Synthesis and Biological Evaluation of 3-Benzisoxazolyl-4-indolylmaleimides as Potent, Selective Inhibitors of Glycogen Synthase Kinase-3β

A series of novel 3-benzisoxazolyl-4-indolyl-maleimides were synthesized and evaluated for their GSK-3β inhibitory activity. Most compounds exhibited high inhibitory potency towards GSK-3β. Among them, compound 7j with an IC50 value of 0.73 nM was the most promising GSK-3β inhibitor. Preliminary structure-activity relationships were examined and showed that different substituents on the indole ring and N1-position of the indole ring had varying degrees of influence on the GSK-3β inhibitory potency. Compounds 7c, 7f, 7j–l and 7o–q could obviously reduce Aβ-induced Tau hyperphosphorylation by inhibiting GSK-3β in a cell-based functional assay.

Staurosporine, a microbial alkaloid, was identified as a potent but nonselective GSK-3β inhibitor. Various bisindolylmaleimides such as GF 109203X and Ro 31-8220 have also been developed as potent GSK-3β inhibitors based on staurosporine ( Figure 1) [14][15][16]. However, most of these bisindolylmaleimides are not suitable for the treatment of diseases such as diabetes and Alzheimer's disease due to their toxicity, poor solubility and low selectivity, especially against PKC family [17,18]. The replacement of one indole with other heteroaryl groups resulted in a series of monoindolylmaleimides such as 4-azaindolyl-indolyl-maleimides, benzofuranyl-indolyl-maleimides and imidazo[1,2-a]pyridinyl-indolyl-maleimides (Figure 1), which showed potent and selective GSK-3β inhibitory activities [19][20][21]. Among them, GSK-3β inhibitor 603281-31-8 developed by Eli Lilly & Co had reached preclinical studies for the treatment of diabetes and was proved efficacy in ZDF rats [19]. In view of these facts and also as a part of our work on the development of potent and selective GSK-3β inhibitors, herein we report the synthesis and biological evaluation of a new series of 3-benzisoxazolyl-4-indolyl-maleimides as GSK-3β inhibitors. Their structure-activity relationship and in silico molecular modeling study are also discussed in this study.

Chemistry
The general synthetic approach to target compounds 7a-q is outlined in Scheme 1.

Scheme 1. Synthetic route to compounds 7a-q.
Reagents and conditions: (a) i (COCl) 2  Indole derivatives 1a-f were reacted with oxalyl chloride in Et 2 O, followed by sodium methoxide to give compounds 2a-f. Reaction of 2a with (Boc) 2 O in the presence of a catalytic amount of DMAP in THF afforded 3a. N-alkylation of 2a-f with different alkyl halides resulted in key intermediates 3b-h and 3j-q. In addition, treatment of indole with 1,4-dibromobutane afforded 4. N-substitution reaction of 4 with morpholine using K 2 CO 3 as acid-trapping agent resulted in 5, which was then treated with oxalyl chloride, followed by sodium methoxide to give another key intermediate 3i. Condensation of glyoxylic esters 3a-q with 2-(benzo[d]isoxazol-3-yl)acetamide 6 [22] in the presence of t-BuOK in THF afforded the target compounds 7a-q.

Enzymatic Activity
The GSK-3 inhibitory potency of all target compounds was examined. In addition, selected compounds 7c, 7j and 7o were also tested for their inhibitory potency against other kinases (PKCepsilon, JAK2, BRAF, IKK2, Drak2) to assess kinase selectivity. Staurosporine, a well known kinase inhibitor was used as the reference compound [19]. The results are listed in Tables 1 and 2. As indicated in Table 1, most of the tested compounds showed similar or more potent GSK-3β inhibitory activity as compared to that of staurosporine. The potency of GSK-3β inhibition of tested compounds was mainly influenced by the substitutions on the indole ring and N 1 -position of the indole ring.
The results of inhibitory activities of compounds 7g-i, 7j and 7k showed that the length of the N 1 -alkyl linker affected GSK-3 inhibitory potency. For example, compound 7j (IC 50 = 0.73 nM) with a (CH 2 ) 3 linker showed better inhibitory activity than compound 7k (IC 50 = 89.8 nM) with a (CH 2 ) 4 linker. The same conclusion could also be drawn from comparison of the inhibitory potency of 7h and 7i.
Interestingly, the introduction of a hydrophobic methyl group on the N 1 -position of the indole ring in 7a resulted in a 15-fold increase in inhibitory potency for GSK-3, while the replacement of the methyl with a large butyl group showed a 3-fold decrease in potency for GSK-3β inhibition.
When comparing the inhibitory activity of 7b-7f with 7h, it suggested that different substituents on the indole ring affected the inhibitory potency for GSK-3β. Compound 7c (IC 50 = 10.2 nM) with bromine at 5-position of the indole ring showed a 14-fold increase in inhibitory activity toward GSK-3 as compared to that of 7h (IC 50 = 137.7 nM). Fluorine at 6-position of the indole ring did not influence activity of 7h, while bromine or chlorine at 6-position or methoxy at the 5-position showed less inhibitory potency.  The data in Table 2 showed that staurosporine was a potent and nonselective kinase inhibitor as reported in the literature [14], which inhibits not only GSK-3β (IC 50 = 72.2 nM) but also many other kinases (e.g., PKC-epsilon, JAK2, BRAF, IKK2 and Drak2). Selected compounds 7c, 7j and 7o with high potency for GSK-3β inhibition were also evaluated for kinase selectivity against PKC-epsilon, JAK2, BRAF, IKK2 and Drak2. The results indicated that they displayed high selectivity for GSK-3β over other tested kinases.

