THE JOURNAL OF BIOOOGICAL CHEMISTRY 0 1994 by The American Society for Biochemistry and Molecular Biology, Inc.

Val. 269, No. 2, Issue of January 14, pp. 1410-1416, 1994 Printed in U.S.A.

A Sarco/endoplasmic Reticulum Ca2’-ATPase 3-Type Ca2+ Pump Is Expressed in Platelets, in Lymphoid Cells, and in Mast Cells*

(Received for publication, September 17, 1993)

Frank WuytackSS, B61a Pappn, Hilde VerboomenS, Luc RaeymaekersS, Leonard DodeS, Egis Boben, Jocelyne EnouFn, Shaila Bokkalall, Kalwant S. Authill, and Rik CasteelsS From the fLaboratorium uoor Fysiologie, Katholieke Universiteit Leuuen, Campus Gasthuisberg, Herestraat, B-3000 Leuuen, Belgium, the Vnstitut National de la Sant.6 et de la Recherche Medicale U348, Hbpital LariboisiBre, 8 rue Guy Patin, 75010 Paris. France. and the IlThrombosis Research Institute, Manresa Road, Chelsea, London SW3 6LR;United Kingdom ”

An organellar-type of Ca2+ pump formerly detected by means of its phosphoprotein intermediate in platelets and in lymphoid cells, and which runs in acid gels at 97 kDa, is now characterized as sarcolendoplasmic reticu- lum Ca2+ATPase 3 (SERCA3). SERCA3 is co-expressed in these cells along with the housekeeping SERCA2b. This conclusion is based on the following observations. 1) Tryptic digestion the phosphoprotein intermediate

of SERCA3 expressed in COS cells yields a phospho- rylated fragment of about 80 kDa, which can be clearly distinguished from the 57-kDa fragments formed in the SERCAl and SERCA2 pumps. This 80-kDa fragment comigrates with a similar phosphoprotein fragment pre- viously observed in human platelets (Papp, B., Enyedi, A., Paszty, K., Kovacs, T., Sarkadi, B., Ghdos, G., Wuy- tack, F., and Enouf, J. (1992) Biochern. J. 288,297402).

2) An antiserum directed against an NH2-terminal SERCA3-specific peptide (N89) reacts with SERCA3 ex- pressed in COS cells and with the 97-kDa protein in rat platelets and the corresponding protein in human plate- lets. Likewise an antiserum against the rat SERCA3 ter- minus (C90) binds to SERCA3 expressed in COS cells and to the 97-kDa band in rat platelets, but it does not rec- ognize the human platelet pump. In conformity with the predicted absence of the TI tryptic cleavage site in SERCA3, the autophosphorylated aspartyl residue and the COOH-terminal epitope were co-localized on the 80- kDa fragment.

3) The co-expression of nearly equal levels of SERCA3 and SERCA2b messengers in human lymphoblastoid Jurkat cells and in proliferating rat mucosal mast cells was also demonstrated by reverse transcriptase poly- merase chain reaction.

Active Ca2+ pumps, localized both in the plasma membrane (1, 2) and in the membranes bordering a number of different intracellular stores (3, 41, cooperate in most cells to bring the free cytosolic Ca2+ concentration a t levels below 0.1 l.m in the resting state. The operation of these pumps is responsible in most types of cells for the termination of cell activation by removing activator Ca2+ from the cytosol. Ca2+ extrusion from the cell is mediated by one or more members belonging to a vast family of plasma membrane Ca2+-transport ATPases, generally working in parallel with one of the two known types of

* 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.

$ To whom correspondence and reprint requests should be addressed. Tel.: 32-16-345730; Fax: 32-16-345991.

14 1’

Na+:Ca2+ exchanger (5). Also for platelets, arguments have been presented in favor of a Ca2+ extrusion from the cell, me- diated both by a Ca2+-transport ATPase (6) and by a Na+:Ca2+ exchanger (7). However, further characterization of a putative Ca2+-transport ATPase in the platelet plasma membrane is momentarily lacking. In particular no data are available indi- cating its possible homology to other members of the plasma membrane Ca2+-transport ATPase family.

Also the picture for the Ca2+ pumps expressed in intracellu- lar platelet membranes, although somewhat less confusing than that of plasma membrane Ca2+-ATPase, is still incom- plete. Most of the known Ca2+-transport ATPases from intra- cellular stores in different mammalian cell types are products of one of three genes SERCA1-3 (3,4), but recently a putative Ca2+ pump belonging to another family and which might be the mammalian homolog of the yeast secretory pathway Ca2+ pump, has been described (8). The transcripts of the SERCAl and SERCA2 genes are subject to tissue and differentiation stage-dependent alternative processing, thus increasing the isoform diversity. In the case of SERCA2, four different mRNAs and two different proteins (SERCA2a and SERCA2b) have been recognized (9). Isoform-specific antisera were obtained for SERCA2a (10) and for SERCA2b (11). Two different Ca2+ pumps appear to be expressed in platelet non-mitochondrial intracellular membranes (12,131. One of both, with a molecular mass of 115 (100 on acid gels), has been immunologically char- acterized as SERCA2b. The other of 100 kDa (97 kDa on acid gels) presents some distinct characteristics. The steady-state level of its Ca2+-dependent phosphoprotein intermediate is in- creased in the presence of La3+. It shows a tryptic phospho- rylated fragment of 80 kDa (not 57 and 35 kDa as is the case for SERCA2b). Papp et al. (13) showed that the formation of its phosphoprotein intermediate indicated a relatively lower sen- sitivity towards the pump inhibitor thapsigargin and a higher sensitivity towards 2,5-di-(t-butyl)-1,4-benzohydroquinone compared to its SERCA2b companion.

