The Sarco/Endoplasmic Reticulum Calcium-ATPase 2b Is an Endoplasmic Reticulum Stress-inducible Protein*

The sarco/endoplasmic reticulum calcium-ATPase (SERCA) translocates Ca2+ from the cytosol to the lumen of the endoplasmic reticulum. This Ca2+ storage is important for cellular processes such as calcium signaling and endoplasmic reticulum (ER)-associated posttranslational protein modifications. We investigated the expression of the SERCA2 and SERCA3 isozymes in PC12 cells exposed to agents interfering with different aspects of the posttranslational protein processing within the ER, thereby activating the ER stress-induced unfolded protein response (UPR). All agents increased the SERCA2b mRNA level 3–4-fold, in parallel with increasing mRNA levels for the ER stress marker proteins BiP/GRP78 and CHOP/GADD153. In contrast, SERCA3 mRNA levels did not change. SERCA2b mRNA stability was not changed, indicating that the mechanism of its up-regulation was transcriptional, in accordance with the presence of ER stress response elements in the promoter region of the SERCA2 gene. SERCA2b was also increased at the protein level upon ER stress treatments. Induction of ER stress by tunicamycin, dithiothreitol, or l-azetidine 2-carboxylic acid did not result in depletion of ER calcium, showing that such depletion was not necessary for up-regulation of SERCA2b expression or UPR activation in general. We conclude that the SERCA2b expression can be controlled by the UPR pathway independently of ER Ca2+depletion.

The Ca 2ϩ storage of endoplasmic reticulum (ER) 1 is crucial for a variety of Ca 2ϩ -dependent processes, such as Ca 2ϩ -mediated cytosolic signaling in response to external stimuli (1), cell growth, and proliferation (2,3), as well as synthesis, posttrans-lational processing, folding and export of proteins synthesized on ER-associated ribosomes (4 -6). The maturation of the newly translocated polypeptide chains within the ER lumen is accomplished by transient interactions with ER chaperones (such as calreticulin, calnexin, BiP/GRP78, GRP94, and others) (7,8). The requirement for Ca 2ϩ as a co-factor for proper ER protein processing and maturation presumably results from the Ca 2ϩ binding capacities and Ca 2ϩ -dependent interactions of the chaperones in the ER lumen (9).
The resting intra-ER free Ca 2ϩ concentration is 3-4 orders of magnitude higher than the cytosolic Ca 2ϩ concentration ([Ca 2ϩ ] i ) (10). This gradient is generated by the sarco/endoplasmic reticulum Ca 2ϩ -ATPases (SERCAs), encoded by three homologous genes (SERCA1, SERCA2, and SERCA3) (11). Transcripts from all three genes undergo alternative splicing in a developmentally and tissue-specific manner, giving rise to a number of isozymes differing in their C-terminal regions. The splicing of the primary transcripts from SERCA1 and SERCA2 genes produces SERCA1a-b (12) and SERCA2a-b isozymes (13,14). SERCA1a-b and SERCA2a expression is restricted to skeletal and cardiac muscles, whereas the SERCA2b isoform is ubiquitously expressed in all nonmuscle tissues (15). Recently, SERCA3 transcripts were found to be spliced too, resulting in three isozymes in mice and humans (SERCA3a, SERCA3b, and SERCA3c) (16) and two isozymes in rats (SERCA3a and SERCA3b/c) (11,17). SERCA3 is expressed in a number of nonmuscle tissues at variable levels, always coexpressed with SERCA2b (15,18,19). At present, the full physiological significance of this heterogeneity among the SERCA isozymes remains unclear.
When unfolded proteins accumulate in the ER, a conserved eukaryotic stress response pathway known as unfolded protein response (UPR) is activated (8,20,21). This response amplifies the protein folding capacity in the ER through an enhanced expression of the ER-resident chaperones (such as BiP), thus counteracting congestion of misfolded protein intermediates and promoting cell survival (22,23). Ire1p, residing in the ER membrane and displaying both serine/threonine kinase and endonuclease activities, appears to function as the sensor of the accumulation of unfolded proteins and to activate the signal that enhances the transcription of chaperone genes in yeast (24) as well as mammals (25,26). Recently, stress inducibility of the chaperone genes in higher eukaryotes was shown to require the presence of a 19-base pair long regulatory elements termed ER stress response elements (ERSEs) in their promoters (27,28). In addition to chaperone up-regulation, several other signals emerge from the stressed ER, including induction of a growth arrest/cell death-promoting transcription factor C/EBP homology protein (CHOP, also known as growth arrest and DNA damage protein 153) (29), and in a number of cases, protein synthesis is transiently down-regulated because of phosphorylation of eukaryotic initiation factor 2␣ by the dedi-cated kinases PKR-like ER kinase (30) and possibly PKR (31).
