Purification and Characterization of Ryanodine Receptor 3 from Mammalian Tissue*

The ryanodine receptors are intracellular Ca2+ release channels that play a key role in cell signaling via Ca2+. There are three isoforms. Isoform 1 from skeletal muscle and isoform 2 from heart have been characterized. Isoform 3 is widely distributed in many mammalian tissues although in minuscule amounts. Its low abundance has hampered its study. We now describe methodology to isolate mammalian isoform 3 in amounts sufficient for biochemical and biophysical characterization. Bovine diaphragm sarcoplasmic reticulum fractions enriched in terminal cisternae containing both isoforms 1 (>95%) and 3 (<5% of the ryanodine binding) served as starting source. Isoform 3 was selectively immunoprecipitated from the 3-[(3-cholamidopropyl)-dimethylammonio]-1-propanesulfonic acid (CHAPS)-solubilized fraction and eluted with peptide epitope. Isoform 3 thus prepared is highly purified as characterized by SDS-polyacryamide gel electrophoresis, Coomassie Blue staining, and by high affinity ryanodine binding. The purified isoform 3 was incorporated into planar lipid bilayers, and its channel properties were studied. Channel characteristics in common with the other two isoforms are slope conductance, higher selectivity to Ca2+ versusK+ (P Ca/K ∼6), and response to drugs and ligands. In its response to Ca2+ and ATP, it more closely resembles isoform 2. The first two-dimensional structure of isoform 3 was obtained by cryoelectron microscopy and image enhancement techniques.

In multicellular organisms, Ca 2ϩ stored in internal compartments is mobilized in response to external signals by way of intracellular calcium release channels (ICRC). 1 There are two subclasses of ICRCs, the ryanodine receptors (RyRs) and the inositol 1,4,5-trisphosphate receptors, with three isoforms of each subclass (1)(2)(3)(4)(5)(6). The genes encoding the three isoforms of the RyR subclass are located in different chromosomes, and each isoform shares about 2 ⁄3 sequence homology (4,7).
The RyRs were first discovered in striated muscle (8,9) as the Ca 2ϩ release machinery that released Ca 2ϩ from intracellular stores to trigger muscle contraction (1)(2)(3)(4)(5)(6). In excitationcontraction coupling in skeletal muscle, all of the Ca 2ϩ is released from the sarcoplasmic reticulum via ryanodine receptor isoform 1 (RyR1). This macroscopic phenomenology is referred to as "depolarization-induced Ca 2ϩ release," as Ca 2ϩ release from internal stores is triggered by T tubule membrane depolarization to be contrasted with "Ca 2ϩ -induced calcium release" in heart. In the cardiomyocyte, extracellular Ca 2ϩ must first enter via voltage-gated Ca 2ϩ channels in the plasma membrane, which in turn activates the ryanodine receptor isoform 2 (RyR2) to release most of the Ca 2ϩ for cell signaling. Purified RyR1 and RyR2 are available in substantial quantity, which has permitted their detailed characterization. The RyRs have a mass of 2.3 million and are the largest channel structures known. Recently, FK506-binding protein 12 (FKBP12) has been found to be associated with the RyR1 in a stoichiometry of 4. RyR1 is a hetero-oligomer with the structural formula (RyR1 protomer) 4 (FKBP12) 4 . The cardiac RyR2 receptor is likewise associated with four FKBPs, albeit with a different isoform, FKBP12.6 (10,11). The three-dimensional structures of RyR1 and RyR2 have been deduced by electron microscopy using quantitative image analysis (12)(13)(14)(15). The receptors have 4-fold symmetry consistent with their subunit stoichiometry.
It is now known that the ICRCs are widely distributed in tissues and that the three RyR isoforms are found in a diversity of tissues including the nervous system, epithelia, and smooth muscle (16 -23). Skeletal muscles from non-mammalian vertebrates, i.e. chicken, frog, and fish, have two ryanodine receptor isoforms, referred to as ␣ and ␤ isoforms (24 -26). Recently, the ␣ and the ␤ isoforms of chicken and frog were reported to be homologous with mammalian RyR1 and RyR3, respectively (27,28). The difficulty in isolating RyR3 from mammalian tissues is that, although widely distributed, RyR3 is present in minuscule amounts.
The use of cloning technology has revolutionized our approach to the study of membrane proteins, especially channels. Indeed, RyR3 was first described and the sequence was deduced from the cDNA (7) without having been isolated. However powerful the cloning technology, the primary structural and functional characterization of receptor proteins still requires their isolation and characterization. This study describes the isolation and characterization of RyR3 from bovine diaphragm SR. A preliminary report has appeared (29).

EXPERIMENTAL PROCEDURES
Keyhole limpet hemocyanin was obtained from Calbiochem. Sulfom-maleimidobenzyl-N-hydroxysuccinimide ester was from Pierce. Freund's complete and incomplete adjuvants were obtained from Life Technologies, Inc. [ 3 H]Ryanodine was obtained from NEN Life Science Products. Reagents for Western blot development were obtained as follows: goat anti-rabbit IgG alkaline phosphatase conjugate was from Rockland (Gilbertsville, PA); nitro blue tetrazolium and 5-bromo-4chloro-3-indolyl phosphate color development reagents were obtained from Promega (Madison, WI); Immobilon-P transfer membrane was from Millipore (Bedford, MA); sodium dodecyl sulfate, acrylamide, methylene bis-acrylamide, and SDS-PAGE prestained molecular weight markers were from Bio-Rad. Phospholipids were obtained from Avanti (Alabaster, WI). Immobilized protein A-Sepharose matrix was obtained from Amersham Pharmacia Biotech (Sweden); and immobilized protein G matrix was from Pierce. All the other reagents were obtained from Sigma or were reagent grade.
