Activation of Retinal Guanylyl Cyclase-1 by Ca2+-binding Proteins Involves Its Dimerization*

Retinal guanylyl cyclase-1 (retGC-1), a key enzyme in phototransduction, is activated by guanylyl cyclase-activating proteins (GCAPs) if [Ca2+] is less than 300 nm. The activation is believed to be essential for the recovery of photoreceptors to the dark state; however, the molecular mechanism of the activation is unknown. Here, we report that dimerization of retGC-1 is involved in its activation by GCAPs. The GC activity and the formation of a 210-kDa cross-linked product of retGC-1 were monitored in bovine rod outer segment homogenates, GCAPs-free bovine rod outer segment membranes and recombinant bovine retGC-1 expressed in COS-7 cells. In addition to recombinant bovine GCAPs, constitutively active mutants of GCAPs that activate retGC-1 in a [Ca2+]-independent manner and bovine brain S100b that activates retGC-1 in the presence of ∼10 μm [Ca2+] were used to investigate whether these activations take place through a similar mechanism, and whether [Ca2+] is directly involved in the dimerization. We found that a monomeric form of retGC-1 (∼110 kDa) was mainly observed whenever GC activity was at basal or low levels. However, the 210-kDa product was increased whenever the GC activity was stimulated by any Ca2+-binding proteins used. We also found that [Ca2+] did not directly regulate the formation of the 210-kDa product. The 210-kDa product was detected in a purified GC preparation and did not contain GCAPs even when the formation of the 210-kDa product was stimulated by GCAPs. These data strongly suggest that the 210-kDa cross-linked product is a homodimer of retGC-1. We conclude that inactive retGC-1 is predominantly a monomeric form, and that dimerization of retGC-1 may be an essential step for its activation by active forms of GCAPs.

In outer segments of vertebrate retinal photoreceptors, rhodopsin absorbs a photon which in turn triggers GTP-dependent activation of cGMP phosphodiesterase. The activated phosphodiesterase hydrolyzes cGMP. The resulting decrease in cytoplasmic [cGMP] leads to reduction in the activity of cGMP-gated cation channels and hyperpolarization of plasma membranes (1)(2)(3)(4). Restoration of the dark membrane potential requires recovery of the dark level of cytoplasmic [cGMP]. Therefore, GC, 1 the enzyme that converts GTP to cGMP, has a crucial role in visual transduction. A retinal membrane GC has been purified from frog, toad, and bovine photoreceptor outer segments (5,6) and identified as a ϳ110 kDa protein. Isozymes of the GC have also been shown biochemically (5,7). Subsequently two forms of membrane GC (retGC-1, ROS-GC1 or GC-E, and retGC-2, ROS-GC2 or GC-F) were cloned from human, bovine, and rat retinal cDNA libraries (8 -12). The structure of these retGCs indicates that the enzyme is a member of the peptide-regulated, membrane-bound GC family, although the retGC is not activated by known peptides. Four functional domains in retGCs have been predicted: a N-terminal extracellular domain, a transmembrane domain, an intercellular protein-kinase-like domain, and a C-terminal catalytic domain. Immunocytochemistry has shown that retGC-1 is localized primarily in cone outer segments and to a lesser extent in rod outer segments (13)(14)(15). RetGC-1 is also detected in the plexiform layers of retina (13,15), leading to speculation that the enzyme is not unique in photoreceptor outer segments. RetGC-2 is in photoreceptors (10); however, detailed localization of the enzyme in the retina has not been demonstrated. RetGC-1 appears to be more abundant than retGC-2 in the retina (10), and only retGC-1 may contribute to the pool of cGMP essential to support phototransduction in cone photoreceptors (16).
