Isoform-dependent Formation of Heteromeric Ca2+ Release Channels (Ryanodine Receptors)*

Three ryanodine receptor (RyR) isoforms, RyR1, RyR2, and RyR3, are expressed in mammalian tissues. It is unclear whether RyR isoforms are capable of forming heteromeric channels. To investigate their ability to form heteromeric channels, we co-expressed different RyR isoforms in HEK293 cells and examined their interactions biochemically and functionally. Immunoprecipitation studies revealed that RyR2 is able to interact physically with RyR3 and RyR1 in HEK293 cells and that RyR1 does not interact with RyR3. Co-expression of a ryanodine binding deficient mutant of RyR2, RyR2 (I4827T), with RyR3 (wt) restored [3H]ryanodine binding to the mutant. Interactions between RyR isoforms were further assessed by complementation analysis using mutants RyR2 (I4827T), RyR2 (E3987A), RyR3 (I4732T), RyR3 (E3885A), and RyR1 (E4032A), all of which are deficient in caffeine response. Caffeine-induced Ca2+release was restored in HEK293 cells co-transfected with mutants RyR2 (I4827T) and RyR3 (E3885A), RyR2 (E3987A) and RyR3 (I4732T), or RyR2 (I4827T) and RyR1 (E4032A), but not with RyR1 (E4032A) and RyR3 (I4732T), indicating that mutants of RyR2 and RyR3, or RyR2 and RyR1, but not RyR1 and RyR3, are able to complement each other. Co-expression of RyR3 (wt) and a pore mutant of RyR2, RyR2 (G4824A), produced regulatable single channels with intermediate unitary conductances. These observations demonstrate that RyR2 is capable of forming functional heteromeric channels with RyR3 and RyR1, whereas RyR1 is incapable of forming heteromeric channels with RyR3.

Ryanodine receptors (RyRs) 1 belong to a superfamily of Ca 2ϩ release channels, which also includes the inositol 1,4,5-trisphosphate receptors (IP 3 Rs). These channels are located in the sarco(endo)plasmic reticulum of muscle or non-muscle cells and play an essential role in muscle contraction and Ca 2ϩ signaling (1). Three RyR isoforms, RyR1, RyR2, and RyR3, have been identified and characterized in mammalian tissues. They are the products of three distinct genes and share 66 -70% amino acid sequence identity. Each RyR isoform exhibits a unique tissue distribution. RyR1 is primarily expressed in skeletal muscle, whereas RyR2 is predominantly expressed in heart and brain. RyR3 expression has been detected, although at a relatively low level, in a variety of tissues, including brain, diaphragm, and smooth muscles (2)(3)(4). However, recent studies using RNase protection assays revealed that all three RyR genes are widely expressed and that some tissues express two or all three RyR isoforms (5). For instance, both RyR1 and RyR3 are expressed in the diaphragm, whereas all three RyR isoforms are detected in brain and smooth muscle tissues.
The finding that multiple RyR isoforms are co-expressed in the same tissue raises a possibility that RyRs may exist as heteromeric channels in addition to homomeric channels. Early biochemical studies showed that the purified RyR1 from rabbit skeletal muscle is a homotetramer composed of four identical subunits each with a molecular mass of ϳ400 kDa (6). Recently, RyR3 was purified from diaphragm skeletal muscle in which RyR1 was also expressed by immunoprecipitation using an RyR3-specific antibody. The purified RyR3 was shown to be devoid of RyR1 even though RyR1 is co-expressed with RyR3 in the same muscle fibers and in large excess (7)(8)(9). These observations indicate that RyR1 and RyR3, although co-localized, exist only in the form of homotetramers. The same has been shown to be true for the ␣and ␤-RyRs, the non-mammalian counterparts of RyR1 and RyR3 (3,10). Immunoprecipitation of RyR3 from brain tissue using an RyR3-specific antibody did not co-precipitate RyR2 (the major RyR isoform expressed in the brain), demonstrating that RyR2 and RyR3 also exist as homotetramers in the brain (11). These observations have led to the general belief that RyRs may exist only in the form of homotetramers in contrast to IP 3 Rs. Immunoprecipitation studies have shown that the type 1, 2, and 3 IP 3 R isoforms can form heteromeric channels among one another (12,13).
