The membrane guanylyl cyclase, retinal guanylyl cyclase-1, is activated through its intracellular domain.

Retinal guanylyl cyclase-1 (RetGC-1) is a membrane guanylyl cyclase found in photoreceptor outer segments. It consists of an apparent extracellular domain (ECD) linked by a single transmembrane segment to an intracellular domain (ICD). Guanylyl cyclase activating protein-2 (GCAP-2) is a Ca-binding protein that activates RetGC-1 in a Ca-sensitive manner. To establish whether GCAP-2 stimulates RetGC-1 through the ECD or ICD, we made deletion mutants lacking either the ECD or both the ECD and transmembrane domains (TMD) of RetGC-1. Recombinant wild type RetGC-1 and both deletion mutants were expressed in HEK 293 cells, and their sensitivities to GCAP-2, Ca, and ATP were compared. Our data demonstrate that both deletion mutants are regulated similarly to wild type RetGC-1 with indistinguishable EC values for Ca and similar K values for activation by GCAP-2. This shows that GCAP-2 functions through the ICD of RetGC-1 and that removal of the ECD and TMD do not significantly alter regulation by these factors. Our data also show that ATP potentiates stimulation of guanylyl cyclase activity by GCAP-2 and that neither the ECD nor the TMD of RetGC-1 participate in its regulation by ATP.

Retinal guanylyl cyclase-1 (RetGC-1) is a membrane guanylyl cyclase found in photoreceptor outer segments. It consists of an apparent extracellular domain (ECD) linked by a single transmembrane segment to an intracellular domain (ICD). Guanylyl cyclase activating protein-2 (GCAP-2) is a Ca 2؉ -binding protein that activates RetGC-1 in a Ca 2؉ -sensitive manner. To establish whether GCAP-2 stimulates RetGC-1 through the ECD or ICD, we made deletion mutants lacking either the ECD or both the ECD and transmembrane domains (TMD) of RetGC-1. Recombinant wild type RetGC-1 and both deletion mutants were expressed in HEK 293 cells, and their sensitivities to GCAP-2, Ca 2؉ , and ATP were compared. Our data demonstrate that both deletion mutants are regulated similarly to wild type RetGC-1 with indistinguishable EC 50 values for Ca 2؉ and similar K1 ⁄2 values for activation by GCAP-2. This shows that GCAP-2 functions through the ICD of RetGC-1 and that removal of the ECD and TMD do not significantly alter regulation by these factors. Our data also show that ATP potentiates stimulation of guanylyl cyclase activity by GCAP-2 and that neither the ECD nor the TMD of RetGC-1 participate in its regulation by ATP.
Photoexcitation of retinal rod cells stimulates hydrolysis of intracellular cGMP. This reduces the activity of cGMP gated cation channels in the rod outer segment plasma membrane, blocks Na ϩ and Ca 2ϩ influx, and allows a Na ϩ /Ca 2ϩ ,K ϩ exchanger to decrease the concentration of free cytoplasmic Ca 2ϩ (reviewed in Lagnado and Baylor (1992)). At low concentrations of free Ca 2ϩ , a soluble factor stimulates guanylyl cyclase (GC) 1 activity in photoreceptor membranes (Koch and Stryer, 1988). It has now been shown that this stimulatory activity is represented by at least two Ca 2ϩ -binding proteins, GCAP-1 and GCAP-2 Gorczyca et al., 1995). It has been proposed that the light-induced decrease in free cytoplasmic Ca 2ϩ concentration stimulates GC activity in vivo. Such a feedback mechanism would stimulate resynthesis of cGMP and enhance photoreceptor recovery and/or light adaptation follow-ing photoexcitation (reviewed in Lagnado and Baylor (1992) and McNaughton (1990)).