Here we present results showing a strong correlation between induction of the UPR pathway and an elevated expression of SERCA2b, but not SERCA3, at the mRNA and protein levels. Depletion of ER calcium was not a necessary condition for UPR activation nor for the up-regulation of SERCA2b. Together with the presence of putative functional ERSE elements in the promoter region of SERCA2 gene, these results provide evidence for a new aspect of SERCA2b function, manifested by its ability to act as a member of the ER stress protein family. EXPERIMENTAL  Use of Stress Agents-Compounds to be added to the cells in culture medium were prepared as stock solutions concentrated 1000-fold in Me 2 SO (thapsigargin, A23187, tunicamycin, brefeldin A, actinomycin D), 100-fold in water (L-azetidine 2-carboxylic acid, EGTA), or 1000-fold in water (cycloheximide, DTT). The EGTA stock was adjusted to pH 9.16 to counter the pH fall upon its addition. Equivalent volumes of Me 2 SO or water were added to control cell batches. Final agent concentrations are given in the figure legends.
Northern Blots and Hybridization-PC12 or RINm5F cells were seeded on rat tail collagen-coated 100-mm dishes (3.5 g collagen/cm 2 ). Total RNA was isolated by a modified acid/phenol procedure (32), where the RNA was solubilized and stored in formamide after drying the final RNA pellet. RNA samples were run on a formaldehyde-agarose gel, transferred to nylon membranes (Hybond N, Amersham Pharmacia Biotech), and fixed by UV light according to standard procedures (33  was then washed by adding a 65°C hot wash solution (50 mM Na 2 HPO 4, pH 7.1, 1 mM EDTA, 2.5% SDS), and the wash was continued at room temperature with four changes, each of 10-min duration. The blot was autoradiographed on the Hyperfilm MP (Amersham Pharmacia Biotech) for 5 h to 2 days. Some of the Northern blots were stripped in boiling 0.1% (w/v) SDS before reprobing. Quantitation was obtained by digital scanning of the films using the NIH Scion Image program.
DNA and Protein Synthesis Rates, Cell Viability, and SERCA Immunostaining-These procedures were each carried out on a portion of the same pool of cells. 3 ϫ 10 6 PC12 cells were treated for 8 h with the indicated stressor, resuspended in 5 ml of the medium in which they had been cultured, and finally divided into four tubes.
Protein Synthesis Rate-2 ml of cells were centrifuged at 100 ϫ g for 10 min at room temperature. The supernatant was discarded, and the remaining cells were centrifuged in leucine-free Dulbecco's modified Eagle's medium and resuspended in 2 ml of leucine-free Dulbecco's modified Eagle's medium with the same stress agent as used prior to harvesting, and 2.5 l (18 pmol) of L-[4,5-3 H]-leucine. Cells were labeled for 7 min at 37°C. Labeling was stopped by addition of 1 ml of 0.1 M NaOH, after which the tube was cooled on ice. Macromolecules were precipitated by 5 ml of ice-cold 10% trichloroacetic acid, and the pellet was washed three times with 5 ml of ice-cold 5% trichloroacetic acid. Finally the pellet was solubilized in 0.25 M NaOH and counted by liquid scintillation.
DNA Synthesis Rate-1 ml of cells were labeled with 2 l (80 pmol) of methyl-[ 3 H]thymidine. The labeling reaction was stopped after 15 min at 37°C by 1 ml of 0.1 M NaOH, after which the tube was transferred to ice. Macromolecules were precipitated and washed, and the radioactivity was counted as described above.
Cell Viability-A 150-l aliquot was taken out for cell counting and viability test. Cells were mixed 1:1 with 0.4% (w/v) trypan blue in PBS. After 2 min at room temperature, dead (blue) and live (not stained) cells were counted in a hemocytometer.