It is clear that RyR1 and RyR3 expressed in mammalian skeletal muscle or ␣-RyR and ␤-RyR expressed in non-mammalian skeletal muscle are incapable of forming heteromeric channels. The question of whether RyR2 and RyR3, or RyR2 and RyR1, are capable of forming heteromeric channel is, however, debatable, because it is unknown whether RyR2 and RyR3, or RyR2 and RyR1, expressed in the brain are co-localized in the same cells. All three RyR isoforms have also been detected in smooth muscle cells. Functional studies have shown that specific inhibition of either RyR1 or RyR2 synthesis abolished Ca 2ϩ -induced Ca 2ϩ release triggered by depolarization and that abolition of RyR3 by gene knock-out enhances the activity of other RyR isoforms in vascular smooth muscle cells (14,15). One possible explanation for these observations is that RyR isoforms form heteromeric channels in these cells. To test this possibility, we co-expressed different RyR isoforms in HEK293 cells and examined the existence of heteromeric RyR channels. To do so, we utilized c-Myc antibody epitope-tagged RyR2 or RyR3 and RyR mutants that are deficient in [ 3 H]ryanodine binding or caffeine response, or have different singlechannel conductances to distinguish the RyR isoforms structurally and functionally. These RyR isoforms, tagged or nontagged, wt or mutants, were co-expressed in HEK293 cells, and the formation of heteromeric RyR channels was examined by immunoprecipitation, [ 3 H]ryanodine binding, and single-channel recordings in planar lipid bilayers. Our results reveal for the first time that RyR2 are capable of forming functional heteromeric channels with RyR3 or RyR1 in HEK293 cells.
Cell Culture and DNA Transfection-HEK293 cells were maintained in Dulbecco's modified Eagle's medium as described previously (16). HEK293 cells grown on 100-mm tissue culture dishes for 18 -20 h after subculture were transfected with 12-16 g of wild type or mutant RyR cDNAs using Ca 2ϩ phosphate precipitation (17).
Site-directed Mutagenesis-Point mutations, RyR1 (E4032A), RyR2 (E3987A), RyR2 (G4824A), and RyR3 (E3885A), were constructed as described previously (18 -21). The RyR2 (I4827T) and RyR3 (I4732T) mutations were generated by the overlap extension method using the polymerase chain reaction (PCR) (22). The NruI (14237)-NotI (vector) fragment that contains the RyR2 (I4827T) mutation was used to replace the corresponding wild type fragment in the full-length mouse RyR2 cDNA. The EcoRI (13422)-NotI (vector) fragment that contains the RyR3 (I4732A) mutation was used to replace the corresponding wild type fragment in the RyR3 cDNA without the RcoRI (11902)-EcoRI (13422) fragment, which was subsequently ligated back to form the full-length RyR3 (I4732A). Insertion of the c-Myc antibody epitope tag, EQKLISEEDL, in RyR3 after glutamate 4318, has been described previously (23). A similar strategy was used to insert the c-Myc tag into RyR2 after glutamate 4414. The Bsu36I (13237)-EcoRV (13873) RyR2 cDNA fragment that contains the c-Myc tag was removed from the PCR product generated by the overlap extension method, and was used to replace the corresponding region in the BsiwI (8864)-NotI (vector) fragment, which was subsequently subcloned into the full-length RyR2 cDNA. All point mutations and insertions were confirmed by DNA sequencing.