RetGC-1 and RetGC-2 are two photoreceptor GCs that were cloned from a human retinal cDNA library (Shyjan et al., 1992;Lowe et al., 1995). 2 Homologues of RetGC-1 have also been cloned from bovine and rat eye cDNA libraries, and a homologue of RetGC-2 was isolated from a rat eye cDNA library (Goraczniak et al., 1994;Yang et al., 1995). RetGC-1 and RetGC-2 are expressed in photoreceptors and associate with the membrane fraction of photoreceptor outer segments (OS) (Shyjan et al., 1992, Dizhoor et al., 1994Lowe et al., 1995). Immunofluorescence studies suggest that RetGC-1 is localized primarily in cone OS and to a lesser extent in rod OS (Dizhoor et al., 1994;Liu et al., 1994). Electron microscopy studies further indicate that RetGC-1 is associated with the membranerich regions of OS (Liu et al., 1994). The GC activities of OS membranes and recombinant RetGC-1 are activated by the photoreceptor Ca 2ϩ -binding proteins GCAP-1 and GCAP-2, and recombinant RetGC-2 is activated by GCAP-2 Gorczyca et al., 1995;Lowe et al., 1995). Activation by either GCAP is inhibited by Ca 2ϩ with an EC 50 for Ca 2ϩ near 200 nM (Dizhoor et al., 1994;Gorczyca et al., 1994aGorczyca et al., , 1995Dizhoor et al., 1995). This value agrees well with the range of bulk free Ca 2ϩ (50 -550 nM) recently measured in intact OS in darkness and following a flash of light (Gray-Keller and Detwiler, 1994). The localization of RetGC-1 and RetGC-2 and their sensitivity to Ca 2ϩ and GCAP-2 suggest that they function in the recovery of photoreceptors from photoexcitation (Dizhoor et al., 1994;Lowe et al., 1995).
A factor that influences the Ca 2ϩ -sensitive stimulation of OS GCs is ATP. It has been reported that ATP or nonhydrolyzable ATP analogues potentiate the Ca 2ϩ -sensitive stimulation of OS GCs in whole OS and in washed OS reconstituted with GCAP-1 (Gorczyca et al., 1994b). Both RetGC-1 and RetGC-2 are members of the membrane GC family that includes the natriuretic peptide receptor-GCs (NPR-A/GC-A and NPR-B/GC-B), the heat stable enterotoxin or guanylin receptor-GC (StaR/GC-C), and the sea urchin sperm GCs (Shyjan et al., 1992;Garbers and Lowe, 1994;Lowe et al., 1995). It has been clearly shown that the stimulation of other members of this family is influenced by adenine nucleotides. The stimulation of NPR-A requires (Chinkers and Garbers, 1991;Marala et al., 1991), and the stimulation of StaR is prolonged by (Vaandrager et al., 1993a) the presence of ATP or nonhydrolyzable ATP analogues.
Each member of the membrane GC family is a type I transmembrane protein that has a ligand-binding extracellular domain (ECD) linked by a single transmembrane domain (TMD) to an intracellular domain (ICD). Within the ICD the membrane-proximal region is homologous to protein kinases (KHD). Adjacent to the KHD is a small domain that is likely to form an amphipathic ␣-helix and for which a clear role in the dimerization of the NPR-A intracellular domain has been established (Wilson and Chinkers, 1995). The C-terminal portion of the ICD contains the cyclase catalytic domain. Based on the homology between RetGC-1, NPR-A, NPR-B, and StaR and on hydropathy analysis, putative assignments have been made for the extracellular and intracellular domains of RetGC-1 (see Fig. 1) (Shyjan et al., 1992;Lowe et al., 1995;Wilson and Chinkers, 1995). The orientation of RetGC-1 in membranes has yet to be determined experimentally.
Previously characterized members of the membrane GC family are activated by the binding of peptide ligands to their extracellular domains (reviewed by Garbers (1992) and Garbers and Lowe (1994)). For example, the membrane GC NPR-A is stimulated by the binding of atrial natriuretic peptide to its ECD. In contrast, the Ca 2ϩ sensitivity of RetGC-1 and RetGC-2 activity stimulated by GCAP-2 (Lowe et al., 1995), the absence of a signal peptide in the GCAP-2 primary sequence , and the presence of Ca 2ϩ -binding sites on GCAP-2 all suggest that regulation of RetGC-1 and RetGC-2 by GCAP-2, Ca 2ϩ , and ATP occurs in the cytoplasm.
To determine experimentally if GCAP-2 acts either through the predicted ICD of RetGC-1 or through the ECD, we expressed deletion mutants of RetGC-1 lacking the ECD or both the TMD and ECD. Our results show that RetGC-1 is regulated by GCAP-2 through the ICD and that removal of the ECD and TMD does not have a significant effect on regulation by GCAP-2, Ca 2ϩ , and ATP.

