Sarco/endoplasmic Reticulum Calcium ATPase-2 Expression Is Regulated by ATF6 during the Endoplasmic Reticulum Stress Response

The recently described transcription factor, ATF6, mediates the expression of proteins that compensate for potentially stressful changes in the endoplasmic reticulum (ER), such as reduced ER calcium. In cardiac myocytes the maintenance of optimal calcium levels in the sarcoplasmic reticulum (SR), a specialized form of the ER, is required for proper contractility. The present study investigated the hypothesis that ATF6 serves as a regulator of the expression of sarco/endoplasmic reticulum calcium ATPase-2 (SERCA2), a protein that transports calcium into the SR from the cytoplasm. Depletion of SR calcium in cultured cardiac myocytes fostered the translocation of ATF6 from the ER to the nucleus, activated the promoter for rat SERCA2, and led to increased levels of SERCA2 protein. SERCA2 promoter induction by calcium depletion was partially blocked by dominant-negative ATF6, whereas constitutively activated ATF6 led to SERCA2 promoter activation. Mutation analyses identified a promoter-proximal ER stress-response element in the rat SERCA2 gene that was required for maximal induction by ATF6 and calcium depletion. Although this element was shown to be responsible for all of the effects of ATF6 on SERCA2 promoter activation, it was responsible for only a portion of the effects of calcium depletion. Thus, SERCA2 induction in response to calcium depletion appears to be a potentially physiologically important compensatory response to this stress that involves intracellular signaling pathways that are both dependent and independent of ATF6.

cent proteins in the ER, which can foster protein denaturation and eventual loss of function. This is known as the unfolded protein response or the ER stress response (1,2). In a few model cell types, such as HeLa, 3T3, and 293 cells, dysfunctional protein folding leads to a reduction in translation rate and increased expression of proteins that enhance the folding of nascent proteins and promote cell survival (3,4). Although these responses appear to be antithetical, they coordinate to promote cellular recovery and viability.
One of the genes induced during the ER stress response is glucose response protein GRP-78, which is targeted to the ER lumen and serves a chaperone function (5)(6)(7). A regulatory element in the GRP-78 gene, the ER stress-response element (ERSE), is required for transcriptional induction during the ER stress response. This element has been found in other ER stress-responsive genes (8).
Recently, it was shown that the transcription factor, ATF6 (10), is critical for induction of GRP-78 through the ERSE (9). Recent studies have demonstrated that the N-terminal domain of ATF6 is responsible for transcriptional activation when fused to a heterologous DNA-binding domain (11), whereas the central region of ATF6 possesses a b-Zip domain that is characteristic of some DNA-binding proteins. It has been shown in HeLa cells that full-length ATF6, composed of 670 amino acids, resides in the ER membrane; the region of ATF6 responsible for anchoring it in the ER membrane is located in the approximate center of the protein, between amino acids 378 and 398 ( Fig. 1) (12). Upon ER stress, the cytosolic, N-terminal portion of ATF6 is released from the ER via proteolysis (12). The proteases responsible for this cleavage are the same as those required for the maturation of sterol regulatory element-binding protein maturation in the ER (13). Given the consensus cleavage sequence for those proteases, it is estimated that ATF6 cleavage probably takes place within the ER transmembrane domain, most likely between leucine 390 and asparagine 391 (13). The N-terminal portion of ATF6 possesses several putative nuclear localization signals at residues 170, 330, and 380 (8). Following proteolytic cleavage, the N terminus of ATF6 translocates to the nucleus where it combines with several other proteins to form an ERSE-binding complex that is responsible for the transcriptional induction of ER stress-response proteins, such as GRP-78 (8,12). Interestingly, it appears as though ATF6 itself does not bind directly to the ERSE but forms a complex with other proteins that do (8).
Like GRP-78, most ER stress-inducible genes encode proteins that are targeted to the ER, where they presumably perform stress response-related functions. In electrically excitable cells, such as myocytes, neurons, or endocrine cells, the sarcoplasmic reticulum (SR) and/or the ER, depending on the cell type, serves a particularly important cellular function. The SR/ER is a critical storage site for the calcium that must be released at appropriate times and in a transient manner to mediate contraction, neurotransmitter, or hormone release, respectively (14 -16). This cytosolic calcium transient requires the rapid reuptake of calcium into the ER, which is usually accomplished by the sarco/endoplasmic reticulum calcium ATPase (SERCA) (reviewed in Ref. 17).
In cardiac myocytes, SERCA2 has been shown to play a central role in the contractile event (18 -19). Relatively small decreases in SERCA2 levels, which are sometimes observed in diseased hearts (20), can make considerable differences in contractile function (21,22). In fact, recent efforts have focused on overexpression of SERCA2 as a potential gene therapy for heart failure (23,24). Thus, if SR/ER calcium levels fall below functionally acceptable limits in a normal cardiac myocyte, it seems possible that the ER stress response could initiate signaling events leading to homeostatic adjustments in the levels of important calcium-handling proteins, such as SERCA2. In this way, the status of SR/ER calcium could be communicated via an intracellular signaling mechanism to other organelles, such as the nucleus, where the initial steps required for the appropriate adjustments would take place. We undertook the present study to pursue this hypothesis, with a focus on determining whether ATF6 could serve a role in cardiac myocytes as an intracellular communicator, relaying information about the status of SR/ER calcium to the nucleus.

