Using [18F]fluorodeoxyglucose μ–positron emission tomography ([18F]FDG μPET), we compared subanesthetic doses of memantine and ketamine on their potential to induce increases in brain activation. We also studied the reversal effect of the well-known metabotropic glutamate receptor (mGluR)-2/3 agonist LY404039 [(−)-(1R,4S,5S,6S)-4-amino-2-sulfonylbicyclo[3.1.0]hexane-4,6-dicarboxylic acid] and the novel mGluR2 positive allosteric modulator (PAM) JNJ-42153605 [3-cylcopropylmethyl-7-(4-phenylpiperidin-1-yl)-8-trifluoromethyl [1,2,4] triazolo[4,3-a]pyridine]. First, rats (n = 12) were subjected to LY404039 (10 mg/kg s.c.) or vehicle, 30 minutes prior to saline, ketamine (30 mg/kg i.p.), or memantine (20 mg/kg i.p.). Second, rats (n = 12) were subjected to 2.5 mg/kg or 10 mg/kg mGluR2 PAM JNJ-42153605 or vehicle (s.c.), 30 minutes prior to memantine (20 mg/kg i.p.) or saline. Fifteen minutes later, [18F]FDG was injected (37 MBq i.v.) followed by a μPET/computed tomography scan. The increase due to memantine is significant for all relevant brain areas, whereas for ketamine this is not the case. Standard uptake values (SUVs) of the LY404039 pretreated and memantine-challenged group display a full reversal. Pretreatment with JNJ-42153605 also dose-dependently decreases SUV with a full reversal as well (for 10 mg/kg). Moreover, specificity of JNJ-42153605 is reached at this dose. In conclusion, this μPET experiment clearly indicates that subanesthetic doses of memantine induce significant increases of [18F]FDG SUVs in discrete brain areas and that the novel mGluR2 PAM has the capacity to dose-dependently and specifically reverse memantine-induced brain activation.
Glutamate is the most abundant neurotransmitter in the central nervous system, and an imbalance in its transmission is implicated in various neurologic and psychiatric disorders (Pilc et al., 2008; Johnson et al., 2009; Lüscher and Huber, 2010; Chiechio and Nicoletti, 2012). Schizophrenia is a severe, disabling chronic disorder affecting approximately 1% of the population (Lindsley et al., 2006). Numerous studies have generated support for the N-methyl-d-aspartate (NMDA) receptor hypofunction hypothesis of schizophrenia (Olney et al., 1999; Lindsley et al., 2006; Gunduz-Bruce, 2009) causing excess glutamate excitotoxicity. Figure 1 illustrates this interplay between the NMDA receptor (NMDAR), GABA, and glutamate, where in nonpathologic conditions, NMDARs on GABAergic neurons mediate glutamate release by excitatory neurons modulating metabotropic glutamate receptor (mGluR)-5 activation on connecting neurons (Fig. 1A). In addition to the aforementioned postsynaptic mGluRs on connecting neurons, the concentration of extracellular glutamate is also regulated by excitatory amino acid or glutamate transporters on neuroglia (Fig. 1A) and by presynaptic mGluRs functioning to limit the release of intracellular glutamate (Fig. 1A). NMDARs are ionotropic receptors that are both ligand-gated, requiring glutamate to bind with coactivation of d-serine or glycine, as well as voltage-dependent through channel blocks by extracellular Zn2+ and Mg2+ ions (Fig. 1B). When the NMDARs function improperly, the inhibitory GABAergic tone is reduced, causing excess glutamate to be released (Fig. 1A).
Therefore, neuropharmacology researchers are especially interested in targeting glutamate receptors to modulate the synaptic efficacy of glutamate (Kew and Kemp, 2005). Recently, mGluR2/3 have proven to be interesting targets. mGluR2/3 are preferentially expressed on presynaptic terminals (Fig. 1A); they negatively modulate glutamate levels and are highly abundant in brain areas, such as the prefrontal cortex and hippocampus, which are implicated in disorders such as anxiety and schizophrenia (Cartmell and Schoepp, 2000; Vinson and Conn, 2012). Therefore, compounds known to activate the mGluR2/3 are considered potential antipsychotics and anxiolytics. The mGluR2 receptor is a G protein–coupled receptor with allosteric binding sites in the 7-transmembrane domain and an extracellular N-terminal domain for binding a ligand through the Venus fly trap mechanism, thereby causing a conformational change in the 7-transmembrane domain, resulting in activation of the G protein of which the α subunit causes a downstream cascade (Fig. 1C). Previously, it was shown that receptor agonists for the mGluR2/3 have robust anxiolytic effects and can improve negative and positive symptoms in schizophrenia in both animal models and humans (Galici et al., 2005; Patil et al., 2007; Nikiforuk et al., 2010; Wierońska et al., 2012).
