PF-3084014 [(S)-2-((S)-5,7-difluoro-1,2,3,4-tetrahydronaphthalen-3-ylamino)-N-(1-(2-methyl-1-(neopentylamino)propan-2-yl)-1H-imidazol-4-yl)pentanamide] is a novel γ-secretase inhibitor that reduces amyloid-β (Aβ) production with an in vitro IC50 of 1.2 nM (whole-cell assay) to 6.2 nM (cell-free assay). This compound inhibits Notch-related T- and B-cell maturation in an in vitro thymocyte assay with an EC50 of 2.1 μM. A single acute dose showed dose-dependent reduction in brain, cerebrospinal fluid (CSF), and plasma Aβ in Tg2576 mice as measured by enzyme-linked immunosorbent assay and immunoprecipitation (IP)/mass spectrometry (MS). Guinea pigs were dosed with PF-3084014 for 5 days via osmotic minipump at 0.03 to 3 mg/kg/day and exhibited dose-dependent reduction in brain, CSF, and plasma Aβ. To further characterize Aβ dynamics in brain, CSF, and plasma in relation to drug exposure and Notch-related toxicities, guinea pigs were dosed with 0.03 to 10 mg/kg PF-3084014, and tissues were collected at regular intervals from 0.75 to 30 h after dose. Brain, CSF, and plasma all exhibited dose-dependent reductions in Aβ, and the magnitude and duration of Aβ lowering exceeded those of the reductions in B-cell endpoints. Other γ-secretase inhibitors have shown high potency at elevating Aβ in the conditioned media of whole cells and the plasma of multiple animal models and humans. Such potentiation was not observed with PF-3084014. IP/MS analysis, however, revealed dose-dependent increases in Aβ11-40 and Aβ1-43 at doses that potently inhibited Aβ1-40 and Aβ1-42. PF-3084014, like previously described γ-secretase inhibitors, preferentially reduced Aβ1-40 relative to Aβ1-42. Potency at Aβ relative to Notch-related endpoints in vitro and in vivo suggests that a therapeutic index can be achieved with this compound.
Amyloid-β (Aβ) peptide is the primary component of senile plaques (Glenner and Wong, 1984) and is the protein product of a gene [amyloid precursor protein (APP)] whose mutation can result in early-onset Alzheimer's disease. The intersection of this genetic and pathologic evidence has led to a strong focus on Aβ as a major culprit in the etiology of Alzheimer's disease. A number of compounds have advanced to the clinic with the goal of either reducing production of this peptide (e.g., β- or γ-secretase inhibitors) or increasing its clearance from the brain (e.g., Aβ vaccines or monoclonal antibodies). Of these approaches, γ-secretase has yielded the greatest diversity of chemical tools that enable the study of Aβ pharmacodynamics in animal models and humans. Bioavailable small-molecule inhibitors of γ-secretase from various chemical series have been shown to rapidly reduce Aβ levels in brain, cerebrospinal fluid (CSF), and plasma from wild-type mice (Yohrling et al., 2007), rats (Best et al., 2005; El Mouedden et al., 2006; Lanz and Schachter, 2006), guinea pigs (Anderson et al., 2005; Lanz et al., 2006), and multiple mutant APP transgenic mouse lines (Lanz et al., 2003, 2004, 2006; Barten et al., 2005; Martone et al., 2009; Portelius et al., 2009; Willuweit et al., 2009). Chronic dosing of a γ-secretase inhibitor in the Tg2576 mouse, which harbors the Swedish familial mutation in a human APP transgene (commonly referred to as APPSw), has been shown to dramatically inhibit age-related plaque deposition (Best et al., 2007). In humans, multiple studies with γ-secretase inhibitors have shown efficacy in reducing Aβ levels in plasma (Siemers et al., 2005, 2007; Martone et al., 2009) and in reducing central Aβ synthesis as measured in CSF with isotope labeling methods (Bateman et al., 2009).
