Further Characterization of the Type 3 Ryanodine Receptor (RyR3) Purified from Rabbit Diaphragm*

We characterized type 3 ryanodine receptor (RyR3) purified from rabbit diaphragm by immunoaffinity chromatography using a specific antibody. The purified receptor was free from 12-kDa FK506-binding protein, although it retained the ability to bind 12-kDa FK506-binding protein. Negatively stained images of RyR3 show a characteristic rectangular structure that was indistinguishable from RyR1. The location of the D2 segment, which exists uniquely in the RyR1 isoform, was determined as the region around domain 9 close to the corner of the square-shaped assembly, with use of D2-directed antibody as a probe. The RyR3 homotetramer had a single class of high affinity [3H]ryanodine-binding sites with a stoichiometry of 1 mol/mol. In planar lipid bilayers, RyR3 displayed cation channel activity that was modulated by several ligands including Ca2+, Mg2+, caffeine, and ATP, which is consistent with [3H]ryanodine binding activity. RyR3 showed a slightly larger unit conductance and a longer mean open time than RyR1. Whereas RyR1 showed two classes of channel activity with distinct open probabilities (P o), RyR3 displayed a homogeneous and steeply Ca2+-dependent activity withP o ∼1. RyR3 was more steeply affected in the channel activity by sulfhydryl-oxidizing and -reducing reagents than RyR1, suggesting that the channel activity of RyR3 may be transformed more precipitously by the redox state. This is also a likely explanation for the difference in the Ca2+ dependence of RyR3 between [3H]ryanodine binding and channel activity.

We characterized type 3 ryanodine receptor (RyR3) purified from rabbit diaphragm by immunoaffinity chromatography using a specific antibody. The purified receptor was free from 12-kDa FK506-binding protein, although it retained the ability to bind 12-kDa FK506binding protein. Negatively stained images of RyR3 show a characteristic rectangular structure that was indistinguishable from RyR1. The location of the D2 segment, which exists uniquely in the RyR1 isoform, was determined as the region around domain 9 close to the corner of the square-shaped assembly, with use of D2directed antibody as a probe. The RyR3 homotetramer had a single class of high affinity [ 3 H]ryanodine-binding sites with a stoichiometry of 1 mol/mol. In planar lipid bilayers, RyR3 displayed cation channel activity that was modulated by several ligands including Ca 2؉ , Mg 2؉ , caffeine, and ATP, which is consistent with [ 3 H]ryanodine binding activity. RyR3 showed a slightly larger unit conductance and a longer mean open time than RyR1. Whereas RyR1 showed two classes of channel activity with distinct open probabilities (P o ), RyR3 displayed a homogeneous and steeply Ca 2؉ -dependent activity with P o ϳ1. RyR3 was more steeply affected in the channel activity by sulfhydryl-oxidizing and -reducing reagents than RyR1, suggesting that the channel activity of RyR3 may be transformed more precipitously by the redox state. This is also a likely explanation for the difference in the Ca 2؉ dependence of RyR3 between [ 3 H]ryanodine binding and channel activity.
Intracellular Ca 2ϩ stores play a critical role in regulation of the cytosolic Ca 2ϩ concentration in various cells. Ryanodine receptors (RyRs) 1 belong to a family of Ca 2ϩ release channels of the Ca 2ϩ stores (1)(2)(3)(4)(5)(6)(7). To date, three isoforms of RyR that are encoded by distinct genes have been identified in mammalian tissues. Type 1 RyR (RyR1) is primarily expressed in skeletal muscles and plays a crucial role in excitation-contraction (E-C) coupling, a process whereby depolarization in the sarcolemma triggers Ca 2ϩ release from sarcoplasmic reticulum (SR) resulting in muscle contraction (6,8,9). Type 2 RyR (RyR2) is the primary isoform in heart and is involved in the E-C coupling in cardiac muscle. cDNA and mRNA for type 3 RyR (RyR3) was first identified in mink lung epithelial cells and in specific regions (hippocampus, thalamus, and corpus striatum) of rabbit brain (3,4) . Recent studies revealed that mRNAs for all these isoforms can be widely detected in various tissues, suggesting potential coexpression of multiple isoforms (5).
RyR1 and RyR2 have been purified from skeletal and cardiac muscles, respectively, and are well characterized (1-3, 5, 7). A homotetramer of the ϳ560-kDa subunit is a functional unit and shows a characteristic structure with 4-fold symmetry that is identical to feet which span the gap between the transverse tubule and terminal cisterna of SR (6,7). A homotetramer can specifically bind one [ 3 H]ryanodine molecule with a K D of nanomolar order, after which it is named RyR (1)(2)(3). When incorporated into planar lipid bilayers, the receptor demonstrates cation channel activity with a large conductance showing properties of Ca 2ϩ -induced Ca 2ϩ release (CICR) (2,3).
Knowledge about RyR3, in contrast, is much less than that about RyR1 or RyR2 because of its minuscule amount. An excess amount of other coexisting isoforms of similar biochemical characteristics has prevented us from isolating RyR3 by conventional purification procedures. To overcome this difficulty, we utilized an antibody specific to RyR3 (anti-RyR3) as a tool for isolation of the protein. Frog skeletal muscle expresses two isoforms of RyRs (␣-and ␤-RyR) in almost equal amounts (10). Cloning and sequencing of their cDNAs revealed that ␣and ␤-RyRs are homologues of RyR1 and RyR3, respectively (11). Taking advantage of the similarity between RyR3 and ␤-RyR, we have successfully raised and purified anti-RyR3 against a peptide of amino acid sequence 4375-4387 of rabbit RyR3 (12). Selective immunoprecipitation with the antibody revealed properties of RyR3 in rabbit brain (12) and diaphragm (13). RyR3 forms a homotetramer of ϳ550-kDa subunit, which is slightly smaller than those of RyR1 and RyR2, as is the case for ␤-RyR. It shows high affinity [ 3 H]ryanodine binding which is activated by micromolar Ca 2ϩ , adenine nucleotides, and caffeine, and inhibited by millimolar concentrations of Ca 2ϩ and Mg 2ϩ , procaine, and ruthenium red, indicating that it may function as a CICR channel. The contents of RyR3 in rabbit diaphragm and brain were estimated to be less than 1 and 0.06%, respectively, of RyR1 in skeletal muscle (12,13).
In this paper, we describe the structural and functional characterization of RyR3 that was successfully purified from rabbit diaphragm by immunoaffinity chromatography. [ 3 H]Ry-anodine binding and single channel recordings of the molecule revealed several unique properties of RyR3. We also defined the location of the differential segment in the assembly of the RyR1 molecule, which could be one of the origins for the difference in function between the two isoforms. During the course of publication of our results, Jeyakumar et al. (14) independently reported the purification of RyR3.
