INTRODUCTION

Exposing cells to environmental stress induces expression of stress proteins in various intracellular compartments including the cytoplasm and the ER¹ (1-6). In addition, prior treatment with a mild insult that is sufficient to induce stress protein expression renders cells tolerant to a subsequent lethal insult (5, 7). For example, inducing HSPs with mild heat shock treatment confers thermotolerance as well as resistance to damage by cytokines, ischemic injury, and chemicals (8-10). The glucose-regulated proteins (GRPs), a family of molecular chaperones and Ca²⁺-binding stress proteins located in the endoplasmic reticulum (ER), are also induced by stress (4, 5). Induction of GRPs by ER stress protects cells against a variety of toxic insults including Ca²⁺ ionophores, oxidative stress, topoisomerase inhibitors, and cytotoxic T-cells (11-19). Thus, multiple stress proteins may be important in the cellular tolerance response.

Chemical toxicants including heavy metals, halogenated hydrocarbons, chemotherapeutic agents, or antibiotics induce stress proteins (1, 3, 5, 6, 20, 21), yet the mechanism(s) by which such a stress response prevents chemical damage in the target organs for these toxicants is not clear. The kidney proximal tubular epithelium is a particularly important target, and much is known about mechanisms of chemically induced cell death in kidney (22) and other cell types (23-26). In general, toxicant exposure initiates a cascade of biochemical events that ultimately cause cell death. For instance, exposing kidney epithelial cells to toxicants that are metabolized to reactive intermediates results in covalent binding of the metabolites to cellular macromolecules, depletion of cellular protein and nonprotein thiols, e.g. glutathione (GSH), increased intracellular Ca²⁺ concentrations, collapse of the mitochondrial membrane potential, and generation of reactive oxygen species (27-33). In LLC-PK1 cells, blocking any of these events with pharmacological agents blocks the toxicity of reactive metabolites and other toxicants (27-29, 34, 35). Taken together, these biochemical perturbations constitute a sequential and highly interrelated cytotoxic signaling cascade that results in cell death.

Despite the integration of the cell death cascade, activation of stress response genes in kidney epithelial cells is linked to specific perturbations suggesting that discrete signals within the cell death pathway are linked to specific genomic responses. For example, activation of hsp70 expression by iodoacetamide (IDAM) or the nephrotoxicant S-(1,2-dichlorovinyl)-L-cysteine is caused by oxidation or depletion of protein and nonprotein thiols and not directly by the covalent binding, Ca²⁺ disturbances, or oxidant production that also occur as part of the cell death pathway (21, 36). On the other hand, c-myc mRNA induction by S-(1,2-dichlorovinyl)-L-cysteine appears to be linked, at least in part, to an the increase in cellular free Ca²⁺ levels (37). Alkylation of cellular macromolecules may be sufficient to induce expression of c-fos and gadd153 (37, 38). Thus, biochemical perturbations caused by toxicant exposure serve both as discrete signals that activate specific stress response genes and as integrated components of a cell death pathway.

Intracellular Ca²⁺ homeostasis has received considerable attention as a cell death signal and as an activator of gene expression, yet consensus has not emerged regarding its role in either process (25, 26, 39, 40). Nevertheless, maintaining intracellular free Ca²⁺ levels at about 100 nM in the face of 1-2 mM extracellular Ca²⁺ is important for cell survival, and toxicant treatment generally causes an increase in free Ca²⁺ levels (26, 39). Membrane pumps in the ER, mitochondria, and plasma membranes work in concert to maintain intracellular Ca²⁺ levels (41, 42). Failure of Ca²⁺ pumping at any of these sites could contribute to an increase in free Ca²⁺ (26, 43). At physiological intracellular Ca²⁺ concentrations, the ER is a major intracellular Ca²⁺ storage site in nonmuscle cells (41, 42), and high lumenal Ca²⁺ is essential for normal ER function (44-46). Abundant ER Ca²⁺-binding proteins, including GRP78, GRP94, calreticulin, and calnexin, may help sequester ER Ca²⁺ (47-50). For example, calreticulin provides up to 45% of the Ca²⁺ buffering capacity in the inositol 1,4,5-trisphosphate-sensitive Ca²⁺ pool (51) and facilitates protein processing in the ER (52). Increasing or decreasing calreticulin expression also modulates physiological Ca²⁺ release from the hormone-sensitive pool (51, 53-55). Thapsigargin or calcium ionophores deplete ER Ca²⁺ thereby inhibiting ER protein processing and cellular protein synthesis in general (45, 46, 56, 57). Induction of ER chaperones renders cells tolerant to Ca²⁺ depletion (4, 5, 19, 56). Thus, a general increase in cellular Ca²⁺ and/or depletion of intracellular Ca²⁺ stores can cause cell death. Because ER chaperones are important both in cellular tolerance and in regulating cellular Ca²⁺, it seems possible that ER stress might protect cells by helping maintain cellular Ca²⁺ homeostasis.