Cell Culture
Primary ventricular myocytes cell cultures were prepared from 1-4day-old Harlan Sprague-Dawley rats as described (25,26). The apical two-thirds of the ventricle were dissected away from the atria, minced, and then washed twice with air-compatible DMEM. Isolation of cells were performed by multiple 10-min rounds of tissue dissociation with 0.001% trypsin. After each trypsin incubation, the supernatant was added to an equal volume of DMEM/Ham's F-12 (1:1) containing 20% fetal bovine serum, and all of the supernatants were combined. The cells were initially suspended in DMEM/Ham's F-12 (1:1) containing 10% fetal bovine serum and plated onto plastic culture dishes for 2 h, during which time most of the noncardiac myocytes attach to the plastic, whereas the myocytes remain in suspension. After 2 h, the suspended myocytes were collected by centrifugation and then resuspended in DMEM/Ham's F-12 (1:1) containing 10% fetal bovine serum. The cardiac myocytes were plated onto plastic culture wells that had been pretreated for at least 1 h with fibronectin at 5 g/ml of DMEM/ Ham's F-12 (1:1). Alternatively, myocytes were plated onto 2-chamber Lab-tek TM glass slides (Nalge Nunc International) that had been pretreated for at least 1 h with fibronectin at 25 g/ml of DMEM/Ham's F-12 (1:1). For immunocytofluorescence, myocytes were plated at a density of 1 ϫ 10 6 cells/Lab-tek TM glass slide chamber. For Western analyses, the myocytes were plated at a density of 1.5 ϫ 10 6 cells/35-mm plastic culture dish. For reporter analyses, the myocytes were plated at a density of 1.0 ϫ 10 6 cells/24-mm plastic culture dish.

Intracellular Calcium Measurements
Cytoplasmic free calcium was measured using indo-1, as described previously (27). Briefly, cells cultured on glass coverslips were loaded with 3 M indo-1/acetoxymethylester for 30 min at 37°C in an atmosphere of 5% CO 2 , 95% air. Following loading, the coverslips were mounted in a Biophysica Technologies (Baltimore, MD) microscope chamber that was maintained at 37°C. Fluorescence measurements were performed on a Nikon Diaphot epifluroescence microscope equipped with a 100ϫ fluor objective that was interfaced to a Photon Technologies Inc. (South Brunswick, NJ) dual emission photometry system. Illumination was provided by a 75-watt zenon lamp, with the excitation wavelength set to 355 nm (1-4-nm band pass) via a monochrometer. Fluorescence emission data at 405 and 485 nm were collected at a rate of 40 Hz and plotted as the ratio of 405/485. An aperture mechanism allowed fluorescence to be collected from a selected portion of the field that was always positioned over the cytoplasmic region of individual cells. Pacing of contractions during fluorescence measurements was performed as previously described (28).

Transfection
For transfection experiments, myocytes obtained after preplating were resuspended in serum-free DMEM/Ham's F-12 (1:1), and between 6 ϫ 10 6 and 9 ϫ 10 6 cells were combined with 12-45 g of the indicated plasmids in a total volume of 300 l. The total quantity of plasmid DNA used in each electroporation was equalized using pCMV6, when necessary. Optimal quantities of each test plasmid were determined in preliminary dose-response experiments. The cells were electroporated at 500 V, 25 microfarad, and 100 ⍀ in a 0.2-cm gap electroporation cuvette (Bio-Rad) using a Gene Pulser II (Bio-Rad). Under these conditions, cardiac myocytes are efficiently transfected, whereas any remaining nonmyocytes are not (25). Following transfection, the cells were plated at a density of 1.0 ϫ 10 6 cells/fibronectin-coated, 24-mm plastic culture dish.

Plasmids
The following plasmids were used as indicated in this study. pGL2-pGL2-Basic (catalog number E1641; Promega, Madison WI) is a vector containing luciferase reporter gene but does not contain eukaryotic promoter or enhancer sequences.
pGL2p-pGL2 promoter (catalog number E1631; Promega) is a vector containing an SV40 promoter inserted into the pGL2-Basic vector upstream of a luciferase reporter gene.
CMV-GFP-pEGFP-C1, which codes for a form of green fluorescent protein (GFP) that has been optimized for brighter fluorescence in mammalian cells driven by the CMV promoter, was obtained from CLONTECH Laboratories, Inc., Palo Alto, CA (code number 6084-1).
ANF638-Luc-This construct codes for a luciferase reporter driven by the rat ANF promoter and 638 nucleotides of the 5Ј-flanking sequence (25).
GRP78-ERSE-Luc-This construct encodes the active ERSE from the human GRP78 gene driving SV-40/luciferase. This construct was based on GRP78-ERSE-1 as identified in Ref. 9 and was prepared by annealing the following oligonucleotides: ctagcTTCACCAATCGGCGG-CCTCCACGACGGa and gatctCCGTCGTGGAGGCCGCCGATTGGTG-AAg, and by cloning the product into the NheI and BglII sites in pGL2 promoter (catalog number E1761; Promega). The restriction sites are indicated with lowercase letters.
GRP78-M-ERSE-Luc-This construct encodes a mutated form of the ERSE from the human GRP78 gene driving SV-40/luciferase. This construct was based on a mutant form of GRP78-ERSE-1, as identified in Ref. 29, that is not inducible by ATF6 or during the ER stress response and that was prepared by annealing the following oligonucleotides: ctagcTTCACCAATCGGCGGCCTTCACGACGGa and gatctCCG- TCGTGAAGGCCGCCGATTGGTGAAg, and by cloning the product into the NheI and BglII sites in pGL2 promoter (catalog number E1631; Promega). The restriction sites are indicated with lowercase letters, and mutated nucleotides are in bold and underlined.
CHOP-ERSE-Luc-This construct encodes the active ERSE from the human CHOP gene driving SV-40/luciferase. This construct was based on CHOP-ERSE-1, as identified in Ref. 29, and was prepared by annealing the following oligonucleotides: ctagcCCTACCAATCAGAAAGT-GGCACGCCGGa and gatctCCGGCGTGCCACTTTCTGATTGGTAG-Gg, and by cloning the product into the NheI and BglII sites in pGL2 promoter (catalog number E1631; Promega). The restriction sites are indicated with lowercase letters.
