The GSM BPN-15606 as a Potential Candidate for Preventative Therapy in Alzheimer’s Disease

Abstract

Background:

In the amyloid hypothesis of Alzheimer’s disease (AD), the dysregulation of amyloid-β protein (Aβ) production and clearance leads to amyloid deposits, tau tangles, neuronal loss, and cognitive dysfunction. Thus far, therapies targeting the enzymes responsible for Aβ production have been found ineffective or having significant side effects.

Objective:

To test whether a γ-secretase modulator, BPN-15606, is an effective disease-modifying or preventative treatment in the PSAPP mouse model of AD.

Methods:

We treated pre-plaque (3-month-old) and post-plaque (6-month-old) PSAPP AD transgenic mice for 3 months and examined behavioral, biochemical, and pathological end points.

Results:

BPN-15606 attenuated cognitive impairment and reduced amyloid plaque load, microgliosis, and astrogliosis associated with the AD phenotype of PSAPP mice when administered to pre-plaque (3-month-old) but was ineffective when administered to post-plaque (6-month-old) mice. No treatment-related toxicity was observed.

Conclusion:

BPN-15606 appears efficacious when administered prior to significant pathology.

Keywords

Alzheimer’s disease amyloid-β BPN-15606 cognitive deficits preventative therapy PSAPP

INTRODUCTION

Alzheimer’s disease (AD) is a degenerative brain disease, and the most common form of dementia in the elderly, which affects about 5.8 million individuals in the US in 2019 (Alzheimer’s Association Report). This number will only continue to grow as the proportion of population aged 65 years or older increases due to extending lifespan resulting from better medical care; currently, 10% of people in this group have AD. Deaths from this disease increased 89% from 2000 to 2014, making AD the sixth leading cause of death in the US [1, 2]. The national cost of the disease is estimated to be $290 billion for 2019, which includes significant unpaid time away from work for assisted care from friends and family of the patient, costs for providing long-term health care, and for currently available palliative treatments which are limited to temporary and mild alleviation of cognitive and behavioral symptoms in a subset of patients [3, 4]. Currently there is no cure, effective prevention, or therapeutic regimen for altering the course of the underlying disease process and thus AD remains an enormous unmet clinical need [5]. Since pathological alterations in the brain appear to commence some 10–20 years prior to cognitive symptoms, there theoretically should be a significant window of time for administering pathologically-targeted therapies capable of delaying the course of the disease [1, 2].

The pathological features of AD include amyloid-β (Aβ) plaques, which accumulate extracellularly, and neurofibrillary tangles, which form intracellularly, in numerous regions of the cerebral cortex and hippocampus in a staged temporal-spatial pattern [6]. The Aβ-containing neuritic plaques result from oligomerization of Aβ₄₂, which is a proteolytic cleavage product of the amyloid-β protein precursor (AβPP). AβPP is processed primarily through two distinct metabolic pathways, one of which is referred to as being “amyloidogenic” and produces Aβ₄₀, the most abundant secreted Aβ peptide, as well as the more fibrillogenic Aβ₄₂, the major peptide found in both diffuse amyloid deposits and in compact neuritic plaques [5]. Familial AD-linked mutations include those within the AβPP transmembrane domain as well as missense mutations within PS1 and PS2 (presenilin 1 and 2) which harbor the catalytically active subunit of γ-secretase, the enzyme that ultimately proteolyzes a carboxy-terminal AβPP fragment to form an altered ratio of the longer Aβ peptides, resulting in increased Aβ₄₂/Aβ₄₀ peptide level ratio (approximately two-fold) over that seen in the absent of these FAD-linked mutations [7 –10].

The amyloidogenic processing of AβPP commences with cleavage by BACE1 (beta-site APP cleaving enzyme 1) which generates a soluble amino-terminal fragment sAβPPβ and a membrane-bound carboxyl fragment APP-C99, also known an APP-CTFβ, which is further proteolyzed by γ-secretase, producing an AICD (AβPP intracellular domain) as well as a number of Aβ peptide variants including Aβ₃₇, Aβ₃₈, Aβ₄₀, and Aβ₄₂. Alternatively, AβPP can be processed through a non-amyloidogenic pathway initiated via α-secretase-mediated cleavage resulting in a soluble amino-terminal fragment termed sAβPPα, along with a membrane bound APP-CTFα fragment, which is further processed by γ-secretase producing non-amyloidogenic products, including p3 and AICD [11].

Since Aβ generation requires the sequential proteolysis of AβPP by BACE1 and γ-secretase, these two aspartyl proteases have been thoroughly explored as therapeutic targets for AD. Many BACE1 inhibitors have failed human trials due to inability to display a benefit in symptomatic disease (lanabecestat, clinical trial NCT02783573, and verubecestat, clinical trial NCT01953601). Atabecestat trials have recently been discontinued due to side effects involving elevated liver enzymes [12]. However, other BACE1 inhibitors are still in clinical trials, such as elenbecestat and CNP520 (clinical trials NCT03036280 and NCT03131453, respectively). There also have been many failed γ-secretase inhibitor (GSI) programs. Semagacestat reached phase III clinical trials and was discontinued after patients exhibited worsening memory and increased risk of skin cancer possibly due to the inhibition of γ-secretase-mediated cleavage of Notch [13]. The following class of GSIs, dubbed “Notch-sparing” because of their reported selectivity for APP-CTFs included begacestat [14] (discontinued for unstated reasons), and avagacestat (which was reported to cause cognitive worsening, along with gastrointestinal and dermatologic complications [15]).

Our approach of testing a γ-secretase modulator (GSM) instead of a GSI negates the issue of Notch-mediated side effects, as it preserves γ-secretase function and does not affect NICD (notch intracellular domain) generation via inhibition of epsilon-site proteolysis of Notch by γ-secretase. BPN-15606 selectively and potently attenuates Aβ₄₂ and to a lesser extent Aβ₄₀, while elevating the levels of less fibrillogenic Aβ₃₈ and Aβ₃₇ [16].

