Amyloid precursor protein selective gamma-secretase inhibitors for treatment of Alzheimer's disease

Abstract

Introduction: Inhibition of gamma-secretase presents a direct target for lowering Aβ production in the brain as a therapy for Alzheimer's disease (AD). However, gamma-secretase is known to process multiple substrates in addition to amyloid precursor protein (APP), most notably Notch, which has limited clinical development of inhibitors targeting this enzyme. It has been postulated that APP substrate selective inhibitors of gamma-secretase would be preferable to non-selective inhibitors from a safety perspective for AD therapy.

Methods: In vitro assays monitoring inhibitor potencies at APP γ-site cleavage (equivalent to Aβ40), and Notch ε-site cleavage, in conjunction with a single cell assay to simultaneously monitor selectivity for inhibition of Aβ production vs. Notch signaling were developed to discover APP selective gamma-secretase inhibitors. In vivo efficacy for acute reduction of brain Aβ was determined in the PDAPP transgene model of AD, as well as in wild-type FVB strain mice. In vivo selectivity was determined following seven days x twice per day (b.i.d.) treatment with 15 mg/kg/dose to 1,000 mg/kg/dose ELN475516, and monitoring brain Aβ reduction vs. Notch signaling endpoints in periphery.

Results: The APP selective gamma-secretase inhibitors ELN318463 and ELN475516 reported here behave as classic gamma-secretase inhibitors, demonstrate 75- to 120-fold selectivity for inhibiting Aβ production compared with Notch signaling in cells, and displace an active site directed inhibitor at very high concentrations only in the presence of substrate. ELN318463 demonstrated discordant efficacy for reduction of brain Aβ in the PDAPP compared with wild-type FVB, not observed with ELN475516. Improved in vivo safety of ELN475516 was demonstrated in the 7d repeat dose study in wild-type mice, where a 33% reduction of brain Aβ was observed in mice terminated three hours post last dose at the lowest dose of inhibitor tested. No overt in-life or post-mortem indications of systemic toxicity, nor RNA and histological end-points indicative of toxicity attributable to inhibition of Notch signaling were observed at any dose tested.

Conclusions: The discordant in vivo activity of ELN318463 suggests that the potency of gamma-secretase inhibitors in AD transgenic mice should be corroborated in wild-type mice. The discovery of ELN475516 demonstrates that it is possible to develop APP selective gamma-secretase inhibitors with potential for treatment for AD.

Figure 1

Schematic illustration of in-vitro assays, and dose response curves of an APP selective gamma-secretase inhibitor. (a) Recombinant E.coli expressed fusion protein substrates APP MBP-C125Sw (left side, numbering relative to Aβ peptide residues, red box) or mNotch100 substrate (right hand side, numbering relative to initiator methionine residue) were incubated in separate reactions with partially purified solubilized gamma-secretase enzyme from IMR-32 cells as detailed in methods. The reactions products were analyzed by ELISA using 2G3/6H9 ELISA for Aβ_X-40 for GammaAPP assay product, or the 9F3/anti-HA ELISA for NICD, the GammaNotch assay product. MBP, N-terminal fusion with maltose binding protein (grey hatched box); TMD, transmembrane domain (blue hatched box); HA, hemagglutnin antigen epitope tag (green hatched box); HIS-6, hexahistidine residue epitope tag (yellow box). (b) Dose response curve demonstrating APP-selective inhibition from a representative screening hit ELN158762. The IC₅₀ values for inhibition of APP and Notch substrate cleavage, and selectivity ratio in the biochemical assays with this compound are reported in Table 1.