General Methods-The protein concentration of SR fractions was estimated by the method of Lowry (30), using bovine serum albumin as standard. SDS-PAGE was performed with a mini-slab gel apparatus (Hoeffer Scientific) using the buffer system described by Laemmli (31). Thick (1.5 mm) slab gels (5.0% resolving gel) were used for SDS-PAGE, and the samples were treated with sample loading buffer at room temperature for about 30 min.
Production and Purification of RyR3-specific Antibody-Polyclonal antibody against ryanodine receptor 3 (RyR3) was raised in rabbits using the synthetic peptide corresponding to the deduced amino acid sequences, 4326 -4336 (11 amino acids) of rabbit RyR3 (7). The epitope peptide with a cysteine residue added at the C terminus was linked to keyhole limpet hemocyanin, using the bifunctional agent, sulfo-m-maleimidobenzyl-N-hydroxysuccinimide ester. The general RyR antibody (number 8) was generated in rabbits using a peptide sequence that is conserved in each of the RyR isoforms (32). The synthetic peptide consisting of 20 amino acid residues with a cysteine residue added at the N terminus corresponds to 4681-4700 of RyR1 (7,33).
The immunizing dose of the epitope was 500 g/rabbit (weighing about 8 pounds). The primary and booster doses were injected at 1-month intervals, with Freund's complete and incomplete adjuvants, respectively. Subcutaneous injections at the back of the neck and intramuscular injections in the back and thigh muscles were given for primary and booster doses, and blood was collected 10 -11 days following each booster dose.
Sequence-specific polyclonal antibody was purified from the antiserum using an affinity matrix, prepared by conjugating the peptide epitope to cyanogen bromide-activated Sepharose B (34).
Preparation of SR Fractions Enriched in RyR from Bovine and Rabbit Diaphragm-SR fractions were prepared from bovine diaphragm muscle by the procedure of Saito et al. (35) as described for rabbit skeletal muscle TC, with the modification that the buffer conditions of Chamberlain et al. (36) were used. The supernatants from blendings 1 and 2 give rise to the I and R series of fractions (Table I). All the buffers used in the preparations contained the protease inhibitors, leupeptin (1.0 g/ml), aprotinin (1.0 g/ml), and phenylmethylsulfonyl fluoride (40 M). The SR fractions were resuspended in a small volume of resuspension buffer, quick-frozen in liquid nitrogen, and stored at Ϫ80°C.
Microsomes from fresh rabbit diaphragms were prepared using muscle/buffer ratio (1 g, 5 ml) as used for bovine diaphragm SR. A Polytron device was used for homogenization, first at low speed setting for 15 s, then at speed setting 6 for 60 s, and for a third time at setting 6 for 15 s. The homogenate was centrifuged at 11,000 ϫ g for 20 min, and the supernatant was filtered through four layers of cheesecloth. The pellet was rehomogenized twice at speed setting 6 for 30 s and processed like the first supernatant. The I and the R series microsomal pellets were obtained from the first and second supernatants after they were sedimented at 150,000 ϫ g for 1 h.
Immunoreactivity of RyRs in Bovine Diaphragm SR Fractions and Specificity of RyR3 Antibody-The immunoreactivities of bovine diaphragm terminal cisternae and the rabbit diaphragm microsomes were probed with both the general RyR (number 8) and RyR3-specific antiserum, respectively (1:1000). Specificity of RyR3 antibody was demonstrated by Western blot analysis, using TC of rabbit fast twitch skeletal muscle (35), canine cardiac microsomes (36), and bovine diaphragm terminal cisternae of SR (R 4 and/or I 4 fractions).
The RyR3-specific Antibody Is an Immunoprecipitating Antibody-Selective immunoprecipitation and quantitative estimation of RyR3 in bovine diaphragm TC was carried out in order to confirm that the RyR3-specific antibody is an immunoprecipitating antibody. Bovine diaphragm TC is enriched in ryanodine receptors and mainly contains RyR1 and some RyR3 (Ͻ5%). The RyR-enriched TC fraction, 6.0 mg/ml, was solubilized with CHAPS buffer (50 mM Tris-Cl, pH 7.4, 1.0 M NaCl, 2.0 mM DTT, and 0.5% CHAPS) at 4°C for 90 min with gentle mixing. The mixture was then centrifuged at 50,000 ϫ g for 30 min to remove insoluble material. The supernatant was diluted 4-fold with 50 mM Tris-Cl, pH 7.4, and incubated with protein A-Sepharose matrix (20-l bed volume) for 30 min at 4°C to remove nonspecific binding components. The protein A-Sepharose was equilibrated with preequilibration buffer (50 mM Tris-Cl, pH 7.4, 0.25 M NaCl, 0.5 mM DTT, and 0.125% CHAPS) for 30 min at 4°C with gentle mixing prior to its use. The supernatant obtained after sedimenting the protein A matrix at low speed was transferred to another container for overnight immunoadsorption with RyR3-specific antiserum (40 l), at 4°C with gentle mixing. Then, preequilibrated protein A-Sepharose (50 l) was added, and incubation was continued for another 4 h at 4°C with gentle mixing. The protein A matrix was then sedimented at low speed. The supernatant was removed, and the pellet was washed five times with preequilibration buffer. After washing, the pellet was solubilized in an equal volume of 2ϫ Laemmli sample loading buffer with boiling for about 5 min. The immunoprecipitate, the starting material, and the supernatant from immunoadsorption were characterized by SDS-PAGE and Western blot analysis using RyR3-specific antiserum ( Fig. 2A).
