Abstract
α-Amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptor (AMPAR) positive allosteric modulation (i.e., “potentiation”) has been proposed to overcome cognitive impairments in schizophrenia, but AMPAR overstimulation can be excitotoxic. Thus, it is critical to define carefully a potentiator’s mechanism-based therapeutic index (TI) and to determine confidently its translatability from rodents to higher-order species. Accordingly, the novel AMPAR potentiator N-{(3R,4S)-3-[4-(5-cyano-2-thienyl)phenyl]tetrahydro-2H-pyran-4-yl}propane-2-sulfonamide (PF-4778574) was characterized in a series of in vitro assays and single-dose animal studies evaluating AMPAR-mediated activities related to cognition and safety to afford an unbound brain compound concentration (Cb,u)–normalized interspecies exposure-response relationship. Because it is unknown which AMPAR subtype(s) may be selectively potentiated for an optimal TI, PF-4778574 binding affinity and functional potency were determined in rodent tissues expected to express a native mixture of AMPAR subunits and their associated proteins to afford composite pharmacological values. Functional activity was also quantified in recombinant cell lines stably expressing human GluA2 flip or flop homotetramers. Procognitive effects of PF-4778574 were evaluated in both rat electrophysiological and nonhuman primate (nhp) behavioral models of pharmacologically induced N-methyl-d-aspartate receptor hypofunction. Safety studies assessed cerebellum-based AMPAR activation (mouse) and motor coordination disruptions (mouse, dog, and nhp), as well as convulsion (mouse, rat, and dog). The resulting empirically derived exposure-response continuum for PF-4778574 defines a single-dose-based TI of 8- to 16-fold for self-limiting tremor, a readily monitorable clinical adverse event. Importantly, the Cb,u mediating each physiological effect were highly consistent across species, with efficacy and convulsion occurring at just fractions of the in vitro–derived pharmacological values.
Introduction
α-Amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptors (AMPARs) mediate fast glutamatergic excitatory neurotransmission throughout the central nervous system. Changes in AMPAR density and subunit composition are principal mechanisms for the dynamic regulation of synaptic efficacy underlying adaptive brain functions (Malinow and Malenka, 2002). Consequently, AMPAR modulation has been targeted to treat numerous psychiatric and neurological diseases (O'Neill et al., 2004; Zarate and Manji, 2008). AMPAR-centric drug discovery (Black, 2005; Grove et al., 2010) has primarily focused on receptor activation via positive allosteric modulators (i.e., “potentiators”) that tend to better match the degree and/or duration of AMPAR opening to synaptic activity than orthosteric agonists. Nonetheless, excessive AMPAR stimulation can cause degeneration of Purkinje cells and hippocampal neurons (Garthwaite and Garthwaite, 1991) and/or convulsion (Yamada, 1998). Therefore, it is imperative to understand the AMPAR potentiator exposures eliciting desired versus deleterious effects to identify a safe and well tolerated molecule for high-confidence clinical evaluation. Because the effect of variations in AMPAR subtypes, splice variants, and binding partners on a potentiator’s therapeutic index (TI) is unknown, as is how the TI may vary across the mammalian hierarchy, a meticulous interspecies exposure-response relationship was developed for the novel AMPAR potentiator N-{(3R,4S)-3-[4-(5-cyano-2-thienyl)phenyl]tetrahydro-2H-pyran-4-yl}propane-2-sulfonamide (PF-4778574; Fig. 1) (Estep et al., 2008). This exposure-response approach is grounded in both the “free drug” hypothesis (Tillement et al., 1988), which stipulates that interstitial fluid potentiator concentrations dictate its interaction with the extracellular ligand binding domains of AMPARs (Shaffer, 2010), and the unbound brain compound concentration (Cb,u) being a valid interstitial fluid potentiator concentration surrogate (Liu et al., 2009; Doran et al., 2012).
For this work, PF-4778574 efficacy was conceptually rooted in AMPAR potentiation overcoming certain cognitive impairments in schizophrenia (Millan et al., 2012; Moghaddam and Javitt, 2012; Collingridge et al., 2013). This rationale stems from the hypothesis that schizophrenia fundamentally results from dysfunction in N-methyl-d-aspartate receptor (NMDAR) glutamatergic neurotransmission (Olney and Farber, 1995; Goff and Coyle, 2001; Javitt, 2007). Clinical studies (Krystal et al., 1994; Javitt, 2007) in which healthy volunteers receiving acute subanesthetic doses of nonselective NMDAR antagonists (e.g., ketamine) presented with schizophrenia-like symptoms, including working memory deficits, support this proposition. Activated AMPARs depolarize neuronal membranes to relieve the Mg2+ block of colocalized NMDARs, which increases NMDAR-mediated Ca2+ gating (Lynch, 2002) to ultimately produce changes in the synaptic morphology and function (Malinow and Malenka, 2002; Lynch and Gall, 2006) believed to underlie learning and memory (Morris, 2003). Moreover, one chief mechanism of NMDAR-dependent increases in synaptic activity is AMPAR insertion into the synapse (Sun et al., 2005). Thus, potentiating AMPARs enhances NMDAR-induced synaptic potentiation. These facts, coupled with decreased AMPAR density in the schizophrenic hippocampus (Meador-Woodruff and Healy, 2000), suggest that augmenting AMPAR activity may assuage cognitive disruptions in schizophrenia. Indeed, AMPAR potentiators have shown procognitive properties in multiple preclinical models of hippocampal/cortical function and working memory (Black, 2005) and in small clinical trials (Marenco and Weinberger, 2006).
In consideration of these collective concepts, PF-4778574 was characterized in a series of in vitro and single-dose in vivo assays assessing AMPAR-contingent activities related to nootropism and safety. Native AMPARs are tetrameric complexes composed of multiple combinations of four monomeric subunits (GluA1–4), each having flip (i) and flop (o) isoforms, that are influenced by transmembrane proteins; this complexity is hypothesized to affect the functional heterogeneity of AMPAR-mediated synaptic transmission (Parsons et al., 2005; Arai and Kessler, 2007). Because it is unknown which AMPAR subtype(s) should (and/or can) be selectively potentiated for an optimal TI, PF-4778574 binding affinity was determined in primary rat cortical tissue while functional potency was assessed in both mouse embryonic stem (ES) cell–derived neurons (McNeish et al., 2010) and primary cultures of rat cortical neurons, as these matrices are expected to express a native mixture of AMPAR subunits and their associated proteins to afford composite pharmacological values. Functional activity was also quantified in recombinant cell lines stably expressing human GluA2i and GluA2o, given that other AMPAR potentiators interact with residues within these flip/flop domains (Fleming and England, 2010; Harms et al., 2013). For efficacy, PF-4778574 was evaluated in two contemporary animal models of pharmacologically induced NMDAR hypofunction (Olney et al., 1999): MK-801 ([5R,10S]-5-methyl-10,11-dihydro-5H-dibenzo[a,d]cyclohepten-5,10-imine (dizocilpine))–induced subiculum-medial prefrontal cortex (PFC) dysfunction in rats (Kiss et al., 2011a) and ketamine-mediated spatial working memory impairment in nonhuman primates (nhp) (Roberts et al., 2010). Safety studies with PF-4778574 looked for cerebellum-based AMPAR activation (mouse) and motor coordination disruptions (mouse, dog, and nhp), as well as convulsion (mouse, rat, and dog). The resulting empirically derived Cb,u-normalized exposure-response continuum for PF-4778574 defines acceptable separation between drug concentrations associated with efficacy, motor coordination disruptions, and convulsion.
