|Home | About | Journals | Submit | Contact Us | Français|
An iterative analogue library synthesis strategy rapidly developed comprehensive SAR for the mGluR5 ago-potentiator ADX-47273. This effort identified key substitutents in the 3-position of oxadiazole that engendered either mGluR5 ago-potentiation or pure mGluR5 positive allosteric modulation. The mGluR5 positive allosteric modulators identified possessed the largest fold shifts (up to 27.9-fold) of the glutamate CRC reported to date as well as providing improved physiochemical properties.
With an onset in late adolescence, schizophrenia, a complex psychiatric disorder characterized by a combination of negative (social withdrawal, blunting of emotional responses, anhedonia) and positive (hallucinations, delusions, paranoia, disorganized behavior) symptoms along with significant cognitive dysfunction is a debilitating disease that requires lifelong, daily maintenance therapy at a cost to society of $65 billion a year. The prevailing dogma by which schizophrenia has been managed for decades states that excessive dopaminergic transmission in the forebrain underlies the disease – the so-called “dopamine hypothesis” or “dopamine hyperfunction hypothesis”. The rationale for this hypothesis is based on the fact that all clinically relevant antipsychotic agents, both typical (haloperidol) and atypical (clozapine, olanzapine), possess significant antagonist activity at the dopamine D2 receptor; however, these agents have a slow onset of action and mainly treat the positive symptoms of schizophrenia, with limited to no effect on the negative and cognitive symptoms, thereby representing a substantial unmet medical need. Moreover, all of these agents bind to a number of neurotransmitter receptors such as dopamine (D1–D4), serotonin (5-HT1A, 1D, 2A, 2C, 6 and 7), adrenergic (α1, α2), histamine (H1) and muscarinic (M1); therefore, the observed efficacy can more accurately be ascribed to polypharmacology.[4–10]
The N-methyl-D-aspartate (NMDA) receptor antagonist phencyclidine (PCP) has been shown to induce the positive, negative and cognitive symptoms of schizophrenia in healthy patients and elicit a resurgence of symptoms in stable schizophrenics. In the clinic, the observation that administration of the NMDA receptor co-agonist glycine provides a modest improvement in schizophrenic patients suggests that increasing NMDA receptor activation may provide a therapeutic benefit. These observations led to the NMDA receptor hypofunction hypothesis as an alternative theory for the underlying cause of schizophrenia. According to this hypothesis, any agent that can potentiate NMDA receptor currents, either directly by action on modulatory sites on the NMDA receptor (i.e., the glycine co-agonist binding site) or indirectly by activation of GPCRs known to potentiate NMDA receptor function (i.e., mGluR5), has the potential to ameliorate the symptoms of schizophrenia.
Glutamate is the major excitatory transmitter in the central nervous system, exerting its effects through either ionotropic or metabotropic glutamate receptors. The metabotropic glutamate receptors (mGluRs) are members of the GPCR family C, characterized by a large extracellular amino-terminal binding domain. To date, eight mGluRs have been cloned, sequenced and assigned to three groups (Group I: mGluR1 and mGluR5; Group II: mGluR2 and mGluR3; Group III: mGluRs 4,6,7,8) based on their structure, coupling to effector mechanisms and pharmacology. Achieving mGluR subtype selectivity has been difficult with orthosteric agonists, which are typically analogs of glutamate, quisqualate or phenyl glycine. Recently, our laboratories have reported on the discovery of positive allosteric modulators (PAMs) of mGluR5, compounds which alone have no effect on mGluR5 function, but potentiate mGluR5 receptor response in the presence of sub-threshold levels of the native agonist glutamate. We have also identified compounds that display allosteric agonist activity at higher concentrations and are more acurately termed mGluR5 ago-potentiators. By virtue of binding at an allosteric site on the receptor, these ligands afford high mGluR5 sub-type selectivity (>100-fold selectivity versus mGluR1,2,3,4,7,8). We have identified (Figure 1) three chemotypes of mGluR5 PAMs, represented by DFB (1), CPPHA (2), and the MPEP-like 3. In addition, we discovered the ago-potentiator CDPPB (4). Interestingly, 1, 3, and 4 have been shown to bind at the MPEP binding site, while 2 binds at a yet undefined second allosteric site on mGluR5.