Construction of RetGC-1 Deletion Mutants-
The wt RetGC-1 cDNA, cloned into the EcoRI site of pBSSK(Ϫ), was removed by digesting with XbaI and HindIII and ligated into a mammalian expression vector, pRC CMV (Invitrogen), digested with the same restriction enzymes. The deletion mutants ⌬ECD RetGC-1 and ICD RetGC-1 were then constructed from wt RetGC-1 cDNA in pRC CMV using signal overlap extension PCR (Horton and Pease, 1991). Two pair of primers (primer pair: overlapping primer and flanking primer), sequential PCR reactions, and two convenient restriction sites were required for the construction of each deletion mutant. Of each pair of primers, a single primer was designed to contain an overlap that could hybridize with sequence contained by the opposing overlapping primer. The first PCR reaction was carried out with separate reactions for each primer pair, and then the resulting PCR product from each of the two reactions was gel purified and mixed in an equimolar ratio to serve as template in the second round of PCR. A second round of PCR was then carried out using only the flanking primers with these templates. Products of the second round of PCR contained the desired deletions. For both ⌬ECD RetGC-1 and ICD RetGC-1, the final PCR products were digested with HindIII and PflMI and ligated into pRC-CMV RetGC-1 digested with the same restriction enzymes. The primers used for ⌬ECD RetGC-1 were forward flanking primer T-7 and reverse overlapping primer 5Ј-GAAGACGAG-GCCCGG/GGAGAGGGCGGGGGGC-3Ј (bp 1468 -1484/216 -234) for pair one and overlapping forward primer 5Ј-GCCCCCCGCCCTCTCC/ CCGGGCCTCGTCTTTC-3Ј (bp 216 -234/1468 -1484) and reverse flanking primer 5Ј- GCGAGGAGGTCCTGAAGAGA-3Ј (bp 1947GCGAGGAGGTCCTGAAGAGA-3Ј (bp -1976 for the second pair. For ICD RetGC-1 the reverse overlap primer was 5Ј-CGGAGACCATTTGCAT/TGCCGGCTTAGGGAAG-3Ј (bp 1560 -1576/66 -81), and the forward overlap primer was 5Ј-CTTCCCTAAGC-CGGCA/ATGCAAATGGTCTCCG-3Ј (bp 66 -81/1560 -1576); the same flanking primers were used for ICD RetGC-1 as for ⌬ECD RetGC-1. pRC CMV expression constructs were amplified in TB1 Escherichia coli strain, purified by a CsCl density gradient and sequenced to ensure no errors had been introduced. All PCR reactions were carried out with either taq (Perkin-Elmer) or pfu (Stratagene) polymerases.
Expression in HEK 293 Cells-10 -20 g of the constructs pRC CMV-wt RetGC-1, pRC CMV-⌬ECD RetGC-1, or pRC CMV-ICD RetGC-1 were transiently transfected into 70 -80% confluent HEK 293 cells using the calcium-phosphate precipitation method (Gorman et al., 1990). Control cells were transiently transfected with the pRC CMV expression vector. The cells were grown in Dulbecco's modified Eagle medium/nutrient mixture F-12 (50/50) (Life Technologies Inc.), 10% heat-treated fetal calf serum, 10 mM Hepes, pH 7.3, in 100-mm tissue culture dishes. DNA precipitates were left on the cells for 4 h followed by removal of the medium, washing the cells once or twice with phosphate-buffered saline and adding fresh medium. After 48 h the medium was again changed, and cells were then left until harvest. At 64 h cells were washed with phosphate-buffered saline and removed from the dish by rocking for 10 min at room temperature in phosphate-buffered saline, 0.5 mM EDTA. The cell suspension was spun down for 2 min at 500 ϫ g in a Beckman tabletop centrifuge, and the pellets were resuspended in 1 ml/dish of homogenization buffer (10 mM MOPS, pH 7.3, 5 mM 2-mercaptoethanol, 20 g/ml leupeptin, 0.5-1 mM phenylmethylsulfonyl fluoride) and allowed to swell on ice for 10 min prior to homogenization with 6 -10 strokes of a Dounce. Large debris were removed from cell homogenates by centrifugation at 2000 ϫ g for 10 min in a Beckman tabletop centrifuge at 4°C. To isolate membranes, homogenates were brought to 0.05 or 0.4 M NaCl and centrifuged for 45 min in a Beckman TLA 100 tabletop ultracentrifuge at 288,000 ϫ g in a TLA 120.3 rotor at 4°C. A pellet derived from one dish of cells was resuspended in 200 -400 l of homogenization buffer. Membrane preparations and homogenates were frozen on dry ice and stored at Ϫ70°C. Protein content of membrane preparations was determined in the absence of reducing agents and in the presence of 0.1% SDS using BCA protein reagent (Pierce) with BSA as a standard. All HEK 293 cell growth and transfection procedures were carried out in a 5% CO 2 , 37°C humidified incubator.
