The Secretory Pathway Ca2+/Mn2+-ATPase 2 Is a Golgi-localized Pump with High Affinity for Ca2+ Ions*

Accumulation of Ca2+ into the Golgi apparatus is mediated by sarco(endo)plasmic reticulum Ca2+-ATPases (SERCAs) and by secretory pathway Ca2+-ATPases (SPCAs). Mammals and birds express in addition to the housekeeping SPCA1 (human gene name ATP2C1, cytogenetic position 3q22.1) a homologous SPCA2 isoform (human gene name ATP2C2, cytogenetic position 16q24.1). We show here that both genes present an identical exon/intron layout. We confirmed that hSPCA2 has the ability to transport Ca2+, demonstrated its Mn2+-transporting activity, showed its Ca2+- and Mn2+-dependent phosphoprotein intermediate formation, and documented the insensitivity of these functional activities to thapsigargin inhibition. The mRNA encoding hSPCA2 showed a limited tissue expression pattern mainly confined to the gastrointestinal and respiratory tract, prostate, thyroid, salivary, and mammary glands. Immunocytochemical localization in human colon sections presented a typical apical juxtanuclear Golgi-like staining. The expression in COS-1 cells allowed the direct demonstration of 45Ca2+ (K0.5 = 0.27 μm) or 54Mn2+ transport into an A23187-releasable compartment.

Accumulation of Ca 2؉ into the Golgi apparatus is mediated by sarco(endo)plasmic reticulum Ca 2؉ -ATPases (SERCAs) and by secretory pathway Ca 2؉ -ATPases (SP-CAs). Mammals and birds express in addition to the housekeeping SPCA1 (human gene name ATP2C1, cytogenetic position 3q22.1) a homologous SPCA2 isoform (human gene name ATP2C2, cytogenetic position 16q24.1). We show here that both genes present an identical exon/intron layout. We confirmed that hSPCA2 has the ability to transport Ca 2؉ , demonstrated its Mn 2؉transporting activity, showed its Ca 2؉ -and Mn 2؉dependent phosphoprotein intermediate formation, and documented the insensitivity of these functional activities to thapsigargin inhibition. The mRNA encoding hSPCA2 showed a limited tissue expression pattern mainly confined to the gastrointestinal and respiratory tract, prostate, thyroid, salivary, and mammary glands. Immunocytochemical localization in human colon sections presented a typical apical juxtanuclear Golgi-like staining. The expression in COS-1 cells allowed the direct demonstration of 45 Ca 2؉ (K 0.5 ‫؍‬ 0.27 M) or 54 Mn 2؉ transport into an A23187-releasable compartment.
In vertebrates, three families of P-type Ca 2ϩ -ATPases control intracellular Ca 2ϩ homeostasis: PMCAs, 1 SERCAs, and SPCAs. All these Ca 2ϩ pumps contribute to the removal of activator Ca 2ϩ from the cytoplasm after stimulation, thus decreasing the cytoplasmic Ca 2ϩ concentration back to baseline levels. PMCAs extrude Ca 2ϩ from the cytoplasm into the extracellular medium, whereas SERCAs accumulate Ca 2ϩ into the endoplasmic reticulum. Ca 2ϩ uptake into the Golgi apparatus is mediated by both SERCA and SPCA Ca 2ϩ pumps (for a review, see Ref. 1). The high concentration of Ca 2ϩ in the lumen of the intracellular organelles is not only a source of activator Ca 2ϩ for cytosolic processes but is also indispensable for proper transcription, translation, translocation, folding, and processing of secreted proteins (2). In the Golgi apparatus, luminal Ca 2ϩ is also required for intra-Golgi membrane transport, transport between the Golgi and ER, and endosome fusion. Although the pivotal role of the ER as a Ca 2ϩ store is well established, the view that also the Golgi apparatus can act as an agonist-releasable Ca 2ϩ store is much more recent (3,4). Also the knowledge of the Golgi-specific SPCA-type Ca 2ϩ -AT-Pases is much more limited than that of the well characterized SERCA and PMCA pumps. The PMR1 gene of Saccharomyces cerevisiae (5) is the first member of the SPCA family that has been described. The Pmr1 protein was localized to the Golgi apparatus or one of its subcompartments (6). Pmr1 is an ionmotive ATPase that supplies the secretory pathway with Ca 2ϩ and Mn 2ϩ ions required for glycosylation, sorting, and ERassociated protein degradation (7,8). Genes homologous to the S. cerevisiae PMR1 have been reported for a number of other fungi.
The characterization of the animal PMR1 homologues is more recent. Although the cDNA of the putative rat PMR1 homologue was cloned in 1992 using a SERCA-derived probe (9), the authors failed to functionally characterize this protein.