The reduction of cGMP-gated channel activity by lowering cytoplasmic [cGMP] blocks Na 2ϩ and Ca 2ϩ influx, and allows a Na 2ϩ /Ca 2ϩ , K ϩ exchanger to decrease cytoplasmic [Ca 2ϩ ] (4,17). When [Ca 2ϩ ] is low, retGC is stimulated (18). In contrast to other membrane-bound GCs that are regulated by binding of peptides to their extracellular domain (19,20), this Ca 2ϩ -sensitive stimulation of retGC is mediated by at least two calmodulin-like Ca 2ϩ -binding proteins termed GCAPs, 1 and 2 (14,(21)(22)(23). GCAPs interact with the intracellular domain of the enzyme (24,25). When free [Ca 2ϩ ] is less than ϳ300 nM, retGC-1 is activated by GCAPs. RetGC-2 has also been reported to be stimulated by GCAP-2 under similar [Ca 2ϩ ] (10,12,26). In addition, GCAPs appear to inhibit the basal GC activity in the presence of the higher [Ca 2ϩ ] (more than ϳ500 nM) (27,28). Thus, the regulatory mechanism of retGC is completely different from that of peptide-regulated GCs. GCAP-1 is mainly detected in cone outer segments, in particular, in disc membrane regions (29 -31). GCAP-1 is also observed in rod outer segments, but the content is much lower than that in cone outer segments. Less GCAP-1 is also found in synaptic regions and inner segments of cones. GCAP-2 is predominantly observed in outer and inner segments of rods and cones (23, 29 -31). Synaptic regions are also labeled by a GCAP-2 antibody. It has also been reported that Ca 2ϩ -binding proteins of the S100 family, especially S100b, activate retGC-1 in the presence of high [Ca 2ϩ ] (32,33). Half-maximal activation was observed at about 40ϳ50 M [Ca 2ϩ ] (33). Therefore, it is believed that S100 proteins are not involved in phototransduction. Additional mechanisms and factors may also be involved in the regulation of retGCs, including phosphorylation (34,35), ATP binding (36 -38), actin binding (39), an inhibitor (guanylyl cyclase-inhibitory protein) of GCAP-activated retGC in amphibian retina (40), and inhibition of retGC-1 by RGS9 (41).
Characterizations of membrane-bound GC and adenylyl cyclase have suggested that the mechanisms for the expression of these enzymatic activities are closely related (42,43), and that at least two cyclase catalytic consensus domains may be required for the activity of these cyclases. Membrane adenylyl cyclase contains two putative catalytic domains that when separately expressed have no activity (44). Peptide-regulated GCs have also been proposed to exist as dimeric or oligomeric forms even in the absence of ligands (45)(46)(47). A recent study has also suggested that retGCs form homodimers in photoreceptor outer segments (48). However, it remains unknown whether dimerization or oligomerization of retGCs is related to the regulation of retGC activity. This question is important especially because the activity of retGCs is regulated so differently from peptideregulated GCs and the regulatory mechanism of retGCs by GCAPs may not be identical to that of peptide-regulated GCs. Thus, we investigated the relationship between activation of retGC-1 by GCAPs and dimerization of the enzyme. In addition to GCAPs, we used constitutively active mutants of GCAPs and S100b to show that dimerization of retGC-1 is related to its activation, not to Ca 2ϩ -binding proteins or [Ca 2ϩ ]. Dimerization of retGC-1 was monitored using a cross-linker under the conditions similar to those used for the measurement of the GC activity. Our results suggest that GCAPs activate retGC-1 by enhancing its dimerization, and that transition of the active form of retGC-1 to its inactive form is caused by either partial or complete dissociation of the dimeric form.
GC Activity Assay-GC activity was measured as described (5). Our preliminary studies indicated that 5 g of protein of a ROS homogenate had high GC activity with negligible hydrolysis of cGMP under assay conditions. Thus, 5 g of protein of ROS homogenate or GCAPs-free ROS membranes were used for all studies. Briefly, ROS membranes (5 g of protein) were incubated with 200 l of Buffer C (50 mM HEPES (pH 7.5), 1 mM GTP, 1 mM cGMP, 2 mM 1-methyl-3-isobutylxanthine, 5 mM MgCl 2 , 15 mM phosphocreatine, 50 g/ml creatine phosphokinase, ϳ5 Ci of [␣-32 P]GTP, and ϳ0.5 Ci of [ 3 H]cGMP). The reaction was initiated by addition of GTP and cGMP. Following incubation (37°C for 10 min), the reaction was terminated by adding 40 l of 1 N HCl and boiling for 2 min. [ 32 P]cGMP derived from [␣-32 P]GTP was isolated by alumina and AG 1-X2 columns (5), and both 3 H and 32 P radioactivities in samples were counted.
