Introduction

Nuclear receptors (NRs) are a large family of ligand-regulated transcriptional factors that control the expression of target genes involved in a range of physiological processes such as development, metabolism, and immunity1. The human NR superfamily is composed of 48 members. Approximately half of them have been characterized as ligand-activated transcription factors regulating the expression of target genes. The remaining receptors are considered to be orphan NRs because they do not have well-characterized ligands2. NRs share a modular structure composed of an N-terminal domain (AF1), a DNA-binding domain (DBD), a hinge region, a ligand binding domain (LBD) and a transcriptional activation function domain 2 (AF2). Binding of ligands in the LBD results in the recruitment of transcriptional co-activators (eg, steroid receptor co-activator 1, SRC1), or co-repressors (eg, nuclear receptor co-repressor, NCoR)3,4.

Retinoic acid receptor-related orphan nuclear receptors (RORs) are examples of orphan nuclear receptors that show constitutive activity in the absence of ligands. There are three subtypes of RORs: RORα (NR1F1), RORβ (NR1F2) and RORγ (NR1F3), which are also known as RORa, b and c in humans5. The three RORs share a high degree of sequence similarity but display distinct tissue distribution patterns and distinct functional roles in the regulation of many physiological processes6. Two forms of RORγ are found in both humans and mice, termed RORγ1 and RORγt; they only differ in their amino terminal domain. RORγ1 is expressed in a variety of tissues, including the thymus, muscle, kidney and liver, whereas RORγt (with a truncated N-terminus) is exclusively expressed in the cells of the immune system, eg, T helper cells and lymphoid tissue inducer cells7,8. RORγt has been shown to have key functions in the differentiation of Th17 cells and the secretion of inflammatory cytokines such as interleukin 17 (IL-17)9. Recent studies indicate that the modulation of IL-17 levels in autoimmune diseases may be beneficial in treating diseases, including multiple sclerosis, psoriasis, rheumatoid arthritis and inflammatory bowel disease. Thus, RORc is a promising therapeutic target for the treatment of Th17-mediated autoimmune diseases10,11.

Since the discovery of the first small molecule T0901317 (1, Figure 1), many RORc ligands with agonistic and inverse agonistic activity have been disclosed in the literature12,13. Some structurally complex natural products, such as digoxin (2) and ursolic acid (UA, 3), have been reported to be RORc inverse agonists14,15,16. Using the T0901317 scaffold as a lead compound, a series of synthetic RORc inverse agonists were developed, including SR1001 (4), SR1555 (5), and SR2211 (6)17,18,19. A team from Genentech identified N-isobutyl-N-((5-(4-(methylsulfonyl)phenyl)thiophen-2-yl)methyl)-1-phenylmethanesulfonamide as RORc inverse agonist (7, Figure 1) via a biochemical screening campaign20. Although the development of RORc inverse agonists has shown significant promise21,22, the development of new RORc selective modulators with therapeutic potential still remains an urgent need.

Figure 1
figure 1

Structures of RORc inverse agonists (1–7) and RORa agonist (8) from the literature.

PowerPoint slide

Full size image

In this paper, we report the discovery of novel 4-(4-(benzyloxy)phenyl)-3,4-dihydropyrimidin-2(1H)-ones as potent RORc inverse agonists. Compound 8 has been previously described as an agonist of RORa23. Considering the sequence and structural similarity of RORa and RORc and their crucial role in the development of Th17-related diseases, compound 8 was assessed for its inhibitory effect on RORc. Compound 8 was identified using an AlphaScreen assay (IC50=20.89 μmol/L). This result provided us with confidence that 8 was a genuine RORc inverse agonist. Thus, compound 8 was selected as the starting structure for further SAR study. Extensive optimization was conducted, which led to the discovery of a potent and selective RORc inverse agonist.

