EP2633066A1 - Methods for identifying notch-sparing gamma secretase inhibitors - Google Patents

Abstract

The invention herein provides a method for identifying a Notch-sparing γ-secretase inhibitor (NS-GSI) by evaluating ε- and γ-cleavage in a single APP-based substrate. The inventive assay utilizes ε-cleavage as surrogate for S3-cleavage of Notch and a substrate comprising the C99 fragment of APP fused to a Gal4/VP16 transcription factor to evaluate the inhibitory activity of compounds on ε and γ-cleavage simultaneously and for the identification of ε -sparing inhibitors that decrease Aβ production without affecting AICD production. In another embodiment, the APP-based substrate incorporates a mutated α-secretase cleavage site for enhanced Aβ production which allows for greater sensitivity of the assay.

Description

TITLE OF THE INVENTION

METHODS FOR IDENTIFYING NOTCH-SPARING GAMMA SECRETASE INHIBITORS

CROSS-REFERENCE TO RELATED APPLICATIONS

This application claims the benefit of priority from U.S. provisional patent application Serial Nos. 61/407,234 and 61/407,289, both of which were filed October 27, 2010, whose contents are incorporated herein by reference in their entireties.

FIELD OF THE INVENTION

The present invention relates to assays for identifying Notch-sparing gamma secretase inhibitors for use in the treatment of Alzheimer's disease.

BACKGROUND OF THE INVENTION

Gamma-secretase (γ-secretase) is a membrane protein complex consisting of at least four integral membrane proteins, i.e. presenilin (PS), nicastrin, APH-1 , and PEN-2. γ- secretase catalyzes the intra-membrane proteolysis of the β-amyloid precursor protein (APP) to generate amyloid β peptides (Αβ), which comprise senile plaques in the brains of Alzheimer's disease (AD) patients. Thus, γ-secretase inhibitors (GSIs) may provide effective therapy for AD (Wolfe, M.S., 2008, Curr. Top. Med. Chem. 8:2-8).

Development of GSIs has been complicated by the role of γ-secretase in the intra- membrane cleavage of other substrates (Lleo, A., 2008, Curr. Top. Med. Chem. 8:9-16). Among these substrates, Notch is the best characterized substrate. Notch undergoes sequential cleavages upon interaction with its ligands, Delta and Jagged, leading to signaling that plays an important role in cell fate determination during development and tissue homeostasis in the adult (Kopan et al., 1996, Proc. Natl. Acad. Sci. USA 93:1683-1688; Schroeter et al., 1998, Nature 393:382-386). There has been accumulating evidence that GSI treatment leads to a variety of toxicities related to Notch inhibition. For example, animals administered potent GSIs show goblet cell

hyperplasia in the intestinal epithelium and changes in the immune system such as decreased lymphocytes in the spleen and thymus (Hyde et al., 2006, J. Pharmacol. Exp. Ther. 319:1133- 1143; Milano et al., 2004, Toxicol. Sci. 82:341-358; Searfoss et al., 2003, J. Biol. Chem.

278:46107-46116; van Es et al, 2005, Nature 435:959-963). Furthermore, adverse events such as diarrhea and hair color change, which may be attributable to Notch inhibition, have been reported

. i . in the Phase 2 clinical trial of LY450139, a potent nonselective GSI (Fleisher et al., 2008, Arch. Neurol. 65:1031-1038).

These findings have led drug discovery efforts in the pharmaceutical industry to the development of "Notch-sparing" GSIs (NS-GSIs), compounds which suppress Αβ production in the brain while sparing Notch processing. To date, several classes of NS-GSIs have been reported (Imbimbo, B. P., 2008, Drug Discovery Today: Therapeutic Strategies 5: 169-175). Among them, begacestat (or GSI-953) and BMS-7081 3 are the most advanced examples, both of which are being tested in clinical trials (Imbimbo, 2008; Kreft et al., 2008, Bioorg. Med.

Chem. Lett. 18: 4232-4236; Martone et al., 2009, J. Pharmacol. Exp. Ther. 331 :598-608; Mayer et al._f 2008, J. Med. Chem. 51 :7348-7351). Begacestat and BMS-708163 have been reported to show a window between their potency for inhibition of APP and Notch cleavages (NICD/Αβ window). Details as to the mechanism for Notch-sparing activity of these compounds still remains unknown or has not been revealed. Recently, Eli Lilly and Company halted clinical studies of its GSI, semagacestat, which is reportedly not a "Notch-sparing" compound, because the compound did not slow disease progression and was associated with worsening of clinical measures of cognition and the ability to perform activities of daily living and with an increased risk of skin cancer (Press Release from Eli Lilly and Company on Tuesday 17, 2010; Lilly Halts IDENTITY Trials as Patients Worsen on Secretase Inhibitor, Alzheimer's Research Forum News, August 18, 2010).

Intra-membrane cleavage of APP and Notch by γ-secretase has several similarities. First, both require removal of the extracellular domain by other proteases, a process called ectodomain shedding. This shedding produces the membrane-tethered substrate for γ- secretase, namely, pCTF/C99 for APP and NEXT for Notch (Brown et al, 2000, Cell 100:391- 398). Second, both CTF/C99 and NEXT undergo cleavage by γ-secretase at two distinct sites, specifically, the ε- and γ-cleavage sites for APP and the S3- and S4-cleavage sites for Notch (Gu et al., 2001, J. Biol Chem. 276:35235-35238; Okochi et al, 2002, EMBO J. 20:5408-5416; Sastre et al, 2001, EMBO Rep. 2(9):835-841 ; Weidemann et al., 2002, Biochemistry 41 :2815- 2835). The first cleavage, e/S3-cleavage, occurs at the boundary of the transmembrane domain and the cytoplasm to liberate the intracellular domain (ICD), i.e. AICD and NICD for APP and Notch, respectively, which translocates into the nucleus to activate its target gene(s). The second cleavage, y/S4-cleavage, occurs in the middle of the transmembrane domain to liberate β peptides, i.e. Αβ and Νβ for APP and Notch, respectively. The current model suggests that s/S3- cleavage precedes y/S4-cleavage, although the precise mechanism of the sequential processing is still unclear (Chandu et al., 2006, J. Neurochem. 96:228-235; Funamoto et al., 2004, Biochemistry 43:13532-13540; Kakuda et al, 2006, J. Biol. Chem. 281 :14776-14786; Qi- Takahara et al., 2005, J. Neurosci. 25:436-445; Sato et al, 2005, FEBS Lett. 579:2907-2912; Takami et al. 2009, J. Neurosci. 29:13042-13052; Yagishita et al. 2008, J. Biol. Chem. 283:733- 738). Third, S4-cleavage of Notch has similar heterogeneity to that observed for γ-secretase cleavage of APP (Okochi et al, 2006, J. Biol. Chem. 281:7890-7898). γ-secretase cleavage yields mainly Αβ40, but several minor species, such as Αβ42 and Αβ38 are also observed. It is well known that mutations in the APP, PSEN1 or PSEN2 genes are linked to familial cases of AD (FAD) and affect the ratio of Αβ species (Golde et al., 2000, Biochim. Biophvs. Acta

1502: 172-187). Similarly, Okochi et al. have reported that the Νβ peptides generated by S4- cleavage are also heterogeneous and that the ratio of short and long Νβ peptides is affected by PSl FAD mutations (Okochi, 2006). Taken together, one skilled in the art would appreciate and recognize that APP and Notch share similar underlying mechanism(s) of intra-membrane cleavage.