Preparation of Cell Lysate from Transfected HEK293 Cells-Preparation of cell lysate from transfected HEK293 cells was performed as described previously (19), with some modifications. HEK293 cells grown for 24 h after transfection were washed three times with PBS (137 mM NaCl, 8 mM Na 2 HPO 4 , 1.5 mM KH 2 PO 4 , 2.7 mM KCl) plus 2.5 mM EDTA and were harvested in the same solution by centrifugation. Transfected HEK293 cells were solubilized in lysis buffer containing 25 mM Tris/50 mM Hepes (pH 7.4), 137 mM NaCl, 1% CHAPS, 0.5% soybean phosphatidylcholine, 2.5 mM dithiothreitol, and a protease inhibitor mix (1 mM benzamidine, 2 g/ml leupeptin, 2 g/ml pepstatin A, 2 g/ml aprotinin, and 0.5 mM phenylmethylsulfonyl fluoride) on ice for 1 h. Cell lysate was obtained after removing the unsolubilized materials by centrifugation twice in a microcentrifuge at 4°C each for 30 min.
Immunoprecipitation and Immunoblotting Analysis-Cell lysates were incubated with protein G-or protein A-Sepharose (30 l) that was pre-washed with PBS and pre-bound with 5-10 g of anti-c-Myc, anti-RyR1 (13C2) (24), anti-RyR2, or anti-RyR3 antibodies at 4°C for 17 -19 h. The anti-RyR2 antibody was produced in rabbits against a synthetic peptide, QKLRQLHTHRYGE, which corresponds to residues 4460 -4472 of rabbit RyR2. Generation of the anti-RyR3 antibody was described previously (11). The protein G-or protein A-agarose beads were washed with ice-cold lysis buffer three times, each time for 10 min. The immunocomplexes bound to the agarose beads were solubilized by addition of 30 l of 2ϫ Laemmli sample buffer (25) plus 5% ␤-mercaptoethanol and were boiled at 100°C for 2 min. The solubilized immunocomplexes (10 -15 l) were separated by 6% SDS-PAGE. The SDS-PAGE-resolved proteins were then transferred to nitrocellulose membranes at 45 mV for 18 -20 h at 4°C in the presence of 0.01% SDS according to Towbin et al. (26). The nitrocellulose membrane was blocked for 0.5 h with PBS containing 0.5% Tween 20 and 5% skim milk powder. The blocked membrane was incubated with primary antibodies, anti-RyR (34C), anti-RyR2, anti-RyR3, or anti-RyR3 (FP8) (16) for 2-4 h, washed three times each time for 15 min with PBS containing 0.5% Tween 20. The membrane was then incubated with the secondary anti-mouse or anti-rabbit IgG (H&L) antibodies conjugated with alkaline phosphatase for 30 -40 min. After washing three times each time for 15 min, the bound antibodies were visualized by the alkaline phosphatase-mediated color reaction using nitro blue tetrazolium and 5-bromo-4-chloro-3-indolyl phosphate as the substrates.
[ 3 H]Ryanodine Binding to Immunoprecipitates-The c-Myc-tagged RyR proteins expressed in HEK293 cells were immunoprecipitated by protein G-agarose that was pre-bound with anti-c-Myc antibody. The immunoprecipitates were washed three times with lysis buffer. Equilibrium [ 3 H]ryanodine binding was carried out in a binding mixture (300 l) containing 15 l of immunoprecipitates (agarose beads), 25 mM Tris/50 mM Hepes (pH 7.4), 500 mM KCl, 0.8 mM CaCl 2 , 5 mM caffeine, 5 nM [ 3 H]ryanodine, and the protease inhibitor mix was incubated at 37°C for 2-3 h. The binding mix was filtered through Whatman GF/B filters presoaked with 1% polyethyleneimine. The filters were immediately washed four times with 5 ml of ice-cold washing buffer containing 25 mM Tris (pH 8.0) and 250 mM KCl. The radioactivity associated with the filters was then determined by liquid scintillation counting, while nonspecific binding was determined by measuring [ 3 H]ryanodine binding in the presence of 50 M unlabeled ryanodine. All binding assays were done in duplicate.