In addition to mGluR2 and mGluR2/3 agonists, compounds that can increase the effect of glutamate by binding to the aforementioned allosteric sites (Conn et al., 2009), called positive allosteric modulators (PAMs), are being investigated (Johnson et al., 2003; Galici et al., 2006; Dhanya et al., 2011; Fell et al., 2011; Lundström et al., 2011; Cid et al., 2012). The use of PAMs compared with agonists has the advantage of a higher selectivity and potentially overcoming the desensitization of G protein–coupled receptors observed after repeated dosing of agonists (Bonnefous et al., 2005). In a recent study, a potent and highly selective mGluR2 PAM JNJ-42153605 [3-cylcopropylmethyl-7-(4-phenylpiperidin-1-yl)-8-trifluoromethyl [1,2,4] triazolo[4,3-a]pyridine] has been identified (Cid et al., 2012). This compound significantly influences rat sleep-wake organization by decreasing rapid eye movement sleep, and it reverses phencyclidine (PCP)-induced locomotor activity in mice, denoting antipsychotic-like effects (Cid et al., 2012).
In developing new antipsychotics, the use of preclinical screening methods using animal models of schizophrenia can be informative for confirmation of pharmacologic effect and dose response. In the search for suitable models, focus is on the similarities of the animal model with the clinical behavioral symptoms of schizophrenia and the increased activation in discrete brain areas associated with the disease. In addition, validity of these animal models is confirmed by determination of the effects of known antipsychotics on the reversal of the pathologic behavior and the increased brain activation. Acute administration of PCP, MK-801 [(5S,10R)-(+)-5-methyl-10,11-dihydro-5H-dibenzo[a,d]cyclohepten-5,10-imine], or subanesthetic doses of ketamine have been used as on-demand models of schizophrenia, as both the behavior resembles the clinical pathology (Javitt and Zukin, 1991; Andiné et al., 1999) as well as the increased activation in distinct brain areas, such as the prefrontal and cingulate cortices and hippocampus (Duncan et al., 1998a). For example, reversal of ketamine-induced effects was evaluated using both the typical antipsychotic haloperidol and the atypical antipsychotic clozapine (Duncan et al., 1998b) and confirms the model’s validity. Besides the NMDAR antagonist ketamine model, administration of another NMDA receptor antagonist, memantine, has also been suggested as an appropriate on-demand model of schizophrenia (Dedeurwaerdere et al., 2011). However, controversy exists on which of these NMDA antagonists is best suited as an animal model. Although memantine and ketamine are similarly potent at human GluN1/GluN2A receptors in electrophysiological assays (Gilling et al., 2009), there are small but clear differences in the biophysical properties of these two uncompetitive NMDA receptor antagonists. Ketamine was shown to have slower kinetics than memantine (Parsons et al., 1995). Both the onset and the offset kinetics for memantine were two times faster than those of ketamine (kon 0.32 ± 0,11 × 106 M−1 s−1, koff 0.53 ± 0.10 second−1 and kon 0.15 ± 0.05 × 106 M−1 s−1, koff 0.22 ± 0.05 second−1, respectively). However, the Kd values (1.65 ± 1.05 μM and 1.47 ± 0.68 μM for memantine and ketamine, respectively) confirm that the two compounds have similar affinity for the receptor despite the difference in kinetics. Memantine is also slightly more voltage-dependent than ketamine, as it binds to a deeper site and senses more of the electrical field (fraction 0.90 ± 0.09 and 0.79 ± 0.04, respectively). Finally, memantine shows a 20% untrapping where ketamine has little or no partial untrapping (Gilling et al., 2009).
