Deficiency of ATP2C1, a Golgi Ion Pump, Induces Secretory Pathway Defects in Endoplasmic Reticulum (ER)-associated Degradation and Sensitivity to ER Stress*

Relatively few clues have been uncovered to elucidate the cell biological role(s) of mammalian ATP2C1 encoding an inwardly directed secretory pathway Ca 2 (cid:1) /Mn 2 (cid:1) pump that is ubiquitously expressed. Deficiency of ATP2C1 results in a human disease (Hailey-Hailey), which primarily affects keratinocytes. ATP2C1-encoded protein is detected in the Golgi complex in a calcium-de-pendent manner. A small interfering RNA causes knockdown of ATP2C1 expression, resulting in defects in both post-translational processing of wild-type thyroglobulin (a secretory glycoprotein) better in identifying an ATP2C1-specific band in transfected COS-7 cells. Immunoglobulins from 161 purified on protein A-agarose. Despite purification, antibody 161 could not reliably detect a strong signal of endogenous ATP2C1 by Western blotting, although it yielded a strong signal by immunofluorescence. Immunofluorescence— plates, cells Coverslips

The secretory pathway calcium pool is multifunctional, regulating activities within the endoplasmic reticulum, Golgi complex, and secretory vesicles, including secretory protein folding and molecular chaperone function, post-translational modifications, and intracellular transport and secretion (1) as well as intracellular signaling and cell viability during ER 1 stress (2)(3)(4)(5)(6)(7)(8). Two P-type ATPases have been identified in yeast that contribute to secretory pathway divalent cation homeostasis (9,10), one of them being Pmr1p, which pumps Ca 2ϩ and Mn 2ϩ from the cytosol into the lumen of the secretory pathway (11)(12)(13)(14)(15). In addition to members of the ER-localized, thapsigarginsensitive SERCA pumps encoded by the ATP2A family of genes, mammalian cells express the thapsigargin-insensitive ATP2C1 gene product (16 -18). The latter protein shares a high level of identity with Pmr1p pumps of yeast and Caenorhabditis elegans that operate at the Golgi complex (12,17,19,20).
In yeast, absence of PMR1 results in a strain that grows well in standard media but exhibits defective growth in the presence of calcium chelators (14) or during various imposed stress conditions (9,10,17). Carbohydrate processing of N-linked glycoproteins is also abnormal in pmr1 mutants (10 -12, 14). In addition, although Pmr1p is localized to the Golgi, pmr1 was identified as der5 in a genetic screen for mutants defective in ER-associated degradation (ERAD) (21), thereby stabilizing the misfolded CPY* glycoprotein (14).
In humans, loss of one functional ATP2C1 allele results in Hailey-Hailey disease (22,23). Affected individuals achieve normal development, fertility, and a normal life span but develop red blistering skin erosions attributable to defective cell adhesion of keratinocytes in the epidermis. Because the ATP2C1 gene product is thought to be ubiquitously expressed, it is unknown why the disease phenotype should be limited to keratinocytes. Using small interfering RNA (siRNA), knockdown of ATP2C1 in mammalian cells has been reported to inhibit calcium accumulation in the Golgi complex (24) and also, significantly, in the ER (25). However, ATP2C1-related secretory pathway phenotypes have remained largely unexplored in mammalian cells.
In this report, we examine endogenous ATP2C1 Ca 2ϩ pump expression by immunofluorescence, the detection of which is markedly increased in cells incubated in low Ca 2ϩ media. Using siRNA-mediated knockdown of ATP2C1, we show that such cells exhibit defects in glycan processing of wild-type thyroglobulin (a secretory glycoprotein), ERAD of mutant thyroglobulin, and hypersensitivity to ER stress. These findings underscore an important role of ATP2C1 in secretory pathway homeostasis.

EXPERIMENTAL PROCEDURES
Materials-MG115 was from Calbiochem; brefeldin A and cycloheximide were from Sigma. Anti-thyroglobulin antibodies have been described elsewhere (26).