[16–19] CDPPB (4) was the first centrally active mGluR5 PAM/ago-potentiator, which allowed us to validate in vivo that allosteric activation of mGluR5 possessed an antipsychotic profile in rat behavioral models. However, CDPPB (4) displayed poor solubility in most vehicles, limiting its utility for further in vivo studies, and a disparity between binding and functional potency. Moreover, due to the intrinsic allosteric agonist activity, though weak, we were unable to validate pure mGluR5 potentiation as the mechanism for the in vivo activity. Lead optimization efforts of the CDPPB (4) scaffold were unable to address these issues.[19,20]
In 2005, Addex disclosed the structurally distinct ago-potentiator ADX-47273 (5) and subsequently produced patents.[21,22] Researchers at both Addex[21,22] and Wyeth have recently reported on the in vivo efficacy of ADX-47273 (5) in a number of preclinical antipsychotic and cognition models, further validating selective mGluR5 activiation as a potential new mechanism to address the complex symptom clusters (positive, negative and cognitive symptoms) of schizophrenia. However, ADX-47273 (Figure 2) is a more potent ago-potentiator (mGluR5 PAM EC50 = 168 nM, 107% Glu Max, 9-fold shift @ 1 μM; Allosteric agonist EC50 = 9.5μM, 65% Glu Max) than CDPPB (4), but still suffers from poor physiochemical properties due to a lack of solublizing moieties.[18–22]
In order to further validate potentiation of mGluR5 as a therapeutic approach for the treatment of the positive, negative and cognitive symptom clusters of schizophrenia and advance the science concerning mGluR5, pure mGluR5 PAMs and mGluR5 ago-potentiators with improved pharmacological and physiochemical properties are required. Little has been disclosed concerning the SAR and pharmalogical profiles of ADX-47273 and its analogues, [21,22] so our lab initiated a campaign to explore the ADX47273 scaffold in an effort to improve the physiochemical properties of ADX-47273, determine if a pure PAM could be identified in this series and address three questions: 1) are alternative aryl/heteroaryl rings tolerated at the 3-position of the oxadiazole, 2) are alternative amides tolerated and 3) is the (S) stereochemistry required for mGluR5 PAM activity (Figure 3)?
We elected to pursue a focused library approach to explore the SAR of ADX-47273. First, we prepared a small library of analogues 6 evaluating known phenyl iosteres to replace the 4-FPh moiety of the 3-position of the oxadiazole within ADX-47273 in an attempt to incorporate solubilizing and/or polar groups while maintaining the 4-FPh amide and the (S)-stereochemistry (Table 1). Pyridine isosteres afforded intriguing SAR. Analogue 6a, a pyridyl congener of the 4-FPh moiety of ADX-47273, lost ~8-fold in potency (EC50 = 1, 460 nM), while the 2-pyridyl analogue 6b lost only 2-fold in potency, but maintained efficacy (EC50 = 348 nM, 109% Glu Max). The 4-pyridyl isomer 6d was comparable to 6a, while the 3-pyridyl 6c lost significant potency (EC50 = 5,000 nM). The 2-thienyl congener 6e (EC50 = 170 nM) was equipotent to ADX-47273, and the 2-pyrimidinyl analogue 6f lost all PAM activity.
These data then led to design of second generation library wherein the optimal 3-position groups with submicromolar EC50s (4-FPh (5), 2-pyridyl (6b) and 2-thienyl (6e) from Table 1) were maintained while evaluating a diverse set of twelve acylating agents. In this second generation library (Figure 4) the (S)-stereochemistry of the 3-piperidine carboxylic acid was again maintained to provide analogues 10a-l, 11a-l and 12a-l.
