The Golgi PMR1 P-type ATPase of Caenorhabditis elegans. Identification of the gene and demonstration of calcium and manganese transport.

In recent years, it has been well established that the Ca(2+) concentration in the lumen of intracellular organelles is a key determinant of cell function. Despite the fact that essential functions of the Golgi apparatus depend on the Ca(2+) and Mn(2+) concentration in its lumen, little is known on the transport system responsible for ion accumulation. The Golgi ion pump PMR1 has been functionally studied only in yeast. In humans, mutations in the orthologous gene ATP2C1 cause Hailey-Hailey disease. We report here the identification of the PMR1 homologue in the model organism Caenorhabditis elegans and after ectopic expression the direct study of its ion transport in permeabilized COS-1 cells. The C. elegans genome is predicted to contain a single PMR1 orthologue on chromosome I. We found evidence for alternative splicing in the 5'-untranslated region, but no indication for the generation of different protein isoforms. C. elegans PMR1 overexpressed in COS-1 cells transports Ca(2+) and Mn(2+) with high affinity into the Golgi apparatus in a thapsigargin-insensitive manner. Part of the accumulated Ca(2+) can be released by inositol 1,4,5-trisphosphate, in agreement with the idea that the Golgi apparatus is an inositol 1,4,5-trisphosphate-sensitive Ca(2+) store.

In eukaryotic cells, cytosolic Ca 2ϩ acts as a second messenger in a large variety of cell functions. The increase of the cytosolic free Ca 2ϩ concentration in response to a stimulus results from the opening of Ca 2ϩ channels present in the plasma membrane and in the membranes of intracellular Ca 2ϩ stores, mainly the endoplasmic reticulum (ER). 1 It has more recently also become clear that the Ca 2ϩ stored in the lumen of many intracellular compartments not only serves as a reservoir of releasable Ca 2ϩ , but also plays a regulatory role in several important cell biological functions. Luminal Ca 2ϩ in the ER as well as in the Golgi or other components of the secretory pathway is required for the proper translation, translocation, folding, and processing of secreted proteins (for review, see Ref. 1). A sufficiently high level of intraorganellar Ca 2ϩ has been implicated also in intra-Golgi membrane transport (2), transport between the Golgi and the ER (3), and in endosome fusion (4). Besides these constitutive actions, luminal Ca 2ϩ controls movements of the Ca 2ϩ ion itself: its release from the intracellular stores, its permeability through the nuclear pore complexes, and capacitative Ca 2ϩ entry at the plasma membrane (for review, see Ref. 5).
The pivotal role of the ER as a Ca 2ϩ store is well established. There is a growing consensus that also the Golgi apparatus can function as an agonist-releasable Ca 2ϩ store (6), the importance of which may, however, greatly differ among the different cell types and ranges from relatively unimportant in Drosophila melanogaster S2 cells (7) to very important in renal LLC-PK 1 cells (8). The storage capacity for Ca 2ϩ in the Golgi is greatly increased by specific proteins like Calnuc/nucleobindin (9,10) and members of the CREC family of Ca 2ϩ -binding proteins (reviewed in Ref. 11). In contrast to the growing unanimity on the role of Golgi as a Ca 2ϩ store, there remains some confusion as to the type of Ca 2ϩ -accumulation system responsible to replenish the store. Ca 2ϩ pumps belonging to the class of P-type ion-motive ATPases have been proposed to take up this task: SERCA2 (6,10), the plasma-membrane Ca 2ϩ pump PMCA en route to the plasma membrane (12) and PMR1 (6,13). The SERCA pumps, which are encoded by three different genes in vertebrates (human gene names: ATP2A1-3), but apparently only by a single gene in invertebrates like C. elegans, are the best characterized members of this class and are responsible for the accumulation of Ca 2ϩ into the ER or into the sarcoplasmic reticulum. The PMCA Ca 2ϩ pumps (corresponding human gene names: ATP2B1-4) responsible for the extrusion of Ca 2ϩ out of the cell, also belong to the same class but are themselves not involved in Ca 2ϩ transport into the stores, with as a possible exception the above mentioned Golgi-based PMCA pumps which are on their way to the plasma membrane. The PMR1-type of Ca 2ϩ -transport ATPases was first identified in the yeast Saccharomyces cerevisiae (14) and localized to the Golgi or one of its subcompartments (15). Genes homologous to the S. cerevisiae PMR1 have been reported for a number of other fungi (see Ref. 16 and