Cells-293, COS-7, NRK, FRT, and HaCat cells (kindly provided by Dr. G. Bokoch, Scripps Institute, La Jolla, CA) and primary human fibroblasts were grown in Dulbecco's modified Eagle's medium (25 mM glucose) supplemented with 10% fetal bovine serum and penicillinstreptomycin (100 units/ml-100 g/ml) except in Fig. 2 in which NRK cells and primary fibroblasts were cultured under serum-free conditions. Primary human keratinocytes were cultured in EpiLife supplemented with "human keratinocyte growth supplement" (cells and media from Cascade Biologics) at a calcium concentration of 0.09 mM. In preparation for experiments Ca 2ϩ was supplemented to 1.25 mM in selected samples for 48 h before fixation and preparation for immunofluorescence.
Transfection of 293 and FRT cells at ϳ80% confluence was performed in individual wells of a 6-well plate using 2 g of 3ϫHA (hemagglutinin)-rATP2C1 plasmid (below) and 4 l of Lipofectamine (Invitrogen) in Opti-MEM for 4 h, and then the cells were supplemented with 10% fetal bovine serum for 12 h, after which it was replaced with fresh growth medium. Individual clones were selected with 800 g/ml G418 and maintained with 400 g/ml G418.
Preparation of Anti-ATP2C1 Antibody-A cDNA (derived from Gen-Bank TM accession number NM131907) encoding residues 369 to 475 of the rat ATP2C1 protein was subcloned into pGEX-3X, encoding a glutathione S-transferase-fusion protein. (More than 40% of the residues in this sequence differ from that in the distinct rATP2C2 gene product (GenBank TM accession number NP604457)). After isopropyl 1-thio-␤-Dgalactopyranoside induction, bacterial lysates were purified on glutathione-Sepharose. Purified fusion protein was used as antigen, and three rabbit antisera (160, 161, and 304) were obtained. Of these, antiserum 161 yielded a better signal in identifying an ATP2C1-specific band in transfected COS-7 cells. Immunoglobulins from 161 were purified on protein A-agarose. Despite purification, antibody 161 could not reliably detect a strong signal of endogenous ATP2C1 by Western blotting, although it yielded a strong signal by immunofluorescence.
3ϫHA Tagging the rATP2C1 cDNA-To create a start codon and N-terminal 3ϫHA tag, a synthetic oligonucleotide primer was designed to include a HindIII site immediately preceding codon 8 of rATP2C1 (GenBank TM accession number NM131907), and a reverse primer including an EcoRI site was designed to begin at nucleotide 894 of the rATP2C1 sequence. The PCR product includes within it a unique ClaI site of rATP2C1. A 2449-bp ClaI fragment of the pMT SPCA vector was subcloned via the ClaI site to generate 3ϫHA-rATP2C1 in pCDNA3.1, encoding a 953-residue gene product in which 41 residues including the 3ϫHA tag replace the first 7 residues of rATP2C1. The final product was confirmed by DNA sequencing.
ATP2C1 Knockdown-A double-stranded RNA oligonucleotide target sequence for rATP2C1 was designed following the AA(N)19 rule (27) and chosen from rATP2C1 positions 384 -406: AAC CAT TAT GGA AGA AGT ACA TT. This sequence differs at four positions (bold letters) from that found in human ATP2C1: AGC CAC TGT GGA AGA AGT ATA TT. The double-stranded oligonucleotide (siRNA) for rATP2C1 was synthesized with and without a fluorescein isothiocyanate tag on the coding strand (Dharmacon). As a control, we also synthesized the scrambled double-stranded oligonucleotide CAC TAT GTG AGA AGA TAA ACA TT. Transfection of the siRNA (0.6 mM oligonucleotide) employed Oligofectamine (Invitrogen) diluted in Opti-MEM. Cells at 50 -80% confluence were incubated with this mixture for 6 h at 37°C and then supplemented with 10% fetal bovine serum for 15 h. Cells were then washed and returned to the complete growth medium. By direct fluorescence of fluorescein isothiocyanate-tagged oligoribonucleotides, a very high fraction of cells (Ͼ70%) incorporated the siRNA in these cell types (not shown). Thus, we did not routinely use fluorescence-activated cell sorting to separate cells bearing or lacking the siRNA.