Cytosolic Calcium Measurements-PC12 cells cultured on polylysinecoated (100 g/ml) glass coverslips were loaded for 30 min at 37°C with 5 M fura-2 acetoxymethyl ester and 0.04% pluronic F-127 (Molecular Probes) in EM solution (145 mM NaCl, 5 mM KCl, 20 mM Hepes, 3 mM CaCl 2 , 1 mM MgCl 2 , 10 mM glucose, and 100 M sulfinpyrazon) . The coverslip was then transferred to a solution consisting of 75% (v/v) EM solution with 25% (v/v) culture medium for additional 30 min at 37°C and 5% CO 2 . The coverslip was placed in a 37°C thermostated chamber (PH1 and TC344A, Warner Instruments, Hamden, CT) with 75% (v/v) EM solution and 25% (v/v) culture medium. For measurements in the absence of extracellular Ca 2ϩ , cells were washed three times in Ca 2ϩfree extracellular solution. Ca 2ϩ -free extracellular solution was EM solution with no Ca 2ϩ added (nominally Ca 2ϩ -free) or nominally Ca 2ϩfree EM solution supplemented with 25 M EGTA, as indicated in the figure legends. Dual excitation single emission fluorescence ratio imaging was performed using a system composed of the following: light from a high intensity source (XBO 75 W, OSRAM, Berlin, Germany) was passed via a monochromator (FL3095, J & M, Analytische Mess, Frankfurt, Germany), programmed to pass light of 340 and 380 nm. This light was transmitted into an inverted microscope (Nikon Diaphot 300), containing a 405-nm dichroic mirror and 510-nm long pass emis-sion filter. Cells were imaged using a 40ϫ lens (Zeiss Fluar, 40/1.30 oil). Emitted light was collected by a high resolution image intensifier coupled to a video camera (GenIIsys, CCD-72EX, DAGE-MTI, Michigan City, IN). The signal output from the camera was connected to a digital image-processing board (Matrox, MVP-AT) controlled by IMAGE-1/FL software (Universal Imaging, West Chester, PA). The digitized signal output from cells excited at each wavelength with an interval of 5 s was processed on-line (four frame average) to yield a ratio image map.
[Ca 2ϩ ] i values were collected from within the camera field as averaged signals from a circular measurement window placed over each cell. The ratio values were converted to intracellular Ca 2ϩ concentration ([Ca 2ϩ ] i ) by means of the formula of Grynkiewicz et al. (37).
Measurement of the SERCA Phosphorylated Intermediate-Following harvesting and centrifugation, cells were washed once in PBS and resuspended in 20 mM Tes (pH 7.0) and sucrose 300 mM (HB). Cells were broken by means of a cell cracker from the European Molecular Biology Laboratory workshop (38), employing the clearance of 12 m and 10 passages. The homogenates were centrifuged for 10 min at 10,000 ϫ g, followed by centrifugation of the supernatants for 60 min at 100,000 ϫ g. The microsomal pellets were resuspended in HB and stored at Ϫ80°C. ATPase phosphorylation was carried out as described earlier ("standard assay") (39).

SERCA2b mRNA Is Induced by ER Stress-
We have challenged PC12 cells with different agents sharing the ability to induce the mammalian UPR and measured the level of SERCA2b and SERCA3 mRNAs on Northern blots. The SERCA2b probe hybridized with the 4.5-, 6-, and 8-kb alternative splice variants of SERCA2b mRNA, differing in their 3Јuntranslated region (11,13). The SERCA2a mRNA was not expressed in PC12 cells (not shown). Fig. 1A shows that after a 6-h exposure, the agents known to perturb and deplete intra-ER Ca 2ϩ (thapsigargin, A23187, or EGTA) all induced an increase in the SERCA2b mRNA level relative to control cells exposed to the appropriate vehicle (Me 2 SO or water). Thapsigargin is an inhibitor of SERCA pumps (40), A23187 is an ionophore known to collapse divalent cation gradients (41), whereas EGTA treatment chelates extracellular Ca 2ϩ , thereby promoting Ca 2ϩ efflux from the cell and ER. The level of SERCA3 mRNA was not changed by these agents (Fig. 1B). In addition, tunicamycin, DTT, L-azetidine-2-carboxylic acid (Azc), and brefeldin A (BFA) also induced an increase in the SERCA2b, but not the SERCA3, mRNA level ( Fig. 1, A and B). Tunicamycin is an inhibitor of N-linked glycosylation (42). DTT perturbs the unique oxidative milieu of the ER lumen, thereby inhibiting the disulphide bond-dependent protein folding (43). Azc is a proline analog that is incorporated into proteins and causes their misfolding (44). BFA is an inhibitor of the transport from ER to Golgi, leading to an absorption of the Golgi apparatus into the ER (45). The treatments did not cause any major changes in the yield of total isolated RNA, except for Azc, which decreased the RNA yield to 50 -60% of controls. Fig. 1C shows the quantitation of the relative levels of mRNA for SERCA2b and SERCA3, as normalized to glyceraldehyde-3phosphate dehydrogenase (GAPDH).
As a positive control of the UPR induction, we monitored BiP and CHOP mRNA levels in PC12 cells exposed to stressors. Panels D and E in Fig. 1 show Northern blots of the same total RNA as analyzed in panels A and B, hybridized to BiP and CHOP probes. The pattern of BiP induction resembled the induction of SERCA2b mRNA, except for the induction magnitude (up to 10-fold, quantitation not shown) compared with 3-4-fold of SERCA2b (Fig. 1C). CHOP mRNA was not detectable in control cells but was induced by all the stressors, with the level of CHOP mRNA induction highly dependent on the particular stressor. CHOP induction by DTT at 6 h was very weak and is better appreciated at 1 or 2 h of DTT treatment (Fig. 2D).
