Possible stimulation of retinal rod recovery to dark state by cGMP release from a cGMP phosphodiesterase noncatalytic site.

Cyclic GMP phosphodiesterase, a key enzyme for phototransduction, contains α, β (Pαβ), and two γ (Pγ) subunits. In addition to catalytic sites, Pαβ has two classes of noncatalytic cGMP binding sites with different affinities (Kd values <100 nM and >1 μM). Pγ regulates Pαβ as an inhibitor of cGMP hydrolysis and as a stimulator of cGMP binding to the high affinity noncatalytic sites. Pγ release from Pαβ by the GTP-bound α subunit of transducin (GTP·Tα) interrupts these two functions. Here we describe a novel regulation of the Pγ release by [cGMP] and its physiological implication. We isolated Pγ mutants that exhibit abnormally one of these two functions, indicating the distinct domains in Pγ are involved to express these functions. When [cGMP] was high (~5 μM), Pγ responsible for the inhibition of cGMP hydrolysis was preferentially released, and cGMP hydrolysis activity of Pαβ was increased about 10 times. When [cGMP] was low (less than ~0.5 μM), Pγ responsible for the stimulation of cGMP binding to the high affinity sites was released. The Pγ release resulted in the decrease of relative affinity of cGMP for the high affinity sites to at least (null)/1;10, followed by the rapid release of cGMP from one of the high affinity sites (apparent t1/2 = 3.8 s). cGMP (~5 μM) inhibited the extraction of Pαβ from rod membranes by a Mg2+-free hypotonic buffer. The inhibition of Pαβ extraction was not affected by Pγ, suggesting that Pαβ detects on the order of micromolar [cGMP] using low affinity noncatalytic sites on Pαβ. Because [cGMP] is ~5 μM in darkness and lowered by photoexcitation and phosphodiesterase concentration is ~30 μM in rod photoreceptors, it is possible that cGMP phosphodiesterase functions to increase cytoplasamic [cGMP] after [cGMP] is reduced to the illuminated level.

Since Bitensky and Miller suggested the involvement of cyclic nucleotides in phototransduction (1), many investigators have contributed to establish the role of cGMP in phototransduction (1)(2)(3). The illuminated rhodopsin stimulates GTP/GDP exchange on T␣, 1 which in turn activates cGMP phosphodiesterase. The resulting decrease of cytoplasmic [cGMP] leads to closure of cGMP-gated channels and hyperpolarization of photoreceptors. The closure of channels also blocks Ca 2ϩ influx, while Ca 2ϩ efflux by a Na ϩ /Ca 2ϩ exchanger continues. The resulting decline in the free [Ca 2ϩ ] is believed to play a major role in the adaptation and recovery processes of photoreceptors by negative feedback regulation by [Ca 2ϩ ] (4). However, regulation of phototransduction by the decrease in cytoplasmic [cGMP] has never been clarified.
Rod cGMP phosphodiesterase contains ␣, ␤ (P␣␤), and two ␥ (P␥) subunits. Previous studies have shown that P␣␤ has two catalytic sites for cGMP hydrolysis (5,6) as well as two classes of noncatalytic cGMP binding sites with different affinities (K d values ϳ100 nM and 1-6 M) (5)(6)(7)(8)(9). These noncatalytic sites are the major cGMP binding sites in ROS, binding more than 90% of the cellular cGMP (7,9). The roles of these noncatalytic sites in phototransduction still remain unclear. However, it is clear in amphibian ROS that P␥ regulates P␣␤ not only as an inhibitor of cGMP hydrolysis (10) but also as a stimulator of cGMP binding to the high affinity noncatalytic sites (11)(12)(13). The same, nonmodified P␥ shows these functions (13). These two P␥ functions are interrupted when GTP⅐T␣ dissociates P␥ from P␣␤ (10 -13). By manipulating ionic strength in the reaction mixtures, we have recently suggested that P␥ expresses only one of these functions when P␥ is complexed with P␣␤ (13).
