The γ Secretase-generated Carboxyl-terminal Domain of the Amyloid Precursor Protein Induces Apoptosis via Tip60 in H4 Cells*

The amyloid precursor protein (APP), a large glycoprotein highly expressed in neurons, is cleaved in its intramembranous domain by γ secretase to generate amyloid-β and a free carboxyl-terminal intracellular fragment (APP-CT), which has previously been suggested to interact with the adapter protein Fe65 and the histone acetyltransferase Tip60. An identical γ secretase activity mediates cleavage of Notch, releasing an intracellular signaling domain that translocates to the nucleus. We examined the effect of an ectopically expressed 58-amino acid APP-CT fragment (APP-C58) on human H4 neuroglioma cells. We demonstrate by confocal microscopy and fluorescence resonance energy transfer analysis that APP-C58 translocates to the nucleus and forms a complex in the nucleus with the Tip60, independent of interactions with Fe65. APP-C58 transfected H4 cells undergo apoptosis within 48–72 h, marked by nuclear blebbing, terminal deoxynucleotidyl transferase-mediated dUTP nick-end labeling (TUNEL) staining, and blockade by a caspase inhibitor. When nuclear access of APP-C58 is prevented by fusing with a strong membrane-targeting farnesylation domain, apoptosis is blocked. APP-C58-induced apoptosis was markedly enhanced by co-transfection with wild type Tip60 and decreased by mutant Tip60 lacking histone acetyltransferase activity, suggesting that Tip60 mediates APP-CT-induced cell death. Thus, γ secretase cleavage of APP may contribute to Alzheimer's disease-related neurodegeneration in two ways: release of amyloid-β and liberation of a bioactive carboxyl-terminal domain from membrane-bound APP.

by ␣ secretase, occurs in the extracellular domain of APP within the amyloid-␤ sequence, precluding its formation (2,3) and leaving an 83-amino acid transmembrane stub. Alternatively, the extracellular domain of APP can be cleaved at the amino terminus of amyloid-␤ by BACE, leaving a 99-amino acid transmembrane stub (4 -7). The carboxyl-terminal stubs undergo intramembrane proteolysis by a presenilin-dependent ␥ secretase activity, releasing p3 and amyloid-␤ respectively, as well as the small cytoplasmic domain of APP (APP-CT). In an analogous fashion, presenilin-dependent ␥ secretase activity also releases the Notch intracellular domain (NICD) into the cytoplasm (8 -14), which interacts with CSL proteins and translocates to the nucleus to regulate gene expression (15) in a cell type-specific fashion.
Several reports suggest that, like NICD, APP-CT can translocate to the nucleus (16,17) and play a role in cellular signaling. Cao and Sudhof (18), using a heterologous signal transduction assay, suggested that APP-CT interacts with Fe65 and histone acetyltransferase Tip60 to mediate transactivation of a reporter gene. Gao and Pimplikar (19) overexpressed ectopic APP-CT59 (C59) and APP-CT57 (C57) (reflecting the predominant cleavage sites of ␥ secretase in APP) and showed that it interacts with and leads to down-regulation of PAT1 in the nucleus and that it represses retinoic acid-responsive gene expression in MDCK cells. Leissring et al. (20) recently reported the role of APP-CT in regulating phosphoinositol-mediated calcium signaling using an ectopically expressed APP-CT fragment. Here we report that ectopic expression of APP-C58 in H4 human neuroglioma cells leads to dramatic nuclear localization and apoptosis. APP-CT-induced apoptosis was dependent on nuclear access; APP-CT modified with a plasma membrane targeting signal did not cause apoptosis. APP-CTinduced apoptosis was dependent on the interaction of APP-CT with Tip60 but not with Fe65. Apoptosis was enhanced by co-transfection with wt Tip60, whereas co-transfection with mutant Tip60 that does not have acetyl transferase activity was protective. Taken together, we demonstrate that APP-CT has potent biological activity and may be responsible for contributing to signal transduction pathways that predispose to apoptotic cell death in Alzheimer's disease.