γ-Secretase is composed of 4 membrane protein subunits: presenilin (PS1 or PS2), nicastrin, Pen-2, and Aph-1. The enzyme is activated by Pen-2-mediated auto-proteolytic cleavage of PS1 or PS2 within the exon 9-coded segment. PS1 is cleaved to generate a PS1-NTF and a PS1-CTF, which form the catalytic center of the γ-secretase complex [17, 18]. Aph-1 and nicastrin assemble into a subcomplex, which then bind to the CTF of presenilin [19, 20]. Mutation of PS1 at Asp257 and Asp385 abolishes both the auto-proteolytic (presenilinase) and γ-secretase activity; thus, it is the catalytic subunit of γ-secretase [21]. A PS1 mutant with the exon 9 domain excised (PS1ΔE9) is constitutively active and does not require the Pen-2-mediated cleavage [10, 22]. PSAPP, a transgenic mouse line that expresses this mutation, as well as the Swedish APP mutation, APP KM670/671NL, which enhances Aβ production, develops accelerated AD-like pathology, including plaque formation and cognitive deficits [23]. In addition, it exhibits synaptic abnormalities, disruption of neuronal connections, and neuroinflammation [24 –27]. Using this transgenic mouse model, the extent of preventative or disease-modifying aspects of the chronic administration of BPN-15606 as potential therapy for AD were explored.

METHODS

Compounds

The novel GSM BPN-15606, (S)-N-(1-(4- fluorophenyl)ethyl)-6-(6-methoxy-5-(4-methyl1H-i midazol-1-yl)pyridin-2-yl)-4-methylpyridazin-3-am ine, was synthesized at Albany Molecular Research Institute (AMRI) Albany, NY using the methods reported in UCSD-MGH published patent application [28]. In addition, all chemical and pharmacological properties BPN-15606 were recently published [16] and provided extensive pharmacokinetics (PK) parameters in several species including mice, as well as brain exposures following oral administration at the dose used in this study (10 mg/kg).

In vitro cell-based assay

The SH-SY5Y-APP cell line was derived by transfecting a human neuroblastoma (SH-SY5Y) cell line with a plasmid expressing wild-type human APP₇₅₁ cDNA. 1 day prior to treatment, the cells were split at 75K/well in a 96 well plate. The next day, either vehicle, 2 concentrations of BPN-15606 (3.17 nM or 10 nM), or a GSI (semagacestat, 100 nM) were added in triplicate. 24-h post-treatment, media was collected and analyzed via Meso Scale Discovery (MSD) 6E10 multiplex kit according to manufacturer’s instructions. Aβ_total levels were measured using a mAb-specific sandwich ELISA assay.

Mice

B6C3-Tg(APPswe,PSEN1dE9)85Dbo/Mmjax (PSAPP) from Jackson labs were used in this study at 3 or 6 months of age (referred to as pre-plaque and post-plaque groups, respectively). All mice were group housed with up to 5 mice per cage in standard 12-h light cycle, with free access to food and water. Mice were randomly assigned to treatment groups (n = 9–11 per group). All experimental procedures were reviewed and approved by IACUC at UC San Diego. Although only female mice were used in this study, previous studies demonstrated similar exposure levels between male and female mice following administration of BPN-15606 [16]. Power analyses were conducted based on previously published studies with BPN-15606 [16] to determine the minimal sample size required to detect a difference of >25% with a probability level of <0.05. This analysis revealed that 9–11 animals per condition were required to observe statistical significance.

Pharmacological treatment and administration

Gamma-secretase modulator BPN-15606 was administered for 3 months via oral gavage at 10 mg/kg/day, 5 days a week. The drug solution was prepared weekly in 80% PEG400/0.1% Tween 20 (v/v) solution.

Behavior

All behavioral and locomotor tests were performed at the end of the treatment period while the mice were still being administered the treatments.

Total activity memory

The Total Activity Memory (TAM) test shows memory retention through habituation to the testing environment. An animal that remembers the testing environment will explore less than an animal with memory impairments. During the test, for the first 3 days, the animal was placed for 10 min into the Kinder Motor Monitor Cage rack system, which has a 7×15 beam configuration to monitor movement. The more beams the animal breaks, the more the animal moves. There is a two-day break before Day 4, when the animal is placed into the same chamber for 10 min and its activity is analyzed. The first day of testing is also analyzed for spontaneous activity, which can be used as a baseline for the Morris water maze test.

Rotarod

In this performance task, the ability of a mouse to balance on a rotating rod at increasing speeds is measured over two days. There are 5 training trials on day 1, varying in speed, starting at 0 RPM ramping up to 10 RPM, then 0–20 RPM, and finally 0–40 RPM. Day 2 has 7 trials ramping to 40 RPM. The animals are allowed a minimum of 5–10 min rest between trials.

Morris water maze

The Morris water maze test was performed in room temperature water. Non-toxic Tempera poster paint is added to the water to increase the contrast between the water and the animal. The mouse was gently placed in the tank at water level, facing the pool wall, at one of the two start positions. The pool contained discrete cues on 4 inner sides, visible to the mice. A video camera was mounted on the ceiling directly above the pool and was used in conjunction with a videotracking system (AnyMaze by San Diego Instruments) to record the swim path of each mouse. The tracking program was launched as soon as the animal is in the water and stopped automatically once the mouse reaches the platform. If the animal cannot locate the platform in the water maze after 90 s, it will be immediately guided to it. This test is performed on the last 2 weeks of drug treatment. On Day 1, the animal is acclimated to the pool and shown where to locate the visible platform to escape the water for four consecutive 90-s trials, with a brief rest between trials starting at the same start location. On Days 2 and 3, the animal will be alternated between two starting points with the same visible escape platform and four 90-s trials each with a brief rest. Days 4 and 5 consist of the same pattern of testing as Days 2 and 3, but with an invisible platform. The animals are then rested for 48 h. Days 6 and 7 are identical to Days 4 and 5. On Day 8, the probe test is administered. The first 40-s trial begins at the same start location as days two through seven, with the platform removed completely, the animal should remember where to go and stay in the vicinity. The second trial is also 40 s in duration; however, the visible platform is placed and the animal starts at the second starting point. Animals that have memory impairments will demonstrate a greater latency to reach the platform location, as well as longer distance traveled. Mice that failed to reach the platform in the visible probe trial were excluded from further analyses.

