Background
Alzheimer’s disease (AD) is a progressive neurodegenerative disease, first described in 1907 [
1]. Functional changes in this devastating disease include early memory deficits and later deficits affecting language, executive functioning, perception, and/or execution of complex motor patterns [
2]. The neuropathology in AD is characterized by the presence of extracellular amyloid (Aβ) plaques and intracellular neurofibrillary tangles in the cerebral cortex, hippocampus, and amygdala, as well as other brain regions associated with memory and other domains of cognition [
3]. The major constituents of the amyloid plaques are well established and include the 4 kDa Aβ peptides, primarily Aβ
42 and Aβ
40[
4‐
7]. There is considerable evidence suggesting that Aβ
42 is deposited early in the amyloid plaques and forms the seed for later deposition of other Aβ species [
5,
8]. This evidence has suggested that an early and constant reduction in Aβ
42 in prodromal AD might delay the onset or slow the progression of the disease by affecting the rate of plaque formation. Aβ
42 is derived from sequential processing of amyloid precursor protein (APP) by two proteases, β-site APP-cleaving enzyme (BACE) and γ-secretase [
9,
10]. Initially, BACE cleaves APP in its ectodomain, followed by γ-secretase cleavage in the APP transmembrane domain [
9,
10], which ultimately produces a number of Aβ peptides of various lengths [
11]. Significantly, most familial AD mutations in the APP gene are found around the BACE or γ-secretase cleavage sites, and thus both BACE and γ-secretase have been targets for therapeutic intervention in AD [
8,
12].
γ-Secretase is a large complex composed of the four polypeptides: presenilin (PS-1 or PS-2), nicastrin (Nct), presenilin enhancer 2 (Pen-2), and anterior pharynx-defective 1 (Aph-1) [
7]. γ-Secretase is responsible for the processing of more than 70 transmembrane proteins involved in normal cellular processes, including regulation of cell fate, cell adhesion, migration, neurite outgrowth, synaptogenesis, calcium homeostasis, transport of membrane proteins, and cell signaling [
13,
14]. Mutations in PS, the putative catalytic site of the γ-secretase complex, and in APP around the γ-secretase cleavage site, were shown to increase the ratio of Aβ
42/Aβ
40, strengthening support for γ-secretase inhibition for disease modification in AD [
8,
12]. While inhibition of γ-secretase would produce the desired Aβ reduction, it would also affect the proteolysis of its other substrates. The Notch receptor is one of these substrates [
15], which is of particular interest since the inhibition of its proteolytic processing by γ-secretase inhibitors (GSIs) has been shown to result in the suppression of intestinal goblet cell differentiation and in immunosuppression [
16]. Several GSIs have entered clinical trials in AD, but unfortunately, have produced toxicities that are presumably mechanism-based. In particular, one compound (Semegacestat) produced drug-related rashes, lightening of hair color, skin cancer, and more importantly, worsening of cognition and the ability to perform activities of daily living [
17‐
19]. These mechanism-based toxicities of GSIs have been attributed to the inhibition of Notch receptor processing and to the accumulation of the APP β-carboxyl-terminal fragment (CTF) [
16,
19].
Neuroinflammation is another pathological hallmark of AD and is characterized by the presence of activated microglia and reactive astrocytes surrounding the amyloid plaques [
20]. The question of whether the gliosis is causative or a compensatory result of the amyloid plaque deposition has been the subject of ongoing discussions and studies since it was first described [
20,
21]. For example, numerous retrospective studies associated a lower incidence of AD in patient populations that were prescribed non-steroidal anti-inflammatory drugs (NSAIDs) for other conditions [
20]. It was therefore assumed that the NSAID therapy exerted positive effects on AD by reducing neurotoxic inflammation through the reduction of cyclooxygenase (COX) activities [
20]. However, Weggen et al. [
22,
23] described a series of
in vitro and
in vivo studies utilizing several NSAIDs that produced a preferential reduction of Aβ
42 compared to Aβ
40. This reduction of Aβ
42 was accompanied by a concomitant increase in Aβ
38, a shorter, less amyloidogenic Aβ peptide [
12], rather than the inhibition of all carboxyl-terminal processing of APP [
22,
23]. Furthermore, they demonstrated that the effects of NSAIDs on the preferential reduction of Aβ
42 peptide levels were not linked to the inhibition of COX or other enzymes, but rather to a specific action on γ-secretase [
22,
24]. The shift in production of Aβ peptides from the longer, toxic forms to the shorter, less toxic forms by NSAIDs has been termed γ-secretase modulation. This has sparked a flurry of activity directed at the development of compounds that modulate APP cleavage by γ-secretase and that could avoid the toxicities arising from the complete enzymatic inhibition of γ-secretase. Several recent publications have described second generation γ-secretase modulators (GSMs) [
25‐
29] and Notch-sparing GSIs [
18,
30]. Here we present the
in vitro and
in vivo characterization of EVP-0015962, a potent, second generation GSM that specifically modulated production of Aβ
42 and Aβ
38 without affecting other γ-secretase substrates. In transgenic mice over-expressing APP, EVP-0015962 was well tolerated following chronic dosing, produced reductions in amyloid plaque burden and neuroinflammation, and improved cognition.
