Discovery of a Novel Class of ERRα Agonists

Abstract

A novel class of estrogen-related receptor α (ERRα) agonists has been discovered. A structure–activity relationship study of high-throughput screening hits 1 and 2 led to the discovery of benzimidazole 3d (DS20362725) and acetophenone analogue 5c (DS45500853). The X-ray crystal structure of the ERRα ligand-binding domain in complex with 5c and PGC-1α coactivator peptide revealed conformational changes in the ligand-binding pocket to accommodate 5c and the key interaction between the protein and ligand. Since both analogues avoided PPARγ transcriptional activity, they can be useful tool compounds for investigating biological ERRα functions.

Estrogen-related receptors (ERRs) are orphan nuclear receptors with three subtypes: ERRα, ERRβ, and ERRγ. Among them, ERRα is expressed mainly in metabolically active tissues such as muscle and adipose tissue and is known as a transcriptional regulator of energy metabolism.^1,2 It is also known that transcription factor peroxisome proliferator-activated receptor γ (PPARγ) coactivator 1α (PGC-1α) is the endogenous coactivator protein for ERRα² and corepressor receptor-interacting protein 140 (RIP140) is the endogenous corepressor protein for ERRα.³ ERRα knockout mice displayed suppressed weight gain, adipose synthesis, and β-oxidation under high-fat-diet feeding.⁴ In addition, ERRα agonist I (Figure 1) has been reported to improve glucose and fatty acid uptake in C2C12 mouse muscle cells,⁵ whereas ERRα inverse agonist II normalized serum triglyceride and insulin levels in vivo.⁶ As it remains controversial whether ERRα upregulation or downregulation is beneficial for metabolic disorders, including type 2 diabetes mellitus (T2DM), a novel class of ERRα ligands is highly demanded.⁷⁻⁹ It has also been reported that the development of ERRα agonists is challenging because small molecules might not induce the conformational changes needed to activate ERRα.¹⁰ In this Letter, the identification of a novel class of ERRα agonists with structural biological properties is described.

Figure 1.

Chemical structures of ERRα agonist I from the Chinese Academy of Sciences,⁵ ERRα inverse agonist II from Johnson & Johnson Pharmaceutical,⁶ and HTS hits 1 and 2 with binding-inhibitory and reporter activities.

Our high-throughput screening (HTS) strategy was to evaluate the suppression of binding between FITC-RIP140 peptide and the GST-hERRα ligand-binding domain (LBD) in a fluorescence polarization (FP) assay, which led to the discovery of HTS hits 1 and 2. Compound 1 inhibited binding of RIP140 corepressor peptide with IC₅₀ = 0.67 μM, while 2 showed IC₅₀ = 0.78 μM. Accordingly, derivatization of these HTS hits was commenced to acquire potent ERRα agonists.

Our initial efforts were focused on the modification of 1 as presented in Table 1. Since compound 1 is a potent PPARγ agonist,^11,12 the avoidance of PPARγ activity was desired. This led us to synthesize non-thiazolidinedione (non-TZD) derivatives because the TZD moiety is known to enhance PPARγ activity.¹³ As presented in Table 1, the potent PPARγ agonist activity of 1 was confirmed (EC₅₀ = 0.41 nM, E_max = 119%). To our delight, non-TZD derivative 3a maintained its inhibitory potency. Compounds with potent binding activity were evaluated in a full-length ERRα luciferase reporter assay to assess the ERRα agonist activity. HTS hit 1 exhibited micromolar potency in the reporter assay, whereas non-TZD derivative 3a lost the activity. The poor correlation between the activities in the reporter assay and those observed in the coactivator binding assay was previously reported.⁶ With regard to the subtype selectivity, compound 1 did not modulate ERRβ or ERRγ transcriptional activity up to 7.9 μM (less than 10% modulation; see the Supporting Information). The phenolic hydroxyl group is essential for the binding activity, as removal of the hydroxyl group led to the loss of binding potency (compound 3b). These results led us further to modify the 2-substituent of the benzimidazole ring. Phenyl derivative 3c retained binding activity, whereas 3c lost transcriptional activity. The phenyl ring on the right-hand side of the molecule was not required for binding activity because methyl analogue 3d retained binding activity. Methyl analogue 3d displayed the same range of potency as HTS hit 1 in the reporter assay (EC₅₀ = 1.1 μM). By the removal of the aromatic ring of the 2-benzimidazole, compound 3d successfully avoided PPARγ agonist activity.