Tissue and organ collection

The mice were sacrificed under isoflurane-induced anesthesia the day after the conclusion of behavioral testing. The brain was dissected and split into hemispheres. The right hemisphere was fixed in 4% PFA in PBS at 4°C for 24 h, then placed in 30% sucrose until sectioning. Sections were cut coronally at 30μm using a freeze-slide microtome, then placed into cryoprotectant (30% ethylene glycol, 30% glycerol, 40% 1× PBS) in –20°C until immunohistochemistry analysis. The left hemisphere was dissected into the hippocampus, cortex, and the rest of the brain, which were snap-frozen in liquid nitrogen for biochemical assays. Organs such as liver, gastrointestinal tract, heart, thymus, lungs, kidneys, spleen, and spinal cord were fixed in 10% formalin and sent to Dr. Kent Osborn in the Pathology Core of the Animal Care Program Diagnostic Laboratory at UCSD for histopathology screening.

Biomarkers and pathology

Quantitation of Aβ₃₈, Aβ₄₀, and Aβ₄₂ in the various brain extractions were determined using the MSD 4G8 multiplex kit according to manufacturer’s instructions. Aβ_total levels were measured using a mAb-specific sandwich ELISA assay [16]. Detergent-soluble brain extractions were performed in RIPA Buffer (Cell Signaling) with the addition of Halt Protease Inhibitor Cocktail (Thermo Fisher Scientific) and PMSF. The samples were homogenized using IKA T8.01 Ultra Turrax Homogenizer, then spun at 14,000 g for 15 min at 4°C. The resulting supernatant was aliquoted and stored at –80°C. Formic acid-soluble extracts were prepared from RIPA-extracted pellets by homogenization in 70% cold formic acid on ice and centrifugation at 100,000 g for 1 h at 4°C. The supernatant was neutralized to pH 7.5–8 with 10 volumes of 2M Tris base, pH 11.5. The samples’ protein concentration was measured via BCA assay (for RIPA-extracted samples) or Bradford Assay (for formic acid-extracted samples) and equal amounts of each sample were loaded onto MSD plates. We also measured Aβ₄₂ levels in formic acid-soluble brain extracts using a different Aβ₄₂-specific sandwich ELISA assay described previously which utilizes a different Aβ₄₂-specific mAb than the one used in the MSD 4G8 multiplex kit [29].

Immunofluorescence

To determine Aβ plaque load, serial coronal sections from PSAPP and wildtype mice (n = 5–6 per group) were washed in MiliQ water, mounted on Fisherbrand Superfrost Plus microscope slides, and stained in 1% Thioflavin S solution. To support our findings, 82E1 (Clontech (10323, Takara Bio USA; 1:1000) staining was also performed on free-floating sections. For assessment of neuronal and synaptic changes, sections were immunolabeled with antibodies against microtubule associated protein-2 (MAP2) (MAB378, Millipore; 1:500) to label neuronal cell bodies and dendrites, GFAP (PA5-16291, Invitrogen; 1:250) and Iba1 (019-19741, Wako; 1:1500). Primary antibody staining was identified with fluorescently-tagged secondary antibodies at 1:1000. Images were taken with Leica LiveImage fluorescence microscope at 10X using the same exposure across all images. Quantification of staining was performed double-blinded, using the ImageJ software from NIH. The threshold was adjusted consistently across all images. Particle count, area, and size were obtained using the Analyze Particles function in hippocampal or cortical regions of interest. The series of sections of each animal was averaged and grouped accordingly prior to statistical analysis. The number of series of section per animal ranged from 5–7 sections, and 3–5 animals were analyzed per group.

Toxicity

Formalin fixed tissues representing major organ systems were submitted to Dr. Kent Osborn, in the Pathology Core of the Animal Care Program Diagnostic Laboratory at UCSD, for light microscopic examination and assessment for abnormalities that might be related to the experimental treatment. The tissue included heart, aorta, lung, trachea, liver, gall bladder, kidney, spleen, thymus, spinal cord, vertebrae (bone), perivertebral muscle, bone marrow, esophagus, stomach, small intestine, cecum, colon, and pancreas. Six-micron paraffin-embedded, H&E-stained sections of these tissues from each animal were examined.

Statistical analysis

All experiments were performed blind-coded. Values in the figures are expressed as means±S.E.M. To determine the statistical significance, values were compared using ANOVA or student’s t-test. The differences were considered to be significant if p-values were <0.05. Data was analyzed using GraphPad Prism (GraphPad Software, La Jolla, CA, USA).

RESULTS

In vitro cell-based assay

To determine the effect of the BPN-15606 on the levels of various Aβ fragment and on total Aβ secretion, we utilized a human neuroblastoma cell line, SH-SY5Y, expressing wild-type human APP₇₅₁ cDNA [29]. The cells were treated with either vehicle or two concentrations of BPN-15606 (3.17 nM or 10 nM). We have found that relative to baseline, BPN-15606 treatment led to a trend towards an increase in levels of Aβ₃₈ at 10 nM and 3.17 nM (Fig. 1A), and significantly decreased levels of Aβ₄₀ at 10 nM (Fig. 1B), and of Aβ₄₂ at 10 nM and 3.17 nM (Fig. 1C) in cell culture supernatant, without altering the total levels of Aβ (Fig. 1D). These results along with previously published studies on this and related GSMs [16, 29] prompted us to pursue a comprehensive in vivo animal study.

Fig.1

GSM treatment attenuates secretion of Aβ₄₀ and Aβ₄₂ species without affecting total Aβ. Levels of Aβ₃₈ (A), Aβ₄₀ (B), Aβ₄₂ (C), and total Aβ (D) in media of SHSY5Y-APP cells after 24-h treatment with GSM, GSI or vehicle. All Aβ measurements utilized triplex MSD ELISA. n = 3. Values presented as mean±SEM, normalized to vehicle.

