Efficacy of SPI-1865, a novel gamma-secretase modulator, in multiple rodent models

Introduction Modulation of the gamma-secretase enzyme, which reduces the production of the amyloidogenic Aβ42 peptide while sparing the production of other Aβ species, is a promising therapeutic approach for the treatment of Alzheimer's disease. Satori has identified a unique class of small molecule gamma-secretase modulators (GSMs) capable of decreasing Aβ42 levels in cellular and rodent model systems. The compound class exhibits potency in the nM range in vitro and is selective for lowering Aβ42 and Aβ38 while sparing Aβ40 and total Aβ levels. In vivo, a compound from the series, SPI-1865, demonstrates similar pharmacology in wild-type CD1 mice, Tg2576 mice and Sprague Dawley rats. Methods Animals were orally administered either a single dose of SPI-1865 or dosed for multiple days. Aβ levels were measured using a sensitive plate-based ELISA system (MSD) and brain and plasma exposure of drug were assessed by LC/MS/MS. Results In wild-type mice using either dosing regimen, brain Aβ42 and Aβ38 levels were decreased upon treatment with SPI-1865 and little to no statistically meaningful effect on Aβ40 was observed, reflecting the changes observed in vitro. In rats, brain Aβ levels were examined and similar to the mouse studies, brain Aβ42 and Aβ38 were lowered. Comparable changes were also observed in the Tg2576 mice, where Aβ levels were measured in brain as well as plasma and CSF. Conclusions Taken together, these data indicate that SPI-1865 is orally bioavailable, brain penetrant, and effective at lowering Aβ42 in a dose responsive manner. With this unique profile, the class of compounds represented by SPI-1865 may be a promising new therapy for Alzheimer's disease.


Introduction
Alzheimer's disease (AD) is a severe neurodegenerative disease that is defined by two pathological features, amyloid plaques and neurofibrillary tangles. Because amyloid plaques appear before the onset of clinicallydefined dementia symptoms, neurodegeneration and subsequent cognitive impairment are hypothesized to be a downstream consequence of β-amyloid (Aβ) peptide dysregulation [1][2][3]. Aβ peptides are small fragments cleaved from a much larger integral membrane protein, the amyloid precursor protein (APP). In the AD cascade, APP is cleaved initially by β-secretase (BACE), leaving the C99 fragment in the membrane, which is then cleaved by gamma-secretase, an aspartyl protease complex [4,5]. Gamma-secretase continues to make sequential cleavages every three to four amino acids [6][7][8][9], resulting in Aβ fragments ranging in size from 49 to fewer than 34 amino acids [10,11]. Much of the focus in AD research has been on Aβ 42 , since it has been shown to be the most amyloidogenic and neurotoxic fragment [12][13][14]. More recently, Aβ 43 has also been shown to have these detrimental properties [15]. To test the hypothesis that lowering Aβ 42 levels may slow the progression of or prevent AD, multiple amyloid-targeted therapeutic approaches have been developed and moved into human clinical trials. These include Aβ clearancedirected immunotherapies as well as inhibitors of BACE or gamma-secretase enzyme activities, both of which are required for Aβ production.
To prevent the production of these neurotoxic Aβ peptides, researchers have focused on developing small molecule inhibitors of BACE and gammasecretase. In preclinical animal models, in vivo administration of gamma secretase inhibitors led to severe side effects, including an increased number of goblet cells in the intestine and decreased intrathymic differentiation and lymphocyte development [16][17][18]. These adverse events were found to be the result of inhibiting gamma-secretase's ability to process other substrates, specifically NOTCH [19][20][21], which is critical for cell development and differentiation [22]. Similar adverse events were also observed in recent clinical trials of semagacestat and avagacestat, further suggesting that complete inhibition of gamma-secretase is not a viable approach [23][24][25]. Much remains unknown about the approach to prevent Aβ production through BACE inhibition, as a subset of those molecules are currently in human clinical trials [26].