The goal of these studies was to address the role of stress proteins in tolerance to chemical damage using the alkylating toxicant IDAM and the renal epithelial cell line LLC-PK1 as a model. These cells have been used extensively to investigate cytotoxicity and stress gene activation (21, 27, 28, 36-38, 58, 59). Herein, we show that conditioning LLC-PK1 cells with mild ER stress, but not heat shock, increases expression of ER stress proteins and prevents IDAM-induced cell death. Increasing expression of ER stress proteins apparently helps control intracellular Ca²⁺ levels following IDAM exposure preventing oxidative stress. The results provide new insights into the role of ER stress proteins in cellular Ca²⁺ homeostasis and cell death as well as in tolerance to chemical damage.

EXPERIMENTAL PROCEDURES

Fetal bovine serum and Dulbecco's modified Eagle's medium (DMEM) were obtained from Life Technologies, Inc. LLC-PK1 cells, a porcine renal epithelial cell line with proximal tubule epithelial characteristics (60, 61), were obtained from American Type Culture Collection (Rockville, MD) at passage 195 and were used from passage 205-215. N,N-Diphenyl-p-phenylenediamine (DPPD) was obtained from Eastman Kodak. The acetoxymethyl ester of EGTA (EGTA-AM) and Fura-2 (Fura-2AM) and Pluoronic F-127 were purchased from Molecular Probes (Eugene, OR). Radiochemicals were obtained from NEN Life Science Products. All other chemicals were obtained from commercial sources.

Cell culture and treatment of LLC-PK1 cells with IDAM were carried out as described (27, 36). LLC-PK1 cells were maintained in Dublecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum (complete medium). Confluent LLC-PK1 cells were treated with IDAM for 15 min in Earle's balanced salt solution (EBSS), then washed with phosphate-buffered saline (PBS), and allowed to recover in complete medium. Where appropriate, the antioxidant DPPD, prepared as a 20 mM stock in ethanol, was added to the medium at a concentration of 20 µM during the treatment period and/or during the recovery period. Cell were treated with DTTox (10 mM) for 2 or 3 h in EBSS and returned to complete medium for 12 h. Cells treated for 12 h in complete medium containing thapsigargin (0.3 µg/ml) or tunicamycin (1.5 µg/ml) were washed with PBS and returned to complete medium. For heat shock treatment, confluent LLC-PK1 cells in 10-cm dishes were incubated for 1 h in a water bath maintained at 43 or 45 °C in a humidified incubator at the same temperature and then returned to 37 °C for either 12 or 24 h.

Cytotoxicity, determined by measuring release of lactate dehydrogenase (LDH), covalent binding of [¹⁴C]IDAM to cellular macromolecules, as well as depletion of protein and nonprotein thiols were measured as described (36). Lipid peroxidation was determined by the formation of thiobarbituric acid-reactive substances (TBARS) as before (27).