Immunoprecipitation of RyR3 was characterized by titration with increasing amounts of RyR3 antibody prebound to immobilized protein A/G. After immunoprecipitation of RyR3, sedimentation, and sequential washing as described under purification of RyR3 (see below), the immune complex was analyzed directly for ryanodine binding.
Immunoaffinity Purification of Ryanodine Receptor 3-Affinity purified RyR3-specific antibody was prebound to immobilized protein A/G matrix by incubating for 30 min at room temperature with gentle mixing. About 400 g of antibody was admixed with protein A or G matrix (about 200-l packed bed volume). RyR-enriched bovine diaphragm TC fractions (R 4 and/or I 4 ) were used as source for the purification of RyR3. The bovine diaphragm TC (ϳ100 mg of protein) was solubilized at 6.0 mg/ml in buffer A (20 mM Na 2 PIPES, pH 7.2, 0.6 M NaCl, 0.1 mM EGTA, 0.2 mM CaCl 2 , 5.0 mM Na 2 AMP, 2.0 mM DTT, and 0.6% CHAPS (w/v), 0.3% SBL) for 90 min at 4°C with gentle mixing. The mixture was then centrifuged at 50,000 ϫ g for 30 min at 4°C to remove insoluble material. The solubilized bovine diaphragm TC supernatant was diluted 4-fold with dilution buffer (buffer B) containing 20 mM Na 2 PIPES, pH 7.2, 0.1 mM EGTA, 0.2 mM CaCl 2 , 5.0 mM Na 2 AMP, 2.0 mM DTT, and 0.3 M sucrose. The mixture was then preabsorbed with 50 l of preequilibrated protein A/G matrix for 15 min at 4°C with gentle mixing to remove nonspecific binding components. After sedimenting to remove the matrix at low speed in the cold, the supernatant was transferred to another container, and RyR3 was selectively immunoadsorbed by incubating, with gentle mixing for 4 h at 4°C, with the appropriate amount of prebound-RyR3 specific antibody, prebound to protein A/G matrix. The prebound RyR3 antibody-protein A/G matrix was preequilibrated with buffer B (containing 0.15 M NaCl, 0.15% CHAPS, 0.075% SBL) for 30 min at 4°C before addition to the solubilized and diluted bovine diaphragm SR. The amount of antibody matrix used to bind RyR3 was determined by the ryanodine binding activity of the bovine diaphragm TC fraction. A 3-4-fold excess of antibody/total RyR binding equivalents was used. The immune complex containing the RyR3 was then sedimented at low speed, and the supernatant was used to purify RyR1 by heparin-agarose column chromatography. The immune complex was washed sequentially, first with buffer B containing 0.15 M NaCl, 0.5% CHAPS, and 0.25% SBL, second with buffer B containing 0.5 M NaCl, 0.5% CHAPS, and 0.25 SBL, and third with buffer B containing 0.1 M NaCl, 0.5% CHAPS, and 0.25% SBL. The elution of the RyR3 was carried out with buffer B containing 0.4 M NaCl, 0.6% CHAPS, 0.3% SBL, and excess of RyR3 epitope peptide (1.0 mg/ ml). The eluate was collected, and the pellet was again washed with elution buffer without epitope peptide (post-eluate). The eluate and post-eluate were quick-frozen in liquid nitrogen and stored at Ϫ80°C.
RyR1 from the supernatant after RyR3 immunoprecipitation was partially purified using a heparin-agarose column essentially as described for RyR2 (9). Buffer B (fortified additionally with 0.1 M NaCl and 0.15% CHAPS, 0.075% SBL) was used for preequilibration, and the linear gradient (0.1 M-0.8 M NaCl) elution buffer B also containing 0.5% CHAPS, 0.25% SBL. Fractions (3.0 ml each) were collected, and the ryanodine binding assay was carried out on 10-l aliquots of the fractions. The RyR1 fractions were quick-frozen and stored at Ϫ80°C. All the buffers used in the purification procedure contained the protease inhibitors, leupeptin (5 g/ml) and aprotinin (5 g/ml).

Characterization of Purified RyR3: SDS-PAGE and Western Blot
Analysis-Bovine diaphragm SR fractions enriched in terminal cisternae and column fractions from the immunoaffinity chromatography were analyzed by SDS-PAGE and visualized by Coomassie Blue staining. For Western blot analysis after SDS-PAGE, the proteins were transferred to an Immobilon-P membrane for 1 h at 24 mA constant current in blot transfer buffer (48 mM Tris, 39 mM glycine, 1.3 mM SDS, pH 9.2) using Trans-Blot SD Semi-Dry Electrophoretic Transfer Cell (Bio-Rad). The vacant binding sites on the membrane were blocked by incubating the membrane in wash buffer (10 mM Tris-Cl, pH 8.0, 0.55 M NaCl, and 0.05% Tween 20) containing 5% non-fat dry milk protein for 1 h. The membrane was probed with the RyR3-specific antibody in blocking buffer for 1 h, washed three times with wash buffer, and then incubated with secondary antibody (goat anti-rabbit IgG) conjugated to alkaline phosphatase in blocking buffer. The membrane was again washed three times with wash buffer and developed with nitro blue tetrazolium and 5-bromo-4-chloro-3-indolyl phosphate substrates.