Isolation of Sarcoplasmic Reticulum (SR) Vesicles-Heavy fraction of SR vesicles was prepared from rabbit diaphragm according to Murayama and Ogawa (13) in the presence of a mixture of protease inhibitors (2 g/ml aprotinin, 2 g/ml leupeptin, 1 g/ml antipain, 2 g/ml pepstatin A, and 2 g/ml chymostatin). The isolated membrane vesicles were quickly frozen in liquid N 2 and stored at Ϫ80°C until used.
Purification of RyR3-RyR3 was purified by a combination of sucrose density gradient ultracentrifugation and immunoprecipitation. Anti-RyR3 beads were prepared by mixing anti-rabbit IgG-agarose beads with the purified anti-RyR3 antibody, followed by washing with a buffer containing 0.15 M NaCl, 20 mM sodium phosphate, pH 7.2. These anti-RyR3 beads can be stored at 4°C for 3 months without a significant decrease in the RyR3-binding activity. Rabbit diaphragm SR vesicles (10 mg) were solubilized in 4% CHAPS and 2% egg lecithin in 2.5 ml of 0.5 M NaCl, 20 mM Tris-HCl, pH 7.4, and 2 mM dithiothreitol (DTT) including the above protease inhibitors (buffer A) to isolate tetrameric RyRs by ultracentrifugation on 5-20% linear gradients of sucrose (10). Fractions around ϳ15% sucrose containing tetrameric RyRs were incubated overnight at 4°C with 100 l of the anti-RyR3 beads. The beads were washed five times with buffer A containing 1% CHAPS and 0.5% egg lecithin, and then incubated overnight at 4°C with 30 M RyR3peptide in the same buffer to dissociate RyR3 from the antibody. Finally, the supernatant containing RyR3 was applied to a small gel filtration column (Centri-sep, Applied Biosystems Inc.) to remove a large amount of the RyR3-peptide.
RyR1 was purified from rabbit back muscle in which virtually no RyR3 protein was detected (13) by sucrose gradients and Mono-Q anion exchange column chromatography as described (10).
Protein Assay-Protein concentrations of the purified RyR1 were determined by the method of Kaplan and Pedersen (19) with Amido Black 10B using bovine serum albumin as a standard. The amount of RyR3 was determined by densitometry of its band on Coomassie Brilliant Bluestained SDS-polyacrylamide gel with a MasterScan densitometer. The purified RyR1 or RyR2 was used as a standard in determinations where it showed a linear relationship up to 100 ng (data not shown).
FKBP12 Binding Assay-Binding of FKBP12 to RyR3 was determined using glutathione S-transferase (GST)-hFKBP12 fusion protein (GST-FKBP) (20). The plasmid construction of GST-FKBP was essentially the same as previously reported (21). A plasmid pFKBP333 for hFKBP12 (15) was subjected to PCR to place BamHI and EcoRI restriction sites at the 5Ј-and 3Ј-end of hFKBP12, respectively. The set of primers was as follows: 5Ј-CGGGATCCAGATGGGAGTGCAGGTGGA-AAC-3Ј (hFKBP12-F) and 5Ј-CGGAATTCTCATTCCAGTTTTAGAAG-CTCC-3Ј(hFKBP12-R). The PCR product was digested with BamHI and EcoRI restriction enzymes and ligated into the BamHI/EcoRI sites of pGEX-3X vector (Amersham Pharmacia Biotech). Expression and purification of the fusion protein was carried out according to the manufacturer's instructions. The purified GST-FKBP was recognized by the anti-FKBP12 antibody (data not shown).
25 l of glutathione-agarose beads was incubated for 2 h at room temperature with 10 g of control GST or GST-FKBP in 0.17 M NaCl, 20 mM Tris-HCl, pH 7.4, 0.1% CHAPS, 2 mM DTT, and 0.3 M sucrose (buffer B) with gentle mixing. After washing twice with buffer B, the beads were incubated for 1 h at 37°C with 1 g of the purified RyR1 or RyR3 in buffer B. The beads were washed three times with buffer B and proteins bound to the beads were subjected to SDS-PAGE.
[ 3 H]Ryanodine Binding Assay-To determine with high sensitivity the [ 3 H]ryanodine binding activity with a small amount of RyR3 protein, we separated protein-bound ryanodine from the free ligand using a small scale gel filtration column (Centri-sep) in a centrifuge. The purified RyR3 was incubated with 8.5 nM [ 3 H]ryanodine for 5 h at 25°C in 30 -50 l of a binding buffer containing 0.17 M NaCl, 10 mM MOPSO/ NaOH, pH 6.8, 2 mM DTT, 1% CHAPS, 0.5% egg lecithin, 4 mM AMP and various concentrations of Ca 2ϩ buffered with 10 mM EGTA. Free Ca 2ϩ concentrations were calculated using the value of 8.79 ϫ 10 5 M Ϫ1 as the apparent binding constant for Ca 2ϩ of EGTA (22). Then, an aliquot of the sample (10 -20 l) was applied to the Centri-sep column that had been equilibrated with 1 M NaCl, 10 mM MOPSO/NaOH, pH 6.8, 1% CHAPS, 0.5% egg lecithin, 2 mM DTT, and 0.1 mM CaCl 2 . The radioactivity of the eluate was determined by a liquid scintillation counter. Nonspecific radioactivity was determined in the presence of 50 M nonradioactive ryanodine. By using this assay system, we were able to determine the ryanodine binding activity with only 20 ng of the purified RyR protein, which is one-tenth the minimum amount for the conventional filtration assay. [ 3 H]Ryanodine binding to the purified RyR1 was determined under the identical condition to the filtration assay with polyethyleneimine-treated glass filters as described (10).
Electron Microscopy-The structure of RyR molecules was examined by electron microscopy after negative staining. It was necessary to completely eliminate phospholipid and to reduce detergent from the preparation, since their presence hampered the visualization of structural details of the proteinaceous receptors. To meet such requirements, the materials for the structural study were treated as follows. The RyR3 immunoprecipitated by anti-RyR3 beads was washed with a buffer containing 0.1% CHAPS, instead of 1% CHAPS and 0.5% egg lecithin, and eluted by the RyR3-peptide. Then, the peptide was removed by passing the solution through a Centri-sep column pre-equilibrated with a buffer containing 0.5 M NaCl, 5 mM sodium phosphate (pH 7.2), 0.1 M sucrose, 0.1% CHAPS, and 2 mM DTT (buffer C). For RyR1, the initial buffer was exchanged with buffer C by gel filtration on a Superose 6 column (Amersham Pharmacia Biotech). Those procedures did not deteriorate the [ 3 H]ryanodine binding activity of RyRs unless the preparations were left standing for a long period. For examination by electron microscopy, 2-3 l of the sample was applied to fresh thin carbon film over a 200-mesh copper grid. After 3 min, the residual solution was thoroughly rinsed off with the above buffer without CHAPS, and the sample was stained negatively with 1% uranyl acetate containing bacitracin (23). Specimens were set in a side entry goniometer for the JEOL 2000EX electron microscope with the carbon side facing upward. Stereo-paired micrographs were taken by tilting Ϯ10°at 80 kV acceleration voltage (23).
