Molecular Cloning of cDNAs from Human Kidney Coding for Two ...

THEJOURNAL

OF

BIOLOGICAL CHEMISTRY

Vol. 263, No. 29, Issue of October 15, pp. 15024-15031,1988 Printed in U.S.A.

0 1988 by The American Society for Biochemistry and Molecular Biology, Inc.

Molecular Cloning of cDNAs from Human Kidney Coding for Two Alternatively Spliced Productsof the Cardiac Ca2+-ATPaseGene* (Received for publication, April 20, 1988)

Jonathan LyttonS and David H. MacLennanQ From the Banting and Best Deuartment of Medical Research, C. H. Best Institute, Universityof Toronto, Toronto, Ontario”5G lL6, Canada

Ca2+-ATPasemolecules present in the microsomal fraction from non-muscle cells were examined immunologically. Rabbit whole brain, cerebellum, liver, kidney, and COS-1 cell microsomes all displayed a polypeptide of about 110 kDa which was immunoreactive with a polyclonal antiserum against the cardiac muscle sarcoplasmic reticulum Ca2+-ATPase molecule, but was not immunoreactive with a monoclonal antibody specific for the fast-twitch muscle Ca2+-ATPase. cDNAs encoding the full length of two Ca2+-ATPase molecules were isolated from a human kidney library using a mixture of nucleotide probes derived from both rabbit fast-twitch and cardiac muscle Ca2+-ATPase cDNAs. The human kidney cDNAs, H K l and H K 2 , are the products of alternative splicing. H K 2 codes for a protein identical to rabbit cardiac muscle Ca2+ATPase, with the exception of 6 scattered amino acid replacements, whereas H K l codes for a protein identical to that encoded by H K 2 , but with the carboxylterminal 4 amino acids replaced by an extended sequence of 49 amino acids. cDNAs of the H K l type are by far the most abundantin the library. The partial structure of a 40-kilobase genomic DNA encoding all but the 5’ end of the human cardiac Ca2+ATPase is described. The exons which give riseto the alternatively spliced products were located by Southern blotting and sequencing, and the alternative splicing patterns were determined.

Changes in the cytosolic concentration of free calcium ion play a key role in controlling a host of different physiological processes in every animal cell. The resting intracellular free calcium concentration in these cells is maintainedin therange of 100-200 nM, but upon stimulation, it can rise as much as 2 orders of magnitude. The concentration of calcium in the extracellular fluid is 4 orders of magnitude higher than the resting intracellular levels. Asa resultof the need to maintain calcium homeostasis in the face of this very large concentration gradient across the plasma membrane, and the central role of calcium in cellular metabolism, it is not surprisingthat

* This research was supported by grants from the Medical Research Council of Canada, The Muscular Dystrophy Association of Canada, and the Heart andStroke Foundation of Ontario (to D. H. M.) The costs of publication of this article were defrayed in part by the payment of page charges. This articlemust therefore be hereby marked “aduertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. The nucleotide sequence(s) reported in thispaperhas been submitted to the GenBankTM/EMBLData Bankwith accession number(s) 504025. $ Recipient of an Alberta Heritage Foundation for Medical Research Postdoctoral Fellowship. To whom correspondence should be addressed.

cells have evolved many ways to control the concentration of intracellular calcium. Channels,both voltage- and ligandgated, allow calcium entry into the cell down its electrochemical gradient, while ATP-coupled pumps, both in the plasma membrane and in cellular organelles, as well as plasma membrane Na+/Ca2+exchange protein function to remove calcium from the cytoplasm (Carafoli, 1987; Reinhart et al., 1984). In recent years interest has focused on intracellular pools of calcium, thought to be associated with the endoplasmic reticulum, which serve as the source for inositol 1,4,5-trisphosphate-induced alterations in cytoplasmic free calcium. These pools are maintained by an ATP-dependent calcium transport system (Streb et al., 1983; Prentki et al., 1985). A microsomal fraction isolated from a variety of cell types and enriched in endoplasmic reticulum components contains both an ATP-dependent calcium transport system and abiochemically correlated calcium-dependent ATPase activity (Prentki et al., 1984; Streb et al., 1984; Spamer et al., 1987). These fractions also contain a polypeptide of about 110 kDa which has been identified as a Ca2+-ATPase the on basis of calciumdependent phosphoenzyme intermediate formation (Heilmann et al., 1984; Spamer et al., 1987) and immunological cross-reactivity (Dean, 1984; Damiani et al., 1988). Although some of the kinetic, enzymatic, and structural featuresof this microsomal Ca2+-ATPaseare remarkably similar to those of the enzyme from skeletal muscle sarcoplasmic reticulum (Heilmann et al., 1984; Heilmann et al., 1985; Spamer et al., 1987),there are also some clear differences, especially for the platelet enzyme (Fischer et al., 1985). Indeed, the existence of a calcium pump, an inositol 1,4,5trisphosphate-gated release channel, and a protein immunologically related to the major calcium-binding protein from skeletal muscle sarcoplasmic reticulum, calsequestrin, all within the same membrane fraction, has fueled the speculation that non-muscle cells may contain a specialized compartment analogous tothe sarcoplasmic reticulum, which would function to take up, store, and release calcium (Damiani et al., 1988; Volpe et al., 1988). As one step toward a detailed molecular characterization of such a compartment, we describe here immunological data in support of earlier claims for the presence in a variety of nonmuscle tissues of a microsomal Ca*’-ATPase similar to that from the sarcoplasmic reticulum. The complete sequence of two cDNA clones from human kidney encoding two alternatively spliced versions of the cardiac Ca2’-ATPase gene confirms the identity of the microsomal calcium pump. MATERIALS AND METHODS

Immunological Procedures-The amino-terminal fifth (amino acid residues 17-215) of the rabbit cardiac muscle Ca*+-ATPase (MacLennan et al., 1985; Brand1 et al., 1987) was subcloned into the trpE fusion vector PATH1 (this derivative of the vectors described by