CHOP-M-ERSE-Luc-This construct encodes a mutated form of the ERSE from the human CHOP gene driving SV-40/luciferase. This construct was based on a mutant form of CHOP-ERSE-1, as identified in Ref. 29, that is not inducible by ATF6 or during the ER stress response and that was prepared by annealing the following oligonucleotides: ctagcCCTACCAGTCAGAAAGTGGCACGCCGGa and gatctCCGGCGT-GCCACTTTCTGCTTGGTAGGg, and by cloning the product into the NheI and BglII sites in pGL2 promoter (catalog number E1631; Promega). The restriction sites are indicated with lowercase letters, and the mutated nucleotides are in bold and underlined.
SERCA-ERSE-1-Luc-In the rat SERCA gene, ERSE-1 is located between positions Ϫ78 and Ϫ60, compared with the transcription start site (30). This construct encodes rat SERCA gene sequences from Ϫ82 to Ϫ56 gene driving SV-40/luciferase and was prepared by annealing the following oligonucleotides: ctagcTCGGCAATGAGCGGCGTCCA-CATGCCa and gatctGGCATGTGGACGCCGCTCATTGCCGAg, and by cloning the product into the NheI and BglII sites in pGL2 promoter (catalog number E1631; Promega).
SERCA-ERSE-1-M-Luc-This construct encodes a mutant form of ERSE-1 from the rat SERCA gene driving SV-40/luciferase and was prepared by annealing the following oligonucleotides: ctagcTCG-GCAATGAGCGGCGTTACAATGCCa and gatctGGCATTGTAACGC-CGCTCATTGCCGAg, and by cloning the product into the NheI and BglII sites in pGL2 promoter (catalog number E1631; Promega). The restriction sites are indicated with lowercase letters, and the mutated nucleotides are in bold and underlined.
SERCA-ERSE-2-Luc-In the rat SERCA gene, ERSE-2 is located between positions Ϫ464 and Ϫ446, compared with the transcription start site (30). This construct encodes rat SERCA gene sequences from Ϫ468 to Ϫ442 driving SV-40/luciferase and was prepared by annealing the following oligonucleotides: ctagcGACACCAGTCCCTGCGCGCCTC-GGTCCa and gatctGGACCGAGGCGCGCAGGGACTGGTGTCg, and by cloning the product into the NheI and BglII sites in pGL2 promoter (catalog number E1631; Promega).
SERCA-ERSE-3-Luc-In the rat SERCA gene, ERSE-3 is located between positions Ϫ999 and Ϫ981, compared with the transcription start site (30). This construct encodes rat SERCA gene sequences from Ϫ1003 to Ϫ976 driving SV-40/luciferase and was prepared by annealing the following oligonucleotides: ctagcTCCGTGTGGCCTCTTTATATTG-GCAAAa and gatctTTTGCCAATATAAAGAGGCCACACGGAg, and by cloning the product into the NheI and Bgl II sites in pGL2 promoter (catalog number E1631; Promega).
SERCA334-M-Luc-This construct codes for a form of SERCA344-Luc where the CCACA sequence within SERCA344-Luc ERSE-1, located at positions Ϫ78 to Ϫ74, has been mutated to TACAA (see Figs. 6C and 8A). This mutation was created using a QuickChange sitedirected mutagenesis kit (Stratagene) using the following primers: CGGCCAATGAGCGGCGTTACAATGCCGCGGCGGCGGCG and CGCCGCCGCCGCGGCATTGTAACGCCGCTCATTGGCCG, where the mutated nucleotides are in bold and underlined.
FLAG-ATF6(670), FLAG-ATF6(373), FLAG-ATF6(366), and FLAG-ATF6(273)-A version of pcDNA3.1(Ϫ) containing a 3ϫ FLAG sequence was constructed by annealing the following oligonucleotides: ctagcGC-CATGGACTACAAAGACCACGACGGTGATTATAAAGATCACGATA-TCGATTACAAGGATGACGATGACAAGt and ctagaCTTGTCATCGTC-ATCCTTGTAATCGATATCGTGATCTTTATAATCACCGTCGTGGTC-TTTGTAGTCCATGGCg. The NheI and XbaI overhangs of the annealed oligonucleotides were used to ligate the product into the XbaI site of pcDNA3.1(Ϫ) (Promega), resulting in the creation of 3ϫ FLAG-pcDNA3.1. Construction of 3ϫ FLAG-ATF6(1-670) began with a 1ϫ FLAG-ATF6(1-670) construct that was provided by Dr. Ron Prywes (Columbia University, New York, NY). Using 1-FLAG-ATF6(1-670) as the template, PCR was carried out to introduce an XbaI site at the first amino acid of ATF6; the resulting product was then ligated into the XbaI site of the 3ϫ FLAG-pcDNA3.1. The 3ϫ FLAG-ATF6(1-373) and 3ϫ FLAG-ATF6(1-366) were created by introducing and XhoI site at amino acid 1, as described above, and by introducing a termination codon and SacI site in the PCR product and then ligating the product into the 3ϫ FLAG vector. The 3ϫ FLAG-ATF6(1-273) was created by introducing an XhoI site at amino acid 1, as described above, and then cutting the PCR product with XhoI and KpnI, the latter of which is a native restriction site located at amino acid 273. The 3ϫ FLAG-ATF6 constructs were prepared to optimize cross-reactivity with the anti-FLAG antibody and, thus, enhance the sensitivity of immunocytofluorescence.