Cognitive performance

We then investigated whether chronic administration of BPN-15606 led to improvements in learning and memory in a mouse model of AD. For this, we utilized 3- or 6-month-old PSAPP mice (pre-plaque-formation and post-plaque-formation groups, respectively) and treated them with 10 mg/kg/day of BPN-15606 for 3 months. In the Morris water maze learning and memory task, BPN-15606-treated pre-plaque PSAPP mice exhibited significantly reduced latency and distance to reach platform compared to vehicle-treated PSAPP mice (Fig. 2A, B). In fact, their performance was not significantly different from wild-type mice except during Day 4, where they were first introduced to the hidden platform; after Day 4, they performed at wild-type mouse levels. Post-plaque mice (6-month-old at start of treatment) did not exhibit significantly different performance in distance (Fig. 2C) and latency (Fig. 2D) to platform compared to untreated transgenic mice. The rotarod performance task (Supplementary Figure 1A, B) and total activity memory test (Supplementary Figure 1C, D) did not reveal any significant differences in balance or activity between any of the groups. The results suggest that BPN-15606 administration led to better learning and memory in PSAPP pre-plaque mice but not post-plaque mice.

Fig.2

GSM treatment is correlated with enhanced performance in Morris water maze in 3-month-old (pre-plaque) PSAPP (TG) mice. Pre-plaque GSM-treated mice (TG GSM) performed significantly better than vehicle-treated (TG) in the distance (A) and time (B) to platform measurements. Post-plaque GSM-treated TG mice performed as well as vehicle-treated TG mice in the distance (C) and time (D) to platform measurements, significantly worse than vehicle-treated wild-type mice (WT). The mice were rested for 48 h between Days 5 and 6 of the test. n = 9–10. Values presented as mean±SEM. *p>0.05, represents significant difference from WT.

Assessment of Aβ load

To determine whether chronic administration of BPN-15606 leads to reduction of plaque load in pre-plaque mice, we stained coronal brain slices with the 82E1 antibody, which recognizes soluble and fibrillar Aβ but not holo-AβPP (Fig. 3A, B). Immunohistochemistry revealed that BPN-15606 reduced the number of plaques relative to vehicle treatment in the cortex, but not in the hippocampus (Fig. 3C) in the pre-plaque mouse group. The average plaque size was also decreased in both the cortex and the hippocampus in the BPN-15606-treated pre-plaque mouse group as compared to vehicle-treated mouse group (Fig. 3D). In the post-plaque BPN15606-treated mice, there was no significant difference between the number (Fig. 3E) or size (Fig. 3F) of plaques compared to the vehicle-treated group.

Fig.3

GSM treatment ameliorates Aβ plaque load and decreases the size and number of Aβ plaques in 3-month-old (pre-plaque) PSAPP (TG) mice. Representative images of pre-plaque (A) vehicle- and (B) GSM-treated TG mice. 82E1 staining revealed that the number of plaques is decreased in cortex (C), and the average size of plaques in hippocampus and cortex (D) were significantly decreased in GSM-treated pre-plaque TG mice versus vehicle-treated pre-plaque TG mice. Scale bar, 1 mm. No differences in number (E) or size (F) of plaques were observed the brains of GSM-treated post-plaque TG mice as compared to vehicle-treated post-plaque TG mice. n = 5. Values presented as mean±SEM. p < 0.05.

Brain levels of Aβ

Next, we wanted to determine whether chronic administration of BPN-15606 leads to the increase of Aβ₃₈, and reduction of Aβ₄₀ and Aβ₄₂ in the brain before and after plaque onset. To do this, we measured Aβ fragments in detergent-soluble and formic acid-soluble brain fractions in both vehicle-and BPN-15606-treated pre-plaque and post-plaque PSAPP mice. Detergent-soluble brain fractions of BPN-15606-treated pre-plaque mice contained significantly reduced levels of Aβ₄₂ and a reduced Aβ₄₂/Aβ₄₀ ratio compared to the vehicle-treated group, while Aβ₃₈ and Aβ₄₀ levels remained unchanged (Fig. 4A). Formic acid-soluble brain fractions from BPN-15606-treated compared to vehicle-treated pre-plaque mice contained only slightly lower levels of all Aβ fragments that did not reach statistical significance and exhibited an unchanged Aβ₄₂/Aβ₄₀ ratio (Fig. 4B). We also measured Aβ₄₂ levels in formic acid-soluble brain extracts from BPN-15606-treated compared to vehicle-treated pre-plaque mice using a sandwich ELISA [29] and also showed lower levels of in the BPN-15606-treated group (∼25%) but this also did not reach statistical significance (data not shown), in agreement with the MSD assays. In the post-plaque treated mice, all Aβ fragment levels were similar between the two treatment groups in both the detergent-soluble (Supplementary Figure 2A) and formic acid-soluble extracts (Supplementary Figure 2B).

Fig.4

GSM treatment reduces detergent-soluble Aβ₄₂ and Aβ₄₂/Aβ₄₀ ratio in the cortex of pre-plaque PSAPP mice. A) RIPA-extracted brain fractions from pre-plaque GSM-treated PSAPP mice have significantly less Aβ₄₂ and a reduced Aβ₄₂/Aβ₄₀ ratio compared to pre-plaque vehicle-treated PSAPP mice. B) Formic-acid soluble brain fractions show no significant differences between treated and untreated groups. Values are expressed as mean±SEM. N = 9–10. p < 0.05.

Microgliosis

In AD brain, Aβ deposits are often associated with microgliosis, the response of microglia to pathogenic insults, which is characterized by an increased number of activated microglia at the site of the insult [30]. This feature is also captured by the PSAPP mouse model [27, 31]. We have assayed the degree of microgliosis through staining mouse brain slices with Iba1, ionized calcium binding adaptor molecule 1, which is a microglia- and macrophage-specific calcium binding protein [32]. Compared to wild-type mice, pre-plaque, vehicle-treated PSAPP mice exhibited a greater percentile of Iba1-positive area and activated microglial counts in cortex (Fig. 5B, D) and hippocampus (Fig. 5C, E). BPN-15606 treatment significantly reduced the Iba1-positive area and the number of activated microglia in both brain regions. In post-plaque mice, BPN-15606 treatment did not reduce microgliosis in either the cortex (Supplementary Figure 3B, D) or hippocampus (Supplementary Figure 3C, E), which parallels our findings regarding its inability to reduce plaque load in this particular AD transgenic mouse aged group.

