Disturbed Activation of Endoplasmic Reticulum Stress Transducers by Familial Alzheimer's Disease-linked Presenilin-1 Mutations*

Recent studies have shown independently that presenilin-1 (PS1) null mutants and familial Alzheimer's disease (FAD)-linked mutants should both down-regulate signaling of the unfolded protein response (UPR). However, it is difficult to accept that both mutants possess the same effects on the UPR. Furthermore, contrary to these observations, neither loss of PS1 and PS2 function nor expression of FAD-linked PS1 mutants were reported to have a discernable impact on the UPR. Therefore, re-examination and detailed analyses are needed to clarify the relationship between PS1 function and UPR signaling. Here, we report that PS1/PS2 null and dominant negative PS1 mutants, which are mutated at aspartate residue 257 or 385, did not affect signaling of the UPR. In contrast, FAD-linked PS1 mutants were confirmed to disturb UPR signaling by inhibiting activation of both Ire1α and ATF6, both of which are endoplasmic reticulum (ER) stress transducers in the UPR. Furthermore, PS1 mutants also disturbed activation of PERK (PKR-like ER kinase), which plays a crucial role in inhibiting translation during ER stress. Taken together, these observations suggested that PS1 mutations could affect signaling pathways controlled by each of the respective ER-stress transducers, possibly through a gain-of-function.

Alzheimer's disease is a progressive neurodegenerative disorder characterized pathologically by deposition of amyloid ␤-protein (A␤), 1 formation of neurofibrillary tangles, and neuronal death in brain regions (1). Some cases show familial Alzheimer's disease (FAD), indicating that genetic factors are involved in the pathogenesis. A subset of early onset cases of FAD are caused by mutations in the amyloid precursor protein (APP) gene located on chromosome 21 (2)(3)(4), presenilin-1 (PS1) found on chromosome 14 (5), and presenilin-2 (PS2) located on chromosome 1 (6 -8). Mutations in the gene encoding PS1 are responsible for many cases of FAD. Although the mechanisms by which mutations in PS1 predispose individuals to FAD have not yet been determined, FAD-linked PS1 mutations have been shown to be associated with altered proteolytic processing of APP, such as increases in the production of the highly amyloidogenic A␤ peptide, A␤42 (9,10). A␤ production accompanied by the accumulation of carboxyl-terminal fragments of APP, which are generated by cleavage of ␤and ␥-secretases for APP, was inhibited in PS1-deficient cells (11). This finding indicated that PS1 is important in the ␥-secretase-mediated cleavage process involved in generating A␤.
Mutations in PS1 increase cellular susceptibility to apoptosis induced by various insults, including the withdrawal of trophic factors and exposure to A␤ (12,13). Previously, we reported that the mechanism by which PS1 mutations promote cell death could contribute to increasing vulnerability to endoplasmic reticulum (ER) stress by altering the signaling pathway of the unfolded protein response (UPR) (14). The accumulation of misfolded polypeptides in the lumen of the ER results in activation of the UPR, inducing transcriptional up-regulation of genes encoding molecular chaperones and folding catalysts present in the ER (15). One of the mediators of the UPR is a transducer of ER stress, the ER-resident transmembrane kinase Ire1 (16,17). Ire1 senses the perturbed environment in the ER and leads to downstream signaling by a process that is thought to depend on oligomerization and autophosphorylation of its kinase domain. The down-regulation of UPR signaling by PS1 mutations is caused by disruption of the function of Ire1, i.e. PS1 mutations decrease the levels of phosphorylated Ire1 (14). Recently, Niwa et al. (18) reported that PS1-deficient cells show a reduced ability to mount the UPR, with a reduction in nuclear accumulation of carboxyl-terminal fragments of Ire1. They speculated that impairment of the UPR is based on a defect in UPR-induced proteolytic processing of Ire1 at its intramembranous region by defects of ␥-cleaving activities associated with PS1. Contrary to these observations, Sato et al. (19) have reported that neither PS1 null nor FAD-linked mutations affect the UPR signaling. Therefore, a re-examination and detailed analyses of relationships between PS1 functions and the UPR should be performed.