Discussion
The presence of neuroinflammation in AD and the seemingly positive effects of chronic NSAID usage on the reduction of the risk of developing AD in retrospective epidemiological studies [
20,
41] suggested that NSAIDs might be operating via the inhibition of COX or activation of the peroxisome proliferator-activated receptor γ. Indeed, studies in Tg2576 mice generated evidence to support this hypothesis. Tg2576 mice that were fed a diet of 375 ppm of ibuprofen for 4–6 months, showed a marked decrease in amyloid plaque load, as well as in inflammatory mediators, such as interleukin-1β, GFAP, and markers of microglial activation [
42,
43]. However, subsequent mechanistic studies suggested that certain NSAIDs were able to produce a preferential reduction of Aβ
42, without altering the levels of other Aβ peptides, presumably through a direct modulation of γ-secretase [
22,
23,
44]. The result of γ-secretase modulation by NSAIDs was an overall shift in the Aβ
42/Aβ
40 ratio, with an accompanying increase in the shorter Aβ
38, both of which were not correlated to COX activity or the activity of other enzymes inhibited by NSAIDs [
22‐
24].
The first generation GSM to be tested in the clinic was Flurizan™, the
R enantiomer of the racemate NSAID, flurbiprofen [
44,
45]. Flurizan™ has the classic GSM signature of lowering Aβ
42 without affecting the total amount of Aβ peptides. However, Flurizan™ was a relatively weak GSM (IC
50 > 300 μM for Aβ
42[
46]), with low brain concentrations in animals and man, and was subsequently withdrawn from clinical evaluation due to the lack of efficacy in a large phase III trial [
47]. More recently, second generation GSMs from multiple chemical classes have been reported, which have improved potency and brain concentrations [
25‐
29,
44,
48]. In the current study, we have characterized EVP-0015962, a potent small molecule compound, which has a substantially improved potency (4,500-fold) and pharmacological profile compared with Flurizan™. EVP-0015962 had a measured LogD of 3.88 and was present in the brain at 1.3 to 4.3 μM after single oral doses of 10 and 30 mg/kg, respectively. Moreover, we have demonstrated that γ-secretase activity is modulated by EVP-0015962 to selectively decrease Aβ
42 levels (IC
50 = 67 nM) in H4-APP
751 cells, and increase levels of the shorter Aβ
38 peptide (EC
50 = 33 nM), without a change in Aβ
Total (composed primarily of Aβ
40[
7]) or in Aβ peptides of other lengths. These changes in the levels of Aβ
42 and Aβ
38 peptides were seen in the
in vitro assay systems, as well as
in vivo in Tg2576 mice after acute treatment. The lack of overall change in Aβ peptide levels and the increase in Aβ
38 by EVP-0015962 suggest selective modulation, rather than inhibition of the γ-secretase enzyme complex, similar to the mechanism of action of other GSMs with similar chemotypes [
48]. In contrast, GSMs of other chemical classes do reduce Aβ
40 peptide levels, albeit with a higher IC
50 than that for Aβ
42, suggesting that they might function differently than EVP-0015962 [
29,
48].
We further substantiated the modulatory effects of EVP-0015962 by determining whether the compound inhibited the S3/ε-cleavage activity of the γ-secretase complex at its other natural substrates. EVP-0015962 did not inhibit the ability of γ-secretase to generate AICD and NICD
in vitro. As expected from a lack of effect on AICD formation, no accumulation of APP α- and β-CTFs was observed in the presence of EVP-0015962 either
in vitro or
in vivo. EVP-0015962 also did not cause the accumulation of EphA4 CTF. The lack of effect on the processing of other γ-secretase substrates is thus also consistent with γ-secretase modulation. This suggests that EVP-0015962 may avoid the mechanism-based side effects that have been previously associated with GSIs (e.g., intestinal goblet cell hyperplasia and immunosuppression due to inhibition of Notch receptor processing and cognitive impairment due to the accumulation of APP β-CTF) [
16,
19]. Furthermore, in our longitudinal efficacy study in Tg2576 mice, in which we observed a significant reduction of Aβ
42, we found no evidence of effects on intestinal goblet cells or on any other organ system examined in satellite animals treated with up to 60 mg/kg/day of EVP-0015962 for 11 weeks. The tolerability of EVP-0015962 was also demonstrated by the similar weight gains at the 20 and 60 mg/kg/day doses of EVP-0015962, when compared to Tg2576 mice on the control diet for 50 weeks. In addition, when survival was examined after 50 weeks of treatment, 25% of the high (60 mg/kg/day) dose group was lost compared to 27% of the control diet group. The observed mortality rates were consistent with the > 20% rates reported for this transgenic model by the supplier (
http://www.taconic.com/wmspage.cfm?parm1=2646).
