Molecular Cloning of cDNA Encoding Human and Rabbit Forms of the Ca2’ Release Channel (Ryanodine Receptor) of Skeletal Muscle Sarcoplasmic Reticulum*

We have cloned cDNAs encoding the rabbit and hu- man forms of the Ca2+ release channel of sarcoplasmic reticulum. The human cDNA encodes a protein of 5032 amino acids, with a molecular weight of 563,584, which is made without an NHz-terminal signal se- quence. Amino acid substitutions between rabbit and human sequences were noted in 163 positions and deletions or insertions in eight regions accounted for ad- ditional sequence differences between the two pro- teins. Analysis of the sequence indicates that 10 poten- tial transmembrane sequences in the COOH-terminal fifth of the molecule and two additional, potential transmembrane sequences nearer to the center of the molecule could contribute to the

We have cloned cDNAs encoding the rabbit and human forms of the Ca2+ release channel of sarcoplasmic reticulum.
The Ca*+ release from fractions of the sarcoplasmic reticulum containing terminal cisternae has been characterized extensively during the past decade (Miyamoto and Racker, 1982;Morii and Tonomura, 1983;Seiler et al., 1984;Meissner, 1984; * This research was suonorted bv grants (to D. H. M.) from the Medical Research Councii-of Canada-and the Muscular Dystrophy Association of Canada (MDAC) and by a grant (to G. M.) from the National Institutes of Health. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
5 Postdoctoral Fellow of the Canadian Heart Foundation. i Postdoctoral Fellow of the Muscular Dystrophy Association of Canada.
$$ Postdoctoral Fellow of Muscular Dystrophy Association.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBankTM/EMBL Data Bank with accession number(s) 505200. Meissner, 1986;Meissner et al., 1986). Rapid Ca2+ release from isolated vesicles is activated by micromolar Ca2+ and millimolar adenine nucleotides and inhibited by millimolar M%+. Calmodulin at micromolar concentrations partially inhibits Ca2+ release, apparently by direct protein-protein interaction with the Ca'+ release channel. Single channel measurements in planar bilayers (Smith et al., 1986) have shown that Ca2+ release from the sarcoplasmic reticulum is mediated by a ligand-gated channel with a conductance greater than 100 pS in 50 mM Ca'+. Identification and isolation of the Ca2+ release channel were facilitated through the use of the plant alkaloid, ryanodine (Jenden and Fairhurst, 1969), which was shown to bind to the protein with high affinity and to modulate its function (Seiler et al., 1984;Fleischer et al., 1985;Pessah et al., 1985;Pessah et al., 1986;Inui et al., 1987a;Campbell et al., 1987;Lai et al., 1987;Lai et al., 1988a). Single channel recordings showed that purified ryanodine receptor preparations, comprised of tetrameric complexes of a single major polypeptide of M, 350,000-450,000, exhibited an intrinsic Ca2+ channel activity that was modulated by Ca'+, ATP, and M%+ Hymel et al., 1988;Smith et al., 1988;Lai et al., 1988a) in a manner similar to native Ca2+ release channels (Smith et al., 1985). Analysis of the stoichiometry and subunit composition of the 30 S ryanodine receptor complex indicates that it is a cooperatively coupled, negatively charged homotetramer .