ATF6 is an ER transmembrane protein, recently isolated as * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
a possible mammalian UPR-specific transcription factor (20 -22). ATF6 is cleaved at or close to the cytosolic face of the membrane in response to ER stress. The amino-terminal cytoplasmic domain, which contains the DNA-binding, dimerization, and transactivation domains, is translocated into the nucleus and activates transcription of ER molecular chaperone genes containing the ER stress response element (ERSE) (20,23), which is thought to be a regulatory element of promoter regions conserved in ER molecular chaperone genes in mammalian cells. However, it has not been determined whether the luminal domain of ATF6 can itself sense the accumulation of unfolded proteins in the ER or whether cleavage of ATF6 requires ER stress-dependent activation of Ire1 proteins. If PS1 null or FAD-linked mutations affect the induction of ER molecular chaperones as described previously, these mutations could disturb the function of ATF6 as well as Ire1.
In the present study, to make it clear whether PS1 is associated with UPR signaling, we re-examined the effects of PS1 on induction of BiP/GRP78 mRNA and the ER stress response using PS1-deficient cells and cell lines expressing dominant negative PS1 mutants, with mutations of aspartate residues located in transmembrane domains 6 and 7 of PS1 (24), and expressing FAD-linked PS1 mutants. Furthermore, to clarify the mechanisms by which mutant PS1 down-regulates signaling of the UPR, we examined whether PS1 mutants affected the signaling pathways mediated by other ER stress transducers such as ATF6 or PKR-like ER kinase (PERK), which is an ER stress transducer and may play a role in inhibiting translation during ER stress (25,26).

EXPERIMENTAL PROCEDURES
Cell Culture and Induction of ER Stress-Human neuroblastoma SK-N-SH cells and mouse embryonic fibroblasts were cultured in ␣-minimal essential medium (Life Technologies, Inc.) supplemented with 10% fetal calf serum and Dulbecco's modified Eagle's medium (DMEM, Life Technologies, Inc.) supplemented with 20% fetal calf serum, respectively. Two days before the stimulation by various ER stressors, 3 ϫ 10 5 cells were plated in 10-cm-diameter dishes (for Western and Northern blotting analyses) or 1 ϫ 10 5 cells in 6-cmdiameter dishes (for cell death assay), with the media changed for fresh ones the next day. In all experiments of ER stress response, we used only culture dishes that were grown at 70 -80% confluence to avoid various stresses induced by overgrowth. On the day of stimulation, cells were placed in fresh media for more than 1 h before treatment with ER stressors to create the same conditions in each dish. We used tunicamycin, thapsigargin, and DTT (all from Sigma) as ER stress inducers at indicated doses and times. The control dishes were treated by just changing the media.
Before collecting, cells were washed in ice-cold phosphate-buffered saline and then lysed in aliquots of 200 l of Triton buffer (20 mM HEPES, pH 7.5, 150 mM NaCl, 1% Triton X-100, 10% glycerol, 1 mM EDTA, 10 mM tetrasodium pyrophosphate, 100 mM NaF, 17.5 mM ␤-glycerophosphate, 1 mM phenylmethylsulfonyl fluoride, 4 mg/ml aprotinin, and 2 mg/ml pepstatin A) for detection of Ire1␣ and PERK. For detection of ATF6, the cells were lysed in equal volumes of hot-SDS buffer (0.9% SDS, 15 mM EDTA, 8 mM methionine, and 1,000 units of Trasylol were incubated in a boiling water bath for 10 min, cooled, diluted to 0.3% SDS, and adjusted to contain 33 mM Tris acetate, pH 8.5, and 1.7% Triton X-100). All samples were centrifuged at 4°C for 10 min after standing on ice for 1 h, and then the supernatants were collected in 1.5-ml Eppendorf tubes.
Antibodies-Rabbit antisera against a synthetic peptide corresponding to residues 1-14 of human PS1 have been described previously (14). Anti-PERK antibody was raised against a synthetic peptide corresponding to residues of mouse PERK (residues 1094 -1114) and was affinitypurified by ProtOn Kit1 (Multiple Peptide System, San Diego). Anti-ATF6 antibodies against recombinant ATF6 (amino acids 6 -307) fused to Escherichia coli maltose-binding protein have also been described previously (21). Anti-Ire1␣ antibodies were provided by Prof. D. Ron (New York University School of Medicine). Anti-phosphorylated eIF2␣ polyclonal and anti-␤-actin monoclonal antibodies were purchased from Research Genetics Co. and Chemicon International Inc., respectively.
