- Email: [email protected]
Pharmacological evaluation of clinically relevant concentrations of (2R,6R)-hydroxynorketamine
Neuropharmacology 153 (2019) 73–81
Contents lists available at ScienceDirect
Neuropharmacology journal homepage: www.elsevier.com/locate/neuropharm
Pharmacological evaluation of clinically relevant concentrations of (2R,6R)hydroxynorketamine
T
Christopher L. Shaffera,1,∗, Jason K. Dutrab, Wei Chou Tsenga, Mark L. Webera, Luke J. Bogarta, Katherine Halesa, Jincheng Panga, Dmitri Volfsona,2, Christopher W. am Endeb, Michael E. Greena, Derek L. Buhla,2,∗∗ a b
Pfizer Worldwide Research & Development, 1 Portland Street, Cambridge, MA, 02139, United States Pfizer Worldwide Research & Development, Eastern Point Road, Groton, CT, 06340, United States
A R T I C LE I N FO
A B S T R A C T
Keywords: Ketamine (2R,6R)-hydroxynorketamine α-amino-3-hydroxy-5-methyl-4isoxazolepropionic acid receptor N-methyl-D-aspartate receptor Depression
Ketamine is a rapid-onset antidepressant whose efficacy long outlasts its pharmacokinetics. Multiple studies suggest ketamine's antidepressant effects require increased α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptor (AMPAR)-dependent currents, which have recently been exclusively attributed to its N-methyl-Daspartate receptor-inactive metabolite (2R,6R)-hydroxynorketamine ((2R,6R)-HNK). To investigate this AMPARactivation claim further, we estimated and evaluated preclinically and clinically relevant unbound brain HNK concentrations (Cb,u). (2S,6S)-HNK and (2R,6R)-HNK were novelly synthesized, and their neuropharmacokinetic profiles were determined to project relevant Cb,u. Using concentrations (0.01–10 μM) bracketing the pertinent cross-species Cb,u, both compounds' AMPAR modulation was assessed in vitro by electrophysiological recordings and GluA1 surface expression. Neither (2S,6S)-HNK nor (2R,6R)-HNK bound orthosterically to or directly functionally activated AMPARs. (2R,6R)-HNK failed to evoke AMPAR-centric changes in any electrophysiological endpoint from adult rodent hippocampal slices. Conversely, time- and concentration-dependent increases in GluA1 expression occurred only with (2R,6R)-HNK (≥0.1 μM at ≥90 min). The (2R,6R)-HNK concentrations that increased GluA1 expression are consistent with its maximal Cb,u (0.92–4.84 μM) at reportedly efficacious doses of ketamine or (2R,6R)-HNK in mouse depression models, but ≥3-fold above its projected maximal human Cb,u (≤37.8 ± 14.3 nM) following ketamine's clinically antidepressant infusion. These findings provide insight into the observed AMPAR-affecting (2R,6R)-HNK concentrations versus its exposures attained clinically at an antidepressant ketamine dose. To optimize any clinical study with (2R,6R)-HNK to fully assess its translational pharmacology, future preclinical work should test (2R,6R)-HNK concentrations and/or Cb,u of 0.01–0.1 μM to parallel its projected human Cb,u at a clinically antidepressant ketamine dose.
1. Introduction Following an acute intravenous infusion (0.5 mg/kg, 40 min), the non-selective N-methyl-D-aspartate receptor (NMDAR) antagonist ( ± )-ketamine (ketamine) reproducibly affords rapid and prolonged antidepressant effects in patients with major depressive disorder (MDD) (Murrough et al., 2017). Ketamine's adverse events (AE) (Newcomer et al., 1999) temporally correlate with ketamine and norketamine (its less-potent NMDAR-antagonist metabolite) exposure-projected NMDAR
occupancy (RO) while its efficacy is sustained long after RO ends (i.e. hysteresis) (Shaffer et al., 2014). These observations suggest ketamine's AE are indeed NMDAR-mediated whereas its antidepressant mechanism (s) remain unclear, but with two predominately debated theories: an NMDAR-dependent process that takes time to manifest and hence lags ketamine/norketamine RO, which merely initiates the requisite intracellular cascade(s); and/or, a ketamine metabolite (other than norketamine) that antagonizes NMDARs well beyond that of ketamine/ norketamine or induces efficacy via an NMDAR-independent pathway.
∗
Corresponding author. External Portfolio Innovation Unit, Research and Development, Biogen Inc., 225 Binney Street, Cambridge, MA, 02142, United States Corresponding author. Translational Neuroscience and Experimental Medicine, Takeda Pharmaceuticals, 40 Landsdowne Street, Cambridge, MA, 02139, United States E-mail addresses: christopher.shaff[email protected] (C.L. Shaffer), [email protected] (D.L. Buhl). 1 Present Address: Biogen Inc., 225 Binney Street, Cambridge, MA 02142. 2 Present Address: Takeda Pharmaceuticals, 40 Landsdowne Street, Cambridge, MA 02139. ∗∗
https://doi.org/10.1016/j.neuropharm.2019.04.019 Received 24 September 2018; Received in revised form 1 April 2019; Accepted 17 April 2019 Available online 20 April 2019 0028-3908/ © 2019 Elsevier Ltd. All rights reserved.
Neuropharmacology 153 (2019) 73–81
C.L. Shaffer, et al.
Fig. 1. (2R,6R)-HNK does not directly functionally activate AMPARs. (A) Chemical structures of (2R,6R)-HNK, (2S,6S)-HNK and PF-4778574. (B) Percent change (mean ± SEM) from baseline in AMPAR-mediated current recorded from rat cortical primary neurons, evoked using SAMPA (30 μM). Application of (2R,6R)-HNK or (2S,6S)-HNK (10 μM) did not increase AMPAR-mediated currents, whereas PF4778574 (3 μM) significantly increased currents (1-way ANOVA followed by Dunnett's post hoc test; F(3,19) = 42.89, P < 0.0001). Inset: Representative current traces for each condition. (C) In the adult rat hippocampal slice, (2R,6R)-HNK (10 μM) application caused no observed increase in field EPSPs recorded in CA1 str. radiatum following Schaffer collateral stimulation (Time: F(32,736) = 0.6858, P = 0.9057; Treatment: F(1,32) = 2.988, P = 0.0973). Representative traces are shown to the right: baseline (dark blue), vehicle (black), (2R,6R)-HNK (green) and DNQX (gray). (D–F) In the adult mouse hippocampal slice, acute whole-cell manual patch recordings from CA1 str. radiatum GFP+ interneurons did not exhibit significant changes in spontaneous AMPAR-mediated EPSC frequency (E; t = 2.008, df = 5, P = 0.1010) or amplitude (F; t = 0.5328, df = 5, P = 0.6170) following (2R,6R)-HNK (10 μM) application. Representative traces from baseline conditions and following application of (2R,6R)HNK (10 μM) are shown in (D).
HNK). Given our interest (Patel et al., 2013; Shaffer et al., 2013, 2015) in AMPAR potentiators, we made (2R,6R)-HNK and (2S,6S)-HNK (Fig. 1A) to better characterize in vitro their AMPAR activity specifically and
The latter is suggested by recent work (Zanos et al., 2016) reporting ketamine's antidepressant mechanism is via α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptor (AMPAR) activation by its NMDAR-inactive metabolite (2R,6R)-hydroxynorketamine ((2R,6R)74
Neuropharmacology 153 (2019) 73–81
C.L. Shaffer, et al.
cells.
broad pharmacology generally. Both enantiomers were synthesized to ensure no inadvertent historical stereocenter misassignment that could ultimately mislead researchers. To enable a stronger translational pharmacology assessment of (2R,6R)-HNK, we generated select in vitro neuropharmacokinetics-based data to supplement published rodent and human exposures to determine unbound brain compound concentrations (Cb,u) that are efficacious preclinically and achieved clinically, following the antidepressant infusion of ketamine. Interstitial fluid compound concentration (CISF) dictates ligand interactions with AMPAR orthosteric or allosteric binding sites (Shaffer, 2010), and a valid CISF surrogate is Cb,u (Doran et al., 2012; Liu et al., 2009). These in vivo-relevant exposures informed test compound concentration(s) in our AMPAR-centric in vitro pharmacology studies. Our findings are intended to help further elucidate the pharmacological mechanism(s) underlying ketamine's distinct antidepressant properties.
2.3.3. Calculation of unbound brain compound concentration (Cb,u) Preclinically and clinically relevant (2R,6R)-HNK Cb,u (nM units) were calculated using established methodology (Doran et al., 2012; Feng et al., 2018; Sawant-Basak et al., 2017; Shaffer, 2010) and published in vivo exposure data. For preclinical studies in which total brain (2R,6R)-HNK concentrations (Cb) are reported, Cb,u was determined by Cb, (2R,6R)-HNK molecular weight (MW; 239.5 ng/nmol), rat fu,b, an assumed brain tissue density of 1 g/mL and Eq. (1):
( C MW ) × 1000 × f b
u,b
= Cb,u
(1)
For clinical studies in which total plasma (2R,6R)-HNK concentrations (Cp) are reported, Cb,u was projected from any Cp using (2R,6R)HNK MW, human fu,p, the rat-determined unbound brain (2R,6R)-HNK concentration-to-unbound plasma (2R,6R)-HNK concentration ratio (Cb,u:Cp,u) (Doran et al., 2012), derived via Eq. (2) using the published (Moaddel et al., 2015) rat total brain-to-total plasma compound concentration ratio (Cb:Cp), and Eq. (3):
2. Materials and Methods 2.1. General All commercially available chemicals, reagents and solvents were used as received. PF-4778574 and Ketaset® (100 mg ( ± )-ketamine/ mL) were purchased from Sigma-Aldrich (St. Louis, MO) and Zoetis (Parsippany, NJ), respectively. Adult male Sprague-Dawley rats (Charles River Laboratories, Wilmington, MA) and Gad1-EGFP mice (The Jackson Laboratory, Bar Harbor, ME) were used for experiments. Rat primary neurons (DIV28─30) were employed for GluA1 expression tests. Animal studies were performed in accordance with the Guide for the Care and Use of Laboratory Animals using methods approved by Pfizer's Institutional Animal Care and Use Committee.
(Cb:Cp) × ⎛ fu,b f ⎞ = Cb,u:Cp,u u,p ⎝ ⎠
(2)
(C
(3)
p
)
MW × 1000 × fu,p × Cb,u:Cp,u = Cb,u
The calculated preclinical and clinical Cb,u guided the selection of (2R,6R)-HNK concentrations for in vitro pharmacology experiments. 2.4. In vitro pharmacology 2.4.1. Rat cortical neuron electrophysiology Rat fetal (E18) cortex primary neurons were isolated and grown on 12 mm round poly-D-lysine/laminin-coated cover slips (BD Biosciences, San Jose, CA). Rat neuronal cultures were prepared using similar methods as previously described for mouse cortical cultures (Shen et al., 2014). Approximately 1 × 105 cells/well were plated on cover slips in 12-well tissue culture plates 2–3 weeks prior to recording. Cover slips with primary neurons were placed in a recording chamber and superfused (1–2 mL/min) with external buffer containing (mM): NaCl (137), KCl (2), MgCl2 (1.5), CaCl2 (2), HEPES (10) and glucose (10), pH 7.40, and 305–310 mOsM at RT. Glass electrodes (1.5 mm O.D.) were filled with internal solution containing (mM): CsCH3SO3 (132), NaCl (2), EGTA (0.6), HEPES (10), MgATP (4), NaGTP (0.4) and QX-314 (5), pH 7.20, and 290–295 mOsM, and had a tip resistance of 3–4 MΩ. Patch clamp experiments were conducted using an Axoclamp 200B amplifier (Molecular Devices, LLC, Sunnyvale, CA) in the whole-cell configuration. Using a pCLAMP 10 software (Molecular Devices, LLC) protocol, cell membrane potential was voltage-clamped at −60 mV. Threesecond compound applications were made via gravity feed using a 3barreled glass tube connected to a SF-77B fast-step perfusion system (Warner Instruments, LLC, Hamden, CT) with a 60-s washout period between applications. For (2R,6R)-HNK, (2S,6S)-HNK, and PF-4778574 experiments, an S-AMPA-evoked (30 μM; Tocris Bioscience, Bristol, UK) current stable baseline was achieved and then S-AMPA (30 μM) + HNK (10 μM) or PF-4778574 (3 μM) was applied for 1–17 min to evaluate modulation of the agonist current by either HNK isomer or the AMPAR potentiator PF-4778574. Data were calculated as % S-AMPA baseline current response. Data were visualized and currents measured using Clampfit 10 software (Molecular Devices, LLC).
2.2. Synthesis of (2R,6R)-hydroxynorketamine hydrochloride ((2R,6R)HNK•HCl) and (2S,6S)-hydroxynorketamine hydrochloride ((2S,6S)HNK•HCl) Both (2R,6R)-HNK•HCl and (2S,6S)-HNK•HCl were enantioselectively generated via a novel synthetic route and chemically characterized (i.e. 1H and 13C NMR, HRMS and single-crystal X-ray crystallography) by Pfizer Medicine Design (Groton, CT). Full synthetic details and characterization data, including pharmacological profiles for both HNK enantiomers (10 μM) in a broad human-based 118-target receptor/transporter/enzyme selectivity panel (Eurofins Cerep SA, Celle-Lévescault, France), are within Supplementary Material (see Appendix A). 2.3. Neuropharmacokinetics 2.3.1. Plasma and brain homogenate nonspecific binding Using a reported equilibrium dialysis procedure (Doran et al., 2012), the unbound fractions of (2R,6R)-HNK and (2S,6S)-HNK (1 μM, N = 3–4/species) in rat and human plasma (fu,p) and in rat brain homogenate (fu,b) were determined. The stability of either HNK enantiomer in each matrix and the optimal incubation/dialysis time were determined separately prior to actual studies. 2.3.2. Blood-brain barrier efflux transporter evaluation Reported (Zhou et al., 2009) transwell assays containing human multidrug resistance protein (MDR1)-transfected Madin-Darby canine kidney (MDCK) cells (human P-glycoprotein (P-gp)) or mouse tetracycline-regulated expression system-mouse breast cancer resistance protein (TREx-Bcrp)-MDCK cells (mouse BCRP) determined if (2R,6R)HNK and (2S,6S)-HNK were substrates for either efflux transporter. In each assay (N = 2), a respective transporter efflux ratio (ER) was determined by dividing the secretory apparent permeability (Papp,B→A) by the absorptive apparent permeability (Papp,A→B) of each HNK enantiomer (2 μM) across contiguous monolayers of the transfected MDCK
2.4.2. Adult rat acute hippocampal slice field recordings Adult male Sprague-Dawley rats (8–16 wk old) were decapitated and their brains were rapidly removed and placed into ice-cold “cutting solution” artificial cerebrospinal fluid (ACSF), which was bubbled with 95% O2/5% CO2, containing (mM): Sucrose (206), KCl (3), NaH2PO4 75
Neuropharmacology 153 (2019) 73–81
C.L. Shaffer, et al.
and once equilibrated in the bath was maintained on-sample for 20 min. The frequency and peak amplitude of AMPAR-mediated spontaneous excitatory postsynaptic currents (sEPSCs) were analyzed using Minianalysis software (Synaptsoft, Inc., Fort Lee, NJ), with any rare amplifier-saturating events (i.e. spontaneous action potentials) being excluded from analysis. Paired t-tests were used to assess the effects of (2R,6R)-HNK.