references therein). The PMR1 ion-motive ATPase supplies the secretory pathway with Ca 2ϩ and Mn 2ϩ ions required for glycosylation, sorting, and ERassociated protein degradation (17,18). A recent study has demonstrated capacitative Ca 2ϩ entry in S. cerevisiae, a mechanism that in higher eukaryotes is thought to be initiated by depletion of intracellular stores that are filled by the SERCA Ca 2ϩ pump (19). The process was stimulated in pmr1 mutants, indicating that in yeast capacitative Ca 2ϩ entry in combination with PMR1 activity supplies the secretory pathway with Ca 2ϩ . Yeast pmr1 mutants do not grow on a medium containing submicromolar concentrations of Ca 2ϩ and show defects in the maturation of secretory proteins which are suppressed by supplying millimolar Ca 2ϩ or micromolar Mn 2ϩ to the growth medium. Mutations in PMR1 were also reported to rescue yeast mutants, which as a result of the lack of superoxide dismutase, show impaired growth in aerobic conditions (20). This effect is ascribed to increased cytosolic levels of Mn 2ϩ resulting from a lack of accumulation of the ion in the Golgi compartment. Mn 2ϩ is known for its capacity to scavenge superoxide ions. Mandal et al. (21) pointed recently to the critical role of transmembrane segment M6 in yeast PMR1 for defining the cation-binding sites in general and in particular of residue Gln 783 in this segment for the Mn 2ϩ selectivity.
Relatively less is known on the PMR1 homologues in animal cells. The cDNA of the putative rat form of the yeast PMR1 was already cloned in 1992 with a SERCA-derived probe (13), but the authors failed in their efforts to show that the corresponding protein, upon its expression in COS cells, was able to catalyze the uptake of Ca 2ϩ into vesicles consisting of fragmented membranes. In the meanwhile homologous cDNAs or genes were reported for D. melanogaster, 2 Bos taurus, 3 and C. elegans. But again until now no direct indication that any of these was involved in Ca 2ϩ uptake has been provided. However, indirect evidence comes from two recent reports which show that Hailey-Hailey disease (MIM 16960), which is manifested by the impaired intercellular adhesion of epidermal keratinocytes, results from mutations in one of the alleles of a human orthologue of yeast PMR1, the ATP2C1 gene (22,23). The symptoms of Hailey-Hailey disease strongly resemble those of Darier-White disease (MIM 124200) which is due to a mutation in one of the alleles of the SERCA2 gene ATP2A2. This, together with the observation that expression of the mammalian SERCA1a prevented the lethality of the pmr1-pmc1 double mutations in yeast (24), strongly suggests that the human PMR1 pump can act as a Ca 2ϩ pump.
We now show for the first time that an animal PMR1 homologue can transport Ca 2ϩ or Mn 2ϩ into the Golgi apparatus of COS-1 cells with high affinity and in a thapsigargin-insensitive manner. The accumulated Ca 2ϩ can be released by IP 3 , in line with the view that the Golgi apparatus is an IP 3 -sensitive Ca 2ϩ store.

EXPERIMENTAL PROCEDURES
Materials-[␥-32 P]ATP and 45 Ca were obtained from Amersham Pharmacia Biotech (Roosendaal, The Netherlands), 54 Mn from PerkinElmer Life Sciences (Boston, MA). Sequencing was done with the AutoRead 200 Sequencing Kit from Amersham Pharmacia Biotech. COS-1 cells were transiently transfected with FuGENE 6 transfection reagent (Roche Molecular Diagnostics, Brussels, Belgium) according to the manufacturer's instructions. Fluorescein isothiocyanate-labeled goat anti-rabbit antibodies were obtained from Sigma (Sigma/Aldrich NV, Bornem, Belgium) and horseradish peroxidase-labeled swine antirabbit antibodies from Dako (Dako A/S, Glostrup, Denmark). The fulllength rabbit SERCA1a clone was kindly provided by J. P. Andersen and B. Vilsen (University of Aarhus, Denmark).
Constructs for Expression-The EST clones yk218a11 and yk334d5 were obtained from Yuji Kohara's database, and sequenced. Both clones contained the complete open reading frame of C. elegans PMR1. It should, however, be remarked that the open reading frame is 63 bp smaller than predicted in the database annotation (see "Results"). Clone yk334d5 was used to make an expression construct. The 3.2-kb PMR1-encoding EcoRI/XhoI fragment of yk334d5 was ligated into the dephosphorylated pcDNA3 expression vector (Invitrogen Co., British Biotechnology Products Ltd., Abingdon, United Kingdom) cut with the same restriction enzymes. The pig SERCA2a expression vector has been described by Verboomen et al. (25). The full-length rabbit SERCA1a was cloned in the pMT2 expression vector (26).