15024

15025

Non-muscle Ca2+-ATPaseIsoform Dieckmann and Tzagoloff (1985) and Kleid et al. (1981) was kindly provided by T. J. Koerner and A. Tzagoloff, Columbia University, New York). The construct was transformedinto Escherichia coli JMlOl cells. Following lysozyme/Triton X-100 lysis of the induced cells (Dieckmann and Tzagoloff, 1985), the insoluble fusion product was separated by SDS' gel electrophoresis (Laemmli, 1970). The appropriate band, running with the expected mobility, was excised, electroeluted according to theprocedure of Hunkapillar et al. (19831, dialyzed extensively to remove SDS, and lyophilized. Rabbits were immunized through to the popliteal lymph node route as described by Sigel et al. (1983) with 50 r g of antigen in 100 pl of phosphatebuffered saline: complete Freund's adjuvant (1:l)in each node. Two boosts were made at 3-week intervals by injecting 100 pg of antigen in a total of 3 ml of phosphate-buffered saline: incomplete Freund's adjuvant (1:l) intramuscularly and at several subcutaneous sites on the back. The animals were bled 3 weeks later, and this serum was used without further purification for the studies described. Immunoblots were performed on proteins separated by SDS slab gel electrophoresis (Laemmli, 1970) and subsequently electrophoretically transferred to nitrocellulose by the method of Towbin et al. (1979). The blots were blocked in 5% (w/v) skim milkpowder in Tris-buffered saline (TBS), and subsequently incubated in a 1:500 dilution of the serum in TBS/l% skim milk powder for 1-2 h at room temperature. After washing three times for 5 min each in 0.1% (w/v) Tween 20 in TBS, an alkaline phosphatase conjugated goat antirabbit secondary antibody (Promega Biotec) was used at a dilution of1:5000-1:7500 in TBS/1% milk powder. The blots were washed again as above and developed using nitro blue tetrazolium/5-bromo4-chloro-3-indolyl phosphate. Membrane Preparations-KC1-washed rabbit sarcoplasmic reticulum was prepared as described by Campbell and MacLennan (1981). Microsomes from rabbit cardiac tissue were prepared according to the procedure of Chamberlain et al. (1983). Crude microsomes from whole rabbit brain or from cerebellum were a kind gift from Susan Treves; the microsomes from brain as well as crude microsomes from rabbit kidney cortex and rabbit liver were prepared by centrifugation (100,000 X g, 60 min) of the post-mitochondrial supernatant (8,000 X g, 20 min) of tissue homogenizedin 0.29 M sucrose, 10 mM imidazole (or 10 mM Tris-Cl), 5 mM P-mercaptoethanol, 0.1 mM phenylmethylsulfonyl fluoride, pH 7.4. Protein concentration was determined by the method of Lowry et al. (1951) with the inclusion of 0.1% SDS (final concentration) in the assay, and using bovine serum albumin as a standard. Molecular Cloning-A cDNA library constructed in XgtlO from human kidney cortex mRNA was a generous gift of Dr. Graeme Bell, University of Chicago (Bell et al., 1986). Probes were obtained by suitable restriction endonuclease digests of cDNA encoding rabbit fast-twitch muscle and rabbit cardiac muscle CaZ+-ATPasemolecules (MacLennan et al., 1985; Brandl et al., 1986). These probes were chosen to maximize the chances of selecting any and all possible Ca2+-ATPasecDNAs present in the library. Thus, segments corresponding to the regions of the molecule having the highest degree of identity with other members of the E1-E' class of ion-pumping ATPases were used.' These were: for the fast-twitch isoform, nucleotides 864-1107, 1490-1760, and 2076-2374; for the cardiac isoform, nucleotides 764-1094,1445-2028, and 2029-2371.Two lots of approximately 400,000 plaques each were screened using 1 pCi/ml of a mixture of these six probes labeled with [ ~ r - ~ ' ] d c T(ICN P or Amersham Corp.) by the method of Feinberg and Vogelstein (1983) using the oligolabeling kit obtained from Pharmacia LKB Biotechnology Inc. Hybridization was carried out overnight a t 42 "C in 25% deionized formamide, 5 X SSCP, 5 X Denhardt's solution, 0.5% SDS, 50 pg/ml sonicated salmon sperm DNA (Sigma) and 10 pg/ml poly(A) (Boehringer Mannheim Canada). Washing was performed a t 50 "C in 2 X SSCP. Following initial isolation of the clones, full-length clones were isolated by screening approximately lo6 library plaques with a probe from the 5' end of one of the initial isolates (corresponding to nucleotides 669-939 in Fig. 3), using conditions of stringent The abbreviations used are: SDS, sodium dodecyl sulfate; Denhardt's solution (1 X), 0.02% (w/v) Ficoll 400 (Pharmacia), 0.02% (w/v) polyvinylpyrrolidone, 0.02% (w/v) bovine serum albumin, fraction V; kb, kilobase pairs; SSCP (1 X), 120 mM NaCl, 15 mM sodium citrate, 13 mM KH'PO,, 1 mM EDTA, pH 7.2; TBS, Tris-buffered saline, 140 mM NaC1, 10 mM Tris-C1, pH 7.5. N. M. Green, W. R. Taylor, and D. H. MacLennan, manuscript in preparation.

hybridization (42 "C, 50% formamide) and washing (65 "C, 0.1 X SSCP). A human genomic DNAlibrary, constructed in the Lorist B vector by Dr. Henry Klamut, Hospital for Sick Children, Toronto, was used for the isolation of the human cardiac Ca'+-ATPase gene. This library was screened using a mixture of two human cardiac cDNA sequences (MacLennan et al., 1987) corresponding to nucleotides 1639-1949 and 3370-3762 in Fig. 3. The longest clone isolated was analyzed further by digestion with all possible combinations of the restriction endonucleases EcoRI, Hind111 and BamHI. A Southern blot of these restriction digests was probed with different labeled fragments from the cDNA clones to generate the gene map. A 6-kb EcoRI fragment which hybridized with the 3' end of the cDNA clones was subcloned into the plasmid vector pTZ 19R (Pharmacia), and analyzed by further restriction endonuclease digests followed by Southern blotting. Nucleic Acid Sequencing-A series of suitable overlapping fragments was subcloned into the pTZ 18R or 19R (Pharmacia) vectors either by standard restriction methods or by the generation of nested deletions (Yanisch-Perron et al., 1985) using the exonuclease III/ mungbean nuclease kit obtained from Stratagene Cloning Systems (La Jolla, CA) according to the instructions of the manufacturer. Sequencing was performed using the dideoxynucleotide chain termination reactions of Sanger et al. (1977) either with the Klenow or with fragment of DNA polymerase I (Pharmacia) and [cY-~'P]~CTP Sequenase (IBI Inc.) and [35S]dCTPaS (Amersham). cDNA clones were sequenced in both directions with the exception of the 3'-most 20 nucleotides of HKl, which were sequenced in only one direction. Genomic sequencing was carried out in only one direction with the exception of exons a and b and theintervening intron, and the3' end of exon b', which were sequenced in both directions. Miscellaneous-All recombinant techniques were carried out essentially according to standard protocols (Maniatis et al., 1982), using the highest quality reagents available. The sheep kidney (Na+,K+)ATPase a subunit clone was a generous gift of Dr. Gary Shull, University of Cincinnati. RESULTS