The discovery of multiple structural classes of compounds that modulate gamma-secretase activity, instead of inhibiting it, offers the potential promise of avoiding the mechanism-based adverse events observed with gamma-secretase inhibitors. Gamma-secretase modulators (GSMs) are observed to decrease the production of the more amyloidogenic Aβ 42 peptide, while preserving total Aβ levels and sparing gamma-secretase cleavage of the other substrates, such as NOTCH. Modulation allows the initial cleavage of substrates, but alters the processivity of the enzyme by shifting the production of Aβ peptides to the shorter, non-amyloidgenic forms without affecting the total level [27]. A first generation GSM, Flurizan from Myraid Genetics was tested in a Phase 3 trial. However, the compound is a weak modulator (IC 50 = 250 μM), lacks brain penetration, and produced side effects [10,28]. The compound failed to show efficacy, and development was halted in 2008 [29]. Since the approach of gamma modulation has not been adequately tested in humans, it is still believed that a more potent, drug-like compound could be a viable therapeutic approach.
Here we describe the in vivo pharmacology for a novel series of compounds, represented by SPI-1865 ( Figure 1). All compounds within the series that have been tested in vivo show a PK/PD relationship in rodents as robust as SPI-1865 (data not shown). The compound scaffold is derived from an initial hit, a triterpene glycoside (SPI-014), which was discovered through a screen of a natural products library for compounds with GSM properties [30]. Following a comprehensive and focused research effort, compounds like SPI-1865 were identified [31,32]. The compounds in this series, including SPI-1865, have a novel and proprietary structure, as well as the unique effect on the Aβ profile of lowering both Aβ 42 and Aβ 38 , without effecting Aβ 40 in cellular systems [33]. Using immunoprecipitation/mass spectrometry (IP/MS) analysis of conditioned media from Satori compound-treated versus control 2B7 cells, it was observed that the total Aβ levels are maintained with concomitant lowering of Aβ 38 and Aβ 42 and increases in Aβ 37 and Aβ 39 [34]. Furthermore, since increasing substrate levels do not result in an IC 50 shift, it is likely that SPI-1865 binds to the gamma-secretase complex as do other GSMs [35][36][37] instead of the APP substrate [33,34]. In the studies described here, the effects of SPI-1865 on Aβ 38 , Aβ 40 and Aβ 42 in both wildtype and transgenic animals were examined. The Aβ changes observed in these studies reflect the changes observed in our cellular systems, resulting in a decrease of both Aβ 38 and Aβ 42 in all three models, suggesting that SPI-1865 maintains its GSM properties in vivo.

Materials and methods
Test compound SPI-1865 was prepared in a manner as described in Bronk et al., patent WO2011109657 A1 20110909. Merck GSM1 was prepared as described in Madin et al., patent WO2007116228 A120071018.
Cell culture and compound treatment CHO-2B7 cells (Mayo Clinic) are Chinese hamster ovary cells stably transfected with human βAPP 695wt [38,39]. The cells were cultured in Ham's F12 media (Thermo Fisher SH30026.01, Waltham, MA, USA) supplemented with 10% FBS, 0.25 mg/mL Zeocin and 90 ug/mL penicillin/streptomycin at 37°C in a 5% CO 2 atmosphere. For compound treatment, cells were plated in 96-well plates at a density of 1.0 × 10 5 cells/mL and allowed grow to 100% confluence over two days. Test compounds in dimethyl sulfoxide (DMSO) were diluted 100-fold directly into the media before being adding to the cells. Immediately prior to adding compound-containing media to the cells, the cells were washed once with 1X PBS. Conditioned media from CHO-2B7 cells were collected after 5 hrs of treatment and the levels of Aβ peptides were assessed as described below. Ab in vitro assay measurement Conditioned media were collected after 5 hrs of treatment and diluted with one volume of MSD blocking buffer (1% BSA in MSD wash buffer). MSD Human (6E10) Aβ 3-Plex plates, which are pre-spotted with three separate spots in each well containing capture antibodies against the unique C-terminal ends of Aβ 38 , Aβ 40 and Aβ 42 , respectively, were blocked with MSD blocking buffer for one hour. Samples were transferred to the blocked plates with 6E10 detection antibody and incubated for 2 hrs at room temperature with orbital shaking followed by washing and reading according to the manufacturer's instructions (SECTOR ® Imager 2400 Meso Scale Discovery, Gaithersburg MD, USA).

