Constitutive cAMP response element binding protein (CREB) activation by Alzheimer's disease presenilin-driven inositol trisphosphate receptor (InsP₃R) Ca²⁺ signaling

Abstract

Mutations in presenilins (PS) account for most early-onset familial Alzheimer's disease (FAD). Accumulating evidence suggests that disrupted Ca²⁺ signaling may play a proximal role in FAD specifically, and Alzheimer's disease (AD) more generally, but its links to the pathogenesis of AD are obscure. Here we demonstrate that expression of FAD mutant PS constitutively activates the transcription factor cAMP response element binding protein (CREB) and CREB target gene expression in cultured neuronal cells and AD mouse models. Constitutive CREB activation was associated with and dependent on constitutive activation of Ca²⁺/CaM kinase kinase β and CaM kinase IV (CaMKIV). Depletion of endoplasmic reticulum Ca²⁺ stores or plasma membrane phosphatidylinositol-bisphosphate and pharmacologic inhibition or knockdown of the expression of the inositol trisphosphate receptor (InsP₃R) Ca²⁺ release channel each abolished FAD PS-associated constitutive CaMKIV and CREB phosphorylation. CREB and CaMKIV phosphorylation and CREB target gene expression, including nitric oxide synthase and c-fos, were enhanced in brains of M146V-KI and 3xTg-AD mice expressing FAD mutant PS1 knocked into the mouse locus. FAD mutant PS-expressing cells demonstrated enhanced cell death and sensitivity to Aβ toxicity, which were normalized by interfering with the InsP₃R–CAMKIV–CREB pathway. Thus, constitutive CREB phosphorylation by exaggerated InsP₃R Ca²⁺ signaling in FAD PS-expressing cells may represent a signaling pathway involved in the pathogenesis of AD.

Alzheimer's disease (AD) is a fatal neurodegenerative disease associated with cognitive decline and progressive neuronal atrophy and death. Although most AD is sporadic with late onset, familial AD (FAD) is early onset due to mutations in three genes: amyloid precursor protein (APP), presenilin 1 (PS1), and presenilin 2 (PS2). PS1 and PS2 homologs are components of the γ-secretase APP cleavage complex. Mutations in PS are associated with AD pathogenesis, including altered γ-secretase–mediated APP cleavage and accumulation of β-amyloid (Aβ) plaques (1). The “amyloid hypothesis” proposes that Aβ accumulation triggers neurodegeneration (1). Nevertheless, whether tau and Aβ aggregations are proximal causes or symptoms of AD is a matter of debate (2). Accumulating evidence implicates disruption of intracellular calcium (Ca²⁺) signaling as a proximal event in AD, suggesting that it could play a role in AD pathogenesis. Many neuronal functions are regulated by intracellular Ca²⁺ signals, and maintenance of their dynamics is critical for proper neuronal activity (3). Several previous studies have demonstrated consistent effects of expression of FAD mutant PS on exaggerated endoplasmic reticulum (ER) Ca²⁺ release in different cell types, including cortical neurons in brain slices from FAD PS1 knock-in mice (2, 4–8) suggesting that it is a fundamental alteration in FAD. Exaggerated ER Ca²⁺ release may be caused by lack of a putative ER membrane Ca²⁺ leak function of PS (9) or by activation of the sarco/ER Ca²⁺-ATPase (SERCA) pump (8). FAD PS1 and PS2 interact biochemically and functionally with the inositol trisphosphate receptor (InsP₃R) Ca²⁺ release channel, increasing its activity in response to low [InsP₃] and allowing it to release excess Ca²⁺, even in resting conditions (10, 11). Despite the uncertainties of molecular mechanisms involved in exaggerated ER Ca²⁺ release in FAD PS-expressing cells, the consequences of chronic excessive Ca²⁺ release are relatively neglected in the “Ca²⁺ hypotheses” of AD. Identification of downstream effects might help discriminate among models proposed for the mechanisms of exaggerated Ca²⁺ signaling, and help define their roles in AD pathogenesis.