Luciferase-After cell lysis and removal of cell debris by centrifugation, as described above, 200-l samples of cell lysate were combined with 100 l of luciferase buffer (25 mM Gly-Gly, pH 7. Calcium Depletion Maneuver-Cultured cardiac myocytes were dissociated and in some cases transfected with pGL2p (Control) or with a test luciferase construct and pCH110 (SV40-␤-gal). Cultures were then maintained for 48 h in DMEM containing 10% fetal calf serum. The cultures were then switched to a calcium depletion medium, consisting of calcium-free medium (DMEM catalog number 21068; Life Technologies, Inc.) containing 10% fetal calf serum, 584 mg/liter L-glutamine, 110 mg/liter sodium pyruvate, 1 mM EGTA, and 10 g/ml ryanodine. Control cultures were maintained in normal calcium medium, consisting of DMEM (catalog number 11960; Life Technologies, Inc.) containing 10% fetal calf serum, 584 mg/liter L-glutamine, and 110 mg/liter sodium pyruvate. Following these treatments, the cultures were extracted and analyzed for luciferase and ␤-galactosidase activities. The relative luciferase (Luc/␤-Gal) values that were obtained for each luciferase test construct following a normal calcium or calcium depletion treatment were then divided by the relative luciferase values that were obtained for pGL2p (control) following normal calcium and calcium depletion treatments, respectively. Thus, by default, all of the relative luciferase values obtained for pGL2p (control) were set to 1.0 for normal calcium and for calcium depletion, and all other values were compared with these as percentages of the control.

Immunocytofluorescence
Cardiac myocytes were transfected with pEGFP-C1 and either 3ϫ FLAG-ATF6(670) or 3ϫ FLAG-ATF6(373) and then plated onto glass slides. The cultures were maintained for 24 h in serum-containing medium and then treated for 4 h with N-acetyl-Leu-Leu-Nle-CHO (catalog number 208719; Calbiochem, La Jolla, CA) to inhibit proteosomal ATF6 degradation, which is necessary to visualize ATF6 by either Western analysis or immunocytofluorescence (13). The cultures were then submitted to calcium depletion for 4 h, as described above, in the presence of N-acetyl-Leu-Leu-Nle-CHO, which we and others have shown has no negative effects on gene induction by ER stress (13). The cultures were then washed twice with phosphate-buffered saline, then fixed with 4% paraformaldehyde for 30 min, permeablized with 0.2% Triton X-100 for 10 min, and then incubated for 16 h at 4°C with anti-FLAG monoclonal antibody, M2 (Sigma). The slides were then washed three times with phosphate-buffered saline and then incubated with goat anti-mouse Texas Red (Molecular Probes, Inc., OR). Transfected (GFP-positive) cells were viewed using a 63ϫ oil immersion objective on a Leica SP2 laser scanning confocal microscope.

RESULTS
In initial experiments, the effects of various forms of ATF6 on selected reporter constructs were examined. Because the cardiac gene encoding ANF is induced in cardiac myocytes in response to some stresses, the effect of ATF6 on a construct harboring 638 nucleotides of the ANF promoter driving luciferase (ANF638-Luc) expression was assessed. Interestingly, none of the ATF6 expression constructs increased ANF promoter activity from this construct ( Fig. 2A). As a control, it was shown that the activated form of MAP kinase kinase 6, MKK6(Glu), which is known to stimulate the p38 stress kinase family in cardiac myocytes (26), induced ANF promoter activation by more than 20-fold, as demonstrated in previous studies (26). In contrast to the results obtained with ANF638-Luc, two of the ATF6 expression constructs tested, ATF6(1-373) and ATF6(1-366), enhanced luciferase production from reporter genes harboring known ERSEs that were derived from two well characterized ER stress-inducible genes, GRP-78 (GRP-78-ERSE/Luc) and CHOP (CHOP-ERSE/Luc) (Fig. 2, B and C). Neither ATF6(1-670) or ATF6(1-273) was able to confer activation of the ERSE-containing reporters; however, these results were anticipated, because ATF6(1-670) is retained in the ER by the transmembrane domain ( Fig. 1 and Ref. 12), and ATF6(1-273), which would not be retained in the ER and would localize to the nucleus, lacks the leucine zipper domain required for forming a complex at the ERSE. The specificity of this response was confirmed by introducing single point mutations in both the GRP-78 and CHOP ERSEs, which would be predicted to block ATF6 binding to the ERSE complex (32,33). Both of these mutated reporters, GRP78-M-Luc and CHOP-M-Luc, displayed significantly reduced luciferase expression in response to ATF6(1-373) and ATF6(1-366) (Fig. 2). These results indicated that both ATF6(1-373) and ATF6(1-366) confer reporter induction through the native ERSE sequences in these constructs. Interestingly, MKK6(Glu) had no effect on reporter induction from either GRP-78-ERSE-Luc or CHOP-ERSE-Luc (Fig. 2), demonstrating that MKK6-activated p38 cannot mediate gene activation through these ERSEs. Taken together, these results indicate that a reporter activation profile similar to that observed using GRP78-ERSE-Luc or CHOP-ERSE-Luc can serve as an accurate indicator of ATF6-inducible, SR/ER stress-response genes in cardiac myocytes.
To test whether expression of the SERCA2 gene is responsive to SR/ER stress in cardiac myocytes, the abilities of the ATF6 expression constructs to activate luciferase driven by 3.2 kb of the rat SERCA2 promoter (SERCA3258-Luc) (31) were assessed. The ATF6 SERCA induction profile mirrored that observed for the prototype GRP78-ERSE-Luc and CHOP-ERSE-Luc reporter constructs (Fig. 2D). SERCA promoter activity was unaffected by ATF6(1-670), or ATF6(1-273) but was activated by about 5-10-fold by either ATF6(1-373) or ATF6 . In comparison with GRP78-ERSE-Luc and CHOP-ERSE-Luc but in contrast to ANF638-Luc, SERCA3258-Luc was not induced by MKK6-activated p38. These results indicate that in cardiac myocytes, the SERCA promoter can be activated by ATF6, as would be expected for a myocardial cell SR/ER stressresponse gene. Thus, it is possible that in cardiac myocytes the SERCA gene might be induced under stress conditions, such as a reduction of SR/ER calcium.