Direct evidence that the Pmr1 homologue in animals is able to transport Ca 2ϩ and Mn 2ϩ was presented for the Caenorhabditis elegans homologue by Van Baelen et al. (10). Ton et al. (11) demonstrated the Ca 2ϩ transport activity of the human homologue hSPCA1a (GenBank TM accession number AF181120). Later studies indicated that hSPCA1d isoform (GenBank TM accession number AY268375) is also capable of transporting Mn 2ϩ ions (12). The ability to transport Mn 2ϩ at an appreciable rate is a characteristic not shared by the SERCA or PMCA Ca 2ϩ transport ATPases. Like Pmr1 in yeast, the SPCA1 protein is localized in the Golgi apparatus.
The human ATP2C1 gene that encodes hSPCA1 was recently identified as the defective gene in Hailey-Hailey disease (Online Mendelian Inheritance in Man (OMIM) accession number 169600) (13,14). Hailey-Hailey disease is an autosomal dominant skin disorder that is characterized by suprabasal acantholysis of keratinocytes, resulting in epidermal blister formation. Up till now a total of 70 different mutations have been described in Hailey-Hailey disease patients (15). These mutations are scattered throughout the entire gene with no apparent clustering. The symptoms of Hailey-Hailey disease strongly resemble those of Darier-White disease (OMIM accession number 124200), which is caused by mutations in one of the SERCA2 gene (ATP2A2) alleles (16). The link between defects in intracellular Ca 2ϩ pumps and epidermal pathogenesis indicates that Ca 2ϩ inside the lumen of intracellular stores plays an essential role in preserving skin integrity.
The human genome also contains a second gene (ATP2C2) encoding a secretory pathway Ca 2ϩ /Mn 2ϩ -ATPase (hSPCA2) that has not been characterized except for a recent study published while the present work was submitted (17). This group revealed hSPCA2 as a Ca 2ϩ pump upon heterologous expression in yeast. Complementation studies in yeast suggested the importance of hSPCA2 in cellular Mn 2ϩ detoxification. hSPCA2 was shown to be expressed in neuronal cells. In contrast to the report of Xiang et al. (17) we showed that hSPCA2 displays a high apparent affinity for Ca 2ϩ similar to those values documented for hSPCA1a expressed in yeast (11) and for hSPCA1d expressed in COS-1 cells (12). Furthermore the Ca 2ϩ -or Mn 2ϩ -dependent activation of phosphoenzyme formation with [␥-32 P]ATP and the ability to pump both Ca 2ϩ and Mn 2ϩ were also demonstrated. Finally the expression of hSPCA2 at both mRNA and protein levels was analyzed showing that hSPCA2 has a much more restricted tissue distribution than hSPCA1 and colocalizes with Golgi-specific markers in colon epithelial cells.

EXPERIMENTAL PROCEDURES
Generation of the Expression Construct-The hSPCA2-encoding expression construct was generated by PCR using Clontech Advantage 2HF polymerase. Human colon cDNA was used as a template. Two overlapping fragments were amplified separately using primers that introduce a HindIII restriction site at the 5Ј-end (indicated in lowercase in the sequence shown below). The forward primer was NewhSPCA2F (5Ј-ATGTaagcttCGCCCGCTCACCATGGTCGAG-3Ј), and the reverse primer was OverlapSPCA2R1 (5Ј-TAATATCACTTAAGTCCATCT-TCATCGCCAG-3Ј). The 3Ј-fragment was amplified using the forward primer OverlapSPCA2F1 (5Ј-CTGGCGATGAAGATGGACTTAAGT-GATATTA-3Ј) and the reverse primer SPCA2FullR2 (5Ј-ATCGTgaat-tcACTACACATCTTCAGGGTGCATCTG-3Ј), which introduces an EcoRI site (sequence in lowercase) at the 3Ј-end. All primers were synthesized by Invitrogen. PCR products were subcloned into pGEM-T easy vectors using A/T cloning. Subsequently the 5Ј-fragment was ligated into pcDNA3 vectors using HindIII and EcoRI (present in the multiple cloning site of the pGEM-T easy vector). To complete the coding sequence, the construct containing the 5Ј-fragment in pcDNA3 was cut with AflII and EcoRI and ligated to AflII/EcoRI-cut 3Ј-fragments. The AflII site is a unique, endogenous restriction site in the overlapping part of the sequences. The sequence of the expression construct was verified before expression studies were attempted.
RNA Extraction and Reverse Transcription-Total RNA was extracted from tissues or cultured cells using TRIzol TM reagent (Invitrogen) using the procedure recommended by the manufacturer. Poly(A) ϩ RNA was purified from cultured cells with the Invitrogen micro-Fast-Track TM kit. Samples of total RNA of human colon tissue and human colon carcinoma tissue were purchased from Stratagene. First strand cDNA synthesis was performed on RNA samples using the Thermoscript TM RT-PCR system (Invitrogen) with oligo(dT) priming.
