Glucose and Endoplasmic Reticulum Calcium Channels Regulate HIF-1β via Presenilin in Pancreatic β-Cells*

Pancreatic β-cell death is a critical event in type 1 diabetes, type 2 diabetes, and clinical islet transplantation. We have previously shown that prolonged block of ryanodine receptor (RyR)-gated release from intracellular Ca2+ stores activates calpain-10-dependent apoptosis in β-cells. In the present study, we further characterized intracellular Ca2+ channel expression and function in human islets and the MIN6 β-cell line. All three RyR isoforms were identified in human islets and MIN6 cells, and these endoplasmic reticulum channels were observed in close proximity to mitochondria. Blocking RyR channels, but not sarco/endoplasmic reticulum ATPase (SERCA) pumps, reduced the ATP/ADP ratio. Blocking Ca2+ flux through RyR or inositol trisphosphate receptor channels, but not SERCA pumps, increased the expression of hypoxia-inducible factor (HIF-1β). Moreover, inhibition of RyR or inositol trisphosphate receptor channels, but not SERCA pumps, increased the expression of presenilin-1. Both HIF-1β and presenilin-1 expression were also induced by low glucose. Overexpression of presenilin-1 increased HIF-1β, suggesting that HIF is downstream of presenilin. Our results provide the first evidence of a presenilin-HIF signaling network in β-cells. We demonstrate that this pathway is controlled by Ca2+ flux through intracellular channels, likely via changes in mitochondrial metabolism and ATP. These findings provide a mechanistic understanding of the signaling pathways activated when intracellular Ca2+ homeostasis and metabolic activity are suppressed in diabetes and islet transplantation.

Inappropriate activation of pancreatic ␤-cell death pathways is involved in the pathogenesis of type 1 diabetes, type 2 diabetes, and several other forms of the disease (1)(2)(3)(4). Increased pro-grammed cell death also plays a critical role in the failure of clinical islet transplantation (5), an otherwise promising therapy for diabetes. We have recently shown that intracellular Ca 2ϩ stores gated by the ryanodine receptor Ca 2ϩ release channel (RyR) 4 are essential for ␤-cell survival in vitro (6), but not for glucose-stimulated insulin release (7). Using islets from genetically engineered mice, we demonstrated that blocking RyR triggers a novel apoptosis pathway dependent on calpain-10 and resembling hypoglycemia-induced cell death (6). Despite the potential importance of this signaling network, other nodes in this pathway remain to be elucidated.
Multiple calcium-dependent modes of cell death have been described (6, 8 -10), but the pathways activated after RyR blockade are incompletely understood. We have recently obtained evidence suggesting that apoptosis subsequent to RyR inhibition does not involve an endoplasmic reticulum (ER) stress response despite the localization of RyR in the ER. 5 Excitotoxicity is unlikely to mediate ryanodine-induced apoptosis, as blocking RyR flux with a high dose of ryanodine suppressed, rather than enhanced, apoptosis caused by chronic hyperglycemia in ␤-cells (6). In neurons, blocking RyR with dantrolene also suppressed excitotoxicity and cell death caused by mutations in presenilin, a gene linked to familial Alzheimer disease (11,12). On the other hand, because Ca 2ϩ flux from the ER to the mitochondria in ␤-cells has been shown to contribute to the generation of ATP (13), chronic block of RyR might cause cell death by reducing metabolic activity. Our previous experiments suggest that calpain-10 represents a common effector for hypoglycemia-induced ␤-cell death and ryanodine-induced apoptosis (6). Other calpains have also been implicated in the response to hypoglycemia and hypoxia (14,15). In this regard, it is interesting that calpain-10 has recently been shown to localize to the mitochondria (16), where it would be strategically positioned to respond to changes in metabolic activity.
