INTRODUCTION

The metabotropic glutamate receptor subtype 5 (mGlu5) has emerged as an exciting new target for treatment of multiple neurological and psychiatric disorders. Selective mGlu5 negative allosteric modulators (NAMs) are advancing in clinical development for treatment of Fragile X syndrome, Parkinson’s disease, and affective disorders (Berg et al, 2011; Emmitte, 2013; Jacquemont et al, 2011) and positive allosteric modulators (PAMs) of mGlu5 may provide a novel approach to treatment of psychosis and cognitive disturbances in schizophrenia patients (Conn et al, 2011; Noetzel et al, 2012).

Advancing mGlu5 allosteric modulators in early clinical studies has been facilitated by the availability of highly selective positron emission tomography (PET) radioligands that allow unambiguous quantitative assessment of receptor occupancy (RO) in the brain after systemic administration of defined doses of test compounds (Majo et al, 2013; Mu et al, 2010). Determining the level of RO necessary for efficacy and adverse effects provides a noninvasive translational biomarker ideal for assessing the dose range required to engage the targeted receptor in preclinical and clinical studies (Bergstrom et al, 2004). The majority of drug candidates that enter development for central nervous system (CNS) disorders fail in late stage development owing to an inability to establish efficacy (Raskin and Casdin, 2011). Late stage failure is most commonly observed when adequate biomarker strategies are not available to allow definitive dose-finding studies (Kola, 2008). In instances where target engagement is not assessed, failure to achieve efficacy may be because of inadequate CNS exposure rather than lack of potential efficacy of the mechanism. Thus, it is critical to take full advantage of available radioligands for advancing novel mGlu5 allosteric modulators into clinical development.

The highly selective mGlu5 allosteric radioligand [18F]FPEB (Hamill et al, 2005) binds to an allosteric site on mGlu5 that is occupied by the prototypical mGlu5 NAM, 2-methyl-6-(phenylethynyl)pyridine hydrochloride (MPEP), and is being used to assess mGlu5 occupancy in multiple clinical studies (Hamill et al, 2005; Russell, 2014; Wang et al, 2007; Wong et al, 2013). It is clear that allosteric modulators can bind to multiple topographically distinct allosteric sites on mGlu5 (Gregory et al, 2014; Gregory et al, 2013). Therefore, many investigators rely on parallel optimization of PET ligands and drug leads from the same chemical scaffold to avoid complications associated with use of PET ligands and drug leads that do not interact with identical sites. However, recent molecular pharmacology, structural and molecular modeling studies provide the surprising finding that closely related mGlu5 allosteric modulators based on a single scaffold can inhibit binding of ligands to the prototypical MPEP/[18F]FPEB site by docking to overlapping but nonidentical sites within a large allosteric binding pocket (Chen et al, 2007; Dore et al, 2014; Gregory et al, 2014; Gregory et al, 2013; Rodriguez et al, 2010; Rodriguez et al, 2005; Wu et al, 2014). Conversely, molecular pharmacology and docking studies suggest that structurally unrelated mGlu receptor modulators can bind to the identical binding site in a fully competitive manner (Gregory et al, 2014; Gregory et al, 2013). These studies raise the possibility that very closely related mGlu5 modulators based on a single chemical scaffold may interact in a noncompetitive manner, which could complicate interpretation of in vivo occupancy studies. We now report that two highly selective mGlu5 PAMs (VU0092273 and VU0360172), both of which are close analogs within the same biaryl acetylenic nicotinamide chemical series as FPEB and MPEP (Noetzel et al, 2012; Rodriguez et al, 2010), show fundamentally different profiles in when assessing in vivo occupancy using [18F]FPEB. Systemic administration of VU0092273 displaced [18F]FPEB, with occupancies that closely align with its in vivo efficacy, whereas VU0360172 does not displace [18F]FPEB at doses well above those required for antipsychotic-like activity. In addition, a novel mGlu5 NAM, VU0409106, that bears no structural similarity to FPEB and MPEP, displaced [18F]FPEB binding in vivo, and displayed in vivo occupancy that fits well with its in vivo efficacy. Interestingly, all three allosteric modulators inhibit binding of a radioligand to the MPEP site in vitro. However, VU0092273 and VU0409106 interact with this site in a fully competitive manner, whereas VU0360172 interacts with this site in a manner that is not fully competitive. Thus, recent structural and modeling studies revealing that minor structural changes within the same chemical scaffold can significantly alter the binding interaction of mGlu receptor allosteric modulators has direct relevance to interpretation of results from in vivo occupancy studies. However, rigorous molecular pharmacology evaluation of binding interactions between individual systemically administered allosteric modulators and specific radioligands accurately predicts utility of ligand pairs for in vivo occupancy studies.