In this study, we refer to the P␥ responsible for the inhibition of cGMP hydrolysis as iP␥ and the P␥ responsible for the stimulation of cGMP binding to the high affinity sites as sP␥. Using data showing that release of these P␥s is regulated by [cGMP], we propose a novel mechanism for the recovery of [cGMP] to the dark level. EXPERIMENTAL PROCEDURES ROS membranes from dark-adapted Rana catesbiana, P␥-less (activated) cGMP phosphodiesterase membranes, and frog P␥ were prepared as described (10). Activity of phosphodiesterase was measured as described (10,13). Amounts of P␥ released by GTP⅐T␣ were determined using a P␥-specific antibody (13). The P␥ content measured by the P␥-specific antibody was similar to the value obtained by the inhibition of cGMP hydrolysis (10,13). Equilibrium binding of [ 3 H]cGMP to P␣␤ in various ROS membranes was assayed as described (7,13 1 The abbreviations used are: T␣, ␣ subunit of transducin; P␣␤, catalytic subunit of phosphodiesterase: P␥, ␥ subunit of phosphodiesterase; iP␥, P␥ responsible for the inhibition of cGMP hydrolysis by P␣␤; sP␥, P␥ responsible for the stimulation of cGMP binding to the high affinity sites on P␣␤; ROS, rod outer segments; IBMX, 1-methyl-3-isobutylxanthine; PCR, polymerase chain reaction; DTT, dithiothreitol; PMSF, phenylmethylsulfonyl fluoride; GTP␥S, guanosine 5Ј-O(3thiotriphosphate). method (11) without nonradioactive cGMP. The filter was washed with 2 ml of 10 mM Tris⅐HCl (pH 7.5) and 2 mM EDTA (ϫ4). Each washing took about 2 s. Data were analyzed with a nonlinear curvefitting program Igor Pro (Wave Metrics, Lake Oswego, OR) (14). All experiments were carried out more than three times, and the results were similar. Data shown were the representative of these experiments.
In order to replace C-terminal 18 amino acids or N-terminal 16 amino acids, a frameshift mutation was introduced into bovine P␥ cDNA (15). For the C-terminal frameshift mutation, the following primers were used for the first PCR reaction: up, 5Ј-GTCATCTGCTTTGG-GAGGCCTTCA-3Ј and down, 5Ј-CGGGGGTCGGATCCTAAGATGC-CGTAC-3Ј. To amplify the mutant P␥ gene, the second PCR reaction was carried out with the purified product from the first PCR reaction and the following primer: 5Ј-GGAGGTTTTACATATGAACTTAGAACC-3Ј. The mutant contains the following amino acid sequence in its Cterminal: FGRPSTTWRCTRWPRTAS. For the N-terminal frameshift mutation, the following primers were used for the first PCR reaction: up, 5Ј-ACAAGCATATGAAACCGGGTGCCACCCAAGGC-3Ј and down, 5Ј-GGGAGTGACGGGCCCCCCATCACC-3Ј. To amplify the mutant P␥ gene, the second PCR reaction was carried out with the purified product from the first PCR reaction and the following primer: 5Ј-GGGGTCG-GATCCTAGATGATGCCGTACTG-3Ј. The mutant contains the following amino acid sequence in its N-terminal: MKPGATQGRDPVGHKGD. These amplified P␥ mutant gene products were digested with NdeI and BamHI and cloned into a NdeI-BamHI-digested pET-IIA (Novagene). The vector was transferred to Escherichia coli BL21 (DE3) (Novagene) for expression of P␥. The mutations were confirmed by double-stranded DNA sequencing using the fmol DNA sequencing system (Promega). The protein expression was induced by addition of 1 mM isopropyl ␤-D-thiogalactopyranoside. Purification of recombinant P␥ was carried out as described (10).