Generation of Expression Constructs for Human APP and Fe65-
Truncated APP constructs encoding the APP carboxyl terminus (APP-C58) were made: one with Myc at the carboxyl terminus (C58-Myc) and the other with EGFP at the carboxyl terminus (C58-GFP). For C58-Myc, an APP770 cDNA template was used to run a PCR on the carboxyl-terminal 58-amino acid coding region without a stop codon. A set of primers that add a Kozak sequence to the 5Ј end was used: 5Ј-GCCTC-GAGGCCACCATGGCGACAGTGATCGTCATCACCTTG-3Ј and 5Ј-GC-GAATTCACCGGTGTTCTGCATCTGCTCAAAGAA-3Ј. The fragment amplified by PCR was digested and ligated into the XhoI and EcoRI site of the pcDNA3.1(Ϫ)B. For C58-GFP, analogous methods were used to ligate the carboxyl-terminal 58-amino acid coding region without a stop codon into the EcoRI and NotI site of pEGFP-N2 (CLONTECH, Pal Alto, CA). The APP-C58-stop construct (without any tags) was made the same way as APP-C58-Myc except that a 3Ј primer containing a stop codon was used: 5Ј-GCGAATTCC-TAGTTCTGCATCTGCTCAAAGAA-3Ј.
The construct of APP-C58 with farnesylation signal (APP-C58-F) was generated by PCR with a set of primers: 5Ј-CAAGCTAGCCGCCAC-CATGGCGACAGTGATCGTCATCACCTTG-3Ј and 5Ј-CTACCGGT-GCGTTCTGCATCTGCTCAAAGAACTT-3Ј, the sense primer containing the Kozak consensus, and a start codon and then subcloned into the NheI and AgeI site of pEGFP-F (CLONTECH) to generate a carboxylterminal GFP-tagged APP-C58 that contains farnesylation signal at the carboxyl terminus. This construct thus contains the 20-amino acid farnesylation signal from c-Ha-Ras fused to the carboxyl terminus of EGFP. This farnesylation signal directs C58-GFP-F to the inner face of the plasma membrane.
To inhibit the interaction of APP-CT with Fe65, mutation of tyrosine residues at codons 682 and 687 of the YXXNPXY motif (of the 695 numbering) to alanine (Y682A/Y687A) was introduced to APP-C58-GFP and APP-C58-Myc constructs using the QuikChange site-directed mutagenesis kit (Stratagene, La Jolla, CA). The generation of carboxylterminal Myc-tagged Fe65 (Fe65-Myc) has been reported elsewhere (29). The authenticity of all PCR-generated constructs was confirmed by DNA sequencing.
For generation of wild type Tip60 (wt Tip60) construct, Tip60 was cut out from the pOZ-Tip60 plasmid (21) with restriction enzymes XhoI and NotI. The expression vector pEGFP-N1 (CLONTECH) was digested with XhoI and NotI, cutting out the GFP protein. The wt Tip60 and histone acetylase-deficient mutant Tip60 (mt Tip60) which has Q377E and G380E, have been reported elsewhere (21); we digested and ligated them into a mammalian expression vector backbone derived from the EGFP-N1 plasmid (CLONTECH), from which the EGFP coding sequence had been deleted.
Antibodies and Reagents-Rabbit polyclonal antibody C8 was raised against the carboxyl-terminal 20 amino acid residues of APP770 (kindly provided by D. Selkoe, Brigham and Women's Hospital, Boston, MA) (17). Mouse monoclonal anti-Myc antibody was purchased from Invitrogen (Carlsbad, CA). The antibody against amino acids 494 -513 of Tip60 was purchased from Upstate Biotechnology, Inc. (Lake Placid, NY). This antibody recognizes both wt and mt Tip60.
Z-VAD(OMe)-FMK, a pan-caspase inhibitor was purchased from Enzyme Systems Products (Livermore, CA). To inhibit caspase activity, 100 M Z-VAD was added to the culture medium 2-3 h after transfection.