One liability inherent in reducing γ-secretase activity is the potential to affect processing of Notch, a γ-secretase substrate that plays a role in cellular differentiation. In particular, chronic treatment with high doses of γ-secretase inhibitors has produced hyperplasia of intestinal goblet cells in rodents and dogs (Wong et al., 2004; Hyde et al., 2006) and reductions in circulating B-cell populations in preclinical models and patients (Wong et al., 2004; Hyde et al., 2006; Henley et al., 2009). Thus close attention must be paid to the therapeutic window to achieve efficacy while minimizing side effects. Fortuitously, inhibition of Notch has an upside; γ-secretase inhibitors have shown efficacy in multiple cancer models by virtue of this activity (Konishi et al., 2007; Plentz et al., 2009; Rao et al., 2009; Watters et al., 2009).
The present work describes a novel tetralin imidazole γ-secretase inhibitor, PF-3084014, which has single-digit nanomolar potency for Aβ reduction in vitro. These in vitro assays demonstrating effects on Aβ production and Notch cleavage will be presented, followed by in vivo pharmacokinetics and pharmacodynamics in guinea pigs and Tg2576 mice. As treatment with various γ-secretase inhibitors have produced paradoxical increases in plasma Aβ at low concentrations in both preclinical species and humans (Lanz et al., 2004, 2006; Siemers et al., 2005, 2007; Martone et al., 2009), and one compound has demonstrated elevations in rat brain Aβ (Burton et al., 2008), in vivo studies in guinea pigs were carefully designed to enable observation of all compartments in the presence of a wide range of drug concentrations both acutely and at steady state. The potentiation phenomenon is much more limited with PF-3084014, although the compound does show some forms of γ-secretase modulation as revealed in the Aβ peptide profiles in multiple preclinical models.
Materials and Methods
In Vitro Assays.
A cell-free assay was developed by using a crude P2 membrane preparation derived from human HeLa cells. The substrate was affinity-purified recombinant human APP-C100-FLAG peptide produced in Escherichia coli. PF-3084014 (Fig. 1) was dosed at log intervals from 0.1 nM to 10 μM. Reactions were assembled and incubated as described previously (Li et al., 2000), and Aβ1-40 was detected by using a DELFIA-based immunoassay (Lanz et al., 2006).
Human H4 cells stably transfected with APPSw were used to test compounds in a whole-cell assay, and Aβ1-X was measured as described previously (Lanz et al., 2006). Compounds were dosed at concentrations ranging from 0.01 to 313 nM.
Fetal thymic organ cultures (FTOCs) were prepared for assessment of compound effects on Notch processing. Intact thymus lobes were dissected out of C57BL/6 mouse embryos at embryonic day 14. For each drug condition, three thymus lobes were placed on a Millicell-CM 0.4 μm well insert (Millipore Corporation, Billerica, MA) in a six-well cellculture plate containing 1.5 ml/well of Dulbecco's modified Eagle's medium containing 10% fetal bovine serum (FBS), penicillin/streptomycin, and glutamate (Invitrogen, Carlsbad, CA). Compound was added in a final concentration of 1% dimethyl sulfoxide. Cultures were incubated for 5 days at 37°C in 5% CO2. Lymphocytes were then harvested by rubbing the lobes between the frosted portions of two glass slides to break apart the lobes without damaging the individual cells. Lymphocytes were collected in 1.5 ml of phosphate-buffered saline (PBS) and centrifuged at 350g for 5 min. Each pellet was resuspended in a mixture containing 2.5 μl of each FACS antibody in a final volume of 100 μl in Dulbecco's PBS containing 2% FBS and 0.02% sodium azide. Antibodies raised to mouse CD4 and CD8a (FITC-labeled), the T-cell marker CD44 (APC-labeled), the B-cell marker B220 (peridinin chlorophyll protein complex-labeled), and the B- and T-cell marker CD25 (phycoerythrin-labeled) were ordered from BD Pharmingen (San Diego, CA). Samples were incubated for 45 min at 4°C, then centrifuged at 350g for 5 min. Each pellet was resuspended in fresh 250 μl of Dulbecco's phosphate-buffered solution containing 2% FBS and 0.02% sodium azide and read by FACS.