RESULTS AND DISCUSSION
P␥ Mutants Devoid of One of Two Functions-P␥ regulates P␣␤ as iP␥ and sP␥. We isolated P␥ mutants in which one of these functions was suppressed (Fig. 1). When the C-terminal 18 amino acids were replaced by a frameshift mutation, the mutant (C-18) stimulated cGMP binding similarly to wild type P␥; however, the inhibitory activity was completely abolished. When the N-terminal 16 amino acids were replaced, the mutant (N-16) retained inhibitory activity like wild type P␥. However, a 200 -300 nM amount of this mutant P␥ was required to reach 50% of the maximal cGMP binding that was achieved with ϳ10 nM wild type P␥. These observations indicate that different domains in P␥ are involved to express these functions. Because P␥ expresses only one of these functions when P␥ is complexed with P␣␤ (Ref. 13 and as shown below), these data suggest that P␥ binds to different sites on P␣␤ to express these functions. It should be emphasized that a C-terminal-deleted (10 amino acids) mutant lost the iP␥ activity as shown by many groups; however, the mutant also showed less stimulatory activity for the cGMP binding due to less interaction with P␣␤ (data not shown). We also note that stimulation of cGMP binding by the C-18 mutant clearly indicates that high phosphodiesterase activity did not affect cGMP binding to the high affinity noncatalytic sites on P␣␤ by reducing [cGMP], since cGMP hydrolysis was maximal for all concentrations of the mutant, while cGMP binding was dependent upon the mutant concentration.
Low Sensitivity of the iP␥ Release to cGMP-Phosphodiesterase is activated by GTP⅐T␣ through iP␥ release from P␣␤ in amphibian ROS membranes (10,16,17). We have shown that the iP␥ release can be distinguished from sP␥ release by changing [NaCl] in a releasing buffer (0 -200 mM) (13). However, the change in the cytoplasmic ionic strength by light in frog ROS is not so drastic (18). We have investigated the possibility that cGMP is a physiological regulator for the release of functionally different P␥ from P␣␤. cGMP (1 mM) inhibited the total P␥ release regardless of [NaCl] in the releasing buffer ( Fig. 2A). When the buffer contained less than 100 mM NaCl, the phosphodiesterase activity in membranes washed without cGMP was consistently higher than that washed with cGMP (Fig. 2B). When the buffer contained 150 mM NaCl, the maximum activation was achieved by the first washing without cGMP; however, in the presence of cGMP the full activation was accomplished after the third washing (Fig. 2C). These data unexpectedly indicate that the release of iP␥, i.e. the phosphodiesterase activation, is sensitive to cGMP. However, we emphasize that, in the presence of 150 mM NaCl, 60% of the maximum phosphodiesterase activity can be retained by the first washing even in the presence of 1 mM cGMP (Fig. 2C). Because less than 5 M cGMP is hydrolyzed for photoexcitation (2), the suppression of phosphodiesterase activation by cGMP may be negligible under the physiological ionic conditions (18). We note that the release of T␣ was not affected by cGMP and that excess amounts of T␣ were released for the P␥ release in each washing (data not shown). We also found that ϳ5 M cGMP inhibited the total P␥ release in a buffer containing 150 mM NaCl and 1 mM IBMX (data not shown). This result is consistent with data published by Arshavsky et al. (17). However, we used 1 mM cGMP in this study to exclude the effect of cGMP hydrolysis by phosphodiesterase activated variously.
In the absence of cGMP, the amount of P␥ released with 200 mM NaCl was 3.4 Ϯ 0.5 times (n ϭ 3) that of P␥ released without NaCl (Fig. 2A). Measurement of P␥ by the inhibition of cGMP hydrolysis also indicated that the amount of P␥ released with 200 mM NaCl was 3.6 Ϯ 0.9 times (n ϭ 3) that of P␥ released without NaCl (data not shown). However, membranes washed with or without NaCl showed the activated, same level of phosphodiesterase activity (Fig. 2B). These observations indicate that only 30% (at most) of the total P␥ released functions as iP␥, and more than 70% of P␥ released is not responsible for phosphodiesterase inhibition. At present, we do not have a simple explanation for such an unequivalent release of the functionally different P␥. We speculate that a portion of P␥ bound P␣␤, especially iP␥, may be GTP⅐T␣-insensitive (10) due to its phosphorylation (19). Phosphorylated P␥ cannot be released by GTP⅐T␣ due to the loss of affinity for GTP⅐T␣ (20).