(1.25), MgCl2 (7), NaHCO3 (26), Glucose (10), CaCl2 (0.5), sodium pyruvate (1) and L-ascorbate (1). Alternatively, some rats were anesthetized with isoflurane and transcardially perfused with ice-cold cutting solution before their brains were removed. Both brain preparation methods yielded equally viable slices as evidenced by similar amplitude fEPSPs. The brain was glued to stage with ethyl-2-cyanoacrylate glue and coronal slices (250 μM) were cut using a Leica VT1200S vibratome (Leica Microsystems, Wetzlar, Germany). The whole-brain slices were hemisected and slices were placed into normal ACSF containing (mM): NaCl (124), KCl (3), NaH2PO4 (1.25), NaHCO3 (26), Glucose (10), MgCl2 (1.3) and CaCl2 (2). External buffer also contained DL-APV (80 μM), CGP-52432 (2 μM), and picrotoxin (100 μM) to isolate AMPAR-mediated fEPSPs. Slice ACSF was continuously bubbled with 95% O2/5% CO2. Slices were incubated at 32 °C for 1 h before being placed onto MED64 (Alpha Med Scientific Inc., Osaka, Japan) electrode arrays (150 μm spacing). Hippocampal slice CA1 region stratum radiatum was placed over the middle two rows of electrodes and held down using nylon mesh and a platinum hold down. Temperature was maintained at 31 °C in the array chambers using inline heaters and controllers (Warner Instruments, LLC, Hamden, CT). Bath volume was reduced as much as possible to ca. 1 mL, and the flow rate was set to 5 mL/min. The inlet was positioned directly above the slice to provide rapid direct access to compounds and fresh oxygenated ACSF, which was recirculated throughout the course of the experiment. Field potentials (fEPSPs) were evoked once every 20 s using 50–75% maximal stimulation (ca. 150 μA, 0.2 ms) of the Schaffer collateral/ commissural fibers adjacent to the pyramidal cell layer. The experimental protocol was: 20-min baseline recording followed by 60 min of (2R,6R)-HNK (10 μM) or control buffer, followed by a 30-min washout and finally a 10-min application of DNQX (50 μM). Field EPSP amplitude was measured and plotted as percent baseline over time. A sample size of 10–15 slices for each treatment group was used.
2.4.4. GluA1 immunofluorescence staining Initial experiments included vehicle (0.1% (v/v) DMSO in dd.H2O), ( ± )-ketamine (1 μM), PF-4778574 (3 μM), (2S,6S)-HNK (0.1, 1 and 10 μM) or (2R,6R)-HNK (0.1, 1 and 10 μM). A subsequent experiment was performed to include 0.01 μM (2R,6R)-HNK, using vehicle and 10 μM (2R,6R)-HNK as negative and positive controls, respectively. As no inter-experimental differences were observed between these controls, the entire dataset was combined for statistical analysis. Rat pyramidal neurons isolated and cultured on cover slips were transfected with lentivirus encoding soluble GFP at DIV7. At DIV28−30, three cover slips per condition were washed twice with conditioned culture media (Neurobasal, B27, and glutamax) before incubation with each compound condition at 37 °C for 0, 30, 90 and 180 min. During the last 30 min of incubation, rat polyclonal antibody against the GluA1 (PC246, Millipore, Billerica, MA) subunit was added 1:10 to cover slips. All cover slips were then washed three times with conditioned culture media and incubated with secondary Alexa fluor 568 antibody 1:100 (Life Technologies, Woburn, MA) at 37 °C for 30 min. Cover slips were then washed three times with PBS, fixed with 4% paraformaldehyde/4% sucrose at RT for 15 min, and then mounted with ProLong Gold solution (Thermofisher, Waltham, MA). Confocal microscopy was performed using a Zeiss inverted microscope. Neurons for analysis (2–4 per cover slip) were selected by a blind investigator. All images were analyzed and quantified using Matlab (Mathworks Inc., Natick, MA; see Supplementary Material for analysis details). Following quantification methods (see Supplementary Material), each treatment group was normalized to the mean of each treatment at Time 0. A 2way ANOVA was used to evaluate significance between the groups, followed by Dunnett's post hoc test (Supplementary Material, Table S3).
2.4.3. Adult mouse acute hippocampal whole-cell manual patch electrophysiology Acute brain slices from adult (2–4 months of age) male GAD1EGFP + mice (Jackson Labs stock #007677 – CB6-Tg(Gad1EGFP)G42Zjh/J") were obtained using standard procedures (Tseng and O'Donnell, 2007). Briefly, mice were anesthetized with isoflurane and perfused transcardially with ice-cold, oxygenated sucrose cutting solution (modified ACSF, containing (mM): Sucrose (206), NaHCO3 (26), NaH2PO4 (1.25), KCl (3), Glucose (10), CaCl2 (0.5), and MgCl2 (7)). Horizontal slices through the dorsal hippocampus were cut 300 μmthick in cutting solution on a Leica VT1200S vibratome, and transferred to an incubation chamber containing oxygenated recording solution (ACSF, containing (mM): NaCl (125), NaHCO3 (25), NaH2PO4 (1.25), KCl (3.5), Glucose (10), CaCl2 (2), and MgCl2 (1)), which was maintained at 34 °C. After a 45-min recovery from cutting, slices were transferred to the recording chamber of an electrophysiology rig superfused with oxygenated ACSF at 2 mL/min and maintained at ca. 31–33 °C, with the muscarinic antagonist atropine (0.5 μM) included in the ACSF. Glass micropipettes (resistance of 3–5 MΩ) were filled with a cesium-based internal solution (containing (mM): cesium methanesulfanate (120), HEPES (20), EGTA (0.4), MgATP (2.5), and NaGTP (0.5)), and EGFP + cells in CA1 str. radiatum and pyramidale of dorsal hippocampus were targeted for patch-clamp recordings. Upon break-in, spontaneous current responses were recorded at −55 mV, with a 250 ms-long step to −58 mV included at the start of each 30-s sweep to monitor cellular access. Signals were acquired using an Axon Instruments MultiClamp 700B amplifier and Axon Digidata 1550 digitizer (AutoMate Scientific, Inc., Berkeley, CA), and were low-pass filtered at 2 kHz during acquisition using Clampex 10.6 software (Molecular Devices, LLC). Picrotoxin (10 μM) and DL-APV (100 μM) were added to the recording solution to block GABAA-receptors and NMDA-type glutamate receptors, respectively. (2R,6R)-HNK (10 μM) was added after recording baseline AMPA-type glutamate receptor-mediated activity,
2.4.5. Statistical analyses Statistical software R version 3.2.3 (R Core Team, 2014) or GraphPad Prism 8.0 (GraphPad Software, San Diego, CA) was used for all statistical analyses. The effects of (2R,6R)-HNK, (2S,6S)-HNK and PF-4778574 on S-AMPA-mediated currents in cultured neurons (Fig. 1B) were assessed using a 1-way ANOVA. For field EPSPs (Fig. 1C), a repeated measures ANOVA was used on the last 10 min of data prior to (2R,6R)-HNK washout. Paired t-tests were used to assess the effects of (2R,6R)-HNK on AMPA-mediated sEPSC frequency and amplitude (Fig. 1E–F). For GluA1 puncta analyses (Fig. 2), puncta area density was the primary endpoint. To quantify changes within each treatment group as well as differences between treatments over time, a longitudinal mixed model two-way ANOVA (Pinheiro et al., 2014; Kuznetsova et al., 2016) was used with time, treatment and their interaction as fixed factors and cover slip as a random factor. Mean baseline levels in each arm at Time 0 were used for normalization. Significant ANOVA findings were followed with post hoc comparisons of the least-squares means estimated using the lsmeans library (Lenth, 2014); comparisons to the 0.1% DMSO group were adjusted for multiple-hypothesis testing via Dunnett′s method, separately for each timepoint. Because of the use of multiple batches, we tested that batch factor was not significant in a more general model considering batch effects. All statistical tests were performed two-tailed at a 5% level of significance.
76
Neuropharmacology 153 (2019) 73–81
C.L. Shaffer, et al.
Fig. 2. (2R,6R)-HNK increases AMPAR surface expression in a time- and concentration-dependent manner. (A) Immunofluorescence staining of surface GluA1 clusters (red) in cultured pyramidal neurons (DIV28−30). Representative rat pyramidal neurons with soluble GFP signal are shown for 0.1% DMSO (vehicle control), ( ± )-ketamine (1 μM), (2S,6S)-HNK (10 μM), and (2R,6R)-HNK (0.01 and 10 μM) at four incubation time points (0, 30, 90 and 180 min). Within this highlighted subset, only 10 μM (2R,6R)-HNK showed a statistically significant (P < 0.0001) increase at 90 and 180 min of incubation. Arrows demarcate GluA1 clusters on the neuronal surface. (B) Quantification (mean ± SEM) of GluA1 puncta for each test-compound concentration and time point. As compared to the 0.1% DMSO condition, a significant (P < 0.0001) increase of surface GluA1 cluster density was only observed for neurons treated with 1 and 10 μM (2R,6R)-HNK at 90 min or 0.1, 1 and 10 μM (2R,6R)-HNK at 180 min. Notably, neither the lowest evaluated concentration of (2R,6R)-HNK (0.01 μM) nor any other experimental compound caused any increase in GluA1 surface expression during the 180-min incubation period. See Supplementary Material Table S3 for complete experimental statistics.
77
Neuropharmacology 153 (2019) 73–81
C.L. Shaffer, et al.
(2S,6S)-HNK bind orthosterically to or directly functionally activate AMPARs. In adult rat or mouse hippocampal slices (Fig. 1C−F), in which only (2R,6R)-HNK (10 μM) was tested, (2R,6R)-HNK failed to evoke AMPAR-centric changes in any electrophysiological endpoint under the conditions assessed. It is important to note that only 10 μM (2R,6R)-HNK was evaluated in our adult rodent hippocampal slice assays because this was the only effective (2R,6R)-HNK concentration reported by Zanos et al. (2016). Therefore, hypothetically, the lack of a more broadly evaluated (2R,6R)-HNK concentration range prohibits concluding that an inverted U-shaped concentration-response curve does not exist in these slice assays. Hormesis has been observed with (2R,6R)-HNK in cultured neurons where it increased phospho-ERK levels from 1 to 50 nM, but reduced levels at ≥100 nM (Fukumoto et al., 2018).
Table 1 In vitro neuropharmacokinetic properties (mean ± SD) of (2R,6R)-HNK and (2S,6S)-HNK. Parameter
(2R,6R)-HNK
(2S,6S)-HNK
RRCK Papp,A→B ( × 10−6 cm/s)a rat fu,pb human fu,pc rat fu,bd hMDR1 NIH ERe mBCRP ERe
47.8 0.875 ± 0.067 0.633 ± 0.046 0.451 ± 0.121 1.018 0.939
31.9 0.769 ± 0.091 0.876 ± 0.065 0.282 ± 0.050 1.110 0.914
a RRCK cells with low transporter activity were isolated from Madin-Darby canine kidney cells and used to estimate passive permeability. b Unbound fraction of compound in rat plasma. c Unbound fraction of compound in human plasma. d Unbound fraction of compound in rat brain homogenate. e Ratio from the MS-based quantification of Papp,B→A by Papp,A→B of the test compound across contiguous monolayers of the transfected MDCK cells.
3.3. (2R,6R)-HNK increases GluA1-containing AMPAR surface expression in a time- and concentration-dependent manner In primary rat cortical neurons (DIV28−30), we tested the capacity of ketamine (1 μM), PF-4778574 (3 μM), (2R,6R)-HNK (0.01, 0.1, 1 and 10 μM) and (2S,6S)-HNK (0.1, 1 and 10 μM) to affect GluA1 surface expression. In these experiments, which used electrophysiological compound concentrations (Fig. 1) and those paralleling the projected human maximal Cb,u of (2R,6R)-HNK (37.8 ± 14.3 nM) and ketamine (974 ± 189 nM) (Shaffer et al., 2014) following ketamine's clinically antidepressant infusion (Zarate et al., 2012), only (2R,6R)-HNK temporally increased GluA1 surface expression in GFP-labelled pyramidal neurons achieving statistical significance (P < 0.0001) after ≥90 min of incubation (Fig. 2A–B; Supplementary Material, Table S3). Importantly, (2R,6R)-HNK was pharmacodynamically active at ≥0.1 μM and inactive at 0.01 μM. Ketamine cannot form (2R,6R)-HNK under the employed culture conditions because they lack the necessary cytochrome P450s (Desta et al., 2012; Zarate et al., 2012).