Materials and Methods
Chemicals, Reagents, and Animals
PF-4778574 (>99% chemical purity, 100% enantiomeric excess; Supplemental Materials and Methods) (Estep et al., 2008) and N-((S)-1-(3,5-[3H]-2-fluoro-4-((S)-5-((1-methylethylsulfonamido)methyl)-4,5-dihydroisoxazol-3-yl)phenyl)pyrrolidin-3-yl)acetamide ([3H]PF-04725379; 98.5% radiochemical purity, 100% enantiomeric excess, 49 Ci/mmol) (Patel N, Schwarz J, Hou X, Hoover D, Xie L, Fliri A, Gallaschun R, Lazzaro J, Bryce D, Hoffmann, W, Hanks A, McGinnis D, Marr E, Gazard J, Hajos M, Scialis R, Hurst R, Shaffer C, Pandit J, O'Donnell C, submitted for publication) were synthesized and fully characterized by Neuroscience Chemistry at Pfizer Worldwide Research and Development (WRD, Groton, CT). PF-4778574 (product no. PZ0214) is commercially available from Sigma-Aldrich (St. Louis, MO). Chemicals and solvents of reagent or high-performance liquid chromatography grade were supplied by Sigma-Aldrich, Thermo Fisher Scientific (Waltham, MA), and Mallinckrodt Baker, Inc. (Phillipsburg, NJ). For control matrices, species-specific plasma was obtained from Bioreclamation, Inc. (Hicksville, NY), and rat brain tissue was obtained at WRD. Male CD-1 mice and Sprague-Dawley rats were bought from Harlan Laboratories (Indianapolis, IN) or Charles River Laboratories, Inc. (Wilmington, MA), male C57BL/6J mice were procured from The Jackson Laboratory (Bar Harbor, ME), and beagle dogs were sourced from Marshall BioResources (North Rose, NY). Non-naïve cynomolgus monkeys (Macaca fascicularis) resided within the WRD nhp colony. All animal studies were performed in accordance with the Guide for the Care and Use of Laboratory Animals using protocols reviewed and approved by the WRD Institutional Animal Care and Use Committee. All blood samples were collected in EDTA-containing tubes and processed immediately to obtain plasma, and the collection of neuromatrices from rodents followed published techniques (Doran et al., 2012). All biomatrices collected for PF-4778574 quantification were stored at −70°C until processing for bioanalysis.
In Vitro Pharmacology
Binding Affinity
Rat Brain Homogenate Radioligand Binding Assay.
Rat brains were purchased from Pel-Freez Biologicals (Rogers, AR). The cortex was dissected, homogenized using a polytron in 50 mM Tris 7.4, and centrifuged (40,000g for 20 minutes). The resulting pellet was resuspended in 50 mM Tris 7.4 and centrifuged (40,000g for 20 minutes); this process was repeated three times, after which the pellet was resuspended in 30 mM Tris 7.4 (100 mg/ml) and stored at −80°C. From a dimethylsulfoxide (DMSO) stock solution (10 or 30 mM), PF-4778574 was titrated, then added (25 μl/well) at 10-fold its final concentration to a 96-well polypropylene plate containing assay buffer (30 mM Tris HCl, pH 7.4) supplemented with l-glutamic acid (25 μl, 5 mM). The AMPAR potentiator N-((3R,4S)-3-(2′-cyanobiphenyl-4-yl)tetrahydro-2H-pyran-4-yl)propane-2-sulfonamide (PF-4697190; 100 μM in assay buffer) (Estep et al., 2008; Shaffer et al., 2012) was added to positive control wells for a final concentration of 10 μM. To each well, an aliquot (175 μl) of a 1 nM stock solution of [3H]PF-04725379 was added; for each experiment, the precise stock solution concentration was determined by liquid scintillation counting. Previously prepared rat cortical tissue was thawed and homogenized (100 mg/ml) before its addition (25 μl, 2.5 mg tissue) to each sample well. Plates were incubated for 2 hours at 37°C and then harvested onto uncoated Filtermat B filters (PerkinElmer Life and Analytical Sciences, Waltham, MA) using a Skatron filter harvester (Skatron Instruments Ltd., Newmarket, UK) and Tris wash buffer (50 mM, pH 7.4, 4°C). Filters were dried overnight, placed in scintillant bags, and read on a Betaplate filtermat reader (PerkinElmer Life and Analytical Sciences). Concentration-response data were fitted with a logistic function using a four-parameter logistic model. The Ki was determined using the ligand concentration of each experiment using the Cheng-Prusoff equation. Geometric means, rounded to two significant figures, and standard errors of the Ki values were calculated from multiple experiments.
Functional Activity
Fluorometric Imaging Plate Reader Functional Assays.
Two distinct cell-based fluorometric imaging plate reader (FLIPR) functional assays were used for compound assessment: mouse ES cell–derived neuronal precursors and recombinant human cell lines. The ES cell–based assay has been fully disclosed (McNeish et al., 2010); thus, only the recombinant cell lines are described here. Final DMSO concentrations were <1% in all assays.
Recombinant Human Cell Lines.
Human embryonic kidney 293 (HEK293) cell lines stably expressing the human AMPAR subunits GluA2i or GluA2o, both expressed in the Ca2+-permeable Q form, were used (Invitrogen, Carlsbad, CA). Cells were maintained in growth media containing Dulbecco’s modified Eagle’s medium high glucose (500 ml; Invitrogen), 10% dialyzed fetal bovine serum (Invitrogen), HEPES (25 mM; Invitrogen), minimum Eagle’s medium nonessential amino acids (Invitrogen), Penicillin Streptomycin (100 μg/ml; Invitrogen ), and blasticidin (50 μg/ml; Invitrogen). Cells were maintained at 37°C and 5% CO2. Two days prior to the assay, cells were lifted from flasks with 0.25% trypsin and plated at a density of 12,500 cells/well on poly-d-lysine–coated black/clear 384-well plates (BD Biosciences, San Jose, CA).
FLIPR Methods.
Assay buffer containing 145 mM NaCl, 5 mM KCl, 10 mM glucose, 10 mM HEPES, 1 mM MgSO4, and 2 mM CaCl2 and pH adjusted to 7.4 with 1 M NaOH, was freshly prepared for each experiment. A Fluo-4 AM (Invitrogen) stock solution (1 mM in DMSO with 10% Pluronic acid) was added to Dulbecco’s modified Eagle’s medium for a 4 μM dye incubation medium, which was then supplemented with probenecid (2.5 mM; Sigma-Aldrich). Probenecid-containing (2.5 mM) assay buffer was also used for cell washing but not compound preparation. Following growth media removal from cell plates, dye solution (50 μl) was added to each well. Plates were then incubated for 1 hour at room temperature (RT) before the dye solution was removed and washed three times with assay buffer, leaving 30 μl assay buffer per well. Plates were left at RT for at least 10 minutes before adding compound. The FLIPR (Molecular Devices, LLC, Sunnyvale, CA) was used to perform simultaneous Ca2+ imaging and drug application. Compounds were initially dissolved in DMSO (10 or 30 mM). After initial compound titration, the compound drug plate was prepared in assay buffer at 4-fold the final concentration. PF-4778574 (15 μl/well) was transferred from the drug plate to the cell plate while imaging the entire 384-well plate at a rate of 1 sample/2 seconds. Five and 40 samples were recorded before and after compound addition, respectively, using excitation and emission wavelengths of 488 and 510 to 570 nm, respectively. After the first read, the plate was incubated at RT for 10 minutes, then placed in the FLIPR for a second addition, during which a 4-fold challenge concentration of S-AMPA (15 μl/well) was added to the cell plate during imaging. The compound addition and read protocol for the second addition was the same as the first. A final S-AMPA concentration of 32 μM was used for the assay. The peak fluorescence in each well was determined by subtracting the maximal fluorescence across the 40 samples after compound addition from the baseline fluorescence. Data analysis for the second addition was performed by determining the percentage effect at each concentration using a reference AMPAR potentiator as the positive control and S-AMPA alone as the negative control. A maximal concentration of the potentiator cyclothiazide (10 or 32 μM; Tocris Bioscience, Bristol, UK) was used as the positive control for the mouse ES cell–derived neurons. Maximal concentrations (32 μM) of the potentiators PF-4697190 (Estep et al., 2008; Shaffer et al., 2012) or 2-(2-(3-(trifluoromethyl)-4,5,6,7-tetrahydro-1H-indaol-1-yl)acetamido)-4,5,6,7-tetrahydrobenzo[b]thiophene-3-carboxamide (CE-382349) (McNeish et al., 2010) were used as positive controls for the human GluA2 cell lines. Concentration-response data were fit as described previously. Geometric means, rounded to two significant figures, and standard errors of the EC50 values were calculated from multiple experiments. The percentage efficacy in each experiment was determined by dividing the maximum asymptote of the fitted PF-4778574 curve by the positive control in the assay. Average percentage efficacy was calculated as an arithmetic mean ± S.E.M.