High Sensitivity of the sP␥ Release to cGMP-When the releasing buffer contained more than 100 mM NaCl, the ratio of the total P␥ released in the absence of cGMP to the P␥ released in the presence of cGMP was approximately 2 to 1 ( Fig. 2A). However, phosphodiesterase activities in these membranes measured after P␥ release were identical (Fig. 2B). These observations indicate that 50% (at least) of the total P␥ released is sensitive to cGMP and that the P␥ is not iP␥. When ROS membranes were washed with a buffer containing only GTP, equilibrium [ 3 H]cGMP binding to the high affinity noncatalytic sites on P␣␤ in these membranes was increased by P␥ added (Fig. 3A). However, if the membranes were washed with a buffer containing GTP and cGMP, approximately 80% of the maximum [ 3 H]cGMP binding was detected without adding P␥ and the added P␥ barely stimulated the [ 3 H]cGMP binding. Both membranes had the same level of phosphodiesterase activity (data not shown). Five M cGMP (with 1 mM IBMX) produced a similar effect, although the effect was smaller due to the possible hydrolysis of cGMP during washing (data not shown). These observations indicate that, when [cGMP] is high, iP␥, but not sP␥, is released readily. When [cGMP] is low, both iP␥ and sP␥ are released.
Without P␥, [ 3 H]cGMP could not be loaded to the high affinity sites on P␣␤ using 1 M [ 3 H]cGMP. However, in the presence of P␥, various concentrations of [ 3 H]cGMP (0.1, 0.5, and 1.0 M) showed the similar level of equilibrium cGMP binding to the sites (data not shown). We note that cGMP binding with a K d value ϳ1 M can be measured under our conditions (7). These data indicate that the relative affinity of cGMP for the high affinity sites is reduced to 1 ⁄10 (at least) by release of sP␥. We measured the time course of cGMP release from the high affinity sites on P␣␤␥ 2 by GTP⅐T␣ (Fig. 3B)  These membranes were divided into 12 portions. P␥ was extracted (at 0°C) from each portion by adding rapidly 0.2 ml of Buffer A containing 400 M GTP and various concentrations of NaCl in the presence (q) or absence of (E) of 1 mM cGMP and spinning (345,000 ϫ g, 5 min, 4°C). These membranes were further washed with 0.5 ml of Buffer A (ϫ2). We note that more than 10% of cGMP added was recovered, even when the extraction was carried out in the presence of the maximal phosphodiesterase activity. The control experiment was carried out without GTP. Ç indicates the enzymatic activity in membranes washed without GTP. A, P␥ content in each supernatant. After the supernatant was incubated (2.5 min, 80°C) and spun (345,000 ϫ g, 30 min, 4°C), P␥ in the supernatant (10 l) was measured using a P␥-specific antibody. B, phosphodiesterase activity in membranes. Following appropriate dilution, phosphodiesterase activity in these membranes was measured. One unit of phosphodiesterase activity indicates 2.0 mol of cGMP hydrolyzed per min. C, phosphodiesterase activity after each washing. Illuminated ROS membranes (12.3 mg) were prewashed with 2 ml of Buffer A (ϫ7). Then, membranes were divided into eight portions, and each portion was washed with 0.5 ml of Buffer A containing 400 M GTP and 150 mM NaCl in the presence or absence of 1 mM cGMP. These membranes were further washed with 0.5 ml of Buffer A (ϫ3). After appropriate dilution, phosphodiesterase activity in membranes was measured. One-hundred percent of phosphodiesterase activity indicates 3.6 nmol of cGMP hydrolyzed per min per tube. relatively slow biphasic dissociation of [ 3 H]cGMP from P␣␤␥ 2 by GTP␥S⅐T␣ (apparent t1 ⁄2 ϭ 25 and 264 s, room temperature) in the presence of cGMP (1 mM). Under their conditions the real rate of [ 3 H]cGMP dissociation by GTP␥S⅐T␣ might be slower than these data, because 1 mM cGMP substantially stimulated cGMP release from P␣␤␥ 2 without GTP⅐T␣ (11). We speculate that under their conditions the release of sP␥ from P␣␤ might be inhibited by 1 mM cGMP as shown above. Therefore, dissociation of [ 3 H]cGMP from the high affinity sites on P␣␤ might be suppressed.