In Vivo Studies.
All animal treatment protocols were approved by Pfizer's Institutional Care and Use Committee and were compliant with Animal Welfare Act regulations. Three-month-old female Tg2576 mice (Hsiao et al., 1996) were dosed subcutaneously with vehicle (saline) or PF-3084014 at doses ranging from 1 to 32 mg/kg (n = 8 per group). Three hours after dose, mice were anesthetized with a ketamine/xylazine cocktail, and CSF was collected as described previously (Lanz et al., 2004). Blood was collected by cardiac puncture, and plasma was isolated. Brains were bisected, and one hemibrain was homogenized in 5 M guanidine/HCl for Aβ analysis.
Male Hartley guinea pigs (Charles River Laboratories, Inc., Wilmington, MA) weighing 225 to 250 g were dosed with vehicle (20% dimethyl sulfoxide, 20% EtOH, 60% PEG400) or PF-3084014 acutely (n = 6 per group) or through osmotic minipump (n = 7 per group). Alzet (Cupertino, CA) 2ML1 pumps with a 10 μl/h flow rate were formulated to deliver PF-3084014 at 0.003, 0.1, 0.3, 1, and 3 mg/kg/day. Pumps were implanted under isoflurane anesthesia, and tissues were collected 5 days after implantation. For acute studies, PF-3084014 (0.03, 0.1, 1, 3.2, and 10 mg/kg) or vehicle was delivered by subcutaneous injection, and guinea pigs were euthanized under CO2 at 0.75, 1.5, 3, 6, 9, 12, 18, 24, or 30 h after dose. An additional group was administered a dose of 32 mg/kg to evaluate the resulting Aβ profile by IP/MS. For both chronic and acute studies, blood was collected by cardiac puncture, and plasma was isolated. CSF was harvested from the cisterna magna and frozen on dry ice. Cerebellum and olfactory bulbs were removed, and brains were then bisected and frozen in tubes in liquid nitrogen. One hemibrain was used for Aβ analysis, and one hemibrain was used to measure drug levels.
Aβ and Drug-Level Analysis.
Aβ1-X, Aβ1-40, and Aβ1-42 were measured in Tg2576 brain, CSF, and plasma by DELFIA as described previously (Lanz et al., 2006). The same assays were used to measure Aβ1-40 and Aβ1-42 in guinea pig CSF and plasma. Guinea pig brains were homogenized in 0.2% diethylamine in 50 mM NaCl, followed by ultracentrifugation. Extracts were analyzed for changes in Aβ1-X (6E10 capture, 4G8 detection) using an IGEN assay (Igen International, Gaithersburg, MD) as described previously (Lanz et al., 2008). Significant differences between groups were detected by one-way analysis of variance followed by Dunnett's post hoc in Prism version 5 (GraphPad Software Inc., San Diego, CA). Treatment effects were considered statistically significant following p < 0.05 at the level of the analysis of variance and in a post-hoc comparison to vehicle.
Aβ was immunoprecipitated from Tg2576 and guinea pig brain extracts and analyzed by matrix-assisted laser desorption/ionization (MALDI) mass spectroscopy as described previously (Du et al., 2007).
Flow Cytometry of Lymphocytes in Blood and Spleen.
Spleens were collected in ice-cold PBS with 2% FBS. The single-cell suspensions were prepared by passage through 70-micron nylon mesh in PBS with 2% FBS and 0.09% sodium azide (stain buffer). Red blood cells were lysed by incubation in Pharm Lyse (BD Biosciences, San Jose, CA) for 5 min at room temperature. Cell suspensions were washed twice with stain buffer and subsequently passed through a nylon mesh. Fresh single-cell suspensions of spleen lymphocytes were incubated for 30 min on ice with fluorochrome-conjugated antibodies: mouse anti-guinea pig CD1b3-FITC and mouse anti-guinea pig T-cells (pan)-APC (AbD Serotec, Raleigh, NC). Stained samples were washed twice and resuspended in cold PBS before flow cytometry analysis.