In vivo study methods
The animal handling and procedures were performed either at Agilux Laboratories in Worcester, MA, USA, or Cerebricon in Kuopio, Finland. All animal handling and procedures were conducted in full compliance to AAALAC International and NIH regulations and guidelines regarding animal care and welfare. These protocols were reviewed and approved by Agilux's or Cerebricon's respective Institutional Animal Care and Use Committees (IACUC) prior to any activities involving animals.
Female transgenic mice (Tg2576, 3 months of age; n = 20), wild-type male CD-1 mice (six weeks of age; n = 8 to 12) or wild-type male Sprague Dawley rats (200 to 225 g body weight; n = 12) were utilized to assess in vivo efficacy.
For wild-type rat and mouse studies, all animals were acclimated to the test facility for a minimum of two days prior to initiation of the study. Compounds were dosed orally in 10:20:70 ethanol/solutol/water or 10:20:70 ethanol/cremaphor/water via oral gavage. Samples were harvested at either at 6 or 24 hrs post dose for Aβ and compound exposure measurements. Blood samples were collected into K 2 ethylenediaminetetraacetic acid (EDTA) and stored on wet ice until processed to plasma by centrifugation (3,500 rpm at 5°C) within 30 minutes of collection. Each brain was dissected into three parts: left and right hemispheres and cerebellum. Brain tissues were rinsed with ice cold PBS (without Mg 2+ or Ca 2+ ), blotted dry and weighed. Plasma and cerebella were analyzed for parent drug via liquid chromatography/tandem mass spectrometry (LC/MS/MS). Parent drug levels were compared to a standard curve to establish the plasma and brain levels.
In the transgenic studies, compound was dosed in 10:20:70 ethanol/solutol/water via oral gavage. Samples were harvested at 24 hrs post dose for Aβ and compound exposure measurement. The mice were subjected to cisterna magna puncture and collection of cerebrospinal fluid (CSF) (approximately 5 μl per mouse). Individual CSF samples were flash-frozen on dry ice and stored at -80°C. Thereafter, the mice were subjected to cardiac puncture and blood samples were collected into K 2 EDTA tubes and stored on wet ice until processed to plasma by centrifugation (2,000 g at 4°C for 10 minutes) within 30 minutes of collection. The plasma was aliquoted for both Aβ measurements and parent drug using LC/MS/ MS. Both sets of tubes were frozen at -80°C. The brains were perfused with non-heparinized saline (the blood flushed away) and removed carefully. Brains were rinsed with ice cold PBS (without Mg 2+ or Ca 2+ ), blotted dry, dissected on ice into three pieces (left and right hemisphere and cerebellum). Samples were frozen in liquid nitrogen immediately prior to storage at -80°C. Cerebella were analyzed for parent drug via LC/MS/MS. Parent drug levels were compared to a standard curve to establish the plasma, brain and CSF levels.

Rodent Ab determination
This protocol is a modification of protocols described by Lanz and Schachter [40] and Rogers et al. [41,42]. Frozen hemispheres were weighed into tared homogenization tubes (MP Biomedicals#6933050 for rat; MP Biomedicals, Solon, OH, USA) and (Simport#T501-4AT; Simport, Beloeil, Qc, Canada) containing one 5-mm stainless steel bead (Qiagen#69989 for mouse). For every gram of brain, 10 mLs of either 6 M guanidine hydrochloride (wild-type rat and mouse) or 0.2% diethyl amine in 50 mM NaCl (transgenic mouse) was added to the brain-containing tubes on wet ice. Rat hemispheres were homogenized for one minute and mouse hemispheres were homogenized for 30 seconds at the 6.5 setting using the FastPrep-24 Tissue and Cell homogenizer (MP Biomedicals#116004500). Homogenates were rocked for 2 hrs at 4°C, then pre-cleared by ultracentrifugation at 100,000 × g for one hour at 4°C. Pre-cleared wild-type rat and mouse homogenates were concentrated over solid phase extraction (SPE) columns (Oasis HLB 96-well SPE plate 30 μm, Waters#WAT058951; Waters Corp., Milford, MA, USA). Briefly, SPE columns were prepared by wetting with 1 mL of 100% methanol followed by dH 2 0 using vacuum to pull liquids through. Brain homogenates were then added to the prepared columns (1.0 mL from rat and 0.7 mL from wild-type mouse). Columns were washed twice with 1 mL of 10% methanol followed by two 1 mL washes with 30% methanol. Labeled eluent collection tubes (Costar cluster tubes #4413; Corning Inc., Corning, NY, USA) were placed under SPE columns and samples were eluted under very mild vacuum with 300 μL of 2% NH 4 OH/ 90% methanol. Eluents