Synthesis and pharmacological evaluation of bivalent tethered ligands to target the mGlu_2/4 heterodimeric receptor results in a compound with mGlu_2/2 homodimer selectivity

Abstract

This Letter details our ongoing efforts to develop selective positive allosteric modulators (PAMs) of the mGlu_2/4 heterodimeric receptor that exists in the CNS and may represent a novel drug target to modulate the glutamatergic system. As multiple hit-to-lead campaigns from HTS hits failed to produce selective small molecule mGlu_2/4 heterodimer PAMs, we were inspired by the work of Portoghese to synthesize and evaluate a set of nine bivalent tethered ligands (possessing an mGlu₂ PAM at one terminus and an mGlu₄ PAM at the other). Utilizing G protein-Inwardly Rectifying Potassium (GIRK) channel functional assays, we found that the tethered ligands displayed PAM activity in a cell line co-expressing both mGlu₂ and mGlu₄ but also in cells expressing mGlu₂ or mGlu₄ alone. In a CODA-RET assay, one of the tethered ligands potentiated mGlu_2/4 heterodimers; however, another compound displayed 75-fold preference for the mGlu_2/2 homodimer over heterodimeric mGlu_2/4 or homomeric mGlu_4/4. This work highlights the developed of mGlu receptor PAMs with homodimer/heterodimer preference and expands the potential for PAMs as tethered ligands beyond the more classical antagonists and NAMs.

Graphical Abstract

With the discovery that numerous Class C G protein-coupled receptors (GPCRs) form functional heterodimers with unique pharmacology beyond that of the canonical homodimer signaling complexes,^1,2 the need for selective ligands to differentiate between homomeric and heteromeric receptor isoforms is paramount. It is now known that the eight metabotropic glutamate (mGlu) receptors can form heterodimers; of note, the mGlu_2/4 heterodimer receptor has been validated in multiple labs^1,3,4 and shown to exist in native neuronal tissues.^5,6 Recently, we described mGlu₄ positive allosteric modulators (PAMs) that also potentiate the mGlu_2/4 heterodimer, as well as compounds that are mGlu₄ homodimer-selective.^4,5 Ligands that only potentiate the mGlu₄ homodimer display robust anti-Parkinsonian activity.⁴ Here, we describe our efforts to develop selective mGlu_2/4 PAMs as tool compounds to further explore the biology of the mGlu_2/4 heterodimer.

Our studies began with modification of a PAM known to bind to mGlu₄ and potentiate the activity of both mGlu₄ homodimers and mGlu_2/4 heterodimers (e.g., VU0155041⁵, 1, Fig. 1); these efforts were unsuccessful in generating selective compounds. To identify new leads, we performed a functional high-throughput screen (HTS) employing an mGlu_2/4 HEK293 cell line expressing G protein-Inwardly Rectifying Potassium (GIRK) channels and thallium flux readout to assess GPCR activation⁷. This HTS afforded a number of diverse chemotype hits (Fig. 1), distinct from our prototypical mGlu_2/4 PAM 1 (VU0155041). We next triaged hits in a cell line expressing GIRK channels but no mGlu receptor, which identified numerous hits, such as 2 and 3, as GIRK activators.⁸ Hit- to-lead efforts around the remaining HTS hits indicated that all of the ligands were active in cells co-expressing mGlu₂ and mGlu₄ and possessed favorable DMPK properties, but these ligands were also active when either mGlu₂ (4–6) or mGlu₄ (7, 8) was expressed alone; no selective mGlu_2/4 heterodimer PAMs were identified.⁸

Figure 1.

Chemotypes optimized from an mGlu_2/4 heterodimer PAM HTS 1–8. All proved to be either GIRK activators (2,3) or compounds that had activity when either mGlu₂ (4–6) or mGlu₄ (1,7,8) was expressed alone. No selective mGlu_2/4 heterodimer PAMs were developed.