We now present evidence indicating that the unidentified Ca2+ pump in platelets is a SERCA3 gene product. We came to this conclusion by the comparison of the phosphoprotein inter- mediate of the native and trypsinized pump in human and rat platelets with that of the rat SERCA3 pump expressed in COS cells. We, furthermore, confirmed this conclusion by immuno- logical evidence. To the best of our knowledge this is together with the evidence presented by Bobe et al. (14) the first direct indication for the expression of SERCA3 at the protein level in any cell type. Reverse transcriptase-PCR1 also allowed a fur-

MOPS, 4-morpholinepropanesulfonic acid; SERCA, sarco/endoplasmic The abbreviations used are: PCR, polymerase chain reaction;

reticulum Ca2+-ATPase; bp, base paids).

0

SERCA3 in Platelets 1411

ther independent confirmation of SERCA3 expression at the mRNA level in human lymphoblastoid Jurkat cells and in rat mucosal mast cells.

MATERIALS AND METHODS Expression of SERCA cDNAs in COS Cells-A clone of rat kidney

cDNA (RK 8-13) for SERCA3 (3) was obtained from Dr. G. E. Shull (Dept. Molecular Genetics, Biochemistry and Microbiology, University of Cincinnati, Cincinnati, OH). A NotI-BglII fragment containing the entire coding sequence was transferred from the original construct in pBR322 into an EcoRI-cut and blunt-ended mammalian expression vec- tor pMT2 (R. J. Kaufman, Genetics Institute, Boston, MA). Pig SERCA2a and SERCA2b were subcloned in an EcoRI-cut mammalian expression vector pSV57 as described earlier (15). The wild-type rabbit SERCAla cDNA was inserted into the EcoRI site of vector pMT2 (16). Transient expression in COS-1 cells by the DEAE-dextran method and preparation of microsomal fractions were done as described earlier (15).

Preparation of Antibodies Against Defined Rat SERCA3 Peptide Se- quences-Antibodies were raised in rabbits against two different syn- thetic SERCA3 peptides conjugated to bovine thyroglobulin. The pep- tides were selected to have a high antigenic potential and to be distinct from the homologous stretches of amino acids in SERCAl or SERCA2. Peptide N89 (VTDARERYGPN) stretches from amino acid residues 29 to 39 in the amino-terminal part of the rat SERCA3 (3), i.e. a zone which is situated NH,-terminally of the first putative membrane-spanning domain of the pump. The peptide was coupled to the carrier by means of glutaraldehyde at a concentration of 0.17 mg of peptide/mg of thyro- globulin (17).

Peptide C90 (cLSRHHVDEKKDLK) corresponds to the extreme COOH terminus of the rat SERCA3. An additional cysteine residue (indicated in small type in the sequence above) was added to its amino terminus in order to allow coupling of 0.5 mg of peptiddmg of thyro- globulin by means of the m-maleimidobenzoyl-N-hydroxysuccinimide ester method (17).

Phosphoprotein Isotyping-Phosphoprotein isotyping of the Ca2+- transport ATPases were done as described earlier (18). In short, phos- phorylation buffer contained (in m ~ ) : 100 KCl, 30 imidazole, pH 6.8,0.5 CaC1,. The phosphorylation reaction was done at 0 "C. It was started by the addition of [y-32P]ATP (3 nM, 3000 mCi/nmol) and stopped with 10% trichloroacetic acid, 0.5 m~ ATP, 50 m~ phosphate. Following pelleting for 10 min at 15,000 x g and three washes with the stop solution, Laemmli-type electrophoresis (0.75-mm thick 7.5% gels), semi-dry im- munoblotting onto Immobilon-P membranes (Millipore, Brussels) were done. When needed, membranes were pretreated for 10 min on ice in phosphorylation buffer with 5 pg of trypsin (from bovine pancreas). Trypsinization was stopped by 20 pg of trypsin inhibitor (from turkey

ATP. egg) and phosphorylation was started immediately by the addition of

Immunostaining of the BZots-Following autoradiographic visual- ization of the 32P-labeled peptides, the blots were quenched in Tris- buffered saline with 0.05% Tween 20, reacted with primary antiserum followed by the appropriate peroxidase-conjugated secondary immuno- globulins and stained in a solution of: 100 mM MOPS, 60 mM Tris, pH 7.2, 1.4 m~ 3',3'-diaminobenzidine, 6 m~ NiC12, 0.0003% H202. When needed the blots could still be stained afterwards with Coomassie Bril- liant Blue. The secondary antibodies were peroxidase-conjugated swine anti-rabbit or peroxidase-conjugated rabbit anti-mouse immunoglobu- lins (DAKO A/S, Glostrup, Denmark). Tris-buffered saline is 150 m~ NaCl, 10 m~ Tris-HC1, pH 7.5.

Preparation of Membrane Fractions-Platelet membrane vesicles were isolated from human and rat platelet-rich plasma as previously described (19).

Microsomes were isolated from COS-1 cells expressing the different SERCA pumps or from cultured A7r5 rat aortic smooth muscle cells (ATCC CRL 1444) or human lymphoblastoid Jurkat cells as described earlier for COS-1 cells (15).

Reverse Danscriptase-PCR Analysis of SERCA3 and SERCA2- Poly(A+) RNA was prepared from rat mucosal mast cells or from human Jurkat cells using the Micro-FastTrack kit (Invitrogen Co., British Bio- technology Products Ltd., Abingdon, United Kingdom). 0ligddT)- primed first strand cDNA was synthesized using Moloney murine leu- kemia virus-reverse transcriptase (Life Technologies Inc.) (20). SERCA3 and SERCA2 were amplified using a single set of PCR primers whose sequences were common for both Ca2+ pumps. The 5' primer (TGCCTGGTAGAGAAGATGAA) corresponded to nucleotides 1470- 1489 in SERCA3 and to nucleotides 1546-1565 in SERCA2 of the rat, the 3' primer (CCC'ITCACAAACATC'ITGCT) to nucleotides 1659-1678

in SERCA3 and to nucleotides 1732-1751 in SERCA2 (numbering for SERCA3 and SERCA2 according to EMBUGenBank data base acces- sion numbers M30581 and X15635, respectively). Amplification was done with Taq polymerase in a Perkin Elmer DNA thermal cycler model 480 for 30 cycles, each cycle consisting of 1 min of denaturation at 94 "C, 1 min annealing at 55 "C, and 1 min extension at 72 "C.

