If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Aβ42-lowering Nonsteroidal Anti-inflammatory Drugs Preserve Intramembrane Cleavage of the Amyloid Precursor Protein (APP) and ErbB-4 Receptor and Signaling through the APP Intracellular Domain*
* This work was supported by National Institutes of Health Grant AG 20206 (to T. E. G. and E. H. K.), an Emmy Noether fellowship from the Deutsche Forschungsgemeinschaft (to S. W.), and a fellowship from the John Douglas French Alzheimer's Foundation (to J. E.). 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. ∥ Present address: Johannes Gutenberg University, 55099 Mainz, Germany.
Epidemiological studies indicate that long term use of nonsteroidal anti-inflammatory drugs (NSAIDs) confers protection from Alzheimer's disease, and some NSAIDs were shown to specifically decrease production of the amyloidogenic Aβ42 peptide, most likely by direct modulation of γ-secretase activity. In contrast to γ-secretase inhibitors, Aβ42-lowering NSAIDs do not impair S3 cleavage in the NOTCH receptor and release of the NOTCH intracellular domain, a finding with conceptual implications for the development of safer drugs targeting Aβ production through γ-secretase modulation. Intramembrane cleavage and release of an intracellular signaling domain has recently been demonstrated in a number of additional γ-secretase substrates. We now show in cell-based assays that intramembrane cleavage of APP and ErbB-4 receptor is not impaired by the Aβ42-lowering NSAIDs, sulindac sulfide and ibuprofen. Generation of the APP intracellular domain (AICD) was further not inhibited in a cell-free assay at concentrations far exceeding those effective in reducing Aβ42 production. Closer inspection of AICD signaling showed that stabilization of the AICD peptide by FE65 and AICD-mediated transcription were also retained at Aβ42-lowering concentrations. These results demonstrate that S3-like/intramembrane cleavage is preserved by Aβ42-lowering NSAIDs in at least three substrates of γ-secretase APP, ErbB-4, and NOTCH and underline the striking specificity by which these drugs target Aβ42 production.
Regulated intramembrane proteolysis of transmembrane proteins has been established as a mechanism for the liberation of cytosolic signaling domains that enter the nucleus and regulate gene transcription (
). A growing number of type-1 membrane proteins including NOTCH and its homologs, the NOTCH ligands Delta-1 and Jagged-2, amyloid precursor protein (APP)
and its homologs, ErbB-4, low density lipoprotein receptor-related protein, E-cadherin, and CD44 have been demonstrated to be cleaved within their transmembrane domain (TMD) by γ-secretase activity, attesting to the importance of this pathway to cell signaling (
). In almost all of the γ-secretase substrates examined to date, intramembrane cleavage is preceded and triggered by shedding of their large extracellular domains. Similarities in these proteolytic cascades have been most thoroughly studied in the γ-secretase substrates NOTCH and APP (
). After ligand binding, the mature NOTCH receptor is cleaved by the metalloprotease tumor necrosis factor α converting enzyme. This cleavage event, which has been termed S2 cleavage, releases the extracellular domain from the cell surface and leaves behind a membrane-bound C-terminal Notch fragment termed NEXT (NOTCH extracellular truncation). NEXT undergoes constitutive S3 cleavage within its TMD close to the cytosolic border. S3 cleavage is carried out by γ-secretase activity, releasing the NOTCH intracellular domain (NICD) to translocate into the nucleus and regulate transcription of genes involved in cell fate decisions (
). Release of the soluble APP ectodomain involves either α- or β-secretase cleavage and generates C-terminal APP fragments that are direct substrates for constitutive cleavage by γ-secretase. Proteolytic processing of APP by β-secretase (beta-site APP cleaving enzyme) followed by γ-secretase cleavage produces the 40–42-amino acid Aβ peptide, which is hypothesized to initiate the cascade of events resulting in Alzheimer's disease (
In contrast to S3 cleavage in Notch receptors, γ-secretase cleavage generating Aβ peptides occurs in the middle of the TMD, and these differences in cleavage topology have raised doubts as to whether NOTCH and APP are cleaved by the same proteolytic activity (
). However, using epitope-tagged substrates, NEXT was recently shown to also undergo cleavage in the middle of the TMD resulting in the release of Aβ-like NOTCH derived peptides (