Fig.5

GSM treatment reduces microgliosis in pre-plaque PSAPP mice. Representative images of Iba1 stain (A). Scale bar represents 500μm. BPN-15606 treatment reduces percentage of Iba1-positive area in cortex (B) and hippocampus (C) of pre-plaque PSAPP mice compared to vehicle-treated pre-plaque PSAPP mice. Microglial count is also reduced after BPN-15606 treatment in cortex (D) and hippocampus (E) compared to vehicle. Values are expressed as mean±SEM. n = 4–5. p < 0.05.

Astrogliosis

Astrogliosis is a common feature of neurodegenerative disease and is defined by an abnormal activation and increased numbers of astrocytes [33, 34]. The PSAPP mouse model also exhibits astrogliosis [27 , 35]. To determine the effect of BPN-15606 on this aspect of neuroinflammation, we stained mouse brain slices for GFAP, using an antibody against glial fibrillary acidic protein, which is expressed in astrocytes [36]. We observed an elevation of GFAP-positive staining in PSAPP transgenic mice compared to wild-type mice, which was reduced in BPN-15606-treated mice (Fig. 6). Specifically, while GFAP-positive area was significantly higher in the cortex of the PSAPP transgenic mice compared to wild-type mice (Fig. 6B), BPN-15606 treated-mice demonstrated a significant reduction in GFAP-positive staining compared to the untreated transgenic group. We did not observe changes in GFAP-positive staining across groups in the hippocampus (Fig. 6C). Similarly, astrocyte size was increased in cortex and hippocampus of PSAPP transgenic mice, indicating astrocyte reactivity (Fig. 6D, E), and BPN-15606 treatment reduced astrocyte size to normal, but only in the cortex. In contrast, the post-plaque PSAPP transgenic mouse group did not exhibit any benefit from BPN-15606 treatment, retaining an elevated GFAP-positive area in the cortex compared to wild-type mice (Supplementary Figure 4B, C) and displaying commensurate levels of GFAP in hippocampus as wild-type mice (Supplementary Figure 4D, E).

Fig.6

GSM reduces astrogliosis in the cortex of pre-plaque PSAPP mice. A) Representative images of GFAP staining in cortex and hippocampus of pre-plaque PSAPP mice. Scale bar represents 500μm. B) Percentage of GFAP-positive area in cortex and (C) hippocampus. D) Average size of individual astrocytes in cortex and (E) hippocampus. Values are expressed as mean±SEM. n = 4–5. p < 0.05.

Dendritic deficits

While dendritic deficits are a feature of AD and PSAPP mice [26], we did not observe significant differences between MAP2-positive staining in cortex or hippocampus of pre-plaque (Supplementary Figure 5A, B) or post-plaque (Supplementary Figure 5D, C) mice.

Toxicity

None of the treatments were associated with weight loss or impaired weight gain (Supplementary Figure 6). The analyzed tissues were all essentially normal, with no indication for apparent toxicity. Lesions that were noted are considered incidental, either as common background lesions in laboratory mice, secondary effects of euthanasia, or lesions in individual animals that, while not common, are not considered to be significant in the context of this study or related to the treatment.

DISCUSSION

This is the first study to comprehensively address the efficacy of a 3-month oral treatment with a potent γ-secretase modulator, BPN-15606, on behavioral, pathological, and biochemical features of AD in the PSAPP mouse model at pre-plaque and the post-plaque ages. We have found that BPN-15606 treatment in vitro increases levels of secreted Aβ₃₈ and decreases the levels of Aβ₄₀ and Aβ₄₂. The longer forms of Aβ are more hydrophobic and fibrillogenic, and are the primary components of amyloid plaques [37]. Thus, BPN-15606 acts by shifting the proteolytic processing of AβPP (cleaving the β-CTF into more benign Aβ species and attenuating deposition of plaques).

We also show an effective intervention in disease progression when BPN-15606 is administered to PSAPP mice before plaque onset. In behavioral tasks, BPN-15606-treated mice performed similarly to age-matched wild-type mice. These cognitive benefits were correlated with reduced plaque pathology, as evidenced by immunohistochemistry, and with reduced Aβ₄₂ levels, as evidenced by ELISA: the level of detergent-soluble Aβ₄₂ fragments and the Aβ₄₂/Aβ₄₀ ratio was decreased in the cortex of 6-month-old mice when treated at the pre-plaque age of 3 months. Detergent-insoluble Aβ₄₂ levels showed a trend toward reduction after BPN-15606 treatment, though it was not statistically significant at the dose, duration, and route of administration; it is possible that using a dose higher than 10 mg/kg/day would lead to a further reduction in detergent-insoluble Aβ₄₂, paralleling our dose-response data from our previous publication on this compound [16].

The theory that alterations in the Aβ₄₂/Aβ₄₀ ratio play a critical role in AD pathogenesis has been supported by multiple studies [38]. While the primary component of an amyloid plaque is insoluble Aβ, soluble Aβ oligomers and cellular components that are expelled after neuronal cell death or injury are also powerful stimuli that lead to astrocytic activation and microglial recruitment; studying these signals and their effect on inflammation separately in vivo is challenging [39, 40]. With the reduction of plaques, we also saw reduced inflammation in terms of microgliosis and area of astrogliosis in the cortex, with a trend towards reduction in size of astrocytes after treatment. In the hippocampus, microgliosis, but not astrogliosis, was ameliorated by the BPN-15606 treatment. Our immunohistochemical data shows that the astrocytes in the brains of the transgenic mice are uniformly present throughout the hippocampus, unlike in the cortex, where they clump around plaque locations (Fig. 8A). Since BPN-15606 does not directly affect neuroinflammation to our knowledge, but rather indirectly ameliorates it through the reduction in plaques, non-plaque associated inflammation will therefore not be reduced with treatment. These results lead us to conclude that BPN-15606 is an effective preventative treatment in this transgenic mouse model.

