Effects of RNA interference-mediated silencing of gamma-secretase complex components on cell sensitivity to caspase-3 activation.

Familial Alzheimer's disease mutations in the presenilin 1 gene (PSEN1) have been previously shown to potentiate caspase activation and apoptosis in transfected cells and transgenic mice. However, the mechanism underlying this effect is not known. We set out to determine whether cellular sensitivity to caspase activation could be affected by modulating presenilin 1 (PS1) processing. PS1 processing was altered using RNA interference (RNAi) aimed at silencing the expression of the genes encoding the four components of the gamma-secretase complex, PSEN1, APH-1, PEN-2, and nicastrin. RNAi for these genes was carried out in naive H4 human neuroglioma cells, as well as H4 cell lines overexpressing either wild-type PSEN1 or the Familial Alzheimer's disease mutant PSEN1-Delta9 (PS1-mutant), that were induced to undergo apoptosis. In wild-type PSEN1 cells, RNAi for PEN-2, as expected, increased levels of full-length PS1 (PS1-FL) and decreased PS1 endoproteolysis. This was accompanied by potentiated caspase-3 activation in response to an apoptotic stimulus. In contrast, nicastrin RNAi, which only decreased levels of PS1-amino-terminal fragment and did not affect PS1-FL levels, had no effect on caspase-3 activation during apoptosis. Surprisingly, in the PS1-mutant cells, RNAi for PEN-2 (and APH-1) did not increase but instead reduced the levels of PS1-FL deleted for exon 9. In turn, this was accompanied by attenuated caspase-3 activation in response to an apoptotic stimulus. Finally, in naive H4 cells, PSEN1 RNAi also attenuated caspase-3 activation in response to an apoptotic stimulus. Collectively, these findings indicate that cellular sensitivity to caspase activation correlates with overall PS1 protein levels, particularly with levels of FL-PS1.

Alzheimer's disease, an insidious and progressive neurodegenerative disorder accounting for the vast majority of dementia and one of the greatest public health problems in the United States, is characterized by global cognitive decline and robust accumulation of amyloid deposits and neurofibrillary tangles in the brain (for review see Ref. 1). The genes encoding the amyloid ␤ protein precursor and presenilins 1 and 2 (PSEN1 and PSEN2) have been shown to harbor autosomal dominant gene defects with near 100% penetrance in up to 50% of early onset (Ͻ60 years of age) cases of familial Alzheimer's disease (FAD) 1 (for review see Ref. 2).
We set out to determine whether cellular sensitivity to caspase activation could be affected by modulating PS1 processing. PS1 processing was manipulated using RNAi to silence the expression of the genes encoding the four known components of the ␥-secretase complex, PSEN1, APH-1, PEN-2, and NCSTN. For these studies, we employed naive H4 human neuroglioma cells as well as H4 cell lines overexpressing either WT PSEN1 (PS1-WT) or the FAD mutant, PSEN1-⌬9 (PS1mutant). To induce apoptosis, we used 1 M staurosporine (STS). Changes in the levels of full-length PS1 (PS1-FL) versus the active PS1-endoproteolytic fragments (for this purpose, the NH 2 -terminal fragment, PS1-NTF) were then monitored relative to levels of caspase activation.

EXPERIMENTAL PROCEDURES
Cell Lines-We employed naive H4 human neuroglioma (H4) cells and H4 cells stably transfected to express low levels of either the PSEN1-⌬9 (PS1-mutant cells) or WT PS1 (PSI-WT cells). These cells have previously been shown to express sufficiently low levels of PS1 so as to avoid spontaneous apoptosis that can result from robust overexpression of PS1 (7). All of the cell lines were cultured in Dulbecco's modified Eagle's medium (high glucose) containing 9% heat-inactivated fetal calf serum, 100 units/ml penicillin, 100 g/ml streptomycin, and 2 mM L-glutamine. Stably transfected H4 cells were additionally supplemented with 200 g/ml G418.