Protein sequencing (Allen et al., 1980; Briggs et al., 1986) as well as cDNA cloning and sequencing (MacLennan et al., 1985; Brandl et al., 1986; Brandl et al., 1987) have demonstrated theexistence in different muscle types of two isozymes of the sarcoplasmic reticulum Ca2+-ATPase:one expressed in fast-twitch skeletal muscle and the other in both cardiac muscle and slow-twitch skeletal muscle. Since the non-muscle microsomal Ca2'-ATPase has properties similar to those of the muscle enzyme (see Introduction), it was of interest to know whether this molecule most closely resembled the fasttwitch or cardiac Ca2+-ATPase.A suitable monoclonal antibody (A52) against the fast-twitch isozyme had already been prepared in this laboratory (Zubrzycka-Gaarn et al., 1984). However, we did not have an antibody which recognized the cardiac isoform with high titre and good specificity. To generate such an antibody, we chose to employ recombinant methods using the PATH vector system constructed by T. J. Koerner and A. Tzagoloff (Hoffman et al., 1987; Dieckmann and Tzagoloff, 1985). The antigen was thus made as aproduct of the fusion between the trpE gene of E. coli and theaminoterminal fifth of the rabbit cardiac Ca2+-ATPasecDNA. A high-titre rabbit polyclonal antibody, C4, was raised. Fig. 1shows that C4 cross-reacted with a single polypeptide of about 110 kDa in both fast-twitch andcardiac sarcoplasmic reticulum preparations. A52, on the other hand, recognized a band of this size only in the fast-twitch sarcoplasmic reticulum lanes. If the C4 serum were first absorbed with fasttwitch sarcoplasmic reticulum, and adherent antibodies removed by sedimentation, the resulting cleared serum recognized only the cardiac sarcoplasmic reticulum polypeptide (data not shown). Thus, C4 appears to recognize epitopes common to both fast-twitch andcardiac Ca2'-ATPase as well as ones specific for the cardiac isozyme. Notice that the

Non-muscle Ca*+-ATPaseIsoform

15026

M,, kDa

50 35 27 -

120 75

Ab C4

A52 MAb

FIG. 1. Immunoblot analysis of Ca*+-ATPaseexpression in various tissues. Samples of the various microsomal preparations (0.1 pg of fast-twitch sarcoplasmic reticulum ( S R ) , 1 pg of cardiac microsomes, 20 pg of all other samples) were separated on a 10% Laemmli gel and transferred to nitrocellulose. The samples on the left were probed with polyclonal antibody C4 a t a dilution of 1:500, those on the right with monoclonal A52 a t a dilution of 1:lOO as described under "Materials and Methods." The numbers indicate the approximate size of the prestained standards (Bio-Rad) in kilodaltons. Note that calibration of the prestained standards with nonstained SDS standards (Bio-Rad) gave different values for the relative molecular mass than those provided by the manufacturer. Our estimates areshown.

immunoreactive band from cardiac sarcoplasmic reticulum runs with a slightly lower apparent molecular mass, and is slightly broader than the bandfrom fast-twitch sarcoplasmic reticulum. When microsomes from a variety of non-muscle tissues were tested with these two antibodies, an immunoreactive band co-migrating with cardiac sarcoplasmic reticulum Ca2'-ATPase, and having the same properties, was observed in all the samples tested (including an extract from whole platelets, data not shown) with C4, but not with A52 (with the possible exception of whole brain). This result suggests that themicrosomal Ca2+-ATPaseexpressed in nonmuscle cells most closely resembles the cardiac isoform of the muscle enzyme. To confirm the exact molecular identity of this non-muscle calcium pump we decided to clone the cDNA coding for the protein from a human kidney library. This source was chosen largely because we wished to identify the natureof the endogenous microsomal calcium pump in primate kidney, since ongoing transfection studies in thelaboratory utilized the monkey kidney-derived cell line, COS-1 (Maruyama and MacLennan, 1988). Our immunological data, plus the experiments of others (Moore et al., 1974), indicated that kidney should express the same, or a very similar, molecule as thatdescribed for other tissues. The overwhelming body of evidence suggested that this molecule must be bothfunctionally and structurallysimilar to the enzyme from the sarcoplasmic reticulum. Thus, it seemed reasonable to use cross-hybridization with sarcoplasmic reticulum Ca2'-ATPase cDNA probes as an approach to the isolation of the human kidney endoplasmic reticulum Ca2'-ATPase clones. A human kidney cortex cDNA library in X g t l O , obtained from Dr. G. Bell (Bell et al., 1986), was screened witha mixture of probes from both rabbit fast-twitch and cardiac Ca2+-ATPasecDNA molecules. Conditions of low stringency (hybridization at 42 "C in 25% deionized formamide, washing at 50 "C in 2 X SSCP) were employed to ensure detection of weakly cross-hybridizing molecules. These conditions resulted in the detection of about 80 positive signals, with a variety of intensities, from 400,000 plaques. Six strong and nine weak positive signals were picked and plaque-purified. In general,

