Design, synthesis, and biological evaluation of a novel series of 2-(2,6-dioxopiperidin-3-yl)isoquinoline-1,3(2H,4H)-dione derivatives as cereblon modulators

Abstract In the current study, we designed and synthesised a novel series of 2-(2,6-dioxopiperidin-3-yl)isoquinoline-1,3(2H,4H)-dione derivatives as cereblon (CRBN) modulators. The results of the CCK8 assay revealed potent antiproliferative activity for the selected compound 10a against NCI-H929 (IC50=2.25 µM) and U239 (IC50=5.86 µM) cell lines. Compound 10a also can inhibit the TNF-α level (IC50=0.76 µM) in LPS stimulated PMBC and showed nearly no toxicity to this normal human cell line. The TR-FRET assay showed compound 10a having potent inhibitory activity against CRBN (IC50=4.83 µM), and the docking study confirmed a nice fitting of 10a into the active sites of CRBN. Further biology studies revealed compound 10a can increase the apoptotic events, arrest the NCI-H929 cells at G0/G1 cell cycle, and induce the ubiquitination degradation of IKZF1 and IKZF3 proteins by CRL4CRBN. These preliminary results suggested that compound 10a could serve as a potential antitumor drug and worthy of further investigation.

The crystal structure of CRBN-DDB1 binding to lenalidomide shows mechanistic insight into how IMiDs act on CRL4 CRBN . The IMiD compounds bind CRBN through their shared glutarimide ring, leaving portions of their variable phthaloyl ring solventexposed 14 . In this study, we describe the discovery of a series of isoquinoline-1,3(2H,4H)-dione derivatives as a type of novel CRBN modulator, which retain the glutarimide group and enlarge the five membered ring in the middle of the compound to six membered ring ( Figure 2). The SAR of all the newly synthesised compounds were studied by the proliferation assay. The TNF-a inhibition ability and toxicity to normal human cells were also investigated. The most potent compound 10a was selected to be further studied through the TR-FRET assay and molecular docking to identify its CRBN binding activity. Furthermore, the effect of 10a on the induction of apoptosis and cell cycle on NCI-H929 cell line were investigated using flow cytometry. The IKZF1 and IKZF3 proteins degradation ability of 10a was also investigated by immunoblot.

Chemistry
The synthetic route for 2-(2,6-dioxopiperidin-3-yl)isoquinoline-1,3(2H,4H)-dione derivatives is depicted in Scheme 1. Briefly, compounds 3a-c were synthesised from the commercial homophthalic anhydride derivatives and reacted with 3-aminopiperidine-2,6-dione hydrochloride under acetic acid. The compounds 9a and 10a-d were prepared from the nitro substituted 2-chlorobenzoic acid 4a-d. The compounds 4a-d first reacted with dimethyl malonate under the CuBr to obtain the compounds 6a-d. Compounds 6a-d amidated with 3-aminopiperidine-2,6-dione hydrochloride, and then decarboxylation under NaOH and cyclised under acetic acid condition to get the compounds 9a-d. The nitro group of the compounds 9a-d were reduced by stannous chloride to obtain the target compounds 10a-d. Compounds 12a, b were prepared from the reaction of compound 9a with alkyl halides and then reduced the nitro group.

Antiproliferative activity
All the new compounds were evaluated for their antiproliferative activities against NCI-H929 and U2932 cancer cell lines and contrasted with lenalidomide using CCK8 assay. The results revealed the ability of the new compounds to inhibit the growth of the selected cancer cell lines with IC 50 values (Table 1).
Replacing the amino group to other groups, such as compounds 3a (hydrogen), 3c (7-methoxy), and 9a (8-nitro) substituted, the activities also decreased, compared with compound 10a. Therefore, we chose compound 10a for further biological activity study and the molecular docking investigation.

PMBC toxicity and TNF-a inhibition assay
Previous studies have reported IMiDs stimulate immunomodulatory activity main through the TNF-a inhibition 9,10 . In this test, the effect of the synthesised compounds on TNF-a in LPS stimulated human peripheral blood mononuclear cell (PBMC) was evaluated ( Table 1). The SAR of TNF-a inhibition was similar to the antiproliferative activity. Compound 10a showed remarkable significant reduction in TNF-a level at IC 50 ¼0.76 ± 0.08 mM, compared with lenalidomide IC 50 ¼0.13 ± 0.02 mM.
And same time, the toxicity of the compounds on the normal human cell line PBMC was also tested by CCK8. All the compounds showed nearly no toxicity at the 20.0 mM concentration. The PMBC cell viability of the most active compound 10a (94%) was better than lenalidomide (86%) at the same concentration.

TR-FRET analysis
To determine the relative binding affinities between lenalidomide and compound 10a, we used a TR-FRET CRBN binding assay to determine the IC 50 values for these compounds. The assay monitors the displacement of the Cy5-labeled thalidomide from the tritrp pocket of CRBN. Under these assay conditions, the IC 50 value for compound 10a was at 4.83 mM, compared with lenalidomide IC 50 ¼1.69 mM (Figure 3). Proved compound 10a having high binding affinity to CRBN protein.