Western Blotting-For Western blotting of Ire1␣ and PERK, cells were lysed in 1% Triton lysis buffer. Samples were loaded onto a 7% SDS-polyacrylamide gels. For Western blotting of ATF6, cells were lysed in Laemmli SDS sample buffer and then loaded onto 10% SDSpolyacrylamide gels. The protein concentration of each sample was quantified by Lowry assay (DC protein assay, Bio-Rad). Protein-equivalent samples were subjected to Western blotting. As an internal control, the levels of ␤-actin were examined on the same blot at the same time as the other proteins of interest.
Northern Blotting-Total RNA was extracted from cells treated with ER stress inducers. Aliquots of 10 g of each total RNA were fractionated by electrophoresis through 1.0% agarose-formaldehyde gels and transferred onto Immobilon-N membranes (Millipore, Bedford, MA). The membranes were hybridized with 32 P-labeled human BiP cDNA probe. After washing in 2ϫ SSC, 0.1% SDS and 0.1ϫ SSC, 0.1% SDS, the membranes were dried and autoradiographed. The human BiP cDNA was a gift from Dr. S. Kaneda (Heat Shock Protein Research Institute).
Fluorescence Microscopy-Fibroblasts from PS1 mutant knock-in mice or SK-N-SH cells stably transfected with each PS1 construct were incubated with or without 3 g/ml Tm for the specified times. Cells were fixed in 4% paraformaldehyde and permeabilized in 0.3% Triton X-100. Anti-ATF6 antibody was used at a dilution of 1:1000. Stained cells were viewed with a confocal microscope (LSM 510, Carl Zeiss). To estimate the relative nuclear fluorescence intensity, cells were co-stained with 4,6-diamidino-2-phenylindole to demarcate the nuclear boundaries. Then, cytoplasmic (C), nuclear (N), or both (N/C) types of staining of ATF6 were measured independently in 10 fields. Each value was represented by the percentage of positive cells in each fraction, and the total of C, N, and N/C was 100%.
Presenilin Null Mice-Mice lacking functional expression of PS1 were generated using homologous recombination to introduce a neomycin cassette flanked by loxP sequences ϳ200 base pairs downstream of exon 5 of the murine PS1 gene. When the loxP flanked neomycin cassette is not removed by cre recombinase, the presence of cryptic splice sites within the neomycin cassette results in the generation of extremely low quantities of unstable transcripts lacking an open reading frame after exon 5. 2 These transcripts result in no detectable PS1 protein expression in embryonic stem cells or in the brains of animals homozygous for this construct. As expected, homozygous PS1-deficient mice (PS1 Ϫ/Ϫ ) displayed the previously reported skeletal abnormalities (27,28). Embryonic fibroblasts from PS1/PS2 double homozygous-deficient mice (29) and the littermates were kindly provided by Dr. B. De Strooper.
Culture of Fibroblasts from PS1-deficient and PS1 Mutant Knock-in Mice-Primary fibroblasts were cultured from PS1 ϩ/Ϫ and PS1 Ϫ/Ϫ embryos on embryonic day 14.5. Fibroblast cultures were also prepared from fetal mice at embryonic day 14.5, generated by mating heterozygous PS1 mutation (I213T) knock-in mice as described (30). These cells were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum and were treated with 2-3 g/ml Tm for the specified times. For evaluation of cell death, lactate dehydrogenase levels in culture medium were quantified as described previously (14).

