Characterization of type 3 ryanodine receptor (RyR3) of sarcoplasmic reticulum from rabbit skeletal muscles.

We investigated type 3 isoform (RyR3) of ryanodine receptor in rabbit skeletal muscles using an antibody specific for RyR3. By Western blot analysis and by immunoprecipitation, a single polypeptide for RyR3 was detected in sarcoplasmic reticulum vesicles from rabbit diaphragm but not in those from back muscle. The molecular mass was slightly smaller than that of RyR1, the major isoform in skeletal muscles. Each of RyR1 and RyR3 formed a homotetramer in rabbit diaphragm. RyR3 had a single class of [3H]ryanodine binding sites of high affinity (KD = 1.6 nM). From the Bmax of the binding, the content of RyR3 was estimated to be only 0.6% of RyR1 in rabbit diaphragm. -3H-Ryanodine binding to RyR3 was biphasically dependent on Ca2+, as is true of RyR1, and was stimulated further by adenine nucleotide, caffeine, or high salt concentration. Procaine and ruthenium red inhibited the binding. RyR3 was more resistant to Mg2+ inhibition than RyR1. Interestingly, RyR3 showed about a 7-fold lower Ca2+ sensitivity for activation than RyR1. Comparison with the counterparts in bullfrog skeletal muscles indicates that the Ca2+ sensitivities of RyR3 homologs are similar to each other, whereas those of RyR1 homologs are species-specific.

We investigated type 3 isoform (RyR3) of ryanodine receptor in rabbit skeletal muscles using an antibody specific for RyR3. By Western blot analysis and by immunoprecipitation, a single polypeptide for RyR3 was detected in sarcoplasmic reticulum vesicles from rabbit diaphragm but not in those from back muscle. The molecular mass was slightly smaller than that of RyR1, the major isoform in skeletal muscles. Each of RyR1 and RyR3 formed a homotetramer in rabbit diaphragm. RyR3 had a single class of [ 3 H]ryanodine binding sites of high affinity (K D ‫؍‬ 1.6 nM). From the B max of the binding, the content of RyR3 was estimated to be only 0.6% of RyR1 in rabbit diaphragm. [ 3 H]Ryanodine binding to RyR3 was biphasically dependent on Ca 2؉ , as is true of RyR1, and was stimulated further by adenine nucleotide, caffeine, or high salt concentration. Procaine and ruthenium red inhibited the binding. RyR3 was more resistant to Mg 2؉ inhibition than RyR1. Interestingly, RyR3 showed about a 7-fold lower Ca 2؉ sensitivity for activation than RyR1. Comparison with the counterparts in bullfrog skeletal muscles indicates that the Ca 2؉ sensitivities of RyR3 homologs are similar to each other, whereas those of RyR1 homologs are species-specific.
Ryanodine receptor (RyR) 1 is one of the Ca 2ϩ release channels of intracellular Ca 2ϩ stores and may play important roles not only in muscles but also in various other cells (1)(2)(3)(4)(5). Molecular cloning of cDNAs encoding mammalian RyRs has shown that there are three distinct isoforms of RyR (RyR1-3) encoded by different genes (6 -11). Although recent studies by RNase protection analysis and by reverse transcription-polymerase chain reaction analysis revealed that mRNA for all RyR isoforms could be widely detected in various tissues (12)(13)(14), these isoforms are highly tissue-specific: RyR1 is expressed primarily in skeletal muscles and cerebellar Purkinje cells, RyR2 in cardiac muscles and ubiquitous regions of brain, and RyR3 in specific regions of brain and various peripheral tissues.
Whereas RyR1 and RyR2 were obtained as purified proteins and their functional properties have been extensively examined (1)(2)(3)(4)(5), molecular characterization of RyR3 remains to be elucidated because of its more miniscule amount. Several studies have been done on the function of RyR3 in tissues where mRNA for RyR3 is expressed exclusively (15)(16)(17), but there have been few investigations on RyR3 protein and its functions. We purified ␣and ␤-RyR from bullfrog skeletal muscles where the two isoforms coexist in almost equal amounts (18) and found them to be homologs of RyR1 and RyR3, respectively, from similarities in the amino acid sequences deduced from their cDNAs (19). Taking advantage of this, we recently produced an antibody (anti-RyR3) that reacts specifically with RyR3 among the three RyR isoforms, and we revealed by immunoprecipitation with anti-RyR3 that a homotetramer of RyR3 protein is expressed in some specific regions of mammalian brain as a caffeine-sensitive Ca 2ϩ -induced Ca 2ϩ release (CICR) channel (20).