Western Blot Analysis of PS1-proteolytic Processing and Caspase-3 Cleavage-Western blot analysis was performed as described by Kovacs et al. (7). 40 g of total protein of each sample was subjected to SDSpolyacrylamide gel electrophoresis using 4 -20% gradient Tris/glycine gels under reducing conditions (Invitrogen). Proteins next were transferred to a polyvinylidene difluoride membrane (Bio-Rad) using a semidry electrotransfer system (Amersham Biosciences). Nonspecific proteins were blocked using 5% nonfat dry milk in Tris-buffered saline with Tween 20 for 1.5 h. Blots were then incubated with a primary antibody followed by a secondary antibody (horseradish peroxidaseconjugated anti-rabbit antibody 1:10,000, Pierce). Blots were washed with 1ϫ Tris-buffered saline with Tween 20 for 30 min between steps. Antibody Ab14 (1:2500) (a generous gift of Dr. Sam Gandy, Thomas Jefferson University) was used to recognize PS1-FL (35-40 kDa in H4 mutant PSEN1-⌬9 cells and 40 -45 kDa in H4 PS1-WT cells) and PS1-NTF (25-30 kDa). Antibody PNT-2 (1:500) (a generous gift of Dr. Gopal Thinakaran, University of Chicago) was used to detect PEN-2. Antibody anti-nicastrin (1:1000) (Calbiochem) was used to detect nicastrin. Antibody anti-mAPH-1a (1:1000) (Oncogene Research Products, San Diego, CA) was used to detect APH-1. A caspase-3 antibody (1:1000 dilution, Cell Signaling, Beverly, MA) was used to recognize caspase-3 fragment (17-20 kDa) resulting from cleavage at aspartate position 175. To visualize full-length caspase-3 in the experiments assessing the effects of PSEN1 siRNA treatment on caspase-3 cleavage, we used a caspase-3 antibody (1:1000 dilution) from Cell Signaling. The intensity of signals was analyzed using an image program (NIH Image 1.62). We quantified the Western blots as follows. We used the levels of ␤-tubulin to normalize the levels of cleaved caspase-3 (i.e. by determining the ratio of cleaved caspase-3 amount to ␤-tubulin amount), PS-FL, and PS1-NTF to control for loading differences in total protein amounts. We present the changes in the levels of cleaved caspase-3, PS-FL, and PS1-NTF in the cells treated with PSEN1, PEN-2, NCSTN, or APH-1 siRNA as the percentage of those in the cells treated with control siRNA.
Statistics-ANOVA with repeated measurements was employed to compare the difference from the control group. The p values Ͻ0.05 were considered statistically significant.

PSEN1 RNAi Decreases Caspase-3 Activation in Naive H4
Cells-We first assessed the effects of RNAi-mediated silencing of PSEN1 in naive human neuroglioma H4 cells treated with 1 M STS to induce apoptosis. The effects on caspase activation then were assessed by monitoring caspase-3 cleavage (activation). 144 h after transfection with PSEN1 siRNA or control siRNA, cells were pretreated with either Me 2 SO or 100 M Z-VAD, a broad-spectrum caspase inhibitor, for 1 h followed by saline or 1 M STS for another 6 h. The cells then were harvested and subjected to Western blot analyses in which antibodies Ab14, caspase-3, and anti-␤-tubulin were used to detect PS1 (PS1-FL and PS1-NTF), cleaved (activated) caspase-3, and the control protein, ␤-tubulin, respectively.
To test for effects of PS1-FL levels on caspase-3 activation following induction of apoptosis, we next carried out PEN-2 RNAi in stably transfected PS1-WT H4 cells. We first documented that PEN-2 siRNA treatment decreased the protein levels of PEN-2 in PS1-WT H4 cells ( Fig. 2A). Luo et al. (42) have shown previously that PEN-2 RNAi increases PS1-FL in cells expressing PS1-WT. We asked whether increased levels of PS1-FL (induced by PEN-2 RNAi) would affect caspase-3 activation in PS1-WT H4 cells induced to undergo apoptosis with STS. Experiments were carried out as described above. Neither control nor PEN-2 siRNA alone induced caspase-3 cleavage (Fig. 2B). STS induced caspase-3 cleavage in both control siRNA-treated and PEN-2 siRNA-treated cells (Fig. 2B). However, caspase-3 cleavage was visibly increased in the cells treated with PEN-2 siRNA (Fig. 2B) as compared with the cells treated with control siRNA. As expected, in comparison to control siRNA, PEN-2 siRNA increased PS1-FL levels with no detectable difference in PS1-NTF or ␤-tubulin levels (Fig. 2B). The quantitation of the levels of cleaved caspase-3, PS1-FL, and PS1-NTF (normalized to ␤-tubulin) revealed PEN-2 siRNA to increase caspase-3 cleavage up to roughly 250% as compared with control siRNA (Fig. 2C, p Ͻ 0.05). This was accompanied by roughly up to a 336% increase in PS1-FL as compared with control siRNA (Fig. 2D, p Ͻ 0.05). Meanwhile, no detectable changes in PS1-NTF levels were observed between the PEN-2 siRNA-treated and control siRNA-treated cells (Fig. 2E). These results indicated that following induction of apoptosis in PS1-WT H4 cells, caspase-3 activation can be dramatically potentiated by PEN-2 RNAi. Because the knock-down of PEN-2 has been shown previously (28) to reduce ␥-secretase activity, these results suggest that increased caspase-3 activation following PEN-2 siRNA could be attributed to either increased levels of PS1-FL or decreased ␥-secretase activity.