Cell Culture and DNA Transfection-For microsome preparation, COS-1 cells were seeded in 100-mm culture dishes at a density of 2.5 ϫ 10 6 cells per plate. For immunocytochemistry 2.0 ϫ 10 4 cells were seeded on gelatin (1%)-coated coverslips. For 45 Ca 2ϩ or 54 Mn 2ϩ fluxes 3.0 ϫ 10 4 cells were seeded in gelatin-coated 12-well plates. For microsome preparation and immunocytochemistry, transfection was performed the day after seeding. For isotope fluxes the period between seeding and transfection was extended to 5 days to allow better attachment of the cells to the plates. After transfection, the cells were incubated for 60 h at 37°C and 5% CO 2 .
Preparation of Antiserum to PMR1 Protein-The immunogen was a recombinant protein corresponding to the putative large cytoplasmic loop between transmembrane segments 4 and 5 of C. elegans PMR1. The protein was expressed in Escherichia coli using the QIAexpress Type IV System (Qiagen, Hilden, Germany). In a first step the cDNA corresponding to the loop region was amplified by PCR with primers PMR1CYTF (5Ј-CAAGCATGCCGTGAAGAAGATGCCAGCAG-3Ј) and PMR1CYTR (5Ј-CTAGTCGACTTTTCCCTCCTCAATCGCC-3Ј) containing, respectively, a SphI and SalI restriction site at their 5Ј end. The PMR1CYTF primer corresponds to nucleotides 26812-26831 of cosmid CECC4 (accession number Z81490) and PMR1CYTR primer to the inverse complement of nucleotides 136 -154 of cosmid CEZK256 (accession number Z82088). PCR was carried out for 20 cycles using the Expand Long Template PCR System from Roche Molecular Diagnostics. Each cycle consisted of 10 s of denaturation at 94°C, 30 s of annealing at 55°C, and 2 min of elongation at 68°C. The SphI/SalI-cut PCR fragment was ligated in the SphI/SalI sites of the pQE-31 bacterial expression vector (Qiagen), containing a 6xHis tag coding sequence 5Ј to the cloning region. The expression of the recombinant protein in E. coli M15 cells was induced by 1 mM isopropyl-1-thio-␤-D-galactopyranoside and purification of the 6xHis-tagged protein was achieved by a Ni-NTA agarose-based method as described in the manufacturer's instructions. The recombinant protein migrated with an apparent molecular mass of about 42 kDa in 12% SDS-polyacrylamide gels and reacted with the monoclonal anti-polyhistidine antibody (clone HIS-1, from Sigma, dilution 1:1000) on Western blots. In a last step the recombinant cytoplasmic loop was concentrated by Centricon TM Plus-20 centrifugal filtering (Millipore, Bedford, MA).
Rabbits were immunized with 0.1 mg of recombinant protein in 0.5 ml of phosphate-buffered saline emulsified with 0.5 ml of complete Freund's adjuvant. Booster injections of the same immunogen with incomplete Freund's adjuvant were given at 4-week intervals. Preimmune serum and serum obtained after 4 boosters were used. The antiserum is designated as Celpmrloop.