MATERIALS AND METHODS

Materials

Dulbecco’s modified Eagle’s medium, fetal bovine serum and antibiotics were purchased from Invitrogen (Carlsbad, CA). [3H]-labeled 3-methoxy-5-pyridin-2-pyridinylethynyl)pyridine ([3H]mPEPy) was purchased from PerkinElmer (Waltham, MA). MPEP was purchased from Tocris Bioscience (Elllisville, MO). VU0092273 (Rodriguez et al, 2010), VU0360172 (Rodriguez et al, 2010), VU0240382 (Williams et al, 2011), 5-methyl-2-phenylethynyl-pyridine (5MPEP) (Rodriguez et al, 2005), 3-((2-Methyl-4-thiazolyl)ethynyl)pyridine (MTEP) (Cosford et al, 2003b), and [18F]FPEB (Ansari et al, 2010; Hamill et al, 2005; Patel et al, 2005) were synthesized as described previously. Unless otherwise stated, all other reagents were purchased from Sigma-Aldrich (St Louis, MO) and were either analytical or HPLC grade.

Animals

All animal experiments were conducted in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals and were approved by the Institutional Animal Care and Use Committee. Male Sprague–Dawley rats (Harlan, Indianapolis, IN) weighing 275–325 g were housed three per cage under a 12-h light/dark cycle with access to food and water ad libitum.

Micro PET Imaging

Rats underwent surgery for carotid and jugular catheter insertion. Approximately 5 days later, rats were given an intraperitoneal (i.p.) injection of vehicle (10% Tween 80) or test compound in a 3-ml/kg volume. Thirty minutes post injection, rats were anesthetized using 1.5% isofluorane and positioned in a micro PET Focus 220 (Siemens, Knoxville, TN). Approximately ∼15 MBq/0.24 ml [18F]FPEB were injected via jugular catheter while simultaneously starting a 60-min dynamic acquisition. Blood samples were drawn via arterial catheter every 15 s for the first 2 min, then at 4, 6, 8, 12, 20, 30, 45, and 60 min. Samples were centrifuged and plasma activity was measured using a well counter (Capintec, Ramsey, NJ). Through acetonitrile extraction and thin layer chromatography, metabolite corrections were carried out on selected blood samples (2, 12, 30 and 60 min) and a metabolite-corrected arterial plasma input function was constructed. Attenuation corrections were carried out using a 57Co transmission scan. Images were reconstructed using OSEM 2D. The dynamic PET images consisted of five 60 s frames, two 150 s frames, two 300 s frames, and three 600 s frames. The data from all possible lines of response were saved in list mode and reconstructed using OSEM 2D with 16 subsets and 4 iterations after applying scatter and attenuation corrections. The reconstructed image matrix consisted of 128 × 128 × 95 voxels with voxel sizes of 0.095 × 0.095 × 0.08 cm3. The PET images were manually co-registered to an MRI rat brain template (Rubins et al, 2003) using Amide software (Loening and Gambhir, 2003). Regions-of-interest were drawn around the cerebellum and striatum of the template and superimposed on the PET images. Time-activity curves were established for these regions over the entire duration of the scan. Total volume of distribution (VT) using graphical analysis (Logan, 2000; Logan et al, 1996) was estimated for the striatum and cerebellum using PMOD (PMOD Technologies, Zurich, Switzerland). Percent occupancy was calculated as: percent occupancy=((VTvehicle−VTtreatment)/VTvehicle) × 100. Data were expressed as mean±SEM Data were normalized using the maximum occupancy of 3 mg/kg MTEP and analyzed using GraphPad Prism 5.0 (GraphPad Software, San Diego, CA).

Amphetamine-Induced Hyperlocomotion

Testing procedures were performed as previously described (Noetzel et al, 2012). In brief, test compounds were dissolved in 10% Tween 80 and double-deionized water, and the pH was adjusted to ∼7.0 using 1 N NaOH. Test compounds were administered i.p. in a 3-ml/kg volume. Vehicle or test compound was administered following a 30-min habituation period to an open-field chamber (KinderScientific, San Diego, CA). After an additional 30 min, rats received a saline vehicle or 1 mg/kg amphetamine injection subcutaneously. Locomotor activity was measured for an additional 60 min and depicted as the number of total photobeam breaks per 5-min intervals using Motor Monitor System software (KinderScientific). Main effects of test compound treatment on the locomotor activity area under the time course curve were evaluated using one-way analysis of variance. Comparisons of treatment group effects relative to the vehicle+amphetamine group were completed across the time interval from t=60−120 min using Dunnett’s post hoc tests with a p value of <0.05 considered significant.

Elevated Plus Maze

On the test day, rats were pretreated with vehicle or a dose of VU0409106 (1, 3, 10, or 30 mg/kg i.p.) and then acclimated in the testing room for 1 h. A 5-min test session began by placing each rat on the central platform facing an open arm of the elevated plus maze. Data were manually scored by an observer blinded to dose and expressed as time spent in the open arms as a percentage of the total testing time. Data were expressed as mean±SEM (n=8). Main effects of test compound treatment on the percent time spent in open arms of the maze were evaluated using one-way analysis of variance. Comparisons of treatment group effects relative to the vehicle group were completed using Dunnett’s post hoc tests with a p value of <0.05 considered significant.