Ca 2ϩ Release Measurements in Transfected HEK293 Cells and Single Channel Recordings-Free cytosolic Ca 2ϩ concentration in transfected HEK293 cells was measured with the fluorescence Ca 2ϩ indicator dye fluo-3-AM as described previously (19). Recombinant RyR proteins were purified from whole cell lysate by sucrose density gradient centrifugation and were used for single-channel recordings as described previously (19). Free Ca 2ϩ concentrations were calculated using the computer program of Fabiato and Fabiato (27).

RyR2 Interacts Physically with RyR3 in HEK293 Cells-To
test the possibility that RyR2 is capable of forming heteromeric channels with RyR3, we first investigated whether RyR2 interacts with RyR3 physically. RyR2 and RyR3 have a comparable molecular weight and migrate similarly in SDS-PAGE. To facilitate specific detection and isolation of RyR2 and RyR3, we inserted a c-Myc tag into RyR2 and RyR3. The c-Myc-tagged RyR2, RyR2 (c-Myc), or c-Myc-tagged RyR3, RyR3 (c-Myc), was expressed in HEK293 cells either alone or in combination with the untagged RyR3 (wt) or untagged RyR2 (wt), respectively. Immunoprecipitation was carried out using the anti-c-Myc antibody, and the immunoprecipitates were separated and probed with either the anti-RyR (34C) antibody, which recognizes all three RyR isoforms, or an RyR3-specific antibody, anti-RyR3 (FP8). Fig. 1 shows that the anti-c-Myc antibody was able to pull down the RyR2 (c-Myc) protein from a lysate of HEK293 cells transfected with RyR2 (c-Myc) (Fig. 1A, lane 1), but not RyR3 (wt) protein from cells transfected with RyR3 (wt) (Fig.  1A, lane 2), indicating that the anti-c-Myc antibody is specific. The anti-c-Myc antibody also pulled down a major band, which should contain the RyR2 (c-Myc) protein, from HEK293 cells co-transfected with RyR2 (c-Myc) and RyR3 (wt) (Fig. 1A, lane  3). The same anti-c-Myc immunoprecipitates were blotted with the anti-RyR3 (FP8) antibody (Fig. 1B). This antibody recognized a major band in the anti-c-Myc immunoprecipitate from HEK293 cells co-transfected with RyR2 (c-Myc) and RyR3 (wt) (Fig. 1B, lane 3). No significant signal was detected by the anti-RyR3 (FP8) antibody in the immunoprecipitate from cells transfected with RyR2 (c-Myc) alone (Fig. 1B, lane 1). These data indicate that the anti-RyR3 (FP8) antibody is specific and that RyR3 (wt) was co-precipitated with RyR2 (c-Myc) by the anti-c-Myc antibody. It should be noted that caffeine and ryanodine induced Ca 2ϩ release from intracellular Ca 2ϩ stores in HEK293 cells transfected with RyR2 (c-Myc) or RyR3 (c-Myc), indicating that both RyR2 (c-Myc) and RyR3 (c-Myc) are functional ( Fig. 1, E and F).
We also performed the reciprocal experiment in which the RyR3 (c-Myc) was immunoprecipitated by the anti-c-Myc antibody and the presence of RyR2 (wt) in the anti-c-Myc immunoprecipitates was detected by Western blotting using an RyR2specific antibody. As shown in Fig. 1C, the anti-c-Myc antibody   binding activity from HEK293 cells transfected with RyR2 (wt) alone (data not shown).