Analysis of its mRNA showed that there were several tissuespecific alternative splicing variants of RyR3, especially between brain and peripheral tissues (11,21). These alternatively spliced variants might generate potential heterogeneity in the function of RyR3 among tissues. mRNA for RyR3 was also found in mammalian skeletal muscles that express primarily RyR1 (14,15,17). Conti et al. (22) recently demonstrated minor amounts of RyR3 protein in mammalian skeletal muscles by Western blot analysis. Interestingly, the content of RyR3 varied among different muscles in rat: higher levels in diaphragm and soleus, lower levels in abdominal muscles and tibialis anterior, and no detectable amounts in the extensor digitorum longus. A particularly high content of RyR3 in the diaphragm was observed in several mammals examined (rat, mouse, rabbit, and cow). To learn whether there are functional differences in RyR3 between brain and skeletal muscles, we identified and characterized here RyR3 expressed in rabbit skeletal muscles using the anti-RyR3 antibody (20). The results of this study show that a homotetramer of RyR3 is expressed in rabbit diaphragm but is undetectable in back muscle. It may function as a CICR channel that is similar to RyR3 in rabbit brain. Further unique properties of RyR3 are also revealed.

Materials-
The peptide corresponding to the amino acid sequence 4375-4387 of the rabbit RyR3 (RyR3-peptide) was synthesized at the Central Laboratory of Medical Sciences, Division of Biochemical Analysis, Juntendo University School of Medicine (20). [ 3 H]Ryanodine (60 -90 Ci/mmol) was purchased from NEN Life Science Products. Goat anti-rabbit IgG-agarose was from Sigma. Egg lecithin (egg total phosphatide extract) was from Avanti Polar Lipids. All other reagents were of analytical grade.
Isolation of Sarcoplasmic Reticulum (SR) Vesicles-Heavy fraction of SR vesicles was prepared from rabbit diaphragm or back muscle according to Murayama and Ogawa (18) 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 membranes were quickly frozen in liquid N 2 and stored at Ϫ80°C until use.
Solubilization of SR Vesicles and Sucrose Gradient Ultracentrifugation-SR vesicles (2-4 mg/ml at the final concentration) were incubated for 15 min on ice with 2% CHAPS and 1% egg lecithin in a buffer containing 0.5 M NaCl, 20 mM Tris-HCl, pH 7.4, 2 mM dithiothreitol, and a mixture of protease inhibitors. The supernatant after centrifugation at 100,000 ϫ g for 30 min was collected and used for detection and characterization of RyR3. Sucrose gradient ultracentrifugation with 5-20% linear gradients was performed as described (18).
Anti-RyR3 Antibody and Western Blot Analysis-Anti-RyR3 antibody was produced in rabbits against the synthetic peptide corresponding to the amino acid sequence 4375-4387 of rabbit RyR3, and was purified with protein-bound polyvinylidene difluoride membranes of ␤-RyR from bullfrog skeletal muscle as described in Murayama and Ogawa (20). The antibody reacted with RyR3 of mammalian brain and ␤-RyR of frog or chicken skeletal muscle, but no cross-reactivity was observed with mammalian RyR1, RyR2, or ␣-RyR of non-mammalian vertebrate skeletal muscles (20). Western blotting was carried out colorimetrically as described previously (18) using peroxidaseconjugated goat anti-rabbit IgG as the secondary antibody and 3,3Јdiaminobenzidine as the substrate.
Immunoprecipitation-Immunoprecipitation of RyR3 was performed using the purified anti-RyR3 antibody and goat anti-rabbit IgG-agarose beads (20). Solubilized SR vesicles (1-2 mg of protein) were incubated overnight at 4°C with 100 l of anti-RyR3 antibody and 30 l of anti-rabbit IgG-agarose beads. For detection of the polypeptide band for RyR3, the beads were washed five times with a buffer containing 0.5 M NaCl, 50 mM Tris-HCl, pH 7.5, 0.05% Tween 20, 0.1% CHAPS, and 2 mM dithiothreitol and were resuspended in 40 l of 2 ϫ Laemmli sample buffer (23) containing 0.1 M dithiothreitol. Aliquots of 15-20 l were subjected to SDS-polyacrylamide gel electrophoresis.