One of the major goals of the longitudinal study was to determine the effects of EVP-0015962 on AD-like pathology in a transgenic mouse model. The Tg2576 mouse model expresses high levels of APP and Aβ with a transgene containing human APP695 bearing the Swedish mutation (K670N/M671L) behind the hamster prion protein gene promoter [
36]. One significant advantage of the Tg2576 model is that diffuse and neuritic plaques begin to appear at approximately 6–7 months and, along with extractable Aβ, increase with age [
37]. Furthermore, these age-related changes are coincident with the appearance of cognitive deficits [
36,
49], suggesting that these mice are a useful tool for studies of Aβ modifying therapies.
In an interim assessment, after 11 weeks of chronic treatment with EVP-0015962, we examined the behavioral effects of GSM treatment. Studies have shown that treatments that lower Aβ can restore cognitive function in these animals [
39,
40]. In the CFC assay, which is primarily hippocampal dependent [
38], the Tg2576 mice displayed the expected cognitive deficit in contextual memory at 30–33 weeks of age. Importantly, the cognitive deficit was reversed and/or prevented at both dose levels following 11 weeks of treatment with EVP-0015962. Separate groups of Tg2576 and WT mice received a single administration of EVP-0015962 (30 mg/kg, p.o. on day 1, 3 h prior to training) in order to evaluate the acute effects of EVP-0015962 treatment on CFC. Overall, differences between the genotypes were observed in this study (i.e., a genotype-associated deficit in the Tg2576 mice). However, acute dosing with EVP-0015962 did not reverse this deficit. This is in contrast to the report that 100 mg/kg, p.o. of the GSI DAPT, administered 3 h before training in Tg2576 mice, lowered Aβ levels and rapidly improved the CFC cognitive deficit [
39]. The WT mice, however, did not have improved CFC performance after DAPT treatment, suggesting that the effect was due to Aβ lowering, rather than to a general cognitive enhancing effect, as observed with rolipram in the same study. It is unclear by what mechanism a GSI improved cognition acutely and why the GSM EVP-0015962 did not. In a Y-maze spontaneous alternation test of 5.5-month-old Tg2576 mice, acute GSI treatments also improved short term spatial memory, but 8 days of treatment did not [
19]. This lack of effect of subchronic treatment with GSIs on spatial memory was attributed to the accumulation of APP β-CTF in hippocampal synapses and was consistent with findings in the clinic of worsening of cognition and the ability to perform activities of daily living in AD patients [
18]. In contrast, GSM-2 improved short term spatial memory after both 1 and 8 days of treatment, suggesting that acute effects of GSMs on cognition can be observed under some experimental conditions [
19].
After 50 weeks of chronic daily treatment with EVP-0015962 in a food formulation, which began before amyloid plaque deposition (17–26 weeks of age), brain levels of soluble Aβ
42 were dose dependently reduced. Additionally, increased levels of Aβ
38 were observed in the TBS soluble pool of Aβ in the 60 mg/kg/day group. Aβ
Total was not changed, similar to what was observed in C57BL/6 mice (data not shown) and Tg2576 mice administered a single dose of EVP-0015962. These findings indicated that EVP-0015962 was modulating Aβ production in the Tg2576 mice in a similar fashion, whether there were Aβ deposits or not. In addition, Aβ peptides, particularly Aβ
42 and Aβ
Total (primarily Aβ
40), were decreased in the formic acid extractable pool, albeit to different extents. Since Aβ deposits contain all lengths of Aβ peptides but are primarily composed of Aβ
42 and Aβ
40[
4‐
7], this finding is consistent with a reduction in Aβ deposition and is supportive of a potential disease modifying effect of EVP-0015962.
As suggested by the reductions in insoluble Aβ peptide levels after chronic treatment with EVP-0015962, Aβ aggregates were also significantly lowered by the high-dose treatment. The degree of reduction in aggregated Aβ (73%) after treatment with 60 mg/kg/day of EVP-0015962 was consistent with the reduced levels of Aβ42 detected by enzyme-linked immunosorbent assay (ELISA) after formic acid extraction of insoluble Aβ (86%) and of Total Aβ detected by Western blotting (81%). Interestingly, more disparity among the measures in amount of Aβ reduction was observed for the 20 mg/kg/day treatment. In the aggregated Aβ assay the reduction was 19%, whereas the reductions by ELISA and Western blotting were larger and similar (about 40–50%).