Annexin V-FITC/PI apoptosis assay
The IMiDs were reported to induce apoptosis through activation of caspase 8 in MM cells 36 . Accordingly, the ability of the selected compound 10a to induce apoptosis in NCI-H929 cells was investigated using annexin V fluorescein isothiocyanate (FITC)/propidium iodide (PI) staining assay. NCI-H929 cells were treated with compound 10a and lenalidomide for 72 h. The results are presented in Figure 4.
The results revealed a significant dose dependent increase of the apoptotic events by compound 10a from 6.0% in the control, to 29.5% and 34.6% at 0, 1, and 5 mM concentration, and be equivalent to lenalidomide from 5.9% in the control, to 31.2% and 43.9%. We can see the significant increase in the percentages of early apoptosis from 3.4% to 13.3% and late apoptosis from 2.6% to 21.3%. Confirmed compound 10a can induce NCI-H929 cells apoptosis.

Cell cycle analysis
To better understand the mechanism by which compound 10a inhibits MM cell growth, cell cycle analysis was performed with the selected compound 10a and contrasted with lenalidomide. The NCI-H929 cells were treated with each of the compounds at 0, 0.5, 1.0, and 5.0 mM for 48 h. Following this treatment, the PIstained cells were analysed and the results are outlined in Figure 5.
The results of the cell cycle analysis revealed that compound 10a and lenalidomide can induce G0/G1 cell cycle arrest. The cell cycle arrest ability of 10a showed a dose-dependent manner in the G0/G1 cell cycle from 34.0% to 42.9% at 0 and 5.0 mM concentrations.

Immunoblot analysis
The earlier studies have revealed that the CRBN modulators thalidomide, lenalidomide, and pomalidomide can induce the ubiquitination of IKZF1 and IKZF3 by CRL4 CRBN . Subsequent proteasome degradation of these transcription factors kills MM cells.
Accordingly, we used the immunoblot assay to measure the IKZF1 and IKZF3 proteins degradation in the current study. The results revealed that compound 10a induced the degradation of IKZF1 and IKZF3 as lenalidomide ( Figure 6(A)). The treatment with compound 10a resulted in the loss of IKZF1 and IKZF3 levels with a

Docking study into CRBN
To elucidate whether compound 10a targeted the CRBN protein, we carried out a molecular docking study to predict the possible binding mode of compound 10a with the CRBN. The outcomes of the molecular docking study showed that the binding mode of compound 10a within the binding pocket of CRBN (binding energy of -7.5 kcal/mol). The glutarimide group is held in a buried cavity between CRBN sheets b10 and b13, which has the similar binding pose with lenalidomide as the literature report ( Figure  7) 14 . Furthermore, the 8-aminoisoquinoline-1,3(2H,4H)-dione carbonyl (C1), the glutarimide carbonyl (C6) and the intervening amide (N1) are in hydrogen-bonding distance to CRBN residues Trp402, Trp382, and His380, respectively. By comparison to lenalidomide, the glutarimide carbonyls (C2 and C6) are in hydrogen-bonding distance to CRBN residues His380 and Trp382, respectively. The molecular docking analyses indicated that compound 10a binds snugly into the active sites of CRBN.

Conclusions
In summary, we designed and synthesised a novel series of 2-(2,6dioxopiperidin-3-yl)isoquinoline-1,3(2H,4H)-dione derivatives as new kinds of CRBN modulators. Study on the SAR of the derivatives based on cell proliferation assay, which resulted in the discovery of compound 10a with considerably antiproliferative potency against MM cell lines NCI-H929 (IC 50 ¼2.25 mM) and U239 (IC 50 ¼5.86 mM), and nearly no toxicity to the normal human cell line PMBC at 20 mM concentration. Compound 10a also can reduce the TNF-a level (IC 50 ¼0.76 mM) in LPS stimulated PMBC. The TR-FRET analysis and molecular docking study results of compound 10a agreed with its ability of inhibiting CRBN and predicted binding mode. Further biology studies revealed compound 10a can increase the apoptotic events, arrest the NCI-H929 cells at G0/G1 cell cycle, and induce the degradation of IKZF1 and IKZF3 proteins degradation by CRL4 CRBN . Our findings suggested that compound 10a could be considered as a potential anti-MM drug candidate or as a novel CRBN modulator which can be used for targeted protein degradation for further development.

Materials and methods
Unless otherwise noted, all reagents and solvents were obtained from commercially available sources and were used without purification. 1 H NMR spectra were tested in CDCl 3 or DMSO-d 6 with TMS as the internal reference on a Bruker AVANCE 400 (Billerica, MA). Mass spectra (MS) were obtained from Agilent 1100 mass spectrometer (Santa Clara, CA) with an electron spray ionisation source.