We reasoned that, if RyR2 is capable of interacting with RyR3, co-expression of RyR3 (wt) with RyR2 (I4827T, c-Myc) should also be able to restore some [ 3 H]ryanodine binding to the mutant. As shown in Fig. 2B, the anti-c-Myc antibody was able to precipitate [ 3 H]ryanodine binding activity from HEK293 cells co-transfected with RyR3 (wt) and RyR2 (I4827T, c-Myc). Because the anti-c-Myc antibody does not precipitate [ 3 H]ryanodine binding activity from cells transfected with RyR3 (wt) alone (data not shown), the [ 3 H]ryanodine binding detected in the immunoprecipitates must result from binding to the RyR3 (wt)⅐RyR2 (I4827T, c-Myc) complexes. These observations indicate that RyR2 and RyR3 can form a functional complex.
RyR2 Is Able to Form Functional Heteromeric Channels with RyR3 in HEK293 Cells-The observed physical and functional RyR2-RyR3 interaction could occur between homomeric RyR2 and homomeric RyR3 channels or via the formation of heteromeric RyR2⅐RyR3 hybrid channels. To distinguish these possibilities, we made use of the RyR2 (I4827T) mutant and a Ca 2ϩ sensing mutant of RyR3, RyR3 (E3885A) (20). Neither mutant when expressed alone in HEK293 cells exhibited caffeine-induced Ca 2ϩ release (Fig. 3, A and B). If RyR2-RyR3 interaction occurs only between homomeric RyR2 and homomeric RyR3 channels, the RyR2 (I4827T)⅐RyR3 (E3885A) mutant complex formed after co-expression in HEK293 cells should remain insensitive to caffeine. On the other hand, if RyR2-RyR3 interaction occurs between monomeric RyR2 and monomeric RyR3, co-expression of RyR2 (I4827T) and RyR3 (E3885A) in HEK293 cells may produce hybrid channels that are caffeine-sensitive, because these two mutations are located in different regions of RyR and one mutant may be able to complement the defect of the other. As shown in Fig. 3, HEK293 cells co-transfected with different ratios of RyR2 (I4827T) and RyR3 (E3885A) displayed caffeine-induced Ca 2ϩ release (Fig. 3, C-E), indicating the formation of heteromeric RyR2⅐RyR3 hybrid channels. The immediate drops in fluorescence signals after addition of caffeine are due to fluorescence quenching by caffeine.
Single RyR2⅐RyR3 Heteromeric Channels Are Sensitive to Modulation-To assess whether RyR2⅐RyR3 heteromeric channels are regulatable, we examined the effect of Ca 2ϩ , ATP, caffeine, Mg 2ϩ , and ryanodine on single RyR3 (wt)/RyR2 (G4824A) hybrid channels with intermediate single-channel conductances. Fig. 5a shows the channel activity of a single RyR3 (wt)/RyR2 (G4824A) heteromeric channel with a unitary conductance of 269-pS (probably comprised of two RyR3 (wt) and two RyR2 (G4824A) subunits). The heteromeric channel was sensitive to Ca 2ϩ (Fig. 5, b, c, and g), activated by ATP (Fig.  5d) and caffeine (Fig. 5e), inhibited by Mg 2ϩ (Fig. 5f), and modulated by ryanodine (Fig. 5h). Similar ligand responses were also observed in other single-hybrid channels with different unitary conductances (not shown). Hence RyR2 and RyR3 are able to form functional and regulatable single heteromeric channels in HEK293 cells.
