Small Molecule Inhibitors of Regulators of G Protein Signaling (RGS) Proteins

G protein-coupled receptors (GPCRs) are widely distributed throughout the body and are an important class of therapeutic targets for drug discovery.¹ The majority of GPCR-targeted drugs act as orthosteric agonists or antagonists at the canonical ligand binding site, but recent attention has focused on drugs acting allosterically to modulate receptor activity.² One advantage of allosteric agents is their ability to modulate ongoing physiological signaling. Other approaches to modulate GPCR-mediated signal transduction may also provide significant therapeutic benefit. To this end, we have focused on targeting a class of proteins that negatively regulates GPCR signaling.

Regulators of G protein signaling (RGS) proteins are potent negative modulators of GPCR signaling. They accelerate the rate of GTP hydrolysis by Gα subunits of heterotrimeric G proteins, shortening the duration and decreasing the magnitude of signal after receptor activation.³ The important role of RGS proteins in GPCR signaling has generated, in recent years, significant interest in RGS proteins as therapeutic targets in their own right.^4,5 One of the attractions is that while many receptors are expressed throughout the body, modulators of the signaling pathway, including RGS proteins, are expressed in a more tissue-specific manner. This may be of particular importance for the development of centrally acting agonist therapeutics. Of over 20 mammalian RGS proteins identified, RGS4 is one of the most extensively characterized. It is broadly, but heterogeneously, expressed in the central nervous system (CNS),^6,7 less so in peripheral tissues, and has been shown to modulate activity of the μ opioid (MOP), δ opioid (DOP), and M3 muscarinic receptors among others.^8,9 The selectivity in regional distribution, coupled with the finding that RGS4 can selectively suppress signaling through DOP receptors as compared to MOP receptors,¹⁰ suggests that a level of specificity for RGS4 inhibitory action will be possible.

The RGS-Gα site is a large, relatively featureless protein–protein interaction (PPI) interface.¹¹ Inhibitors of RGS proteins must act by disrupting this interaction either directly or through an allosteric mechanism. While a number of PPI inhibitors are known against a variety of targets, most of these compounds are large molecules that lack access to the CNS.⁵ The development of cell- and blood–brain barrier-permeable small molecules that can act as selective PPI inhibitors is a nontrivial task but one that could ultimately provide significant clinical benefit.

Our group recently discovered small molecule inhibitors of RGS4¹²⁻¹⁵ using a high-throughput flow cytometry protein interaction assay (FCPIA).¹⁶ As part of this screening process, CCG-50014 (1a, Figure 1) was identified as a selective inhibitor of RGS4 that acted by forming a covalent adduct to cysteine residues in the RGS protein.^15,17 With an IC₅₀ of 30 nM, it is the most potent RGS inhibitor reported to date. In an attempt to further define the structural requirements for high-potency inhibition of this protein, analogues of 1a have been synthesized with variation in both the N2 and the N4 side chains. In addition, a set of these newly synthesized RGS4 inhibitors have been evaluated for their effects on calcium mobilization, an off-target activity displayed by 1a.

The synthesis of CCG-50014 (1a) and its thiadiazolidinone (TDZD) analogues is shown in Scheme 1. Commercially available isothiocyanates were reacted with isocyanates in the presence of sulfuryl chloride (Scheme 1).¹⁸ This allowed a range of R¹ and R² substituents, having varying lipophilic, electronic, and steric properties, to be evaluated. While chlorine gas¹⁹ or N-chlorosuccinimide²⁰ are also used in literature procedures for making TDZDs, we found the use of sulfuryl chloride straightforward and consistent. The resulting S-chloroisothiocarbamoyl chloride, proposed by Slomczyńska and Barany,¹⁸ was subsequently oxidized in atmospheric oxygen to the desired products. This reaction was easily carried out in parallel, with typically 11 different reactions running simultaneously. In total, 75 TDZD analogues were synthesized, with 39 (1a–i, 2a–i, 3–8, 9a,b, 10a–d, 11a–d, 12a–d, and 13) reported in Table 1. Data for all compounds are reported in the Supporting Information.

In addition to the large series of TDZD compounds synthesized, an imidazolidine-2,4-dione (14) and a maleimide (15) were both prepared to compare their activity to the TDZDs. The synthesis of 14 was carried out in three steps, starting from ethyl bromoacetate and p-toluidine (Scheme 2a). After reaction in the presence of sodium acetate, the resulting amino acetate 16 was stirred at reflux in the presence of p-methylbenzyl isocyanate to provide the corresponding ureido acetate 17. This was cyclized using sodium hydride, providing 14. Compound 15 was synthesized by first reacting bromomaleic anhydride with benzyl amine in the presence of acetic acid. The desired product resulted from a Suzuki reaction coupling p-tolyl boronic acid to 18.

All compounds were evaluated using the FCPIA assay as a primary screen to determine IC₅₀ values for inhibition of Gα_o binding to both RGS4 and RGS8 (the closest relative to RGS4 based upon sequence homology). In preliminary studies, 1a was also found to suppress Ca²⁺ responses to GPCRs in a manner unrelated to its activity at RGS proteins. The most interesting of the new ligands was also assessed for this off-target effect.