were dried to films under vacuum with no heat in a speed vacuum microcentrifuge. Films were resuspended in 150 μL of Meso Scale Discovery (MSD, Gaithersburg, MD, USA) blocking buffer (1% BSA in MSD wash buffer) for one hour at room temperature with occasional vortexing. Transgenic mouse plasma (50 μL) was extracted in 500 μL of 6 M guanidine hydrochloride briefly at room temperature and then 450 μL was concentrated over SPE columns and dried to films as described above. Transgenic plasma films were resuspended in 225 μL of MSD blocking buffer. Pre-cleared transgenic mouse brain homogenates were diluted and neutralized as follows. A volume of 45 μL of pre-cleared transgenic mouse brain homogenates were diluted into 450 μL of blocking buffer and were neutralized with 5 μL of 0.5 M Tris pH 6.8. For Aβ 38 , Aβ 40 and Aβ 42 measurements, MSD 96well multi-spot Human/Rodent (4G8) Aβ triplex ultrasensitive ELISA plates, which are pre-spotted with three separate spots in each well containing capture antibodies against the unique C-terminal ends of Aβ 38 , Aβ 40 and Aβ 42 , respectively, were blocked with MSD blocking buffer for one hour at room temperature with orbital shaking. A volume of 25 μL of neat resuspended wildtype rat or mouse brain homogenates were added in duplicates to the blocked 3-plex Aβ MSD plates with SULFO-TAG 4G8 antibody (MSD). Diluted and neutralized transgenic mouse brain homogenates, neat resuspended transgenic plasma samples or transgenic mouse CSF samples (diluted 1:10 in MSD blocking buffer) were added as described above to blocked MSD 96-well multi-spot Aβ triplex ultra-sensitive ELISA plates with SULFO-TAG 6E10 antibody (MSD). The Aβ 3-Plex plates were incubated for 2 hrs at room temperature with orbital shaking followed by washing and reading according to the manufacturer's instructions (SECTOR ® Imager 2400, MSD). The average Aβ concentrations from duplicate measurements of each animal were converted to percent vehicle values and the treatment group averages were statistically compared by analysis of variance (ANOVA). Statistical significance was defined as P < 0.01 in all experiments.

SPI-1865 decreases Ab 42 and Ab 38 in 2B7 cells
CHO-2B7 cells, which over-express human wild-type APP, were treated with increasing concentrations of SPI-1865. Conditioned media from CHO-2B7 cells were collected after 5 hrs of treatment and the levels of Aβ peptides were assessed using the MSD 3-Plex assay for Aβ 42 , Aβ 40 and Aβ 38 . As shown in Figure 2, SPI-1865 reduces both Aβ 38 and Aβ 42 with an IC 50 of 259 nM and 106 nM, respectively. Aβ 40 was found to have an IC 50 of 2.8 μM, resulting in > 20-fold selectivity for Aβ 42 over Aβ 40 . Total Aβ only decreased at doses where cytotoxicity was observed, indicating that SPI-1865 is capable of modulating gamma-secretase processivity, not inhibiting enzyme activity.

SPI-1865 reduces Ab 42 and Ab 38 with a single oral dose in Sprague Dawley rats
The compound was assessed for efficacy in Sprague Dawley rats. SPI-1865 has a delayed Tmax and half-life in excess of 24 hrs following a single oral dose in rats (Table 1). Based on this profile, compound efficacy was examined using a single oral dose and tissues were   40 and Aβ 42 (pg Aβ/g of brain) for the vehicle-treated group of rats were 312 ± 21, 3,094 ± 192 and 682 ± 61, respectively. As shown in Figure 3, at all doses, a significant lowering of brain Aβ 42 and Aβ 38 levels was observed compared to vehicle-treated animals. Aβ 40 was significantly reduced at only the highest dose by 22 ± 5% (average percent lowering ± standard error of the mean, SEM). The decreases in Aβ 42 levels were dose-responsive and correlated with the exposures in both brain and plasma. In the rats dosed with 10 mg/kg SPI-1865, brain levels reached 2.8 ± 0.3 μM and plasma levels were 3.3 ± 0.1 μM, which resulted in a lowering of Aβ 42 by 21 ± 6% relative to vehicle control. The compound levels increased with the 30 mg/kg dose to 11 ± 1 μM in the brain and 8.5 ± 0.3 μM in the plasma and resulted in a 37 ± 5% decrease in Aβ 42 . In the 100 mg/kg dose group, brain levels reach 33 ± 2 μM and plasma levels were 14 ± 1 μM of SPI-1865, leading to an Aβ 42 reduction of 50 ± 5%. Similar changes occurred with brain Aβ 38 levels, resulting in dose-responsive reductions of 26 ± 5, 36 ± 3 and 47% ± 5 upon the administration of 10, 30 and 100 mg/kg SPI-1865, in that order. These data demonstrate that SPI-1865 is capable of modulating gamma-secretase in vivo and results in a similar Aβ profile as observed in vitro.