the weaker signals turned out to be false-positive signals, which in several cases gave positive signals with the phage plaques, but negative signals with DNA isolated from those phage. In one case, the weak positive signal was the result of cross-hybridization to a (Na',K')-ATPase cDNA (confirmed by hybridization under stringent conditions to coding region probes from the sheep kidney (Na',K')-ATPase a subunit). The six strongly hybridizing cDNAs fell into threecategories according to theirrestriction endonuclease maps (HKl, HK2, HK3; see Fig. 2, HK3 is not shown). Four of these cDNAs belonged to group H K l , one to group HK2, and one to group HK3. Further screening of the library at reduced stringency produced 11more strongly positive clones, all identical to the first isolate of group H K l . The library was then rescreened under stringent conditions to obtain full-length clones, using an EcoRI-PstI fragment from the 5' end of cDNA XHK2a (see Fig. 2; this corresponded to nucleotides 669-939 in Fig. 3). Six more positive signals were obtained four in group HK1, and two in group HKZ. A t least one cDNA from each of these groups was full-length (see Fig. 2). Partial amino acid sequence deduced from analysis of the cDNA belonging to the HK3 group revealed a novel protein with 70% identity to either the rabbit fast-twitch or cardiac Ca2+-ATPases. This protein is not considered here, but will be the subject of a forthcoming paper." The complete nucleotide sequence of cDNAs HKI and HK2 was determined in both directions(see Fig. 2 for the strategy) and is shown in Fig. 3. Both sequences end in poly(A) tracts greater than 50 bases in length. These two sequences are identical up tonucleotide 2980, after which they diverge. This suggests that the two cDNAs are the result of alternative splicing of the same primary transcript. HK2 encodes a protein of 997amino acids which is identical to the human muscle cell line partial cDNA cloned and sequenced by MacLennan et al. (1987), and virtually identical to therabbit cardiac Ca2+ATPase cDNA (MacLennan et al., 1985; Brandl et al., 1987). There are six scattered changes: (human amino acid, residue number, rabbit amino acid) Arg7*Lys,G l ~ ~ ~ ~ Asn403Lys, Asp, Sers38Ala,Aspa4G1u, SerMgAla, was originally reported as Lys, but see MacLennan et at., 1987. H K l codes for a protein identical to that of HK2 with the exception of the replacement (as aresult of the alternative splice) of the carboxyl-terminal 4 amino acids with an extended sequence of 49 amino acids. The fast-twitch muscle Ca2'-ATPase has also been demonstrated to undergo alternative splicing, although in that case it results in a change from a carboxyl-terminal glycine residue to the short, highly charged sequence, Asp-Pro-GluAsp-Glu-Arg-Arg-Lys(Brandl et al., 1986; Brandl et al., 1987; Korczak et al., 1988). Within the extended carboxyl-terminal sequence of the kidney isoform is a stretch of20 uncharged and relatively hydrophobic amino acids. Fig. 4 shows ahydropathyplot (Kyte and Doolittle, 1982) for the carboxyl-terminal quarter of H K l . Withinthis sequence are 6 of the 10 proposed membrane-spanning domains for the muscle Ca2'-ATPase isoform (MacLennan et al., 1985; Brandl et al., 1986). The length and magnitude of the hydropathy index for the nonpolar region of the extended kidney sequence suggest that it may be an 11th membrane-spanning segment. Fig. 4 also shows the positions of three potential glycosylation signals within the carboxyl-terminal quarter of the Ca2'-ATPase molecule, one of which is found in the extended sequence. It was of interest to determine the structureof the human cardiac Ca2'-ATPase gene in order to confirm, and tounderJ. Lytton and D. H. MacLennan, unpublished data.

15027

Non-muscle Ca2+-ATPaseIsoform

FIG. 2. Partial restriction endonuclease maps and sequencing strategies for the human kidney Ca2+ATPase cDNA clones. The length of the cDNAs H K l and HK2 is shown by the solid lines.Open boxes denote the coding regions. The lengths of the differXHKlb, xHK2a, ent clones, XHKla, XHK2b are shown as solid lines beneath the map of their respective cDNAs. Arrows indicate the direction and extent of sequencing.

0

1

2

3

4

5 kb

I

I

I

I

I

I

HK1 WKla

HK2

.

i

.

...

.. . .

.

. .

.

J poly(A) Tail

“

-

-

“-

”

-

-

” ”

a ” ”

stand better, the alternative splicing event. A genomic clone containing more than 40 kb was isolated from a cosmid library in the Lorist B vector, using a fragment from the human cardiac cDNA (MacLennan et al., 1987) asa probe. The approximate distribution of exons within the gene was determined by Southern blotting using probes from the HK1 and HK2 cDNAs (see Fig. 5). This distributionappears to be similar to the rabbit fast-twitch Ca2’-ATPase gene (Korczak et al., 1988) in that exons encoding the 3‘ half of the cDNA are clustered in a relatively small segment of the gene, whereas the otherexons are more spread out inthe 5’ direction. Since it did not hybridize with probes from the 5‘ end of the cDNA, this genomic DNA must lack the 5’ end of the gene, despite being more than 40 kb in length. In contrast, the rabbit fasttwitch Ca2+-ATPasegene is only 16.5 kb long. A 6-kbEcoRI fragment of the human cardiac Ca2’-ATPase gene which hybridized with probes from both of the alternatively spliced 3’ termini was subcloned for further analysis, as shown in Fig. 5. The positions of exons within this region were determined by restriction endonuclease digestion and Southern blotting. They were then confirmed by sequencing through the exons and across all of the exon/intron boundaries in at least one direction (Fig. 6). The arrangement of these exons allows us to see how the alternatively spliced messages are made. HKl is formed when exons a and bb’ are spliced together and thesplice site within exon bb’ remains cryptic. Polyadenylation occurs after nucleotide 3966, although the precise site is unclear because A nucleotides are found at positions 3967,3968, and 3969 in the gene. Definition of the site is further complicated by the fact that an isolated G nucleotide is found in our cDNA clone HKI at position 3972, preceded by five A nucleotides and succeeded by about 50 more. Since this G nucleotide was not derived from the gene sequence (Fig. 6), we assume that it has occurred in thecDNA as a result of a reverse transcriptase misincorporation error. A perfect polyadenylation consensus sequence (Proudfoot and Brownlee, 1976) starts 13 base pairs upstream from the site of poly(A) addition in exon bb’. HK2 is formed when the splice site within exon bb’ is used, resulting in the excision of segment b’ together with the contiguous sequence as far asexon c, and thus bringing exon c adjacent to b. Again, a perfect polyadenylation signal begins 19 base pairs upstream from the site of poly(A) addition (nucleotide 3778) in exon c. This mechanism of alternative splicing has been referred to as “internal donor site splicing,” and differs from the “cassette” type (Breitbart et al., 1987) used to generate thetwo fast-twitch muscle Ca2’-ATPase cDNAs (Brand1 et al., 1987; Korczak et al., 1988). In that case, the penultimate

WK2a WK2b

“ “

exon either remains in the message together with the antepenultimate and ultimate exons (adult) or is excised in entirety, bringing the antepenultimate and ultimate exons into juxtaposition (neonatal). In addition to the exon boundary defined by the alternatively spliced cDNAs, the genomic sequence defined two other boundaries which delineate exon a (see Figs. 3 and 6). Interestingly, these are in the identical positions as boundaries 19 and 20 for the rabbit fast-twitch Ca2’-ATPase gene (Korczak et al., 1988). On the other hand, the 3’-most splice site in the human cardiac Ca2’-ATPase gene is shifted 3 nucleotides to the 3‘ side compared to the same exon boundary in the fasttwitch gene. It is possible that thisshift allows for the different mechanisms of alternate splicing which exist for the fasttwitch and the cardiac Ca2’-ATPase genes. DISCUSSION