Radioligand Binding Assays

The interaction of novel mGlu5 PAMs with the allosteric antagonist MPEP analog [3H]mPEPy (Cosford et al, 2003a) was assessed using membranes preparations from HEK293A cells stably expressing rat mGlu5 as previously described (Rook et al, 2012). In brief, compound concentration–response curves were diluted into assay buffer and added to each well of a 96-well plate, along with 50 μg/well cell membrane and 5 nM [3H]mPEPy. Following a 60-min incubation period at room temperature, the membrane-bound ligand was separated from free ligand by filtration through glass fiber 96-well filter plates (Unifilter-96, GF/B; PerkinElmer Life and Analytical Sciences, Boston, MA). Forty microliters of scintillation fluid was added to each well, and the membrane-bound radioactivity was determined by scintillation counting (TopCount; PerkinElmer Life and Analytical Sciences). Nonspecific binding was estimated using 10 μM MPEP.

Fluorescence-Based Calcium Assays in Rat mGlu5 Cells

HEK293A cells lines stably expressing rat mGlu5 or polyclonal mutant rat mGlu5 derived from the wild-type rat mGlu5 gene were maintained and site-directed point mutations in rat mGlu5-pCl:Neo (A809V and F585I) were generated as previously described (Gregory et al, 2012; Noetzel et al, 2012; Rook et al, 2012). Quantification of mGlu5-mediated intracellular Ca2+ mobilization was performed using Fluo-4, a Ca2+ sensitive dye, and a Flexstation II (Noetzel et al, 2012; Rodriguez et al, 2010; Rook et al, 2012). For 5-methyl-6-(phenylethynyl)-pyridine (5MPEP) assays, a 210-s protocol was used. 5MPEP was added at 20 s, followed by a PAM concentration–response curve at 80 s. An EC20 of glutamate was added at 170 s. Fold shift data was determined using a 130-s protocol where calcium assay buffer or 10 μM PAM was added at 20 s followed by a glutamate concentration–response curve at 90 s. Peak fluorescence response was obtained following the last compound addition and normalized to maximum glutamate response for each assay. Data were analyzed using GraphPad Prism 5.0 (GraphPad Software, San Diego, CA).

RESULTS

In Vivo RO/Efficacy Relationship of mGlu5 PAMs

Based on the previous findings that mGlu5 PAMs display antipsychotic-like activity and cognition enhancement in preclinical models, we sought to test the hypothesis that there will be a close relationship between mGlu5 in vivo RO in the CNS and efficacy of mGlu5 PAMs in reversing amphetamine-induced hyperlocomotion (AHL), a common preclinical assay utilized to assess antipsychotic-like activity of novel compounds. VU0092273 and VU0360172 are highly selective mGlu5 PAMs that possess robust efficacy in reversing AHL (Noetzel et al, 2012) and provide excellent tools to assess the mGlu5 RO/efficacy relationship of mGlu5 PAMs. Also, VU0092273 and VU0360172 are close structural analogs based on the same biaryl acetylenic nicotinamide scaffold as [18F]FPEB, MPEP, and MTEP (Figure 1). Finally, VU0092273 and VU0360172 potentiate responses to EC20 concentrations of glutamate in cells expressing mGlu5 with similar EC50 values of 35±5 and 13±2 nM (mean±SEM), respectively (Noetzel et al, 2012; Rodriguez et al, 2010). Thus, we performed PET imaging studies to determine whether [18F]FPEB is appropriate for evaluation of the in vivo RO of these two closely related mGlu5 PAMs. For comparison, we also evaluated the effect of MTEP as a prototypical biaryl acetylenic mGlu5 NAM ligand at this site.

Figure 1
figure 1

Structure activity relationship and binding interaction of mGlu5 (metabotropic glutamate receptor subtype 5) positive allosteric modulators (PAMs). VU0092273 and VU0360172 are mGlu5 PAMs and closely related structural analogs of the prototypical mGlu5 negative allosteric modulator (NAM), MPEP (methyl-6-(phenylethynyl)pyridine hydrochloride), and the mGlu5 NAM PET ligand [18F]FPEB. VU0409106 is structurally distinct from both mGlu5 PAM and NAMs.