Materials and methods

General chemistry

All reagents were purchased from commercial sources and used without further purification. For the synthesized compounds described below, flash chromatography was performed using silica gel (300–400 mesh). All of the reactions were monitored by TLC using silica gel plates (fluorescence F254, UV light). 1H NMR spectra were recorded on a Bruker AV-400 spectrometer (Bruker, Fallanden, Switzerland) at 400 or 500 MHz. 13C NMR spectra were recorded at 400 or 500 MHz. CDCl3 is deuterochloroform, and DMSO-d6 is hexadeuterodimethylsulfoxide. All coupling constants were measured in hertz, and the chemical shifts (δH and δC) were quoted in parts per million (ppm) relative to TMS (δ 0), which was used as the internal standard. Abbreviations for NMR data are as follows: s=singlet, d=doublet, t=triplet, q=quartet, m=multiplet, dd=doublet of doublets, dt=doublet of triplets, app=apparent, and br=broad. The low or high resolution mass spectra (LRMS and HRMS) were measured on an Agilent 1200 HPLC-MSD mass spectrometer or an Applied Biosystems Q-STAR Elite ESI-LC-MS/MS mass spectrometer, respectively. The experimental procedures and characterization of all compounds are provided in the Supplementary information.

Biological assays

Protein and peptide preparation

The human RORc LBD (residues 262–507) was expressed as a His6-fusion protein using the pET24a expression vector (Novagen, Madison, WI, USA)24. BL21 (DE3) cells that were transformed with this expression plasmid were grown in LB broth at 25 °C until an OD600 of approximately 1.0 was reached, and the cells were then induced with 0.1 mmol/L isopropyl-β-D-1-thiogalactopyranoside (IPTG) at 16 °C overnight. Cells were harvested, resuspended, and high-pressure homogenized in 200 mL of extract buffer (20 mmol/L Tris (pH 8.0), 500 mmol/L NaCl, 10% glycerol, and 25 mmol/L imidazole) per 6 L of cells. The lysate was centrifuged at 20 000 rounds per minute for 30 min, and the supernatant was loaded onto a 5-mL NiSO4-loaded HisTrap HP column (GE Healthcare, Piscataway, NJ, USA). The column was washed with extract buffer, and the protein was eluted with a 25–500 mmol/L imidazole gradient. A gel filtration column (HiLoad S75, 16/60, GE Healthcare) was used for a second purification.

Transient transfection assays

Transient transfection assays were performed as described previously11,24. 293T cells were maintained in DMEM containing 10% fetal bovine serum (FBS). Cells were transiently transfected in Opti-MEM using Lipofectamine 2000 (Invitrogen), according to the manufacturer's protocol. The 96-well plates were plated 24 h prior to transfection (15 000 cells per well). For all of the experiments, each well of cells was transfected in Opti-MEM with 25 ng of reporter plasmids and 5 ng of Renilla luciferase expression plasmids. For Gal4-driven reporter assays, the cells were transfected with 25 ng of Gal4-RORg LBD (residues 262–507) and 25 ng of pG5Luc reporter. For native promoter reporter assays, the cells were co-transfected with the Pcp2/RORE-Luc reporter along with the plasmids encoding full-length RORc. The cells were then transfected with the indicated expression and reporter plasmids. The media was changed 24 h after transfection, and the compounds were added. Cells were incubated for another 24 h followed by harvesting for a luciferase assay using the dual-luciferase reporter assay system (Promega). Luciferase data were normalized to Renilla luciferase data, which was used as an internal transfection control.

AlphaScreen assay

Interactions between RORc and ligands were assessed by luminescence-based AlphaScreen technology (Perkin Elmer) using a histidine detection kit from PerkinElmer (Norwalk, CT, USA) as previously described24,25. All of the reactions contained 200 nmol/L receptor LBD bound to nickel acceptor beads (5 μg/mL) and 50 nmol/L biotinylated SRC1-4 peptide bound to streptavidin donor beads (5 μg/mL) in the presence or absence of the indicated concentration of control compound SR2211 or candidate compounds. The N-terminal biotinylated coactivator peptide SRC1-4 sequence was QKPTSGPQTPQAQQKSLLQQLLTE. Compound concentrations varied from 150 nmol/L to 200 μmol/L in the dose-response assay.