Whole blood was collected by cardiac puncture in microtainer tubes with EDTA (BD Biosciences) and mixed at room temperature. One-hundred microliters of whole blood was incubated for 30 min on ice with fluorochrome-conjugated antibodies: anti-CD1b3-FITC, anti-T cells (pan)-phycoerythrin, and mouse anti-guinea pig CD45-APC (AbD Serotec). Red blood cells were lysed by incubating samples with Pharm Lyse for 5 min at room temperature. Stained samples were washed twice and resuspended in cold PBS before flow cytometry analysis. Antibodies were conjugated to the appropriate fluorochromes by using Phycolink conjugation kits (ProZyme, San Leandro, CA) and purified. Flow cytometric analysis was performed with a FACSCalibur cytometer (BD Biosciences).
In Vitro γ-Secretase Inhibition.
The structure of PF-3084014 is shown in Fig. 1 (Brodney et al., 2009). This compound inhibits γ-secretase with single-digit nanomolar potency as assessed by inhibition of Aβ production in a broken-cell enzyme preparation and a whole-cell assay (Table 1). Effects on Notch processing were evaluated in FTOC B- and T-cell populations. PF-3084014 had an IC50 on B- and T-cell reductions of 1.3 to 3 μM with a mean EC50 of 2.1 μM. This represents >300-fold separation from the broken-cell Aβ IC50 and >1500-fold separation from the whole-cell IC50. In contrast, in a separate set of experiments, the potent γ-secretase inhibitor LY-411575 [N2-[(2S)-2-(3,5-difluorophenyl)-2-hydroxyethanoyl]-N1-[(7S)-5-methyl-6-oxo-6,7-dihydro-5H-dibenzo[b,d]azepin-7-yl]-l-alaninamide] exhibited a mean FTOC EC50 of 3.95 nM, with a broken-cell and whole-cell IC50 of 700 and 21 pM, respectively.
Acute γ-Secretase Inhibition in Tg2576 Mice.
Young, plaque-free Tg2576 mice were dosed acutely with PF-3084014 at doses ranging from 1 to 18 mg/kg s.c., and tissues were collected at 3 h for Aβ measurement in brain, CSF, and plasma by DELFIA. Additional cohorts were dosed with 3.2 or 32 mg/kg s.c. PF-3084014, and brain samples were processed for IP/MS analysis of Aβ fragments. Aβ levels were reduced in a dose-dependent manner in brain, CSF, and plasma (Fig. 2). Mean plasma drug levels ranged from 88 ± 25 nM at 1 mg/kg to 1049 ± 419 nM at the 18 mg/kg dose. Plasma Aβ1-X was significantly reduced at all doses. Brain Aβ1-X was significantly reduced from 3 to 18 mg/kg, with CSF Aβ changes reaching statistical significance at doses of 9 and 18 mg/kg. Analysis of Aβ1-40 and Aβ1-42 levels after treatment revealed more inhibition of Aβ1-40 compared with Aβ1-42 (Fig. 3). All doses significantly reduced brain Aβ1-40, whereas only 9 to 18 mg/kg significantly reduced brain Aβ1-42. Changes in additional brain Aβ peptides were quantified by IP/MS. In addition to Aβ1-40 and Aβ1-42, Aβ1-34, Aβ1-37, and Aβ1-38 showed dose-dependent reduction in Tg2576 brain extracts after acute PF-3084014 treatment (Fig. 4).
Chronic γ-Secretase Inhibition in Guinea Pigs.