). Furthermore, a novel γ-secretase cleavage site within the TMD of APP was similarly identified that is located close to the cytosolic border and resembles the S3 cleavage site in NOTCH (
). This cleavage event, which was termed ϵ cleavage, releases the 50-amino acid APP intracellular domain (identified as AID, AICD, or CTFγ; hereon referred to as “AICD”) into the cytoplasm, and accumulating data suggest a role for this fragment in the regulation of gene transcription (
). Therefore, γ-secretase activity apparently catalyzes very similar cleavages in the TMDs of APP and NOTCH as well as CD44, with major cleavage sites both in the middle as well as close to the cytosolic border of the TMD (see Fig. 1). Whether these findings further extend to the other γ-secretase substrates is unclear at this time (
Fig. 1Schematic showing location of known intramembrane cleavage sites in the γ-secretase substrates APP, NOTCH-1, and ErbB-4. The gray boxes represent the transmembrane domains. Human APP is cleaved after Val40 and Ala42, which generates Aβ40 and Aβ42 peptides. NSAID treatment decreases Aβ42 but increases Aβ peptide ending at residue 38. The predominant ϵ cleavage site between Leu49 and Val50 generates the AICD. The topologically similar S3 cleavage site in human NOTCH-1 between Gly1743 and Val1744 generates the NICD. γ-Secretase-dependent cleavage also occurs near the middle of the NOTCH-1 transmembrane domain with a predominant cleavage site between Ala1731 and Ala1732. The exact γ-secretase cleavage site within the transmembrane domain of the ErbB-4 receptor is currently unknown.
Because of its essential role in generation of the Aβ peptide, targeting γ-secretase activity remains a viable and important option for drug development in Alzheimer's disease therapeutics (
). γ-Secretase is a multiprotein complex consisting of at least four identified membrane-bound proteins: presenilin (PS), nicastrin, APH-1, and PEN-2 (
). Small molecule γ-secretase inhibitors efficiently suppress Aβ production in both cultured cells and in APP-transgenic mouse models of amyloid pathology (
). However, these inhibitors indiscriminately block all cleavages within the TMD of γ-secretase substrates and prevent formation of the intracellular signaling domains (
). Furthermore, a recent report demonstrated that γ-secretase inhibitors also block proteolytic activity of signal peptide peptidase, an intramembrane-cleaving aspartic protease with distant homology to PS (
We recently showed that certain nonsteroidal anti-inflammatory drugs (NSAIDs) specifically reduce production of the amyloidogenic Aβ42 peptide without apparent inhibition of NOTCH processing, specifically NICD formation (
Weggen, S., Eriksen, J. L., Sagi, S. A., Pietrzik, C. U., Ozols, V., Fauq, A., Golde, T. E., and Koo, E. H. (June 12, 2003) J. Biol. Chem. 10.1074/jbc.M303592200.
and full elucidation of their mechanism of action could facilitate the development of improved γ-secretase inhibitors that only target generation of a disease relevant subset of Aβ peptides. However, based only on the result that NSAIDs preserve S3 cleavage in NOTCH, it cannot be concluded with confidence that Aβ42-lowering NSAIDs will not impair intramembrane cleavage in other γ-secretase substrates. We therefore investigated APP and ErbB-4 processing after NSAID treatment, specifically assaying γ-secretase-mediated cleavages and signaling events with more sensitive measurements. Our results demonstrated that intramembrane γ-secretase cleavage of APP and ErbB-4 as well as downstream signaling through AICD were unaffected by treatment with Aβ42-lowering NSAIDs. These findings significantly strengthen the concept that these drugs preferentially target Aβ42 production but not other vital γ-secretase mediated cleavage and signaling events.
EXPERIMENTAL PROCEDURES
Drugs and Antibodies—The NSAIDs sulindac sulfide and ibuprofen were purchased from Biomol, and γ-secretase inhibitor L685,458 was from Bachem. All other chemicals were from Sigma except when otherwise indicated. Monoclonal antibody 9E10 against the Myc epitope sequence was purchased from Calbiochem, and polyclonal antibodies against the C terminus of human c-ErbB-4 were from Santa Cruz (C-18) and Neomarkers. The polyclonal antibody CT15 against the C-terminal 15 amino acid residues of APP has been described (
Cell Lines and Cell Culture—293T cells, 293 cells stably transfected with human APP695 harboring the “Swedish” mutation, CHO cells, APP-PS1ML CHO cells stably transfected with both wild type human APP751 and human mutant PS1 (M146L), and T47-14 cells (kindly provided by M. Kraus and G. Carpenter) were all maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum (Invitrogen).
