BRI2 inhibits amyloid beta-peptide precursor protein processing by interfering with the docking of secretases to the substrate

Abstract

Genetic alterations of amyloid beta-peptide (Abeta) production caused by mutations in the Abeta precursor protein (APP) cause familial Alzheimer's disease (AD). Mutations in BRI2, a gene of undefined function, are linked to familial British and Danish dementias, which are pathologically and clinically similar to Alzheimer's disease. We report that BRI2 is a physiological suppressor of Abeta production. BRI2 restrict docking of gamma-secretase to APP and access of alpha- and beta-secretases to their cleavage APP sequences. Alterations of BRI2 by gene targeting or transgenic expression regulate Abeta levels and AD pathology in mouse models of AD. Competitive inhibition of APP processing by BRI2 may provide a new approach to AD therapy and prevention.

Figure 1.

BRI2 is a physiological inhibitor of Aβ generation in cell lines. A, RNAi-mediated reduction of BRI2 expression with BRI2-specific short hairpin constructs sh5 and sh10 in HEK293APP cells. Interference was verified by real-time quantitative RT-PCR. Quantitative PCR analysis of BRI2 mRNA of HEK293APP cells transfected with control (m1) or BRI2 shRNAs (sh5 and sh10). The abundance of BRI2 mRNA is normalized by that of β-actin, and is displayed as a percentage of normalized BRI2 mRNA of untransfected cells. B, WB of the same samples indicates the reduction of both mature (m) and immature (im) BRI2 in cells transfected with BRI2 shRNAs. Also, whereas α-tubulin, mAPP, imAPP, and C83 levels are similar in all samples, sAPPα and sAPPβ are increased by sh5 and sh10. The numbers above the WBs indicate the relative amounts of proteins detected. The amount present in the M1 sample was given an arbitrary value of 1. C, D, Aβ ELISA from supernatants of HEK293APP cells shows that reducing BRI2 expression results in increased Aβ40 and Aβ42 production. Error bars indicate SEM.

Figure 2.

BRI2 is a physiological inhibitor of APP processing and Aβ generation in the mouse brain. A, WB analysis of brain membranes from Bri2^+/+, Bri2^+/−, and Bri2^−/− mice shows lack or reduced levels of Bri2 expression in Bri2^−/− and Bri2^+/− mice, respectively. Calnexin is used as a control to verify equal loading of protein samples. B, Analysis of brain membrane extracts from Bri2^−/− and APP^−/− mice. Total lysates were analyzed for Bri2 and APP expression (left panel). Brain lysates were immunoprecipitated with the αBRI2, αAPPct, and rabbit polyclonal (RP) control antibody (right panel). Precipitates were analyzed for APP and Bri2 proteins. Bri2^−/− and wild-type mice express equal amounts of APP. IP of endogenous APP with the αBRI2 antibody is specific because APP is precipitated in wild-type mice but neither APP^−/− nor Bri2^−/− mice. C, The brains of six 11-month-old male Bri2^−/− and wt littermates were analyzed for APP-derived fragments in the brain. Calnexin (Caln.) is used as a control to verify equal loading of protein samples. D, Quantification of the WBs shown in C. The increases in sAPPβ and -sAPPα observed in Bri2^−/− mice are statistically significant when compared with wt littermates. Error bars indicate SEM.

Figure 3.

Reducing BRI2 levels increases processing of a FAD APP mutant in vivo. Bri2^+/− mice were crossed to APP-PS1 tag APP and PS1 mutant transgenic mice to obtain Bri2^+/−/APP-PS1 and Bri2^+/+/APP-PS1 animals. WB analysis of postnuclear supernatants of 8-month-old (2 Bri2^+/−/APP-PS1 and 2 Bri2^+/+/APP-PS1) and 4-month-old (1 Bri2^+/−/APP-PS1 and 1 Bri2^+/+/APP-PS1) shows the following: total APP (mouse plus transgenic human APP), C83, C99, Nct, and human PS1DE9 levels are similar among mice in each age group. However, sAPPα is increased in the Bri2^+/−/APP-PS1 mice compared with the Bri2^+/+/APP-PS1 littermates. As for total Aβ, the peptide is only detectable in 8-month-old mice. Interestingly, total Aβ is increased in the two Bri2^+/−/APP-PS1. Aβ is detected by two different monoclonal antibodies, 6E10 and 4G8. The numbers above the WBs indicate the relative amounts of proteins detected. The amount present in the first Bri2^+/+ sample was given an arbitrary value of 1.

Figure 4.