In order to determine the ratio of the SERCA2 to the SERCA3 mes- sages, the PCR products corresponding to SERCA2 and SERCA3 were discriminated by using restriction digestion at unique and specific sites (for rat we used: BsaHI for SERCA2 and Sty1 for SERCA3; for human: XbaI for SERCA2 and PstI for SERCA3). ARer 30 cycles, the amplified mixture was diluted 20-fold in fresh amplification mixture containing 15 nCi/pl [a-32PldCTP and two additional cycles were given. The labeled amplication products or their restriction fragments were separated on a 6% polyacrylamide gel and quantified by means of a PhosphorImager model 425 (Molecular Dynamics, Sunnyvale, CA). For the calculation of the SERCMSERCA3 ratio from the radioactivity of the PCR frag- ments, their respective differences in length, and CG content was taken into consideration. The latter was necessary since dCTP was the only labeled nucleotide in the assay.

In order to confirm their identity, the amplification products were separated on a 1% agarose gel and the DNA mixture recovered from a band of approximately 200 base pairs by means of the QIAEX extraction kit (Qiagen Inc., Chatsworth, CA). The DNA was blunt-ended, phos- phorylated, and cloned in a SmaI-cut dephosphorylated pGEM-7Z(D+ vector (Promega Co.) as described earlier (15). Clones containing SERCA3 inserts were discriminated from those with SERCA2 inserts by a digest by means of RsaI. Three individual SERCA2 clones and three SERCA3 clones were sequenced in both directions by means of dideoxy sequencing using the Sequenase enzyme (U. S. Biochemical Carp.). DNA to protein sequence translations, DNA and protein se- quence alignments, and the construction of the dendrogram were all done with the PC/GENE program (IntelliGenetics, Inc., Mountain View, CA).

RESULTS

SERCA3 Shows a Qpical Tryptic Fragmentation Pattern of Its Phosphoprotein Intermediate-All P-type Ca2+-transport ATPases present a characteristic Ca2+-dependent phosphopro- tein intermediate. It has been shown earlier that the tryptic fragmentation pattern of the phosphoprotein intermediates can be used to discriminate the plasma membrane Ca2+ pumps from the SERCA pumps (181, and within the group of SERCA pumps to distinguish between SERCAl and SERCA2 gene products. Fig. 1 indicates that by the same method, also SERCA3 pumps can easily be recognized. The figure shows the phosphoprotein intermediates of SERCA2b or SERCAS overex- pressed in COS-1 cells. The SERCA2b which is endogenously expressed in COS cells and in the smooth muscle cell line A7r5 can also be seen. The Ca2+-dependent phosphoprotein interme- diates were formed in microsomes with or without trypsin pre- treatment. In the case of SERCA2, tryptic digestion rapidly creates a fragment of 57 kDa. Upon protracted digestion, this 57-kDa fragment is further converted into a 33-kDa fragment (not shown in the figure, but see Ref. 18). This confirms the well known breakdown of SERCA2 into, respectively, the A and Al fragments as a result of the successive use of the tryptic cleav- age sites TI and T2 (21). The situation is different for SERCA3. The unproteolyzed SERCA3 pump presents a slightly higher electrophoretic mobility compared to the SERCA2b isoform (97 and 100 kDa in acid gels, or 100 kDa and of 115 kDa in Laemmli gels, respectively). Moreover in the case of SERCA3, a 80-kDa (same mobility in both acid and Laemmli gels) tryptic phosphoprotein intermediate fragment is formed.

The Pattern of SERCA3 Dyptic Fragmentation Can be Ex- plained by the Absence of TI-The most likely explanation for the formation of a 80-kDa tryptic SERCAS fragment versus a 57-kDa one, both in SERCAl and SERCA2, lies in the predicted absence of the T1 tryptic cleavage site from SERCA3 (see Fig. lb), whereas this site is present in both other pumps (3). This idea is corroborated by simultaneous phosphoprotein isotyping and immunoblotting, which allowed a very precise comparison

SERCA3 in Platelets 1412

a

n -66

-29 "

I

2b

I

N89

c (801ux87W FIG. 1. SERCA2 and SERCM present distinct tryptic-hgmen-

tation patterns of their phosphoprotein intermediates. a, an au- toradiogram of the Ca"-pump phosphoprotein intermediates on a semi- dry blot is shown. Microsomal membranes (3 pg of proteidane) from COS-1 cells transiently transfected with SERCA2b cDNA (lanes 1 and 2 ) or SERCA3 cDNA (lunes 3 and 4 ) and of A7r5 smooth-muscle cells (lunes 5 and 6 ) were phosphorylated as indicated under "Materials and Methods." The membranes in lunes 2, 4, and 6 were pretreated with trypsin before phosphorylation, those in lunes I , 3, and 5 served as controls. Closed urrowheuds indicate SERCA2 or its fragments, open arrowheads indicate SERCA3 or its fragments. b, a scheme is shown of the SERCA2 and SERCA3 proteins. The phosphorylation site (P) and the tryptic cleavage sites (TI and T,) are indicated. The epitopes for the SERCA2b- (2b) and for the SERCA3 (N89 and CSO)-specific antisera are also indicated.

of immunostaining and phosphorylation patterns. Antibodies N89 and C90, respectively, directed against an NH2-terminal and a COOH-terminal SERCA3 epitope, both recognized the unproteolyzed rat SERCA3 pump expressed in COS cells. The SERCA3 specifity of the N89 and C90 antisera is illustrated in Fig. 2 where it is shown that the antisera did not bind to the SERCAla, SERCMa, or SERCA2b proteins expressed from the corresponding cDNAs in COS cells.