SERCA2b and BiP mRNA Induction Follows Similar Kinet-ics-To further characterize the SERCA2b mRNA induction we compared the time course with that of BiP and CHOP. Fig. 2A shows an autoradiogram of a SERCA2b-probed Northern blot with total RNA isolated from cells treated for 1, 2, 4, or 8 h with EGTA or DTT. Fig. 2C shows a corresponding Northern blot hybridized with a BiP probe. Both BiP and SERCA2b were induced in 1-2 h, consistent with the described pattern of BiP induction (46). For both SERCA2b and BiP maximum was reached in about 2 h with DTT and 4 h with EGTA. The same overall kinetics was found for tunicamycin and A23187 (not shown). Longer stress periods (up to 24 h) did not increase SERCA2b mRNA level further. Fig. 2B shows an autoradiogram of SERCA3-probed Northern blot, demonstrating that SERCA3 mRNA level did not change in response to DTT or EGTA after 1, 2, 4, or 8 h.
The time course of the CHOP mRNA induction (Fig. 2D) appeared faster than for the BiP or SERCA2b mRNA. CHOP mRNA level reached a maximum in the presence of DTT within 1 h and was decreasing thereafter to become undetectable after 8 h. However, addition of fresh DTT after 4 h led to an increase in the CHOP mRNA after 8 h to the same level as at 4 h (not shown), indicating that the decreasing CHOP mRNA response to DTT, at least in part, was the result of the inactivation of DTT.
The SERCA2b Induction Is Reversible and Reinducible-To test whether the increase in SERCA2b mRNA levels was reversible and reinducible, cells were treated with EGTA for 4 h, after which the medium was changed to control medium for 20 h, to be followed again by EGTA treatment for 4 h. As shown in Fig. 3, SERCA2b, CHOP, and BiP responses were reversible and could be reinduced after 20 h of reconstitution in control medium. Both BiP and SERCA2b mRNA levels remained elevated for 2 and 4 h after the withdrawal of the stressor (Fig. 3, time points 6 -8 and 30 -32 h), after which they returned to the basal level (time point 24 h).
In contrast to BiP and SERCA2b, CHOP mRNA level responded much more promptly to the application and withdrawal of EGTA (Fig. 3). Notably, for all three mRNA types, the second response to the stressor appeared less pronounced than the first one.
The SERCA2b mRNA Increase Is Dependent on Protein Synthesis-A characteristic feature of the UPR is the requirement of protein synthesis for the transcriptional induction of BiP mRNA (46). Fig. 4A shows that neither tunicamycin nor A23187 were able to induce the SERCA2b mRNA in the presence of cycloheximide, demonstrating that protein synthesis was necessary for stress-mediated SERCA2b mRNA induction.
The Stress-induced Increase in SERCA2b mRNA Level Is Dependent on Transcription, and SERCA2b mRNA Is Not Stabilized under Conditions of ER Stress- Fig. 4B shows that actinomycin D, an inhibitor of transcription (47), could prevent the A23187 or tunicamycin-mediated elevation of SERCA2b mRNA. To test whether SERCA2b mRNA is stabilized under the conditions of ER stress, cells were incubated in the presence of tunicamycin or the vehicle for 4 h, following which (Fig.  4C, time 0) each medium was replaced with a similar one except for the addition of actinomycin D. Fig. 4C shows that after 6 h in the presence of actinomycin D, the SERCA2b mRNA decreased in both control and tunicamycin-treated cells. Quantitation showed a decline to about 40% of the level before addition of actinomycin D. Without the addition of actinomycin D the SERCA2b mRNA was increasing throughout the tunicamycin treatment (not shown). Total RNA yield decreased after 6 h of actinomycin D treatment to 60 -80% of control in both the vehicle and the tunicamycin-treated cells. Thus, SERCA2b mRNA was not stabilized in response to tunicamycin-induced ER stress.
Up-regulation of SERCA2b, but Not SERCA3, mRNA in RINm5F Insulinoma Cells in Response to ER Stress-To test whether the differential effect of ER stress on the 2 SERCA isoforms was present in other cell types, we tested RINm5F insulinoma cells (derived from rat pancreatic 74 cells) known to have a high expression level of SERCA3 (48). Fig. 5 shows that the various stress-inducing agents caused an elevation of SERCA2b, but not of SERCA3, mRNA in RINm5F cells, suggesting that the selective SERCA2b mRNA induction under ER stress conditions is a general response displayed by cells of different lineages. An elevation of SERCA2b mRNA in response to ER stressors was also found in EL4 lymphoma cells (not shown).