Membrane Preparations and Immunoblotting Analysis-Microsomes were isolated from COS-1 cells as described by Verboomen et al. (25). Membranes from C. elegans were isolated as described by Baylis et al. (28). Protein concentrations were determined by the bicinchoninic acid method (Pierce, Rockford, IL). Denaturing gel electrophoresis and Western blotting were done as described earlier (29). 45 Ca 2ϩ / 54 Mn 2ϩ Fluxes-COS-1 cells were grown on 12-well plates. Loading with 45 Ca 2ϩ and efflux were done essentially as described earlier (30). Cells were treated for 10 min with 20 g/ml saponin at 25°C and loaded for the indicated lengths of time in 120 mM KCl, 30 mM imidazole-HCl (pH 6.8), 5 mM ATP, 10 mM NaN 3 , 0.44 mM (for Ca 2ϩ fluxes) or 0.1 mM (for Mn 2ϩ fluxes) EGTA. MgCl 2 , CaCl 2 , and MnCl 2 were added to obtain a calculated free Mg 2ϩ concentration of 0.5 mM and the indicated concentrations of free Ca 2ϩ and Mn 2ϩ . Free concentrations of Ca 2ϩ and Mn 2ϩ were calculated based on the stability constants for EGTA and ATP given by Fabiato and Fabiato (31) (for Ca 2ϩ ) and by Martell and Smith (32) (for Mn 2ϩ ). Thapsigargin (2 M) was added if inhibition of pumping by the SERCA Ca 2ϩ pump was needed. Efflux was performed in 120 mM KCl, 30 mM imidazole (pH 6.8), and 1 mM EGTA. All experiments on Mn 2ϩ transporting activity or Mn 2ϩ effects on Ca 2ϩ uptake activity were performed with Chelex 100 (Bio-Rad, Eke, Belgium)-treated solutions. To test the sensitivity of the phosphoprotein to hydroxylamine, the samples were additionally incubated for 20 min at room temperature in 0.2 M acetic acid-NaOH with or without 0.2 M hydroxylamine. The samples were then dissolved in a modified SDS loading buffer and subjected to SDS-gel electrophoresis in acid gels as described by Sarkadi et al. (33). The gels were dried between gel drying sheets (Promega) and exposed to screens for quantification of the radioactive bands on a Storm840 scanner in combination with the ImageQuant TM software (Molecular Dynamics, Sunnyvale, CA).
Immunocytochemistry-Cells grown on coverslips were fixed in 4% paraformaldehyde in phosphate-buffered saline for 10 min at room temperature, washed in phosphate-buffered saline, and permeabilized in 0.2% Triton X-100. Primary and secondary antibodies were diluted in phosphate-buffered saline containing 3% bovine serum albumin at the indicated dilutions.

RESULTS
Gene Structure of C. elegans PMR1 and Alternative Transcript Processing-The C. elegans genome is predicted to contain a single PMR1 orthologue on chromosome I, which is annotated ZK256.1 by the C. elegans sequencing consortium (34). The genomic sequence can be found on the two overlapping cosmids CECC4 (accession number Z81490) and CEZK256 (accession number Z82088). Two EST clones from Yuji Kohara's database, yk218a11 and yk334d5, representing cDNA clones of the C. elegans PMR1 gene, were completely sequenced. The deduced C. elegans PMR1 protein would contain 901 amino acids instead of the 922 predicted in the database annotation (protein identification number CAB04015.1). This is because a sequence of 63 bp is wrongly assigned by the Genefinder program to the putative exon 8 (303 bp). We found this sequence to be absent from both cDNA clones (see Fig. 1A) and hence the actual protein should correspondingly be 21 amino acids shorter. Furthermore, it was clear from clone yk334d5 that the 5Ј UTR of C. elegans PMR1 is at least 257 bp long and contains, besides exon 1, two extra exons, comprising only untranslated sequences and designated exon Ϫ1 and exon Ϫ3 for reasons discussed below (Fig. 1B, second line). Exon 1 (399 bp) starts at position Ϫ87 relative to the ATG start codon, which corresponds to nucleotide 23653 on cosmid CECC4. Exon Ϫ1 (105 bp) and exon Ϫ3 (65 bp) correspond, respectively, to nucleotides 23111-23215 and 15957-16021. This implies that exon Ϫ3 is located more than 7 kb upstream from exon Ϫ1. EST clone yk218a11 does obviously not contain the entire 5Ј UTR, as only the last part of exon 1 is represented in this clone.
A comparison of the exon/intron layout of CePMR1 to that of the D. melanogaster (accession number AC014929) and human orthologues (gene ATP2C1, Refs. 22 and 23) shows that the human gene contains the most elaborate exon/intron layout, D. melanogaster the simplest, whereas C. elegans takes an intermediate position (Fig. 1B). Apparently C. elegans and to a larger degree D. melanogaster have lost most of the introns during evolution. The worm and human have in total 5 conserved exon boundaries, the worm and the fly only one. The localization of the exon 5/6 boundary in CePMR1 corresponds to the exon 1/2 boundary in D. melanogaster and to the exon 18/19 junction in the human gene. There are four extra conserved exon/intron positions between C. elegans and the human ATP2C1 gene (asterisk in Fig. 1B) while there is no other exon/intron junction at homologous positions between the C. elegans and D. melanogaster PMR1 genes. One extra exon boundary is conserved between exons 2 and 3 in D. melanogaster and between exons 21 and 22 in the human ATP2C1 gene. Among the exon borders of the PMR1 family, only one (exon 1/2 in CePMR1, exon 5/6 in the human ATP2C1) is conserved in the mammalian SERCA genes (exon 4/5 in human ATP2A1-3). Compared with the mammalian PMCA family, none of the exon/intron borders is conserved.
