Cyclic GMP-specific, High Affinity, Noncatalytic Binding Sites on Light-activated Phosphodiesterase*

T w o classes of high affinity, cGMP-specific binding sites have been found in association with a peripheral membrane protein in rod outer segments. [3H]cGMP and a photoaffinity label, 8-N3-[32P]cIMP, have been used to study these cGMP binding sites. The cGMP binding sites co-migrated with rod outer segment phosphodiesterase (EC 3.1.4.17) upon Bio-Gel A-0.5m column chromatography, sucrose density gradient cen- trifugation, and isoelectric focusing (PI 5.35). Upon sodium dodecyl sulfate-polyacrylamide gel electrophore- sis, the 8-N3-[32P]cIMP-labeled protein also migrated in a position identical with that of purified phosphodies- terase. Scatchard analysis, using purified phosphodiesterase, revealed the presence of two classes of cGMP binding sites with apparent KO values of 0.16 and 0.83 CL” A number of observations indicated that these high affinity, cGMP-specific binding sites are distinct from the phosphodiesterase catalytic site. CAMP, which is a substrate for phosphodiesterase, did not bind to the high affinity cGMP specific sites. Limited tryptic proteolysis of phosphodiesterase resulted in a striking ac- tivation of the catalytic activity and a 96% loss of cGMP binding.

A number of observations indicated that these high affinity, cGMP-specific binding sites are distinct from the phosphodiesterase catalytic site. CAMP, which is a substrate for phosphodiesterase, did not bind to the high affinity cGMP specific sites. Limited tryptic proteolysis of phosphodiesterase resulted in a striking activation of the catalytic activity and a 96% loss of cGMP binding. 1-Methyl-3-isobutylxanthine inhibited phosphodiesterase activity and enhanced the specific binding of cGMP. M$' was necessary for phosphodiesterase activity, but not for high affinity cGMP binding. Finally, phosphodiesterase activity and the cGMP-specific high affinity sites showed different stabilities on storage in phosphate buffer. These specific high affinity cGMP binding sites may be involved in the regulation of phosphodiesterase activity.
It is now known that the interactions of a group of disc membrane proteins support a light and GTP-dependent activation of rod outer segment (ROS) cGMP phosphodiesterase (1-7). Action spectra for ROS' phosphodiesterase reveal that rhodopsin is the photopigment which participates in this lightand GTP-dependent cascade. The activation of phosphodies-* This research was supported by National Institutes of Health Grants AM-20179 and NS-08440. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "adoertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
These enzyme reactions mediate a rapid and striking lightinduced decline in ROS cGMP levels. It has been suggested that this light-mediated decline in cGMP might play a role in photoreceptor transduction or sensitivity control (1,11,12). It is not yet possible to conclude that all of the components of the light-activated reactions have been identified, nor is the mechanism of phosphodiesterase activation by light and GTP fully understood. In view of the apparent importance of cGMP for the function of vertebrate rods, we have examined extracts of ROS for the presence of specific high affinity, cGMP receptor sites. While this work was in progress a report appeared by Hamet and Coquil(l3) demonstrating specific, high affinity cGMP-binding activity in an extract of rat platelets; this cGMP-binding activity co-chromatographed with cGMP phosphodiesterase activity.