Efficacy in Sprague Dawley rats is improved following multiple day dosing
To further investigate the effects of SPI-1865 in the rat, a multiple-day study was performed. Male Sprague Dawley rats were orally dosed once a day for six days with 10, 30 or 60 mg/kg of SPI-1865. The average amounts  of Aβ 38 , Aβ 40 and Aβ 42 (pg Aβ/g of brain) for the vehicle-treated group of rats were 98 ± 47, 2,690 ± 92 and 840 ± 32, respectively. As shown in Figure 4, the 10 mg/kg dose resulted in brain levels of 4.4 ± 0.2 μM, plasma levels of 8.0 ± 0.4 μM, and approximately a 25% reduction in both brain Aβ 38 and Aβ 42 levels with no significant alteration in Aβ 40 as compared to vehicle control. The 30 mg/kg dose lowered both Aβ 38 and Aβ 42 levels in the brain by roughly 44%, with brain exposures of 16 ± 1 μM and plasma levels 13 ± 1 μM, once again without significantly affecting Aβ 40 . At the highest dose, Aβ 42 was reduced by 66 ± 1% at exposures of approximately 45 μM (brain) and 19 ± 1 μM (plasma). At this highest dose, Aβ 40 was significantly lowered by 26 ± 2%. For all doses, Aβ 38 levels were lowered to a similar degree as Aβ 42 . This demonstrates that multiple-day administration of SPI-1865 can result in higher compound exposures and enhanced lowering of Aβ 42 levels.

SPI-1865 reduces Ab 42 and Ab 38 in both brain and plasma of Tg2576 mice
The use of transgenic mouse models, which overexpress human APP, allows for the measurement of Aβ peptides in three compartments: the brain, plasma and CSF. In this study, three month old female Tg2576 mice (n = 20 per group) were administered SPI-1865 for six days, which is sufficient to reach steady state exposures based on previous murine pharmacokinetic analysis (data not shown). The mice received 10, 30, 60 or 90 mg/kg of SPI-1865 orally once a day for six days or a positive control, Merck GSM-1, orally once on day six. Samples were harvested 24 hrs post dose for SPI-1865 and 6 hrs post dose for GSM-1. CSF and blood were directly harvested and the brain was perfused prior to collection.
As shown in Figure 5, dose-responsive decreases in Aβ 38 and Aβ 42 were observed in plasma with SPI-1865. In the brain, there was a trend towards lowering Aβ 38 and Aβ 42 levels, but the changes did not reach statistical significance. Aβ 42 seemed to be somewhat lower in the 90 mg/kg dose group in the CSF, but the change was also statistically insignificant. The average amounts of Aβ 38 , Aβ 40 and Aβ 42 in the brain (pg Aβ/g of brain), for the vehicle-treated group of Tg mice were 4,202 ± 547, 44,052 ± 5,262 and 6,470 ± 812, respectively. In the plasma of the vehicle-treated Tg2576 mice, average Aβ 38 , Aβ 40 and Aβ 42 levels were found to be 412 ± 21 6,266 ± 251 and 1,294 ± 66 pg Aβ/g and 4,384 ± 446, 35,137 ± 4,111 and 5,057 ± 785 pg Aβ/g in the CSF, respectively. At the highest dose, an Aβ 42 reduction of 76 ± 2% in the plasma was observed, with a plasma exposure of 6.4 ± 0.5 μM as measured 24 hrs post the final dose. The 10 mg/kg-treated group had no significant effect on any brain Aβ levels with drug levels in brain of 0.5 μM and plasma drug levels of 1 μM. The 30 and 60 mg/kg treatment groups had more pronounced reductions, which correlated with the increased brain and plasma exposures. For example, in the 60 mg/kg group, compound levels reached 5.4 ± 0.5 μM in the plasma and 3.9 ± 0.4 μM in the brain that resulted in a decrease in Aβ 38 by 50 ± 2.4% and Aβ 42 by 71 ± 2% in the plasma, respectively. Merck GSM-1 significantly reduced Aβ 42 and increased Aβ 38 levels. In this study, large variability was observed in both the brain and CSF samples, possibly due to blood contamination in both tissues. The variability was also observed in the vehicle and positive control groups, indicating that this is not a consequence of SPI-1865 treatment.