Many neuronal processes regulated by Ca²⁺ involve changes in gene expression. The Ca²⁺-sensitive transcription factors Ca²⁺/cAMP response element binding protein (CREB) can be activated by various kinases in response to electrical activity, neurotransmitters, hormones, and neurotrophins, among others, promoting expression of many genes that contain cAMP response elements (CREs) (12, 13). Multiple signaling cascades converge onto CREB phosphorylation, including Ca²⁺/calmodulin kinase (CaMK), ras/MAPK ERK1/2 (14), and protein kinases A and C (15). CREB plays a central role in memory formation (16). Despite the loss of cognitive ability in AD, the relationship of FAD PS mutations and CREB activity has received relatively little attention (17). In the present work, we examined the consequences of FAD mutant PS1 and PS2 expression on CREB activation. Our results, obtained in neural cells and brain neurons, reveal that FAD mutant PS causes constitutive CREB activation and CREB target gene expression as a result of constitutive InsP₃R-mediated activation of CaMK pathways. This signal transduction pathway contributes to increased apoptosis observed in FAD PS-expressing cells, and it sensitizes cells to Aβ-induced death.

Results

Mutant FAD PS1 Enhances InsP₃R-Dependent Spontaneous [Ca²⁺]_i Signaling and Constitutive CREB Phosphorylation.

We previously showed that mutant FAD PS1 increases the frequency of spontaneous [Ca²⁺]_i oscillations in human B lymphoblasts derived from patients with FAD (10). In the present work, we observed a similar phenotype in neuronal cells. Human SH-SY5Y neuroblastoma cells stably expressing either PS1 WT or mutant PS-M146L were loaded with fura-2 and perfused with DMEM containing 10% FBS at 37 °C. Spontaneous Ca²⁺ signals were rare in control and PS1-WT cells, whereas constitutive [Ca²⁺]_i spiking activity was observed in nearly 50% of the PS1-M146L–expressing cells (Fig. 1 A–C and Table S1), with higher amplitude and frequency than those seen in control and PS1-WT cells (Table S1). Baseline [Ca²⁺]_i was similar in all lines. Exaggerated Ca²⁺ signaling specifically in FAD-PS1 cells also was observed in sympathetic neuron-derived PC12 cells (Fig. S1). FAD PS1 and PS2 interact with InsP₃R and stimulate its gating, particularly in response to low levels of InsP₃ (10, 11). Such an effect would be expected to result in more frequent and possibly higher-amplitude Ca²⁺ spiking, as we observed (Fig. 1 A–C). To address the role of InsP₃R in the enhanced spontaneous Ca²⁺ signaling observed in PS1-M146L–expressing SH-SY5Y cells, we pharmacologically inhibited InsP₃R with xestospongin B (XeB), a potent membrane-permeable specific inhibitor of InsP₃R-mediated Ca²⁺ release (18), or genetically with siRNA against InsP₃R-1, the predominant isoform in neurons (19). Both approaches strongly reduced spontaneous Ca²⁺ signaling in PS1-M146L–expressing cells (Fig. 2 A–C and Table S2), suggesting that FAD PS1-M146L–induced exaggerated Ca²⁺ signaling in SH-SY5Y cells is caused by its stimulatory effects on InsP₃R gating and Ca²⁺ release.

Fig. 1.

Constitutive CREB phosphorylation in mutant PS1-M146L–expressing cells. (A–C) Representative spontaneous [Ca²⁺]_i signals demonstrating the spectrum of signals observed among different untransfected and stable PS1-WT– and PS1-M146L–expressing SH-SY5Y cells, respectively. (D) Summary and representative Western blots showing the effects of PS1 on P-CREB in stable SH-SY5Y cells under basal conditions. P-CREB was normalized to T-CREB. n = 7 experiments. (E) Effects of PS1 RNA_i on basal P-CREB in SH-SY5Y–stable lines. CTL was exposed to scrambled siRNA.

Fig. 2.