Location of the D2 segment in the RyR1 architecture was determined similarly by negative staining but with use of a specific anti-D2-region polyclonal antibody as a probe. Since the dimension of the IgG molecule is quite small and thin (ϳ150 kDa) as compared with the gigantic RyR molecule (ϳ2,200 kDa) standing upright on the carbon-support, it was anticipated that the IgG probe bound to the receptor molecule might be hard to identify by itself, in the thick stain layer embedding the total receptor assembly. Thus, we used a two-step procedure (24,25) to search for the epitope site of the antibody, tentatively with IgG conjugated with a probe large enough to be easily found, and later with the antibody by itself, to finally pinpoint the precise site. The antibody against the D2 region (amino acid sequence 1358 -1413) of RyR1 was produced with rabbits using GST-fusion protein as an antigen. A cDNA fragment (residues 4073-4239) was obtained from a cassette pBS-RyR1cs4 (26) and cloned into the pGEX-3X vector (Amersham Pharmacia Biotech). Expression and purification of the GST-fusion protein was carried out according to the manufacturer's instructions. The IgG was affinity-purified by the RyR1-bound PVDF membranes (10). This anti-body stained a single band of RyR1 on Western blot analysis of SR vesicles which, in turn, were specifically immunoprecipitated by the same antibody (data not shown). In order to produce a probe with a large marker, the antibody was biotinylated and conjugated with oligomeric avidin, a linear rod-like assembly. Biotinylation of the antibody was carried out by incubation of the antiserum with sulfo-NHS-LCbiotin (Pierce). An oligomeric avidin was prepared by addition of a divalent biotin (generous gift from Dr. Kazuo Sutoh to E. K.) to neutravidin (Pierce) at a molar ratio of 2:1 (divalent biotin/neutravidin) (27) and was mixed with the biotinylated antibody before use. For negative staining, purified RyR1 receptor was first applied onto a carbon film and left standing for 3 min. The grid was briefly washed and the solution containing the antibody with or without avidin conjugate was added. After 3-5 min incubation period, it was exhaustively washed with the buffer and stained negatively with uranyl acetate, as described above.
Planar Lipid Bilayer Experiments and Single Channel Data Acquisition-Single channel recordings were carried out as described previously (28). Briefly, lipid bilayer consisting of a mixture of L-␣-phosphatidylethanolamine, L-␣-phosphatidyl-L-serine, and L-␣-phosphatidylcholine (5:3:2 by weight) in decane (40 mg/ml) was formed across a hole (ϳ200 m in diameter) in a polystyrene partition separating two chambers referred to as cis (volume 1 ml) and trans (volume 1.5 ml). The trans chamber was held at virtual ground potential, and the cis chamber was voltage-clamped at Ϫ40 mV relative to the ground, unless noted otherwise. Incorporation of the purified RyR channel was performed in asymmetrical KCl solutions containing 500:50 mM (cis/ trans) KCl, 20 mM HEPES/Tris, pH 7.4, and 0.1 mM CaCl 2 . The protein was added to the cis chamber. After confirming the channel incorporation by the occurrence of flickering currents, further incorporation of the protein was prevented by supplement of an aliquot of 3 M KCl dissolved in 20 mM HEPES/Tris to the trans chamber. Single channel currents were recorded in symmetrical solutions containing 500 mM KCl, 20 mM HEPES/Tris, pH 7.4, and various concentrations of free Ca 2ϩ buffered with 1 mM EGTA. Free Ca 2ϩ was calculated using the apparent binding constant of EGTA for Ca 2ϩ by Harafuji and Ogawa (22) and was confirmed by potentiometry with a handmade ETH1001based Ca 2ϩ electrode. Only bilayers containing a single channel were used in this study. Experiments were carried out at room temperature (18 -22°C). We found that the purified RyRs could be incorporated into the bilayers in either orientation. The sidedness of the single channel was determined by response to Ca 2ϩ , because the RyR channel sensitively responded to cytoplasmic Ca 2ϩ . About 90% of the RyR channels was blocked by lowering the cis free Ca 2ϩ to 10 -100 nM with EGTA, indicating that the cytoplasmic side of most channels faced the cis chamber.
Single channel currents amplified by an Axopatch 1D patch clamp amplifier (Axon Instrument, CA) were displayed on an oscilloscope, filtered at 1 kHz using an eight-pole low-pass Bessel filter, and digitized at 5 kHz for analysis. Data were saved on the hard disk of an IBM personal computer. P o and the lifetime of open and closed events from records of duration Ͼ2 min were calculated by 50% threshold analysis using pClamp (version 6.0.4) software. The results were presented as means Ϯ S.E.

RESULTS
Purification of RyR3 from Rabbit Diaphragm SR-RyR3 was purified from rabbit diaphragm using immunoprecipitation with an antibody specific to RyR3 (anti-RyR3) (12,13) which was raised against a synthetic peptide corresponding to residues 4375-4387 of rabbit RyR3 (RyR3-peptide) (see "Experimental Procedures"). The diaphragm was found to express about 10-fold more abundant RyR3 than brain in rabbits (13). As shown in Fig.  1A, SR protein prepared from rabbit diaphragm contained a large amount of RyR1 that was easily seen on Coomassie Brilliant Blue-stained gel and positively detectable on Western blot with anti-RyR1 monoclonal antibody XA7 and a low level of RyR3 that was identified on anti-RyR3 blot. After solubilization with CHAPS, SR proteins were incubated with anti-RyR3-agarose beads to precipitate RyR3 (12,13). The band for RyR3 was clearly detected in the precipitate on both protein staining and anti-RyR3 blot (Fig. 1B, left lanes). Addition of 30 M RyR3-peptide during incubation failed to precipitate RyR3 (Fig. 1B, right lanes), indicating specific interaction of RyR3 with the anti-RyR3. RyR1 band was detected neither on protein staining (Fig.   1B, left panel) nor on anti-RyR1 blot (data not shown, but see Fig.  2B). RyR3 was thus effectively concentrated and separated from the other SR proteins. The immunoprecipitated RyR3 was then dissociated from the antibody by incubation with excess RyR3peptide. Preliminary attempts of treatment with 0.1 M glycine HCl (pH 2.8) or 5 M potassium thiocyanate, which is conventionally used to release antigen from antibody, caused total loss of [ 3 H]ryanodine binding activity, indicating irreversible degeneration of RyR3 (data not shown). As shown in Fig. 1C, RyR3 (arrowhead) was detected in supernatant after incubation with 30 M RyR3-peptide, whereas it did not come out from the beads without the peptide. Finally, the pure RyR3 was eluted from a small gel filtration column that held therein the RyR3-peptide. By this procedure, 0.5-0.7 g of RyR3 was purified from 10 mg of SR vesicles.