A robust criterion for a suitable schizophrenia animal model is a pronounced increase of glucose usage in delineated brain structures as the NMDAR hypofunction causes hyperactive glutamatergic neurons. The energy empowering these excited glutamatergic neurons is mainly lactate, which is anaerobically generated in astrocytes that take up glucose from the plasma in circulation via the glucose transporter 1, and partly direct aerobic glycolysis in the neuron receiving plasma glucose via glucose transporter 3 (Fig. 1A). The radiotracer [18F]fluorodeoxyglucose ([18F]FDG) is a radiolabeled glucose analog and is used to visualize changes in glucose consumption. [18F]FDG becomes phosphorylated to [18F]FDG-6-PO4 and, due to the molecule modifications in contrast to glucose, gets trapped once transported within cells. The concentration of [18F]FDG within cells of specific brain regions is considered an indirect measure of neural activity (Bailey et al., 2004).
Therefore in the present study, we used in vivo [18F]FDG molecular imaging positron emission tomography (PET) in the rat, using a volume-of-interest (VOI)–based approach to compare both memantine and ketamine: 1) on their potential to induce increases in activation in discrete brain regions, and 2) on the reversal effects in these models induced by the well-characterized mGlu2/3 agonist LY404039 [(−)-(1R,4S,5S,6S)-4-amino-2-sulfonylbicyclo[3.1.0]hexane-4,6-dicarboxylic acid] (Patil et al., 2007; Rorick-Kehn et al., 2007). Second, we determined the effect of the novel mGluR2 PAM JNJ-42315605 (Cid et al., 2012) in the best schizophrenia model of the latter two (memantine/ketamine). The major advantage of using μPET for screening of novel antipsychotics in comparison with other preclinical screening techniques, such as autoradiography, microdialysis, and ex vivo tissue analysis, is its noninvasiveness, allowing longitudinal studies and intra-animal comparisons with significantly lower numbers of animals (Lancelot and Zimmer, 2010).
Materials and Methods
Twenty-four male Sprague-Dawley rats (Charles River Laboratories, Lyon, France) weighing 275–350 g and aged 9–12 weeks were treated in accordance with the European Ethics Committee (decree 86/609/CEE), and the study protocol was approved by the local Animal Experimental Ethical Committee of the University of Antwerp, Belgium (2011–67). The animals were kept in individually ventilated cages under environmentally controlled conditions (12-hour normal light/dark cycles, 20–23°C, and 50% relative humidity) with food and water ad libitum.
For ketamine, a concentration of 30 mg/kg was chosen, based on literature (Duncan et al., 1998b; Dedeurwaerdere et al., 2011) as higher concentrations (>40mg/kg) are reported to show loss of excitation (Miyamoto et al., 2000). For memantine, a concentration of 20 mg/kg was chosen based on the findings in literature (Dedeurwaerdere et al., 2011).
Both memantine (memantine hydrochloride; Sigma-Aldrich, St. Louis, MO) (20 mg/kg) and ketamine (Ketalar; Pfizer, New York, NY) (30 mg/kg) were dissolved in saline (NaCl 0.9%; B. Braun Medical, Melsungen, Germany). The mGluR2/3 agonist LY404039 (10 mg/kg; provided by Janssen Pharmaceutica NV, Beerse, Belgium) was dissolved in H2O + NaOH (3.5 < pH < 9) as the LY-vehicle. The mGluR2 PAM agonist JNJ-42153605 (provided by Janssen Pharmaceutica NV) was dissolved in 10% cyclodextrin + HCl (3.5 < pH <9) as the JNJ-vehicle.