Rat Cortical Neurons Electrophysiological Studies.
Rat fetal (E18) frontal cortex primary neurons were grown on 12-mm round poly-d-lysine–coated cover slips (~1×105 cells/well plated 2–3 weeks prior to recording in 12-well tissue culture plates). Cover slips with primary neurons were placed in a recording chamber and superfused (~1–2 ml/min) with external buffer containing the following (in mM): NaCl (130), KCl (2), MgCl2 (1.5), CaCl2 (2), HEPES (10), and glucose, pH 7.4, and approximately 305–310 mOsm at RT. One-millimeter outer diameter glass electrodes were filled with internal solution containing the following (in mM): CsCH3SO3 (126), CsCl (10), NaCl (4), MgCl2 (1), EGTA (8), CaCl2 (0.5), HEPES (10), Na2ATP (3), NaGTP (0.3), and phosphocreatine (4), pH 7.2, and approximately 295 mOsm and had a tip resistance of 3–4 MΩ. Patch-clamp experiments were conducted at RT using an Axoclamp 200B amplifier (Molecular Devices, LLC) in the whole-cell configuration. By use of a pCLAMP 10 software (Molecular Devices, LLC) protocol, cell membrane potential was voltage-clamped at −60 mV. For S-AMPA (±300 nM PF-4778574) concentration-response studies, S-AMPA was added in increasing concentrations (0.1, 0.3, 1, 3, 10, 30, and 100 μM) to elicit a current. Four-second compound applications were performed via gravity feed using a 3-barreled glass tube connected to an SF-77B fast step perfusion system (Warner Instruments LLC, Hamden, CT) with a 120-second washout period between applications. For PF-4778574 concentration-response experiments, S-AMPA (30 μM) with increasing PF-4778574 concentrations (3, 30, 300, and 3000 nM) was added using 20-second applications with a 120-second washout period between applications. Data were calculated as the percentage of maximum current (for S-AMPA concentration-response) or percentage of S-AMPA control current (for PF-4778574 concentration-response), and EC50 values were calculated using a logistic nonlinear regression fit. Data were analyzed and visualized using Clampfit 10 software (Molecular Devices, LLC).
Pharmacokinetics Studies
Plasma and Brain Homogenate Nonspecific Binding
Using a reported (Doran et al., 2012) equilibrium dialysis procedure, the unbound fraction of PF-4778574 (1 μM, N = 3/species) in plasma (fu,p) and in brain homogenate (fu,b) were determined. The stability of PF-4778574 in each matrix and the optimal dialysis time were determined separately prior to actual studies.
In Vivo Pharmacokinetics
Before in vivo pharmacology experiments, studies with PF-4778574 characterizing its pharmacokinetics (peripheral and/or central) were conducted in the relevant preclinical species. These studies determined the following: total plasma (Cp) and/or total brain (Cb) compound concentrations at select times after a specific dose; pharmacokinetic parameters [particularly maximal matrix PF-4778574 concentration (Cmax) and area under the matrix compound concentration-time curve (AUC)] and their linearity at doses of pharmacological interest, which allowed the correlation of pharmacodynamics to Cp and/or Cb; the first time of Cmax (Tmax), which determined PF-4778574 pretreatment time in certain assays; and PF-4778574 plasma half-life (t1/2) that dictated washout periods between doses in the nhp spatial delayed response (SDR) task. Individual animal doses were calculated based on respective dose solution concentrations (milligrams per milliliter), pretreatment body weights (kilograms), and dose volume (milliliters per kilogram). The actual amount of dose solution administered to each animal was determined by weighing the loaded syringe before and after it was dispensed. All study animals were fasted overnight and for approximately 4 hours after dose. For each species, the specific vehicle and dosing route used in its pharmacokinetics study were also used in its pharmacodynamics experiment.
Mouse.
A single dose (0.178 mg/kg) of PF-4778574 in 5:5:90 (v/v/v) DMSO:Cremophor EL:deionized H2O (0.0178 mg/ml) was administered (10 ml/kg s.c.) to CD-1 (N = 2) or C57BL/6J (N = 3) mice. At 0.5 hour after dose, animals were placed under isoflurane anesthesia for the collection of blood and brain samples.
Rat: Intravenous Dose.
From jugular vein–cannulated Sprague-Dawley rats (N = 3) receiving a single bolus (1 ml/kg i.v.) of PF-4778574 (0.2 mg/kg) in 2:98 (v/v) DMSO:20% hydroxypropyl-β-cyclodextrin (0.2 mg/ml), blood samples (0.5 ml) were serially collected just before dosing and at 0.083, 0.167, 0.25, 0.5, 1, 2, 4, 6, 8, and 12 hours after dose.
Neuropharmacokinetics.
As described (Doran et al., 2012), a single dose (1 mg/kg, s.c.) of PF-4778574 in 2:98 (v/v) DMSO:20% hydroxypropyl-β-cyclodextrin (1 mg/ml) was injected (1 ml/kg) into each male Sprague-Dawley rat. Blood, cerebrospinal fluid (CSF), and brain samples were collected from each animal after euthanasia by CO2 asphyxiation at 0.25, 0.5, 1, 2, 4, and 8 hours after dose (N = 2/time point).
Dog.
A single dose (0.1 mg/kg) of PF-4778574 in 0.5% methylcellulose (0.1 mg/ml) was given (1 ml/kg PO) to each dog (N = 1/sex). Blood samples (2 ml) were obtained via the jugular vein just before dosing and at 0.25, 0.5, 1, 2, and 4 hours after dose. During this time period, dogs were continuously monitored by trained laboratory staff for any readily apparent adverse events.
Nonhuman Primate.
A single dose (0.1 or 0.32 mg/kg s.c.) of PF-4778574 in 1:9 (v/v) Cremophor EL:deionized H2O (0.3 or 1 mg/ml) was administered (0.33 ml/kg delivered to the dorsal thoracolumbar area) to each nhp (N = 2/dose). Blood samples (3 ml) were obtained via a vascular access port just before dosing and at 0.25, 0.5, 1, 2, 4, 6, and 24 hours after dose. After remaining chaired for 2 hours after dose, animals were returned to their home cage. They were again chaired just prior to blood collection at 4, 6, and 24 hours after dose and released back to their home cage after each blood sampling. Trained laboratory staff constantly observed each nhp for any readily apparent adverse events for 6 hours after dose.
Quantitative Analysis of PF-4778574 in Biologic Matrices
The quantification of PF-4778574 in plasma, CSF, or brain tissue samples from rodents, dogs, or nhp was achieved using published liquid chromatography–tandem mass spectrometry methodologies (Doran et al., 2012). The dynamic ranges were 0.488–500 ng/ml for plasma samples and 0.488–1000 ng/ml for CSF and brain tissue samples.
Pharmacokinetic Calculations
Pharmacokinetic parameters were calculated by noncompartmental analyses using WinNonlin version 5.2 (Pharsight Corp., Mountain View, CA). The area under the matrix compound concentration-time curve (AUC0–Tlast) was calculated using the linear trapezoidal method, the elimination rate constant (kel) was determined by linear regression of the log concentration versus time data during the last observable elimination phase, t1/2 was calculated as 0.693/kel, and AUC0–∞ was calculated as the sum of AUC0–Tlast and AUCTlast–∞, which was determined by dividing the Cp at Tlast by kel. Both Cmax and Tmax were taken directly from the matrix compound concentration versus time data. Means and S.D. were computed when half or more of the values exceeded 0.488 ng/ml. A value of 0 was used if a measured value was <0.488 ng/ml. Volume of distribution (V) was calculated by the equation V = CL × (AUMC/AUC), where AUMC is the total area under the first moment-time curve and systemic clearance (CL) was determined by dividing the measured dose by plasma AUC0–∞. All AUC values were subject weight normalized.
