|Home | About | Journals | Submit | Contact Us | Français|
This letter describes the re-exploration of the mGlu1 PAM Ro 07-11401 scaffold through a multi-dimensional, iterative parallel synthesis approach. Unlike recent series of mGlu1 PAMs with robust SAR, the SAR around the Ro 07-11401 structure was incredibly steep (only ~6 of 200 analogs displayed mGlu1 PAM activity), and reminiscent of the CPPHA mGlu5 PAM scaffold. Despite the steep SAR, two new thiazole derivatives were discovered with improved in vitro DMPK profiles and ~3- to 4-fold improvement in CNS exposure (Kps 1.01 to 1.19); albeit, with a ~3-fold diminution in mGlu1 PAM potency, yet comparable efficacy (~5-fold leftward shift of the glutamate concentration-response curve at 10 μM). Thus, this effort has provided additional CNS penetrant mGlu1 PAM tools in a different chemotype than the VU0486321 scaffold. These compounds will permit a better understanding of the pharmacology and therapeutic potential of selective mGlu1 activation, while highlighting the steep SAR challenges that can often be encountered in GPCR allosteric modulator discovery.
Efforts towards the development of positive allosteric modulators (PAMs) of the metabotropic glutamate receptor subtype 1 (mGlu1) were pioneered by Knoflach and co-workers at Roche, resulting in 1-4 (Figure 1).1–3 These small molecule PAMs, coupled with data generated with negative alosteric modualtors (NAMs) of mGlu1,4,5 highlighted issues with species differences due to a single amino acid in rat versus human mGlu1,6,7 and thus 4, a PAM active on both human and rat mGlu1, emerged as a valuable tool compound, despite modest CNS penetration (Kp = 0.29 and high protein binding (fu< 0.01). For over a decade, 4 was the only in vivo tool compound to study selective mGlu1 activation.8–11 Based on recent genetic data implicating GRM1 in schizophrenia,12–14 coupled with data showing that the adverse effect liabilities of group I metabotropic glutamate receptors (mGluRs) are mediated by mGlu5 and not mGlu1,15 our lab has launched a program to develop the next generation of mGlu1 PAMs.14–18 In the past year, we have reported on the discovery and optimization of novel mGlu1 PAMs 5-7 with improved potency (EC50s <20 nM), DMPK profiles (fus >2.0% unbound) and CNS penetration (Kps >1) to afford new avenues for target validation and to assess the therapeutic potential of selective mGlu1 activation.14–18
Previously, revisiting the mGlu4 PAM (−)-PHCCC scaffold led to the discovery of improved tool compounds.19 Therefore, over a decade after its discovery, we felt it was prudent to revisit the Ro 07-11401 scaffold in an effort to develop an in vivo tool compound within this series with improved disposition to account for any chemotype or ligand-biased pharmacology and expand the repertoire of validation tools for mGlu1.
In our functional assays, Ro 07-11401 (4) was an equipotent mGlu1 PAM on both rat (EC50 = 276.5 nM, pEC50 = 6.56±0.08, 109±3% Glu Max) and human (EC50 = 246.0 nM, pEC50 = 6.61±0.08, 96±3% Glu Max), but with a modest disposition profile (vide infra).14–18 Thus, we pursued a mutlidimensional optimization plan (Figure 2) to explore SAR around 4 in an attempt to improve disposition by identifying replacements for the weakly basic oxadiazole (effectively neutral with the pendant CF3 moiety) with analogs 9 and the lipophilic 9H-xanthene moiety in analogs 8. Once idenitifed, optimal moieties would then be combined.
The synthesis was straighforward (Scheme 1). A variety of commercially available 2-amino-4-substiuted oxazoles 10 were coupled under HATU conditions with a diverse array of carboxylic acids to provide analogs 8 in yields ranging from 18–54%. Similarly, the 9H-xanthene-9-carboxylic acid 11 was coupled under HATU conditions to a variety of 5- and 6-membered heterocycic amines to deliver analogs 9 in 11–94% yields. For both series, many of the amines and acids coupled poorly due to a combination of steric and stereoelectronic effects, as well as a range of poor solubilty.