SERCA2b Expression Is Down-regulated by Heat Shock-A response to unfolded/malfolded proteins in the cytosolic compartment is another major stress-induced pathway in the cell. This pathway is highly activated upon heat exposure and is known as the heat shock response (49), whereas the UPR is normally not induced by heat shock. We tested whether SERCA2b mRNA level changed upon activation of the heat shock response. PC12 cells were incubated for 4 h at 42°C, and RNA was isolated. Fig. 6 shows that the SERCA2b mRNA was down-regulated upon heat shock, whereas mRNA for the heat shock protein 70 (HSP70), a classical heat stress-responsive chaperone, was highly induced. On the other hand, HSP70 was not induced upon treatment of the cells with A23187 and tunicamycin (not shown), showing a clear dissociation of the ER stress and heat shock-activated response pathways.
Stressed Cells Show Higher Levels of SERCA2b, but Not of SERCA3 Protein-Western blots of PC12 cell lysates stained with the SERCA2b-specific antibody (35) showed one band at 115 kDa in accordance with the known size of SERCA2b (Fig.  7A, upper part). In PC12 cell lysates, proteins with molecular mass of 105 and 116 kDa were stained with an antibody directed against an N-terminal epitope shared by, and specific for, all SERCA3 isoforms (36) (Fig. 7A, lower part). The 105-kDa immunoreactive protein exhibited a tissue distribution profile in agreement with that described for SERCA3 mRNA (Ref. 15 and results not shown), most likely reflecting the presence of SERCA3a (999 amino acids) isozyme. The size of the 116-kDa band suggests that it corresponds to the novel SERCA3b/c (1067 amino acids), recently identified (17). Fig. 7 (A and C) shows that the SERCA2b immunostaining increased after exposure to thapsigargin, A23187, EGTA, tunicamycin, and DTT for 8 h, whereas SERCA3 staining intensity remained at a steady level. Longer stress treatments (16 h) did not enhance the level of SERCA2b further. No effect was found of BFA or Azc on the SERCA2b level at 8 h (Fig. 7C).
Total Active SERCA Is Up-regulated after 7 h of Stress-An impaired folding and processing of proteins is at the heart of the ER stress, with an obvious potential for an impairment of function. To test whether the stress-induced increase in the synthesis of SERCA2b resulted in a functional protein, we measured the level of the phosphorylated Ca 2ϩ -ATPase intermediate in microsomes from PC12 cells. We have previously shown that the single radioactive band obtained in this phosphorylation assay represents a sum of SERCA2b and SERCA3 EϳPs (Ref. 39 and results not shown). Fig. 7 (B and C) shows that treatment of PC12 cells with A23187, EGTA, or DTT for 7 h did produce an increase in the amount of the phosphorylatable enzyme intermediate.
Protein and DNA Synthesis Are Differentially Inhibited by ER Stressors in PC12 Cells-The ER stress response has been shown to be accompanied by an inhibition of translation because of the activation of PKR-like ER kinase and/or PKR protein kinases (30,31). DNA synthesis and cell cycle progression were also found to be perturbed in stressed cells through an inhibition of cyclin D1 mRNA translation (50). To see whether the increased synthesis of SERCA2b protein during the ER stress occurred against a background of a decrease in the global protein and DNA synthesis, we tested whether these processes were perturbed in stressed PC12 cells. From Fig. 8 it appears that the inhibition of protein synthesis was very dependent on the particular type of stressor. Treatment with A23187 and thapsigargin resulted in only little or no reduction in the global protein synthesis rate. Treatments with DTT, tunicamycin, EGTA, or Azc resulted in an intermediate inhibi-tion of protein synthesis to about 50 -70%, whereas BFA inhibited protein synthesis to about 30% of the control. In contrast, the rate of DNA synthesis was depressed by all ER stressors, decreasing to about 20 -40% of the controls.