Studies of the morphology of the ryanodine receptor (Inui et al., 1987a;Lai et al., 1988a;Saito et al., 1988;Wagenknecht et al., 1989) have shown it to have an exquisite quatrefoil structure, with hydrophobic segments of the four identical subunits forming a putative membrane-spanning baseplate structure, and hydrophilic segments forming a cytoplasmic domain that surrounds and decorates the central baseplate. Three-dimensional image reconstruction (Wagenknecht et al., 1989) suggests the presence of four internal channels which branch from a common origin above the baseplate and open into vestibules in the four quarters of the tetramer. The morphology of the purified Ca2+ release channel shows that it makes up the "feet" structures (Franzini-Armstrong, 1970; ization, and involvement in disease. Takeshima et al. (1989) have reported the cloning and expression of cDNA encoding the rabbit skeletal muscle ryanodine receptor and Marks et al. (1989) have reported the isolation of partial clones encoding the rabbit protein. We have reported the cloning of cDNA encoding full length rabbit and human skeletal muscle ryanodine receptors and the localization of the ryanodine receptor gene to the proximal long arm of human chromosome 19 (Zorzato et al., 1989a;MacLennan et al., 1989;MacKenzie et al., 1989). We have shown that the ryanodine receptor gene is not involved in myotonic dystrophy but that it is a candidate for the defective gene in malignant hyperthermia. We now report the full nucleotide and deduced amino acid sequences for the human ryanodine receptor cDNA. These sequences will be of great interest as studies of the involvement of the ryanodine receptor gene in human muscle disease unfolds. Other manipulations of mRNA and DNA were carried out using standard protocols (Maniatis et al., 1982) with the exception of DNA transfers to nitrocellulose. In this case, DNA was transferred electrophoretically to Zeta-Probe nylon membranes for 60 min at 50 V using the transfer apparatus described by Gershoni et al. (1980) (Zorzato et al., 198913) specific for the Ca2+ release channel.
Screening of the library was carried out by the method of Young and Davis (1983), essentially as described by Leberer et al. (1989a). The screening of 3 X lo6 recombinants led to the isolation of two cDNA clones in the region defined by nucleotides 14280-14629 and 13434413758 in Fig. 1. Analysis of the sequences of these clones showed that both were rearranged when compared with the linear sequence of the human cDNA, which was by then available in our laboratory.
Potential calmodulin-binding sites were identified as predicted a-helices containing clusters of 2-4 positive charges, separated by a predominantly hydrophobic region, which are the usual requirements for a calmodulin-binding site (Buschmeier et al., 1987;Harris et al., 1988;Lear et al., 1988). There is no sequence generally diagnostic of a nucleotide-binding site, although many contain a glycine-rich loop GXGXXG(KT) following the first pstrand of a P-a-@ alternation (Wierenga and Hol, 1983). Glycine-rich loops are found in many other situations and, since 90% of known nucleotide sites belong to a family of parallel p sheets, many irrelevant candidates were eliminated by restricting the search to predicted @ol-/3 regions.
Potential transmembrane segments were identified by the methods of Kyte and Doolittle (1982) and Engelman et al. (1986)  brary, we isolated clones whose expressed product reacted with an antibody specific for the Ca*+ release channel protein (Zorzato et al., 1989b). The fusion protein expressed by these isolated clones also reacted with a second antibody raised against the purified 30 S ryanodine receptor . As supporting evidence that we had isolated the correct clones, both rabbit and human probes from the coding region of the DNA hybridized to a message of about 15 kb in rabbit muscle mRNA (Fig. 2, A and B). These observations provided evidence that we had cloned cDNA encoding the ryanodine receptor. As we extended these cDNAs and analyzed their sequences, we found four deduced amino acid sequences that corresponded to the sequences of peptides isolated from the purified ryanodine receptor. These sequences are underlined in Fig. 1B. As further supporting evidence that the clones encoded the ryanodine receptor, we noted that the deduced amino acid sequence would give rise to a protein with several transmembrane passages at the carboxyl-terminal end and that the bulk of the protein was hydrophilic.
