Identification of a novel selective and potent inhibitor of glycogen synthase kinase-3

Abstract

Glycogen synthase kinase-3 (GSK-3) is a multitasking protein kinase that regulates numerous critical cellular functions. Not surprisingly, elevated GSK-3 activity has been implicated in a host of diseases including pathological inflammation, diabetes, cancer, arthritis, asthma, bipolar disorder, and Alzheimer’s. Therefore, reagents that inhibit GSK-3 activity provide a means to investigate the role of GSK-3 in cellular physiology and pathophysiology and could become valuable therapeutics. Finding a potent inhibitor of GSK-3 that can selectively target this kinase, among over 500 protein kinases in the human genome, is a significant challenge. Thus there remains a critical need for the identification of selective inhibitors of GSK-3. In this work, we introduce a novel small organic compound, namely COB-187, which exhibits potent and highly selective inhibition of GSK-3. Specifically, this study 1) utilized a molecular screen of 414 kinase assays, representing 404 unique kinases, to reveal that COB-187 is a highly potent and selective inhibitor of GSK-3; 2) utilized a cellular assay to reveal that COB-187 decreases the phosphorylation of canonical GSK-3 substrates indicating that COB-187 inhibits cellular GSK-3 activity; and 3) reveals that a close isomer of COB-187 is also a selective and potent inhibitor of GSK-3. Taken together, these results demonstrate that we have discovered a region of chemical design space that contains novel GSK-3 inhibitors. These inhibitors will help to elucidate the intricate function of GSK-3 and can serve as a starting point for the development of potential therapeutics for diseases that involve aberrant GSK-3 activity.

INTRODUCTION

Protein kinases are enzymes that transfer a phosphate group from a high-energy donor like adenosine triphosphate (ATP) to select amino acids on target proteins (33, 53). To date, 518 protein kinases have been identified in the human genome (33). These enzymes play a critical role in signal transduction pathways and control many cellular functions including transcription, cell division, metabolism, and apoptosis (33, 53). One of the most active kinases is glycogen synthase kinase-3 (GSK-3). GSK-3 is a multitasking kinase that catalyzes the addition of phosphate groups onto serine and threonine amino acid residues (19, 25). GSK-3 was discovered over 35 years ago and named based on its initially identified role of phosphorylating and inactivating glycogen synthase, the final enzyme involved in glycogen synthesis (19, 25).

There are two isoforms of GSK-3, namely GSK-3α and GSK-3β (38), which share nearly identical sequences in their catalytic domains but differ significantly in their NH₂- and COOH-terminal regions (19, 49, 59). For instance, GSK-3α has a glycine-rich extension at its NH₂-terminal region, which results in its higher molecular mass of 51 kDa relative to GSK-3β that has a molecular mass of 47 kDa (49). GSK-3 is constitutively active (12) and plays a critical role in many cellular functions including Wnt/β-catenin signaling, insulin regulation of glucose, neural development, and cell division (7, 19, 25, 49). GSK-3 activity is optimal when tyrosine-279 of GSK-3α or tyrosine-216 of GSK-3β is phosphorylated. In contrast, phosphorylation of GSK-3β at serine-9 (Ser9) and/or GSK-3α at serine-21 (Ser21) inhibits GSK-3 activity (49, 56, 58).

Since GSK-3 was originally identified as a kinase for glycogen synthase, over 500 proteins have been suggested as potential GSK-3 substrates (12, 19) with over 40 proteins being identified as bona fide GSK-3 substrates (12). These substrates can be broadly classified into two categories namely, unprimed and primed substrates (11), with primed being more numerous than unprimed substrates (7). Primed substrates are prephosphorylated by a priming kinase at serine or threonine residues before being phosphorylated by GSK-3 (11). The prephosphorylated substrate binds to a specific binding site on GSK-3, which contains arginine-96, arginine-180, and lysine-205 residues (11). This binding aligns GSK-3 and the substrate in a way that facilitates the phosphorylation of additional serine or threonine residues on the primed substrate (11). More specifically, the prephosphorylated (primed) sequence S/T-X-X-X-S/T(P) is the most common target for which GSK-3 phosphorylates a serine or threonine, four residues in the NH₂-terminal direction from the pre-phosphorylated serine or threonine (7). This priming mechanism also provides for a clever way in which GSK-3 activity is regulated. Namely, phosphorylation of Ser21 on GSK-3α and Ser9 on GSK-3β leads to a phosphorylated motif that can bind to the GSK-3 primed substrate binding site, sterically blocking this site and reducing the activity of GSK-3 (11).

Given the plethora of known substrates for GSK-3 (19), the ubiquitous nature of GSK-3 (7, 19, 25, 49), and the fact that elevated expression/activity of GSK-3 can perturb critical signaling pathways (35), it is not surprising that GSK-3 has been implicated in a host of human diseases including inflammatory diseases (e.g., sepsis, arthritis), diabetes, cancer, asthma, neurodegenerative diseases (e.g., Alzheimer’s disease), and psychiatric disorders (e.g., bipolar mood disorder) (17, 25, 27). These observations raise two somewhat paradoxical considerations. First, GSK-3 appears to be an attractive therapeutic target due to its involvement in numerous diseases. Second, GSK-3 may not be an ideal therapeutic target since it is a highly active kinase, and thus its inhibition could cause numerous side effects. The latter point, related to adverse side effects, has been addressed by Martinez (37) where the author highlights preclinical studies and the clinical use of lithium (a GSK-3 inhibitor), which indicate that GSK-3 inhibitors can be used without significant adverse side effects. Additionally, GSK-3 inhibitors have progressed through Phase II clinical trials demonstrating that they are relatively safe (23, 62). Thus there remains an ongoing intense effort to identify and develop GSK-3 inhibitors.