RyR2 Interacts Physically with RyR1 in HEK293 Cells-The finding that RyR2 interacts physically with RyR3 raises the question of whether RyR2 is also able to interact with RyR1. To address this question, we co-expressed RyR2 (c-Myc) and RyR1 in HEK293 cells and examined possible interactions between them by immunoprecipitation and Western blotting. Because RyR2 migrates faster than RyR1 in SDS-PAGE, RyR1-RyR2 interaction could be readily detected by immunoprecipitation followed by SDS-PAGE. Fig. 6A shows that the anti-c-Myc antibody did not pull down the RyR1 protein from HEK293 cells transfected with RyR1 alone (Fig. 6A, lane 2), but an anti-RyR1 (13C2) antibody did (Fig. 6A, lane 1). The anti-c-Myc antibody also did not cross-react with RyR1 on Western blot (Fig. 6A, lane 3). On the other hand, the anti-c-Myc antibody precipitated two RyR bands from HEK293 cells co-transfected with RyR2 (c-Myc) and RyR1 (Fig. 6B, lane 2). The bottom band (indicated by an arrowhead), which was recognized by both the anti-RyR (34C) and the anti-c-Myc antibodies, corresponds to RyR2 (c-Myc), whereas the top band (indicated by an arrow), which was recognized by the anti-RyR (34C) antibody but not by the anti-c-Myc antibody, corresponds to RyR1 (Fig. 6B, lanes  2 and 4). These data indicate that RyR1 was co-precipitated with RyR2 (c-Myc) by the anti-c-Myc antibody. It should be noted that the anti-RyR1 (13C2) antibody, although raised against RyR1-specific sequence, exhibits some cross-reactivity with the RyR2 and RyR3 isoforms in both immunoprecipitation and immunoblotting (data not shown). Thus, the presence of RyR2 (c-Myc) in the anti-RyR1 (13C2) immunoprecipitate (Fig.  6B, lane 3) is due, in part, to direct immunoprecipitation of RyR2 (c-Myc) by the anti-RyR1 (13C2) antibody in addition to co-precipitation.
We also examined whether RyR1 interacts with RyR3 when co-expressed in HEK293 cells. Unlike the double bands seen in Fig. 6B (lane 2), only a single RyR band was detected by the anti-RyR (34C) antibody in the anti-c-Myc immunoprecipitate from cells co-transfected with RyR1 and RyR3 (c-Myc) (Fig. 6C,  lane 2), although a comparable amount of RyR1 protein was present in the lysate (Fig. 6C, lane 1). This band corresponds to RyR3 (c-Myc), because it was also recognized by the anti-c-Myc antibody (Fig. 6C, lane 4). These results indicate that RyR1 was not co-precipitated with RyR3 (c-Myc) by the anti-c-Myc antibody. A faint band, corresponding to RyR3 (c-Myc), was detected in the anti-RyR1 (13C2) immunoprecipitate (Fig. 6C,  lane 3). This is most likely due to direct immunoprecipitation of RyR3 (c-Myc) by the nonspecific anti-RyR1 (13C2) antibody.

C-E).
Although the caffeine-induced Ca 2ϩ transients in cotransfected cells are relatively small, they are easily differentiated from the flat signals observed in individually transfected cells. The appearance of caffeine-sensitive Ca 2ϩ release in cotransfected cells most likely arises from hybrid channels formed by these two mutants. These observations suggest that RyR2 can also form functional heteromeric channels with RyR1 in HEK293 cells.
RyR1 Does Not Form Functional Heteromeric Channels with RyR3 in HEK293 Cells-To test the specificity of the HEK293 cell expression system for detecting functional interaction between RyR isoforms, we carried out similar co-expression studies using the mutants RyR3 (I4732T), which is equivalent to RyR2 (I4827T), and RyR1 (E4032A). Both mutants RyR3 (I4732T) and RyR1 (E4032A) when expressed alone in HEK293 cells displayed no caffeine-sensitive Ca 2ϩ release (Fig. 8, A and  B). Co-expression of these mutants in different ratios did not restore caffeine-induced Ca 2ϩ release (Fig. 8, C-E), indicating that mutants RyR3 (I4732T) and RyR1 (E4032A) are unable to complement each other. On the other hand, the same RyR3 (I4732T) and RyR1 (E4032A) mutants were able to restore the caffeine response of mutants of RyR2, RyR2 (E3987A) (Fig. 8, F-I) and RyR2 (I4827T) (Fig. 7), respectively. These observations confirm the functional interaction between RyR2 and RyR3 and indicate that RyR1 and RyR3 expressed in HEK293 cells do not form heteromeric channels as in diaphragm skeletal muscle. DISCUSSION It is commonly believed that RyRs exist in the form of homotetramers. This is unexpected, given the existence of multiple RyR isoforms and their similar tetrameric structure, and the fact that another type of Ca 2ϩ release channels related to RyRs, IP 3 Rs, are able to form heteromeric complexes (3,4,12,13). Structural incompatibility for forming heteromers or nonoverlapping cellular distribution of RyR isoforms may account for a lack of detectable heteromeric RyR channels. The purpose of this study is to investigate the ability of RyR isoforms to form heteromeric channels when co-expressed in HEK293 cells. Our results demonstrate for the first time that RyR2 is able to form heteromeric channels with RyR3 and RyR1.