A single band of purified RyR3 moved slightly ahead of RyR1 on a Coomassie Brilliant Blue-stained SDS-polyacrylamide gel ( Fig. 2A). Western blots by isoform-specific antibodies are shown in Fig. 2B. Monoclonal antibody 34C, which can recognize all three mammalian RyR isoforms, clearly detected both RyR1 and RyR3. Anti-RyR1 antibody XA7 positively reacted with the purified RyR1 but did not detect any bands in the RyR3 preparation, indicating no contaminating RyR1 in the purified RyR3. Anti-RyR3 reacted with RyR3 but not with the purified RyR1. These findings confirm our previous results of homotetrameric formation of each isoform (13).
FKBP12 Binding of RyR3-It has been demonstrated that FKBP12 is tightly associated with RyR1 with a stoichiometry of 1 mol/mol of the monomer and modulates channel activity (24,29,30). To know whether RyR3 associated with FKBP12 as well, we carried out Western blotting using an antibody against FKBP12. As shown in Fig. 3A, rabbit diaphragm SR vesicles contained a band recognized with anti-FKBP12 monoclonal antibody 3F4 whose mobility was almost identical with human recombinant FKBP12 (hFKBP12) (lower panel). This corresponds to previous results that RyR1 associates with FKBP12 in skeletal muscles (29,31). In contrast, no band was detectable in either RyR1 or RyR3 preparation. Because the charged amount of the RyR in these experiments was comparable to that in the SR vesicles on the basis of the density of RyR bands (upper panel), negative immunoreactivity is not due to a small amount of charged proteins. These results indicate that FKBP12 may be dissociated from RyR1 or RyR3 during purification procedures. This is consistent with the results of Wagenknecht et al. (24) who reported a substantial loss of FKBP12 from the RyR1 preparation on addition of CHAPS.
To address whether RyR3 possesses an ability to bind FKBP12, we carried out FKBP12 binding assays using glutathione S-transferase (GST)-hFKBP12 fusion protein (GST-FKBP) (Fig. 3B). RyR1 (left panel) and RyR3 (right panel) were incubated with glutathione-agarose beads that had been incubated with GST-FKBP or control GST in a buffer containing 0.17 M NaCl and 0.1% CHAPS. After washing the beads, proteins bound to the beads were detected by SDS-PAGE. When incubated with GST-FKBP beads, RyR1 and RyR3 were bound to the beads (lane 2). When incubated with control GST beads (lane 1), in contrast, bound RyRs were hardly detectable. In the presence of 10 M FK506, which was known to dissociate FKBP12 from RyR1, RyRs were also severely prevented from binding the GST-FKBP beads (lane 3). These results strongly indicate that RyR3 can specifically bind FKBP12, as is true of RyR1.
Electron Microscopy of Purified RyR3- Fig. 4 exhibits nega-tively stained images of RyR3 particles, in comparison with those of RyR1. As shown in the general view (Fig. 4A), RyR3 preparation showed homogeneous distribution of the particles of rectangular shape similar to the reported image of RyR1. A gallery of the selected particles shows that both isoforms are indistinguishable in shape or in overall dimension (about 25 nm on each side) from each other (compare Fig. 4, B with C). They showed 4-fold symmetry reflecting their tetrameric nature, and the substructures were well defined under higher magnification. The central core with a crossed groove and outer regions with several cavities or pockets were the features commonly found in both isoforms, although closer examination revealed delicate differences in appearance from particle to particle, probably owing to uncontrollable differences in stain depth. It is also notable that both isoforms shared a pinwheellike appearance with the same rotational handedness. We also obtained several images of RyR3 particles with less contribution of phospholipid and/or detergent. Four pairs of stereo-micrographs for RyR3 (alternate columns in each row of Fig. 4D) show the tubular network-like structure forming several stain-filling pockets, which appears very similar to the published solid models of RyR1 (32,33). Despite careful comparison of the two isoforms, we could not find any significant difference between them which exceeds the image variance due to staining conditions. Thus, we concluded that the actual difference in the molecular architecture of the isoforms might be subtle, even if it exists.
We then switched to another approach to find the structural difference in a more positive way, using a structural probe specific to the segments unique to each isoform. One of such candidates is naturally the antibody used to isolate RyR3 in the present study. However, since the epitope of this antibody in the amino acid sequence of RyR is close to the membranespanning region, it might not be an easy task to identify the epitope site in the geometry of the total RyR assembly that is attached to the carbon support through the membrane-spanning region. So, we decided to use the other antibody probe which is not only easier to visualize but also specific to some functionally important region.