Changes in cellular metabolic activity induce the expression of hypoxia-inducible factors, transcription factors that can activate both pro-survival and pro-apoptotic programs. HIF-1␤ was reported to be the most significantly decreased gene in islets from type 2 diabetics, and mice lacking this gene had impaired glucose homeostasis (17,18). Arntl2, a homologue of HIF-1␤, has been identified as a type 1 diabetes candidate gene in the Iddm6 locus of the NOD mouse (19). However, the signals that regulate hypoxia-inducible factors in ␤-cells are poorly characterized.
Another family of proteins that has been implicated in the response to reduced cellular metabolic activity is the presenilins. Presenilin mutations lead to familial Alzheimer disease (20). Presenilin is a critical component of the ␥-secretase complex, which we have recently shown can regulate ␤-cell survival via the cleavage of Notch (21). Presenilins have also been implicated in intracellular Ca 2ϩ homeostasis, apoptosis, and the response to hypoglycemia in neurons (12,22,23), but their functional role in pancreatic islet cells has not been studied in detail. Moreover, although the effects of mutant presenilins have been extensively researched, there are few studies on the endogenous regulators of these critical proteins. Interestingly, hypoxia has been shown to regulate the expression and the alternate splicing of presenilin-2 (24 -26), possibly via binding sites in the presenilin-2 promoter specific for HIF-1 (25). Conversely, HIF-1␣ mRNA was identified as one of the most upregulated targets of presenilin-1 by both microarray (27) and differential display (28). It is not known whether ryanodineinduced ␤-cell death or hypoglycemia (6) triggers this pathway.
In the present study, we sought to identify novel components of a ␤-cell survival gene network linked to RyR. We also sought to clarify the mechanisms associated with ryanodine-induced ␤-cell death and compare the effects of blocking RyR with inhibition of IP 3 R and SERCA. Analogies between ryanodine-and hypoxia-induced cell death caused us to focus our study on two key pathways that have been implicated in programmed cell death, namely the hypoxia-inducible pathway and the presenilins. We found that blocking RyR and/or IP 3 R, but not SERCA pumps, induced the expression of presenilins, as well as a hypoxia-inducible factor. We also determined that overexpression of presenilin-1 increased the levels of HIF-1␤ protein.
Together these results provide the first evidence of a RyR-presenilin-HIF genetic network controlling ␤-cell survival.

EXPERIMENTAL PROCEDURES
Reagents and Cell Culture-Ryanodine was purchased from Molecular Probes (Eugene, OR) or Tocris (Ellisville, MO) or Calbiochem (La Jolla, CA). Xestospongin C was purchased from AG Scientific (San Diego, CA). Thapsigargin was purchased from Calbiochem or Sigma. Carbonyl cyanide m-chlorophenyl-hydrazone (CCCP) was from Calbiochem. All drugs were kept as 1000ϫ stocks dissolved in Me 2 SO. Human islets were provided by Dr. Garth Warnock as part of the Centre for Human Islet Transplantation and Beta-cell Regeneration Core and cultured as described (29) in RPMI 1640 medium containing 5 mM glucose, 10% fetal bovine serum, penicillin/streptomycin (Invitrogen). Islets were kept in a 37°C incubator with 5% CO 2 and saturated humidity and were used within 3 days of their isolation from donors.
MIN6 cells were cultured in 5 or 25 mM glucose Dulbecco's modified Eagle's medium with 10% fetal bovine serum and 2% penicillin and streptomycin (Invitrogen) as indicated. Cells were treated at a confluency of ϳ80%.
Single Cell Imaging-MIN6 cells were imaged through the ϫ100, 1.45 numerical aperture objective of a Zeiss 200M microscope. Images were deconvolved using Slidebook from Intelligent Imaging Innovations (Boulder, CO). For organelle studies, cells were transfected with fluorescent proteins targeted to ER using KDEL and calreticulin motifs or mitochondria using the pOCT sequence (gift from H. McBride, University of Ottawa). The ratiometric mitochondrial pericam was a gift from Tullio Pozzan, University of Padua. D1ER was a gift from Amy Palmer and Roger Tsien (University of California San Diego).