Cell Culture Conditions and Transient Transfection-H4 cells derived from human neuroglioma cells (American Type Culture Collection, Manassas, VA) were cultured in OPTI-MEMI with 10% fetal bovine serum. Transient transfection of H4 cells was performed using a liposome-mediated method (FuGene 6; Roche Molecular Biochemicals). The cells were plated onto 4-well chambers 1 day before the transfection. First a mixture of 1 g of plasmid DNA and 3 l of FuGENE 6 was made in 100 l of Dulbecco's modified Eagle's medium and left for 15-30 min at room temperature, and then 25 l of this mixture was added to the medium in each well. The incubation time was 24 -72 h. The same protocol was used for double transfection.
Immunohistochemistry-Immunostaining was performed 24 -48 h post-transfection. The cells were fixed in 4% paraformaldehyde for 10 min, washed in Tris-buffered saline (pH 7.3), permeabilized with 0.5% Triton X-100 for 20 min, and blocked with 1.5% normal goat serum for 1 h. The transfected cells were then incubated with appropriate primary antibodies for 1 h at room temperature: mouse anti-Myc monoclonal antibody (Invitrogen 1:1000) to label C58-Myc or Fe65-Myc, rabbit polyclonal C8 antibody (1:500) to label carboxyl terminus domain of APP, or rabbit anti-Tip60 antibody to label endogenous or overexpressed Tip60. The cells were then washed three times in Tris-buffered saline and labeled with Cy3-conjugated anti-mouse or anti-rabbit IgG antibody (Jackson Immunoresearch Laboratories, West Grove, PA; 10 g/ml) for 1 h at room temperature. For the FRET experiments, the immunostained cells were stored in Tris-buffered saline at 4°C. For the morphological analysis of the nuclei, the immunostained cells were mounted with DAPI-containing mounting medium (Vector Laboratories, Burlingame, CA) and covered with a coverslip.
Assessment of Nuclear Change and Apoptosis-Cells with abnormal nuclear morphology (lobulation, multinucleation, condensation, and fragmentation) were identified under ultraviolet visualization by DAPI staining (Vector Laboratories). Cells with abnormal nuclei in the C58-GFP-transfected, C58-GFP-F-transfected cells, or pEGFP-transfected cells 48 h after transfection were counted under the microscope in five random fields from each well of cultured cells (about 300 -700 cells) in triplicate for each condition. The results are indicated as percentages of dysmorphic nuclei (lobulation, multinuclei, condensation, and fragmentation) divided by the total number of transfected cells.
To measure the cell death ratio, the number of transfected cells was also counted in five random fields from each well of cultured cells of C58-transfected and pEGFP-transfected cells, starting 24 h after transfection until 72 h. The results are normalized as percentage ratios compared with the number of cells at 24 h after transfection. This is shown as a survival ratio (see Fig. 2B).
TUNEL staining was performed according to the manufacturer's protocol in C58-GFP-, C58-GFP-F-, and pEGFP-transfected cells in the presence or absence of 100 M Z-VAD(Ome)-FMK (in situ cell death detection kit, TMR red; Roche Molecular Biochemicals). TUNEL reaction was labeled with TMR red and analyzed by confocal microscopy with appropriate filters. The number of TUNEL positive cells in five random fields was counted in three sets of experiments, and the results were normalized as percentage ratios compared with the total number of the transfected cells.
LDH was measured as an indicator of cell death by using a cytotoxicity detection kit (Roche Molecular Biochemicals). Conditioned medium was collected 48 h post-transfection from C58-GFP-or pEGFPtransfected cells in the presence or absence of 100 M Z-VAD-, C58-GFP-F-, or pEGFP-F-transfected cells and from C58-GFP-transfected cells in the presence of either wt Tip60 or mt Tip60. LDH was measured in a 96-well plate by a Wallack plate reader according to the manufacturer's protocol in triplicate for each condition. The result was normalized to the value of the mock transfected cells and shown as a percentage of increase.
Fluorescence Resonance Energy Transfer-FRET measurements were observed using a Bio-Rad 1024 confocal microscope mounted on a Nikon Eclipse TE300 inverted microscope. The krypton-argon laser (emission, 488 and 568 nm) was used to excite the fluorescein or EGFP and Cy3, respectively.