In the first part, animals (n = 12) were subjected to subcutaneous injection of the test compound, i.e., LY404039 (10 mg/kg) or vehicle, 30 minutes prior to challenge with the NMDA receptor antagonist, i.e., ketamine (30 mg/kg i.p.), memantine (20 mg/kg i.p.) or saline, each of which, in turn, was administered 15 minutes prior to [18F]FDG injection (37 MBq i.v. under short anesthesia) as schematically illustrated by Fig. 2A. In the second part, animals (n = 12) were subjected to subcutaneous injection of JNJ-42153605 (2.5 mg/kg or 10 mg/kg) or vehicle, 30 minutes prior to challenge with memantine. Memantine (20 mg/kg i.p.) injection is performed 15 minutes prior to [18F]FDG injection (37 MBq i.v. under short anesthesia) as shown in Fig. 2B. In both studies, [18F]FDG injection was followed by a 30-minute awake uptake period to reach steady state. Anesthesia with isoflurane (mixture with medical oxygen: 5% for induction, 1.5% during the experiment) was started 25 minutes after tracer injection; the animal was positioned onto the μPET scan bed and scanned (PET) for 20 minutes, followed by a micro–computed tomography (CT) scan. For these scans the animals were fasted for at least 12 hours, and during the entire procedure their body temperature was controlled to reduce variability. [18F]FDG was prepared using a cassette-based GE Fastlab synthesis module (GE Healthcare, Diegem, Belgium). [18F]Fluoride was produced by bombarding 18O-enriched water using an 11 MeV proton beam in a Eclipse HP cyclotron (Siemens, Knoxville, TN). The purified [18F]FDG was diluted with 0.9% NaCl (Baxter, Eigenbrakel, Belgium) and sterile filtered through a 0.22 µM filter. Quality control was performed according to European Pharmacopoeia 7.1, while radiochemical identity was confirmed by high-pressure liquid chromatography (Dionex; ThermoFisher, Waltham, MA) and radio–thin layer chromatography. Radiochemical purity was also determined by high-pressure liquid chromatography (Dionex; ThermoFisher), and radionuclidic identity and purity were confirmed by γ spectrum analysis (Multichannel analyzer; Canberra, Meriden, CT).
Imaging was performed on a Siemens Inveon PET-CT scanner (Siemens Preclinical Solutions, Knoxville, TN). For quantitative analysis, μPET images were reconstructed using the two-dimensional filtered backprojection (FBP). Before image reconstruction, three-dimensional sinograms were converted into two-dimensional sinograms by Fourier rebinning. For the FBP reconstruction, a ramp filter with cut-off at one-half the Nyquist criteria (maximum sampling frequency) was set. Normalization, dead time, random, CT-based attenuation, and single scatter stimulation (Watson, 2000) scatter corrections were applied. The reconstructed spatial resolution approached 1.4 mm in the center of the field of view, and images are displayed in transverse, coronal, and sagittal planes with a matrix size of 128 × 128 × 159. The image pixel size in the FBP-reconstructed images is 0.77 mm with a slice thickness of 0.79 mm.
Each individual reconstructed PET image was transformed into the space of a standard [18F]FDG PET template (Schiffer et al., 2006) using brain normalization in PMOD v3.3 (PMOD Technologies, Zurich, Switzerland). The normalized images were then overlaid with a magnetic resonance imaging–based rat-brain VOI template (Schiffer et al., 2006) available in the same software package (Fig. 2, C–E) and by default already coregistered with the PET normal database. Averaged activity concentrations for the different VOIs Ai (kBq/ml) were extracted and the standard uptake values (SUVs) for these VOIs was calculated as SUVi = Ai/ID × mi, where i is the VOI index, ID (kBq) is the injected activity, and m (g) is the weight of the animal. The SUV values for eight clearly identifiable and relevant brain areas were determined for the six experimental conditions, for all animals. The rationale was for caudate putamen: coherent intrinsic activity of the dorsal striatum increases and correlates with positive symptoms, such as delusion and hallucination (Sorg et al., 2013); cingulate cortex: receives input from the thalamus and is an integral part of the limbic system, which makes it important in cognitive function (Adams and David, 2007); frontal cortex and medial prefrontal cortex: hyperfrontality (Soyka et al., 2005; Whitfield-Gabrieli et al., 2009); motor cortex: motor abnormalities are frequently part of the pathophysiology (Walker et al., 1994); parietal cortex: involved in sensory integration; (anterodorsal) hippocampus: increased hippocampal drive to the ventral tegmental area (Adams et al., 2013), and thalamus: abnormality in thalamo-prefrontal cortical circuitry (Lewis et al., 2001).
Data were expressed as mean ± S.E.M. To determine the effect of challenges in comparison with saline, and for treatments in comparison with vehicle we performed a repeated measures analysis of variance, taking into account a Bonferroni adjustment (SPSS20; SPSS, Chicago, IL). For statistical comparison of the two treatment doses for JNJ-42153605, however, we performed repeated measures analysis of variance, taking into account a Dunnett correction (SPSS20). For determining the specificity of LY404039 and JNJ-42153605, paired t tests were performed. Significance is reached at P < 0.05.