Cross-linking of retGC-1 by BS 3 -In order to obtain the exact relationship between dimerization of retGC-1 and its activity, cross-linking reaction of retGC-1 was carried out under conditions similar to that for the measurement of GC activity, although the protein concentrations used were different because each experiment was performed in its linear range. A bovine ROS homogenate, GCAPs-free ROS membranes or COS-7 cell membranes containing recombinant bovine retGC-1 were used. The preparation was suspended in 50 l of Buffer D (50 mM HEPES (pH 7.5), 1 mM GTP, 1 mM cGMP, and 5 mM MgCl 2 ). Pilot experiments indicated that the optimal conditions for the cross-linking reaction by BS 3 were the following: BS 3 concentration, 50 M (Fig. 1); temperature for the cross-linking reaction, 0°C; incubation period for the cross-linking reaction, 30 min; and protein, 50 g. We note that the GTP regeneration system and proteinase inhibitors added in the GC assay mixture did not affect the cross-linking reactions. We also note that the presence of 1 mM GTP in the cross-linking reaction mixture slightly (less than 10%) inhibited the formation of the 210-kDa crosslinked product, but 1 mM cGMP did not. The cross-linking reaction was terminated by addition of SDS-sample buffer and boiling for 5 min. The cross-linked products were immediately separated by SDS-PAGE (5-20% acrylamide gradient). After electrophoresis, proteins were blotted to PVDF membranes and cross-linked products of retGC-1 were detected with a retGC-1-specific antibody (13) and chemiluminescent autoradiography using ULTRA Super Signal substrate. The bands of cross-linked products were scanned by Paragon 1200A3 Pro Scanner and the relative density (mm 2 ϫ OD) was calculated by Molecular Analyst Software (Bio-Rad). It should be emphasized that the R f value of the 210-kDa cross-linked product was slightly different in each SDS-PAGE; however, the molecular mass of the 210-kDa product was constantly monitored by myosin (205 kDa) as a molecular standard. The molecular mass of the Ͼ400-kDa cross-linked product(s) was estimated using high molecular mass (97,400 -584,400) standards.
Gel Filtration Column Chromatography-The freshly solubilized retGC was prepared by incubation of ROS membranes with 1 ml of Buffer E (20 mM HEPES (pH 7.5), 1 mM dithiothreitol, 5 mM MgCl 2 , 0.1 mM phenylmethylsulfonyl fluoride, 5 M leupeptin, 5 M pepstatin A, and 150 mM NaCl) containing 5% n-dodecyl-␤-D-maltoside for 2 h at 0°C (5). After centrifugation (100,000 ϫ g, 4°C, 30 min), a portion of the supernatant (1.5 mg of protein, 200 l) was immediately applied to a Sephacryl S-200 HR column (9 ϫ 600 mm) which had been equilibrated with Buffer E containing 0.1% n-dodecyl-␤-D-maltoside. Another portion of the supernatant was stored on ice overnight, and applied to the column under the same conditions. Purified retGC (8 g) was also applied to the column under the same conditions. Column chromatography conditions were as follows: flow rate, 1.5 ml/10 min; and fraction volume, 1.0 ml. For column calibration, proteins with known Stokes radii were chromatographed under identical conditions.
Analytical Methods-SDS-PAGE was performed as described (51). Protein concentrations were assayed with bovine serum albumin as standard (52). Protein concentration in samples containing a detergent was monitored as described (5). Ca-EGTA buffers were calculated (26) and prepared (53). Free [Ca 2ϩ ] was verified by Ca 2ϩ -electrode and Rod-2 titration. It should be emphasized that all experiments were carried out more than two times, and the results were similar. Data shown are representative of these experiments.

Detection of a retGC-1 Dimer with Molecular Mass ϳ210
kDa-Dimerization or oligomerization of peptide-regulated GCs have been monitored using several methods including cross-linking (45), measurement of molecular weight (46), and the yeast two-hybrid system (47). In this study, to obtain the exact relationship between dimerization of retGC-1 and its activity, we attempted to fix dimeric (or oligomeric) forms of retGC-1 by cross-linking under a condition similar to these used for measurement of GC activity. We found that two oligomeric forms of the retGC-1, ϳ210 and Ͼ400 kDa, were fixed by less than 60 M BS 3 in a ROS homogenate (Fig. 1, upper inset). The amounts of both cross-linked products were increased in a [BS 3 ]-dependent manner; however, the 210-kDa product was detected by a much lower [BS 3 ] than the Ͼ400-kDa product. These cross-linked products in ROS homogenates were also formed with different cross-linkers, such as 3,3Ј-dithiobis(sulfosuccinimidyl propionate) and disuccinimidyl suberate.
