Optimized Chemical Probes for REV-ERBα
J Med Chem. Author manuscript; available in PMC 2015 Mar 3.
Published in final edited form as:
J Med Chem. 2013 Jun 13; 56(11): 4729–4737.
Published online 2013 May 23.
In short, it would seem that GSK-4112 is a promising compound with a higher rate of bioavailability than SR-9009 (can be orally digested) with a longer half-life. It also demonstrates efficacy in combatting gastric cancers. The following paper tested this product against several analogues of SR-9009. It should be noted that there was little stastical deviation in mechanism of action between SR-9009 and SR-9011. The only drawback with GSK-4112 is that it currently costs $7.2 million dollars per kilo, based off Sigma Aldrich figures.
Abstract
REV-ERBα has emerged as an important target for regulation of circadian rhythm and its associated physiology. Herein, we report on the optimization of a series of REV-ERBα agonists based on GSK4112 (1) for potency, selectivity, and bioavailability. Potent REV-ERBα agonists 4, 10, 16, and 23are detailed for their ability to suppress BMAL and IL-6 expression from human cells while also demonstrating excellent selectivity over LXRα. Amine 4 demonstrated in vivo bioavailability after either IV or oral dosing.
Introduction
The circadian clock aligns all the tissues in most organisms with the day and night cycle of our planet. Through a transcriptional mechanism, the clock controls many important biological pathways, such as metabolism, inflammation, and sleep-wake cycles.1–3 REV-ERBα is a nuclear receptor that has been demonstrated to, upon activation by heme, form a complex with co-factors that represses the transcription of target genes.4,5 REV-ERBα is at the heart of the circadian clock and is a mechanism by which the circadian clock gates inflammatory response and controls the metabolic state of the organism.6–9 Investigation into the function of REV-ERBα will provide further insights into circadian biology and may identify the receptor as a therapeutic target for a variety of diseases.
Pharmacological investigations into the biological role of the REV-ERBα have used the sub-optimal chemical probe GSK4112 (1) and, its analogs SR9011 (2) and SR9009 (3) (Figure 1).10,11 All three compounds are closely related by a common tertiary amine core and two of three identical substituents off the amine. Use of these compounds to interrogate REV-ERBα biology is complicated by high metabolic clearance rates that necessitate high dosing to achieve meaningful levels of exposure in vivo. In addition, the tertiary amine chemotype has known activity on the nuclear receptor LXRα, a potential liability for interpretation of results from cell-based and animal pharmacology.12–14 Herein we describe our optimization of the tertiary amine series of REV-ERBα agonists to address these liabilities and allow further exploration of the role of REV-ERBα in circadian control and inflammation.
Reported REV-ERBα agonists 1-3.
Results and Discussion
To deliver a tool suitable for in vivo studies we evaluated the literature compounds 1–3 to assess opportunities for improvement. First we assessed the ability of the literature compounds 1–3 to bind to both REV-ERBα and LXRα. The tertiary amine 1 induced recruitment of co-factor NCOR peptide fragment to purified REV-ERBα protein, indicating that the compound binds to the protein and induces a conformational change (Table 1). Interestingly, neither compounds 2 or 3 increased recruitment of the NCOR peptide fragment to REV-ERBα in vitro. Amines 2 and 3 were reported to have other activities associated with REV-ERBα activation, so we investigated the compounds ability to activate the receptor by a different mechanism.11 To determine if these two compounds were recruiting a different set of peptides to the REV-ERBα protein, we performed a scan of co-factor derived peptide fragments recruited to REV-ERBα in the presence of amine 1 (See Supporting Information Figure 1). Aside from NCOR and SMRT, the only other tested peptide that showed significant interaction with the REV-ERBα was a fragment of PGC1-β. Full curve analysis of the ability of compounds 1-3 demonstrated that the compounds were dose-dependently able to induce recruitment of PGC1-β peptide to REV-ERBα (Figure 2). This result demonstrated that amines 2 and 3 do bind to the REV-ERBα protein to induce a conformational change and that the induced conformation is different from that induced by amine 1. These data raise the intriguing possibility that the different peptide recruitment profiles of the compounds may result in distinct biological activities. They also indicate that PGC1-β may have a yet uninvestigated role in REV-ERBα pharmacology. The investigation of these possibilities is beyond the scope of this paper.
Effect of compounds 1-3 on PGC1-β peptide fragment recruitment to REV-ERBα.