As such, there exists a need to be able to screen for compounds for the treatment of AD that selectively inhibit the effect of γ-secretase cleavage without detrimentally altering Notch processing.

SUMMARY OF THE INVENTION

The invention herein is directed to a "dual cleavage" assay, i.e. the C99GVP assay, in which a compound's effect on γ- and ε-cleavage can be evaluated together to identify selective γ-secretase inhibitors (GSIs), and specifically inhibitors that affect the production of Αβ without altering the production and processing of Notch, i.e. Notch-sparing GSIs (NS-GSIs).

The C99GVP assay, in which ε-cleavage is used as a surrogate for S3-cleavage, uses an APP substrate, comprising the C99 fragment of APP fused to a Gal4 VP16 transcription factor, which allows for the identification of inhibitors that decrease Αβ production without affecting APP intracellular domain (AICD), as a surrogate for Notch intracellular domain (NICD), production.

The C99GVP assay may further comprise a substrate with mutations at the a-cleavage site, i.e.

HHQK. (SEQ ID NO: 1) to DHQE (SEQ ID NO: 2).

In one embodiment, the invention is a method for identifying a Notch-sparing γ- secretase inhibitor (NS-GSI) comprising: (a) contacting a cell, expressing an APP-based substrate having an ε- and a γ- cleavage site and a reporter gene under transcriptional control of an intracellular domain (ICD), with γ-secretase and a test compound under conditions that allow for γ-secretase activity;

(b) determining the potency of the test compound for inhibiting cleavage of Αβ and the ICD;

(c) determining the test compound window by dividing the potency of the ICD by the potency of Αβ; and

(d) determining the magnitude of difference between the test compound and a control by comparing the test compound window to a control compound window;

wherein a NS-GSI has a magnitude of difference that is the same or greater than the control.

In another embodiment of the invention, the APP-based substrate is C99GVP (SEQ ID NO: 14). In an alternative embodiment of the invention, the APP-based substrate is C99GVP DHQE (SEQ ID NO: 16). In still another embodiment of the invention, the magnitude of difference between the test compound window and a control compound window is 3- to 7-fold greater than the control.

BRIEF DESCRIPTION OF THE DRAWINGS

Figures 1 A- IE are representations of the C99GVP assay. Figure 1 A is a schematic representation of the C99GVP assay methodology in which the substrate, C99GVP, undergoes cleavages at both the ε- and γ-sites to produce the fragments AICD-GVP and Αβ, respectively. In turn, AICD-GVP induces the expression of firefly luciferase via the UAS promoter. Figure IB is a graphical representation of the luciferase signal for HEf 293 cells transiently transfected with the C99GVP and the pGL4.31 reporter plasmids and treated with DMSO (open column) or 10 μΜ DAPT (closed column). Luciferase and Αβ signals were measured from cell lysates and cultured media, respectively (mean ± S.D., n = 3). Figures 1C-1E are graphical representations of the effect of varying concentrations of L-685,458 (Figure 1C), DAPT (Figure ID), and GSM-1 (Figure IE) on the production of AICD-GVP (□ open square), Αβ40 (■ closed square), and Αβ42 (■ closed square). Luciferase (AICD-GVP) activity and Αβ were measured with luciferase and Meso Scale Αβ assays, respectively. Data were represented as % of controls (DMSO - 100%, 10 μΜ DAPT = 0%). Figure 2 are representative structures of γ-secreatase inhibitors (GSIs) (L-685,458 and DAPT), γ-secretase modulators (GSMs) (GSM-1, Compound C, and Compound D), and Notch-sparing γ-secretase inhibitors (NS-GSIs) (begacestat, BMS708163, Compound A, and Compound B).

Figure 3 A and 3B are graphical representations of the effect of varying concentrations of begacestat (Figure 3 A) and BMS-708163 (Figure 3B) in the C99GVP assay. AICD-GVP (□ open square), Αβ40 (m closed square), and Αβ42 (■ closed square) were measured with luciferase and Meso Scale Αβ assays. Data were represented as % of controls (DMSO - 100%, 10 μΜ DAPT = 0%).

Figure 4 is a graphical representation of correlation of the potency of nineteen compounds tested in a HeLa Notch split luciferase and C99GVP assay in order to examine whether AICD-GVP production can be used as a surrogate for NICD production. IC50 values against AICD-GVP production in the C99GVP assay were significantly correlated with those against NICD production in the HeLa Notch split luciferase assay with an R.2 value of 0.78 (p < 0.01).

Figure 5A-5D are graphical representations of the effects of Compound A (Figure 5A), Compound B (Figure 5B), Compound C (Figure 5C), and Compound D (Figure 5D) at varying concentrations in the C99GVP assay. AICD-GVP (□ open square), Αβ40 (■ closed square), and Αβ42 (■ gray square) were measured with luciferase and Meso Scale Αβ assays. Data were represented as % of controls (DMSO = 100%, 10 μΜ DAPT - 0%).

Figures 6A-6F are graphical representations of the effect of the DHQE mutation in the C99GVP substrate. Figure 6 A represents the N-terminal sequences of C99GVP wild type (C99GVP/WT) (SEQ ID NO: 3) and C99GVP/DHQE (SEQ ID NO: 4) with the a-cleavage site mutation (DHQE)(SEQ ID NO: 2) (in bold). Residues with an underline represent the transmembrane domain. The arrowhead indicates the α-cleavage site. Figure 6B represents an immunoblot analysis of HEK293 cells transiently transfected with mock plasmids (-),

C99GVP/WT or C99GVP DHQE. Each transfection was examined in duplicate. Lysates were probed with anti-VP 16 or anti-actin antibody. "M" represents a molecular size marker. Figure 6C represents the effects of the DHQE mutation on relative signal intensity in luciferase and Meso Scale Αβ assays. HE 293 cells were transfected with C99GVP/WT or C99GVP DHQE and treated with DMSO or 10 μΜ DAPT. Relative signal intensity was represented as % signal of WT in each assay (mean ± S.D., n = 3) [***, p < 0.0001] versus WT in Student's t-test. Figures 6D-6F represent the effect of L-685,458 (Figure 6D)₅ BMS708163 (Figure 6E), and GSM-1 (Figure 6F) on the production of AICD-GVP (o open circle), Αβ40 (· closed circle), and Αβ42 (· gray circle). HEK293 cells were transiently co-transfected with C99GVP/WT or C99GVP/DHQE and pGL4.31 and treated with each compound at varying concentrations.

Luciferase (AICD-GVP) activity and Αβ were measured with luciferase and Meso Scale Αβ assays, respectively. Data are represented as % of controls (DMSO = 100%, 10 μΜ DAPT = 0%).

Figure 7 is a graphical representation of the correlation of the ε/γ windows for C99GVP/WT and C99GVP/DHQE. Eight compounds were tested in C99GVP/WT and

C99GVP/DHQE assays at a range of concentrations to determine the ε/γ window for each substrate. The ε/γ windows for the two substrates were significantly correlated, with an R² value of 0.94 (p < 0.01). The diagonal solid line represents identical ε/γ windows (y = x).

Figure 8 is a graphical representation of the increase in Αβ40 resulting from the DHQE (SEQ ID NO: 2) mutation introduced into an APP NFEV substrate. SH-SY5Y cells were transfected with APP NFEV with or without the -site mutation, DHQE(SEQ ID NO: 2) or

HHQK (SEQ ID NO: 1), respectively. Secreted Αβ40 was measured with Meso Scale Αβ assay. *, p < 0.05 (mean ± S.D., n =3).