FRET was measured using a method developed for laser scanning confocal microscopy (22,23). The energy transfer was detected as an increase in donor fluorescence (EGFP) after complete photobleaching of the acceptor molecules (Cy3). The amount of energy transfer was calculated as the percent increase in donor fluorescence after acceptor photobleaching; the initial scan was obtained at low laser energy using the 488 line of the krypton-argon laser to record the fluorescein (or EGFP) signal. A second scan was performed with the 568 line, and the area of co-localization was noted. A small part of the cells (ϳ5 ϫ 5 m) was then photobleached with intense 568 nm light (laser power, 100%) to destroy the acceptor molecules. The cells were rescanned using 488-nm light. An increase of the EGFP signal within the photobleached area was used as a measure of the amount of FRET present. Exposing singly EGFP labeled cells to 568-nm light for equivalent times did not alter the amount of fluorescein emission. The ratio of donor fluorescence after photobleaching to donor fluorescence before photobleaching was compared with the null hypothesis value of 1.0 by one-group t tests.
For the FRET between C58 and endogenous Tip60, H4 cells were transfected with C58-Myc, fixed and immunostained by mouse anti-Myc (1:1000, Invitrogen) and rabbit anti-Tip60 antibody (1:200), and visualized by Cy3 and fluorescein isothiocyanate, respectively. In all settings, Cy3 signal was photobleached to see changes of GFP or fluorescein isothiocyanate signal. When the acceptor fluorophore was omitted, no enhancement of donor signal after photobleaching was observed. As a negative control, we examined possible FRET between the NICD (24) and Tip60, two proteins that are both known to be in the nucleus but that are not known to interact directly. A complete description of these techniques is provided by Kinoshita et al. (23).

RESULTS
When full-length APP is overexpressed, it has been reported that APP-CT can be detected only if Fe65 is co-expressed to stabilize the APP-CT fragment (17), and we have confirmed this observation in H4 cells as well. By contrast, Gao and Pimplikar (19) ectopically expressed APP-CT-GFP constructs in MDCK cells and observed strong nuclear localization. We expressed APP-C58 as a fusion protein with EGFP (analogous to the construct of Gao and Pimplikar), tagged with V5 or Myc on the carboxyl terminus, or without a tag (C58-stop; detected by C8), in H4 cells. In all instances APP-C58 was observed strongly in the nucleus (Fig. 1, A and B). Co-transfection of APP-C58 with Fe65 showed strong co-localization of APP-C58 and Fe65 in the nucleus, as expected (Fig. 1, D and E). We then constructed a mutant form of APP-C58 (Y682A/Y687A)-GFP or C58 (Y682A/Y687A)-Myc that does not interact with Fe65. The mutant APP-C58 (Y682A/Y687A) is also found predominantly in the nucleus (Fig. 1C). These observations suggest that interaction of APP-CT with Fe65 is not critical for nuclear translocation per se. We then tested the co-transfection of Fe65 and both wild type and YXNPXY mutant of APP-C58 with farnesylation signals, which make the APP-CT anchor to the membrane. When we co-transfected those C58-F (or C58-F-NPXY) and Fe65, we observed a differential localization. Fe65 remained mostly in the cytoplasm, co-localized with APP when it is co-transfected with wild type C58-F, whereas Fe65 is found in the nucleus, not co-localized with APP, when transfected with C58-F-NPXY (data not shown). This result indicates that APP interacts with Fe65 via NPXY region, the interaction being unnecessary for APP to be translocated to the nucleus.