An antisense grp78 expression vector was constructed in pcDNA3 (Invitrogen). A 524-base pair fragment from a hamster grp78 cDNA (62), a gift from Dr. Amy Lee, was digested with NaeI (+145 to +669) and inserted into the EcoRV site of pcDNA3 in a 3 to 5 orientation to create the antisense grp78 expression plasmid pASgrp78. pASgrp78 or the pcDNA3 empty vector was transfected into LLC-PK1 cells using Lipofectin (Life Technologies, Inc.), and a mass culture of cells that expressed the 0.5-kb antisense RNA (pkASgrp78 cells) was selected in 800 µg/ml G418 (Sigma) and maintained in 500 µg/ml G418. Multiple clones of pkASgrp78 were selected from the mass culture by ring cloning. Empty vector clones, termed pkNEO cells, were selected at the same time. Five pkASgrp78 clones were screened further for expression of GRPs following DTTox treatment by [³⁵S]methionine and [³⁵S]cysteine metabolic labeling (see below). Bands on autoradiograms representing ³⁵S-labeled GRP78 were quantitated by densitometric scanning using a BioImage Densitometer (BioImage, Ann Arbor, MI) as described previously (36). The integrated optical densities were normalized by taking the ratio of the GRP78 and actin signals in each lane, and the data were expressed as the fold increase in GRP78 relative to untreated cells. Three clones, pkASgrp78-5, -8, and -10, showed markedly reduced GRP78 synthesis and were further tested for the presence of 0.5-kb grp78 cDNA fragment by Southern blot analysis. Genomic DNA (20 µg) was digested with ApaI and BamHI; fragments were separated by electrophoresis, transferred to nitrocellulose membranes, and blotted with a hamster grp78 cDNA probe according to standard procedures. In experiments in which the response of pkNEO and pkASgrp78 clones was compared, three pkNEO clones, 2, 9, and 10, were compared with three pkASgrp78 clones, 5, 8, and 10. The response for the individual clones was determined in at least two separate experiments, and the mean of each clone was used as a single data point to calculate the mean of the three clonal lines.

An expression vector, pRC/CMV, containing a full-length (1.9 kb) human calreticulin cDNA (63) was provided by Dr. S. Dedhar. After transfection, calreticulin overexpressing cells (pkCRT) were selected for G418 resistance and were ring cloned as described above. Again, pkNEO cells were selected under identical conditions. Individual clones were tested for the expression of calreticulin by immunofluorescence and Western blot analysis using an antibody against calreticulin (StressGen, Vancouver, British Columbia). Clones overexpressing calreticulin were analyzed further for sensitivity to IDAM. Biological responses in three pkNEO clones, 1, 2, and 3, were compared with the pkCRT clones, 2, 3, and 5, as described above for pkASgrp78 cells.

Intracellular free Ca²⁺ was determined with the Ca²⁺-sensitive fluorescent dye Fura-2 according to Chen et al. (28) with modifications. Cells grown on coverslips coated with bovine collagen type I were rinsed with PBS and loaded with Fura-2AM in EBSS to achieve a final concentration of 3 µM. A 1:1,000 (v/v) dilution of 20% Pluoronic F-127 was added to EBSS to dissolve Fura-2AM and facilitate cell loading. In addition, probenecid, an inhibitor of organic ion transport, was included at a concentration of 2 mM to prevent intracellular transport or extrusion of Fura-2 free acid (33). Loading with Fura-2 was carried out at room temperature. After loading cells with Fura-2AM for 1 h, cells were washed four times with EBSS in the presence of 2 mM probenecid to prevent leakage. The coverslips were positioned in a quartz cuvette containing 3.5 ml of EBSS with probenecid for fluorescence analysis using a Shimadzu RF-5000 spectrofluorophotometer (Shimadzu, Columbia, MD). The calcium concentration was calculated as K_d (224 nM)× (R  R_min)/(R_max  R) according to Grynkiewicz et al. (64) as described previously (28). R is the ratio (F₁/F₂) of the fluorescence at excitation (ex) 340 nm, emission 505 nm over that of the fluorescence at excitation 380 nm. In some experiments, Ca²⁺ concentrations were also determined using digital fluorescence imaging as described (30).

When spectrofluorometric measurements were used to quantitate intracellular free Ca²⁺, the distribution of Fura-2 between the cytosol and intracellular compartments was determined in cells loaded as described above. Cytoplasmic Fura-2 was released by adding buffer A (250 mM sucrose, 20 mM KCl, 3 mM EGTA, 10 mM K₂HPO₄, 5 mM MgCl₂, 5 mM succinate) containing 50 µM digitonin for 5 min to permeabilize the plasma membrane. The supernatant was collected, and the cells were lysed with 0.1% Triton X-100 in buffer A. Fura-2 fluorescence in the digitonin (cytosolic Fura-2) and Triton X-100 fractions (total remaining) were monitored at the calcium-independent wavelength _ex = 362 nm. Using this procedure, we found that over 75% of the Fura-2 was in the cytosol, i.e. released by digitonin.

