Roles of IP₃R and RyR Ca²⁺ Channels in Endoplasmic Reticulum Stress and β-Cell Death

MIN6 cells were cultured and transfected as described previously (23). Cells were imaged 48–72 h after transfection. Pancreatic islets were obtained from 8- to 16-week-old male C57BL6/J mice by collagenase digestion and filtration and cultured as described (28,29). For high-throughput imaging–based cell death assays, islets were hand-picked the next day and dispersed into single cells and plated on 96-well plates (see below).

Thapsigargin (Tg) was purchased from Calbiochem (La Jolla, CA) or Sigma (St. Louis, MO) and was kept as a 1,000× DMSO stock. Tetramethylrhodamine ethyl ester perchlorate (TMRE) (Sigma), xestospongin C (AG Scientific, San Diego, CA; Calbiochem), ryanodine (Molecular Probes, Eugene, OR; Tocris, Ellisville, MO; Calbiochem), dantrolene, CGP-37157, and carbonyl cyanide m-chlorophenyl-hydrazone (CCCP) (Calbiochem) were dissolved in DMSO. Carbachol from Calbiochem was dissolved in water.

Single-cell imaging was performed in Ringer's solution containing (in mmol/l): 5.5 KCl, 2 CaCl₂, 1 MgCl₂, 20 HEPES, 141 NaCl, and 3 glucose. Cytosolic Ca²⁺ was imaged in fura-2-AM–loaded cells as described previously (4,30). Preheated solutions were applied by stable perifusion at 1 ml/min, and complete solution changes were achieved in <30 s.

ER luminal and mitochondrial Ca²⁺ was imaged using the fluorescence resonance energy transfer (FRET)-based D1ER and mt4D3cpv cameleons, respectively (31,32). The cyan fluorescent protein (CFP) component of the probes was excited using a S430/25x filter (Chroma). CFP and FRET (i.e., yellow fluorescent protein) emission were alternately collected using S470/30m and S535/30m filters mounted in a Sutter Lambda 10-2 filter wheel. Changes in ER and mitochondrial Ca²⁺ were expressed as the FRET-to-CFP emission ratio. There was no correlation between the apparent Ca²⁺ levels and the intensity of the FRET probe in the cells used for this study.

Single-cell, FRET-based imaging of caspase-3 activity was performed using the MiCy-DEVD-mKO probe (33). Activation of caspase-3 cleaves this probe and results in a loss of FRET between a CFP (MiCy) and an orange fluorescent protein (mKO). MiCy excitation and emission was controlled by 436/20x and 490/40m filters, respectively. FRET with mKO was measured using a 585/60m filter and normalized to MiCy emission intensity.

For kinetic analysis of cell death, MIN6 cells or dispersed mouse islet cells were plated onto glass-bottom 96-well microplates (ViewPlate-96; Perkin Elmer) in culture media (see above) and treated as indicated. Cell death was monitored by the incorporation of propidium iodide (PI) (250–500 ng/ml in each well). PI fluoresces brightly only once it passes through the compromised plasma membrane and binds to DNA. It labels cells in the last stages of apoptosis that follow caspase activation, as well as cells undergoing necrosis (34). Approximately 30 min after treatment, 96-well plates were imaged at 37°C and 5% CO₂ using a Cellomics KineticScan (Pittsburgh, PA). Two to four nonoverlapping images were taken from each well at 30- or 60-min intervals. PI-positive cells were automatically identified and counted using the Target Activation Bioapplication (Cellomics, Pittsburgh, PA). Cell death is presented as an absolute count of PI-positive cells and quantified by calculating the incremental area under the curve. Plots of representative cell death profiles show the means ± SE of three independent cultures imaged simultaneously. For statistical comparisons, we used a conservative approach in which the average response of three similarly treated, independent cell cultures on a microplate was treated as a separate “n”. At least three of these replicate sets were performed on separate days for each study.