The 80-kDa phosphoprotein fragment of rat SERCA3 does not react with the N89 antiserum, whereas it does react with the C90 antiserum. This is illustrated in Fig. 3. The figure shows in lanes 2 and 3 a partial tryptic proteolysis of the rat SERCA3 cDNA expressed in COS cells. The phosphorylations in panels a and c show about equal amounts of undigested SERCA3 (upper arrow) and of the 80-kDa tryptic fragment (lower arrow; f), but whereas both the native ATPase and the

a b 1 2 9 4 1 2 9 4

C d 1 2 9 4 1 2 3 4

-66 - . , 0' 0

.- 29 -

identical semi-dry Western blots (a and b and c and d ) from 7.5% Laemmli gels are shown. Each lane contained approximately 3 pg of microsomal membrane protein. Different cDNAs were expressed in COS-1 cells: lane 1, rabbit SERCAla; lune 2, pig SERCA2a; lune 3, pig SERCA2b; lune 4, rat SERCA3. Blot u l b was first reacted with the SERCA3-directed antiserum N89 (1/1000 dilution) as shown inpanel a. The same blot was subsequently reacted with the SERCAl-specific

other SERCA3-directed antiserum C90 (V200 dilution) and subse- monoclonal antibodyA52 (panel b). Blot c ld was first reacted with the

quently with the SERCA2-specific monoclonal antibody IID8.

80-kDa fragment are stained with a nearly equal intensity by the C90 antibody (panel b) , the N89 does not recognize the 80-kDa fragment (panel d ) . This indicates that the phospho- rylation site and the COOH terminus in SERCA3 are still on the same fragment after the first tryptic cleavage. The use of tryptic cleavage site TI, provided it was present in SERCA3, would separate the phosphorylation site from the C90 epitope. On the other hand, in some experiments a faint rat SERCA3 tryptic fragment of about 20 kDa was specifically recognized by the N89 antiserum (results not shown). This presumably cor- responds to the NH2-terminal fragment proximal to the T2 cleavage site.

SERCA3 in Platelets 1413

a 1 2 5 4

C 1 2 5 4

b d 1 2 3 4 1 2 3 4

FIG. 3. Western blot analyoin of the SERCAS tryptic fragments. Autoradiograms of the Ca2+-pump phosphoprotein intermediates on a semi-dry blot from 7.5% Laemmli gels are shown (panels a and c). Microsomal membranes from human platelets (lanes 1; 3 pg of protein) and from COS-1 cells transiently transfected with rat SERCA3 cDNA (lanes 2-4; 3 pg of protein) were pretreated with trypsin and then phosphorylated as indicated under “Materials and Methods.” Panel b shows an immunostaining of blot a with the C90 SERCA3-specific an- tiserum. Panel d shows an immunostaining of blot c with the N89 SERCA3-specific antiserum. In lanes 1 and 2, the membranes were trypsinized and subsequently phosphorylated in the presence of 50 p~ Ca2+. Lanes 3 and 4 show the effect of Ca2+ or EGTAon tryptic digestion. In lanes 3 the membranes were trypsinized in the presence of 50 p~ Ca2+ and phosphorylation was done in the presence of 150 p~ Ca2+, 100 p~ EGTA. In lanes 4 the membranes were trypsinized in the presence of 100 p~ EGTA and as in 3, phosphorylation was done in the presence of 150 PM Ca2+, 100 p~ EGTA. The arrowheads indicate human SERCA3, the arrows rat SERCA3, the arrowheadslurrows marked with fpoint to their respective approximate 80-kDa fragments. Note the absence of undigested SERCA2 phoshoprotein intermediates in panels a and c, because the use tryptic cleavage site TI in SERCA2 is much faster than that of T2 in SERCA3.

A further argument supporting the view that in SERCA3 TI is absent and hence that trypsin first attacks the pump at Tar comes from the Ca2+ sensitivity of the formation of the 80-kDa SERCA3 fragment. Indeed for SERCAl it has been shown that the exposure of Tz is Ca2+ dependent (22). In the presence of Ca2+, the pump enzyme resides in the El conformation and Tz is readily accessible to trypsin. EGTAinduces the E2 conforma- tion and the tryptic digestion is much slower in this condition. Fig. 3, lanes 3 and 4, shows that the formation of the 80-kDa SERCA3 fragment is slowed down in the presence of EGTA.

Platelets Express SERCA3”Both human and rat platelets express SERCA2-b along with SERCA3. This can now be shown by analysis of the phosphoprotein intermediates and their re- spective tryptic fragments and by immunoblot analysis (Figs. 4 and 5). Whereas both human and rat platelet SERCA2b pre- sent on Laemmli gels the same mobility corresponding to 115 kDa, a remarkable difference in mobility is observed between human (103 kDa) and rat (100 kDa) SERCA3. The mobility of rat SERCA3 in platelets and that of rat SERCA3 cDNA ex- pressed in COS cells is the same.

Rat and human SERCA3 are also immunologically distinct. Whereas the N89 antiserum reacts with both the rat and the human pump, the C90 antiserum does not recognize human SERCA3. The monoclonal antibody PMM 430 (23) does in human platelet membranes apparently bind to the same band as N89 (Fig. 4). It does not recognize the SERCA3 protein in rat platelets.