UPR Signaling in PS1 Mutant Cells-PS1
Ϫ/Ϫ cells have been reported to show impaired induction of BiP/GRP78, an ER molecular chaperone (18), suggesting that native PS1 is involved in UPR signaling. We re-examined the induction of BiP mRNA in PS1 Ϫ/Ϫ , PS1/PS2 Ϫ/Ϫ fibroblasts, and SK-N-SH cell lines expressing PS1 dominant negative mutants (D257A and D385A). The cells were stimulated with 3 g/ml tunicamycin (Tm), which induces ER stress by preventing protein glycosylation, for specified times. Northern blotting showed that there was no difference in the induction of BiP mRNA between PS1 Ϫ/Ϫ and PS1 ϩ/Ϫ fibroblasts, PS1/PS2 Ϫ/Ϫ and PS1/PS2 ϩ/ϩ fibroblasts, or SK-N-SH cells expressing wild-type PS1 and dominant negative mutants (Fig. 1). No effects of these mutations on the induction of BiP mRNA were confirmed at concentrations from 0.5 to 10 g/ml of Tm. In contrast, fibroblasts from FAD-linked PS1 mutant knock-in mice, which express the PS1 mutant (I213T) at normal physiological levels (30), and SK-N-SH cells expressing PS1 mutant (delta exon 9, PS1⌬9) revealed that the induction of BiP mRNA was delayed and attenuated compared with wild-type cells at each dose from 0.5 to 3 g/ml Tm (Fig. 2). Furthermore, we examined the changes in induction of CHOP, which is also induced during ER stress (31,32). We detected no differences in the induction of CHOP mRNA in PS1 Ϫ/Ϫ fibroblasts (Fig. 1) or aspartate mutant- Northern blotting of BiP/GRP78 and CHOP mRNAs. Primary cultured fibroblasts from PS1 ϩ/Ϫ and Ϫ/Ϫ (a) and PS1/2 ϩ/ϩ and Ϫ/Ϫ (b) mice were exposed to Tm at a concentration of 3 g/ml for the indicated times. Total RNA was isolated from each culture and subjected to Northern blotting with probes for BiP, CHOP, or ␤-actin mRNA. Lower panel, left, the amounts of BiP mRNA in each cell are shown. Quantitative analyses of BiP mRNA levels as shown in autoradiographs were performed as described previously (14). Values are arbitrary intensities and represent the mean Ϯ S.D. of data from four independent experiments. Right, the expression of PS1 in fibroblasts from PS1-deficient mice. Full-length and amino-terminal fragments of PS1 were absent in PS1-deficient cells. c, SK-N-SH cells were stably transfected with dominant negative PS1 mutants mutated at aspartate residues 257 or 385. Wild, cells stably transfected with wild-type PS1. The lower panel shows quantitative analyses of BiP mRNA levels as shown in autoradiographs and Western blotting of PS1 in these cells. Note that mutant cells showed accumulation of full-length PS1 (arrow). expressing neuroblastomas (data not shown). Cells expressing FAD-linked PS1 mutants exhibited a delay in CHOP mRNA induction similar to that in BiP mRNA induction (Fig. 2). The effects of FAD-linked PS1 mutations on the induction of BiP and CHOP mRNA were observed at each dose of Tm from 0.5 to 5 g/ml, but they could not be detected at 10 g/ml because of the toxic effects of Tm, suggesting that higher doses of ER stress inducers could mask the down-regulation of UPR induced by PS1 mutation. Down-regulation of BiP mRNA induction by FAD-linked PS1 mutation was also observed in treatments with thapsigargin (TG) and dithiothreitol (DTT) (data not shown). These results indicated that although native PS1 is not involved in UPR signaling, FAD-linked PS1 mutants do have an effect that may be attributable to a gain of aberrant functions.
Vulnenerability to ER Stress in Cells Expressing PS1 Mutants-We previously demonstrated that cells expressing PS1 mutants exhibited increased vulnerability to various ER stresses, the mechanism of which was the down-regulation of UPR signaling (14). If native PS1 is not relevant to UPR signaling, sensitivity to ER stress in PS1-deficient cells should be equivalent to that in wild-type cells. Therefore, we carried out cell death assays using various mutant cells treated with Tm. Cell death induced by Tm was significantly promoted in PS1 mutation knock-in fibroblasts and SK-N-SH cells expressing PS1⌬9 (Fig. 3c), findings consistent with those observed previously. In contrast, PS1 Ϫ/Ϫ fibroblasts and SK-N-SH cells expressing PS1 dominant negative mutants showed no differences in cell death induced by ER stress as compared with the respective controls (Fig. 3, a and b). Thus, we confirmed that a deficiency of PS1 does not affect susceptibility to ER stress and that native PS1 is not associated with UPR signaling.