While our small molecule approach proved unsuccessful, based on the success of numerous labs with a tethered ligand approach for selective modulation of heterodimers (employing tethered antagonists or NAMs),^9–13 we elected to pursue this strategy for the mGlu_2/4 heterodimer. Tethered ligands rely on a phenomenon known as avidity, the additive effect of multiple affinities. One molecule binding to half of the receptor results in a local environment in which the other ligand has a very high concentration. This then results in a high probability of activation of both monomers simultaneously. Our tethering ligand strategy employed PAMs for both mGlu₂ and mGlu₄. With an enormous wealth of chemically diverse mGlu₂ and mGlu₄ PAMs, the appropriate selection of PAMs with the desired avidity (i.e., the accumulation of multiple affinities) was essential.^14,15 Pin and co-workers demonstrated that the mGlu₄ PAM VU0415374 (9) and the classical mGlu₂ PAM BINA (10) exhibited cooperativity when co-administered in a mGlu_2/4 TR-FRET assay⁶; therefore, our tethering efforts were derived from these PAMs (Fig. 2).

Figure 2.

Previous work with a TR-FRET assay demonstrated that mGlu₄ PAM 9 and mGlu₂ PAM 10 exhibited cooperativity when co-administered in an mGlu_2/4 heterodimer cell line, suggesting these as PAMs for a tethered ligand approach.

We explored historical SAR to determine where the PAM ligands 9 and 10 could be modified for tether incorporation (Fig. 3). For the mGlu₄ PAM 9, only the 4-position of the Eastern phenyl ring was tolerant, and for 10, mGlu₂ PAMs 11–13 indicated that a substitution as in 13 would allow for viable tethering. Thus, our target mGlu₂ PAM- mGlu₄ PAM tethered ligands would be based on 14.

Figure 3.

Determining linkage points for a tether on PAMs 9 and 10, and envisioned tethered ligands 14 to activate the mGlu_2/4 heterodimer receptor.

The route to access the requisite mGlu₄ PAM tether component 18 is shown in Scheme 1. Starting from aniline 15, N-Boc protection, nitro reduction, acylation with picolinoyl chloride and Boc deprotection affords aniline 16. Then, acylation with 3-chloro-4-nitrobenzoyl chloride and iron-catalyzed nitro reduction provides 17 in 80% yield for the two steps. Finally, reaction with 1,4-dioxane-2,6-dione delivers carboxylic acid 18.

Scheme 1.

Synthesis of the mGlu₄ PAM tether component 18^a

^aReagents and conditions: (a) Boc₂O, DMAP, THF, rt; (b) Pd/C, H₂, EtOH, rt; (c) picolinoyl chloride, DIEA, DCE, 80 °C, 5 hr; TFA, DCM, rt, ~10% over 4 steps; (e) 3-chloro-4-nitrobenzoyl chloride, DIEA, DCM, rt, 16 hr; (f) Fe⁰, AcOH, EtOH, 80 °C, 16 hr, 80% over 2 steps; (g) l,4-dioxane-2,6-dione, THF, 60 °C, 16 hr, 16%.

The route to access the requisite mGlu₂ PAM tether component 25 is shown in Scheme 2 and is based on established literature precedent.¹⁶ First, Friedel-Crafts acylation with the acyl chloride 19 the desired adduct 20 in 85% yield. A Mannich reaction with 20, dimethylamine hydrochloride and para-formaldehyde in acetic acid followed by dilution in DMF and reflux gave the enone 21 in 96% yield. Then, a Nazarov cyclization in concentrated sulfuric acid resulted in 22 in 78% yield. Next, demethylation of the methoxy group gave phenol 23, which was taken on crude into an S_N2 reaction with 3-pinacolboranebenzyl bromide yielding 24. A subsequent Suzuki coupling between 24 and methyl 5-bromosalicylate, followed by alkylation of the phenol with tert-butyl bromoacetate gave the desired adduct in 65% yield over the four steps. Finally, selective deprotection of the tert-butylcarboxylate with TFA yielded key mGlu₂ PAM tether 25 in 90% yield.

Scheme 2.

Synthesis of the mGlu₂ PAM tether component 25^a

^aReagents and conditions: (a) 1-methoxy-2,3-dimethylbenzene, A1C1₃, DCM, rt, 85%; (b) formaldehyde, dimethylamine hydrochloride, DMF, 80 °C, 16 hr, 96%; (c) H₂SO₄, rt, 16 hr, 78%; (d) AlCl₃, PhMe, 110 °C, 1 hr; (e) 2(3-(bromomethyl)phenyl)-4,4,5,5-tetramethyl-l,3,2-bioxoborolane, K₂CO₃, rt, 6 hr, 73% over 2 steps; (f) methyl 5-bromo-2hydroxybenzoate, Pd(dppf)₂Cl₂, Na₂CO₃, MeCN:H₂O, 100 °C, 1 hr; (g) (i) tert-butyl bromoacetate, Cs₂CO₃, DMF, rt, 16 hr, (ii) TFA, DCM, rt, 1hr, 58% over 2 steps.