DETAILED DESCRIPTION OF THE INVENTION

Inhibiting γ-secretase activity is a promising strategy to intervene in the pathogenesis of Alzheimer's disease. One of the major challenges in developing γ-secretase inhibitors (GSIs) is adverse events associated with Notch signaling inhibition, including gastrointestinal and immunological abnormalities. Recently, several classes of Notch-sparing GSIs (NS-GSIs) have been reported, which inhibit Αβ production but do not impair Notch signaling, i.e. Notch-sparing. However, the mechanism(s) of this Notch-sparing activity remains unknown. Specifically, it is unclear whether Notch-sparing activity derives from selectivity between substrates (i.e., β-amyloid precursor protein (APP) versus Notch) or selectivity between cleavage sites (s/S3- versus γ/84-cleavages).

Applicants herein have developed a dual cleavage assay system using an APP- based substrate fused with the Gal4/VP16 transcription factor (C99GVP) to evaluate inhibitory activity of compounds on ε- and γ-cleavages simultaneously. In this assay, two known NS-GSIs, begacestat and BMS708163, showed greater selectivity between ε- and γ-cleavage sites, as compared to a transition-state analog inhibitor (GSI), L-685,458, suggesting that Notch-sparing activity is derived, at least partly, from cleavage site selectivity. Screening using the C99GVP assay identified compounds with high cleavage site selectivity, one of which (Compound A) has a large window in vivo, making it suitable for use as a therapeutic agent. Applicants also found that specific mutations at the a-cleavage site of APP, i.e. HHQK (SEQ ID NO: 1) to DHQE (SEQ ID NO: 2), increased Αβ production without affecting AICD-GVP production as introduced in the C99GVP substrate, suggesting that the extracellular region of APP plays a role in regulating γ- versus ε-cleavage of APP, which in turn indicates the possibility that molecules modulating the effect of this region may function as novel NS-GSIs/e-sparing-GSIs. Moreover, The DHQE mutation affects the potency of GSIs (L-685,458 and DAPT) and NS-GSIs

(begacestat, BMS708163, Compound A, and Compound B) for Αβ, but does not alter the potency for AICD-GVP. When normalized to the effect of the non-selective GSI, L-685,458, ε/γ windows measured from C99GVP/DHQE (SEQ ID NO. 16) are correlated well with those measured from C99GVP/WT (SEQ ID NO. 16), suggesting that the effect of the DHQE mutation on compound potency is independent from the mechanism of ε-sparing activity. Therefore, C99GVP/DHQE may provide a more sensitive and reliable assay with higher signal intensity of Αβ for identifying Notch-Sparing GSIs as compared to C99GVP/WT.

The term "C99" or "βΟΤΡ" as used herein refers to the C-terminal fragment resulting from cleavage of the β-amyloid precursor protein (APP) by β-secretase, corresponding to amino acid positions 597-695 (numbered relative to APP695). Conversely, the term "βΑΡΡβ" as used herein refers to the N-terminal fragment resulting from cleavage of the β-amyloid precursor protein (APP) by β-secretase, corresponding to amino acid positions 1-596 (numbered relative to APP695).

The term "C99GVP" or "C99GVP substrate" as used herein refers to a substrate, having the nucleic acid sequence of SEQ ID NO: 13 and the amino acid sequence of SEQ ID NO: 14, comprising the C99 fragment of APP, which incorporates a Gal4 DNA-binding VP16 tranactivation (GVP) domain (SEQ ID NO: 9). Following cleavage at the γ-secretase site, the intracellular domain containing the GVP domain translocates to the nucleus and initiates expression of luciferase reporter gene through the UAS promoter sequence.

The term "C99GVP assay" as used herein refers to the assay of Example 2, 4, and 5 that comprise assays determining the potency of compounds for inhibiting the production of Αβ and AICD-GVP by measuring the effect on γ- and ε-secretase cleavage sites. The term " -secretase cleavage site" as used herein refers to the site at which APP is proteolytically cleaved by the enzyme a-secretase and corresponds to amino acid positions 612-613 (numbered relative to APP695).

The term "β-secretase cleavage site" as used herein refers to the site at which APP is proteolytically cleaved by the enzyme β-secretase and corresponds to amino acid positions 596-597 (numbered relative to APP695).

The term "γ-secretase cleavage site" as used herein refers to the site at which APP is proteolytically cleaved by the enzyme γ-secretase and corresponds to amino acid positions 635- 640 (numbered relative to APP695).

The term "ε-secretase cleavage site" as used herein refers to the site at which APP is proteolytically cleaved by the enzyme ε-secretase and corresponds to amino acid positions 646-647 (numbered relative to APP695).

The term "γ-secretase inhibitor" or "GSI" as used herein refers to compounds that inhibit Αβ production, AICD production, and NICD production.

The term "γ-secretase modulator" or "GSM" as used herein refers to compounds that inhibit Αβ42 production more than Αβ40 production without affecting AICD or NICD production.

The term "non-selective γ-secretase inhibitor" or "non-selective GSI" as used herein refers to compounds that inhibit Αβ production, AICD production and NICD production, i.e. γ-, ε- and S3-cleavage, with the same or similar potency.

The term "Notch- Sparing γ-secretase inhibitor" or "NS-GSI" as used herein refers to compounds that inhibit Αβ production, i.e. γ-cleavage, more than NICD production, i.e. S3- cleavage. Such compounds have a NICD/ Αβ window that is greater than known or reference non-selective GSIs.

The term "ε-sparing γ-secretase inhibitor " or " ε-sparing-GSI" as used herein refers to compounds that inhibit Αβ production, i.e. γ-cleavage, more than AICD production, i.e. ε-cleavage. Such compounds have a window that is greater than known or reference nonselective GSIs.

The term "window" or "ε/γ window" as used herein refers to a ratio of the inhibitory potency of a compound for AICD or NICD as reflected by their IC50 values relative to Αβ40/42 ("Αβ") production, for example, by dividing the AICD-GVP IC50 value by the Αβ40 IC50 value. The window is indicative of a compound's ability to decrease or inhibit Αβ production relative without affecting either AICD or NICD production.

The term "potency" as used herein refers to a compound's ability to inhibit the production of a specific fragment of APP or Notch as measured by its IC50 value. For example, a compound that is a NS-GSI has increased potency for Αβ, but did not alter the potency for NICD.

Reducing Αβ by inhibiting γ-secretase activity is an important approach for therapeutic intervention in AD. Because γ-secretase is responsible for the cleavages of other physiologically important substrates such as Notch, a major focus in the pharmaceutical industry has been the development of Notch-sparing γ-secretase inhibitors (NS-GSIs), which inhibit Αβ production while sparing NICD production and subsequent Notch signaling (Wolfe, M.S., 2008, Curr. Top. Med. Chem. 8:2-8). Although NS-GSIs have been reported (Imbimo, B.P., 2008, Drug Discovery Today: Therapeutic Strategies 5: 169-175), the underlying mechanism of Notch- sparing activity still remains unknown. To investigate the mechanism of NS-GSIs Applicants developed an assay using a combination substrate that enables the analysis of both the ε- and y- cleavage in a single cell that allows for more accurate determination of selectivity. In this assay system, a C99GVP substrate transiently overexpressed in HEK293 cells was readily processed and γ-secretase-dependent production of Αβ and AICD-GVP were observed (Figures 1A-1E), consistent with similar systems previously reported for monitoring γ-secretase cleavage

(Karlstom et al., 2002, J. Biol. Chem. 277:6763-6766; Ren et al, 2007, J. Biol. Chem.