Given that 6-month-old mice already have plaques, and that other memory tests, plaque measurements, and Aβ fragment levels in cortex did not reflect an effect from the treatment, BPN-15606 does not appear to have a disease-modifying effect in amyloid-laden mice. The reason why the brain Aβ measurements do not reflect any treatment-based differences is likely because in the PSAPP mouse model, rate of amyloid deposition from 6 to 9 months is significantly increased compared to 3 to 6 months [41]. At the dose tested, we observed significant mild to moderate reduction in the younger mouse group, in which treatment started at 3 months. If, according to Takeuchi et al., the amyloid burden in a PSAPP mouse brain is 1% at 3 months, 5% at 6 months, but rises to about 25% at 9 months, it appears that 10 mg/kg/day of BPN-15606 p.o. is insufficient to significantly interrupt the accelerated deposition of amyloid which occurs during this later time frame. This is consistent with our previous study which showed only about a 50% attenuation of plaque load when administered from 3–9 months in this same PSAPP model [16].

In light of multiple failures of γ-secretase-targeting therapies, which have primarily focused on inhibition strategies, we propose an alternative method of modulating the activity of γ-secretase via a GSM similar to BPN-15606, preferably one of its more potent (Ames negative) analogues which lack evidence of any potential genotoxicity (BPN-15606 recently showed positivity in the Ames assay against a single T98 salmonella strain T98 but only in the presence of rat S9 liver extracts; data not shown).

Given that γ-secretase has close to 90 different type 1 membrane protein potential substrates, which marks its involvement in multiple cell processes such as cell fate determination, neurite outgrowth, cell adhesion and migration, and synapse formation [42, 43], consideration to the enormous potential for side effects for any therapy that targets γ-secretase must be recognized. Using BPN-15606, we did not detect any treatment-associated abnormalities following a thorough toxicological analysis. Additionally, taking into account the substantial evidence that Aβ has important physiological functions, such as antimicrobial/antiviral activity [44, 45] and synapse regulation [43], aiming for blanket attenuation of all Aβ species may have multiple deleterious consequences. The data presented demonstrates the efficacy and safety of the γ-secretase modulator BPN-15606 as treatment for some features of AD-like pathophysiology in PSAPP transgenic mice, and only when used in a preventative manner (prior to significant Aβ plaque deposition). These results may help to explain, at least in part, the lack of efficacy of numerous anti-amyloid-targeted therapeutics which have thus far been administered primarily, if not exclusively, to prodromal subjects or patients already presenting with evidence of significant cerebral amyloidosis [46 –48].

Footnotes

ACKNOWLEDGMENTS

This work was supported by National Institute of Neurologic Disorders and Stroke/National Institute of Aging [Grant numbers U01-NS074501, U01-AG048986]; and Cure Alzheimer’s Fund (all to S.L.W.). The authors thank A. Adame, B. Spencer, J. Florio, and F. Sarsoza. Pathology was performed by Dr. Kent Osborn in the Pathology Core of the Animal Care Program Diagnostic Laboratory at UCSD. R.E.T and S.L.W. are shareholders and cofounders of a privately held company (Neurogenetic Pharmaceuticals) that holds rights to a γ-secretase modulator, distinct from the GSM BPN-15606 described herein, currently in clinical development. The contents do not represent the views of the U.S. Department of Veterans Affairs or the United States Government.

Authors’ disclosures available online ().

The supplementary material is available in the electronic version of this article: .

References

Bateman

, Xiong

, Benzinger

, Fagan

, Goate

, Fox

, Marcus

, Cairns

, Xie

, Blazey

, Holtzman

, Santacruz

, Buckles

, Oliver

, Moulder

, Aisen

, Ghetti

, Klunk

, McDade

, Martins

, Masters

, Mayeux

, Ringman

, Rossor

, Schofield

, Sperling

, Salloway

, Morris

, Dominantly Inherited Alzheimer Network (2012) Clinical and biomarker changes in dominantly inherited Alzheimer’s disease. N Engl J Med 367, 795–804.

Reiman

, Quiroz

, Fleisher

, Chen

, Velez-Pardo

, Jimenez-Del-Rio

, Fagan

, Shah

, Alvarez

, Arbelaez

, Giraldo

, Acosta-Baena

, Sperling

, Dickerson

, Stern

, Tirado

, Munoz

, Reiman

, Huentelman

, Alexander

, Langbaum

, Kosik

, Tariot

, Lopera

(2012) Brain imaging and fluid biomarker analysis in young adults at genetic risk for autosomal dominant Alzheimer’s disease in the presenilin 1 E280A kindred: A case-control study. Lancet Neurol 11, 1048–1056.

Alzheimer’s Association (2016) Alzheimer’s disease facts and figures. Alzheimers Dement 12, 459–509.

Collins

(2014) Taking a Toll on Families and the Economy: The Rising Cost of Alzheimer’s in America. National Institutes of Health, https://www.nih.gov/sites/default/files/institutes/olpa/20140226-senate-testimony-collins.pdf, February 26, 2014.

Wagner

, Tanzi

, Mobley

, Galasko

(2012) Potential use of gamma-secretase modulators in the treatment of Alzheimer disease. Arch Neurol 69, 1255–1258.

Serrano-Pozo

, Frosch

, Masliah

, Hyman

(2011) Neuropathological alterations in Alzheimer disease. Cold Spring Harb Perspect Med 1, a006189.

Borchelt

, Thinakaran

, Eckman

, Lee

, Davenport

, Ratovitsky

, Prada

, Kim

, Seekins

, Yager

, Slunt

, Wang

, Seeger

, Levey

, Gandy

, Copeland

, Jenkins

, Price

, Younkin

, Sisodia

(1996) Familial Alzheimer’s disease-linked presenilin 1 variants elevate Abeta1-42/1-40 ratio in vitro and in vivo. Neuron 17, 1005–1013.