The tethered ligands 14 were prepared by a HATU mediated coupling between 25 and a series of aliphatic mono-N-Boc alkyl diamines (2- to 12-carbon chain) to generate the amide linkage. Hydrolysis of the methyl ester to the acid and TFA deprotection of the Boc group delivered amines 26a-i. Nine additional HATU-mediated couplings between 26a-i and the mGlu₄ PAM tether component 18 affords mGlu₂/mGlu₄ PAM tethered ligands 14a-i.

With the tethered ligands 14a-i in hand, we assessed their ability to potentiate responses in cells co-expressing mGlu₂ and mGlu₄ in our HEK293 cell line/thallium flux assay (Table 1). Gratifyingly, the tethered ligands proved to be potent PAMs in cells co-expressing both mGlu₂ and mGlu₄, with EC₅₀s ranging from 90 nM (90% Glu Max) to 2 μM (71% Glu Max). The 3, 4 and 5-carbon spacers (14b-d) were the most potent PAMs, while the 12-carbon spacer (14i) was the least potent. Ligands 14b-f, with a 3- to 7-carbon tether, exhibited decreases Hill slopes (HS, Table 1, example in Fig. 4), which was thought to be potentially indicative of dimeric interaction or which could represent overlapping activity at homodimers. Interestingly, and to our surprise, the majority of these large, tethered PAMs 14a-i were similarly potent (EC₅₀s of 33 nM to 1 μM) and efficacious (80 to 114% Glu Max) PAMs in cells expressing only mGlu₂ or mGlu₄, although the longer linker compounds 14g-i exhibited decreased activity in cells expressing just mGlu₄. It is rather remarkable that such large molecules can gain access to a PAM binding site in the receptor transmembrane domain and that the mGlu₂ and mGlu₄ PAM selective moieties can bind and individually potentiate the homodimeric receptors.

Table 1.

Structures and PAM activities of tethered ligands 14.

Cmpd	n	mGlu2/4 EC50 (nM)a [% Glu max], HS	mGlu2 EC50(nM)a [% Glu Max]	mGlu4 EC50 (nM)a [% Glu max]
9		2760 [42.0], 1.2	Inactive	34 [94.4]
13		300 [64.5], 2.4	84.3 [88.2]	Inactive
1:1 Mix 9:13		33 [81.6], 0.74	9.3 [95.4]	70 [114.1]
14a	2	463 [70], 1.2	281 [92]	657 [104]
14b	3	90 [90], 0.55	54 [100]	38 [114]
14c	4	150 [88], 0.44	98 [98]	33 [113]
14d	5	128 [89], 0.46	92 [102]	35 [110]
14e	6	680 [101], 0.34	62 [102]	32 [106]
14f	7	524 [86], 0.75	122 [101]	30 [85]
14g	8	371 [83], 1.2	94 [97]	32 [52]
14h	10	350 [86], 1.4	134 [102]	> 10,000 [41]
14i	12	2,070 [71], 1.3	1,180 [102]	> 10,000 [45]

^aThallium mobilization assays with HEK293/GIRK cells co-expressing rat mGlu₂ and mGlu₄, mGlu₂ alone, or mGlu₄ alone performed in the presence of an EC₂₀ fixed concentration of glutamate; values represent N=3 experiments performed in triplicate for linked compounds in cells expressing mGlu_2/4 and N=1 experiment performed in triplicate for cells expressing mGlu₂ or mGlu₄ alone or for 9, 13 and a 1:1 mixture of 9:13. EC₅₀=potency, nM=nanomolar, %Glu Max=maximal response relative to a saturating concentration of glutamate, HS=Hill slope.

Figure 4.

Examples of concentration-response curves of tethered ligands, with 14d (red) generating a Hill slope of 0.46 versus 14h, (red, 1.4).