The fact that RyR3 lacks a stretch of about 100 amino acid residues corresponding to 1305-1405 in RyR1 is one of the most substantial differences between them (34). The region including this stretch is generally termed D2 because of its high divergency among RyR1-3 (35). As an appropriate candidate to search for the subtle difference between RyR1 and RyR3 in the local architecture, we made an attempt to define the location of the D2 region in the organized structure of the RyR1 assembly by negative staining. We utilized an antibody specific to the D2 region of RyR1 (anti-D2 antibody) that was produced against the amino acid sequence 1358 -1413 of rabbit RyR1 (see "Experimental Procedures"). Because this antibody could immunoprecipitate native RyR molecules in solution, it is clear that the antibody can attach to RyR without significantly affecting its organized architecture. In order to visualize unambiguously the bound antibody in negatively stained images, we initially used the antibody conjugated with an oligomeric avidin which might form a characteristic rod-like shape and would be easily distinguished from the other structures (27). Fig. 5A exhibits a general view of such specimen. Most of the receptor molecules in the field were accentuated with strongly stain-excluding dots and many of them with short rod-like structures (indicated by black arrows), both close to the corners of the square-shaped assembly. This indicates that the anti-D2 antibody should be associated with the RyR1 assembly at the sites close to the corner of the square. This probe, however, was too large to pin-point the precise epitope site on the assembly. With the knowledge that the D2-specific antibody might bind close to the corners of the RyR1, we then carried out similar experiments with the unconjugated antibody. Fig. 5B shows the images obtained in such way. After incubation with antibody solution and exhaustive washing, many receptor molecules appeared as square-shaped particles with small but strongly stain-excluding spots close to the corners. A substantial fraction of them also had the protrusions from the same position close to the corner of the square assembly. These spots often consisted of a triangular or tripodal structure that would represent the anti-D2 IgG molecule (free IgG molecules are shown in the insets as reference). They were reproducibly found at several nanometers apart counterclockwise from the exact corners of the rectangular receptor molecules. More convincing evidence for the location of the epitope site was the presence of cross-linked particles through fine Y-or V-shaped pieces, the dimension of which is similar to that of free IgG, connected to the sites equivalent to those suggested above as the epitope site. It is notable that the disposition of the two adjacent receptors seen here was opposite to the model motif II for self-association of the purified RyR which was suggested by Wagenknecht et al. (25). Hence, it seems unlikely that these connected receptors were due to self-association without the antibody. These lines of evidence strongly suggest that the epitope site of anti-D2 antibody, i.e. the D2 region, exists at the position several nanometers apart counterclockwise from the corner of the tetrameric receptor particles. It has been shown that RyR adheres almost exclusively with its transmembrane side toward the carbon film (36). Because the images in this paper are printed so that the receptor molecules are observed from their cytoplasmic side, the location of the D2 region may be safely interpreted to be at or close to "domain 9" in the three-dimensional archi- FIG. 4. Negatively stained images of purified receptor molecules. All the images are printed so that the particles are seen from their top surface, i.e. cytoplasmic side. A, general view of RyR3 preparation indicating the homogeneity of the purified molecules in their size (25 nm on each side) and overall square shape. Scale bar represents 50 nm. B and C, galleries of selected images of RyR3 and RyR1 molecules, respectively. Both receptor isoforms share a pinwheel-like appearance with the same rotational handedness. A crossed groove in the central core and the stain-filling pockets in the outer regions were also commonly observed and indistinguishable from each other. D, four pairs (alternate columns in each row) of stereo images of the RyR3 molecules including some oblique views. These particles seem to show the details of polypeptide folding probably because of less contribution of lipid and/or detergent in the polypeptide images. One might trace a complexed organization of the skeleton-like polypeptide chain similar to that visualized in the reconstructed three-dimensional model (32). tecture of RyR1 molecule (32), close to the top of the cytoplasmic assembly (see also the side view in Fig. 5B).
[ 3 H]Ryanodine Binding to the Purified RyR3-[ 3 H]Ryanodine binding to the purified RyR3 was determined by gel filtration method (see "Experimental Procedures"). The preliminary results indicate that properties of [ 3 H]ryanodine binding to the purified RyR3 are consistent with those already reported with the immunoprecipitated receptors (12,13); it was activated by micromolar Ca 2ϩ , adenine nucleotides, and caffeine, inhibited by millimolar concentrations of Ca 2ϩ and Mg 2ϩ , procaine, and ruthenium red, and affected by salt concentration of the medium (data not shown). In this study, we focused on some important characteristics that so far remained to be determined or were controversial. Fig. 6 shows dose-dependent [ 3 H]ryanodine binding to the purified RyR3. The amounts of the bound [ 3 H]ryanodine in the medium containing 1 M NaCl, 1 mM AMP-PCP, and 0.1 mM Ca 2ϩ increased with increases in [ 3 H]ryanodine concentrations and approached the asymptotic value. The Scatchard plot (inset) gave a straight line within the range of 0.5-21 nM [ 3 H]ryanodine, indicating a single class of high affinity ryanodinebinding sites with an apparent K D of 1.5 nM, which is similar to that of the immunoprecipitated RyR3 (1.6 nM) (13). In the previous study (13), we were unable to estimate the stoichiometry of ryanodine bound to RyR3, because we could not determine the amount of the receptor. The B max of the purified RyR3 (522 pmol/mg protein) shown here revealed that it binds [ 3 H]ryanodine with a stoichiometry of 1 mol of ryanodine per 1 mol of the homotetramer, as is true of the other isoforms (3). Based on the binding sites of [ 3 H]ryanodine binding to RyR3 in rabbit diaphragm SR (B max ϭ 0.065 pmol/mg) (13), the amount of RyR3 in 10 mg of the SR is calculated to be 1.2 g (10 ϫ 0.065 Ϭ 522 ϫ 1,000). We obtained 0.5-0.7 g of the purified RyR3 from 10 mg of the SR vesicles. These results show that FIG. 5. Location of the D2 segment in the molecular architecture of RyR1 probed with anti-D2 antibody. A, a general view of the purified RyR1 molecules mixed with the anti-D2 antibody which was conjugated to oligomeric avidin. Antibody probes are observed typically as short stain-excluding rods and sometimes merely as spots, both attached near the corners of the quatrefoil structure (receptor particles with rods are indicated by black arrows). Scale bar exhibits 40 nm. B, gallery of the negatively stained images of the purified RyR1 complexed with the unconjugated anti-D2 antibody. Note that small stain-excluding particles (arrowheads) are consistently bound to the edge of the receptor at the identical site close to the corner. These particles look similar in shape and size to those of the free IgG molecules (two small insets). The images indicated in the panels of the right column exhibit RyR molecules that are cross-connected by small V-or Y-shaped pieces corresponding to IgG. The IgG probe is also seen bound near the other corner of the molecules. It should be noted that the cross-linked sites are equivalent to those decorated by the spots shown in the panels of the left column. The bottom panel of the left column of Fig. 5B represents the cross-linked receptors whose side and oblique views are seen. Antibody binding occurs at the same site as above. Scale bar represents 20 nm. the yield of RyR3 was 42-58% and that the extent of purification was more than 8,000-fold.
We previously reported that RyR3 in rabbit diaphragm showed unique characteristics with regard to sensitivity to Ca 2ϩ and Mg 2ϩ ; it was less sensitive to activating Ca 2ϩ than RyR1 and resistant to inhibition by Mg 2ϩ (13). Since these conclusions were obtained from the results of [ 3 H]ryanodine binding with the immunoprecipitated receptors, one might assume that binding of anti-RyR3 would modify the cation sensitivity of RyR3. We therefore determined Ca 2ϩ and Mg 2ϩ sensitivities of RyR3 with the purified receptors. Fig. 7A shows the Ca 2ϩ dependence of the [ 3 H]ryanodine binding to the purified RyR1 and RyR3 in an isotonic medium containing 0.17 M NaCl. The Ca 2ϩ sensitivity of RyR3 for activation (open circles) (EC 50 ϭ 14 M) was 4-fold lower than that of RyR1 (filled circles) (EC 50 ϭ 3.5 M). On the other hand, it was apparent that both receptors were effectively inactivated by millimolar Ca 2ϩ ; IC 50 values were calculated to be 1.9 and 2.9 mM for RyR1 and RyR3, respectively. These results confirm the previous conclusion (13) that RyR3 shows a lower sensitivity to activating Ca 2ϩ than RyR1, whereas they are similar with regard to the Ca 2ϩ sensitivity in inactivation.