Changes in mitochondrial membrane potential were estimated by flow cytometry of MIN6 cells stained with TMRE (35). After the indicated treatments, all floating and adherent cells were collected, spun down at 500g for 10 min, and loaded with 50 nmol/l TMRE in PBS with 2% fetal bovine serum (FBS) for 30 min at 37°C. The cells were washed again and kept in PBS for ∼30 min before a total of 10⁵ events were collected using the FL2 channel of a Becton Dickinson FACScan. Cellular debris was identified by forward- and side-scatter criteria and excluded from analysis. Every set of measurements included a depolarized control sample pretreated for 30 min with 10 μmol/l of the mitochondrial uncoupler CCCP.

Western blots were carried out as described (23). Rabbit monoclonal antibody to cleaved caspase-3 was from Cell Signaling (Danvers, MA). Rabbit polyclonal anti–C/EBP homologous protein (CHOP) antibody was from Santa Cruz (Santa Cruz, CA). Rabbit monoclonal antibody to phospho–PKR-like ER kinase (PERK) and rabbit polyclonal antibody to phospho–eukaryotic initiation factor-2α (eIF2α) were from Cell Signaling. The antibody to total eIF2α was a mouse monoclonal from Abcam (Cambridge, MA). After three washes, membranes were incubated with secondary antibodies diluted in I-block (caspase-3, CHOP) or 5% BSA (PERK, eIF2α) for 1 h. In the case of phospho-PERK and phospho-eIF2α (detection), phosphatase inhibitor (Calbiochem) was added to the lysis buffer. After three washes with 0.1% Tween-PBS, immunoreactivity was visualized using chemiluminescence. Densitometric analysis was performed using Photoshop (Adobe Systems, San Jose, CA).

Unless otherwise indicated, data are presented as means ± SE. Differences between means were evaluated using Student's t test and were considered significant if P < 0.05.

The ER is involved in the regulation of multiple cell death pathways (7–9). Although luminal ER Ca²⁺ levels are thought to play critical roles in many apoptotic cascades, measurements of Ca²⁺ dynamics within the ER under pro- and antiapoptotic conditions have remained technically challenging. Using fura-2 and FRET-based imaging, respectively, we measured the changes in cytosolic and ER luminal Ca²⁺ caused by inhibition of SERCA pumps and activation of IP₃Rs. Thapsigargin is a specific SERCA inhibitor known to induce apoptosis in many cell types including β-cells (12). In agreement with other studies (36), blocking ER Ca²⁺ uptake with thapsigargin evoked transient cytosolic Ca²⁺ rises in MIN6 β-cells, although not in every cell (data not shown). This suggests that there is a substantial steady-state leak from the ER or that SERCA pumps are a component of a critical Ca²⁺ buffering system in MIN6 cells. In agreement with other reports (12,37), we also confirmed that carbachol, a cholinergic agonist that causes IP₃ formation, rapidly mobilized intracellular Ca²⁺ to evoke a cytosolic Ca²⁺ peak in MIN6 and primary β-cells (data not shown). The initial Ca²⁺ spike was often followed by a lower, but sustained, Ca²⁺ elevation that depended on extracellular Ca²⁺ influx, suggesting a possible role for a Ca²⁺ release–activated Ca²⁺ current (i.e., CRAC channel) (38,39).

To extend these findings, we analyzed the dynamics of luminal ER Ca²⁺ directly by using the D1ER cameleon. This probe has an optimal sensitivity range, excellent signal-to-noise characteristics (31,32), and was designed to mitigate interference with endogenous Ca²⁺ signaling and reduce pH-based artifacts (31). D1ER was localized to the ER by virtue of both KDEL and calreticulin sequences and was observed in a reticular pattern that matches ER-targeted monomeric red fluorescent protein (Fig. 1A). Direct measurements of luminal Ca²⁺ demonstrated that treatment with 1 μmol/l thapsigargin resulted in a robust decrease in ER Ca²⁺ (Fig. 1B) and that carbachol stimulated Ca²⁺ release from the ER in a reversible and repeatable manner (Fig. 1C). Notably, the thapsigargin-induced lowering of ER Ca²⁺ was more gradual and characterized by larger cellular heterogeneity than that evoked by carbachol (Fig. 1B and C). In some cells, ER Ca²⁺ levels fell immediately upon SERCA inhibition, whereas a significant delay was seen in others, suggesting cell-to-cell differences in basal ER Ca²⁺ release rates (i.e., leak). Similar results were seen with 20 μmol/l 2,5-Di-(t-butyl)-1,4-hydroquinone, a structurally distinct SERCA inhibitor (data not shown). Together, these experiments validate the use of the D1ER cameleon in β-cells and demonstrate the dynamics of ER Ca²⁺ emptying when ER Ca²⁺ release channels are activated and ER Ca²⁺ uptake blocked, respectively.