Protein Determination-Small quantities of protein such as purified receptor were determined by scanning densitometry. Digital images of gels were obtained with a HP 6100P scanner (Hewlett-Packard), and Visioneer PaperPort 3.0.1 software (Visioneer Communications, Inc., Palo Alto, CA). Densitometry was carried out using the DNA ProScan A13 software (Technology Resources, Nashville, TN) on the RyR1 and RyR3 bands after separation in 7.5% resolving gel (SDS-PAGE) and staining with Coomassie Blue. Bovine serum albumin was used for protein calibration.
[ 3 H]Ryanodine Binding-Routine ryanodine binding was determined at 70 nM ryanodine. Ryanodine binding to particulate fractions made use of sedimentation to separate bound from free ryanodine as described previously, or for solubilized fractions by polyethylene glycol precipitation with equine gamma globulin as carrier protein (37,38). High affinity ryanodine binding parameters were measured by ryanodine binding isotherms using [ 3 H]ryanodine (ϳ15,000 cpm/pmol, obtained from NEN Life Science Products). Radioactivity was measured in a Beckman LS 5000TD scintillation counter. Nonspecific binding was measured in the presence of 20 M cold ryanodine. Data of the binding isotherms were analyzed by Sigmaplot 2.0 (Jandel Scientific, Corte Madera, CA).
For fusion of purified RyR3 to the bilayer, solubilized RyR3 (1-4 l) was added to the cis chamber while stirring. Fusion of channels was detected by the occurrence of Ca 2ϩ or K ϩ currents. [Ca 2ϩ ] free was adjusted using 1 mM 1,2-bis(2-amino-5-bromophenoxy)ethane-N,N,NЈ,NЈtetraacetic acid and/or EGTA, and various amounts of CaCl 2 , yielding the concentrations depicted (40,41). Data were filtered through a low pass Bessel filter at 0.8 -1 kHz and digitized at 2-5 kHz with an analog to digital converter (Digidata 1200, Axon Instruments) for computer analysis. Unitary conductances were estimated from the slope current/ voltage relationship at 0 mV. The relative calcium permeability of the channels for Ca 2ϩ versus monovalent ions (M ϩ : potassium and Tris) was estimated from the reversal potential of the currents (V cis reversal ), and relative permeability (P) was estimated using the Fatt-Ginsborg Equation.
where P Ca 2ϩ and P M are the permeability coefficients for Ca 2ϩ and monovalent ions through the channel, R is the gas constant, T is the temperature in K, and F is the Faraday constant (42). The mean open probabilities (P o ) of the channels at V cis ϭ 0 mV were used as an index of activity to study the response of the channels to the following known modulators of RyRs: ATP, caffeine, calcium, magnesium, ruthenium red, and ryanodine. In most cases, P o were calculated from the 50% threshold analysis. Cryoelectron Microscopy Studies-Cryoelectron microscopy was carried out as described previously (12,13) except in the manner in which the specimen was applied to the grid. Since RyR3 was isolated in the presence of 3 mg/ml phospholipid, which interferes with cryoelectron microscopy, the receptor was diluted into the same buffer lacking phospholipid. To minimize the time of exposure to low levels of lipid, the dilution was done on the specimen grid itself by first applying 4.5 l of the lipid-free buffer and then applying 0.5 l of RyR3 (0.2 mg/ml). Further dilution of lipids was achieved by removing 2.5 l of sample from the grid and replacing it with 2.5 l of lipid-free buffer, followed by blotting and plunging into cryogen. Freeze-drying of RyR3 was done by raising the temperature of a specimen grid (mounted in a Gatan model 626 cryoholder) containing frozen-hydrated RyR3 while in the microscope (Philips EM420) from the usual operating temperature of Ϫ175 Ϯ 2°C to higher than Ϫ120°C for a few minutes (43) and then lowering the temperature back to Ϫ175°C.
Computer image analysis, which involved correlation alignment and averaging of individual images of 4-fold symmetric RyR3 complexes, was accomplished essentially as described by Radermacher et al. (12,13). Micrographs were recorded at a magnification of 36,000 and digitized for image analysis on a Perkin-Elmer PDS flatbed scanning microdensitometer using a 20-m square scanning raster (corresponding to 5.6 Å on the specimen).

Ryanodine Binding Characteristics of Bovine Diaphragm SR Fractions
Bovine diaphragm SR fractions were prepared and subfractionated by equilibrium isopycnic centrifugation. The binding characteristics of bovine diaphragm SR fractions are presented in Table I. The fractions at highest density, R 4 and I 4 , analogous to the TC of rabbit skeletal muscle (35) are enriched in RyR. The specific ryanodine binding was 10.7 Ϯ 0.55 pmol/mg of protein for R 4 (mean Ϯ S.E., n ϭ 4) and 9.73 Ϯ 1.10 pmol/mg of protein for I 4 (n ϭ 4). The two fractions together amounted to 160 mg of protein obtained from processing 600 g of diaphragm ground muscle. Both R 4 and I 4 fractions had similar characteristics with regard to source for preparation of RyR3 and are referred to hereafter as bovine diaphragm terminal cisternae (TC).