The Simultaneous Expression of SERCA2 and SERCA3 mRNAs in Human Lymphoblastoid Jurkat Cells and in Rat Mast Cells-Previous phosphoprotein analysis had suggested that platelets and lymphoid cells including Jurkat cells, co- express another organellar Ca2+ pump along with the house- keeping SERCA2b (13). We have now confirmed the predicted expression of SERCA3 in Jurkat cells at the RNA level by

a 1 2 1 2 b

r c 1 2 1 2

d

1 7

” FIG. 4. Comparison of SERCMb and SERCAS in human and

rat platelet membranes. Panel a shows an immunoblot of a 7.5% Laemmli gel immunostained with the N89 SERCA3-specific antiserum; panel b shows the same blot as in panel a , but overstained with the SERCA2b-specific antiserum. Panel c shows another blot stained with the monoclonal antibody PMM 430 (23);panel d shows the same blot as in panel c, but overstained with the antiserum N89. Lane 1 contains 10 pg of protein of human platelet membranes, lane 2, 10 pg of protein of rat platelet membranes. The filled arrowhead indicates SERCA2b; open arrowheads, human SERCA3; and open arrows, rat SERCA3.

a b 1 2 3 4 6 1 2 3 4 6

SERCA2b and SERCA3 in human platelets. Panel a shows an au- FIG. 5. Phosphoprotein intermediates and immunostaining of

toradiogram of a semi-dry blot from a 7.5% Laemmli gel presenting the Ca2+-pump phosphoprotein intermediates in human platelet mem- branes. Each of lanes 1-5 received 8.5 pg of protein. Following autora- diography, the blot was cut into five strips, which were each immuno- stained with a different antibody and the results are shown in panel b. Lane 1 shows staining with the N89 SERCA3-specific antiserum; lane 2 with a SERCA2b-specific antiserum (11); lane 3 with a mixture of both

clonal antibody IID8 (26); lane 5 with a SERCA2a-specific antiserum antisera used in lanes 1 and 2; lane 4 with the SERCM-specific mono-

(10). SERCA2b is indicated by the filled arrowhead; SERCA3 by the open arrowhead.

means of reverse transcriptase-PCR (Fig. 6). For the PCR anal- ysis we first compared the published cDNA sequences for rat SERCA2 and SERCA3 and selected a couple of PCR primers with the prerequisite that, when present, both SERCA2 and SERCA3 mRNAs would be amplified. Fig. 6a shows a phosphor- image of the amplification products and their restriction frag- ments obtained from the Jurkat cell RNA. For rat, and hence presumably also for human, the predicted lengths of the am- plified DNA fragments would be: 209 for SERCA3 and 206 for SERCA2-ah. After subcloning and sequencing the amplified cDNA mixture, two classes of cDNA could indeed be discrimi- nated on the basis of restriction analysis (see “Materials and Methods”). Based on the known sequence of human SERCA2 (24) one of the classes appeared to represent SERCA2, by elimi- nation the other might then be SERCA3. The presumed iden- tity of both classes of cDNA was indeed confirmed by sequenc- ing (Figs. 6b and 7). Digesting the PCR-amplified fragments

SERCA3 in Platelets 1414

a

2 0 6 / 2 0 9 bo 168 b

114 b

92 b

b

1 2 3 4

C

I 111631

I s n t x moo 16300Y

FIG. 6. Reverse transcriptase-PCR amplification of SERCAS and SERCA2 fragments from Jurkat T lymphocytes. a shows a phosphorimage of reverse transcriptase-PCR fragments separated on a 6% polyacrylamide gel. Primers were chosen to co-amplify SERCA2 and SERCA3. Lane 1 shows the combined SERCA2/3 amplification product;

XbaI; lane 3, the SERCA3-specific fragments obtained after digestion lane 2, the SERCA2-specific fragments obtained after digestion with

with PstI; lane 4, the combination of digestion with both restriction enzymes. Open arrowheads represent SERCA3-related fragments. Filled arrowheads represent SERCA2-related fragments. The length in base pairs is given at the left (SERCA2/SERCA3). b, alignment of the human SERCA3 and SERCA2 peptide fragments deduced from the PCR sequencing with the corresponding peptides from rat. c, dendro- gram comparing all the peptide sequences in the EMBUGenBank data bases which are homologous to the sequences given in B . The percent- age of amino acid identity at each branch point is indicated above the branching point. The accession numbers of the DNA sequences from which the peptide sequences were derived is also given in the right column.

with XbaI converted the 206-bp SERCA2 product into 114- and 92-bp fragments, leaving the 209-bp fragment undigested. Di- gestion with PstI reduced the 209-bp SERCA3 product into 168- and 41-bp fragments, leaving the 206-bp SERCA2 fragment unaffected.

A comparison of the sequence of the amplified cDNA frag- ment for SERCA2 cDNA which lies in between the primers (i.e. with the exclusion of the primer sequences proper) and which comprises a total of 166 bp, shows a 100% identity to that reported for human kidney (23) and a 86.1% identity to the corresponding stretch in rat (3). However, the corresponding peptide sequences are 100% identical between human and rat. For SERCA3, the difference is much more pronounced. Of the 169-bp long stretch in between the primers, only 81.1% are identical between human and rat. At the protein level this corresponds to 85.7% of identity.

SKRCA3 HUMAN- CGTGTTCGACACCGACCKCAffiCTCl'GTCCCGGGTGGAGCGAGC -45

SKRCA3 RAT - CGTGT'XTCACACAGACCTGAAAGGACTGTCGCGGGTCGAGCGCGC -1534 I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I

SKRCA3 W- n;GCGCCTGTMCACff iTC~GCAGCl'GATGCff iMff iAGTT -90

SKRCA3 RAT - TGGTGCCTGCAATTCCGTCATCMGCMCTCATGCAGMGGAGm -1579 I l l I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I

s R C A 3 HUMAN- CACCCTGGAGTTCTCCCGAGACCGGAAATCCATGTCCGTGTACTG -135

SKRCAB RAT - CACCCTCGAATTCTCCCGCCGGAAGTCCATGTCTGTGTACTG -1624 I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I

SKRCA3 HljWN- CACGCCCACCCGCCCTCACCCTACEGCCA-C -169

SKRCA3 RAT - CACACCCACTCGTGCTGACCCCMGGCCCAGGGC -1658 I l l I I I I I I I I1 I I I I I I I I I I I I I

FIG. 7. The sequence of the PCR fragment from human SERCA3. The PCR fragment amplified from human lymphoblastoid Jurkat cells was cloned in a SmaI-cut pGEM-72(0+ vector and se- quenced as indicated under "Materials and Methods." The 169-bp long cDNA stretch of human SERCA3 sequence lying within the primers (i.e. excluding the primers proper) is represented in the upper line and aligned with a part of the published rat SERCA3 sequence (3) repre- sented in the lower line and which stretches from position 1490 to 1658 (numbering according to data base accession number M30581). Identi- cal bases are connected by vertical lines.