ER Ca²⁺ homeostasis is regulated by both pumps and channels (19). To analyze the time course of cell death in response to blockers of β-cell Ca²⁺ pumps and channels, we utilized a high-throughput imaging platform to monitor propidium iodide incorporation for 24–48 h under normal incubated culture conditions. Both the time course and the degree of cell death were dose dependent in thapsigargin-treated cells (Fig. 2A–C). Cell death was associated with induction of ER stress, as demonstrated by increased expression of the transcription factor CHOP/GADD153 (Fig. 2D), an essential component of ER stress–mediated apoptosis in β-cells (40). Western blotting also demonstrated that blocking ER Ca²⁺ pumps with thapsigargin activated caspase-3 (Fig. 2D) and increased cleaved caspase-7 by sixfold at 24 h (P < 0.05, n = 4; data not shown). We then used real-time single-cell FRET-based imaging to examine the dynamics of this process. Caspase-3–dependent loss of FRET was observed 2–4 h following the addition of thapsigargin (Fig. 2E). Hence, apoptosis of MIN6 cells is triggered relatively quickly following SERCA inhibition. Cyclopiazonic acid and 2-aminoethoxydiphenyl borate, two structurally distinct SERCA blockers (19,41), similarly induced β-cell ER stress and apoptosis (data not shown). These observations strongly suggest that thapsigargin triggered the ER stress response by specific inhibition of SERCA and provide a detailed kinetic analysis of β-cell apoptosis caused by ER stress.

We next investigated whether ER stress and apoptosis might be induced when ER-resident Ca²⁺ release channels (i.e., IP₃Rs and RyRs) are blocked under conditions of normal ER Ca²⁺ uptake. Inhibition of RyRs for 24 h with 100 μmol/l ryanodine did not induce death in MIN6 cells cultured in high glucose (Fig. 3A), in agreement with our previous findings (17). When compared with thapsigargin, ryanodine did not evoke a similarly robust increase in CHOP expression or caspase-3 cleavage (Fig. 3), although preliminary analysis indicated that calpain-10 protein levels were upregulated (not shown), consistent with our previous results (17,23). We also utilized the IP₃R inhibitor, xestospongin C, previously demonstrated to block IP₃-dependent Ca²⁺ release in pancreatic β-cells (42,43). As was the case with ryanodine, the effects of 1 μmol/l xestospongin C on CHOP, caspase-3, and cell death were modest. Simultaneous inhibition of RyRs and IP₃Rs also did not cause cell death, ruling out the possibility of compensatory Ca²⁺ flux through one class of channel when the other type was blocked (Fig. 3). CGP-37157, a drug that has been reported to indirectly interfere with ER Ca²⁺ uptake by blocking mitochondrial Na⁺/Ca²⁺ exchange (30,44), had little effect on ER stress or caspase-3 activation. It should be noted, however, that we have previously demonstrated that CGP-37157 also inhibits voltage-gated Ca²⁺ entry in β-cells (30). There was a tendency for caspase-3 cleavage to be reduced by changing the glucose concentration from 5 to 25 mmol/l (Fig. 3B and C). Together, these experiments demonstrated that marked ER stress and caspase-3–dependent apoptosis were induced specifically by inhibition of ER Ca²⁺ uptake but not by inhibition of ER Ca²⁺ release via RyR and IP₃R channels.