Identity of RyRs in Bovine Diaphragm TC and Specificity of RyR3 Antibody
Bovine diaphragm terminal cisternae and rabbit diaphragm microsomes (I and R fractions) are compared with rabbit skeletal muscle terminal cisternae by Coomassie staining of SDS-PAGE gels (Fig. 1A). Each fraction has prominent high molecular weight bands referable to ryanodine receptors. Both RyR1 and RyR3 can be observed since RyR3 has somewhat faster mobility (44). RyR1 is the main RyR component, whereas RyR3 I Characteristics of subfractions from bovine diaphragm SR Bovine diaphragm sarcoplasmic reticulum was prepared from 600 g of ground diaphragm muscle as described under "Experimental Procedures" and subfractionated by isopycnic centrifugation on a step gradient (27,32,38, and 45% sucrose (w/w)) as for rabbit skeletal muscle SR (35). The terminal cisternae fraction (R 4 and I 4 ) were most enriched in RyR, similar to the TC from rabbit skeletal muscle. The data are presented as mean Ϯ S.E. (n ϭ 4). is a minor component by comparison (lanes 1 and 2). Under these conditions, RyR3 is not observed in TC from rabbit skeletal muscle SR. Similar gels were also characterized by Western blot analysis probed with the general RyR antibody (number 8) (Fig. 1B) and the RyR3-specific antibody (Fig. 1C). The general RyR antibody detects three bands in the high molecular region of the diaphragm fractions (lanes 1-3) referable to RyR1 (major band with slowest mobility), RyR3 a relatively minor band with slightly faster mobility, and fragmented RyR1 (F-RyR1) with yet faster mobility (25). The amount of F-RyR1 was quite variable. The RyR3 antibody selectively detects only RyR3 in terminal cisternae fractions of bovine diaphragm SR as well as in the rabbit diaphragm microsomes (Fig. 1, C and  D). RyR3 was not detected in rabbit skeletal muscle terminal cisternae. RyR3 antibody reacts specifically with RyR3 (lane 3), as shown in Fig. 1D but not with RyR1 of bovine diaphragm TC or rabbit terminal cisternae (lane 1) or with RyR2 from dog cardiac SR (lane 2). Thus, the RyR3 antibody is specific for RyR3.

Purification of RyR3 by Selective Immunoprecipitation
RyR3 can be selectively immunoprecipitated from CHAPSsolubilized bovine diaphragm terminal cisternae as documented in Fig. 2, A and B. The other protein components including RyR1 remained in the supernatant. RyR3 was selectively concentrated in the immunoprecipitate. RyR3 can be discerned from RyR1 by its more rapid mobility ( Fig. 2A) and by Western blot analysis using RyR3-specific antiserum (Fig.  2B, lane 3).
Thus, the basis of a method to purify RyR3 from bovine diaphragm terminal cisternae was now available. We next carried out a titration with affinity purified RyR3 antibody to characterize the concentration dependence of RyR3 antibody for affinity purification of RyR3 from bovine diaphragm termi-nal cisternae (Fig. 2C). Increasing concentrations of RyR3 antibody prebound to protein A/G were added to aliquots of solubilized bovine diaphragm TC. Ryanodine binding activity was measured in the sedimented, sequentially washed immune complex (see "Experimental Procedures"). Initially, there is a near-linear correlation of RyR3 binding with increasing amounts of RyR3 antibody which then plateaus in a narrow concentration range (Fig. 2C). This study indicates the following: 1) high affinity binding of RyR3 antibody to RyR3 (K D ϳ1.4 nM); 2) RyR3 accounts for approximately 4.4% of the total ryanodine binding in bovine diaphragm TC; 3) the addition of peptide epitope blocks the association of RyR3 with the antibody, indicating low nonspecific binding of RyRs to the antibody. These results indicated that we can selectively bind and separate RyR3 from solubilized bovine diaphragm TC and that the epitope peptide can be used for the elution of RyR3 from the antibody-protein A/G complex.
The purification procedure of RyR3 from bovine diaphragm SR is illustrated in Fig. 3. The immunoprecipitation procedure described above and in Fig. 2 was repeated on a large scale with some modifications (Fig. 3A). The binding of RyR3 to the antibody-protein A/G matrix was similar. The new step is the elution of purified RyR3 from the immune complex with the use of peptide epitope. The purified eluate of RyR3 is practically homogeneous (Fig. 3A, lane 6). The mobility of the affinity purified RyR3 matched that in the bovine diaphragm SR (lane 1). The supernatant after immunoadsorption of RyR3 contained RyR1; the RyR3 was not detected by Western blot analysis (Fig. 3B, lane 2). Western blot analysis confirmed that the peptide eluates from the immune precipitate contained practically pure RyR3, i.e. devoid of RyR1 (lane 6 in Fig. 3, A and B). We did not detect RyR1 in a similar blot using RyR general antibody 8 (not shown). The elution of RyR3 from the affinity matrix appears to be quantitative. No RyR3 was observed on the immune complex after elution with peptide epitope.

Biochemical and Functional Characterization of Purified RyR3
The purification and yield of RyR3 by the immunoaffinity procedure from the bovine diaphragm terminal cisternae for five separate experiments are summarized in Table II. The yield of RyR3 protein, determined by densitometry, was 52.7 Ϯ 6.2 g of protein (n ϭ 5) from approximately 100 mg of bovine diaphragm TC. The value measured by ryanodine binding assay was 20 pmol of RyR3 or 46 g of receptor protein. This relatively small difference (46 g versus 53 g) is further indication that RyR3 is highly purified.