The effect of Mg 2ϩ on [ 3 H]ryanodine binding of the purified RyR3 is shown in Fig. 7B. The amount of binding to RyR3 that had been activated by 100 M Ca 2ϩ was dose-dependently reduced by Mg 2ϩ with IC 50 of 2.1 mM, which was similar to that of RyR1 (2.5 mM). The IC 50 value for Mg 2ϩ was consistent with that for inactivating Ca 2ϩ (see Fig. 7A). Because dose-dependent inhibition by Mg 2ϩ was also observed on single channel recordings (Fig. 9C), it is concluded that RyR3 is as effectively inhibited by Mg 2ϩ as RyR1. This is in contrast to our previous results with the immunoprecipitated receptors (13). It is unlikely that binding of anti-RyR3 antibody may alter the sensitivity to Mg 2ϩ of RyR3, because the inhibition by Mg 2ϩ of the purified RyR3 was unchanged, irrespective of the presence or absence of the antibody (data not shown). The reason for the discrepancy remains to be solved.
Single Channel Recordings of the Purified RyR3 Channel-Single channel currents through the purified RyR3 were recorded in a solution containing symmetrical 500 mM KCl, 20 mM HEPES/Tris, pH 7.4. As shown in the right column in Fig.  8A, the RyR3 channel was activated by exposure to cis pCa 6.7 (P o ϭ 0.236). The P o remained unchanged at voltages between  Fig. 8A. It should be noted that the subconductance opening was not as common as claimed to be characteristic of FKBP12-free RyR, although the frequency of subconductance opening was only slightly higher with the purified RyR1 than that with the receptor in SR vesicles. The current-voltage plot for RyR3 (open circles in Fig. 8B) gave a linear relationship between Ϫ60 and ϩ60 mV, as is true of RyR1 (filled circles in Fig. 8B). This indicates the non-rectifying property of either channel. The conductance of the RyR3 channel ranged from 719 to 794 pS with an average of 743 Ϯ 11 pS (mean Ϯ S.E., n ϭ 6). This value was significantly larger than 630 Ϯ 9 pS (n ϭ 8) for the RyR1 channel (ranging from 582 to 669 pS), when K ϩ was used as current carrier.
The gating kinetics for RyR1 and RyR3 channels Modulation of the RyR3 channels by Ca 2ϩ , caffeine, ATP, Mg 2ϩ , ryanodine, and ruthenium red is demonstrated in Fig. 9. At 0.1 M Ca 2ϩ (control), the RyR3 channel showed no open events (Fig. 9A). The channel was partially activated by 0.2 M cis Ca 2ϩ . Addition of 1 mM caffeine increased P o from 0.007 to 0.101, being consistent with the result of [ 3 H]ryanodine binding (12,13). ATP also enhanced the channel activity; P o increased from 0.007 to 0.498 by 1 mM ATP in the presence of 0.25 M free Ca 2ϩ (Fig. 9B) (Fig. 9C).   (Fig. 10A). The channel was further activated to a great extent (P o ϭ 0.817) at pCa 6.3 and showed almost full opening (P o ϳ1.0) over the wide range of pCa 6 -3. Around 3 mM Ca 2ϩ , on the contrary, the channel activity markedly declined (P o ϭ 0.191). P o values that were obtained from 4 to 7 separate experiments are plotted against varied cis Ca 2ϩ concentrations in Fig. 10C. At 0.1 M Ca 2ϩ , five of six determinations showed no open events, whereas one experiment showed P o ϭ 0.181. Thus, the threshold of Ca 2ϩ concentration for the RyR3 channel opening seemed to be around 0.1 M, and the channel was already fully activated at 1 M or more Ca 2ϩ . The pCa-P o relationship was very steep and appeared to be in almost all-or-none fashion. This Ca 2ϩ dependence is similar to the results reported with RyR2 (1, 2) as well as those with RyR3 which were determined by lipid bilayer experiments (14,37). The RyR3 channel was inactivated by Ca 2ϩ concentrations above 1 mM. The dose dependence of inactivation by high Ca 2ϩ appeared similar to that of inhibition by Mg 2ϩ (see Fig. 9).
We determined Ca 2ϩ dependence of the RyR1 channel activity to compare with that of the RyR3 channel (Fig. 10B). RyR1 channels demonstrated the following two distinct populations of single channel current fluctuations: high and low open probability (high and low P o channel, respectively), just as recently revealed with terminal cisterna of rabbit skeletal muscle SR (38,39). The high P o channel (7 of 19 determinations) was activated by cis Ca 2ϩ concentrations up to 10 M and inhibited at higher than 100 M Ca 2ϩ (right column in Fig. 10B). In contrast, the low P o channel (12 of 19 determinations) showed only a few open events even at the optimum Ca 2ϩ , although it showed biphasic Ca 2ϩ dependences (left column in Fig. 10B). P o values against pCa values for high P o (n ϭ 4 -7) and low P o (n ϭ 9 -12) channel are also plotted in Fig. 10C. Their Ca 2ϩ dependences appeared to be similar, although the relationship for low P o channel was less clear. Whereas the RyR3 channel showed homogeneous channel activity with nearly full opening (P o ϳ1), RyR1 displayed heterogeneous populations of the channel activity with high P o (ϳ0.3) and low P o (Ͻ0.05), respectively, at the optimum Ca 2ϩ . Ca 2ϩ concentrations that would give half the maximum activation were estimated to be ϳ1 and ϳ0.3 M for RyR1 and RyR3 channels, respectively. In the presence of 0.1 M Ca 2ϩ , interestingly, the high P o RyR1 showed a larger P o than the RyR3 channel.