282;3535O-35360). Using this C99GVP dual cleavage assay Applicants demonstrate herein that (i) ε-cleavage of APP can be a surrogate for the S3-cleavage of Notch, (ii) selectivity between ε- and γ-secretase cleavage sites serves as a mechanism for Notch-sparing activity, and (iii) the DHQE mutation at the a-cleavage site plays a role in regulating ε- and γ-secretase cleavages.

There has been accumulating evidence showing the similarity between APP and

Notch, with respect to both cleavage of the substrate and the fragments generated. Both APP and Notch undergo sequential dual intramembrane cleavage at the e/S3- and y/S4-sites to produce an ICD and a secreted β peptide, respectively (Okochi et al., 2002, EMBO J. 21, 5408-5416;

Chandu et al, 2006, J. Neurochem. 96: 228-235; Qi-Takahara et al., 2005, J. Neurosci 25:436- 445). Applicants' observation that IC50 values for AICD-GVP in the C99GVP assay showed a strong correlation with IC50 values for NICD in the HeLa Notch split luciferase assay (Figure 4) corroborates the notion that APP and Notch share the same underlying mechanism of intramembrane cleavages and that the ε-cleavage of APP can be a surrogate of S3 -cleavage of Notch. In addition, it has been reported that ICDs of other γ-secretase substrates also translocate into the nucleus and regulate transcription of target genes, although the physiological role of these ICDs remains to be elucidated (Lleo, A., 2008, Curr. Top. Med. Chem, 8:9-16). For example, CD44 undergoes cleavages by γ-secretase, and the released CD44-ICD translocates into the nucleus and activates transcription of a subset of genes including CD44 itself (Okamoto et ah, J. Cell Biol. 155:755-762). As such, the inventive C99GVP dual cleavage assay can be used to identify compounds sparing not only ε-cleavage of APP but also the ICD-producing cleavages of other substrates, which could consequently lower the risk of adverse events.

Based on current knowledge on intramembrane cleavage by γ-secretase, one can hypothesize two not mutually exclusive mechanisms by which Notch-sparing GSIs exert Notch- sparing activity: substrate selectivity and cleavage site selectivity. In the former mechanism, a Notch-sparing GSI selectively inhibits APP cleavages, ε- and γ-cleavages, but does not affect Notch cleavages, S3- and S4-cleavages. In the latter mechanism, a Notch-sparing GSI selectively inhibits β peptide-generating cleavages, γ- and S4-cleavages, but does not affect ICD-generating cleavages, ε- and S3-cleavages. Simultaneous monitoring of ε- and γ-cleavage using the C99GVP assay enabled Applicants to investigate the ε/γ windows for various GSIs, NS-GSIs and GSMs and to examine the latter hypothesis, i.e. whether a compound had selectivity between ε~ and γ-cleavages. A transition-state analog GSI, L-685,458, inhibited Αβ and AICD-GVP production with similar potency in the C99GVP assay (Figure 1 and Table 1), consistent with previous reports that L-685,458 binds to the catalytic center of the γ-secretase complex and inhibits all cleavage non-selectively (Li et a., 2000, Nature 405:689-694; Wrigley et al, 2004, X, Neurochem. 90:1312-1320). Making the assumption that L-685,458 inhibits ε- and γ-cleavages equally, ε/γ window of each compound were normalized to that of L-685,458 for comparing "ε- sparing activities". Another non-selective GSI, N-[N-(3,5-di-fluorphenacetyl)-L-alanyl]-(S)- phenylglycine t-butyl ester (DAPT) (Dovey et al., 2001) also inhibited Αβ and AICD-GVP production with similar potency in the C99GVP assay (Figure I , Table 1). The normalized ε/γ window of DAPT was 0.9-fold (Table 1), suggesting non-selective inhibition of ε- and γ- cleavages by DAPT. The Notch-sparing activity of begacestat and BMS708163, two known NS- GSIs (Imbimbo, B. P., 2008, Drug Discovery Today: Therap. Strategies 5: 169-175; reft et at,

2008, Bioorg. Med. Chem. Lett. 18: 4232-4236; Martone et al., 2009, J. Pharmacol. Exp. Ther.

331 :598-608; Mayer et al., 2008, J. Med. Chem. 51 :7348-7351), was confirmed in the assays herein. The Notch/A 40 windows of begacestat and BMS7081 3 were 517-fold and 739-fold, respectively (Table 2) in the Notch/Αβ assay, while in the C99GVP assay, both NS-GSIs showed preferential inhibition of Αβ40 and Αβ42 over AICD-GVP with ε/γ windows of 5.8- and 14-fold, respectively (Figure 3 and Table 1). When normalized to the effect of the non-selective GSI, L- 685,458, begacestat and BMS-708163 showed a 3.1- and 7.3-fold normalized ε/γ window, respectively, both of which were greater than normalized ε/γ windows of two reference nonselective GSIs, L-685,458 and DAPT, demonstrating that the C99GVP assay can differentiate the effect of non-selective GSIs and NS-GSIs on ε- and γ-cleavages. Taken together, these results suggest that these NS-GSIs have selectivity between ε- and γ-cleavage sites and that cleavage site selectivity (i.e. ε-sparing activity) could be one of the underlying mechanisms of their Notch- sparing activity.

The ability to identify Notch-sparing GSIs in the assays herein was also demonstrated by the results with Compound A and Compound B, both of which showed

Notch/Αβ windows in Notch/Αβ assays (Table 1) and ε/γ windows in the C99GVP assay (Table 2) that were comparable to or exceeded those for begacestat and BMS708163, which were statistically different from those for the non-selective GSI, L-685,458. Moreover, Compound A has been reported to reduce brain Αβ without showing Notch-related toxicity in rodents (Best et al., 2007, J. Pharmacol. Exp. Ther. 320:552-558). Taken together, these results suggest that the Notch/Αβ window in the Notch/Αβ assay is equivalent to the ε/γ window in the C99GVP assay, and that selectivity between ε- and γ-cleavage sites is one of the mechanisms through which NS- GSIs exhibits their Notch-sparing activity. Another class of molecules affecting γ-secretase activity, GSMs, lower Αβ production, specifically Αβ42 production, without affecting Notch cleavage. Accumulating evidence suggests that the mechanism by which GSMs modulate γ~ secretase activity without affecting Notch is different from that of NS-GSIs (Tomita, 2009, Expert Rev. Neurother. 9:661-679). For example, NS-GSIs inhibit γ-secretase cleavage, while GSMs shift γ-secretase cleavage sites resulting in lowering of specific Αβ species (Weggen, et al., 2001, Nature 414:212-216; ounnas et al., 2010, Neuron 67:769-780). Because it still remains unknown which of the two mechanisms, NS-GSIs and GSMs, provides effective therapy for AD in the clinic, it is important to differentiate distinct mechanisms of action of compounds, i.e., GSIs, Notch-sparing GSIs and GSMs, in drug discovery for AD. Applicants herein were able to successfully identified novel GSMs, Compound C and Compound D, which preferentially inhibited Αβ42 over Αβ40 with no or little effect on AICD-GVP production (Figure 5), demonstrating that the C99GVP assay can be used to evaluate and differentiate compounds with distinct mechanisms of action, i.e., GSIs, Notch-sparing GSIs, and GSMs. Moreover, Applicants herein found that the use of the DHQE mutation at the a- cleavage site of APP promoted γ-cleavage without affecting ε-cleavage in the C99GVP substrate (Figure 6C). This increase in Αβ was not specific to C99GVP and HEK293 cells in that