cDNA Constructs—A construct containing the C-terminal 50 amino acids of APP was generated using PCR, cloned into pCIneo (Promega), and verified by sequencing. The construct (APP C50-Myc) was designed to include an additional methionine at the 5′-end as well as a Myc tag sequence between the methionine and the first valine residue. pCDNA3-FE65 was a gift from J. Buxbaum (
). Plasmids encoding the GAL4 DNA-binding domain (pMST), the GAL4 DNA-binding domain engineered into the cytoplasmic tail of APP695 (pMST-APP), the APP-GAL4 fusion protein with a mutation in the NPTY motif of the APP cytoplasmic tail (pMST-APPΔ), and a GAL4 reporter plasmid encoding firefly luciferase (pG5E1B-luc) have been described (
) and were kindly provided by T. Sudhof. Plasmid pRL-TK encoding Renilla luciferase was from Promega, and pCDNA3 was from Invitrogen.
AICD Detection—APP-PS1ML CHO cells (10-cm dish) were treated overnight with 2.5 μm γ-secretase inhibitor L685,458, 500 μm ibuprofen, 60 μm sulindac sulfide or Me2SO vehicle. Crude membrane extracts were then prepared as described (
) with modifications. The cells were collected and resuspended in 1 ml of hypotonic buffer (10 mm Tris, pH 7.4, 1 mm EDTA, 1 mm EGTA) containing 1× protease inhibitor mixture (Roche Applied Science) and homogenized by passing five times through a 27-gauge needle and five times through a 30-gauge needle. To prepare a postnuclear supernatant, the homogenate was centrifuged at 1,000 × g for 15 min at 4 °C. The membranes were then isolated from the supernatant by centrifugation at 20,000 × g for 45 min at 4 °C. The membranes were resuspended in 100 μl of RIPA buffer and cleared by a spin at 20,000 × g for 10 min at 4 °C. 10 μl of membrane extract was separated on 10–20% Tricine gels, and AICD was detected by Western blotting with antibody CT-15.
Detection of the Soluble ErbB-4 C-terminal Fragment—Fractionation of NIH 3T3 cells stably overexpressing ErbB-4 (T47-14 cells) and detection of the soluble ErbB-4 C-terminal fragment was performed according to published procedures with modifications (
). T47-14 cells were starved overnight in Dulbecco's modified Eagle's medium with 0.5% fetal bovine serum followed by treatment with 2.5 μm L685,483, 60 μm sulindac sulfide, 500 μm ibuprofen, or Me2SO vehicle for 2 h. This was followed by stimulation with 100 nm phorbol 12-myristate 13-acetate for 1 h in the presence of drug or Me2SO vehicle. Postnuclear supernatants were prepared from T47-14 cells as described above, and NaCl was added to a final concentration of 150 mm. The membranes were isolated by centrifugation at 100,000 × g for 60 min at 4 °C. Nonidet P-40 was added to the cytosolic supernatant at a final concentration of 1%. The membranes in the pellet were resuspended in 1 ml of RIPA buffer and cleared by a spin at 20,000 × g for 10 min at 4 °C. The cytosol and membranes were immunoprecipitated with 4 μg of polyclonal ErbB-4 antibody (C-18, 1:50 dilution). The immunoprecipitated material was separated on 4–12% Bis-Tris gels, and ErbB-4 was detected by Western blotting with polyclonal antibody against amino acids 1285–1308 of the ErbB-4 C terminus (5 μg/ml; Neomarkers).
AICD in Vitro Generation—In vitro generation of AICD was performed as described (
) with modifications. Postnuclear supernatants were prepared from 293 cells stably transfected with APP695 harboring the Swedish mutation as described above, and the membranes were isolated by centrifugation at 20,000 × g for 45 min at 4 °C. The membranes were washed once with 200 μl of homogenization buffer and pelleted again at 20,000 × g for 30 min at 4 °C. The membranes corresponding to cells from half of a 10-cm dish were then resuspended in 25 μl of assay buffer (150 mm sodium citrate, pH 6.4) containing 2.5 μm γ-secretase inhibitor L685,458, ibuprofen, or sulindac sulfide (prepared by serial dilution of the highest drug concentration in assay buffer) or Me2SO. The membrane preparations were incubated for 2 h at 37 °C to allow in vitro generation of AICD. In parallel, control samples were kept at 4 °C. The membranes were then pelleted by centrifugation at 20,000 × g for 30 min at 4 °C, and the supernatants were separated on 4–12% Bis-Tris gels. AICD was detected by Western blotting with antibody CT-15.