We have demonstrated that the microsomal fraction from many different non-muscle tissues contains amolecule of 110 kDa which is immunoreactive with an antiserum against the cardiac sarcoplasmic reticulum Ca2+-ATPase, whereas a monoclonal antibody specific for the fast-twitch Ca2+-ATPase isozyme shows reactivity only with an ATPase from fasttwitch muscle. This suggests that thepredominant organellar Ca2’-ATPasemolecule expressed in non-muscle tissues is similar to the cardiac sarcoplasmic reticulum isoform. This conclusion is supported by the tryptic digestion patterns observed for cardiac muscle sarcoplasmic reticulum (Kirchberger et al., 1986), for liver microsomes (Heilmann et al., 1984; Spamer et al., 1987), and for platelet membranes (Enyedi et al., 1986) using calcium-dependent covalent phosphoenzyme intermediate formation to identify the Ca2’-ATPase molecule. These patterns are virtually identical with those observed for the fast-twitch Ca2’-ATPase. The situation is more complicated in platelets, however, where Fischer et al. (1985) found entirely different digestion patterns from fast-twitch Ca2’-ATPase using a polyclonal antibody against the fast-twitch enzyme to recognize the platelet Ca2+-ATPase molecule. The explanation for these discrepancies could lie in the expression of another undetected Ca2+-ATPaseisoform in platelets, such as our clone HK3.3 It is worth noting, however, that we find the highest concentration of immunoreactive Ca2’-ATPase in cerebellum, where Worley et al. (1987) find the highest concentration of inositol 1,4,5-trisphosphate-binding sites. This reinforces the proposal by Damiani et al. (1988) and Volpe et al. (1988) that the functions of calcium uptake, sequestration, and release

Non-muscle Ca2+-ATPaseIsoform -121

-1 120 1

240 41 360

81 480 121

600 161 120 201 840 24 1 960 281

1080 32 1

1200 361 1320 401 1440 441 1560 481

1680 521

1800 561 1920

601 2040 64 1 2160 681 2280 721 2400

161 2520 801 2640 841 2760 881 2880

92 1

961 3120

1001 3240 1041

3360 3180 3600 3720 3840 3960 3911 3000 3120 3240 3360

3480

3600 3720 3778

sequence above the skid line is that of H K l . The slash in the sequenceat nucleotide 2980 is the site at which the sequence of HK2 diverges (shown below the solid line). The inverted triangle at nucleotide -3 indicates the 5' extent of HK2. Arrows a t nucleotides 2741 and 2859 indicate the exon boundaries which were mapped from the gene. Clone XHKla extends from nucleotide 711 to 3951; XHKlb from -163 to 3971; XHK2a from 669 to 3778 XHK2b from -3 to 3647.

15029

Non-muscle Ca2'-ATPase Isoform

4 qaattcacagtttqtcctqcattaqqacattctcttcaacttt~ccactqtaqaaaqtqqaq qtaqgtcaqcqqatqqtqccacattaacaqccqccttactqaagtgtagtccaacagqqtct tactqccactgtqacacgtqccttgccttqqqgqtqcqtttcccacctctccttqctctqc 2142 Exon a

3

aqCTTGTCCG~CCAGTCCTTGCTGAGGATGCCCCCCTGGGAGMCATCTGGCTCGTG

SerLeuSerGluAsnGlnSerLeuLeuCIrgMetProProTrpGluAsnIleTrpLeuVa1 914 2859 GGCTCUTCTGCCTGTCCATGTCACTCCACTTCCTGATCCTCTATGTC~CCTTGCCA

GlySerIleCysLeuSerMetSerLe~i~PheLeulleLeuTyrValGl~roLeuPro 953

gtaactqgttqqgtqqqqcttqqqaccaqccacFtccttccaqqqqaqqctqqaqqcgtqa

2060 Exon b cacqtcttccctgtgtqtcaqCTCATCTTCCAGATCACACCGCTGMCGTGACCCAGTGG LeUIlePheGlnIleThrProLeuAsnValThrGlnTrp 954 CTGATGGTGCTG~TCTCCTTGCCCGTGATTCTCATGGATGAGACGCT~GTTTGTG

850

800

950

900

1000

Amino Acid Residue FIG.4. A hydropathy plot for residues 746-1042 of H K 1 . The hydropathic index (Kyte and Doolittle, 1982) averaged over a window of 19 amino acids is shown. The threepotential glycosylation signals in this region of the protein, along with their sequence in single letter code, are indicated by the arrows. They are located at positions 918, 962, and 1035. The triangle marks the position of the carboxyl-terminal end of the HK2 protein. The transmembrane segments proposed by MacLennan et al. (1985) and Brandl et al. (1986) correspond approximately to amino acids 760-780 (segment 5 ) , 790810 (segment 6), 830-855 (segment 7), 895-915 (segment 81,930-955 (segment 9), and965-985 (segment 10).

LeuHetValLeuLySIleSerteuProValIleLeeWeUSpC1UThr~uLy.PheVal 2901 Exon b' GCCCGCMCTACCTGGMCCTGIGTAAAGMTGTGTGCAGCCTGC~C~TCCTGCTCG

~~r~snTyrLeuGluProG/lyLysGluCysValGlnPro~~ThrLyaSerCyJSer 994

TTCTCGGCAT GC... ( 1 0 5 bp) ...TTAGGATTGTGATGGTTCGTTCTGTTTACAG PheSerAlaCys.. . TTTTMCGAGAGGTATGCCTGTACTCGCTTGTGCAGAAMCTTCMTCG

ACTGGGTTTATGTCCCTTCACATAGTTTTTMGGTTATTTATTT~TGTCTMTGTATT TATTGTMCAGACATTGTTTTGCCM~TTGCCTATTTCAGTGGCACGTCAT~AGTTTT

3966 AWAAAATAAAAcATTTTaaatqqacaqaqaaaaataactqtcttgtctttaactcgtaa gtqqcttacctqqqactaacacatgtccaactttctctccagttcttaqctcaqaacttt aqttqtactctqcttqaqqqqaagaaqqctcctqctctqctgtgtaqqtagt~taqqaa ttgtattcttaatqtacaqqcactaattqtcatctqtqatqtauttttatq~agtttc tqctqqcctqgtataqaqatcataaqqqcaaqaqagtatqtgtgtgtatqtqtqtqqttq taaatctgtaataqcacatqaccaaacctqaacatattgtataqaactattttattqaat gtqqcactaaccaccacuccqttactacqatcaatgtttqcqcat~tcqaqatqagtc tcaccaauqtqtqtaagtcattaacaqtcctaactgtqgtgtttcct~aatqccttcc aacatccatcaactaa qaqtattttcttcctqqqatttqqatqctttaqcctaaaqq

'tg'

tqactqccaccaaqtqaqataactgtatqtcactaacttataaqccqcctccatqqcaqa

tqctgctgtqctccctqatqccctgtgaqcaccqqqgttqcctqtqqcqcctqccatqtq actcqqqcqcagcatcaqctq (2.2 kb)...tctagatqctaccctqtqtqqqcqg cacctcaqqqacaqtaaatcaqaaatqctqqtcttqaaaccttqaaaaqatcaaqctqaa tqttccttttcatctgtcqctqttqatcttcatctatttaaataqqtattctaacqtttc 2901 Exon c ctctctqtatttcatqaaqctqatttcctctctctttccttttcaq/CMTACTGGAGTM A/laIle~UGlU***

...