Table 1
REV-ERBα and LXRα activity of GSK4112 analogs.
 | |||||||||
|---|---|---|---|---|---|---|---|---|---|
| Entry | R1 | R2 | R3 | cLogP | REV-ERBαEC50a | LXRα | BMAL1 | ||
| IC50(μM) | Fold Sel. | Supp.b | Phase Delayc | ||||||
| 1 | tBuO2C- | 4-Cl-C6H4– | 5-Nitro-thiophen-2-yl | 5.2 | 0.50 | 5.0 | 10 | 6% | 0 |
| 2 | 1-Ethylcarboxylate-pyrrolidin-3-yl | 4-Cl-C6H4– | 5-Nitro-thiophen-2-yl | 4.8 | >50 | 6.3 | 0.13 | 20% | 1.3 |
| 3 | 1-N-Pentylcarboxamide-pyrrolidin-3-yl | 4-Cl-C6H4– | 5-Nitro-thiophen-2-yl | 5.3 | >50 | 13 | 0.25 | 14% | 1.3 |
| 4 | 4-Cl-2-Me-C6H3– | 4-Cl-C6H4– | 5-Nitro-thiophen-2-yl | 6.8 | 0.050 | 63 | 1,259 | 21% | 0 |
| 5 | Ph- | 4-Cl-C6H4– | 5-Nitro-thiophen-2-yl | 5.6 | 0.10 | >100 | >1,000 | NT | NT |
| 6 | 2-F-C6H4– | 4-Cl-C6H4– | 5-Nitro-thiophen-2-yl | 5.7 | 0.063 | 20 | 320 | 38% | 0 |
| 7 | iso-Butyl- | 4-Cl-C6H4– | 5-Nitro-thiophen-2-yl | 5.9 | 0.16 | 16 | 100 | 21% | 0 |
| 8 | 3-Pyridyl- | 4-Cl-C6H4– | 5-Nitro-thiophen-2-yl | 4.1 | 0.16 | 13 | 79 | NT | NT |
| 9 | 3-Pyridyl- | 4-Me-C6H4– | 5-Nitro-thiophen-2-yl | 3.9 | 0.16 | 40 | 250 | 23% | 0 |
| 10 | 3-Pyridyl- | 4-F-C6H4– | 5-Nitro-thiophen-2-yl | 3.5 | 0.16 | 40 | 250 | 14% | 0 |
| 11 | 3-Pyridyl- | 2-Me-C6H4– | 5-Nitro-thiophen-2-yl | 3.8 | 0.20 | 32 | 160 | 37% | 0 |
| 12 | 3-Pyridyl- | 3-Me-C6H4– | 5-Nitro-thiophen-2-yl | 3.9 | 0.20 | 25 | 130 | 41% | 0.8 |
| 13 | 3-Pyridyl- | 4-OMe-C6H4– | 5-Nitro-thiophen-2-yl | 3.3 | 0.20 | 32 | 160 | NT | NT |
| 14 | 3-Pyridyl- | 3-OMe-C6H4– | 5-Nitro-thiophen-2-yl | 3.3 | 0.40 | 10 | 25 | 42% | 0.5 |
| 15 | 3-Pyridyl- | 3-CF3-C6H4– | 5-Nitro-thiophen-2-yl | 4.3 | 0.40 | 16 | 40 | NT | NT |
| 16 | 3-Pyridyl- | 4-Cl-C6H4– | 5-Nitrile-thiophen-2-yl | 3.8 | 0.20 | 20 | 100 | 23% | 0 |
| 17 | 3-Pyridyl- | 4-Cl-C6H4– | 4-Br-5-Me-thiophen-2-yl | 5.7 | 0.25 | 32 | 130 | 16% | 0 |
| 18 | 3-Pyridyl- | 4-Cl-C6H4– | Benzo[b]thiophen-2-yl | 5.7 | 0.40 | 40 | 100 | 24% | 0 |
| 19 | 3-Pyridyl- | 4-Cl-C6H4– | 3-(1H-Pyrrol-1-yl)thiophen-2-yl | 5.5 | 0.63 | 40 | 63 | 30% | 0 |
| 20 | 3-Pyridyl- | 4-Cl-C6H4– | Thiazol-2-yl | 3.0 | >50 | 50 | <1 | NT | NT |
| 21 | 3-Pyridyl- | 4-Cl-C6H4– | 1H-Imidazol-2-yl | 2.5 | >50 | >100 | NT | NT | NT |
| 22 | 3-Pyridyl- | 4-F-C6H4– | 5-Nitrile-thiophen-2-yl | 3.2 | 0.20 | NT | NT | NT | NT |
| 23 | 3-Pyridyl- | 3,4-F2-C6H3– | 5-Nitrile-thiophen-2-yl | 3.3 | 0.20 | 16 | 79 | 37% | 0 |
| 24 | 3-Pyridyl- | 4-OMe-C6H4– | 5-Nitrile-thiophen-2-yl | 3.0 | 0.63 | >100 | >160 | NT | NT |
| 25 | 3-Pyridyl- | 2,5-F2-C6H3– | 5-Nitrile-thiophen-2-yl | 3.4 | 0.63 | 6.3 | 10 | 27% | 0 |
| 26 | 3-Pyridyl- | 2-F-C6H4– | 5-Nitrile-thiophen-2-yl | 3.3 | >50 | 32 | <1 | 6% | 0 |
aAll active compounds demonstrated a 30% – 60% increase in NCOR peptide recruitment.