[ 3 H]Ryanodine Binding Assay-Assays were carried out as described previously (20). SR vesicles (0.2 mg) were incubated with [ 3 H]ryanodine (2-21 nM) for 4 h at 25°C in 0.5 ml of a binding buffer containing 0.3 or 1 M NaCl, 10 mM MOPSO/NaOH, pH 6.8, 2 mM dithiothreitol, 1% CHAPS, 0.5% egg lecithin, the mixture of protease inhibitors, and a specified concentration of free Ca 2ϩ buffered with 10 mM EGTA. Then RyR3 was immunoprecipitated with 100 l of anti-RyR3 and 30 l of anti-rabbit IgG-agarose beads. The radioactivity of the beads after five washings with a buffer (1 M NaCl, 10 mM MOPSO/NaOH, pH 6.8, 1% CHAPS, 0.5% egg lecithin, 2 mM dithiothreitol, and 0.1 mM Ca 2ϩ ) was determined as the activity for RyR3. The nonspecific radioactivity was determined in the absence of anti-RyR3 in each experiment. The value was similar to that determined with the addition of 30 M RyR3-peptide (see Fig. 5). The nonspecific activity was decreased as the salt concentration of the medium was increased and approached a value similar to the result on the addition of excess unlabeled ryanodine (10 -50 M) at 0.3 M NaCl. These findings indicate that it may be caused probably by direct binding of [ 3 H]ryanodine and RyR1 to the agarose beads (see Fig.  2) rather than weak binding to anti-RyR3. By compensating for the nonspecific radioactivity thus determined, [ 3 H]ryanodine binding specific to RyR3 can be obtained. In this study, the [ 3 H]ryanodine binding to RyR3 could not be determined accurately in an isotonic medium containing 0.17 M NaCl because of a high nonspecific radioactivity. Instead, assays were carried out in a medium containing 0.3 M NaCl where the properties are assumed to be more physiological than those in 1 M NaCl (see Figs. 8 and 9, Tables I and II). The binding for RyR1 was measured by filtering an aliquot of the remaining supernatants after immunoprecipitation through polyethyleneimine-treated Whatman GF/B glass filters (18). Free Ca 2ϩ was calculated using values of 8.79 ϫ 10 5 M Ϫ1 and 1.82 ϫ 10 3 M Ϫ1 as the apparent binding constants for Ca 2ϩ of EGTA (24) and of AMP-PCP (25), respectively.

Identification of RyR3 in Rabbit Skeletal
Muscles-To identify RyR3 in skeletal muscles, we prepared terminal cisternaerich fractions of SR vesicles from rabbit skeletal muscles where RyR3 is reported to be localized (22). Because the content of RyR3 varied among different muscles in rat (22), we used diaphragm (which was reported to express the highest content of RyR3 among skeletal muscles examined) and back muscle (material commonly used for SR vesicles) as materials. Fig. 1A shows a Coomassie Brilliant Blue-stained SDS-polyacrylamide gel of SR vesicles from diaphragm and back muscle. The two specimens showed very similar patterns of protein composition. Mammalian skeletal muscles express primarily RyR1, in contrast to skeletal muscles of non-mammalian vertebrates which have nearly equal amounts of two RyR isoforms (␣and ␤-RyR). Consistently, single bands for RyR1 of nearly equal density were clearly detected at the low mobility range of the gel (arrowhead) in both SR preparations, and the band for RyR3 could not be identified in diaphragm or back muscle SR on the Coomassie-stained gel.
When the transferred membrane was probed with anti-RyR3, the antibody raised against a synthetic peptide (RyR3peptide) corresponding to 4375-4387 of rabbit RyR3 which reacts highly specifically with mammalian RyR3 among the three isoforms (20), a single band for RyR3 was faintly but significantly reacted just below the RyR1 band in diaphragm SR, but not in back muscle SR (Fig. 1B). Thus, a minute amount of RyR3 in addition to dominant RyR1 was expressed in rabbit diaphragm, whereas no RyR3 was detected in back muscle SR.