Pulse-Chase and Western Blotting-For pulse-chase experiments, control and siRNA-treated cells were preincubated for 30 min in Met/ Cys-deficient Dulbecco's modified Eagle's medium at 37°C. Labeling for 30 or 60 min employed Expre 35 S 35 S (PerkinElmer Life Sciences). At various chase times, cells were lysed in 150 mM NaCl, 0.1% SDS, 1% Nonidet P-40, 2 mM EDTA, and 10 mM Tris-HCl, pH 7.4, plus complete protease inhibitor mixture (Roche Applied Science). Lysates were precleared with 30 l of zysorbin (Zymed Laboratories Inc.), and then primary antibody was added to the supernatant for overnight incubation at 4°C and immunoprecipitation. When necessary, immunoprecipitates were diluted into endoglycosidase H denaturing buffer (New England Biolabs) and heated to 100°C for 1 min. Zysorbin was then pelleted and the supernatant split in two equal portions to be either digested with 1,000 units of recombinant endo H (New England Biolabs) or mock-digested with buffer for 1 h at 37°C. Finally, with or without endo H digestion, immunoprecipitates were solubilized in gel sample buffer and boiled before SDS-PAGE and fluorography or phosphorimaging. For mutant insulin, cells were lysed, immunoprecipitated, and analyzed as described previously (28).
ER Stress Induction and Crystal Violet Staining-In 6-well plates, FRT cells without or with ATP2C1 knockdown were exposed to different concentrations of tunicamycin (Calbiochem) for 48 h, washed three times in PBS, and stained with 0.2% (w/v) crystal violet (Sigma) in 2% ethanol.
Reverse Transcription-PCR for XBP-1-Briefly, after incubation with tunicamycin, 293s22 cells were washed twice with PBS, and total RNA was isolated with RNeasy (Qiagen). Forward 5Ј-CCT TGT AGT TGA GAA CCA GG-3Ј and reverse 5Ј-GGG GCT TGG TAT ATA TGT GG-3Ј (GenBank TM accession numbers AB076384 and AB076383) primers were used with 5 l of cellular RNA suspension for reverse transcription-PCR using the one-step reverse transcription-PCR (Invitrogen) kit. Reverse transcription was carried out at 50°C for 30 min, and PCR was performed in 35 cycles, using 57°C in the annealing step. After PCR, the synthesized products were resolved by 4% agarose gel electrophoresis and stained with ethidium bromide.

RESULTS
Localization and Expression of the Endogenous ATP2C1 Protein-A rabbit polyclonal antiserum was prepared against a glutathione S-transferase fusion protein encoding residues 369 -475 in a large cytoplasmic loop of the polytopic membrane protein encoded by ATP2C1 (Fig. 1, inset) (16). Upon characterization of the antibody we found it most suitable for immunofluorescence studies, revealing a Golgi localization (i.e. selectively co-localizing with Golgi mannosidase II and redistributing to a pattern similar to that of ER-localized protein disulfide isomerase in brefeldin

ATP2C1 in the Secretory Pathway
A-treated cells) as has been reported previously (17,20,30,31). It was therefore of interest to examine endogenous expression of ATP2C1 in keratinocytes relative to other cells, as patients with ATP2C1 haploinsufficiency exhibit a keratinocyte cell adhesion defect (32). Cells maintained in serum-free medium (as is standard for keratinocyte cultures) were examined after a 2-day exposure to calcium added to a level of 1.25 or 0.09 mM (the latter also standard for primary keratinocyte cultures). Primary human keratinocytes had very bright immunofluorescent expression of the ATP2C1 gene product in comparison with other cells, and addition of 1.25 mM Ca 2ϩ to the culture medium markedly decreased this (Fig. 2). A low-calcium, serum-free culture environment is more challenging for growth of other cell types; nevertheless, similar behavior was also observed in primary human fibroblast cultures and NRK cells under these conditions. HaCat keratinocyte cell cultures grow in the presence of 10% fetal bovine serum (free Ca 2ϩ ϳ0.09 mM), and in these too, the addition of 1.25 mM calcium provoked loss of immunofluorescent ATP2C1 expression (whereas Golgi GM130 expression persisted). Together these results indicate an especially robust expression of the ATP2C1 gene product in nonmutant keratinocytes as well as a clear relationship between calcium availability and immunofluorescent ATP2C1 expression.
