On the contrary, PDE5 and Chi4 at concentrations up to 100 nM failed to change P␥83BC fluorescence (not shown). These results support lack of P␥ binding to these PDEs.
Comparison of Noncatalytic cGMP Binding by Chi4 and PDE6␣Ј-Analysis of noncatalytic cGMP binding by chimeric PDEs revealed that, under the conditions of the assay, Chi1 and Chi2 did not bind cGMP (not shown). Even though a detectable noncatalytic cGMP binding was observed for Chi3 (not shown), the stoichiometry was very low (Ͻ0.1 mol of cGMP/subunit). Chi3 was not analyzed further considering the low stoichiometry of cGMP binding and a lack of the catalytic activity. Chi4 was fully competent to bind cGMP at the noncatalytic binding sites. The cGMP-binding curve exhibited a single class of binding sites with a K d value of 0.45 M and a binding stoichiometry of 0.9 mol of cGMP/subunit (Fig. 4A). Because Chi4 contains noncatalytic cGMP-binding sites from the bovine PDE6␣Ј, we have compared noncatalytic cGMP binding by Chi4 to that of PDE6␣Ј purified from bovine retinas. Preparation of tPDE6␣Ј bound approximately 0.3 mol of cGMP/ subunit with a K d of 2.1 M. Addition of P␥ to the assay led to a substantial increase in the affinity (K d 0.38 M) and the stoichiometry (0.9 mol/subunit) of cGMP binding (Fig. 4B). The latter increase is likely a result of an incomplete saturation of cGMP binding in the absence of P␥ rather than of an emergence of additional cGMP-binding sites. Affinity of noncatalytic cGMP binding by Chi4 was greater than that for tPDE6␣Ј but lower than that for tPDE6␣Ј reconstituted with P␥. Predictably, P␥ had no effect on cGMP binding of Chi4 (Fig. 4A). We further examined effects of P␥ on cGMP binding of PDE6␣Ј by utilizing two synthetic peptides, P␥-24 -45 and P␥-63-87, corresponding to the major P␥ sites of interaction with PDE6 catalytic subunits (36 -39). P␥-24 -45 potently stimulated cGMP binding by tPDE6␣Ј(K d 0.96 M), whereas the P␥ Cterminal peptide, P␥-63-87 critical for inhibition of PDE activity was ineffective (Fig. 4B). Together, peptides P␥-24 -45 and P␥-63-87 elicited a weak synergistic effect on the noncatalytic cGMP binding (K d 0.71 M) (Fig. 4B). In addition, the noncatalytic cGMP binding to PDE6␣Ј might be affected by the enzyme association with ROS membranes. Chi4 contains the PDE5 C-terminal domain and lacks a consensus sequence for isoprenylation required for attachment of PDE6 to membranes (3,21). Therefore, we have not analyzed effects of ROS membranes on noncatalytic cGMP binding in this study.
Mutational Analysis of the Noncatalytic cGMP-binding Sites in Chi4 -Functional analysis of Chi4 suggested that this enzyme represents an appropriate target for the mutational analysis of the PDE6␣Ј noncatalytic cGMP-binding sites. The conserved Asn residues Asn 193 and Asn 402 from the two putative noncatalytic cGMP-binding sites were replaced by Ala residues. Both Chi4 mutants were expressed as soluble proteins in High Five cells, although their yields were lower (ϳ1-2 mg/100 ml of culture) than those for Chi4. The Chi4 Asn 193 3 Ala and Chi4 Asn 402 3 Ala mutants hydrolyzed cGMP with K m values of 1.5 and 13 M and V max values of 9.6 and 9.2 mol cGMP⅐mol PDE Ϫ1 ⅐s Ϫ1 , respectively (Fig. 5A). Zaprinast competitively in-hibited catalytic activity of Chi4 Asn 193 3 Ala and Chi4 Asn 402 3 Ala with IC 50 values of 0.5 and 1.7 M, respectively (Fig. 5B). Surprisingly, both Chi4 mutants exhibited no noncatalytic cGMP binding (not shown). DISCUSSION Rod and cone cGMP-phosphodiesterases (PDE6) serve as key effector enzymes in the visual transduction cascade of vertebrate photoreceptor cells. Progress in understanding the structure and function of PDE6 has been hindered by difficulties in developing an efficient functional expression system for PDE6 that allows mutational analysis of the enzyme (20,21). PDE6 enzymes share several important characteristics with PDE5, including a common general domain organization, a high homology (45-48% identity) between their catalytic domains, a strong specificity for cGMP relative to cAMP, and sensitivity to the PDE inhibitors zaprinast and dipyridamole (17,40). Expression of fully functional PDE5 has been recently reported using the baculovirus system (22). Our strategy to develop a system for PDE6 expression has been to construct chimeric PDE6␣Ј/PDE5 enzymes with a maximal structural and functional resemblance to native PDE6. This study demonstrates an efficient functional expression of chimeric PDE, Chi4, that contains a catalytic domain from PDE5 and noncatalytic cGMP-binding sites from cone PDE6␣Ј. Thus far, all tested PDE6␣Ј/PDE5 chimeras containing the PDE6␣Ј catalytic domain have been catalytically inactive. Chi4 had catalytic properties such as K m and V max for cGMP hydrolysis as well as IC 50 for inhibition by zaprinast very similar to those of PDE5. Our recent results have demonstrated that the P␥ C terminus binds to the PDE6 catalytic domain and blocks the access of cGMP to the catalytic site (30). The P␥ C terminus binding was also competitive with zaprinast (30). These findings suggest that catalytic domain residues that participate in the binding and hydrolysis of cGMP, and those that bind the P␥ C terminus are in very close proximity in the three-dimensional structure of PDE or may even overlap. This coupled with shared sensitivity of PDE6 and PDE5 to zaprinast have opened up an intriguing possibility that P␥ may inhibit PDE5 or Chi4. However, our data suggest that P␥ has no effect on the activities of PDE5 or Chi4. The PDE6␣Ј sequence that binds the P␥ C terminus corresponds to PDE6␣Ј-749 -761 and is adjacent to the putative NKXD motif for binding cGMP (25,28). We speculate that the absence of the segment homologous to PDE6␣Ј-749 -761 in the PDE5 sequence accounts for the inability of P␥ to inhibit PDE5. This conclusion does not necessarily contradict the existence of putative protein modulators of PDE5 that may have a domain homologous to the polycationic region of P␥ (35).