The Effect of Memantine/Ketamine.
In the first part, after both vehicle and memantine or ketamine administration, widespread increases in [18F]FDG uptake are observed in neuroanatomically distinct regions in comparison with the vehicle + saline controls. The mean (± S.E.M.) percentage changes in compared SUVs are indicated in Table 1.
Statistically significant increases in [18F]FDG SUVs are observed between vehicle + memantine and vehicle + saline treatment in all of the eight relevant brain areas (Fig. 3). The percentage increase due to this memantine challenge in comparison with vehicle + saline treatment is between 31.0% (±8.9%) for thalamus and 50.1% (±12.1%) for frontal cortex (Fig. 4, A and C; Table 1). No statistically significant increases in [18F]FDG SUVs are observed in any of the eight relevant brain areas between vehicle + ketamine and vehicle + saline treatment (Fig. 3). The percentage increase due to ketamine challenge in comparison with vehicle + saline treatment is between 13.2% (±6.9%) for thalamus and 22.4% (±9.8%) for frontal cortex (Fig. 4, A and B; Table 1). Memantine induces on average a 16.6, 25.0, 24.8, 22.9, 24.0, 23.3, 18.0, and 17.0% greater [18F]FDG uptake in comparison with ketamine, for the caudate putamen, cingulate cortex, frontal cortex, medial prefrontal cortex, motor cortex, parietal cortex, anterodorsal hippocampus, and thalamus, respectively.
The Effect and Specificity of LY404039 Treatment.
Comparison between vehicle + memantine and LY404039 + memantine SUVs reveals significant decreases for all eight relevant brain areas due to treatment with LY404039, similarly as for the decrease in [18F]FDG uptake between vehicle + ketamine and LY404039 + ketamine (Fig. 3). Percentage decreases due to LY404039 pretreatment are between −30.2% (±3.9%) for caudate putamen and −38.5% (±3.4%) for cingulate cortex (Fig. 4, C and F; Table 2) in the memantine-challenged group and are between −29.2% (±5.8%) for anterodorsal hippocampus and −37.9% (±5.8%) for frontal cortex (Fig. 4, B and E; Table 2) in the ketamine group. However, the mGlu2/3 agonist LY404039 also causes a statistically significant decrease in [18F]FDG uptake in all eight relevant brain areas (LY404039 + saline) relative to the basal uptake values in vehicle + saline of between −32.5% (±3.2%) for anterodorsal hippocampus and −38.5% (±4.4%) for frontal cortex (Fig. 4, A and D; Table 2). There are no statistically significant differences between the three decreases reported in Table 2 in any of the eight relevant brain areas, indicating lack of specificity.
Animals that received a pretreatment with LY404039 and a memantine challenge show a decrease between −0.3% (±7.1%) for motor cortex and −10.9% (±5.2%) for thalamus compared with vehicle + saline treatment, indicating a full reversal (Fig. 4, A and F; Table 1). On the other hand, animals receiving LY404039 and ketamine challenge show a decrease between −20.7% (±4.6%) for anterodorsal hippocampus and −27.1% (±5.7%) for cingulate cortex (Fig. 4, A and E; Table 1), proving that the ketamine enhancement is not as high as with memantine, resulting in an overshoot of the reversal.
The Effect and Specificity of JNJ-42153605 Treatment.
As the second part of this study entails another treatment (and thus also another vehicle), we first confirm again that administration of memantine replicates the results of the first part of the study, namely widespread significant increases in [18F]FDG uptake in neuroanatomically distinct regions in comparison with the corresponding vehicle + saline controls (Figs. 4, G and H, and 5). The increases in this second part of the study range similarly between 25.4 ± 7.6% for thalamus and 51.0 ± 10.6% for frontal cortex (first columns of Tables 1 and 3 are similar).