Thermal stability shift assay

Thermal stability shift assay (TSA) is an increasingly popular method to identify small molecule ligands. In this study, TSA were carried out using the Bio-Rad CFX96 Real-Time PCR system. All reactions were buffered in 10 mmol/L HEPES, pH 7.5, 150 mmol/L NaCl and 5% (v/v) glycerol at a final concentration of 10 μmol/L protein and 200 μmol/L compounds. The 20 μL reaction mix was added to the wells of 96-well PCR plates. SYPRO Orange (ABI, Sigma) was added as a fluorescence probe at a dilution of 1:1000 and incubated with compounds on ice at least 30 min. Total DMSO concentration was restricted to 2% or less. The temperature was raised with a step of 0.5 °C per minute from 30 °C to 80 °C, and fluorescence readings were taken at each (0.5 °C) interval. All experiments were performed in triplicates for the RORc inverse agonist screen. Melting temperatures (Tm) were calculated by fitting the sigmoidal melt curve to the Boltzmann equation using GraphPad Prism. ΔTm is the difference in Tm values calculated for reactions with and without compounds26,27.

Molecular docking simulation

The crystal structure of RORc in complex with compound 7 (PDB code: 4QM0) was used as the reference structure in the docking study. Protein structure preparation for docking studies included water deletion, hydrogen atom addition and protonation state adjustment. All of the ligand and protein preparations were performed in Maestro (version 9.4, Schrödinger, LLC, New York, NY, USA. 2013), implemented in the Schrödinger program (http://www.schrodinger.com). The Glide docking program with SP score was used for binding mode prediction28.

Results

Design rationality

In this study, we try to develop new RORc inverse agonists starting from compound 8 which was identified as RORa agonist23. In order to predict RORc-8 binding modes, we analyzed known RORc-ligand complex structures. In a previous study from Genentech, compound 7 (Figure 1) was demonstrated to be a potent RORc inverse agonist. In the RORc-LBD-7 complex structure (PDB code: 4QM0), the benzylic ring of 7 formed a π–π interaction with His479 and disturbed the secondary structure for helices 11′–12 in the RORc-LBD co-structure, which may account for the RORc inverse agonist profile of 7. The N-isobutyl group of 7 resided in a shallow hydrophobic pocket lined with Val376, Phe388, and Phe401. It is worth noting that access to the shallow hydrophobic pocket may help to achieve potency improvement20.

Because there is an asymmetric center in this scaffold, we first investigated the differences in the RORc binding modes of the S and R configurations of 8. The molecular docking result showed that the (S) enantiomer had a repeatable and stable binding mode. Thus, we selected 8 (S) for a docking study, and all further docking studies were carried out with 8 (S). In the following discussion, we explored the SAR of this series, with a concentration on the right-hand side of the molecule, which does not involve the asymmetric center. Therefore, compounds were synthesized as racemates to rapidly produce structure-activity data.

The molecular docking study of 8 (S) (Figure 2A) and 8 (S) overlaid with 7 (Figure 2B) was performed. In compound 8, the carbonyl oxygen of the urea and ester moiety interacted with the amino acid residues Gly380 and Gln286 through H-bonds. Two π-π interactions exist between compound 8 and His323 and Phe388. Upon careful examination of the binding mode of 8 (S) overlaid with 7, we considered the effects of increasing the size of the terminal aromatic ring of 8 by introducing substituents that would interfere with the packing of helix12. In addition, we also considered the effects of introducing substituents on the terminal aryl ring of 8 to access the hydrophobic pocket lined with Val376, Phe388, and Phe401. We investigated whether this modification (Figure 3) on the terminal aryl ring of 8 could improve the activity of RORc.

Figure 2
figure 2

Predicted binding modes of ligands with RORc. (A) The docking models of compound 8 (green) in the RORc ligand binding pocket (orange). (B) The superimposition of the binding modes of 7 (cyan) and 8 (green). The ligands and important residues are shown as sticks. Hydrogen bonds are depicted as red dashed lines, and π–π interactions indicted by green dashed lines.

PowerPoint slide

Full size image
Figure 3
figure 3

Strategy for structural optimization.

PowerPoint slide

Full size image

Synthesis

Compounds 15a-l and 18a-l were synthesized as shown in Scheme 1. The key intermediates 12a-c were synthesized from 4-hydroxybenzaldehyde (9a-c), urea (10) and ethyl acetoacetate (11) with anhydrous ZnCl2 by means of a Biginelli multicomponent reaction23. Compounds 13a-l and 16a-l were reduced with NaBH4 in the presence of ethanol, followed by treatment with phosphorus tribromide in DMF, which generated intermediates 14a-l and 17a-l, respectively. Condensation of intermediates 14a-l and 17a-l with intermediates 12a-b was accomplished using the Williamson ether synthesis to afford the desired target compounds 15a-l and 18a-l29.