Spontaneous Ca²⁺ release in PS1-M146L–expressing cells is responsible for constitutive CREB phosphorylation. (A–C) Representative spontaneous single-cell [Ca²⁺]_i oscillations, representing the spectrum of behaviors observed in PS1-M146L–expressing SH-SY5Y cells under basal conditions (A), after 30 min of XeB (2 μM) (B), and after 48 h of transient transfection with InsP₃R-1 siRNA (C). (D–F) Densitometric analyses and representative Western blots of P-CREB in SH-SY5Y cells after 30 min of incubation in the absence or presence of InsP₃R inhibitor XeB (XB, 2 μM; n = 3 experiments) (D), after 48 h of transient transfection with InsP₃R-1 siRNA (R) or nontargeting siRNA (NT) (E), and in the absence or presence of rapamycin (Rap, 100 nM) (F). SH-SY5Y cells stably expressing WT or FAD PS1 were transfected with pmFRB-FKBP12-5ptase. After 48 h, cells were preincubated with rap for 30 min. Some gels are digitally rearranged for presentation purposes. n = 3 experiments. Data are expressed as mean ± SEM. *^,#P < 0.05; **^,##P < 0.01.

FAD PS-mediated enhanced InsP₃R channel gating and Ca²⁺ release have been observed in neural and other cells, but the cellular biological consequences of these effects are unknown. Many neuronal processes are regulated by Ca²⁺-dependent changes in gene expression (20); for example, CREB is essential for long-term memory formation, a feature that is severely impaired in patients with AD (17, 21). Accordingly, we evaluated CREB activation, inferred from the level of ser133 phosphorylation (22) detected by immunoblot analysis of phospho-ser133 (P-CREB), normalized to total CREB (T-CREB). Notably, CREB phosphorylation was constitutively elevated in the FAD PS1-M146L–expressing cells, whereas PS1-WT–expressing cells exhibited much lower basal CREB phosphorylation, similar to controls (Fig. 1D). Comparable studies in transiently transfected PC12 cells yielded similar results (Fig. S2A). P-CREB accumulated in nuclei in basal conditions specifically in cells expressing mutant PS1 (Fig. S2B), consistent with the constitutive hyperphosphorylation seen in Western blot analyses. In contrast, localization was largely cytoplasmic in PS1-WT–expressing cells, similar to that observed in EGFP controls (Fig. S2B). These data suggest that expression of FAD PS1-M146L specifically enhances constitutive CREB phosphorylation. In agreement, siRNA knockdown of PS1 (by ∼70% after 48 h; Fig. S2C) in SH-SY5Y cells stably expressing PS1-M146L or PS1-WT blocked constitutive CREB hyperphosphorylation in PS1-M146L–expressing cells (Fig. 1E). Constitutively elevated P-CREB also was observed in cells expressing FAD PS2. FAD PS2-N141I–expressing PC12 cells had strongly enhanced constitutive P-CREB, whereas P-CREB in PS2-WT–expressing cells was similar to that observed in controls (Fig. S2D). Similar experiments in chicken DT40 cell lines recapitulated the results observed in SH-SY5Y and PC12 cells (Fig. S2E). These data reveal a direct relationship between FAD mutant PS expression and constitutively enhanced CREB activity in several cell lines.

We explored the relationship between FAD PS-mediated spontaneous [Ca²⁺]_i signaling and enhanced CREB phosphorylation by inhibiting the InsP₃R with XeB (2 μM for 30 min). Notably, XeB reduced constitutive CREB hyperphosphorylation in PS1-M146L–expressing SH-SY5Y cells to normal levels, whereas P-CREB levels were unaffected in PS1-WT and control cells (Fig. 2D). siRNA knockdown of InsP₃R-1 (>70%; Fig. S2F) had a similar effect, whereas a pool of nontargeting siRNAs had no effect (Fig. 2E). Similar effects of depression of InsP₃R activity by XeB or siRNA knockdown also were seen in transiently transfected PC12 cells specifically expressing PS1-M146L (Fig. S3 A and B). To further confirm the role of InsP₃R-mediated Ca²⁺ signaling in constitutive CREB phosphorylation, we used a molecular approach to rapidly deplete the InsP₃ precursor PIP₂ (23). SH-SY5Y cells stably expressing PS1 were transfected with bicistronic plasma membrane (PM)-recruitable 5-phosphatase fused to FKBP12 and a PM-targeted fragment of mammalian target of rapamycin (FRB) that binds FKBP12 in the presence of rapamycin. On the addition of rapamycin, 5-ptase rapidly accumulates at the PM by binding to FRB, localizing it in proximity to its substrate PIP₂, which it rapidly depletes (23). At 48 h after transfection, cells were incubated for 30 min in the presence or absence of rapamycin (100 nM). Constitutive CREB phosphorylation was significantly reduced by rapamycin, with the exaggerated levels in PS1-M146L–expressing cells normalized to control levels (Fig. 2F). These results further support a model in which FAD PS-mediated signaling through the InsP₃R is responsible for the constitutive activation of CREB.