Cellular Activity
It has been implicated that GSK-3β is involved in multiple cellular processes, and its ability to hyperphosphorylate Tau protein and induce neurofibrillary tangle was intensively studied. Therefore, the cell-based assay examining Tau phosphorylation at Serine 396 represents a direct functional assay to measure the cellular activity of GSK-3β inhibitors [23]. Compounds 7c, 7f, 7j-l and 7o-q were tested for the ability to reduce Tau phosphorylation at Ser 396 in human neuroblastoma SH-SY5Y cells. LiCl, a well-known inhibitor of GSK-3β [24], was used as a positive control in this assay. As shown in Figure 2, all selected compound significantly reduced Aβ 25-35 -induced Tau hyper-phosphorylation, showing that these compounds can inhibit GSK-3β activity at the cellular level. The Aβ-induced Tau hyperphosphorylation results in neurofibrillary tangle formation, which plays an important role in Alzheimer's disease. Our data suggests that these novel GSK-3β inhibitors may have potential actions on inhibition of neurofibrillary tangle formation and would be tested for the treatment of Alzheimer's disease. In addition, the predictions about these compounds' brain permeability were performed using ADME module with Discovery Studio 2.1 software package. According to the prediction results (data not shown), most of them exhibited moderate blood-brain barrier permeability and would be further investigated for the treatment of Alzheimer's disease.

Molecular Modeling
To examine possible binding modes of compounds bearing different side chains at N 1 -position of the indole ring (e.g., 7j and 7n) with GSK-3β, a docking analysis utilizing the C-DOCKER program within the Discovery Studio 2.1 software package was performed. The published X-ray crystal structure of GSK-3β (PDB ID: 1Q3D) [14] was used for the docking calculation. Figure 3 shows that both 7j and 7n could occupy the ATP binding site of GSK-3β with similar binding modes as a few other ATP-competitive inhibitors of GSK-3β [14,25], and they could thus serve as ATP-competitive inhibitors of GSK-3β. The NH and carbonyl group in maleimide ring of 7j and 7n could form two key hydrogen bonds with Asp133 and Val135 of GSK-3β. Besides, the 3-position nitrogen atom of the imidazole ring of 7j could form another hydrogen bond with Lys-183, which was not observed in 7n. Furthermore, the CDOCKERENERAGE of 7j (−20.821 kcal/mol) was much lower than that of 7n (2.971 kcal/mol). Thus, the molecular docking study results could explain the fact that 7j showed significantly improved potency compared to that of 7n.

General
Melting points were determined with a BÜCHI Melting Point B-450 apparatus (Büchi Labortechnik, Flawil, Switzerland). The 1 H-NMR spectra were recorded in DMSO-d 6 or CDCl 3 on Bruker Avance DMX 500 at 500 MHz (chemical shifts are expressed as δ values relative to TMS as internal standard). ESI spectra (positive ion mode) were recorded on an Esquire-LC-00075 spectrometer. Elemental analyses were performed on an Eager 300 instrument. All reactions were monitored by thin-layer chromatography (TLC). All reagents were obtained from commercial sources and used without further purification unless stated. Et 2 O and THF were distilled from sodiumbenzophenone. DMF was distilled from calcium hydride.