The expression of RyR3 was also examined by immunoprecipitation using antibody-conjugated agarose beads (see "Experimental Procedures"). Fig. 2 shows Coomassie Brilliant Blue-stained gels of the proteins immunoprecipitated with the anti-RyR3 from CHAPS/egg lecithin-solubilized SR. In diaphragm SR, a single high molecular weight band (arrowhead) was observed, which specifically disappeared by the addition of 30 M RyR3-peptide during immunoprecipitation (panel A). This band was reacted with anti-RyR3 on Western blot analysis (data not shown). These results indicate that RyR3 is definitely expressed in rabbit diaphragm. A band seen above RyR3 band was a minute contamination of RyR1 because of its positive reaction with anti-RyR1 antibody (data not shown). Because the band did not disappear by the RyR3-peptide, the precipitation of RyR1 may be the result of nonspecific binding to the agarose beads rather than weak binding to anti-RyR3. No specific bands, in contrast, were immunoprecipitated from back muscle SR (panel B), suggesting that there are no detectable amounts of RyR3 in back muscle, the same conclusion as shown in Fig. 1B. The varied contents of RyR3 in contrast to similar amounts of RyR1 between diaphragm and back muscle corresponded well to the results with rat skeletal muscles (22). The following experiments to characterize RyR3 in skeletal muscles were therefore carried out with diaphragm SR vesicles.
Subunit Molecular Mass and Tetramer Formation of RyR3-The molecular mass of rabbit RyR3 protein is estimated to be 552 kDa from its predicted amino acid sequence (10), which is slightly smaller than that of rabbit RyR1 (565 kDa) (6, 7). As shown in Fig. 3, the mobility of immunoprecipitated RyR3 was slightly but significantly larger than that of RyR1 on the Coomassie Brilliant Blue-stained SDS-polyacrylamide gel. The mobilities of RyR1 and RyR3 were similar to those of bullfrog ␣and ␤-RyR, the homologs of RyR1 and RyR3, respectively, in non-mammalian vertebrates (18,19).
Formation of tetramer is one of the typical characteristics of RyRs. As described above, rabbit diaphragm expresses considerable amounts of RyR1 and minor levels of RyR3 (see Fig. 1). It is therefore important to determine whether RyR3 forms a homotetramer of its subunit or heterotetramer with RyR1. The tetramer formation of RyR is detected easily by the sedimentation pattern of sucrose density gradients: a tetrameric RyR sediments to higher density fractions of the gradient because of its large sedimentation coefficient of about 30 S, whereas monomeric RyR subunits remain in lower density fractions as is true with the other proteins (26,27). Fig. 4A shows the low mobility range of Coomassie Brilliant Blue-stained gels of total proteins (30 l of each fraction) of fractions that were divided after centrifugation through 5-20% sucrose density gradient of solubilized rabbit diaphragm SR. The band for RyR1 was detected in higher density fractions (fractions 6 and 7) of the gradient. This pattern corresponds well with our previous results with ␣and ␤-RyR of bullfrog skeletal muscle (18) and RyR2 and RyR3 of rabbit brain microsomes (20) under identical conditions, as is true of RyR of rabbit skeletal muscle SR (26,27). The sedimentation pattern of RyR3 through the sucrose gradient was similarly examined on the Coomassie Brilliant Blue-stained gel after immunoprecipitation of a large volume of the gradient fractions (4 ml of each fraction) with anti-RyR3. As shown in Fig. 4B, the bands for RyR3 were detected in the identical fractions of nos. 6 and 7. It should be noted that only a single RyR3 band was detected in the immunoprecipitated products on each lane of the gel. If RyR3 forms a heterotetramer with RyR1, the two should be coprecipitated, resulting in the detection of dual bands on the Coomassie Brilliant Bluestained gel because of their different mobilities (see Fig. 3). The absence of an RyR1 band clearly excludes the possibility of heterotetramer formation of RyR3 with RyR1. These results indicate that each of RyR1 and RyR3 coexpressed in rabbit diaphragm exclusively forms a homotetramer. Homotetramer formation of RyR3 is also observed in rabbit brain, which expresses a high content of RyR2 (20).

Estimation of the Amounts of RyR3 in Rabbit Diaphragm by [ 3 H]Ryanodine
Binding-The functional properties of RyR3 in rabbit skeletal muscles were determined through the [ 3 H]ryanodine binding. Solubilized SR was incubated with [ 3 H]ryanodine to achieve ryanodine binding, and then RyR3 was specifically immunoprecipitated with anti-RyR3 as described in Murayama and Ogawa (20). Fig. 5 shows the radioactivity immunoprecipitated with anti-RyR3 from SR of rabbit diaphragm or back muscle. In diaphragm SR, significant radioactivity was precipitated with the antibody. Thirty M RyR3peptide, which completely prevented RyR3 from being immunoprecipitated (see Fig. 2), reduced the radioactivity to the background level. A similar reduction in the radioactivity was also observed in determination without anti-RyR3 (data not shown). Therefore, the radioactivity is caused by RyR3 itself, but not by minute amounts of contaminating RyR1. The addition of excess concentrations (10 -50 M) of unlabeled ryanodine caused almost total loss of the radioactivity (data not shown). These results suggest that RyR3 in diaphragm has specific [ 3 H]ryanodine binding activity. In back muscle SR, in contrast, no significantly detectable radioactivity was observed, consistent with the absence of the immunoprecipitable RyR3 (see Fig. 2).