Stabilization of AICD by FE65—The APP C50-Myc construct was co-transfected into 293T cells (10-cm dish) with either pCMS-GFP control plasmid or pCDNA3-FE65. 36 h after transfection, the cells were treated overnight with Me2SO, 60 μm sulindac sulfide, or 500 μm ibuprofen. 48 h after transfection, the cells were extracted with Nonidet P-40 lysis buffer (1% Nonidet P-40, 50 mm Tris, 150 mm NaCl containing 1× protease inhibitor mixture) for 30 min and centrifuged at 20,000 × g for 10 min at 4 °C. Insoluble material was re-extracted in 100 μl of 1% SDS (
), sonicated, and then diluted to 0.1% SDS and cleared by centrifugation at 20000 × g at 4 °C for 20 min. Nonidet P-40 and SDS extracts were immunoprecipitated with monoclonal anti-Myc antibody 9E10. Immunoprecipitated material was separated on 4–12% Bis-Tris gels, and the APP C50-Myc fragment was detected by Western blotting with antibody CT-15.
AICD Reporter Assay—Subconfluent CHO cells were transiently transfected in 12-well plates using FuGENE 6 with the following combinations of plasmids (0.3 μg each): 1) pMST + pG5EIB-luc + pCDNA3; 2) pMST-APP + pG5EIB-luc + pCDNA3; 3) pMST-APP + pG5EIB-luc + pCDNA3-FE65; and 4) pMST-APPΔ + pG5EIB-luc + pCDNA3-FE65. 20 ng of pRL-TK was added to each plasmid mix to control for transfection efficiency. 24 h after transfection, the cells were treated for additional 24 h with 2.5 μm γ-secretase inhibitor L685,458, 500 μm ibuprofen, 60 μm sulindac sulfide, or Me2SO vehicle. The cells were then lysed, and firefly and Renilla luciferase activities were quantified using a dual luciferase reporter assay system (Promega) and a dual injector luminometer (EG&G Berthold). Firefly luciferase values were standardized to the corresponding Renilla luciferase values. To control for nonspecific drug effects, these values were then normalized for transactivation observed in cells expressing the GAL4 DNA-binding domain alone and treated with the respective drug. The values expressed as fold induction over GAL4 are averages of triplicate measurements, and one of two representative experiments is shown.
RESULTS
Aβ42-lowering NSAIDs Do Not Impair AICD or ErbB4 C-terminal Fragment Generation—To gain further insight into the specificity of Aβ42-lowering NSAIDs, we investigated intramembrane proteolysis of APP and ErbB-4 by examining the corresponding C-terminal fragments after γ-secretase cleavage. These specific γ-secretase substrates were chosen for several reasons. First, emerging evidence supported a physiological function for the intracellular signaling domains of both of these proteins (
). Second, the TMD sequences of other γ-secretase substrates including ErbB-4 are entirely different from APP and NOTCH, and the γ-secretase cleavage sites are unknown (
) (Fig. 1). Consequently, the degree to which the pharmacological properties of these other cleavage substrates are similar to the Aβ42 scissile bond in the APP TMD has not been established. Third, sulindac sulfide was shown to modulate AICD generation in vitro, and ibuprofen suppressed AICD-mediated activation of the KAI1/CD82 promotor (
), indicating that Aβ42-lowering NSAIDs might interfere with AICD function.
To analyze AICD generation, APP-PS1ML CHO cells stably transfected with both wild type APP751 and the PS1 mutant M146L were treated with sulindac sulfide (60 μm), ibuprofen (500 μm), γ-secretase inhibitor L685,458, or Me2SO vehicle. We previously showed that at these concentrations of NSAIDs, Aβ42 secretion was reduced by 60–70% (
). Crude membrane extracts were then prepared and immunoblotted with an antibody against the C terminus of APP. As expected, treatment with the γ-secretase inhibitor strongly suppressed AICD formation but markedly elevated APP C-terminal fragments. However, neither sulindac sulfide nor ibuprofen reduced AICD production as compared with Me2SO-treated control CHO cells (Fig. 2).