To assess efficacy of a γ-secretase inhibitor under conditions of chronic dosing, guinea pigs were implanted with subcutaneous minipumps delivering vehicle or PF-3084014 at 0.03 to 3 mg/kg/day for 5 days. A clear relationship between plasma drug concentration and efficacy in brain, CSF, and plasma was observed (Fig. 5). Plasma drug levels were below the limit of detection at doses of 0.03 and 0.1 mg/kg/day, and no significant Aβ changes could be detected at either dose. At 0.3 mg/kg/day plasma exposure was 19.6 ± 3.5 nM, which produced a significant (16%) reduction in brain Aβ1-X. At 1 to 3 mg/kg/day, drug levels were 81 ± 4 and 226 ± 28 nM, and Aβ1-X was significantly reduced in all three compartments. The estimated ED50 in brain was calculated to be 0.7 mg/kg/day. No overt adverse events were noted in this experiment, although pathologic endpoints were not evaluated.
Concentration-Dependent Changes in Brain, CSF, and Plasma Aβ in Guinea Pigs over Time.
To obtain a more comprehensive understanding of Aβ changes over time in brain, CSF, and plasma, guinea pigs were dosed acutely with PF-3084014 at 0.03 to 10 mg/kg, and tissues were collected at regular intervals from 0.75 to 30 h. Brain and plasma drug levels were analyzed at most time points, and B-cell counts in spleen and circulating blood were assessed at 3 to 10 mg/kg starting at the 3-h time point. The full data set is available in Supplemental Table 1. Efficacy-exposure relationships are presented in Figs. 6 to 7.
At 0.75 h after dose, the drug was readily measured in brain, but no significant alterations in brain Aβ could be detected (Fig. 6A). A significant reduction in CSF Aβ (25–35%; Supplemental Table 1), however, could be detected at this time point; CSF Aβ1-X was reduced at doses of 0.32, 1, 3.2, and 10 mg/kg. Significant reductions in Aβ1-40 and Aβ1-42 were also detected in CSF at the 10 mg/kg dose (Supplemental Table 1). By 1.5 h, brain exposure reached 95.5 nM, brain Aβ1-X was reduced by 30%, and CSF Aβ was reduced by 51% (p < 0.05; Fig. 6B). Plasma Aβ1-40 and Aβ1-42 were also significantly reduced at 10 mg/kg at 1.5 h (Supplemental Table 1). By 3 h after dose peak brain exposure had been reached (293 nM at 10 mg/kg), and Aβ1-X was significantly reduced in brain and plasma at doses of 3.2 and 10 mg/kg (Fig. 6C). At 3 h, CSF Aβ1-X, Aβ1-40, and Aβ1-42 and plasma Aβ1-42 were also significantly reduced at both doses (Supplemental Table 1). At 6 h after dose, drug levels were starting to decline, but significant reductions in brain Aβ1-X could be detected at all doses from 0.32 to 10 mg/kg (Fig. 6D). Plasma Aβ1-X, Aβ1-40, and Aβ1-42 were significantly reduced at doses of 3.2 and 10 mg/kg, and CSF Aβ1-X, Aβ1-40, and Aβ1-42 were significantly reduced at 10 mg/kg.
Brain Aβ1-X reductions continued to be significant from 1 to 10 mg/kg at the 9- and 12-h time points (Fig. 7), and significant reductions in CSF and plasma Aβ were detectable at 3.2 to 10 mg/kg at both time points. Twelve hours seemed to be the time of maximal efficacy for both of these doses. At the 1 mg/kg dose, however, plasma Aβ was significantly elevated at 12 h (141% of vehicle), whereas brain Aβ1-X was still 37% reduced (Fig. 7B). At 18 h, brain Aβ1-X was still significantly reduced at doses of 3.2 and 10 mg/kg (38 and 71% reductions, respectively; Fig. 7C), but by 24 h the 3.2 mg/kg dose had returned to vehicle levels (Fig. 7D). The 10 mg/kg doses still showed efficacy in brain, CSF, and plasma at 24 h, and drug was still measurable in all compartments (30.7 nM in brain, 1.9 nM in CSF, and 112.8 nM in plasma). This dose continued to be efficacious in brain (70% reduction, p < 0.05) and plasma (61% reduction, p < 0.05) 30 h after a single dose (Supplemental Table 1).