The level of Ca²⁺ in the ER lumen reflects the balance between influx and efflux. We sought to establish the role of channel-mediated efflux in β-cell ER stress and apoptosis. First, we tested the hypothesis that reducing ER Ca²⁺ release might ameliorate the effects of thapsigargin. Indeed, the RyR blocker dantrolene protected MIN6 cells from thapsigargin-induced death (Fig. 4A and B). Ryanodine (100 μmol/l), alone or in combination with xestospongin C (1 μmol/l), also suppressed thapsigargin-induced propidium iodide incorporation, although this trend did not reach statistical significance (Fig. 4B). The protective effect of dantrolene was also associated with a reduction in the amount of cleaved caspase-3 observed after 8 h of treatment with thapsigargin (Fig. 4C and D). Similarly, the combination of ryanodine and xestospongin C protected cells from thapsigargin-induced caspase-3 cleavage (Fig. 4E–G). The observation that thapsigargin-induced caspase-3 activation, but not total PI incorporation, was significantly reduced by the xestospongin C and ryanodine combination may reflect the fact that PI labeling is not strictly specific for apoptotic death. Thapsigargin-induced CHOP expression at 8 h was also reduced by inhibition of RyRs and IP₃Rs (Fig. 4G). Together, these experiments suggest that ER Ca²⁺ release through RyRs and IP₃Rs contribute to short-term ER stress and caspase-3–mediated cell death following SERCA inhibition.

In light of these findings, we hypothesized that chronic activation of ER Ca²⁺ release channels might exacerbate the effects of SERCA inhibition. Indeed, thapsigargin-induced cell death was significantly increased when IP₃ receptors were concurrently activated by either 1 μmol/l (not shown) or 100 μmol/l carbachol (Fig. 5A and C). The additional cell death was associated with a dramatic increase in CHOP expression at 8 h (Fig. 5D and E). We further tested this hypothesis using a stimulatory concentration of ryanodine (18,42) and found that 1 nmol/l ryanodine also augmented thapsigargin-induced MIN6 cell death (Fig. 5B and C). Taken together, these data demonstrate that ER Ca²⁺ release channels can both negatively and positively modulate ER stress.

Next, we looked for changes in ER luminal Ca²⁺ flux that correlated with the augmentation of β-cell death and ER stress by IP₃R activation. Imaging ER Ca²⁺ levels revealed that a combination of thapsigargin and carbachol resulted in a more rapid and homogeneous depletion of ER Ca²⁺ stores, compared with thapsigargin alone (Fig. 6A and B). Following a response to carbachol, the addition of thapsigargin further depleted ER stores (Fig. 6C). When thapsigargin was added first there was a complete depletion and no additional response to carbachol in the majority of cells (Fig. 6D). However, in thapsigargin-treated cells that had not yet fully depleted, carbachol accelerated the ER Ca²⁺ loss (Fig. 6D, arrow).

What rapid events might link ER Ca²⁺ depletion kinetics to the transcriptional induction of CHOP? Recent work has implicated the eIF2α kinase PERK in the induction of Ca²⁺-dependent ER stress (24). In our experiments, thapsigargin caused sustained PERK and eIF2α phosphorylation that was detectable as early as 5 min (Fig. 6E and F). There was a tendency for carbachol to augment the thapsigargin-induced phosphorylation of PERK at the 5-min time point, potentially linking rapid ER Ca²⁺ depletion to the subsequent amplification of CHOP expression (45). Taken together, these findings suggest that rapid events in the ER lumen can have profound and immediate effects on the unfolded protein response, ER stress, and apoptosis in β-cells.