Recombinant GCAP-2: Expression and Preparation-A cDNA of GCAP-2 (p24)  was cloned into the bacterial expression vector pet 11d (Novagen). GCAP-2 was coexpressed in BL21 E. coli with yeast N-myristoyl transferase with 100 g/ml myristic acid in Luria broth . Preparation and purification of active myristoylated recombinant GCAP-2 from E. coli will be described elsewhere. 3 Briefly, recombinant myristoylated GCAP-2 was insoluble and therefore was dissolved in 6 M urea and then refolded by overnight dialysis against 20 mM Tris, 1 mM EDTA, and 1 mM dithiothreitol. The recombinant GCAP-2 used in these studies had a specific activity indistinguishable from that of purified preparations of retinal GCAP-2. The purification of retinal GCAP-2 has been described . Protein concentrations of both retinal and recombinant GCAP-2 were determined using the Bio-Rad Protein assay reagent and protocol.
GC Assay-The measurement of GC activity was carried out essentially as described in Dizhoor et al. (1995). In brief, transiently transfected 293 cell homogenates or membranes were added to equal volumes of 4 ϫ GC buffer (400 mM KCl, 200 mM MOPS, 28 mM 2-mercaptoethanol, 40 mM MgCl, 32 mM NaCl, 4 mM EGTA). 12.5 l of this mixture were added to purified retinal or recombinant GCAP-2 or BSA to reach a volume of 20 l. The reaction was started by adding 5 l of a 5 ϫ substrate solution (5 mM GTP, 25 mM cGMP,ϳ 2 Ci of [␣-32 P]GTP, ϳ100,000 dpm [8-3 H]cGMP) and then incubated at 30°C for 30 min for wt RetGC-1 and ⌬ECD RetGC-1 or 1 h for ICD RetGC-1 because of its lower specific activity. The reaction was stopped by heating in a 100°C heating block for 2 min. After centrifugation at 10,000 ϫ g for 10 min to pellet the heat-denatured proteins, 32 P-labeled cGMP was separated from [␣-32 P] GTP by chromatographing 6 l of the reaction mix on a polyethyleneimine cellulose thin layer chromatography plate in 0.2 M LiCl. Spots corresponding to cGMP were visualized on a short wavelength UV illuminator, excised, and eluted by gentle shaking for 10 min in 1 ml of 2 M LiCl in a 20-ml scintillation vial, and both 3 H and 32 P were counted in 10 ml of Ecolume scintillant (ICN) in a Beckman model LS 3801 scintillation counter. For all assays less than 10% of the substrate was depleted. For experiments involving the effects of adenine nucleotides on GC activity, an appropriate concentration of adenine nucleotides was added to the substrate mix. (For the ATP titration experiment the value of the basal GC activity of wt RetGC-1 for each of the two experiments was 1.57 Ϯ 0.422 and 0.2 Ϯ 0.03 nmol cGMP/min/mg protein and for the stimulated GC activity (without ATP) was 4.61 Ϯ 0.77 and 0.77 Ϯ 0.004 nmol cGMP/min/mg protein. The basal GC activity of ⌬ECD RetGC-1 for each of the two experiments was 0.367 Ϯ 0.026 and 0.187 Ϯ 0.013 nmol cGMP/min/mg protein and for stimulated GC activity was 1.307 Ϯ 0.02 and 0.53 Ϯ 0.047 nmol cGMP/min/mg protein.) Ca-EGTA Buffers-Ca-EGTA buffers were prepared from solutions of EGTA (Sigma) and EGTA saturated with CaCl 2 (Fluka) by pH titration in strict accordance with the method of Tsien and Pozzan (1989). Free Ca 2ϩ concentrations under the assay conditions were calculated using a multi-factor program (Marks and Maxfield, 1991) and verified by Ca 2ϩ electrode and by titration with Rhod-2 fluorescent dye (Calbiochem). In experiments where the free Ca 2ϩ concentration was controlled by the addition of a Ca-EGTA buffer, no EGTA was present in the 4 ϫ GC buffer, and 2 l of a 20 mM Ca-EGTA buffer stock was added to the assay mixture to reach a final volume of 27 l.