3. Results 3.1. Preclinically and clinically relevant (2R,6R)-HNK Cb,u Select in vitro neuropharmacokinetics data were generated for (2R,6R)-HNK to enable the calculation of its Cb,u from published rodent and clinical exposures. The resulting Cb,u guided the selection of (2R,6R)-HNK concentrations for in vitro pharmacology experiments. Both enantiomers' rat and human fu,p and rat fu,b were determined, in addition to their susceptibility to the two major blood-brain barrier (BBB) efflux transporters P-gp and BCRP. Each enantiomer demonstrated high fu,p (≥0.633) and moderate fu,b (≥0.282), high absorptive permeability (Papp,A→B ≥ 31.9 × 10−6 cm/s) and no P-gp or BCRP efflux (Table 1). For (2R,6R)-HNK specifically, these data, combined with a Cb:Cp of 0.73 in rats (Moaddel et al., 2015), afford a rat-derived Cb,u:Cp,u of 0.38 (see Materials and Methods, Eq. (2)). Small molecules with this neuropharmacokinetic profile typically have analogous (≤2fold different) interspecies Cb,u:Cp,u allowing the confident projection of Cb,u from a measured Cp in large animals (Doran et al., 2012; Feng et al., 2018), including humans (Sawant-Basak et al., 2017), using Eq. (3); Cb,u is an established surrogate of CISF (Doran et al., 2012; Liu et al., 2009), which interacts with AMPAR orthosteric and allosteric binding sites. Hence, at ketamine's clinically antidepressant infusion, the projected maximal human Cb,u for (2R,6R)-HNK, using its human fu,p (0.633), rat Cb,u:Cp,u (0.38) and mean ( ± SD) maximal (2S,6S;2R,6R)HNK Cp (37.6 ± 14.2 ng/mL) (Zarate et al., 2012), is 37.8 ± 14.3 nM, which is 24- to 265-fold lower than the (2R,6R)-HNK concentrations reported to affect AMPAR-mediated currents in vitro (10 μM) (Zanos et al., 2016) and to show efficacy in mouse depression models (maximal Cb,u of 0.92–4.84 μM, per Eq. (1)) (Pham et al., 2018; Zanos et al., 2016). This maximal (2R,6R)-HNK Cb,u is a best-case value as it assumes the (2S,6S;2R,6R)-HNK Cp is exclusively (2R,6R)-HNK, which is highly unlikely following a ( ± )-ketamine dose.
4. Discussion It is imperative to decipher the pharmacological mechanism(s) causing ketamine's distinctive, and consistently clinically reproducible, rapid and prolonged antidepressant effect. To enable such research, we previously reported (Shaffer et al., 2014) an interspecies NMDAR occupancy normalization approach to enable translational pharmacology studies with ketamine and norketamine, leveraging their respective neuropharmacokinetics, in the hope that their future preclinically studied doses afford the human NMDAR occupancy-time relationship at ketamine's clinically antidepressant infusion (0.5 mg/kg, 40 min). Based on the purported role of ketamine's metabolite (2R,6R)-HNK in its antidepressant activity (Zanos et al., 2016), we have again applied established translational neuropharmacokinetics methodology (Doran et al., 2012) to estimate currently clinically relevant Cb,u of (2R,6R)HNK and subsequently investigated both these and reported preclinically effective exposures in an AMPAR-centric fashion. First, our work uniquely addresses the highly debated (Abdallah, 2017; Collingridge et al., 2017; Suzuki et al., 2017; Zanos et al., 2017) uncertainty around clinically relevant central (2R,6R)-HNK exposures at ketamine's clinical antidepressant dose (Zarate et al., 2012). A combination of our in vitro analyses (Table 1) and published mouse (1.2–1.4) (Yamaguchi et al., 2018; Zanos et al., 2016) and rat (0.73) (Moaddel et al., 2015) Cb:Cp suggests (2R,6R)-HNK has a Cp,u-favoring asymmetry at the BBB in rats (Cb,u:Cp,u of 0.38). Per our small-molecule experience (Doran et al., 2012; Feng et al., 2018; Sawant-Basak et al., 2017; Zhang et al., 2014), (2R,6R)-HNK's neuropharmacokinetic profile suggests its Cb,u:Cp,u will be analogous (≤2-fold difference) in primates, including humans. Accordingly, we estimate that ketamine's clinically antidepressant infusion affords a maximal human (2R,6R)-HNK Cb,u of 37.8 ± 14.3 nM, which is 24- to 265-fold lower than the (2R,6R)-HNK concentrations reported to affect AMPAR-mediated currents in vitro
3.2. (2R,6R)-HNK does not pharmacologically activate AMPARs We next determined if each enantiomer (10 μM) (Zanos et al., 2016) bound to and functionally activated native AMPARs. In a 118-target human-based receptor/transporter/enzyme selectivity panel (Cerep) that contained an AMPAR orthosteric-site binding assay, neither HNK enantiomer significantly interacted with any target, including NMDARs (Supplementary Material, Tables S1 and S2). Furthermore, neither enantiomer had any AMPAR-dependent functional effects ( ± S-AMPA) in rat cortical primary neurons, whereas the well-characterized AMPAR potentiator PF-4778574 (Shaffer et al., 2013) (3 μM) significantly and immediately potentiated AMPAR-mediated currents (Fig. 1B). Holistically, these data suggest that at 10 μM neither (2R,6R)-HNK nor 78
Neuropharmacology 153 (2019) 73–81
C.L. Shaffer, et al.
(10 μM) (Zanos et al., 2016) and to show efficacy in mouse depression models (maximal Cb,u of 1.13–4.84 μM (Zanos et al., 2016) or Cb,u,30min of 0.92–3.67 μM (Pham et al., 2018); per Eq. (1)). Critically, this projected maximal Cb,u is a best-case value as it assumes the (2S,6S;2R,6R)HNK Cp from which it is derived is exclusively (2R,6R)-HNK, a highly unlikely scenario given both HNK enantiomers arise via inconsequential enantioselective hydroxylation of ( ± )-norketamine (Desta et al., 2012). Based on this and the definition of a racemate, only half of the reported (2S,6S;2R,6R)-HNK Cp is expected to be (2R,6R)-HNK. Therefore, the most likely clinically relevant (2R,6R)-HNK projected maximal human Cb,u is 18.9 ± 7.2 nM, which is 48- to 530-fold below preclinically effective Cb,u. Importantly, these calculations greatly enable the back-translation of currently clinically relevant (2R,6R)-HNK exposures to preclinical mechanistic studies to best elucidate the diverse actions of (2R,6R)-HNK (Collo et al., 2018; Fukumoto et al., 2018; Wray and Rasenick, 2018). Second, our work further investigated the suggestion that (2R,6R)HNK activates AMPARs (Zanos et al., 2016). In a 118-target humanbased receptor/transporter/enzyme selectivity panel (Cerep) that contained an AMPAR orthosteric-site binding assay, (2R,6R)-HNK did not significantly interact with any target, including NMDARs (Supplementary Material, Table S1). Furthermore, (2R,6R)-HNK lacked any AMPAR-dependent functional effects ( ± S-AMPA) in rat cortical primary neurons or adult rat or mouse hippocampal slices (Fig. 1B−F). These data unequivocally demonstrate that at 10 μM (2R,6R)-HNK does not bind orthosterically to or have immediate functional activity at AMPARs, suggesting direct AMPAR activation by (2R,6R)-HNK at ≤10 μM is likely not an antidepressant mechanism of ketamine. A similar conclusion applies to a proposed (Suzuki et al., 2017) direct NMDAR-dependent mechanism for (2R,6R)-HNK as it consistently lacks NMDAR binding/activity at 10 μM (Morris et al., 2017; Suzuki et al., 2017; Zanos et al., 2016). However, we did observe the ability of (2R,6R)-HNK, but not ketamine (at its clinical-antidepressant maximal Cb,u of ca. 1 μM), PF-4778574 or (2S,6S)-HNK, to temporally (≥90 min of incubation) increase GluA1 expression in pyramidal neurons (Fig. 2) at concentrations (0.1–10 μM) paralleling its maximal Cb,u (0.92–4.84 μM) at efficacious doses of either ketamine or (2R,6R)-HNK itself in mouse depression models (Pham et al., 2018; Zanos et al., 2016). This finding suggests (2R,6R)-HNK may indeed influence ketamine's antidepressant effects via indirect AMPAR manipulation. These data imply that such metaplasticity does take time and may help explain ketamine's aforementioned hysteresis between NMDAR occupancy and efficacy (Shaffer et al., 2014) as well as elevated GluA1/ GluA2 levels in hippocampal synaptosomes 24 h, but not 1 h, after dosing mice ketamine or (2R,6R)-HNK (Zanos et al., 2016). Although reports (Yamaguchi et al., 2018) of (2R,6R)-HNK being ineffective at maximal Cb,u of 8.1–19.6 μM in specific mouse depression models has prompted questioning (Hashimoto and Shirayama, 2018) of (2R,6R)HNK's essential role (if any) in ketamine's antidepressant mechanism, ketamine's molecular and behavioral effects are persistently blocked by AMPAR-antagonist pretreatment in preclinical studies (Li et al., 2010; Maeng et al., 2008; Zanos et al., 2016), which implicates an essential role of AMPAR activation in ketamine's antidepressant mechanism(s). Accordingly, our data allow the hypothesis that (2R,6R)-HNK increases AMPAR synaptic insertion/stabilization, a signature of changing plasticity, which leads to enhanced AMPAR-mediated currents using the same endogenous glutamatergic tone as before ketamine or (2R,6R)HNK treatment. Such an effect could lead to enhanced and prolonged ketamine- (but not NMDAR inhibition-) dependent antidepressant efficacy due not to ketamine itself, but to its biotransformation to (2R,6R)-HNK. Due to the different experimental conditions between electrophysiological (represented in Fig. 1) and GluA1 surface expression experiments (represented in Fig. 2), we were unable to determine whether increases in GluA1 surface expression indeed result in changes in functional activity. Rodent hippocampal recordings were performed
from adult (i.e. 8−16-wk old) animals (Fig. 1), whereas the experiments assessing the impact of treatment on GluA1 expression were in hippocampal-cortical coculture (Fig. 2). Additionally, in the GluA1 surface expression experiments, cultures were incubated with vehicle, ketamine, PF-4778574, (2S,6S)-HNK or (2R,6R)-HNK for 0, 30, 90 or 180 min, with GluA1 levels significantly increasing (vs. 0 min) for only (2R,6R)-HNK at ≥90 min in a concentration-dependent manner (Fig. 2B). These longer time points are difficult to achieve in slice electrophysiology due to the challenges in maintaining the health of the neurons and tissue obtained from adult animals. Instead, for the electrophysiological experiments (summarized in Fig. 1) we focused on replicating previously published application times for 10 μM (2R,6R)HNK that demonstrated an effect on electrophysiological signals (i.e. 60 min (Fig. 1C) and 20 min (Fig. 1E–F) per Zanos et al., 2016). Thus, assuming a similar inter-system temporal relationship for such pharmacodynamic changes at 10 μM (2R,6R)-HNK, if (2R,6R)-HNK indeed ultimately increased GluA1 in the studied rodent hippocampal system then this phenomenon would not be expected to occur within the application times of the electrophysiological assessments. Furthermore, our analyses were unable to determine if longer incubation times (i.e. > 180 min) of 0.01 μM (2R,6R)-HNK in hippocampal-cortical coculture would also eventually increase GluA1 surface expression. While the in vitro-determined (2R,6R)-HNK concentrations (0.1–10 μM) increasing GluA1 surface expression are consistent with its preclinically efficacious Cb,u (Pham et al., 2018; Zanos et al., 2016), our neuropharmacokinetic calculations suggest this is not true for (2R,6R)HNK's projected maximal human Cb,u (best-case and likely values of 37.8 ± 14.3 nM and 18.9 ± 7.2 nM, respectively), which are 3-fold below and essentially equal to the minimally effective (0.1 μM) and completely ineffective (0.01 μM) (2R,6R)-HNK concentrations, respectively. Although the “clinically best-case” (2R,6R)-HNK concentration of 37.8 nM was not technically evaluated in this assay, we believe the clinically relevant exposures were indeed tested based on known (2R,6R)-HNK interindividual pharmacokinetic (PK) variability. In other words, if the true clinical (2R,6R)-HNK Cb,u is most likely between 11.8 and 26.0 nM, then 1.2-to-2.6-fold above the ineffective (2R,6R)-HNK concentration of 0.01 μM is of in vivo relevance considering the lowvariability interindividual exposure of (2R,6R)-HNK (Zarate et al., 2012). In the GluA1 surface expression experiments (Fig. 2), we covered a concentration range from 10 μM, which is the relevant “AMPAR activation” (2R,6R)-HNK concentration reported by Zanos et al. (2016), to just at or below the most likely projected clinical (2R,6R)-HNK Cb,u of 11.8–26.0 nM following an antidepressant infusion of ketamine (Zarate et al., 2012). Thus, using standard log-concentration differences to explore the full exposure-response of (2R,6R)-HNK, we selected the (2R,6R)-HNK concentrations of 10, 1, 0.1 and 0.01 μM. These concentrations, along with all the necessary controls (i.e. vehicle, PF4778574, ketamine and 3 similarly preclinically/clinically relevant (2S,6S)-HNK concentrations) resulted in a robust comparison that demonstrated only (2R,6R)-HNK concentration-dependently increased GluA1 surface expression. Our conclusions assume a one-to-one (and interspecies) in vitro-in vivo pharmacological correlation for these rat pyramidal neuron-based observations. In our experience (Patel et al., 2013; Shaffer et al., 2013, 2015) with AMPAR potentiators, mechanism-based in vivo animal activity (both desired and deleterious) were religiously observed at Cb,u that are magnitudes lower than potentiator concentrations eliciting functional activity in vitro, but effectively identical to projected human Cb,u that are clinically effective and safe/well-tolerated (Ranganathan et al., 2017; Shaffer, 2018). Thus, to inform such translational pharmacology appropriately and to optimize any decisions on the relevance of these in vitro-derived (2R,6R)-HNK-mediated GluA1 expression data to actual clinical exposures, future laboratory work should test (2R,6R)HNK doses that afford Cb,u of 0.01–0.1 μM and evaluate more pharmacodynamic time points than just 1 and 24 h post-dose. Carefully determining (2R,6R)-HNK's in vivo-based exposure-response continuum 79
Neuropharmacology 153 (2019) 73–81
C.L. Shaffer, et al.
is vital, particularly due to the intriguing temporal parallelism of the (2R,6R)-HNK-mediated increase in GluA1 surface expression reaching statistical significance (vs. 0 min) after 90–180 min of incubation relative to the typical time (i.e. 190 min after the 40 min ketamine infusion) (Zarate et al., 2012) of the first clinical detection of ketamine's antidepressant activity; (2S,6S;2R,6R)-HNK clinical exposure peaks 70–190 min after infusion cessation (Zarate et al., 2012) because of the time required to metabolize ketamine to (2S,6S;2R,6R)-HNK. Such preclinical data will dictate not only any further implications in the indirect increase in AMPAR signaling by (2R,6R)-HNK in ketamine's antidepressant activity, but also the potential clinical testing of (2R,6R)HNK itself by allowing the careful determination of its projected human efficacious concentration for the calculation of high-confidence therapeutic indices.