Fig. 2Aβ42-lowering NSAIDs preserve intramembrane cleavage in APP and generation of the APP intracellular domain (AICD). Solubilized membrane preparations of APP-PS1ML CHO cells stably transfected with both wild type APP751 and mutant PS1 (M146L) were probed with an antibody against the C terminus of APP. Treatment with the γ-secretase inhibitor L685,458 strongly suppressed AICD formation and induced accumulation of APP C-terminal fragments (CTF). In contrast, neither sulindac sulfide nor ibuprofen reduced AICD production as compared with dimethyl sulfoxide-treated (DMSO) control cells. A construct encoding the C-terminal 50 amino acids of APP, which correspond to the AICD fragment, was expressed in 293T cells and used as a standard shown on the right. Note that this fragment (C50-myc) runs slightly higher than AICD because of the attached Myc tag.
). This subsequently triggers intramembrane cleavage of the remaining membrane-bound C-terminal fragment and release of a soluble 80-kDa intracellular domain (
). To investigate whether γ-secretase-mediated intramembrane cleavage of the ErbB-4 receptor would be compromised by NSAID treatment, T47-14 NIH3T3 cells stably transfected with ErbB-4 were stimulated with phorbol 12-myristate 13-acetate in the presence of the indicated drugs. Cytosolic and membrane fractions were obtained and analyzed by combined immunoprecipitation/Western blotting with antibodies against the C terminus of ErbB-4. Low amounts of the soluble ErbB-4 intracellular domain, but not full-length ErbB-4, were present in the cytosolic fraction of vehicle-treated control cells (Fig. 3A). This fragment was undetectable after treatment with γ-secretase inhibitor. However, as with AICD generation, neither sulindac sulfide nor ibuprofen had an effect on the formation of the soluble ErbB-4 domain as compared with vehicle control (Fig. 3A). There was also no effect on the levels of the membrane-bound C-terminal fragment or full-length ErbB-4 in the membrane fraction after treatment with NSAIDs or γ-secretase inhibitor (Fig. 3B). These results indicated that sulindac sulfide and ibuprofen did not interfere with intramembrane cleavage in APP and ErbB-4 at concentrations that caused significant reduction in Aβ42 production in previous studies (
Fig. 3Aβ42-lowering NSAIDs preserve intramembrane cleavage in the ErbB-4 receptor and generation of the soluble ErbB-4 intracellular domain. Ectodomain shedding and intramembrane cleavage of the ErbB-4 receptor was induced by treatment of T47-14 cells with phorbol 12-myristate 13-acetate. The cells were then fractionated into cytosolic and membrane fractions, and ErbB-4 was detected by combined immunoprecipitation/Western blotting with antibodies against the C terminus of ErbB-4. Low amounts of the soluble ErbB-4 intracellular domain (sCTF), but not full-length ErbB-4, were detected in the cytosolic fraction, and this fragment was abolished by treatment of cells with γ-secretase inhibitor. Sulindac sulfide or ibuprofen treatment did not impair formation of the soluble ErbB-4 domain as compared with vehicle control (A). Neither NSAIDs nor γ-secretase inhibitor had any effect on levels of the membrane-bound C-terminal ErbB-4 fragment (mCTF) or full-length ErbB-4 in the membrane fraction (B). DMSO, dimethyl sulfoxide.