Because in C. elegans the majority of the mRNAs are transspliced, we have investigated whether this is also the case for transcripts of CePMR1 in the worm. 5Ј Rapid amplification of cDNA ends experiments were performed on whole worm RNA. PCR reactions with primer pairs SL1/CePMR1Rev and SL2/ CePMR1Rev resulted in the amplification of an 824-and 818-bp fragment, respectively. The 818-bp PCR product corresponded to a cDNA in which splice leader SL2 was transspliced to exon Ϫ3 and exon Ϫ3 was in turn spliced to exon Ϫ1, i.e. an exon layout as found in EST clone yk334d5. By sequencing the SL1/CePMR1Rev fragment of 824 bp, we found that SL1 was spliced to a novel exon (Ϫ2) in the 5Ј UTR of the gene, which corresponds to nucleotides 22950 -23020 on cosmid CECC4 (Fig. 1B, upper line). Thus it appears that the C. elegans transcripts are both alternatively spliced and transspliced at their 5Ј end. Both types of splicing are coupled: a mRNA transcript containing exon Ϫ1 and exon Ϫ2 is transspliced to SL1, a mRNA containing exon Ϫ1 and exon Ϫ3 is trans-spliced to SL2.
We also performed 3Ј rapid amplification of cDNA ends with primer pair CePMR1For/468 to check for the possibility of 3Ј alternative processing at the 3Ј end of the gene's transcript. PCR amplification gave a product of about 600 bp. By subcloning and sequencing of several individual clones, it became clear that the gel band actually consisted of a mixture of three products with small differences in length and corresponding to three polyadenylation isoforms. The isoforms are the result of the use of three polyadenylation sites (pA 1 , pA 2 , and pA 3 ) in the 3Ј UTR of the gene (Fig. 1B). Both yk218a11 and yk334d5 apparently used pA 2 . However, no indication was found for 3Ј alternative splicing, neither in our 3Ј rapid amplification of cDNA ends experiments nor by database searching.
In summary, the C. elegans PMR1 gene consists of 12 exons (Ϫ3 to 9) of moderate length and spans a region of more than 19 kb of genomic DNA. The transcripts are both alternatively and trans-spliced at their 5Ј end. At their 3Ј end, however, no alternative splicing could be documented but alternative polyadenylation occurs.
The C. elegans PMR1 Protein- Fig. 2 shows the predicted amino acid sequence of C. elegans PMR1 and its major domains together with an alignment with the corresponding sequences of three distant species, Homo sapiens, D. melanogaster, and S. cerevisiae. In the sequences of all species 10 hydrophobic segments can be identified, which like in the SERCA Ca 2ϩ pumps, presumably form the transmembrane domain. The highest degree of sequence similarity occurs around the phos-phorylation site, in regions demonstrated in SERCA to contribute to the ATP-binding site or to form structurally important loops, and in transmembrane segments M4, M6, and M8, which have been documented in SERCA to form the binding sites for Ca 2ϩ (35,36). The overall amino acid sequence is 37% identical to rat SERCA2a. The percentage identity with the PMR1 sequences of human, D. melanogaster, and S. cerevisiae is, respectively, 59, 57, and 49%. The amino acids that form the binding site for one of the transported Ca 2ϩ ions in SERCA, more specifically the site II, are conserved in the PMR1 homologues in all species. The amino acids belonging to site I are not conserved in the PMR1 protein. In S. cerevisiae, Gln 783 has been demonstrated to define the Mn 2ϩ selectivity of the PMR1 ion pump (21). Also this residue is conserved in all species.
The N-terminal region upstream of the first transmembrane domain of C. elegans PMR1 has the same length as in the human PMR1 sequence reported by Sudbrak et al. (23), whereas the human sequence predicted by Hu et al. (22) is 16 residues longer. The length of the C-terminal part is more similar to the product of the human splice variant ATP2C1a than to the shorter ATP2C1b described by Hu et al. (22). As mentioned above, there is no evidence for the generation of C-terminal PMR1 protein variants in C. elegans. The C-terminal sequence does not contain an eleventh hydrophobic region as in the SERCA2b splice variant. The EF hand-like domain near the N terminus of the S. cerevisiae sequence has been shown to play a role in modulating ion transport (37). Because its primary structure is poorly conserved, it remains to be demonstrated whether a similar function occurs in other species.