Ectopic expression of APP-CT in MDCK cells has been reported to affect gene transcription (19). However, in H4 cells, we observed a striking change: H4 cells transfected with APP-CT demonstrated nuclear lobulation and fragmentation by DAPI staining and underwent cell death within 1-2 days (Fig. 2). The percentage of cells with dysmorphic nuclei 24 -72 h after transfection assessed by DAPI staining was markedly increased in APP-C58-transfected cells (greater than 60% of transfected cells in each of four experiments) compared with mock transfected cells, and cell death as assayed by LDH was 42% greater in the C58 transfected cells (p Ͻ 0.0001) (Fig. 3A). Quantification of survival rate after transfection in the APP-C58-transfected and control (pEGFP-transfected) groups was assessed by counting the number of the viable transfected cells in five random fields. Transfection of APP-C58 causes significant cell death compared with the control transfected H4 cells. The percentage of viable cells 48 and 72 h after APP-C58 transfection are 53 Ϯ 3.2 and 34 Ϯ 5.5% (means Ϯ S.E.) compared with that of 24 h after transfection, whereas those of control transfection were 92 Ϯ 2.5 and 81 Ϯ 4.5%, showing a significant difference in the number of the viable cells (p Ͻ 0.0001) (Fig. 2B).
In a separate experiment, C58 and C58 co-transfected with Fe65 were equally toxic, judged by the abnormal morphology of the nucleus (for C58 alone, 62.3%; for C58 plus Fe65, 59.4%; both p Ͻ 0.0001, in each of three experiments, compared with EGFP; C58 alone and C58 plus Fe65 were not different from one another).
The TUNEL assay showed robust staining in APP-C58-transfected cells but rarely in mock transfected cells, suggesting the initiation of an apoptotic pathway (Fig. 3, B and C). To test the idea that APP-C58 induced apoptosis, we treated the transfected cells with the caspase-3 inhibitor Z-VAD. Treated cells accumulated APP-C58 in the nucleus, but cell death, as assessed by TUNEL staining and DAPI staining as well as LDH release, was essentially eliminated (Fig. 3, A, D, and E).
The LDH value of C58-transfected cells against control trans- whereas that in the presence of Z-VAD was 2.1 Ϯ 0.5% (p Ͻ 0.0001). Next we examined the possibility that APP-C58 induced apoptosis indirectly, by causing the release of a toxic factor. Conditioned medium from APP-C58 transfected cells was applied to native H4 cells without consequence.
We then tested whether nuclear localization of APP-C58 was necessary for apoptosis. To block the nuclear translocation of APP-C58, we used the C58-F-GFP construct, encoding a tagged C58 peptide with a farnesylaton plasma membrane localization signal. The addition of farnesylation signal made the APP-CT fragment primarily (although not entirely) anchored to the plasma membrane (compare the C58-GFP-F signal in Fig. 4A with that of C58-GFP signal in Fig. 4C). The expression level of APP-C58-GFP-F was comparable with that of APP-C58-GFP. The number of TUNEL-positive cells 24 h after transfection is markedly decreased in C58-GFP-F-transfected cells (9.4 Ϯ 1.4% of C58-GFP-F versus 25.7 Ϯ 3.4% of C58-GFP; p Ͻ 0.01; Fig. 4), suggesting that preventing the APP-C58 fragment from translocating to the nucleus greatly diminishes APP-CT induced apoptosis. Of note, we continued to observe some lobulation of the nucleus in APP-C58-F-GFP-transfected cells, (61.0% of the transfected cells) to an extent comparable with that of C58-GFP (62.3%) in three sets of experiments; whether this is due to residual nuclear access of C58-GFP-F or to a process that does not involve nuclear translocation remains to be clarified.
As noted above, the C58 mutant with Y682A/Y687A that does not interact with Fe65 still translocates to the nucleus. The C58 (Y682A/Y687A)-GFP-transfected H4 cells showed TUNEL-positive cells and high LDH values that are comparable with that of the wild type C58-GFP transfected cells (49.9 Ϯ 5.6% increase compared with the pEGFP-transfected cells; n ϭ 3; p Ͻ 0.01), suggesting that an interaction with Fe65 is not necessary to induce apoptosis.