_ _ ...,.../-= g2 $ 8g [I

1 1

kb:0

a b b'

I l D

HKI

"-_I"*I

I1 I

1

2i 2 8 $j& ,.,. 1111 4

a/''

1 1 Ill 6

cj

994 Exons

SpllW 1 rnANA

ab b B

b b'

t l a

H e

1

...

bp) CMTACAMTATTGT GATGCATTTATCTTMTAAAATGCT~TGTCMTTTATCACTGCGCATGTTTGACTTTA

T A G C M C ~ A T C T A C C M C C C T G T G C A T G ~. . ( 5 2 5

GACTGT~TAGAGATCAGTTGTTTCTTTCTGTGCTGGTMCMTGA~GTCGCACAGA

0

I/ /

" - 8 l a A l ab C

CCGCTTCCTMACCATTTTGCAGhMTGTMGGGTGTMffiGTGTTCGGTT~GTG~TGTGC~TTT

Exons mRNA

splice 2

FIG. 5. Partial human cardiac Ca2+-ATPasegene structure and alternative splicing. The top line shows a map of the cDNA, with the restriction sites delimiting the different probes indicated. The second line illustrates the partial organization of the gene. The axes in kb are notto scale. B, BarnHI; E, EcoRI; H, HindIII; P, PstI; Pu, PuuII, S, SphI, X,XhoI. The shaded bars between the first and second lines illustrate the fragments of the gene which hybridized to the indicated cDNA probes. Note that the 5' cDNA probe did not hybridize to the genomicDNA clone, indicating that the gene is longer than 45 kb. An enlargement of the 6-kb EcoRI piece which hybridized to probes from the alternatively spliced 3' ends of both HKl and HK2 is shown below the gene. The positions of the exons within this fragment are indicatedbelow it, together with the splicing pattern which generates the different messages. HKI, exons a,b,b'; HK2, exons a,b,c. Exons b and b' are contiguous in the gene sequence, exon b' becoming a part of the excised intronic sequence as a consequence of the HK2 splice.

3770 UTGGTTTCAGGT~TAMTCTATTCTATGATaaattctcaqtqtqqtqgtqactqttc tqtqqqaqqqqqactgatqqgcaccaaqqqcctccactccqcacccqqq

FIG.6. Partial nucleotide sequences of the human cardiac Ca2+-ATPasegene which undergo alternative splicing. Exon sequences are shown in uppercase letters, and intron sequences are shown in lowercase. The positions within the cDNA of the nucleotides a t the exon/intron boundaries are indicated above the nucleotide sequence, and thepositions within the coding sequence of the protein are shown below the amino acid sequence. The slashes preceding nucleotides 2981 in exons bb' and c demarcate the region of the DNA excised to generate the HK2 splice which results in exon e being joined to exon b. Nucleotide 3966 is polyadenylated in clone HKI (but see the textfor discussion) and nucleotide 3778 is polyadenylated in clone HK2. The arrow in the intron separating exons 6' and c is the point homologous to the site of divergence of clones RB2-5 and RB4-17 (see Fig. 5 of Gunteski-Hamblin et al. (1988)).

tical to the sarcoplasmic reticulum CaZ+-ATPase,HK3: are both expressed but at much lower levels. This corroborates our antibody data showing that a cardiac-like isoform of the are coupled within a unique physical structure. Ca2'-ATPase is expressed in kidney, and suggests that a We have addressed the question of the precise molecular similar Ca2'-ATPase isozyme may be expressed in other nonnature of the Ca2'-ATPase molecules in kidney through muscle tissues. Additional support for this idea is provided by analysis of a human kidney cDNA library. We find that the the study of Gunteski-Hamblin et al. (19881, who have cloned predominantly expressed molecule, HKI (based on the num- cDNAs virtually identical to those corresponding to HKl and ber of independently isolated clones), is identical to the car- HK2 from rat kidney, brain and stomach. diac sarcoplasmic reticulum Ca2'-ATPase, except for an alIt is clear that both human and rat cDNAs code for the tered and extended carboxyl terminus present as a result of same proteins. The deduced amino acid sequence of the rat an alternative splicing event. A molecule representing the brain clone RB2-5 (Gunteski-Hamblin et al., 1988) differs at other splice product, HK2 (the one expressed in cardiac tis- only seven positions from the protein deduced from human sue), as well as a previously undescribed molecule 70% iden- cDNA HKI. Three of these changes are amino acid replace-