Based upon these data, we evaluated two tethered ligands (14d and 14h) employing CODA-RET, a technique that allows for the isolation of signaling exclusively from a defined receptor complex (Fig. 5). These studies used a 5 μM concentration of each PAM along with a concentration range of agonists that are selective for either mGlu₂ or mGlu₄, or a combination of the two. This strategy was undertaken as mechanistic studies performed during the validation of the CODA-RET assay showed that co-application of both agonists, which occupies both protomers, elicits the same response as glutamate. The CODA-RET assay relies upon complementation and Bioluminescence Resonance Energy Transfer (BRET) and allows the detection of a signal induced only by a defined mGlu_2/4 heteromer without contribution from homodimers that may be present¹⁷. In the CODA-RET experiments presented here, different subtypes of receptor (in this case mGlu₂ and mGlu₄) are labeled with N- or C-terminal fragments of a luminescent donor molecule, rLuc8, respectively.^4,17 When two different receptor subunits form a heterodimer, these two halves complement, forming a functional luciferase molecule. In contrast, in a homodimer composed of mGlu_4/4 or mGlu_2/2, only one half of the donor molecule would be present and a luminescent signal would not be generated in response to substrate. In parallel, a defined G protein, in this case G_αi, is labeled with the acceptor molecule, mVenus. Activation of the complemented luminescent heterodimeric receptor then leads to recruitment of this acceptor-labeled G_αi protein; the resulting proximity of the complemented donor and the acceptor molecule leads to BRET.^4,17 We assessed the ability of these compounds to potentiate mGlu_4/4 homodimers, mGlu_2/2 homodimers, and mGlu_2/4 heterodimers using either an mGlu₂ agonist (DCG-IV) or an mGlu₄ agonist (L-AP4). In these studies, 14d displayed PAM pharmacology at both the mGlu_2/2 (Fig. 5A) and mGlu_4/4 (Fig. 5B) homodimeric receptors, whereas 14h only potentiated homodimeric mGlu_2/2. Neither of the tethered ligands 14 potentiated the mGlu_2/4 heterodimer in the CODA-RET assay when an agonist for either receptor was used independently (Fig. 5C, D). However, when both L-AP4 and DCG-IV were added together, 14d potentiated the response, but 14h did not. As 14d is active at both mGlu_2/2 and mGlu_4/4, the most likely explanation for this finding is that the configuration of 14d allows one molecule to occupy each promoter, regardless of hetero or homodimerization. PAM activity, however, requires that both promoters be engaged by an orthosteric agonist. This also suggests that the mGlu_2/4 heterodimer is conformationally distinct from the mGlu_2/2 homodimeric receptor as the ability of 14h to potentiate mGlu_2/4 responses is lost. Therefore, while these efforts did not identify a compound selective for the mGlu_2/4 heterodimer, they do indicate that, below a 5 μM concentration, 14h is a PAM that is selective for mGlu_2/2 homodimers versus mGlu_2/4 heterodimers, providing a potentially useful tool to tease apart the functions of these receptor combinations in more complex systems, such as electrophysiological examination of receptor activity in the CNS.

Figure 5.

CODA-RET profiles of tethered ligands. A. Both 14d (VU6023800) and 14h (VU6023804) potentiate mGlu₂ homodimers using the mGlu₂ agonist DGC-IV. B. 14d acts as a PAM at mGlu₄ homodimers using the mGlu₄ agonist L-AP4, while 14h does not. C,D. None of the tested compounds show significant efficacy versus vehicle when mGlu_2/4 heterodimers are complemented using either DCG-IV (C) or L-AP4 (D). E. When DCG-IV and L-AP4 are combined, 14d can potentiate the mGlu_2/4 heterodimer, but 14h cannot.

Clearly, the discovery and development of selective mGlu_2/4 PAMs is a major challenge that has yet to be overcome with either a multitude of small molecule chemotypes or a tethered ligand approach employing mGlu₂ and mGlu₄ PAMs. However, this exercise has now identified a m new molecule, 14d (VU6023804) with preference for mGlu_2/2 homodimers over mGlu_4/4 or mGlu_2/4 receptors and has demonstrated unique pharmacological profiles of large, flexible tethered ligands that are still able to access receptor binding sites. Additional studies with the mGlu_2/4 heterodimer are underway and will be reported in due course.

Scheme 3.

Synthesis of the mGlu₂ PAM-mGlu₄ PAM tethered ligands 14^a

^aReagents and conditions: (a) H₂N-(CH₂)_nNHBoc, HATU, DIEA, DMF, rt, 16 hr; (b) LiOH, THF/H2O, rt, 16hr; (c) HCl, dioxanes, rt, 16 hr; (d) 18, HATU, DIEA, DMF, rt, 16 hr, 2–12% over 4 steps.