BRI2 is a specific inhibitor of APP processing. A, γ30 cells, which express high levels of endogenous NCT and overexpress the γ-secretase components PS1, Pen2 (Flag-tagged), and Aph1, were transfected with Myc-tagged BRI2. IP of Flag-Pen2 isolates the γ-secretase complex but not BRI2. Conversely, precipitation of Myc-BRI2 did not pull-down any γ-secretase component. TL, Total lysates. B, HeLa cell were transfected with β-secretase (BACE1), with or without BRI2. BRI2 IP does not isolate BACE1. HeLa cells were transfected with APP (C), APLP1 (D), APLP2 (E), Notch1, or NotchΔE (F, G) alone or plus FLAG-tagged BRI2. Proteins were immunoprecipitated with a αFLAG antibody and eluted with the FLAG epitope peptide (IP). BRI2 interacts with both mAPP and C99 (C) but not APLP1 (or C-terminal APLP1β and APLP1α-stubs), APLP2 (or C-terminal APLP2β and APLP2α-stubs), Notch, or NotchΔE (D–F). BRI2 affects APP processing and generation of C99 (C), but it does not impact on the levels of APLP1 and APLP2 processing (D, E), because the levels of C-terminal stubs are not altered. G, BRI2 does not inhibit cleavage of NotchΔE by γ-secretases and generation of NICD, whereas, in the same transfected cells, it inhibits endogenous APP processing as demonstrated by increased C99 levels. The bottom two panels were probed with either αAPPct (fourth panel from the top), which detects both C83 and C99, or 6E10 (bottom panel), which reacts only with C99.

Figure 5.

BRI2 masks the α- and β-secretase cleavage sites of APP and competes with NCT for binding to C99. A, Schematic representation of APLP1/2-APP²⁰ APLP1/2-APP³³ chimeric molecule. The sequences on the far left represent the APP regions inserted in APLP molecules for the APLP1/2-APP²⁰ (top sequence) and APLP1/2-APP³³ (bottom sequence) chimeras. The α- and β-cleavage sites are indicated. TM, Transmembrane. B, HeLa cells were transfected with the indicated APLPs-APP constructs, with or without FLAG-BRI2. Total lysates (TL) and αFLAG-immunoprecipitates were analyzed by WB. mAPLP1-APP³³ and mAPLP2-APP³³ chimeras bind BRI2 and BRI2 increases the β-CTF produced by mAPLP1-APP³³ and mAPLP2-APP³³. C, Analysis of supernatants derived from transfected HeLa cells shows that BRI2 inhibits production of sAPLP1-APPα, sAPLP1-APPβ, and sAPLP2-APPα. D, HeLa cells were transfected with the APPΔC31 construct, with or without FLAG-BRI2. Total lysates (TL), αFLAG-immunoprecipitates, and cell culture supernatants were analyzed as in B and C. mAPPΔC31 binds BRI2, and BRI2 increases C99ΔC31, whereas it inhibits production of sAPPα (APPβ was not detected) (data not shown). E, HeLa cells were transfected with C99 and NCT. Two NCT bands are visible: the bottom band is the predominant specie present in total lysates, whereas the top band is visible only after longer exposure (second panel). IP of C99 isolates both NCT species but enriches for the higher molecular weight form. When BRI2 is coexpressed, C99 preferentially interacts with BRI2 over mNCT. F, Reverse IP shows that NCT coprecipitates C99. BRI2 expression reduces the amount of C99 bound to NCT, and NCT does not copurify BRI2. G, EndoH (H) and PNGase F (F) digestion of C99 immunoprecipitates shows that the top NCT form represents mature NCT because it is resistant to endoH digestion but sensitive to PNGase F, whereas the bottom molecular species is immature NCT. Again, expression of BRI2 inhibits the interaction between C99 and mNCT.

Figure 6.

BRI2 overexpression inhibits APP processing in normal animals and APP mutant transgenic mice. A, WBs of two wild-type animals and the progeny of lines BRI2-8.4, BRI2-8.5, BRI2, and ABri (Pickford et al., 2006) with the αBRI2 antibody indicate that the BRI2 protein is overexpressed in tgBRI2 animals. B, WBs of five BRI2-8.4 transgenic mice and wild-type littermates (mice were all 2-month-old males) show that (1) overexpression of BRI2 does not appreciably affect the total levels of APP or the mAPP:imAPP ratio in the brain; (2) the levels of C83 are not consistently changed (the levels of C99/89 are below detection); (3) BRI2 overexpression reduces the levels of sAPPβ and sAPPβ-sAPPα (total sAPP, detected by 22C11 in the soluble protein fraction, which is deprived of full-length APP). Calnexin is used as an internal control for protein loading. C, Quantification of the WBs shown in B. The decreases in sAPPβ and total-sAPP (tsAPP) as well as the increase in BRI2 observed in tgBRI2-8.4 mice are statistically significant when compared with wt littermates. Cal., Calnexin. Error bars indicate SEM.

Figure 7.

BRI2 reduces AD pathology in APP mutant transgenic mice. Cortical sections of 6-month-old mice stained with αAβ 6E10 show a reduction in amyloid pathology in BRI2 transgenic-CRND8 mice.