[ 3 H]Ryanodine Binding
The ryanodine binding of the purified RyR3 receptor was 392 Ϯ 44 pmol/mg of protein (n ϭ 5), measured at 70 nM ryanodine (Table II) or 459 Ϯ 51 when calculated for B max . This value indicates that the RyR3 is highly purified and functional with respect to ryanodine binding. Homogeneously pure RyR3, calculated from the known sequence (7) 2). In addition to the abbreviations used in Fig. 1, CS, calsequestrin; and IgG (HC), immunoglobulin heavy chain. B, Western blot analysis with RyR3 antiserum as probe. The conditions for electrophoresis and blot transfer are as in Fig. 1. C, titration of RyR3 immunoprecipitation as function of the concentration of specific antibody to RyR3. Aliquots of solubilized bovine diaphragm terminal cisternae (R 4 fraction) were immunoprecipitated with increasing amounts of RyR3-specific antibody prebound protein A/G affinity matrix. The K D was obtained by curve fitting using a single binding site equation and is expressed as molar ratio (unitless). Its value of 0.084 is equivalent to 1.37 nM anti-RyR3 antibody. Note that peptide epitope blocks the immunoprecipitation.

TABLE II
Immunoaffinity purification of RyR3 from bovine diaphragm TC In a typical preparation of RyR3 from bovine diaphragm terminal cisternae-enriched fractions, ϳ100 mg of protein containing ϳ900 pmol of ryanodine receptor was used. The RyR concentration was adjusted to ϳ14 nM (total ϭ RyR1 ϩ RyR3), and RyR3-specific antibody was added in ϳ3.5-fold excess (concentration of affinity purified antibody was ϳ0.4 mg/ml in 0.1 M glycine/Tris, pH ϳ7.2, and 0.15 M NaCl). The data are presented as means Ϯ S.E. (n ϭ 5). The yield of RyR3 protein was measured by densitometry using SDS-PAGE, 7.5% resolving gels, and staining with Coomassie Blue. The recovery of RyR3 protein was found to be 2.5 Ϯ 0.4% (n ϭ 5), based on total RyR protein calculated from ryanodine binding measured at 70 nM, using 1 pmol ϭ 2.3 g. The procedure was generally reproducible in the five separate experiments as shown. That is, the recovery calculated from the ryanodine binding is 46.0 g of RyR3/preparation. b The specific ryanodine binding is 459 Ϯ 51 when corrected to B max , using the K D of 12 nM (Fig. 4C). and a B max of one binding site/ryanodine receptor (Fig. 4C). Increasing the concentration of ryanodine to 2 M indicates the presence of lower affinity binding in the micromolar range (see inset in Fig. 4C). Thus, RyR3 is similar to RyR1 and RyR2 regarding one high affinity binding site and lower affinity binding in the micromolar concentration range (3,9,37).
The partially purified RyR1 was obtained from the supernatant after immunoprecipitating RyR3 and then enriched by heparin-agarose column chromatography. The high affinity K D for ryanodine binding of the partially purified RyR1 of 11.0 Ϯ 2.5 nM (Fig. 4B) is essentially the same as obtained for RyR3.
The K D for ryanodine binding by bovine diaphragm TC is shown for comparison (Fig. 4A).

Single Channel Measurements of Purified Ryanodine Receptor 3
Ion Selectivity and Conduction of Purified RyR3-Purified RyR3, solubilized with CHAPS and SBL, was reconstituted in planar lipid bilayers, and the channel characteristics were studied using different ionic conditions. High conductance channels were observed in symmetrical 250 mM KCl (Fig. 5A, K ϩ /K ϩ ) which displayed a reversal potential of 0 mV (Fig. 5B,  E). The slope conductance at 0 mV was 530 Ϯ 32 pS (n ϭ 3). Much lower channel amplitudes were observed with Ca 2ϩ as current carrier and with Tris ϩ in the cis solution (Fig. 5A, Ca 2ϩ /Tris ϩ , and Fig. 5B, q). The slope conductance at 0 mV was 85 Ϯ 5 pS (n ϭ 4). In this condition, the current approached zero (reversal potential) at V cis of ϳ40 mV (Fig. 5B, q). When 250 mM KCl is used instead of Tris ϩ (50 mM Ca 2ϩ /250 mM KCl), the current approached zero at V cis ϳ20 mV (not shown). From these reversal potentials, the permeability coefficient (P) of the RyR3 channels to Ca 2ϩ relative to monovalent ions are P Ca2ϩ/Trisϩ ϳ13 and P Ca2ϩ/Kϩ ϳ6.
With Ca 2ϩ /Tris ϩ solutions, the amplitude of the channels at 0 mV in Ca 2ϩ /Tris ϩ was 3.6 Ϯ 0.10 pA (range from 3.3 to 4.1 pA, n ϭ 8). Conductance substrates of ϳ25 and 50% of the full channel conductance were observed with rare frequency compared with full openings (Յ1%; not shown). As found with other RyR isoforms, in the presence of ryanodine (1 M), the channels mostly displayed long openings with current amplitude of ϳ0.4, relative to the full openings observed in control conditions (Fig. 5C).