Ratio RT-PCR-cDNA reverse transcribed from 1 g of poly(A) ϩ mRNA or from total RNA was subjected to PCR using primers that amplify a 176-bp homologous fragment of both ATP2C1-and ATP2C2derived messengers. The primer sequences were: PMRRatioF, 5Ј-CT-GAAG(T or G)CTGCAGACATTGG-3Ј; and PMRRatioR, 5Ј-CTCGT-GCTCAGCTGGAATC-3Ј. The PCR program consisted of a 2-min denaturation step at 94°C followed by 27-30 cycles of the sequence 94°C for 1 min, 53°C for 1 min, and 74°C for 1 min. The final step was an additional elongation at 74°C for 2 min. The distinction between ATP2C1 and ATP2C2 amplicons was made by digesting the PCR fragments with MseI and/or ScrFI. MseI cuts the ATP2C1 but not the ATP2C2 amplicon in two fragments of 142 and 34 bp, respectively, whereas ScrFI digests only ATP2C2 and generates two fragments of 100 and 76 bp. After digestion, all PCR samples were subjected to 7.5% PAGE. The gels were run in Tris borate-EDTA buffer at 250 V. The different bands were visualized by fluorescent staining with Vistra Green TM (Amersham Biosciences) and scanning using a STORM TM 840 scanner (Amersham Biosciences). Quantification was performed using ImageQuant TM software (Amersham Biosciences).
Northern Dot-blot-The DNA hybridization probe was generated using Pfu Ultra polymerase (Stratagene) in 20-l PCRs using 10 ng of plasmid DNA of the hSPCA2 expression construct as template. PCR products were 32 P-labeled by the inclusion of 5 l of [␣-32 P]dCTP (10 Ci/l). The primers used for ATP2C2 were ATP2C2NBlotF (5Ј-CGC-CCGCTCACCATGGTCG-3Ј) and ATP2C2NBlotR (5Ј-CGCTCTGGC-CAAATCCTCTTTC-3Ј) and generated a 216-bp fragment corresponding to positions 78 -293 of the published sequence. For ATP2C1, the primers were ATP2C1NBlotF (5Ј-GGGGGCTTCTCTTCCTTGTC-3Ј) and ATP2C1NBlotR (5Ј-CCATGAAAGGCTCGCCTATG-3Ј). The hybridization probe was separated from unincorporated 32 P-labeled nucleotides using mini-Quick Spin columns (Roche Diagnostics). A human MTE TM (multiple tissue expression) array was purchased from Clontech. Probes were hybridized according to the manufacturer's instructions, and the labeling was detected by exposure to Biomax film (Eastman Kodak Co.).
Cell Culture-COS-1 cells were obtained from the European Collection of Cell Culturing and cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, 3.8 mM L-glutamine, 85 IU/ml penicillin, 85 g/ml streptomycin, and 1% non-essential amino acids. HeLa cervical carcinoma cells were obtained from the ATCC (Manassas, VA) and grown in Ham's F-12 medium supplemented with 10% fetal calf serum, 85 IU/ml penicillin, and 85 g/ml streptomycin. LNCaP and T47D cells were purchased from the European Collection of Cell Culturing and cultured in RPMI 1640 medium supplemented with 2 mM L-glutamine, 85 IU/ml penicillin, 85 g/ml streptomycin, and 10% fetal calf serum. 16HBE14oϪ bronchial epithelial cells were a gift from Dr. D. C. Gruenert (University of Vermont, Colchester, VT) and were cultured in a 50% Dulbecco's modified Eagle's medium, 50% Ham's F-12 medium supplemented with 20% fetal calf serum, 3.8 mM L-glutamine, 85 IU/ml penicillin, and 85 g/ml streptomycin. Cultured human colon carcinoma cells (Caco2) were purchased from the ATCC and maintained in Dulbecco's modified Eagle's medium supplemented with 20% fetal bovine serum, 2 mM L-glutamine, 2 IU/ml penicillin, and 2 mg/ml streptomycin. Human primary keratinocytes were isolated from human foreskin obtained from circumcision operations. Briefly the tissue was cut into small pieces and incubated in dispase. The epidermis was peeled from the dermis and incubated in Ca 2ϩ -and Mg 2ϩ -free PBS before collecting the cells by centrifugation (18). Keratinocytes were cultured in Keratinocyte-SFM medium (Invitrogen) supplemented with bovine pituitary extract (25 g/ml) and epidermal growth factor (0.1-0.2 ng/ml). To induce differentiation, the extracellular Ca 2ϩ concentration was raised from 0.09 mM in the normal medium to 1.2 mM, and the cells were allowed to grow to postconfluence.
For 45 Ca 2ϩ fluxes, COS-1 cells were seeded in 12-well gelatin-coated dishes at a density of ϳ30,000 cells/cm 2 and investigated 8 days later. For microsomal preparation, cells were seeded in 100-mm culture dishes at 2.5 ϫ 10 6 cells/plate.
Transfections-COS-1 cells were transiently transfected using Gene-Juice transfection reagent (Novagen). Transfection was performed the day after seeding except for 45 Ca 2ϩ fluxes in which case the period between seeding and transfection was 5 days to allow better attachment of the cells to the plates.
SPCA Antisera-A hSPCA1 antiserum, designated hSPCA1cytl, has been described previously by our group (19). The hSPCA2-specific antiserum was generated by Eurogentec. Rabbits were immunized with a mixture of two internal peptides coupled to keyhole limpet hemocyanin as a carrier protein. The internal peptides of hSPCA2 were SLKT-EDQEDIY (positions 514 -524) and VDSVEKGELADRVGK (positions 646 -659) and were chosen in regions with low sequence homology to hSPCA1. The resulting antiserum was designated XIB.