Evidence indicates that the intrinsic mitochondrial pathway of apoptosis may be triggered during ER stress–associated cell death (46,47). To evaluate the effects on mitochondria, we first analyzed the time-dependent changes in mitochondrial membrane potential by flow cytometry. Interestingly, SERCA inhibition initially caused marked mitochondrial hyperpolarization, suggestive of increased metabolic flux (Fig. 7A and B). However, mitochondrial polarization ultimately collapsed in the majority of cells, and this depolarization was augmented when IP₃Rs were simultaneously stimulated by carbachol. The activation of mitochondrial apoptosis was further suggested by a significant increase in cleaved caspase-9 after 24 and 48 h thapsigargin treatment (Fig. 7C). Functional cross-talk between ER and mitochondria was evidenced by acute increases in mitochondrial Ca²⁺ following ER Ca²⁺ mobilization by carbachol or thapsigargin (Fig. 7D). Together these results provide compelling evidence for the involvement of mitochondria in the regulation and execution of β-cell apoptosis induced by ER Ca²⁺ depletion.

We also examined the effects of SERCA and Ca²⁺ channel blockers on dispersed mouse islet cells. Thapsigargin caused progressive dose- and time-dependent death, though it was quantitatively less than what we observed with MIN6 β-cells (Fig. 8A and B). Nevertheless, carbachol significantly augmented the death of primary cells following SERCA block, in agreement with our MIN6 cell findings. There was also a small degree of cell death in response to carbachol alone (Fig. 8C). Combined application of 100 μmol/l ryanodine and 1 μmol/l xestospongin C protected primary β-cells from the potentiating effect of carbachol, verifying that it was due to release of intracellular stores. These results suggest that similar, but not identical, mechanisms are involved in MIN6 cell apoptosis and primary β-cell apoptosis in the context of SERCA inhibition.

The goal of this study was to determine the specific roles of luminal ER Ca²⁺ influx and efflux mechanisms in the control of β-cell ER stress and apoptosis. To this end, we used a combination of “online” cell death measurements and real-time imaging of ER Ca²⁺ dynamics. First, we determined that blocking ER-resident Ca²⁺ release channels alone was not sufficient to induce substantial ER stress–related cell death. Among the perturbations of β-cell ER Ca²⁺ handling tested, only direct inhibition of SERCA pumps was associated with robust induction of ER stress and activation of caspase-3. Second, we found that thapsigargin-induced apoptosis was attenuated by drugs that reduce channel-mediated Ca²⁺ release from the ER. Third, we established that the effects of thapsigargin on ER Ca²⁺ depletion, ER stress activation, and cell death were accelerated by stimulation of IP₃Rs or RyRs. Together, these data provide the first evidence that β-cell ER stress is regulated by the activity of ER-resident Ca²⁺ release channels.

The pathophysiological importance of ER stress and apoptosis in the pancreatic β-cell is becoming increasingly clear (2,5,10). Our identification of conditions that slow and speed up ER stress and cell death elucidates the underlying mechanisms of and may promote strategies for counteracting ER stress associated with cytokines in type 1 diabetes (13), obesity-related type 2 diabetes (10), and rare disorders such as Wolcott-Rallison syndrome and Wolfram's syndrome (5). Indeed, ion channels such as the IP₃Rs and RyRs are useful drug targets. The finding that β-cell ER stress might be exacerbated by increased IP₃R activity is important because pancreatic β-cells are innervated by cholinergic neurons that likely mobilize ER Ca²⁺ via these channels (48). It is therefore possible that increased nervous tone in the pancreas could aggravate ER stress during the development of diabetes. The expression of type 2 IP₃ receptors in β-cells has been shown to be upregulated by chronic hyperglycemia (15), suggesting that β-cell ER stress could also be an indirect complication of diabetes. Given the potential importance of this deleterious positive feedback loop, additional in vivo studies of the role of IP₃Rs in ER stress would be valuable.