As shown in Fig. 6, the specifically immunoprecipitated ryanodine binding activity in 0.2 mg of the solubilized diaphragm SR increased with added amounts of anti-RyR3 and saturated around 0.044 pmol/mg of protein with 100 l of antibody or more. The activity was no longer immunoprecipitated from the supernatant after the immunoprecipitation with 100 l of the antibody (data not shown). This indicates that 100 l of the antibody was enough to precipitate almost all RyR3 from 0.2 mg of diaphragm SR. The [ 3 H]ryanodine binding activity in the supernatant after immunoprecipitation is ascribed to RyR1, which can be determined by filtering an aliquot of the supernatant through polyethyleneimine-treated glass filters as described (18). Thus, we could measure the activity for each of RyR1 and RyR3 separately under identical conditions. Fig. 7 shows dose-dependent [ 3 H]ryanodine binding to RyR1 and RyR3 of rabbit diaphragm SR under optimum conditions. The amounts of the binding to RyR1 (upper panel) and RyR3 single class of binding sites. RyR1 showed a dissociation constant (K D ) for ryanodine of 2.3 nM and B max of 11.4 pmol/mg protein. The K D and B max of RyR3 were calculated to be 1.6 nM and 0.065 pmol/mg protein, respectively. The K D values correspond to those with RyR from rabbit skeletal muscle (26,28) and with the purified ␣and ␤-RyR from bullfrog skeletal muscle under similar conditions (18). The similarity in K D for the high affinity ryanodine binding sites between RyR3 and RyR1 suggests that they are generally similar to each other and allows us to assume that the stoichiometry of the site for RyR3 may be identical to that for RyR1 (1 mol of ryanodine/ tetramer) (26). This assumption was also verified with ␣and ␤-RyRs from bullfrog skeletal muscle SR (18). Thus, the B max value would directly express the content in the SR vesicles. From the ratio of B max for RyR3 to that for RyR1, the amount of RyR3 was estimated to be only 0.6% of that of RyR1 in rabbit diaphragm SR. Table I summarizes the effect of well known CICR activators on [ 3 H]ryanodine binding to RyR3. The experiments were carried out in a medium containing 0.3 M NaCl, albeit a higher salt concentration than physiological medium, because a higher nonspecific radioactivity prevents sensitive detection of the specific binding in an NaCl concentration of less than 0.3 M. In the absence of Ca 2ϩ (10 mM EGTA), no significant ryanodine binding was detected. The Ca 2ϩ -activated binding of 6.3 fmol/mg of protein was observed at 10 M Ca 2ϩ . Addition of 1 mM of AMP-PCP, a nonhydrolyzable ATP analog, further increased the binding to 18.1 fmol/mg of protein. Caffeine (10 mM), a Ca 2ϩ sensitizer of CICR, also enhanced the binding by about 3.7-fold. The binding was greatly enhanced (5.7-fold) by increasing the NaCl concentration from 0.3 to 1 M. These properties are consistent with those of mammalian RyR1 and nonmammalian vertebrate ␣and ␤-RyRs (3).

Effect of CICR Modulators on [ 3 H]Ryanodine Binding to RyR3-
We examined in detail the effect of Ca 2ϩ on the [ 3 H]ryanodine binding to RyR3. In a medium of a low salt concentration containing 0.3 M NaCl, the effect of Ca 2ϩ was biphasic: it increased the binding in a dose-dependent manner at lower than 0.1 mM Ca 2ϩ , whereas it inhibited the binding above 0.1 mM (open circles in Fig. 8A). The Ca 2ϩ concentrations that would give half the maximal binding for activation (EC 50 ) and for inactivation (IC 50 ) were 7.9 M and 4.9 mM, respectively. The increase in NaCl concentration from 0.3 M to 1 M enhanced the binding at every Ca 2ϩ concentration (closed circles in Fig.  8A) as described above (see Table I). Under this condition, the effect of Ca 2ϩ appeared to be monophasic; only slight inactivation was observed up to 10 mM Ca 2ϩ . Furthermore, the Ca 2ϩ sensitivity for activation was enhanced. The binding was already at the maximum at 10 M Ca 2ϩ , which corresponds to the EC 50 value in 0.3 M NaCl medium, and the EC 50 in 1 M NaCl medium was calculated to be 2.6 M. These characteristic modifications by high concentrations of salt were also observed with RyR3 in rabbit brain (20) and with ␤-RyR from bullfrog skeletal muscle, the homolog of RyR3 in non-mammalian vertebrates, but not with ␣-RyR, the homolog of RyR1 (29).