Characterization and Functional Analysis of C. elegans PMR1 Protein-The Celpmrloop antibody, raised against the large cytosolic loop between transmembrane segments 4 and 5 of PMR1 of C. elegans, clearly demonstrated the expression of the protein in COS-1 cells transfected with the corresponding cDNA, both by Western blot analysis and immunocytochemistry (Fig. 3). On Western blots, the immunoreactive band migrated slightly below the predicted theoretical M r value of 98,505. A strong immunoreaction was also seen on blots of fragmented membranes prepared from whole worms. Hence, it is clear that our antibody is able to recognize the PMR1 protein both in C. elegans and after its ectopic expression in COS-1 cells. The vertebrate PMR1 homologue found in untransfected COS-1 cells appears not to react with the antiserum as shown by the controls of untransfected COS-1 cells. Furthermore, immunocytochemistry of PMR1-overexpressing and control cells reveals the correct targeting of PMR1 to the Golgi compartment of the COS-1 cells (Fig. 3B). In conclusion, by using our polyclonal anti-PMR1 antiserum it became clear that PMR1 is expressed in the worms, that it can be overexpressed in COS-1 cells and that it contains all the information needed to target the PMR1 protein to the Golgi membranes.
Because the fraction of Golgi-derived membranes in microsomes of COS-1 cells is relatively small compared with that of ER, we could not rely on conventional techniques used to measure Ca 2ϩ transport, like those for the ER-based SERCA transport ATPases (25). Instead we took advantage of the 45 Ca 2ϩflux system utilizing detergent-permeabilized cells (30).
In a first series of experiments we tested PMR1 of C. elegans for its ability to transport Ca 2ϩ . Control COS-1 cells, cells overexpressing rabbit SERCA1a (as a positive control for Ca 2ϩ pumping), and cells overexpressing PMR1 were permeabilized with saponin in a medium mimicking a cytosolic composition and loaded with 45 Ca 2ϩ for 45 min in the presence of NaN 3 to prevent mitochondrial Ca 2ϩ uptake. Ca 2ϩ transport via SERCA-type Ca 2ϩ -ATPases was determined by comparing the uptake in the presence and absence of 2 M thapsigargin. The loading of the cells was followed by an efflux for 20 min in a Ca 2ϩ -free medium. IP 3 (10 M) was administered after 10 min of efflux. Fig. 4A shows that control and SERCA1a-transfected cells exhibit a Ca 2ϩ uptake, which is blocked by 2 M thapsigargin. In control COS-1 cells and in PMR1-expressing cells, the difference in Ca 2ϩ content with thapsigargin and without thapsigargin represents the Ca 2ϩ pump activity of the endogenous SERCA2b. In SERCA1a-transfected cells, it represents that of endogenous SERCA2b plus overexpressed SERCA1a. In contrast, COS-1 cells overexpressing PMR1 show an additional Ca 2ϩ pump activity even in the presence of thapsigargin. This suggests that PMR1 is a Ca 2ϩ -transporting protein residing in the Golgi and that it is insensitive to thapsigargin. Fig. 4A also shows that Golgi membranes contain IP 3 receptors, since 10 M IP 3 induced a more rapid decrease in the store Ca 2ϩ content. Fig. 4B shows the time course of Ca 2ϩ uptake by PMR1expressing cells. Transfected cells were loaded with 45 Ca 2ϩ for different time intervals in the presence of thapsigargin to block Ca 2ϩ uptake by endogenous SERCA2b. Ca 2ϩ pump activity reaches almost a plateau after 20 min of Ca 2ϩ loading. Subsequent experiments were performed after 10 min of loading with Ca 2ϩ . Fig. 4C shows the Ca 2ϩ dependence of Ca 2ϩ uptake by PMR1. The Ca 2ϩ concentration needed for half-maximal activation was 0.25 M Ca 2ϩ .