Depletion of ER Calcium Is Not an Obligatory Step for Activation of UPR or Up-regulation of SERCA2b by ER Stress-
Although depletion of Ca 2ϩ from the ER by thapsigargin, A23187, or EGTA is known to activate UPR (8), it has not been established whether initiation of the UPR response is always associated with such depletion. Resolution of this issue was of interest for at least two reasons. First, if induction of UPR were invariably accompanied by ER Ca 2ϩ loss, one might suspect that such loss was obligatorily linked to the UPR activation. Second, independently of the precise relationship between UPR initiation and ER Ca 2ϩ loss, such loss might be expected to be the key factor controlling the up-regulation of SERCA2b, in view of the obvious homeostatic rationale for such a mechanism. We therefore carried out experiments to estimate the size of the lumenal ER Ca 2ϩ pool following treatment with various stress-inducing agents, emphasizing those agents whose ability to interfere with ER Ca 2ϩ homeostasis has not been generally established, i.e. DTT, tunicamycin, BFA, and Azc. In these experiments (Fig. 9), the cells were exposed to the stressor, followed by a removal of extracellular Ca 2ϩ and an application of a test stimulus (a Ca 2ϩ -mobilizing agonist or thapsigargin) to release the ER Ca 2ϩ . PC12 cells in a resting state had a cytosolic Ca 2ϩ concentration ([Ca 2ϩ ] i ) between 40 and 50 nM. They responded to stimulation with acetylcholine with a transient increase in [Ca 2ϩ ] i (Fig. 9A, panel a), demonstrating the operation of a functional Ca 2ϩ signaling system. Treatment with thapsigargin ( Fig. 9, A, panel b, and B, panel b, Tg addition at Ϫ4 h) elicited a long lasting elevation of [Ca 2ϩ ] i as expected from the activation of store-operated Ca 2ϩ channels in the plasma membrane (51). Fig. 9A (panels d and e) shows that no perturbation of [Ca 2ϩ ] i within 18 -20 min from addition of tunicamycin or Azc was observed, and addition of thapsigargin followed by ionomycin in Ca 2ϩ -free medium produced increases in the cytosolic free Ca 2ϩ not different from those observed in control cells (Fig. 9A, panel c). The use of DTT as stressor gave analogous results. Therefore, tunicamycin, Azc, or DTT did not act as immediate or direct releasers of ER Ca 2ϩ . To test whether any such Ca 2ϩ releasing action of these compounds would occur in the time frame necessary for ER

FIG. 3. Reversible and repeated induction of SERCA2b, BiP and CHOP mRNAs.
Autoradiograms of Northern blots hybridized with SERCA2b, BiP and GAPDH, and CHOP and GAPDH probes. PC12 cells were exposed to medium with 2.5 mM EGTA for 4 h, followed by a control medium for 20 h. The cells were then re-exposed to the 2.5 mM EGTA-containing medium for 4 h, followed again by 4 h of control medium.Total RNA was isolated at the indicated times, and 10 g/gel lane was analyzed. Representative of three independent experiments. stress induction, cells were treated with these compounds for 4 h. Fig. 9B shows that 4-h exposures to tunicamycin (panel c), DTT (panel d), or Azc (panel e) did not result in a significant depletion of Ca 2ϩ from ER. Resting [Ca 2ϩ ] i , before removal of extracellular Ca 2ϩ , was at a level equal to that in the control (Me 2 SO-exposed) cells or slightly lower, indicating that storeoperated Ca 2ϩ channels were not activated, as would have been the case had the ER Ca 2ϩ been depleted and as seen in Fig. 9  (A, panels a and b, and B, panel b). Importantly, in comparison with controls, thapsigargin was able to release similar amounts of Ca 2ϩ (corresponding to a rise in [Ca 2ϩ ] i of about 20 nM) in Azc-, tunicamycin-, or DTT-treated cells. In contrast, Fig. 9B  (panel b) shows that an equivalent 4-h exposure to thapsigargin as a stressor abolished a subsequent effect of this drug applied as a test stimulus. Panel b also shows that thapsigar-gin-mediated emptying of ER Ca 2ϩ store resulted in the expected capacitative Ca 2ϩ entry, evidenced by an elevated [Ca 2ϩ ] i (ϳ75 nM, a value before introduction of Ca 2ϩ -free medium), when compared with control cells as well as to cells exposed to Azc, tunicamycin, or DTT (ϳ40 -45 nM, before introduction of Ca 2ϩ -free medium). The response to acetylcholine appeared somewhat smaller following these treatments in some experiments. It should be noted that the results with BFA (Fig. 9B, panel f) present a special case; although there was a clearly elevated [Ca 2ϩ ] i level, the ER Ca 2ϩ store was not depleted, as evidenced by a large increase of [Ca 2ϩ ] i upon application of thapsigargin test stimulus (Fig. 9B, panel f, open  arrowhead). (The reasons for this BFA-mediated elevation of [Ca 2ϩ ] i were not investigated.) DISCUSSION The parameters that need to be set correctly and coordinated for the ER-lumenal protein processing to proceed successfully include the total available capacity of the chaperones and folding enzymes and the physico-chemical parameters within the ER lumen, such as the oxidative milieu, oligosaccharide avail- ability, and the high Ca 2ϩ concentration. A number of agents and conditions may interfere with these parameters to evoke the ER stress response (8).