These two cross-linked products of retGC-1 were also observed in a purified retGC preparation (Fig. 1, lower inset), indicating that these oligomers are made from retGC. Because the molecular mass (ϳ210 kDa) of the cross-linking product is similar to the calculated molecular mass of the retGC-1 dimer (ϳ220 kDa), we conclude that the 210-kDa product is a dimer of retGC-1. We note that the 210-kDa cross-linked product was detected in ROS homogenates only when [Ca 2ϩ ] was low (data not shown), and that the amount of 210-kDa product in the purified retGC-1 preparation was much less than that in ROS homogenates.
RetGC-1 has been shown to be regulated by proteins, such as GCAPs (14,(21)(22)(23) and RGS9 (41), in photoreceptor outer segments. It is possible that the 210-kDa cross-linked product of retGC-1 in the ROS homogenate (Fig. 1) is a complex of retGC-1 with these retGC regulators. As shown in Fig. 2, the ROS homogenate contained GCAPs; however, GCAPs were not detected in the 210-kDa cross-linked product although under these conditions the formation of the 210-kDa cross-linked product was stimulated by GCAPs, as described below. In addition, the ROS homogenate also contained RGS9; however, RGS9 was not present in the 210-kDa product. These observations support our conclusion that the 210-kDa cross-linked product is a homodimer of retGC-1, but not a monomeric form of retGC-1 cross-linked with other proteins, such as GCAPs and RGS9.
Effects of Ca 2ϩ and GCAPs on the GC Activity and the Formation of the 210-kDa Cross-linked Product of retGC-1-The GC activity in a ROS homogenate was high in the presence of low [Ca 2ϩ ], but inhibited to its basal activity when [Ca 2ϩ ] was higher than 400 nM (Fig. 3A). This behavior of retGCs is consistent with the interpretation that GCAPs in the ROS homogenate function as activators of retGCs in the presence of low [Ca 2ϩ ]. Under these conditions, the 210-kDa cross-linked product of retGC-1 was clearly formed; however, the product was drastically decreased when [Ca 2ϩ ] was higher than 500 nM (Fig. 3, B and C). In other words, if the GC activity was basal, the 110-kDa retGC-1 was distinctly observed; however, the 210-kDa product was increased when the GC activity was increased. These data strongly indicate a relationship between the formation of the 210-kDa cross-linked product of retGC-1 and the retGC activity. In this experiment, the amounts of the retGC-1 cross-linked product(s) with molecular mass Ͼ400-kDa appear to be parallel to the retGC activity. The possible mechanism for the formation of this high molecular weight complex and meanings of this observation will be discussed later.
In order to strengthen our conclusion that the 210-kDa crosslinked product of retGC-1 was formed when the activity of retGC-1 was increased by GCAPs, the effect of recombinant GCAP-1 on the GC activity and the formation of the 210-kDa cross-linked product of retGC-1 in GCAPs-free ROS membranes were investigated (Fig. 4). A constitutively active mutant of GCAP-1, Y99C, was also used not only to strengthen our The cross-linked protein products were immediately separated by SDS-PAGE, transferred to a PVDF membrane, and detected by Western immunoblotting analysis with a chemiluminescent substrate using a retGC-1-specific antibody. The upper inset shows the profile of cross-linked products of retGC-1 in ROS homogenates. The 210-kDa product was scanned and its relative density was calculated as described under "Experimental Procedures." One hundred % indicates the amounts of the 210-kDa product formed in the presence of 60 M BS 3 . Purified retGC (ϳ0.5 g) was also incubated with or without 50 M BS 3 under the same conditions. The lower inset shows the profile of cross-linked products of retGC-1 in the purified preparation. A ϳ110-kDa retGC-1 is a monomeric form.  (0°C and 30 min). The cross-linking reaction was quenched by the addition of SDS sample buffer and boiling for 5 min. The cross-linked products were separated by SDS-PAGE and transferred to PVDF membranes. The cross-linked products of retGC-1, RGS9, and GCAPs were detected by Western immunoblotting analysis using rabbit antibodies specific to these proteins and a chemiluminescent substrate. To avoid a larger figure, parts of gels are shown. There is no chemiluminescent band between 55 and 110 kDa.