Preparation of mRNA was carried out as described previously (21). cDNA probes were labeled with [³²P]dCTP (NEN Life Science Products) by random priming using a kit (Boehringer Mannheim). Blots were probed with a hamster grp78 cDNA probe and then with -actin cDNA as an internal control. Western blot analysis for stress-inducible HSP70, also called HSP72, was carried out essentially as described (36) using a monoclonal antibody (Amersham Corp.). For detection of calreticulin, anti-calreticulin polyclonal antibody (StressGen) was used. Nitrocellulose membranes were blocked with 5% nonfat milk and probed with antibody in the presence of 5% nonfat milk. Detection of endogenous calreticulin by immunoblotting required an anti-calreticulin antibody dilution of 1:250, but with overexpressing cells a 1:5000 dilution was used. Appropriate secondary antibodies and the enhanced chemiluminescence system (Amersham Corp.) were used to develop the blots.

Immunofluorescence analysis of calreticulin was done using the same polyclonal anti-calreticulin antibody. Confluent cells on collagen-coated glass coverslips were rinsed in PBS and fixed with methanol at 20 °C for 10 min. After blocking with 2% horse serum in PBS for 45 min, the coverslips were incubated for 1 h with anti-calreticulin antibody (1:50) followed by dichlorotriazinyl aminofluorescein-conjugated goat anti-rabbit IgG (Jackson ImmunoResearch, West Grove, PA), diluted 1:250 in PBS containing 1% bovine serum albumin. Coverslips were mounted on slides and observed with a Nikon episcopic fluorescence microscope using a 60 × objective.

Porcine GRP78 did not cross-react with any available GRP78 antibodies tested; therefore, increased synthesis of stress proteins was determined by [³⁵S]methionine and [³⁵S]cysteine labeling. For short term labeling, confluent LLC-PK1 cells were incubated with methionine- and cysteine-free DMEM for 20 min followed by a 1-h incubation with [³⁵S]methionine and [³⁵S]cysteine (100 µCi/ml) in methionine- and cysteine-free DMEM. For the long term labeling, cells were incubated with [³⁵S]methionine and [³⁵S]cysteine (50 µCi/ml) in normal DMEM for 4 h. After radiolabeling, cells were lysed in hypotonic buffer (0.25 M sucrose, 25 mM Tris, pH 7.4, 2.5 mM magnesium acetate, 2.0 mM dithiothreitol), and proteins were solubilized in SDS sample preparation buffer. Radiolabeled proteins were resolved by SDS-polyacrylamide gel electrophoresis and protein bands visualized by autoradiography.

Student's t test was used to determine if there was a significant difference between the two groups (p < 0.05). When multiple means were compared, significance (p < 0.05) was determined by ANOVA followed by the Student-Newman-Keul's test. For ANOVA analysis, letter designations are used to indicate significant differences. Means with a common letter designation are not different, and those with a different letter designation are significantly different from all other means with different letter designations. Means with more than one letter designation are not different from groups with either letter designation. In cases where statistical analysis is shown for two different parameters in a single table or figure, i.e. Ca²⁺, thiobarbituric acid-reactive substances or LDH release, letters indicating significant differences apply only within that measurement group.

RESULTS

IDAM treatment increases expression of hsp70 in LLC-PK1 cells (36). Since induction of HSP expression is linked to tolerance, we evaluated the effect of heat shock on IDAM cytotoxicity. Although heat shock induced HSP70 in LLC-PK1 cells (data not shown), it did not protect against IDAM-induced cell death (Fig. 1). IDAM treatment also increased expression of the mRNA for prototypical ER stress protein grp78 in a time- and concentration-dependent manner (Fig. 2). Treating cells with DTTox, tunicamycin, or thapsigargin, agents that cause ER stress (5, 65), increased mRNA for grp78 and synthesis of both GRP78 and GRP94 proteins (Fig. 3, A and B) in LLC-PK1 cells. Pretreatment with all three agents prevented IDAM-induced cell death without altering [¹⁴C]IDAM covalent binding to macromolecules (Fig. 3C). There was also a good correlation between the peak of GRP78 and GRP94 biosynthesis and the onset of the tolerant phenotype after DTTox treatment (Fig. 4, A and B). The cells maintained the tolerant phenotype up to 24 h, probably due to the long half-life (>36 h) of ER stress proteins such as GRP78 (18). Thus, conditioning cells with ER stress protected them against IDAM toxicity without affecting toxicant entry and covalent binding.