In a second series of experiments we explored the possibility of PMR1 of C. elegans to function as a Mn 2ϩ pump, since previous reports based on the activation of ATP hydrolysis by Mn 2ϩ and on the inhibition of Ca 2ϩ transport by Mn 2ϩ suggested that yeast PMR1 could act as a Mn 2ϩ pump. Control cells and PMR1-expressing cells were loaded with radioactive Mn 2ϩ for 10 min in the presence of thapsigargin. The efflux was followed for 10 min (Fig. 5A). Fig. 5, A and C, provide direct evidence that PMR1 can indeed act as a Mn 2ϩ -transporting protein. PMR1-overexpressing cells show an enhanced uptake of Mn 2ϩ . The accumulated Mn 2ϩ was released by the ionophore A23187 (10 M), demonstrating that the Mn 2ϩ has been transported into a membrane-delineated compartment. However, the addition of IP 3 did not have a significant effect on the rate of efflux. The Mn 2ϩ uptake was inhibited by Ca 2ϩ (Fig. 5C), and conversely the Ca 2ϩ uptake by PMR1 was inhibited by Mn 2ϩ (Fig. 5B). It is clear from Fig. 5B that at higher Ca 2ϩ concentrations more Mn 2ϩ was needed to inhibit the transport of Ca 2ϩ . Half-maximal inhibition was observed at 1, 0.5, and 0.25 M Mn 2ϩ for loading at, respectively, 1.0, 0.32, and 0.1 M Ca 2ϩ .
Formation of the Phosphoenzyme Intermediate-A determining characteristic of all P-type ion-transport ATPases is the transient transfer of the ␥-phosphate of ATP to the protein, forming a covalent bond with the carboxyl group of a conserved aspartic acid residue in the large cytosolic domain. The radioactively labeled phosphoprotein can be preserved during SDS-gel electrophoresis by quenching the reaction in acid and maintaining acid conditions throughout electrophoresis. Fig. 6 shows the radioactively labeled phosphointermediate in an SDS gel of microsomes from COS-1 cells overexpressing PMR1. The labeling was completely removed by treatment with hydroxylamine, demonstrating that the phosphate was bound to a carboxyl and not to a hydroxyl group (data not shown). As for the protein detected on Western blots, the phosphoprotein migrated slightly faster relative to the markers than expected from the predicted M r of 98,505. The anomalous migration is probably due to the gel system because a similar shift is also observed for the SERCA2a Ca 2ϩ -transport ATPase, which has a predicted M r of 109,720 (Fig. 6A). The phosphointermediate formation was stimulated by Ca 2ϩ and Mn 2ϩ . The maximum levels were observed below 1 M for both Ca 2ϩ and Mn 2ϩ and these maximum levels were not significantly different (data not shown). There remains a small residual amount of phosphoprotein also in the presence of EGTA without added Ca 2ϩ or Mn 2ϩ . At present we do not have a straightforward explanation for this background labeling. Possibly, a small fraction of the transport-protein molecules in the COS-1 cell membranes is able to reach a conformational state that allows phosphorylation without occupation of the transport sites. The Ca 2ϩ -or Mn 2ϩ -dependent phosphoprotein formation was strongly inhibited by 50 M La 3ϩ (Fig. 6B). The phosphorylation experiments thus confirm the high affinity of the C. elegans PMR1 transporter for both Ca 2ϩ and Mn 2ϩ . DISCUSSION The analysis of the C. elegans genome has resulted in the previous identification of several members of the superfamily of P-type Ca 2ϩ -transport ATPases. A single gene encoding a member of the SERCA-type subfamily is found on chromosome III (cosmid K9D11) and three genes (mca1-3) encoding members of the PMCA subfamily reside on chromosome IV (38). In this work, another gene (designated CePMR1) is identified encoding a P-type Ca 2ϩ -transport ATPase that is located on chromosome I. The conservation of some intron positions, the overall sequence similarity of the encoded protein, and the conservation of major domains and critical motifs of the pri- mary sequence unequivocally place it in the PMR1 subfamily of P-type Ca 2ϩ -transport ATPases. Besides the CePMR1 genomic sequence, we have in this work also characterized its transcripts and protein product.
With respect to the number of exons, the CePMR1 gene takes an intermediate position between the human and D. melanogaster orthologues. Only some of the exon/intron borders are conserved between these different species. All splice sites follow the GT . . . AG rule. Surprisingly, with the exception of the intron between exon 2 and 3, all introns in this gene are relatively long considering the fact that most of the introns in C. elegans genes have a length of only about 50 nucleotides (39). A particularly long intron (ϳ7 kb) is that between exons Ϫ3 and Ϫ2, both located in the 5Ј UTR. Interestingly, we observed the possibility of alternative splicing in the 5Ј UTR, which was coupled to trans-splicing to SL1 or SL2. The meaning of such an alternative trans-splicing which only affects the 5Ј UTR remains unknown. It should be noted that for the human PMR1 orthologue (gene ATP2C1), 5Ј-end alternative splicing has also been suggested but here it affects the open reading frame (23), whereas in C. elegans it does not result in the formation of distinct protein isoforms. At the 3Ј end of the CePMR1 gene three polyadenylation sites were predicted and also experimentally detected by PCR analysis. However, there was neither any predicted nor any experimental indication for alternative splicing. Both for the corresponding human gene and for the rat gene alternative splicing and different protein tails have been suggested. However, the alternative splice sites in human and rat appear not to be conserved.