GSK-3 inhibitors from either natural or synthetic sources can be divided into four categories of 1) cations, 2) ATP-competitive inhibitors, 3) non-ATP competitive inhibitors, and 4) allosteric inhibitors (17, 34, 49). Lithium is a relatively weak cation inhibitor of GSK-3, which has been used clinically for the treatment of bipolar disorders (49, 54). Lithium inhibits GSK-3 via competition with Mg²⁺ ions (48) and enhancement of Ser9/21 phosphorylation (26, 61). Even though ATP-competitive inhibitors [e.g., indirubins (29), AR-A014418 (9), and AZD-1080 (17)] are quite potent (17), the presence of the ATP binding pocket on all the protein kinases reduces the specificity of these inhibitors (2, 5). The lack of specificity of ATP-competitive inhibitors has led to an effort to identify non-ATP competitive inhibitors [e.g., manzamines (46), thiadiazolidinones (17), L803-mts (45), and L807 (31)]. For example, substrate competitive inhibitors block GSK-3 activity through binding to the GSK-3 substrate binding site rather than the ATP binding pocket (34). Allosteric inhibitors [e.g., VP0.7 (43)] bind to an allosteric binding site of GSK-3, presumably distinct from both the ATP and substrate binding site, that results in conformational changes in GSK-3, which consequently reduces GSK-3 activity (34, 43).

Currently, several inhibitors of GSK-3 are undergoing preclinical studies [e.g., 6-bromoindirubin-3′-oxime (6-BIO), hymenialdisine, kenpaullone, alsterpaullone, cazpaullone, SB415286, and L803-mts] or clinical trials [e.g., Tideglusib (49)]. Among these GSK-3 inhibitors, Tideglusib is arguably the most clinically advanced inhibitor of GSK-3. Dominguez et al. (16), provided a detailed investigation into the mechanism of action of Tideglusib and found it to be an irreversible inhibitor of GSK-3β. Further they reported that Tideglusib’s mechanism of action appears to involve Cys-199 and is specific. They did not find unequivocal evidence that Tideglusib binds covalently to GSK-3β. Tideglusib has been used in Phase II clinical trials for Alzheimer’s disease (51), progressive supranuclear palsy (57), myotonic dystrophy (57a), and autism spectrum disorders (57b). Tideglusib is currently undergoing clinical trials for congenital myotonic dystrophy (57c). While GSK-3 inhibitors have been identified and evaluated in clinical trials, no specific GSK-3 inhibitor, to date, has made it to market (12, 49). Thus GSK-3 remains an important therapeutic target, and there is a critical need for the identification and development of novel GSK-3 inhibitors.

Our group had previously worked with a compound termed C10 (15, 22). In an attempt to find more effective compounds, we generated a plethora of small organic molecules and investigated their unique pharmacological properties in a series of assays, including molecular-based kinase screens. This effort led to the identification of compounds, described herein, that are potent and selective inhibitors of GSK-3. These compounds could be the starting point for the development of therapeutics for diseases involving aberrant GSK-3 activity and may provide insights into the mechanism of action of GSK-3 and GSK-3 inhibitors.

MATERIALS AND METHODS

Synthesis of novel compounds.

Compounds COB-152 and COB-187 were prepared by the reaction of the appropriate amine, α-bromoacetophenone, and carbon disulfide (21). COB-197 was prepared by the reaction of the appropriate amine, 2-bromo-1-(2,5-dimethoxyphenyl)ethan-1-one, and carbon disulfide (21). Compounds were purified to >95% purity as determined by HPLC. A schematic of synthetic routes for COB-152, COB-187, and COB-197 is presented in Supplemental Figure S1 (Supplemental Material is available at https://doi.org/10.6084/m9.figshare.9750845). Tideglusib was purchased from Sigma-Aldrich (St. Louis, MO).

Kinase assay.

The kinase screens, utilizing up to 414 FRET-based kinase assays, were performed via contract research with Life Technologies/Thermo Fisher Scientific (Waltham, MA). The complete list of tested kinases and kinase assays used in this study are provided in Supplemental Table S1 (https://doi.org/10.6084/m9.figshare.9546998). The kinase screen consisted of Z’-LYTE, Adapta, and Lantha Eu assays. The Z’-LYTE and Adapta assays measure kinase activity via detection of phosphorylation of the peptide and ADP, respectively, while the LanthaScreen Eu assay measures the binding of a tracer to the ATP site of the kinase.

To determine IC₅₀ values for Tideglusib, COB-187, and COB-152, inhibition of GSK-3 using the Z’-LYTE GSK-3 kinase assay was performed with varying levels of each compound and 10 µM, 0.64 nM, and 0.87 nM of ATP, GSK-3α, and GSK-3β, respectively. A peptide, based on human glycogen synthase I containing Ser641 (16), was the substrate for the GSK-3 kinase assay. Note that both kinase and development reactions were carried out for 60 min at room temperature in 384-well plates. The resulting inhibition versus compound concentration data were used to determine the IC₅₀ for each compound. This data (see Fig. 2) was generated via contract with Life Technologies/Thermo Fisher Scientific.

Fig. 2.

COB-187 and COB-152 inhibit the activity of GSK-3 with greater potency than Tideglusib. The inhibitory effect of Tideglusib (A), COB-187 (B), and COB-152 (C) on GSK-3α and GSK-3β was separately evaluated in a non-cell-based Z’-LYTE molecular assay. The IC₅₀ for Tideglusib, COB-187, and COB-152 was 908, 22, and 77 nM for GSK-3α, and 502, 11, and 132 nM for GSK-3β, respectively. Results are the average of two replicates. Error bars represent the SE.

Cell culture and treatment.