The Effect of [Ca 2ϩ ] cis and Voltage (V cis ) on RyR3 Channel Activity-The activity of RyR3 is biphasically modulated by calcium. As shown in Fig. 6A, RyR3 channels had low open probabilities (P o ) at low [Ca 2ϩ ] cis , activated in the micromolar Ca 2ϩ range (0.5-10 M) reaching P o Ͼ0.6, and inactivated at high [Ca 2ϩ ] (mM). We determined the parameters for the activation of RyRs with the Hill equation. For the three channels shown in Fig. 6A, the EC 50 ϭ 2.7 Ϯ 1.2 M (ranged from 1.2 to 5.1 M) and n H a ϭ 1.9 Ϯ 0.5 (ranged from 1.6 to 2.8). The channel reached a peak of activity that decreased at high [Ca 2ϩ ] cis Ն5 mM. The [Ca 2ϩ ] cis for half-maximal inhibition appears to be greater than 5 mM. As shown in Fig. 6B, maximally activated channels ([Ca 2ϩ ] cis at 100 -500 M) did not have a marked voltage dependence in the range Ϫ50 to 50 mV, although the open probabilities decreased somewhat at positive V cis .
Modulation by Different Agonists-At low [Ca 2ϩ ] cis (ϳ50 nM), RyR3s were activated by caffeine as shown in Fig. 7A, where the P o of 0.02 is increased by caffeine to P o of 0.4. ATP (2 mM) did not activate the channels under similar conditions of low [Ca 2ϩ ] cis (Fig. 7B). As found with other RyRs, RyR3 channels were inhibited by [Mg 2ϩ ] cis (Fig. 7, c1 and c2) although much less than by RyR1 (41). The addition of 1 mM [Mg 2ϩ ] cis decreased P o by 20 -40% (n ϭ 3), whereas the addition of ruthenium red (10 M) completely blocked the channels (Fig. 7c3). Channel characteristics of RyR3 as a function of [Ca 2ϩ ] and response to ATP (Figs. 6A and 7B3) resemble the cardiac RyRs (or the ␤-RyRs of non mammalian) more than those of RyR1 (41).

Cryoelectron Microscopy of RyR3
Purified, detergent-solubilized RyR3 was characterized by cryoelectron microscopy. Fig. 8B shows a field of receptors that had been adsorbed onto a carbon support film and flash-frozen in a thin film of buffer. Typically, images of individual frozenhydrated receptors are square in overall shape, appearing much like the images obtained previously for the skeletal muscle isoform of the ryanodine receptor (12,13). The contrast in such micrographs is inherently low, and the presence of detergent and residual phospholipid further decreases it (see "Experimental Procedures" for details of specimen preparation). To show more clearly that the receptors are structurally intact (e.g. not aggregated or dissociated to subunits) and their overall square shape, a field of freeze-dried receptors is also shown (Fig. 8A) which was obtained by raising the temperature of a grid containing frozen-hydrated receptors above the vitrification temperature, approximately Ϫ120°C (45) and then cooling it back down to Ϫ175°C. Freeze-drying greatly enhances the contrast of RyR3, but much of the ultrastructure is lost making the images unsuitable for image enhancement.
The ultrastructure of frozen-hydrated RyR3 is readily apparent in averaged images of the 4-fold symmetric molecules, such as the one shown in the inset of Fig. 8B. The pattern of protein density (dark regions) in the averaged RyR3 appears to be very similar to that in previously described averages of the skeletal muscle isoform of the ryanodine receptor (12,13). This similarity is not unexpected, given the extensive amino acid homology of the two isoforms (7). DISCUSSION In this study, methodology was developed to prepare RyR3 from a mammalian tissue in sufficient quantity for biochemical and biophysical study. The essence of our study is that we could isolate substantial quantities of RyR3 from mammalian dia-phragm which allows detailed characterization. We describe the preparation of bovine diaphragm SR enriched in TC which served as large scale source for RyR3. These bovine diaphragm TC fractions were enriched in ryanodine binding (ϳ10 pmol/mg of protein), albeit the ryanodine binding is mainly referable to RyR1 (ϳ95%), with RyR3 accounting for only ϳ4.4%, i.e. 0.44 pmol/mg of protein. Nonetheless, this is the highest enrichment of RyR3 thus far reported in a mammalian membrane fraction. In a single purification approximately 100 mg of bovine diaphragm TC was used to purify (Table II) 20 pmol of RyR3 (46 g). RyR3 was selectively immunoprecipitated and then eluted with peptide epitope. The procedure developed to isolate the RyR3 made use of a new sequence-specific polyclonal antibody which selectively binds and immunoprecipitates the RyR3 from solubilized bovine diaphragm TC. This means that RyR1 and RyR3 do not exist as a hetero-oligomer of RyR1 and RyR3 isoforms. The RyR3 preparation is functional and highly purified.
In previous studies (44,46,47), only minuscule amounts of mammalian RyR3 were isolated which was not sufficient to be observed by Coomassie staining on SDS-PAGE. Thus, the contamination and purity could not be assessed. Our RyR3 is highly purified as assessed by Coomassie staining and with respect to RyR isoforms. Only RyR3 was detected by Western blot analysis by probing with a general RyR antibody and RyR3-specific antibody (Fig. 3). Furthermore, the purity and functionality was also assessed by ryanodine binding. The specific high affinity ryanodine binding (K D ϳ10 nM) of 459 Ϯ 51 pmol of ryanodine/mg of protein (Table II and Fig. 4C) also confirms that we have highly purified RyR3 receptor. This binding is experimentally indistinguishable from the calculated high affinity binding for the purified receptor (assuming one high affinity binding site, 450 pmol/mg of protein), based on the known molecular weight obtained from the primary sequence (7). RyR3 also has a low affinity binding in the micromolar range. Thus, each of the RyR isoforms is similar in that they have one high affinity binding site/ryanodine receptor, and they have a much lower affinity binding site. In this respect, the ryanodine receptors have in common that they are highly cooperative systems.