Measurement of Apoptosis-Apoptosis was measured in single dispersed human islet cells, cultured overnight as indicated, using the ApoPercentage kit (Biocolor). This kit uses a dye that stains cells as they undergo the membrane "flip-flop" event that mediates phosphatidylserine translocation to the outer leaflet (28). This event is considered diagnostic of apoptosis, but not necrosis. Apoptotic cells appeared bright pink against the white background of phenol red-free RPMI 1640 medium. Apoptosis in MIN6 cells was assayed using terminal deoxynucleotidyltransferase-mediated dUTP nick end-labeling (TUNEL) (Roche Applied Science) according to the manufacturer's instructions. MIN6 cells were counterstained with hematoxylin.
ATP/ADP Ratio Assay-MIN6 cells were grown on tissue culture-treated 96-well luminometer assay plates and exposed to drugs as indicated. Following culture, cells were lysed and ATP/ADP ratios were measured using a commercial kit (Alexis, Axxora, San Diego, CA) according to the manufacturer's instructions. A luminometer (Molecular Devices, Sunnyvale, CA) was used to read the plates.
Reverse transcriptase PCR was carried out as described (21) for 35 cycles. Primer sequences are available upon request.
Gene Overexpression in MIN6 Cells-A full-length presenilin-1 clone in a CMV promoter vector (pCMV6-XL5-PSEN1) was purchased from Origene (Rockville, MD). MIN6 cells were transfected with 1 g of pCDNA3 or green fluorescent protein (as controls) or pCMV6-XL5-PSEN1 using Lipofectamine 2000 (Invitrogen) according to the manufacturer's instructions. Cells were incubated for 48 h in a 37°C incubator with 5% CO 2 and were subsequently harvested and subjected to Western blot analysis as described above.

RESULTS
Relationship between ␤-Cell ER and Mitochondria-A close physical and functional relationship between the ER and mitochondria has been demonstrated in many cell types but has received less attention in ␤-cells. To assess the subcellular distribution of both organelles simultaneously, MIN6 cells were co-transfected with fluorescent proteins targeted to each organelle. ER was distributed widely throughout the cytoplasm, and mitochondria were typically found nearby (Fig. 1A). A sim-␤-Cell Metabolic Signaling ilar pattern was observed when mitochondria were labeled with a Ca 2ϩ -sensitive probe, the ratiometric pericam. Apparent mitochondrial Ca 2ϩ hotspots were observed near to ER (Fig.  1B). Together, these data suggest that ER and mitochondria have many sites of close proximity to each other in ␤-cells and illustrate the physical possibility of ER-to-mitochondria communication in these cells.
In other cell types, ER calcium can be mobilized through both IP 3 R and RyR Ca 2ϩ release channels. We employed the Ca 2ϩsensitive fluorescent protein-based probe D1ER cameleon (30) to determine whether these channels are functional on ␤-cell ER. Indeed, both a stimulatory dose of ryanodine and the IP 3 -generating agonist carbachol could release Ca 2ϩ from the ER lumen (Fig. 1C). This result provided further functional evidence that both RyRs and IP 3 Rs can be present in MIN6 ␤-cells.
Identification of RyR Channel Isoforms in ␤-Cells-The presence of RyR mRNA in pancreatic ␤-cells remains controversial (31). With the knowledge that RyRs can be subject to alternate splicing (32), we used PCR primers targeting multiple regions of the mRNA for each of the three RyR genes. These experiments demonstrated that all three RyR genes are expressed in human islets and MIN6 cells (Fig. 2, A and B). Exons 6 and 45 of the RyR2 were also detected using additional primers (not shown).

FIGURE 1. Relationship between endoplasmic reticulum (ER) and mitochondria in MIN6 cells.
