Processing of β-Amyloid Precursor-like Protein-1 and -2 by γ-Secretase Regulates Transcription
2002; Elsevier BV; Volume: 277; Issue: 46 Linguagem: Inglês
10.1074/jbc.m208110200
ISSN1083-351X
AutoresMeir H. Scheinfeld, Enrico Ghersi, Karen Laky, B. J. Fowlkes, Luciano D'adamio,
Tópico(s)Prion Diseases and Protein Misfolding
ResumoThe familial Alzheimer's disease gene product β-amyloid (Aβ) precursor protein (APP) is processed by the β- and γ-secretases to produce Aβ as well as AID (APP Intracellular Domain) which is derived from the extreme carboxyl terminus of APP. AID was originally shown to lower the cellular threshold to apoptosis and more recently has been shown to modulate gene expression such that it represses Notch-dependent gene expression while in combination with Fe65 it enhances gene activation. Here we report that the two other members of the APP family, β-amyloid precursor-like protein-1 and -2 (APLP1 and APLP2), are also processed by the γ-secretase in a Presenilin 1-dependent manner. Furthermore, the extreme carboxyl-terminal fragments produced by this processing (here termed APP-likeIntracellular Domain or ALID1 and ALID2) are able to enhance Fe65-dependent gene activation, similar to what has been reported for AID. Considering that only APP and not the APLPs have been linked to familial Alzheimer's disease (AD), this data should help in understanding the physiologic roles of the APP family members and in differentiating these functions from the pathologic role of APP in Alzheimer's disease. The familial Alzheimer's disease gene product β-amyloid (Aβ) precursor protein (APP) is processed by the β- and γ-secretases to produce Aβ as well as AID (APP Intracellular Domain) which is derived from the extreme carboxyl terminus of APP. AID was originally shown to lower the cellular threshold to apoptosis and more recently has been shown to modulate gene expression such that it represses Notch-dependent gene expression while in combination with Fe65 it enhances gene activation. Here we report that the two other members of the APP family, β-amyloid precursor-like protein-1 and -2 (APLP1 and APLP2), are also processed by the γ-secretase in a Presenilin 1-dependent manner. Furthermore, the extreme carboxyl-terminal fragments produced by this processing (here termed APP-likeIntracellular Domain or ALID1 and ALID2) are able to enhance Fe65-dependent gene activation, similar to what has been reported for AID. Considering that only APP and not the APLPs have been linked to familial Alzheimer's disease (AD), this data should help in understanding the physiologic roles of the APP family members and in differentiating these functions from the pathologic role of APP in Alzheimer's disease. The APP 1The abbreviations used are: APP, Aβ precursor protein; Aβ, β-amyloid; APLP, Aβ precursor-like protein; AD, Alzheimer's disease; AID, APP intracellular domain; ALID, APP-like intracellular domain; GST, glutathione S-transferase; YFP, yellow fluorescent protein; PS, Presenilin; DAPI, 4′,6-diamidino-2-phenylindole family consists of three family members, APP, APLP1, and APLP2 (1Wasco, W., Bupp, K., Magendantz, M., Gusella, J. F., Tanzi, R. E. & Solomon, F. Proc. Natl. Acad. Sci. U. S. A., 89, 10758–10762.Google Scholar, 2Paliga K. Peraus G Kreger S. Durrwang U. Hesse L. Multhaup G. Masters C.L. Beyreuther K. Weidemann A. Eur. J. Biochem. 1997; 250: 354-363Crossref PubMed Scopus (49) Google Scholar, 3Wasco W. Gurubhagavatula S. Paradis M., D. Romano D.M. Sisodia S.S. Hyman B.T Neve R.L. Tanzi R.E. Nat. Genet. 1993; 5: 95-100Crossref PubMed Scopus (322) Google Scholar, 4Bayer T.A. Cappai R. Masters C.L. Beyreuther K. Multhaup G. Mol. Psychiatry. 1999; 4: 524-528Crossref PubMed Scopus (81) Google Scholar). Most research on the family has been focused on APP because it has been directly implicated in Alzheimer's disease (5Sisodia S.S. St George-Hyslop P.H. Nat. Rev. Neurosci. 2002; 3: 281-290Crossref PubMed Scopus (487) Google Scholar, 6Haass C. De Strooper B. Science. 