R., O'Loughlin, K., Wainer, I.W., 2015. The distribution and clearance of (2S,6S)hydroxynorketamine, an active ketamine metabolite, in Wistar rats. Pharmacol Res Perspect 3, e00157. Morris, P.J., Moaddel, R., Zanos, P., Moore, C.E., Gould, T., Zarate Jr., C.A., Thomas, C.J., 2017. Synthesis and N-methyl-D-aspartate (NMDA) receptor activity of ketamine metabolites. Org. Lett. 19, 4572–4575. Murrough, J.W., Abdallah, C.G., Mathew, S.J., 2017. Targeting glutamate signalling in depression: progress and prospects. Nat. Rev. Drug Discov. 16, 472–486. Newcomer, J.W., Farber, N.B., Jevtovic-Todorovic, V., Selke, G., Melson, A.K., Hershey, T., Craft, S., Olney, J.W., 1999. Ketamine-induced NMDA receptor hypofunction as a model of memory impairment in psychosis. Neuropsychopharmacology 20, 106–118. Patel, N.C., Schwarz, J., Hou, X.J., Hoover, D.J., Xie, L., Fliri, A.J., Gallaschun, R.J., Lazzaro, J.T., Bryce, D.K., Hoffman, W.E., Hanks, A.N., McGinnis, D., Marr, E.S., Gazard, J.L., Hajos, M., Scialis, R.J., Hurst, R.S., Shaffer, C.L., Pandit, J., O'Donnell, C.J., 2013. Discovery and characterization of a novel dihydroisoxazole class of αamino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptor potentiators. J. Med. Chem. 56, 9180–9191. Pham, T.H., Defaix, C., Xu, X., Deng, S.-X., Fabresse, N., Alvarez, J.-C., Landry, D.W., Brachman, R.A., Denny, C.A., Gardier, A.M., 2018. Common neurotransmission recruited in (R,S)-ketamine and (2R,6R)-hydroxynorketamine-induced sustained antidepressant-like effects. Biol. Psychiatry 84, e3–e6. Pinheiro, J., Bates, D., DebRoy, S., Sarkar, D., R Core Team, 2014. Nlme: Linear and Nonlinear Mixed Effects Models. R package version 3.0.1. http://CRAN.R-project. org/package=nlme. R Core Team, 2014. R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria. http://www.R-project.org/. Ranganathan, M., DeMartinis, N., Huguenel, B., Gaudreault, F., Bednar, M.M., Shaffer, C.L., Gupta, S., Cahill, J., Sherif, M.A., Mancuso, J., Zumpano, L., D'Souza, D.C., 2017. Attenuation of ketamine-induced impairment in verbal learning and memory in healthy volunteers by the AMPA receptor potentiator PF-04958242. Mol. Psychiatr. 22, 1633–1640. Sawant-Basak, A., Chen, L., Shaffer, C.L., Palumbo, D., Schmidt, A., Tseng, E., Spracklin, D.K., Gallezot, J.-D., Labaree, D., Nabulsi, N., Huang, Y., Carson, R.E., McCarthy, T., 2017. Quantitative projection of human brain penetration of the H3 antagonist PF03654746 by integrating rat-derived brain partitioning and PET receptor occupancy. Xenobiotica 47, 119–126. Shaffer, C.L., 2010. Defining neuropharmacokinetic parameters in CNS drug discovery to determine cross-species pharmacologic exposure-response relationships. In: Macor, J.E. (Ed.), Annual Reports in Medicinal Chemistry. Elsevier, New York, pp. 55–70. Shaffer, C.L., 2018. Reverse and forward translational pharmacology in psychiatric drug discovery. Clin. Pharmacol. Ther. 103, 193–195. Shaffer, C.L., Hurst, R.S., Scialis, R.J., Osgood, S.M., Bryce, D.K., Hoffmann, W.E., Lazzaro, J.T., Hanks, A.N., Lotarski, S., Weber, M.L., Liu, J., Menniti, F.S., Schmidt, C.J., Hajos, M., 2013. Positive allosteric modulation of AMPA receptors from efficacy to toxicity: the interspecies exposure-response continuum of the novel potentiator PF4778574. J. Pharmacol. Exp. Ther. 347, 212–224. Shaffer, C.L., Osgood, S.M., Smith, D.L., Liu, J., Trapa, P.E., 2014. Enhancing ketamine translational pharmacology via receptor occupancy normalization. Neuropharmacology 86, 174–180. Shaffer, C.L., Patel, N.C., Schwarz, J., Scialis, R.J., Wei, Y., Hou, X.J., Xie, L., Karki, K., Bryce, D.K., Osgood, S.M., Hoffmann, W.E., Lazzaro, J.T., Chang, C., McGinnis, D.F., Lotarski, S.M., Liu, J., Obach, R.S., Weber, M.L., Chen, L., Zasadny, K.R., Seymour, P.A., Schmidt, C.J., Hajos, M., Hurst, R.S., Pandit, J., O'Donnell, C.J., 2015. The discovery and characterization of the α-amino-3-hydroxy-5-mehtyl-4-isoxazolepropionic acid (AMPA) receptor potentiator N-{(3S,4S)-4-[4-(5-cyano-2thienyl)phenoxy]tetrahydrofuran-3-yl}propane-2-sulfonamide (PF-04958242). J. Med. Chem. 58, 4291–4308. Shen, W., Henry, A., Paumier, K., Li, L., Mou, K., Dunlop, J., Berger, Z., Hirst, W.D., 2014. Inhibition of glucosylceramide synthase stimulates autophagy flux in neurons. J. Neurochem. 129, 884–894. Suzuki, K., Nosyreva, E., Hunt, K.W., Kavalali, E.T., Monteggia, L.M., 2017. Effects of a ketamine metabolite on synaptic NMDAR function. Nature 546, E1–E3. Tseng, K., O'Donnell, P., 2007. D2 dopamine receptors recruit a GABA component for their attenuation of excitatory synaptic transmission in the adult rat prefrontal cortex. Synapse 61, 843–850. Wray, N.H., Rasenick, M.M., 2018. NMDA-receptor independent actions of ketamine: a new chapter in a story that's not so old. Neuropsychopharmacology 44, 220–221. Yamaguchi, J.-i., Toki, H., Qu, Y., Yang, C., Koike, H., Hashimoto, K., Mizuno-Yasuhira, A., Chaki, S., 2018. (2R,6R)-Hydroxynorketamine is not essential for the antidepressant actions of (R)-ketamine in mice. Neuropsychopharmacology 43, 1900–1907. Zanos, P., Moaddel, R., Morris, P.J., Georgiou, P., Fischell, J., Elmer, G.I., Alkondon, M., Yuan, P., Pribut, H.J., Singh, N.S., Dossou, K.S.S., Fang, Y., Huang, X.-P., Mayo, C.L., Wainer, I.W., Albuquerque, E.X., Thompson, S.M., Thomas, C.J., Zarate Jr., C.A., Gould, T.D., 2016. NMDAR inhibition-independent antidepressant actions of ketamine metabolites. Nature 533, 481–498. Zanos, P., Moaddel, R., Morris, P.J., Wainer, I.W., Albuquerque, E.X., Thompson, S.M., Thomas, C.J., Zarate Jr., C.A., Gould, T., 2017. Reply to: antidepressant actions of ketamine versus hydroxynorketamine. Biol. Psychiatry 81, e69–e71. Zarate Jr., C.A., Brutsche, N., Laje, G., Luckenbaugh, D.A., Vattem Venkata, S.L., Ramamoorthy, A., Moaddel, R., Wainer, I.W., 2012. Relationship of ketamine's plasma metabolites with response, diagnosis, and side effects in major depression. Biol. Psychiatry 72, 331–338. Zhang, L., Balan, G., Barreiro, G., Boscoe, B., Chenard, L., Cianfrogna, J., Claffey, M., Chen, L., Coffman, K., Drozda, S., Dunetz, J., Fonseca, K., Galatsis, P., Grimwood, S., Lazzaro, J., Miller, E., Mancuso, J., Reese, M., Rogers, B., Sakurada, I., Skaddan, M.,
Declaration of interest When this work was performed, each author was a full-time employee of Pfizer Inc. and owned and/or held options/restricted stock units for the company's publicly traded shares. Acknowledgments The authors acknowledge Kewa Mou for generating cortical cell cultures for experiments. Appendix A. Supplementary data Supplementary data related to this article can be found at https:// doi.org/10.1016/j.neuropharm.2019.04.019. References Abdallah, C.G., 2017. What's the buzz about hydroxynorketamine? Is it the history, the story, the debate, or the promise? Biol. Psychiatry 81, e61–e63. Collingridge, G.L., Lee, Y., Bortolotto, Z.A., Kang, H., Lodge, D., 2017. Antidepressant actions of ketamine versus hydroxynorketamine. Biol. Psychiatry 81, e65–e67. Collo, G., Cavalleri, L., Chiamulera, C., Merlo Pich, E., 2018. (2R,6R)Hydroxynorketamine promotes dendrite outgrowth in human inducible pluripotent stem cell-derived neurons through AMPA receptor with timing and exposure compatible with ketamine infusion pharmacokinetics in humans. Neuroreport 29, 1425–1430. Desta, Z., Moaddel, R., Ogburn, E.T., Xu, C., Ramamoorthy, A., Venkata, S.L.V., Sanghvi, M., Goldberg, M.E., Torjman, M.C., Wainer, I.W., 2012. Stereoselective and regiospecific hydroxylation of ketamine and norketamine. Xenobiotica 42, 1076–1087. Doran, A.C., Osgood, S.M., Mancuso, J.Y., Shaffer, C.L., 2012. An evaluation of using ratderived single-dose neuropharmacokinetic parameters to project accurately large animal unbound brain drug concentrations. Drug Metab. Dispos. 40, 2162–2173. Feng, B., Doran, A.C., Di, L., West, M.A., Osgood, S.M., Mancuso, J.Y., Shaffer, C.L., Tremaine, L., Liras, J., 2018. Prediction of human brain penetration of P-glycoprotein and breast cancer resistance protein substrates using in vitro transporter studies and animal models. J. Pharm. Sci. 107, 2225–2235. Fukumoto, K., Fogaca, M.V., Liu, R.-J., Duman, C., Kato, T., Li, X.-Y., Duman, R.S., 2018. Activity-dependent brain-derived neurotrophic factor signaling is required for the antidepressant actions of (2R,6R)-hydroxynorketamine. Proc. Natl. Acad. Sci. Unit. States Am (in press). Hashimoto, K., Shirayama, Y., 2018. What are the causes for discrepancies of antidepressant actions of (2R,6R)-hydroxynorketamine? Biol. Psychiatry 84, e7–e8. Kuznetsova, A., Brockhoff, P.B., Christensen, R.H.B., 2016. lmerTest: tests in linear mixed effects models. R package version 2.0‐32. http://CRAN.R-project.org/package= lmerTest. Lenth, R.V., 2014. Lsmeans: Least-Squares Means. R package version 2.11. http://CRAN. R-project.org/package=lsmeans. Li, N., Lee, B., Liu, R.-J., Banasr, M., Dwyer, J.M., Iwata, M., Li, X.-Y., Aghajanian, G., Duman, R.S., 2010. mTOR-Dependent synapse formation underlies the rapid antidepressant effects of NMDA antagonists. Science 329, 959–964. Liu, X., Van Natta, K., Yeo, H., Vilenski, O., Weller, P.E., Worboys, P.D., Monshouwer, M., 2009. Unbound drug concentration in brain homogenate and cerebral spinal fluid at steady state as a surrogate for unbound concentration in brain interstitial fluid. Drug Metab. Dispos. 37, 787–793. Maeng, S., Zarate Jr., C.A., Du, J., Schloesser, R.J., McCammon, J., Chen, G., Manji, H.K., 2008. Cellular mechanisms underlying the antidepressant effects of ketamine: role of α-amino-3-hydroxy-5-methylisoxazole-4-propionic acid receptors. Biol. Psychiatry 63, 349–352. Moaddel, R., Sanghvi, M., Dossou, K.S.S., Ramamoorthy, A., Green, C., Bupp, J., Swezey,
80
Neuropharmacology 153 (2019) 73–81
C.L. Shaffer, et al.
Zhou, L., Schmidt, K., Nelson, F.R., Zelesky, V., Troutman, M.D., Feng, B., 2009. The effect of breast cancer resistance protein and P-glycoprotein on the brain penetration of flavopiridol, imatinib mesylate (Gleevac), prazosin, and 2-methoxy-3-(4-(2-(5methyl-2-phenyloxazol-4-yl)ethoxy)phenyl)propanoic acid (PF-407288) in mice. Drug Metab. Dispos. 37, 946–955.
Smith, D., Stepan, A., Trapa, P., Tuttle, J., Verhoest, P., Walker, D., Wright, A., Zaleska, M., Zasadny, K., Shaffer, C.L., 2014. The discovery and preclinical characterization of 1-methyl-3-(4-methylpyridin-3-yl)-6-(pyridin-2-ylmethoxy)-1H-pyrazolo[3,4-b]pyrazine (PF470) as a highly potent, selective and efficacious mGluR5 negative allosteric modulator. J. Med. Chem. 57, 861–877.