Such a protein would match very well with the structure of the ryanodine receptor in which the bulk of the protein is cytoplasmic and only a small segment is transmembrane (Wagenknecht et al., 1989). The simultaneous sequencing of both rabbit and human cDNAs gave us confidence that cloning artifacts did not arise in these very long sequences. cDNA Sequence-In Fig. lA we present the restriction map and sequencing strategy for the human ryanodine receptor cDNA. In Fig. lB, we present the nucleotide and deduced amino acid sequences. The sequence was determined from linear clones 1,3,4,5,6, and 7 in Fig. L4, which abutted each other at EcoRI sites between clones 1 and 3, 3 and 4, and 5 and 6. The junction between clones 1 and 3 was sequenced in clone 2, which contained an intact EcoRI restriction site, and Poly(A)' RNA from neonatal rabbit tissues skeletal muscle was separated in formaldehyde-agarose gels and transferred to Hybond nylon membranes. Ribosomal RNA bands corresponding to 4.7 and 1.9 kb are indicated. A, hybridization of neonatal rabbit skeletal muscle mRNA to a human ryanodine receptor genomic DNA fragment covering the cDNA clone g-clone 7 boundary, as indicated in Fig. 1. The arrowhead indicates the size of the dystrophin transcript (14 kb, Koenig et al., 1987) which was analyzed in the same blot. B, hybridization of neonatal rabbit skeletal muscle mRNA to a rabbit ryanodine receptor cDNA probe (residues 8612-9215). junctions between clones 3 and 4, 4 and 5 (a gap was introduced in this segment of cDNA by a second oligo(dT) primer initiation site), and 5 and 6 were obtained through sequencing of genomic DNA clones isolated from the human chromosome 19 library. The region around the EcoRI site in clones 6 and 7 was also sequenced in genomic DNA.
The 3'-untranslated region, beginning after the TGA termination codon, was 142 bp long. A canonical AAAATAAA polyadenylation signal (Proudfoot and Brownlee, 1976) was found 19 bases upstream of the polyadenylation site and this was followed closely by the TG-rich sequence TCTGTCGT-ACG, characteristic of sequences between the polyadenylation signal and the polyadenylation site (McLauchlan et al., 1985). The initiator methionine was found 15,096 bp upstream of the termination codon. The initiator methionine codon was present in the longer sequence ACATCATGG which closely resembles the consensus initiation sequence, CCA(G)CCA-TGG (Kozak, 1984). Although the 105 bp upstream of the initiator methionine were in frame, the sequence contained about 75% G+C residues, characteristic of 5'-untranslated sequences in other sarcoplasmic reticulum protein cDNAs (MacLennan et al., 1985;Brand1 et al., 1986). We were unable to obtain an NH*-terminal amino acid sequence from the ryanodine receptor protein, which would have been helpful in defining the position of the initiator methionine in the cDNA sequence, and we conclude that the NHz-terminal methionine is blocked in the mature protein. The study of Takeshima et al. (1989), in which the cap site for the rabbit mRNA was found to lie 138 residues upstream of the initiator methionine codon, is our final guide to the placement of the initiator methionine in Fig. 1. Amino Acid Sequence Analysis-Our human cDNA sequence encoded a protein of 5,032 amino acids with a molecular weight of 563,584. This deduced molecular mass is considerably larger than that previously predicted for the ryanodine receptor, on the basis of its mobility in sodium dodecyl sulfate gels. It is, however, consistent with the large mass of the foot protein (Saito et al., 1988) and with measurements of ryanodine binding to the purified tetrameric protein (Lai et al., 1988).
In Fig. 3, third lane, we present the deduced amino acid sequence of the human ryanodine receptor in a single-letter code. In the second line (above the linear sequence), we have indicated positions where we found differences with the rabbit sequence. In the first line (two above the linear sequence) we have noted discrepancies between our rabbit cDNA sequence and that presented by Takeshima et al. (1989). Two discrepancies are recorded. We did not find the sequence -Ala-Gly-Asp-Ala-Gln-, recorded by Takeshima et al. (1989) as residues 3481-3485, in either our rabbit or human cDNA sequences. Residue 2015 was recorded as Glu by Takeshima et al. (1989), but we find Asp at this position in rabbit and Gly at this position in the human sequence.