Administration of 2.5 mg/kg and 10 mg/kg JNJ-42153605 in memantine-challenged animals causes a decrease in SUV in comparison with memantine-challenged animals under vehicle pretreatment. The decrease in [18F]FDG uptake ranges from −22.4% (±4.5%) for the caudate putamen to −31.1% (±4.1%) for the cingulate cortex under the 10 mg/kg JNJ-42153605 dose (Table 4) and is significant for all relevant brain areas (Figs. 4, H and L, and 5). In the 2.5 mg/kg dose, the decrease in SUV is only significant for cingulate cortex, frontal cortex, medial prefrontal cortex, motor cortex, parietal cortex, and anterodorsal hippocampus (Fig. 5) and ranges from −5.2% (±6.5%) for the caudate putamen to −11.2% (±6.3%) for the cingulate cortex (Fig. 4, H and J; Table 4). Moreover, administration of 2.5 mg/kg and 10 mg/kg JNJ-42153605 in saline-challenged animals induces a decrease in SUV in comparison with saline-challenged animals under vehicle pretreatment (Fig. 5). In the 10 mg/kg dose, the decrease ranges from −15.8% (±3.8%) for the caudate putamen to −24.3% (±4.3%) for the frontal cortex (Fig. 4, G and K; Table 4) and is significant for all relevant brain regions (Fig. 5). For the 2.5 mg/kg dose, the decrease in SUV is only significant for the motor cortex (Fig. 5), as it ranges from 0.2% (±6.4%) for the caudate putamen to −6.4% (±5.7%) for the cingulate cortex (Fig. 4, G and I; Table 4). There is no statistically significant difference between the SUV-decreasing effects induced by JNJ-42153605 in the 2.5 mg/kg concentration for memantine-challenged versus saline-challenged animals (Fig. 6; Table 4). However, for the 10 mg/kg concentration, there is a statistically significant difference between the effect induced by JNJ-42153605 in the memantine-challenged animals versus the saline-challenged animals for the cingulate cortex, the motor cortex, the parietal cortex, and the anterodorsal hippocampus (Fig. 6; Table 4), showing the specific effect of JNJ-42153605 at this dose of 10 mg/kg.
Animals that received memantine as challenge and 2.5 mg/kg of JNJ-42153605 as a treatment still show an increase of SUV ranging from 11.4% (±2.9%) for the thalamus to 28.4% (±3.9%) for the frontal cortex compared with vehicle + saline (Fig. 4, G and J; Table 3). However, for the 10 mg/kg treatment dose of JNJ-42153605, this ranges from −8.9% (±3.9%) for the thalamus to 2.7% (±5.0%) for the frontal cortex in memantine-challenged animals achieving full reversal (Fig. 4, G and L; Table 3).
First, this μPET experiment clearly indicates that both memantine and ketamine induce increases in rat brain [18F]FDG uptake. Moreover, our study reveals that memantine is more potent at inducing statistically significant regional increases in SUV [18F]FDG in the rat brain than ketamine. This finding strengthens the hypothesis that both memantine and ketamine animal models are suitable on-demand animal models for schizophrenia, but that the memantine-challenged model necessitates lower numbers of animals to reach statistically significant results in comparison with ketamine-challenged animals.
It has been previously shown that ketamine and memantine have equipotent binding affinities (Kd ∼ 1.5 μM) at the human NMDA receptor (Gilling et al., 2009) and have nearly indistinguishable synaptic pharmacodynamics (Emnett et al., 2013). However, the pronounced differences in pharmacokinetics between memantine and ketamine, both in humans and in rats, might explain the difference in resulting increase in [18F]FDG SUV, as memantine has a half-life in humans of around 100 hours (Periclou et al., 2006), whereas ketamine is extremely rapidly eliminated, with a redistribution half-life of around 15 minutes (Clements and Nimmo, 1981) making it very difficult to achieve a long steady-state for ketamine. In this [18F]FDG μPET study, the overall degree of memantine-induced [18F]FDG increases correlates with the literature on memantine activation in rodent brain (Dedeurwaerdere et al., 2011). Minor disparities between the findings in literature and the current findings might be explained by differences in animal species and/or route of administration, etc. In addition, the difference in VOI determination can contribute to minor disparities. Other investigators (Duncan et al., 1998a; Dedeurwaerdere et al., 2011) use manually outlined VOIs in comparison with our fully automated predefined Schiffer rat brain VOI template available from the PMOD v3.3 software, which results in larger regions as a whole, which could eventually even out very regional changes such as the [18F]FDG uptake increases in the dentate gyrus, CA3 stratum radiatum, presubiculum, and stratum lacunosum moleculare of the hippocampus, etc. However, it is worth highlighting the advantage of our approach: the use of predefined and standardized VOI templates enables very fast and robust analysis of many regions of interest, which is a major advantage for the high-throughput screening of pharmacological compounds and radiotracers using in vivo functional imaging with μPET.