Additional APP cleavage events were monitored by IP/MS analysis 3 h after dose. Aβ spectra for guinea pigs treated with vehicle or PF-3084014 at 1 to 32 mg/kg are shown in Fig. 8. As initially observed in Tg2576 mice, there was a preference for reduction in Aβ1-40 (up to 52% reduction) relative to Aβ1-42 (up to 7% reduction). This pattern could be observed in plasma and by DELFIA (Supplemental Table 1). Dose-dependent reductions in multiple shorter γ-cleaved Aβ fragments were also detected in brain by IP/MS; Aβ1-33, Aβ1-34, Aβ1-35, Aβ1-36, Aβ1-37, Aβ1-38, Aβ1-39, and Aβ2-40 all were reduced relative to vehicle levels. Aβ1-42 showed little effect, whereas Aβ1-43 seemed to be elevated relative to vehicle. It is noteworthy that the N-terminal-truncated peptide Aβ11-40 exhibited a large increase relative to vehicle.
Effect of Acute γ-Secretase Inhibition on B-Cell Endpoints in Guinea Pigs.
To evaluate potential Notch-related toxicity in relationship to efficacy and exposure over time in guinea pigs, B-cell populations were analyzed by FACS in whole-blood and spleen preparations. The 10 mg/kg dose of PF-3084014 produced a significant reduction in B-cells by 18 to 24 h, but by 30 h counts had returned to vehicle levels (Fig. 9). FACS analysis of B-cells in whole blood demonstrated significant effects in advance of changes in spleen, with 45% reduction at 12 h. B-cell changes in spleen and blood, however, were much lower in terms of both maximal effect and area under the curve relative to Aβ lowering at 10 mg/kg. B-cell counts were also analyzed in spleen at a dose of 3.2 mg/kg, but no significant alterations were observed at any time point (Supplemental Table 1).
PF-3084014 is a novel small-molecule inhibitor of γ-secretase with an in vitro IC50 for Aβ lowering under 10 nM as measured in both cell-free and whole-cell systems. The potency for inhibition of Notch is much lower as measured by effects on B- and T-cell populations in an FTOC assay. This Notch-sparing selectivity for Aβ was 7- to 10-fold greater for PF-3084014 relative to the well characterized γ-secretase inhibitor LY-411575. Whereas the latter compound has previously been shown to affect Notch processing with functional consequences on the thymus and intestine, a 3- to 5-fold therapeutic window could be observed in CRND8 mice (Hyde et al., 2006). CRND8 and Tg2576 mice dosed with LY-411575 exhibited as much as 40% larger reductions in plasma Aβ relative to brain Aβ (Lanz et al., 2004; Hyde et al., 2006). Whereas PF-3084014 still showed greater efficacy in plasma relative to brain, the difference was on the order of 15 to 20%. As these models both overexpress APPSw, additional characterization was carried out in guinea pigs, whose Aβ sequence is identical to human (Johnstone et al., 1991), and whose APP substrate levels are not artificially elevated. In guinea pigs dosed with PF-3084014, plasma and brain Aβ levels were inhibited to a similar magnitude both after acute dosing and under steady-state conditions, probably owing to a high brain-to-plasma ratio of compound exposure. Similar compartmental differences in Aβ lowering between mice and guinea pigs have been observed with other γ-secretase inhibitors (Anderson et al., 2005).