A, analysis of MIN6 cells co-transfected with ECFP (blue) targeted to mitochondria and mRFP (red) targeted to ER. Note that mitochondria are often observed near ER structures, particularly in the "holes" lacking ER. B, co-transfection of MIN6 cells with ER-targeted mRFP and Ca 2ϩ -sensitive mitochondrial pericam. The rightmost panel indicates hot spots of high mitochondrial calcium (arrows). C, direct measurement of calcium in the lumen of the ER with the D1ER cameleon probe. Ryanodine and carbachol were able to liberate Ca 2ϩ from the same ER calcium pool in MIN6 cells (13 of 22 carbachol-responsive cells responded to ryanodine).
Overall, the expression of RyR2 appeared to be stronger and more consistent than that of RyR1 or RyR3. Together, these studies suggest RyRs are present in islet cells.
Next we employed pan-RyR antibodies to study the subcellular localization of these channels in MIN6 ␤-cells. Consistent with our previous studies in human ␤-cells (7), RyR immunoreactivity was not observed in insulin secretory granules but was found in endosomes (Fig. 2C). As expected, we also observed RyR staining in ER, especially in regions close to mitochondria (Fig. 2C). Together, these studies further point to a role for RyR in ␤-cell function.
Blocking ER Ca 2ϩ Channels Induces Apoptosis-We demonstrated in a previous study that chronically blocking RyR Ca 2ϩ release channels led to a calpain-10-dependent form of programmed cell death in mouse islets (6). In the present study, we sought to confirm and extend these findings using other assays and ␤-cell models. Treatment of dispersed human islet cells for 48 h with 10 M ryanodine increased the percentage of cells with exposure of phosphatidylserine on the outer plasma membrane to ϳ50% from ϳ14% in control cells (Fig. 3, A-C). Previous studies in primary T lymphocytes have demonstrated that this event can occur in caspase-3-independent modes of apoptosis (33). We also confirmed that MIN6 cells underwent programmed cell death in response to 48-h treatment with ryanodine, as indicated by increased terminal deoxynucleotidyltransferase-mediated dUTP nick end-labeling (TUNEL) staining (Fig. 3, D and E).
Our previous investigation also pointed to increased calpain-10 mRNA in ryanodine-treated primary mouse islets (6). In the present study, we examined calpain-10 protein levels in human islets. We found that a blocking dose of ryanodine increased calpain-10 protein, predominantly a lower molecular weight splice variant preferentially expressed in islets (6, 34) (Fig. 3F). Similar results were observed in MIN6 cells (not shown). Blocking ER IP 3 R Ca 2ϩ release channels with 1 M xestospongin C increased calpain-10 levels in human islets (Fig.  3F), but not MIN6 cells (not shown). Together, these results show that calpain-10 protein levels are increased when intracellular Ca 2ϩ release is blocked in human islets.
Effects of Intracellular Ca 2ϩ Channels and Pumps on ATP/ADP-Ca 2ϩ flux through intracellular release channels has been implicated in the control of mitochondrial function in many cell types, including ␤-cells (13). Our previous studies suggested the possibility that Ca 2ϩ flux from RyR-gated intracellular Ca 2ϩ stores was required to prevent a form of cell death that resembled hypoglycemia (6). To directly test the hypothesis that RyRs control ␤-cell metabolism, we examined the effects of Ca 2ϩ channel inhibitors on the ATP to ADP ratio using a luciferase-based assay in MIN6 ␤-cells. At both high and low glucose, 24-h treatment with a blocking dose of ryanodine significantly decreased the ATP to ADP ratio (Fig. 4). Thapsigargin increased the ATP to ADP ratio, suggesting the possibility that ATP was being spared from hydrolysis in the absence of SERCA activity. The mitochondrial uncoupler carbonyl cyanide m-chlorophenyl-hydrazone was used as a positive control, and its effect likely marks the lower limit of the ATP to ADP ratio. Thus, blocking Ca 2ϩ flux via RyR reduced the ATP to

␤-Cell Metabolic Signaling
ADP ratio by approximately half of what was seen with carbonyl cyanide m-chlorophenyl-hydrazone. In the low glucose condition, the ATP to ADP ratio was increased by opening RyR with a stimulatory low dose of ryanodine (6, 7) (Fig. 5). Activating RyR did not induce an additional increase in the ATP to ADP ratio in the presence of high glucose. Together, these studies indicate that normal Ca 2ϩ flux through RyR is critical for the control of cellular metabolism in ␤-cells.