The main findings of this study focus on the regulation of ER Ca 2ϩ -ATPase expression under conditions evoking the ER stress response pathway known as UPR. We found a 3-4-fold selective up-regulation of the SERCA2b mRNA level in response to an array of UPR inducing agents. In contrast, there was no change in the SERCA3 mRNA level (Fig. 1). An earlier study has suggested that the intra-ER Ca 2ϩ content may be an important determinant of SERCA expression, because depletion of ER Ca 2ϩ resulted in an increased production of SERCA mRNA (however, no distinction was made between SERCA2 and SERCA3 mRNAs) (52). Our present findings agree with these earlier results. However, the present study substantially enlarges the context of SERCA expression regulation to that of the ER stress response in general, because of the inclusion of several stress conditions not expected a priori to affect the ER Ca 2ϩ content. We confirmed such lack of any significant ER Ca 2ϩ perturbation upon either short term or longer exposure of PC12 cells to DTT, tunicamycin, or Azc, by direct measurements of the thapsigargin-releasable ER Ca 2ϩ (Fig. 9). Therefore, we propose that the induction of SERCA2b mRNA was linked to the activation of the UPR by the stressed ER rather than representing an adjustment to the depletion of the ER Ca 2ϩ . More generally, these results also indicate that although the Ca 2ϩ -releasing agents have been employed in studies of the UPR (8), ER Ca 2ϩ depletion is not a necessary requirement to trigger this pathway. Several independent lines of evidence support the up-regulation of SERCA2b mRNA as reflecting the activation of the UPR pathway. First, several features of the SERCA2b mRNA up-regulation corresponded closely to those observed for the mRNA of BiP, the well studied UPR-induced ER chaperone measured in parallel in our system. Thus, the increase in SERCA2b mRNA occurred at a time scale similar to that for BiP (2-4 h), was similarly reversible upon the removal of the stress agent, and could be reinduced upon the reintroduction of this agent. Second, it is well established that the ER stressinduced increase of BiP mRNA requires protein synthesis (46) and is due to a transcriptional activation (53). We have shown that these characteristics did apply to the increase in the amount of the SERCA2b mRNA, based on the inhibition of the response by cycloheximide and actinomycin D and the lack of mRNA stabilization. Third, an analysis of the promoter region of SERCA2 genes revealed, within the first 1000 nucleotides upstream to the transcription initiation site, up to three copies of the nucleotide sequences showing at least 80% identity with the recently identified ERSE consensus sequence (Fig. 10A). were treated for 7 h with the indicated agents, the microsomes were isolated, and the phosphorylation carried out using [␥-32 P]ATP at 0°C for 20 s, followed by TCA precipitation, and analysis on acidic SDS-PAGE. The indicated band with an apparent molecular mass of 105 kDa represents SERCA2b and SERCA3 proteins, not separated because of a lower resolution of this acidic SDS-PAGE gel. C, quantitation of the immunoreactive SERCA2b protein (after 8 h of stress exposure) and of SERCA EϳP (7 h of stress). The optical density of the bands was measured and expressed relative to untreated controls (means Ϯ S.D.; n ϭ 5 for Tun, DTT, Tg, A23187, and EGTA; n ϭ 3 for BFA; n ϭ 2 for Azc). Statistical significance of differences were relative to control p Ͻ 0.05 (*) and p Ͻ 0.01 (**) as measured by Student's t test. DMSO, dimethyl sulfoxide. ERSE regulatory elements have recently been shown to be sufficient and necessary for the induction of the ER stress genes in higher eukaryotes (27,28). The ERSE consensus se-quence encompasses 19 base pairs and has an overall structure CCAAT(N 9 )CCACG (Fig. 10A), proposed on the basis of promoter comparison between ER stress-regulated genes as well  a and b, 300  as functional studies (27,28). It may be seen in Fig. 10A that the SERCA2 ERSE1 was conserved between species and agreed with the consensus ERSE, except for the substitution of adenine for guanine in position 19. ERSE position 19 is also substituted to adenine in the ERSE-like element of GRP58 (54) (Fig. 10A) and is the least conserved position in the ERSE consensus sequence. Several ERSE2 and ERSE3 sequences of SERCA2 genes showed additional divergences from the ERSE consensus. Importantly, SERCA2 ERSE1 from all species analyzed contained the GGC motif within the nine intervening nucleotides. On the other hand, this GGC motif was missing in the single ERSE-like element identifiable within the SERCA3 gene, presumably invalidating the SERCA3 ERSE, because the functional importance of this nucleotide triplet has been demonstrated by mutagenesis of the rat the BiP promoter (28). Similarly, the presence of only a single ERSE-like element in SERCA3 gene, in contrast to the multiple copies in SERCA2, may also constitute a functionally important difference (27,28). Yeast ER stress gene promoters contain the unfolded protein response element (UPRE, defined by the core sequence GA-CAGCGTGTC), which mediates ER stress gene up-regulation through binding of the transcription factor Hac1 (55). Recently, binding of yeast Hac1 to an UPRE-like sequence (GC-CAGCTTG) in the rat BiP promoter was reported, indicating that a putative mammalian Hac1 homolog might enhance the mammalian ER stress response (56). The SERCA2 promoter contains an UPRE-like sequence resembling the yeast BiP UPRE and rat BiP UPRE (Fig. 10B). Thus, SERCA2 promoters contain conserved sequences (ERSE and UPRE) known to mediate up-regulation of ER stress proteins.