Fig. 6c shows a phylogenetic tree (dendrogram) for the SERCA pump proteins. The dendrogram was constructed tak- ing into account each of the 10 peptide sequences for SERCA represented in the EMBUGenBank data bases, supplemented with the human SERCA3 sequence obtained in the present work. Only the peptide sequences homologous to the SERCA3 stretch specified by the PCR-amplified fragment were consid- ered in this analysis. Although the sequence information con- cerns only a limited part of the pumps, it is clear from this analysis that the novel Jurkat sequence resembles most to the rat SERCA3 sequence, but that rat and human SERCA3 differ more from each other then, respectively, the SERCA2 or SERCAl sequences differ in between the species.

SERCA2 and SERCA3 messengers could also be detected in RBL-2H3 rat mucosal mast cells. Fig. 8 shows the reverse transcriptase-PCR amplification product in rat mast cells, which comprises a mixture of SERCA2 and SERCA3. About 50% of the amplified DNA (on a molecular basis) was cut by the restriction enzyme BsaHI and is thereby characterized as a SERCA2 DNA fragment. The remaining 50% must be a SERCA3 fragment since all the DNA product not digested by BsaHI was digested by StyI, yielding a fragment of the pre- dicted length and in the predicted quantity.

DISCUSSION SERCA Isoform Qping by Means of Phosphoprotein Frag-

ments-For the analysis of the phosphoprotein intermediates, like in a previous report (111, we have preferred to conduct the phosphorylation reactions at a high specific activity of [y-32P]ATP and to separate the radioactive peptides by means of Laemmli-type electrophoresis followed by semi-dry blotting. Although by this technique a considerable fraction (>75%) of the phosphoprotein intermediate is hydrolyzed, enough of i t remains to allow a fast (3-16 h exposure at room temperature) autoradiographic analysis of the blot. Most important for our purpose, however, the Laemmli-type of electrophoresis pro- vides a resolving power unparalleled by the alternative acid gel systems which were designed to maximize phosphoprotein lev- els. An additional advantage of the approach followed in the present work is that, for cells like A7r5 which express also plasma membrane Ca2+ pumps, the phosphoenzyme of this latter pump does not show up after Laemmli-type electropho- resis.

From previous work by Papp et al. (13) it became clear that platelets and a number of lymphoid cells co-express two types of SERCA pumps: i.e. SERCMb, the "housekeeping" Ca2+

SERCA3 in Platelets 1415

1 2 3 4

FIG. 8. The ratio of SERCA!USERCAS messengers in rat muco- sal mast cells. A phosphorimage is shown of reverse transcriptase- PCR fragments amplified from RBL-2H3 rat mucosal mast cell RNA and separated on a 6% polyacrylamide gel. Lane 1 shows the combined SERCA2/3 amplification product. Lane 2 shows the SERCA2-specific digestion using BsaHI. Lane 3 shows the SERCA3-specific digestion using StyI. Lane 4 shows the combined digestion with BsaHI and StyI. Open arrowheads indicate SERCA3 fragments. Closed arrowheads in- dicate SERCA2 fragments. The length in base pairs of the amplified fragments is indicated at the left (SERCA2/SERCA3). Note that the restriction enzymes differ from the ones used in Fig. 6. This is because of the different animal species (rat and human) used in both experi- ments.

pump found in nearly all cells and a second one of unknown identity. The typical proteolysis pattern of this latter pump’s phosphoprotein intermediate suggested, however, that this could be the SERCA3 pump. Indeed from the amino acid se- quence alignment of SERCAl, SERCA2, and SERCA3 (3), it was apparent that of these three pumps only SERCA3 would present a critically modified first tryptic cleavage site T1, whereas all three of the pumps would contain a normal cleav- age site Tz. The Ar$05 residue which determines the tryptic cleavage site T1, in both SERCAl and SERCA2, is replaced in SERCA3 by Pro505. I t should be remarked, however, that in SERCA3 this Pro505 is followed by a Lys506 residue that poten- tially could specify an alternative T1 cleavage site. But appar- ently, at least under the experimental conditions used in our work, this putative alternative TI site in SERCA3 is not used. Therefore the specific organization of the TI and T2 sites would, upon tryptic digestion of SERCAl or SERCA2, lead to the rapid formation of a phosphoprotein fragment of about 57 kDa (frag- ment A, T1 cleavage) that then would be further converted to a size of about 33 kDa (fragment Al, T2 cleavage). In contrast, in view of the absence of T1, digestion of SERCA3 would produce a 80-kDa fragment.

We now confirmed these predictions by expressing rat SERCA3 in COS cells and carefully comparing the tryptic phos- phoprotein fragmentation of the expressed pump with this in platelets.

Production of SERCA3-specific Antisera and Comparison with Other SERCA-specific Antibodies-An important part of the present work involves the production of SERCA3-specific antisera. Two different SERCA3-specific antisera were ob- tained. One (N89) directed against a peptide stretch running from amino acid residues 29-39 in the amino-terminal part of rat SERCA3, the other ((290) against the 13 amino acid long COOH-terminal tail of rat SERCA3. Identical or similar pep- tides (i.e. allowing 5 4 mismatches in N89; 5 6 mismatches in C90) were found to be absent from all the amino acid sequences translated from any of the SERCA entries in the EMBU GenBank data bases, except for SERCA3.