Accordingly, SR/ER calcium was depleted using a previously published procedure where cardiac myocytes are maintained in calcium-free medium with EGTA and ryanodine (34). In initial studies it was determined that treating cardiac myocytes for up to 16 h with this calcium depletion maneuver did not result in a significant change in the numbers of viable cells or the quantity of protein in each culture (not shown). However, this maneuver effectively abolished calcium transients and unloaded the SR of calcium, as demonstrated by the lack of calcium transients and the relatively small release of calcium upon treatment with caffeine (Fig. 3A). Thus, the maneuver was adopted as a method to reduce SR and ER calcium in cardiac myocytes.
To characterize whether calcium depletion could activate ATF6, immunocytofluorescence was carried out to localize ATF6 before and after the treatment. It was shown that ATF6(373) was constitutively localized to the nucleus, as expected (12) (Fig. 3, B and C). However, ATF6(670) displayed a diffuse staining pattern under normal conditions, consistent with an SR/ER localization (Fig. 3D) but translocated to the nucleus following calcium depletion (Fig. 3E). This indicated that the calcium depletion maneuver was capable of fostering ATF6 translocation in a manner consistent with a role for ATF6 in mediating SERCA induction during SR/ER stress. The effects of calcium depletion on the induction of reporter genes from various test constructs were evaluated. As expected, calcium depletion resulted in no apparent induction of luciferase from a construct that lacks an ERSE (SV-40/Luc) (Fig. 4A, Con). However, calcium depletion induced luciferase expression from GRP78-ERSE-Luc by about 4 -5-fold (Fig. 4A,  GRP78-ERSE, compare open and filled bars). Although of lower magnitude, a similar result was observed when CHOP-ERSE-Luc was tested (Fig. 4A, CHOP-ERSE). The lack of luciferase induction from the ERSE mutants GRP78-M-ERSE-Luc or CHOP-M-ERSE-Luc indicated that the transcriptional activation observed upon calcium depletion possessed a profile similar to that observed upon ATF6 overexpression (see Fig. 2). Thus, the calcium depletion maneuver resulted in cellular re-FIG. 3. Effects of calcium depletion on intracellular calcium and subcellular distribution of ATF6. A, intracellular calcium measurements. Cardiac myocytes were plated onto glass coverslips and maintained for 48 h in serum-containing medium. The cultures were then maintained in medium fostering normal (Normal Ca) or depleted intracellular calcium (Ca Depleted) for 16 h, as described under "Materials and Methods." The cells were then analyzed for intracellular calcium, also as described under "Materials and Methods." In this experiment, the normal calcium and calcium-depleted cells were paced to contract during the first 30 s of each recording. The stimulator was then shut off, and the slides were incubated with 20 mM caffeine to elicit calcium release from the SR, as reported previously (46). Shown are traces of two cells that produced results that were representative of least 10 cells submitted to each treatment. Further experiments indicated that this maneuver effected complete calcium depletion within 15 min (not shown). B-E, ATF6 immunocytofluorescence. Cultured cardiac myocytes were co-transfected with CMV-GFP (pEGFP-C1) and either 3ϫ FLAG-ATF6(1-373) or 3ϫ FLAG-ATF6(1-670), as shown, and then maintained in serum-containing medium for 48 h. The cultures were then switched to medium fostering normal calcium or calcium-depleted conditions for 4 h, as described under "Materials and Methods." The cultures were then fixed and stained using a FLAG antibody, as described under "Materials and Methods." Transfected cells were identified by green fluorescence and then viewed under red fluorescence and photographed. The bar in B represents 16 m.

FIG. 4. Effects of calcium depletion on gene induction.
A, cardiac myocytes were transfected with pGL2p (Con), GRP78-ERSE-Luc, GRP78-M-ERSE-Luc, CHOP-ERSE-Luc, CHOP-M-ERSE-Luc, SERCA3258-Luc, or ANF638-Luc, as shown, and SV40-␤-galactosidase (pCH110) and maintained in serum-containing medium for 48 h. The cultures were then switched to medium fostering normal calcium or calcium-depleted conditions for 16 h, as described under "Materials and Methods." The cultures were then extracted, reporter assays were carried out, and the relative luciferase values of each test construct compared with pGL2p (Con) were plotted, as described under "Materials and Methods" under "Calcium Depletion Maneuver." Shown are the mean values Ϯ S.E. for triplicate cultures. B, duplicate cultures of cardiac myocytes were maintained in serum-containing medium for 48 h after dissociation and then switched to medium fostering normal calcium or calcium-depleted conditions for the times shown, as described under "Materials and Methods." The cultures were then extracted and following fractionation by SDS-PAGE, the samples were submitted to Western blotting using a SERCA2-or an ␣-actininspecific antibody, as described under "Materials and Methods." C, the bands from the Western blot shown in B were quantified on a densitometer using MD ImageQuant, and the mean densities Ϯ S.D. were plotted. This experiment was performed at least three times, and representative results are shown. sponses possessing the hallmark features of the ER stress response. Calcium depletion resulted in an approximate 6-fold increase in luciferase in cells transfected with SERCA3258-Luc (Fig. 4A, SERCA), which was consistent with the hypothesis that SERCA is an SR/ER stress-responsive gene in cardiac myocytes. As a control, it was shown that there was no detectable luciferase induction from ANF638-Luc upon calcium depletion (Fig. 4A, ANF), consistent with the absence of any ERSEs in the ANF promoter.
To evaluate whether calcium depletion has an effect on endogenous SERCA levels, Western blot analyses were carried out. Within 4 h of calcium depletion there was a clear increase in the level of SERCA2 protein to about 2.5-fold over control; this level of SERCA2 expression was maintained through 16 h of treatment (Fig. 4, B and C). These results are consistent with those obtained using the SERCA2 promoter and indicate that calcium depletion can induce SERCA2 transcription and lead to increased quantities of SERCA2 protein, demonstrating that SERCA2 gene expression is responsive to SR/ER stress in cardiac myocytes.