Ultimately, we followed two synthetic routes to access ADX-47273 analogs 10–12 (Scheme 1). In route I, three (Z)-N′-hydroxyimidamides 7 (R = 4-FPh, 2-thienyl, 2-pyridyl) were coupled under standard EDCI/HOBt conditions with (S)-(tert-butoxycarbonyl)piperidine-3-carboxylic acid, followed by refluxing in 1,4-dioxane to afford oxadiazoles 8. The Boc group was removed with 4 N HCl/dioxane to provide 9, followed by typical acylation chemistry with 12 diverse acid chlorides to provide ADX-47273 analogs 10 (R = 4-FPh), 11 (R = 2-thienyl) and 12 (R = 2-pyridyl). Alternatively, analogs 10–12 could be accessed according to route II, wherein the oxadiazole is installed in the final step. In this scenario, (S)-piperidine-3-carboxylic acid 13 is converted to the corresponding methyl ester 14, followed by typical acylation chemistry to deliver analogs 15 in good yields. Saponification with LiOH provides the corresponding acids which were then converted into the corresponding analogs 10 (R = 4-FPh), 11 (R = 2-thienyl) and 12 (R = 2-pyridyl). We employed route I for the initial library generation of 36 ADX-47273 analogs 10–12, and generally relied on route II for scale-up of interesting compounds.
Robust SAR was observed for ADX-47273 analogs 10–12, but the most striking finding was that a subtle change in the nature of the substituent in the 3-position of the oxadiazole afforded either potent mGluR5 ago-potentiators (10 and 11) or pure mGluR5 positive allosteric modulators (12), with little or no detectable agonist activity. As shown in Table 1, analogs 10a–10f and 10j were potent with submicromolar EC50s (133–714 nM) and efficacious (92–107% glutamate max) mGluR5 ago-potentiators when the amide moiety was a either a fluorinated or cyano-containing benzamide or a 2-thienyl amide. Other amides (heterocyclic, aryl and cycloalkyl) possessed EC50s > 1 μM, and were therefore not useful as potential in vivo candidates.
A striking similar trend of activity was noted with amide analogues in the 2-thienyl series 11 (Table 3). Both potency and efficacy paralleled the 4-FPh series 10 for analogues 11a-g (EC50s 144–1,170 nM) and efficacious (90–109% glutamate max). In this instance as well, other amides (heterocyclic, aryl and cycloalkyl) possessed EC50s > 1 μM, and were therefore not useful as potential in vivo candidates.
All of the analogs 10a–10f, 10j and 11a–11g possessed significant mGluR5 allosteric agonist activity as well as positive allosteric modulator activity, and are therefore more accurately described as ago-potentiators, like the parent ADX-47273 (5/10a). With many of these analogues, the agonist activity was so strong, that it precluded fold-shift data from being calculated, or forced these experiments to be conducted at very low doses of compound (i.e., fold shifts at 370 nM versus the typical 1–10 μM). Figure 5 highlights a prototypical ago-potentiator from this series. The raw calcium fluorescence trace clearly illustrates the intrinsic mGluR5 agonsist activity as well as the potentiation of an EC20 concentration of glutamate. Full CRCs confirm an EC50 for potentiation of 133 nM and an agonst EC50 of 5 μM for 10c. Like most of the analogues 10a–10f, 10j and 11a–11g, the intrinsic agonist activity required that fold shift experiments be conducted at compound concentrations of only 370 nM; however, 4- to 5-fold shifts of the glutamate CRC were still observed. Moreover, all of these analogues were selective for mGluR5 (no activity at mGluRs 1, 2, 3, 4, 7 or 8). As with the parent, ADX-47273 (5), none of these analogues offered an improvement in solubility across pharmaceutically acceptable vehicles (clogPs were >4.5) and overall physiochemical properties were poor. Still, this effort provided potent and efficacious ago-potentiators for further studies.
Analogues 12a–12d, 12f and 12g, which contained a 2-pyridyl moiety in the 3-position of the oxadiazole, afforded an unanticipated pharmacological profile. The potency and efficacy of these analogues 12 were comparable to, or slightly less potent (EC50s 244–757 nM) and efficacious (89–109% glutamate max) than 10a–10f, 10j and 11a–11g (Table 4). In addition to providing a basic nitrogen atom capable of salt formation, analogues 12a–12d, 12f and 12g demonstrated an unexpected profile–pure mGluR5 positive allosteric modulation. Indeed, all of these analogues either displayed no mGluR5 agonism, or only a trace (<10% at 30 μM) at high compound concentrations.