Reduction of PS1-NTF Levels Does Not Attenuate Caspase-3 Activation in PS1-WT H4 Cells-To test whether a reduction in PS1 endoproteolysis (and ␥-secretase activity) affects cellular sensitivity to caspase activation, we next used NCSTN RNAi in PS1-WT H4 cells and then monitored the levels of PS1-NTF and caspase-3 cleavage following the induction of apoptosis with STS. Experiments were carried out as described above. We first documented that NCSTN siRNA treatment decreased nicastrin protein levels in PS1-WT H4 cells (Fig. 3A). As expected, neither control siRNA nor NCSTN siRNA alone induced caspase-3 cleavage. ␤-Tubulin levels also were not significantly affected by either NCSTN or control siRNA (Fig. 3B). STS induced caspase-3 cleavage in both control siRNA-treated and NCSTN siRNA-treated cells. However, no visible difference in caspase-3 cleavage was observed between NCSTN siRNA-treated and control siRNA-treated cells (Fig. 3B). As expected, NCSTN siRNA treatment led to a decrease in PS1-NTF levels (as compared with control siRNA). However, in these cells, PS1-FL levels did not change (Fig. 3B). Quantitation of cleaved caspase-3, PS1-FL, and PS1-NTF (normalized to ␤-tubulin) revealed no change in caspase-3 cleavage (Fig. 3C). However, as expected, NCSTN siRNA decreased PS1-NTF as compared with control siRNA (Fig. 3D, p Ͻ 0.05). NCSTN RNAi did not alter PS1-FL in these cells (Fig. 3E). Collectively, these data show that a decrease in the active PS1-endoproteolytic fragment, PS1-NTF, in the absence of any effect on PS1-FL has no effect on cellular sensitivity to caspase activation in PS1-WT H4 cells.
Attenuated Caspase-3 Cleavage Is Associated with Lower Levels of PS1-FL in H4 Mutant PSEN1-⌬9 Cells-We next treated H4 mutant PSEN1-⌬9 cells with PEN-2 RNAi and assessed the effects on caspase activation following induction of apoptosis with STS. We chose the ⌬9 mutant form of PS1 in the following experiments because it is uniquely not cleaved by presenilinase (40,46,47), making it particularly suitable for the present studies aimed at assessing the effects of FL-PS1 in caspase activation. All of the experiments were carried out as described above. We first documented that PEN-2 siRNA treat- There was no significant difference in the amounts of ␤-tubulin in the PSEN1 siRNA-treated and control siRNA-treated cells. B, caspase-3 cleavage as assessed by quantifying ratio of cleaved caspase-3 to FL caspase-3 in naive H4 cells. PSEN1 siRNA decreased caspase-3 cleavage (*, p Ͻ 0.05) normalized to ␤-tubulin. C, PS1-proteolytic processing as assessed by quantifying PS1-NTF levels in naive H4 cells. PSEN1 siRNA decreased PS1-NTF levels (*, p Ͻ 0.05) normalized to ␤-tubulin. ment reduced the protein level of PEN-2 in H4 mutant PSEN1-⌬9 cells (Fig. 4A). Control siRNA or PEN-2 siRNA alone did not induce caspase-3 cleavage, and no significant differences were observed for ␤-tubulin. Interestingly, whereas PEN-2 RNAi increased PS1-FL in WT-PS1 cells (Fig. 2), the levels of PS1-FL were decreased by PEN-2 siRNA in the H4 mutant PSEN1-⌬9 cells. This reduction in ⌬9-PS1-FL was accompanied by reduced caspase-3 cleavage in the PEN-2 siRNAtreated versus control siRNA-treated H4 mutant PSEN1-⌬9 cells (Fig. 4B). Quantitation of cleaved caspase-3 and PS1-FL (normalized to ␤-tubulin) revealed PEN-2 siRNA to attenuate caspase-3 cleavage to roughly 17% (Fig. 4C) and to decrease PS1-FL to roughly 34% (Fig. 4D) as compared with control siRNA (p Ͻ 0.05). These results indicate that reductions in the levels of the ⌬9-FL-PS1 using PEN-2 RNAi attenuate sensitivity to caspase-3 activation in the H4 mutant PSEN1-⌬9 cells.
