Nucleotide Sequences of Avian Cardiac and Brain SR/ER Ca2+-ATPases and Functional Comparisons with Fast Twitch Ca2+-ATPase CALCIUM AFFINITIES AND INHIBITOR EFFECTS*

Two similar forms of the cardiac/slow Ca2+-ATPase (SERCA2a and SERCAQb), differing in sodium dodecyl sulfate-polyacrylamide gel electrophoresis mobility, are expressed in chicken heart and brain (Kaprielian, Z., Campbell, A. M., and Fambrough, D. M. (1989) Mol. Brain Res. 6, 55-60). In the current study, cDNAs encoding each form were cloned and sequenced. minus 38 49 amino acids homologues. SERCA2a sequence within lated Chicken genomic DNA sequence reveals within culture washed with 12 ml of 10 mM MOPS, pH 7.0,2 mM LaCL. A liquid scintilation counter was used to determine amounts of 45Ca2+ in each aliquot. For the cyclopiazonic acid and thapsigargin studies, the inhibitors were present during the 5-min incubation period. For the calcium concen- tration-dependent studies, free calcium concentration was varied as estimated from total CaClz and EGTA in solution (Fernandez-Belda et al., 1984). The data points in Figs. 6 and 7 represent averaged results obtained from two or three independent transfections and microsome preparations.

**TO whom correspondence should be addressed.
The nomenclature used here was adopted from Burk et al., (1989). 1989). The two subtypes are identical except that the four carboxyl-terminal amino acids of SERCA2a are replaced by 49 or 50 different residues in SERCA2b. The significance of these two forms, generated from alternate splicing of primary transcripts from the same gene, is unknown.
If the different carboxyl termini found in mammals have functionally significant roles, one might expect this to be evolutionarily conserved. Based on immunological data, our laboratory found that two slightly different forms of SERCA2 are expressed in the chicken heart and brain . One purpose of this study was to identify these different forms of chicken SERCA2 at the molecular level. Specifically, are the differences in the avian Ca2+-ATPase subtypes homologous to the alternately spliced products seen in mammals? If so, are there any functional differences between the two alternate forms of SERCA2 (e.g. inhibitor sensitivity or Ca2+ affinity)?
cDNA Library Construction and Screening-Total RNA was extracted from the heart and brain of one adult chicken in guanidine thiocyanate as described in Taormino and Fambrough (1990). The poly(A) RNA was converted to oligo (dT)-primed double-stranded cDNA, methylated, coupled with EcoRI linkers, ligated into lambdaZAP phage vector and packaged (Stratagene Cloning Systems). (However, based on examination of numerous clones encoding several different proteins, it has become apparent that the cDNA used to make this library had been incompletely methylated. Many clones terminate at internal EcoRI sites and some have unrelated sequences ligated adjacent to the clones of interest.) For screening the libraries, the coding region of rat stomach SERCA2a (Gunteski-Hamblin et al., 1988) was excised with PstI and isolated as described (Davis et al., 1986). The resulting probe was labeled with 32P by the method of Feinberg and Volgelstein (1983). Hybridization of probe to nitrocellulose filter lifts of the plated library was performed overnight in a solution of 120 mM Tris, pH 8,600 mM NaCl, 4 mM EDTA, and 50% formamide at 68 "C. Forty and 20 positive clones from the brain and heart cDNA libraries, respectively, were rescreened with the same rat probe. Five clones from each library were isolated by an in uiuo excision method involving the helper phage R408 (Stratagene Cloning Systems).
cDNA Sequencing and Analysis-All but the 107 5' most nucleotides (noncoding) of SERCA2 were sequenced on both strands by the dideoxynucleotide termination method (Sanger et al., 1977) with the U. S. Biochemical Corporation Sequenase kit. Some clones were sequenced after subcloning restriction fragments. Synthetic oligonucleotide sequencing primers were made on the 391 DNA Synthesizer (Applied Biosystems) and used to sequence other clones. By the procedure in Sambrook et al. (1989), nested deletions were produced with exonuclease 111 to sequence one clone. The GAP program of GCG (a software package from the University of Wisconsin) was used for computer comparisons of cDNA sequence similarity across species.