Role of Specific ER Ca 2ϩ Channels in Hypoxia-inducible Factor Expression-
The observation that intracellular Ca 2ϩ release channels played a role in ␤-cell energy metabolism prompted us to examine the expression of hypoxia-inducible factors, which are also activated by non-hypoxic stresses, including hypoglycemia, in other cell types. HIF-1␤ was recently shown to be the most significantly altered mRNA in islets from type 2 diabetic patients and was necessary for ␤-cell function in mice (17). In human islets, blocking RyR with 100 M ryanodine markedly increased HIF-1␤ protein levels (Fig. 6A). We also compared the role of IP 3 receptors using the inhibitor xestospongin C and found that it did not alter HIF-1␤ levels in human islets. In MIN6 cells, blocking IP 3 R caused an increase in HIF-1␤ protein expression, whereas ryanodine had no consistent effect at this time point (Fig. 6B). Blocking SERCA with thapsigargin reduced HIF-1␤ expression in all cases. Together, these results indicate that when intracellular Ca 2ϩ flux is compromised in ␤-cells a hypoxia-associated pathway is affected. We also directly examined the effects of different glucose concentrations on HIF-1␤ protein levels. The expression of HIF-1␤ was markedly induced at low glucose and suppressed at high glucose in MIN6 cells (Fig. 6C). (25,27). Microarray analysis of MIN6 cells treated with 100 M ryanodine 6 pointed to the possibility that presenilin mRNA was increased when RyRs were blocked. To confirm the presence of presenilin proteins in ␤-cells, we performed immunofluorescent staining on MIN6 cells. Using deconvolution microscopy we observed that presenilin-1 was found in a punctate cytoplasmic and nuclear pattern in MIN6 cells (Fig. 7A) consistent with the observation that, in ␤-cells, presenilins are predominately found in the form of C-terminal fragments (see Fig. 8A below). Presenilin-2 was observed in a cytoplasmic pattern reminiscent of the Golgi apparatus (Fig. 7B). These findings are in agreement with the known subcellular location of presenilins in other cells types (35,36). In human islets, reducing the glucose or blocking RyR dramatically increased presenilin-1 levels, whereas blocking IP 3 R or SERCA did not (Fig. 7, C and E). Presenilin-2 levels were not consistently altered in human islets by hypoglycemia or Ca 2ϩ manipulation. In the MIN6 ␤-cell model, blocking either RyR or IP 3 R led to an increase in both presenilin-1 and presenilin-2 levels (Fig. 7F). This was observed at either 5 or 25 mM glucose (data not shown) and suggests that both channels play an important role in the regulation of this pathway in MIN6 cells. Interestingly, blocking SERCA pumps with thapsigargin decreased presenilin levels, especially in MIN6 cells cultured in low glucose. This response to thapsigargin was in line with its effects on HIF-1␤. These findings also correlate with the opposing effects of channel inhibitors and pump inhibitors on the ATP to ADP ratio. Together, these studies show that presenilin levels can be affected by Ca 2ϩ flux through intracellular Ca 2ϩ channels in pancreatic ␤-cells.