Antisense and ribozyme strategies directed against grp78 and grp94, respectively, have been effective in probing the role of ER stress proteins in tolerance and protein secretion (12, 13). Selective targeting of grp78 with antisense interferes with induction of both grp78 and grp94 and disrupts the ER stress response (12). We targeted grp78 using a 0.5-kb antisense grp78 fragment that spanned the translation start site. After transfection, G418-resistant pkASgrp78 clones were tested for induction of GRP78 and GRP94. In pkASgrp78 clones, GRP78 synthesis after DTTox treatment was attenuated compared with empty vector pkNEO clones (Fig. 5, A and B). All the pkASgrp78 clones had integrated the antisense fragment (Fig. 5C). Although it appeared that induction of ³⁵S-labeled GRP94 was also reduced (Fig. 5B), the band could not be quantitated accurately by densitometry due to its proximity to other bands.

The pkASgrp and pkNEO clones were tested for IDAM sensitivity. Covalent binding of [¹⁴C]IDAM was equivalent in pkASgrp78 and pkNeo cells, 407 ± 113 versus 448 ± 14 pmol/mg protein, respectively, indicating that both took up IDAM equally well. LDH release 1-2 h after IDAM treatment was higher in pkASgrp78 clones compared with pkNeo clones, but there was no difference in maximum LDH release observed at 6 h (Fig. 6A). Unlike pkNEO cells, pkASgrp78 cells had a reduced capacity to develop tolerance after DTTox (Fig. 6B), nor did they develop tolerance after treatment with thapsigargin and tunicamycin (Fig. 7). Thus, expression of grps is important for tolerance to IDAM. The data clearly suggest that GRP78 is important in the ER stress response and cytoprotection, but we cannot exclude a role for GRP94 as well.

Having established a role for ER stress proteins in cellular tolerance, we went on to investigate the mechanism of protection. As shown in Fig. 3C, and in previous reports (29, 36), IDAM covalently modifies cysteinyl thiol groups in proteins. However, IDAM also elicits secondary effects in LLC-PK1 cells including depletion of GSH and oxidation of protein thiols (29, 36). Since ER stress did not affect covalent binding of [¹⁴C]IDAM (Fig. 3C), we determined if it diminished thiol-disulfide redox perturbations. However, depletion of cellular nonprotein and protein thiols after IDAM treatment was not altered by DTTox (Table I). Similar results were obtained in cells rendered tolerant by thapsigargin or tunicamycin treatment (data not shown).

Table I. Effect of ER stress on IDAM-induced loss of protein (PSH) and nonprotein (NPSH) thiols Cells were treated with IDAM for 15 min with (DTTox) and without (EBSS) DTTox pretreatment, as described in the legend to Fig. 3, and the levels of PSH and NPSH were determined as described under "Experimental Procedures." The data are presented as the mean ± S.D. of the data collected in three separate experiments (n = 3). Significant differences were determined by ANOVA as described under "Experimental Procedures." Means with a different letter designation are significantly different (p < 0.05) and apply only within that column of data, i.e. statistical comparisons were not made between PSH and NPSH values. Pretreatment IDAM PSH NPSH µM nmol/mg nmol/mg EBSS 0 55  ± 3a 15  ± 3a DTTox 0 50  ± 3a 16  ± 3a EBSS 75 43  ± 1b 3  ± 1b DTTox 75 44  ± 1b 5  ± 1b

Elevation of cytosolic Ca²⁺ is important in toxicant-induced cell death in renal epithelial cells (28, 30, 37), and other cell types (26, 39). Therefore, we determined if the cellular free Ca²⁺ surge observed after IDAM treatment was attenuated in tolerant cells. There was a sustained increase in intracellular free Ca²⁺ after IDAM treatment (Fig. 8, and data not shown). DTTox pretreatment blocked the increase in intracellular Ca²⁺. Lipid peroxidation also increased within 30 min after IDAM treatment followed by LDH release; both were prevented by DTTox pretreatment (Fig. 9). Thus, conditioning cells with ER stress blocked the IDAM-induced Ca²⁺ surge, lipid peroxidation, and cell death.

Loss of membrane integrity due to lipid peroxidation can cause extracellular Ca²⁺ influx (40). If this were the case, then the antioxidant, N,N-diphenyl-p-phenylenediamine (DPPD), which blocks lipid peroxidation after IDAM treatment (29), should block Ca²⁺ entry. DPPD treatment blocked much of the increase in intracellular Ca²⁺; however, Ca²⁺ still increased 3-fold from 64 to 189 nM (Fig. 10). When DPPD and DTTox treatments were combined, Ca²⁺ remained at control levels (Fig. 10). Removing extracellular Ca²⁺ also prevented the increase in free Ca²⁺ after IDAM treatment (data not shown), consistent with a role for oxidative stress in influx of extracellular Ca²⁺.