To confirm these findings, we next investigated the effects of APH-1 RNAi on the H4 mutant PSEN1-⌬9 cells. Experiments   FIG. 3. Effects of NCSTN RNAi on PS1 processing and caspase-3 cleavage in PS1-WT H4 cells. In the PS1-WT H4 cells, NCSTN siRNA decreased PS1-NTF levels but not caspase-3 cleavage induced by STS. A, NCSTN siRNA treatment decreased the protein levels of nicastrin (NCT) as compared with control siRNA. There was no significant difference in the amounts of ␤-tubulin in the NCSTN siRNA-treated and control siRNA-treated cells. B, control siRNA or NCSTN siRNA alone did not cause caspase-3 cleavage. STS caused caspase-3 cleavage in both control siRNA and NCSTN siRNA groups with similar levels. PS1 immunoblotting showed decreases in PS1-NTF levels in the cells treated with NCSTN siRNA as compared with control siRNA. However, NCSTN siRNA did not decrease PS1-FL levels as compared with control siRNA. There was no significant difference in the amounts of ␤-tubulin in the NCSTN siRNA and control siRNA cells. C, caspase-3 cleavage as assessed by quantifying cleaved caspase-3 in PS1-WT H4 cells. NCSTN siRNA did not decrease caspase-3 cleavage normalized to ␤-tubulin. D, PS1-proteolytic processing as assessed by quantifying PS1-NTF levels in PS1-WT H4 cells. NCSTN siRNA decreased PS1-NTF levels (*, p Ͻ 0.05) normalized to ␤-tubulin. E, PS1-proteolytic processing as assessed by quantifying PS1-FL levels in PS1-WT H4 cells. NCSTN siRNA did not alter PS1-FL levels normalized to ␤-tubulin.
Finally, as an overall control for the RNAi experiments, we compared the effects of the transfection agent, TKO, in the absence of any siRNA on caspase-3 activation and PS1-NTF levels with other RNAi controls including scrambled siRNA (30,42) and luciferase GL2 siRNA. PSEN1 siRNA was used as a positive control in this experiment. As expected, PSEN1 siRNA decreased PS1-NTF but neither TKO nor scrambled siRNA and luciferase GL2 siRNA altered PS1-NTF levels (Fig.  6). None of the above treatments induced caspase-3 activation in the absence of STS (Fig. 6). PSEN1 siRNA, scrambled siRNA, luciferase GL2 siRNA, and TKO alone had no significant effects on ␤-tubulin levels (Fig. 6). DISCUSSION FAD mutations in the PSEN1 gene have been previously shown to potentiate caspase activation and apoptosis in transfected cells and transgenic mice (7, 10 -13). However, the mechanism underlying this effect has not been elucidated. We asked whether cellular sensitivity to caspase activation could be affected by modulating PS1 processing using RNAi aimed at silencing the expression of the genes encoding the four protein components of the ␥-secretase complex, PSEN1, APH-1, PEN-2, and NCSTN. These experiments were carried out in naive H4 human neuroglioma cells as well as H4 cell lines overexpressing either PS1-WT or PS1-mutant.
We first found that lowering overall PS1 protein levels via PSEN1 RNAi in naive H4 human neuroglioma cells induced to undergo apoptosis led to attenuated caspase-3 activation (Fig.  1). Reduced caspase activation in these cells could have been attributed to either lower PS1 protein levels and/or reduced ␥-secretase activity. To further address this issue, we first employed RNAi-mediated silencing of PEN-2 in PS1-WT H4 cells to increase PS1-FL and decrease PS1-endoproteolytic fragments required for ␥-secretase activity. Following the induction of apoptosis, caspase-3 activation was potentiated in these cells (Fig. 2). These findings indicated that increased caspase-3 activation in the PS1-WT cells treated with PEN-2 siRNA could have been due to either increased levels of PS1-FL or reduced ␥-secretase activity. To further discern between these two possibilities, we next employed NCSTN RNAi to specifically reduce PS1 endoproteolysis (as evidenced by decreased PS1-NTF) in the PS1-WT cells. In this case, PS1-NTF levels were reduced while PS-FL levels were unchanged, and sensitivity of these cells to caspase-3 activation was also unchanged (Fig. 3). These results suggested that the extent of caspase activation in response to an apoptotic stimulus can be specifically modulated by protein levels of PS1 and primarily by PS1-FL (as opposed to levels of active PS1-endoproteolytic fragments).