argument but also to indicate that [Ca 2ϩ ] does not directly regulate the formation of the 210-kDa product. The GC activity of these membranes was low even when [Ca 2ϩ ] was low (Fig.  4A), indicating that functional amounts of GCAPs had been washed out from membranes. Addition of GCAP-1 increased GC activity of membranes if [Ca 2ϩ ] was low; however, the GC activation was drastically diminished when [Ca 2ϩ ] was increased (Fig. 4A). On the other hand, even without Ca 2ϩ , the 210-kDa product in GCAPs-free membranes was reduced to ϳ35% of that found in the ROS homogenate (Fig. 4, B and C). Addition of GCAP-1 recovered the 210-kDa product to ϳ85% of that found in the ROS homogenate if [Ca 2ϩ ] was low; however, the recovery was not detected in the presence of more than 500 nM [Ca 2ϩ ] (Fig. 4, B and C). When the constitutively active mutant of GCAP-1 was added to these membranes, the GC activity in the membranes was constantly high, although the GC activity was slightly reduced if [Ca 2ϩ ] was increased (Fig.  4A). Under these conditions, the amount of the 210-kDa product was also consistently high although the 210-kDa product decreased slightly if [Ca 2ϩ ] was increased (Fig. 4, B and C). These observations indicate that the 210-kDa cross-linked product of retGC-1 is less when the retGC activity is basal, and . The cross-linking reaction (30 min, 0°C) was carried out with 50 M BS 3 and quenched by the addition of SDS sample buffer and boiling for 5 min. As controls, cross-linking reaction of ROS homogenate (50 g) was also performed with or without BS 3 . The cross-linked products were immediately isolated by SDS-PAGE, transferred to PVDF membrane, and detected by Western immunoblotting analysis with a chemiluminescent substrate using a retGC-1-specific antibody. The 210-kDa cross-linked product in B was scanned and the relative density was shown in C. One hundred % indicates the amounts of the 210-kDa product formed without Ca 2ϩ in the ROS homogenate. ROS, ROS homogenate; Ca-W-Me, GCAPs-free membranes; R*, ROS homogenate without BS 3 ; R, ROS homogenate with BS 3 ; C, GCAPs-free ROS membranes with BS 3 ; ϩGCAP-1, GCAPs-free ROS membranes and GCAP-1 in the presence of BS 3 ; ϩGCAP-1m, GCAPs-free ROS membranes and GCAP-1m in the presence of BS 3 . that the formation of the 210-kDa product is stimulated when the GC activity was stimulated by GCAP-1 or its mutant. Moreover, [Ca 2ϩ ] is not directly involved in the regulation of the 210-kDa product formation. We note that, when [Ca 2ϩ ] was more than 500 nM, GCAP-1 did not inhibit the basal activity of retGC in membranes (Fig. 4A), although amounts of the 210-kDa product were reduced to ϳ20% of the basal level (Fig. 4, B and C). We speculate that the ability of GCAP-1 to inhibit retGC basal activity is weak. Alternatively, the final concentration of GCAP-1 in the cross-linking mixture may be higher than that in the mixture for the assay of enzymatic activity because larger amounts of membranes were used for the crosslinking reaction; that is, the residual amounts of GCAP-1 in membranes may be much higher than that for the enzyme assay. We also note that the cross-linked product(s) of retGC-1 with Ͼ400-kDa molecular mass was not detected under these conditions. GCAP-2 and its constitutively active mutant, E80Q/ E160Q/D158N, showed similar effects on the GC activity and the formation of the 210-kDa cross-linked product of retGC-1 in the GCAPs-free ROS membranes (Fig. 5). In addition, both the GC activity and the formation of the 210-kDa product were reduced to less than these basal levels if [Ca 2ϩ ] was higher than 1 M. These observations clearly indicate the relationship between the GC activation and the formation of the 210-kDa cross-linked product of retGC-1 in ROS membranes.