Figure 2

Novel sulfonamides exhibit properties in common with classic gamma-secretase inhibitors. Western blots of cell extracts (a) and media (b, c) after treatment of cultures in duplicate of with gamma-secretase inhibitors indicated above the lanes. Compounds were tested at two doses, approximately 1 X ED₅₀ and approximately 10 X ED₅₀ based on cellular assay results. (a) Full length and c-terminal fragments of APP total cellular lysates recognized by an antibody against the C-terminus of APP (Sigma A8717). The reference gamma-secretase inhibitors L-685,458 and ELN46719 (analog of LY411575, see methods), as well as the novel sulfonamides tested stabilize both α-CTF and β-CTF in a dose-dependent manner, and none of the compounds significantly affect steady-state levels of full-length APP. (b, c) Western blots of the conditioned media samples from this same experiment probed with antibody 8E5 against total secreted APP (b), and 192Sw against the β-secreted APP formed by BACE cleavage of the Swedish mutant APP over-expressed in these HEK293 cells (c). No significant effect on either total sAPPα or sAPPβ was seen by any of the gamma-secretase inhibitors. (d) Quantification of Aβ_1-40 in the conditioned media. The results suggest that the sulfonamide only affects gamma-secretase cleavage of APP, and does not appear to have nonspecific effects on overall APP metabolism or cell viability in HEK293 cells.

Figure 3

Binding site analysis by competitive displacement of biotinylated active site isostere. A competitive binding assay employing a biotin conjugated active site isostere to capture gamma secretase from a partially purified cell extract, plus or minus co-incubation with increasing concentrations of reference or test gamma secretase inhibitors, was developed to characterize inhibitor binding site. The eluted enzyme captured on neutravidin beads was detected on western blots with antibodies recognizing either PS-1 amino-terminal fragment (PS1-NTF), or Nicastrin (Nct). The binding assay was carried out using 50 nM affinity probe together with different concentrations of competing compounds. Serial dilutions of affinity captured enzyme were included on the gels to provide a standard curve for densitometric quantitation of test samples following autoradiography of the western blots. (a) The biotinylated isostere is displaced by its non-biotinylated analog L-685,458 in a concentration dependent manner. In subsequent experiments, a 200-fold excess of L-685,458 was employed as a positive control. (b) LY411575 was tested for its ability to displace active site isostere at concentrations ranging up to 1000-fold its IC₅₀ in the Gamma APP enzyme assay. There is no significant displacement of the active site directed compound by at concentrations below 200-fold its IC₅₀ in the enzyme assay. At higher concentrations, a modest dose-dependent effect of LY411575 to displace the active-site-directed compound is observed on both Western blots. Previous observations revealed substrate concentration affects the potency of sulfonamides in cell and enzyme assays (unpublished). Hence, the ability of sulfonamides to displace the active site directed probe was tested in the absence (c and d, left side) or presence (c and d, right side) of 1 Km substrate (MBP-C125sw). Substrate was added to enzyme concurrently with test compound and affinity ligand. In the presence of added APP, the sulfonamides displace the active site probe in a dose-dependent manner. (c) No displacement of active site probe is observed by ELN-318463 in the absence of substrate. In the presence of substrate, ELN318463 displaces approximately 50% of the active site probe at a concentration of approximately 2,000-fold its IC₅₀ in the Gamma APP assay. (d) ELN-475516 does not displace the active site probe at concentrations ranging up to 2,000X its IC₅₀ in the gammaAPP assay. However, in the presence of substrate, ELN475516 displaces 50% of the active site probe at a concentration equivalent to approximately 67-fold its Gamma APP IC₅₀. The results shown in (c) and (d) suggest displacement of active site isostere from gamma-secretase by benzene caprolactam and fused pyrazolo-bicyclic sulfonamide is influenced by the presence of substrate.

Figure 4

Acute reduction of brain Aβ by ELN318463 in PDAPP (a) and FVB mice (b). The mice were dosed orally with vehicle, or ELN318463 at 30 mg/kg or 100 mg/kg and sacrificed 3 h post treatment. Brain Aβ_1-x in PDAPP (a), or Aβ_x-40 in FVB (b) was quantified by ELISA following extraction in guanidine buffer as described in methods. ** denotes P < 0.01, *** denotes P < 0.001 by Mann-Whitney in A, and two-tailed Students t-test in (b).