Synthesis of 2-(1,3-dimethoxy-1,3-dioxopropan-2-yl)-6-nitrobenzoic acid (6a)
To a solution of dimethyl malonate, 15 mL was added 2-chloro-6nitro-benzoic acid 4a (1.0 g, 5.0 mmol) and sodium methanolate (0.8 g, 15.0 mmol) under nitrogen. The mixture was stirred at room temperature for 30 min, then added cuprous bromide (0.12 g, 0.8 mmol). The resulting mixture was heated at 80 C for 24 h. After cooling to room temperature, water 50 mL was added to the mixture followed by hexanes 50 mL. The aqueous layer was separated and then added toluene 50 mL, and the biphasic mixture was filtered through celite to remove insolubles. Then the aqueous layer was separated and acidified with 6 N aqueous HCl to pH 2-3, and then extracted twice with ethyl acetate. The combined organic phase was washed with brine and dried over anhydrous Na 2 SO 4 . After filtration and evaporation, the crude residue was purified by flash column chromatography (ethyl acetate:petroleum ether ¼ 1:2) to obtain 6a (0.9 g, 61%) as a yellow oil. 1

2-(1,3-Dimethoxy-1,3-dioxopropan-2-yl)-4-nitrobenzoic acid (6c)
It was prepared as for 6a as a yellow solid, 68% yield. 1 4 g, 10.8 mmol), and 3-aminopiperidine-2,6-dione hydrochloride (0.7 g, 4.0 mmol) were added into dichloromethane 50 mL, the mixture reacted under room temperature overnight. Then water 50 mL was added to the solution and the aqueous layer was extracted twice with dichloromethane. The combined organic phase was combined and dried to give the crude compound 7a, which was used for the next reaction without further purification.
For the cell proliferation experiments, 1.0 Â 10 5 cells were first seeded in the 96-well culture plate and incubated overnight. Compounds tested were serially diluted in the corresponding media and then added to the well of 96-well plate in a final volume of 100 mL and then the cells were incubated for 72 h at 37 C. Finally, 10 mL of CCK-8 was added to each well for 1-4 h, and the absorbance values were read at 450 nm in a Microplate Reader (ELx-800, BioTek Instruments, Winooski, VT). The IC 50 value for each compound was calculated at the basis of the number of viable cells using GraphPad Prism 5.0 software (San Diego, CA).
Flow cytometry was used to analyse effects of the compounds tested on cell cycle (PI staining) and apoptosis (Annexin V-FITC and PI staining). Stained cells were analysed on a Flow Cytometer (BD Biosciences, Franklin Lakes, NJ), and the data were analysed using the Cell Quest software. 40 PBMC (4 Â 10 5 cells) were incubated in the 96-well culture plate and stimulated by 2 mg/mL LPS. The level of TNF-a was checked by standard ELISA (R&D Systems, Minneapolis, MN). Analysis was performed by following the manufacture's procedure for each ELISA kit. The test samples were assessed in triplicate and absorbance was taken on a Microplate Reader (ELx-800, BioTek Instruments, Winooski, VT) at 450 nm. The IC 50 value for each compound was calculated by comparison with standard curves with purified recombinant TNF-a using GraphPad Prism 5.0 software (San Diego, CA).

Immunoblot analysis
For immunoblot analysis, NCI-H929 cells were treated with the tested compounds at the indicated concentrations for various times, collected, and lysed in RIPA buffer in the presence of protease inhibitors on ice for 30 min. The protein from each sample was quantitated using a BCA protein assay kit, and was separated by SDS-PAGE and then transferred onto the PVDF membrane. The membranes were probed with specific primary antibodies at 4 C overnight, followed by incubation with IRDye V R 680 goat antimouse secondary antibodies. The signals were acquired using the Odyssey Infrared Imaging System (LI-COR, Lincoln, NE). 41,42 His-tagged CRBN-DDB1 complex (Abcam, Cambridge, UK, catalogue no. ab235611) 60 nM was mixed with Eu-anti-His Tag antibody 3 nM (Thermo Fisher, Waltham, MA, catalogue no. PV5596) in a final buffer containing 20 mM HEPES pH 7.0, 150 mM NaCl, and 0.005% Tween-20. The solution was then mixed with Cy5labeled thalidomide 10 nM and various concentrations of compounds, and then was incubated at room temperature for 1 h. FRET signals were measured on an EnVision plate reader by exciting at 340 nm and recording emission at both 615 nm (no FRET control) and 665 nm (FRET signals) with a 60 ms delay. FRET efficiency was calculated by the ratio of 665 nm/615 nm. Quantitative loss of FRET efficiency as a function of compound concentrations was fitted by GraphPad Prism 5.0 (San Diego, CA) and calculated the IC 50 .

Molecular docking studies
The docking studies were done using AutoDock 4.2.6. During the docking simulations, the pdb file of CRBN (pdb: 4CI2) was downloaded from protein data bank (http://www.rcsb.org/pdb) and the ligand and its single bonds were moved freely within the potential binding pocket. Discovery studio and PyMOL softwares were used to visualise the binding interaction. The result of the docking study of the compound 10a is represented in Figure 7.