Immunocytochemistry-Cryosections (5 m) of human colon tissue or cells grown on gelatin-coated coverslips were washed several times with PBS and fixed with 3% paraformaldehyde in PBS for 15 min at room temperature. The cells were permeabilized by incubation in 0.2% Triton X-100 for 2 min at room temperature. After three washes with PBS, nonspecific binding was blocked by incubation for 1 h in PBS containing 3% bovine serum albumin and 1:100 normal goat serum. Primary antibodies were diluted in 1% bovine serum albumin in PBS and incubated for 1 h followed by three washes with PBS. As negative controls, coverslips were incubated with preimmune serum at the same dilution as the immune serum. Secondary antibodies were added in 1% bovine serum albumin in PBS and incubated for 1 h. Secondary antibodies were goat anti-rabbit Alexa Fluor 488 or 594 or goat anti-mouse Alexa Fluor 488 or 594 (Molecular Probes). Finally cells were washed, and the coverslips were mounted in Vectashield (Vector Laboratories) to inhibit photobleaching, sealed with nail polish, and examined with a fluorescence microscope (Olympus Cell R ). Subcellular structures could be identified using monoclonal antibodies directed against components of the ER or the Golgi. A SERCA2 antibody (IID8, Affinity Bioreagents) was used to visualize the ER, whereas the Golgi apparatus was identified using anti-TGN46 antibodies (Serotec).
Preparation of Crude Cell Extracts and Isolated Membranes-For total protein extractions, cells were washed twice in PBS before lysis in freshly prepared SDS extraction buffer (150 mM NaCl, 50 mM Tris-HCl, pH 8, 5 mM EDTA, and 2% SDS supplemented with the protease inhibitors 0.28 mM phenylmethylsulfonyl fluoride, 0.83 mM benzamidine, and a 1 g/ml concentration of each of leupeptin, aprotinin, and pepstatin A). After a 30-min incubation on ice, the insoluble material was pelleted at 6000 ϫ g and discarded. The supernatant was used for further experiments. Microsomes from transfected COS-1 cells were prepared as described by Verboomen et al. (20). Microsomes from tissues were prepared as follows. Fresh or frozen tissue was rinsed in cold PBS supplemented with 1 mM EDTA and allowed to equilibrate for 10 min on ice in hypotonic Buffer A (10 mM Tris-HCl, pH 7.5, 1 mM MgCl 2 , 2 mM DTT, 1 mM EDTA, and the protease inhibitor mixture given above). Homogenization was performed using a Teflon/glass Potter homogenizer. An equal volume of Buffer B (10 mM Tris-HCl, pH 7.5, 2 mM DTT, 1 mM EDTA, 300 mM KCl, and 500 mM sucrose) was added, and homogenization was continued. Cellular debris and nuclei were centrifuged at 4000 ϫ g for 15 min at 4°C. The supernatant was recentrifuged at 100,000 ϫ g for 40 min at 4°C followed by resuspension of the microsomal pellet in Buffer C (10 mM Tris-HCl, pH 7.5, 2 mM DTT, 150 mM KCl, and 250 mM sucrose) (21). Quantification of total protein was done using the bicinchoninic acid method (Pierce) using the manufacturer's instructions. When compounds in the samples interfered with this method, the protein was precipitated with trichloroacetic acid prior to quantification by the Lowry method.
Western Blotting-Microsomes or crude cell extracts were loaded on NuPage TM bis-Tris 4 -12% gradient gels or NuPage 7% Tris acetate gels (Invitrogen). After electrophoresis, the separated proteins were transferred to Immobilon-P membranes. The blots were quenched in TBST (Tris-buffered saline containing 0.05% Tween) supplemented with 5% (w/v) nonfat dry milk. When necessary, the primary antibody solution was preincubated with 10 g of the immunogenic peptides for 30 min. After three washes in TBST, incubation with the primary antibody in TBST containing 1% nonfat dry milk was performed for 1 h at room temperature. After labeling with the secondary antibody coupled to alkaline phosphatase (dilution, 1:8000), the labeled bands were detected using Vistra ECF (Amersham Biosciences) as substrate. Detection of the fluorescence was performed using a STORM 840 scanner (Amersham Biosciences) in combination with ImageQuant software (Amersham Biosciences).