Figure 5

Reduction of brain Aβ and thymic Hes-1 in wild-type FVB mice after treatment with gamma-secretase inhibitor. Female FVB mice were treated orally with a single dose of ELN475516 (30, 100 or 300 mg/kg), LY411575 (10 mg/kg) or vehicle and sacrificed at three hours post dose. (a) Cortical Aβ_x-40 levels, estimated by ELISA from guanidine extracted brain homogenates. (b) Thymic Hes-1 mRNA levels, estimated by TaqMan RT-PCR. Statistical significance between treatment groups and vehicle was determined by ANOVA and Dunnett's test. *** P < 0.001, * P < 0.05

Figure 6

Time course of brain Aβ reduction in wildtype FVB mice after treatment with a single dose of ELN475516. Female FVB mice were treated orally with a single dose of ELN475516 (30, 100 or 300 mg/kg, n = 5/dose group) or vehicle and sacrificed at 1, 3, 6, 10, 14 and 24 hours post dose. (a) Cortical Aβ_x-40 levels, estimated by ELISA from guanidine extracted brain homogenates. (b) ELN475516 levels in plasma and brain estimated by LC/MS/MS. (c) ELN475516 plasma and brain exposure over 24 hours (AUC_0-24).

Figure 7

Reduction of Aβ and Math1 in FVB mice after treatment with ELN475516 or LY411575 for seven days. Female FVB mice were treated orally twice daily with ELN475516, LY411575 or vehicle and sacrificed at three hours post last dose. (a) Cortical Aβ_x-40 levels, estimated by ELISA from guanidine brain homogenates. Statistically significant (P <.0001) Aβ reduction (33%) was achieved at the lowest dose tested (15 mg/kg/dose). (b) Relationship of ELN475516 levels in plasma (estimated by LC/MS/MS) vs. brain Aβ_x-40 levels (estimated by ELISA from guanidine brain homogenates). Duodeneum (c) and Ileum (d) Math1 mRNA levels, estimated by TaqMan RT-PCR. Treatment with 5 mg/kg/dose LY411575 BID for seven days resulted in a robust elevation of Math1 gene expression in the duodeneum and ileum (P < 0.001). No effects on Math1 in the duodeneum or ileum were observed after treatment of ELN475516 BID for seven days up to the highest dose tested, 1,000 mg/kg/dose (2,000 mg/kg/day). Statistical significance between treatment groups and vehicle was determined by ANOVA and Dunnett's test. ***, P < 0.001

Figure 8

Assessment of systemic safety signals in FVB mice after treatment with ELN475516 or LY411575 for seven days. Female FVB mice were treated orally twice daily with ELN475516, LY411575 or vehicle and sacrificed at three hours post last dose. Mice treated with 5 mg/kg LY411575 BID for seven days showed consistent signs of toxicity while there were no observed signs of toxicity in any group treated with ELN475516, even up to the highest dose of 1000 mg/kg (2000 mg/kg/day) for seven days. (a) Body weight (b) Neutrophils (c) Relative thymus weight and (d) Relative small intestine weight. Statistical significance between treatment groups and vehicle was determined by ANOVA and Dunnett's test. *** P <0.001, ** P < 0.01.

Figure 9

Photomicrographs of ileum sections (10X magnification) from mice treated seven days x b.i.d. with gamma-secretase inhibitors noted below. Selected sections represent the range of goblet cell hyperplasia observed in this study. In H&E stained sections (a, c, e and g), goblet cells appear as cells with clear or foamy cytoplasm (arrows) while in PAS stained sections (b, d, f and h) they appear as dark magenta stained cells (arrow heads). A normal population of goblet cells was observed in all vehicle control mice (a, b). All positive control LY411575 treated mice had a moderate increase in goblet cells (c, d). There was no apparent dose-related increase in goblet cells across ELN475516 treated groups (see Table 2); among all mice in these groups, 21 had no increase in goblet cells, 8 had a questionable increase (illustrated by a single animal from 100 mg/kg dose group g, h). A single mouse from the 600 mg/kg dose group had a mild increase in goblet cells, which is shown for comparison purposes in panels (e and f). Scale bar = 100 μm.