The RAW 264.7 mouse macrophage, human embryonic kidney (HEK293), and THP-1 human monocytic cell lines were purchased from American Type Culture Collection (ATCC, Manassas, VA) and cultured according to the manufacturer’s protocol. Briefly, RAW 264.7 cells were cultured in Dulbecco’s modified eagle’s medium (DMEM) (ATCC) supplemented with 10% nonheat inactivated fetal bovine serum (FBS; ATCC) and 5% penicillin/streptomycin (Lonza, Walkersville, MD), at 37°C and 5.0% CO₂. HEK293 cells were cultured in DMEM medium (ATCC) containing 10% nonheat inactivated FBS (ATCC) at 37°C and 5.0% CO₂. For transfection, 50–60% confluent HEK-293 cells cultured in 0.1% gelatin-coated six-well plates (Sigma-Aldrich) were transfected with 2.5 µg of the longest human tau [tau441; NM_005910, GFP-tagged human microtubule-associated protein tau (MAPT); OriGene Technologies; Rockville, MD] using lipofectamine 2000 (Thermo Fisher Scientific) for 24 h according to the manufacturer’s protocol. The expression of tau was then evaluated by Western blotting. THP-1 cells were cultured in RPMI-1640 medium (ATCC) supplemented with 10% nonheat inactivated FBS (ATCC), 0.05 mM of 2-mercaptoethanol (Millipore, Billerica, MA), and 5% penicillin/streptomycin (Lonza), at 37°C and 5.0% CO₂. To differentiate THP-1 monocytes to the macrophage phenotype, THP-1 cells were stimulated with 50 ng/mL of phorbol-12-myristate-13-acetate (PMA; Sigma-Aldrich) for 48 h. Afterwards, differentiated cells were washed with PMA-free fresh medium twice and incubated for 24 more hours with PMA-free medium at 37°C and 5.0% CO₂. THP-1 cells differentiated in this manner are referred to as THP-1 macrophages throughout this paper.

To confirm the differentiation of THP-1 monocytes to macrophages, as indicated by an increase in macrophage molecular markers (CD44 and CD11b) and a decrease in monocyte molecular marker (CD15s), the expression levels of these markers on PMA-stimulated THP-1 cells and unstimulated THP-1 cells were evaluated using flow cytometric analysis (41). In brief, PMA stimulated THP-1 cells were harvested with TrypLE Express (Gibco, Gaithersburg, MD) and washed, along with unstimulated THP-1 cells, in Dulbecco’s phosphate-buffered saline with Ca²⁺ and Mg²⁺ (DPBS⁺; Thermo Fisher Scientific) containing 2% FBS (ATCC). Washed cells were aliquoted (~2 × 10⁵/ aliquot) and treated with a mouse anti-human CD44 monoclonal antibody (mAb) [515; IgG₁; cat. no. 550988; BD Bioscience; San Jose, CA; Research Resource Identifier (RRID): AB_393999], mouse anti-human CD15s mAb (CSLEX1; IgM; cat. no. 551344,; BD Bioscience; RRID: AB_394156), rat anti-mouse CD11b FITC-conjugated mAb (M1/70; IgG; cat. no. 11-0112-82; eBioscience; San Diego, CA; RRID: AB_464935), or appropriate isotype controls. Mouse IgG1к (cat. no. 550878; BD Bioscience; RRID: AB_10052281), mouse IgM (cat. no. 0101-01; Southern Biotech; Birmingham, AL; RRID: AB_2629437), and FITC-conjugated rat IgG2bк (cat. no. 11-4031-82; eBioscience; RRID: AB_470004) were isotype controls for the mAbs to CD44, CD15s, and CD11b, respectively. Cells treated with mAbs to CD44, CD15s or the appropriate isotype controls were subsequently treated with FITC-conjugated goat anti-mouse IgG (H⁺L) (cat. no. 1031-02; RRID: AB_2794304) or FITC-conjugated rat anti-mouse IgM (cat. no. 1140-02; RRID: AB_2794627) secondary antibodies from Southern Biotech. All antibody treatments were with diluted antibodies (10 µg/mL) in complete growth medium containing 2% FBS. Incubations with antibodies were for 30 min on ice followed by washes. After the final wash, the THP-1 cells were fixed in DPBS⁺ with 1% paraformaldehyde (Electron Microscopy Sciences, Hatfield, PA) and analyzed using a FACSAria Special Order Research Product flow cytometer/cell sorter (BD Biosciences).

For compound treatment, a 200-mM stock solution of COB-187 in in dimethyl sulfoxide (DMSO) (Sigma-Aldrich) was diluted in appropriate culture medium to the final desired concentrations. The final solution contained 0.1% DMSO (vol/vol). Cells were treated with fresh medium, fresh medium containing 0.1% DMSO (solvent control), or fresh medium containing different concentrations of COB-187 for 5 h.

Cell metabolic assay (MTS).

Undifferentiated THP-1 cells were placed on 96-well plates (4 × 10⁵ cells/mL) and differentiated into macrophages as described above. RAW 264.7 were seeded on 96-well plates (2 ×10⁵ cells/well) and incubated at 37°C and 5.0% CO₂ until they reach 60–70% confluency. HEK-293 cells were cultured on 0.1% gelatin-coated 96-well plates until they reached 50–60% confluency. Cells were then treated with varying concentrations of COB-187 and incubated at 37°C and 5.0% CO₂ for 5 h. [Note that the upper limit concentration of COB-187 we could test was 200 μM due to solubility issues.] Subsequent to the 5-h incubation of THP-1 macrophages and RAW 264.7 cells with COB-187, the old medium was replaced with 100 µl of fresh medium and then 20 µl of the CellTiter 96 AQ_ueous One Solution Cell Proliferation Assay solution (Promega, Madison, WI) was added to each well. Subsequent to the 5-h incubation of HEK-293 cells with COB-187, 20 µl of the Cell Proliferation Assay solution was directly added to the old medium to avoid detachment of the cells from surface. The treated THP-1 macrophages, RAW 264.7, and HEK-293 cells were incubated for 3–4 h at 37°C and 5.0% CO₂. Subsequently, the absorbance of the solution within each well at 490 nm was determined using a Synergy HT Multi-Mode Microplate Reader (BioTek, Winooski, VT) (40). Note that any background signal resulting from the presence of compound in the old medium was deducted from the HEK-293 absorbance signal. Three independent experiments were performed in triplicate. Presented results are normalized to THP-1 macrophages treated with 0.1% DMSO control.