Chronic treatment of Tg2576 mice with other GSMs has demonstrated reductions in amyloid deposition [
25‐
27]. Similarly, the histological analyses of the chronically treated Tg2576 mice demonstrated reductions in percent area and number of amyloid plaques in the neocortex and hippocampus. Although there was only a small non-significant difference between the two doses of EVP-0015962, there was a general trend towards a more pronounced effect at the 60 mg/kg/day dose. Significant reductions were observed in the aggregated Aβ of the plaque cores (visualized with ThioflavinS) and in the more diffuse Aβ surrounding the plaque cores (visualized with the 6E10 antibody) [
50‐
53]. Although both staining procedures produced significant EVP-0015962 treatment effects for percent area and number of plaques, plaque size was only reduced in the ThioflavinS analysis. Since the cores of amyloid plaques are initially seeded by aggregated Aβ
42[
5], and this is the peptide most reduced by EVP-0015962 treatment, it is likely that the higher number of significant findings with ThioflavinS relates to its preferential staining of the plaque cores, while 6E10 antibody recognizes Aβ of various lengths and states of aggregation, which were less altered overall by the treatment. This suggests that EVP-0015962 might mediate its effect through a reduction of the amount of Aβ
42 available for the initial seeding and growth in the size of amyloid plaque core, and maybe less through the subsequent growth of plaques by the addition of Aβ peptides of other lengths.
The effect on reactive gliosis was pronounced, and may have been the result of the significant reduction in the amyloid plaque load and/or a direct anti-inflammatory activity of EVP-0015962. In enzyme assays for COX-1 and COX-2, the IC
50 of EVP-0015962 for COX-1 was > 10 μM (n = 2), and the percent inhibition for COX-2 at 30 μM was 56% (n = 2), well above the
in vitro IC
50 for Aβ
42 (67 nM) and the brain exposures of EVP-0015962 after chronic treatment. However, effects of EVP-0015962 on other inflammatory pathways have not been excluded. Nevertheless, since the reactive gliosis in Tg2576 mice on the control diet was localized around the amyloid plaques and Aβ can activate astrocytes and microglia, as well as induce inflammatory responses [
20], it is likely that the EVP-0015962-induced decrease in Aβ
42 and amyloid plaques at least partially led to the attenuated reactive gliosis. These data suggest that a therapy which reduces Aβ
42 levels will effectively reduce amyloid plaque formation and the concomitant reactive gliosis.
Methods
EVP-0015962
EVP-0015962 was synthesized and chirally separated, as described by Shapiro and Chesworth [
54]. For
in vitro experiments, EVP-0015962 was dissolved in dimethyl sulfoxide (DMSO, Sigma-Aldrich, St. Louis, MO) as a 30 mM stock solution for dilution into medium. For food formulation studies, EVP-0015962 (20 and 60 mg/kg/day) was milled into standard mouse chow pellets (AIN-93M, Research Diets, Inc., New Brunswick, NJ). Diet doses of EVP-0015962 were calculated based on the desired single daily dose, body weight, and daily food intake. The food formulations of 20 and 60 mg/kg/day were equivalent to 171 and 514 ppm of EVP-0015962 in the diet, respectively.
COX enzyme assays
COX-1 and COX-2 enzyme assays were performed by Cerep (Redmond, WA, Catalog # 0726 and 0727). Briefly, the assays used human recombinant Sf9 cells and arachidonic acid (4 μM for COX-1 and 2 μM for COX-2) as the substrate. Production of PGE2 was detected by enzyme immunoassay.
Animals
Animals were maintained on 12/12 h light/dark cycle with food available ad libitum. All procedures were performed with approval from the Institutional Animal Care and Use Committee and were in accordance with the guidelines in the Guide for the Care and Use of Laboratory Animals from the U.S. Department of Health and Human Services.
Cell-based assay for measuring secreted Aβ peptides
Human neuroglioma H4 cells (ATCC, Manassas, VA) were transfected with a pcDNA3.1 (Invitrogen, Carlsbad, CA) plasmid expressing human wild type APP
751 cDNA; and a stable cell line was generated using G418 (Invitrogen) selection [
55]. Prior to experimentation, H4-APP
751 cells were maintained in Dulbecco’s modified Eagle’s high glucose medium with 10% fetal bovine serum, 1% penicillin-streptomycin, 2 mM L-glutamine, and 0.4 mg/mL G418. All culture reagents were from Invitrogen.
Cells were plated at 15,000 cells/well in Costar 96-well plates (Corning, Corning, NY) and placed at 37°C and 5% CO2. Six hours after plating, cells were washed three times with Pro293™ chemically defined medium (Lonza, Walkersville, MD) with 1% penicillin-streptomycin and 2 mM L-glutamine, followed by addition of EVP-0015962 (0.003–10 μM, final DMSO concentration of 0.33%). Plates were incubated overnight (16–18 h) and supernatant was removed for quantification of Aβ peptides by sandwich ELISA. Cytotoxicity was evaluated using CellTiter 96® AQueous One Solution Cell Proliferation Assay (MTS Assay, Promega, Madison, WI) according to the manufacturer’s protocol.