In order to confirm the isoform specificity of the antisera, Western blots of SERCAla, SERCMa, SERCMb, and SERCA3

cDNAs expressed in COS cells were analyzed. Neither of the SERCA3 antisera recognized SERCAla, nor SERCA2a or SERCA2b. These results strongly suggest that our antisera are SERCA3 specific. It should be mentioned, however, that these observations do not unequivocally demonstrate the isoform specificity because the cDNAs were obtained from different species and since the same isoforms of mammalian SERCA pumps typically present a 1-3% amino acid difference among the different species. An alignment of the SERCA2a protein sequences from cat, human, rat, pig, and rabbit translated from the EMBUGenBank data base entries for instance, indicates that 976 out of 997 residues (97.9%) are identical in these species. The possibility of a false-negative reaction of the SERCA3 antisera with other SERCA pumps, simply due to species differences, although unlikely, can therefore not be fully excluded. A mouse monoclonal antibody P M M 430, raised against human platelet intracellular membranes and known to inhibit Ca2+ uptake (23) in platelet organelles, appears to bind to exactly the same protein band as our SERCA3-specific N89. This strongly suggests that PMM 430 is directed against SERCA3. However, the monoclonal antibody shows a species- specificity. Indeed, immunoblot analysis showed that while this antibody does bind to SERCA3 in human platelets, it does neither react with rat SERCA3 expressed in COS cells, nor with the corresponding band in rat platelets.

As was shown earlier (13,25), our SERCA2b-specific antise- rum can be used to visualize the second organellar-type of Ca2+ pump in human and rat platelets. Our SERCA2a-specific an- tiserum does not detect a corresponding pump in platelets. The monoclonal antibodies IID8 (26) and A52 (27) react specifically with, respectively, SERCA2 and SERCA1. Monoclonal antibody IID8 binds neither SERCA3 (data not shown) nor SERCAl (Fig. 2d) , A52 only binds to SERCAl (data not shown). Hence a plethora of SERCA isoform-specific antibodies are presently available.

SERCA3 Shows an Unusually Large Interspecific Sequence Diversity-One interesting observation concerns the higher ap- parent molecular mass of human (103 kDa) versus rat (100 kDa) platelet SERCA3 on Laemmli-type electrophoresis. This is observed for the unproteolyzed pumps by phosphoprotein analysis and it is confirmed by immunostaining. A similar dis- crepancy is also seen for the tryptic phosphoprotein fragments. In the latter case immunological confirmation is, however, pre- cluded by the fact that our COOH-terminal-specific antiserum (C90) only binds to rat SERCA3.

Such a considerable interspecific difference in mobility is not observed for SERCA2b. In fact, this is not surprising because a comparison of the amino acid sequences of all other Ca2+-pump isoforms present in the EMBUGenBank data bases learns that the same Ca2+-pump isoforms from different mammalian spe- cies usually are 97% or more identical.

It is remarkable that SERCA3 differs from the other Ca2+ pumps by the fact that it shows an unusually large sequence variation between the different animal species. This conclusion is corroborated by our sequencing data which show that, at least in the limited part of SERCA2 and SERCA3 that was PCR amplified in the present study, SERCA2 presented a 100% se- quence identity between rat and human at the amino acid level, whereas SERCA3 was only 85.7% identical. One could of course consider the possibility that the novel PCR-amplification prod- uct belongs not to SERCA3 but instead to another as yet un- known SERCA pump. However, the phylogenetic analysis showed that it most resembles rat SERCA3. Hence, this infor- mation reinforces the conclusion drawn from phosphoprotein and immunological analysis.

The Expression of SERCA3 in Different Tissues-A study of the expression of SERCA3 in the different tissues, which was

1416 SERCA3 in Platelets

hitherto based only on Northern blot analysis (3), gave a rather confused picture. SERCA3 clearly showed a much more re- stricted expression pattern compared to SERCA2, but no clear embryological or functional relationship could be discerned be- tween the SERCA3-positive tissues. The present results, com- bined with similar observations made by Bobe et al. (141, may help to throw some new light on this topic. We now know that SERCA3 is expressed in platelets, probably in levels compa- rable to those of SERCA2b as can be deduced from the relative intensity of both pump's phosphoprotein intermediates. The fact that spontaneous hypertensive rats express higher levels of SERCA3 in their platelets compared to normotensive con- trols, presents an interesting observation (28). Previous phos- phorylation data from Papp et al. (13) and data presented here can now be interpreted to indicate that SERCA3 is also ex- pressed in lymphoid cells. We have confirmed this by means of immunoblot and PCR analysis to be the case for human T- lymphoblastoid Jurkat cells. Also in proliferating mast cells nearly equal levels of SERCA3 and SERCA2 messengers could be amplified by means of reverse transcriptase-PCR. This might indicate that, besides cells of the lymphoid series, also cells of the myeloid series are expressing SERCA3. The expres- sion pattern of SERCA3 might even be broader and include the embryologically related endothelial cells. Indeed in situ hybrid- ization, supplemented by RNase protection assays, further- more, indicated that in rat embryos SERCA3 is also expressed in endothelial cells of arterial vessels.2

Further data are needed in order to find out whether SERCA3 and SERCA2b in a given cell co-localize in the same compartment andor whether they serve each a distinct role. The documented lower affinity for Ca2+ of SERCA3 compared to SERCA2 (4), at least when both are artificially expressed in COS cells, still remains an interesting and enigmatic property.