RNA Blot Analyses-Total RNA was isolated from adult and embryonic heart and brain tissues by the RNAzol B method (Cinna/ Biotecx Laboratories International Inc.). 30 or 45 pg of total RNA were loaded onto a 0.6% agarose gels containing formaldehyde. Probe A in Fig. 3A was derived from the 1.3-kb EcoRI fragment of clone B13' and labeled with 32P. The remaining probes were generated by a PCR (Perkin-Elmer Cetus)' procedure with primers which were designed to amplify specific regions of chicken SERCA2. The primers are as follows: probe B, 5' primer AAGAAAACAAAAGCAT (bases 3520-3535 as numbered in Fig. 2A), 3' primer GAAACAATCTGA-CACAA (reverse complement of bases 4001-4017); probe C, 5' primer GTAATCACTTCCTAAAC (4408-4424), 3' primer TACA-TAAGCTGTTATAG (reverse complement of bases and 4820-4836); and probe D, 5' primer CTGGCGTGTTA'MTGATGCAC (bases 5183-5203), 3' primer GAGGGATTTACAAACAATG (reverse complement of bases 5442-5460). Reaction conditions were 10 mM Tris-HCl, pH 8.3 (at 25 "C), 50 mM KCl, 1.5 mM MgCl, 0.001% gelatin, 2.5 units of Amplitaq DNA polymerase (Perkin Elmer-Cetus), with 0.07 mM non-radioactive deoxynucleotides, and 125 pCi of (u-~'Plabeled dATP and dCTP. 30 temperature cycles of 95, 50, and 72 "C for 1 min each were performed to produce each probe. The reaction products were then passed through two Sephadex G-50 spin columns to purify the radiolabeled probes. The 5' end-labeled probe E in Fig.  3 is an oligonucleotide 30 bases long with the sequence of AT-TACTCCAGTATTGCAGGTTCCAGGTAGT. This sequence is comprised of 15 nucleotides on either side of the alternate splice site, half of which are specific for the SERCA2a terminal encoding sequence. Gene Structure-The primers used to subclone the chicken genomic DNA that included the alternate splice site were 5' GGAATT-CATCTGGCTGGTGGAGC (an EcoRI linker plus bases 2788-2804) and 3' GGAATTCATATCACTAAAGTAG (an EcoRI linker plus reverse complement of bases 3101-3117). 1.5 pg of chicken genomic DNA were used as template in reaction conditions outlined above but without radiolabled nucleotides. The PCR product was digested with EcoRI and cloned into the vector pBluescript. Another pair of primers was designed to amplify the intron downstream of the SERCA2b unique sequence: 5' AAGAAAACAAAAGCAT (bases 3520-3535) and CAACCTCACATTTCTGC (reverse compliment to bases 4431-4447). When the latter pair of primers was used, no product was seen when genomic DNA was used as template. However, cDNA template yielded a band of the correct size. The inability to amplify this portion of the SERCA2 chicken gene is consistent with the human gene structure which has a 3-kb intron in this region (Lytton and Mac-Lennan, 1988).
Expression in Tissue Culture-Full-length cDNAs of SERCA2a and SERCA2b were constructed by ligating the appropriate fragments of clones B13, B14, and H14. Two "false-start" ATGs in the 5'-UT region were deleted by a PCR method. The 5'-oligo contained a KpnI site, a translation initiation consensus sequence (Kozak, 1989) and the first nine coding nucleotides (TGTGTGGTACCCCGAC-CATGGAGAACG). The other PCR primer was based on sequence down stream of the SpeI site at position 463. The PCR conditions were as described above with an extention time of 30 s. The resulting product was digested with SpeI and KpnI and cloned into pBluescript and sequenced. This was then ligated onto the 5' end of both SERCA2a and SERCA2b. SERCA2b was digested with Ssp1 and recloned into pBluescript in order to delete the sequence containing the SERCA2a carboxyl-terminal encoding nucleotides. The modified cDNAs of SERCA2a and SERCA2b were cloned into the KpnI site of the expression vector pcDL-SRa296 (Takebe et al., 1988). Chicken SERCAl (Karin et al., 1989) was cloned into the EcoRI site of pcDL-SRa296. COS-1 cells were transfected using DEAE dextran (Clarke ' The abbreviations used are: PCR, polymerase chain reaction; kb, kilobase; SDS-PAGE, sodium dodecyl sulfate-polyacrylamide gel electrophoresis; MOPS, 4-morpholinepropanesulfonic acid; 3"UT and 5'UT, 3'-and 5"untranslated; HEPES, 4-(2-hydroxyethyl)-l-piperazineethanesulfonic acid; [ethylenebis(oxyethylenenitrilo)]tetraacetic acid; TG, thapsigargin; CPA, cyclopiazonic acid; ER, endoplasmic reticulum; SR, sarcoplasmic reticulum; Clones beginning with the letter "B" are from the brain cDNA library while those with an "H" are derived from the heart library.