It has previously been shown that introducing mutations into the putative DNA-binding domain of ATF6(373), such as to change amino acids 315-317 from KNR to TAA, can result in a form of ATF6 that acts in a dominant-interfering manner on ER stress-inducible genes in HeLa cells (35). Accordingly, the effects of ATF6 possessing this mutation were assessed in cardiac myocytes subjected to calcium depletion. Dominantnegative ATF6 partially blocked SERCA2 promoter induction and luciferase induction from GRP78-ERSE-Luc in response to calcium depletion (Fig. 5). These results indicated that in calcium-depleted cardiac myocytes, ATF6 is essential for maximal gene induction through the GRP78 ERSE and through the SERCA promoter. Accordingly, further experiments were carried out to identify the element(s) within the SERCA promoter responsible for induction by calcium depletion.
The rat SERCA promoter has three putative ERSEs located at positions Ϫ999, Ϫ464, and Ϫ78 relative to the transcriptional start site (Fig. 6A). A series of truncation constructs was prepared and evaluated for inducibility by ATF6(373) or by calcium depletion. Interestingly, the SERCA2 promoter could be truncated down to position Ϫ334 and retain full inducibility in response to either stimulus, indicating that the putative ERSEs located at Ϫ999 and Ϫ464 were not critical for SERCA2 induction in response to SR/ER stress (Fig. 6, B and C). However, truncating the SERCA2 promoter further to remove the ERSE located at position Ϫ78 led to a complete loss of reporter induction, suggesting a requirement for ERSE-1 (Fig. 6A).
To confirm the critical role of ERSE-1 in SERCA2 promoter induction, the abilities of the isolated SERCA2 ERSEs 1, 2, and 3 to confer reporter induction in response to ATF6(373) or calcium depletion were determined. Each SERCA2 ERSE (Fig.  7A) was cloned into pGL2p to create SERCA-ERSE-1-Luc, SERCA-ERSE-2-Luc, and SERCA-ERSE-3-Luc, which are comparable with GRP-78-ERSE-Luc. As expected, both the GRP-78-ERSE and the SERCA-ERSE-1 were inducible by ATF6, whereas the mutant forms of these displayed essentially no inducibility (Fig. 7B). The putative SERCA-ERSE-2 and SERCA-ERSE-3 were not ATF6-inducible, which is consistent with the truncation analyses. The GRP-78-ERSE conferred a ϳ3-fold induction of reporter expression following calcium depletion, as demonstrated in earlier experiments (Figs. 4 and 5), and the mutant form of GRP-78-ERSE was not induced, as expected. However, the construct containing the isolated SERCA-ERSE-1 did not confer any induction following calcium depletion (Fig. 7C). Thus, although the isolated GRP-78-ERSE possessed all the information capable of conferring induction in response to both ATF6 and calcium depletion, the isolated SERCA-ERSE-1 was only responsive to ATF6 and apparently lacked the information needed to respond to calcium depletion. Thus, it seemed possible that in addition to the ERSE-1, other sequences within SERCA334-Luc might be required to confer full inducibility of the SERCA promoter in response to calcium depletion. To test this hypothesis, ERSE-1 in SERCA334-Luc was mutated in a manner predicted to disrupt induction by ATF6 (Fig. 8A). Compared with the wild type SERCA334-Luc, the mutant construct SERCA334-M-Luc displayed a nearly complete loss of induction in response to ATF6(373) (Fig. 8B), as expected, but only a partial decrease of induction following calcium depletion (Fig. 8C). These results indicate that even though ERSE-1 is essential for ATF6-mediated SERCA2 induction, it is only partially responsible for induction following calcium depletion. Sequences outside of ERSE-1 are therefore required for maximal inducibility by calcium depletion, and these sequences may mediate this induction independently of ATF6.
It seemed possible that calcium depletion can activate stressresponsive genes through both ATF6-dependent and -independent mechanisms, whereas other forms of SR/ER stress, which do not result in SR/ER calcium depletion, might use primarily ATF6-dependent pathways. To test this hypothesis, tunicamycin was used as a treatment that would activate the SR/ER stress (unfolded protein) response by inhibiting N-linked protein glycosylation but would presumably not affect SR/ER calcium. As expected, tunicamycin did not alter cardiac myocyte calcium transients or the ability of caffeine to release calcium from SR/ER stores (not shown). Also as expected, tunicamycin was an effective activator of SERCA334-Luc induction, conferring an approximately 7-fold increase in luciferase expression compared with untreated cells (Fig. 9A). However, in contrast to calcium depletion, tunicamycin was unable to effect any reporter induction from SERCA334-M-Luc. Tunicamycin effectively activated reporter expression from GRP-78-ERSE-Luc, and in contrast to calcium depletion, tunicamycin increased reporter expression from SERCA-ERSE-1-Luc by about 2-fold (Fig. 9B). Thus, the isolated SERCA ERSE-1 is able to respond to ATF6 and tunicamycin but not to calcium depletion, consistent with the hypothesis that in the SERCA2 promoter, ERSE-1 confers only a portion of the gene induction in response to SR/ER calcium reduction.

DISCUSSION
This study has shown that in cardiac myocytes, the depletion of calcium in the SR, and probably the ER as well, activates a cellular stress response that is similar to the ER or unfolded protein stress response observed in other cell types (2). Importantly, the findings in this study demonstrate that ATF6, which was recently found to play a critical role in the ER stress response in other cell types (9,12,33,35), also plays a central role in mediating the induction of the SERCA2 gene in cardiac myocytes following SR/ER calcium depletion.