The present work also demonstrates a selective up-regulation of SERCA2b (but not SERCA3) immunoreactive protein under the conditions of ER stress (Fig. 7). Unlike the relatively uniform effect of all the tested agents on the level of SERCA2b mRNA (3-4-fold increase), the resulting protein level varied from a 2.5-fold increase with thapsigargin or A23187, to no increase at all with BFA or Azc. Notably, a general correlation appeared to exist between the relative strength of the effect of each agent on SERCA2b immunoreactive protein and its effect on the global rate of protein synthesis (Figs. 7 and 8). These data suggest that the SERCA2b protein synthesis was reflecting the overall state of protein synthesis in the cell. This is at a variance from BiP, whose preferential synthesis even in the face of a strong translation inhibition is enabled by a unique, 5Ј-cap-independent mechanism of ribosomal loading for BiP mRNA (31).
The formation of the EϳP upon incubation with [ 32 P]ATP was used to quantitate SERCA pumps in an activity-dependent manner (39). As seen in Fig. 7C, the four tested ER stressinducing agents did cause a statistically significant increase in SERCA EϳP, albeit by a factor smaller than the corresponding increase in the immunoreactive SERCA2b. This discrepancy was at least in part due to the fact that the acid SDS-PAGE system used for the EϳP analysis (as distinct from the SDS-PAGE according to Laemmli (34)) was unable to resolve SERCA2b and SERCA3 EϳPs. Hence, the stress-unresponsive SERCA3 contribution to the EϳP band at 105 kDa (Fig. 7C) must have partly obscured the stress-induced response because of SERCA2b. In addition, the level of SERCA EϳP was measured in microsomes, in contrast to whole cell extracts used to measure SERCA2b immunostaining.
In summary, the experimental data (Figs. 1-6), as well as the analysis of the SERCA2 gene promoters (Fig. 10) strongly suggest that SERCA2 transcription is selectively activated by the UPR pathway, along with other ER stress-inducible genes and that this gives rise to an elevated level of active SERCA2b protein in the cell following ER stress (Fig. 7).
Given the universal nature of UPR, the observation of the specific association of SERCA2b, rather than SERCA3, with the ER stress-induced UPR agrees well with the ubiquitous nature of SERCA2b distribution. Indeed, SERCA2b has been regarded as a "householding" Ca 2ϩ -ATPase present in all tissues except cardiac and skeletal muscle, whereas SERCA3 displays a somewhat narrower distribution (11,15,16).
From the point of view of Ca 2ϩ homeostasis it seems intriguing that significant up-regulation of SERCA2b should take place with no apparent change in the ER content of Ca 2ϩ upon ER stress induced by agents like tunicamycin, DTT, or Azc. It may be speculated that an enhanced Ca 2ϩ uptake activity into the ER might act to shorten the period of a relative ER Ca 2ϩ depletion subsequent to a stimulus-induced ER Ca 2ϩ release, in this way responding to an increased demand for Ca 2ϩ by the ER chaperone system under conditions where unfolded proteins are accumulated.
One example of a possible pathophysiological relevance of the SERCA2b regulation by the UPR came from the recent identification of the presenilin-1 as a regulator of the mammalian UPR. Presenilin-1 mutations, known to underlie a large proportion of familial Alzheimer's disease cases, down-regulate UPR (57). Interestingly, such a presenilin-1 mutation was also shown to cause a destabilization of cellular Ca 2ϩ homeostasis (58). It is tempting to speculate that this destabilization of Ca 2ϩ homeostasis resulted from a low SERCA2b level, because For the individual genes, the light gray background indicates nucleotides identical to those within the consensus ERSE flanking base quintets, and dark gray indicates nucleotides identical to the consensus middle segment GGC motifs (28). For each ERSE, the distance in base count of the first C from the initiation of transcription (set to ϩ1) is indicated by a negative number. For each gene, up to three ERSEs found are numbered 1-3 in the order of increasing distance upstream from the transcription initiation site. (L) indicates that the element is found on the noncoding lower strand. B, alignment of UPRE-like sequences of SERCA2 and SERCA3 genes with those of yeast and rat BiP. Nucleotides shown on light or dark gray backgrounds are important for yeast BiP induction (55). Nucleotides on a dark gray background were found important for the binding of yeast UPR-activating factor Hac1p to rat BiP UPRE (56). Negative numbers indicate positions relative to transcription start, as for ERSE elements in A. GenBank TM accession numbers: human, SERCA2, AC006088.1; murine, AF029982.1; rabbit, M33834.1; rat, AF031937; yeast BiP, S40310.1; rat BiP M14866.1; SERCA3, Y15724. of a down-regulated UPR, in agreement with SERCA2b expression being linked to the UPR as reported here.