Table 1. Binding and Transcriptional Activities of Compounds 3.

^aInhibition of the binding between 10 nM F-RIP140 and 1.2 μM GST-ERRα LBD in 10 mM HEPES, 0.10 mM EDTA, 5% bovine gamma globulin, 2.0 mM DTT, and buffer (pH 6.8). Each value represents the mean of at least two independent experiments run in quadruplicate, unless otherwise noted.

^bTranscriptional activity of full-length ERRα in MG63 cells. Mean values of at least two independent experiments run in sextuplicate are shown, unless otherwise noted.

^cTranscriptional activity of full-length PPARγ. Values from single experiments run in quadruplicate are shown.

^dValue from a single experiment.

^eNT = not tested.

^fND = not determined.

We then focused on the modification of HTS hit 2 (Table 2). Although 2 exhibited the same level of binding potency as 1, it displayed poor transcriptional activity in the reporter assay. Since it appeared that the diphenyl ether moiety was essential for the activity, our derivatization strategy involved minor modifications of the X, Y, and Z moieties as indicated in Table 2. Elongation of the acetyl group effectively enhanced binding activity, as compounds 4a and 4b had 2-fold higher activity than 2. However, neither compound displayed transcriptional activity. In a similar manner, although removal of the chloro group from the right-hand side of the molecule maintained the binding activity, compound 4c exerted no transcriptional activity. The chloro group on the naphthyl ring plays a pivotal role in the binding activity, as compound 4d exhibited 5-fold lower activity. The structural similarity between HTS hits 1 and 2 led us to replace the naphthyl ring with a tert-butylphenol group, giving 5a. Modification of 5a was required because the compound showed modest binding activity and almost no transcriptional activity in the reporter assay. Replacement of the acetamide moiety with an acetyl group enhanced both binding and transcriptional activity (compound 5b). A close analogue of 5b also retained binding and transcriptional activity (compound 5c). A reporter assay indicated that 5b and 5c did not possess PPARγ transcriptional activity.

Table 2. Binding and Transcriptional Activities of Compounds 4 and 5.

				ERRα bindinga			ERRα reporterb			PPARγ reporterc
compd	X	Y	Z	IC50 (μM)	Imax (%)	Hill slope	EC50 (μM)	Emax (%)	slope	EC50 (μM)	Emax (%)	slope
2	Cl	Cl	NHAc	0.78	48	1.7	NDe	149	0.90	>10d	–5d	NDe
4a	Cl	Cl	NHCOEt	0.34	53	1.6	>10d	99d	NDe	>10d	12d	NDe
4b	Cl	Cl	NHCOn-Pr	0.35d	49d	2.1	>10d	97d	NDe	>10d	18d	NDe
4c	Cl	H	NHAc	0.42d	58d	1.4	>10d	101d	NDe	>10d	10d	NDe
4d	H	Cl	NHAc	3.8d	45d	1.1	>10d	102d	NDe	>10d	19d	NDe
5a	OH	Cl	NHAc	1.0	38	1.0	>10d	115d	NDe	NDe	62d	NDe
5b	OH	Cl	Ac	0.12d	51d	1.5	5.9d	278d	1.9	>10d	6d	NDe
5c	OH	H	Ac	0.80	40	1.4	5.4	354	1.3	>10d	15d	NDe

^aInhibition of the binding between 10 nM F-RIP140 and 1.2 μM GST-ERRα LBD in 10 mM HEPES, 0.10 mM EDTA, 5% bovine gamma globulin, 2.0 mM DTT, and buffer (pH 6.8). Each value represents the mean of at least two independent experiments run in quadruplicate, unless otherwise noted.

^bTranscriptional activity of full-length ERRα in MG63 cells. Mean values of at least two independent experiments run in sextuplicate are shown, unless otherwise noted.

^cTranscriptional activity of full-length PPARγ. Values from single experiments run in quadruplicate are shown.

^dValue from a single experiment.

^eND = not determined.