Effects of GCAPs on the GC Activity of Recombinant retGC-1 and the Formation of the 210-kDa Cross-linked Product of retGC-1-Recombinant retGC-1 expressed in COS-7 cells was
also used to check the relationship between the activation of retGC-1 and the formation of the 210-kDa cross-linked product (Fig. 6). The GC activity in these membranes was low and the activity was not changed with or without Ca 2ϩ , indicating that GCAPs were not expressed (Fig. 6B). Addition of GCAPs to these membranes activated retGC-1 in the absence of Ca 2ϩ ; however, in the presence of 1.5 M Ca 2ϩ , the GC activation was drastically reduced. Addition of constitutively active mutants of GCAPs, Y99C, and E80Q/E160Q/D158N, to these membranes enhanced the retGC-1 activity and this enhancement cross-linking reaction of ROS homogenate (50 g) was also performed with or without BS 3 . The cross-linked products were immediately isolated by SDS-PAGE, transferred to PVDF membrane, and detected by Western immunoblotting analysis with a chemiluminescent substrate using a retGC-1-specific antibody. The 210-kDa cross-linked product in B was scanned and the relative density was shown in C. One hundred % indicates the amounts of the 210-kDa product formed without Ca 2ϩ in the ROS homogenate. ROS, ROS homogenate; Ca-W-Me, GCAPs-free ROS membranes; R*, ROS homogenate without BS 3 ; R, ROS homogenate with BS 3 ; C, GCAPs-free ROS membranes with BS 3 ; ϩGCAP-2, GCAPs-free ROS membranes and GCAP-2 in the presence of BS 3 ; ϩGCAP-2m, GCAPs-free ROS membranes and GCAP-2m in the presence of BS 3 . was virtually unaffected by the high [Ca 2ϩ ]. These results indicate that active forms of GCAPs affect the retGC-1 activity in these membranes the same as in photoreceptor membranes. Under these conditions, when GCAPs were added to membranes and Ca 2ϩ was not present, the formation of the 210-kDa product was clearly detected; however, the 210-kDa product was not observed in the presence of 1.5 M Ca 2ϩ (Fig. 6A). When constitutively active mutants of GCAPs were added to membranes, the formation of the 210-kDa product was detected and the formation was not sensitive to Ca 2ϩ . These results indicate the clear relationship between the activation of retGC-1 by GCAPs and the formation of the 210-kDa crosslinked product of retGC-1. Moreover, these observations demonstrate that Ca 2ϩ is not directly involved in the formation of the 210-kDa product. We note that the retGC-1 complex(s) with molecular mass Ͼ400 kDa was constantly detected in these membranes even without BS 3 , and that formation of the complex was not related to the GC activity.
The Effect of S100b on the GC Activity and the Formation of the 210-kDa Cross-linked Product of retGC-1-The GC activity in ROS membranes is activated by the S100 family proteins in the presence of very high [Ca 2ϩ ] (32,33). The relationship between the activation of retGC and the formation of 210-kDa product was also investigated using GCAPs-free membranes and S100b (Fig. 7). The GC activity in GCAPs-free membranes used was ϳ27% of the GC activity in the original ROS homogenate (Fig. 7A), indicating that GCAPs have been washed out from these membranes. Under these conditions, the GC activity in membranes was increased by the addition of 2 M S100b in a [Ca 2ϩ ]-dependent manner. The half-maximal activation was observed to be ϳ40 M [Ca 2ϩ ]. Under the same conditions, the amounts of the 210-kDa product were increased when [Ca 2ϩ ] was increased (Fig. 7, B and C). We note that without S100b such high [Ca 2ϩ ] stimulated neither the GC activity nor the formation of the 210-kDa product (data not shown). These results indicate that the clear relationship exists between the activation of retGC by S100b and the formation of the 210-kDa product of retGC-1. Since GCAPs also stimulate the formation of the 210-kDa product in the presence of low [Ca 2ϩ ] (Figs. 4 -6), these observations imply that the 210-kDa product is formed whenever retGC is activated by any activator, and that Ca 2ϩ is only involved in the regulation of activators.