Analysis of the noncatalytic cGMP-binding sites on Chi4 has indicated that their affinity for cGMP (K d of 0.45 M) is comparable to the affinity of cGMP binding at the noncatalytic sites of native PDE6␣Ј assessed in this study. Earlier estimates have resulted in a lower K d value for the noncatalytic cGMP binding to bovine PDE6␣Ј (9). The comparative analysis is complicated by the fact that the affinity of noncatalytic cGMP binding to PDE6␣Ј depends on whether PDE6␣Ј subunits are complexed with inhibitory P␥ subunits or not. This study is the first to document that the interaction of PDE6␣Ј with P␥ significantly (ϳ6-fold) enhances the affinity of cGMP binding to the PDE6␣Ј noncatalytic sites. Similar effects of P␥ have been previously shown for frog rod PDE6 (18,19). Interestingly, the K d values for cGMP binding to tPDE6␣Ј alone (2.1 M) and in the presence of P␥ (0.38 M) correlate with the low affinity (Ͼ1 M) and high affinity (60 nM) states for nonactivated and transducinactivated frog PDE6, respectively (19). It appears that reciprocal relationships exist between P␥-binding sites and noncatalytic cGMP-binding sites, whereby occupancy of the one class of sites modulates the interaction at the other sites (18,19). A model has been proposed such that, when cGMP levels drop upon light activation of frog rod PDE, cGMP dissociates from the noncatalytic sites leading to a physical removal of P␥ by transducin from PDE6. Inactivation of PDE then occurs faster due to acceleration of GTPase activity in the transducin ␣-P␥ complex (19). Two regions of P␥ participate in the interaction with PDE6, the central polycationic region P␥-24 -45 and the C terminus of P␥ (36 -39). The P␥ C terminus is most important for PDE inhibition. The results presented herein suggest that the polycationic region of P␥ is primarily responsible for the effects of P␥ on noncatalytic cGMP binding by PDE6. Hence, these effects are not coupled to the inhibitory action of P␥ on PDE6 catalytic activity, which is consistent with findings of Yamazaki et al. (41).
Each catalytic subunit of PDE6 as well as PDE5 has two internally homologous repeats (a and b) containing structural elements for cGMP binding (11,17). Repeats a and b participate in formation of two different allosteric cGMP-binding sites in PDE5 (26). However, direct binding studies have shown that only 2 molecules of cGMP are bound to dimeric PDE5 and PDE6 molecules, even though each has four cGMP binding segments (15,16,34). Perhaps each cGMP-binding site is formed by two motifs from both PDE subunits (26). The critical structural element for noncatalytic cGMP binding to PDE5 appears to be a N(K/R)X n D motif (22). A similar motif (NKXD) is critical for GTP binding to G proteins (42). To determine whether one or both a and b repeats participate in binding of cGMP in PDE6, we mutated Asn residues from the two NKXDlike motifs of Chi4, 193 NKVD 196 and 402 NRKD 405 , within the a and b repeats, respectively. Surprisingly, each of the two mutations abolished noncatalytic cGMP binding as measured by the filter binding assay. It is possible that these mutants retain some affinity for cGMP at the noncatalytic sites, but it is too weak to be detected under conditions of the assay. A similar point mutant in the PDE5 a repeat, N276A, had a dramatically reduced but still detectable affinity for cGMP (K d 60 M) (22). Perhaps, the Asn residues in PDE6␣Ј make a greater contribution to the overall affinity of the noncatalytic cGMP binding. Our data suggest that both a and b repeats are important for the noncatalytic cGMP binding by PDE6␣Ј. Furthermore, elimination of one noncatalytic cGMP-binding site leads to loss of cGMP binding at the second site. Alternatively, if repeats a and b, rather than two a or two b repeats, form a single noncatalytic cGMP-binding site, this would explain the loss of cGMP binding by the mutants.
Chi4 may have retained the conformational link between the noncatalytic cGMP-binding sites and the catalytic domain which apparently exists in PDE6␣Ј. An increase in the K m value exhibited by the Chi4 Asn 402 3 Ala mutant indicates that occupancy of the noncatalytic cGMP-binding sites in PDE6 may not only affect the affinity of PDE6 for P␥ subunits but the catalytic properties of PDE6 as well. This would be consistent with earlier observations that noncatalytic cGMP binding controls catalytic activity of another cGMP-binding PDE, PDE2 (43). Although mutations of noncatalytic cGMP-binding sites did not change the enzyme catalytic properties, it is possible that the enzymatic activity of the catalytic domain within the native PDE5 structure is affected by the noncatalytic cGMPbinding sites. Alternatively, noncatalytic cGMP-binding sites may have different functions in different PDE families.