As reported by Cao and Sudhof (18), APP-CT is associated with Fe65 and with the histone acetyltransferase Tip60 in the nucleus. Because Tip60 is known to be involved in apoptotic pathways (21), we tested the hypothesis that APP-C58 interaction with Tip60-induced apoptosis in H4 cells. First, we examined whether APP-C58 and Tip60 form a stable complex by performing FRET experiments. FRET is a sensitive biophysical technique that can detect protein-protein interactions within ϳ10 nm. FRET experiments were performed between APP-C58-Myc and endogenous Tip60 in the nucleus. The FRET ratio increase between APP-C58 and Tip60 was 26.7 Ϯ 2.7% (n ϭ 10; p Ͻ 0.0001) (Fig. 5). Like APP-C58, APP-C58 (Y682A/Y687A) co-localized with Tip60 immunostaining in the nucleus. Interestingly, the FRET ratio increase between C58 (Y682A/Y687A)-GFP and endogenous Tip60 in the nucleus was 29.0 Ϯ 4.4% (p Ͻ 0.001; n ϭ 13), not different from that of the wild type APP-CT and endogenous Tip60. As a negative control for FRET between noninteracting proteins, we examined the FRET ratio between Tip60 and NICD. NICD is localized to the nucleus but is not known to interact with Tip60. The FRET ratio increase was 4.1 Ϯ 2.2% (n ϭ 10); there is a significant statistical difference between APP-Tip60 interaction and NICD-Tip60 interaction (p Ͻ 0.0001), suggesting that APP and Tip60 tightly interact with each other. We interpret these data to suggest that APP-CT can translocate to the nucleus, interact with endogenous Tip-60, and cause apoptosis independent of its interaction with Fe65.
Because APP-C58 interacts strongly with Tip60 in the nu- cleus, we tested the possibility that APP-C58-triggered cell death is mediated by Tip60. The cells were co-transfected with APP-C58-GFP and either wild type Tip60 (wt Tip60) or mutant Tip60 (mt-Tip60), which does not have histone acetyltransferase activity (21). The cells were fixed 24 -48 h after transfection with APP-C58-GFP and immunostained for Tip60 to identify the subset of cells that expressed both constructs.
Analysis of the transfected cells showed that wt Tip60 cotransfection enhanced the apoptosis triggered by APP-C58, whereas almost no apoptotic cells were found in the mt Tip60expressing cells (Fig. 6). This observation was confirmed by LDH assay. The LDH value in the wt Tip60 and APP-C58-GFP transfected cells was 50.4 Ϯ 4.5% higher than that of APP-C58-GFP-singly transfected cells (p Ͻ 0.01; n ϭ 3), whereas mutant Tip60 blocked LDH release to 20 Ϯ 2.0% (p Ͻ 0.05; n ϭ 3) (Fig.  6). This result indicates that Tip60 interactions with APP-C58 may be critical for APP-CT-mediated apoptotic events. DISCUSSION Cao and Sudhof (18) reported that a fusion protein of APP containing Gal4-or LexA-DNA-binding domains within the carboxyl-terminal region stimulated transcription using a heterologous signal transduction assay. Reporter gene induction depended on the presence of Fe65 and the histone acetyltransferase Tip60. Gao and Pimplikar (19) ectopically expressed the ␥ secretase cleaved carboxyl-terminal fragment of APP (APP-CT) and showed that it is directed to the nucleus and, in MDCK cells, can alter gene expression. Leissring et al. (20) recently reported the role of APP-CT in regulating phosphoinositolmediated calcium signaling using ectopically expressed APP-CT fragment. Our current data confirm and extend these observations: We show that, in H4 cells, ectopically expressed APP-CT localizes to the nucleus and forms a complex with Tip60, independently of its association with Fe65. However, surprisingly, APP-CT dramatically induces apoptosis in H4 cells. Apoptosis is dependent on nuclear access of the peptide, because plasma membrane anchoring prevents apoptosis. APP-CT-induced apoptosis is mediated via Tip60; co-transfection with wild type Tip60 enhances apoptosis, whereas co-transfection with a mutant Tip60 that does not have histone acetyltransferase activity reduces apoptosis. Taken together, these data suggest that APP-CT translocation to the nucleus is associated with Tip60 activation and gene transcription that ultimately leads H4 cells to an apoptotic path.