Isotope Fluxes-45 Ca 2ϩ fluxes were performed on cells permeabilized with saponin. Cells grown in 12-well, gelatin-coated plates were treated for 10 min with 20 g/ml saponin in permeabilization buffer (120 mM KCl, 30 mM imidazole-HCl, pH 6.8, 2 mM MgCl 2 , 1 mM ATP, and 1 mM EGTA) at 25°C. After a wash with the same buffer without saponin, the non-mitochondrial stores were loaded with 45 Ca 2ϩ (3 Ci/ml) or 54 Mn 2ϩ (10 Ci/ml) for 90 min in a buffer containing 120 mM KCl, 30 mM imidazole-HCl, pH 6.8, 5 mM ATP, 10 mM NaN 3 , 0.44 mM EGTA, and 100 nM thapsigargin (SERCA-specific inhibitor). To obtain a free Mg 2ϩ concentration of 0.5 mM and the indicated free Ca 2ϩ and Mn 2ϩ concentrations, the total MgCl 2 , CaCl 2 , and MnCl 2 concentrations were calculated using the CaBuf program (ftp.cc.kuleuven.ac.be/pub/droogmans/ cabuf.zip) developed by G. Droogmans in our laboratory. These calculations are based on the dissociation constants given by Fabiato and Fabiato (22) for Ca 2ϩ and by Martell and Smith for Mn 2ϩ (23). After the loading phase, cells were washed twice with 1 ml of efflux medium (120 mM KCl, 30 mM imidazole-HCl, pH 6.8, and 1 mM EGTA). Subsequently the efflux medium was replaced every 2 min to monitor the leak of Ca 2ϩ . At the end of each experiment, the remaining 45 Ca 2ϩ in the cells was released by the addition of 1 ml of 2% (w/v) SDS for 30 min. 45 Ca 2ϩ or 54 Mn 2ϩ was quantified by liquid scintillation counting.
Formation of the Phosphoenzyme Intermediate-Microsomes (10 g) were phosphorylated on ice in a 100-l reaction mixture containing 1 l of 2 Ci/l [␥-32 P]ATP (Amersham Biosciences), 160 mM KCl, 17 mM Hepes, pH 7.0, 1 mM DTT, 5 mM NaN 3 , and the appropriate concentrations of Ca 2ϩ , Mn 2ϩ , and EGTA. The reaction was stopped after 20 s by the addition of ice-cold stop solution (6% trichloroacetic acid, 10 mM phosphoric acid, and 1 mM ATP). The protein was allowed to precipitate on ice for 30 min and pelleted by centrifugation at 4°C for 3 min at 13,200 rpm. Pellets were washed two times with stop solution and finally with 0.2 M sodium acetate before solubilization in modified SDS-PAGE sample buffer (150 mM Tris-HCl, pH 7.4, 8 mM EDTA, 3% SDS, 20% sucrose, 0.14 mg/ml bromphenol blue, and 10 mM DTT). Phosphorylated proteins were separated by acid 7.5% SDS-PAGE according to Sarkadi et al. (24). After fixation of the gels in 7.5% acetic acid they were dried between sheets of gel drying film (Promega) and exposed to a PhosphorImager TM screen (Amersham Biosciences) for quantification. Background phosphorylation due to endogenous Ca 2ϩ pumps in this expression system is negligible (25,26).

RESULTS
ATP2C2 Encodes a P-type ATPase-The GenBank TM entry of the cDNA sequence KIAA0703 (GenBank TM accession number AB14603, Ref. 27) was the first indication of a putative second isoform of the human secretory pathway Ca 2ϩ -ATPase. The existence of a second hSPCA isoform has been mentioned previously in the literature (11,28), but no further studies were published until the recent study by Xiang et al. (17). Our initial studies showed that overexpression of the protein encoded by the KIAA0703 cDNA in COS-1 cells did not result in the formation of a functional Ca 2ϩ -ATPase (data not shown). It is now clear that the reason behind this failure lies in the faulty N terminus. The KIAA0703 sequence contains three putative inframe ATG start codons at positions 441, 501, and 606. We assembled expression constructs for each of the three putative protein products. All three proteins could be overexpressed, but this did not result in an enhanced Ca 2ϩ transport capacity, indicating that non-functional proteins were formed (data not shown). Moreover these proteins were not localized in the Golgi but instead were found in the ER, and they were susceptible to enhanced degradation because Western blotting showed an intense additional band corresponding to a smaller proteolytic fragment (data not shown).
Further analysis of chromosome 16 suggested the existence of two alternative exons located more upstream in the genomic sequence. The first two exons of the KIAA0703 cDNA sequence were replaced by these two new exons, thus generating a new cDNA clone whose nucleotide sequence was deposited in Gen-Bank TM under accession number AY791884. In contrast to the corresponding sequence in KIAA0703, the revised 5Ј-end of the new cDNA clone was highly similar to that of the SPCA2 mRNA sequence from rat found in the data base (GenBank TM accession number AF484685). In addition, the first two exons of ATP2C2 are separated by a large intron of about 30 kb, which is similar in size to the corresponding intron of ATP2C1.
The complete genomic exon/intron layout is shown in Fig. 1. expression pattern of the novel gene with that of its previously identified family member ATP2C1 (Fig. 2). Whereas there was little variation in the level of the ubiquitous signal of ATP2C1 (Fig. 2B), the new paralogue presented large differences in the relative expression levels among various tissues ( Fig. 2A). The strongest signals were detected throughout the gastrointestinal tract from the stomach (B5) to the rectum (C6) with the most intense signal in the rectum (C6). Other positive tissues include trachea (H7), fetal lung (G11), prostate tissue (E8), thyroid gland (D9), salivary gland (E9), and mammary gland (F9).