bSuppression of BMAL1 expression after 40 h of 20 μM compound treatment.
cDelay of peak of second cycle in hours of 20 μM treated cells relative to DMSO treated cells.
Each of the compounds 1-3 was able to displace a radio ligand from the LXRα binding site, confirming the compounds bind to the LXRα at ~10 μM (Table 1). We investigated if the observed binding of the compounds to both REV-ERBα and LXRα induced related effects in a living cell. To investigate REV-ERBα cellular efficacy, we measured the ability of agonists 1 and 2 to inhibit IL-6 production from human THP-1 cells following LPS stimulation (Figure 3a). To assess the cellular activity of the compounds in an LXRα-dependent system, we compared the effects of the compounds 1 and 2 on expression of ABCA1 in THP-1 cells (Figure 3b). Both compounds were able to significantly inhibit LPS-stimulated IL-6 expression and up-regulate ABCA1 levels indicating significant cellular activity of each compound in both REV-ERBα and LXRα dependent phenotypes that correlated with the ability of the compounds to bind to target proteins. The original report on compound 2 described a lack of activity in an LXRα Gal4 reporter gene assay,11 but our results indicated that the compounds bound directly to LXRα and induced cellular affects through LXRα activation. As activation of LXRα has been reported to have diverse effects on multiple pharmacological pathways, including inflammation, cholesterol metabolism, and insulin resistance, a key goal for this project was to improve the selectivity profile for the series.12–14
THP-1 cells were stimulated with LPS 10 μg/mL in the presence of DMSO, 1 (10 μM) or compound 2 (10 μM). After six hours the cells were lysed and the RNA analyzed using RT-qPCR. Both compounds significantly repressed IL-6 but increased ABCA1 transcript expression (**<0.01, T test). A representative experiment of three experiments is shown.
In addition to the observed off-target activity, compounds 1- 3 have relatively high cLogP values of > 5 that may contribute to off-target binding, particularly with nuclear receptors, and the high clearance observed in reported mouse DMPK studies.15
The goal of our synthetic efforts was therefore to provide an effective and readily accessible in vivoprobe for REV-ERBα that could be used as an efficient tool to explore the biological profile of this fascinating receptor. To improve on the significant short falls of 1–3 it was crucial to increase selectivity over LXRα to allow clear and precise interpretations of subsequent data and secondarily reduce the lipophilicity of the compounds. Preparation of the target tertiary amines was through a series of reductive amination procedures (Scheme 1). The overall process could be carried out under ambient reaction conditions or accelerated through microwave irradiation. It also proved possible to carry out the entire reaction process through a one-pot two-step protocol allowing rapid access to range of tertiary amine scaffolds for evaluation. Pertinent data in this ligand development on a selection of the amines prepared are collected in Table 1.
Synthesis of tertiary amine compounds reported.
Initial efforts focussed on replacing the tert-butyl ester arm (R1) of compound 1. Alternative substitution at this position provided multiple options for improved REV-ERBα activity and LXRα selectivity, but often at the cost of an increase in compound cLogP (Table 1), however, significant improvements in the overall compound profile proved possible. Replacement of the tert-butyl ester with both substituted benzyl (4–6) and alkyl (7) groups was not only well tolerated but led to stunning improvements in LXRα selectivity. Although many of these changes resulted in an increase in cLogP, it showed that LXRα activity could ultimately be removed from this tertiary amine series through simple structural modification. In order to address the lipophilicity issue a variety of heterocyclic substituents were examined in this position, the vast majority of which were not tolerated (data not shown), however, introduction of a 3-pyridyl group (e.g. 8) provided a modest increase in REV-ERBα activity (EC50 = 0.16), significant improvement in LXRα selectivity (79 fold), and importantly a meaningful lowering in cLogP (4.1).