CaMKIV Is Constitutively Phosphorylated in FAD Mutant PS1-Expressing Cells.

A major Ca²⁺-sensitive mechanism that can mediate CREB ser133 phosphorylation is Ca²⁺/calmodulin-dependent protein kinase (CaMK), in particular CaMKIV, which induces fast and strong CREB activation during neuronal stimulation (14, 15, 24, 25). To examine whether CaMKIV might couple enhanced FAD PS-mediated InsP₃R Ca²⁺ signaling to constitutive CREB phosphorylation, we evaluated CaMKIV activity by measuring levels of phospho-CaMKIV, normalized to total levels of CaMKIV (T-CaMKIV). PS1-M146L–expressing SH-SY5Y cells exhibited elevated P-CaMKIV under basal conditions, reminiscent of CREB constitutive hyperphosphorylation, whereas there was little difference between PS1-WT and control cells (Fig. 3A). P-CaMKIV also was markedly elevated under basal conditions in PC12 cells specifically expressing FAD mutants PS1-M146L and PS2-N141I (Fig. S3 C and D). Constitutive activation of both P-CREB and P-CaMKIV in basal conditions specifically in FAD PS-expressing cells suggested that CaMKIV might be an upstream activator of CREB in these cells. To explore the relationship between CaMKIV and CREB, we examined responses of P-CREB to CaMKIV inhibition in stable SH-SY5Y cells. Notably, the specific CaMK inhibitor KN93 reduced constitutive CREB phosphorylation in FAD mutant PS1-expressing cells to levels observed in control cells, whereas the inactive KN92 analog had no effect (Fig. 3B). Similar results were observed in PS1-M146L–expressing PC12 cells (Figs. S3E and 4B). The CaM kinase-signaling cascade contains upstream kinases, including Ca²⁺/CaM kinase kinase β (CaMKKβ) (26), which phosphorylates CaMKIV and enhances its ability to phosphorylate CREB. The specific CaMKKβ inhibitor STO-609 completely suppressed constitutive hyperphosphorylation of both CREB and CaMKIV in PS1-M146L–expressing SH-SY5Y cells (Fig. 3F and Fig. S4D) and PC12 cells (Figs. S3F and S4C). These results indicate that expression of FAD mutant PS1 results in constitutive activation of a CaMK cascade involving CaMKKβ and CaMKIV that leads to enhanced CREB phosphorylation.

Fig. 3.

Constitutive CaMKIV phosphorylation is upstream of enhanced CREB activation in FAD PS-expressing cells. (A) Basal CaMKIV phosphorylation in SH-SY5Y cells stably transfected with PS1-WT or PS1-M146L, and representative Western blots of T-CaMKIV and P-CaMKIV. n = 4 experiments. (B) CREB phosphorylation in stable SH-SY5Y cells preincubated for 30 min in the presence of CaMK inhibitor KN93 or inactive analog KN92 (10 μM). n = 3 experiments. (C–E) Densitometric analyses and representative Western blots of P-CaMKIV in SH-SY5Y cells after 30 min of incubation in absence or presence of InsP₃R inhibitor XeB (XB, 2 μM; n = 3 experiments) (C), after 48 h of transient transfection with InsP₃R-1 siRNA (R) or nontargeting siRNA (NT) (D), and in the absence or presence of rapamycin (rap, 100 nM) (E). Stable SH-SY5Y cells were transfected with pmFRB-FKBP12-5ptase (E). After 48 h, cells were preincubated with rapamycin for 30 min. n = 3 experiments. (F) Effects on P-CREB in the absence or presence of the CaMKKβ inhibitor STO-609 (10 μM for 30 min). Results are summarized from three experiments for each cell type. Some gels are digitally rearranged for presentation. Data are expressed as mean ± SEM. *^,#P < 0.05; **^,##P < 0.01.