15030

Non-muscle Ca2'-ATPase Isoform

ments within the sequence common to HKI and HK2: (human contain this sequence. Similarly, our clones do not contain amino acid, residue number, rat amino acid) H i ~ ~ ~ ~ Gthe l n sequence , suggested by Poruchynsky et al. (1985) to be Asp403Lys,Ad61Ser. Within the alternatively spliced car- responsible for maintaining the rotavirus VP7 protein in the boxyl-terminal extension, clone RB2-5 contains an additional endoplasmic reticulum. Nor do they contain the carboxylproline residue at position 1004 as well as three conservative terminal sequence Lys-Asp-Glu-Leu proposed by Munro and hydrophobic replacements: Valgg8Ala,PhelW7Leu,Ile'027Val. Pelham (1987) to result in the retention of several otherwise While all of our clones of the HKl type had 3' ends close unrelated soluble luminal proteins within the endoplasmic to thatof clone RK7-12, Gunteski-Hamblin et al. (1988) found reticulum. Thus, if a signal for targeting exists in the carboxylthat the majority of their clones were extended in the 3' terminal extension of the non-muscle Ca2'-ATPase isoform, direction. Examination of our genomic sequence (Fig. 6) re- it is one which has not yet been described. This wouldbe veals a homology to this 3' region, particularly around the consistent with the suggestion of Volpe et ai. (1988)that the site of polyadenylation of clone RK7-12 and about the site of calciosome is a compartment distinct from the endoplasmic divergence of clones RB2-5 and RB4-17. The latter is clearly reticulum. Were this the case, then the Ca2'-ATPase would contiguous with the gene at this point, and must be spliced not contain any signal which would cause it to stop within out as part of an intron to form RB2-5. We presume that the the endoplasmic reticulum, and prevent it from movingto the 3' exon of RB2-5 (if present in the human gene) would be found within the more than 2 kb of unsequenced genomic calciosome. There is evidence, although almost entirely from platelets, DNA separating exons b' and e. for the existence of systems which act toregulate the activity As in this study, Gunteski-Hamblinet al. (1988) found that cDNAs of the HKl type were the most abundant clone. of calcium pumping into the microsomal stores. Yoshida and However, the true relative abundance of each isoform in Nachmias (1987) have demonstrated that phorbol esters stimdifferent tissues will only come from analysis of the mRNA, ulate calcium sequestration in saponin-permeabilized plateand from analysis of the protein using isoform-specific anti- lets,and CAMP has also been demonstrated to stimulate calcium uptake into microsomes isolated from platelets (KBsera. The HK1 molecule expresses a carboxyl terminus with an ser-Glanzmann et al., 1977; Kaser-Glanzmann et al., 1979; additional 45 amino acids, compared with the isozyme ex- Fox et a/., 1979). Moore et al. (1984) have shown that calcimedins, a family of calcium-dependent phospholipid-binding pressed in cardiac muscle. This sequence contains a stretch of 20relatively hydrophobic amino acids which mayconstitute proteins, are capable of stimulating the Ca2+-ATPaseof liver an additional transmembranesegment. In addition to thetwo microsomes. One might speculate that the alternatecarboxyl potential glycosylation signals of the cardiac Ca2+-ATPase(at terminus of the Ca2+-ATPaseexpressed in non-muscle cells residues 918 and 962), an additional one lies within the provides the binding site for a cellular factor which could carboxyl-terminal extension of the kidney enzyme at position influence the function of the protein, and provide a potential 1035. According to the proposed membrane topology of the mechanism for these regulatory events. enzyme (MacLennan et al., 1985; Brand1 et al., 1986) the site Both CAMP-dependent protein kinase and protein kinase at position 918 wouldface the cytoplasm, whereas both of the C, as well as calcium/calmodulin-dependent protein kinase, others would face the lumen of the endoplasmic reticulum, have effects on the activity of the cardiac sarcoplasmic reticand thus representpotentialsites for the addition of an ulum Ca2'-ATPase, which are known to be mediated by a oligosaccharide chain. Although there are otherglycosylation specific interaction with the phosphorylatable molecule, phossignals within the sequence of the Ca2'-ATPase molecule (at pholamban (Tada et al., 1974; Movsesian et al., 1984; Kirchpositions 19, 421,589,661, 738), these are unambiguously berger and Antonetz, 1982). Although there isno evidence for cytoplasmic, since the first is close to theamino terminus and the existence of phospholamban in tissues otherthan cardiac, lies before the firstmembrane-spanning segment, whereas the slow-twitch, and smooth muscle (Jorgensen and Jones, 1987; other four are within the large cytoplasmic segment compris- Jorgensen and Jones, 1986; Raeymaeckers and Jones, 19861, ing the phosphorylation and nucleotide binding domains. it is possible that thealtered carboxyl terminus of the kidney Hence these signals could not act as oligosaccharide acceptor Ca2'-ATPase isozyme expresses a binding site for an analosites. Studies are in progress to determine whether any of gous regulatory molecule. these possible sites are actually used. In a recent paper, de la Bastie et al. (1988) demonstrated, The functional significance of these potential structural using S1 analysis, that transcriptsencoding the Ca2+-ATPase features is unclear. For example, no functional significance in smooth muscle only partially protected probes from the has been found yet for the alternative splicing event in fasttwitch Ca2'-ATPase. When expressed in COS-1 cells (Maruy- cardiac Ca2+-ATPasecDNA. The 700-base pair sequence that ama and MacLennan, 1988), both neonatal and adult forms was not protected corresponded to the3"untranslated region of the fast-twitch Ca2'-ATPase had identical calcium trans- of the cardiac Ca2+-ATPasecDNA. It is possible that the port activities with respect to rate, extent, and calcium de- smooth muscle transcript is identical to the predominant pendence. However, the fast-twitch alternative splicing event alternatively spliced transcript, HKl, described in thispaper. produces products which are much less different than those Further study will demonstrate whether this or another alterproduced asa consequence of alternative splicing of the natively spliced product makes up the smooth muscle Caz'ATPase isoform. cardiac Ca2+-ATPasegene. In view of the "calciosome" hypothesis of Volpe et al. (1988) Acknowledgments-We would like to thank Vijay Khanna, Stella and Damiani et al. (1988), it is tempting to suggest that the de Leon, and Kazimierz Kurzydlowski for their expert and invaluable extended carboxyl terminus of the non-muscle Ca2+-ATPase technical assistance, Dr. Graeme Bell, University of Chicago, for his isoform acts to target the protein to such a structure within gifts of human kidney cDNA library, and Dr. Henry J. Klamut, the cell. Paablo et al. (1987) have suggested that ashort Hospital for Sick Children, Toronto, for the opportunity to screen sequence at the carboxyl terminus of the adenovirus E19 his human cosmid library. We are grateful to Dr. Gary Shull, Univerprotein is responsible for its retention in the endoplasmic sity of Cincinnati, for sharing information in his manuscript prior to reticulum. None of the Ca2'-ATPase clones we have identified its publication (Gunteski-Hamblin et al., 1988).