PowerPoint slide

Figure 2a depicts representative images of [18F]FPEB binding in rat brains following pretreatment with either VU0092273, VU0360172, or MTEP. Consistent with previous studies (Anderson et al, 2003; Wang et al, 2007), MTEP (3 mg/kg, i.p.) induced a reduction of [18F]FPEB binding in the striatum of rats assessed 30 min after administration, with a maximal occupancy of 89%. Initial studies with MTEP demonstrated a 20% reduction in binding within the cerebellum, confirming significant mGlu5 expression within this brain region (data not shown). Therefore, arterial plasma input function and graphical analysis was utilized for PET imaging data analysis as opposed to a reference tissue model. Increasing doses of VU0092273 (3–100 mg/kg, i.p.) induced a similar reduction of [18F]FPEB binding in the striatum with an ED50 value of 17.3 mg/kg and maximal occupancy of 80.8% following 100 mg/kg administration i.p. (Figure 2a and b). Surprisingly, the closely related mGlu5 PAM, VU0360172, did not significantly reduce [18F]FPEB binding in striatum in vivo at doses up to 100 mg/kg (Figure 2a and c).

Figure 2
figure 2

VU0092273 displays dose-dependent in vivo receptor occupancy (RO) of the mGlu5 (metabotropic glutamate receptor subtype 5) MPEP (methyl-6-(phenylethynyl)pyridine hydrochloride) site. (a) Positron emission tomography representative images of [18F]FPEB binding in vehicle-, VU0092273-, VU0360172-, and MTEP-treated rat brains. (b) Pretreatment with VU0092273 (3–100 mg/kg, i.p.) dose-dependently displaces [18F]FPEB binding in the striatum with an ED50 of 17.3 mg/kg and maximum percent occupancy of 80.1±8. (c) In addition, VU0360172 does not significantly displace [18F]FPEB binding to mGlu5 in vivo, demonstrating that RO does not predict in vivo efficacy for this mGlu5 PAM. Time-activity curve of [18F]FPEB binding in the cerebellum and striatum following vehicle and VU0360172 (100 mg/kg, 30 min, i.p.). Data represent mean standard uptake value±SEM (n=5–7).

PowerPoint slide

To relate the in vivo occupancy studies to a common measure of efficacy of mGlu5 PAMs, the effect of increasing doses of VU0092273 and VU0360172 (10–56.6 mg/kg, i.p.) on AHL was evaluated. Both mGlu5 PAMs induced dose-dependent reversal of increases in locomotor activity caused by administration of 1 mg/kg amphetamine (subcutaneously, p<0.0001). The ED50 value of VU0092273 was 11.6 mg/kg with a maximum reversal of 67.3% (Figure 3a), demonstrating robust antipsychotic-like efficacy. VU0360172 demonstrated similar dose-dependent in vivo efficacy in this assay with an ED50 value of 15.2 mg/kg and maximum reversal of 60.7% (Figure 3b). Importantly, the in vivo potencies and maximal efficacies of these two mGlu5 PAMs were similar. The potency of VU0092273 in occupying the allosteric site on mGlu5 labeled by [18F]FPEB (ED50=17.3 mg/kg; Figure 2b) is in close agreement with the potency of this mGlu5 PAM in reversing AHL (ED50=11.6 mg/kg). Furthermore, the in vivo RO achieved with the minimal effective dose (10 mg/kg) of VU0092273 in AHL was 26%. The occupancy achieved with the lowest dose required for full efficacy (30 mg/kg) was ∼60%. These data reveal a close occupancy/efficacy relationship for this compound with relatively low RO required to achieve in vivo efficacy. In addition, full occupancy is not necessary to obtain maximal efficacy in this behavioral assay. However, the lack of effect of VU0360172 on [18F]FPEB binding suggests a more complex relationship between in vivo [18F]FPEB site binding and efficacy for these two closely related mGlu5 NAMs. Importantly, the mGlu5 antagonist, MPEP, blocked in vivo efficacy of the PAM in reversing AHL, supporting an mGlu5-mediated effect (data not shown).

Figure 3
figure 3

Despite differential displacement of [18F]FPEB in vivo, both VU0360172 and VU0092273 exhibit an antipsychotic-like profile in rats. (a) VU0092273, an mGlu5 (metabotropic glutamate receptor subtype 5) positive allosteric modulator, displays antipsychotic-like activity. Male Sprague–Dawley rats treated with increasing doses of VU0092273 (10–56.6 mg/kg, i.p.) exhibited dose-dependent reversal of amphetamine-induced hyperlocomotion with an ED50 of 11.6 mg/kg and maximum efficacy observed at a dose of 30 mg/kg. Data represent mean±SEM (n=5–7). (b) VU0360172 displayed antipsychotic-like activity similar to VU0092273. Male Sprague–Dawley rats treated with increasing doses of VU0360172 (10–56.6 mg/kg, i.p.) exhibited dose-dependent reversal of amphetamine-induced hyperlocomotion with an ED50 of 15.2 mg/kg. Data represent mean±SEM (n=5–8).