At the protein level, the major domains described in other P-type transport ATPases can be recognized in C. elegans PMR1. Also sequence motifs demonstrated to be critical for function in other P-type transport ATPases are conserved in the C. elegans sequence (Fig. 2). On the basis of these comparisons, it can be firmly concluded that the coding sequence identified in the C. elegans genome and whose protein product has been investigated in this study is a member of the family of PMR1 ion-transport ATPases. This conclusion is further substantiated by the transport and phosphorylation studies on the protein expressed in COS-1 cells.
Functional data on the PMR1 transporter are up till now available only for the yeast S. cerevisae (21,37,40). In the present work the first characterization of the ion transporting activity of an animal PMR1 enzyme is presented. The C. elegans PMR1 protein overexpressed in COS-1 cells showed a predominantly Golgi-like distribution as shown by immunocytochemistry. ATP-dependent uptake of 45 Ca and 54 Mn was demonstrated in cells permeabilized with saponin. Cells overexpressing PMR1 accumulated more Ca 2ϩ and Mn 2ϩ than control cells, and this additional uptake was not diminished by the SERCA-specific inhibitor thapsigargin, demonstrating that PMR1 is not only able to transport Ca 2ϩ but also Mn 2ϩ . Part of this thapsigargin-insensitive Ca 2ϩ pool was released by IP 3 , which is compatible with existing evidence that the Golgi apparatus is an IP 3 -sensitive Ca 2ϩ store (6). The rate of 45 Ca 2ϩ uptake as determined in our experimental system was halfmaximal at 0.25 M Ca 2ϩ , which is slightly higher than the value determined from the ATPase activity of the purified PMR1 protein of S. cerevisiae (21). The 45 Ca 2ϩ uptake was progressively inhibited by increasing concentrations of Mn 2ϩ . Half-maximal inhibition occurred at a Mn 2ϩ concentration that is about the same as the half-maximal Ca 2ϩ concentration for stimulating the uptake, indicating that the affinity of the C. elegans PMR1 transport ATPase for Ca 2ϩ and Mn 2ϩ is approximately the same. A similar competition between Ca 2ϩ and Mn 2ϩ has been observed for the ATPase activity of S. cerevisiae PMR1, but the yeast showed a somewhat higher affinity for Mn 2ϩ than for Ca 2ϩ (40). Conversely, the uptake of Mn 2ϩ was inhibited by Ca 2ϩ , further confirming the competition between Ca 2ϩ and Mn 2ϩ for binding to the transport sites.
The phosphorylation experiments further validated the conclusion that the high-affinity transport sites can be activated either by Ca 2ϩ or Mn 2ϩ . Phosphoprotein formation was maximal at submicromolar concentrations of Ca 2ϩ or Mn 2ϩ , confirming the high affinity for both cations derived from the competition between Ca 2ϩ and Mn 2ϩ for transport. The phosphoprotein formation was strongly inhibited by La 3ϩ . This effect is in the same direction but more pronounced than that observed with the SERCA-type Ca 2ϩ pumps, whereas the phosphoprotein level of PMCA-type Ca 2ϩ -transport ATPases is increased by La 3ϩ (41).
In conclusion, we have identified and characterized the C. elegans homologue of the PMR1 ion-transport ATPase previously characterized only in yeast. We have shown that the PMR1 protein overexpressed in COS-1 cells shows a predominantly Golgi-like distribution and that its ion transport activity can be measured following permeabilization of the plasma membrane. These experiments directly demonstrate that C. elegans PMR1 is able to accumulate Ca 2ϩ and Mn 2ϩ with high affinity into the Golgi membranes. Part of the accumulated Ca 2ϩ can be released by IP 3 , confirming the observation of Pinton et al. (6) that the Golgi apparatus is an IP 3 -sensitive Ca 2ϩ store.