Toxic Concentrations of NSAIDs Do Not Inhibit AICD Generation in Vitro—We next proceeded to examine AICD generation and downstream signaling in more detail. Takahashi et al. (
) recently reported that production of AICD was modulated by sulindac sulfide in an in vitro γ-secretase assay with an unusual biphasic response: at low concentrations (1–50 μm), a severalfold increase in AICD production was observed, but at 100 μm, AICD generation was almost completely inhibited. This result is in contrast to the lack of any observable effect on AICD generation after intact cells were treated with 60 μm sulindac sulfide (Fig. 2). To encompass the same drug dosages used by Takahashi et al. (
), we therefore tested a wide range of sulindac sulfide and ibuprofen concentrations using a previously described cell-free in vitro assay for AICD generation (
). When given to living cells, sulindac concentrations above 150 μm and ibuprofen concentrations above 1 mm are associated with cellular toxicity, which could mask any effect on AICD generation. However, the in vitro assay allowed us to circumvent this toxicity issue and to evaluate whether high NSAID concentrations would eventually inhibit AICD generation. AICD levels were assessed after incubation of membrane preparations of 293 cells stably expressing APP695 with the Swedish mutation for 2 h at 37 °C in the presence of increasing concentrations of sulindac sulfide or ibuprofen. Additional samples were kept at 4 °C to control for background AICD levels. After the 2-h incubation, high levels of AICD production were readily seen in samples incubated at 37 °C but not at 4 °C in the presence of Me2SO vehicle (Fig. 4). γ-Secretase inhibitor L685,458 reduced AICD generation to background levels. However, neither sulindac sulfide at concentrations from 10 to 400 μm (Fig. 4A) nor ibuprofen at concentrations from 100 to 2,000 μm (Fig. 4B) displayed any effect on AICD production in vitro. Therefore, at NSAID concentrations that far exceeded what was necessary to lower Aβ42 in cell-based assays and indeed would have caused significant toxicity, there was no evidence of any impairment of AICD production in vitro.
Fig. 4Aβ42-lowering NSAIDs do not impair AICD production in vitro. Membrane preparations of 293 cells stably overexpressing APP695 with the Swedish mutation were incubated for 2 h at 37 °C to allow in vitro generation of AICD. The membranes were then pelleted by centrifugation, and the supernatants were probed with an antibody against the C terminus of APP. Very low amounts of AICD were detected in control samples incubated at 4 °C, but robust AICD production was observed in samples incubated at 37 °C in the presence of dimethyl sulfoxide (DMSO) vehicle. γ-Secretase inhibitor L685,458 diminished AICD generation to almost undetectable levels. Neither sulindac sulfide (A) nor ibuprofen (B) displayed any effect on AICD production in vitro at all concentrations tested.
NSAIDs Do Not Impair AICD/FE65-mediated Transcription—AICD is rapidly degraded in vivo, but upon binding to the adapter protein FE65, AICD is stabilized and accumulates in both the cytosol and nucleus, the latter presumably to activate downstream nuclear transcription (
). This suggested that NSAIDs might influence downstream signaling by AICD at some step distal to generation of AICD by γ-secretase. We therefore investigated whether Aβ42-lowering NSAIDs would impair stabilization of AICD by FE65 or AICD-mediated transcription. To analyze the former, 293T cells were transiently transfected with a plasmid encoding the last 50 amino acid residues of APP containing a Myc tag (APP C50-Myc) with or without FE65. The cells were treated with sulindac sulfide, ibuprofen, or Me2SO vehicle and then sequentially extracted with 1% Nonidet P-40 followed by 1% SDS to assay for AICD in cytosolic or nuclear fractions as described (
). When expressed alone, low amounts of the APP C50-Myc fragment were present in the Nonidet P-40-soluble fraction, but it was undetectable in the SDS-soluble fraction (Fig. 5). Co-expression of FE65 stabilized APP C50-Myc in the Nonidet P-40-soluble fraction and further allowed its detection in the SDS-soluble fraction. However, neither sulindac sulfide nor ibuprofen treatment perturbed APP C50-Myc levels to any detectable degree as compared with Me2SO vehicle, demonstrating that Aβ42-lowering NSAIDs did not hinder stabilization of an APP C-terminal fragment corresponding to AICD by FE65 (Fig. 5).
Fig. 5Aβ42-lowering NSAIDs do not prevent AICD stabilization by FE65. 293T cells were transiently transfected with either a plasmid encoding the C-terminal 50 amino acids of APP alone, or co-transfected with plasmid encoding FE65. Cells were then treated with drugs or dimethyl sulfoxide (DMSO) vehicle and sequentially extracted with 1% Nonidet P-40 (upper panel) followed by 1% SDS (lower panel). The Myc-tagged C50-fragment (APP C50-Myc) was immunoprecipitated with anti-Myc antibody 9E10 and detected by Western blotting using an antibody against the C terminus of APP. When expressed alone, low amounts of the APP C50-Myc fragment were present in the Nonidet P-40-soluble fraction, but it was undetectable in the SDS-soluble fraction. Co-expression of FE65 stabilized APP C50-Myc in the Nonidet P-40-soluble fraction and allowed detection in the SDS-soluble fraction. Sulindac sulfide and ibuprofen treatment had no effects on APP C50-Myc stabilization as compared with dimethyl sulfoxide vehicle.