Based on the following observations, it is unlikely that the observed interactions between RyR2 and RyR3 and between RyR2 and RyR1 in HEK293 cells are the result of overexpres-sion. First of all, the interaction is isoform-specific. Co-expression of RyR1 and RyR3 in HEK293 cells did not result in the formation of RyR1⅐RyR3 complexes, whereas RyR2⅐RyR3 and RyR2⅐RyR1 complexes were detected after co-expression. Second, the RyR2⅐RyR3 complexes are capable of binding [ 3 H]ryanodine, indicating that they are functional. Third, co-expression of mutants RyR2 (I4827T) and RyR3 (E3885A), RyR2 (E3987A) and RyR3 (I4732T), or RyR2 (I4827T) and RyR1 (E4032A) rescued each other, indicating that these mutants are able to complement each other functionally. On the other hand, co-expression of mutants RyR1 (E4032A) and RyR3 (I4732T) in HEK293 cells did not rescue caffeine-induced Ca 2ϩ release. Fourthly, co-expression of RyR2 and RyR3 led to the formation of functional and regulatable single heteromeric channels. Thus, it is most unlikely that these heteromeric RyR channel complexes observed in HEK293 cells are the result of a nonspecific aggregation of RyR monomers of different isoforms caused by overexpression.
The lack of detectable RyR1⅐RyR3 complexes after co-expression in HEK293 cells is consistent with the observation that RyR1 and RyR3 exist as homotetrameric channels in diaphragm skeletal muscle (7,8). On the other hand, our finding that RyR2 is able to form heteromeric channels with RyR3 and RyR1 appears to be inconsistent with the observation that RyR2 and RyR3 exist as homotetramers in the brain where all three RyR isoforms are expressed (11). One possible explanation for this apparent discrepancy is that RyR2 and RyR3 may be localized in different brain cells. It has been shown by in situ hybridization analysis that each RyR isoform displays distinct but overlapping patterns of expression in different brain regions (5). However, within the overlapping regions it is unknown whether different RyR isoforms are expressed in the same brain cells. An alternative explanation is that formation (E4032A)(4 g) (E). Fluorescence intensity of the fluo-3-loaded cells was monitored continuously before and after addition of 5 mM caffeine (letter C) as described in the legend to Fig. 3. A small transient increase in fluorescence following the initial sharp reduction was detected in C, D, and E, but not in A and B after addition of caffeine. Similar results were obtained from three separate experiments.
Recently, it has been shown that several immune cells, including U937, SKW6.4, SupT1, and Jurkat treated with transforming growth factor-␤, express multiple RyR isoforms (30). We have attempted to determine whether heteromeric RyR channels exist in these cells. We have performed immunoprecipitation studies on cell lysates prepared from these immune cells using RyR2-and RyR3-specific antibodies. We found that the levels of RyR expression in these cells were too low to allow us to determine whether the cross-activity was due to co-precipitation or nonspecific antibody binding (data not shown). Similar results were also observed using microsomal membranes isolated from bovine uterus. Better RyR isoform-specific antibodies would be required for determining the existence of heteromeric RyR channels in native cells or tissues.