Antibody Preparation-Anti-peptide antibody RetGC-1-IC (intracellular) was generated in rabbits against a synthetic peptide corresponding to Ala 642 -Gln 655 from the intracellular domain of RetGC-1 and an additional cysteine added at the the N terminus. The peptide was coupled to keyhole limpet hemocyanin with the cross-linking reagent, n-maleimidobenzoyl-N-hydroxysuccimide ester (Pierce). This peptide is derived from the KHD of RetGC-1 and is distinct from any sequence of RetGC-2. The antibody was purified by specific binding to the peptide coupled to CNBr-activated Sepharose, eluted in 10 mM glycine, 0.5 M NaCl, 0.05% Tween 20 at pH 2.3, and the eluate was immediately neutralized to pH 7.5 with an aliquot of 1 M Na 2 HPO 4 . The affinity purified antibody recognizes only a single band in immunoblot analysis of unwashed photoreceptor OS (data not shown).
Demonstration of Expression in HEK 293 Cells by Immunoblot Analysis-Proteins mixed with Laemmli sample buffer were electrophoresed on either 7.5 or 10% SDS tris-glycine gels and transferred to nitrocellulose membrane (Schleicher & Schuell) using the Bio-Rad Mini-PRO-TEIN system. The membrane was blocked in TTBS (Tris-buffered saline, 0.05% Tween 20) containing 10% dry milk for 1 h at room temperature and then probed with 4 nM of affinity purified RetGC-1-IC antibody in blocking buffer for 3 h at room temperature. After extensive washing in TTBS, the membrane was probed for 1 h at room temperature with a donkey anti-rabbit antibody coupled to horseradish peroxidase (Amersham Corp.), and after extensive washing with TTBS, the membrane was developed using chemiluminescent reagents (Amersham Corp.). With both transfected and untransfected HEK 293 cell homogenates or membranes isolated in low salt, three or four nonspecific bands showed up on immunoblots probed with the RetGC-1-IC antibody. However, washing the HEK 293 cell membranes with 0.4 M NaCl removed all but a single band for transfected cells and all bands for control cells. A single band corresponding to the predicted size of the recombinant variants of RetGC-1 and not these additional bands could be competed away with the peptide used to generate the antibody (data not shown).

RESULTS
To determine whether the ECD or TMD of RetGC-1 play a role in stimulation of GC activity by GCAP-2, deletion mutants were constructed that lacked either the ECD (⌬ECD RetGC-1) or both the ECD and TMD domains (ICD RetGC-1) of RetGC-1 (Fig. 1). We then determined if the truncated proteins could be regulated in a Ca 2ϩ -sensitive manner by GCAP-2. Both recombinant and purified retinal GCAP-2 were used to carry out these experiments. We also compared the effect of ATP on the truncated and wt forms of RetGC-1 to further gauge how these truncations affected the regulation of RetGC-1 activity.
Expression of wt RetGC-1 and ⌬ECD RetGC-1-HEK 293 cells were transfected with wt RetGC-1 or ⌬ECD RetGC-1 expression constructs, and expression of the recombinant proteins was demonstrated by immunoblot analysis (Fig. 2A). Immunoreactivity of the RetGC-1-IC antibody reveals a single band for both wt RetGC-1 and ⌬ECD using recombinant membranes washed in 0.4 M NaCl. Membranes were washed in 0.4 M NaCl to remove proteins that were recognized nonspecifically by the primary or secondary antibody (see "Materials and Methods"). The molecular weights of 114,656 and 71,545, predicted from the coding region of each construct, are similar to the relative molecular weights of the immunoreactive polypeptides. Both recombinant proteins are localized to the particu- DD indicates a putative dimerization domain. The 41 amino acids shown to be the minimum requirement for dimerization of NPR-A (Wilson and Chinkers, 1995) correspond to Ile 766 -Leu 806 of RetGC-1 and share 46% identity and 71% homology. Furthermore, helical wheel analysis of this stretch of amino acids for NPR-A (Wilson and Chinkers, 1995) and RetGC-1 shows that it is likely that both sequences form an amphipathic ␣-helix. CT indicates a hydrophilic C terminus for which there is no counterpart in the NPRs. StaR, however, has a hydrophilic C-terminal extension, but it is longer than and has little homology to that of RetGC-1. ICD indicates the entire intracellular region. The bar denotes the approximate location of the Ala 642 -Gln 655 peptide used to generate the RetGC-1-IC antibody. late fraction in HEK 293 cells. If we assume that the immunoreactivity of wt and ⌬ECD RetGC-1 to this antibody are similar, then estimates of relative expression levels by densitometry of Immunoblots indicate that wt RetGC-1 is expressed at 10 -50-fold higher levels than ⌬ECD RetGC-1.