Most interestingly, when examining the total data set from the transgenic mouse study, the plasma reductions stand out as being significantly greater than those in brain and CSF. It has been reported that the Tg2576 mice do produce peripheral Aβ and SPI-1865 could be impacting those sources directly [43]. Other studies with SPI-1865 indicate that the compound is highly protein bound, with 97.6% bound in the plasma and approximately 99.9% in the brain. When we calculate the free fraction in plasma from the 90 mg/kg dose group where there was 76% lowering of plasma Aβ 42 , the free plasma fraction is calculated to be 154 nM which exceeds the IC 50 of 106 ± 19 nM. If we perform a similar comparison focusing on plasma levels using the 30 mg/kg group, where 58% lowering was observed, the free plasma levels are found to be 60 nM, which with inherent assay variability is reasonably close to the in vitro IC 50 of 106 nM. These data suggest that free plasma concentrations of SPI-1865 are driving the efficacy that was observed in the periphery. In the brain, the highest dose resulted in a C CSF AE Levels  non-significant 30% lowering and a calculated brain free fraction of 6.9 ± 6 nM, which is in the range of the IC 25 of 33 ± 18 nM. With the difficulty in measuring brain protein binding accurately when the free fraction is quite small, the high variability in measuring Tg2576 brain Aβ 42 in vivo and the fact that there were no significant data points for brain Aβ 42 lowering, we needed another study investigating free brain versus free plasma concentrations as the cause of Aβ 42 reduction in the central compartment.

SPI-1865 lowers Ab 42 and Ab 38 in wild-type CD-1 mice
To further examine the effects of SPI-1865, an improved study design in wild-type mice, which have much less variability in Aβ levels, was performed. However, due to very low plasma levels of Aβ in wild-type mice, we are unable to measure efficacy in plasma. Based on the halflife of 8 hrs and a Tmax of approximately 4 hrs in the mouse, samples were collected 6 hrs post dose to optimize the measurement of Aβ and compound exposures. A twice-daily (BID) dosing regimen for six days was selected to ensure steady state was attained and maximize the time SPI-1865 was engaged with the enzyme complex. The results of this treatment are shown in Figure 6. The average amounts of Aβ 38 , Aβ 40 and Aβ 42 in the brain (pg Aβ/g of brain) for the vehicle-treated group of wild-type mice were 324 ± 8.5, 4,465 ± 98 and 1,012 ± 18, respectively. A significant lowering of brain Aβ 42 was observed at all doses. Brain Aβ 38 was also significantly lowered at the two top doses and a slight, but significant lowering of 11% was observed for brain Aβ 40 levels at the 30 mg/kg dose. The total compound exposures in the 50 mg/kg BID group were found to be 7.6 ± 0.3 μM in the plasma and 22 ± 1 μM in the brain where brain Aβ 42 was decreased by 47 ± 2%. The 30 mg/kg BID treatment reduced brain Aβ 42 by 39 ± 2% with a total plasma exposure of 4.4 ± 0.3 μM and brain levels of 8.7 ± 1 μM. In the 15 mg/kg BID-treated animals, brain exposures of 3.6 ± 0.1 μM and a plasma exposure of 3.4 ± 0.2 μM to reduce brain Aβ 42 by 22%. The percent Aβ 42 lowering of each dosed animal was plotted against the plasma ( Figure 6B) and brain exposures ( Figure 6C) in μM. Best-fit lines were superimposed over the data revealing the roughly hyperbolic nature of the PK/PD relationship in BID-dosed mice. From these lines, the plasma and brain exposure levels required for 25% and 50% Aβ 42 lowering were extrapolated. We found that 25% Aβ 42 lowering correlated with a 2.3 μM exposure in the plasma and 3.8 μM exposure in the brain and the exposure levels associated with 50% Aβ 42 lowering were 5.9 and 25.2 μM in the plasma and brain, respectively. We compared brsain free-fraction levels to the efficacy and once again found the exposures to be below the IC 25 (33 ± 18 nM) even at the highest dose where 22 ± 1 nM was achieved and corresponded with 47% reduction. Taken together with the Tg2576 data, we cannot determine if brain free-fraction plays a role in SPI-1865 efficacy. Conversely, plasma free-fraction, and the observed efficacy in the brain, appear to be strongly correlated. At the highest dose, plasma free-fraction was calculated to 182 nM, which resulted in a 47% reduction in brain Aβ 42 , while the 30 mg/kg dose group was determined to have a plasma free-fraction of 105 nM (39% lowering of brain Aβ 42 ) and the low dose, a plasma free-fraction of 82 nM (22% lowering of brain Aβ 42 ). With an in vitro Aβ 42 IC 50 of 106 ± 19 nM and an IC 25 of 33 ± 18 nM, there is a significant correlation between plasma free-fraction and brain efficacy. At all doses, the plasma free-fraction was slightly higher than the amount of compound expected to induce the same response in vitro, but the numbers are within the range of variability. Taken together these data further support the Tg2576 data that plasma freefraction of SPI-1865 correlates with brain efficacy.

Discussion
The studies described here demonstrate that SPI-1865 is a novel modulator of gamma-secretase, which is capable of lowering both Aβ 42 and Aβ 38 in multiple animal models. In Sprague-Dawley rats, the compound effectively lowered Aβ 42 levels following a single dose. With once a day dosing for six days, the Aβ 42 and Aβ 38 response was improved. Given that the half-life of SPI-1865 is greater than 24 hrs in the rat, improved efficacy was anticipated, since accumulation of compound in the plasma occurs following multiple days of dosing. However, the compound levels in the brain did not show the same degree of accumulation with multiple dosing, even though brain efficacy was improved. Based on multiple in vitro studies (data not shown), SPI-1865 is not a substrate for transporters nor does it cause CYP induction, either of which may lower the brain exposure. Together these data suggest that sustained exposure to SPI-1865 levels over multiple days contributes to enhanced efficacy of the compound to modulate gamma-secretase.
SPI-1865 was next examined for efficacy in 3-monthold, female Tg2576 mice. These mice produce human APP with the SWE mutation via the prion promoter [44]. The transgene expression has been shown to be highest in the brain, but the transgene is also expressed in peripheral tissues [43]. The peripheral production of Aβ in the Tg2576 mouse may influence the effects of SPI-1865 on brain Aβ versus plasma Aβ levels when comparing efficacy in transgenic and wild-type models.
Data from this study indicate that SPI-1865 readily crosses the blood-brain barrier, as evident in the levels of SPI-1865 measured in brain. The ability to measure efficacy in both the brain and plasma compartments in the transgenic mice allows us to probe whether total or free concentrations of compound drives efficacy. If total compound were responsible for efficacy in both tissue compartments, one would expect to see similar decreases in Aβ 42 and Aβ 38 in both the plasma and brain in the Tg2576, as the exposure in each compartment is similar. However, what is observed is a significantly improved lowering of Aβ 42 and Aβ 38 in the plasma relative to the brain. The most likely reason for the difference in efficacy between the two compartments is the level of free drug available in the plasma versus the brain. The level of compound binding in the plasma is 97.6% based on in vitro studies, leaving 2.4% free compound to interact with gamma-secretase. This is a significantly higher free fraction than the estimates for free fraction in the brain, where protein binding was measured at approximately 99.9%. When the average free fraction for each compartment is compared for the 90 mg/kg dose (brain total = 6.9 μM; plasma total = 6.4 μM), the plasma free-fraction levels of 153 nM exceed the in vitro IC 50 of SPI-1865 (106 ± 19 nM) and plasma Aβ 42 levels were lowered by 76 ± 2%. The average brain free-fraction was only 6.9 nM, below the IC 25 of 33 ± 18 nM, and while a trend toward lowering of Aβ 42 was observed, it was not statistically significant . It is important to note the high variability of Aβ 42 measurement in the Tg2576 brains impacted the ability to see statistically significant decreases in brain Aβ 42 , even with a large number of animals per group (n = 20) and may affect our determination of free versus bound in the brain. Moreover, it is challenging to get an accurate measure of free concentration in brain when brain protein binding is greater than 99%. Even in the face of these technical challenges, the data presented here indicate that the plasma free-fraction of SPI-1865 correlates most closely with Aβ lowering in that compartment.