Applicants observed similar increases in Αβ production in experiments using an APP/NEFV substrate and SH-SY5Y cells (Figure 8). The APP/NEFV substrate is an APP substrate in which the wild type β-secretase cleavage site, KMDA (SEQ ID NO: 5), positions 595-598 in APP695, have been replaced with the variant cleavage site, NFEV (SEQ ID NO: 6). Immunoblot analysis using C99GVP/WT and C99GVP/DHQE showed similar levels of expression, suggesting that differences in transcription rates or stability were not the cause of the Αβ-specific increase (Figure 6B). Considering the current model of γ-secretase cleavage, in which ε-cleavage precedes γ-cleavage, one possible mechanism of the Αβ-specific increase is that the DHQE mutation may facilitate the transition from the ε-site to the γ-site. Moreover, two recent reports have suggested that the extracellular domain of the βΟΤΡ 099 fragment of APP may have an affect on ε- and γ-cleavages (Ren et al, 2007, J. Biol. Chem. 282:35350-35360; Tian et al., 2010, Nat. Struct. Mol. Biol. 17: 151-158). Ren et al . have reported that the luminal juxtamembrane domain of APP plays an important regulatory role in differentiating γ- and ε-cleavages by showing a reduction in the production of Αβ, but not AICD, upon site-directed mutation at K28 and S26 (Αβ numbering). Tian et al. identified the region of Αβ corresponding to Αβ 17-23 as a substrate inhibitory domain that suppresses Αβ production more so than AICD and NICD production. Further, it has been reported that mutations in APP can alter the conformation of PS1 and the resulting interaction of the substrate and the enzyme (Herl et al., 2009, Mol. Cell Neurosci. 41: 166-174; Tesco et al., 2005, J. Neurochem. 95:446-456). Thus, without wishing to be bound by any theory, Applicants believe that the DHQE mutation altered the structure around the active site upon binding to the γ-secretase complex, which in turn facilitated γ-cleavage sequentially after ε-cleavage. It should be noted that the normalized ε/γ windows of the ε-sparing GSIs were not affected by the DHQE mutation (Figure 7), suggesting that the mechanism by which the DHQE affected γ- and ε-cleavages was independent from that of ε-sparing GSIs.

In summary, Applicants have shown herein that ε-cleavage and γ-cleavage by γ- secretase are distinct and separate processes, which can be affected by specific compounds (for example, ε-sparing GSIs) or by a mutation in the extracellular domain of a substrate (i.e., the DHQE mutation). As such, small molecule compounds or peptides targeting the extracellular domain adjacent to the -cleavage site of APP might provide a new class of therapeutics for selective γ-secretase inhibition. C99GVP dual cleavage assay

The invention herein comprises a C99GVP dual cleavage assay that uses a single, combination substrate that enables analysis of both the ε- and γ- cleavage site simultaneously in a single cell for more accurate determination of cleavage site selectivity. Previous methods for monitoring s/S3- and γ/84-cleavage have done so through the use of two separate substrates for APP and Notch. Such assays cannot differentiate two mechanisms by which Notch-sparing activity is achieved, i.e. substrate selectivity and cleavage site selectivity, and cannot determine the exact window for a given compound between the two substrates because the compound's potency is affected by the expression level of the substrate. The use of a single, combination substrate that evaluates both cleavage sites eliminates the non-specificity from the dual substrates and enables a higher throughput.

This inventive assay uses an APP-based substrate, C99GVP (SEQ ID NO: 13 and SEQ ID NO: 14), comprising the N-terminal signal sequence, pCTF/C99, and the Gal4/VP16 transcription factor sequence inserted at the C-terminus of the transmembrane domain of

PCTF/C99 (Figure 1 A). Cleavage of C99GVP by γ-secretase yields an Αβ peptide and an APP- ICD with the GVP domain (AICD-GVP). The AICD-GVP fragment can in turn activate a luciferase reporter under the control of a Upstream Activation Sequence (UAS) promoter ( arlstrom et al., 2002, J. Biol. Chem. 277:6763-6766; Ren et al._; 2007, J. Biol. Chem.

282:35350-35360).

To evaluate the utility of the C99GVP substrate, Applicants transiently transfected HEK293 cells with the C99GVP and a pGL4.31 reporter plasmid, and treated the cells with DMSO or 10 μΜ DAPT, a potent GSI (Dovey et al., 2001, J. Neurochem. 76:173-181). As shown in Figure IB, transient co-transfection resulted in robust luciferase and Αβ40/42 signals (Figure IB). Treatment with DAPT abolished these signals, indicating that C99GVP is a functional substrate of γ-secretase, resulting in the production of Αβ peptides and AICD-GVP. The signal-to-background ratios of AICD-GVP, Αβ40, and Αβ42 were 13.6, 7.6, and 2.5-fold, respectively. Applicants then compared the Αβ- and AICD-G VP-lowering activity of two reference GSIs, L-685,458 and DAPT, which have been well characterized and known to inhibit γ-secretase activity non-selectively (Dovey, 2001 ; Shearman et al., Biochemistry 39:8698-8704). As determined by the luciferase and Αβ assays that follow, L-685,458 inhibited both Αβ and AICD-GVP production from C99GVP at similar potency. As shown in Table 1 below, the IC50 values for AICD-GVP, Αβ40, and Αβ42 were 36.4 nM, 19.7 nM, and 29.1 nM, respectively (Figure 1C, Table 1), consistent with the notion that L-685,458 targets the active site of the γ- secretase complex (Li et al., 2000, Nature 405:689-694). Table 1

IC₅0 (nM) normalized ε/γ window

AICD-GVP Αβ40 Αβ42 ε/γ window

L-685,458 36.4 ± 2.0 19.7 ± 2.0 29.1 ± 2.9 1.9 1.0

316 ± 17 196 ± 16 283 ± 27 1.6 0.9

GSM-l > 10000 >10000 108 ± 18 - -

Begacestat 201 ± 15 34.9 ± 4.1 71.4 ± 21.1 5.8 3.1⁺

BMS-708163 41.9 ± 5.7 3.10 ± 0.46 5.26 ± 1.03 14 7 3**

Compound A 473 ± 37 15.0 ± 3.5 21.6 ± 5.1 32 17*

Compound B 206 ± 22 8.89 ± 0.81 16.1 ± 4.5 23 13*

Compound C >10000 89.5 ± 12.4 5.23 ± 0.74 - -

Compound D > 10000 56.9 ± 6.5 9.01 ± 2.03 - -

Potencies of compounds are from n = 3-7 independent experiments. Values are reported as mean ± S.E. ⁺, p = 0.06 versus L-685,458; *, p < 0.05 versus L-685,458; **, p < 0.01 versus L-685,458 in Student's t-test. The ε/γ window in this assay, calculated by dividing the AICD-GVP IC50 value by the Αβ40 IC₅o value, was 1.9-fold. Making the assumption that L-685,458 binds to the active site and inhibits ε- and γ-cleavages equally, for comparison of the "ε-sparing activity", i.e. Notch- sparing activity achieved by cleavage site selectivity, the ε/γ window of each compound was normalized to that of L-685,458, that is, it was set equal to 1.9-fold. DAPT also inhibited Αβ and AICD-GVP production with similar potency. IC50 values for AICD-GVP, Αβ40, and Αβ42 were 316 nM, 196 nM, and 283 nM, respectively (Figure ID, Table 1). The normalized ε/γ window of DAPT was 0.9-fold (Table 1), suggesting that the compound is a non-selective inhibitor of ε- and γ-cleavage.