Taking advantage of the simultaneous determination of [ 3 H]ryanodine binding to RyR1 and RyR3 under identical conditions, we compared their Ca 2ϩ dependences in 0.3 M NaCl medium (Fig. 8B). The [ 3 H]ryanodine binding activity of RyR1 was bell-shaped against the Ca 2ϩ concentration. The Ca 2ϩ concentrations for EC 50 and IC 50 were 1.2 M and 2.9 mM, respectively. This EC 50 value was consistent with that reported previously with rabbit skeletal muscle SR in [ 3 H]ryanodine binding (28) or in Ca 2ϩ release experiments (30). The normalized curves of Ca 2ϩ dependences of RyR1 and RyR3 showed that RyR3 was about 7-fold less sensitive to Ca 2ϩ required for activation than RyR1. Although the inactivation effect of Ca 2ϩ appeared to be slightly weaker on RyR3 than on RyR1, this difference was unlikely to be significant. These results suggest that RyR3 has properties of CICR channel of lower Ca 2ϩ sensitivity than RyR1 in mammalian skeletal muscles. Table II summarizes the effect of CICR inhibitors on [ 3 H]ryanodine binding to RyR3. Ruthenium red (1 M) reduced the binding to 24% of the control that had been activated by the optimal concentration of Ca 2ϩ (0.1 mM). The binding was effectively decreased to 18% in the presence of procaine (10 mM). It is noted that the inhibition by Mg 2ϩ appeared to be weak: nearly two-thirds of the control binding still remained even in the presence of 10 mM Mg 2ϩ .
The Mg 2ϩ sensitivity of RyR3 was examined further and  Table I  compared with that of RyR1 (Fig. 9). Mg 2ϩ dose-dependently inhibited the [ 3 H]ryanodine binding to RyR3 to about 75% of the control up to 5 mM. However, no more significant reduction in the binding was observed even in the presence of 10 or 20 mM Mg 2ϩ . In contrast, the [ 3 H]ryanodine binding to RyR1 was remarkably decreased with Mg 2ϩ concentration: the IC 50 of Mg 2ϩ was around 2 mM. These results indicate that RyR3 is more resistant to inhibition by Mg 2ϩ than RyR1.

DISCUSSION
In this study we identified homotetrameric RyR3 expressed in rabbit diaphragm. RyR3 demonstrated the [ 3 H]ryanodine binding that was activated by micromolar to submillimolar Ca 2ϩ and inactivated by millimolar or more Ca 2ϩ . It is stimulated further by adenine nucleotides and caffeine and inhibited by ruthenium red and procaine. These results indicate that RyR3 may function as a CICR channel in rabbit diaphragm.
[ 3 H]Ryanodine binding of RyR3 was determined by immunoprecipitation of the RyR3 protein that had been incubated with [ 3 H]ryanodine. In this case, the RyR3 showing [ 3 H]ryanodine binding was the antibody-bound protein. It was reported that single-channel properties of RyR were modified by some anti-RyR antibodies (31,32). Therefore, one might argue that "antibody-bound RyR3" is different in its properties from the "free RyR3." Our results indicate that the anti-RyR3 antibody had no significant effects on the ryanodine binding properties of bullfrog ␤-RyR, a homolog of RyR3, which is also precipitated with the antibody (data not shown). Therefore it is unlikely that immunoprecipitation might affect RyR3, although we cannot completely exclude the possibility. We are currently trying to determine the [ 3 H]ryanodine binding activity with the purified RyR3 protein to address this question directly.