Expression of Epitope-tagged Rat ATP2C1-To follow more sensitively the ATP2C1 gene product by Western blotting or immunoprecipitation, we constructed a rat ATP2C1 cDNA that encoded an N-terminal 3ϫHA epitope tag (i.e. 3ϫHA-rATP2C1). To establish that the tagged construct is functional, a yeast GAL1 promoter was subcloned behind the 3ϫHA-rATP2C1 cDNA, which was transformed into pmr1 yeast. When transformed only with the empty high copy vector (pYES2), pmr1 cells are unable to grow on an EGTA plate; this phenotype is complemented by expressing PMR1 on a plasmid as demonstrated previously (14). Moreover, in multiple independent transformants in which 3ϫHA-rATP2C1 was expressed from the pYES2 vector, pmr1 cell growth on an EGTA plate was rescued (Fig. 3); similar rescue was observed by rATP2C1 without the HA epitope tag (not shown). Although not quantitatively assessing enzyme function, these crude complementation experiments clearly established that 3ϫHA-rATP2C1 encodes a bioactive protein. When the 3ϫHA-rATP2C1 construct was subcloned into a mammalian expression vector, the anti-HA immunofluorescence was confined to a juxtanuclear compartment, which co-localized with the endogenous ATP2C1 gene product in transiently transfected 293 cells (Fig. 4, upper panels) as well as in stably transfected 293s22 cells in which anti-HA immunofluorescence co-localized with Golgi mannosidase II and GM130 (Fig. 4, lower panels). Thus, the epitope-tagged rATP2C1 construct both functions and is localized normally in mammalian cells.
Knockdown of Rat ATP2C1 and Human ATP2C1 Expression-For knockdown of ATP2C1 protein expression, we selected a synthetic 21-mer oligoribonucleotide duplex (corresponding to nucleotides 385-408 of the rat ATP2C1 coding sequence). We proceeded to test the siRNA duplex in either rat-derived FRT cells, human-derived 293 cells, or the 293s22 clone that stably expresses 3ϫHA-rATP2C1. The siRNA was not expected to have an impact in the 293 cells (27) because the human mRNA in this region differs at four positions from that found in rodents (see "Experimental Procedures"). Nevertheless, because of the 100% identity that exists along many stretches shared between the rat and human mRNA sequences (overall 97% identity), we were curious as to the possibility that 293s22 cells made to express rATP2C1 would be rendered susceptible to knockdown of total ATP2C1 (endogenous hATP2C1 as well as rATP2C1 after siRNA treatment (33)).
Five days after siRNA treatment of FRT cells, we observed a dramatic loss of immunofluorescence expression of the rATP2C1 gene product (Fig. 5, upper left panels). By immunoblotting with anti-HA in FRT or 293s22 cells, we also observed profound knockdown of rATP2C1 at 2 days, 5 days, or 8 days post-siRNA treatment (Fig. 5, upper right). Control 293 cells that had not been transfected to express rATP2C1 were not susceptible to knockdown of endogenous hATP2C1, which showed a normal distribution in the Golgi complex (Fig. 5,  lower left panel). However in 293s22 cells, although GM130 remained properly localized (Fig. 5, lower middle) as did Golgi mannosidase II (not shown), anti-HA immunofluorescence was rendered virtually undetectable (lower right), and the endogenous hATP2C1 gene product could no longer be detected as well (lower middle and right). Although the extent to which hATP2C1-specific mRNA fragments might be generated by dicer activity upon treatment of 293s22 cells with rATP2C1specific siRNA is unknown, the utility of the above finding is that the persistence of hATP2C1 expression in control 293 cells provides an outstanding control for specificity of siRNA effects in 293 cells. (Independent controls comprised of no oligoribonucleotide, single-stranded oligoribonucleotide, or scrambled duplex oligoribonucleotide sequence all worked successfully as controls as well.)