In Vivo Pharmacology
PF-4778574 was studied in select in vivo pharmacology models to understand the relationship between its Cb,u and a particular AMPAR-mediated effect, which together allowed comparison with its in vitro pharmacology values. Collectively, this strategy provided dose-dependent, temporally normalized, and assay-dependent Cb,u that could be directly compared across all in vivo models. For each species, the pharmacodynamics evaluation of PF-4778574 used the same dosing vehicle, volume, and route as its pharmacokinetics studies. For terminal assays in which brain tissue was collected from an animal undergoing the assessment, Cb,u (nanomolar units) was determined by measured Cb, PF-4778574 molecular weight (MW 390.5 g/mol), rat fu,b, an assumed brain tissue density of 1 g/ml, and eq. 1:
(1)The experimentally determined rat fu,b was used for mouse Cb,u calculations since interspecies fu,b are highly consistent (Di et al., 2011).
For nonterminal assays, if plasma samples were collected for PF-4778574 quantification during the evaluation (e.g., low-dose dog studies), then Cb,u was projected from each Cp using compound molecular weight (MW), species-specific fu,p, species-specific ratio of unbound brain compound concentration to unbound plasma compound concentration (Cb,u:Cp,u) (Doran et al., 2012), and eq. 2:
(2)If plasma samples were not collected for PF-4778574 quantification in the pharmacodynamic test (e.g., nhp SDR task), Cp at any time point for a specific dose within that species’ satellite pharmacokinetics study was converted to Cb,u using eq. 2. The resulting Cb,u versus time data were then used to extrapolate linearly Cb,u at a specific time point for each dose tested in the in vivo pharmacology model.
Mechanism
Mouse Cerebellum Cyclic Guanosine Monophosphate Assay.
Paralleling the procedure of Ryder et al. (2006), adult male CD-1 mice (25–30 g, acclimated to the vivarium 7 days before use, N = 5–6/dose) receiving vehicle or PF-4778574 (0.1, 0.32, or 1.0 mg/kg s.c.) were placed in an open shoebox with sawdust bedding and observed for 0.5 hour. Mice were then individually placed head first in a Plexiglas restraining device and euthanized using a beam of microwave radiation focused on the skull for 0.88 second at 100% power in a Cober Electronics microwave (Cober Electronics, Inc., Norwalk, CT). After cooling for 1 minute, the cerebellum was dissected, weighed, and frozen on dry ice. Tissue cGMP levels were determined using a competitive enzyme immunoassay (EIA) using a commercially available Cyclic GMP EIA kit (Cayman Chemical Company, Ann Arbor, MI). All buffers and cGMP standards were prepared exactly as described in the package insert. Each cerebellum sample was homogenized (10 μl of buffer per mg of tissue) in lysis buffer (10 mM Tris HCl, 100 mM NaCl, 1 mM EDTA, and 0.3% Nonidet P-40, pH 7.4) using a tip sonicator. Homogenate was added to a 1.5-ml Eppendorf tube (Eppendorf, Hamburg, Germany) and centrifuged (18,000g for 20 minutes at 4°C), and the supernatant was diluted 1:25 with EIA buffer and added in triplicate to a 96-well plate. Samples were processed according to kit directions for 12 hours at 4°C. A cGMP standard curve (0.023–3.0 pmol/ml) was included on each assay plate. After incubation, plates were washed, developed using Ellman’s reagent for 2 hours at RT, and then read at 420 nm on a SpectraMax Microplate Reader (Molecular Devices, LLC). Data were compiled and reduced using the SpectraMax data reduction software, which calculates sample cGMP levels via standard curve extrapolation. For each dose group, data were reported as group means ± S.E.M. using X-fold changes in tissue cGMP levels relative to the vehicle-treated group. PF-4778574 doses were compared with the concurrent vehicle control group using one-way analysis of variance (ANOVA) followed by Dunnett’s t test (SigmaStat version 3.5; Systat Software, Inc., Chicago, IL). A P value <0.05 indicated a statistically significant effect.
Safety
Mouse Rotarod Assay.
For 2 weeks prior to testing, C57BL/6J male mice (7–8 weeks old) were group housed (N = 4/isolator) with free access to food and water in a temperature- and humidity-controlled environment on a 6:00 AM/6:00 PM light/dark schedule. On test day 1, mice were acclimated to the test room 1 hour prior to the training session. Four programmable SmartRod chambers with SmartRod software version 1.70 (AccuScan Instruments, Columbus, OH) were used for testing. Each chamber was equipped with a rotating rod (3 cm in diameter, spanned the 11-cm width of the chamber) horizontally affixed 32.5 cm on-center above the grid floor. Two infrared beams, located 2 cm above the floor, detected the animal when it fell from the rods and deactivated the timer to automatically record fall latency. Rods were programmed to accelerate at a rate of 0.25 rev/s to a maximum speed of 20 rpm for 62 seconds, after which the rod decelerated for an additional 5 seconds to the stop position. The maximum cycle time for one trial was 67 seconds. The training session consisted of six consecutive trials on the rod with a 10-second intertrial interval for each mouse. Mice had to complete two of the six training trials with a fall latency >10 seconds to be accepted for testing the following day. Mice that did not meet the criteria were eliminated from the study, and mice that met the criteria were returned to their home cage and remained in the test room under conditions consistent with the housing room environment. On test day 2, mice were randomly assigned to groups of 12 and dosed with vehicle or PF-4778574 (0.178, 0.32, or 0.56 mg/kg s.c.). For testing, each mouse was subjected to three consecutive trials on the rods with a 10-second intertrial interval. The mean fall latency for three trials was calculated and data were reported as group means ± S.E.M. PF-4778574 doses were compared with the concurrent vehicle control group using one-way ANOVA followed by Dunnett’s t test (SigmaStat version 3.5; Systat Software, Inc.). A P value <0.05 indicated a statistically significant effect.
Convulsion Dose-Response Studies: Rodents.
PF-4778574 was administered (N = 6/dose) to adult male CD-1 mice (1, 1.78, or 3.2 mg/kg s.c.) or Sprague-Dawley rats (1.78, 3.2, or 5.6 mg/kg s.c.), which were observed for up to 2 hours after dose. Individual animals were euthanized by CO2 asphyxiation at the time of convulsion (if any), and blood and brain were collected for PF-4778574 quantification.
Dogs.
These studies were conducted in the Department of Drug Safety and Research at WRD by scientists highly skilled in monitoring and treating serious adverse events in research animals. Adult beagle dogs were administered PF-4778574 [0.2 (N = 1 female), 0.25 (N = 1 male), or 0.5 (N = 1/sex) mg/kg PO] and observed from 0 to 4 hours after dose. Blood samples (2 ml) were scheduled for collection at 0.25, 0.5, 1, 2, and 4 hours after dose. In the event of a convulsion, a blood sample was taken at the time of its first observation.
Efficacy
Rat MK-801–Disrupted Cortical Oscillation and Paired-Pulse Facilitation Assay.