1999; 286: 916-919Crossref PubMed Scopus (366) Google Scholar). APP undergoes extensive proteolytic processing along two major pathways. APP can be cleaved extracellularly by the α-secretase creating a C83 membrane-bound intermediate, which is subsequently cleaved by the γ-secretase releasing a non-amyloidogenic fragment termed p3. Alternatively, APP can be cleaved extracellularly by the β-secretase forming a C99 membrane-bound intermediate, which is subsequently cleaved by the γ-secretase releasing the amyloidogenic Aβ fragment. Aβ is the major component of the amyloid plaques found in the brains of patients with AD and is considered to be the major underlying cause of the disease. An additional peptide termed AID, extending from the γ-secretase cleavage site to the carboxyl terminus of APP, is also produced following γ-secretase cleavage regardless of whether it was preceded by α or β cleavage. This AID peptide was first identified in the brains of patients with AD and normal controls, and was shown to either induce or sensitize cells to apoptosis (7Passer B. Pellegrini L. Russo C. Siegel R.M. Lenardo M.J. Schettini G. Bachmann M. Tabaton M. D'Adamio L. J. Alzheimers Dis. 2000; 2: 289-301Crossref PubMed Scopus (202) Google Scholar, 8D'Adamio L. Israel A., De Strooper B. Checler F. Christen Y. Notch from Neurodevelopment to Neurodegeneration: Keeping the Fate. Fondation IPSEN, Springer, Berlin2002: 101-108Google Scholar). More recently AID has been found to participate in transcription (9Cao X. Sudhof T.C. Science. 2001; 293: 115-120Crossref PubMed Scopus (1054) Google Scholar, 10Roncarati R. Sestan N. Scheinfeld M.H. Berechid B.E. Lopez P.A. McGlade J.C. Meucci O. Rakic P. D'Adamio L. Proc. Natl. Acad. Sci. U. S. A. 2002; 99: 7102-7707Crossref PubMed Scopus (181) Google Scholar, 11Gao Y. Pimplikar S.W. Proc. Natl. Acad. Sci. U. S. A. 2001; 98: 14979-14984Crossref PubMed Scopus (210) Google Scholar, 12Baek S.H. Ohgi K.A. Rose D.W. Koo E.H. Glass C.K. Rosenfeld M.G. Cell. 2002; 110: 55-67Abstract Full Text Full Text PDF PubMed Scopus (495) Google Scholar) by activating gene expression in combination with Fe65 and repressing activation of genes induced by Notch. These data have opened the important possibility that just as Notch undergoes regulated intramembranous proteolysis (13Brown M.S., Ye, J. Rawson R.B. Goldstein J.L. Cell. 2000; 100: 391-398Abstract Full Text Full Text PDF PubMed Scopus (1149) Google Scholar) by the γ-secretase, so does APP. Much less is known about APLP1 and APLP2, the two other APP family members. The human APLP1 gene is located on chromosome 19q13.1 (14Lenkkeri U. Kestila M. Lamerdin J. McCready P. Adamson A. Olsen A. Tryggvason K. Hum. Genet. 1998; 102: 192-196Crossref PubMed Scopus (18) Google Scholar), and APLP2 is located on chromosome 11q23-q25 (15von der Kammer H. Loffler C. Hanes J. Klaudiny J. Scheit K.H. Hansmann I. Genomics. 1994; 20: 308-311Crossref PubMed Scopus (22) Google Scholar). Although no clear neuronal functions have been found for these molecules, APLP1 has been implicated in synaptogenesis (16Kim T.W., Wu, K., Xu, J.L. McAuliffe G. Tanzi R.E. Wasco W. Black I.B. Brain Res. Mol. Brain Res. 1995; 32: 36-44Crossref PubMed Scopus (61) Google Scholar), and recombinant APLP2 possess the ability to promote neurite outgrowth (17Cappai R. Mok S.S. Galatis D. Tucker D.F. Henry A. Beyreuther K. Small D.H. Masters C.L. FEBS Lett. 1999; 442: 95-98Crossref PubMed Scopus (50) Google Scholar). There is also evidence that the APLPs undergo processing in the ectodomain. Secretory APLP1 consisting of the APLP1 extracellular domain has been detected in human cerebrospinal fluid and in the conditioned media of cells overexpressing APLP1 (2Paliga K. Peraus G Kreger S. Durrwang U. Hesse L. Multhaup G. Masters C.L. Beyreuther K. Weidemann A. Eur. J. Biochem. 1997; 250: 354-363Crossref PubMed Scopus (49) Google Scholar). APLP2 also sheds it extracellular domain in multiple cell culture systems (18Xu K.P. Zoukhri D. Zieske J.D. Dartt D.A. Sergheraert C. Loing E. Yu F.S. Am. J. Physiol. Cell Physiol. 2001; 281: C603-C614Crossref PubMed Google Scholar, 19Lo A.C. Thinakaran G. Slunt H.H. Sisodia S.S. J. Biol. Chem. 1995; 270: 12641-12645Abstract Full Text Full Text PDF PubMed Scopus (20) Google Scholar). Carboxyl-terminal fragments of APLP1 and APLP2 have also been identified in rat brains (20Gu Y. Misonou H. Sato T. Dohmae N. Takio K. Ihara Y. J. Biol. Chem. 2001; 276: 35235-35238Abstract Full Text Full Text PDF PubMed Scopus (269) Google Scholar). Both APLP1 (21Galvan V. Chen S., Lu, D. Logvinova A. Goldsmith P. Koo E.H. Bredesen D.E. J. Neurochem. 