Molecular Mass of Various GC Preparations from Bovine ROS Measured by Gel Filtration-The purified retGC has been shown to behave as a very large molecular mass complex or an aggregated form (5). We confirmed that a large portion of retGC purified from ROS membranes appeared to be aggregated (Fig.  8A). This observation is also supported by the presence of substantial amounts of the oligomeric form(s) of retGC-1 with molecular mass Ͼ400-kDa in the purified GC even without a cross-linker (Fig. 1). Since the activity of purified retGC is basal (5) and the amounts of the large molecular mass complex(es) was not proportional to the retGC activity (data not shown), the large molecular mass complex should not be related to the activation of retGC. In order to estimate the formation mechanism of the large molecular mass complex, we measured the molecular mass of the retGC solubilized freshly from photoreceptor membranes. As shown in Fig. 8B, the enzyme solubilized freshly was eluted in a peak corresponding to a molecule with a Stoke's radius of 48.9 Å and an estimated molecular mass of ϳ150 kDa. The size of the retGC is consistent with that of a monomeric form of retGC with detergent bound and/or that of elongated retGC. However, when the same retGC preparation was stored overnight and applied to the column, the retGC was eluted in fractions similar to that of the purified retGC (Fig. 8B). Aparicio and Applebury (54) also reported similar data. These results suggest that retGC can be present as a monomeric form in a fresh preparation, and that FIG. 7. Effect of S100b on GC activity and formation of the 210-kDa cross-linked product in GCAPs-free ROS membranes. A, GC activity. With 2 M S100b, the GC activity in GCAPs-free ROS membranes (equal to 5 g of ROS homogenate) was measured in 200 l of Buffer C containing various [Ca 2ϩ ]. Results represent mean values from three separate experiments performed in duplicate. As a control, GC activity of ROS homogenate (5 g) was also measured without Ca 2ϩ . E, GCAPs-free ROS membranes; q, ROS homogenate. B and C, all cross-linked products (B) and relative amounts of the 210-kDa product (C). With or without 5 M S100b, GCAPs-free ROS membranes (equal to 50 g of ROS homogenate) were incubated in 50 l of Buffer D containing various [Ca 2ϩ ]. The cross-linking reaction (30 min, 0°C) was carried out with 50 M BS 3 and quenched by the addition of SDS sample buffer and boiling for 5 min. As a control, cross-linking reaction of ROS homogenate (50 g) was also performed in the absence of Ca 2ϩ . The cross-linked products were immediately isolated by SDS-PAGE, transferred to PVDF membrane, and detected by Western immunoblotting analysis with a chemiluminescent substrate using a retGC-1-specific antibody. The 210-kDa cross-linked product in B was scanned and the relative density was shown in C. One hundred % indicates the amounts of the 210-kDa product formed without Ca 2ϩ in the ROS homogenate. ROS, ROS homogenate; Ca-W-Me, GCAPs-free ROS membranes. the storage of retGC preparations makes the enzyme aggregated. These observations imply that the aggregated form of the purified retGC may be formed during its purification. DISCUSSION In retinal photoreceptor cells, retGC is known to be activated by GCAPs when [Ca 2ϩ ] is less than 300 nM. The activation is believed to be essential for the recovery of photoreceptors to the dark state, although the molecular mechanism of the activation is unknown. In this study, we have shown that retGC-1 exists in a monomeric form (ϳ110 kDa) whenever GC activity is at basal or low level, and the 210-kDa cross-linked product is increased whenever the GC activity is stimulated. We have also indicated that the 210-kDa cross-linked product is a homodimer of retGC-1, and the [Ca 2ϩ ] directly regulates neither the GC activity nor formation of the 210-kDa cross-linked product. Thus, we conclude that dimerization of retGC-1 is involved in its activation by active forms of GCAPs, as summarized in Fig. 9. To reach these conclusions, we monitored the GC activity and formation of the 210-kDa cross-linked product of retGC-1 under similar conditions. Moreover, we used constitutively active mutants of GCAPs and S100b, not only to strengthen our argument but also to indicate that [Ca 2ϩ ] does not directly regulate the formation of the 210-kDa product. In addition to retGC-1 expressed in COS cells, ROS homogenates and GCAPs-free ROS membranes were used. Then, the total GC activity in these membranes was compared with the amounts of cross-linked products detected by a retGC-1-specific antibody. Although these ROS membranes contain two retGCs (retGC-1 and -2) (8 -12, 48), our results show a clear relationship between retGC activity and the 210-kDa cross-linked product of retGC-1. It is possible that retGC-2 is also dimerized when activated by GCAPs because retGC-2 is similar to retGC-1 in its activation by GCAPs (8 -12). We could not use retGC-2 expressed in COS cells to investigate this possibility because of low activity. Alternatively, the GC activity measured may be mainly contributed by retGC-1 because retGC-2 may be less abundant (10).