The data suggested that there might be a connection between ER stress, induction of Ca²⁺ binding chaperone proteins, and blockade of an IDAM-induced Ca²⁺ surge linked to oxidative stress. If the mechanism underlying this effect was dependent on an increase in Ca²⁺-binding proteins in the ER, then artificially increasing the level of ER Ca²⁺-binding proteins might produce the same effect. Overexpression of calreticulin, the major ER Ca²⁺-binding protein in nonmuscle cells (49), has been shown to increase ER Ca²⁺ stores and to modulate ER Ca²⁺ release (53, 55); therefore, we determined the effect of calreticulin overexpression on IDAM toxicity. We prepared three clones of LLC-PK1 cells, designated pkCRT-2, -3, and -5, all of which expressed high levels of calreticulin (Fig. 11A) in the ER (Fig. 11B). Compared with pkNEO cells, pkCRT cells were less sensitive to IDAM-induced cell death (Fig. 11C), although covalent binding of [¹⁴C]IDAM was unchanged; i.e. pkCRT, 398 ± 11 pmol/mg/protein; pkNEO clones, 399 ± 14 pmol/mg protein. Thus, enforced expression of calreticulin produced a tolerant phenotype indicating that ER proteins other than GRP78 could participate in cellular tolerance. Although we could not determine GRP78 levels by Western blotting due to a lack of antibodies (see "Experimental Procedures"), CRT expression did not alter the basal level of GRP94 (data not shown), indicating that CRT expression may not have a global effect on other ER stress proteins.

We also determined the effect of calreticulin overexpression on intracellular Ca²⁺ and oxidative stress after IDAM treatment (Table II). Without IDAM treatment, there was no difference in resting Ca²⁺ levels in pkCRT and pkNEO clones. However, after IDAM treatment, there was a significant increase in intracellular Ca²⁺ in pkCRT cells, but not nearly to the level seen in pkNEO cells. In addition, lipid peroxidation was prevented in pkCRT but not pkNEO clones after IDAM exposure. Thus, overexpression of calreticulin blocked the IDAM-induced increase in intracellular Ca²⁺ and oxidative stress indicating that the presence of Ca²⁺-binding proteins in the ER was important in preventing both responses.

Table II. Effect of calreticulin expression on intracellular Ca2+ and lipid peroxidation after IDAM treatment Calreticulin overexpressing clones (pkCRT) and the empty vector expressing clones (pkNEO) were treated with IDAM (75 µM) for 15 min and then incubated in complete medium for 1 h, at which time samples were collected for the assay of intracellular free Ca2+ and lipid peroxidation (see "Experimental Procedures"). The data are the mean ± S.D. from three different clones for lipid peroxidation (TBARS) or the average ± the range from two different clones for Ca2+ determinations. Statistical analysis of the Ca2+ or TBARS was done by ANOVA as described in Table I. Cells IDAM [Ca2+]i TBARS µM nM nmol/well pkNEO 0 75  ± 0a 0.3  ± 0.02a pkCRT 0 104  ± 71a 0.3  ± 0.01a pkNEO 75 963  ± 77b 1.6  ± 0.4b pkCRT 75 218  ± 24a 0.4  ± 0.07a

Although prior ER stress blocked the increase in cellular Ca²⁺ and prevented oxidative stress, much of the Ca²⁺ surge was due to entry from the extracellular pool, i.e. outside-in Ca²⁺ flux. To address the role of Ca²⁺ influx in IDAM toxicity, we compared the effect of removing extracellular Ca²⁺ on cell death caused by treatment with IDAM or the Ca²⁺ ionophore, ionomycin. Removing extracellular Ca²⁺ blocked cell death caused by the Ca²⁺ ionophore ionomycin but had no effect on IDAM-induced cell death (Table III). We next determined if DTTox pretreatment or calreticulin overexpression had any effect on toxicity due to influx of extracellular Ca²⁺ caused by ionomycin. pkCRT cells were less sensitive to ionomycin, indicating that pkCRT cells had an enhanced capacity to buffer extracellular Ca²⁺ (Fig. 12). However, the protection was not as dramatic as observed for IDAM (Fig. 11). DTTox treatment had no effect on ionomycin toxicity (data not shown).