Western blotting.

Differentiated THP-1 cells (2 × 10⁶ cells/well of 6-well plate) and 70–80% confluent RAW 264.7 macrophages were treated with COB-187 at varying concentrations for 5 h. The effect of the treatments on the phosphorylation state and total level of β-catenin and glycogen synthase were assessed by Western blotting (13, 36). In brief, the whole cell lysates were collected as previously described (36) and the total protein content quantified using the micro-BCA protein assay kit (Thermo Fisher Scientific) and Nanodrop 2000 Micro-volume UV-Vis Spectrophotometer (Thermo Fisher Scientific). Whole cell lysates, containing 20 µg of total protein, were then electrophoresed on 4–12% Bis-Tris SDS-PAGE gradient gels using the NuPage system (Invitrogen, Carlsbad, CA) and transferred to nitrocellulose membranes (Invitrogen). Odyssey blocking buffer (LI-COR Biosciences, Lincoln, NE) was used for blocking the membranes and to dilute the antibodies. Rabbit anti-phospho-β-catenin (Ser33/37/Thr41; cat. no. 9561; RRID: AB_331729) polyclonal antibody (pAb) was used to detect phosphorylation of β-catenin at serines 33 or 37 or threonine 41 of RAW 264.7 and THP-1 macrophages. Blots were also probed with rabbit anti-β-catenin pAb (cat. no. 9562; RRID: AB_331149) and rabbit anti-β-actin (pAb, internal control; cat. no. 4967; RRID: AB_330288). Phosphorylation of glycogen synthase in THP-1 macrophages was examined using a rabbit anti-phospho-glycogen synthase (Ser641; cat. no. 3891; RRID: AB_2116390) pAb. Blots were also probed with rabbit anti-glycogen synthase (15B1; cat. no. 3886; RRID: AB_2116392) mAb and rabbit anti-β-actin (internal control). All primary antibodies were purchased from Cell Signaling Technology (Danvers, MA). IRDye 680LT (cat. no. P/N 925-68021; RRID: AB_2713919) and 800CW (cat. no. 925-32211; RRID: AB_2651127) goat anti-rabbit IgG (H⁺L) polyclonal (LI-COR Biosciences) secondary antibodies were used for primary antibody detection, and the LI-COR Odyssey Infrared Imaging System was used for visualization and quantification of signals (13). Three independent biological samples were prepared and analyzed. Total protein levels for glycogen synthase and β-catenin were ratioed to β-actin to control for sample variability. Phosphorylated glycogen synthase and β-catenin levels were ratioed to total glycogen synthase and β-catenin levels, respectively. These ratios were used in the statistical analyses.

To evaluate the effect of the COB-187 treatments on the total level of tau and its phosphorylation state at Ser396 and 404, HEK-293 cells were transfected with the tau expression vector described above for 24 h in six-well plates, then treated with COB-187 at varying concentrations for 5 h, and finally assessed by Western blotting. Mouse anti-tau mAb (TAU-5; cat. no. ab80579; Abcam; RRID: AB_1603723), rabbit anti-phospho-tau mAb (Ser396; cat. no. ab109390; Abcam; RRID: AB_10860822), rabbit anti-phospho-tau mAb (Ser404; cat. no. ab92676; Abcam; RRID: AB_10561457), and rabbit anti-β-actin pAb (internal control; cat. no. 4967; Cell Signaling; RRID: AB_330288) were used as primary antibodies. IRDye 680LT goat anti-mouse IgG (H⁺L) (cat. no. P/N 925-68020; LI-COR Biosciences; RRID: AB_2687826) and 800CW goat anti-rabbit IgG (H⁺L) (cat. no. 925-32211; LI-COR Biosciences; RRID: AB_2651127) polyclonal secondary antibodies were used for primary antibody detection. Three independent biological samples were prepared and analyzed.

Immunocytochemistry.

Whole cell staining was conducted on THP-1 macrophages and RAW 264.7 cells to investigate the total level of β-catenin and its localization to the nucleus. In brief, cells were cultured on Lab-Tek II chamber slides (Thermo Fisher Scientific) and then treated with different concentrations of COB-187 for 5 h. Treated cells were fixed on the surface using 4% paraformaldehyde (Electron Microscopy Sciences) in PBS for 10 min and permeabilized with 0.2% (vol/vol) Triton-X-100 (Sigma-Aldrich) for another 10 min. Afterwards, cells were blocked to reduce background fluorescence, in PBS containing 0.3M glycine (Fisher Scientific, Hampton, NH), 10% (vol/vol) normal goat serum (Abcam, Cambridge, UK), 0.5% (wt/vol) BSA (Sigma-Aldrich), and 0.1% Tween-20 (Thermo Fisher Scientific). Blocked cells were incubated with rabbit anti-β-catenin mAb (E247; 1:150; Abcam) overnight at 4°C and then with goat anti-rabbit IgG H&L (Alexa Flour 488) secondary antibody (1:750; Abcam) for 1 h at room temperature. Monoclonal rabbit IgG (Abcam) was used as an isotype control. Note that cells were washed on a shaker three times with PBS (5 min each) after labeling with primary and secondary antibodies. Finally, cells were mounted in proLong gold antifade mountant with DAPI (Life Technologies, Carlsbad, CA) and evaluated with a Zeiss LSM 510 confocal microscope (Carl Zeiss MicroImaging, Thornwood, NY). Two independent experiments were performed.

Reverse transcription quantitative real-time polymerase chain reaction.