Although the immunoblotting approaches described above have been consistently validated by a number of laboratories, it can be argued that these are not very sensitive assays, which could miss smaller changes. Therefore, we examined potential effects of Aβ42-lowering NSAIDs on AICD-mediated transcription using a previously described heterologous reporter assay system. In this approach, a GAL4 DNA-binding domain engineered into the cytoplasmic tail of APP695 is released by γ-secretase-mediated intramembrane cleavage and binds to a GAL4-dependent promotor driving a luciferase reporter gene (
). Expression of the APP-GAL4 fusion protein in CHO cells resulted in only minimal transactivation of the reporter gene as compared with expression of the GAL4 DNA-binding domain alone (Fig. 6). However, co-expression of FE65 with the APP-GAL4 fusion protein induced a >16-fold stimulation of reporter transcription activity. This induction was abolished when an APP-Gal4 construct with a mutation in the NPTY motif (APPΔ-GAL4), which prevented interaction with FE65, was expressed (
). As expected, induction of reporter activity was also largely eliminated when cells were treated with γ-secretase inhibitor L685,458. However, neither sulindac sulfide nor ibuprofen demonstrated any effect on reporter gene transactivation by APP-Gal4 alone or by co-expression of APP-Gal4/FE65 (Fig. 6). Taken together, these results demonstrated that Aβ42-lowering NSAIDs did not interfere with either AICD generation in vitro and in vivo, stabilization by FE65, or AICD-mediated transcription of a heterologous reporter gene.
Fig. 6Aβ42-lowering NSAIDs do not modulate AICD-mediated transcription in a heterologous reporter assay system. The results are expressed as fold induction in transactivation over control cells expressing the GAL4 DNA-binding domain alone and treated with the respective drug. Expression of an APP-GAL4 fusion protein induced only minimal transactivation of the GAL4-dependent reporter gene. Co-expression of FE65 with the APP-GAL4 fusion protein strongly stimulated transcription of the reporter, and this induction was abolished when APP-Gal4 with a mutation in the NPTY motif (APPΔ-GAL4) was used. Induction was also reduced when cells were treated with γ-secretase inhibitor L685,458. Sulindac sulfide and ibuprofen demonstrated no significant effects on reporter gene transactivation by APP-Gal4 alone or by co-expression of APP-Gal4/FE65. DMSO, dimethyl sulfoxide.
Epidemiological studies indicated that chronic users of NSAIDs have a lower risk to develop Alzheimer's disease, and high dose ibuprofen treatment was shown to reduce amyloid pathology in aging APP-transgenic mice (
), these results suggested that some NSAIDs, in addition to their anti-inflammatory properties, might directly ameliorate amyloid pathology in Alzheimer's disease. We showed that the NSAID effect on Aβ42 was fully retained in cells lacking cyclooxygenase activity, thereby excluding the primary pharmacological targets of NSAIDs as mediators of the Aβ42 reduction (
).2 However, definitive identification of this cellular target will have to await further photo or chemical affinity binding studies.
The degree of specificity to which NSAIDs target Aβ42 production over other γ-secretase-mediated cleavages is an important issue for clinical and conceptual reasons, especially if future drug development efforts are based upon existing Aβ42-lowering NSAIDs or their mechanism of action. NSAIDs are pleiotropic compounds with numerous cellular targets (
). In the limited studies to date, however, NSAIDs have shown remarkable specificity on APP processing. The reduction in the Aβ42 level after NSAID treatment was accompanied by an increase in shorter Aβ species, particularly Aβ38, suggesting that these compounds, rather than inhibiting overall γ-secretase activity, induce a subtle shift in the cleavage pattern. No generalized effects on APP expression, APP half-life, secretion of the soluble APP ectodomain, internalization of APP from the cell surface, or accumulation of C-terminal fragments were observed (
). However, in the absence of a fully defined mechanism of action and based solely on the observed preservation of NOTCH cleavage, it cannot be assumed that Aβ42-lowering NSAIDs will not affect intramembrane cleavages in other γ-secretase substrates. In particular, generation of the intracellular signaling domains of APP and other γ-secretase substrates has not been investigated in detail. In this respect, our new findings that, in cell-based assays, sulindac sulfide and ibuprofen did not affect ϵ cleavage in the APP TMD and intramembrane cleavage of the ErbB-4 receptor provided additional support for the unexpected specificity of these compounds in their actions on γ-secretase activity.