Table III. The effects of extracellular Ca2+ on IDAM and ionomycin-induced cell death LLC-PK1 cells were treated with IDAM (75 µM) for 15 min in EBSS in the presence of various concentrations of Ca2+ and then returned to EBSS containing various concentrations of Ca2+. After 6 h, the release of LDH was determined. For ionomycin treatment, cells were treated with ionomycin at 10 µM in EBSS with various concentrations of Ca2+ for 6 h and then the release of LDH was determined. The data are the mean ± S.D. from triplicate samples in a single experiment and are representative of two individual experiments (n = 2). [Ca2+]ex % LDH release EBSS IDAM Ionomycin mM 0.0 13.7  ± 1.0 69.9  ± 4.1 14.1  ± 1.4 0.5 10.0  ± 2.0 67.7  ± 9.8 40.0  ± 3.0 1.0 6.0  ± 0.7 67.7  ± 15.9 82.5  ± 2.3 1.5 4.0  ± 0.9 63.1  ± 11.7 105.5  ± 2.5 1.8 7.64  ± 1.5 71.4  ± 12.4 103.5  ± 2.6

Several lines of evidence suggest that inside-out Ca²⁺ flux also might be important in cell death (39). Thapsigargin releases ER Ca²⁺ and causes apoptosis in LLC-PK1 cells, but prior ER stress blocks this response.² Since increasing cytosolic Ca²⁺ results in mitochondrial Ca²⁺ uptake and increased oxidant production (30), efflux of ER Ca²⁺ could stimulate mitochondrial oxidant production providing an inside-out mechanism of Ca²⁺ flux in cell death. Agents that buffer intracellular Ca²⁺ (EGTA-AM) or prevent mitochondrial Ca²⁺ uptake (ruthenium red) prevent oxidative stress and cell death in renal epithelial cells (28, 30). Loading cells with EGTA, using EGTA-AM, or adding ruthenium red prevented IDAM-induced cell death (Table IV). Thus, a disturbance of the ER Ca²⁺ pool (inside-out signaling) may be more important in the cell death pathway than the influx of extracellular Ca²⁺ (outside-in signaling).

Table IV. Effect of EGTA-AM and ruthenium red on IDAM-induced cell death Cells were incubated with EBSS alone or IDAM (75 µM) for 15 min and returned to complete medium in the presence or absence of EGTA-AM (50 µM) or ruthenium red (50 µM). LDH release was determined 6 h after removing IDAM. The data are the mean ± S.D. of data collected in three separate experiments (n = 3). Significant differences were determined by ANOVA as described under "Experimental Procedures." Means with different letter designations were significantly different (p < 0.05). Treatment % LDH release EBSS 3  ± 1a IDAM 64  ± 6b IDAM + EGTA-AM 6  ± 2a IDAM + ruthenium red 7  ± 4a

DISCUSSION

A number of useful conclusions can be drawn from these studies. First, induction of ER stress proteins protects cells against alkylating chemicals. Preliminary studies show that ER stress also protected LLC-PK1 cells from nephrotoxic cysteine conjugates (65), t-butylhydroperoxide, and thapsigargin toxicity as well.³ Second, multiple ER proteins may be important since blocking induction of GRPs prevented tolerance while overexpressing calreticulin protected cells. To our knowledge, calreticulin has not been shown to play a role in tolerance to chemical damage. A role for GRP78 also seems clear, but GRP94 may play a role as well, nor can we exclude the possibility that altering expression of one ER stress protein has an indirect but significant effect on another. Third, ER tolerance depends in part on maintaining cellular Ca²⁺ homeostasis and preventing oxidative stress. Although the importance of ER Ca²⁺ in protein processing (45, 46, 57), translational control (56), and regulation of grp78 transcription (66) is well known, the role of the ER in regulating cellular Ca²⁺ homeostasis and oxidative stress after chemical damage has not been shown previously. Thus, our data shed new light on the role of the ER in control of cellular Ca²⁺ and cytotoxicity.