81
Contents lists available at ScienceDirect
Neuropharmacology journal homepage: www.elsevier.com/locate/neuropharm
Pharmacological evaluation of clinically relevant concentrations of (2R,6R)hydroxynorketamine
T
Christopher L. Shaffera,1,∗, Jason K. Dutrab, Wei Chou Tsenga, Mark L. Webera, Luke J. Bogarta, Katherine Halesa, Jincheng Panga, Dmitri Volfsona,2, Christopher W. am Endeb, Michael E. Greena, Derek L. Buhla,2,∗∗ a b
Pfizer Worldwide Research & Development, 1 Portland Street, Cambridge, MA, 02139, United States Pfizer Worldwide Research & Development, Eastern Point Road, Groton, CT, 06340, United States
A R T I C LE I N FO
A B S T R A C T
Keywords: Ketamine (2R,6R)-hydroxynorketamine α-amino-3-hydroxy-5-methyl-4isoxazolepropionic acid receptor N-methyl-D-aspartate receptor Depression
Ketamine is a rapid-onset antidepressant whose efficacy long outlasts its pharmacokinetics. Multiple studies suggest ketamine's antidepressant effects require increased α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptor (AMPAR)-dependent currents, which have recently been exclusively attributed to its N-methyl-Daspartate receptor-inactive metabolite (2R,6R)-hydroxynorketamine ((2R,6R)-HNK). To investigate this AMPARactivation claim further, we estimated and evaluated preclinically and clinically relevant unbound brain HNK concentrations (Cb,u). (2S,6S)-HNK and (2R,6R)-HNK were novelly synthesized, and their neuropharmacokinetic profiles were determined to project relevant Cb,u. Using concentrations (0.01–10 μM) bracketing the pertinent cross-species Cb,u, both compounds' AMPAR modulation was assessed in vitro by electrophysiological recordings and GluA1 surface expression. Neither (2S,6S)-HNK nor (2R,6R)-HNK bound orthosterically to or directly functionally activated AMPARs. (2R,6R)-HNK failed to evoke AMPAR-centric changes in any electrophysiological endpoint from adult rodent hippocampal slices. Conversely, time- and concentration-dependent increases in GluA1 expression occurred only with (2R,6R)-HNK (≥0.1 μM at ≥90 min). The (2R,6R)-HNK concentrations that increased GluA1 expression are consistent with its maximal Cb,u (0.92–4.84 μM) at reportedly efficacious doses of ketamine or (2R,6R)-HNK in mouse depression models, but ≥3-fold above its projected maximal human Cb,u (≤37.8 ± 14.3 nM) following ketamine's clinically antidepressant infusion. These findings provide insight into the observed AMPAR-affecting (2R,6R)-HNK concentrations versus its exposures attained clinically at an antidepressant ketamine dose. To optimize any clinical study with (2R,6R)-HNK to fully assess its translational pharmacology, future preclinical work should test (2R,6R)-HNK concentrations and/or Cb,u of 0.01–0.1 μM to parallel its projected human Cb,u at a clinically antidepressant ketamine dose.
1. Introduction Following an acute intravenous infusion (0.5 mg/kg, 40 min), the non-selective N-methyl-D-aspartate receptor (NMDAR) antagonist ( ± )-ketamine (ketamine) reproducibly affords rapid and prolonged antidepressant effects in patients with major depressive disorder (MDD) (Murrough et al., 2017). Ketamine's adverse events (AE) (Newcomer et al., 1999) temporally correlate with ketamine and norketamine (its less-potent NMDAR-antagonist metabolite) exposure-projected NMDAR
occupancy (RO) while its efficacy is sustained long after RO ends (i.e. hysteresis) (Shaffer et al., 2014). These observations suggest ketamine's AE are indeed NMDAR-mediated whereas its antidepressant mechanism (s) remain unclear, but with two predominately debated theories: an NMDAR-dependent process that takes time to manifest and hence lags ketamine/norketamine RO, which merely initiates the requisite intracellular cascade(s); and/or, a ketamine metabolite (other than norketamine) that antagonizes NMDARs well beyond that of ketamine/ norketamine or induces efficacy via an NMDAR-independent pathway.
∗
Corresponding author. External Portfolio Innovation Unit, Research and Development, Biogen Inc., 225 Binney Street, Cambridge, MA, 02142, United States Corresponding author. Translational Neuroscience and Experimental Medicine, Takeda Pharmaceuticals, 40 Landsdowne Street, Cambridge, MA, 02139, United States E-mail addresses: christopher.shaff[email protected] (C.L. Shaffer), [email protected] (D.L. Buhl). 1 Present Address: Biogen Inc., 225 Binney Street, Cambridge, MA 02142. 2 Present Address: Takeda Pharmaceuticals, 40 Landsdowne Street, Cambridge, MA 02139. ∗∗
https://doi.org/10.1016/j.neuropharm.2019.04.019 Received 24 September 2018; Received in revised form 1 April 2019; Accepted 17 April 2019 Available online 20 April 2019 0028-3908/ © 2019 Elsevier Ltd. All rights reserved.
Neuropharmacology 153 (2019) 73–81
C.L. Shaffer, et al.
Fig. 1. (2R,6R)-HNK does not directly functionally activate AMPARs. (A) Chemical structures of (2R,6R)-HNK, (2S,6S)-HNK and PF-4778574. (B) Percent change (mean ± SEM) from baseline in AMPAR-mediated current recorded from rat cortical primary neurons, evoked using SAMPA (30 μM). Application of (2R,6R)-HNK or (2S,6S)-HNK (10 μM) did not increase AMPAR-mediated currents, whereas PF4778574 (3 μM) significantly increased currents (1-way ANOVA followed by Dunnett's post hoc test; F(3,19) = 42.89, P < 0.0001). Inset: Representative current traces for each condition. (C) In the adult rat hippocampal slice, (2R,6R)-HNK (10 μM) application caused no observed increase in field EPSPs recorded in CA1 str. radiatum following Schaffer collateral stimulation (Time: F(32,736) = 0.6858, P = 0.9057; Treatment: F(1,32) = 2.988, P = 0.0973). Representative traces are shown to the right: baseline (dark blue), vehicle (black), (2R,6R)-HNK (green) and DNQX (gray). (D–F) In the adult mouse hippocampal slice, acute whole-cell manual patch recordings from CA1 str. radiatum GFP+ interneurons did not exhibit significant changes in spontaneous AMPAR-mediated EPSC frequency (E; t = 2.008, df = 5, P = 0.1010) or amplitude (F; t = 0.5328, df = 5, P = 0.6170) following (2R,6R)-HNK (10 μM) application. Representative traces from baseline conditions and following application of (2R,6R)HNK (10 μM) are shown in (D).
HNK). Given our interest (Patel et al., 2013; Shaffer et al., 2013, 2015) in AMPAR potentiators, we made (2R,6R)-HNK and (2S,6S)-HNK (Fig. 1A) to better characterize in vitro their AMPAR activity specifically and
The latter is suggested by recent work (Zanos et al., 2016) reporting ketamine's antidepressant mechanism is via α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptor (AMPAR) activation by its NMDAR-inactive metabolite (2R,6R)-hydroxynorketamine ((2R,6R)74
Neuropharmacology 153 (2019) 73–81
C.L. Shaffer, et al.
cells.
broad pharmacology generally. Both enantiomers were synthesized to ensure no inadvertent historical stereocenter misassignment that could ultimately mislead researchers. To enable a stronger translational pharmacology assessment of (2R,6R)-HNK, we generated select in vitro neuropharmacokinetics-based data to supplement published rodent and human exposures to determine unbound brain compound concentrations (Cb,u) that are efficacious preclinically and achieved clinically, following the antidepressant infusion of ketamine. Interstitial fluid compound concentration (CISF) dictates ligand interactions with AMPAR orthosteric or allosteric binding sites (Shaffer, 2010), and a valid CISF surrogate is Cb,u (Doran et al., 2012; Liu et al., 2009). These in vivo-relevant exposures informed test compound concentration(s) in our AMPAR-centric in vitro pharmacology studies. Our findings are intended to help further elucidate the pharmacological mechanism(s) underlying ketamine's distinct antidepressant properties.
2.3.3. Calculation of unbound brain compound concentration (Cb,u) Preclinically and clinically relevant (2R,6R)-HNK Cb,u (nM units) were calculated using established methodology (Doran et al., 2012; Feng et al., 2018; Sawant-Basak et al., 2017; Shaffer, 2010) and published in vivo exposure data. For preclinical studies in which total brain (2R,6R)-HNK concentrations (Cb) are reported, Cb,u was determined by Cb, (2R,6R)-HNK molecular weight (MW; 239.5 ng/nmol), rat fu,b, an assumed brain tissue density of 1 g/mL and Eq. (1):
( C MW ) × 1000 × f b
u,b
= Cb,u
(1)
For clinical studies in which total plasma (2R,6R)-HNK concentrations (Cp) are reported, Cb,u was projected from any Cp using (2R,6R)HNK MW, human fu,p, the rat-determined unbound brain (2R,6R)-HNK concentration-to-unbound plasma (2R,6R)-HNK concentration ratio (Cb,u:Cp,u) (Doran et al., 2012), derived via Eq. (2) using the published (Moaddel et al., 2015) rat total brain-to-total plasma compound concentration ratio (Cb:Cp), and Eq. (3):
2. Materials and Methods 2.1. General All commercially available chemicals, reagents and solvents were used as received. PF-4778574 and Ketaset® (100 mg ( ± )-ketamine/ mL) were purchased from Sigma-Aldrich (St. Louis, MO) and Zoetis (Parsippany, NJ), respectively. Adult male Sprague-Dawley rats (Charles River Laboratories, Wilmington, MA) and Gad1-EGFP mice (The Jackson Laboratory, Bar Harbor, ME) were used for experiments. Rat primary neurons (DIV28─30) were employed for GluA1 expression tests. Animal studies were performed in accordance with the Guide for the Care and Use of Laboratory Animals using methods approved by Pfizer's Institutional Animal Care and Use Committee.
(Cb:Cp) × ⎛ fu,b f ⎞ = Cb,u:Cp,u u,p ⎝ ⎠
(2)
(C
(3)
p
)
MW × 1000 × fu,p × Cb,u:Cp,u = Cb,u
The calculated preclinical and clinical Cb,u guided the selection of (2R,6R)-HNK concentrations for in vitro pharmacology experiments. 2.4. In vitro pharmacology 2.4.1. Rat cortical neuron electrophysiology Rat fetal (E18) cortex primary neurons were isolated and grown on 12 mm round poly-D-lysine/laminin-coated cover slips (BD Biosciences, San Jose, CA). Rat neuronal cultures were prepared using similar methods as previously described for mouse cortical cultures (Shen et al., 2014). Approximately 1 × 105 cells/well were plated on cover slips in 12-well tissue culture plates 2–3 weeks prior to recording. Cover slips with primary neurons were placed in a recording chamber and superfused (1–2 mL/min) with external buffer containing (mM): NaCl (137), KCl (2), MgCl2 (1.5), CaCl2 (2), HEPES (10) and glucose (10), pH 7.40, and 305–310 mOsM at RT. Glass electrodes (1.5 mm O.D.) were filled with internal solution containing (mM): CsCH3SO3 (132), NaCl (2), EGTA (0.6), HEPES (10), MgATP (4), NaGTP (0.4) and QX-314 (5), pH 7.20, and 290–295 mOsM, and had a tip resistance of 3–4 MΩ. Patch clamp experiments were conducted using an Axoclamp 200B amplifier (Molecular Devices, LLC, Sunnyvale, CA) in the whole-cell configuration. Using a pCLAMP 10 software (Molecular Devices, LLC) protocol, cell membrane potential was voltage-clamped at −60 mV. Threesecond compound applications were made via gravity feed using a 3barreled glass tube connected to a SF-77B fast-step perfusion system (Warner Instruments, LLC, Hamden, CT) with a 60-s washout period between applications. For (2R,6R)-HNK, (2S,6S)-HNK, and PF-4778574 experiments, an S-AMPA-evoked (30 μM; Tocris Bioscience, Bristol, UK) current stable baseline was achieved and then S-AMPA (30 μM) + HNK (10 μM) or PF-4778574 (3 μM) was applied for 1–17 min to evaluate modulation of the agonist current by either HNK isomer or the AMPAR potentiator PF-4778574. Data were calculated as % S-AMPA baseline current response. Data were visualized and currents measured using Clampfit 10 software (Molecular Devices, LLC).
2.2. Synthesis of (2R,6R)-hydroxynorketamine hydrochloride ((2R,6R)HNK•HCl) and (2S,6S)-hydroxynorketamine hydrochloride ((2S,6S)HNK•HCl) Both (2R,6R)-HNK•HCl and (2S,6S)-HNK•HCl were enantioselectively generated via a novel synthetic route and chemically characterized (i.e. 1H and 13C NMR, HRMS and single-crystal X-ray crystallography) by Pfizer Medicine Design (Groton, CT). Full synthetic details and characterization data, including pharmacological profiles for both HNK enantiomers (10 μM) in a broad human-based 118-target receptor/transporter/enzyme selectivity panel (Eurofins Cerep SA, Celle-Lévescault, France), are within Supplementary Material (see Appendix A). 2.3. Neuropharmacokinetics 2.3.1. Plasma and brain homogenate nonspecific binding Using a reported equilibrium dialysis procedure (Doran et al., 2012), the unbound fractions of (2R,6R)-HNK and (2S,6S)-HNK (1 μM, N = 3–4/species) in rat and human plasma (fu,p) and in rat brain homogenate (fu,b) were determined. The stability of either HNK enantiomer in each matrix and the optimal incubation/dialysis time were determined separately prior to actual studies. 2.3.2. Blood-brain barrier efflux transporter evaluation Reported (Zhou et al., 2009) transwell assays containing human multidrug resistance protein (MDR1)-transfected Madin-Darby canine kidney (MDCK) cells (human P-glycoprotein (P-gp)) or mouse tetracycline-regulated expression system-mouse breast cancer resistance protein (TREx-Bcrp)-MDCK cells (mouse BCRP) determined if (2R,6R)HNK and (2S,6S)-HNK were substrates for either efflux transporter. In each assay (N = 2), a respective transporter efflux ratio (ER) was determined by dividing the secretory apparent permeability (Papp,B→A) by the absorptive apparent permeability (Papp,A→B) of each HNK enantiomer (2 μM) across contiguous monolayers of the transfected MDCK
2.4.2. Adult rat acute hippocampal slice field recordings Adult male Sprague-Dawley rats (8–16 wk old) were decapitated and their brains were rapidly removed and placed into ice-cold “cutting solution” artificial cerebrospinal fluid (ACSF), which was bubbled with 95% O2/5% CO2, containing (mM): Sucrose (206), KCl (3), NaH2PO4 75
Neuropharmacology 153 (2019) 73–81
C.L. Shaffer, et al.
and once equilibrated in the bath was maintained on-sample for 20 min. The frequency and peak amplitude of AMPAR-mediated spontaneous excitatory postsynaptic currents (sEPSCs) were analyzed using Minianalysis software (Synaptsoft, Inc., Fort Lee, NJ), with any rare amplifier-saturating events (i.e. spontaneous action potentials) being excluded from analysis. Paired t-tests were used to assess the effects of (2R,6R)-HNK.