2002; 82: 283-294Crossref PubMed Scopus (83) Google Scholar) and APLP2 (22Cowan C.M. Thai J. Krajewski S. Reed J.C. Nicholson D.W. Kaufmann S.H. Roskams A.J. J. Neurosci. 2001; 21: 7099-7109Crossref PubMed Google Scholar) are also cleaved by caspases within their cytoplasmic domains at the conserved VEVD motif. Still, the significance of these cleavages are unknown. Work using knockout mice has revealed an aspect of the relationships between the various APP family members. Single disruptions of APP, APLP1, or APLP2 cause minor abnormalities, which are distinct for the different family members (23von Koch C.S. Zheng H. Chen H. Trumbauer M. Thinakaran G. van der Ploeg L.H. Price D.L. Sisodia S.S. Neurobiol. Aging. 1997; 18: 661-669Crossref PubMed Scopus (287) Google Scholar, 24Heber S. Herms J. Gajic V. Hainfellner J. Aguzzi A. Rulicke T. von Kretzschmar H. von Koch C. Sisodia S. Tremml P. Lipp H.P. Wolfer D.P. Muller U. J. Neurosci. 2000; 20: 7951-7963Crossref PubMed Google Scholar). Some of the differences in phenotype are probably related to the expression of APLP1 being restricted mostly to the nervous system (14Lenkkeri U. Kestila M. Lamerdin J. McCready P. Adamson A. Olsen A. Tryggvason K. Hum. Genet. 1998; 102: 192-196Crossref PubMed Scopus (18) Google Scholar). Furthermore, although mice with both APP and APLP1-disrupted are normal, APP plus APLP2 or APLP1 plus APLP2 double knockout mice die soon after birth (24Heber S. Herms J. Gajic V. Hainfellner J. Aguzzi A. Rulicke T. von Kretzschmar H. von Koch C. Sisodia S. Tremml P. Lipp H.P. Wolfer D.P. Muller U. J. Neurosci. 2000; 20: 7951-7963Crossref PubMed Google Scholar). These observations indicate that although there are unique roles for each of the different APP family members during development, there is also functional redundancy among them that allows compensation in single knockout mice. To gain more insight into the APP family, we sought to identify aspects of the APLPs that are similar to or different from APP. In this report we show that APLP1 and APLP2 are substrates of the γ-secretase and are processed to release ALID1 and ALID2. This processing is also shown to be Presenilin 1 (PS1)-dependent. We further show that the carboxyl terminal fragments of the APLPs bind Fe65 and may function in combination with Fe65 to activate gene expression. Finally, we show that full-length APLPs restrict Fe65 from entering the nucleus prior to γ-secretase processing. Gal4-APP and Gal4-APPCT44 constructs were obtained from Dr. Thomas Sudhof. Gal4-APLP1CT44 and Gal4-APLP2CT44 were made by PCR on the carboxyl terminal of APLP1 with primers sense: 5′-AAG GGG ATC CCA CCG GGA AAG CCC TAC GGG GCT ATC AGC-3′ and antisense: 5′-TCG CAG TCG ACT CAG GGT-3′ 5′-CGT TCC TCC AGG AAG or APLP2-3′ with primers sense: 5′-AAG GGG ATC CCA CCG GGA CAG TAT GGC ACC ATC AGC CAC-3′ and antisense: 5′-TCG CAG TCG ACC TAA ATC TGC ATC TGC TCC AG-3′ and replacing the APPCT44 fragment used by Cao and Sudhof (9Cao X. Sudhof T.C. Science. 2001; 293: 115-120Crossref PubMed Scopus (1054) Google Scholar). Gal4-APLP1 and Gal4-APLP2 were constructed by adding the cytoplasmic and transmembrane domains before the Gal4 tag by PCR into the NheI site. For APLP1 the following primers were used. Sense: 5′-TTT TTT GCT AGC GAT GGG GCC CGC CAG CCC CGC TGC TC-3′ and antisense: 5′-TTT TTT GCT AGC GGC TTC CTG CGC AGG AGC AGC ATG GAG-3′ and for APLP2 sense: 5′-TTT TTT GCT AGC GAT GGC GGC CAC CGG GAC CGC GGC CG-3′ and antisense: 5′-TTT TTT GCT AGC GGC CTC TTC CTC AGC ATC ACC AGG CTG-3′. Mutants were made by PCR as above with mutations incorporated into the primers. Cloning of full-length APP was described previously (7Passer B. Pellegrini L. Russo C. Siegel R.M. Lenardo M.J. Schettini G. Bachmann M. Tabaton M. D'Adamio L. J. Alzheimers Dis. 2000; 2: 289-301Crossref PubMed Scopus (202) Google Scholar). Full-length APLP1 was expressed directly from the expressed sequence tag I.M.A.G.E. Clone ID: 4180020 in pCMV-SPORT6 (under control of the cytomegalovirus promoter). APLP2 full-length was cloned from expressed sequence tag I.M.A.G.E. Clone ID: 