Effect of Presenilin Overexpression on HIF-1␤ Levels-
The experiments reported above demonstrated that blocking intracellular Ca 2ϩ release channels or hypoglycemia leads to the induction of presenilin-1 and HIF-1␤. To test whether the increase in presenilin-1 could be responsible for the rise in HIF-1␤ protein, we overexpressed presenilin-1 in the MIN6 6 J. D. Johnson, unpublished data.   APRIL 11, 2008 • VOLUME 283 • NUMBER 15 ␤-cell line. The resulting increase in presenilin-1 led to an upregulation of HIF-1␤ protein (Fig. 8A). This was not simply a nonspecific stress response to protein overexpression, because overexpression of green fluorescent protein did not induce an increase in HIF-1␤ protein (Fig. 8B). These results suggest that presenilin-1 is an upstream regulator of HIF-1␤ expression in ␤-cells, as in other cell types (27,28). This observation is consistent with the possibility that ryanodine-dependent HIF-1␤ up-regulation involves presenilin-1 in ␤-cells.

DISCUSSION
The present study was undertaken to elucidate the links between ER calcium channels, cellular metabolism, and gene expression in the context of programmed ␤-cell death pathways. Our study has three major findings. First, we demonstrated that RyRs are present in MIN6 cells and human islets and that blocking these channels reduced the bulk ATP/ADP ratio. This provides evidence that RyRs regulate cellular metabolism to influence ␤-cell function and survival. These effects on metabolism led to the second major finding, that blocking intracellular Ca 2ϩ channels or exposing ␤-cells to hypoglycemia induced the expression of HIF-1␤, an essential gene implicated in ␤-cell function (17). To our knowledge, this is the first evidence that intracellular Ca 2ϩ stores can alter HIF-1␤ levels in any cell type. The third key observation was that intracellular Ca 2ϩ channels and glucose controlled the expression of presenilin proteins and that an increase in presenilin-1 was sufficient to up-regulate HIF-1␤. Presenilins have previously been shown to alter Ca 2ϩ flux through RyR and IP 3 R, but this is the first evidence of ER Ca 2ϩ feedback on presenilins. Together, these studies elucidate a previously undescribed signaling network controlled by ER Ca 2ϩ in human islets and MIN6 ␤-cells. Differences in the relative roles of RyR and IP 3 R in human islets and MIN6 cells likely reflect distinct expression profiles of these channels between each cell type (6,7).
Altered intracellular Ca 2ϩ homeostasis and perturbed ␤-cell metabolism are hallmarks of diabetes (37,38). In addition, islet transplant failure is associated with hypoxia and related stresses during isolation, culture, and the post-transplantation period (39,40). In the present study, we show that blocking Ca 2ϩ flux   4). C, working model of the interactions between intracellular Ca 2ϩ stores, metabolism, hypoxia-inducible factors, and presenilins in the pancreatic ␤-cell. Ca 2ϩ release from either RyR can regulate mitochondrial ATP production. When cellular metabolism is sufficiently high, calpain-10, HIF-1␤, and presenilin levels remain low. Attenuation of cellular metabolism activates this signaling network, which influences the path to programmed cell death in ␤-cells.

␤-Cell Metabolic Signaling
from intracellular Ca 2ϩ channels induced the expression of HIF-1␤, a key member of the hypoxia-inducible factor family. The finding that HIF-1␤ levels could be acutely modulated in ␤-cells is particularly interesting because HIF-1␤ was recently shown to be dramatically reduced in islets from type 2 diabetic patients (17). ␤-Cell-specific knock out of HIF-1␤ led to diabetes in the mouse (17). Interestingly, inhibition of SERCA pumps with thapsigargin, which leads to ER stress-mediated apoptosis, down-regulated HIF-1␤ expression in our experiments. Although both thapsigargin and high ryanodine induce programmed ␤-cell death, we have found that these agents act via distinct mechanisms (6,41). It is possible that HIF-1␤ is up-regulated in an attempt to protect ␤-cells from ryanodine-induced death. It is also possible that an excess of HIF-1␤ may be deleterious in the ␤-cell. Several studies in other cell types have demonstrated pro-apoptotic effects of HIF-1␤, in addition to its anti-apoptotic actions (42,43). Similarly, although up-regulation of the related protein, HIF-1␣, is associated with the expression of protective genes, HIF-1␣ is also required for hypoxia-and hypoglycemia-induced apoptosis (44). Together, these studies point to a dual role for HIFs in the control of apoptosis.