Rat primary neocortical cultures
Primary cultures were established from the neocortex of E17 rat embryos obtained from timed pregnant CD rats (Charles River Laboratories, Wilmington, MA). Following tissue dissection and trituration, the cultures were suspended in Neurobasal™ medium (Invitrogen) supplemented with 10% horse serum (Sigma-Aldrich) and 520 μM L-glutamine [
56,
57]. Cells were plated at 50,000 cells/well in Costar 96-well poly-
d-lysine-coated plates. Following incubation at 37°C and 5% CO
2 for 4–5 h, the plating medium was exchanged with Neurobasal® medium with 2% B-27® supplement (Invitrogen), 520 μM of L-glutamine, and 1% penicillin-streptomycin. Assays were performed at day eight
in vitro (DIV8) after replacement of one-half of the medium and addition of EVP-0015962 (0.01–30 μM, final DMSO concentration of 0.1%). Cultures were incubated with compound for 24 h for analysis of Aβ peptides by sandwich ELISA and cytotoxicity by MTS assay.
Aβ measurements
Aβ peptide levels were quantified by sandwich ELISA. Plates were coated overnight at 4°C with either human anti-Aβ
Total (amino acids [aa] 27–37) or rodent anti-Aβ
1-x (aa 1–8) rabbit polyclonal antibodies. For detection of Aβ
38 and Aβ
42, plates were coated with antibodies specific for the carboxyl-terminal neoepitope of each peptide. Antibodies were from Dr. Pankaj Mehta (Institute for Basic Research in Developmental Disabilities, Staten Island, NY) [
58].
Freshly collected samples of cultured cell supernatant or synthetic peptides (American Peptide Co., Sunnyvale, CA or California Peptide Research, Inc., Napa, CA) for standard curves were diluted with media, and loaded into anti-Aβ antibody-coated plates and incubated at 4°C for about 24 h. Aβ peptides were detected with the 4G8 antibody (Aβ, aa 17–24) conjugated to horseradish peroxidase (Covance, Inc., Princeton, NJ), and SureBlue 3,3’,5,5’-tetramethylbenzidine (TMB) peroxidase substrate (KPL, Inc., Gaithersburg, MA). Following addition of TMB stop solution (KPL, Inc.), plates were read for absorbance at 450 nm (SpectraMax M5e Microplate Reader, Molecular Devices, Inc., Sunnyvale, CA). EVP-0015962-treated samples were normalized to samples treated with DMSO alone (no inhibition) and to samples treated with DAPT (5 μM, Sigma-Aldrich). IC50 and EC50 values were calculated from values reported as percent of DMSO controls using nonlinear regression, based on a sigmoidal dose–response (variable slope) model. All data were analyzed in Prism 4 (GraphPad Software, Inc., San Diego, CA).
MALDI-TOF mass spectrometry
Immunoprecipitation of carboxyl-terminally truncated Aβ peptides from 4 mL of H4 cell media was conducted using Aβ-specific antibodies coupled to magnetic beads as described elsewhere [
59]. The anti-Aβ antibodies 6E10 and 4G8 (Covance, Inc.) were used. After elution of the immune-purified Aβ peptides, the detections were performed on an UltraFlextreme MALDI-TOF/TOF instrument (Bruker Daltonics, Bremen, Germany).
AICD production
The effects of increasing concentrations of EVP-0015962 on cleavage products of APP were tested in a cell-free system. DAPT was used as a reference positive control for the assay. For each assay condition, a 5-μL microsomal fraction, containing 1% CHAPSO (3-[(3-cholamidopropyl)dimethylammonio]-2-hydroxy-1-propanesulfonate) from murine embryonic fibroblasts (MEF) lacking the Aph-1A, Aph-1B, and Aph-1C genes [
60] (triple knock out) and rescued with Aph-1A
L, was prepared in a reaction mixture of 5 μL substrate (recombinant APP β-CTF 3-Flag-tagged) at 0.8 μM, 0.0125% phosphatidylethanolamine, 0.1%, phosphatidylcholine, 0.05% sphingomyelin, and 8.9 μL PIPES (piperazine-N,N′-bis[−ethanesulfonic acid]) buffer. EVP-0015962 (0.001–100 μM) or DAPT (10 μM) in 0.5 μL of DMSO was added to the reaction mixture and incubated at 37°C for 3 h. For detection of AICD, one-half of the reaction mixture was purified by a methanol-chloroform extraction in order to remove lipids and hydrophobic proteins (e.g., APP β-CTF). This purified fraction was further analyzed by SDS-PAGE and Western blotting. AICD was detected with an ANTI-FLAG® antibody (Sigma-Aldrich). The other half of the reaction mixture was used to analyze the levels of Aβ
1-40 and Aβ
1-42 by AlphaLisa (Perkin Elmer, Waltham, MA).