Figure 6 highlights a prototypical mGluR5 positive allosteric modulator, (12b) from this series. The raw calcium fluorescence trace clearly illustrates a complete lack of intrinsic mGluR5 agonsim, but a robust potentiation of an EC20 concentration of glutamate. Full CRCs confirm an EC50 for potentiation of 244 nM with no agonism by compound alone up to 30 μM, a finding in sharp contrast to 10a–10f, 10j and 11a–11g. As there was no agonist activity, fold shift experiments could be conducted at standard concentrations, and (12b) afforded a strong 14-fold shift of the glutamate CRC at 1 μM. Other analogs in this series provided similar fold shifts, but (12c) elicited an unprecedented 27.9-fold shift of the glutamate CRC at 1 μM – the largest fold-shift reported to date for an mGluR5 PAM. Moreover, all of these analogues were selective for mGluR5 (no activity at mGluRs 1, 2, 3, 4, 7 or 8). Unlike the parent, ADX-47273 (5) and 10a–10f, 10j and 11a–11g, the corresponding HCl salts of 12a–12d, 12f and 12g offered an improvement in solubility across pharmaceutically acceptable vehicles (clogP <3.6 – a full log improvement) and overall physiochemical properties were improved, such that homogeneous dosing solutions could be obtained in tox-friendly vehicles (saline (5–25 mg/mL), β-cylcodextrin (10–20 mg/mL) as opposed to the 10 and 11 series which only afforded homogeneous solutions/microsuspensions in toxic PEG/DMSO vehicles (~ 5mg/mL).
Surprisingly, one analogue 12k, with a cyclobutyl amide, demonstrated a ‘switch’ in mode of pharmacology and was found to be an mGluR5 negative allosteric modulator (NAM). While weak (IC50 = 8.7 μM, 23% Glu Max), this is the first time we have observed this ‘switch’ of modes of pharmacology in a non-MPEP chemotype (Figure 7).[16,18]
Finally, we wanted to evaluate the stereochemistry at the C3 position of the piperidine-3-carboxylic acid 13, as ADX-47273 (5) was the first reported mGluR5 PAM possessing a chiral center, and we wanted to determine if there was enantioselective potentiation. Following the synthetic routes in Scheme 1 and substituting (R)-piperidine-3-carboxylic acid for the previously employed (S)-congener 13, allowed us to synthesize and evaluate the corresponding (R)-enantiomers of representative analogs from the 10–12 series and address this issue. As shown in Table 5, the (R)-enantiomers are uniformly 9- to 10-fold less potent than the corresponding (S)-enantiomers, but equally efficacious. This constitutes the first reported example of enantioselective potentiation of mGluR5, and while Addex has only reported on the (S)-enantiomers, this work quantifies the importance of the (S)-stereochemistry for activation of mGluR5.
In summary, we explored the chemical space surrounding the mGluR5 ago-potentiator ADX-47273 (5) employing an iterative library design comprised of three small libraries (~60 compounds). This effort identified potent mGluR5 ago-potentiators 10a–10f, 10j and 11a–11g which possessed either a 4-FPh or 2-thienyl moiety in the 3-position of the oxadiazole core. Quite unexpectedly, when the 3-position was substituted with a 2-pyridyl moiety 12a–12d, 12f and 12g, a new series of potent mGluR5 positive allosteric modulators resulted, which lacked the intrinsic agonist activity of 10a–10f, 10j and 11a-g, and afforded 14-to 27.9-fold shifts – the largest ever observed for mGluR5. Moreover, analogues 12a–12d, 12f and 12g could form HCl salts which displayed improved solubility and physiochemical properties. Quite unexpectedly, we identified one ADX-47273 analogue 12k, demonstrated a ‘switch’ in mode of pharmacology to a negative allosteric modulator – an observation previously reserved to MPEP-like scaffolds. Finally, we discovered that the (S)-enantiomer for analogues 10–12 is required for mGluR5 activation, and represents the first example of enantioselective potentiation. With these new tools, we are poised to evaluate, in vivo, the effects of pure mGluR5 potentiation versus ago-potentiation of mGluR5 in preclinical antipsychotic and cognition models. These experiments are in progress and will be reported in due course.
Full experimental details for representative mGluR5 PAMs, general details for analog synthesis and full experimental details for the in vitro assays are available in the Experimental Section.
The authors warmly thank the NIH, NIMH, Stanley Medical Research Institute and Vanderbilt University Medical Center for support of our programs in Drug Discovery.