EXPERIMENTAL PROCEDURES
Materials-['HICAMP and ['HIcGMP were purchased from New England cIMP was prepared as previously described (14, 15). Bio-Gel A-0.5m was obtained from Bio-Rad. Other reagents were of the highest purity available.
Preparation of EDTA-washed Disc Membranes a n d Crude Phosphodiesterase-EDTA-washed disc membranes and crude phosphodiesterase (EDTA-solubilized disc membrane proteins) were prepared from Rana catesbiana as previously described (8-10). ROS prepared by sucrose flotation were suspended in 1 mM EDTA, 1 mM dithiothreitol (pH 8.1), and passed successively through a No. 21 needle (three times) and a No. 25 needle (three times). This crude disc suspension was kept on ice for 60 min (to permit EDTA solubilization of peripheral membrane proteins including phosphodiesterase) and centrifuged for 60 min in a Beckman SW 27.1 rotor (65,000 X g, 4'C).
The supernatant, referred to as "crude phosphodiesterase," was lyophilized and stored at -80°C. The disc membrane pellet was washed in 30 volumes of 20 mM Tris.HCI (pH 10) containing EDTA (1 mM) and dithiothreitol (1 mM). Finally, the disc membrane pellet was washed (three times) in 30 volumes of 200 m Tris.HCI (pH 7.5) containing MgS04 (20 mM) and dithiothreitol (1 mM). The pellet ("EDTA-washed membranes") was resuspended in the same buffer (2 to 3 mg of protein/ml), quick-frozen with acetone and dry ice, and stored at -80°C.
Assay of Phosphodiesterase Activity-Phosphodiesterase activity was measured as previously described (4, 16) with 1.25 mM cGMP as substrate. The incubation medium (in a final volume of 40 pl) contained 50 mM Tris.HC1 (pH 7.5), 6.5 mM MgS04, 0.1 PM [:H]cGMP (6.22 X lo4 cpm), 1.25 mM unlabeled cGMP, 40pg of protamme sulfate and enzyme protein. For assay of purified phosphodiesterase activity, about 0.05 pg of enzyme was added/tube. After incubation at 30°C for 3 min, the reaction was stopped by boiling for 2 min. Determinations were carried out on quadruplicate samples. Results agreed within 5% and average values are given.
Phosphodiesterase Purification-The lyophilized crude phosphodiesterase prepared from five large R. catesbiana was suspended in a solution containing 10 mM Tris-HCI (pH 7.5). 1 mM dithiothreitol, and 5 mM MgSOd (Buffer A). The magnesium concentration was then supplemented to a final concentration of 20 mM prior to loading this material onto a Bio-Gel A-0.5m column (6.35 X 1.1 cm, bed volume 5 ml) which had been equilibrated with the same buffer. The purpose of adding magnesium was to ensure that the binding of phosphodiesterase to the rhodopsin-containing vesicles which are present in the crude phosphodiesterase preparation was not prevented by the EDTA in this preparation.
Most of the disc membranes and disc membrane fragments did not enter the Bio-Gel A-0.5m column matrix but could be visualized as a narrow zone of bleached photopigment at the very top of the column. The column was washed with Buffer A, and membrane-bound phosphodiesterase was then eluted with a solution containing 10 mM Tris. HCI (pH 7.5), 1 mM dithiothreitol, and 2 mM EDTA (Buffer B).
Fractions (0.7 ml) were collected. Column fractions were concentrated by loading into washed viscane dialysis tubing which was sealed, covered with unimbibed Sephadex G-200, wrapped in aiuminum foil, and kept on ice. When the crude phosphodiesterase was enriched with additional amounts of EDTA-washed disc membranes (not done routinely) and this sample loaded onto a Bio-Gel A-0.5m column, such material yielded a larger amount of phosphodiesterase during the elution step with Buffer B. This was so because a larger amount of phosphodiesterase was initially retained on the column in the form of a membrane complex which was stable during the wash step with Buffer A.  Eighty micrograms of crude enzyme preparation were added to each channel. Band 1 corresponds to the high affinity cGMP binding site.

Determination of cGMP
Bands 2 and 3 could correspond to type I1 and type I CAMPdependent protein kinases, respectively.
incubation medium (final volume, 100 pl) contained 25 mM Tris. HCI (pH 7.5). 0.1 mM EGTA, 0.5 mM l-methyl-3-isobutylxanthine, 1 p~ ['HlcGMP (3.11 X lo5 Fpm), and crude or purified phosphodiesterase. Mixtures were incubated routinely for 30 min on ice, and 80-p1 aliquots were then applied to a Millipore filter (HA, pore size 0.45 pm) and washed with 20 ml of 20 mM potassium phosphate buffer (pH 6.5). Control experiments indicated that the pH optimum for cGMP binding was 7.5, that maximum binding (about 1.38 of the added counts) was reached within 20 min, that binding was stable for a t least 2 h, and that the counts were virtually unaffected (less than 4% displaced) by extensive washing. (The filter itself exhibited less than 0.1% of the cGMP-binding capacity of 5 pg of purified phosphodiesterase.) The filter was dissolved (1 h) in Formula 963 (New England Nuclear) and the amount of cGMP bound was determined by liquid scintillation spectrometry in a Beckman LS-200. Determinations were camed out in duplicate and were accompanied by an additional sample to which a 100-fold excess of unlabeled cGMP was added. Addition of unlabeled cGMP inhibited binding of the labeled material by greater than 95%. All data have been corrected for nonspecific binding.
Photoaffinity Labeling-Photoaffinity labeling experiments with 8-N.,-["2P]cIMP were performed as described previously (15)  saturation a t about 0.09 p~ cGMP. The specificity of this binding for cGMP was demonstrated by the inability of 100 PM unlabeled cAMP to prevent the binding of 1 p~ labeled cGMP. We have also observed that photoactivated incorporation of 8-N3-[:"P]cIMP (1 p~) into a band of M, = 120,000 could be prevented by 1 p~ unlabeled cGMP but not by 1 p~ cAMP (Fig. 2). Incorporation of 8-N:5-[:'2P]cIMP into two additional bands of about M, = 54,000 (Bund 2 ) and 47,000 (Bund 3) can also be observed in Fig. 2. These bands, whose labeling was prevented by cAMP but not cGMP, correspond to the regulatory subunits of type I (Band 3) and type I1 (Band 2) CAMP-dependent protein kinases. Bands 2 and 3 might be attributable partially to contamination by rod inner segments, or non-photoreceptor retinal cells, since their staining intensity was quite variable in different preparations and since there is little cAMP in purified rod outer segments (22).
Evidence that Phosphodiesterase is Responsible for this Specific cGMP Binding-The cGMP-binding activity behaved similarly to ROS phosphodiesterase in that it co-sedimented with the disc membranes in the presence of Mg"+, could be eluted from the disc membranes with EDTA, and failed to emerge from a DEAE-Sephadex G-50 column (data not shown). These results led us to suspect that the cGMP binding sites might actually be located on ROS phosphodiesterase. In order to characterize further the protein responsible 1 min (18). Samples were analyzed by polyacrylamide slab gel electrophoresis on 7.5% acrylamide gels (18) (20) using bovine serum albumin as standard. In some cases (e.g. Bio-Gel column fractions) proteins were estimated by the procedure of Bradford (21).