A number of interesting parallels between neurons and ␤-cells have been established, including shared transcription factor networks during embryonic development (45). Diabetes and Alzheimer disease are risk factors for each other, and it is possible that these two diseases may share common molecular defects (46,47). The Alzheimer disease genes, presenilin-1 and presenilin-2, are key mediators in neurodegeneration (20,47) but their role in ␤-cell degeneration remains unclear. Previous studies have shown that presenilin-1 and -2 are present in human and rodent pancreas and that they localize to ␤-cells (48,49), but nothing was known about stimuli that might increase or decrease their expression or the signaling pathways they control in the endocrine pancreas. Our results confirm the presence of presenilin-1 and presenilin-2 in ␤-cells. In the case of presenilin-1, the bulk of this protein exists as nuclear and cytosolic C-terminal fragments. We also show that presenilin-1 levels can be controlled by glucose and Ca 2ϩ signals and that presenilin-1 can regulate HIF-1␤ expression. Like HIF-1␤, both an excess of presenilin activity and the loss of presenilin have been shown to increase cell death in other cell types (12,20,50).
Although it is not clear whether presenilins are involved in the generation of amyloid plaques seen in type 2 diabetes (51), it is known that presenilins can regulate cell differentiation and survival in both ␥-secretase-dependent and ␥-secretase-independent ways (52). We have recently shown that ␥-secretase activity is essential for adult ␤-cell survival (21). Multiple studies have linked the ␥-secretase-independent presenilin activity to the control of ER Ca 2ϩ release under basal conditions and during caspase-3-mediated apoptosis (53)(54)(55). In general, the presenilin-1 gain-of-function mutations in Alzheimer disease lead to an overload of releasable ER Ca 2ϩ , while presenilin-2 mutations have been reported to decrease Ca 2ϩ stores (56). Considerable evidence points to a direct link between presenilin and RyR (57)(58)(59). For example, presenilin-2 mutations have been shown to increase apoptosis and decrease differentiation by enhancing RyR-mediated Ca 2ϩ release (59,60). The hearts of presenilin-2 knock-out mice demonstrated greater amplitude of Ca 2ϩ transients and increased contractility, specifically under conditions of low Ca 2ϩ (57). In humans, presenilin mutations have been linked to cardiomyopathy and altered Ca 2ϩ homeostasis (61). Although previous studies have demonstrated that presenilins can have significant effects on intracellular Ca 2ϩ signaling, we show for the first time in any cell type a reciprocal effect, blocking intracellular Ca 2ϩ channels significantly increases presenilin levels. In the ␤-cell, presenilins appear to participate in a signaling network implicated in mitochondrial function and cell fate. Interestingly, presenilin-1 has been localized to the mitochondria in other cell types (62,63) and mitochondrial respiration was decreased in knock-in mice with a presenilin-1 gain-of-function mutation (64). Studies also point to the up-regulation of presenilins in hypoxic conditions (12,24,27,65), similar to our observation that hypoglycemia increases presenilin-1 expression. Together, these findings indicate that presenilins interact with intracellular Ca 2ϩ homeostasis and metabolism on many levels. Our studies mark an important step toward understanding the role of presenilins in pancreatic ␤-cells. The regulation of presenilin proteins by ␤-cell Ca 2ϩ flux may also shed light on the regulation of these proteins in other cell types.
In conclusion, we have shown that blocking intracellular Ca 2ϩ release channels can reduce ␤-cell metabolism and induce expression of HIF-1␤ via presenilin up-regulation (Fig.  8C). This novel ␤-cell signaling network is likely to play a key role in the process of islet transplantation, during which ␤-cells are subjected to considerable stress. The implication of HIF-1␤ in glucose homeostasis and human type 2 diabetes suggests possible roles for this pathway in vivo.