Analysis of its mRNA revealed that there could be several tissue-specific alternative splicing variants of RyR3 especially in brain and peripheral tissues (11,21). These alternatively spliced variants might generate potential heterogeneity in the function of RyR3 among tissues. We have recently identified and characterized RyR3 expressed in rabbit brain (20). Brain and diaphragm RyR3 proteins demonstrated similarity in several points (mobility on SDS-polyacrylamide gel, homotetramer formation, biphasically Ca 2ϩ -activated [ 3 H]ryanodine binding, and caffeine sensitivity), and no significant differences were found between them. Thus, the properties of RyR3 are similar between brain and skeletal muscle in rabbit. This is consistent with the results by Marziali et al. (11) that the spliced variants of RyR3 might not be distinct between brain and diaphragm in mink.
In mammalian skeletal muscles, RyR1 plays an important role in excitation-contraction coupling because the mice lacking RyR1 (dyspedic mice) failed normal excitation-contraction coupling (33). In addition to RyR1, we demonstrated here that RyR3 is expressed in rabbit diaphragm as a functional CICR channel. Recently, Takeshima et al. (34) concluded that RyR3 may not play a critical role in mammalian skeletal muscles on the basis of the observation of normal excitation-contraction coupling in muscles from the RyR3-null mice. We estimated the amount of RyR3 to be only about 0.6% of RyR1 in rabbit diaphragm, which is reported to be the highest level of RyR3 among skeletal muscles examined (22). It is unlikely that RyR3 has extraordinary larger unit conductance or higher open probability than RyR1 because frog or chicken ␤-RyR showed single-channel kinetics similar to ␣-RyR and mammalian RyRs (18,35). Therefore, the contribution of RyR3 to total Ca 2ϩ release channel activity in mammalian skeletal muscles cannot be much greater in magnitude than that expected from its relative content. The insignificant contribution of RyR3 to Ca 2ϩ signaling in the skeletal muscles may thus be explained by its minute amount.
The content of RyR3 in rabbit diaphragm is estimated to be only 0.6% of RyR1. Our previous results showed that the amount of RyR3 in rabbit brain was 1.6 -2% of total RyRs (20). This might lead to a misunderstanding that RyR3 is less expressed in diaphragm than in brain of rabbit. From the amounts of specific [ 3 H]ryanodine binding sites, the total RyR in rabbit brain microsomes (primarily RyR2) is estimated to be about 3% of RyR1 in rabbit diaphragm SR. Therefore, the content of RyR3 in the brain would be at most 0.06% of RyR1 in diaphragm, which is only one-tenth of that in diaphragm. This corresponds well with the clear detection of RyR3 on Western blot analysis in rabbit diaphragm SR (Fig. 1), whereas no bands were observed in rabbit brain microsomes (20).
The results of this study demonstrate that each of RyR1 and RyR3 coexisting in rabbit diaphragm forms a homotetramer, excluding the possibility of heterotetramer formation. Rabbit diaphragm consists of several different types of muscle cells as is true of the diaphragm of other animals (36 -38). If RyR1 and RyR3 were expressed separately in different types of cells, homotetramer formation of RyR1 and RyR3 should be simply the result of the lack of opportunity for heterotetramer formation. Preliminary results by immunohistochemistry with anti-RyR3 show that RyR3 is detected almost homogeneously in rat diaphragm. 2 Consistently, homogeneous distribution of mRNA for RyR3 in rat diaphragm was also reported (22). These findings indicate that RyR1 and RyR3 are coexpressed in the same cell, and thereby they exclusively form a homotetramer even in the presence of different subunits. Homotetramer formation of RyR3 is also observed in rabbit brain, which expresses all three isoforms (20). Although homotetramer formation of RyR1 was believed in mammalian skeletal muscles where only the isoform was formerly acknowledged to occur, potential coexpression of RyR3 and RyR1 requires more strict evidence to prove it. In cardiac muscles where RyR2 is the primary RyR isoform, 2 T. Hijikata, personal communication. the situation is similar. Our present results are the first clear evidence of homotetramer formation of RyR1 in mammalian skeletal muscles where multiple RyR isoforms may be coexpressed. On the other hand, it has been clearly proved that the two isoforms coexpressed in the same cells of non-mammalian vertebrate skeletal muscles (␣-and ␤-RyR, homologs of RyR1 and RyR3, respectively) form only a homotetramer of each subunit (18,39,40). Homotetramer formation may be one of the typical characteristics of vertebrate RyRs. This is in marked contrast to the inositol 1,4,5-trisphosphate receptor, another Ca 2ϩ release channel of intracellular Ca 2ϩ stores which potentially forms both a homotetramer and heterotetramer of three isoforms (41,42).