The objective of this study was to quantify the effects of systemically administered PF-4778574 on the cortical disruption of electroencephalogram (EEG) and paired-pulse facilitation (PPF) by the noncompetitive NMDAR antagonist MK-801 in urethane-anesthetized Sprague-Dawley rats (Kiss et al., 2011b). This is a preclinical model, which evaluates the subiculum-medial PFC pathway, proposed for studying NMDAR hypofunction of schizophrenia. Experimental procedures were those published by Kiss et al. (2011a) using male Sprague-Dawley rats (250–320 g, N = 5) under urethane anesthesia (1.5 mg/kg i.p.). For the intravenous dose-response effects of PF-4778574, control waveform averages were first computed using 60 consecutive paired stimuli over a 10-minute period. MK-801 (0.05 mg/kg i.v.; Tocris Bioscience) was then administered, and after a 5-minute waiting period, paired-pulse averaging was again conducted over the next two consecutive 5-minute time bins. The total MK-801 response was then computed by combining these two averages. At the end of the second 5-minute data collection time bin (i.e., 15 minutes after the MK-801 injection), cumulative intravenous administration of PF-4778574 (0.03 and 0.1 mg/kg) was initiated. Similar to how the data were handled following MK-801 dosing, 5 minutes were allowed between each successive PF-4778574 injection and at the start of the subsequent averaging. If no effect was observed before the end of the second 5-minute averaging period, the two 5-minute averages were combined as a single value for that dose and the next intravenous dose was given. If an effect was observed at any point during the first two averages following a given dose, two additional 5-minute averages were combined as the data point for that dose. The reversal of the MK-801 effects by PF-4778574 did not occur in a typical cumulative dose-response manner, but rather as an all-or-none effect with variable threshold doses between animals. Since reversal was an all-or-none effect once the threshold dose in a particular animal was reached, no additional cumulative doses of PF-4778574 were given. Instead, responses to the paired-pulse stimulations were recorded continuously and averaged over each consecutive 10-minute time bin for 60 minutes to obtain the overall time course of PF-4778574 activity. Mean effects in the PF-4778574 doses were compared with both the control and MK-801 doses using a two-tailed Student’s t test (Microsoft Excel; Microsoft Corporation, Redmond, WA). A P value <0.05 indicated a statistically significant effect.
Nonhuman Primate Ketamine-Disrupted Spatial Delayed Response Task.
The objective of this study was to quantify the effects of PF-4778574 on ketamine-induced spatial working memory impairments in nhp, a proposed model of cognitive impairment associated with schizophrenia. PF-4778574 (0.001, 0.01, and 0.1 mg/kg s.c.) was evaluated as previously described (Roberts et al., 2010).
Results
In Vitro Pharmacology
Displacement studies using [3H]PF-04725379 in rat cortical tissue determined a Ki of 85 nM for PF-4778574. Functional potency was assessed by measuring effects induced by S-AMPA alone and following PF-4778574 pretreatment using either changes in intracellular Ca2+ concentration (HEK293 cells and mouse ES cell–derived neuronal precursors) or whole-cell current (rat primary cortical neurons) to measure the S-AMPA–dependent responses. When added alone (up to 32 and 3 μM in the FLIPR-based assays and rat cortical neuron electrophysiological assay, respectively), PF-4778574 produced no detectable response in any of the tested cell types expressing functional AMPARs. However, PF-4778574 concentration-dependently increased S-AMPA–evoked responses in HEK293 cells expressing human GluA2i or GluA2o, mouse neuronal precursors, and rat primary neurons. Depending on the cell type, PF-4778574 EC50 ranged from 45 to 919 nM. Table 1 summarizes the geometric means for pKi, Ki, pEC50, EC50, the arithmetic means for percent efficacy, and the number of assay replicates (N). Intrinsic efficacy was evaluated in primary cultures of rat cortical neurons using whole-cell patch clamp electrophysiology. When preapplied to neurons, PF-4778574 increased the maximal S-AMPA–evoked current by approximately 9-fold at a maximally effective concentration (Fig. 2, A and C). PF-4778574 also increased the potency of S-AMPA for activating AMPARs; for example (Fig. 2B), the EC50 of S-AMPA was shifted from 4.38 μM in the absence of PF-4778574 to 0.67 μM in the presence of 300 nM PF-4778574 (near its EC50 of 282 nM).
Pharmacokinetics Studies
Species-dependent single or multiple time point neuropharmacokinetics and/or pharmacokinetics studies were conducted prior to in vivo pharmacology tests to optimize PF-4778574 dose selection and pharmacodynamic time point(s), as well as to ensure accurate Cb,u extrapolation at select time points for those doses. All pharmacokinetics data are within the Supplemental Results.
Plasma and Brain Homogenate Nonspecific Binding.
PF-4778574 had consistently low fu,p for CD-1 mice (0.0570), Sprague-Dawley rats (0.0476), beagle dogs (0.0685), and cynomolgus monkeys (0.0876), and a lower rat fu,b (0.0195). Because of fu,b being species-independent (Di et al., 2011), the rat fu,b was used to convert mouse Cb to Cb,u.
Neuropharmacokinetics
Mouse.
Thirty minutes after a single dose (0.178 mg/kg s.c.) of PF-4778574, CD-1 mice had a mean Cp of 15.7 ng/ml and Cb of 18.8 ng/g (equating to a Cb,u of 0.94 nM), resulting in a ratio of Cb to Cp (Cb:Cp) of 1.20 and a Cb,u:Cp,u of 0.41. For the same time point and dose, C57BL/6J mice had a mean ± S.D. Cp of 26.1 ± 6.5 ng/ml and Cb of 30.5 ± 6.6 ng/g (equating to a Cb,u of 1.52 ± 0.33 nM), which afforded a Cb:Cp of 1.27 ± 0.63. On the basis of these data and the pretreatment time of 0.5 hour in both the mouse cGMP and rotarod assays, doses within each model were assumed to have the same dose-Cb,u relationship as in this pharmacokinetics study and, thus, their Cb,u were extrapolated linearly from these data.
Rat.
After a single dose (1 mg/kg s.c.), PF-4778574 demonstrated similar t1/2 within plasma (0.680 hour), CSF (1.29 hours), and brain (0.804 hour), and a mean Tmax within each neuromatrix of approximately 0.25 hour after dose (Fig. 3A). Consistent with it being a highly permeable non–P-glycoprotein and non–breast cancer resistance protein substrate (Pfizer, unpublished data), PF-4778574 afforded AUC0–Tlast–derived ratios suggesting it is readily and rapidly brain penetrant (Cb:Cp of 1.07) with interneurocompartmental equilibrium, and a Cb,u:Cp,u of 0.44 (Doran et al., 2012).
Dog and nhp.
In independently conducted neuropharmacokinetics studies (Doran et al., 2012) in dogs (0.5 mg/kg PO) and nhp (0.1 mg/kg s.c.), PF-4778574 had a Cb:Cp of 1.35 and 2.82 and a Cb,u:Cp,u of 0.51 and 0.73, respectively. Rodent and large animal Cb,u:Cp,u showed a ≤1.7-fold difference (Table 2).
Pharmacokinetics
Rat.
After an intravenous bolus (0.2 mg/kg), PF-4778574 demonstrated very high CL (250 ml/min/kg) and V (4.59 l/kg) with a very short t1/2 (0.237 hour). Generated Cp versus time data were converted to Cb,u versus time (Fig. 3B) using eq. 2, and this dose-Cb,u-time relationship was used to linearly extrapolate Cb,u versus time for each dose evaluated in the rat PPF assay.
Dog.
Following administration of the highest dose (0.1 mg/kg PO) not associated with any readily apparent untoward effects, PF-4778574 revealed moderate apparent CL (8.01 ml/min/kg), high apparent V (2.1 l/kg), and moderate t1/2 (3.1 hours), with a mean Cmax (56.0 ng/ml) at 0.75 hour after dose and AUC0–∞ of 210 ng⋅h/ml. Through eq. 2, Cp versus time data were transformed to Cb,u versus time (Fig. 3C). The highest individual projected Cb,u were 4.1 and 5.9 nM.
Nonhuman Primate.
As reported (Roberts et al., 2010) for single doses (0.1 and 0.32 mg/kg s.c.), PF-4778574 had a moderate mean t1/2 (3.36 hours) with mean Tmax occurring approximately 2 hours after dose; mean Cmax (25.5 and 102 ng/ml, respectively) and AUC0–∞ (85.0 and 340 ng⋅h/ml, respectively) were linear with dose, suggesting stationary pharmacokinetics across this 3.2-fold dose range. By use of eq. 2, Cp versus time data were translated to Cb,u versus time (Fig. 3D), and this dose-Cb,u-time relationship was used for the linear projection of Cb,u versus time for the doses evaluated in the nhp SDR model. PF-4778574 was innocuous at 0.1 mg/kg, but at 0.32 mg/kg, movement-related tremors, ataxia, and decreased activity were observed from approximately 2 to 4 hours after dose, consistent with its Tmax and exposure plateau over this time period. On the basis of individual animal pharmacokinetics and the side effects observed at 0.32 mg/kg s.c., movement-related tremors and ataxia occurred in nhp at projected Cb,u of 8.0–24.1 nM (Fig. 4A).