As seen in vitro, a greater magnitude of effect was observed in the reduction of Aβ1-X relative to Notch-related endpoints. In the minipump and time course studies, sustained plasma exposures of approximately 60 nM correlated with brain and plasma Aβ1-X reductions of 50%. This drug concentration is 10- to 50-fold of the broken-cell and whole-cell IC50. Effects on B-cell endpoints were observed under dosing conditions where maximum plasma exposure measured was 203 nM (approximately 10-fold lower than the FTOC EC50 for PF-3084014), suggesting that the in vitro assay is less sensitive than in vivo manifestations. It should be noted, however, that making predictions about a therapeutic index among these different models is confounded by substantial differences between the models. The in vitro efficacy assays included one in which exogenous C100 cleaved substrate was added, and one in which the substrate was APPSw. Both in vitro efficacy assays were performed in human cells, whereas the FTOC assay was murine. Additional variables are added when considering comparisons to and between the animal models used in these experiments. The Tg2576 model expresses APP with the human Swedish mutation, although the rate of overexpression may not be identical to that of the cell line used in the whole-cell assay. In addition to the production of more Aβ, the panoply of Aβ peptides and Aβ40/42 ratios produced in Tg2576 mouse differ from those of guinea pig (Du et al., 2007). Thus confirmation of both safety and efficacy endpoints in the same species is vital to establishing a therapeutic window. The B-cell endpoints measured in the current studies are early and reversible indications of Notch inhibition. Although the currently described experiments evaluated these endpoints acutely, similar levels of inhibition of both Aβ and B-cells have been observed after 5 to 16 days of dosing with PF-3084014 (unpublished work). These biomarkers of efficacy and toxicity have been validated in human subjects and used in clinical studies evaluating γ-secretase inhibitors (Henley et al., 2009).
Biomarkers have been critical to the interpretation of past clinical trials with γ-secretase inhibitors, although plasma Aβ in particular has exhibited complex behavior.
PF-3084014 has been unique among γ-secretase inhibitors evaluated in the presently described models in that the low-dose potentiation phenomenon characteristic of γ-secretase inhibitors from other series has been difficult to observe. We have previously reported potentiation of Aβ1-X in media from the APPSw cellular assay described herein with LY-450139 [(S)-2-hydroxy-3-methyl-N-[(S)-1-((S)-3-methyl-2-oxo-2,3,4,5,-tetrahydro-1H-benzo[d]azepin-1-carbamoyl)-ethyl]-butyramide] (Lanz et al., 2006). Efforts to find concentrations of PF-3084014 that elevated Aβ in vitro have been unsuccessful. In guinea pigs, LY-450139 significantly elevated plasma Aβ1-X and Aβ1-40 at subefficacious doses in animals implanted with subcutaneous minipumps and significantly elevated (in some cases doubled) Aβ1-X, Aβ1-40, and Aβ1-42 in plasma at low concentrations after acute dosing (Lanz et al., 2006). PF-3084014 did not show any sign of Aβ potentiation at any of the subefficacious concentrations evaluated by using minipumps, consistent with the in vitro data. To more systematically look for evidence of potentiation after PF-3084014 administration, a six-point dose response was evaluated at nine different time points from 45 min to 30 h after dose. Modest increases in plasma Aβ were observed at low doses of PF-3084014, but they failed to achieve statistical significance, and the magnitude of potentiation paled in comparison with other γ-secretase inhibitors (Lanz et al., 2006; Burton et al., 2008). As doses escalated, a period of Aβ inhibition would be expected to be followed by plasma Aβ potentiation (when drug levels are waning), as has also observed in human subjects (Siemers et al., 2005, 2007; Martone et al., 2009). In guinea pigs, multiple doses of LY-450139 exhibited this phenomenon, and its duration enabled detection at multiple time points. Additional time points and doses were evaluated with PF-3084014, yet only a single dose at a single time point produced significant elevation of plasma Aβ (141% of vehicle at 1 mg/kg at 12 h). As the mechanism of potentiation is still under debate, these data do not exclude the possibility that PF-3084014 may potentiate Aβ under certain circumstances, but the compound seems to behave differently than γ-secretase inhibitors from other series that have been reported.