Ca 2ϩ -sensitive Regulation of wt RetGC-1 and ⌬ECD RetGC-1 by GCAP-2-Both wt RetGC-1 and ⌬ECD RetGC-1 are stimulated by GCAP-2 in the presence of low but not high Ca 2ϩ concentrations (Fig. 2, B, C, and D). This shows that GCAP-2 activates RetGC-1 through the intracellular or transmembrane domains. The specific activity of stimulated recombinant wt RetGC-1, estimated on the basis of GC activity and total protein, ranged from 2-10 times greater than that of ⌬ECD RetGC-1 in eight separate transfections. In general, the specific activity did not correlate well with expression levels estimated from densitometry of immunoblots. This may reflect variability of the fraction of recombinant GC that is active or folded correctly. Because our primary goal was to study the regulation of RetGC-1 by Ca 2ϩ and GCAP-2, the variability in specific activity does not affect our conclusion that stimulation by GCAP-2 does not require the ECD.
Effect of Adenine Nucleotides on Regulation of wt RetGC-1 and ⌬ECD RetGC-1-GCAP-2 stimulation of recombinant wt RetGC-1 and ⌬ECD RetGC-1 is potentiated by ATP and the nonhydrolyzable analogue AMP-PNP (Fig. 3). The ATP/GCAP-2-stimulated activity is approximately 2.7-fold above the effect of GCAP-2 alone. Adenosine nucleotides do not stimulate GC activity in the absence of GCAP-2. The potentiating effect of ATP was previously reported by a group using a different guanylyl cyclase activating protein, GCAP-1, reconstituted with washed OS membranes (Gorczyca et al., 1994b). In contrast to those results we detected only a small (ϳ20%) potentiation by ATP using GCAP-2 when we used washed OS membranes rather than recombinant RetGC-1 (data not shown).
To examine whether the ECD influences the effect of ATP, we examined the concentration dependence of the ATP effect on wt RetGC-1 and ⌬ECD RetGC-1 (Fig. 4). No significant difference in ATP dependence was detected, indicating that the ECD is not involved in the mechanism by which ATP potentiates activation of RetGC-1. For both wt RetGC-1 and ⌬ECD RetGC-1, the effect of ATP reaches a maximum near 0.5 mM ATP then decreases. The decrease in activity above 0.5 mM ATP may be due to competition for binding at the catalytic site of RetGC-1 between ATP and the substrate, GTP. The nucleotide TTP had no effect on the stimulation of catalytic activity by GCAP-2 (data not shown).
Expression and Solubility of ICD RetGC-1-We also examined the deletion mutant, ICD RetGC-1, which lacks both the ECD and TMD. Its expression in transiently transfected HEK 293 cells is demonstrated by immunoblot analysis using the RetGC-1-IC antibody (Fig. 5A). We expected that ICD RetGC-1 would be a soluble enzyme, but the immunoreactivity of the recombinant protein resided entirely within the particulate fraction. Treatment with 1 M NaCl or 2% Triton X-100 did not remove the immunoreactivity from the particulate fraction (data not shown). However, ICD RetGC-1 can be partially solubilized (ϳ50%) in 6 M urea without detergent. This suggests that ICD RetGC-1 no longer contains a TMD. Urea without detergent does not solubilize wt RetGC-1 and ⌬ECD RetGC-1 (data not shown). These properties of ICD RetGC-1 may reflect the same interaction that makes bovine RetGC-1 insoluble in nonionic detergents (Hakki and Sitaramayya, 1990;Koch, 1991). The detergent insolubility of RetGC-1 both with and without the TMD suggests that RetGC-1 is either a large oligomeric structure or associates with the cytoskeleton.
Regulation of ICD RetGC-1 by GCAP-2/Ca 2ϩ and Adenine Nucleotides-The catalytic activity of ICD RetGC-1 is regulated by GCAP-2, Ca 2ϩ , and adenine nucleotides (Fig. 5, B and  C). However, its specific activity appears to be lowered significantly by the removal of the TMD. Even when immunoblot analysis suggests that ICD RetGC-1 is expressed at levels equal to or greater than recombinant wt RetGC-1 and ⌬ECD RetGC-1, its specific activity remains significantly lower.