APP CTF assays
To evaluate the accumulation of APP α- and β-CTFs, H4-APP
751 cells were treated for 16 h with EVP-0015962 or the GSI LY-411,575 (Alchem Laboratories Corp., Alachua, FL) at 0.001–10 μM. Cells were lysed
in situ and the resulting lysate cleared of nuclei by centrifugation, as described elsewhere [
61]. Samples were fractionated by SDS-PAGE on a 12% NuPage gel (Invitrogen), transferred to a nitrocellulose membrane, and immunoblotted with an anti-APP-CTF antibody (Sigma-Aldrich). Following incubation with an IRDye secondary antibody (Li-Cor Biosciences, Lincoln, NE), detection was performed using an Odyssey scanner (LiCor Biosciences).
For determination of APP α- and β-CTFs in Tg2576 mouse brain, hemisphere samples were thawed in ice-cold homogenization buffer (150 mg of tissue/mL, 50 mM Tris and 150 mM NaCl, pH 7.4) with protease inhibitor cocktail (Mini Complete™, Roche Applied Science, Penzberg, Germany). Homogenized samples were spun at 435,000
g for 50 min at 4°C. The supernatant was discarded and the pellet resuspended in the same volume of homogenization buffer containing 0.5% Triton®-X 100, 1% deoxycholate, and 3% SDS. The pellets were triturated with a pipette until resuspended and incubated on a rotator for 1 h at 4°C. The samples were then centrifuged at 435,000
g for 50 min at 4°C and the detergent-soluble supernatant was collected. Protein concentration for membrane (detergent soluble) fraction was determined using the BCA method (Pierce, Rockford, IL). Thirty micrograms of total protein were separated on a Tricine 10-20% gel (Invitrogen) and transferred to a nitrocellulose membrane. Blots were probed with an anti-APP-CTF antibody (Sigma-Aldrich) or anti-Aβ antibody (clone 6E10) followed by incubation with IRDye secondary antibodies (Li-Cor Biosciences). Detection was performed using an Odyssey scanner (LiCor Biosciences) and blots quantified using ImageJ software (available at
http://rsbweb.nih.gov/ij).
NICD production
HEK293 cells (700,000 cells per assay condition) were transiently transfected with a NotchΔE, myc-tagged construct to produce a truncated Notch1 substrate, which is constitutively cleaved by γ-secretase (S3 cleavage) [
62]. Control transfections were performed with an empty vector. Twenty-four hours after transfection, the cells were treated with EVP-0015962 (0.001–30 μM) or DMSO (control) for 20 h. Lactacystin (10 μM, Sigma-Aldrich) was used to prevent degradation of NICD. Cell lysates were analyzed by SDS-PAGE and Western blotting using a c-myc monoclonal antibody (9E10, Santa Cruz Biotechnology, Santa Cruz, CA) and a cleaved Notch1 antibody (Cell Signaling Technology, Beverly, MA). The amounts of NICD and total Notch were quantified and NICD was expressed as a percentage of total Notch. The DMSO control condition was set at 100%. Cell viability was determined using the CellTiter-Blue® Cell Viability Assay (Promega).
EphA4 CTF assay
Rat primary neocortical cultures (DIV14) were treated with 0.1–10 μM of EVP-0015962 or LY-411,575 for 24 h. Cells were lysed with RIPA buffer (50 mM Tris–HCl, pH 7.4, 150 mM NaCl, 0.25% deoxycholic acid, 1% NP-40, and 1 mM EDTA) (Upstate, Temecula, CA) containing protease inhibitors and samples centrifuged for 8 min at 16,200
g to remove cell debris. Protein concentration of the lysate was determined by the BCA method, and 30 μg of total protein were loaded on a 10% Bis-Tris gel (Invitrogen). EphA4 CTF was visualized using an anti-EphA4-CTF antibody (Abnova, Walnut, CA) and an IRDye800CW® secondary antibody (Rockland Immunochemicals, Inc., Gilbertsville, PA). Detection was performed using an Odyssey scanner (LiCor Biosciences) and blots quantified using ImageJ software (available at
http://rsbweb.nih.gov/ij).