Amino acid differences between the rabbit and human sequences are scattered throughout the molecule and involve several deletions and/or additions. This is pronounced in the Glu-rich region lying between residues 1872 and 1923. In this region of the rabbit sequence, a single stretch of 18 Glu is found and the whole region has 39 Glu and 4 Asp residues. In the comparable human sequence there are 35 Glu and 4 Asp residues. A deletion of 3 Glu in a row and then 1 Glu in the human sequence are later compensated for by the inclusion of the sequence Thr-Ala-Gln and later still by the inclusion of a Gly so that, over the course of about 40 residues, the sequences realign. Overall, we noted 163-amino acid substitutions between the two proteins, while deletions or insertions in eight different positions accounted for many more differences between the two proteins.
Trunsmembrune Sequence Predictions-The hydropathy of the deduced amino acid sequence of the human ryanodine receptor is presented in Fig. 4. We identified 11 sequences which were long enough to span the membrane, gave mean hydropathy indices for a window of I7 amino acids ranging from 1.2 to 2.9, and are candidates for transmembrane sequences located in the baseplate of the protein. A 12th potential transmembrane sequence (sequence 9) is also included in Figs. 3 and 5. Although its hydropathy index is only 0.8, largely because of a cluster of 4 glycine residues which do not score highly in hydropathy plots but are compatible with a hydrophobic environment, this sequence resembles the other putative transmembrane sequences in other respects. M' and M" are isolated in the central region of the molecule in residues 3123-3134 and 3187-3205, while sequences Ml to Ml0 are distributed in the COOH-terminal fifth of the molecule between residues 3978 and 4932. In the amino acid sequences of these potential transmembrane sequences listed in Fig. 5, the charged and amide-containing residues are boxed. These sequences could form six pairs of transmembrane passages with loops in the lumen of the sarcoplasmic reticulum, 44, 18, 40, 67, 16, and 19 amino acids in length, progressing from NH, to COOH termini.
Homology Searches-A search of the EMBL/GenBank data base has revealed relatively little sequence identity between the ryanodine receptor and any other protein.
The most conspicuous identities were with sequences from the predicted transmembrane regions of the ryanodine receptor and the Na' (Noda et al., 1984) and Ca2+ (Tanabe et al., 1987) channels and the acetylcholine receptor (Noda et al., 1983) (Fig. 3). Repeated Segments-We observed a repeating sequence motif of 114-120 amino acids, averaging 28% identity, and occurring four times in two doublets. The first and second repeats are 114 residues long and include residues 841-954 and 955-1068. The third and fourth repeats are 120 residues long and include residues 2725-2844 and 2845-2958. In order to make the appropriate alignment, a B-residue gap was introduced into repeat sequence 4 between residues 2899 and 2900. A 16-residue segment of the longer motif is repeated twice more in residues 1344-1359 and 1371-1386. The repeated segments are aligned in Fig. 6. The sequence p-al pattern, with the possibility of a third p-a unit if the final identity is not high, but it is sufficient to imply a common helix is extended and interrupted by a short @ strand. TWO tertiary structure (Chothia and Lesk, 1986). Although the units in tandem would be predicted to give a viable parallel B predicted secondary structure (Fig. 3) for the four repeats sheet domain. differ in some segments, the consensus (Fig. 6) shows a ,&I- The "profile" method of Gribskov et al. (1987) was used to FIG. 4. Hydropathy profile of the human ryanodine receptor.
The hydropathy plot (Kyte and Doolittle, 1982)   scan the NBRF data base to see if any similar segments occurred elsewhere. The highest score of 37 (0.34 of the maximum score) was given by a segment of vinculin, but the resemblance was not significant. Ligand Rinding Sites-The Ca*+ release channel is modulated by four physiologically relevant molecules; Ca'+, MC, ATP, and calmodulin (Meissner, 1986;Meissner et al., 1986;Morii and Tonomura, 1983). Therefore, it is of interest to determine where in the sequence these ligands bind, especially since consensus high affinity binding sequences are available for three of them.