THP-1 and RAW 264.7 macrophages were cultured in six-well tissue culture plates (2 × 10⁶ cells/well) and treated with different concentrations of COB-187 for 5 h. Treated THP-1 and RAW 264.7 macrophages were then harvested, and the level of β-catenin mRNA was quantified via reverse transcription quantitative real-time polymerase chain reaction (RT-qPCR). In brief, THP-1 and RAW macrophage lysates were homogenized using QIAshredder microcentrifuge spin-column homogenizers (Qiagen, Valencia, CA), and RNA was isolated using the RNeasy Mini Kit (Qiagen). Afterwards, RNA was treated with DNase to remove any possible contaminating genomic DNA using the RNase-free DNase Set (Qiagen) and the purity and concentration of mRNA samples were quantified using the Nanodrop 2000 Micro-Volume UV-Vis Spectrophotometer (Thermo Fisher Scientific). cDNA was subsequently synthesized with the High Capacity cDNA Reverse Transcription Kit (Applied Biosystems, Foster City, CA). RT-qPCR was then performed using Taqman Gene Expression Assays CTNNB1 (human, Hs00355045_m1, FAM-MGB; Thermo Fisher Scientific), ctnnb1 (mouse, Mm00483039_m1, FAM-MGB; Thermo Fisher Scientific), and CCND1 (human, Hs00765553_m1, FAM-MGB; Thermo Fisher Scientific), Taqman Gene Expression Master Mix (Applied Biosystems), and the StepOnePlus Real-Time PCR System (Applied Biosystems). Gene expression data for THP-1 macrophages and RAW 264.7 cells were normalized to the housekeeping gene ACTB (human, Hs99999903_m1, VIC-MGB) and Hprt1 (mouse, Mm00446968_m1, VIC-MGB) from Thermo Fisher Scientific, respectively and the ∆∆CT method was used for gene expression comparisons. Three independent experiments were performed in triplicate.

RESULTS

Structures of compounds used in this study.

In the present study, Tideglusib [4-benzyl-2-(naphthalen-1-yl)-1,2,4-thiadiazolidine-3,5-dione], which has been and is currently being evaluated in clinical trials, was utilized along with the novel compounds, COB-152 [4-hydroxy-4-phenyl-3-(pyridin-3-ylmethyl)thiazolidine-2-thione], COB-187 [4-hydroxy-4-phenyl-3-(pyridin-4-ylmethyl)thiazolidine-2-thione], and COB-197 [4-(2,5-dimethoxyphenyl)-4-hydroxy-3-(pyridin-3-ylmethyl)thiazolidine-2-thione] (Fig. 1). As shown in Fig. 1, COB-152, COB-187, and COB-197 contain a five-member heterocyclic ring with a thione at C₂, a pyridine substituted methyl at N₃, and a phenyl and hydroxyl at C₄. The novel compounds compared with Tideglusib are distinctly different in terms of molecular structure. Specifically, the differing orientation of the two aromatic substituents on the five-membered ring (attached on adjacent atoms for the novel compounds and on atoms separated by another atom for Tideglusib) leads to differing molecular shapes, the five-member ring of Tideglusib is more planar, and the novel compounds have a hydroxyl group on one of the Sp3-hybridized carbons, which, due to the shape of the five-membered ring, points up at ~90° from the ring while Tideglusib lacks any such substituent. Note that COB-187 and COB-152 differ only in the position of the N in the pyridine ring (COB-152 meta and COB-187 para) and COB-197 is identical to COB-152 except for the two methoxy substitutions on the phenyl ring of COB-197. COB-152, COB-187, and COB-197 were synthesized as described in materials and methods.

Fig. 1.

The chemical structures of compounds used in this study. Tideglusib is a non-ATP competitive glycogen synthase kinase-3 (GSK-3) inhibitor (16) that has been and is being used in clinical trials. COB-152, COB-187, and COB-197 are the novel compounds introduced in this study. COB-152 and COB-187 differ only in the position of the N in the pyridine ring. COB-197 is identical to COB-152 except for the 2 methoxy substitutions on the phenyl ring of COB-197. Molecular mass of Tideglusib, COB-152, COB-187, and COB-197 are 334.39, 302.41, 302.41, and 364.48 g/mol, respectively.

Identification of a structurally related set of compounds that appear to inhibit GSK-3 with a high level of selectivity.

During our ongoing compound development program, we identified several structurally related compounds (COB-152, COB-187, and COB-197; Fig. 1) that appeared to inhibit signal transduction. Thus we decided to screen these compounds for kinase activity utilizing a robust set of 54 kinase assays. The results of this screen are given in Table 1. [Note that we fully screened COB-152 on all the 317 kinase assays available at the time. The results for the full COB-152 screen are given in Supplemental Table S1 (https://doi.org/10.6084/m9.figshare.9546998) and discussed in the next section; the results with the 54 kinases are discussed here.] Remarkably, all three of these compounds inhibited both GSK-3α and GSK-3β activity with COB-187 appearing to be the most potent and COB-197 the least (Table 1). The inhibition was highly selective (Table 1). Indeed, only two other kinases, besides GSK-3α and GSK-3β, were inhibited by ≥40% (Table 1). Specifically, MAPKAPK5 (PRAK) was inhibited by both COB-187 and COB-152 but not COB-197 and KDR (VEGFR2) was inhibited by COB-152 but not by either COB-187 nor COB-197. Thus we began to focus on the effect of these compounds on GSK-3. Given that COB-197 appeared to be less potent than COB-152 and COB-187 (Table 1), we proceeded with these latter two compounds.

Table 1.