In all, over 200 analogs of 8 and 9 were synthesized and triaged via a 10 μM single point screen on human mGlu1, using an EC20 concentration of glutamate, prior to full concentration-response cruves (CRCs) on both human and rat mGlu1. Surprisingly, all analogs 8 (Figure 3) were uniformly inactive mGlu1 PAMs (no potentiation of an EC20 of glutamate at a concentration of 10 μM), indicating that the 9H-xanthene was a critical pharmacophore. The SAR was remarkably steep, and reminiscent of the steep SAR encountered with the non-MPEP, mGlu5 PAM CPPHA,20 wherein virtually any modification led to a complete loss of PAM activity. Obviously, these data cast doubt on the success of the campaign with analogs 9, wherein the 9H-xanthene amide was held constant. While SAR once again was steep, active analogs did result (Table 1); however, functionalized pyrazoles, oxazoles, oxadiazoles, thiophenes, piperidines, azetidines, cycloalkyl, a structurally diverse array of tertiary amides and functionalized aryl/benzyl derivatives were also devoid of mGlu1 PAM activity (inactive up to 10 μM). Only a sub-set of thiazole amides proved to be active as mGlu1 PAMs. Here, the 2-thiazole regioisomer 9c (human EC50 = 1.4 μM, pEC50 = 5.85±0.07, 75±3% Glu Max and rat EC50 = 3.6 μM, pEC50 = 5.43±0.17, 111±7% Glu Max) proved active on both human and rat mGlu1, while the regiosiomeric 5-thiazole congener 9e was a weak PAM (EC50>10 μM). Derivatives of 9c with substitution in the 4-position, such as the 4-CN 9a and the 4-iPr 9b congeners, displayed submicromolar activity on human (mGlu1 EC50 = 0.711 μM, pEC50 = 6.1±0.07, 110±6% Glu Max and mGlu1 EC50 = 0.90 μM, pEC50 = 6.04±0.07, 89±3% Glu Max, for 9a and 9b, respectively) and low micromolar potency on rat (mGlu1 EC50 = 1.8 μM, pEC50 = 5.73±0.13, 124±8% Glu Max and mGlu1 EC50 = 2.4 μM, pEC50 =5.62±0.11, 111±7% Glu Max, for 9a and 9b, respectively). Replacement of the thiazole 9e for the benzothiazole 9f caused almost a 3-fold increase in potency, whereas a 4,5-diMe thiazole 9g displayed weak PAM activity. Finally, a 2-pyridyl analog (9h) showed moderate activity as an mGlu1 PAM. In total, we had synthesized and assayed ~200 analogs based on 8 and 9, and only eight compounds (4% hit rate) displayed significant mGlu1 PAM activity, highlighting why 4 was advanced early on as an in vivo tool in this series. Moreover, no ‘molecular switches’ were noted,21 as compounds were either PAMs or inactive.
Having identified 9a and 9b, more basic thiazole analogs of 4, we decided to further profile them, despite the diminution in PAM potency. We assessed their ability to leftward shift a glutamate CRC (an assessment of efficacy), to gauge their ability to render glutamate a more potent agonist (Figure 4). All three compounds induced a comparable, robust leftward shift (4.6 to 6.9-fold) of the glutamate CRC at a 10 μM concentration.
Then, we assessed the in vitro DMPK profiles of 4, 9a and 9b, as well as determined CNS penetration (Table 2).15 The prototypical PAM in this series, 4, displayed high intrinsic clearance (at hepatic blood flow) in both rat and human microsomes, very high protein and brain homogenate binding, a mixed CYP profile (with potent, 600 nM inhibition of 2C9) and modest CNS penetration (Kp = 0.29). PAM 9b looked identical across all of the assays, except for CNS penetration, where the Kp was 1.01 (~3-fold improvement over 4). In contrast, PAM 9a displayed an improved overall profile, except for CYPs, where as 9b showed an improved CYP profile. In general, inhibition of CYP 2C9 appears to be an issue with this chemotype. Human microsomal intrinsic clearance was reduced by ~50%, predicting a moderate hepatic clearance (CLhep = 9.85 mL/min/kg), and there was measureable free drug in human plasma (Fu = 0.027) and rat brain homogenate (Fu = 0.017). Importantly, CNS penetration was also improved for 9a, affording a Kp of 1.19 (an ~4-fold increase compared to 4). While rat intrinsic clearance did not improve beyond that of 4, the enhanced free drug in brain (>10-fold), coupled with diminished peripheral exposure, provide advantages for this chemical class of mGlu1 PAMs as CNS tool compounds. Thus, 9a emerged as an improved mGlu1 PAM tool compound within the Ro 07-11401 (4) series; however, the ideal in vivo tool compound from this scaffold remains elusive due to steep SAR.
In conclusion, we revisited the Ro 07-11401 (4) series of mGlu1 PAMs in an effort to develop an improved in vivo tool compound within this scaffold. Unlike the recent VU0486321 series of mGlu1 PAMs which exhibited robust SAR, the SAR of the Ro 07-11401 series was incredibly steep (only ~8 of 200 analogs displayed mGlu1 PAM activity), and reminiscent of the CPPHA mGlu5 PAM scaffold. Despite the steep SAR, two new thiazole derivatives (9a, VU6000799, and 9b, VU6000790) were discovered with improved in vitro DMPK profiles and ~3- to 4-fold improvement in CNS exposure (Kps 1.01 to 1.19), albeit an ~3-fold diminution is mGlu1 PAM potency, yet comparable efficacy (~5 left fold shift of a glutamate CRC at 10 μM). A modeling effort is underway to provide further insight to direct future chemical optimization of this scaffold. Studies are ongoing and will be reported in due course.
We thank William K. Warren, Jr. and the William K. Warren Foundation who funded the William K. Warren, Jr. Chair in Medicine (to C.W.L.). P.M.G. would like to acknowledge the VISP program for its support. This work was funded by the William K. Warren, Jr. Chair in Medicine and the NIH (U54MH084659).
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.