It has been demonstrated that the RyR channel is modulated by the redox state of RyR molecules that could be attained by sulfhydryl-oxidizing and -reducing reagents (39 -43). We therefore examined effects of sulfhydryl reagents on single channel activity of RyR1 and RyR3 channels. Typical data are depicted in Fig. 11. Addition of 50 M DTT to the cis side of the high P o RyR1 channel led to a remarkable reduction in P o (from P o ϭ 0.217 Ϯ 0.061 in controls to P o ϭ 0.071 Ϯ 0.046, n ϭ 7; see Fig.  11A for a typical result). Increase in DTT to 500 M induced a further decrease in P o (P o ϭ 0.035 Ϯ 0.002, n ϭ 7). All DTTtreated channels were modified by 10 M ryanodine and blocked by 5 M ruthenium red (data not shown). The channel activity inhibited by application of 100 -500 M DTT (from P o ϭ 0.104 Ϯ 0.051 in controls to P o ϭ 0.002 Ϯ 0.001, n ϭ 3) was overshot to P o ϭ 0.313 Ϯ 0.006 by subsequent cis addition of 100 -500 M p-chloromercuriphenylsulfonic acid (pCMPS), a specific organic sulfhydryl reagent (bottom traces in Fig. 11A). On the other hand, the low P o RyR1 channel was activated by the addition of 10 -50 M cis pCMPS (from P o ϭ 0.004 Ϯ 0.001 in controls to P o ϭ 0.116 Ϯ 0.034, n ϭ 9), which was similar to the high P o state (Fig. 11B). Subsequent exposure of such pCMPS-activated channel to 25-50 M DTT reversed the P o to a low level (0.019 Ϯ 0.008, n ϭ 4). In 3 of 14 experiments, cis-pCMPS elicited irreversible closure after transient increase in P o , which could not be recovered by any ligands including a large amount of DTT (data not shown), suggesting the existence of at least two different sulfhydryls associated with activation and inactivation of the channel in the RyR1 molecule. Similar results on sulfhydryl reagents were observed with RyR channels in frog skeletal muscle SR (43). The conversion between low P o and high P o channels occurred immediately after addition of these reagents. During channel activation or inactivation by these sulfhydryl-modifying reagents, the channel conductance remained unchanged. These findings suggest that the two populations of the RyR1 channel showing different P o against pCa values might reflect the extent of their redox state as follows: the high P o channel at a more oxidized state and the low P o channel at a more reduced state. Fig. 11, C and D, exhibits the effects of sulfhydryl-oxidizing and -reducing reagents on the RyR3 channels. When 250 M DTT was added to the cis side, four of five RyR3 channels that had been activated by 0.15-0.3 M cis-Ca 2ϩ (P o ϭ 0.466 Ϯ 0.175, n ϭ 5) suddenly switched to the closed state after a latent period (2-10 min) (Fig. 11C). In some channels, the addition of DTT resulted in alternation between the burst of opening and the closed state in the duration of several tens of seconds to minutes (data not shown). One channel in five, however, did not respond to DTT even at 1 mM (data not shown). Subsequent addition of 100 M pCMPS to channels that had been inhibited by DTT treatment recovered usual current fluctuations with high P o (0.543 Ϯ 0.161, n ϭ 3). The RyR3 channel showing low P o (0.054 Ϯ 0.026, n ϭ 4) at 0.13-0.3 M Ca 2ϩ , on the other hand, was markedly activated by the addition of 10 M pCMPS to the cis side to high P o (0.599 Ϯ 0.192) (Fig. 11D). Thus, the effect of sulfhydryl reagents on the RyR3 channel was similar to that on the RyR1 channel; a sulfhydryl-oxidizing reagent activates the channel activity, whereas a sulfhydryl-reducing reagent inhibits it. The channel activity of RyR3, however, appears to depend steeply on the redox state showing almost all-or-none fashion, whereas the dependence is more gradual with RyR1. This may be reflected in the observation that the RyR3 channel showed no opening in the closed state after addition of DTT, whereas some opening was still detected with RyR1 even at P o Ͻ 0.001.

DISCUSSION
In this study, we purified homotetrameric RyR3 from rabbit diaphragm by immunoprecipitation with highly specific antibody to RyR3 and by dissociation under gentle conditions using an epitope peptide. The purified RyR3 demonstrated the characteristic structure of a quatrefoil under electron microscopy. It also showed Ca 2ϩ -dependent high affinity [ 3 H]ryanodine binding and cation channel activity. Thus, RyR3 purified by this procedure retained structural and functional integrity, which enabled us to characterize the protein.
Our recent studies with the immunoprecipitated RyR3 indicate that RyR3 forms a homotetramer of a single polypeptide with a mobility slightly larger than that of RyR1 on SDS-PAGE (12,13). It showed Ca 2ϩ -dependent high affinity [ 3 H]ryanodine binding which was modulated by several ligands, indicating that it constitutes a Ca 2ϩ release channel (12,13). The results shown here with the purified receptor confirmed most of these previous results. In the previous results, [ 3 H]ryanodine binding was partially inhibited by Mg 2ϩ (13). As shown in Fig. 7B in this report, however, it was fully inhibited by Mg 2ϩ , and its IC 50 was similar to that for Ca 2ϩ . There is no satisfactory explanation for this discrepancy at the present time. Furthermore, several new findings were added. RyR3 was free of FKBP12, although it retained the ability to bind FKBP12. This is also the case with RyR1. Observations with negative staining electron microscopy revealed that RyR3 forms a characteristic structure with 4-fold symmetry, which is morphologically very similar to RyR1. RyR3 had a single class of [ 3 H]ryanodinebinding sites with a stoichiometry of 1 mol of [ 3 H]ryanodine per 1 mol of a homotetramer. When incorporated into planar lipid bilayers, RyR3 displayed a cation channel current with a large conductance, which is activated by micromolar Ca 2ϩ , caffeine, and ATP, inhibited by millimolar concentrations of Ca 2ϩ , and Mg 2ϩ , and ruthenium red, and modified by ryanodine. Thus, RyR3 constitutes a homotetrameric Ca 2ϩ release channel that shares several basic properties with RyR1.
In addition to the overall similarities as mentioned above, some functional differences between RyR1 and RyR3 were also noted. First, RyR3 displays a significantly larger monovalent cation conductance (743 Ϯ 11 pS in 500 mM KCl) than RyR1 (630 Ϯ 9 pS) (Fig. 8). Second, the purified RyR3 channel differs in gating kinetics from the RyR1 channel. The RyR3 channel showed longer durations of channel opening than the RyR1 channel under similar P o values as reflected in a larger fraction of the component of the slow open time constant ( o2 ) (Fig. 8).
Single channel recordings of RyR3 have recently been reported with recombinant rabbit uterine RyR3 that was expressed in HEK293 cells (37). The properties of the purified RyR3 channel shown here well correspond to those of the recombinant RyR3 channel in several respects as follows: (i) a large monovalent cation conductance, pendence for channel activation and inactivation. Our results are also consistent with more recent reports by Jeyakumar et al. (14) using RyR3 channels isolated from bovine diaphragm.