It is possible that the response of cardiac myocytes to SR/ER calcium depletion allows the cells to compensate for a stress that could severely reduce contractile ability. Indeed, the increased SERCA2 expression observed in this study would constitute such a compensatory response. In vivo, changes in SR/ER calcium would most likely be relatively small; however, it seems possible that even these small changes could be "sensed" by this intracellular signaling system involving ATF6. This would then drive moderate, compensatory increases in SERCA2 expression, which could be sufficient to restore proper levels of calcium to the SR and ER, the end effect of which would be to restore optimal contractility. Consistent with the findings in the present report is a recent study demonstrating that ER stress, such as reduction of protein N-glycosylation or reduction of ER calcium, can lead to increased levels of SERCA mRNA and protein in PC-12 cells (36).
The mechanism of ATF6-dependent gene induction during the ER stress and unfolded protein responses has been intensively investigated recently in several labs. The canonical ERSE through which ATF6 can enhance gene expression bears the sequence CCAATX 9 CCAC(G/A) (9). However, ATF6 binding to such sequences requires the presence of other DNAbinding proteins, such as NF-Y/CBF, YY-1, and ERSE factor (32). Apparently, NF-Y, also known as CBF, binds to the CCAAT box of the ERSE site, and ATF6 interacts with NF-Y while also making contact with the CCAC(G/A) sequences of the ERSE. YY-1 can also interact with ATF6 and the canonical ERSE, along with a protein known as the ERSE factor. In a previous study (36), sequence analysis showed that there are three conserved putative ERSE motifs in the 1.5-kb 5Ј-flanking sequence of the SERCA promoter from human, mouse, rat, and rabbit SERCA2. In the rat SERCA2 promoter, both ERSE-1 and ERSE-2, but not ERSE-3, possess the CCAAT and the CCAC(G/A) motifs that are required for ER stress inducibility (Fig. 7). Moreover, neither ERSE-2 or ERSE-3 possess the CCG motif in the core region that is required for ERSE factor binding (32). Accordingly, it was implied by Caspersen et al. (36) that ERSE-1 might be responsible for SERCA2 induction in response to ER stress. This prediction was borne out in the present study, where it was demonstrated that ERSE-1 plays a critical role in SERCA2 gene induction by the ER stress, tunicamycin, and a truncated, constitutively active form of ATF6, ATF6(1-373). It has been shown that the isolated bZIP domain of ATF6 can bind directly to DNA (35), indicating the existence of an element within some genes that could potentially bind ATF6 in the absence of other DNA-binding proteins. In that study, the consensus DNA sequence for this ATF6-binding site was shown to be TGACGTG(G). However, there are no TGACGTG(G) sequences in the known 3258 nucleotides of the rat SERCA2 5Ј-flanking sequence used in the present study, indicating that it is unlikely that ATF6 could bind directly to any regions of the SERCA2 promoter. Accordingly, it appears that SERCA2 gene induction in response to ATF6 takes place through the ERSE-1 and, therefore, most likely requires other accessory DNA-binding proteins, such as CBF and YY-1.
The mechanism of ATF6-independent gene induction during the ER stress response remains to be determined. In the present study, the increase in SERCA2 expression following tunicamycin treatment was fully dependent on ERSE-1 and ATF6. In contrast, the increase in SERCA2 expression following calcium depletion was only partly dependent on ERSE-1 and ATF6. Evidence for this was that SERCA334-M-Luc, which harbored a mutated ERSE, displayed an only a 50% reduction in activity compared with the native SERCA334-Luc construct FIG. 6. Truncation mutation analysis of SERCA promoter activation by ATF6 and by calcium depletion. A, rat SERCA2 gene. Shown is a diagram of the rat SERCA2 5Ј-flanking sequence from Ϫ3258 to ϩ78 with the locations of the truncation mutants and the putative ERSEs. B, effect of ATF6 on SERCA2 truncation mutants. Cardiac myocytes were transfected with pGL2 (Con) or one of the truncated forms of SERCA2-luciferase shown and SV40-␤-galactosidase (pCH110). The cultures were also transfected with either 3ϫ FLAG-pcDNA3.1 (Vector Con), or with 3ϫ FLAG-ATF6(1-373) (ATF6(373)). After 48 h in serum-free medium, the cultures were extracted, the reporter assays were carried out, and the relative luciferase values were plotted as fold of pGL2 (fold of control). C, effect of calcium depletion on SERCA truncation mutants. Cardiac myocytes were transfected with pGL2 (Con) or one of the truncated forms of SERCA2-luciferase shown and SV40-␤-galactosidase (pCH110). The cultures were then maintained in serum-containing medium for 48 h and then switched to medium fostering normal calcium or calcium-depleted conditions for 16 h, then treated, and analyzed as described in the legend to Fig. 4A. following calcium reduction but lost all ability to be activated in response to ATF6 (Fig. 8). It therefore appears that the elements of the SERCA2 gene that mediate ATF6-independent induction lie within the Ϫ334 to ϩ78 region of the 5Ј-flanking sequence of that gene.
It is possible that ERSE-1 and ATF6-independent induction of SERCA2 following calcium depletion involves non-ATF6related signaling pathways that are activated by ER stress. A number of kinases are activated upon ER stress. For example, in yeast, a transmembrane, ER-resident protein kinase, IRE1p, is activated by ER stress and is thought to transmit stress signals from the ER to other locations, such as ribosomes where translation is reduced by ER stress (2,37). Two mammalian homologs of IRE1p, IRE1␣ and IRE1␤, have been identified (38,39), and recent findings indicate that in addition to transmitting ER stress to the translational machinery, as does IRE1p, these IREs dimerize upon following ER stress and that their cytoplasmic domains, which then mimic the oligomerized cytosolic domains of cell surface cytokine and growth factor receptors, can activate the stress mitogen-activated protein kinase, c-Jun N-terminal kinase (40). This c-Jun N-terminal kinase activation apparently requires the binding of the adaptor protein, TRAF2, to the dimerized, activated IREs. Because TRAF2 and perhaps other adaptor proteins could potentially bind to activated IREs, it is possible that through this mechanism other TRAF2-mediated events could also be activated and then participate in SR/ER stress-response gene induction. For example, NFB has been shown to be activated by TRAF2 (41), and preliminary studies from our lab have demonstrated that SR/ER calcium depletion can activate NFB-dependent gene expression in cardiac myocytes (not shown). Moreover, there exist several putative NFB binding sites within SERCA-334 that could potentially serve as elements that might mediate SERCA2 induction in an ATF6-independent manner in response to SR/ER calcium depletion. Other kinases, such as protein kinase R and PERK, a protein kinase R homolog, are activated by ER stress (42)(43)(44). Protein kinase R and PERK are known to phosphorylate the ribosomal elongation factor, eIF-2␣, and thus contribute to translational inhibition (44,45), but it is also possible that they could have alternate roles in mediating ATF6-independent SERCA2 induction.