PowerPoint slide

As a negative control, we also performed studies with VU0240382, a close structural analog of VU0092273, that is a potent mGlu5 PAM that has been shown to interact with the MPEP/FPEB site in a fully competitive manner (Noetzel et al, 2012; Rook et al, 2011). However, VU0240382 does not achieve appreciable brain exposure. We previously reported that the unbound fraction of VU0240382 in rat brain after systemic administration is 0.01%, resulting in low brain exposure after systemic administration (Noetzel et al, 2012). Consistent with this, a relatively high dose of VU0240382 (56.6 mg/kg, i.p.), did not significantly reverse AHL (Figure 4a). In contrast to MTEP and VU0092273, VU0240382 did not significantly displace [18F]FPEB binding in vivo (Figure 4b). This is consistent with the low brain exposure and lack of in vivo efficacy in the AHL model achieved with systemic administration of this compound.

Figure 4
figure 4

Subtle structural changes to mGlu5 (metabotropic glutamate receptor subtype 5) PAMs (positive allosteric modulators) result in significant alterations in in vivo efficacy and/or receptor occupancy. (a) Unlike VU0092273 and VU0360172, VU0240382 does not significantly reverse amphetamine-induced hyperlocomotion in rats (56.6 mg/kg, i.p.) Data represent mean±SEM (n=5–8). (b) The highly potent, selective mGlu5 PAM, VU0240382, does not exhibit in vivo occupancy of mGlu5. VU0240382 does not significantly displace [18F]FPEB binding in vivo. Time-activity curve of [18F]FPEB binding in the cerebellum and striatum following vehicle and VU0240382 (100 mg/kg, 30 min, i.p.) administration. Data represent mean standard uptake value±SEM (n=5–7). (c, d) Pretreatment with structurally diverse mGlu5 negative allosteric modulator, VU0409106 (1–30 mg/kg, i.p.), dose-dependently displaces [18F]FPEB binding in the striatum with an ED50 of 7.5 mg/kg and maximum percent occupancy of 87.7%±2. Data represent mean±SEM (n=5–7). (e) VU0409106 demonstrated anxiolytic-like behavior in the elevated plus maze in male Sprague-Dawley rats. VU0409106 dose-dependently (1–30 mg/kg, i.p.) increased time spent in open arms with an ED50 of 8.2 mg/kg. Data represent mean±SEM. ***p< 0.0001 (n=8).

PowerPoint slide

In addition to evaluating the in vivo efficacy/occupancy relationship of closely related members of the biaryl acetylenic mGlu5 allosteric modulator series, we also evaluated the effects of a novel mGlu5 NAM, VU0409106 (Felts et al, 2013). This mGlu5 NAM bears no structural resemblance to FPEB and the other mGlu5 modulators studied (Figure 1). However, recent molecular modeling and docking studies suggest that this mGlu5 NAM binds to the MPEP/FPEB site in a manner similar to MPEP (Gregory et al, 2013). Interestingly, VU0409106 induced a dose-dependent displacement of in vivo [18F]FPEB binding with an ED50 of 7.5 mg/kg and maximum occupancy of 87.7% at 30 mg/kg (Figure 4c and d). Consistent with the established anxiolytic-like activity of mGlu5 NAMs, VU0409106 dose-dependently enhanced the percent time spent in the open arms of the elevated plus maze (p<0.0001) with an ED50 of 8.2 mg/kg and a maximum increase of 29.2% (Figure 4d). Thus, the potency of VU0409106 in occupying the MPEP site of mGlu5 aligns closely with its potency in the elevated plus maze assay (8.2 mg/kg). In addition, these data are in line with occupancies of other mGlu5 NAMs that are required for efficacy in the elevated plus maze assay.

Differential Interaction of mGlu5 PAMs with the Prototypical Allosteric MPEP Site

Based on the close structural similarities of VU0092273 and VU0360172 to MPEP and FPEB, we originally postulated that each of these compounds would interact with a previously defined allosteric site that is shared by multiple mGlu5 modulators, including MPEP and FPEB. However, the finding that VU0360172 does not reduce in vivo [18F]FPEB binding at doses that have full efficacy in rodent models suggests an added complexity to the nature of interactions between these ligands or the relationship between mGlu5 occupancy and efficacy. Interestingly, mutagenesis and structural docking studies suggest that VU0360172 may interact with the MPEP/FPEB site in a manner that is overlapping but not identical with the interaction by MPEP (Gregory et al, 2013). Thus, it is conceivable that the lack of effect of VU0360172 on in vivo [18F]FPEB binding reflects a more complex interaction at this site that has been predicted by structural modeling studies. To further understand the nature of interactions of these ligands with the MPEP site, we performed equilibrium radioligand binding studies in membranes from cells expressing mGlu5 using [3H]methoxy-PEPy as a well-characterized radioligand for this site (Cosford et al, 2003a). VU0092273, VU0360172, VU0409106, FPEB, and MPEP all inhibited [3H]methoxy-PEPy binding to mGlu5-expressing cell membranes with rank order potencies of FPEB (0.2 nM) >MPEP (51 nM) >VU0409106 (118 nM) >VU0360172 (203 nM) >VU0092273 (2.7 μM) (Figure 5a). This is in line with previous studies demonstrating that mGlu5 PAMs typically have higher potencies in functional assays than apparent affinities at the allosteric site owing to the combined influence of affinity and cooperativity in determining PAM functional potencies (Conn et al, 2009; Conn et al, 2014). However, in contrast to each of the other mGlu5 modulators, which fully displaced binding, VU0360172 only partially inhibited [3H]methoxy-PEPy binding with a maximum inhibition of 83% (Figure 5a). Although FPEB and MPEP are known to interact with the MPEP site in a fully competitive manner, this finding raised the possibility that VU0360172 may interact with this site in a manner that is distinct from that of the other mGlu5 allosteric modulators and is not fully competitive. In addition, increasing the concentration of the radioligand resulted in a concentration-dependent decrease in [3H]methoxy-PEPy binding by VU0360172 (data not shown), further supporting a noncompetitive interaction with the MPEP site.

