|Home | About | Journals | Submit | Contact Us | Français|
We previously reported the discovery of BRD0476 (1), a small molecule generated by diversity-oriented synthesis that suppresses cytokine-induced β-cell apoptosis. Herein, we report the synthesis and biological evaluation of 1 and analogs with improved aqueous solubility. By replacing naphthyl with quinoline moieties, we prepared active analogs with up to a 1400-fold increase in solubility from 1. In addition, we demonstrated that compound 1 and analogs inhibit STAT1 signal transduction induced by IFN-γ.
Type-1 diabetes (T1D) is a chronic metabolic disorder that 78,000 children worldwide develop each year.1 This disease is characterized by autoimmune destruction of pancreatic β-cells, resulting in decreased insulin production. The secretion of pro-inflammatory cytokines by macrophages in the pancreatic islets of Langerhans is believed to trigger intracellular signaling cascades leading to β-cell apoptosis.2 Small molecules that restore β-cell viability in the presence of cytokines may have potential to augment insulin replacement therapy. However, relatively few small molecules are known to possess protective effects from cytokines.3–6
In a campaign to discover new therapies and probes for T1D, our group developed a phenotypic assay using the rat INS-1E β-cell line to screen for small-molecule suppressors of apoptosis in the presence of the cytokines tumor necrosis factor-α (TNF-α), interleukin-1β (IL-1β), and interferon-γ (IFN-γ).7 High-throughput screening identified several hits that increased β-cell viability. Subsequent medicinal-chemistry optimization of a screening hit belonging to a library derived from diversity-oriented synthesis (DOS)8–10 led to the discovery of BRD0476 (1, ML18711) (Figure 1), a small molecule with sub-micromolar activity (EC50 = 0.78 μM).12 1 represents a novel chemotype with a highly functionalized and stereochemically rich medium-sized (8-membered) lactam ring.13–14
Compound 1 exhibits poor aqueous solubility, which would adversely affect the bioavailability of 1 in animals and limit its use in vivo. In this study, we describe our efforts toward the synthesis and biological evaluation of analogs of 1 with improved solubility. In addition, we show that the biological activity of 1 and analogs largely results from inhibiting cell-signaling pathways activated by IFN-γ.
To rapidly access compound 1 and urea analogs, we developed an efficient, solution-phase synthesis of aniline 4 in four steps from 2, a DOS-generated intermediate previously prepared using an aldol-based “build/couple/pair” strategy (Scheme 1).12–13 Treatment of 2 with TBSOTf resulted in a protecting group switch from a N-Boc-carbamate to a N-silyl-carbamate intermediate, which was desilylated (TBAF) to afford a secondary crude amine. Direct cleavage of the Boc-carbamate to a free amine under acidic conditions was not feasible due to competing deprotection of the p-methoxybenzyl (PMB) ether. Capping the amine intermediate with 1,4-benzodioxan-6-sulfonyl chloride provided sulfonamide 3 in 74% isolated yield for three steps. Hydrogenation (H2, Pd/C) of the nitro moiety generated 4 in gram quantities.
Two strategies were pursued to install the urea moiety for compound 1 and analogs. One strategy involved addition of isocyanates directly to aniline 4. However, these reactions were sluggish and required microwave heating under basic conditions to achieve moderate yields at best. Furthermore, quinolyl isocyanates were not commericially available for the preparation of quinoline urea analogs, and their syntheses proved problematic using standard protocols for isocyanate formation. Instead, the urea moiety could be installed using a two-step protocol by subjecting 4 to diphosgene (ClCO2CCl3) to form isocyanate intermediate 5,15 followed by addition of various aryl and alkyl amines (Scheme 2). Isocyanate 5 did not require purification for further reactions and could be stored at −20 °C for relatively short periods of time (at least one week). Moreover, addition of amines to 5 was facile and proceeded at room temperature under neutral conditions to afford PMB-protected urea derivatives 6 in moderate to excellent yields (36–99%). Deprotection of the PMB ether with DDQ or TFA provided 1 and urea analogs 7 (see Table 1 for complete structures).