In our study the RyR3 channel was reconstituted into planar phospholipid bilayers so that single channel behavior could be characterized. From this and other recent studies (44,47,49,50,53), a general picture of the characteristics of mammalian RyR3 is emerging. RyR3 has a number of channel characteristics in common with all RyRs (3,41,48) and some that categorize RyR3 channels closer to cardiac RyR2 channels. Characteristics in common with RyR1 and RyR2 include the following: 1) slope conductances with K ϩ are approximately 6-fold higher than with Ca 2ϩ ; 2) the channels have higher selectivity to Ca 2ϩ than monovalent cations (K ϩ ), i.e. P Ca 2ϩ/P K ϩ ϳ6; 3) the channel activity is not markedly influenced by membrane potential; 4) biphasic gating by Ca 2ϩ , i.e. activation by M Ca 2ϩ and inhibition at high [Ca 2ϩ ] (mM); 5) diagnostic pharmacology, both high and low affinity binding of ryanodine, activation by caffeine, and inhibition by procaine, Mg 2ϩ , and ruthenium red. On the other hand, RyR3 channels differ from RyR1 and behave more like RyR2 (3-5, 41, 48) in the following ways: 1) channel population is relatively homogeneous in response to Ca 2ϩ ; such differences were also noted for non-mammalian isoforms (␣ and ␤, corresponding to RyR1 and RyR3) of chicken (27); 2) RyR3 is less sensitive to inhibition at high [Ca 2ϩ ] (mM), also noted for ␣ and ␤ isoforms of chicken (27), frog (54), and fish (55); 3) ATP does not activate RyR3 at low [Ca 2ϩ ] (this study and Ref. 47). Again, comparable differences in ATP responses were found between chicken ␣ and ␤ RyR channels (27); and 4) Ca 2ϩ release is characteristic of Ca 2ϩ -induced Ca 2ϩ release (50 -53, 56).
The first structural analysis of RyR3 is presented in our study. The two-dimensional image of RyR3 presented here, obtained by cryoelectron microscopy and image enhancement analysis, provides a structure that is quite similar to that of RyR1 and RyR2 (12). It will require more detailed structural analysis to pinpoint any differences. Studies are currently in progress to determine whether any structural differences between the two isoforms are detectable by cryoelectron microscopy and quantitative image analysis.
Only recently has there been limited progress in the study of RyR3 from mammalian sources. Ogawa's laboratory used antibody specific for RyR3, raised against a synthetic peptide corresponding to 4375-4387 of rabbit RyR3 (7). The antibody was used to immunoprecipitate RyR3 from rabbit brain and diaphragm where RyR3 was estimated to account for 2 and 0.6% of the ryanodine binding, respectively. The mobility of RyR3 was observed to be slightly faster than RyR1 on SDS-PAGE. Ryanodine binding was enhanced by activators of calcium-induced calcium release (44,46). In another approach Chen et al. (47) expressed RyR3 from rabbit uterus in transfected HEK 293 cells and characterized the channel by single channel recordings in bilayers. Our findings are largely in accord with these recent studies (44,46,47).
The role of RyR3 has been explored by preparing mutant mice lacking RyR3. Such mice mature to adulthood, albeit they display increased motor activity suggesting abnormal Ca 2ϩ signaling in some neurons (49). By contrast, mutant mice lacking RyR1 (dyspedic mice, homozygous offspring) do not survive past birth, since lack of RyR1 results in respiratory paralysis and death (50 -52). Clearly, RyR3 does not play a comparable, indispensable role for the survival of the mammal. Important differences between RyR1 and RyR3 in their mechanisms of excitation-contraction coupling were discerned in the mutant mice lacking RyR1 or RyR3 (50,53). Cells expressing RyR1 channels release calcium from intracellular stores, induced by depolarization, caffeine, or by calcium (50,53). However, depolarization-induced calcium release is not observed in cells expressing RyR3, even though there is release of calcium in response to caffeine and to increasing cytosolic [Ca 2ϩ ] (49 -51). Thus, the physiological mechanism for excitation-contraction coupling of RyR3 resembles that of RyR2 channels, i.e. calcium-induced calcium release.
RyR3 is widely distributed in diverse tissues although at low levels (16 -23, 57-59). RyR3 is relatively enriched in some regions of the brain, i.e. the Purkinje cells in mouse cerebellum and the basal ganglia and the hippocampus of rabbit brain (17,46). Yet, the physiological role of RyR3 such as control of muscle movement and behavior associated with memory remains to be defined. Does RyR3 contain FKBP-binding sites, and if so which isoforms of FKBP bind and what is their role (60)? Do RyR3 channels interact with other molecules (61,62) or ligands which modulate their response, e.g. cADPR scripts of RyR3, which are differentially expressed in some tissues (65,66), generate functionally different channels? Will the detailed three-dimensional structure of RyR3 reveal unique characteristics or differences from RyR1 and RyR2? The availability of RyR3 should facilitate more detailed study especially in the mammalian system.