Mutant PS1 Alters Proteolytic Cleavage and Translocation of ATF6 -ATF6 is cleaved at or close to the cytosolic face of the membrane, and its amino-terminal portion is translocated into the nucleus during ER stress. The amino-terminal domain activates transcription of ER molecular chaperones such as BiP and GRP94 (20,21,22). We examined whether FAD-linked PS1 mutants also impaired activation of the other UPR transducer protein, ATF6, similarly to Ire1 as has been reported previously (14). Fibroblasts from PS1 mutant knock-in mice were stimulated by the addition of 2 g/ml Tm. Cells were lysed, and Western blotting was performed using anti-ATF6 antibodies raised against the amino-terminal region of ATF6 (amino acids 6 -307). ATF6 was constitutively expressed in wild-type fibroblasts as a 90-kDa protein (p90ATF6) (Fig. 4b). The p90ATF6 levels decreased, and instead a new band of 50-kDa (p50ATF6) appeared 2 h after treatment with Tm, with p50ATF6 gradually accumulating after this time. The homozygous fibroblasts from the PS1 mutant knock-in mice showed that the production of p50ATF6 was delayed and that the levels of the fragments were decreased by ϳ50% compared with those from wild-type mice from 2 to 6 h after treatment with Tm (Fig. 4b). We also confirmed that the conversion of p90ATF6 to p50ATF6 was inhibited in SK-N-SH neuroblastoma cells stably transfected with FAD-linked PS1 mutants, such as PS1⌬9 or PS1 with an alanine-to-glutamate mutation (A246E) (Fig. 4c). These findings indicated that PS1 mutations disturb the conversion of ATF6 under conditions of ER stress. In contrast, these effects on ATF6 were not observed in either PS1-deficient cells (Fig.  4a) or in PS1 dominant negative-expressing cells (Fig. 4c). Thus, native PS1 is not considered to play a role in the processing of ATF6 at or close to the cytosolic face of the ER membrane, findings that are likely to be different in the case of processing at the ␥-site of APP.
ATF6 has been reported to be constitutively expressed at the perinuclear region under non-ER stress conditions. ER stress causes the conversion of p90ATF6 to p50ATF6, which is translocated into the nucleus. We performed an immunocytochemical analysis of the subcellular localization of ATF6 in PS1 mutant cells under normal or ER stress conditions. For evaluation of ATF6 translocation, the fluorescence signals of cells stained with anti-ATF6 antibodies were divided into cytoplasmic (C), nuclear (N), or both regions (N/C). N/C shows that ATF6 was shifting from the ER to the nucleus (Fig. 5a). Under non-inducing conditions, the distribution of ATF6 (p90ATF6) in PS1 mutant knock-in cells resembled that seen in fibroblasts expressing wild-type PS1 (Fig. 5b). p90ATF6 was localized FIG. 4. Conversion of ATF6 to p50ATF6 after ER stress. Primary cultured fibroblasts from PS1 knock-out (a) and PS1 mutation knock-in (b) mice were treated with 2 g/ml Tm for the indicated times, and cells were lysed as described under "Experimental Procedures." Then, aliquots of lysates were subjected to Western blotting of ATF6. Wild-type cells showed accumulation of p50ATF6 band (arrowhead) after treatment with Tm. a, the amounts of p50ATF6 in PS1 Ϫ/Ϫ cells were almost equivalent to those in PS1 ϩ/Ϫ cells. predominantly in the ER, and a small fraction of cells (ϳ10%) showed nuclear staining. Nuclear immunoreactivity of ATF6, which has been demonstrated previously as p50ATF6 fragments (21), increased in wild-type cells from 2 h after the induction of ER stress by treatment with Tm and reached a maximum within 4 h (Fig. 5, b and c). The changes in ATF6 immunoreactivity were consistent with the results of Western blotting. PS1 mutant knock-in cells showed slight nuclear accumulation of p50ATF6 2 h after induction (Fig. 5, b and c). After 4 h, it gradually accumulated in the nucleus, and after 8 h, the immunoreactivity was almost equivalent to that in wild-type fibroblasts. Similar findings were obtained in SK-N-SH cells expressing PS1⌬9 (Fig. 5c). Thus, mutation in PS1 obviously delayed nuclear accumulation of p50ATF6, which could be ascribed to the delayed processing of p90ATF6 at the ER membrane (see Fig. 4b). In PS1-deficient cells, subcellular localization of ATF6 was not altered compared with that of control cells (Fig. 5c).