In Vivo Pharmacology
PF-4778574 was assessed in select in vivo pharmacology models to define the relationship between its Cb,u and AMPAR-mediated effects, and the cross-species translatability of these correlations. Holistically, this afforded a Cb,u-normalized interspecies exposure-response continuum for PF-4778574 across multiple models of AMPAR-dependent mechanism, safety, and procognitive properties (“efficacy”) to quantify its single-dose TI.
Mechanism
Mouse Cerebellum cGMP Assay.
PF-4778574 dose dependently increased cGMP levels in CD-1 mouse cerebellum, with statistically significant elevations (versus vehicle-treated animals) 0.5 hour after receiving doses of 0.32 (P < 0.05) and 1.0 (P < 0.01) mg/kg s.c. (Fig. 5A). On the basis of the satellite CD-1 mouse neuropharmacokinetics study (0.178 mg/kg s.c.) and assuming stationary pharmacokinetics at all tested doses, the projected mean PF-4778574 Cb,u at the cGMP-increasing doses were 1.7 and 5.3 nM, respectively.
Safety
Mouse Rotarod Assay.
PF-4778574 induced statistically meaningful (P < 0.05) motor deficits in C57BL/6J mice undergoing an accelerating rotarod test 0.5 hour after 0.56 mg/kg s.c. (Fig. 5B). Assuming a linear dose-exposure relationship across the doses evaluated, the C57BL/6J mouse neuropharmacokinetics study (0.178 mg/kg s.c.) projected a mean PF-4778574 Cb,u of 4.8 nM to decrease animal fall latency.
Convulsion Dose-Response Studies.
Because of the concern of convulsions resulting from AMPAR potentiation and the necessity to understand both the consistency (if any) of interspecies convulsion-eliciting Cb,u and their separation from those inducing other effects, PF-4778574 underwent convulsion dose-response studies in CD-1 mice, Sprague-Dawley rats, and beagle dogs.
Rodents.
In CD-1 mice, general convulsions were observed at both 1.78 (N = 5 of 6 animals) and 3.2 (N = 6/6) mg/kg s.c.; animals receiving 1 mg/kg s.c. had no readily apparent adverse events. Following dosing, convulsions occurred later (23 ± 11 minutes) in the 1.78 mg/kg dose group than in the 3.2 mg/kg dose group (7 ± 2 minutes), which is consistent with their concentration “threshold” nature. Mean ± S.D. convulsion-causing Cb,u were 9.6 ± 0.9 and 13.2 ± 4.9 nM in the 1.78 and 3.2 mg/kg groups, respectively. Including all convulsing animals (N = 11), a Cb,u of 11.6 ± 4.0 nM was deemed the convulsive exposure threshold in CD-1 mice (Fig. 4B). In rats, convulsions were observed in both the 3.2 (N = 6/6) and 5.6 (N = 5/6) mg/kg s.c. groups; although no convulsions were observed at 1.78 mg/kg s.c., these animals (N = 5/6) became stationary approximately 20 minutes after dose. As in mice, convulsions occurred later (54 ± 27 minutes) in the 3.2 mg/kg dose group than in the 5.6 mg/kg dose group (27 ± 17 minutes). Mean ± S.D. convulsion-causing Cb,u were 10.7 ± 1.5 and 11.8 ± 1.9 nM in the 3.2 and 5.6 mg/kg groups, respectively. Collectively (N = 11), a Cb,u of 11.2 ± 1.6 nM defined the convulsive exposure threshold in Sprague-Dawley rats (Fig. 4B).
Dogs.
Doses were selected based on the initial dog pharmacokinetics study (0.1 mg/kg PO) and rodent dose-convulsion work. After 0.5 mg/kg PO (N = 2), the first dose tested, each animal experienced repetitive generalized convulsions at 14 and 18 minutes after dose that dictated their euthanasia; Cp at these time points were 182 and 207 ng/ml, respectively, which through eq. 2 and the dog Cb,u:Cp,u of 0.51 (Table 2) afford respective projected Cb,u of 16.3 and 18.5 nM (Fig. 4B). In the next dose (0.25 mg/kg PO; N = 1), whole-body tremors began approximately 15 minutes (projected Cb,u of 8.8 nM) after dosing and resolved by 1 hour after dose (Cb,u of 13.8 nM); ataxia was observed from approximately 15 minutes to 2 hours after dose (Cb,u of 12.1 nM). Although these adverse events were self-limiting and the animal survived the dose, the severity of signs was considered dose-limiting for subsequent studies. Last, after 0.2 mg/kg PO (N = 1), whole-body tremors and ataxia presented from 0.5 to 2 hours after dose (projected Cb,u of 10.9–6.0 nM) with signs being less severe than those observed in the animal receiving 0.25 mg/kg. Taken together, in dogs, PF-4778574 induced generalized tremors and ataxia at Cb,u of 6.0–15.6 nM (Fig. 4A) and convulsions at a mean Cb,u of 17.4 nM (Fig. 4B).
Efficacy
Rat MK-801–Disrupted Cortical Oscillation and PPF Assay.
This study quantified the effects of PF-4778574 on the disruption of cortical EEG and PPF by the NMDAR antagonist MK-801 in urethane-anesthetized Sprague-Dawley rats (Kiss et al., 2011a). Consistent with prior data (Kiss et al., 2011a,b), MK-801 significantly altered EEG delta activity (i.e., the regular 2-Hz delta oscillation was changed to an irregular 0.5–1.5-Hz delta rhythm) while simultaneously reducing PPF elicited by electrical stimulation of the subiculum; these MK-801–mediated effects were both sustained for 1 hour after dose and unaffected by an intravenous bolus of vehicle. Intravenous administration of PF-4778574 at 0.1 mg/kg, but not 0.03 mg/kg, significantly reversed the MK-801–induced increase in the low delta component of the local field potential and decrease in PPF (Fig. 6). After the 0.1 mg/kg bolus, these significant effects of PF-4778574 began at 5.0 ± 1.2 minutes and lasted until 20 ± 6 minutes, equating to projected Cb,u of 0.98 ± 0.23 and 0.35 ± 0.09 nM, respectively, based on linear dose-Cb,u extrapolation from the rat intravenous pharmacokinetics study (Fig. 3B). Although the ability to block the PF-4778574 effects with a selective AMPAR antagonist (e.g., CP-465022 [(S)-3-(2-chlorophenyl)-2-[2-(6-diethylaminomethyl-pyridin-2-yl)-vinyl]-6-fluoro-3H-quinazolin-4-one] [Menniti et al., 2003]) was not explored, other studies of this model have shown such antagonism to completely inhibit these same restorative effects of other AMPAR potentiators (Kiss et al., 2011a; Pfizer, unpublished data). Conceptually, results in this assay provided the electrophysiological rationale for studying PF-4778574 in the nhp ketamine-disrupted SDR task, since monosynaptic projection from the hippocampus/subiculum to the medial PFC contributes to working memory (Goldman-Rakic, 1994), and such dysfunction is implicated in the decreased cognitive performance of patients with schizophrenia (Tamminga et al., 2010; Lisman, 2012).
Nonhuman Primate Ketamine-Disrupted SDR Task.