RyR3 in rabbit diaphragm is sensitive to caffeine as is true of rabbit brain RyR3 (20). Similar results were also observed with skeletal muscles of the dyspedic mice (17). Caffeine sensitivity may therefore be a common property of RyR3 and all mammalian RyR isoforms. The lack of caffeine response of RyR3 was reported previously in mink lung epithelial cells (15) and human Jurkat T-cells (16). Later, Giannini et al. (14) acknowledged that the lack of caffeine sensitivity was probably because of the minimal content of RyR3. However, RyR3 in epithelial cells is likely to be a differently spliced variant from the one in skeletal muscle or brain (11). There might be some spliced variants that are caffeine-insensitive in the RyR3 group. It is also possible that the experimental conditions may be responsible for the discrepancy. The lack of caffeine response was obtained with intact cells (15,16), whereas the caffeine response was shown with skinned fibers in the absence of ATP (17). It seems somewhat difficult to observe the Ca 2ϩ release induced by caffeine in the presence of ATP where released Ca 2ϩ is actively reaccumulated to Ca 2ϩ stores by the Ca 2ϩ pump, resulting in failure to increase in [Ca 2ϩ ] i . Consistently, an early study by Weber and Herz (43) reported occasional failure in showing Ca 2ϩ release by caffeine from several preparations of rabbit SR vesicles in the presence of ATP, whereas they succeeded with frog SR vesicles.
The Ca 2ϩ sensitivity is one of the most important factors for functions of CICR channels. RyR3 in rabbit diaphragm showed biphasic Ca 2ϩ dependence as is true of RyR1; low concentrations of Ca 2ϩ were stimulatory, and Ca 2ϩ above 0.1 mM was inhibitory. One notable point is that RyR3 was about 7-fold less sensitive to Ca 2ϩ required for activation than RyR1 (Fig. 8B). This may be consistent with the results by Takeshima et al. (17) that the Ca 2ϩ sensitivity of RyR3 in mouse skeletal muscles was 10 times lower than that of RyR1. However, it is in contrast to the situation in frog skeletal muscles where ␣and ␤-RyR, the homologs of RyR1 and RyR3, respectively, showed similar Ca 2ϩ sensitivities (29). The EC 50 values for activation of ␣and ␤-RyR were 9.1 and 8.7 M, respectively, in 0.17 M NaCl medium (29). Interestingly, the EC 50 for rabbit diaphragm RyR3 (7.9 M) was similar to that for bullfrog ␤-RyR, whereas the EC 50 for rabbit RyR1 (1.2 M) was about 8-fold lower than that for bullfrog ␣-RyR. This is consistent with the findings that EC 50 for CICR in skinned mammalian skeletal muscle was about 1 M, whereas it was about 10 M in skinned frog skeletal muscle (44). This may indicate that the Ca 2ϩ sensitivities for RyR1 and its homologs vary among species, whereas those for RyR3 and its homologs are similar among vertebrates. The possibility of diversity among RyR1 homologs was also suggested by more strict immunological cross-reactivities among vertebrates, whereas there was broader cross-reactivity among RyR3s (45).
RyR3 was more resistant to inhibition by Mg 2ϩ than RyR1 (Fig. 9). The weak inhibition by Mg 2ϩ was also observed with RyR3 in rabbit brain (data not shown). This was in marked contrast to ␤-RyR of bullfrog which was as sensitive to the inhibitory effect of Mg 2ϩ as ␣-RyR (29). Interestingly, the amount of [ 3 H]ryanodine binding to RyR3 decreased with Mg 2ϩ concentration up to 5 mM and reached the maximum attenuation of about 25% around 5 mM Mg 2ϩ (Fig. 9). High concentrations of Ca 2ϩ , on the other hand, inhibited RyR1 and RyR3 in their similar Ca 2ϩ dependences to very low activities (Fig. 8). It is assumed that Ca 2ϩ and Mg 2ϩ bind to the same inactivating sites of low affinity with similar affinities to inhibit the Ca 2ϩ release channel (30,44). The results shown in Figs. 8 and 9, however, indicate that the cation binding sites for inactivation may be different between Ca 2ϩ and Mg 2ϩ in mammalian RyR3 and that Mg 2ϩ inhibits only partially, whereas Ca 2ϩ does so nearly completely. Another possible explanation is that the diaphragm RyR3 might be composed of two or more heterogeneous populations in its properties: one sensitive to Mg 2ϩ as is true of RyR1 and the other insensitive to Mg 2ϩ . Further characterization should provide some insights into the properties of RyR3.