2820109 into pcDNA3. Robust expression was confirmed by Western blotting (data not shown). pGEX5–1 constructs to produce GST fusion proteins and cyan fluorescent protein-tagged constructs both coding for the terminal 57 amino acids of APLP1 and APLP2 were generated by PCR. For ALID1 sense: 5′-AAA CTC GAG GCC GCC ATG ATC GTC CTC TCC ATG CTG CTC-3′ and antisense: 5′-TTA CCG GTC GGG GTC GTT CCT CCA GGA AGC G-3′, and for ALID2 sense: 5′-AAA CTC GAG GCC GCC ATG ATC GTC ATC AGC CTG GTG ATG-3′ and antisense: 5′-TTT ACC GGT CGA ATC TGC ATC TGC TCC AGG-3′. Mutants were made by PCR with mutations introduced into the primers. All constructs obtained using PCR were confirmed by sequencing. YFP-Fe65 was obtained from Dr. Tommaso Russo. FLAG-Fe65 and GFP-Fe65 were constructed by moving the Fe65 insert from YFP-Fe65 to the appropriate vectors. PS1 D385A was obtained from Dr. Dennis J. Selkoe. PS1+/− mice obtained from Dr. S. S. Sisodia were used. PS1+/− mice were intercrossed to obtain PS1 wild type (PS1+/+, PS1+/−) and PS1 knockout pups. E17.5–18.5 pregnant females were sacrificed, and PS1−/− pups were identified by skeletal defects and the presence of cerebral hemorrhages. Fetal tissues (from 2–3 embryos pooled) were harvested directly into, and then homogenized in, lysis buffer (20 mm Tris, pH 7.5, 150 mm NaCl, 1 mm EDTA, 1% Nonidet P-40, 1 mm Na3VO4) with protease inhibitors (Roche Molecular Biochemicals complete tablets). Samples were incubated on ice for more than 30 min, inverting every 10 min to mix. The samples were then spun for 5 min at 12,000 rpm at 4 °C. The supernatants were transferred to fresh tubes and stored at −80. HEK293T and N2a cells were grown in RPMI 1640 medium (Invitrogen) with 10% heat-inactivated fetal calf serum (Biofluids; Rockville, MD). N2a cells were differentiated by allowing them to grow 3 days without changing the media leading to low serum-induced differentiation. Compound E (25Seiffert D. Bradley J.D. Rominger C.M. Rominger D.H. Yang F. Meredith Jr., J.E. Wang Q. Roach A.H. Thompson L.A. Spitz S.M. Higaki J.N. Prakash S.R. Combs A.P. Copeland R.A. Arneric S.P. Hartig P.R. Robertson D.W. Cordell B. Stern A.M. Olson R.E. Zaczek R. J. Biol. Chem. 2000; 275: 34086-34091Abstract Full Text Full Text PDF PubMed Scopus (303) Google Scholar) was used at a concentration of 3 × 10−8m and DAPT (26Dovey H.F. John V. Anderson J.P. Chen L.Z. de Saint Andrieu P. Fang L.Y. Freedman S.B. Folmer B. Goldbach E. Holsztynska E.J., Hu, K.L. Johnson-Wood K.L. Kennedy S.L. Kholodenko D. Knops J.E. Latimer L.H. Lee M. Liao Z. Lieberburg I.M. Motter R.N. Mutter L.C. Nietz J. Quinn K.P. Sacchi K.L. Seubert P.A. Shopp G.M. Thorsett E.D. Tung J.S., Wu, J.S. Yang S. Yin C.T. Schenk D.B. May P.C. Altstiel L.D. Bender M.H. Boggs L.N. Britton T.C. Clemens J.C. Czilli D.L. Dieckman-McGinty D.K. Droste J.J. Fuson K.S. Gitter B.D. Hyslop P.A. Johnstone E.M., Li, W.Y. Little S.P. Mabry T.E. Miller F.D. Audia J.E. J. Neurochem. 2001; 76: 173-181Crossref PubMed Scopus (797) Google Scholar) was used at a concentration of 2 × 10−6m, both were obtained from Dr. T. Golde. For luciferase assays, transfections were done using FuGENE 6 (Roche Molecular Biochemicals) with all samples done in 96-well plates in triplicate. For Fig. 2, b, c, e, andf and Fig. 3 a 0.06 μg of luciferase reporter plasmid, 0.01 μg of plasmid encoding β-galactosidase, 0.06 μg of YFP-Fe65/YFP, and 0.06 μg of DNA encoding Gal4 fusion proteins were used. For Fig. 2 d 0.03 μg of luciferase reporter plasmid, 0.01 μg of plasmid encoding β-galactosidase, 0.02 μg of YFP-Fe65, 0.04 μg of DNA encoding Gal4 fusion proteins, and 0.1 μg of PS/pcDNA3.1 were used. Cells were lysed after 24–48 h using reporter lysis 5× buffer (Promega, Madison, WI), and luciferase substrate (Promega) was used for the assay. Luciferase values were normalized using a β-galactosidase assay (Tropix, Bedford, Massachusetts).Figure 3Point mutation of the first tyrosine of the conserved YENPTY motif of the ALIDs abolish nearly completely their interaction with Fe65 with concomitant abolishment of transcriptional activation. a, when the wild type (wt) terminal 44 amino acids of APP, APLP1, or APLP2 (APPCT44, APLP1CT44 and APLP2CT44, respectively) is co-expressed along with