Interestingly, similar studies carried out in H4 cells expressing the FAD mutant PSEN1-⌬9 (Figs. 4 and 5) revealed that RNAi for APH-1 and PEN-2 did not increase but instead decreased ⌬9-PS1-FL. The reduction in ⌬9-PS1-FL was accompanied by reduced caspase activation in response to STS. Given that ⌬9-PS1-FL is functional (analogous to the PS1-WT NTFcarboxyl-terminal fragment complex), the experiments carried out in the PS1-mutant cells did not, by themselves, clarify whether the reduction in caspase-3 activation was the result of reduced levels of ⌬9-PS1-FL or reduced ␥-secretase activity. However, when taken together with the experiments employing PEN-2 and NCSTN RNAi in the PS1-WT cells, the combined data suggest that cell sensitivity to caspase-3 activation is mainly affected by protein levels of full-length PS1.
Our observation that PEN-2 RNAi increases wild-type PS1-FL while reducing levels of PS1 endoproteolysis and ␥-secretase activity is in agreement with previously published studies (28,42). However, we also show for the first time that, in H4 cells stably expressing the ⌬9 FAD PSEN1 mutation, PEN-2 RNAi does not increase but decreases PS1-FL. Collectively, these studies involving the experimental manipulation of PS1-FL levels revealed a significant correlation between the levels of WT PS1-FL or ⌬9 mutant PS1-FL and the amount of There was no significant difference in the amounts of ␤-tubulin in the APH-1 siRNA-treated and control siRNA-treated cells. B, control siRNA or APH-1 siRNA alone did not cause caspase-3 cleavage. STS caused caspase-3 cleavage in both control siRNA and APH-1 siRNA groups but with less caspase-3 cleavage in the APH-1 siRNA group. Z-VAD blocked STS effects on inducing caspase-3 cleavage. PS1 immunoblotting showed reductions in PS1-FL levels in the cells treated with APH-1 siRNA as compared with control siRNA. There was no significant difference in the amounts of ␤-tubulin in the APH-1 siRNA-treated and control siRNA-treated cells. C, caspase-3 cleavage as assessed by quantifying cleaved caspase-3 in H4 mutant PSEN1-⌬9 cells. APH-1 siRNA decreased caspase-3 cleavage (*, p Ͻ 0.05) normalized to ␤-tubulin. D, PS1-proteolytic processing as assessed by quantifying PS1-FL levels in H4 mutant PSEN1-⌬9 cells. APH-1 siRNA decreased PS1-FL levels (*, p Ͻ 0.05) normalized to ␤-tubulin.
One possible molecular mechanism by which PSEN1 mutations may increase cellular sensitivity to caspase activation and apoptosis is that the FAD mutant forms of PS1 act as proapoptotic effectors (gain-of-function) or interfere with antiapoptotic molecules (dominant negative effect of mutant PS1) (7). Our current findings argue that caspase activation can be potentiated simply by increasing the accumulation of PS1-FL without the need for higher levels of active PS1-endroproteolytic fragments required for ␥-secretase activity. Thus, our data would favor the hypothesis that the FAD mutant ⌬9 form of PS1 may be proapoptotic as a function of cellular accumulation of PS1-FL. Increasing evidence suggests a role for caspase activation and apoptosis in Alzheimer's disease (3)(4)(5)(6)(7)(8)(9). A detailed understanding of the molecular mechanisms by which Alzheimer's disease and other neurodegenerative disease-related proteins participate in the apoptotic pathway and affect cellular sensitivity to caspase activation should facilitate the design of future therapeutic strategies for these devastating disorders.
FIG. 6. Effects of control siRNA on PS1-proteolytic processing and caspase-3 activation in naive H4 cells. In naive H4 cells, control siRNA alone did not decrease PS1-NTF levels and did not induce caspase-3 activation. PSEN1 siRNA decreased PS1-NTF levels, whereas control siRNA, scrambled siRNA, and luciferase GL2 siRNA did not decrease the levels. Caspase-3 activation was not seen in the cells treated with PSEN1 siRNA, scrambled siRNA, luciferase GL2 siRNA, or control siRNA. There was no significant difference in the amounts of ␤-tubulin in the PSEN1 siRNA-treated, control siRNAtreated, scrambled siRNA-treated, and luciferase GL2 siRNA-treated cells.