In our earlier report,12 we determined the potency and maximum activity of compound 1 and analogs using the CellTiter-Glo viability assay with INS-1E β-cells.16 In the present study, caspase-3/7 activity was measured as a more direct readout of apoptosis. 1 demonstrated a small protective effect with standard concentrations of cytokines in the caspase assay (condition A, Figure 2). Lowering the concentrations of IL-1β and TNF-α by 20- and 10-fold, respectively, as well as increasing IFN-γ 4-fold resulted in higher rescue efficacy of 1, despite lowering caspase-3/7 activity for this mixture (condition B). Further increasing the concentration of IFN-γ to 500 ng/mL resulted in equal levels of cytokine-induced caspase-3/7 activation, but elicited the strongest protective effects from 1, reducing caspase activity by 74% (condition C). Using these cytokine concentrations, 1 generated a dose-response with an IC50 value of 1.66 μM and a maximum activity of 71% (Table 1). We used this assay for the biological evaluation of additional analogs.
Using a thermodynamic solubility assay, the aqueous solubility of compound 1 was determined to be 0.1 μM in phosphate buffered saline (pH 7.4) (Table 1). This low solubility may be attributed to the lipophilic naphthyl moiety. In previous SAR studies, this group was found to be a crucial structural requirement in suppressing cytokine-induced β-cell apoptosis.12 An approach to increase aqueous solubility while retaining the naphthyl moiety involved disrupting molecular planarity between the naphthyl and urea groups to lower crystal packing energy of solutes.17 1-Naphthylmethyl urea 7a was designed to disrupt planarity by interrupting delocalization of the naphthyl- and urea-conjugated pi system with a methylene spacer. However, 7a did not show enhanced solubility and was inactive.
N-Methyl-N-1-naphthyl urea 7b was also designed to increase aqueous solubility by disrupting molecular planarity. The addition of the methyl group to the urea nitrogen may prevent access to planar conformations, which should be hindered by a steric interaction between the N-methyl and C(2)- or C(8)-hydrogen of the quinoline moiety. In fact, 7b exhibited a 110-fold increase in solubility (11.2 μM), which correlated to a decreased melting point for 7b (144–146 °C) compared to 1 (170–172 °C). However, 7b was less efficacious in inhibiting apoptosis (maximum activity = 15%)
Alternatively, quinoline analogs 7c–i should retain the steric properties of the naphthyl urea moiety crucial for biological activity, while increasing aqueous solubility. Introduction of nitrogen atoms would decrease the lipophilicity of the naphthyl component and provide an additional hydrogen bond acceptor that should enhance solubility.18–19 Indeed, quinoline analogs 7c–i demonstrated an increase in solubility, in which 5-quinolyl urea 7f showed a remarkable 1400-fold increase in solubility compared to 1. 7f was equally potent (1.62 μM) to 1 in the caspase assay but had a lower maximum activity (41%). Likewise, the exceptionally soluble 4-isoquinolyl (7d) and 5-isoquinolyl (7g) ureas exhibited similar biological activity. 1-Isoquinolyl (7c) and 8-quinolyl (7i) ureas were less soluble (0.7 and 1.8 μM, respectively) than other quinoline analogs, which may be due to hydrogen bonding of the urea hydrogen (N-H) and quinoline nitrogen. However, 7i possessed the best activity among the quinoline analogs, inhibiting apoptosis similarly to 1 with a favorable maximum activity of 69%.
As indicated in the optimization of cytokine concentrations for the caspase-3/7 assay, compound 1 and active analogs may directly perturb cell-signaling pathways induced by IFN-γ. This cytokine activates the JAK/STAT signaling pathway,20–21 and STAT1 has previously been shown to regulate apoptosis in INS-1E β-cells.22–23 Binding of IFN-γ to cell surface receptors leads to activation of JAK1/JAK2, which subsequently phosphorylate cytoplasmic STAT1 at Tyr701. Upon phosphorylation, STAT1 dimerizes and translocates to the nucleus to regulate gene transcription.
When INS-1E β-cells were treated with IFN-γ, rapid phosphorylation of STAT1 at 30 min was observed by Western blotting (Figure 3). In addition to Tyr701, phosphorylation was also observed at Ser727.20–21 Co-treatment with IFN-γ and compound 1 led to a complete reduction of STAT1 phosphorylation, while inactive analog 7a did not result in inhibition. Further, total STAT1 levels were slightly increased by IFN-γ alone or in combination with 1 and analogs. Taken together, these results suggest that STAT1 phosphorylation may be an excellent biomarker for compound activity. To this end, the 8-quinolyl urea analog 7i with improved aqueous solubility also significantly decreased levels of pSTAT1 in an analogous manner to 1.