CCG-50014 (1a) was confirmed as a potent inhibitor of RGS4 with excellent selectivity over RGS8. Retaining the 4-fluorobenzyl R¹ group and varying R² led to both more (e.g., 1d) and less (e.g., 1g) potent and selective compounds. It appeared that a 3-substituent on the R² aryl ring was associated with reduced RGS4 potency as compared to unsubstituted and 4-substituted analogues (e.g., 1f, 1g, 1h cf. 1b, 1c, 1d). While not completely consistent, this trend is repeated across a number of the series where R¹ is held constant and R² is varied. A 3,4-dichlorophenyl group as R² generally resulted in low potency at RGS4 and relatively low selectivity (e.g., 1e, 2e). In contrast, a 4-methyl substituent was more often associated with high affinity and high selectivity at RGS4, with a number displaying >1000-fold selectivity versus RGS8 (e.g., 3, 4, 5, and 7). Replacement of the phenyl group by benzyl at R² (1i, 2i) did not improve activity at RGS4 but did reduce RGS8 activity, resulting in each compound having near 3 orders of magnitude selectivity. In fact, of the compounds discussed so far, that is, retaining a benzyl or substituted benzyl at R¹, 1i and 2i were the most selective.

Variation in the aryl groups of R¹ and R² has therefore led to the discovery of a number of ligands with high potency and excellent selectivity. However, the uniformly high lipophilicity (Clog P typically >4) of these ligands resulted in only moderate solubility in aqueous solution, and they were therefore less than ideal for consideration for more in-depth study. To address this problem, analogues in which one or both R groups were replaced with short alkyl chains were prepared. In the former series, where one aryl group was replaced by alkyl (9a,b, 10a, 11a, and 12a), potency and selectivity at RGS4 were retained. Making both R groups short alkyl chains (10b–d, 11b–d, and 12b–d) substantially improved solubility (complete solubility at 500 μM) while also providing the most consistently selective group of compounds yet developed (all >1000-fold selective). The potency at RGS4 (IC₅₀ 14.4 nM), near 6000-fold selectivity, and high solubility of 11b mean that it is an ideal candidate for further evaluation, including in vivo studies. As a means to even further enhance solubility of this compound, analogues containing ether side chains were considered, and the ether analogue of 11b was prepared. This compound (13) retained good potency (56 nM) and excellent selectivity (>600-fold).

The effects of 1a, 11b, and 13 were tested on the Ca²⁺ transient induced by M3 muscarinic receptors in HEK293T cells. Compound 1a at 10 μM nearly completely abolished the carbachol-induced Ca²⁺ transient (Figure 2), while 11b and 13 had no effect. The action of 1a on this response cannot be through effects on RGS proteins since HEK cells express minimal levels of functional RGS proteins.²¹

We have previously published our studies that indicate that the lead compound (1a) reacts to form an adduct with a cysteine residue on the RGS protein through disulfide bond formation.¹⁵ The proposed mechanism (Scheme 3, 19b) is analogous to that proposed by Nasim and Crooks for the ring-opening of TDZDs with PPh₃.²⁰ To help confirm the importance of disulfide bond formation to the activity of this series of ligands, analogues 14 and 15 were prepared. Compound 14 is the imidazolidine-2,4-dione analogue of 4, while 15 is the maleimide analogue of 2i; 4 and 2i being two of the most potent inhibitors discovered. As expected, neither 14 or 15 displayed activity at RGS4. Also supporting the disulfide bond-forming mechanism, the reaction of propane thiol with 1a appears to give efficiently and cleanly the expected adduct 19a (Scheme 3). Importantly, 1a is not a general cysteine alkylator, failing to inhibit the cysteine protease papain, suggesting selectivity for RGS4.¹⁵

Previously, thiadiazolidine-3,5-diones have been reported as having a number of biological effects,²²⁻²⁴ including being glycogen synthase kinase 3β (GSK-3β) inhibitors with activities in the micromolar range.¹⁹ This latter activity has been suggested to account, at least in part, for the antidepressant-like effects in mice of the TDZD NP031115.²⁵ Interestingly, 11b was evaluated as part of that study and was found to be one of the weaker inhibitors (GSK-3β IC₅₀ 70 μM) meaning that it has significant selectivity (almost 5000-fold) for RGS4 over GSK-3β. As such, 11b should prove to be an invaluable tool in defining the physiological role of RGS4 in vivo, including a potential role in 5-HT1A-mediated antidepressant effects.²⁶

In summary, a series of RGS4 inhibitors have been synthesized with improved selectivity over RGS8 and lacking the off-target calcium mobilization activity of the lead 1a. One compound, 11b, combines potency (RGS4 IC₅₀ 14 nM) and selectivity (5800-fold over RGS8 and no calcium transient) with excellent aqueous solubility and should prove an invaluable tool for better defining the role of RGS4 and its potential as a therapeutic target. Its ether analogue (13) had further improved solubility while retaining good potency and selectivity. Analogues 11b and 13 are now being evaluated in vivo with positive preliminary data, and the results of this latter work will be reported separately.