Ca2+ is bound with high affinity to EF hand structures in many proteins (Kretsinger, 1987). No sequence with the precise characteristics of an EF hand is present in either the human or rabbit ryanodine receptors, although several sequences meet the requirements in part. The maximum score obtained for all potential sequences was 750, well below the score that we consider to be significant. The glutamate-rich region between residues 1872 and 1923 is of interest as a potential low affinity Ca2+-binding domain, in light of the capacity of such sequences to bind Ca2+ with low affinity (Fliegel et al., 1987;Leberer et al., 1989b). Magnesium-binding sites have not yet been defined and cannot be predicted in the primary structures of proteins.
We were also unable to detect a convincing consensus sequence for adenine nucleotide binding in the primary sequences of the rabbit and human receptors. GXGXXG motifs were found at positions 1194-1199 and 4447-4452 in Fig. 1. The first lies in a region that is predicted to contain several @ strands. The second occurs in a glycine-proline-rich region where proline bends might give a rather rigid structure, unlike that of a typical nucleotide-binding site.
Potential calmodulin-binding sites made up from an amphipathic helix, with two clusters of positive charges separated by a hydrophobic region, were found between residues 2807 and 2840,2909 and 2930, and 3031 and 3049. These sites are indicated in Fig. 3. One of these (2909-2930) showed some sequence homology to the calmodulin-binding site of /3 spectrin (Harris et al., 1988).
The sites of CAMP and calmodulin-dependent protein kinase phosphorylation have been well characterized (Feramisco et al., 1980). Potential phosphorylation sites were found in the protein at residues 3940-3945 and 4314-4317. Glycosylation sites are composed of the sequence N-X-T(S). We found N-X-T(S) sequences which would be glycosylated at positions 1064, 2773, 3127, 3943, 4142, and 4859 in the sequence, provided they were luminally located.

DISCUSSION
In this paper we have described the cloning and sequence analysis of cDNAs encoding one of the largest proteins analyzed to date, the rabbit and human forms of the ryanodine receptor. The rabbit and human sequences were found to be very similar. Major differences were found in an extremely acidic region of the protein (residues 1872-1923 in Fig. 1) in which several deletions and insertions were noted between the two sequences. We found several long amino acid sequences (residues 2948-3293, 3764-4096, 4534-5032) which exhibited complete identity with the corresponding rabbit sequence. The first conserved region contains putative calmodulin binding sequences, whereas the last two sequences contain putative transmembrane sequences Mi and Mz and Ms to MiO. In other regions, the differences were minor and fully consistent with species differences in the same gene, as opposed to differences between different genes. We noted only one significant difference between our rabbit cDNA sequence and that published by Takeshima et al. (1989). A stretch of five amino acids reported in their sequence was absent from ours. We cannot readily explain this discrepancy as an allelic variation. It may represent an alternative splicing of a small exon or the retention of an unexcised intron in the Takeshima sequence. This sequence would not appear to be essential to function, since the rabbits used in both studies were considered normal. The region is predicted to form a turn (Fig. 3), so its loss or inclusion would not be likely to disrupt a helix or strand domain. The Asp for Glu replacement that we found at position 2015 is conservative and could represent an allelic variant.
A major goal in obtaining the primary sequence of a membrane protein is to deduce features relating to the structure and function of the protein. Hydropathy plots (Fig. 4) illustrate that, with the exception of the sequences labeled M' and M" in Fig. 3 (residues 3123-3143 and 3187-3205), the first 4000 amino acids are hydrophilic and are likely to constitute the cytoplasmic domain of the ryanodine receptor. The clearest boundaries in this portion of the molecule are provided by the four 114 or 120 residue repeats (Fig. 6), occurring in two tandem pairs, and three regions rich in runs of glutamic acid residues (1870-1930, 2025-2090, and 3675-3750). In between these fairly well-defined segments are regions, typical of globular proteins, in which predicted a-turn-a, p-turn-p, or p-turn-a-turn-p supersecondary motifs predominate. Some of these are indicated in Fig. 3.