To examine the effect of inhibiting STAT1 phosphorylation on subsequent nuclear translocation, cells were treated with IFN-γ in the presence or absence of compounds, fixed, and immunofluorescence was performed to monitor subcellular localization of STAT1 (Figure 4).24 STAT1 was located in both the cytoplasm and nucleus of untreated cells, while treatment with IFN-γ induced a substantial accumulation of STAT1 in the nucleus. Remarkably, compounds 1 and 7i nearly abolished this nuclear localization of STAT1. As a negative control, compound 7a did not inhibit nuclear translocation of STAT1 in the presence of IFN-γ.
In summary, we developed an efficient synthesis of compound 1 and analogs, which suppress cytokine-induced β-cell apoptosis. Quinoline-containing analogs showed improved aqueous solubility from 1, of which 8-quinolyl urea 7i retained optimal biological activity. In addition, we have demonstrated that these small molecules inhibit the JAK-STAT signaling pathway by inhibiting IFN-γ-induced STAT1 phosphorylation and nuclear translocation. Studies to improve the pharmacokinetic properties of 1 and 7i, as well as develop a quantitative immunofluorescence assay for STAT1 nuclear translocation, are ongoing and will be described in subsequent reports.
The purity of all compounds was >95% as determined by UPLC-MS. The synthesis of 3, 4, and 7i is described below. See Supporting Information for detailed syntheses and spectral characterization of all compounds.
TBSOTf (1.98 mL, 8.61 mmol) was added dropwise to a solution of N-Boc-protected amine 212 (1.6 g, 2.87 mmol) and 2,6-lutidine (1.35 mL, 11.59 mmol) in CH2Cl2 (29 mL) at rt. The reaction mixture was stirred at rt for 1h, diluted with sat. NH4Cl, and extracted into CH2Cl2. The organic layers were dried over MgSO4, filtered, and concentrated in vacuo to provide a N-silyl carbamate intermediate. LCMS (ESI+) m/z: 616.39 (M+H). This crude intermediate was dissolved in THF (57 mL) and cooled to 0 °C. To the resulting solution was added AcOH (0.181 mL, 3.16 mmol) followed by dropwise addition of TBAF (1.0 M in THF, 3.16 mL, 3.16 mmol). The reaction mixture was further stirred for 30 min at 0 °C, diluted with sat. NH4Cl, and extracted into EtOAc. The organic layers were washed with brine, dried over MgSO4, filtered, and concentrated in vacuo to provide a secondary amine. LCMS (ESI+) m/z: 458.41 (M+H). This crude intermediate and 2,6-lutidine (1.67 mL, 14.35 mmol) were dissolved in CH2Cl2 (29 mL), and 1,4-benzodioxan-6-sulfonyl chloride (808 mg, 3.44 mmol) was added to the resulting solution at rt. The reaction mixture was further stirred for 20 h at rt, diluted with sat. NH4Cl, and extracted into CH2Cl2. The organic layers were dried over MgSO4, filtered, and concentrated in vacuo. Purification on silica gel (gradient of 10–50% EtOAc in hexanes) provided 1.39 g (2.12 mmol, 74% yield for 3 steps) of 3 as a white foam. 1H NMR (300 MHz, CDCl3) δ 7.86 (dd, J = 8.1, 1.2 Hz, 1H), 7.70 (dd, J = 7.5, 1.5 Hz, 1H), 7.30-7.21 (ovrlp m, 5H), 6.95 (d, J = 8.4 Hz, 1H), 6.84 (d, J = 8.7 Hz, 2H), 4.55 (d, J = 11.7 Hz, 1H), 4.47 (d, J = 11.7 Hz, 1H), 4.39 (m, 1H), 4.29 (m, 4H), 3.90-3.77 (ovrlp m, 3H), 3.77 (ovrlp s, 3H), 3.66 (dd, J = 9.9, 4.5 Hz, 1H), 3.38-3.20 (ovrlp m, 2H), 3.11 (d, J = 12.6 Hz, 1H), 2.81 (s, 3H) 2.70 (m, 1H), 1.38 (d, J = 6.9 Hz, 3H), 0.86 (d, J = 6.6 Hz, 3H); LCMS (ESI+) m/z: 656.38 (M+H).