Fe65 there is strong activation of the reporter gene. When the first tyrosine of the conserved YENPTY motif is mutated (Y-G), activation is almost completely abolished. Note that the activation values are greater than in Fig. 2 e because untagged Fe65 causes greater activation than Fe65 fused to YFP. b, lysates from 293 cells overexpressing FLAG-Fe65 were incubated with Sepharose beads bound to GST alone (−), the terminal 57 amino acids of APP (AID) or the terminal 57 amino acids of APLP1/2 wild type (ALID1/2 w.t.), or with the first tyrosine of the conserved YENPTY motif mutated to glycine (ALID1/2 Y-G). The total lysate is also indicated (T.L.). Fe65 only binds wild type proteins and not proteins with the first tyrosine of the YENPTY motif mutated.c, immunoprecipitation was carried out on lysates containing FLAG-Fe65 and either wild type (wt) or mutant (Y-G) ALIDs. Mutation of the first tyrosine of the conserved YENPTY motif causes a decrease in the amount of ALID immunoprecipitated by Fe65.View Large Image Figure ViewerDownload (PPT) Cell lysis, immunoprecipitation, and Western blotting were carried out as described previously (27Scheinfeld M.H. Roncarati R. Vito P. Lopez P.A. Abdallah M. D'Adamio L. J. Biol. Chem. 2002; 277: 3767-3775Abstract Full Text Full Text PDF PubMed Scopus (148) Google Scholar). APP specific antibody was from Zymed Laboratories Inc., and APLP1 and APLP2 antibodies were fromCalbiochem. Antibody B10, which cross-reacts with all APP family members and the PS1 carboxyl-terminal antibody was a gift of Dr. Bart De Strooper. Recombinant GST-fusion proteins were expressed inEscherichia coli strain BL21 (Stratagene) using the pGEX system (Amersham Biosciences). Proteins were purified using glutathione-Sepharose beads according to the manufacturer's instructions. Equal amounts of GST fusion proteins were incubated with 293 cell lysates containing FLAG-Fe65 for 1 h at room temperature. Following five washes the beads were boiled with loading buffer and separated by SDS-PAGE. Pulled down protein was detected by Western blotting with anti-FLAG antibody (Sigma). Cells were transfected using FuGENE 6 such that the ratio of APP/APLP1/APLP2 to YFP/GFP-Fe65 was 4:1. The Gal4BD epitope was detected with anti-Gal4BD monoclonal antibody (Santa Cruz) and Cy5-conjugated secondary antibody (Jackson ImmunoResearch, West Grove, PA). Endogenous APLP1 and APLP2 were detected using an antibody dilution of 1:2500. Overexpressed wild type APLP1 and APLP2 were detected with an antibody dilution of 1:10,000. Secondary antibody Alexa-594 conjugated to goat anti-rabbit IgG (Molecular Probes, Eugene, OR) was used at a dilution of 1:250. Nuclei were stained with DAPI (Sigma). Images were captured using a charge-coupled device camera mounted on an Olympus Provis AX70. Since APP signals by releasing the biologically active AID peptide, we sought to investigate whether APLP1 and APLP2 also release biologically active ALIDs due to processing by the γ-secretase. HEK293 and N2a neuroblastoma cells were incubated either in the absence or presence of the γ-secretase inhibitors Compound E or DAPT. Western blotting using antibodies specific for the carboxyl termini of APP, APLP1, or APLP2 showed that just as APP produces C83 and C99 intermediates, APLP1 and APLP2 also produce C83/C99-like intermediates that accumulate when the γ-secretase activity is inhibited (Fig.1 a). This suggested that APLP1 and APLP2 are substrates of the γ-secretase. It is important to note that we did not detect the endogenous ALID peptides, perhaps because they are unstable and have a short half-life similar to the AID peptide (28Cupers P. Orlans I. Craessaerts K. Annaert W. De Strooper B. J. Neurochem. 2001; 78: 1168-1178Crossref PubMed Scopus (217) Google Scholar). We next wanted to determine whether this γ-secretase processing of APLP1 and APLP2 is PS1- dependent. Fig. 1 bshows that in brain lysates from PS1 knockout mice there is accumulation of intermediates of APP, APLP1, and APLP2 processing but not in the lysates from wild type controls. These data indicate that all three members of the APP family are processed by the γ-secretase in a PS1- dependent manner. In the recent report by Cao and Sudhof (9Cao X. Sudhof T.C. Science. 