In conclusion, the present study has demonstrated that in cardiac myocytes, the depletion of SR/ER calcium activates SERCA2 expression. At least a portion of this activation requires ATF6 and is mediated by a canonical ERSE, ERSE-1, located at position Ϫ78 in the rat SERCA2 promoter. However, significant SERCA2 induction by calcium depletion occurs independently of ATF6 and ERSE-1. The mechanism by which calcium depletion can lead to SERCA2 promoter activation in FIG. 7. Analysis of the induction of isolated ER stress-response elements from the GRP78 and SERCA genes by ATF6 and by calcium depletion. A, ERSE sequences. Shown is a diagram of the ERSEs from the GRP78 gene and the three putative ERSEs from the rat SERCA2 gene. The bold and underlined letters of each sequence represent the boundaries of each ERSE. Also shown are the nucleotides in each ERSE that were altered (circled and italic) in creating mutants that were predicted to lose responsiveness to ER stress. B, effect of ATF6 on isolated ER stress elements. Cardiac myocytes were transfected with pGL2 (Con) or one of the GRP78-ERSE-luciferase or SERC-A2-luciferase constructs shown and SV40-␤-galactosidase (pCH110). Cultures were also transfected either with 3ϫ FLAG-pcDNA3.1 (Vector Con) or with 3ϫ FLAG-ATF6(1-373) (ATF6(373)). The cultures were then treated and analyzed as described in the legend to Fig. 6B. C, effect of calcium depletion on isolated ER stress elements. Cardiac myocytes were transfected with pGL2p (Con) or one of GRP78-ERSE-luciferase or SERCA2-luciferase constructs shown and SV40-␤-galactosidase (pCH110). The cultures were then maintained in serum-containing medium for 48 h and then switched to normal calcium or calcium-depleted conditions for 16 h, then treated, and analyzed as described in the legend to Fig. 4A.   FIG. 8. Point mutation analysis of SERCA promoter activation by ATF6 and by calcium depletion. A, diagram of rat SERCA334-Luc constructs. Shown is a diagram of the wild type and mutated forms of rat SERCA334-Luc. In the region of the rat SERCA2 gene spanning positions Ϫ334 to ϩ78, there is only one ERSE, ERSE-1, which is located at position Ϫ78. Shown is the mutation in ERSE-1 that was prepared in a manner expected to render it inactive in the ER stress response; bold letters indicate the boundaries of ERSE-1, and underlined letters represent the mutated positions. B, effect of ATF6 on wild type and mutated rat SERCA334-Luc. Cardiac myocytes were transfected with pGL2 (Con) or one of the SERCA2-luciferase constructs shown and SV40-␤-galactosidase (pCH110). The cultures were also transfected with either 3ϫ FLAG-pcDNA3.1 (Vector Con) or with 3ϫ FLAG-ATF6(1-373) (ATF6(373)). The cultures were then treated and analyzed as described in the legend to Fig. 6B. C, effect of calcium depletion on wild type and mutated rat SERCA334-Luc. Cardiac myocytes were transfected with pGL2p (Con) or one of the SERCA2-luciferase constructs shown and SV40-␤-galactosidase (pCH110). The cultures were then maintained in serum-containing medium for 48 h and then switched to medium fostering normal calcium or calcium-depleted conditions for 16 h, then treated, and analyzed as described in the legend to Fig. 4A. an ATF6/ERSE-independent manner remains to be determined, but preliminary results suggest a role for NFB in this process. The present study is the first to demonstrate the ER stress response in cardiac myocytes and to show that a downstream target of this response, SERCA2, is induced in a manner consistent with a compensatory, physiological role in maintaining optimal myocardial cell contractility during times of reduced SR/ER calcium. Future studies focusing on the other mechanisms by which SERCA2 is induced during calcium depletion and on the identities of other cardiac specific genes that are responsive to this form of stress will reveal interesting new genetic programs that could be critical for maintaining optimal cardiac cell function during times of stress. FIG. 9. Effects of tunicamycin on SERCA induction and on induction through isolated GRP-78-and SERCA ER stress-response elements. A, effect of tunicamycin on SERCA334-Luc. Cardiac myocytes were transfected with pGL2p (Con) or one of the SERCA2luciferase constructs shown and SV40-␤-galactosidase (pCH110). After 48 h in serum-free medium, the cultures were treated for 16 h with or without tunicamycin (TN) (2 g/ml) and then extracted, the reporter assays were carried out, and the relative luciferase values were plotted as fold of pGL2 (fold of control). B, effect of tunicamycin on isolated GRP78 and SERCA2 ER stress elements. Cardiac myocytes were transfected with pGL2 (Con) or one of the GRP78-or SERCA2-luciferase constructs shown and SV40-␤-galactosidase (pCH110). After 48 h in serum-free medium, the cultures were treated for 16 h with or without tunicamycin (2 g/ml) and then extracted, the reporter assays were carried out, and the relative luciferase values were plotted as fold of pGL2p (fold of control).