Acknowledgments-We thank Dr. Gary E. Shull for the gift of the SERCA3 cDNA clone RK8-13, Dr. Jens Peter Andersen, Institute of Physiology, University ofAarhus, Denmark, for the SERCAl cDNA, and Dr. Randal J. Kaufman, Genetics Institute, Boston, for the gift of ex- pression vector pMT2. The monoclonal antibodies IID8 and A52 were kindly provided by, respectively, Dr. Kevin Campbell, Howard Hughes Medical Institute, Iowa City, and Dr. David H. MacLennan, Banting and Best Department of Medical Research, Toronto, Canada. The RBL- 2H3 rat mucosal mast cells were kindly provided by Dr. Israel Pecht, Department Chemical Immunology, Weizmann Institute of Science, Re- hovot, Israel. We also thank Dr. Stefan De Reys, Center for Molecular

Anger, M., Samuel, J.-L., Marotte, E , Wuytack, F., Rappaport, L., and Lompr6, A.-M. (1993) FEBS Lett., in press.

and Vascular Biology, University of Leuven, Belgium, for the prepara- tion of human platelets, and Dr. Humbert De Smedt, Laboratorium voor Fysiologie, for help with the cell cultures.

REFERENCES

2. Wuytack, F., and Raeymaekers L. (1992) J. Biwnerg. Biomembr: 24,285-300 1. Carafoli, E. (1991) Physiol. Reu. 71, 129-153

3. Burk, S. E., Lytton, J., MacLennan, D. H., and Shull, G. E. (1989) J. Biol. Chem. 264, 18561-18568

4. MacLennan, D. H., Toyofuku, T., and Lytton, J. (1992) Ann. N.Y Acad. Sci. 871,l-10

5. Missiaen, L., Wuytack, F., Raeymaekers, L., De Smedt, H., Droogmans, G., De Jaegere, S., and Casteels, R. (1992) in International Encyclopedia of Phar- macology and Therapeutics (Taylor C. W., ed) Vol. 139, pp. 347-405, Per-

6. Enouf, J., Bredoux, R., Bourdeau, N., Sarkadi, B., and Levy-Toledano, S. (1989) gamon keas , United Kingdom

7. Valant, P. A., Adjei, P. N., and Haynes, D. H. (1992) J. Membr: Biol. 130,63-82 8. Gunteski-Hamblin, A"., Clarke, D. M., and Shull, G. E. (1992) Biochemistry

9. Eggermont, J. A., Wuytack, F., and Casteels, R. (1990) Biochem. J. 268.901- 31, 7W7608

907 10. Eggermont, J. A., Wuytack, F., Verbist, J., and Casteels, R. (1990) Biochem. J.

271,649-653 11. Wuytack, E , Eggermont, J. A,, Raeymaekers, L., Plessers, L., and Casteels, R.

(1989) Biochem. J. 264,765-769 12. Enouf, J., Bredoux, R., Papp, B., Djaffar, I., Lompr6, A M., Kieffer, N., Gayet,

O., Clemetson, K, Wuytack, F., and Rosa, J.-P. (1992) Biochem. J. 286, 13. Papp, B., Enyedi, A., Pbszty, K, Kovbcs, T., Sarkadi, B., Gbrdos, G., Magnier,

135-140

14. Bobe, R., Bredoux, R., Wuytack, F., Quarck, R., Kovbcs, T., Papp, B., Corvazier, C., Wuytack, F., and Enouf, J. (1992) Biochem. J. 288,297-302

15. Verboomen, H., Wuytack, E , De Smedt, H., Himpens, B., and Casteels, R. E., Magnier, C., and Enouf J. (1994) J. Biol. Chem. 289, 1417-1424

16. Andersen, J. P., Vilsen, B., and MacLennan, D. H. (1992) J. Biol. Chem. 267, (1992) Biochem. J. 286,591-596

17. Harlow, E., and Lane, D. (1992) in Antibodies: A Laboratory Manual, Cold 2767-2774

18. Wuytack, F., Kanmura, Y., Eggermont, J. A, Raeymaekers, L., Verbist, J., Spring Harbor Laboratory, Cold Spring Harbor, NY

19. Papp. B., Enyedi, A,, Kovbcs, T., Sarkadi, B., Wuytack, F., Thastrup, O., Gbr- Hartweg, D., Gietzen, K., and Casteels, R. (1989) Biochem. J . 257,117-123

dos, G., Bredoux, R., Uvy-lbledano, S., and Enouf, J. (1991) J. Bwl. Chem. 286, 14593-14596

20. De Jaegere, S., Wuytack, F., De Smedt, H., Van Den Bosch, L., and Casteels, R. (1993) Biochim. Biophys. Acta 1173,218%2194

21. Stewart, P. S., MacLennan, D. H., and Shamoo, A. E. (1976) J. Biol. Chem. 251, 712-719

22. Andersen, J.-P., Vilsen, B., Collins, J. H., and Jorgensen, P. L. (1986) J. Membr: Biol. 93, 85-92

23. Hack, N., Wilkinson, J. M., and Crawford, N. (1988) Biochem. J. 250,355-361 24. Lytton, J., and MacLennan, D. H. (1988) J. Biol. Chem. 263,15024-15031 25. Enouf, J., Bredoux, R., Papp, B., DjafTar, I., LomprB, A,"., Kieffer, N., Gayet,

O., Clemetson, K, Wuytack, F., and Rosa, J.-P. (1992) Biochem. J. 286, 26. Jorgensen, A. O., Arnold, W., Pepper, D. R., Kahl, S. D., Mandel, F., and

135-140

27. Zubrzycka-Gaarn, E., MacDonald, G., Phillips, L., Jorgensen, A. O., and Ma- Campbell, K. P. (1988) Cell. Motil. Cytoskeleton 9, 164-174

28. Papp, B., Corvazier, E., Magnier, C., Kovlcs, T., Bourdeau, N., Uvy-Toledano, cLennan, D. H. (1984) J. Bioenerg. Biomembr: 18,441462

S., Bredoux, R., Uvy, B., Poitevin, E, Lompr.5 A.-M., Wuytack, F., and Enouf, J. (1993) Biochem. J. , 295,685-690

Bwchem. J . 263,547452