Using compound 8 as the basis for continued exploration of scaffold modification we examined variation of the 4-chlorobenzyl arm (R2) of compound 1. In contrast to substitution of R1, both alkyl chains and heterocyclic substituents were not tolerated in this position with all changes resulting in compounds with no significant activity at REV-ERBα (data not shown). It was possible to replace this substituent with a variety substituted benzyl groups (e.g. 9–15) that provided multiple options that maintained REV-ERBα activity and concurrently improved LXRα selectivity and reduced cLogP. For example, introduction of a 4-fluorobenzyl unit (10) lowered the cLogP significantly (3.5) and improved selectivity over LXRα to >250 fold. Other changes at R2 such as the introduction of a 4-methoxybenzyl substituent (13) provided a similar and substantial improvement in overall compound profile.
In parallel to these modifications, an array of compounds was prepared replacing the nitro-substituent on the thiophene ring (R3). Interestingly, the majority of replacements for the nitrothiophene ring system with alternative polar 5-membered ring heterocycles as exemplified by compounds 20 and 21, resulted in complete loss of activity in the primary NCOR peptide recruitment assay. It did prove possible to introduce different substituents to the thiophene ring and maintain REV-ERBα potency and improve LXRα selectivity as exemplified by compounds 17 and 18, however, these changes often proved detrimental to the cLogP. One crucial exception to this trend was introduction of the 5-nitrile thiophene in compound 16, which not only maintained REV-ERBα selectivity (EC50 = 0.2) but also resulted in a significant drop in cLogP (3.8) and a slight improvement in LXRα selectivity (100 fold).
Combination of the optimal substituents discovered for R1, R2, and R3 resulted in the tertiary amines 22–26 which on evaluation revealed the key compound 23 which maintained potency in the NCOR peptide recruitment assay and also provided the desired LXRα selectivity profile and reduced lipophilicity.
Initially, to confirm the ability of the compounds to induce REV-ERBα-driven phenotypes in a cellular system, profiling for the ability to affect the circadian expression of BMAL in synchronized U2OS cells was examined. By stably expressing a BMAL promoter-driven luciferase reporter in U2OS cells, we were able to capture real time bioluminescent oscillations of synchronized cells. A single endpoint assay of BMAL-luciferase suppression by REV-ERBα agonists (including compound 1) has been described in the literature.16 We confirmed that comparable data (a statistically significant 15% suppression) can be generated with compound 1 through our in-house U2OS reporter assay. Further, we showed that the natural ligand for REV-ERBα (heme), produced a >50% suppression in the same assay (data not shown). A representative data set from the assay for amine 4 is shown in Figure 5. Additional data from the assay are also shown in Table 1 and Supporting Information Figure 2. As can be seen, the increase in potency for recruitment of the NCOR peptide promoted by compound 4 correlates with the suppression and shift of the BMAL oscillation curve in a dose-dependent manner.
Oscillation of BMAL-Luc gene over time in synchronized U2OS cells with and without treatment with compound 4.
During the evolution of this SAR knowledge, compounds 4, 10, 16, and 23 (Figure 4) were selected for further profiling and evaluation of their suitability to be adopted as in vivo probes for REV-ERBα.
Key compounds selected during study for further evaluation as potential probes.
To assess the cellular efficacy of REV-ERBα activation and selectivity over a LXRα driven pathway, amines 4, 10, 16, and 23 were profiled for their ability to inhibit LPS induction of IL-6 production and to upregulate expression of ABCA1 in human THP-1 cells. All compounds demonstrated a significant reduction of IL-6 secretion following treatment with 4, 10, 16, and 23 at 1 μM with no measurable effect on ABCA1 levels (Figure 6). Additionally, none of the compounds showed any toxicity in measurement of ATP levels in THP-1 cells (See Supporting Information Figure 3). These data indicated that each of these tertiary amines was able to potently activate REV-ERBα in cells, that suitable selectivity over LXRα had been achieved, and that no general toxicity was skewing the results.