We have identified two potential transmembrane sequences near the middle of the molecule and 10 in the COOH-terminal fifth of the molecule which are candidate sequences for the transmembrane channel of the baseplate. The size of the baseplate, made up from transmembrane sequences of tour subunits (Wagenknecht et al. (1989), is about 140 X 140 A. A structure of this size Fould accommodate up to 150 transmembrane helices of 11 A diameter (Engelman et al., 1980), provided it were free of lipid or of polar segments of the protein embedded in the hydrophobic helices, while each monomeric subunit could accommodate about 36 transmembrane sequences. If sequences M', M", and Ml-Ml0 were all transmembrane, then six transmembrane hairpin loops, one near the center of the molecule and five in the COOH-terminal fifth of the molecule would anchor each monomer to the membrane and the total number of transmembrane sequences in the tetramer would be 48.
Of the 12 sequences proposed to be transmembrane, four, labeled Mb, Me, Ma, and Ml0 in Fig. 3, have mean hydropathy indices over 17 residues ranging from 2.0 to 2.9. The remainder have mean hydropathy indices ranging from 0.8 to 1.6 and are less clear candidates for transmembrane sequences. In our earlier analysis of transmembrane segments of the Ca2+ ATPase (MacLennan et al., 1985), we predicted that 10 transmembrane sequences with mean hydropathy indices ranging from 1.3 to 2.7 would exist in this protein, and these included sequences which were relatively rich in polar and charged amino acids. We have obtained evidence recently (Clarke et al., 1989) that it is, indeed, these charged polar residues that are involved in forming the Ca'+-binding sites and the Ca2+ channel in the transmembrane domain of the Ca*+ ATPase. By analogy, we believe that it is unlikely that the Ca2+ release channel of the ryanodine receptor would be made up only of very hydrophobic sequences. Lodish (1988) has presented a similar view of the structure of membrane transport proteins.
The COOH-terminal fifth of the ryanodine receptor molecule contains highly charged sequences in addition to the major hydrophobic stretches. In the folding model that would result from the assignment of sequences M', M", and Ml to Ml0 (Fig. 3) to the transmembrane sector, rather highly charged sequences would lie in the lumen of the terminal cisternae. The sequences RRRVRRLRR (residues 4307-4314) and EEAEGDEDE (residues 4612-4620) represent especially concentrated regions of positive and negative charges. A high density of charge surrounding the luminal mouth of the Ca*+ release channel might influence the gating properties of the channel or act as an ion selective screen at the channel entrance. It might also influence the interaction of luminal proteins such as calsequestrin (MacLennan and Wong, 1970) or calsequestrin-binding proteins (Mitchell et al., 1988) with the ryanodine receptor.
Sequences labeled Ma, Mq and Ms are rich in glycine and alanine residues and Ms has a low hydropathy index, largely due to a cluster of glycine residues in the sequence. Glycine and alanine residues are compatible with transmembrane sequences, however, and one of the transmembrane sequences in subunit C of the bacterial FIFo ATPase complex (Walker et al., 1984;Senior, 1988) and transmembrane sequence D in the P-glycoprotein (Gras et al., 1987) are glycine and alanine rich.
In a search for homology of the ryanodine receptor sequence with that of other known proteins, we noted amino acid identities between our proposed transmembrane segments Ms and Ma and segments Mz and Ms of the nicotinic acetylcholine receptor which, according to the model of Noda et al. (1983) has only four hydrophobic transmembrane segments. These identities are confined to two hydrophobic regions in the two molecules. While the regions of identity are contiguous in the acetylcholine receptor, they are separated by 160 residues encompassing putative transmembrane sequence M7 in the ryanodine receptor sequence. Another match in the region containing putative transmembrane sequences Ms-Mio can be made to the S4, S5 region of the Na+ channel (Noda et al., 1984) and to the corresponding segment of the dihydropyridine-sensitive Ca*+ channel (Tanabe et al., 1987). One of these matches is shown in Fig. 3. Since the matches are mainly hydrophobic in a region rich in such residues, they are unlikely to be significant. The lack of significant homology of the ryanodine receptor with other channel proteins suggests that the ryanodine receptor may be the first member of a novel family of channel proteins that might contain other intracellular Ca2+ release channels such as the cardiac form of the ryanodine receptor (Lai et al., 1988b;Inui et al., 1987b) or the inositol trisphosphate receptor (Supattapone et al., 1988), which have been identified, but not yet cloned.