Under hydrogen (H2) atmosphere, a solution of 3 (1.39 g, 2.12 mmol) and palladium (10% on activated carbon, 226 mg, 0.212 mmol) in EtOH (42 mL) was stirred at 35°C for 3.5 h. The reaction mixture was cooled, filtered, and washed through a pad of celite, and concentrated in vacuo. Purification on silica gel (gradient of 10–60% EtOAc in hexanes) provided 1.21 g (1.93 mmol, 91% yield) of 4 as a white foam. 1H NMR (300 MHz, CDCl3) δ 7.34-7.26 (ovrlp m, 4H), 7.02 (d, J = 8.4 Hz, 1H), 6.97 (d, J = 7.5 Hz, 1H), 6.85 (d, J = 8.7 Hz, 2H), 6.79 (d, J = 7.8 Hz, 2H), 4.67 (m, 1H), 4.54 (d, J = 11.4 Hz, 1H), 4.47-4.43 (ovrlp m, 3H), 4.36-4.33 (ovrlp m, 4H), 3.81-3.74 (ovrlp m, 2H), 3.79 (ovrlp s, 3H), 3.67 (dd, J = 10.2, 4.5 Hz, 1H), 3.49 (dd, J = 15.6, 10.5 Hz, 1H), 3.30 (d, J = 13.8 Hz, 1H), 3.06 (d, J = 15.0 Hz, 1H), 2.94-2.86 (ovrlp m, 1H), 2.90 (ovrlp s, 3H), 2.01 (m, 1H), 1.32 (d, J = 6.9 Hz, 3H), 0.83 (d, J = 6.9 Hz, 3H); HRMS (ESI+) m/z calculated for C32H39N3O8SNa (M+Na): 648.2356, found 648.2352.
To a solution of disphosgene (11 μL, 0.095 mmol) in CH2Cl2 (0.5 mL) at 0°C was added dropwise a solution of aniline 4 (100 mg, 0.160 mmol) and proton sponge (67 mg, 0.314 mmol) in CH2Cl2 (0.5 mL). The reaction mixture was warmed to rt, stirred for 15 min, and concentrated in vacuo. The residue was redissolved in CH2Cl2, washed with 1M HCl and 1M NaOH, dried over MgSO4, filtered, and concentrated in vacuo to provide 120 mg (0.184 mmol, 96% yield) of isocyanate 5. LCMS (ESI+) m/z: 652.25 (M+H). Crude 5 (80 mg, 0.123 mmol) and 8-aminoquinoline (89 mg, 0.615 mmol) were dissolved in toluene (2.1 mL) and stirred at rt for 2.5 h. The reaction mixture was concentrated in vacuo and directly purified on silica gel (gradient of 0–5% MeOH in CH2Cl2) to provide 88 mg (0.111 mmol, 90% yield) of quinolyl urea 6i. LCMS (ESI+) m/z: 796.37 (M+H). To a solution of 6i (53 mg, 0.067 mmol) in CH2Cl2 (2.3 mL) was added TFA (1.1 mL) dropwise at rt. The reaction mixture was further stirred at rt for 15 min, concentrated in vacuo, and directly purified on silica gel (gradient of 0–5% MeOH in CH2Cl2) to provide the 39 mg of 7i (0.058 mmol, 87% yield). 1H NMR (300 MHz, CDCl3) δ 9.63 (s, 1H), 8.86 (d, J = 2.7 Hz, 1H), 8.49 (d, J = 7.2 Hz, 1H), 8.27 (d, J = 8.1 Hz, 1H), 8.18 (s, 1H), 8.07 (dd, J = 6.6 Hz, 3.0 Hz, 1H) 7.53 (ovrlp m, 3H), 7.32 (d, J = 1.8 Hz, 1H), 7.26 (ovrlp dd, J = 8.1 Hz, 1.5 Hz, 1H), 7.16 (ovrlp m, 2H), 6.85 (d, J = 8.4 Hz, 1H), 5.65 (br s, 1H), 4.22 (ovrlp m, 4H), 4.10 (m, 1H), 3.78 (dd, J = 11.1 Hz, 2.1 Hz, 1H), 3.79-3.47 (ovrlp m, 4H), 3.31 (d, J = 13.5 Hz, 1H), 3.02 (ovrlp d, J = 15.6 Hz, 1H), 2.95 (ovrlp s, 3H), 2.26 (m, 1H), 1.41 (d, J = 6.9 Hz, 3H), 0.83 (d, J = 6.3 Hz, 3H); HRMS (ESI+) m/z calculated for C34H38N5O8S (M+H): 676.2441, found 676.2441.
Financial support from the NIH-NIDDK (Type 1 Diabetes Pathfinder Award, to B.K.W.) is gratefully acknowledged. K.P. was sponsored in part by a NIH/NIGMS MARC U*STAR T34 08663 National Research Award and the Howard Hughes Medical Institute (HHMI) Undergraduate Science Education Program at UMBC. We thank Tamara Gilbert for experimental assistance as well as Dr. Danny Chou, Dr. Jeremy Duvall, and Prof. Stuart Schreiber (Broad Institute) for helpful discussions.