Inhibited Activation of Endogenous ER Stress Transducers by PS1 Mutants-The above findings demonstrated that PS1 mutants affect the activation of endogenous ATF6 after ER stress. To assess whether functions of the other ER stress transducers, Ire1␣ and PERK, are disturbed by PS1 mutants, we examined the activation of these endogenous molecules during ER stress in cells expressing PS1 mutation. Ire1 and PERK activation during ER stress were correlated with autophosphorylation of their cytoplasmic kinase domains. Initially, to examine the effects of PS1 mutants on the activation of Ire1␣, Western blotting was performed using lysates of knock-in fibroblasts stimulated with ER stress inducers. Phosphorylation of Ire1␣ retards their mobility on SDS-polyacryl-amide gels and serves as a convenient marker of their activation status (33). As treatment with Tm inhibits glycosylation of Ire1␣, the retardation of their mobility cannot be detected during ER stress. Therefore, we employed 0.5 mM DTT and 1 M TG as ER stress inducers in this experiment. In wild-type fibroblasts, Ire1␣ was completely phosphorylated within 15 min after treatment with DTT (Fig. 6a). In PS1 mutant knock-in cells, almost no phosphorylation of Ire1␣ was observed at 15-30 min after treatment. The same findings were observed when cells were treated with 1 M TG (data not shown). Furthermore, the activation of Ire1␣ was also inhibited in SK-N-SH cells stably transfected with PS1 mutants when cells were treated with 1 M TG (PS1⌬9, Fig. 6b). These results were consistent with the previous findings of 293T cells transiently co-transfected with Ire1 and PS1 mutants (14), showing that FAD-linked PS1 mutants inhibited autophosphorylation of Ire1␣. The effects of PS1 mutants on inhibited activation of Ire1␣ were detected at concentrations from 0.5 to 3 M TG, but no changes were detected at 5 M TG compared with those in cells expressing wild-type PS1 (Fig. 6b). This suggests that higher doses of ER stress inducers could mask the effects of PS1 mutation on the inhibited activation of Ire1, similar to the findings that down-regulation of induction of BiP mRNA was not detected in FAD-linked PS1 mutant-expressing cells at higher doses of ER stress inducers.
Second, we examined the effects of PS1 mutations on activation of PERK. PERK is known as an ER stress signal transducer, which is autophosphorylated and phosphorylates eukaryotic initiation factor-2␣ eIF2␣ during ER stress (25,26). PERK was phosphorylated within 15 min after treatment with 0.5 mM DTT in wild-type fibroblasts (Fig. 6a)  not shown). In PS1 mutant knock-in cells, about 50 or 30% of the PERK molecules remained nonphosphorylated at 15 or 30 min, respectively, after ER stress (Fig. 6a). In SK-N-SH cells stably transfected with PS1 mutants (PS1⌬9), phosphorylation of PERK was severely inhibited within 30 min after treatment with 1 M TG compared with that in cells expressing wild-type FIG. 6. Inhibited activation of endogenous ER stress transducers. a, fibroblasts from PS1 mutant knock-in mice were treated with 0.5 mM DTT for the indicated times. Upper panel, cells were lysed in 1% Triton buffer, and then Western blotting of Ire1␣ was carried out. The bands of Ire1␣ were shifted to a higher apparent molecular weight after treatment with DTT in wild-type (Wt) fibroblasts. The higher molecular weight bands of Ire1␣ were demonstrated to be derived from phosphorylated forms, as described previously (33). P-Ire1, phosphorylated-Ire1␣. Phosphorylation of Ire1␣ was completely inhibited in homozygous mutant fibroblasts (Ho). Lower panel, Western blotting of PERK after treatment of fibroblasts with DTT. Phosphorylation of PERK was also inhibited in homozygous mutant fibroblasts. P-PERK, phosphorylated-PERK. b, SK-N-SH cells stably transfected with each PS1 construct were stimulated with TG (1 and 5 M) for the indicated times, and Western blotting of Ire1␣ was performed. When cells were treated with 1 M TG (upper panel), the bands of Ire1␣ were completely shifted (phosphorylated-Ire1␣) within 15 min in cells expressing wildtype PS1 (PS1W). In contrast, no shift of Ire1␣-immunoreactive bands was seen in PS1⌬E9 cells treated with the same dose of TG. When cells were treated with 5 M TG (lower panel), both cells showed that phosphorylation of Ire1␣ occurred within 15 min, and there were no differences of the phosphorylation kinetics. c, phosphorylation of PERK and eIF2␣ in SK-N-SH stable cell lines expressing PS1 constructs. The changes in levels of phosphorylated PERK (upper and middle panels) and eIF2␣ (lower panel) after treatment of cells with 1 M (upper panel) and 5 M (middle) TG are shown. PS1 mutation inhibited or delayed the phosphorylation of both PERK and eIF2␣ after ER stress when cells were treated with 1 M TG treatment, but the differences in phosphorylation kinetics of PERK between wild-type and mutant cells disappeared when cells were treated with 5 M TG. ␤-Actin was used as an internal control. PS1 (Fig. 6c) similar to the findings in PS1 mutation knock-in cells. At 60 min after ER stress, PERK was completely phosphorylated in both wild-type and mutant PS1-expressing cells. These results indicated that activation of PERK was delayed by PS1 mutation at the dose of 1 M TG. When cells were treated with 5 M TG, PERK was completely phosphorylated in these cells within 15 min, indicating that excessive doses of ER stress inducers could mask the effects of PS1 mutation on the delay of activation of PERK.