RESULTS AND DISCUSSION
Demonstration of cGMP Binding-The crude disc suspension was found to contain specific, high affinity, cGMP binding sites. In the presence of Mg'+, this cGMP-binding activity sedimented with the disc membrane pellet upon centrifugation (data not shown). Furthermore, this cGMP-binding activity was extracted from the disc membranes by EDTA, a procedure which has been found to extract phosphodiesterase 4. The preparations were layered on the appropriate gradients, and centrifuged at 94,000 X g for 44 h at 4°C. Fractions of 0.47 ml were collected and phosphodiesterase (t".) activity and ['HIcCMP binding (o"-o) were assayed. Upper panel, EDTA; lower panel, MgS04. for cGMP binding, we undertook its purification. It was found that the high affinity, cGMP binding sites co-purified with, and appeared to be located on, ROS phosphodiesterase. The evidence for this conclusion includes the following: The cGMP-specific binding activity co-eluted with the peak of phosphodiesterase activity upon Bio-Gel A-0.5m column chromatography (Fig. 3). Moreover, when the phosphodiesterase peak fractions (19-21) were pooled and an aliquot examined on an SDS (7.5%) polyacrylamide gel for incorporation of 8-Na-["P]cIMP into protein, a single band of radioactivity was observed, which corresponded to a single band of Coomassie blue staining (Fig. 4).
The protein content (5 pg/lOO pl) of the purified phosphodiesterase activity peak (Fractions 19 to 21 in Fig. 3) corresponds to 20 pmol of phosphodiesterase/lW p1. The observed phosphodiesterase activity (225 pmol of cGMP hydrolyzed/ min/mg of protein) gives a catalytic constant of 54,000 mol of cGMP hydrolyzed/min/mol of phosphodiesterase, in good agreement with that (48,000) calculated previously (4). The observed [JH]cGMP-binding activity was 16 pmo1/100 pl, indicating that about 80% of the phosphodiesterase molecules bind ['HIcGMP.
A comparison of the isoelectric points for phosphodiesterase activity and cGMP-binding activity is shown in Fig. 5. It was found necessary to include the nonionic detergent Srij 35 in the ampholyte system to prevent aggregation of ROS proteins, which otherwise occurred in the region of pH 4.6. Incorporation of 0.5% Brij 35 prevented the formation of such aggregates without inhibiting phosphodiesterase or cGMP-binding activ-ity. Under these conditions, both phosphodiesterase and cGMP-binding activities focused at pH 5. 35.
When crude phosphodiesterase was layered onto a continuous sucrose gradient, catalytic activity sedimented as a peak with a molecular weight of 240,000 (4). When a similar preparation was mixed with EDTA-washed disc membranes, in the presence of 2 mM EDTA, the same pattern for phosphodiesterase sedimentation was observed (Fig. 6). If, in the place of EDTA, 5 nm Mg" was added together with the disc membrane suspension, the phosphodiesterase activity sedimented to the bottom of the gradient with the disc membrane pellet (Fig. 6). Under either condition, the cGMP-binding activity sedimented in the same position as the phosphodiesterase activity. The small amount of phosphodiesterase activity and cGMP-binding activity, found at the top of the sucrose gradient when the sedimentation was carried out in the presence of MgS04, is attributable to phosphodiesterase binding to minute fragments of the disc membrane which did not sediment into the sucrose gradient (data not shown). A Scatchard analysis of cGMP binding to purified phosphodiesterase indicated the presence of two classes of binding sites with Kd values of 0.16 p~ and 0.83 p~, respectively (Fig. 7). ["HIcGMP to its specific high affinity site. Thus, while cAMP competed with cGMP for hydrolysis at the catalytic site, it did not appear to compete with cGMP for the high affinity binding sites.