Our data also support a model of IDAM-induced cell death in which an increase in cytoplasmic Ca²⁺ leads to mitochondrial Ca²⁺ uptake, induction of oxidative stress, membrane peroxidation, and cell death (Fig. 13), a model supported by data from other studies in renal epithelial cells (27, 28, 30). Mitochondria sense cytosolic Ca²⁺ fluctuations by accumulating Ca²⁺ and thus tune energy production to meet the biological responses initiated by Ca²⁺ signaling (41). However, Ca²⁺ buffering by mitochondria must be coupled to extrusion across the plasma membrane and/or re-uptake of Ca²⁺ into the ER, otherwise mitochondria will accumulate a lethal load of Ca²⁺ (30, 67, 68). If the latter happens, membrane potential collapses, reduced pyridine nucleotide pools are depleted, phospholipases are activated, and large pores open in the mitochondrial inner membrane (67). When cellular GSH has been depleted, mitochondrial Ca²⁺ overload can cause excess oxidant production, oxidative stress, and plasma membrane rupture. Thus, buffering intracellular Ca²⁺ and preventing mitochondrial Ca²⁺ accumulation and/or cycling blocks cell death following chemical exposure by uncoupling Ca²⁺ perturbations from oxidative stress (30, 67).

Although we did not directly assess the changes in ER Ca²⁺ stores and the effect of chaperone expression on ER Ca²⁺ stores during toxicant treatment, it may be that the ability of the ER to release or buffer intracellular Ca²⁺ modulates cell death (Fig. 13). The ER is the major intracellular Ca²⁺ storage site in nonmuscle cells and could be a target for toxic damage. The ER Ca²⁺-ATPase is inhibited by carbon tetrachloride treatment in vivo (69), and toxicants have been shown to impair operation of ER Ca²⁺ release channels (26, 70, 71). Moreover, thapsigargin treatment causes apoptosis suggesting that loss of ER Ca²⁺ is a cell death signal (72-75). Interestingly, bcl-2 expression can block thapsigargin-induced cell death in some cells (73, 74). Mitochondria from bcl-2 overexpressing cells have an increased capacity to accumulate Ca²⁺ (68) indicating that there could be a link between ER and mitochondrial Ca²⁺ pools in cell death. In addition, toxicants that modify protein sulfhydryls release ER Ca²⁺ (70, 71) and prevent extrusion of Ca²⁺ across the plasma membrane (43) generally perturbing Ca²⁺ signaling. Depleting ER Ca²⁺ also disrupts ER protein processing and general protein synthesis and activates expression of grps genes (41, 45, 46, 66), effects that are blocked by expression of ER chaperones (5, 11, 19, 56). Thus, in naive cells, disrupting ER Ca²⁺ buffering may contribute to cell death, whereas inducing ER chaperones and Ca²⁺-binding proteins prevents cell death. Although it is not clear from our studies if the protection caused by induction of ER stress proteins is due to their ability to modulate ER Ca²⁺ stores directly or indirectly, it is apparent that induction of ER stress proteins helps control general intracellular Ca²⁺ homeostasis preventing toxicant-induced cell death.

It has been suggested that accumulation of intracellular Ca²⁺ is merely a secondary effect of membrane damage and influx from the extracellular pool (40). Indeed, in our studies removing extracellular Ca²⁺ or adding antioxidants blocked Ca²⁺ accumulation after IDAM treatment. However, this outside-in Ca²⁺ surge was not responsible for IDAM-induced cell death. Yet, buffering intracellular Ca²⁺ by treating cells with EGTA-AM prevented IDAM toxicity arguing that Ca²⁺ does play a role. Here again the model in Fig. 13 accounts for these observations since release of intracellular Ca²⁺ would lead to secondary influx from the extracellular pool due to oxidative stress and membrane damage (43). Cooperation between ER Ca²⁺ efflux and extracellular Ca²⁺ influx is well know during hormone-induced capacitative Ca²⁺ entry (76).

Although, we addressed toxicant-induced necrosis, our data may provide general support for an ER-mitochondrial Ca²⁺ axis in cell death. Disturbances in Ca²⁺ pumping in the ER and mitochondria also cause apoptotic as well as necrotic cell death (26, 39). bcl-2 prevents this apoptosis, perhaps by increasing the capacity of mitochondria to buffer Ca²⁺ (68). This connection between bcl-2 and mitochondrial Ca²⁺ is particularly important given that mitochondrial damage is linked to activation of proapoptotic protease cascades (77, 78). Our recent finding that prior ER stress or overexpression of calreticulin protects against thapsigargin-induced apoptosis in LLC-PK1 cells² further supports the notion that efflux of ER Ca²⁺ is an early event in cell death. Even if this model does not hold for all cells (79), when taken in context, our data point toward a link between ER and mitochondrial Ca²⁺ handling and cell death signals.