Peptide-regulated GCs have been proposed to be present as oligomeric forms even without activation by peptides (42)(43)(44). In this study, we have also shown that retGC-1 complexes with molecular mass Ͼ400 kDa was observed in several GC preparations. As summarized in Fig. 9, we believe that the high molecular mass complex(es) of retGC-1 is formed artificially from monomeric and/or dimeric forms of retGC-1 for the following reasons. (a) The detection of the high molecular mass complex was not constant in ROS membranes. For example, the complex was detected in ROS homogenates shown in Figs. 3, 5, and 7, but not in Figs. 2 and 4. In addition, these complexes were detected in the GCAP-free membranes shown in Fig. 5, but not in Fig. 4. Moreover, the presence or absence of these complexes appears not to be related to the GC activity. Thus, the high molecular mass complex may be formed during preparation of membranes. (b) We have shown that a retGC preparation becomes aggregated if stored, and that a purified retGC preparation behaves in a gel filtration column as the retGC preparation stored overnight (Fig. 8). Moreover, the purified preparation contains substantial amounts of the retGC complex with molecular mass Ͼ400 kDa (Fig. 1). These observations imply that the high molecular mass complex in the purified sample is self-aggregated by storage or purification procedures. (c) In recombinant retGC-1 expressed in COS-7 cells, the high molecular mass complex was detected without cross-linker (Fig. 6). The high molecular mass complex was also observed even when the GC activity was basal and the content of the complex was not changed even when the activity of retGC-1 was changed. The high molecular mass complex was also detected in retGC-2 expressed in COS cells (data not shown). These results suggest that the high molecular mass complex is not related to the GC activity. It is possible that the complex is a self-aggregated form and/or a form of misfolding retGC-1 during its expression.
Although it is difficult to rule out completely the possibility that the 210-kDa cross-linked product is a retGC-1 complexed with proteins other than retGC-1, we anticipate that the 210-kDa product is a dimer form of retGC-1 for the following reasons. (a) The molecular mass of the cross-linked product (ϳ210 kDa) is similar to the calculated molecular mass (ϳ220 kDa) of the retGC-1 dimer. (b) The 210-kDa product was detected in a preparation of purified retGC-1 (Fig. 1). This retGC preparation did not contain GCAPs and RGS9 (data not shown). Formation of the 210-kDa product in the purified retGC preparation was much less than that in ROS homogenates because the purified preparation did not contain any activator and its GC activity is low. Even if GCAPs are contaminated in the purified preparation, its GC activity should be basal because GCAPs do not function as retGC activators in the presence of a detergent (6). (c) The 210-kDa product was also detected in a preparation of recombinant retGC-1 expressed in COS-7 cells (Fig. 6). (d) The 210-kDa product is increased whenever the GC activity is stimulated (Figs. 1 and 3-7). These observations exclude the possibility that inhibitory regulators, such as guanylyl cyclaseinhibitory protein (40) and RGS9 (41), are complexed with retGC-1 to form the 210-kDa product. Using a RGS9-specific antibody (41) we have shown that RGS9 is not involved in the 210-kDa product (Fig. 2). (e) The 210-kDa product does not contain GCAPs even when the formation of the 210-kDa product was stimulated by GCAPs. Western blotting of the 210-kDa product using GCAP-specific antibodies has indicated that the 210-kDa product does not contain GCAPs under our conditions (Fig. 2). (f) The 210-kDa product in ROS homogenates was also observed by different cross-linkers, such as 3,3Ј-dithiobis(sulfosuccinimidyl propionate) and disuccinimidyl suberate, when the GC activity in the preparation was stimulated by lowering GCAPs were not found in the ϳ210-kDa cross-linked product of retGC-1 when the dimerization of retGC-1 was stimulated by GCAPs (Fig. 2). However, it should be emphasized that this study does not exclude the possibility that the real retGC dimer contains Ca 2ϩ activators if the formation of the retGC dimer is enhanced by Ca 2ϩ activators. As depicted in Fig. 9, we rather believe that Ca 2ϩ activators were contained in the retGC dimer before cross-linking reaction if the dimer formation was stimulated by Ca 2ϩ activators. We believe that our conditions for the cross-linking reaction fix the interaction between retGCs, but not between retGC and GCAPs. A previous study has already shown that a high molecular mass cross-linked product contains retGC-1 and GCAP-1 (24). Previous studies have suggested that a different region of retGC-1 appears to be required for its interaction with each Ca 2ϩ activator (24,25,55). Thus, it is possible that each Ca 2ϩ activator forms a retGC-1 dimer in a different way, and that each retGC-1 dimer may be complexed with a different Ca 2ϩ activator. A slight difference in the conformation of the retGC-1 dimers may be important for the fine regulation of free [cGMP] in photoreceptor outer segments. FIG. 9. Relationship between monomer, dimer, and oligomer of retGC-1 in the molecular mechanism of retGC-1 activation by GCAPs.