Applicants also examined the effect of an NSAID-like γ-secretase modulator (GSM), GSM-l , which has been reported to reduce Αβ42 while increasing Αβ38 in cultured cells (Page et al., J. Biol. Chem. 283:677-683). As shown in Figure IE, GSM-l inhibited Αβ42 with an IC50 of 108 nM and was much less potent against Αβ40 and AICD-GVP production from C99GVP (Figure IE). This was consistent with reported characteristics of GSMs (Weggen et al., Nature 414:212-216). Thus, these results suggest that the C99GVP substrate was a functional substrate of γ-secretase and that the C99GVP assay can be used to evaluate the effect of GSIs and GSMs on γ- and ε-cleavage.

Effects of known Notch-sparing compounds

Known NS-GSIs, such as, begacestat and BMS708163, both of which are currently undergoing clinical evaluation (Figure 2) (Imbimbo, B. P., 2008, Drug Discovery Today: Therapeutic Strategies 5:169-175; Kreft et al., 2008, Bioore. Med. Chem. Lett. 18:4232- 4236; Martone et al., 2009, J. Pharmacol. Exp. Ther. 331:598-608; Mayer et al., 2008, J. Med. Chem. 51 : 7348-7351) were used as reference compounds in the conventional Notch and Αβ assays, i.e. HeLa Notch split-luciferase and SH-SY5Y/SPA4CT assays, respectively (hereinafter 'Notch/ Αβ assays"). The inhibitory potency of theses compounds for NICD and Αβ40/42 generated in the HeLa Notch split luciferase and SH-SY5Y/SPA4CT Αβ assays, respectively, are shown in Table 2. Not unexpectedly, the two compounds showed selectivity between the two substrates. The Νοΐΰη/Αβ40 window, calculate by dividing the NICD IC50 value by the Αβ40 IC50 value, of begacestat and BMS708163 were 517-fold (Αβ40 IC50 - 1.6 nM; Notch IC50 = 833 nM) and 739-fold (Αβ40 IC50 = 0.08 nM; Notch IC50 = 59 nM), respectively.

Table 2

IC50 (nM) Ν Ί)/Αβ40

NICD Αβ40 Αβ42 window

L-685,458 24.7 ± 2.4 6.89 ± 1.30 19.48 ± 2.84 3.6

DAPT 39.1 ± 4.4 19.4 ± 2.6 18.8 ± 3.0 2.0

GSM-1 >10000 >5000 208 ± 12 -

Begacestat 833 ± 155 1.61 ± 0.37 3.26 ± 0.51 517

BMS-708163 59.1 ± 6.6 0.08 ± 0.01 0.24 ± 0.03 739

Compound A 95.7 ± 6.6 0.14 ± 0.02 0.22 ± 0.01 683

Compound B 446 ± 69 0.14 ± 0.04 0.33 ± 0.08 3185

Compound C > 10000 101 ± 4 12.9 ± 1.2 -

Compound D > 10000 30.33 ± 0 3.36 ± 0 - Potencies of compounds are from n = 1-36 independent experiments. Values are reported as mean ± S.E.

Begacestat and BMS708163 were also evaluated in the C99GVP assay in order to examine their ε/γ windows. Begacestat and BMS708163 inhibited Αβ40 and Αβ42 more potently than AICD-GVP (Figure 3 A and B and Table 1) and showed a larger ε/γ window compared to L-685,458. Normalized ε/γ windows were 3.1-fold for begacestat (P = 0.06) and 7.3-fold for BMS-708163 (P < 0.01). These results suggest that Notch-sparing activity of the two known NS-GSIs may be attributable, at least partly, to the selectivity between ε- and γ- cleavage sites, i.e. ε-sparing activity.

Evaluation of novel ε-sparing GSIs and GSMs

The results presented above demonstrate that the inventive C99GVP dual cleavage assay offers a good platform for differentiating distinct mechanisms of γ-secretase inhibition/modulation and that can be used to identify ε-sparing GSIs. To determine whether ε- sparing activity can be used as a surrogate for Notch-sparing activity, Applicants compared the potency of nineteen reference compounds from different structural classes to inhibit AICD-GVP and NICD production using the C99GVP and HeLa Notch split-luciferase assays, respectively. As shown in Figure 4, the IC50 values for AICD-GVP and NICD correlated well (R2 = 0.78, P<0.01) and each compound evaluated showed comparable potency in the two assays (Figure 4), suggesting that the inhibitory effect on NICD production (i.e., S3-cleavage) can be evaluated by monitoring the inhibitory potency on AICD-GVP production (i.e., ε-cleavage).

Putative ε-sparing compounds were evaluated using the inventive C99GVP assay. Compound A (Figure 2) is a potent GSI in conventional Notch/Αβ assays (Αβ40 IC50 in SH- S Y5 Y = 0.14 nM, Notch IC50 in HeLa = 96 nM) and has been reported to lower brain Αβ without showing Notch-related pathology in rodent models (Best et al., 2006, J. Pharmacol. Exp. Ther. 317:786-790; Best et al., 2007, J. Pharmacol. Exp. Ther. 320:552-558) (Table 2). In the inventive C99GVP dual cleavage assay, Compound A inhibited Αβ40 and Αβ42 production with an IC50 values of 15.0 nM and 21.6 nM, respectively, while showing less potency against AICD- GVP (IC50 value = 473 nM) (Figure 5A and Table 1). The normalized ε/γ window of Compound A was 17.0-fold, which is greater than that of begacestat and BMS-708163, and is consistent with in vivo Notch-sparing activity (Best et al., 2007). A novel pyrazole compound, Compound B (Figure 2), was found to be a potent GSI with Notch-sparing activity, with an IC50 value for Αβ40 and Notch of 0.14 nM and 447 nM, respectively, in the Notch Αβ assays (Table 2). In the C99GVP assay, Compound B inhibited Αβ40 and Αβ42 production with an IC50 value of 8.89 nM and 16.1 nM, respectively, while showing less potency against AICD-GVP (IC50 value = 206 nM) (Figure 5B and Table 1). The normalized ε/γ window was 13-fold, demonstrating that Compound B is a novel class of ε-sparing GSI. Similarly, Applicants evaluated the effect of GSMs with novel scaffolds, Compound C and Compound D (Figure 2), which have triazole and purine moieties, respectively. In the Notch Αβ assays, these GSMs lowered Αβ42 with higher potency than for Αβ40 and NICD (Table 2). In the C99GVP assay, the two GSMs also showed preferential inhibition of Αβ42 over Αβ40 (Κ¾ο for Αβ40 and Αβ42: 89.5 nM and 5.23 nM for Compound C; 56.9 nM and 9.01 nM for Compound D) (Figure 5C and 5D and Table 1). AICD- GVP production was not inhibited within the concentration range used, suggesting that the inventive C99GVP assay can be used to identify GSMs by measuring Αβ40 and 42 production. Taken together, these results demonstrate that the inventive C99GVP dual cleavage assay can be used to identify novel ε-sparing GSIs and GSMs.

Effect of a-secretase cleavage site mutation on γ-secretase activity

In another embodiment of the C99GVP dual cleavage assay, the substrate incorporates mutations at the α-secretase cleavage site that increase Αβ production and detection. Detection of Αβ is often hampered by insufficient production from the substrate, a problem that often required the use of sodium butyrate to augment expression levels of the transfected cD As.