Figure 5
figure 5

VU0092273 and VU0409106 display differential interactions than VU0360172 with the MPEP (methyl-6-(phenylethynyl)pyridine hydrochloride) binding site of mGlu5 (metabotropic glutamate receptor subtype 5). (a) VU0092273 and VU0409106, but not VU0360172, fully inhibit [3H]methoxy-PEPy binding with potencies of 2.7 μM and 118 nM in mGlu5 rat cell membranes. (b) Addition of increasing concentrations of the neutral ligand 5MPEP (5-methyl-2-phenylethynyl-pyridine) results in a parallel rightward shift in VU0092273 and VU0409106 concentration–response curve with no change in maximum glutamate response. However, 5MPEP decreases in the maximum glutamate response evoked by VU0360172, and Schild regression analysis depicts a slope of 0.73, suggesting a noncompetitive interaction. (c) Introduction of the single-point mutation A809V, which abolishes activity at the mGlu5 allosteric MPEP-site, eliminated the shift in the glutamate concentration–response curves evoked by VU0092273 or VU0409106 observed in polyclonal wild-type rat mGlu5 cells, while only partially attenuating the VU0360172-induced fold shift (4.4 to 2.8). Data represent the mean±SEM of three independent experiments performed in duplicate.

PowerPoint slide

Based on these radioligand binding data, we performed further functional studies to rigorously evaluate the nature of the interaction between VU0092273, VU0360172, and VU0409106 at the MPEP site. Multiple previous studies have taken advantage of the MPEP-site neutral allosteric ligand, 5MPEP, to perform Schild analyses to determine whether specific mGlu5 PAMs potentiate mGlu5-mediated responses by competitive interaction at the canonical MPEP site. Increasing concentrations of 5MPEP (30–300 μM), induced concentration-dependent parallel rightward shifts of the VU0409106 or VU0092273 concentration–response curves at inhibiting or potentiating responses of glutamate, respectively, with no effect on the maximum glutamate response (Figure 5b). This is consistent with previous studies with other mGlu5 allosteric modulators (Chen et al, 2008; Chen et al, 2007; Rodriguez et al, 2010; Rodriguez et al, 2005; Rook et al, 2012) and supports the hypothesis that VU0092273 potentiates mGlu5-mediated responses by competitive interaction at the MPEP site. However, consistent with the radioligand binding studies, pretreatment of mGlu5-expressing cells with 5MPEP (30–300 μM) significantly reduced the maximum response induced by VU0360172. Furthermore, Schild regression analysis of these data revealed a slope of unity for VU0409106 and VU0092273 but not VU0360172 (Figure 5b), suggesting that VU0409106 and VU0092273 elicit their effects by a competitive interaction with the MPEP allosteric binding site, whereas the interaction of VU0360172 with the MPEP site is noncompetitive in nature.

Additional studies utilized single-point mutagenesis of mGlu5 to further evaluate the interaction of VU0092273, VU0360172, and VU0409106 with the MPEP binding site. A single-point mutation (A809V) was introduced into rat mGlu5, which has been previously shown to result in a loss of function of ligands that bind competitively to the MPEP site (Pagano et al, 2000; Rook et al, 2012). PAM activity of VU0092273 was completely abolished in cells expressing mGlu5-A809V. In addition, NAM activity of VU0409106 was also absent in mGlu5-A809V cells (Figure 5c). However, VU0360172-induced potentiation of the mGlu5-mediated response remained intact in the A809V mutant cells (Figure 5c). A similar effect with VU0360172 is observed in the S808A mGlu5 mutation, which has also been shown to completely abolish MPEP site activity (data not shown) (Gregory et al, 2013). Taken together with the incomplete displacement of [3H]methoxy-PEPy binding and results from the Schild analysis, these data suggest that while structurally similar, VU0360172 does not interact with the MPEP site of mGlu5 in a fully competitive manner. Interestingly, while the mGlu5 NAM, VU0409106, is structurally distinct from MPEP, these data demonstrate a fully competitive interaction.