COB-152, COB-187, and COB-197 were utilized in a kinase screen containing 54 assays

			% Inhibition Caused By
Kinase	Assay	ATP	COB-152	COB-187	COB-197
GSK3A (GSK-3α)	Z	Km	80 ± 1.7	89 ± 1.1	59 ± 1.5
GSK3B (GSK-3β)	Z	Km	74 ± 0.2	90 ± 0.7	50 ± 0.0
ABL1 E255K	Z	Km	3 ± 0.6	6 ± 0.4	5 ± 0.8
ACVR2B	L	–	31 ± 7.1	10 ± 9.6	9 ± 7.4
AKT2 (PKB-β)	Z	Km	9 ± 0.4	0 ± 0.0	−1 ± 1.4
AKT3 (PKB-γ)	Z	Km	7 ± 0.0	3 ± 0.8	2 ± 0.2
AURKA (Aurora A)	Z	Km	17 ± 0.3	5 ± 2.7	2 ± 0.4
BRSK1 (SAD1)	Z	Km	12 ± 2.8	0 ± 0.9	9 ± 6.6
CAMK1 (CaMK1)	A	10 µM	1 ± 9.9	5 ± 8.5	−14 ± 9.7
CAMK2B (CaMKII-β)	Z	Km	9 ± 5.9	17 ± 0.5	16 ± 0.6
CAMK2D (CaMKII-δ)	Z	Km	13 ± 1.7	4 ± 0.9	7 ± 5.3
CDK5/p35	Z	Km	7 ± 0.7	3 ± 0.5	5 ± 4.0
CSF1R (FMS)	Z	Km	3 ± 4.1	1 ± 8.6	−3 ± 0.7
DYRK3	Z	Km	6 ± 1.9	1 ± 2.0	1 ± 1.0
EGFR (ErbB1) T790M L858R	Z	Km	4 ± 2.5	3 ± 0.5	1 ± 1.5
FER	Z	Km	−2 ± 0.8	−4 ± 0.1	1 ± 0.4
FGR	Z	Km	27 ± 2.2	9 ± 1.7	7 ± 2.7
FLT3 D835Y	Z	Km	7 ± 1.2	8 ± 0.5	7 ± 1.9
GRK4	Z	Km	19 ± 0.8	18 ± 1.2	17 ± 1.8
IKBKB (IKK-β)	Z	Km	11 ± 2.6	1 ± 0.5	−2 ± 1.4
IKBKE (IKK-ε)	Z	Km	4 ± 0.3	0 ± 1.9	1 ± 4.3
JAK3	Z	Km	12 ± 0.5	7 ± 0.4	7 ± 2.1
KDR (VEGFR2)	Z	Km	64 ± 1.1	11 ± 2.1	32 ± 1.8
MAP2K2 (MEK2)	L	–	27 ± 4.7	9 ± 2.0	8 ± 0.8
MAP3K14 (NIK)	L	–	17 ± 2.4	9 ± 6.1	18 ± 11.0
MAP4K2 (GCK)	Z	Km	7 ± 0.7	7 ± 2.0	14 ± 5.2
MAP4K4 (HGK)	Z	Km	6 ± 2.6	12 ± 3.5	16 ± 4.3
MAPK14 (p38-α)	Z	100 µM	27 ± 0.2	17 ± 1.0	19 ± 0.4
MAPK8 (JNK1)	L	–	22 ± 1.4	10 ± 1.0	5 ± 0.6
MAPKAPK2	Z	Km	2 ± 0.2	5 ± 0.5	7 ± 1.9
MAPKAPK3	Z	Km	14 ± 1.8	7 ± 0.8	5 ± 1.9
MAPKAPK5 (PRAK)	Z	Km	57 ± 0.2	55 ± 3.4	21 ± 0.9
MELK	Z	Km	15 ± 0.9	4 ± 4.8	11 ± 0.8
MINK1	Z	Km	12 ± 1.5	6 ± 4.4	−1 ± 2.8
NEK1	Z	Km	23 ± 2.0	0 ± 8.5	1 ± 0.1
PASK	Z	Km	2 ± 2.0	−5 ± 1.3	−3 ± 0.5
PIK3CD/PIK3R1 (p110-δ/p85-α)	A	Km	−14 ± 3.8	−5 ± 7.4	−3 ± 5.4
PLK3	Z	Km	10 ± 0.0	9 ± 14.2	13 ± 0.6
PRKACA (PKA)	Z	Km	−1 ± 0.1	−3 ± 0.0	−1 ± 1.2
PRKCA (PKC-α)	Z	Km	4 ± 4.0	−4 ± 14.7	−1 ± 2.9
PRKCB2 (PKC-βII)	Z	Km	12 ± 5.8	3 ± 2.4	−1 ± 2.3
PRKCG (PKC-γ)	Z	Km	4 ± 3.0	12 ± 2.5	8 ± 6.2
PRKCH (PKC-η)	Z	Km	8 ± 0.7	1 ± 4.0	3 ± 0.6
PRKCQ (PKC-θ)	Z	Km	11 ± 0.5	−8 ± 15.9	−6 ± 7.4
PRKG1	Z	Km	4 ± 1.0	3 ± 0.0	3 ± 1.8
PRKX	Z	Km	8 ± 0.3	1 ± 2.2	4 ± 10.3
ROCK2	Z	Km	1 ± 3.6	6 ± 0.7	7 ± 2.7
RPS6KA3 (RSK2)	Z	Km	11 ± 0.5	5 ± 0.7	8 ± 3.3
RPS6KA6 (RSK4)	Z	Km	21 ± 1.3	8 ± 4.6	3 ± 0.5
SGK (SGK1)	Z	Km	12 ± 2.7	7 ± 0.6	5 ± 2.2
SGK2	Z	Km	15 ± 1.1	2 ± 1.6	1 ± 1.2
SGKL (SGK3)	Z	Km	11 ± 4.7	4 ± 3.2	8 ± 9.2
SPHK1	A	Km	17 ± 2.1	20 ± 4.3	18 ± 2.2
WNK2	L	–	10 ± 6.7	4 ± 10.7	6 ± 4.7