Acute treatment study in Tg2576 mice
EVP-0015962 was prepared in a vehicle of 10% DMSO, 15% Solutol® HS 15 (BASF, Ludwigshafen, Germany), 10% ethyl alcohol, and 65% water. The 10 and 30 mg/kg doses were a solution and a suspension, respectively. Male Tg2576 mice (B6;SJL-Tg(APPSWE)2576Kha, model 1349, Taconic, Hudson, NY) were used at 21 weeks of age [
36]. Mice (24–35 g body weight) were administered 10 or 30 mg/kg, p.o. of EVP-0015962 without prior fasting, and euthanized by CO
2 asphyxiation 4 h post-dosing. After extraction of the brain, olfactory bulbs and hindbrain were removed and the cerebral hemispheres were cut into 4 pieces. Cerebellum was removed for bioanalysis of EVP-0015962 concentrations. All tissue samples were weighed, frozen in liquid nitrogen, and stored at −80°C until analysis.
Chronic treatment studies in Tg2576 mice
Male C57BL/6 mice (Charles River Laboratories, 10–12 weeks of age, 24–32 g body weight) were group housed (3–5 per cage) and used in pilot studies to establish the concentrations of EVP-0015962 for the food formulations for the chronic treatment studies. Mice were placed on the food formulations for 3 or 7 days and were euthanized, as described above, approximately 1 h after the onset of the light cycle.
Male Tg2576 and WT mice were both derived from crosses of male Tg2576 mice and female B6SJLF1 mice (Taconic) and were singly housed upon arrival. At 16–23 weeks of age, mice were placed on the control diet (AIN-93M), and one week later, were switched to the experimental diets (EVP-0015962 at 20 or 60 mg/kg/day) or maintained on the control diet. Tg2576 mice were maintained on the food formulations containing EVP-0015962 for 11, 28, or 50 weeks, and were sacrificed at 2–4 h after the onset of the light cycle. Some animals were euthanized as described above. For histological studies, a subset of mice were euthanized by CO2 asphyxiation and transcardially perfused with cold 0.9% NaCl. The right hemispheres were immersion fixed in fresh 4% paraformaldehyde in 0.1 M phosphate buffer, pH 7.4 for 1 h, transferred to a 15% sucrose cryoprotectant for 24 h at room temperature, and on the next day, frozen in isopentane chilled by dry ice for 1 min. The left cerebral hemisphere was cut into 2 pieces, weighed, and frozen.
Measurement of EVP-0015962 in brain
Cerebellums were homogenized in a Mini-BeadBeater (Biospec Products, Inc., Bartlesville, OK). Brain samples and calibration standards in brain homogenate were prepared for liquid chromatography-tandem mass spectroscopy (LC-MS/MS) by precipitating proteins with acetonitrile and vacuum filtration in the presence of an internal standard (EVP-0015962-D6). EVP-0015962 was resolved by HPLC (Shimadzu, Kyoto, Japan) using a reverse-phase Xterra® C18 column (100 × 2.1 mm i.d.) (Waters Corp., Milford, MA). Following separation, the column effluent was introduced into a hybrid triple quadrupole/linear ion trap mass spectrometer (API 3200 Q-Trap, Applied Biosystems, Foster City, CA), optimized for detection of EVP-0015962 and using multiple reaction monitoring with mass transition of 479.300>296.700. Concentration of EVP-0015962 was measured as nanograms per gram of brain and expressed as micromolar.
In order to examine both soluble and insoluble pools of Aβ, sequential extraction of Tg2576 mouse brain samples was performed to produce 4 pools of Aβ, essentially as described elsewhere [
37]. Briefly, brains were homogenized in 25 mM TBS, pH 7.4 with protease inhibitors (Mini Complete™) to release the soluble Aβ into the resulting supernatants after high speed ultracentrifugation. The above procedure was then repeated with the pellets from the previous extraction step, using 1% Triton®-X 100 in TBS with protease inhibitors, 2% SDS with protease inhibitors, and 68.75% formic acid. The extracted Aβ from each step was subjected to ELISA for human Aβ
42, Aβ
38, and Aβ
Total, as described above.
Detection of aggregated Aβ in Tg2576 mouse brain
The A4 assay (Amorfix Life Sciences, Toronto, Canada) was employed to quantify levels of aggregated Aβ in Tg2576 mouse brains. Brain samples were homogenized in 10 volumes of 2% NP-40 in phosphate buffered saline (PBS, pH 7.4) containing 1 mM phenylmethanesulfonylfluoride (PMSF) and protease inhibitors (Mini Complete™). The homogenates were further diluted in PBS with 0.05% Tween® 20 and 1% bovine serum albumin to a final concentration that would provide a signal within the linear range of the immunoassay. A sample enrichment protocol, proprietary to Amorfix Life Sciences, was used to specifically isolate oligomeric and aggregated Aβ rather than monomeric Aβ. Following enrichment, samples were eluted and disaggregated to allow detection of now monomeric Aβ [
63]. Aβ was detected by immunoassay following incubation at 37°C with europium-fluorescent beads coupled to the mouse monoclonal 4G10 antibody (N-terminal Aβ, aa 1–17) and magnetic beads coupled to the antibodies 1F8 (C-terminal Aβ, aa 3–40) and 2H12 (C-terminal Aβ, aa 3–42). 4G10, 1F8, 2H12 were produced by Amorfix Life Sciences. Following incubations the samples were placed on a magnet to isolate the immune complex. The intensity of the europium fluorescent signal was measured using time resolved fluorescence on each sample in triplicate and was taken as being directly proportional to the concentration of aggregated Aβ in the sample. The limit of detection using this assay was 50 fg per well. The fluorescent signal for each mouse relative to the background noise for the assay was then expressed as percent of the control diet group.