It was shown that FKBP12 tightly bound to RyR1 with a stoichiometry of 1 mol/mol of the RyR1 monomer and affected the function (24,29,30,44). We found that FKBP12 was associated with SR vesicles isolated from rabbit diaphragm but was not detected either in RyR1 or RyR3 preparations (Fig. 3). Wagenknecht et al. (24) recently reported that FKBP12 was considerably dissociated from CHAPS-solubilized RyR1 preparation. Considering the abundance of RyR1 in the SR vesicles (Ͼ99%), these results indicate that FKBP12 may be dissociated from RyR1 during solubilization and purification procedures. A binding study with GST-FKBP12 fusion protein revealed that RyR3 can bind FKBP12, as is true of RyR1 (Fig. 3). Because FKBP12 is widely and abundantly expressed in mammalian tissues including skeletal muscles (31), this suggests that RyR3 may bind FKBP12 in situ. It should also be noted that functional properties of RyR1 and RyR3 shown here were determined with receptors lacking FKBP12. It has been reported that removal of FKBP12 from RyR1 makes the channel unstable, resulting in increase in the number of openings to subconductance levels (30). Although we observed channels showing subconductance states in both RyR1 and RyR3 preparations, populations of such channels were minor (see Figs. 8 -11). We are currently investigating the effect of FKBP12 on the RyR3 channel activity.
With the hope of correlating the functional differences of RyR1 and RyR3 to the specific structural features, we examined the negatively stained images of the two isoforms. Although we could not recognize any significant differences in their submolecular architecture, we found that a site-directed antibody against the D2 region of RyR1 which is absent in RyR3 binds to domain 9 or its vicinity (Fig. 5). Considering the three-dimensional architecture of the receptor molecule, the epitope site may correspond to the "clamp-shaped" domain according to the terminology of Orlova et al. (45), in which they found the occurrence of certain conformational change accompanied with opening and closing events of the channel. Several attempts were recently made to determine the interaction sites of the receptor molecules with some specific proteinaceous ligands, e.g. calmodulin (46) and FKBP12 (24). Although such studies have successfully pin-pointed the binding sites of those proteins in the three-dimensional architecture of the receptor molecule, none of them has determined the position of those sites in the primary amino acid sequence. Our result thus provides the first case to link the three-dimensional architecture of RyR with its primary structure.
Since the distance between two adjacent voltage sensors of a tetrad is almost equal to the dimension of a "foot" structure, it has been postulated that the corners of a quatrefoil including domain 9 or clamp-shaped domain might interact directly with dihydropyridine receptor of the T-tubules (45,47,48). On the other hand, the crucial role of the D2 region in the E-C coupling of skeletal muscle was claimed based on the fact that the deletion of the D2 region causes the functional loss of Ca 2ϩ release activity triggered by the electrical stimulation while preserving caffeine-induced Ca 2ϩ release (49). Our assignment of the D2 region to domain 9 or its vicinity may reasonably link those two issues and could suggest the importance of the D2 region in the interaction of RyR1 with dihydropyridine receptor. Recently, two additional regions of RyR1 were also shown to interact with dihydropyridine receptor (50,51). It is also of interest where these stretches are localized in the threedimensional structure of the RyR molecule since these regions are relatively close to the D2 region in the primary structure of the receptor.
It has been demonstrated that a number of sulfhydryl-oxidizing reagents activate RyR channels (40,41,43). Sulfhydrylreducing reagents such as glutathione and DTT have been shown to reduce the RyR channel activity (39,42). On single channel recordings, we found that a sulfhydryl-oxidizing reagent pCMPS activated both RyR1 and RyR3 channels, whereas a sulfhydryl-reducing reagent DTT inhibited them (Fig. 11). The channel activity stimulated by the oxidizing reagent was reversed by the subsequent addition of the reducing reagent, and vice versa. Transition between the two states is reversible. We observed two populations of channel activity (high P o and low P o channel) in RyR1 preparation (Fig. 10). Multiple populations of RyR1 channel activity were also reported with SR vesicles prepared from skeletal muscles (38,39). Because the channel activity of RyR is dependent on the redox state of the receptor molecule (Fig. 11), distinct populations of RyR1 channels may be attributed to the heterogeneity of the redox state of the receptor in the lipid bilayer, as recently suggested by Marengo et al. (39). In contrast, only a single population of current fluctuations with P o ϳ1 was detected in RyR3 preparation (Fig. 10). Consistently, no heterogeneity of RyR3 channels was so far reported (14,37). Homogeneous channel activity was also shown in cardiac RyR (RyR2) (38). These results indicate that RyR3 may be much more steeply dependent on the redox state in the channel activity than RyR1, showing almost all-or-none fashion. This may be reflected in homogeneous channel activity of RyR3 in contrast to heterogeneous population of RyR1.
The results of the Ca 2ϩ sensitivity of RyR3 for activation and inactivation were so far controversial (13,37,52). We determined the Ca 2ϩ dependence of the purified RyR3 both in [ 3 H]ryanodine binding and in single channel recordings, and surprisingly we found that they were entirely different. EC 50 of RyR3 for Ca 2ϩ in [ 3 H]ryanodine binding was about 14 M which was about 4-fold larger than that of RyR1 (Fig. 7A), confirming the previous results with the immunoprecipitated RyR3 (13). This is consistent with the CICR activity of RyR1knock-out (dyspedic) mice skeletal muscle which is probably attributed to RyR3 (52). In contrast, the RyR3 channel in single channel recordings was steeply activated to P o ϳ1 between 0.1 and 1 M Ca 2ϩ in almost all-or-none fashion (Fig. 10C). A very similar relationship was obtained with the recombinant rabbit uterine RyR3 channel (37) and the RyR3 purified from bovine diaphragm (14). Although a higher salt concentration in the channel recordings (500 mM KCl) than in the ryanodine binding (170 mM NaCl) might be a possible cause, it is unlikely because similar results of the single channel experiments with the recombinant RyR3 were obtained in relatively low salt concentration (250 mM KCl) (37). These findings indicate that the difference in Ca 2ϩ dependence is simply attributed to the method adopted. RyR3 channel showed almost full opening (P o ϳ1) at 10 M Ca 2ϩ in the absence of adenine nucleotide or caffeine (Fig. 10) (37). [ 3 H]Ryanodine binding activity at the same Ca 2ϩ concentration, in contrast, was further stimulated by adenine nucleotides or caffeine (13). It was so low in the presence of Ca 2ϩ alone that a supplement of 4 mM AMP was required to obtain the precise results (Fig. 7A). Dithiothreitol (2 mM), on the other hand, was included in the medium for ryanodine binding but not for channel recordings. The low P o activity of RyR1 appears to correspond well to [ 3 H]ryanodine binding in the presence of Ca 2ϩ alone. Marengo et al. (39) reported that the Ca 2ϩ -dependent channel activity of RyR was sensitively dependent on the redox state. We observed that the channel activity of RyR3 was likely to be varied in almost all-or-none fashion by the redox state, whereas that of RyR1 was gradually dependent. The steep increase in the channel activity of RyR3 between 0.1 and 1 M Ca 2ϩ (Fig. 10C) may indicate that some Ca 2ϩ -dependent process(es) could also be involved in the abrupt transition of the channel activity in RyR3. This steep dependence may be a probable explanation for the difference in the Ca 2ϩ dependence.