Effects of ATP2C1 on Post-translational Modifications Including Acquisition of Endoglycosidase H Resistance and Sorting in the Golgi
Complex-In addition to pumping Ca 2ϩ , ATP2C1 pumps Mn 2ϩ (17,19), which is thought to allow for optimal activity of a variety of enzymes involved in the posttranslational processing of glycans on glycoproteins (34 -36).
To check for phenotypes related to these activities, FRT-thyroglobulin cells (37) treated or untreated with ATP2C1 siRNA were pulse-labeled with 35 S-amino acids and chased for either 15 min or 5 h. Acquisition of endoglycosidase H resistance for ϳ80% of the many N-linked glycans on each thyroglobulin molecule occurs upon the arrival of thyroglobulin at the medial Golgi complex (38); normally, this is accompanied by a slowing of thyroglobulin band mobility upon SDS-PAGE. However, upon knockdown of ATP2C1, acquisition of endo H resistance for newly synthesized thyroglobulin was not accompanied by the same degree of slowing of thyroglobulin band mobility, indicating alterations in the fidelity of secretory pathway posttranslational glycan processing (Fig. 6A). (As an aside, thyroglobulin acquisition of endo H resistance occurred a little more quickly in siRNA-treated cells; however, these effects were subtle, and the ultimate extent of thyroglobulin export did not differ.) Effects of ATP2C1 on ERAD, ER Stress Response, and ER Stress-induced Cell Toxicity-We have previously shown that the cog thyroglobulin mutation encodes a mutant glycoprotein that serves as a substrate for ERAD via the ubiquitin-proteasome system (26). To examine the role of ATP2C1 in this process, FRT or 293s22 cells expressing the cog thyroglobulin cDNA were transfected with the ATP2C1 siRNA. In mock-  5. Knockdown of ATP2C1. Upper left panels, 2 days after rATP2C1 siRNA treatment, rat-derived FRT cells exhibited a profound loss of immunofluorescent expression of the endogenous gene product. N denotes nuclei. Upper right panels, the FRT(A11) or 293s22 clones, both of which stably express 3ϫHA-rATP2C1, were transfected with rATP2C1 siRNA. At the times indicated, the cells without prior fluorescence-activated cell sorting were lysed, and equal aliquots of cellular protein were analyzed by SDS-PAGE and Western blotting with anti-HA. Below the line, either control 293 cells or 293s22 cells were transfected with ATP2C1 siRNA. In control 293 cells, the ATP2C1 siRNA was without effect, resulting in continued expression of endogenous human ATP2C1 that co-localized with the Golgi marker GM130 (panel at left). In identically treated 293s22 cells, GM130 was unchanged, but detection of ATP2C1 with either anti-ATP2C1 or anti-HA was lost. DAPI was used to stain nuclei.

ATP2C1 in the Secretory Pathway
transfected FRT cells, labeled cog thyroglobulin present in the cells 1 h after synthesis (Fig. 6B, lane 1) had completely disappeared at 24 h (Fig. 6B, lane 3), whereas in siRNA-treated FRT cells, labeled cog thyroglobulin persisted even after 1 day of chase (Fig. 6B, lane 4). In control 293 cells bearing only endogenous hATP2C1, treatment with the rat-specific siRNA had no effect on the complete degradation of cog thyroglobulin (Fig.  6B, lanes 9 -12). However, in 293s22 cells as in the rat-derived FRT cells, knockdown of ATP2C1 dramatically stabilized the cog thyroglobulin mutant, persisting even after 1 day of chase (Fig.  6B, lane 8). Quantitative recovery data from this representative experiment (Fig. 6B, bar graph at right) were replicated in three independent experiments, and no mutant thyroglobulin was detectably secreted from any of the samples (not shown).
For comparison, we examined intracellular degradation of a mutant single-chain insulin, which lacks glycans and has been shown to be completely defective for secretion because of the absence of one of three evolutionarily conserved disulfide bonds (28). Although modest slowing of ERAD could be observed by the addition of the proteasome inhibitor MG115, ATP2C1 knockdown afforded no protection from ERAD for the nonglycosylated substrate (Fig. 6C). These data indicate that ATP2C1 knockdown does not inhibit the general ERAD machinery but has specificity to distinct ERAD substrates.