Histology
The right hemispheres were sectioned at 20 μm in the sagittal plane. From 12 levels, 7 sections were retained and processed. Aβ plaque load was quantified in sections doubly stained by immunohistochemistry with 6E10 antibody (N-terminal human Aβ, aa 1–16) and by ThioflavinS. Astrocytes were immunostained using a primary rabbit polyclonal antibody against GFAP (Dako, Glostrup, Denmark) and a fluorescently labelled secondary Cy2 antibody (Jackson ImmunoResearch Laboratories, West Grove, PA). Activated microglia were immunostained using a rat anti-murine CD11b primary antibody (AbD Serotec, Kidlington, UK) and a fluorescently labelled secondary Cy3 antibody (Jackson ImmunoResearch Laboratories). Astrocytes and microglia were labelled in a double incubation and sections were counterstained with DAPI to visualize cell nuclei.
Hippocampus and neocortex were measured separately, and the percent area of immunoreactivity or ThioflavinS staining per brain region was evaluated using automated image analysis software (Image Pro Plus, v. 6.2, Media Cybernetics, Bethesda, MD). Numbers and sizes of Aβ deposits were also counted automatically. Measurements from 5 sections per mouse were averaged and the individual mouse average was used to calculate the group means.
Contextual fear conditioning assay
A 2-day CFC paradigm was used, and the effects of acute and chronic dosing of EVP-0015962 were assessed. For the acute study, EVP-0015962 was prepared as a suspension in a vehicle of 4.5% DMSO, 15% Solutol® HS 15, 10% ethyl alcohol, and 70.5% water. On day 1, acutely dosed mice were administered 30 mg/kg, p.o. of EVP-0015962, 3 h prior to training. Animals in the food formulation groups were maintained on the compound formulated diets. The training session on day 1 consisted of the following sequence: 3 min acclimation to the test chamber (Kinder Scientific, Poway, CA), 2 s 1.5 mA foot shock, 2 min wait period, another 2 s 1.5 mA foot shock, followed by a final 1 min wait period. On day 2, testing consisted of returning the animals to the chamber for 5 min and the total time spent freezing was measured by an automated software system (Kinder Scientific). Freezing was defined as the absence of movement (i.e., zero beam breaks) and was measured every second for the 5-min trial duration (maximum freezing score of 300). Percentage of time spent freezing for each mouse was calculated using the formula: (number of seconds with zero beam breaks/300) X 100.
Statistics
In vivo data were expressed and graphed as mean ± SEM. Group differences for Aβ peptide levels, APP processing fragments, histological measures, and behavioral measures were assessed by one-way analyses of variance (ANOVA) followed by Newman-Keuls post-hoc testing or by Student’s t-test. Differences in aggregated Aβ were assessed by non-parametric Kruskal-Wallis ANOVA followed by Dunn’s test since the groups did not have equal variances. The level of significance was set at p < 0.05 in all tests.
Competing interests
EnVivo Pharmaceuticals, Inc. has a financial interest in EVP-0015962 and funded this research. KR, KMF, LH, ZT, FA, WL, SH, MJM, DS, ZY, HH, SN, MW, DC, J-FB, EF, MA, DF, LL, HP, RC, and GK received compensation and stock and/or options as current or former employees of EnVivo Pharmaceuticals, Inc. GS, BD, VV, HZ, and EP received compensation as consultants of EnVivo Pharmaceuticals, Inc. LS was an employee of Amorfix, Inc. and received compensation from EnVivo Pharmaceuticals, Inc. BHP and DH are employees of QPS Austria GmbH and received compensation from EnVivo Pharmaceuticals, Inc.
Authors’ contributions
KR contributed to the experimental design and interpretation of all studies and wrote the manuscript. RC and GS designed EVP-0015962 and oversaw its synthesis. KMF, LH, ZT, FA, WL, SH, MM, DS, ZY, HH, DC, J-FB, EF, BDS, VV, LS, HZ, EP, BHP, DH, LL, and HP contributed to the design, execution and interpretation of the studies. SN and MW performed the bioanalytical analyses. DF performed statistical analyses and contributed to the preparation of the manuscript. MA, RC, and GK contributed to the experimental design and interpretation of studies. All authors have read and approved the final manuscript.