2001; 293: 115-120Crossref PubMed Scopus (1054) Google Scholar), a Gal4BD (binding domain) fusion protein reporter system was used to show that following γ-secretase cleavage of APP, a multimolecular complex consisting of AID, Fe65, and Tip60 forms in the nucleus to cause reporter gene activation. Once we found that APLP1 and APLP2 also release fragments via the γ-secretase, we sought to determine whether APLP1 and APLP2 may also have a role in gene activation. Full-length constructs of APP, APLP1, and APLP2 with the Gal4BD inserted into the cytoplasmic domain (Fig.2 a) were co-transfected with a Gal4-luciferase reporter construct with or without Fe65. Fig.2 b shows that in two cell lines all the APP family members cause transcriptional activation of the reporter gene with an Fe65 dependence. We next wanted to determine whether this activation was due to release of an active fragment by the γ-secretase. Performance of a similar assay with and without γ-secretase inhibitors revealed (Fig.2 c) that reporter gene activation is indeed inhibited by γ-secretase inhibitors. Furthermore, to establish that APLP processing is PS1-dependent, reporter gene assays were carried out with co-transfection of the dominant negative PS1 mutant D385A. This aspartate mutation has been shown to replace endogenous Presenilins and abolish Presenilin-dependent γ-secretase activity (29Wolfe M.S. Xia W. Ostaszewski B.L. Diehl T.S. Kimberly W.T. Selkoe D.J. Nature. 1999; 398: 513-517Crossref PubMed Scopus (1688) Google Scholar, 30Kim S.H. Leem J.Y. Lah J.J. Slunt H.H. Levey A.I. Thinakaran G. Sisodia S.S. J. Biol. Chem. 2001; 276: 43343-43350Abstract Full Text Full Text PDF PubMed Scopus (90) Google Scholar). Fig. 2 d shows that for all APP family members, the PS1 D385A mutant depresses reporter gene activation. For APP, only the terminal 44 amino acids containing the Fe65 binding motif (31Fiore F. Zambrano N. Minopoli G. Donini V. Duilio A. Russo T. J. Biol. Chem. 1995; 270: 30853-30856Abstract Full Text Full Text PDF PubMed Scopus (269) Google Scholar) (YENPTY) is required for reporter gene activation (9Cao X. Sudhof T.C. Science. 2001; 293: 115-120Crossref PubMed Scopus (1054) Google Scholar). Since the second phosphotyrosine binding domain of Fe65 has been shown to bind directly with all APP family members via the conserved YENPTY motif (32Duilio A. Faraonio R. Minopoli G. Zambrano N. Russo T. Biochem. J. 1998; 330: 513-519Crossref PubMed Scopus (87) Google Scholar), we questioned whether a similar 44 amino acid stretch from APLP1 and APLP2 (APLP1CT44 and APLP2CT44, respectively, Fig. 2 a) could also cause reporter gene activation. Additionally, this 44-amino acid fragment, which comprises most of the ALID fragment released upon γ-secretase cleavage, should be the minimal sequence needed for transcriptional activation. In Fig. 2 e we show that both the terminal 44 amino acids of APLP1 and APLP2 are sufficient to cause activation of the reporter gene, whereas their deletion from full-length constructs (APLP1ΔCT and APLP2ΔCT, Fig. 2 a) abolishes the activation (Fig. 2 f). To further show that gene activation requires interaction between the APLPs and Fe65, we created point mutants of APLP1CT44 and APLP2CT44 with the first tyrosine in the conserved YENPTY motif mutated to glycine. As Fig.3 a shows, for these point mutants, reporter gene activation drops nearly to baseline. We then performed GST pull down assays on fusion proteins containing the carboxyl termini of APLP1 and APLP2 with the same tyrosine mutated. Fig. 3 b shows that although wild type APP, APLP1, and APLP2 bind Fe65, APLP1, and APLP2 containing the point mutation binds Fe65 much less efficiently. Last, to confirm this interaction in vivo, we performed immunoprecipitations. Fig. 3 c shows that the amount of ALID immunoprecipitated by Fe65 decreases when the first tyrosine of the YENPTY motif is mutated. Thus, these data show a correlation between gene activation and Fe65-APLP interaction. In addition to the AID-Fe65 complex-stimulating reporter gene activation, full-length APP interacts with Fe65 to regulate its distribution between the nucleus and cytoplasm (9Cao X. Sudhof T.C. Science. 