Figure 6
figure 6

Reagents and conditions: (1a) ZnCl2, 80 °C, 1 h; (1b) 1. NaBH4, EtOH, rt, 1 h; 2. PBr3, DCM, rt, 1 h; (1c) 12, K2CO3, DMF, 80 °C, 4 h.

PowerPoint slide

Full size image

Compounds 22a and 22b were synthesized as shown in Scheme 2. Compounds 19a-b were reacted with phenol by using the Williamson ether synthesis in the presence of Cs2CO3 to give intermediates 20a-b. Compounds 20a-b were reduced with NaBH4, followed by treatment with phosphorus tribromide, which resulted in intermediates 21a-b. Condensation of the obtained intermediates 21a-b with the intermediate 12b was accomplished using the Williamson ether synthesis to afford the desired target compounds 22a and 22b.

Figure 7
figure 7

Reagents and conditions: (2a) phenol, Cs2CO3, DMF, 75 °C, 4 h; (2b) 1. NaBH4, EtOH, rt, 1 h; 2. PBr3, DCM, rt, 1 h; (2c) 12b, K2CO3, DMF, 80 °C, 4 h.

PowerPoint slide

Full size image

SAR analysis

We first explored the introduction of additional steric bulk on the terminal aryl ring of 8 in an effort to interfere with the packing of helix12. As demonstrated in Table 1, the introduction of a fluorine atom (15a) at the 4-position of the terminal phenyl ring of 8 resulted in a significant loss of potency. Naphthalene (15b) and quinoxaline (15c) derivatives also resulted in a complete loss of activity, whereas benzodioxole (15d) derivatives showed better activity than 8. To further investigate the impact of introducing larger substituents onto the terminal aromatic ring, compound 15e was synthesized. However, compound 15e had no activity. Replacement of the phenyl ring in compound 15e with a thiophene (15f) resulted in a slight potency decrease compared to 8. Interestingly, inserting an oxygen atom between biphenyl groups at the R1 position dramatically improved RORc inhibition (15g, 15h, 15i vs 15e). Compound 15g (IC50=3.47 μmol/L) demonstrated a 6-fold improved inhibition compared with 8 (IC50=20.89 μmol/L), whereas adding one flexible carbon atom between the oxygen atom and the terminal aryl ring led to a slight reduction in RORc potency (15j vs 15g). These results indicated that the presence of two aromatic rings linked with a flexible atom, such as an oxygen atom, at the R1 position increases the potency of the compounds, which is presumably due to the π-π stacking interaction between the terminal aryl ring of the compounds (15g, 15h, 15i) and His479 residues, which interfere with the packing of helix12 in the AF2 region. We further synthesized compounds 15k and 15l. The results demonstrated that both compounds resulted in a complete loss of activity.

Table 1 SAR at the R1 position.
Full size table

We next investigated the modification of substituent R2 (Table 2) to probe the hydrophobic pocket lined with Val376, Phe388, and Phe401. Taking into account the synthetic challenges, we also selected 8 as the starting point for further structural optimization. We first examined the importance of the methoxy substitution at the R3 position. When comparing compounds 18a and 8, it appeared that methoxy substitution at the R3 position favored efficacy. We then explored a range of substituents at the R2 position (Table 2). With the methoxy group maintained at the R3 position, R2 was modified from H to halogen, small alkyls, and small alkoxyl groups. The results of these modifications showed that the small substituents at the R2 position did not provide any potency improvement (18b, 18c, 18d, 18e vs 8). In the R2 position, substituents with different groups that were linked by an oxygen atom were attached to the aryl ring as a means of probing the intended cavity. Compounds 18f (O-nPr), 18g (O-nBu) and 18h (OBn) were synthesized and exhibited improved activity. Compound 18f exhibited the best activity, with an IC50 of 6.38 μmol/L and a temperature shift of 5.8 °C in the thermal shift assay. This result indicated that the propoxy group at the R2 position is the optimal size and fits very well in the hydrophobic pocket. Additionally, when comparing compounds 18b, 18f, 18g and 18h with compounds 18i, 18j, 18k and 18l, respectively, it appeared that the introduction of a chlorine atom at the R4 position increased RORc potency.