Evidence that the cGMP Binding Sites on Phosphodiesterase are Distinct from the Phosphodiesterase Catalytic
2. When crude phosphodiesterase was exposed to 20 pg/ml of bovine pancreatic trypsin, the activity of phosphodiesterase was enhanced about 67%. Maximal activation was observed at about 4 min. Trypsin had an opposite effect on cGMP binding. Thus, there was a 96% loss of binding at a time when enhancement of phosphodiesterase activity was maximal (Fig. 8). In addition, the photoactivated incorporation of 8-N3-["2P]cIMP into the 120,000-dalton band was abolished by prior exposure of the crude phosphodiesterase preparation to trypsin (data not shown).
3. When 1 p~ cGMP was added to the crude phosphodiesterase preparation, it substantially decreased the photoactivated incorporation of 8-Ns-[:"P]cIMP into the 120,000-dalton protein, under conditions designed to prevent cGMP hydrolysis by phosphodiesterase (Fig. 9, ChannelsA and B ) but not under conditions designed to permit cGMP hydrolysis by phosphodiesterase (Fig. 9, Channels C and D ) . These results are consistent with the conclusion (23, 24) that 8-substituted cyclic nucleotides are poor substrates for certain phosphodiesterases, whereas the unsubstituted cyclic nucleotides are readily hydrolyzed in the absence of a phosphodiesterase inhibitor. The results also demonstrated that 8-N&"P]cIMP binding, and presumably cGMP binding, to the high affinity site was not inhibited by 1-methyl-3-isobutylxanthine per se and did not require Mg' + ( Fig. 9, compare Lanes B and D ) , in apparent contrast to binding of cGMP at the catalytic site. In fact, we have found that 1-methyl-3-isobutylxanthine enhanced cGMP binding in the presence of 1 p~, but not of 10 PM, added cGMP under conditions in which 1-methyl-3-isobutylxanthine markedly inhibited phosphodiesterase catalytic activity (data not shown). Hammet and Coquil also found, using 0.1 PM ["HI cGMP in their assay mixture, that I-methyl-3-isobutylxanthine enhanced cGMP binding.

4.
When the crude phosphodiesterase preparation was stored in phosphate buffer at ice temperature (pH 6.0 to pH 7 . 9 , catalytic activity was stable for at least 9 days. Under these same conditions, cGMP-binding activity deteriorated (Fig. 10). This difference in stability between phosphodiesterase activity and cGMP-binding activity was not seen in Tris. HC1 buffer (pH 7 to 8.5), in which both phosphodiesterase activity and cGMP binding were stable at ice temperature for up to 9 days. In acetate buffer (pH 4.0 to 6.0), both phosphodiesterase activity and cGMP binding deteriorated rapidly, and were lost after storage at ice temperature for 48 h (data not shown).
Concluding Remarks-The data presented here indicate the existence of two classes of specific, high affinity, saturable cGMP binding sites on photoreceptor phosphodiesterase. A variety of experimental procedures described here suggest that these sites are distinct from the catalytic site. Clearly, further studies are needed to elucidate the physiological role of these sites. One attractive possibility would be that the high affinity cGMP binding sites are involved in regulating the catalytic activity of the phosphodiesterase molecule. Preliminary evidence2 indicates that two physiological regulators of phosphodiesterase activity (light and GTP) are capable of decreasing the binding of cGMP to the high affinity sites.
We note that binding of cGMP to crude phosphodiesterase ( Fig. 1) half-saturates at somewhat lower concentrations than binding to highly purified phosphodiesterase (Fig. 7). We A. Yamazaki and M. Bitensky, unpublished observations, believe this difference reflects a factor(s) in the crude material which enhances the binding of cGMP to phosphodiesterase.' This factor may also contribute to the modest differences in binding/activity ratios found between more highly purified phosphodiesterase preparations (Figs. 3 and 5) and cruder phosphodiesterase preparations (Figs. 6 and 10). These differences in cGMP binding, which are associated with the degree of purification of the phosphodiesterase, may also reflect the fact that the cGMP binding sites are much more sensitive to proteolysis than is the catalytic activity (Fig. 8).
After the completion of this manuscript, S. H. Francis, T. M. Lincoln, and J. D. Corbin kindly informed us that they have also characterized a specific cGMP binding protein from rat lung which co-purifies with a cGMP phosphodiesterase (25). Their results with the lung phosphodiesterase differed in certain respects from the results obtained with the lightsensitive enzyme. Nevertheless, the specific cGMP binding site of the lung enzyme was distinct from the phosphodiesterase catalytic site (25), in agreement with the results obtained with rat platelets by Hamet and Coquil (13) and with rod outer segments in the present study.