Microsome Preparation-As in Clarke, et al. (1990), twenty 150 X 25-mm plates of transfected COS-1 cells were washed twice with 10 ml of phosphate-buffered saline, harvested in 80 ml of 5 mM EDTA in phosphate-buffered saline, and washed with 40 ml of phosphatebuffered saline. The cells were resuspended in 16 ml of 10 mM MOPS, pH 7.0, 0.5 mM MgC12, 200 KIU/ml aprotinin, 1 mM phenylmethylsulfonyl fluoride and, after 10 min to allow for hypotonic swelling of cells, homogenized with Dounce homogenizer. The suspension was diluted with 0.5 M sucrose/6 mM 2-mercaptoethanol/40 p M CaClz/ 300 mM KC1/20 mM MOPS, pH 7.0, and centrifuged at 10,000 X g for 20 min. The supernatant was adjusted to 600 mM KC1 and centrifuged at 100,000 X g for 60 min. The pellet was resuspended with 50 mM MOPS, pH 7.0,10% sucrose and frozen in liquid nitrogen for later use.
Calcium Uptake Assay-5.0 ml of the reaction mixture (20 mM MOPS, pH 7.0, 60 mM KCl, 5 mM MgC12, 5 mM sodium oxalate, 0.2 mM EDTA, 2.5 mM ATP, 0.2 mM CaC1, (with 0.4 pCi/ml "Ca*+) and 10 pg/ml of microsomal protein) was equilibrated at 25 "C for 5 min. 1.0 ml aliquots were filtered (0.45-pm pore size; Millipore) and washed with 12 ml of 10 mM MOPS, pH 7.0,2 mM LaCL. A liquid scintilation counter was used to determine amounts of 45Ca2+ in each aliquot. For the cyclopiazonic acid and thapsigargin studies, the inhibitors were present during the 5-min incubation period. For the calcium concentration-dependent studies, free calcium concentration was varied as estimated from total CaClz and EGTA in solution (Fernandez-Belda et al., 1984). The data points in Figs. 6 and 7 represent averaged results obtained from two or three independent transfections and microsome preparations.

RESULTS
cDNA Sequence and Analysis-A cDNA probe encoding rat SERCA2a was used to screen chicken heart and brain cDNA libraries at high stringency (68 "C, 50% formamide). A number of overlapping cDNA clones from each library were excised in vivo (see "Materials and Methods") and sequenced to yield complete nucleotide sequences encoding chicken SERCA2a and SERCA2b. Clone B112 did not contain the complete coding sequence but it was of particular interest since it was the only clone isolated from the brain cDNA library which encoded the SERCA2a carboxyl terminus. No poly(A) tails were found in any of the clones, perhaps due to the presence of an EcoRI site between the polyadenylation signal and the poly(A) tail in combination with partial methylation of the cDNA during construction of both libraries (see Materials and Methods).