Normalizing excess glutamate levels by mGluR2/3 agonists has led to identification of potential anxiolytics and antipsychotic drugs. Studies in animals have shown that mGluR2 agonists are able to improve cognitive impairment produced by psychotomimetics and are active in several models related to the positive symptoms of schizophrenia (Lorrain et al., 2003; Patil et al., 2007; Pehrson and Moghaddam, 2010; Seeman, 2013). Pharmacological reversal testing showed that the mGlu2/3 receptor agonist LY404039 significantly attenuates both the memantine- and ketamine-induced [18F]FDG uptake. LY404039 almost fully reverses the memantine-induced [18F]FDG increase, while for the ketamine-induced increase, there is an overshoot in the reversal. These findings again illustrate the superiority of the memantine model, which was also highlighted in a previous 2DG autoradiography study with mice (Dedeurwaerdere et al., 2011). However, the interpretation of the reversal effect of this mGlu2/3 receptor agonist becomes potentially biased due to a similar decrease in [18F]FDG uptake in the LY404039 + saline condition as in the memantine and ketamine models. This raises the question of whether LY404039 is specifically reversing the ketamine- and memantine-induced [18F]FDG increases. LY404039 may have a polypharmacology: in addition to being an mGlu2/3 agonist (Fell et al., 2008), it also affects serotonin turnover in a GluR2/3-mediated way (Lowe et al., 2012), and it may inhibit the binding of ligands to dopamine D2 and D3 receptors (Seeman and Guan, 2009; Seeman, 2013) although this is controversial (Fell et al., 2009; Zysk et al., 2011) and unlikely to alter FDG uptake. Therefore, teasing out the specific pharmacology responsible for the decrease in LY404039 + saline condition and its effect on memantine-induced changes in [18F]FDG uptake is very complex. Our findings on the nonspecific effect of LY404039 are in contrast with a previous 2DG autoradiography (Dedeurwaerdere et al., 2011), where no effects of the compound on baseline 2DG uptake was observed. Differences in imaging techniques used or species differences may be an explanation.
Positive allosteric modulation (i.e., compounds that bind a site other than the orthosteric site) of the mGluR2 provides a novel way to regulate glutamatergic function. The use of mGluR2 PAMs in comparison with agonists to improve the functional effects of glutamate has several advantages, such as the possibility of higher selectivity, the lower risk of potential tolerance and desensitization, and improved safety, as the mGluR2 PAM will only activate the receptor in the presence of increased glutamate. A range of potent mGluR2 PAMs has been discovered; i.e., BINA [potassium 3′-([(2-cyclopentyl-6-7-dimethyl-1-oxo-2,3-dihydro-1H-inden-5-yl)oxy]methyl)biphenyl l-4-carboxylate] (Jin et al., 2010), LY487379 [N-(4-(2-methoxyphenoxy)-phenyl-N-(2,2,2-trifluoroethyl-sulfonyl)-pyrid-3-ylmethylamine] (Nikiforuk et al., 2010), THIIC [N-(4-((2-(trifluoromethyl)-3-hydroxy-4-(isobutyryl)phenoxy)methyl)benzyl)-1-methyl-1H-imidazole-4-carboxamide] (Fell et al., 2011), and JNJ-42153605 (Cid et al., 2012), among others. The mGluR2 PAM JNJ-42153605 showed the best overall profile among 28 structurally related compounds being screened, with an optimal in vitro profile combining potency, metabolic stability, and acceptable preliminary cardiovascular profile. Furthermore, it has acceptable pharmacokinetics, high selectivity for mGluR2, and excellent mGluR2 PAM activity (Cid et al., 2012). At a dose of 3 mg/kg (orally administered), the JNJ-42153605 suppresses rapid eye-movement sleep and prolongs rapid eye-movement sleep onset latency but does not affect other sleep-wake stages. In addition, the JNJ-42153605 compound displayed a dose-dependent and significant attenuation of PCP-induced hyperlocomotion, further confirming its potential antipsychotic effects. Moreover, JNJ-42153605 did not inhibit hyperlocomotion induced by amphetamine administration in mice, indicating that it lacks an inhibitory effect on dopaminergic neurotransmission, underlining a different mechanistic pathway compared with currently available antipsychotics. Once we selected the memantine challenge as the best animal model we showed, in the second part of our study, that the mGluR2 PAM JNJ-42153605 is able to dose-dependently reverse the memantine-induced increases in [18F]FDG uptake in the rat brain in vivo. Of importance, the JNJ-42153605 component in the presence of a memantine challenge is significantly different from the effect of the JNJ-42153605 component in the presence of saline, which may indicate that JNJ-42153605 is specific for memantine-pretreated animals.