As described above, the disturbed function of PERK is known to cause the down-regulation of phosphorylation of eIF2␣ (25,26). Therefore, we examined the levels of phosphorylated eIF2␣ after ER stress in SK-N-SH cells expressing PS1 mutant. As expected, PS1 mutation inhibited phosphorylation of eIF2␣ (Fig. 6c). Thus, activation of PERK and the resultant phosphorylation of eIF2␣ were disturbed by expression of a PS1 mutation. Taken together with the results of inhibited activation of Ire1␣, it is possible to conclude that PS1 mutants disturb the signaling pathways from ER stress transducers such as Ire1␣, ATF6, and PERK simultaneously.

DISCUSSION
Our recent study showed that a missense mutation in PS1 leads to attenuation of the induction of BiP under ER stress conditions (14). In the present study, to determine whether the function of wild-type PS1 is essential for or is associated with the UPR signaling system, we analyzed the expression of BiP and CHOP, which are known to be induced by ER stress, using PS1-deficient cells. However, we found no attenuation of BiP or CHOP induction in fibroblasts from PS1-deficient mice in contrast to PS1 mutant knock-in cells. Furthermore, the UPR was not down-regulated in cells expressing dominant negative PS1, which was mutated in aspartate at residues 257 or 385. These findings suggest that wild-type PS1 is not associated with UPR signaling. Previously, PS1 Ϫ/Ϫ cells were reported to show reduced induction of BiP after ER stress (18). In this paper, it was speculated that the mechanisms of attenuated induction of BiP were due to the disturbance of UPR-dependent proteolytic processing of Ire1␣ by a defect in the ␥-secretase activity of PS1. We tried to detect the proteolytic processing of Ire1␣ and the nuclear accumulation of Ire1␣ fragments after ER stress in both normal and PS1-deficient cells. However, we did not observe the cleaved fragments of Ire1␣ in both cells. The cause may be attributed to the instability of Ire1␣ fragments, i.e. such fragments might be short-lived and therefore difficult to detect. The other possible reason for the nondetection of cleaved fragments of Ire1␣ is the differences in experimental conditions, such as methods of cell culture and stimuli by ER stress inducers.
Presenilin proteins seem to be important in the cleavage of APP at the ␥-site because inhibition of A␤ production is accompanied by the accumulation of carboxyl-terminal fragments of ␤-APP in PS1 Ϫ/Ϫ cells (11). ATF6 is synthesized as an ER transmembrane protein and is cleaved at or close to the cytosolic face of the membrane. The biochemical characteristics of ATF6 suggest that PS1 may be associated with the cleavage of ATF6 on the ER membrane, similar to the cleavage of APP at the ␥-site. However, the conversion of p90ATF6 to p50ATF6 was intact in PS1 Ϫ/Ϫ cells, suggesting that PS1 is not associated with the proteolytic process of ATF6, although PS1 has been shown recently to be required for processing of some of the type 1 transmembrane proteins such as APP (11), Notch (34), and Ire1 (18), and that the cleavage is different from the processing of APP at the ␥-site. This conclusion was supported by the lack of changes in the processing of ATF6 in cell lines expressing dominant negative PS1 that was mutated in aspartate at residues 257 or 385. In contrast to native PS1, FADlinked PS1 mutants inhibited the cleavage of p90ATF6. Al-though the detailed mechanisms are unclear, mutation of PS1 may affect the processing enzymes of ATF6 on the ER membrane. Recently, ATF6 was demonstrated to be processed by Site-1 protease (S1P) and Site-2 protease (S2P), which are processing enzymes known as sterol regulatory element-binding proteins (SREBPs) (35). FAD-linked PS1 mutations have not been reported to inhibit the processing of SREBPs. Therefore, it is inconceivable to consider that FAD-linked PS1 mutants inhibit the activities of S1P and S2P. Ire1␣ is known to be sufficient to activate the ATF6 reporter gene, whereas a dominant negative form of Ire1␣ blocks ER stress activation (36), suggesting that ATF6 is downstream of Ire1␣ in the ER stress signaling pathway. Our study also showed that FAD-linked PS1 mutants inhibited activation of Ire1␣. Taken together, it is likely that disturbances of processing and nuclear transports of ATF6 could be caused by the dysfunction of Ire1␣.