The nucleotide and deduced amino acid sequences of chicken SERCA2a are shown in Fig. lA. The amino acid sequence is 94% identical to mammalian homologues , Lytton and MacLennan, 1988, and Gunteski-Hamblin et al., 1988, Eggermont et al., 1989. Of the variant amino acids, conservative changes account for nearly half of the substitutions. In addition to the cDNA clones that encode SERCA2a, the avian cDNA homologue of mammalian SERCA2b was also sequenced. The nucleotide and deduced amino acid sequences of SERCASb are presented in Fig. lA

L V R I L L L A A C I S F V L A W F E E G E E T I T A F V E F F V I L L I L V A N A I V G V W Q E R 1 l 0 N A E N A I E A L K E Y E P E M G K V Y R Q D R K S V Q R I K A R D I V P G D I V E V A V G D K V P 160 A D I R I T S I K S T T L R V D Q S I L T G E S V S V I K H T D P V P D P R A V N Q D K K N M L F S 210 G T N I A A G K A M G V V I A T G V N T E I G K I R D E M V A T E Q E R T P L Q Q K L D E F G E Q L 260 S K V I S L I C I A V W I I N I G H F N D P V H G G S W I R G A I Y Y F K I A V A L A V A A I P E G 310
1081 1 1231 T 1381 3

M F V K G A P E G V I D R C T H V R V G N A K I F L S S G I K Q K I M S V I R E W G T G R D T L R C 560 -A L A T H D N P P R K E E M N L E D S S N F I N Y E T N L T F V G C V G M L D P P R I E V A S S I 610 1831
1981 3 2131 2281 2431  the various chicken cDNA clones, there are three base substitutions occurring in the coding region. These changes do not alter the primary structure of the protein and are probably caused by two different alleles being expressed by a heterozygous animal used for construction of the libraries. The changes are from C' ' " to A, TI'"' to C and CZ7"' to A. Previous publications of SERCA2 sequences (Lytton and MacLennan, 1988, Gunteski-Hamblin et al., 1988, and Eggermont et al., 1989 had not shown the correct relationship between SERCA2a and SERCA2h mRNA. The fact that the 3"UT region of SERCA2h message also contained the sequence which encodes the SERCA2a carboxyl terminus was not realized. But as shown in Fig. 1, A and R, the SERCA2a terminal encoding cDNA is downstream of the SERCA2b cDNA. In order to translate SERCA2a, the primary transcript must be spliced so that all of the SERCA2b unique sequence is excised. By removing the SERCA2h unique sequence, the encoding sequence of the four terminal amino acids of SERCA2a becomes contiguous with the bulk of the coding region, thus allowing SERCA2a translation. Therefore, primary transcripts of the SERCA2 gene contain the encoding sequences for both SERCA2a and SERCA2b. After processing the RNA, the 3"UT region of SERCA2b mRNA still contains the SERCA2a-terminal encoding sequence while SERCA2a mRNA has had its SERCABb unique portion excised.

L N V T Q W L M V L K I S L P V I L L D E T L K Y V A R N Y L E P G K D S V Q P A T K P C S L S
mRNA Processing-To delineate exact intronlexon boundaries surrounding the chicken alternate splice site within the gene, PCR was employed to amplify a portion of the gene. The deduced gene structure and RNA splicing pattern are shown in Fig. 2. Primers were designed to amplify hoth the alternate splice site and the intron upstream of the SERCA2h unique sequence. The resulting PCR product was cloned and sequenced. Between nucleotides G2"''9 and AZR" is a 119-base pair intron:' which begins with GT and ends in AG. An intron occurs at the homologous position in the human SERCA2 gene (Lytton and MacLennan, 1988). There is no intron at the alternate donor splice site used to generate SERCA2a mRNA.
A series of probes to different regions of the 3'-UT regions of SERCAZ were hybridized to RNA blots to determine which of three potential polyadenylation signals are used in mRNA processing. Probe A was the EcoRI fragment depicted in the diagram in Fig. 3. The resulting bands (Fig. 3, blot A ) show the relative amounts of heart and brain mRNA loaded in blots A through E. Probe R, which hybridizes to a SERCA2b specific portion of the sequence, does not hybridize to the heart RNA hut does detect high levels of SERCA2b mRNA isolated from brain tissue. Embryonic brain also contains mRNAs that include the sequence unique to SERCA2b; only SERCA2a mRNA was detected in embryonic heart (data not This autoradiograph is an R N A blot prohed with '.'l'-lnt)led prnbes. One 0.6% agarose gel was loaded with replicates of 45 p g of adult hrain and 80 pg of adult heart RNA as indicated. The approximate sizes of the prohes are 1 .2 kh ( A ) . .TOO (13). 400 (('1, 250 ( 1 ) ) . and 30 ( E l nucleotides. The molecular size markers are 4.G and 1.8 kt). The location of each prohe is indicated in the finlre helow with the three potential polyadenylation signals indicated hy crozsvs.
shown). The first consensus sequence for polyadenylation occurs at nucleotide 4048, within the 3"UT region unique to SERCA2h. Little if any SERCA2h mRNA uses the consensus sequence a t 4048 since the band detected with prohe H has an apparent size 1 kh greater than that predicted for messages terminating a t 4048. In RNA blots from hoth heart and hrain, bands are evident when prohe C was used hut not prohe D.