THP-1 cells were stimulated with LPS 10 mcg/mL in the presence of DMSO or compounds 4, 10, 16, 23 (1 μM), in triplicate. After six hours the cells were lysed and the RNA analyzed using RT-qPCR. All compounds repressed IL-6 without affecting ABCA1 transcript expression (*=p<0.05,**<0.01, One way ANOVA with post-hoc Dunnett’s).
The IV and oral DMPK profiles of compounds 3, 4, 10, 16, and 23 were evaluated in a 1 mg/Kg cassette dose experiment in C57Bl/6 mice. Four of the compounds demonstrated essentially identical profiles with short half-lives, high clearance, and low oral bioavailability. Compound 4 (GSK2945) was differentiated from the group with a longer half-life of 2.0 h and an oral bioavailability of 23%, despite the higher cLogP of the compound (Table 2 and Supporting Information Figure 4). The profile of amine 4 is suitable to achieve plasma concentrations around the IC50 of the compound with oral doses around 20-30 mg/Kg, a dosing regimen that is compatible with long term in vivo studies. The remaining four compounds (3, 10, 16, 23) would be suitable for acute time-of-day dosing by injection where short exposure of the compound at meaningful levels in desired. The specific advantage of 10, 16, and 23 over the previously reported tertiary amine 3 is the selectivity profile observed for REV-ERBα over LXRα.
Table 2
Pharmokinetic Parameters.
| Dose Route (1 mg/Kg) | PK Parameter | 3 | 4 | 10 | 16 | 23 |
|---|---|---|---|---|---|---|
| IV | CL (L/h/Kg) | 2.7 | 0.4 | 2.8 | 2.8 | 2.6 |
| Vss (L/Kg) | 1.6 | 0.3 | 1.5 | 1.8 | 1.4 | |
| Terminal t1/2 (h) | 0.55 | 1.65 | 0.74 | 1.27 | 0.74 | |
| AUClast(h*ng/mL) | 365 | 2,538 | 354 | 361 | 375 | |
| Oral | Tmax (h) | 0.50 | 2.0 | 0.25 | 0.25 | 0.25 |
| Cmax (ng/mL) | 7.41 | 180 | 8.98 | 9.14 | 9.94 | |
| AUClast(h*ng/mL) | 7.93 | 594 | 8.39 | 10.2 | 9.39 | |
| F (%) | 2.2 | 23.4 | 2.4 | 3.5 | 3.0 |
Conclusion
Herein we report on the discovery of REV-ERBα agonists from the GSK4112 series that demonstrated > 1,000 fold selectivity over LXRα and have improved DMPK properties. In addition, we have identified PGC1-β as a novel co-factor for REV-ERBα and have noted that reported ligands 1-3 recruit different co-factor peptides to REV-ERBα and may induce distinct pharmacology as a result. GSK2945 (4) is a potent agonist with a profile suitable for chronic in vivo dosing via both oral and IV routes. Other compounds (10, 16, and 23) are identified with a shorter half-life in rodents that could be used for acute time-of-day dosing studies. These compounds represent valuable additions to the pharmacological toolbox to investigate the biology of REV-ERBα without the complication of activity on LXRα.
Experimental Section
General
All chemical reagents were purchased and used as received. 1H NMR spectra were recorded on a Varian Unity-300 or Varian Unity Plus-400. Chemical shifts are expressed in parts per million (ppm, δ units). Coupling constants are in units of hertz (Hz). Splitting patterns describe apparent multiplicities and are designated as s (singlet), d (doublet), t (triplet), q (quartet), m (multiplet), or br (broad).
Unless otherwise noted, the LCMS system used to determine purity was a UPLC analysis was conducted on a Waters Acquity system with BEH C18, 2 × 50 mm, 1.7 micron column at 40 °C 95% H2O, 5% MeCN to 99% MeCN in 1.1 minutes, holding at 100% MeCN for 40 seconds. Water contained 0.2% v/v formic acid. MeCN contained 0.15% v/v formic acid. The flow rate was 1 mL/min with 5 μL of solution injected. Mass spectra were recorded utilizing electrospray ionization (ESI) or atmospheric pressure chemical ionization (APCI) switching between positive and negative modes with DAD scanning from 210 to 350 nm. All compounds tested were of ≥95% purity.
Normal phase chromatography was accomplished using either Isco or Biotage equipment using pre-packed silica columns.
Reverse phase HPLC was accomplished using Agilent 110 series prep HPLC systems using C18 Phenomenex Luna, 75 × 30 mm, 5-micron column using the gradient described. The flow rate was 70 mL/min and the product was collected based on UV detection at 220 or 254 nm.