In this study, we further investigated the effects of Aβ42-lowering NSAIDs on AICD generation in vitro and on downstream signaling mediated by the AICD fragment. Takahashi et al. (
) reported that sulindac sulfide modulated AICD generation in a γ-secretase in vitro assay in an unusual biphasic manner. They observed an increase in AICD production at sulindac sulfide concentrations up to 50 μm and near complete inhibition at 100 μm. Using a slightly different γ-secretase in vitro assay, we were not able to confirm either an increase in AICD generation at low concentrations of sulindac sulfide or inhibition at higher concentrations. In fact, we did not detect any inhibitory effect up to 400 μm sulindac sulfide or 2,000 μm ibuprofen, concentrations that are severalfold in excess of those required to lower Aβ42 production in cell-based assays (
). We suspect that differences in the respective in vitro assay conditions may account for the discrepancies between our results and those of Takahashi et al. In particular, the latter study utilized solubilized membranes as an enzyme source and a recombinant peptide consisting of the 100 C-terminal amino acids of APP as substrate (
). In contrast, we used membrane preparations derived from intact cells containing both the enzymatic activity and the substrate APP for in vitro generation of AICD (
). It is conceivable that our assay system more closely resembles the conditions in intact cells and that, consequently, our results in the cell-free system are more concordant with the cell-based studies. Regardless, our results from both cell-based and in vitro assays indicate that inhibition of AICD generation is unlikely to occur under in vivo conditions with nontoxic concentrations of Aβ42-lowering NSAIDs.
Finally, our results showed that ibuprofen and sulindac sulfide did not interfere with either the stabilization of AICD by FE65 or AICD-mediated transcription using a heterologous reporter gene system. The lack of effect on transactivation in a sensitive reporter system reinforced the notion that AICD generation is not compromised by Aβ42-lowering NSAIDs, because transcriptional activation of the reporter also requires γ-secretase-mediated release of the APP C terminus fused to the GAL4 DNA-binding domain. In a recent study, ibuprofen treatment interfered with activation of the KAI1/CD82 promoter by a trimeric AICD-FE65-Tip60 complex (
), suggesting that NSAIDs might impair AICD-mediated transcription. Our results with the reporter assay indicated that ibuprofen and sulindac sulfide did not prevent formation of this AICD-FE65-Tip60 signaling complex, because the same ternary complex assembles to drive the reporter gene (
), and our co-transfection studies clearly showed that stabilization of an APP fragment corresponding to AICD by FE65 was not prevented by ibuprofen or sulindac sulfide treatment. However, if future studies reveal more in vivo target genes of AICD, then a potential influence of NSAIDs on these specific AICD-mediated transcription pathways will have to be investigated in further detail.
In conclusion, the results presented in this study provide important additional evidence for the unprecedented specificity with which some NSAIDs target Aβ42 production. This feature strongly separates Aβ42-lowering NSAIDs from published γ-secretase inhibitors, which indiscriminately inhibit cleavages in the TMD of γ-secretase substrates including NOTCH and APP. If currently available NSAIDs or newly developed compounds with a similar mechanism of action were ever to be used clinically as Aβ42-lowering agents, then they might avoid serious side effects that are likely associated with global inhibition of γ-secretase function.
Acknowledgments
We are indebted to Drs. Graham Carpenter and Matthias Kraus for the T47-14 cells and for help with analysis of ErbB-4 cleavage. We also thank Drs. Thomas Sudhof and Xinwei Cao for the generous gift of the AICD reporter plasmids and Dr. Joseph Buxbaum for the FE65 construct.
84106-4&id=gr1.jpg){kind=link}
84106-4&id=gr2.jpg){kind=link}
84106-4&id=gr3.jpg){kind=link}
84106-4&id=gr4.jpg){kind=link}
84106-4&id=gr5.jpg){kind=link}
84106-4&id=gr6.jpg){kind=link}