(1.25), MgCl2 (7), NaHCO3 (26), Glucose (10), CaCl2 (0.5), sodium pyruvate (1) and L-ascorbate (1). Alternatively, some rats were anesthetized with isoflurane and transcardially perfused with ice-cold cutting solution before their brains were removed. Both brain preparation methods yielded equally viable slices as evidenced by similar amplitude fEPSPs. The brain was glued to stage with ethyl-2-cyanoacrylate glue and coronal slices (250 μM) were cut using a Leica VT1200S vibratome (Leica Microsystems, Wetzlar, Germany). The whole-brain slices were hemisected and slices were placed into normal ACSF containing (mM): NaCl (124), KCl (3), NaH2PO4 (1.25), NaHCO3 (26), Glucose (10), MgCl2 (1.3) and CaCl2 (2). External buffer also contained DL-APV (80 μM), CGP-52432 (2 μM), and picrotoxin (100 μM) to isolate AMPAR-mediated fEPSPs. Slice ACSF was continuously bubbled with 95% O2/5% CO2. Slices were incubated at 32 °C for 1 h before being placed onto MED64 (Alpha Med Scientific Inc., Osaka, Japan) electrode arrays (150 μm spacing). Hippocampal slice CA1 region stratum radiatum was placed over the middle two rows of electrodes and held down using nylon mesh and a platinum hold down. Temperature was maintained at 31 °C in the array chambers using inline heaters and controllers (Warner Instruments, LLC, Hamden, CT). Bath volume was reduced as much as possible to ca. 1 mL, and the flow rate was set to 5 mL/min. The inlet was positioned directly above the slice to provide rapid direct access to compounds and fresh oxygenated ACSF, which was recirculated throughout the course of the experiment. Field potentials (fEPSPs) were evoked once every 20 s using 50–75% maximal stimulation (ca. 150 μA, 0.2 ms) of the Schaffer collateral/ commissural fibers adjacent to the pyramidal cell layer. The experimental protocol was: 20-min baseline recording followed by 60 min of (2R,6R)-HNK (10 μM) or control buffer, followed by a 30-min washout and finally a 10-min application of DNQX (50 μM). Field EPSP amplitude was measured and plotted as percent baseline over time. A sample size of 10–15 slices for each treatment group was used.
2.4.4. GluA1 immunofluorescence staining Initial experiments included vehicle (0.1% (v/v) DMSO in dd.H2O), ( ± )-ketamine (1 μM), PF-4778574 (3 μM), (2S,6S)-HNK (0.1, 1 and 10 μM) or (2R,6R)-HNK (0.1, 1 and 10 μM). A subsequent experiment was performed to include 0.01 μM (2R,6R)-HNK, using vehicle and 10 μM (2R,6R)-HNK as negative and positive controls, respectively. As no inter-experimental differences were observed between these controls, the entire dataset was combined for statistical analysis. Rat pyramidal neurons isolated and cultured on cover slips were transfected with lentivirus encoding soluble GFP at DIV7. At DIV28−30, three cover slips per condition were washed twice with conditioned culture media (Neurobasal, B27, and glutamax) before incubation with each compound condition at 37 °C for 0, 30, 90 and 180 min. During the last 30 min of incubation, rat polyclonal antibody against the GluA1 (PC246, Millipore, Billerica, MA) subunit was added 1:10 to cover slips. All cover slips were then washed three times with conditioned culture media and incubated with secondary Alexa fluor 568 antibody 1:100 (Life Technologies, Woburn, MA) at 37 °C for 30 min. Cover slips were then washed three times with PBS, fixed with 4% paraformaldehyde/4% sucrose at RT for 15 min, and then mounted with ProLong Gold solution (Thermofisher, Waltham, MA). Confocal microscopy was performed using a Zeiss inverted microscope. Neurons for analysis (2–4 per cover slip) were selected by a blind investigator. All images were analyzed and quantified using Matlab (Mathworks Inc., Natick, MA; see Supplementary Material for analysis details). Following quantification methods (see Supplementary Material), each treatment group was normalized to the mean of each treatment at Time 0. A 2way ANOVA was used to evaluate significance between the groups, followed by Dunnett's post hoc test (Supplementary Material, Table S3).
2.4.3. Adult mouse acute hippocampal whole-cell manual patch electrophysiology Acute brain slices from adult (2–4 months of age) male GAD1EGFP + mice (Jackson Labs stock #007677 – CB6-Tg(Gad1EGFP)G42Zjh/J") were obtained using standard procedures (Tseng and O'Donnell, 2007). Briefly, mice were anesthetized with isoflurane and perfused transcardially with ice-cold, oxygenated sucrose cutting solution (modified ACSF, containing (mM): Sucrose (206), NaHCO3 (26), NaH2PO4 (1.25), KCl (3), Glucose (10), CaCl2 (0.5), and MgCl2 (7)). Horizontal slices through the dorsal hippocampus were cut 300 μmthick in cutting solution on a Leica VT1200S vibratome, and transferred to an incubation chamber containing oxygenated recording solution (ACSF, containing (mM): NaCl (125), NaHCO3 (25), NaH2PO4 (1.25), KCl (3.5), Glucose (10), CaCl2 (2), and MgCl2 (1)), which was maintained at 34 °C. After a 45-min recovery from cutting, slices were transferred to the recording chamber of an electrophysiology rig superfused with oxygenated ACSF at 2 mL/min and maintained at ca. 31–33 °C, with the muscarinic antagonist atropine (0.5 μM) included in the ACSF. Glass micropipettes (resistance of 3–5 MΩ) were filled with a cesium-based internal solution (containing (mM): cesium methanesulfanate (120), HEPES (20), EGTA (0.4), MgATP (2.5), and NaGTP (0.5)), and EGFP + cells in CA1 str. radiatum and pyramidale of dorsal hippocampus were targeted for patch-clamp recordings. Upon break-in, spontaneous current responses were recorded at −55 mV, with a 250 ms-long step to −58 mV included at the start of each 30-s sweep to monitor cellular access. Signals were acquired using an Axon Instruments MultiClamp 700B amplifier and Axon Digidata 1550 digitizer (AutoMate Scientific, Inc., Berkeley, CA), and were low-pass filtered at 2 kHz during acquisition using Clampex 10.6 software (Molecular Devices, LLC). Picrotoxin (10 μM) and DL-APV (100 μM) were added to the recording solution to block GABAA-receptors and NMDA-type glutamate receptors, respectively. (2R,6R)-HNK (10 μM) was added after recording baseline AMPA-type glutamate receptor-mediated activity,
2.4.5. Statistical analyses Statistical software R version 3.2.3 (R Core Team, 2014) or GraphPad Prism 8.0 (GraphPad Software, San Diego, CA) was used for all statistical analyses. The effects of (2R,6R)-HNK, (2S,6S)-HNK and PF-4778574 on S-AMPA-mediated currents in cultured neurons (Fig. 1B) were assessed using a 1-way ANOVA. For field EPSPs (Fig. 1C), a repeated measures ANOVA was used on the last 10 min of data prior to (2R,6R)-HNK washout. Paired t-tests were used to assess the effects of (2R,6R)-HNK on AMPA-mediated sEPSC frequency and amplitude (Fig. 1E–F). For GluA1 puncta analyses (Fig. 2), puncta area density was the primary endpoint. To quantify changes within each treatment group as well as differences between treatments over time, a longitudinal mixed model two-way ANOVA (Pinheiro et al., 2014; Kuznetsova et al., 2016) was used with time, treatment and their interaction as fixed factors and cover slip as a random factor. Mean baseline levels in each arm at Time 0 were used for normalization. Significant ANOVA findings were followed with post hoc comparisons of the least-squares means estimated using the lsmeans library (Lenth, 2014); comparisons to the 0.1% DMSO group were adjusted for multiple-hypothesis testing via Dunnett′s method, separately for each timepoint. Because of the use of multiple batches, we tested that batch factor was not significant in a more general model considering batch effects. All statistical tests were performed two-tailed at a 5% level of significance.
76
Neuropharmacology 153 (2019) 73–81
C.L. Shaffer, et al.
Fig. 2. (2R,6R)-HNK increases AMPAR surface expression in a time- and concentration-dependent manner. (A) Immunofluorescence staining of surface GluA1 clusters (red) in cultured pyramidal neurons (DIV28−30). Representative rat pyramidal neurons with soluble GFP signal are shown for 0.1% DMSO (vehicle control), ( ± )-ketamine (1 μM), (2S,6S)-HNK (10 μM), and (2R,6R)-HNK (0.01 and 10 μM) at four incubation time points (0, 30, 90 and 180 min). Within this highlighted subset, only 10 μM (2R,6R)-HNK showed a statistically significant (P < 0.0001) increase at 90 and 180 min of incubation. Arrows demarcate GluA1 clusters on the neuronal surface. (B) Quantification (mean ± SEM) of GluA1 puncta for each test-compound concentration and time point. As compared to the 0.1% DMSO condition, a significant (P < 0.0001) increase of surface GluA1 cluster density was only observed for neurons treated with 1 and 10 μM (2R,6R)-HNK at 90 min or 0.1, 1 and 10 μM (2R,6R)-HNK at 180 min. Notably, neither the lowest evaluated concentration of (2R,6R)-HNK (0.01 μM) nor any other experimental compound caused any increase in GluA1 surface expression during the 180-min incubation period. See Supplementary Material Table S3 for complete experimental statistics.
77
Neuropharmacology 153 (2019) 73–81
C.L. Shaffer, et al.
(2S,6S)-HNK bind orthosterically to or directly functionally activate AMPARs. In adult rat or mouse hippocampal slices (Fig. 1C−F), in which only (2R,6R)-HNK (10 μM) was tested, (2R,6R)-HNK failed to evoke AMPAR-centric changes in any electrophysiological endpoint under the conditions assessed. It is important to note that only 10 μM (2R,6R)-HNK was evaluated in our adult rodent hippocampal slice assays because this was the only effective (2R,6R)-HNK concentration reported by Zanos et al. (2016). Therefore, hypothetically, the lack of a more broadly evaluated (2R,6R)-HNK concentration range prohibits concluding that an inverted U-shaped concentration-response curve does not exist in these slice assays. Hormesis has been observed with (2R,6R)-HNK in cultured neurons where it increased phospho-ERK levels from 1 to 50 nM, but reduced levels at ≥100 nM (Fukumoto et al., 2018).
Table 1 In vitro neuropharmacokinetic properties (mean ± SD) of (2R,6R)-HNK and (2S,6S)-HNK. Parameter
(2R,6R)-HNK
(2S,6S)-HNK
RRCK Papp,A→B ( × 10−6 cm/s)a rat fu,pb human fu,pc rat fu,bd hMDR1 NIH ERe mBCRP ERe
47.8 0.875 ± 0.067 0.633 ± 0.046 0.451 ± 0.121 1.018 0.939
31.9 0.769 ± 0.091 0.876 ± 0.065 0.282 ± 0.050 1.110 0.914
a RRCK cells with low transporter activity were isolated from Madin-Darby canine kidney cells and used to estimate passive permeability. b Unbound fraction of compound in rat plasma. c Unbound fraction of compound in human plasma. d Unbound fraction of compound in rat brain homogenate. e Ratio from the MS-based quantification of Papp,B→A by Papp,A→B of the test compound across contiguous monolayers of the transfected MDCK cells.
3.3. (2R,6R)-HNK increases GluA1-containing AMPAR surface expression in a time- and concentration-dependent manner In primary rat cortical neurons (DIV28−30), we tested the capacity of ketamine (1 μM), PF-4778574 (3 μM), (2R,6R)-HNK (0.01, 0.1, 1 and 10 μM) and (2S,6S)-HNK (0.1, 1 and 10 μM) to affect GluA1 surface expression. In these experiments, which used electrophysiological compound concentrations (Fig. 1) and those paralleling the projected human maximal Cb,u of (2R,6R)-HNK (37.8 ± 14.3 nM) and ketamine (974 ± 189 nM) (Shaffer et al., 2014) following ketamine's clinically antidepressant infusion (Zarate et al., 2012), only (2R,6R)-HNK temporally increased GluA1 surface expression in GFP-labelled pyramidal neurons achieving statistical significance (P < 0.0001) after ≥90 min of incubation (Fig. 2A–B; Supplementary Material, Table S3). Importantly, (2R,6R)-HNK was pharmacodynamically active at ≥0.1 μM and inactive at 0.01 μM. Ketamine cannot form (2R,6R)-HNK under the employed culture conditions because they lack the necessary cytochrome P450s (Desta et al., 2012; Zarate et al., 2012).