Numerous studies have established that Fe65, a cytosolic adapter protein, interacts with the carboxyl terminus of fulllength APP (see Ref. 23 for review). Our current data are in accord with these observations, and suggest that Fe65 also interacts with APP-CT. In contrast to the observations that APP-CT interacts with Fe65 and that APP-CT may require Fe65 for stabilization (17,18), our data do not support an FIG. 5. APP-C58 accumulates in the nucleus and forms a heterodimer with Tip60 confirmed by FRET. H4 cells were transfected with APP-C58-Myc (labeled by Cy3), and endogenous Tip60 was detected by anti-Tip60 antibody and then labeled with fluorescein isothiocyanate. Tip60 is strikingly localized to the nucleus (C) and co-localizes with APP-C58 (A). Photobleaching of the Cy3 label of APP-C58 in the nucleus (B) leads to enhanced Tip60 fluorescent signal within the photobleached area, demonstrating FRET (D). The FRET ratio increase between APP-C58 and Tip60 was 26.7 Ϯ 2.7% (n ϭ 10; p Ͻ 0.0001). obligatory role for Fe65 in nuclear translocation of APP-CT or for interaction of Fe65 with Tip60, because APP-CT (Y682A/ Y687A), which does not interact with Fe65, nonetheless is localized in the nucleus, shows strong FRET with Tip60, and also induces apoptosis. Cao and Sudhof (18) showed that mutant APP that does not bind Fe65 was ϳ10-fold less efficient in inducing the reporter gene read-out, and they suggested that a heterotrimeric complex is necessary for transcriptional activation of the Gal4 or LexA reporters. This apparent discrepancy may simply reflect different assays but could also be interpreted to reflect the important role Fe65 appears to play in stabilizing the APP-CT fragment generated from full-length APP, insofar as APP-CT that was not stabilized by Fe65 may well be less available for transcriptional activation of the reporter gene.
We also demonstrate that APP-C58 induces apoptosis in H4 cells. Apoptosis is dependent on nuclear translocation of the APP-C58, because directing APP-CT toward a membrane localization (and away from the nucleus) with a farnesylation signal diminishes apoptosis. We demonstrate a molecular mechanism for APP-CT-induced apoptosis via its interaction with Tip60, because co-transfection with wild type Tip60 enhances, whereas co-transfection with mutant Tip60 that does not have histone acetyltransferase activity reduces apoptosis. Taken together, we suggest that APP-C58 interacts with Tip60 in the nucleus to either cause or predispose cells toward apoptotic cell death. Tip60 activity appears to be involved in apoptotic pathways after other apoptotic stimuli as well (21), suggesting a role for Tip60 (and hence APP-CT) in regulating cell survival.
By contrast with these results, expression of APP-CT in MDCK cells (19) or COS cells (17) leads to different cell responses. A broad literature in which various carboxyl terminal fragments of APP (e.g. CT105, CT100, or CT31) lead to apoptosis in some but not all cell types (25) or studies (26,27) suggests that APP-CT might be pro-apoptotic under some cellular conditions; the observation that in human neuroglioma cells APP-CT can induce apoptosis raises the possibility that a similar phenomenon also occur in neural cells in vivo.
While these studies were being performed, several reports appeared that the stable product of ␥ secretase action on APP might be a shorter, ϳ49-amino acid fragment (28). Other studies suggest that a smaller, caspase cleaved carboxyl-terminal fragment of ϳ31 amino acids is pro-apoptotic (26,27). Although on Western blot we do not observe any smaller fragments after transfection with APP-C58-Myc (data not shown), we cannot rule out the possibility that small amounts of a smaller and highly toxic product derived from C58 is generated in H4 cells, and the exact domains within C58 that lead to interactions with Tip60 and to apoptosis remain to be studied. Nonetheless, our results support the hypothesis that presenilin-related ␥ secretase cleavage of APP could lead to neurodegeneration both by generation of amyloid-␤ leading to deposition in senile plaques and by production of bioactive carboxyl-terminal domain fragments.