To better quantify the relative expression levels of ATP2C1 and ATP2C2 mRNA in different tissues and cell lines, we used a ratio RT-PCR protocol. In this method, a common set of primers is used to co-amplify homologous fragments of related sequences (30). Fig. 3A shows a typical experiment in which the relative amount of both messengers is determined by restriction analysis of the PCR products. Ratio RT-PCR was performed on mRNA samples from a number of human tissues (lung, colon, and colon carcinoma) that tested positive for ATP2C2 mRNA expression ( Fig. 2A), from cultured cells derived from positive tissues (Caco2, colon; LNCaP, prostate; 16HBE14oϪ, lung; and T47D, mammary gland), and from other cultured human cell lines (SH5Y-5Y, human embryonic kidney, EA, and HeLa). The latter cell lines expressed only the ATP2C1 mRNA (data not shown), whereas Caco2, LNCaP, 16HBE14oϪ, and T47D cells did co-express the ATP2C2 and ATP2C1 mRNAs. Both transcripts were detected in all the above mentioned human tissues (Fig. 3B). The highest ATP2C2/ATP2C1 ratio (ϳ1:1) was found in colon tissue and in cell lines derived from colon, lung (16HBE14oϪ), or prostate (LNCaP). Additionally human keratinocytes were also included in this study because of the central role of hSPCA1 in Hailey-Hailey disease and because skin tissue was not represented on the master blot (Fig. 2). Human keratinocytes tested negative for the ATP2C2 mRNA when cultured under proliferating conditions (data not shown). However, when keratinocytes were allowed to differentiate (by growing them to confluency in the presence of high extracellular Ca 2ϩ concentration), the expression of both genes could be demonstrated (Fig. 3B).
Characterization of the hSPCA2 Protein-The hSPCA2specific antibodies (XIB) were generated against a mixture of two SPCA2-specific peptides that are not conserved in hSPCA1. The antiserum labeled a band on Western blots of hSPCA2-overexpressing COS-1 cells but not of control COS-1 cells (Fig. 4A). The specificity of the antiserum was proven by the absence of the immunoreaction with preimmune serum and its suppression by preincubation of the serum with the immunogenic peptides. The immunoreactive protein migrated slightly below the predicted M r value of 103,293. A similar anomalous faster migration has also been observed for the C. elegans SPCA homologue (10). Immunostaining of Western blots for other organellar Ca 2ϩ pumps demonstrated that overexpression of hSPCA2 did not alter the expression levels of the endogenous Ca 2ϩ -ATPases SERCA2b and hSPCA1 (Fig.  4A). The overexpressed protein was localized in a juxtanuclear region in COS-1 cells. This region corresponded to the Golgi apparatus as shown by the colocalization experiments depicted in Fig. 4B, middle panels. SERCA2, which predominantly stains the ER, showed a much more diffuse, reticular staining pattern (Fig. 4B, bottom panels).
In the next series of experiments, the endogenous hSPCA2 protein was studied in human colon. Fig. 5 shows a Western blot performed with microsomes isolated from human colon and lung. The protein could only be demonstrated in colon, a tissue with one of the highest hSPCA2 mRNA expression levels. The immunoreactive band migrated slightly higher than the protein overexpressed in COS-1 cells. This could be due to posttranslational modifications. An additional band of lower mobility was present in colon microsomes, but this band was nonspecific because it was not inhibited by preincubation of the antibody with the immunogenic peptides (Fig. 5). Immunohistochemical staining of human colon cryosections resulted in hSPCA2-specific labeling in close proximity of the apical pole of the nuclei of colon epithelial cells. This juxtanuclear staining colocalized with the Golgi marker TGN46 (Fig. 6). hSPCA1 showed a similar juxtanuclear staining, indicating that hSPCA1 and hSPCA2 reside in the same or closely juxtaposed subcellular compartments. The Golgi marker TGN46 appeared to label all epithelial cells because all nuclei were associated with a Golgi-like staining. Additionally cells stained for the Golgi marker were also positive for both hSPCA1 and hSPCA2.
hSPCA2 Is a Functional Ca 2ϩ -and Mn 2ϩ -transporting Enzyme-The hSPCA2 protein was functionally characterized by heterologous overexpression in COS-1 cells. The transient formation of a phosphorylated intermediate is a key feature of all P-type ion transport ATPases. To visualize the phosphorylated hSPCA2 intermediate, control cells (transfected with empty vector) and hSPCA2-overexpressing COS-1 cells were phosphorylated in vitro using [␥-32 P]ATP. The phosphorylated intermediate can be preserved on SDS-polyacrylamide gels in acidic conditions. Overexpressing cells clearly showed a phosphorylated enzyme intermediate in the presence of either Ca 2ϩ or Mn 2ϩ (Fig. 7). This phosphorylated intermediate was also formed when the phosphorylation reactions were performed in the presence of a 100 nM concentration of the SERCA-specific inhibitor thapsigargin, thus preventing the phosphointermediate formation from the endogenous SERCA2b. The specificity of the reactions was illustrated by the absence of signals when the experiments were done in the absence of Ca 2ϩ or Mn 2ϩ (EGTA). The bound radioactive phosphate was completely removed in the presence of hydroxylamine, demonstrating that the phosphate group was bound to a carboxyl and not to a hydroxyl group (Fig. 7B).