General Procedure for the Preparation of 2a-f
Oxalyl chloride (3.40 g, 26 mmol) in Et 2 O (5 mL) was added dropwise to a solution of indole adducts 1a-f (26 mmol) in Et 2 O (30 mL) at 0~5 °C [26,27]. The reaction mixture was stirred under the same conditions for 1 h, and a 20 wt% solution of CH 3 ONa in MeOH (14.05 g, 52 mmol) was added dropwise at −30 °C~−20 °C. After addition, the mixture was stirred for 30 min at room temperature, and poured into cold water (100 mL). The product was isolated with filtration, washed with dichloromethane and dried to afford 2a-f.

General Procedure for the Preparation of 3b-h, 3j-n and 3q
70% NaH (0.51 g, 14.8 mmol) was added portionwise to a solution of 2a-f (14.8 mmol) in DMF (30 mL) at 0~5 °C. The reaction mixture was warmed to room temperature and stirred for 30 min.

General Procedure for the Preparation of 3o and 3p
70% NaH (0.17 g, 4.9 mmol) was added portionwise to a solution of 2a (1.0 g, 4.9 mmol) in DMF (10 mL) at 0~5 °C. The reaction mixture was warmed to room temperature and stirred for 30 min. After that, iodomethane or 1-bromobutane (5.9 mmol) was added at 0 °C and stirred for 1 h at room temperature. The mixture was then poured into cold water (100 mL) and extracted with ethyl acetate (3 × 50 mL). The organic phase was combined, washed with brine (3 × 150 mL), dried over Na 2 SO 4 and concentrated in vacuo. The residue was purified by flash column chromatography on silica gel using petroleum ether/ ethyl acetate (3:1, v/v) as eluent to afford 3o and 3p.

General Procedure for the Preparation of 7a-q
A solution of t-BuOK (94.0 mg, 0.84 mmol) in THF (5mL) was added dropwise to a solution of 3a-q (0.36 mmol) and 2-(benzo[d]isoxazol-3-yl)acetamide 6 (0.28 mmol) in dry THF (10 mL) at −10~0 °C. After stirring for 2 h at room temperature, concentrated hydrochloric acid (5 mL) was added and the result mixture was stirred for 30 min at room temperature, then poured into 10% NaHCO 3 aqueous solution (100 mL) and extracted with ethyl acetate (3 × 50 mL). The organic phase was combined and washed with brine (3 × 150 mL), dried over Na 2 SO 4 and concentrated in vacuo. The residue was purified by flash column chromatography on silica gel using dichloromethane/methanol (50:1, v/v) as eluent to afford 7a-q.

Kinase Assay
Human kinome, consisting of 518 kinases, is classified into CMGC, AGC, TK, TKL, CAMK, STE and other 7 subfamilies, according to the DNA sequence and evolution. GSK3β belongs to CMGC family. Here, the other family members were used to evaluate the selectivity of GSK3β inhibitors, including PKC-epsilon (AGC family), JAK2 (TK family), Braf (TKL family), DRAK2 (CAMK family) and IKKβ (other family).The recombinant GST-GSK-3β protein was expressed in Escherichia coli strain BL21-Codon Plus (DE3), purified by GSTrap affinity chromatography, and cleaved by thrombin. The GSK-3β kinase assay was carried out with the Z´-LYTETM Kinase Assay kit Ser/Thr 9 Peptide substrate (Invitrogen, Grand, NY, USA) in 10 μL reaction volume containing 50 nM enzyme, 30 μM ATP and 2 μM substrate peptide. Drak2 Proteins were presented by professor Jiang-ping Wu (Laboratory of Immunology, Research Centre, CHUM, Notre Dame Hospital, Pavilion DeSève). The Drak2 kinase reaction was performed in a final assay volume of 3.4 μL using the ADP-GLO Kinase Assay Kit (Promega, Madison, WI, USA), according the ADP-GLO protocol and read on an EnVision plate reader. The recombinant PKC-epsilon, IKK, and JAK2 with N-terminal His-tag were expressed using baculovirus expression system and purified with Ni-Beads. BRAF protein was from Carna Biosciences, Inc. (Kobe Port Island, Japan). And the related kinase reactions were performed in a final assay volume of 10 μL using the related HTRF Assay Kit (Cisbio, Codolet, France). Reactions were according the HTRF protocol and read on an EnVision plate reader. All reactions were carried out in triplicate. IC 50 values (concentration at which a 50% of enzyme inhibition is shown) were derived from a nonlinear regression model (curvefit) based on sigmoidal dose response curve (variable slope) and