DISCUSSION

The use of PET imaging has emerged as a critical tool that allows quantitative assessment of interactions of novel pharmacological agents with their intended target in the CNS and other tissues to ensure accurate human dose projections (Bergstrom et al, 2004; Castner et al, 2014; Facklam et al, 1992). However, while direct measures of RO are viewed as more reliable than the use of functional biomarkers (EEG, fMRI, evoked potentials, and so on) for assessing target engagement, interactions of ligands with specific binding sites can be complex. A lack of understanding of the nature of these ligand–receptor interactions can complicate studies using selective ligands as imaging reagents to assess RO. Thus, it is critical to rigorously evaluate the interaction between each PET ligand and the drug lead or drug candidate being investigated.

Availability of [18F]FPEB and related ligands have played a critical role in guiding early clinical studies for mGlu5 NAMs and may also provide a viable approach for establishing CNS occupancy for mGlu5 PAMs. Two other mGlu5 radioligands that have been reported, [11C]ABP688 (Ametamey et al, 2006) and [18F]SP203 (Shetty et al, 2008), are structurally closely related to [18F]FPEB and developed from the MPEP scaffold (Andersson et al, 2013). As new allosteric modulators advance, it is hoped that developing drug candidates and PET ligands based on the same chemical scaffold will mitigate problems associated with complex interactions between different ligands acting at GPCR allosteric sites. Based on their close structural similarity, it is commonly assumed that these compounds bind to the same site as MPEP and FPEB in a fully competitive manner. However, recent advances in understanding the structural and molecular basis for allosteric modulator interactions with mGlu receptors reveal that closely related mGlu5 allosteric modulators can differentially interact with a single binding pocket. The present studies reveal that these in vitro complexities can lead to fundamentally different profiles when assessing in vivo occupancy with a closely related PET ligand. Thus, allosteric modulators that are close structural analogs of a given PET ligand may inhibit radioligand binding in vitro, but show fundamentally different profiles in reducing binding of a PET ligand in vivo. Thus, while [18F]FPEB provided an excellent measure of in vivo occupancy for VU0092273, this PET ligand was not useful for assessing occupancy of VU0360172. This is especially interesting in light of the finding that VU0360172 was actually more potent than VU0092273 in inhibiting equilibrium radioligand binding to this site in vitro using membrane preparations from cells expressing mGlu5. However, while both MPEP/FPEB analogs inhibit in vitro binding, detailed molecular pharmacological analysis reveals that VU0092273 does so in a fully competitive manner, whereas VU0360172 displays a noncompetitive interaction with this site. Furthermore, a structurally unrelated mGlu5 NAM, VU0409106, interacts with the FPEB/MPEP site in a fully competitive manner and displaces [18F]FPEB binding in vivo with a predicted occupancy that fits well with its in vivo efficacy. Thus, rigorous molecular pharmacological analysis was successful in accurately predicting utility of a given PET ligand for assessing occupancy of a test compound. These data are consistent with recent molecular modeling and docking studies that suggest that VU0360172 binds to the allosteric site on mGlu5 with a pose that is distinct from that of MPEP (Gregory et al, 2013). In contrast, docking studies predicted that VU0409106, while structurally distinct, binds to this site in a manner that is overlapping with the binding of MPEP or FPEB (Gregory et al, 2014). At present, it is not known whether [18F]FPEB, [11C]ABP688, and [18F]SP203 bind to mGlu5 in the same manner, or whether they will assess different binding poses of allosteric modulators to this site. More recently, [11C]AZD9272 was reported as the first mGlu5 PET radioligand developed from a scaffold that is chemically distinct (Andersson et al, 2013). In future studies, it will be important to develop a full understanding of the nature of each of these radioligands with mGlu5.

Although VU0360172 interaction with the FPEB/MPEP site is not strictly competitive, it is still somewhat surprising that this compound inhibits binding to this site in vitro with a higher potency than VU0092273 but does not inhibit binding in vivo. VU0360172 achieved high brain concentrations (Noetzel et al, 2012) and had robust in vivo efficacy in reversal of AHL with an in vivo potency similar to that of VU0092273. This suggests that this mGlu5 PAM clearly enters the brain at concentrations sufficient to potentiate mGlu5 and illustrates the critical issue of context dependence of complex noncompetitive ligand interactions that can impact in vivo studies. In cases where two ligands interact with a site in a noncompetitive manner, the interactions between the two ligands can vary widely depending on the specific cellular and membrane environment in different cell populations. The finding that VU0360172 was unable to displace [18F]FPEB binding in vivo provides a clear demonstration of the critical impact of subtle variation of the nature of interactions of individual drug leads with specific PET ligands and the need to rigorously evaluate the nature of this interaction for each ligand pair before advancing to in vivo RO studies.