In some instances, ERAD defects have been found to be associated with the induction of ER chaperone synthesis as part of the unfolded protein response (UPR) (39 -42). We compared BiP levels in 293 cells treated with the rat-specific siRNA (a negative control) to 293s22 cells with or without knockdown of ATP2C1. As shown in Fig. 7A, knockdown of ATP2C1 expression did not induce an increase in BiP and also had no effect on intracellular levels of calreticulin or PDI. We considered the possibility that biosynthesis of BiP might indeed be induced but might fail to be properly retained in the ER and become secreted, leaving steady-state levels unaffected (43). However as shown in Fig. 7B, BiP synthesis was unaltered by knockdown of ATP2C1, and BiP was efficiently retained within cells over a 5-h period. These findings exclude the possibility that knockdown of ATP2C1 itself significantly activates the ER stress pathways leading to ER chaperone induction.
Deletion of the PMR1 gene renders yeast cells hypersensitive to ER stress. In FRT cells exposed for 16 h to various low doses of tunicamycin (Fig. 7C), cells transfected with the control oligoribonucleotide remained in culture and could be stained with crystal violet (wells at left). By contrast, as the tunicamycin dose was increased up to 1 g/ml, almost no cells with knockdown of ATP2C1 remained (Fig. 7C, wells at right). After challenge with 100 nM thapsigargin, hypersensitivity was again observed selectively in cells with insufficiency of ATP2C1 (not shown). Thus, expression of ATP2C1 appears to facilitate survival in the face of ER stress.
Because failure to activate ER stress pathways (most notably PERK (44)) can result in cellular toxicity (45), we considered the possibility that ATP2C1 deficiency might impair UPR activation. We examined that question using either low dose or high dose tunicamycin treatment protocols (Fig. 8). We could dismiss the notion that ATP2C1 knockdown prevents either proteolytic cleavage and disappearance of full-length ATF6 or PERK-mediated phosphorylation of the eIF2␣ subunit of the eIF2B translational initiation factor because these activation events were clearly detected after tunicamycin treatment (Fig.  8A). We did find it more difficult to detect the active cleaved ATF6 transcription factor in cells treated with the ATP2C1 siRNA, but this was not a consistent finding (Fig. 8B, upper  panel). Additionally, ATP2C1 knockdown did not prevent Ire1mediated splicing of the XBP-1 mRNA that was seen after high-dose tunicamycin treatment (Fig. 8B, lower panel). Thus, increased toxicity from ER stress in cells with insufficiency of ATP2C1 cannot be explained by an inability to activate the ER stress response pathways. DISCUSSION ATP2C1 encodes a polytopic ion pump that resides in the Golgi complex of all eukaryotic cells. Despite the fact that the FIG. 6. Phenotypic effects of ATP2C1 knockdown. A, FRT-thyroglobulin (Tg) cells Ϯ ATP2C1 knockdown were pulse-labeled with 35 S-amino acids for 60 min and chased as indicated. After 5 h in control cells without ATP2C1 knockdown, the newly synthesized thyroglobulin shifted to a position of slower mobility by SDS-PAGE, which includes the appearance of an endo H-resistant form. In cells with ATP2C1 knockdown, thyroglobulin did not develop the same degree of slowing of band mobility, indicating differences in thyroglobulin post-translational processing. B, FRT cells, 293 cells, and 293s22 cells were transfected or not transfected with ATP2C1 siRNA. After 3 days, 293 and 293s22 cells were transfected again to express cog thyroglobulin whereas the FRT cells were already a stable clone expressing cog thyroglobulin). 2 days after this transfection, the cells were pulse-labeled with 35 S-amino acids and chased either for 1 or 24 h. No cog thyroglobulin was secreted (not shown). In control 293 cells, the newly synthesized cog thyroglobulin was completely degraded by 24 h (quantified in bar graph at right). After ATP2C1 knockdown, cog thyroglobulin persisted in either FRT cells (open bars at right) or 293s22 cells (closed bars at right). C, 293s22 cells were transfected either with control oligoribonucleotide or siRNA and were subsequently transfected with a cDNA encoding a single-chain insulin with a 7-amino acid peptide sequence (-MGGGGGM-) that links the insulin B-and A-chains and additionally carries the mutations C7S,C44S that abrogate the B7-A7 disulfide bond (28). 2 days after this transfection, the cells were pulse-labeled with 35 S-amino acids and chased in the presence or absence of 108 M MG115. No mutant insulin was secreted at any chase time (not shown). MG115 inhibited ERAD, but there was no effect of ATP2C1 knockdown on the rate or extent of ERAD of this nonglycosylated protein substrate.