Sensitivity of ICD RetGC-1 to Salt Concentration-Like wt RetGC-1 and ⌬ECD RetGC-1, recombinant ICD Ret GC-1 membranes were washed in 0.4 M NaCl to remove nonspecific immunoreactive bands. These same membranes were initially used to test ICD Ret GC-1 for Ca 2ϩ -sensitive stimulation by GCAP-2. We discovered that unlike wt RetGC-1 and ⌬ECD RetGC-1, ICD RetGC-1 is inactive at high NaCl concentrations (data not shown). Homogenates and membranes isolated in 0.05 M NaCl are stimulated by GCAP-2, but membranes washed in 0.4 M NaCl lose most or all GCAP-2-stimulated activity. The lowered specific activity of ICD RetGC-1 relative to wt RetGC-1 and its heightened sensitivity to NaCl may reflect instability in its structure induced by removal of the TMD.
Effect of RetGC-1 Deletions on K 1/2 of GCAP-2 and EC 50 for Ca 2ϩ -To determine if the ECD or TMD influenced the ability of GCAP-2 to stimulate GC activity we compared the EC 50 values for Ca 2ϩ and for GCAP-2 stimulation of wt RetGC-1, ⌬ECD RetGC-1, and ICD RetGC-1. The Ca 2ϩ sensitivity for stimulation of wt RetGC-1 and the deletion mutants overlap closely, with an EC 50 for Ca 2ϩ near 280 nM (Fig. 6). This value is in the range of previously reported EC 50 values for Ca 2ϩ of 90, 240, or 144 nM for the GC activity of OS membranes (Koch and Stryer, 1988;Dizhoor et al., 1991;Wolbring and Schnetkamp, 1995) and 200 nM for recombinant wt RetGC-1 (Dizhoor et al., 1994). The half-saturation (K1 ⁄2 ) values for stimulation by GCAP-2 were also measured. Mean K1 ⁄2 values and standard deviations of 5.42 Ϯ 2.31, 8.03 Ϯ 2.25, and 2.27 Ϯ 0.031 M were obtained for wt RetGC-1, ⌬ECD RetGC-1, and ICD RetGC-1, accordingly (Fig. 7). The K1 ⁄2 values obtained for the deletion mutants were each statistically different from the values obtained for wt RetGC-1 in two out of three experiments with a 95% confidence interval. However, it is clear that no gross alterations of the K1 ⁄2 value have been introduced by the removal of the ECD and TMD. Interestingly, these K1 ⁄2 values are 10 -40-fold higher than the value obtained when recombinant GCAP-2 is reconstituted with washed OS membranes, but the general characteristics of regulation by GCAP-2/Ca 2ϩ are clearly represented. The higher K1 ⁄2 values may reflect that recombinant RetGC-1 is differentially modified than the OS GCs, that HEK 293 cell membranes have a different lipid composition than bovine OS membranes, or that an additional protein, present in OS membranes, which increases the potency of GCAP-2 is absent in our reconstituted system.

DISCUSSION
The data we present here show that the ECD and TMD of RetGC-1 are not required for its regulation by Ca 2ϩ and GCAP-2. GCAP-2 must exert its action through the intracellular domain of RetGC-1. RetGC-1 associates with the OS membranes of photoreceptors, is glycosylated, and is predicted from the cDNA to have a signal peptide for membrane localization on the N terminus (Dizhoor et al. 1994;Koch et al., 1994;Shyjan et al., 1992;Lowe et al., 1995). These properties of RetGC-1 suggest that the ECD is either extracellular or intradiscal and that a new mechanism for regulating membrane  6. Comparison of the Ca 2؉ sensitivity of GCAP-2 activation for wt RetGC-1, ⌬ECD RetGC-1, and ICD RetGC-1. GC activity was measured on HEK 293 cell homogenates as described under "Materials and Methods." Assays of wt RetGC-1 (closed triangle), ⌬ECD RetGC-1 (open circle), and ICD RetGC-1 (open square) in the presence of retinal GCAP-2 and 0.5 mM ATP were carried out at Ca 2ϩ concentrations ranging from 7 nM to 2.6 M free Ca 2ϩ . A Ca 2ϩ -EGTA buffering system was used to achieve the desired free Ca 2ϩ concentrations. Duplicate data points are plotted from a single experiment and are representative of two independent experiments. GC activities were normalized for the comparison.