In the Tg2576 study, the changes in CSF Aβ levels were examined along with the plasma and brain. While treatment with SPI-1865 trended towards a decrease in brain Aβ 38 and Aβ 42 and significantly lowered the plasma levels of Aβ 42 at the three highest doses tested, in the CSF the Aβ peptide levels were not significantly decreased. There are several possible explanations for the lack of a significant effect. The high variability in CSF Aβ measurements within this study may mask an effect on the Aβ 38 and Aβ 42 levels. In addition, the timing of sample collection relative to dosing may have been more optimal to assess plasma changes than to assess changes in brain and CSF Aβ levels .
In wild-type mice, we further investigated the in vivo activity of SPI-1865 once steady state plasma levels were achieved. With a half-life of 8 hrs and a Tmax of approximately 4 hrs in the mouse (Table 1), a six-day BID dosing study was designed to measure exposure and Aβ levels 6 hrs post dose, allowing maximal efficacy to be observed by capturing exposures near the maximal plasma concentration and providing nearly continuous engagement with the enzyme. This is different from the Tg2576 study where once-a-day dosing was utilized for six days and samples were harvested 24 hrs post dose. In this study, there was much less variability of Aβ levels among animals than the Tg2576 mice and six days of BID dosing resulted in a significant lowering of brain Aβ 42 at all doses compared to vehicle-dosed animals. Plasma and CSF levels were not measured in this experiment. When we examined plasma free-fraction The pharmacokinetic and pharmacodynamic properties of SPI-1865 were assessed in mouse and rat model systems. *Statistically significant lowering of the indicated β-amyloid (Aβ) species compared to vehicle-treated animals (P < 0.01). CSF, cerebrospinal fluid. Values are given as the mean ± standard error of the mean.
versus efficacy, an Aβ 42 lowering of 22 ± 4%, 39 ± 3% and 47 ± 2% occurred as the doses increased, and these changes corresponded with free plasma concentrations of 82, 105 and 182 nM, respectively. Together, this wild-type mouse study in combination with the Tg2576 model and rat data, demonstrate the ability of SPI-1865 to lower both Aβ 38 and Aβ 42 in vivo.

Conclusions
Taken together the data shown in these studies demonstrate that SPI-1865 is an efficacious gamma-modulator in vivo (all in vivo data are summarized in Table 2). This is demonstrated in multiple rodent models using a single dose, multiple-day dosing or a multiple-day BID dosing regimen. These data indicate that SPI-1865 is orally bioavailable, brain penetrant, and has a different Aβ profile from other GSMs, both in vitro and in vivo. SPI-1865 lowers Aβ 42 and Aβ 38 while sparing Aβ 40 levels. This novel GSM pharmacology is dose-responsive, driven by the plasma free-fraction of the compound and is capable of reducing both Aβ 42 and Aβ 38 levels in APP over-expressing mice. Overall, SPI-1865 exemplifies the unique Aβ profile and good drug-like properties of SPI compounds, and further suggests they may be novel therapeutic approaches for Alzheimer's disease. Authors' contributions RML carried out the processing and analysis of the study samples, aided in study design, and contributed to the drafting of the manuscript. JAD designed the studies and coordinated with the contract research organizations to ensure studies were performed as designed and contributed to the drafting of the manuscript. TDM developed the in vitro assay used for compound assessment and produced the in vitro data shown here and edited manuscript drafts. WFA, NOF, JLH, RS and BSB designed and produced the chemical matter utilized in the described studies. JJ and JI provided support and advice for these studies. BT oversaw the biology efforts, ensured data quality and edited manuscript drafts. All authors read and approved the final manuscript.