Single doses of PF-4778574 (0.001, 0.01, and 0.1 mg/kg s.c.) were evaluated in nhp for its capacity to attenuate the working memory deficits induced by acutely disrupting glutamatergic synaptic transmission by the NMDAR antagonist ketamine. Complete details and context of this study are discussed in Roberts et al. (2010). Using the nhp pharmacokinetics studies for dose selection, pretreatment time, and the projection of evaluated Cb,u, across a 100-fold dose range PF-4778574 did not alter animal function in the absence of ketamine coadministration but did provide significant protection against ketamine-impaired performance at 0.01 mg/kg s.c. From an exposure perspective (Fig. 3D), the data suggest that a mean projected Cb,u of 0.38 nM is effective in this model, whereas 10-fold lower (0.038 nM) or higher (3.8 nM) Cb,u are not. Consistent with procognitive phenomena (Lidow et al., 1998; Calabrese, 2008; Hutson et al., 2011), PF-4778574 demonstrated a hormetic exposure-response effect in this nhp model, although the true width of the “inverted U”–shaped dose-response curve is unknown without more granular dose selection between active and inactive doses.
Discussion
The AMPAR potentiator PF-4778574 was characterized in a series of in vitro assays and acute-dose animal studies evaluating AMPAR-mediated mechanism, safety, and nootropism. Potentiator-induced animal effects were likely purely AMPAR-dependent since PF-4778574 (10 μM) only affected the dopamine transporter (IC50 of 910 nM) in a broad human-based receptor/enzyme selectivity panel (Cerep, 118 targets). For each animal dose, Cb,u was calculated to define a Cb,u-normalized AMPAR-regulated interspecies exposure-response continuum for PF-4778574 to understand separations between Cb,u enhancing cognition, disturbing motor coordination, and triggering convulsion, as well as in vitro–in vivo pharmacological associations. Ultimately, these data sets provided a single-dose, target-based TI for determining if PF-4778574 might be tested safely in humans.
Since it is unclear which AMPAR subtypes should (and/or can) be selectively targeted to balance optimal safety and efficacy, the in vitro characterization of PF-4778574 used a primary strategy of quantifying its pharmacological properties in rodent tissues that likely express native populations of AMPARs and their associated proteins. This subtype-independent approach was supported by 19 AMPAR potentiators having similar functional potencies in mouse and human ES cell–derived neurons (McNeish et al., 2010), and it was evaluated further by determining if PF-4778574 caused equivalent effects at similar Cb,u in rodents and large animals. Using the proprietary radioligand [3H]PF-04725379 that binds at the same AMPAR allosteric site as biarylpropylsulfonamide LY451646 [(R)-N-[2-(4′-cyano[1,1′-biphenyl]-4-yl)propyl]propane-2-sulfonamide] (Quirk and Nisenbaum, 2002), PF-4778574 had a mean Ki of 85 nM in a rat cortical homogenate. Functionally, PF-4778574 demonstrated glutamate-dependent AMPAR positive allosteric modulation in mouse ES cell–derived neuronal precursors, rat primary cortical neurons, and human GluA2-expressing homotetrameric systems (Fig. 2; Table 1). For perspective, the benzamide potentiator CX516 (piperidin-1-yl(quinoxalin-6-yl)methanone) (Arai and Kessler, 2007) did not displace [3H]PF-04725379, and its functional assays’ EC50 were >32 μM. Divergently, LY451646 had a Ki of 285 nM, a mouse neuronal EC50 of 1.9 μM (McNeish et al., 2010), and a human GluA2i EC50 of 117 nM. The cumulative in vitro data suggest PF-4778574 and LY451646 have a common AMPAR binding site, but PF-4778574 has similar functional potency at GluA2i and GluA2o.
Because of the convulsive risk of AMPAR potentiation (Yamada, 1998) and the unknown species-dependent dose-exposure correlation before exploratory experiments, animal studies predominantly used subcutaneous administration to control the variability in (and rise to) Cmax and the potential confounds of a first-pass-generated active metabolite. Only dogs were orally dosed since these studies determined a maximally tolerated single dose for long-term safety studies. In rodents, dogs, and nhp, PF-4778574 demonstrated interneurocompartmental equilibria (Table 2), and the experimentally determined Cb,u:Cp,u allowed the confident conversion of measured Cp,u to Cb,u when brain tissue was not excised for compound quantification. Additionally, in all species, only PF-4778574 was detected in brain homogenate, suggesting it solely caused in vivo effects.
From a mechanism and safety perspective, PF-4778574 was first evaluated in mice for its ability to elevate cerebellar cGMP, a pharmacological event linked to AMPAR potentiation (Ryder et al., 2006). In this assay, PF-4778574 showed progressively greater statistically significant effects at mean Cb,u ≥1.7 nM (Fig. 5A). Equivalent cerebellar cGMP elevations at identical Cb,u occurred in mice and rats with LY451646 (Shaffer et al., 2009), suggesting similar AMPAR-mediated exposure-effect relationships in rodents. In mice, harmaline increases the potency of AMPAR potentiators to elevate cerebellar cGMP purportedly by enhancing excitatory neurotransmitters within the olivary-cerebellar synapse (Ryder et al., 2006). Based on this and the fact that harmaline causes essential tremor by disturbing olivocerebellar rhythmicity (Deuschl and Elble, 2000), it became important to evaluate the tremorgenic potential of PF-4778574, which would be clinically dose-capping. Thus, before large animal dosing, PF-4778574 was evaluated in mice undergoing an accelerating rotarod test in which it caused motor deficits at a mean Cb,u of 4.8 nM, approximately 3-fold the minimal cGMP-elevating Cb,u (Fig. 5B). Subsequently, PF-4778574 caused general tremors in dogs at a Cb,u of 11.1 ± 3.2 nM and movement-related tremors/ataxia in nhp at a Cb,u of 15.9 ± 8.4 nM (Fig. 4A). Tremorgenic Cb,u were consistent across dogs and nhp, and their mean values were approximately 2- to 3-fold the mouse rotarod Cb,u, which effectively overlapped with the lowest tremor-causing Cb,u in dogs. Doses beyond those causing movement-related tremors/ataxia were tested in mice, rats, and dogs, but not nhp, to evaluate convulsion liability (Fig. 4B). Mice (Cb,u of 11.6 ± 4.0) and rats (Cb,u of 11.2 ± 1.6) had uniform convulsion thresholds, which were conserved in dogs (mean Cb,u of 17.4 nM). Of note, no convulsion was self-limiting, and rodent convulsions were not preceded by observable tremors. Holistically, the data suggest that PF-4778574 causes analogous adverse events at similar Cb,u across species, and progressively higher Cb,u lead from cerebellar cGMP elevation to motor coordination disruptions to convulsion.
AMPAR potentiators have demonstrated procognitive effects in numerous animal models (Black, 2005) and small clinical trials (Marenco and Weinberger, 2006). For this work, PF-4778574 was specifically evaluated in two models of pharmacologically induced NMDAR hypofunction to explore its ability to assuage glutamatergic dysregulation as it pertains to working memory impairment in schizophrenia. In a rat model (Kiss et al., 2011a,b) exploring the subiculum-medial PFC pathway, PF-4778574 overcame the disruptive effects of MK-801 on both cortical EEG and PPF at a projected Cb,u of 0.715 ± 0.276 nM (Fig. 6, A and B). This reversal is likely a direct effect of PF-4778574, as the initial MK-801–induced disruption returned once the mean Cb,u fell below that (0.3 nM) maximally projected for the ineffective dose of 0.03 mg/kg (Fig. 6C). Interestingly, genetically modified NMDAR hypomorphic mice display identical cortical and hippocampal EEG activities as observed in this rat model, and as in rats, the AMPAR potentiator LY451395 [(R)-N-(2-(4′-(2-(methylsulfonamide)ethyl)-[1,1′-biphenyl]-4-yl)propyl)propane-2-sulfonamide (mibampator)] normalized these electrophysiological signals at comparable Cb,u (Kiss et al., 2013). This demonstrates that AMPAR potentiation improves the impaired synaptic transmission and neuronal network oscillations in both pharmacological and genetic models of hypoglutamatergia. These rat electrophysiological data further rationalized studying PF-4778574 in the nhp ketamine-disrupted SDR task since monosynaptic projection from the hippocampus/subiculum to the PFC contributes to working memory (Goldman-Rakic, 1994), and such dysfunction may underlie the diminished cognitive performance of patients with schizophrenia (Tamminga et al., 2010; Lisman, 2012). In this nhp model (Roberts et al., 2010), PF-4778574 did not affect spatial working memory performance in the absence of ketamine but did block its impairing effects at a projected mean Cb,u of 0.38 nM. Furthermore, PF-4778574 did not hinder the ketamine working memory disruptions at mean Cb,u of 0.038 or 3.8 nM, mirroring a hormetic exposure-response relationship typical of nootropics (Lidow et al., 1998; Calabrese, 2008; Hutson et al., 2011). Data from both models suggest that effects deemed procognitive (i.e., “efficacy”) either electrophysiologically or behaviorally occur at essentially equivalent Cb,u.