Fig. 4.

CREB and CaMKIV are constitutively activated in presymptomatic 3xTg-AD mouse brain. Expression of T-and P-CREB (A; n = 5) and CaMKIV (B; n = 4), and levels of CREB target genes c-fos (C), nNOS (D), and BDNF (E) and control gene GAPDH (F) in whole-brain extracts from 4- to 6-wk-old 3xTg-AD mice. Mice of the same background strain, sex, and age were used as controls (n = 4). Protein levels were normalized to β-tubulin (β-tub). Data are expressed as mean ± SEM. *P < 0.05; **P < 0.01.

Because CREB hyperphosphorylation observed in cells expressing FAD mutant PS required InsP₃R activity (Fig. 2 D and E), we speculated that FAD PS-mediated constitutive activation of the CaMK cascade had a similar dependence. To test this, we blocked the InsP₃R with XeB (Fig. 3C), knocked down InsP₃R-1 expression (Fig. 3D), or depleted PIP₂ in SH-SY5Y cells (Fig. 3E) and PC12 cells (Fig. S5). All three approaches reduced constitutive CaMKIV hyperphosphorylation in PS1-M146L–expressing cells but had little effect in either PS1-WT or control cells (Fig. 3 C–E and Fig. S5). These results suggest that FAD PS-associated signaling through the InsP₃R is responsible for the constitutive activation of a CaMK cascade that impinges on CREB phosphorylation.

FAD PS1 Enhances CREB Phosphorylation and Transcriptional Activity in 3xTg-AD and M146V-PS1_ki Mouse Brains.

Our results indicate that constitutive CREB phosphorylation is a consistent feature associated with FAD PS expression in different cell lines. To explore the relevance of these findings in animal models, we used M146V-PS1_ki and 3xTg-AD (M146V-PS1_ki, APP_swe, Tau_P301L) mice. The PS1 mutation is the predominant “calciopathic” factor in cortical neurons, given that exaggerated ER Ca²⁺ signals are similar in M146V-PS1_ki and 3xTg-AD mice (4). Because exaggerated ER-derived Ca²⁺ signals are detected at all ages, whereas Aβ and tau accumulations are observed in 3xTg-AD mice only at age >6 mo, we examined P-CREB in whole-brain homogenates from 4- to 6-wk-old presymptomatic 3xTg-AD mice and non-Tg mice of the same background. Both P-CREB and P-CaMKIV were highly elevated (by more than threefold and fivefold, respectively) in the 3xTg-AD mice compared with controls (Fig. 4 A and B), in striking agreement with results obtained in SH-SY5Y and PC12 cells. Because CREB is a key mediator of Ca²⁺-inducible transcription, we examined the expression of three CREB target genes chosen based on the presence of two or more CRE sites in their promoter regions. Notably, 3xTg-AD mouse brains demonstrated significantly higher expression of c-fos, nNOS, and BDNF (Fig. 4 C–E). GAPDH expression, the transcription of which is independent of CRE sites, did not differ in 3xTg-AD and nontransgenic mice (Fig. 4F). To ensure that the observations in 3xTg-AD mice were not complicated by expression of APP and tau transgenes, we conducted a similar set of experiments using 4- to 6-wk-old M146V-PS1_ki mice that never develop amyloid or tau accumulation. The findings in these mice are in remarkable agreement with those in the 3xTg-AD mice (Fig. S6). Furthermore, c-fos and nNOS also were elevated specifically in PC12 cells expressing PS1-M146L (Fig. S7 A–C). These results suggests that constitutive hyperphosphorylation of CREB in the FAD mutant PS-expressing cells is associated with changes in normal gene expression in vivo and in vitro.

Enhanced Cell Death and Aβ Toxicity in FAD PS Cells Is Abolished by Inhibition of Constitutive InsP₃R-CaMKIV-CREB Signaling.