2001; 293: 115-120Crossref PubMed Scopus (1054) Google Scholar, 33Minopoli G. de Candia P. Bonetti A. Faraonio R. Zambrano N. Russo T. J. Biol. Chem. 2001; 276: 6545-6550Abstract Full Text Full Text PDF PubMed Scopus (122) Google Scholar). We therefore wanted to determine whether full-length APLP1 and APLP2 are able to accomplish the same task. We first performed immunocytochemistry on differentiated N2a cells to determine the localization of endogenous APLP1 and APLP2. Fig.4 a shows that both APLP1 and APLP2 are present in the perinuclear area (agreeing with previous data from immunohistochemistry on brain sections) (34McNamara M.J. Ruff C.T. Wasco W. Tanzi R.E. Thinakaran G. Hyman B.T. Brain Res. 1998; 804: 45-51Crossref PubMed Scopus (33) Google Scholar) with some accumulation at the tips of neurites. To test whether the full-length APLP1 and APLP2 constructs used above for the luciferase assays were also able to regulate the distribution of Fe65, we expressed YFP-Fe65 along with either Gal4-tagged APLP1, APLP2, APLP1ΔCT or APLP2ΔCT and used immunocytochemistry to locate the Gal4BD epitope. We first wanted to confirm that the distribution of overexpressed APLP was similar to that of endogenous APLP. Fig. 4 b shows that the APLPs are located predominantly outside the nucleus similar to endogenous APLP. Fig. 4 b further shows that only the full-length proteins are able to exclude Fe65 from the nucleus; the deletion mutants, however, although they are localized mostly outside the nucleus, do not restrain Fe65 from moving into nucleus. Because of concerns that the Gal4-tagged APLPs would act different from the wild type proteins, we performed a similar analysis on wild type APP, APLP1, or APLP2 expressed along with GFP-Fe65. (Note that we changed Fe65 from the YFP to GFP fluorophore to allow compatibility with Alexa-594 used to stain the APLPs.) We found that Fe65 was restrained mostly to the cytoplasm when expressed along with APP, APLP1, or APLP2 but not with empty vector (data not shown). We speculated that the small amount of Fe65 present in the nucleus was due to Fe65 binding free AID or ALIDs. To eliminate processing nearly entirely so that no AID or ALID/Fe65 complexes would be released, we added 2 × 10−6m DAPT to the growth medium to inhibit γ-secretase processing. Fig. 4 c shows that nearly complete exclusion from the nucleus was obtained with APP, APLP1, or APLP2 but not with pcDNA3 empty vector. This data would be consistent with Fe65 binding the carboxyl terminal of the APLPs and it being constrained to the perinuclear space in the absence of γ-secretase processing. Upon γ-secretase cleavage, ALID probably along with bound Fe65 can travel to the nucleus to modulate gene expression. Finally, probably due to the instability of untagged AID and ALID peptides, we were unable to detect these endogenous fragments by Western blotting (Fig. 1). Still, to detect production of these fragments, we used the Gal4BD-tagged full-length constructs (so that tagged AID and ALIDs would be produced). Additionally, co-transfection of Fe65 was used to stabilize the ALID fragments as has been shown for AID (35Kimberly W.T. Zheng J.B. Guenette S.Y. Selkoe D.J. J. Biol. Chem. 2001; 276: 40288-40292Abstract Full Text Full Text PDF PubMed Scopus (397) Google Scholar). Fig. 5 shows that in the absence of γ-secretase inhibitors there is an extra lower molecular weight band present, presumably tagged ALID1 and ALID2. Furthermore, in the presence of γ-secretase inhibitors, production of this band is inhibited, and there is an increase in accumulation of the APP intermediates C99/C83 as well as the APLP (C99/C83-like) intermediate fragments. Fig. 5 also shows that the size of the extra band is larger than the size of the Gal4BD-tagged constructs, which contain only the terminal 44 amino acids (APPCT44, APLP1CT44, and APLP2CT44). The larger size indicates that the γ-cleavage product is larger than 44 amino acids, perhaps large enough to account for an intramembranous cleavage site. Indeed, it is plausible that the APLPs may be cleaved, just like APP, in two distinct positions in the intramembranous portion. One of these cleavage sites may correspond to that recently characterized by Gu et al. (20Gu Y. Misonou H. Sato T. Dohmae N. Takio K. Ihara Y. J. Biol. Chem. 2001; 