3. Results 3.1. Preclinically and clinically relevant (2R,6R)-HNK Cb,u Select in vitro neuropharmacokinetics data were generated for (2R,6R)-HNK to enable the calculation of its Cb,u from published rodent and clinical exposures. The resulting Cb,u guided the selection of (2R,6R)-HNK concentrations for in vitro pharmacology experiments. Both enantiomers' rat and human fu,p and rat fu,b were determined, in addition to their susceptibility to the two major blood-brain barrier (BBB) efflux transporters P-gp and BCRP. Each enantiomer demonstrated high fu,p (≥0.633) and moderate fu,b (≥0.282), high absorptive permeability (Papp,A→B ≥ 31.9 × 10−6 cm/s) and no P-gp or BCRP efflux (Table 1). For (2R,6R)-HNK specifically, these data, combined with a Cb:Cp of 0.73 in rats (Moaddel et al., 2015), afford a rat-derived Cb,u:Cp,u of 0.38 (see Materials and Methods, Eq. (2)). Small molecules with this neuropharmacokinetic profile typically have analogous (≤2fold different) interspecies Cb,u:Cp,u allowing the confident projection of Cb,u from a measured Cp in large animals (Doran et al., 2012; Feng et al., 2018), including humans (Sawant-Basak et al., 2017), using Eq. (3); Cb,u is an established surrogate of CISF (Doran et al., 2012; Liu et al., 2009), which interacts with AMPAR orthosteric and allosteric binding sites. Hence, at ketamine's clinically antidepressant infusion, the projected maximal human Cb,u for (2R,6R)-HNK, using its human fu,p (0.633), rat Cb,u:Cp,u (0.38) and mean ( ± SD) maximal (2S,6S;2R,6R)HNK Cp (37.6 ± 14.2 ng/mL) (Zarate et al., 2012), is 37.8 ± 14.3 nM, which is 24- to 265-fold lower than the (2R,6R)-HNK concentrations reported to affect AMPAR-mediated currents in vitro (10 μM) (Zanos et al., 2016) and to show efficacy in mouse depression models (maximal Cb,u of 0.92–4.84 μM, per Eq. (1)) (Pham et al., 2018; Zanos et al., 2016). This maximal (2R,6R)-HNK Cb,u is a best-case value as it assumes the (2S,6S;2R,6R)-HNK Cp is exclusively (2R,6R)-HNK, which is highly unlikely following a ( ± )-ketamine dose.
4. Discussion It is imperative to decipher the pharmacological mechanism(s) causing ketamine's distinctive, and consistently clinically reproducible, rapid and prolonged antidepressant effect. To enable such research, we previously reported (Shaffer et al., 2014) an interspecies NMDAR occupancy normalization approach to enable translational pharmacology studies with ketamine and norketamine, leveraging their respective neuropharmacokinetics, in the hope that their future preclinically studied doses afford the human NMDAR occupancy-time relationship at ketamine's clinically antidepressant infusion (0.5 mg/kg, 40 min). Based on the purported role of ketamine's metabolite (2R,6R)-HNK in its antidepressant activity (Zanos et al., 2016), we have again applied established translational neuropharmacokinetics methodology (Doran et al., 2012) to estimate currently clinically relevant Cb,u of (2R,6R)HNK and subsequently investigated both these and reported preclinically effective exposures in an AMPAR-centric fashion. First, our work uniquely addresses the highly debated (Abdallah, 2017; Collingridge et al., 2017; Suzuki et al., 2017; Zanos et al., 2017) uncertainty around clinically relevant central (2R,6R)-HNK exposures at ketamine's clinical antidepressant dose (Zarate et al., 2012). A combination of our in vitro analyses (Table 1) and published mouse (1.2–1.4) (Yamaguchi et al., 2018; Zanos et al., 2016) and rat (0.73) (Moaddel et al., 2015) Cb:Cp suggests (2R,6R)-HNK has a Cp,u-favoring asymmetry at the BBB in rats (Cb,u:Cp,u of 0.38). Per our small-molecule experience (Doran et al., 2012; Feng et al., 2018; Sawant-Basak et al., 2017; Zhang et al., 2014), (2R,6R)-HNK's neuropharmacokinetic profile suggests its Cb,u:Cp,u will be analogous (≤2-fold difference) in primates, including humans. Accordingly, we estimate that ketamine's clinically antidepressant infusion affords a maximal human (2R,6R)-HNK Cb,u of 37.8 ± 14.3 nM, which is 24- to 265-fold lower than the (2R,6R)-HNK concentrations reported to affect AMPAR-mediated currents in vitro
3.2. (2R,6R)-HNK does not pharmacologically activate AMPARs We next determined if each enantiomer (10 μM) (Zanos et al., 2016) bound to and functionally activated native AMPARs. In a 118-target human-based receptor/transporter/enzyme selectivity panel (Cerep) that contained an AMPAR orthosteric-site binding assay, neither HNK enantiomer significantly interacted with any target, including NMDARs (Supplementary Material, Tables S1 and S2). Furthermore, neither enantiomer had any AMPAR-dependent functional effects ( ± S-AMPA) in rat cortical primary neurons, whereas the well-characterized AMPAR potentiator PF-4778574 (Shaffer et al., 2013) (3 μM) significantly and immediately potentiated AMPAR-mediated currents (Fig. 1B). Holistically, these data suggest that at 10 μM neither (2R,6R)-HNK nor 78
Neuropharmacology 153 (2019) 73–81
C.L. Shaffer, et al.
(10 μM) (Zanos et al., 2016) and to show efficacy in mouse depression models (maximal Cb,u of 1.13–4.84 μM (Zanos et al., 2016) or Cb,u,30min of 0.92–3.67 μM (Pham et al., 2018); per Eq. (1)). Critically, this projected maximal Cb,u is a best-case value as it assumes the (2S,6S;2R,6R)HNK Cp from which it is derived is exclusively (2R,6R)-HNK, a highly unlikely scenario given both HNK enantiomers arise via inconsequential enantioselective hydroxylation of ( ± )-norketamine (Desta et al., 2012). Based on this and the definition of a racemate, only half of the reported (2S,6S;2R,6R)-HNK Cp is expected to be (2R,6R)-HNK. Therefore, the most likely clinically relevant (2R,6R)-HNK projected maximal human Cb,u is 18.9 ± 7.2 nM, which is 48- to 530-fold below preclinically effective Cb,u. Importantly, these calculations greatly enable the back-translation of currently clinically relevant (2R,6R)-HNK exposures to preclinical mechanistic studies to best elucidate the diverse actions of (2R,6R)-HNK (Collo et al., 2018; Fukumoto et al., 2018; Wray and Rasenick, 2018). Second, our work further investigated the suggestion that (2R,6R)HNK activates AMPARs (Zanos et al., 2016). In a 118-target humanbased receptor/transporter/enzyme selectivity panel (Cerep) that contained an AMPAR orthosteric-site binding assay, (2R,6R)-HNK did not significantly interact with any target, including NMDARs (Supplementary Material, Table S1). Furthermore, (2R,6R)-HNK lacked any AMPAR-dependent functional effects ( ± S-AMPA) in rat cortical primary neurons or adult rat or mouse hippocampal slices (Fig. 1B−F). These data unequivocally demonstrate that at 10 μM (2R,6R)-HNK does not bind orthosterically to or have immediate functional activity at AMPARs, suggesting direct AMPAR activation by (2R,6R)-HNK at ≤10 μM is likely not an antidepressant mechanism of ketamine. A similar conclusion applies to a proposed (Suzuki et al., 2017) direct NMDAR-dependent mechanism for (2R,6R)-HNK as it consistently lacks NMDAR binding/activity at 10 μM (Morris et al., 2017; Suzuki et al., 2017; Zanos et al., 2016). However, we did observe the ability of (2R,6R)-HNK, but not ketamine (at its clinical-antidepressant maximal Cb,u of ca. 1 μM), PF-4778574 or (2S,6S)-HNK, to temporally (≥90 min of incubation) increase GluA1 expression in pyramidal neurons (Fig. 2) at concentrations (0.1–10 μM) paralleling its maximal Cb,u (0.92–4.84 μM) at efficacious doses of either ketamine or (2R,6R)-HNK itself in mouse depression models (Pham et al., 2018; Zanos et al., 2016). This finding suggests (2R,6R)-HNK may indeed influence ketamine's antidepressant effects via indirect AMPAR manipulation. These data imply that such metaplasticity does take time and may help explain ketamine's aforementioned hysteresis between NMDAR occupancy and efficacy (Shaffer et al., 2014) as well as elevated GluA1/ GluA2 levels in hippocampal synaptosomes 24 h, but not 1 h, after dosing mice ketamine or (2R,6R)-HNK (Zanos et al., 2016). Although reports (Yamaguchi et al., 2018) of (2R,6R)-HNK being ineffective at maximal Cb,u of 8.1–19.6 μM in specific mouse depression models has prompted questioning (Hashimoto and Shirayama, 2018) of (2R,6R)HNK's essential role (if any) in ketamine's antidepressant mechanism, ketamine's molecular and behavioral effects are persistently blocked by AMPAR-antagonist pretreatment in preclinical studies (Li et al., 2010; Maeng et al., 2008; Zanos et al., 2016), which implicates an essential role of AMPAR activation in ketamine's antidepressant mechanism(s). Accordingly, our data allow the hypothesis that (2R,6R)-HNK increases AMPAR synaptic insertion/stabilization, a signature of changing plasticity, which leads to enhanced AMPAR-mediated currents using the same endogenous glutamatergic tone as before ketamine or (2R,6R)HNK treatment. Such an effect could lead to enhanced and prolonged ketamine- (but not NMDAR inhibition-) dependent antidepressant efficacy due not to ketamine itself, but to its biotransformation to (2R,6R)-HNK. Due to the different experimental conditions between electrophysiological (represented in Fig. 1) and GluA1 surface expression experiments (represented in Fig. 2), we were unable to determine whether increases in GluA1 surface expression indeed result in changes in functional activity. Rodent hippocampal recordings were performed
from adult (i.e. 8−16-wk old) animals (Fig. 1), whereas the experiments assessing the impact of treatment on GluA1 expression were in hippocampal-cortical coculture (Fig. 2). Additionally, in the GluA1 surface expression experiments, cultures were incubated with vehicle, ketamine, PF-4778574, (2S,6S)-HNK or (2R,6R)-HNK for 0, 30, 90 or 180 min, with GluA1 levels significantly increasing (vs. 0 min) for only (2R,6R)-HNK at ≥90 min in a concentration-dependent manner (Fig. 2B). These longer time points are difficult to achieve in slice electrophysiology due to the challenges in maintaining the health of the neurons and tissue obtained from adult animals. Instead, for the electrophysiological experiments (summarized in Fig. 1) we focused on replicating previously published application times for 10 μM (2R,6R)HNK that demonstrated an effect on electrophysiological signals (i.e. 60 min (Fig. 1C) and 20 min (Fig. 1E–F) per Zanos et al., 2016). Thus, assuming a similar inter-system temporal relationship for such pharmacodynamic changes at 10 μM (2R,6R)-HNK, if (2R,6R)-HNK indeed ultimately increased GluA1 in the studied rodent hippocampal system then this phenomenon would not be expected to occur within the application times of the electrophysiological assessments. Furthermore, our analyses were unable to determine if longer incubation times (i.e. > 180 min) of 0.01 μM (2R,6R)-HNK in hippocampal-cortical coculture would also eventually increase GluA1 surface expression. While the in vitro-determined (2R,6R)-HNK concentrations (0.1–10 μM) increasing GluA1 surface expression are consistent with its preclinically efficacious Cb,u (Pham et al., 2018; Zanos et al., 2016), our neuropharmacokinetic calculations suggest this is not true for (2R,6R)HNK's projected maximal human Cb,u (best-case and likely values of 37.8 ± 14.3 nM and 18.9 ± 7.2 nM, respectively), which are 3-fold below and essentially equal to the minimally effective (0.1 μM) and completely ineffective (0.01 μM) (2R,6R)-HNK concentrations, respectively. Although the “clinically best-case” (2R,6R)-HNK concentration of 37.8 nM was not technically evaluated in this assay, we believe the clinically relevant exposures were indeed tested based on known (2R,6R)-HNK interindividual pharmacokinetic (PK) variability. In other words, if the true clinical (2R,6R)-HNK Cb,u is most likely between 11.8 and 26.0 nM, then 1.2-to-2.6-fold above the ineffective (2R,6R)-HNK concentration of 0.01 μM is of in vivo relevance considering the lowvariability interindividual exposure of (2R,6R)-HNK (Zarate et al., 2012). In the GluA1 surface expression experiments (Fig. 2), we covered a concentration range from 10 μM, which is the relevant “AMPAR activation” (2R,6R)-HNK concentration reported by Zanos et al. (2016), to just at or below the most likely projected clinical (2R,6R)-HNK Cb,u of 11.8–26.0 nM following an antidepressant infusion of ketamine (Zarate et al., 2012). Thus, using standard log-concentration differences to explore the full exposure-response of (2R,6R)-HNK, we selected the (2R,6R)-HNK concentrations of 10, 1, 0.1 and 0.01 μM. These concentrations, along with all the necessary controls (i.e. vehicle, PF4778574, ketamine and 3 similarly preclinically/clinically relevant (2S,6S)-HNK concentrations) resulted in a robust comparison that demonstrated only (2R,6R)-HNK concentration-dependently increased GluA1 surface expression. Our conclusions assume a one-to-one (and interspecies) in vitro-in vivo pharmacological correlation for these rat pyramidal neuron-based observations. In our experience (Patel et al., 2013; Shaffer et al., 2013, 2015) with AMPAR potentiators, mechanism-based in vivo animal activity (both desired and deleterious) were religiously observed at Cb,u that are magnitudes lower than potentiator concentrations eliciting functional activity in vitro, but effectively identical to projected human Cb,u that are clinically effective and safe/well-tolerated (Ranganathan et al., 2017; Shaffer, 2018). Thus, to inform such translational pharmacology appropriately and to optimize any decisions on the relevance of these in vitro-derived (2R,6R)-HNK-mediated GluA1 expression data to actual clinical exposures, future laboratory work should test (2R,6R)HNK doses that afford Cb,u of 0.01–0.1 μM and evaluate more pharmacodynamic time points than just 1 and 24 h post-dose. Carefully determining (2R,6R)-HNK's in vivo-based exposure-response continuum 79
Neuropharmacology 153 (2019) 73–81
C.L. Shaffer, et al.
is vital, particularly due to the intriguing temporal parallelism of the (2R,6R)-HNK-mediated increase in GluA1 surface expression reaching statistical significance (vs. 0 min) after 90–180 min of incubation relative to the typical time (i.e. 190 min after the 40 min ketamine infusion) (Zarate et al., 2012) of the first clinical detection of ketamine's antidepressant activity; (2S,6S;2R,6R)-HNK clinical exposure peaks 70–190 min after infusion cessation (Zarate et al., 2012) because of the time required to metabolize ketamine to (2S,6S;2R,6R)-HNK. Such preclinical data will dictate not only any further implications in the indirect increase in AMPAR signaling by (2R,6R)-HNK in ketamine's antidepressant activity, but also the potential clinical testing of (2R,6R)HNK itself by allowing the careful determination of its projected human efficacious concentration for the calculation of high-confidence therapeutic indices.