FIG. 7. Concentration dependence of GC activation by GCAP-2. The activity of wt RetGC-1, ⌬ECD RetGC-1, and ICD RetGC-1 was measured in 1 mM EGTA, 0.5 mM ATP, and the indicated concentrations of recombinant GCAP-2. GC activity was plotted against the GCAP-2 concentration, and a curve was fit to each plot using the equation v ϭ [GCAP-2] V max /K1 ⁄2 ϩ [GCAP-2] (Abelbeck Kaleidagraph software). Data shown are from a single experiment and are representative of three independent experiments. GC activity was normalized to the predicted V max . K1 ⁄2 (Ϯ standard error) values of 7.48 (Ϯ 0.59), 10.09 (Ϯ 0.88), and 2.28 (Ϯ 0.27) M for wt RetGC-1, ⌬ECD RetGC-1, and ICD RetGC-1, respectively, were calculated from the above experiment. Average K1 ⁄2 values generated from this and two additional experiments, both carried out with duplicate measurements at each , are reported in the results section.
GCs from the cytoplasm has now been established. This does not, however, preclude RetGC-1 from also having a yet to be discovered extracellular ligand analogous to those of NPR-A, NPR-B, and StaR.
We also show that ATP potentiates but is not necessary for the stimulatory effect of GCAP-2 on RetGC-1 and that neither the ECD nor TMD are necessary for this effect. The observation that AMP-PNP also potentiates stimulation of RetGC-1 by GCAP-2 indicates that the effect of ATP is not due to phosphorylation or ATPase activity. In vivo, a 2-3-fold increase in the rate of cGMP synthesis could potentially have a large effect on the inward current of a photoreceptor OS because cGMP binding to the cGMP gated channels is cooperative (Fesenko et al., 1985). However, it has yet to be experimentally determined whether intracellular ATP levels vary enough to regulate RetGC-1 activity in vivo.
Based on available published data, it appears that the nonobligatory effect of adenine nucleotides on RetGC-1 differentiate it from NPR-A, for which ATP is a necessary cofactor for stimulation by atrial natriuretic peptide (Chinkers and Garbers, 1991;Marala et al., 1991). In contrast, ligand activation of StaR activity does not require ATP but is potentiated approximately 2-fold by ATP in a manner similar to our findings with RetGC-1 (Vaandrager et al., 1993a). The activation of immunoaffinity-purified NPR-A and StaR by ATP and their extracellular ligands has been reported (Vaandrager et al., 1993b;Wong et al., 1995). This strongly suggests that ATP regulates both receptors through direct binding. Interestingly, both RetGC-1 and StaR lack a glycine-rich nucleotide-binding motif, which is conserved in protein kinases and in the KHDs of both NPR-A and NPR-B (Koller et al., 1992;Shyjan et al., 1992). Mutations in the glycine-rich sequence make NPR-A insensitive to ATP and atrial natriuretic peptide (Goraczniak et al., 1992;Duda et al., 1993). These data taken together suggest that the effect of ATP on membrane GCs may be mediated by binding of ATP to the KHDs.
It is unclear why the effect of ATP we observe for washed OS membranes is not as pronounced as when we use recombinant RetGC-1. One possibility is that the state of phosphorylation of RetGC-1 influences the effect of adenine nucleotides. Recently, it was reported that ATP has a 2-fold stimulatory effect on Ca 2ϩ -sensitive GC activity in fresh intact OS (Wolbring and Schnetkamp, 1995). However, the effect of ATP was lost by washing the OS or by adding inhibitors of protein kinase C. The ATP effect could be restored by treating washed OS with a purified preparation of brain protein kinase C. Together with our data, these findings suggest that phosphorylation by protein kinase C may be a prerequisite for a noncatalytic role of ATP in the potentiation of RetGC-1 activation. Different states of phosphorylation of OS GC and recombinant RetGC-1 might explain the different magnitude of ATP effects observed with recombinant RetGC-1 and OS GCs.
Although we have shown that the intracellular domain of RetGC-1 is sufficient for activation, we have not shown whether or not GCAP-2 functions through direct binding to an intracellular domain of RetGC-1. The reconstitution of Ca 2ϩ sensitive regulation of recombinant RetGC-1 using recombinant GCAP-2 supports this model but does not rule out the involvement of an additional factor found in both OS and in HEK 293 cell membranes. For example, if the detergent insolubility of both OS and recombinant RetGC-1 is indicative of a cytoskeletal association, then we cannot rule out the involvement of a cytoskeletal protein as an intermediate for stimulation by GCAP-2. It also remains to be determined through which intracellular subdomain(s) of RetGC-1 that GCAP-2 transduces its direct or indirect stimulatory effect on catalytic activity.
RetGC-2 is another photoreceptor specific membrane GC that has been shown to be stimulated by GCAP-2 in vitro (Lowe et al., 1995). It is highly homologous to RetGC-1, and together these proteins may define a new subfamily of membrane GCs that respond to intracellular activators.