However, sodium butyrate is not an optimal solution in that the compound can interfere with the test compound effect in screening. Applicants used an α-secretase cleavage site mutation,

DHQE (SEQ ID NO: 2), in place of the wild type α-secretase cleavage site, HHQK (SEQ ID NO: 1) increased Αβ production by 2-5 fold without affecting the ε/γ window for the compounds evaluated. This further confirmed the appropriateness of the proposed substrate for use in identifying Notch-sparing GSIs.

To further optimize the inventive C99GVP dual cleavage assay, specifically with respect to the signal intensity of Αβ, Applicants used an α-secretase cleavage site mutated derivative of the C99GVP substrate (SEQ ID NO: 13 and SEQ ID NO: 14), C99GVP/DHQE

(SEQ ID NO: 15 and SEQ ID NO: 16), in which the two amino acid residues adjacent to the a- cleavage site, HHQK (SEQ ID NO: 1), are mutated to DHQE (SEQ ID NO: 2) (Figure 6A). The DHQE mutations were similar to the murine APP mutations R609D and 612E which, when used in combination, have been shown to result in a less efficient substrate for -secretase (De Strooper et al., 1993, J. Cell Biol. 121 :291-304). Immunoblot analysis revealed that both substrates provided similar levels of expression with the plasmid amount used (7.7 μg) (Figure 6B). Secretion of Αβ40 and 42 was increased by the DHQE mutation by 2- and 3.5-fold, respectively (Figure 6C), as compared to C99GVP/WT, which was consistent with the results obtained from APP NFEV substrate in SH-SY5Y cells. In contrast, C99GVP/WT and

C99GVP/DHQE showed comparable levels of luciferase activity (Figure 6C).

When the a-secretase cleavage site mutations were introduced into a human APP/NFEV substrate (U.S. Patent No. 7,196,163, Majercak et al., 2006, Proc. Natl. Acad. Sci. USA 103:1 967-17972), in which the wild type β-secretase cleavage site, MDA (SEQ ID NO: 5), has been altered to a mutated β-secretase cleavage site, NFEV (SEQ ID NO: 6), secretion of Αβ40 from SH-SY5Y cells increased by 2-fold (Figure 8). These results suggest that the <x- secretase site DHQE mutation increases γ-cleavage with no or little effect on ε-cleavage.

Applicants next examined whether the DHQE mutation affected the potency of compounds with respect to ε- and γ-cleavages. HEK293 cells transiently transfected with the C99GVP/DHQE substrate (SEQ ID NO: 15 and SEQ ID NO: 16) were treated with L-685,458 at a range of concentrations and AICD-GVP and Αβ production were measured. As shown in Figure 6D, L-685,458 inhibited Αβ40 and Αβ42 more potently than AICD-GVP with IC50 values for AICD-GVP, Αβ40 and Αβ42 of 36.3 nM, 4.85 M, and 2.29 nM, respectively (Table 3). When compared with the potency observed with C99GVP/WT, L-685,458 exhibited a 5 to 10-fold increased potency against Αβ production as compared to that in the C99GVP/DHQE substrate, while showing similar potency against AICD-GVP. Accordingly, the ε/γ window for the α-secretase cleavage site mutated substrate was 7.5-fold, which is 4-fold larger than that observed with C99GVP/WT (p - 0.00008, compare Table 1 and Table 3). The same trend was observed for DAPT (Figure 6E) and the reference ε-sparing GSIs (Table 3). These results suggest that the a-site DHQE mutation increases the potency of GSIs and ε-sparing GSIs for Αβ with no or little effects on the potency for AICD-GVP. Moreover, the normalized ε/γ window of each compound was comparable to that observed with C99GVP/WT (Figure 7), suggesting that the effect of the DHQE mutation on compound potency is independent from the mechanism of ε- sparing activity.

Similarly, Applicants investigated the effect of the DHQE mutation on the potency of GSMs. Table 3 below shows the inhibitory potency and the ε/γ window for AICD- GVP, Αβ40 and Αβ40 generated from C99GVP DHQE for representative compounds. As expected, GSM-1 (Figure 6F)_; Compound C and Compound D showed little or no effect on AICD-GVP production from C99GVP/DHQE (Table 3). The inhibitory effect on Αβ production was also similar to those observed with C99GVP/WT, that is, all three GSMs preferentially inhibited Αβ42 over Αβ40, demonstrating that the DHQE mutation does not alter γ-secretase modulating activity of these compounds. It was notable, that unlike GSis and NS-GSIs, the IC50 value for Αβ40 and Αβ42 were almost identical to those observed with C99GVP/WT (compare Table 1 and Table 3). These results suggest that the DHQE mutation does not alter the effect of GSMs and that GSMs exert their modulating activity via a different mechanism from that of GSis and NS-GSIs.

Table 3

IC50 (nM) normalized ε/γ window

AICD-GVP Αβ40 Αβ42 ε/γ window

L-685,458 36.3 ± 3.5 4.85 ± 0.40 2.29 ± 0.27 7.5 1.0

DAPT 378 ± 24 49.5 ± 3.6 37 _*2^* r— 2-»3 7.6 1.0

GSM-1 >10000 > 10000 106 ± 9 - - begacestat 276 ± 23 9.82 ± 0.60 6.98 ± 0.69 28.2 3.8*

BMS-708163 38.4 ± 5.9 0.68 ± 0.11 0.51 ± 0.05 56.6 7.6*

Compound A 646 ± 72 2.01 ± 0.33 1.19 ± 0.21 322 43*

Compound B 272 ± 29 1.10 ± 0.16 0.71 ± 0.14 246 33**

Compound C >10000 50.1 ± 6.8 6.02 ± 0.28 -

Compound D >10000 49.9 ± 7.1 6.00 ± 0.41 - -

Potencies of compounds are from n = 3-7 independent experiments. Values are reported as mean

± S.E. *, p < 0.05 versus L-685,458; **, p < 0.01 versus L-685,458 in Student's t-test.

EXAMPLES

Example 1

Cell culture, plasmids, and antibodies

HEK293 cells were obtained from ATCC (Manassas, VA) and maintained in Dulbecco's Modified Eagle Medium (Invitrogen, Carlsbad, CA) containing 10% heat-inactivated fetal bovine serum (Sigma-Aldrich, St. Louis, MO). SH-SY5Y neuroblastoma cells stably overexpressing SPA4CT were maintained as previously described (Beher et al, 2001, J. Biol. Chem. 276:45394-45402). For the Notch split-luciferase assay, HeLa cells stably overexpressing Notchl ΔΕ and RBPJ fused to the N- and C-terminal fragments of firefly luciferase, respectively, were maintained in Dulbecco's Modified Eagle Medium containing 5% fetal bovine serum.

The C99GVP construct (SEQ ID NO: 13 and SEQ ID NO: 14), pCTF with the signal sequence of APP (SEQ ID NO: 17) at the N-terminus and the Gal4 VP 16 transcription factor sequence (SEQ ID NO: 9) inserted at the C-terminus of the transmembrane domain, was constructed from an APP-GVP construct in pcDNA3 (kindly provided by Dr. Gang Yu,

University of Texas Southwestern Medical Center, Dallas, TX) by the ExSite™ PCR-based site- directed mutagenesis kit (Stratagene, Agilent Technologies, Santa Clara, C A) using a pair of primers directed to the DNA sequence corresponding to the N terminus of the Αβ sequence and the DNA sequence corresponding to the signal sequence of APP (SEQ ID NOS: 9 and 10). The DHQE mutations were introduced into the C99GVP construct by the Quick Change™ XL system (Stratagene, Agilent Technologies, Santa Clara, CA) using a pair primers directed to the DNA sequence corresponding to the amino acid sequence around the a-secretase cleavage site (SEQ ID NOS: 11 AND 12).