AD is associated with significant neuronal atrophy and death (27, 28). We consistently observed an approximately fivefold increase in apoptotic cell death in PS1-M146L–expressing SH-SY5Y cells compared with either control or PS1-WT–expressing cells (Fig. 5A), in agreement with previous observations (27–30). We asked whether constitutive InsP₃R-CaMKIV-CREB signaling contributed to the enhanced death of FAD PS-expressing cells. Inhibition of CaMK activity (Fig. 5B) or RNAi knockdown of either CREB (Fig. 5D and Fig. S7D) or InsP₃R-1 (Fig. 5F) each significantly reduced enhanced apoptosis in PS1-M146L–expressing cells to levels observed in PS1-WT and control cells. Inhibition of this signaling axis enhanced apoptosis of PS1-WT and untransfected cells, likely due to the requirement for InsP₃R signaling to maintain normal cell bioenergetics (31).

Fig. 5.

Enhanced spontaneous and Aβ-induced cell death in FAD PS1-expressing cells is abolished by inhibition of InsP₃R-CaMKIV-CREB signaling. (A) Viability of control and stable PS1-M146L– or PS1-WT–expressing SH-SY5Y cells after 48 h in the presence of Aβ 1–40 oligomers or scrambled peptide as control (10 μM each; n = 4). (B–G) Effects of CaMKKβ inhibitor STO-609 (2.5 μM) (B and C), CREB siRNA (D and E), or InsP₃R-1 siRNA (F and G) on basal (B, D, and F) and Aβ-induced (C, E, and G) cell death (n = 3 each). Controls in D–G were exposed to nontargeting siRNA. Data are expressed as mean ± SEM. *P < 0.05; **P < 0.01.

Exposure to Aβ 1–40 oligomers (10 μM) for 48 h was cytotoxic for all cells, with FAD PS1 cells the most vulnerable (Fig. 5A). Strikingly, the enhanced Aβ toxicity in FAD-PS cells was normalized by inhibition of CaMK activity (Fig. 5C) or RNAi knockdown of CREB (Fig. 5E) or InsP₃R-1 (Fig. 5G). These data suggest that FAD PS expression enhances apoptosis and susceptibility to Aβ toxicity by constitutive activation of InsP₃R-CaMKIV-CREB signaling.

Discussion

The etiology of AD remains under debate, with age the main known risk factor but major molecular mechanisms still unclear. The identification of three components in FAD (PS1, PS2, and APP) that are linked in a biochemical pathway that impinges on Aβ production has strongly influenced acceptance of the amyloid hypothesis (1). Nevertheless, there is also compelling evidence of altered Ca²⁺ signaling associated with in vivo expression of FAD mutant PS long before the appearance of plaques or tangles and in a variety of heterologous expression systems (2, 4–8). The relationships among mutant PS-mediated Ca²⁺ signaling, Ca²⁺ regulation of effector systems, and subsequent AD-related pathophysiology have been largely ignored. Constitutive Ca²⁺ signaling may alter transcriptional programs that shape cell physiology and responses to stress (15). Here we focused on the Ca²⁺-regulated transcription factor CREB, which has multiple roles in neuronal physiology, and evaluated in cell and animal models whether FAD PS-induced dysregulation of Ca²⁺ homeostasis affects its activity. Our results indicate that CREB is constitutively activated specifically in cells expressing FAD PS, through a pathway involving constitutive activation of a Ca²⁺-CaMK cascade driven by Ca²⁺ release from internal stores by the InsP₃R. This pathway alters gene expression in the brain and enhances cell death and the cytotoxic effects of Aβ oligomers.

Our results demonstrate that P-CREB is strongly increased in cells expressing FAD PS. Expression of either PS1-M146L or PS2-N141I in cultured cells or PS1-M146V at physiological levels in two transgenic mouse models had strong and similar effects on P-CREB that were not observed in cells expressing WT PS or in control cells, or in brain lysates from control mice. A remarkable finding was the constitutive hyperphosphorylation of CREB in cells expressing the FAD PS in different sets of data and in different cells lines, as well as in brain lysates from transgenic mice. Several kinases can phosphorylate CREB, with the Ca²⁺/CaM cascade and MAPK pathway the best described (15, 25). Although FAD PS expression has been associated with altered MAPK-ERK responses to agonists (32, 33), we found no evidence that constitutively phosphorylated CREB in FAD PS-expressing cells is correlated with ERK activity (Fig. S8 A and B). In contrast, our results strongly implicate a Ca²⁺–CaMK pathway. Constitutive CREB hyperphosphorylation was associated with constitutive CaMKIV hyperphosphorylation specifically in FAD PS-expressing cells, and CaMKIV and CaMKKβ inhibitors reduced the constitutive P-CREB observed in FAD PS-expressing cells.