These data show that for hoth SERCA2a and SERCA2h mRNAs, the AATAAA at 5158 is used as the predominant polyadenylation signal rather than the consensus sequence at position 5473. It is interesting to note that the rarely used polyadenylation signal sequence at 5473 is not present in homologous mammalian cDNAs. Heart transcripts occasionally use the signal sequence at 5473 (or at some position further downstream) since one cardiac cDNA clone (H14) was found to contain the sequence downstream of the predominantly used polyadenylation signal at hase 5158. (A faint signal in the heart RNA lane was detected when prohe D WAS used hut was too faint to appear in the photograph.) There is no evidence that hrain messages ever use the polyadenylation signal a t nucleotide 5473. Prohe E, an oligonucleotide which spans the alternate splice site and is specific for SERCA2a mature mRNA, hybridizes to RNA of similar size in hoth brain and heart lanes. This verifies that hrain does express SERCA2a but a t a much lower level than SERCA2h. The brain SERCA2a mRNA was not detected in ponds A or (' probably hecause of the low expression level and smeared signal. In summary, hoth brain and heart transrrihe messages which predominantly use the polyadenylation simal at nucleotide 5158, though heart infrequently uses the polyadenylation signal at nucleotide 5473. SERCA2a mRNA was detectahle in hoth hrain and cardiac lanes while SERCA2h message was seen in hrain RNA only. Expression and Anal.vsi.7 of cDNA Cloncs in Tissuc Cuiture-In order to examine functional differences among Ca"' pumps, cDNAs encoding each SERCA:! suht-ype as well a chicken SERCAl were expressed in tissue culture (see "Material and Methods"). Full-length constructs encoding either SERCA2a. SERCAPb, or SERCA1 were transiently expressed in COS-1 cells. The transfected cells were fixed, permeahilized, and labeled with a chicken specific anti-SERCA'L or anti-SERCAl monoclonal antihodv and rhodamine-conjugated secondary antihodv. High levels of expression were of Avian Ca2+-ATPaqe.s obtained for all three avian Ca2+-ATPases. An immunofluorescent staining pattern indicative of the endoplasmic reticulum was observed. This is best seen at the thin edges of cells as shown in Fig. 4, A-C. These results show that SERCAl and the SERCA2 subtypes are capable of targeting to the appropriate organelle when transfected into non-muscle tissue cultured cells. Microsomes made from similarly transfected cells were analyzed by SDS-PAGE and immunoblots. Using protein blots from 6% polyacrylamide gel electrophoresis and probing with avian-specific monoclonal antibodies, it was possible to demonstrate clearly the expressed avian Ca2+-ATPases (Fig. 4 0 ) .
T o ensure that the ER localization was not merely due to accumulation of misfolded protein, microsomes of cells transfected with SERCAl, SERCA2a, or SERCA2b were assayed for their ability to sequester ' "Ca". Fig. 5 shows that all three pumps are functional. The apparent lower activity of SERCA2a is due to lower yields of SERCA2a protein/milligram of total microsomal protein (see Fig. 4 0 ) . Equal amounts of total microsomal protein were analyzed by immunoblots to quantify relative amounts of SERCA2a and SERCA2b. There is 1.7-fold less SERCA2a than SERCA2b in the respective microsomes (data not shown). Therefore, with a factor of 1.7 to correct for the lower expression of SERCA2a, all three pumps sequester about 1100 nmol of Ca2+/mg protein/h. This is an order of magnitude greater than the rate of There are two toxins reported to be SR/ER Ca'+ pump inhibitors. Both thapsigargin (TG, Thastrup el af., 1990) and cyclopiazonic acid (CPA, Seidler at al., 1989) are believed to act upon SERCA-type ATPases but not the plasma membrane calcium pumps. When calcium uptake was measured for SERCA2a, SERCA2b, and SERCAl over a range of inhibitor concentrations, no significant differences were detected in the sensitivity of the three isoforms to the inhibitors (Fig. 6).