276: 35235-35238Abstract Full Text Full Text PDF PubMed Scopus (269) Google Scholar). In this report we show that APLP1 and APLP2, the other two members of the APP gene family, are cleaved by a Presenilin-dependent γ-secretase. This cleavage, just as for APP, releases the intracellular peptides ALID1 and ALID2. We also show that APLP1 and APLP2 can bind Fe65 and sequester Fe65 in the cytoplasm. In addition, our data suggest that ALID molecules released after γ-secretase cleavage enter the nucleus and can presumably activate gene transcription in an Fe65-dependent manner. These data indicate that just as APP undergoes regulated intramembranous proteolysis (13Brown M.S., Ye, J. Rawson R.B. Goldstein J.L. Cell. 2000; 100: 391-398Abstract Full Text Full Text PDF PubMed Scopus (1149) Google Scholar) by the γ-secretase to release a biologically active intracellular peptide, so too do APLP1 and APLP2. This study provides the first functional aspect of APP found to be conserved throughout the APP family and may partially explain the functional redundancy that exists among them during development. The possible transcriptional role for AID previously described by Cao and Sudhof (9Cao X. Sudhof T.C. Science. 2001; 293: 115-120Crossref PubMed Scopus (1054) Google Scholar) and for the ALIDs described here is derived using an artificial system involving Gal4BD-fusion APP/APLPs proteins and a Gal4-dependent reporter gene. Discovering physiological, endogenous target genes for the three APP family members will be essential in determining whether the results from this reporter system represent a true physiologic phenomenon. The recent implication of AID in the regulation of an endogenous gene (KAI-1) (12Baek S.H. Ohgi K.A. Rose D.W. Koo E.H. Glass C.K. Rosenfeld M.G. Cell. 2002; 110: 55-67Abstract Full Text Full Text PDF PubMed Scopus (495) Google Scholar) suggests that the next step of our research should be to find genes that are regulated by the ALIDs. It is possible that AID, ALID1, and ALID2 may activate transcription of identical, partially overlapping, or distinct genes. The discovery of such genes may account for the different development phenotypes described by Heber et al. (24Heber S. Herms J. Gajic V. Hainfellner J. Aguzzi A. Rulicke T. von Kretzschmar H. von Koch C. Sisodia S. Tremml P. Lipp H.P. Wolfer D.P. Muller U. J. Neurosci. 2000; 20: 7951-7963Crossref PubMed Google Scholar). Finally, with the growing list of γ-secretase substrates including Notch (36De Strooper B. Annaert W. Cupers P. Saftig P. Craessaerts K. Mumm J.S. Schroeter E.H. Schrijvers V. Wolfe M.S. Ray W.J. Goate A. Kopan R. Nature. 1999; 398: 518-522Crossref PubMed Scopus (1799) Google Scholar), ErbB4 (37Ni C.Y. Murphy M.P. Golde T.E. Carpenter G. Science. 2001; 294: 2179-2181Crossref PubMed Scopus (756) Google Scholar, 38Lee H.J. Jung K.M. Huang Y.Z. Bennett L.B. Lee J.S. Mei L. Kim T.W. J. Biol. Chem. 2002; 277: 6318-6323Abstract Full Text Full Text PDF PubMed Scopus (263) Google Scholar), low density lipoprotein receptor-related protein (39May P. Reddy Y.K. Herz J. J. Biol. Chem. 2002; 277: 18736-18743Abstract Full Text Full Text PDF PubMed Scopus (260) Google Scholar), E-cadherin (40Marambaud P. Shioi J. Serban G. Georgakopoulos A. Sarner S. Nagy V. Baki L. Wen P. Efthimiopoulos S. Shao Z. Wisniewski T. Robakis N.K. EMBO J. 2002; 21: 1948-1956Crossref PubMed Scopus (623) Google Scholar), and now APLP1 and APLP2, there is the added complication of designing γ-secretase inhibitors for the prevention and treatment of AD. Further work will be necessary to localize the site of the γ-cleavage of APLP1 and APLP2 and to determine whether an APP-specific inhibitor found by Petit et al. (41Petit A. Bihel F. Alves da Costa C. Pourquie O. Checler F. Kraus J.L. Nat. Cell Biol. 2001; 3: 507-511Crossref PubMed Scopus (193) Google Scholar) inhibits APLP1 and/or APLP2 processing too. We gratefully thank Drs. T. Sudhof, T. Russo, D. J. Selkoe, B. De Strooper, S. S. Sisodia, and T. Golde for materials provided. For technical assistance we thank Yong Mei Zhao, and for helpful advice and discussions we thank the members of the D'Adamio laboratory. We express our appreciation to Dr. T. Dragic for extensive usage of equipment.
Referência(s)