Ca²⁺ from distinct sources can activate CREB. Most studies in the nervous system have focused on Ca²⁺ influx through voltage-gated Ca²⁺ channels or NMDA receptors (25, 34), but Ca²⁺ released from intracellular stores also can activate CREB (35–37). Our results implicate Ca²⁺ release from ER stores as the major mechanism for constitutive CREB and CaMKIV phosphorylation in FAD PS-expressing cells. Depletion of stores normalized CaMKIV and CREB constitutive phosphorylation to control levels, whereas inhibition of voltage-gated Ca²⁺ channels had no effect (Fig. S8 C and D). Identification of intracellular stores as the source of the Ca²⁺ that drives constitutive CaMKIV and CREB phosphorylation is notable, because the best-known manifestation of altered Ca²⁺ homeostasis by FAD PS is enhanced Ca²⁺ release during cell stimulation (2, 5–7, 10, 11, 38). FAD PS1 or PS2 enhance InsP₃R sensitivity to low concentrations of InsP₃, stimulating low-level Ca²⁺ signaling in resting conditions (10, 11). Our results reported here suggest that this is the molecular basis for constitutive CREB phosphorylation in FAD PS cells. FAD PS1-expressing cells exhibited spontaneous [Ca²⁺]_i oscillations that were suppressed by inhibiting or reducing expression of the InsP₃R. Importantly, both maneuvers suppressed constitutive hyperphosphorylation of both CREB and CaMKIV. Furthermore, depletion of PIP₂, the substrate for phospholipase C that generates the InsP₃R ligand InsP₃, also normalized constitutive CREB and CaMKIV hyperphosphorylation. These results support the proposal that FAD PS causes exaggerated Ca²⁺ signaling through an InsP₃R-dependent mechanism (10, 11) that impinges on a CaMK cascade that results in constitutive CREB phosphorylation. Of note, because the InsP₃R is present on both nuclear membranes (39, 40), excessive InsP₃R activity is poised to strongly influence nuclear [Ca²⁺], which is critical for the activation of nuclear proteins CaMKIV and CREB.

Constitutive CREB hyperphosphorylation in FAD PS-expressing cells was associated with enhanced expression of c-fos and nNOS. Notably, the same two proteins were up-regulated in brains of 3xTg-AD and M146V-PS1_ki mice. Recapitulation of cell culture observations of constitutively elevated P-CREB and P-CaMKIV, as well as CREB target gene expression in vivo, suggest that the proposed mechanisms involving FAD PS-enhanced InsP₃R Ca²⁺ release are relevant in vivo. The 3xTg-AD mice used in these studies were much younger than the age at which Aβ accumulation, plaque deposition, and tangles occur. The M146V-PS1_ki mice do not display such pathology at any age. Thus, the observed constitutively activated CaMK cascade and CREB transcriptional programs are early, proximal manifestations that precede and are independent of the appearance of classical AD features.

AD is associated with significant neuronal atrophy and death (27). Previous studies have found that the expression of FAD PS enhances apoptosis, with altered Ca²⁺ homeostasis and oxidative stress proposed to play important roles (41, 42), but the molecular mechanisms have remained unclear. We also observed that FAD PS enhanced basal levels of cell death. Surprisingly, we found that inhibiting constitutive CREB activity enhanced cell survival to control levels. These results suggest that FAD PS-induced constitutive activation of the InsP₃R–CaMKIV–CREB pathway could possibly play a role in enhanced cell death in FAD. CREB can be activated by many stimuli, including growth and survival factors and in response to stress (43). CREB-mediated transcription plays important roles in neuronal survival (44) and learning and memory (45). In general, CREB levels are decreased in the brains of patients with AD (17) and in old 3xTg-AD mice (46). However, the diminished CREB in mice has been shown to be a consequence of Aβ accumulation (46), suggesting that reduced CREB observed in human AD brains may be a consequence of AD pathology. In contrast, our findings demonstrate an early proximal effect that is independent of and precedes AD-like pathology. The effects of CREB depend on its level and the duration of activation, as well as on the developmental stages of specific brain regions (25, 47). Chronic CREB activation leads to memory deficits (48), cognitive decline (49), interference with information retrieval (48), and neurodegeneration (50). It is possible that the constitutive CREB activation caused by FAD PS expression reported here could contribute to the premature cognitive deficits normally associated with aging.