To obtain structural information on the interaction of the protein with the ligand responsible for activity, the X-ray crystal structure of the ERRα LBD complexed with compound 5c and PGC-1α coactivator peptide was determined at 2.7 Å resolution, as shown in Figure 2. Two ERRα LBD monomers in an asymmetric unit form a dimer and display the canonical fold of nuclear receptor LBDs. The nomenclature of the secondary structure elements is based on the RXR LBD crystal structure¹⁴ (Figure 2B). The ERRα LBD adopts an agonist conformation with the bound coactivator peptide, as judged by the position of H12.¹⁵ The crystal structure revealed that 5c binds in the same ligand-binding pocket (LBP) of the ERRα LBD as inverse agonist II.^6,16 Inside the LBP, 5c forms extensive hydrophobic interactions with the protein, including 50 van der Waals contacts (Figure 2C). In addition, the phenolic hydroxyl group of 5c has a hydrogen-bonding interaction mediated through a water molecule with the side-chain nitrogen atom of Arg-372 and the main-chain oxygen atom of Phe-382, which are the only hydrophilic interactions between the protein and 5c. This explains why the phenolic hydroxyl group is essential for activity, as described in Table 1 (compound 3b). Because this pocket is lipophilic, only lipophilic substituents were tolerated except for the substituents involved in this hydrogen-bonding network.

Figure 2.

Crystal structure of the ERRα LBD in complex with 5c and PGC-1α coactivator peptide. (A) σA-weighted 2F_o – F_c electron density map around 5c contoured at 1.0σ. The protein (dark blue) and 5c (cyan) are shown as a stick model and a ball-and-stick model, respectively. (B) Schematic representation of the ERRα LBD (dark blue) with 5c (cyan) and PGC1α peptide (yellow). Helix 12 is colored in green. The N-terminus, C-terminus, and secondary structure assignment are labeled. (C) Details of the interaction of 5c with ERRα. Residues of ERRα involved in the binding of 5c are depicted as stick models. Hydrogen bonds are marked as dotted lines. (D) Superposition of the 5c-bound (dark blue) and apo (pink) ERRα LBDs. Significant conformational changes are induced to accommodate the binding of 5c.

To better understand the structural rearrangement required for the binding of 5c to the ERRα LBD, the LBD structure in complex with 5c was compared with that of the apo ERRα LBD (PDB accession code 1XB7)¹⁷ (Figure 2D). The overall structures of the 5c-bound LBD and the apo LBD were similar; however, there were significant differences in the LBPs. Although residues from Val-321 to Leu-324 form helix H3 in apo ERRα, the same residues are unwound in the 5c-bound LBD. The N-terminal part of H3 moves away from the LBP to create the necessary space for 5c. In particular, the Cα atoms of Leu-324, Cys-325, and Phe-328 in H3 move by 5.4, 3.6, and 1.7 Å, respectively, to create the binding site for 5c. Furthermore, 5c induces multiple conformational changes within the LBP, including the side-chain atoms of Phe-328, Leu-365, Phe-382, and Phe-495.

In summary, a novel class of ERRα agonists has been discovered. From HTS hit 1, a slight enhancement of RIP140 corepressor peptide binding inhibitory potency and agonistic activity was achieved to give methylbenzimidazole 3d. Since a strong requirement was the elimination of PPARγ activity, the derivatization strategy involved the removal of the TZD moiety. From HTS hit 2, a structure–activity relationship study identified acetyl derivative 5c. The X-ray crystal structure of the ERRα LBD complexed with 5c revealed that the LBD adopts an agonist conformation and 5c induces multiple conformational changes in the LBP. This crystal structure also explains the importance of the phenolic hydroxyl group for binding potency. As compounds 3d (DS20362725) and 5c (DS45500853) did not activate PPARγ, both compounds can be useful tool compounds to investigate the biological roles of ERRα agonists. Pharmacological evaluations of these compounds are underway, and the findings will be reported when available.

Glossary

Abbreviations

ERR

estrogen-related receptor

PGC-1α

PPARγ coactivator 1α

PPAR

peroxisome proliferator-activated receptor

RIP140

receptor-interacting protein 140

HTS

high-throughput screening

TZD

thiazolidinedione

LBD

ligand-binding domain

LBP

ligand-binding pocket

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsmedchemlett.1c00100.

• Experimental procedures, characterization data for all synthesized compounds, procedures for pharmacological activities, and X-ray crystallography (PDF)

The authors declare no competing financial interest.

Notes

The atomic coordinates have been deposited in the Protein Data Bank (www.rcsb.org) with the accession code 7E2E.