Scheuner

, Eckman

, Jensen

, Song

, Citron

, Suzuki

, Bird

, Hardy

, Hutton

, Kukull

, Larson

, Levy-Lahad

, Viitanen

, Peskind

, Poorkaj

, Schellenberg

, Tanzi

, Wasco

, Lannfelt

, Selkoe

, Younkin

(1996) Secreted amyloid beta-protein similar to that in the senile plaques of Alzheimer’s disease is increased in vivo by the presenilin 1 and 2 and APP mutations linked to familial Alzheimer’s disease. Nat Med 2, 864–870.

Citron

, Eckman

, Diehl

, Corcoran

, Ostaszewski

, Xia

, Levesque

, St George Hyslop

, Younkin

, Selkoe

(1998) Additive effects of PS1 and APP mutations on secretion of the 42-residue amyloid beta-protein. Neurobiol Dis 5, 107–116.

10.

Ahn

, Shelton

, Tian

, Zhang

, Gilchrist

, Sisodia

, Li

(2010) Activation and intrinsic gamma-secretase activity of presenilin 1. Proc Natl Acad Sci U S A 107, 21435–21440.

11.

Zheng

, Koo

(2006) The amyloid precursor protein: Beyond amyloid. Mol Neurodegener 1, 5.

12.

Johnson J, Update on Janssen’s BACE Inhibitor Program, https://www.janssen.com/update-janssens-bace-inhibitor-program, May 17, 2018.

13.

Doody

, Raman

, Farlow

, Iwatsubo

, Vellas

, Joffe

, Kieburtz

, He

, Sun

, Thomas

, Aisen

, Alzheimer’s Disease Cooperative Study Steering Committee, Siemers

, Sethuraman

, Mohs

, Semagacestat Study Group (2013) A phase 3 trial of semagacestat for treatment of Alzheimer’s disease. N Engl J Med 369, 341–350.

14.

Martone

, Zhou

, Atchison

, Comery

, Xu

, Huang

, Gong

, Jin

, Kreft

, Harrison

, Mayer

, Aschmies

, Gonzales

, Zaleska

, Riddell

, Wagner

, Lu

, Sun

, Sonnenberg-Reines

, Oganesian

, Adkins

, Leach

, Clarke

, Huryn

, Abou-Gharbia

, Magolda

, Bard

, Frick

, Raje

, Forlow

, Balliet

, Burczynski

, Reinhart

, Wan

, Pangalos

, Jacobsen

(2009) Begacestat (GSI-953): A novel, selective thiophene sulfonamide inhibitor of amyloid precursor protein gamma-secretase for the treatment of Alzheimer’s disease. J Pharmacol Exp Ther 331, 598–608.

15.

Coric

, van Dyck

, Salloway

, Andreasen

, Brody

, Richter

, Soininen

, Thein

, Shiovitz

, Pilcher

, Colby

, Rollin

, Dockens

, Pachai

, Portelius

, Andreasson

, Blennow

, Soares

, Albright

, Feldman

, Berman

(2012) Safety and tolerability of the gamma-secretase inhibitor avagacestat in a phase 2 study of mild to moderate Alzheimer disease. Arch Neurol 69, 1430–1440.

16.

Wagner

, Rynearson

, Duddy

, Zhang

, Nguyen

, Becker

, Vo

, Masliah

, Monte

, Klee

, Echmalian

, Xia

, Quinti

, Johnson

, Lin

, Kim

, Mobley

, Rissman

, Tanzi

(2017) Pharmacological and toxicological properties of the potent oral gamma-secretase modulator BPN-15606. J Pharmacol Exp Ther 362, 31–44.

17.

Takasugi

, Tomita

, Hayashi

, Tsuruoka

, Niimura

, Takahashi

, Thinakaran

, Iwatsubo

(2003) The role of presenilin cofactors in the gamma-secretase complex. Nature 422, 438–441.

18.

Steiner

, Winkler

, Haass

(2008) Chemical cross-linking provides a model of the gamma-secretase complex subunit architecture and evidence for close proximity of the C-terminal fragment of presenilin with APH-1. J Biol Chem 283, 34677–34686.

19.

Kaether

, Capell

, Edbauer

, Winkler

, Novak

, Steiner

, Haass

(2004) The presenilin C-terminus is required for ER-retention, nicastrin-binding and γ-secretase activity. EMBO J 23, 4738–4748.

20.

Liu

, Waltz

, Woodruff

, Ouyang

, Israel

, Herrera

, Sarsoza

, Tanzi

, Koo

, Ringman

, Goldstein

LSB

, Wagner

, Yuan

(2014) Effect of potent γ-secretase modulator in human neurons derived from multiple presenilin 1–induced pluripotent stem cell mutant carriers. JAMA Neurol 71, 1481–1489.

21.

Berezovska

, Jack

, McLean

, Aster

, Hicks

, Xia

, Wolfe

, Kimberly

, Weinmaster

, Selkoe

, Hyman

(2000) Aspartate mutations in presenilin and gamma-secretase inhibitors both impair notch1 proteolysis and nuclear translocation with relative preservation of notch1 signaling. J Neurochem 75, 583–593.

22.

Lessard

, Wagner

, Koo

(2010) And four equals one: Presenilin takes the gamma-secretase role by itself. Proc Natl Acad Sci U S A 107, 21236–21237.

23.

Borchelt

, Ratovitski

, van Lare

, Lee

, Gonzales

, Jenkins

, Copeland

, Price

, Sisodia

(1997) Accelerated amyloid deposition in the brains of transgenic mice coexpressing mutant presenilin 1 and amyloid precursor proteins. Neuron 19, 939–945.

24.

Tsai

, Grutzendler

, Duff

, Gan

(2004) Fibrillar amyloid deposition leads to local synaptic abnormalities and breakage of neuronal branches. Nat Neurosci 7, 1181–1183.

25.

Rutten

, Van der Kolk

, Schafer

, van Zandvoort

, Bayer

, Steinbusch

, Schmitz

(2005) Age-related loss of synaptophysin immunoreactive presynaptic boutons within the hippocampus of APP751SL, PS1M146L, and APP751SL/PS1M146L transgenic mice. Am J Pathol 167, 161–173.