Results given are the average of two data points ± SD. Cells shown in boldface in the results columns demarcate inhibition ≥80%; underlined cells demarcate inhibition between 79% and 40%; and unmarked cells demarcate inhibition <40%. Each compound inhibited GSK-3α and GSK-3β with COB-187 appearing to be the most potent and COB-197 the least. Only 2 other kinases were inhibited by ≥40%; specifically, MAPKAPK5 (PRAK) was inhibited by both COB-187 and COB-152, but not COB-197, and KDR (VEGFR2) was inhibited by COB-152, but not by either COB-187 or COB-197. The Kinase, Assay, and ATP columns give the details of the assay. Z indicates Z′-LYTE activity assay, A indicates Adapta activity assay, and L indicates Lantha binding assay. ATP gives the concentration of ATP as either K_m apparent, a concentration given in μM, or a dash, which indicates that no ATP is used in the Lantha assay. Concentrations of compounds were 2 μm COB-152, 1 μm COB-187, and 2.5 μm COB-197. [The concentrations were varied based on potency observed in the compound development assay, i.e., COB-187 was the most potent and COB-197 the least.] Note that COB-152 was screened using all of the kinases available at the time (a total of 317), and the complete results are given in Supplemental Table S1 (Supplemental Material is available at https://doi.org/10.6084/m9.figshare.9546998). The COB-152 data from that screen were also used for the comparison with Tideglusib given in Table 2.

In molecular assays, COB-187 and COB-152 inhibit GSK-3 activity with a greater selectivity and potency than Tideglusib.

We next compared the potency of COB-187 and COB-152 to Tideglusib. To make this comparison, the inhibitory effect of Tideglusib, COB-187, and COB-152 on GSK-3α and GSK-3β kinase activity was determined in the Z’-LYTE assay. The results are shown in Fig. 2 from which IC₅₀ values for each compound were determined. The IC₅₀ values for Tideglusib, COB-187, and COB-152 on GSK-3α were 908, 22, and 77 nM, and on GSK-3β were 502, 11, and 132 nM, respectively. Thus COB-187 and COB-152 appear to be more potent than Tideglusib at inhibiting the activity of both GSK-3 isoforms.

With this finding in hand, we investigated the selectivity of COB-187 and COB-152 relative to Tideglusib. Tideglusib and COB-187 were each evaluated using 414 kinase assays, which represents 404 unique kinases. [Note that from the time we had screened COB-152 to the time we screened Tideglusib and COB-187, additional kinase assays had become available.] Table 2 lists all the kinases, except for GSK-3α and GSK-3β, that Tideglusib, COB-187, and/or COB-152 were observed to inhibit by ≥40%. As shown, Tideglusib inhibited 50 kinases at ≥40%. In contrast, COB-187 inhibited MAPKAPK5 (PRAK) (as reported in Table 1) and MAPK15 (ERK7) while COB-152 did not inhibit any kinases beyond those identified in Table 1, namely KDR (VEGFR2) and MAPKAPK5 (PRAK). [Note that COB-152 was not tested in the MAPK15 (ERK7) assay.] The complete results of the COB-187 screen, along with the previously mentioned COB-152 screening results, are provided in Supplemental Table S1 (see https://doi.org/10.6084/m9.figshare.9546998). Thus it appears that the newly identified compounds are significantly more selective than Tideglusib. Given these observations, we probed the effect of COB-187 on cellular GSK-3 activity. COB-187 was chosen for these studies since it appeared to be more potent than COB-152 (Fig. 2).

Table 2.