Possible Role of the Low Affinity Noncatalytic Sites on P␣␤-We studied the possibility that P␣␤ detects [cGMP] by its low affinity noncatalytic cGMP binding site. When P␣␤ was extracted from ROS membranes with a magnesium-free hypotonic buffer, cGMP specifically inhibited the P␣␤ release (Fig.  4A). cAMP (100 M) did not inhibit the release (data not shown). This inhibition was detectable as low as 5 M cGMP and was not dependent upon P␥, since this inhibition was not changed even after membranes were washed with GTP or GTP and cGMP (Fig. 4, B and C). These data indicate that P␣␤ detects, directly or indirectly, on the order of micromolar [cGMP] and that cGMP binding to the high affinity sites is not involved in this detection. Previous data have shown that under similar conditions only noncatalytic sites on P␣␤ are detected as cGMP binding sites by photoaffinity labeling (6, 10), the K d value for cGMP binding to low affinity noncatalytic sites is about 1-6 M (6, 8), and the K m value of the P␣␤ catalytic sites is about 0.3-1 mM. On the basis of these data, we propose that the low affinity noncatalytic sites on P␣␤ serve as a sensor of the cytoplasmic [cGMP]. The binding of cGMP to these low affinity sites appears to inhibit the sP␥ release by GTP⅐T␣ from P␣␤ through changing of P␣␤ conformation, because a similar [cGMP] inhibits the release of sP␥ as described above.
The present data indicate that sP␥ is released by GTP⅐T␣ when [cGMP] becomes low, resulting in the rapid release of cGMP from one of the high affinity noncatalytic sites on P␣␤. We hypothesize that this cGMP release functions to promote recovery of [cGMP] to the dark level in amphibian ROS. We emphasize that hydrolysis of GTP bound to T␣ measured by biochemical methods is slow (21) and that GTP⅐T␣ is still present after the turnoff of GTP⅐T␣-activated phosphodiesterase by P␥ phosphorylation (19,20). Therefore, GTP⅐T␣ is expected to be present at the late stage of phototransduction. GTP⅐T␣ may not release iP␥ at the stage. We found that release of iP␥ by GTP⅐T␣ is inhibited by endogenous ADP-ribosylation of P␥ complexed with P␣␤ (22). We speculate that this kind of mechanism may be functional when [cGMP] becomes low. Since [cGMP] is ϳ5 M in darkness and P␣␤ is estimated to be ϳ30 M in concentration, cGMP released from one of the high affinity sites by GTP⅐T␣ may be enough to increase cytoplasmic [cGMP] to the dark level, and less than 1.2 s is required to release ϳ5 M cGMP from the site. Thus, it is possible that cGMP phosphodiesterase functions to increase cytoplasmic [cGMP] to the dark level when [cGMP] is reduced to the illuminated level. At present, we cannot compare the contribution of cGMP release from the high affinity site on P␣␤ to that of cGMP synthesis by guanylyl cyclase for the recovery of [cGMP] to the dark level, because the precise data of the velocity of cGMP synthesis are not available. [cGMP] required for the inhibition of P␣␤ release. Illuminated ROS membranes (12 mg) were washed with 2 ml of Buffer A (ϫ7) and then these membranes were divided into two portions. Each portion was washed with 2 ml of Buffer A containing 400 M GTP and 150 mM NaCl in the presence (C) or absence (B) of 1 mM cGMP. Then, the portion was further divided into three portions. P␣␤ was extracted from each portion by 0.5 ml of Buffer B with various concentrations of cGMP, as indicated, in the presence of 1 mM IBMX (ϫ3). P␣␤ in each supernatant (0.15 ml) was isolated by SDS-gel electrophoresis and measured by densitometric scanning.