Hailey-Hailey disease phenotype is restricted to the skin of heterozygous individuals, no higher eukaryotic models have been identified to date in which both functional alleles of ATP2C1 are lacking, suggesting the possibility that homozy-gotes may suffer from a more general phenotype, which may be lethal. With this in mind, it is interesting to note that most of the cell biological role(s) of the ATP2C1 gene product in the mammalian secretory pathway have yet to emerge.
In this report, immunofluorescent expression of the ATP2C1 gene product is most apparent when extracellular calcium availability is limiting (Fig. 2). Teleologically, given that the ATP2C1 encoded protein becomes less detectable when extracellular calcium is increased, it seems unlikely that ATP2C1 subserves a primary purpose in the sequestration/lowering of Ca 2ϩ from cytosolic pools (which would instead predict more ATP2C1 protein when more Ca 2ϩ enters the cytosol from the extracellular environment) but rather in uphill concentration/ accumulation of Ca 2ϩ (and potentially other divalent ions (17,19)) within the lumen of the secretory pathway. Thus, the present data strongly suggest that as keratinocytes migrate to more peripheral skin, farther from the vascular and interstitial supply of extracellular calcium, their differentiation program is likely to include strong ATP2C1 expression that facilitates proper delivery of cellular adhesive proteins via the secretory pathway (32). This may make keratinocytes especially vulnerable to secretory pathway dysfunction in states of ATP2C1 haploinsufficiency. Additionally, our data (Fig. 2) suggest that similar ion-dependent regulation of the abundance of the ATP2C1 gene product occurs in many cell types, suggesting a general homeostatic role for ATP2C1 in the secretory pathway of all cells.
To explore this role, we have engineered an siRNA that is effective in ATP2C1 knockdown in rat cells but not human cells (despite the fact that the gene products are 97% identical), providing a valuable specificity control, and we have engineered a cDNA encoding epitope-tagged rat ATP2C1 for expression in human-derived (293) cells. The construct clearly encodes a functional rATP2C1 protein (Fig. 3), which co-localizes with endogenous ATP2C1 and Golgi markers (Fig. 4). The siRNA treatment results in knockdown of ATP2C1 in rat-derived (FRT) cells but not 293 cells. Interestingly, however, expression of rat ATP2C1 in human-derived 293s22 cells renders these cells susceptible to knockdown of the endogenous gene product in trans upon treatment with the rat-specific FIG. 7. ER chaperone levels, secretion, and sensitivity to ER stress in cells with knockdown of ATP2C1. A, 293s22 cells, either expressing ATP2C1 or at 5 days after siRNA treatment, were lysed, normalized for total protein, and immunoblotted with anti-HA (upper row), anti-BiP, anti-calreticulin, anti-PDI, or anti-glycerol-3 phosphate dehydrogenase (G3PDH, used as a loading control). B, FRT cells Ϯ ATP2C1 knockdown were pulse-labeled with 35 S-amino acids and chased for 5 h as in Fig. 8. Both the cell lysates (C) and media (M) were analyzed by immunoprecipitation with anti-BiP. ATP2C1 knockdown did not increase BiP secretion from the cells. C, 293s22 cells were transfected either with single-stranded sense oligoribonucleotide or double-stranded siRNA. Prior to the 5th day post-transfection, cells were exposed overnight to the doses of tunicamycin (Tun) indicated. Cells surviving this treatment were fixed and stained with crystal violet, and the entire plate was scanned to generate the image shown. Cells pretreated with control oligo survived the ER stress, whereas after knockdown of ATP2C1, almost no cells remained after exposure to 1 g/ml tunicamycin.