Ca2+-ATPase Non-muscle

Isoform

REFERENCES Allen, G., Trinnaman, B. J., and Green, N.M. (1980) Biochem. J . 187,591-616 Bell. G. I.. Fone. N. M.. Stemoien. M. H.. Wormsted. M. A.. CaDut. D:, Ku,’L., erdea, M . S., Ral1,’L. B., and Sanchez-Pescador, R. (1986) Nucleic Acids Res. 14, 8427-8446 Brandl, C. J., Green, N. M., Korczak, B., and MacLennan, D. H. (1986) Cell 44,597-607 Brandl, C. J., deLeon, S., Martin, D. R., and MacLennan, D. H. (1987) J. Biol. Chem. 262,3768-3774 Breitbart, R. E., Andreadis, A., and Nadal-Ginard, B. (1987) Annu. Reu. Biochem. 56,467-495 Briggs, F. N., Cable, M. B., Geisow, M. G., and Green, N. M. (1986) Biochem. Biophys. Res. Commun. 135,864-869 Campbell, K. P., and MacLennan, D. H. (1981).J. Biol. Chem. 2 5 6 , 4626-4632 Carafoli, E. (1987) Annu. Reu. Biochem. 56, 395-433 Chamberlain, B. K., Levitsky, D. O., and Fleischer, S. (1983) J. Biol. Chem. 258,6602-6609 Damiani, E., Spamer, C., Heilmann, C., Salvatori, S., and Margreth, A. (1988) J. Biol. Chem. 2 6 3 , 340-343 Dean, W. L. (1984) J. Biol. Chem. 259, 7343-7348 de la Bastie, D., Wisnewsky, C., Schwartz, K., and Lompre, A.-M. (1988) FEBS Lett. 229,45-49 Dieckmann, C. L., and Tzagoloff, A. (1985) J. Biol. Chem. 260,15131520 Enyedi, A., Sarkadi, B., Foldes-Papp, Z., Monostory, S., and Gardos, G. (1986) J. Biol. Chem. 261,9558-9563 Feinberg, A. P., and Vogelstein B. (1983) Anal. Biochem. 132,6-13 Fischer, T. H., Campbell, K. P., and White, G. C. I1 (1985). J. Bwl. Chem. 260,8996-9001 Fox, J. E. B., Say, A. K., and Haslam, R. J. (1979) Biochem. J . 184, 651-661 Gunteski-Hamblin, A.-M., Greeb, J., and Shull, G. (1988). J. Biol. Chem. 2 6 3 , 15032-15040 Heilmann, C., Spamer, C., and Gerok. W. (1984) J.Bwl. Chem. 259, 11139-11144 Heilmann, C., Spamer, C., and Gerok, W. (1985) J.Biol. Chem. 260, 788-794 Hoffman, E. P., Brown, R. H., Jr., and Kunkel. L. M. (1987) Cell 5 1 , 919-928 Hunkapillar, M. W., Luan, E., Ostrander, F., and Hood, L. E. (1983) Methods Enzymol. 91, 227-236 Jorgensen, A. O., and Jones, L. R. (1986). J. Biol. Chem. 2 6 1 , 37753781 Jorgensen, A. O., and Jones, L. R. (1987) J. Cell Bid. 1 0 4 , 13431352 Kaser-Glanzmann, R., Jakabova, M., George, J. N., and Luscher, E. F. (1977) Biochim. Biophys. Acta 466,429-440 Kaser-Glanzmann, R., Gerber, E., and Liischer, E. F. (1979) Biochim. Biophys. Acta 558,344-347 Kirchberger, M.A., and Antonetz, T. (1982) J. Bid. Chem. 257, 5685-5691 Kirchberger, M.A., Borchman, D., and Kasinathan, C. (1986) Biochemistry 25,5484-5492 Kleid, D. G., Yansura, D., Small, B., Dowbencko, D., Moore, D. M., Grubman, M. J., McKercher, P. D., Morgan, D. O., Robertson, B. H., and Bachrach, H. L. (1981) Science 2 1 4 , 1125-1129 Korczak, B., Zarain-Herzberg, A., Brandl, C. J., Ingles, C. J., Green, I

-

I

15031

N. M., and MacLennan, D. H. (1988) J. Biol. Chem. 2 6 3 , 48134819 Kyte, J., and Doolittle, R. F. (1982) J. Mol. Biol. 1 5 7 , 105-132 Laemmli, U. K. (1970) Nature 2 2 7 , 680-685 Lowry, 0. H., Rosebrough, N. J., Farr,A. L., and Randall, R. J . (1951) J. Biol. Chem. 193,265-275 MacLennan, D. H., Brandl, C. J., Korczak, B., and Green, N. M. (1985) Nature 3 1 6 , 696-700 MacLennan, D. H., Brandl, C. J., Champaneria, S., Holland, P. C., Powers, V. E., and Willard, H. F. (1987) Somatic Cell Mol. Genet. 13,341-346 Maniatis, T., Fritsch,E. F., and Sambrook, J. (1982) Molecular Cloning; A Laboratory Manual. Cold Spring Harbor Laboratory, Cold Spring Harbor, NY Maruyama, K., and MacLennan, D. H. (1988) Proc. Natl. Acad. Sci. U. S. A. 85,3314-3318 Moore, L., Fitzpatrick, D. G., Chen, T. S., and Landon, E. J . (1974) Biochim. Biophys. Acta 345,405-418 Moore, P. B., Kraus-Friedmann, N., and Dedman, J. R. (1984) J. Cell Science 72,121-133 Movsesian, M. A., Nishikawa, M., and Adelstein, R. S. (1984) J. Biol. Chem. 259,8029-8032 Munro, S., and Pelham, H. R. B. (1987) Cell 48,899-907 Paablo, S., Bhat, B. M., Wold, W. S. M., and Peterson, P. A. (1987) Cell 5 0 , 311-317 Poruchynsky, M. S., Tyndall, C., Both, G. W., Sato, F., Bellamy, A. R., and Atkinson, P. H. (1985) J. Cell. Biol. 1 0 1 , 2199-2209 Prentki, M., Biden, T. J., Janjic, D., Irvine, R. F., Berridge, M. J., and Wollheim, C. B. (1984) Nature 3 0 9 , 562-564 Prentki, M., Corkey, B. E., and Matschinsky, F. M. (1985) J. Biol. Chem. 260,9185-9190 Proudfoot, N. J., and Brownlee, G. G. (1976) Nature 2 6 3 , 211-214 Raeymaekers, L., and Jones, L. R. (1986) Biochim. Biophys. Acta 882,258-265 Reinhart, P. H., Taylor, W. M., and Bygrave, F. L. (1984) Biochem. J. 223,l-13 Sanger, F., Nicklen, S., and Coulson, A. R. (1977) Proc. Natl. Acad. Sci. U. S. A. 74,5463-5467 Sigel, M. B., Sinha, Y. N., and Vanderlaan, W. P. (1983) Methods Enzymol. 93, 3-12 Spamer, C., Heilmann, C., and Gerok, W. (1987) J. Biol. Chem. 262, 7782-7789 Streb, H., Irvine, R. F., Berridge, M. J., and Schulz, I. (1983) Nature 306,67-69 Streb, H., Bayerdorffer, E., Haase, W., Irvine, R. F., and Schulz, I. (1984) J. Membr. Biol. 8 1 , 241-253 Tada, M., Kirchberger, M.A., Repke, D. I., and Katz, A. M. (1974) J. Bwl. Chem. 249,6174-6180 Towbin, H., Staehelin, T., and Gordon, J. (1979) Proc. Natl. Acad. 5%. U. S. A. 76,4350-4354 Volpe, P., Krause, K.-H., Hashimoto, S., Zorzato, F.. Pozzan, T., Meldolesi, J., and Lew, D. P. (1988) Proc. Natl. Acad: Sci. U. S . A: 85,1091-1095 Worley, P. F., Baraban, J. M., Supattapone, S., Wilson, V. S., and Snyder, S. H. (1987) J. Biol. Chem. 2 6 2 , 12132-12136 Yanisch-Perron, C., Vieira, J., and Messing, J . (1985) Gene (Amst.) 33, 103-119 Yoshida, K., and Nachmias, V. T.(1987) J. Biol. Chem. 262, 1604816054 Zubrzycka-Gaarn, E., MacDonald, G., Phillips, L., Jorgensen, A. O., and MacLennan, D. H. (1984) J.Bioenerg. Bwmembr. 16,441-464