Finally, the apparent Ca'+ affinity for each isoform was determined. Equal amounts of microsomes were incubated with varying free Ca" concentrations. There WAS no appreciable difference in the Ca" activation patterns of the Ca"-ATPases as shown in Fig. 7.

DISCUSSION
Previous work has shown that two suht-ypes of SERCA2 with different M , are expressed in the chicken . This paper demonstrates that the difference is due to alternate splicing at an intraexonic donor site in the primary transcript. SERCA2a can only be expressed when a splice site donor, which occurs within the exon coding for the carboxyl terminus of SERCAW, is used for RNA processing. Only SERCA2a was detected in heart, while both forms of SERCA2 were expressed in brain with SERCA2h being the predominant form. This means that the internal RNA splice site donor is used much less often in brain and that the carboxyl-terminal coding sequence of SERCA2a usually appears within the 3"UT region of SERCAZb mRNA. A similar, though more complex, splicing pattern of RNA has recently been reported for mammals (Plessers et al., 1991). Unlike mammalian SERCAZ mRNA expresssion, there are only two forms of avian SERCAZ mRNA. The only detected SERCAZb mRNA always contained within its 3'-UT region the SERCAZa terminal encoding nucleotides. Therefore, alternate splicing via an internal donor site appears to be the mechanism to produce alternate carboxyl termini in avian as well as mammalian SERCA2.
When SERCAZa and SERCAZb were expressed in COS-1 cells, the Ca*+-ATPases were targeted to the endoplasmic reticulum as evident by the immunofluorescent staining pattern (Fig. 4,A-C). This localization is not due to accumulation of misfolded protein since similarly transfected cells synthesized functional enzymes. In the photomicrographs, there is some punctate staining in addition to the reticular network. This could be due to incomplete fixation and vesicularization of the ER and/or capping of the Ca*+-ATPases within the ER. The antibody's epitope is glutaraldehyde and methanol sensitive so other fixation protocols were unsuccessful. The possibility of lateral mobility within the ER of SERCAZ proteins is under investigation.
We have compared the expression and activities of the three chicken isoforms in a number of ways. When analyzed by immunoblots (Fig. 4 0 ) , the bands in the SERCAZ lanes appear as broad bands. These data, in addition to some prelimonary data, suggest that the calcium pump might be a glycoprotein. Functionally, the three Ca2+-ATPases are very similar in their sensitivity to Ca2+ as an activator and to CPA and TG as inhibitors. Since SERCA2a and SERCA2b differ only at their carboxyl termini, it is not surprising that the ATPases are indistinguishable in their apparent Ca2+ affinities and inhibitor sensitivities. Although there is a 15% amino acid sequence difference between SERCAl and SERCAZ, the similar effects of TG and CPA suggest that neither inhibitor interacts with isoform-specific residues. By comparing primary sequences and pharmacological sensitivities of SERCAtype pumps from a variety of species, it may be possible to predict which regions interact with TG and CPA.
In order to understand Ca*+-ATPases more fully, it is helpful to compare primary sequences across a wide range of species. Chicken SERCA2a is 94% identical to its mammalian homologue while the carboxyl terminus of SERCA2b is also highly conserved. A series of mutagenesis studies has furthered our understanding of the structure/function relationship (e.g. Clarke et al., 1989aClarke et al., , 1989bMaruyama et al., 1989;Vilsen et al.;1989Clarke, et al., 1990). Of the residues shown by other laboratories to be required for the function of the Ca2+-ATPase, all are completely conserved in chicken SERCA2. There is no evidence which shows that the carboxyl terminus of a calcium pump is vital for function and yet diverse species have conserved, through millions of years, alternate SERCA2a and SERCA2b termini. It remains to be determined why there is a selective advantage for birds and mammals to retain multiple isoforms of the Ca2+-ATPase.