For test compounds whose occupancy could be accurately assessed using [18F]FPEB, these studies also provide new insights into the RO/efficacy relationship. Maximum in vivo efficacy of the mGlu5 PAM VU0092273 in the AHL model is observed at relatively low RO. Unlike NAMs, which demonstrate a close correlation between potency and affinity at mGlu5, mGlu5 PAMs have been shown to possess significant positive cooperativity between the functional potency in potentiating the glutamate response at mGlu5 and their affinity for the receptor (Chen et al, 2007; Conn et al, 2014; Gregory et al, 2012). Although this phenomenon has been extensively evaluated in in vitro systems, it remained unknown as to whether this was an artifact of the artificial systems used. These data provide the first direct evidence that the relationship of positive cooperativity exists in vivo. Therefore, these data suggest that when designing a dosing regimen for clinical studies to assess the antipsychotic-like efficacy of mGlu5 PAMs, doses chosen do not need to achieve full RO in humans.

In conclusion, the current findings support the utility of [18F]FPEB and PET imaging as a useful biomarker for the development of mGlu5 PAMs. These studies support a close relationship between in vivo RO and efficacy and suggest that RO of mGlu5 can be used as a valuable tool to accurately predict doses of PAMs required for antipsychotic efficacy. However, subtle differences in ligand docking suggested by recent structural and modeling studies can lead to fundamental differences in ligand interactions of mGlu5 modulators, even for modulators that are closely related structural analogs. Detailed molecular pharmacology analysis is essential for establishing a fully competitive interaction of two ligands before advancing to in vivo occupancy studies.

FUNDING AND DISCLOSURE

Dr Rook’s work has been funded by NIH, Alzheimer’s Drug Discovery Foundation, and Thome Alzheimer’s Disease Drug Discovery Foundation. Dr Felts has received compensation from Seaside Therapeutics. Dr Stauffer’s work has been funded by the NIH and by an industry sponsored research contract from Janssen. He is an inventor on patents that protect different classes of mGlu5 PAMs. Dr Emmitte’s work has been funded by the NIH and collaborative research agreements with Seaside Therapeutics. He is an inventor on patents that protect different classes of mGlu5 NAMs. Dr Kessler’s work has been funded by Novo Nordisc and by the NIH. In addition, he is a stockholder in PHarmRx Therapeutics and a consultant for Shire Pharmaceuticals. Dr Niswender’s work has been funded by the NIH, the International Rett Syndrome Foundation and Autism Speaks. She has received licensing royalties from Johnson and Johnson, Bristol–Myers Squibb, and Seaside Therapeutics, and is an inventor on patents that protect different classes of metabotropic glutamate and muscarinic receptor allosteric modulators. Dr Daniels’ work has been funded in part by the NIH and the Molecular Libraries Probe Production Centers Network. He has received compensation from Johnson and Johnson, Bristol-Myers Squibb, Seaside Therapeutics and as a member of the scientific advisory board of the Sigma-Aldrich company and through consulting for Agios Pharmaceuticals Company and the Michael J. Fox Foundation. He is an inventor on patents that protect different classes of metabotropic glutamate and muscarinic receptor allosteric modulators. Dr Jones received research support from Bristol–Myers Squibb, Johnson and Johnson, and AstraZeneca. Dr Jones also received funding through the Michael J. Fox Foundation, the Barrus Foundation, and NIH. She is an inventor on patents that protect different classes of metabotropic glutamate and muscarinic receptor allosteric modulators. Dr Lindsley's work has been funded by the NIH, Bristol–Myers Squibb, AstraZeneca, Michael J. Fox Foundation, as well as Seaside Therapeutics. He has consulted for AbbVie and received compensation. He is an inventor on patents that protect different classes of metabotropic glutamate and muscarinic receptor allosteric modulators. Dr Conn has been funded by NIH, Johnson and Johnson, AstraZeneca, Bristol–Myers Squibb, Michael J. Fox Foundation, and Seaside Therapeutics. He has consulted over the past 3 years for Pfizer, Cambridge, and Millipore Corporation, and received compensation. Over the past 3 years he has served on the Scientific Advisory Boards and received compensation from Seaside Therapeutics, Michael J. Fox Foundation, Stanley Center for Psychiatric Research Broad Institute (MIT/Harvard), Karuna Pharmaceuticals, Lieber Institute for Brain Development Johns Hopkins University, Clinical Mechanism and Proof of Concept Consortium, and Neurobiology Foundation for Schizophrenia and Bipolar Disorder He is an inventor on patents that protect different classes of metabotropic glutamate and muscarinic receptor allosteric modulators. Dr Tantawy, Dr Ansari, and Dr Felts declare no potential conflict of interest.