PF-3084014 may not readily raise Aβ1-X, but there are signs that it does modulate γ-secretase in ways that may not be expected for a compound that is simply inhibiting global γ-secretase enzymatic activity. Selected nonsteroidal anti-inflammatory drugs have been shown to preferentially reduce Aβ1-42 relative to Aβ1-40 (Takahashi et al., 2003), and brain-penetrant γ-secretase modulators of multiple chemical series have been described with this profile (Hall et al., 2010; Stanton et al., 2010). A γ-secretase modulator usually refers to a compound that selectively reduces Aβ1-42 and may spare Notch to a greater extent than a γ-secretase inhibitor, but often results in elevation of smaller Aβ fragments such as Aβ1-38 (Hall et al., 2010). Most classic γ-secretase inhibitors described to date, however, exhibit some form of modulation of γ-secretase activity with respect to effects on the pool of Aβ peptides produced. At doses that reduce Aβ1-40 and Aβ1-42, DAPT [N-[N-(3,5-difluorophenacetyl)-l-alanyl]-S-phenylglycine t-butyl ester] increases the production of Aβ1-43, Aβ1-46, and other longer Aβ fragments in vitro (Qi-Takahara et al., 2005; Yagishita et al., 2006). PF-3084014 similarly exhibited an elevation of Aβ1-43 at doses that reduced levels of multiple other Aβ peptides, a finding previously reported in guinea pig brain after LY-411575 treatment (Du et al., 2007). DAPT, LY-411575, and LY-450139 all have been shown to preferentially inhibit Aβ1-40 over Aβ1-42 and elevate Aβ levels at low concentrations (Lanz et al., 2003, 2004, 2006; Burton et al., 2008). PF-3084014 shares at least the former feature of these γ-secretase inhibitors. IP/MS analysis revealed that LY-450139 (Lanz et al., 2006), and now PF-3084014 as well, produced elevations in Aβ11-40 at doses that inhibited full-length Aβ1-40. Although increased β-secretase cleavage could explain the N-terminal truncation (Cai et al., 2001), γ-secretase is still required for the C-terminal cleavage at amino acid 40. It is possible that the N-terminally truncated APP is treated as a different substrate depending on β- or α-cleavage. Substrate selectivity is the foundation behind the idea that a γ-secretase modulator might be safer than an inhibitor with regard to toxicity associated with Notch processing. PF-3084014 seems to show some preference for processing of APP over Notch as assessed in vitro and in vivo. Thus PF-3084014, and most γ-secretase inhibitors described to date, might be more accurately categorized as 40-selective γ-secretase modulators, a class with distinct properties from the 42-selective γ-secretase modulators.
With respect to which Aβ fragment might be most therapeutically relevant, this question cannot be answered with satisfaction until the Aβ hypothesis is mechanistically evaluated in patients. Many animal models with Aβ levels in different quantities and proportions have been developed, and some of them have been reported to show some cognitive dysfunction. Preclinical Aβ models, however, do not exhibit the robust synaptic loss characteristic of the Alzheimer's disease brain. No preclinical model is a perfect recapitulation of the human disease, and until cognitive improvement can be demonstrated with an Aβ-relevant therapeutic, no preclinical model can claim to be predictive of efficacy in humans. Ultimately, clinical studies will be necessary to define what level of Notch inhibition can be tolerated, what level of Aβ inhibition is needed to produce efficacy, which Aβ fragments are most highly linked with disease, and whether inhibition of Aβ production can even modify the course of disease progression once symptoms have manifested.
Article, publication date, and citation information can be found at http://jpet.aspetjournals.org.
- amyloid precursor protein
- fetal thymic organ culture
- mass spectrometry
- matrix-assisted laser desorption/ionization
- cerebrospinal fluid
- enzyme-linked immunosorbent assay
- dissociation-enhanced lanthanide fluorescent immunoassay
- fetal bovine serum
- fluorescein isothiocyanate
- fluorescence-activated cell sorting
- phosphate-buffered saline
- N-[N-(3,5-difluorophenacetyl)-l-alanyl]-S-phenylglycine t-butyl ester.
- Received February 17, 2010.
- Accepted April 1, 2010.
- Copyright © 2010 by The American Society for Pharmacology and Experimental Therapeutics