R., O'Loughlin, K., Wainer, I.W., 2015. The distribution and clearance of (2S,6S)hydroxynorketamine, an active ketamine metabolite, in Wistar rats. Pharmacol Res Perspect 3, e00157. Morris, P.J., Moaddel, R., Zanos, P., Moore, C.E., Gould, T., Zarate Jr., C.A., Thomas, C.J., 2017. Synthesis and N-methyl-D-aspartate (NMDA) receptor activity of ketamine metabolites. Org. Lett. 19, 4572–4575. Murrough, J.W., Abdallah, C.G., Mathew, S.J., 2017. Targeting glutamate signalling in depression: progress and prospects. Nat. Rev. Drug Discov. 16, 472–486. Newcomer, J.W., Farber, N.B., Jevtovic-Todorovic, V., Selke, G., Melson, A.K., Hershey, T., Craft, S., Olney, J.W., 1999. Ketamine-induced NMDA receptor hypofunction as a model of memory impairment in psychosis. Neuropsychopharmacology 20, 106–118. Patel, N.C., Schwarz, J., Hou, X.J., Hoover, D.J., Xie, L., Fliri, A.J., Gallaschun, R.J., Lazzaro, J.T., Bryce, D.K., Hoffman, W.E., Hanks, A.N., McGinnis, D., Marr, E.S., Gazard, J.L., Hajos, M., Scialis, R.J., Hurst, R.S., Shaffer, C.L., Pandit, J., O'Donnell, C.J., 2013. Discovery and characterization of a novel dihydroisoxazole class of αamino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptor potentiators. J. Med. Chem. 56, 9180–9191. Pham, T.H., Defaix, C., Xu, X., Deng, S.-X., Fabresse, N., Alvarez, J.-C., Landry, D.W., Brachman, R.A., Denny, C.A., Gardier, A.M., 2018. Common neurotransmission recruited in (R,S)-ketamine and (2R,6R)-hydroxynorketamine-induced sustained antidepressant-like effects. Biol. Psychiatry 84, e3–e6. Pinheiro, J., Bates, D., DebRoy, S., Sarkar, D., R Core Team, 2014. Nlme: Linear and Nonlinear Mixed Effects Models. R package version 3.0.1. http://CRAN.R-project. org/package=nlme. R Core Team, 2014. R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria. http://www.R-project.org/. Ranganathan, M., DeMartinis, N., Huguenel, B., Gaudreault, F., Bednar, M.M., Shaffer, C.L., Gupta, S., Cahill, J., Sherif, M.A., Mancuso, J., Zumpano, L., D'Souza, D.C., 2017. Attenuation of ketamine-induced impairment in verbal learning and memory in healthy volunteers by the AMPA receptor potentiator PF-04958242. Mol. Psychiatr. 22, 1633–1640. Sawant-Basak, A., Chen, L., Shaffer, C.L., Palumbo, D., Schmidt, A., Tseng, E., Spracklin, D.K., Gallezot, J.-D., Labaree, D., Nabulsi, N., Huang, Y., Carson, R.E., McCarthy, T., 2017. Quantitative projection of human brain penetration of the H3 antagonist PF03654746 by integrating rat-derived brain partitioning and PET receptor occupancy. Xenobiotica 47, 119–126. Shaffer, C.L., 2010. Defining neuropharmacokinetic parameters in CNS drug discovery to determine cross-species pharmacologic exposure-response relationships. In: Macor, J.E. (Ed.), Annual Reports in Medicinal Chemistry. Elsevier, New York, pp. 55–70. Shaffer, C.L., 2018. Reverse and forward translational pharmacology in psychiatric drug discovery. Clin. Pharmacol. Ther. 103, 193–195. Shaffer, C.L., Hurst, R.S., Scialis, R.J., Osgood, S.M., Bryce, D.K., Hoffmann, W.E., Lazzaro, J.T., Hanks, A.N., Lotarski, S., Weber, M.L., Liu, J., Menniti, F.S., Schmidt, C.J., Hajos, M., 2013. Positive allosteric modulation of AMPA receptors from efficacy to toxicity: the interspecies exposure-response continuum of the novel potentiator PF4778574. J. Pharmacol. Exp. Ther. 347, 212–224. Shaffer, C.L., Osgood, S.M., Smith, D.L., Liu, J., Trapa, P.E., 2014. Enhancing ketamine translational pharmacology via receptor occupancy normalization. Neuropharmacology 86, 174–180. Shaffer, C.L., Patel, N.C., Schwarz, J., Scialis, R.J., Wei, Y., Hou, X.J., Xie, L., Karki, K., Bryce, D.K., Osgood, S.M., Hoffmann, W.E., Lazzaro, J.T., Chang, C., McGinnis, D.F., Lotarski, S.M., Liu, J., Obach, R.S., Weber, M.L., Chen, L., Zasadny, K.R., Seymour, P.A., Schmidt, C.J., Hajos, M., Hurst, R.S., Pandit, J., O'Donnell, C.J., 2015. The discovery and characterization of the α-amino-3-hydroxy-5-mehtyl-4-isoxazolepropionic acid (AMPA) receptor potentiator N-{(3S,4S)-4-[4-(5-cyano-2thienyl)phenoxy]tetrahydrofuran-3-yl}propane-2-sulfonamide (PF-04958242). J. Med. Chem. 58, 4291–4308. Shen, W., Henry, A., Paumier, K., Li, L., Mou, K., Dunlop, J., Berger, Z., Hirst, W.D., 2014. Inhibition of glucosylceramide synthase stimulates autophagy flux in neurons. J. Neurochem. 129, 884–894. Suzuki, K., Nosyreva, E., Hunt, K.W., Kavalali, E.T., Monteggia, L.M., 2017. Effects of a ketamine metabolite on synaptic NMDAR function. Nature 546, E1–E3. Tseng, K., O'Donnell, P., 2007. D2 dopamine receptors recruit a GABA component for their attenuation of excitatory synaptic transmission in the adult rat prefrontal cortex. Synapse 61, 843–850. Wray, N.H., Rasenick, M.M., 2018. NMDA-receptor independent actions of ketamine: a new chapter in a story that's not so old. Neuropsychopharmacology 44, 220–221. Yamaguchi, J.-i., Toki, H., Qu, Y., Yang, C., Koike, H., Hashimoto, K., Mizuno-Yasuhira, A., Chaki, S., 2018. (2R,6R)-Hydroxynorketamine is not essential for the antidepressant actions of (R)-ketamine in mice. Neuropsychopharmacology 43, 1900–1907. Zanos, P., Moaddel, R., Morris, P.J., Georgiou, P., Fischell, J., Elmer, G.I., Alkondon, M., Yuan, P., Pribut, H.J., Singh, N.S., Dossou, K.S.S., Fang, Y., Huang, X.-P., Mayo, C.L., Wainer, I.W., Albuquerque, E.X., Thompson, S.M., Thomas, C.J., Zarate Jr., C.A., Gould, T.D., 2016. NMDAR inhibition-independent antidepressant actions of ketamine metabolites. Nature 533, 481–498. Zanos, P., Moaddel, R., Morris, P.J., Wainer, I.W., Albuquerque, E.X., Thompson, S.M., Thomas, C.J., Zarate Jr., C.A., Gould, T., 2017. Reply to: antidepressant actions of ketamine versus hydroxynorketamine. Biol. Psychiatry 81, e69–e71. Zarate Jr., C.A., Brutsche, N., Laje, G., Luckenbaugh, D.A., Vattem Venkata, S.L., Ramamoorthy, A., Moaddel, R., Wainer, I.W., 2012. Relationship of ketamine's plasma metabolites with response, diagnosis, and side effects in major depression. Biol. Psychiatry 72, 331–338. Zhang, L., Balan, G., Barreiro, G., Boscoe, B., Chenard, L., Cianfrogna, J., Claffey, M., Chen, L., Coffman, K., Drozda, S., Dunetz, J., Fonseca, K., Galatsis, P., Grimwood, S., Lazzaro, J., Miller, E., Mancuso, J., Reese, M., Rogers, B., Sakurada, I., Skaddan, M.,
Declaration of interest When this work was performed, each author was a full-time employee of Pfizer Inc. and owned and/or held options/restricted stock units for the company's publicly traded shares. Acknowledgments The authors acknowledge Kewa Mou for generating cortical cell cultures for experiments. Appendix A. Supplementary data Supplementary data related to this article can be found at https:// doi.org/10.1016/j.neuropharm.2019.04.019. References Abdallah, C.G., 2017. What's the buzz about hydroxynorketamine? Is it the history, the story, the debate, or the promise? Biol. Psychiatry 81, e61–e63. Collingridge, G.L., Lee, Y., Bortolotto, Z.A., Kang, H., Lodge, D., 2017. Antidepressant actions of ketamine versus hydroxynorketamine. Biol. Psychiatry 81, e65–e67. Collo, G., Cavalleri, L., Chiamulera, C., Merlo Pich, E., 2018. (2R,6R)Hydroxynorketamine promotes dendrite outgrowth in human inducible pluripotent stem cell-derived neurons through AMPA receptor with timing and exposure compatible with ketamine infusion pharmacokinetics in humans. Neuroreport 29, 1425–1430. Desta, Z., Moaddel, R., Ogburn, E.T., Xu, C., Ramamoorthy, A., Venkata, S.L.V., Sanghvi, M., Goldberg, M.E., Torjman, M.C., Wainer, I.W., 2012. Stereoselective and regiospecific hydroxylation of ketamine and norketamine. Xenobiotica 42, 1076–1087. Doran, A.C., Osgood, S.M., Mancuso, J.Y., Shaffer, C.L., 2012. An evaluation of using ratderived single-dose neuropharmacokinetic parameters to project accurately large animal unbound brain drug concentrations. Drug Metab. Dispos. 40, 2162–2173. Feng, B., Doran, A.C., Di, L., West, M.A., Osgood, S.M., Mancuso, J.Y., Shaffer, C.L., Tremaine, L., Liras, J., 2018. Prediction of human brain penetration of P-glycoprotein and breast cancer resistance protein substrates using in vitro transporter studies and animal models. J. Pharm. Sci. 107, 2225–2235. Fukumoto, K., Fogaca, M.V., Liu, R.-J., Duman, C., Kato, T., Li, X.-Y., Duman, R.S., 2018. Activity-dependent brain-derived neurotrophic factor signaling is required for the antidepressant actions of (2R,6R)-hydroxynorketamine. Proc. Natl. Acad. Sci. Unit. States Am (in press). Hashimoto, K., Shirayama, Y., 2018. What are the causes for discrepancies of antidepressant actions of (2R,6R)-hydroxynorketamine? Biol. Psychiatry 84, e7–e8. Kuznetsova, A., Brockhoff, P.B., Christensen, R.H.B., 2016. lmerTest: tests in linear mixed effects models. R package version 2.0‐32. http://CRAN.R-project.org/package= lmerTest. Lenth, R.V., 2014. Lsmeans: Least-Squares Means. R package version 2.11. http://CRAN. R-project.org/package=lsmeans. Li, N., Lee, B., Liu, R.-J., Banasr, M., Dwyer, J.M., Iwata, M., Li, X.-Y., Aghajanian, G., Duman, R.S., 2010. mTOR-Dependent synapse formation underlies the rapid antidepressant effects of NMDA antagonists. Science 329, 959–964. Liu, X., Van Natta, K., Yeo, H., Vilenski, O., Weller, P.E., Worboys, P.D., Monshouwer, M., 2009. Unbound drug concentration in brain homogenate and cerebral spinal fluid at steady state as a surrogate for unbound concentration in brain interstitial fluid. Drug Metab. Dispos. 37, 787–793. Maeng, S., Zarate Jr., C.A., Du, J., Schloesser, R.J., McCammon, J., Chen, G., Manji, H.K., 2008. Cellular mechanisms underlying the antidepressant effects of ketamine: role of α-amino-3-hydroxy-5-methylisoxazole-4-propionic acid receptors. Biol. Psychiatry 63, 349–352. Moaddel, R., Sanghvi, M., Dossou, K.S.S., Ramamoorthy, A., Green, C., Bupp, J., Swezey,
80
Neuropharmacology 153 (2019) 73–81
C.L. Shaffer, et al.
Zhou, L., Schmidt, K., Nelson, F.R., Zelesky, V., Troutman, M.D., Feng, B., 2009. The effect of breast cancer resistance protein and P-glycoprotein on the brain penetration of flavopiridol, imatinib mesylate (Gleevac), prazosin, and 2-methoxy-3-(4-(2-(5methyl-2-phenyloxazol-4-yl)ethoxy)phenyl)propanoic acid (PF-407288) in mice. Drug Metab. Dispos. 37, 946–955.
Smith, D., Stepan, A., Trapa, P., Tuttle, J., Verhoest, P., Walker, D., Wright, A., Zaleska, M., Zasadny, K., Shaffer, C.L., 2014. The discovery and preclinical characterization of 1-methyl-3-(4-methylpyridin-3-yl)-6-(pyridin-2-ylmethoxy)-1H-pyrazolo[3,4-b]pyrazine (PF470) as a highly potent, selective and efficacious mGluR5 negative allosteric modulator. J. Med. Chem. 57, 861–877.
81