To study the ion transport activity in overexpressing cells we performed 45 Ca 2ϩ fluxes on permeabilized cells (10). In a first series of experiments we measured the Ca 2ϩ accumulation in the stores of overexpressing COS-1 cells as a function of time. Permeabilized COS-1 cells were incubated for different time periods in a medium resembling the cytosolic ionic composition and containing 45 Ca 2ϩ . To exclude the interference of the endogenous SERCA2b pump, the experiments were performed in the presence of 100 nM thapsigargin. Fig. 8A shows that the time-dependent loading of 45 Ca 2ϩ in hSPCA2-overexpressing COS-1 cells was higher than in control cells. This difference increased with time both in absolute and relative terms. Because the cells are more prone to detach during prolonged incubation time, the loading time was restricted to 90 min as described previously (12).
To determine the Ca 2ϩ dependence of the hSPCA2-mediated Ca 2ϩ uptake, permeabilized cells were loaded at different free Ca 2ϩ concentrations for a fixed time of 90 min, and the difference in the level of 45 Ca 2ϩ uptake of hSPCA2-overexpressing cells and control cells was measured (Fig. 8B). Half-maximal activation (K 0.5 ) of the enzyme was observed at 0.27 M free Ca 2ϩ concentration.
Because previous reports on the SPCA1 pump of humans (12) and C. elegans (10) have shown that SPCA1 can act as a Mn 2ϩ pump, we tested whether hSPCA2 also functions as a Mn 2ϩ -transporting enzyme. Control cells and hSPCA2-overexpressing cells were loaded with 54 Mn 2ϩ for 90 min in the presence of 100 nM thapsigargin, and the efflux of 54 Mn 2ϩ was followed for 18 min. The higher uptake of 54 Mn 2ϩ in hSPCA2-overexpressing cells provided direct evidence that hSPCA2 can act as a Mn 2ϩ -transporting ATPase (Fig. 8C). The efflux of accumulated Mn 2ϩ was accelerated by the ionophore A23187 (10 M), demonstrating that the Mn 2ϩ ions had been transported into a membrane-delineated compartment.
In the presence of A23187, the efflux curves of control cells and hSPCA2-overexpressing cells converged, showing that the amount of background 54 Mn 2ϩ binding was identical and that the difference between the two curves was specifically due to an enhanced Mn 2ϩ transporting capacity of hSPCA2expressing cells. DISCUSSION In this study we confirmed the finding of Xiang et al. (17) that human tissues can express a second isoform of secretory pathway Ca 2ϩ -ATPase, named hSPCA2, encoded by the ATP2C2 gene. We further explored its tissue-specific and cellular expression pattern and present its functional characteristics upon heterologous expression in COS-1 cells. A TBLASTN search with an SPCA signature peptide sequence (IQEYRSEKSLEELTK) revealed that mammalian genomes (Homo sapiens, Pan troglodytes, Canis familiaris, Bos taurus, Rattus norvegicus, and Mus musculus) all contain two different SPCA-encoding genes. Also the chicken genome (Gallus gallus) appears to contain two related genes. We did not find any indication for the presence of a second SPCA isoform in the fish Danio rerio, Fugu rubripes, and Tetraodon nigroviridis. This is remarkable because in general, as a result of an additional genome duplication, euteleost fishes seem to have larger gene families than tetrapods (31). Whereas euteleost genomes contain more SERCA genes than mammals, this is apparently not the case for the SPCA genes. Invertebrates (Drosophila melanogaster, Anopheles gambiae, Apis mellifera, and C. elegans) also seem to possess only one SPCA1 gene, but these lower organisms also contain only a single SERCA gene. Given the complete description of the human genome, the ATP2C2 gene is probably the last functional gene to be identified in the multigene family of type 2 Ca 2ϩ -transport ATPases. This superfamily consists of three SERCA-encoding genes (ATP2A1-3), four PMCA genes (ATP2B1-4), and two SPCA-encoding genes (ATP2C1 and -2). hSPCA1 and hSPCA2 mRNAs present different tissue distributions. hSPCA1 was recognized as a housekeeping enzyme (13), and this observation was confirmed and extended by the result shown in Fig. 2B. The expression pattern of hSPCA2 mRNA is much more heterogeneous. The highest levels occur in the gastrointestinal tract and in a number of secretory tissues as determined by Northern dot-blot hybridization and by ratio RT-PCR. Also the tissue distribution of expressed sequence tag clones in the data base is compatible with this pattern. These results are different from the data published by Xiang et al. (17) who showed abundant hSPCA2 protein expression in brain and testes. In our studies, the mRNA expression of hSPCA2 in brain and testes was in the lower range (Fig. 2).