Table 2 SAR at the R2, R3, and R4 positions.
Full size table

Finally, all the beneficial changes were combined, and compounds 22a and 22b were designed and synthesized (Table 3). Furthermore, we also evaluated their cell activity in a reporter gene assay. As expected, both compounds showed good RORc inhibition activity. Compound 22b was the most potent RORc inverse agonist among all the compounds. The IC50 of 22b was 2.39 μmol/L and 0.82 μmol/L in the protein-based AlphaScreen assay and the reporter gene assay, respectively (Figure 4). The TSA assay also showed a stabilizing effect on the RORc protein with a temperature shift of 5.2 °C.

Table 3 SAR at the R2 position.
Full size table
Figure 4
figure 4

Dose-response curves for suppression of the derivative activity of RORc by 22b in the Dual Luciferase assay (A) and AlphaScreen assay (B). The IC50 values are 2.39 and 0.82 μmol/L, respectively.

PowerPoint slide

Full size image

Selectivity profile

To assess the selectivity profile for the three isoforms of ROR, we profiled the potent inverse agonist compounds with 293T cell-based assays using Gal4-NR constructs (Table 4). In addition, the selectivity profiles of these compounds for the farnesoid X receptor (FXR) and liver X receptor (LXR)-α were investigated using the same cell-based luciferase reporter gene assay30.

Table 4 RORc potency and selectivity profiles in Gal4 human transcription assaysa.
Full size table

Compounds 15g, 18f, 18k, 18l and 22b exhibited potent RORc activity in the reporter gene assay and excellent selectivity for RORc versus the other NRs in our selectivity panel. One highlight from this work was that 22b was the most potent RORc inverse agonist, with more than a 120-fold selectivity over the other NRs. 22b showed weak activity at FXR and LXRα. The results indicated that 22b gave a significantly improved selectivity window over the other NRs.

Discussion

To rationalize the SAR between these novel inverse agonists and RORc, a molecular docking study was performed with the most potent compound 22b (S) (Figure 5). The binding model of 22b (S) (Figure 5A) was similar to that of 8 (S) (Figure 2A). For compound 22b, the carbonyl oxygen of the urea and ester moiety form two hydrogen bonds with the backbone of Gly380 and Gln286. Three π-π interactions exist between compound 22b and His323, Phe388 and His479. The propoxy group of compound 22b resided in a shallow lipophilic pocket lined with Val376, Phe388 and Phe401. Docking results demonstrated that the SAR was consistent with the binding mode of compound 22b. Therefore, the docking simulations further confirmed that 22b is a favorable RORc inverse agonist. As discussed above, we showed that two aromatic rings linked with a flexible atom at the R1 position were crucial for RORc inhibition and that the size of the substituent at the R2 position was correlated with potency. These findings provide an important structural basis for further ligand optimization and suggest that inverse agonists for RORc with even higher pharmacological utility can be obtained in the near future.

Figure 5
figure 5

Predicted binding modes of ligands with the RORc protein. (A) 3D presentation of the predicted binding mode of 22b (magenta) superimposed with 8 (green) in the ligand binding pocket. (B) 2D schematic presentation of the predicted binding mode of 22b in the ligand binding pocket. The ligands and important residues are shown as sticks. Hydrogen bonds are depicted as red dashed lines, and π-π interactions indicted by green dashed lines.

PowerPoint slide

Full size image

In summary, we designed and optimized a new series of RORc inverse agonists comprising a 4-(4-(benzyloxy)phenyl)-3,4-dihydropyrimidin-2(1H)-one scaffold. A total of 27 target compounds were synthesized and evaluated for their inhibitory activity against RORc. Compound 22b was the most potent RORc inverse agonist in this series and displayed more than a 120-fold selectivity for RORc over other NRs as assessed by cellular assays. It is notable that 22b lacked inhibitory activity for RORa. The most potent and selective compound may serve as a useful tool for developing RORc inverse agonists. Further structural optimization and in vivo studies are currently in progress and will be reported in due course.

Author contribution

Yong XU, Jian-qin JIANG, and Yu-lai Zhou designed the experiments; Xi-shan WU, Rui WANG, and Yan-li XING, performed the experiments. All authors participated in data acquisition, analysis, and interpretation. Xi-shan WU, Rui WANG, Yan-li XING, Xiao-qian XUE, Yan ZHANG, and Yong XU wrote the paper. All authors reviewed the final manuscript.