The pGL4.31 reporter plasmid was purchased from Promega Corporation

(Madison, WI). Antibodies used were obtained from the following sources: mouse anti^-actin and rabbit anti-VP 16 (Abeam, Cambridge, MA); biotinylated 6E10 and mouse anti-Ap42, 12F4 (Signet™, Covance, Princeton, NJ); anti-mouse IgG (H+L) IRDye® 800 and anti-rabbit IgG (H+L) IRDye® 700 (LI-COR Biosciences, Lincoln, NE). Αηίΐ-Αβ40 G210 was made by Strategic Biosolutions (now SDIX, Newark, DE).

Representative γ-secretase compounds were purchased as follows. The γ- secretase inhibitors, L-685,458 and N-rN-(3,5-d9-fluorophenacetyl)-L-alanyl]-(S)-phenylglycine t-butyl ester (DAPT), were purchased from Calbiochem® (EMD4Biosciences, Merck KGaA,

Darmstadt, Germany). Other compounds were synthesized in-house at Merck Research

Laboratories (West Point, PA). Compound A is a Notch-sparing GSI with the cyclohexyl sulfone moiety. Compound B is a Notch-sparing GSI with the pyrazole moiety. Compound C is a GSM with the triazole moiety. Compound D is a GSM with the purine moiety. The chemical structure of the representative compounds used in this study is shown in Figure 2. Example 2

Transient transfection and compound treatment

Transfection was performed using FuGene® HD transfection reagent (Roche Applied Sciences, Indianapolis, IN) following the manufacturer's instructions. Briefly, cells were plated at the density of 15 xlO⁶ cells/plate in 10-cm dishes. The next day, cells were transfected with plasmids using the FuGene® HD transfection reagent (Roche Applied Sciences,

Indianapolis, IN). Twenty-four hours after transfection, cells were trypsinized and re-plated into 96-weIl plates at the density of 6000 cells/well. Compounds were added after the cells were attached to the plates and the cells were incubated for another twenty-four hours. Example 3

Immunoblot analysis

Forty-eight hours after transient transfection, cells were washed briefly with ice- cold phosphate buffered saline (PBS) and lysed in RIPA buffer (50 mM Tris-HCl pH 7.4, 150 mM NaCl, 1% NP-40, 0.5% sodium deoxycholate, 0.1% SDS) containing Complete Protease Inhibitor Cocktail (Roche Applied Sciences, Indianapolis, IN). All samples were solubilized at 4°C for 30 minutes and cleared by centrifugation. Aliquots of the supernatant were heated at 67°C for 30 minutes in NuPAGE® LDS Sample Buffer (Invitrogen, Carlsbad, CA) containing NuPAGE® Sample Reducing Agent (Invitrogen, Carlsbad, CA) and resolved on a NuPAGE® 4- 12% Bis-Tris gel (Invitrogen, Carlsbad, CA). The gels were then immunoblotted with appropriate antibodies and analyzed with the Odyssey® Infrared Imaging System (LI-COR Biosciences, Lincoln, NE).

Example 4

Luciferase assay

Twenty-four hours after compound treatment, the cells were harvested in 100 μΐ/well of reporter lysis buffer (Promega Corporation, Madison, WI). After adding 100 μΐ/well of the Steady-Glo® luciferase assay substrate (Promega, Corporation, Madison, WI), luciferase signals were detected with a microplate reader (EnVision®, Perkin Elmer, Waltham, MA). Example 5

Αβ quantitation in cultured media

Twenty-four hours after compound treatment, culture media was harvested. For detection of the Αβ peptides in culture media of SH-SY5 Y/SPA4CT cells, AlphaScreen® assays (Perkin Elmer, Waltham, MA) were performed using a variety of anti-Αβ antibodies and a microplate reader (EnVision®, Perkin Elmer, Waltham, MA). The G210 and 12F4 antibodies were conjugated to AlphaLlSA® acceptor beads (Perkin Elmer, Waltham, MA). Αβ40 was detected with biotinylated 6E10, G210-conjugated acceptor beads and streptavidin IgG donor beads (Perkin Elmer, Waltham, MA). Αβ42 was detected with biotinylated 6E10, 12F4- conjugated acceptor beads, and streptavidin IgG donor beads. For detection of the Αβ peptides in culture media of HE cells transiently transfected with C99GVP and pGL4.31 , Meso Scale Discovery® Assay Systems (Meso Scale Discovery, Gaithersburg, MD) were performed using biotinylated 4G8 and ruthenium-labeled G210 or 12F4 for Αβ40 or 42, respectively.

Example 6

HeLa Notch split-luciferase assay

In this assay, HeLa cells are made to express a Notch protein lacking its extracellular domain (NotchAE) fused to an N-terminal fragment of luciferase. The same cells also expressed a C-terminal fragment of luciferase fused to the immunoglobulin J kappa recombination signal sequence binding protein (RBP). Upon NotchAE cleavage by gamma secretase, a Notch intracellular domain (NICD)-N terminal luciferase protein is generated which translocates to the nucleus and binds the RBP-C terminal luciferase fusion, bringing two independently nonfunctional halves of luciferase together to form a functional luciferase enzyme.

The HeLa Notch split-luciferase cell line was plated in 384-well plates and treated with compound for 5 hours. For measurement of reconstituted luciferase activity, cells were incubated with britelite plus substrate (Perkin Elmer, Waltham, MA), and luminescence was measured with a ViewLux™ ultraHTS Microplate Imager (Perkin Elmer, Waltham, MA). Example 7

Data analysis

All experiments were repeated at least three times. All inhibitory effects of compounds were normalized to controls by setting the effects of DMSO and 10 μΜ DAPT as 100% and 0%, respectively. The potency of the reference compounds (i.e. IC50 value) was calculated using GraphPad Prism® version 5.02 software (GraphPad Software, Inc., San Diego, CA).

Claims

WHAT IS CLAIMED:

1. A method for identifying a Notch-sparing γ-secretase inhibitor (NS-GSI) comprising:

(a) contacting a cell, expressing an APP-based substrate having an ε- and a γ- cleavage site and a reporter gene under transcriptional control of an intracellular domain (ICD), with γ-secretase and a test compound under conditions that allow for γ-secretase activity;

(b) determining the potency of a test compound for inhibiting cleavage of Αβ and the ICD;

(c) determining the test compound window by dividing the potency of the

ICD by the potency of Αβ; and

(d) determining the magnitude of difference between the test compound and a control by comparing the test compound window to a control compound window;

wherein a NS-GSI has a magnitude of difference that is the same or greater than the control.

2. A method of claim 1 where the reporter gene is Gal4 VP 16 transcription factor (SEQ ID NO: 9).

3. A method of claim 2 where the APP-based substrate is C99GVP (SEQ ID

NO: 14).

4. A method of claim 2 where the APP-based substrate is C99GVP DHQE (SEQ ID NO: 16).

5. A method of claim 1 where the magnitude of difference between the test compound window and the control compound window is 3- to 7-fold greater than the control.