FIG. 8. ER stress response pathways in cells with knockdown of ATP2C1. A, 293s22 cells Ϯ ATP2C1 knockdown were treated with 0.5 g/ml tunicamycin for the times indicated. Samples were lysed and immunoblotted for ATF6␣, eIF2␣-Ser51-phosphate, and total eIF2␣. In cells with ATP2C1 knockdown, loss of preexisting full-length ATF6␣ was comparable in kinetics and efficiency. Increased phosphorylation of eIF2␣-occurred in both sets of cells with the same timing, initially appearing in the 5-h sample. unglycos., unglycosylated. B, 293s22 cells Ϯ ATP2C1 knockdown were treated with 10 g/ml tunicamycin for the times indicated. Samples were lysed and analyzed both for loss of preexisting full-length ATF6␣ by immunoblotting and for splicing of the XBP-1 mRNA as measured by reverse transcription-PCR. siRNA (Fig. 5). This produces secretory pathway defects, including differences in N-linked glycosylation of newly synthesized thyroglobulin (Fig. 6A). Furthermore, despite the fact that the ATP2C1 gene product is localized to the Golgi complex, this pump has impact on regulation within the ER. Specifically, after ATP2C1 knockdown, we note a major defect in the ERAD of misfolded thyroglobulin encoded by the cog mutant (Fig. 6B) that occurs without secretion of cog thyroglobulin. Indeed, as mentioned in the Introduction, a mutant of the yeast homolog, pmr1, is synonymous with der5, being defective in the degradation of the ER misfolded CPY* glycoprotein. The severity of the ERAD defect after ATP2C1 knockdown may be related to the presence of N-glycans, as it is not at all detectable for a mutant form of insulin, which is nonglycosylated (Fig. 6C).
The effect of the Golgi-localized pump encoded by ATP2C1 on glycoprotein ERAD may entail several possibilities that are not mutually exclusive. First, ER-Golgi traffic in both anterograde and retrograde directions may facilitate the ERAD of misfolded secretory glycoproteins (46 -49), possibly giving substrates at least transient exposure to the Golgi ionic environment. Second, the ATP2C1 gene product might pump ions directly into the ER either when it is newly synthesized or if the pump should cycle between the Golgi and ER compartments. Third, divalent cations pumped into the Golgi lumen might be delivered by retrograde trafficking to the lumen of the ER. Certainly, yeast cells lack the SERCA class of pumps and nevertheless have ER luminal calcium that is likely to be maintained by the Golgi-localized PMR1 gene product. This supports the recent conclusion that ATP2C1, the mammalian homolog of PMR1, is at least a contributor to ER luminal calcium levels (25). Nevertheless, ATP2C1 knockdown is distinct from thapsigargin treatment (which disables SERCA pumps). Our characterization of ATP2C1 knockdown shows that it leads neither to marked UPR activation (Fig. 7A), nor does it block the ability of cells to activate UPR pathways in response to other ER stresses (Fig. 8). At the same time, the ATP2C1 gene product is important for cell survival after ER stress. This was demonstrated both in FRT cells (Fig. 7C) and in 293s22 cells (by increased annexin V staining, not shown) indicating a general finding that is also reported in yeast cells lacking PMR1 (14). Although proximal steps in ER stress signaling occur in ATP2C1-deficient cells, more work is needed to identify which if any downstream targets of UPR activation may be either hyporesponsive or hyperresponsive in ATP2C1 deficiency, thereby predisposing them to cellular toxicity.
One function of secretory pathway calcium accumulation is to provide an adequate ionic environment for ER luminal chaperones that are known calcium-binding proteins (5). ER chaperones such as BiP require calcium binding for their full function (50,51) and are necessary for their anti-apoptotic activity (2,52). Thus, based on the studies presented here, we propose that the ion pump activity of the ATP2C1 gene product in mammalian cells, like that of the PMR1 gene product in yeast cells (14), can account for its role in homeostatic maintenance of secretory pathway function, including cellular resistance to ER stress.