Neither PS1 null nor dominant negative mutants affected the signaling of the UPR. In contrast, FAD-linked PS1 mutants did affect it, suggesting that its abilities could be caused by gaining aberrant function. However, it is unclear whether the effects of mutant PS1 on the UPR could be associated with its ␥-secretase activities. To address this issue, it a study will need to be made of the ability of PS1 mutants to attenuate the UPR after introducing second mutations into the aspartic acid residues in the catalytic portion of the protein.
The present study has shown that down-regulation of BiP induction by FAD-linked PS1 mutant is due to attenuated signaling of the UPR through decreased levels of phosphorylated Ire1 and inhibition of activation of ATF6 under ER stress conditions. Moreover, PS1 mutants also inhibited the phosphorylation of PERK, which is another ER stress transducer. Therefore, it is possible that mutant PS1 perturbs the functions of each ER stress transducer and inhibits its downstream signal. However, it remains unclear why mutant PS1 inhibits the activation of ER stress transducers. A recent study suggested that activation of the signaling-mediated ER stress transducers could be triggered by dissociation of BiP from stress transducers (33,36,37). This dissociation would lead to oligomerization of stress transducers inducing autophosphorylation and the resultant activation of downstream signaling. Previously, we reported that PS1 was associated with Ire1 on the ER membrane (14). If FAD-linked PS1 mutants form malfolded structures, BiP may constitutively bind to PS1 molecules to promote its folding. The complex formation of PS1, Ire1 (and/or PERK), and BiP may inhibit the dissociation of BiP from Ire1 (and/or PERK) under ER stress conditions. However, to clarify the detailed mechanisms responsible for perturbation of ER stress signaling by PS1 mutants, further studies focusing on the structures of PS1 molecules bearing missense mutations and the regulation of ER stress sensing are needed.
Sato et al. (19) recently reported that FAD-linked PS1 mutations do not have a discernible impact on the ER stress response, a finding that is not consistent with the present study. Their inability to reproduce the previous results, in which we showed that FAD-linked PS1 mutation causes the down-regulation of UPR signaling, is likely to be due to their use of cells (HEK293 and Neuro2a cells) that were, in comparison, not sensitive to ER stress and to their use of time points that were too late. They examined activation of PERK at 5 h after ER stress. As described in our previous study (38) and present studies, FAD-linked PS1 mutants delayed the activation of stress transducers during ER stress, but the effects of PS1 mutations could be masked by treatment with excessive doses of ER stress inducers or stimulation for too long a time. In addition, it is most important to pre-incubate for more than 1 h with fresh medium before treatment with agents to gain constant data in the ER stress response (as has been described above under "Experimental Procedures"), because the induction levels of BiP mRNA are altered considerably if the changes are not conducted. The decrease of BiP mRNA induction in cells expressing FAD PS1 mutations was ϳ30 or 50% compared with the controls in the PS1 mutation knock-in fibroblasts or stably transfected SK-N-SH cells, respectively. To detect these subtle defects in the UPR in PS1 mutation-expressing cells, the cells should be handled carefully under the same experimental conditions.
In summary, wild-type PS1 does not play a role in signaling of the UPR, but FAD-linked PS1 mutants do affect it, suggesting that PS1 mutations cause perturbation of the ER stress response through a gain-of-function. The possible mechanism of a perturbed ER stress response could be the inhibition of activation of ER stress transducers such as Ire1␣, ATF6, and PERK. The disturbance of these protective systems to ER stress in PS1 mutation-bearing cells may lead to susceptibility to various cellular stresses.