Because CREB regulates numerous genes, its constitutive activation as a consequence of FAD PS expression may alter gene expression in the brains of patients with AD that may be related to disease pathogenesis. Constitutive CREB activation caused chronic expression of the immediate early gene c-fos, as well as nNOS, in cell cultures and two different AD mouse models with mutant PS1 knocked in. c-fos levels are increased in the brains of patients with AD (51–53), and whereas transient c-fos expression is associated with differentiation, proliferation, and excitability (54), its constitutive (55) or strong (56) expression drives cell degeneration and death. nNOS accounts for most nitric oxide in the nervous system, where it regulates gene regulation, plasticity, and excitotoxic injury (57). Nevertheless, hippocampal nNOS up-regulation is associated with chronic depression and impaired neurogenesis (58) and precedes memory loss and Aβ processing alterations (59). Moreover, high nitro-oxidative stress can initiate a cascade of redox reactions that can have cytotoxic effects. The brains of patients with AD exhibit evidence of reactive oxygen species– and reactive nitrogen species–mediated injury, supporting the idea that oxidative stress is a feature of AD brain pathology (60). Whether constitutive expression of c-fos or nNOS contributes to InsP₃R-CaMKIV-CREB–mediated enhanced cell death remains to be determined, however.

A diagnostic feature of all AD is an accumulation of Aβ plaques. Neuronal exposure to Aβ oligomers may have detrimental consequences, including synapse disruption, oxidative stress, altered [Ca²⁺]_i signaling, and cell death (41, 42, 61–64). Here we have confirmed that FAD PS enhances susceptibility to Aβ toxicity and have demonstrated a molecular mechanism through which altered [Ca²⁺]_i homeostasis plays a role in this phenotype. Thus, XeB and InsP₃R knock down each normalized cell death responses to Aβ to control levels, indicating that Ca²⁺ release from the ER plays a critical role. Surprisingly, in view of the general prosurvival functions in the nervous system (but see the discussion above), our results suggest that constitutive CREB activation may be involved in mediating Aβ toxicity, as demonstrated by the significantly decreased vulnerability to Aβ resulting from pharmacologic or molecular inhibition of each component of the InsP₃R–CaMKIV–CREB pathway.

In summary, we found that CREB is constitutively activated in neuronal cell lines and in AD mouse brains expressing FAD mutant PS. FAD PS promotes CREB-dependent gene transcription by activating a CaMK pathway driven by exaggerated InsP₃R-mediated Ca²⁺ release. This pathway enhances cell death and susceptibility to Aβ-induced toxicity. Our results support previous reports that FAD PS expression enhances intracellular Ca²⁺ release, and specifically support the hypothesis that this effect is mediated by InsP₃R (10, 11).

Materials and Methods

See SI Materials and Methods for detailed descriptions of cell culture, transfections, Western blot analyses, Ca²⁺ imaging, immunofluorescence, brain protein extraction, and cell viability assays.

Materials.

Antibodies included ser133-phosphorylated CREB, c-fos, nNos, and CaMKIV (Cell Signaling Technology); CREB and P-CaMKIV (Santa Cruz Biotechnology); PS1, PS2, and BDNF (Chemicon); β-tubulin (Affinity Bioreagents). KN93, KN92, U73122, and U73433 were from Calbiochem. siRNAs were designed by Dharmacon. Aβ peptide 1–40 and scrambled Aβ were obtained from R peptide. Propidium iodide and Hoescht dyes were obtained from Invitrogen.

Statistical Analyses.

Data are summarized as mean ± SEM. Statistical significance was assessed using the unpaired t test or ANOVA for repeated measures at the 95% level (P < 0.05).