Because skin tissue was not represented in the RNA blots used and because of the central role of the hSPCA1 pump in Hailey-Hailey disease, we investigated the co-expression of hSPCA1 and hSPCA2 in a separate set of experiments on isolated keratinocytes using PCR. We demonstrated that hSPCA2 is expressed at the mRNA level in differentiated keratinocytes. The co-expression of hSPCA2 and hSPCA1 could help to explain the relatively high contribution of thapsigargininsensitive Ca 2ϩ accumulation in the Golgi of keratinocytes (33) and lung-derived 16HBE14oϪ cells (34). The proper function of the secretory pathway in keratinocytes may be crucial for correct delivery of cell adhesion components (15). The fact that hSPCA2 is not able to compensate for the decreased level of hSPCA1 in keratinocytes of Hailey-Hailey disease patients may suggest a specific function for this pump distinct from that of hSPCA1. This hypothesis fits with the observation that a complete knock-out of the ATP2C1 gene in mice results in an embryonic lethal phenotype. 2 The yeast Pmr1 pump has been located to the medial Golgi compartment (6,7). hSPCA1 is also a Golgi-resident protein (11) and comigrates with markers for the trans-Golgi compartment (32). hSPCA2 shares the property of Golgi localization with hSPCA1, but the exact localization inside the Golgi has not been studied yet. In epithelial cells of human colon tissue, both hSPCA isoforms show the same Golgi-like localization. However, the resolution of immunofluorescence microscopy may not be adequate to determine whether both pumps are in the same subcompartments of the Golgi. Reports on the mouse SPCA1 indicate that it can also be present in post-Golgi compartments (35).
The co-expression of hSPCA2 with hSPCA1 suggests that hSPCA2 supplements the role of hSPCA1 in the thapsigargininsensitive Ca 2ϩ /Mn 2ϩ accumulation pathway in the Golgi of a number of cell types. The expression in secretory cells suggests that hSPCA2 assists in the function of hSPCA1 or that it performs an as yet unknown specialized function in those cell types in which the Golgi apparatus is important not only for the basic cell biological maintenance of cell function but also for the specific secretory task these cells perform in the organism. The large number of epithelial cells expressing hSPCA1 and hSPCA2 in colon epithelium cannot be accounted for by the number of mucus-secreting cells only, indicating that these Ca 2ϩ pumps also are present in enterocytes and play an important role in their function.
Xiang et al. (17) have reported a slightly different localiza-2 T. Doetschman and G. E. Shull, unpublished data. tion of hSPCA2. They found hSPCA2 in TGN-derived vesicles in rat hippocampal neurons. We showed a predominantly juxtanuclear Golgi-like distribution in human colon epithelial cells. This difference most probably reflects variations in the architecture of the secretory pathway in different cell types.
Our functional studies of the overexpressed protein in mammalian cells confirmed the findings of Xiang et al. (17) that hSPCA2 is a P-type ATPase capable of transporting both Ca 2ϩ and Mn 2ϩ with high affinity. Half-maximal (K 0.5 ) stimulation of the enzyme occurred at a free Ca 2ϩ concentration of 0.27 M, which is in the same range as that described for the human SPCA1a protein (0.26 M) (11), hSPCA1d (0.20 M) (11), and its homologue in C. elegans (0.25 M) (10). The yeast homologue Pmr1 shows a slightly higher apparent affinity for Ca 2ϩ (Յ0.1 M) (8). When hSPCA2 is overexpressed in yeast, a lower halfmaximal activation of the enzyme for Ca 2ϩ is observed (K 0.5 ϭ 1.35 M) (17). This discrepancy can be explained by the use of different expression systems, i.e. yeast versus mammalian cells. The Golgi apparatus of yeast cells is less structured than that of mammalian cells, and the membranes may have a different lipid composition, which may affect transport activity. The involvement of additional essential cofactors present in mammalian cells, but not in yeast, can also not be excluded.
Analogous to hSPCA1, Ca 2ϩ transport by the hSPCA2 protein was insensitive to thapsigargin (SERCA-specific inhibitor). We also demonstrated in this study that hSPCA2 is capable of transporting Mn 2ϩ . The phosphorylation properties of hSPCA2 were further investigated by testing its ability to be phosphorylated in the presence of Ca 2ϩ and Mn 2ϩ . In each case, a phosphoprotein intermediate could be demonstrated. Furthermore the formation of the phosphointermediate was sensitive to hydroxylamine, showing that the phosphorylation occurs on the catalytic carboxyl group of an aspartate residue, thus excluding the possibility that it was the result of protein kinase activity.
In conclusion, we functionally characterized a second isoform of SPCA-type ATPase. Both hSPCA2 overexpressed in COS-1 cells and hSPCA2 endogenously present in colon epithelial cells displayed a Golgi-like juxtanuclear distribution. hSPCA2 was shown to be capable of transporting Ca 2ϩ with high affinity in a thapsigargin-independent way. Further studies are required to characterize the specific properties of hSPCA2 and its function in cellular divalent ion homeostasis.