Because of the consistent Cb,u-effect relationships across multiple species, the data sets were integrated to afford a species-normalized AMPAR-mediated exposure-response continuum for PF-4778574 that provided an overall preclinical TI (Table 3). Cognitive effects occurred at Cb,u of 0.38–0.72 nM but were lost behaviorally in nhp between mean Cb,u of 0.38 and 3.8 nM. Amid improved working memory emergence and disappearance, PF-4778574 significantly activated AMPARs within the cerebellum, as measured by cGMP elevation (Cb,u of 1.7 nM), leading to motor coordination disruptions manifested as decreased rotarod fall latency in mice (Cb,u of 4.8 nM), general tremors in dogs (Cb,u of 11.1 ± 3.2 nM), or movement-related tremors/ataxia in nhp (Cb,u of 15.9 ± 8.4 nM). Intriguingly, FDG-PET (2-deoxy-2-[18F]fluoro-D-glucose positron emission tomography) studies with LY451646 in rats (Zasadny et al., 2009) and nhp (Williams et al., 2009) demonstrated increased cerebral glucose metabolism rotating from cortical regions to the cerebellum at Cb,u paralleling the behavioral transition from improved cognition to motor coordination disruption. Closely thereafter (mean Cb,u ≥ 11.2 nM), the continuum completes with generalized convulsions in all species. From a mean exposure-response perspective, the data suggest a convulsion-based TI of 16–30 in rodents and 24–46 in dogs; for motor coordination disruptions, PF-4778574 has a TI of 7–13, 16–29, and 22–42 in mice, dogs, and nhp, respectively.
Interestingly, the Cb,u mediating physiological effects are just fractions of the concentrations required to produce half-maximal responses in the receptor-based in vitro assays (Table 1). By use of the rat Ki and respective mean Cb,u (Table 3), efficacy and convulsion happen at projected AMPAR occupancies of <1% and 12–17%, respectively, and 30 nM PF-4778574 was the lowest tested concentration that clearly enhanced S-AMPA–evoked currents in rat cortical neurons (Fig. 2, A and C). A potential explanation for this phenomenon is that animals, unlike the used in vitro systems, have fully intact, synergistic neural networks in which only minimal AMPAR potentiation affords enough excitatory neurotransmission for the nonlinear activation of downstream pathways.
Assuming that the single-dose-defined TI is not compressed upon long-term dosing and that both preclinical efficacy (mean Cb,u of 0.38–0.72 nM) and the lowest large animal tremorgenic Cb,u (6.00) are translatable, the described animal data suggest that PF-4778574 has an 8- to 16-fold TI in humans for self-limiting tremor, a readily monitorable clinical adverse event. This TI, supplemented with projected human pharmacokinetic properties and their intersubject variability at a forecasted clinically efficacious dose (Shaffer et al., 2012), suggests that PF-4778574 may be evaluated safely as a cognitive enhancer in patients with schizophrenia.
Acknowledgments
The authors acknowledge Dr. Kimberly Estep, Dr. Anton Fliri, and Randall Gallaschun for the synthesis and purification of PF-4778574; Nandini Patel for compiling the synthesis and characterization data for PF-4778574; Curt Christoffersen for conducting the in-life portion of the nhp pharmacokinetics study; and numerous colleagues within the Department of Drug Safety and Research for performing the dog single-dose escalation studies.
Authorship Contributions
Participated in research design: Bryce, Hajós, Hanks, Hoffmann, Hurst, Lotarski, Menniti, Schmidt, Scialis, Shaffer, Weber.
Conducted experiments: Bryce, Hanks, Hoffmann, Lazzaro, Liu, Lotarski, Osgood, Scialis, Weber.
Performed data analysis: Bryce, Hajós, Hanks, Hoffmann, Hurst, Lazzaro, Liu, Lotarski, Menniti, Osgood, Schmidt, Scialis, Shaffer, Weber.
Wrote or contributed to the writing of the manuscript: Hajós, Hurst, Lazzaro, Menniti, Schmidt, Shaffer, Weber.
Footnotes
- Received March 12, 2013.
- Accepted July 26, 2013.
↵1 Current affiliation: Amgen Inc., Cambridge, Massachusetts.
↵2 Current affiliation: Bristol-Myers Squibb Company, Wallingford, Connecticut.
↵3 Current affiliation: Mnemosyne Pharmaceuticals, Inc., Providence, Rhode Island.
↵4 Current affiliation: Yale School of Medicine, New Haven, Connecticut.
↵This article has supplemental material available at jpet.aspetjournals.org.
Abbreviations
- AMPA
- α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid
- ANOVA
- analysis of variance
- AUC
- area under the matrix compound concentration-time curve
- Cb
- total brain compound concentration
- CL
- systemic clearance
- Cp
- total plasma compound concentration
- Cb,u
- unbound brain compound concentration
- Cb,u:Cp,u
- ratio of unbound brain compound concentration to unbound plasma compound concentration
- CP-465022
- (S)-3-(2-chlorophenyl)-2-[2-(6-diethylaminomethyl-pyridin-2-yl)-vinyl]-6-fluoro-3H-quinazolin-4-one
- CSF
- cerebrospinal fluid
- CX516
- piperidin-1-yl(quinoxalin-6-yl)methanone
- DMSO
- dimethylsulfoxide
- EEG
- electroencephalogram
- EIA
- enzyme immunoassay
- ES
- embryonic stem
- FDG-PET
- 2-deoxy-2-[18F]fluoro-d-glucose positron emission tomography
- FLIPR
- fluorometric imaging plate reader
- fu,b
- unbound fraction of PF-4778574 in brain homogenate
- fu,p
- unbound fraction of PF-4778574 in plasma
- HEK
- human embryonic kidney
- [3H]PF-04725379
- N-((S)-1-(3,5-[3H]-2-fluoro-4-((S)-5-((1-methylethylsulfonamido)methyl)-4,5-dihydroisoxazol-3-yl)phenyl)pyrrolidin-3-yl)acetamide
- nhp
- nonhuman primate
- kel
- elimination rate constant
- LY451395
- (R)-N-(2-(4′-(2-(methylsulfonamide)ethyl)-[1,1′-biphenyl]-4-yl)propyl)propane-2-sulfonamide (mibampator)
- LY451646
- (R)-N-[2-(4′-cyano[1,1′-biphenyl]-4-yl)propyl]propane-2-sulfonamide
- MK-801
- [5R,10S]-5-methyl-10,11-dihydro-5H-dibenzo[a,d]cyclohepten-5,10-imine (dizocilpine)
- NMDAR
- N-methyl-d-aspartate receptor
- PF-4778574
- N-{(3R,4S)-3-[4-(5-cyano-2-thienyl)phenyl]tetrahydro-2H-pyran-4-yl}propane-2-sulfonamide
- PFC
- prefrontal cortex
- PPF
- paired-pulse facilitation
- RT
- room temperature
- SDR
- spatial delayed response
- t1/2
- half-life
- TI
- therapeutic index
- Tmax
- first time of Cmax
- V
- volume of distribution
- WRD
- Pfizer Worldwide Research and Development
- Copyright © 2013 by The American Society for Pharmacology and Experimental Therapeutics