A few limitations should be acknowledged. It should be noted that [18F]FDG is not a specific tracer for the mGluR2/3 receptor and that we thus only indirectly show that JNJ-42153605 has a higher specificity for the mGluR2/3 receptor than LY404039. Also, we could not work with relative [18F]FDG quantification versus a reference region given the wide distribution of NMDAR and especially the downstream effects, as the [18F]FDG readout is an integrated signal over time, so that the net effect can be distant to the NMDA-rich area of initial pharmacology. Finally, although we show a statistically significant specific effect of JNJ-42153605 in four out of eight relevant brain regions, it should be noted that JNJ-42153605 also causes significant decreases in both control and memantine-challenged animals in the other brain regions, albeit not statistically different in terms of specificity.
In summary, this study shows that memantine, more pronounced than ketamine, induces significant regional activation in the rat brain, which can be effectively reversed by a pharmacologic challenge with the mGlu2/3 receptor agonist LY404039. Further, these experiments indicate that the mGluR2 PAM JNJ-42153605 has the capacity to dose-dependently reverse such memantine-induced brain activation with specific effects at the 10 mg/kg dose.
The authors thank their coworkers from the Molecular Imaging Center Antwerp, University of Antwerp, Belgium: Joke Parthoens and Philippe Joye for technical assistance and Lauren Kosten and Stijn Servaes for the artwork for Fig. 1; and Hilde Lavreysen from the Department of Neuroscience, Janssen Pharmaceutica NV, Beerse, Belgium, for discussions of mGluR2.
Participated in research design: Wyckhuys, wyffels, Langlois, Schmidt, Stroobants, Staelens.
Conducted experiments: Wyckhuys, wyffels.
Performed data analysis: Wyckhuys, Staelens.
Wrote or contributed to the writing of the manuscript: Wyckhuys, wyffels, Langlois, Schmidt, Stroobants, Staelens.
- Received February 13, 2014.
- Accepted June 2, 2014.
This work was funded by Antwerp University, Belgium, through a full-time associate professor position for St.S., a part-time full professor position for Si.S., and a postdoctoral position for T.W.; and by Antwerp University Hospital, Belgium, through a part-time departmental position for Si.S. and a full-time position for L.w. M.S. and X.L. are with Janssen Research and Development, Beerse, Belgium.
Part of this work was presented as follows: Wyckhuys T, wyffels L, Langlois X, Schmidt M, Stroobants S, and Staelens S (2013) Evaluation of mGluR2 positive allosteric modulator JNJ-42153605 in an animal model of glutamatergic dysfunction using [18F]FDG microPET. Society for Neuroscience Annual Meeting; 2013 Nov 9–13; San Diego, CA.
- potassium 3′-([(2-cyclopentyl-6-7-dimethyl-1-oxo-2,3-dihydro-1H-inden-5-yl)oxy]methyl)biphenyl l-4-carboxylate
- computed tomography
- filtered backprojection
- 3-cylcopropylmethyl-7-(4-phenylpiperidin-1-yl)-8-trifluoromethyl [1,2,4] triazolo[4,3-a]pyridine
- (−)-(1R,4S,5S,6S)-4-amino-2-sulfonylbicyclo[3.1.0]hexane-4,6-dicarboxylic acid
- metabotropic glutamate receptor
- NMDA receptor
- positive allosteric modulator
- positron emission tomography
- standard uptake value
- volume of interest
- Copyright © 2014 by The American Society for Pharmacology and Experimental Therapeutics