26.

Zhang

, Kuo

, Moghadam

, Monte

, Campbell

, Rice

, Sawchenko

, Masliah

, Rissman

(2016) Corticotropin-releasing factor receptor-1 antagonism mitigates beta amyloid pathology and cognitive and synaptic deficits in a mouse model of Alzheimer’s disease. Alzheimers Dement 12, 527–537.

27.

Matsuoka

, Picciano

, Malester

, LaFrancois

, Zehr

, Daeschner

, Olschowka

, Fonseca

, O’Banion

, Tenner

, Lemere

, Duff

(2001) Inflammatory responses to amyloidosis in a transgenic mouse model of Alzheimer’s disease. Am J Pathol 158, 1345–1354.

28.

Wagner

SL MW

, Tanzi

, Johnson

, Buckle

, Mayhew

, Herr

, Rynearson

(2016) Potent gamma-secretase modulators. Regents of the University of California and General Hospital Corporation. International Patent Classification: WO2016/070107.

29.

Kounnas

, Danks

, Cheng

, Tyree

, Ackerman

, Zhang

, Ahn

, Nguyen

, Comer

, Mao

, Yu

, Pleynet

, Digregorio

, Velicelebi

, Stauderman

, Comer

, Mobley

, Li

, Sisodia

, Tanzi

, Wagner

(2010) Modulation of gamma-secretase reduces beta-amyloid deposition in a transgenic mouse model of Alzheimer’s disease. Neuron 67, 769–780.

30.

Selkoe

(1994) Cell biology of the amyloid beta-protein precursor and the mechanism of Alzheimer’s disease. Annu Rev Cell Biol 10, 373–403.

31.

Schwab

, Klegeris

, McGeer

(2010) Inflammation in transgenic mouse models of neurodegenerative disorders. Biochim Biophys Acta 1802, 889–902.

32.

Ito

, Imai

, Ohsawa

, Nakajima

, Fukuuchi

, Kohsaka

(1998) Microglia-specific localisation of a novel calcium binding protein, Iba1. Brain Res Mol Brain Res 57, 1–9.

33.

Eng

, Ghirnikar

(1994) GFAP and astrogliosis. Brain Pathol 4, 229–237.

34.

Sofroniew

(2014) Astrogliosis. Cold Spring Harb Perspect Biol 7, a020420.

35.

McGowan

, Sanders

, Iwatsubo

, Takeuchi

, Saido

, Zehr

, Yu

, Uljon

, Wang

, Mann

, Dickson

, Duff

(1999) Amyloid phenotype characterization of transgenic mice overexpressing both mutant amyloid precursor protein and mutant presenilin 1 transgenes. Neurobiol Dis 6, 231–244.

36.

Bignami

, Eng

, Dahl

, Uyeda

(1972) Localization of the glial fibrillary acidic protein in astrocytes by immunofluorescence. Brain Res 43, 429–435.

37.

Selkoe

(2001) Alzheimer’s disease: Genes, proteins, and therapy. Physiol Rev 81, 741–766.

38.

Tanzi

, Bertram

(2005) Twenty years of the Alzheimer’s disease amyloid hypothesis: A genetic perspective. Cell 120, 545–555.

39.

D’Andrea

, Cole

, Ard

(2004) The microglial phagocytic role with specific plaque types in the Alzheimer disease brain. Neurobiol Aging 25, 675–683.

40.

Serrano-Pozo

, Muzikansky

, Gomez-Isla

, Growdon

, Betensky

, Frosch

, Hyman

(2013) Differential relationships of reactive astrocytes and microglia to fibrillar amyloid deposits in Alzheimer disease. J Neuropathol Exp Neurol 72, 462–471.

41.

Takeuchi

, Irizarry

, Duff

, Saido

, Hsiao Ashe

, Hasegawa

, Mann

, Hyman

, Iwatsubo

(2000) Age-related amyloid beta deposition in transgenic mice overexpressing both Alzheimer mutant presenilin 1 and amyloid beta precursor protein Swedish mutant is not associated with global neuronal loss. Am J Pathol 157, 331–339.

42.

Jurisch-Yaksi

, Sannerud

, Annaert

(2013) A fast growing spectrum of biological functions of gamma-secretase in development and disease. Biochim Biophys Acta 1828, 2815–2827.

43.

Puzzo

, Privitera

, Fa

, Staniszewski

, Hashimoto

, Aziz

, Sakurai

, Ribe

, Troy

, Mercken

, Jung

, Palmeri

, Arancio

(2011) Endogenous amyloid-beta is necessary for hippocampal synaptic plasticity and memory. Ann Neurol 69, 819–830.

44.

Eimer

, Vijaya Kumar

, Navalpur Shanmugam

, Rodriguez

, Mitchell

, Washicosky

, Gyorgy

, Breakefield

, Tanzi

, Moir

(2018) Alzheimer’s disease-associated beta-amyloid is rapidly seeded by herpesviridae to protect against brain infection. Neuron 100, 1527–1532.

45.

Gosztyla

, Brothers

, Robinson

(2018) Alzheimer’s amyloid-beta is an antimicrobial peptide: A review of the evidence. J Alzheimers Dis 62, 1495–1506.

46.

Henley

, Raghavan

, Sperling

, Aisen

, Raman

, Romano

(2019) Preliminary results of a trial of atabecestat in preclinical Alzheimer’s disease. N Engl J Med 380, 1483–1485.

47.

Egan

, Kost

, Voss

, Mukai

, Aisen

, Cummings

, Tariot

, Vellas

, van Dyck

, Boada

, Zhang

, Li

, Furtek

, Mahoney

, Harper Mozley

, Mo

, Sur

, Michelson

(2019) Randomized trial of verubecestat for prodromal Alzheimer’s disease. N Engl J Med 380, 1408–1420.

48.

Knopman

(2019) Lowering of amyloid-beta by beta-secretase inhibitors - some informative failures. N Engl J Med 380, 1476–1478.