Little is known about the role of intracellular Ca²⁺ channels in β-cell survival. Recently, we have shown that long-term (>2 days) inhibition of β-cell RyRs activates a programmed cell death pathway that is distinct from the one triggered by SERCA inhibition and instead shares traits with hypoglycemia-induced cell death (17,23). Instead of requiring caspase-3, ryanodine-induced apoptosis requires calpain-10, a human diabetes susceptibility gene (17). In contrast to thapsigargin-induced apoptosis, cell death resulting from RyR inhibition is associated with a decrease in the ATP-to-ADP ratio (23), and increasing the β-cell metabolic rate with high glucose completely abrogates ryanodine-induced programmed cell death (17). For this reason, studies designed to look at the effects of ryanodine on thapsigargin-induced cell death were typically performed in high glucose and on a time scale that precedes the deleterious effects of RyR block (17). Thus, it appears that disruption of β-cell ER Ca²⁺ handling can trigger multiple apoptosis pathways, but only SERCA inhibition leads to robust CHOP induction and caspase-3 activation. The same ER Ca²⁺ pool can thus control several types of cell death, depending in part on whether the filling or release of this pool is disrupted. Our results do not preclude the possibility that intracellular Ca²⁺ channels localized to non-ER compartments, such as insulin granules or endosomes (18,49), might also affect β-cell survival.

The present study provides insight into the complex sequence of events initiated by reduced SERCA activity in β-cells. We demonstrate clear roles for both ER Ca²⁺ levels and mitochondrial membrane potential in ER stress–induced β-cell apoptosis. Our results support that SERCA dysfunction can induce ER stress by rapidly depleting the luminal Ca²⁺ required for proper ER chaperone function and protein folding (11). No difference was observed in the absolute depletion of the ER between cells treated with thapsigargin alone and thapsigargin plus carbachol. Combined SERCA inhibition and cholinergic activation, however, did result in increased cell death and more rapid and homogeneous depletion kinetics. This suggests that absolute Ca²⁺ levels and the kinetics of luminal Ca²⁺ depletion together dictate the degree of ER stress, perhaps by rapidly modulating PERK phosphorylation. Subsequent to the initial ER-mediated events, there was an early mitochondrial hyperpolarization, in agreement with our previous observation that thapsigargin increased ATP production (23). In agreement with previous work, we found that ER-derived cytosolic Ca²⁺ signals could be sensed by nearby mitochondria (43,50). This might underlie the increased ATP synthesis, as Ca²⁺ in the mitochondrial matrix stimulates respiration via calcium-dependent dehydrogenases (43). On the other hand, excessive mitochondrial Ca²⁺ can trigger the loss of mitochondrial integrity and cell death (9,46,50). We found that β-cell mitochondria eventually fell into a depolarized, “dead” state, but the seemingly transient nature of the thapsigargin-induced Ca²⁺ rise (Fig. 7D) suggests this involves factors other than mitochondrial Ca²⁺ overload or direct ER-mitochondria signaling, per se. However, augmentation of this mitochondrial demise by carbachol indicates some role for Ca²⁺ in this late component of ER stress and CHOP-dependent apoptosis.

In conclusion, we have performed measurements of Ca²⁺ dynamics in the lumens of β-cell organelles, combined with real-time measurements of caspase-3 activation and dynamic assays of β-cell death, to provide a detailed picture of the events mediating β-cell apoptosis associated with disrupted ER Ca²⁺ homeostasis. Our study revealed the involvement of mitochondria in the execution of β-cell ER stress and cell death that results from reduced SERCA activity. Most importantly, we demonstrated a role for the ER Ca²⁺ release channels in the regulation of this important β-cell ER stress and apoptosis pathway. Together, these observations provide new insight into the mechanisms by which β-cells die in diabetes, knowledge that is essential for therapeutic efforts to reduce β-cell death.

This work was supported by operating grants (to J.D.J.) from the Canadian Institutes of Health Research (CIHR), the Canadian Diabetes Association (CDA), and the Natural Sciences and Engineering Research Council. D.S.L. was supported by a Juvenile Diabetes Research Foundation (JDRF) Fellowship. J.D.J. was supported by scholarships from Michael Smith Foundation for Health Research, the CDA, and JDRF and is a CIHR New Investigator.

No potential conflicts of interest relevant to this article were reported.

We thank Drs. Roger Tsien and Amy Palmer for the D1ER and mt4D3cpv constructs and Dr. Atsushi Miyawaki for the MiCy-DEVD-mKO construct.