Our attempts to identify nucleotide-binding sites in the primary sequence were not successful. Known nucleotidebinding sites are almost invariably in parallel /3 sheet domains and perhaps half of them include a glycine-rich bend in the first (~$3 unit of the sheet. The glycine-rich sequence GLGDMG, (residues 4447-4452) noted earlier (Takeshima et al., 1989), is present in a glycine-proline-rich region which would not be predicted to form a typical nucleotide-binding domain. Our present information on the structural features of the nucleotide-binding site derives largely from proteins such as muscle adenylate kinase (Fry et al., 1986) which has a high affinity for ATP of the order of 30 FM (Noda, 1973). Half-maximal activation of the Ca*+ release channel, either in planar bilayers (Smith et al., 1986;Smith et al., 1988) or in intact heavy sarcoplasmic reticulum vesicles (Morii and Ton omura, 1983;Meissner et al., 1986) occurs at about 2 mM ATP, indicating that the activation of Ca*+ release by ATP is due to the interaction of the ligand with low affinity binding sites which may not be closely related to those with high affinity.
The search for high affinity Ca*+-binding sites of the EF hand type was also unsuccessful. Since high affinity Ca2+ binding occurs in many proteins of known structure which do not contain EF hand structures, this is not surprising. For example, in a recent study, Fliegel et al. (1989) have defined the primary sequence of the high affinity Ca*+-binding protein (calreticulin) of the sarcoplasmic reticulum (Ostwald and MacLennan, 1974). Although this protein of 400 amino acids binds 1 mol of Ca*+/mol with high affinity and 25 mol of Ca'+/mol with low affinity, no clear EF hand sequences were present in the molecule. In analogy with the long acidic sequence in the ryanodine receptor (residues 1872(residues -1923, 32 out of 40 amino acids near the COOH terminus of calreticulin were acidic. Thus, in both calreticulin and the ryanodine receptor, high affinity Ca2+ binding occurs in the absence of a clear EF hand sequence, but in the presence of a long acidic sequence. The determination of the Ca*+ binding properties of both of these long acidic sequences will be of great interest. Our search for calmodulin-binding sites revealed a number of candidate sequences lying near the center of the molecule. If these sites are functional, they would indicate that modulation of the channel could occur in domains that are distant in the primary sequence from the sequences most likely to make up the release channel itself. Potential phosphorylation sites were found in the sequence just upstream from the probable channel forming sequences in regions that we predict to be cytoplasmic. Although phosphorylation of the channel has been reported (Seiler et al., 1984), regulation of the channel by phosphorylation has not been reported. Several consensus glycosylation sites were found in the sequence but all were in regions that we predict to be cytoplasmic. Since glycosylation reactions occurred in the lumen of the endoplasmic reticulum (Lennarz, 1987), which has the same orientation as the sarcoplasmic reticulum, none of these would be predicted to be glycosylated. An alternative folding pattern (Takeshima et al., 1989) could lead to the luminal location and glycosylation of residue 4859, however. To date, no detailed study of glycosylation of the ryanodine receptor has been published.
The primary sequence of the ryanodine receptor provides us with the first important clues to the understanding of the structure/function relationships in the molecule. Further information is likely to arise from detailed investigation of the biochemical properties of altered forms of the molecule such as those which can be obtained through site-specific or naturally occurring mutations.