Although no heteromeric RyR channels have been isolated biochemically, their presence has been functionally implicated in vascular smooth muscle cells where the expression of all three RyR isoforms has been detected (14,31). For instance, in rat portal vein myocytes, which express all three RyR isoforms, inhibition of either RyR1 or RyR2 by isoform-specific antisense oligonucleotides blocked depolarization-induced Ca 2ϩ sparks and global Ca 2ϩ response (14). These observations suggest that both RyR1 and RyR2 are required to form the functional Ca 2ϩ release units in these cells, although how these Ca 2ϩ release units are formed by two different RyR isoforms is unclear. These functional Ca 2ϩ release units could be composed of homotetrameric RyR1 and RyR2 channels somehow coupled both structurally and functionally, or of heteromeric RyR1/RyR2 channels. Our finding that RyR2 is able to form heteromeric channels with RyR1 when expressed together makes the latter possibility likely. Functional Ca 2ϩ release units comprised of multiple RyR isoforms have also been implicated in cerebral artery smooth muscle cells. It was reported that the frequency of Ca 2ϩ sparks and spontaneous transient outward currents in cerebral artery smooth muscle cells isolated from RyR3-deficent mice was significantly higher than that in wild type cells (15). These data suggest that RyR3 is part of the functional Ca 2ϩ release units formed by RyR2 or RyR1 and regulates their activity negatively.
These functional studies of Ca 2ϩ release units in smooth muscles not only implicate the existence of heteromeric RyR channels, but also provide some clues as to their physiological roles. RyR1 is known to interact physically with the voltagedependent L-type Ca 2ϩ channel and to be activated by voltage directly (2). On the other hand, RyR2 is activated by Ca 2ϩ via a mechanism known as Ca 2ϩ -induced Ca 2ϩ release (CICR) (32). RyR1 can also be activated by Ca 2ϩ , but CICR in RyR1 appears to be suppressed in vivo (33). Considering these different properties between RyR1 and RyR2, it is reasonable to speculate that heteromeric RyR1/RyR2 channels would be able to interact with the L-type Ca 2ϩ channels and be activated by Ca 2ϩ -induced Ca 2ϩ release. In this regard, it is of interest to know that, although depolarization-induced Ca 2ϩ sparks and global Ca 2ϩ response were abolished by suppressing either RyR1 or RyR2 in vascular myocytes, caffeine is still able to trigger Ca 2ϩ response in the same cells and that this depolarization-induced Ca 2ϩ response is dependent on Ca 2ϩ influx (14). These observations suggest that the RyR1/RyR2 Ca 2ϩ release units may be preferentially localized to the superficial sarcoplasmic reticulum that is in close association with the plasma membrane and be activated by Ca 2ϩ influx through the L-type Ca 2ϩ channels upon depolarization. Thus, by forming heteromeric RyR1/RyR2 channels, functional Ca 2ϩ release units with strong CICR (a property of RyR2) could be targeted specifically to the plasma membrane via physical interaction with the L-type Ca 2ϩ channels (a property of RyR1), and thus coupled to depolarization.
The physiological role of RyR3 in smooth muscle is largely undefined. The observation that abolition of RyR3 increases the frequency of Ca 2ϩ sparks in smooth muscles suggests that RyR3 may normally inhibit the activity of Ca 2ϩ release units (15). RyR3 expressed in smooth muscle appears to function differently from that expressed in skeletal muscle. It does not produce Ca 2ϩ sparks (14,34) and becomes active only under the condition of sarcoplasmic reticulum Ca 2ϩ overload (35). Considering these unique properties of smooth muscle RyR3 and our observation that RyR3 is able to form heteromeric channels with RyR2, it is possible that RyR3 in some smooth muscles may function as a suppressor of RyR2-mediated Ca 2ϩ release by forming non-sparking RyR2⅐RyR3 heteromeric channels with a decreased sensitivity to activation by luminal Ca 2ϩ . Further investigation is required to determine the molecular basis that confers these unique properties of smooth muscle RyR3 and whether smooth muscle RyR3 suppresses the activity of RyR2.