List of kinase assays inhibited ≥40% by Tideglusib, COB-187, and/or COB-152

			% Inhibition Caused By
Kinase	Assay	ATP	Tideglusib	COB-187	COB-152
ABL1 E255K	Z	Km	86 ± 0.8	1 ± 4.0	3 ± 0.6
ABL1 Y253F	Z	Km	64 ± 0.3	3 ± 1.2	5 ± 2.0
AMPK A1/B1/G1	Z	Km	67 ± 5.1*	3 ± 0.7	14 ± 0.9
AMPK A2/B1/G1	Z	Km	53 ± 3.7*	6 ± 0.4	16 ± 1.1
BLK	Z	Km	53 ± 2.8	5 ± 4.4	−6 ± 2.6
CAMK1 (CaMK1)	A	10 µM	99 ± 0.2	−2 ± 16.2	1 ± 9.9
CAMK2A (CaMKII-α)	Z	Km	90 ± 1.6	−1 ± 5.9	14 ± 6.7
CAMK2D (CaMKII-δ)	Z	Km	62 ± 1.1	9 ± 3.5	13 ± 1.7
CAMK2G (CaMKII-γ)	L	–	67 ± 9.1	−1 ± 10.6	NP
CSF1R (FMS)	Z	Km	85 ± 1.0	21 ± 2.1	3 ± 4.1
DAPK1	A	Km	48 ± 3.8	−10 ± 8.7	−4 ± 5.0
DDR2 N456S	L	–	70 ± 7.3	−4 ± 2.4	NP
EPHB4	Z	Km	64 ± 2.2	−3 ± 0.4	5 ± 3.2
FER	Z	Km	74 ± 0.3	−7 ± 1.4	−2 ± 0.8
FES (FPS)	Z	Km	42 ± 2.2	2 ± 1.3	0 ± 1.3
FGFR1 V561M	L	–	69 ± 6.1	−12 ± 17.0	NP
FGR	Z	Km	98 ± 0.1	16 ± 4.6	27 ± 2.2
FLT3	Z	Km	82 ± 3.4	8 ± 12.6	−5 ± 0.5
FLT3 D835Y	Z	Km	92 ± 0.1	14 ± 0.4	7 ± 1.2
FYN	Z	Km	51 ± 9.2	9 ± 1.8	8 ± 5.5
GRK4	Z	Km	96 ± 1.4	7 ± 2.2	19 ± 0.8
IKBKB (IKK-β)	Z	Km	77 ± 4.3	−4 ± 1.2	11 ± 2.6
JAK2	Z	Km	85 ± 3.2	8 ± 2.8	7 ± 0.1
JAK2 JH1 JH2	Z	Km	61 ± 0.6	6 ± 1.3	−2 ± 3.8
JAK3	Z	Km	75 ± 3.4	14 ± 4.2	12 ± 0.5
KDR (VEGFR2)	Z	Km	31 ± 5.5	17 ± 0.0	64 ± 1.1
LIMK1	L	–	46 ± 2.8	3 ± 3.8	7 ± 13.0
MAP2K1 (MEK1) S218D S222D	L	–	100 ± 1.1	8 ± 2.2	29 ± 1.2
MAP2K6 (MKK6)	L	–	67 ± 1.7	14 ± 0.2	29 ± 2.1
MAP4K4 (HGK)	Z	Km	79 ± 3.1	8 ± 3.5	6 ± 2.6
MAPK14 (p38-α)	Z	100 µM	98 ± 0.8	23 ± 4.7	27 ± 0.2
MAPK15 (ERK7)	L	–	7 ± 0.5	42 ± 0.4	NP
MAPK8 (JNK1)	L	–	96 ± 0.4	17 ± 6.0	22 ± 1.4
MAPK9 (JNK2)	L	–	75 ± 1.8	5 ± 1.5	20 ± 1.9
MAPKAPK2	Z	Km	97 ± 2.5	−1 ± 0.5	2 ± 0.2
MAPKAPK3	Z	Km	100 ± 0.4	9 ± 0.3	14 ± 1.8
MAPKAPK5 (PRAK)	Z	Km	50 ± 7.2	64 ± 0.3	57 ± 0.2
MELK	Z	Km	97 ± 0.5	23 ± 3.4	15 ± 0.9
MKNK2 (MNK2)	L	–	57 ± 3.8	3 ± 0.1	7 ± 5.2
MYLK2 (skMLCK)	Z	Km	70 ± 4.8	3 ± 0.1	6 ± 3.0
PIK3CA/PIK3R1 (p110-α/p85-α)	A	Km	62 ± 0.3	−14 ± 5.4	2 ± 8.2
PIK3CD/PIK3R1 (p110-δ/p85-α)	A	Km	66 ± 3.6	−8 ± 11.5	−14 ± 3.8
PLK3	Z	Km	98 ± 1.3	10 ± 1.1	10 ± 0.0
PRKCQ (PKC-θ)	Z	Km	71 ± 9.7	4 ± 14.0	11 ± 0.5
PRKD2 (PKD2)	Z	Km	88 ± 0.5	−2 ± 5.7	17 ± 3.9
ROCK2	Z	Km	76 ± 0.3	11 ± 5.5	1 ± 3.6
RPS6KA6 (RSK4)	Z	Km	91 ± 0.1	5 ± 0.1	21 ± 1.3
SPHK1	A	Km	76 ± 0.3	12 ± 1.2	17 ± 2.1
SRC N1	Z	Km	72 ± 5.4	6 ± 2.3	6 ± 1.1
SYK	Z	Km	53 ± 4.2	1 ± 2.7	2 ± 1.1
TEC	L	–	52 ± 2.9	5 ± 0.2	14 ± 8.9
YES1	Z	Km	41 ± 2.0	9 ± 1.2	10 ± 1.4

Results given are the average of two data points ± SD. Cells shown in boldface in the results columns demarcate inhibition ≥80%; underlined cells demarcate inhibition between 79% and 40%; and unmarked cells demarcate inhibition <40%. Cells labeled with NP in the COB-152 column indicate that the assay was not performed. Tideglusib and COB-187 were each evaluated using 414 kinase assays, which represents 404 unique kinases, and COB-152 was screened using a total of 317 kinase assays. All compounds were used at a concentration of 2 µM. Listed are all of the kinases, except GSK-3α and GSK-3β, that Tideglusib, COB-187, and/or COB-152 inhibited by ≥40%. Tidegluib inhibited 50 kinases at ≥40%. In contrast, COB-187 inhibited MAPKAPK5 (PRAK) (also reported in Table 1) and MAPK15 (ERK7), while COB-152 did not inhibit any kinases beyond those identified in Table 1, namely KDR (VEGFR2) and MAPKAPK5 (PRAK). [Note that COB-152 was not tested in the MAPK15 (ERK7) assay.] The complete results of the screens for COB-187 and COB-152 are provided in Supplemental Table S1 (Supplemental Material is available at https://doi.org/10.6084/m9.figshare.9546998). The Kinase, Assay, and ATP columns give the details of the assay. Z indicates Z′-LYTE activity assay, A indicates Adapta activity assay, and L indicates Lantha binding assay. ATP gives the concentration of ATP as either K_m apparent, a concentration given in μM, or a dash, which indicates that no ATP is used in the Lantha assay.

^*These cells in the Tideglusib column demarcate possible interference of the test compound with the assay. Such interference can suppress the inhibitory effect of the compound (i.e., the actual inhibitory effect may be higher than that reported).

Five-hour treatment with COB-187 has a limited, if any, effect on the metabolic activity of THP-1 macrophages.

To investigate the effect of COB-187 on cellular GSK-3 function we chose to use human THP-1 cells differentiated into the macrophage phenotype by PMA, since this cell line has been used by previous investigators for this purpose (6, 39, 52). To differentiate THP-1 cells, human THP-1 monocytic cells were treated with 50 ng/mL PMA for 48 h followed by 24-h incubation in PMA-free culture medium (10, 44). Subsequently, the expression of macrophage and monocyte markers on the PMA-treated THP-1 cells, relative to untreated THP-1 cells, was investigated using flow cytometric analysis. Elevated expression of CD11b and CD44 macrophage molecular markers and lower expression of CD15s monocyte molecular marker on PMA-treated THP-1 cells, relative to untreated THP-1 cells, confirmed the differentiation of THP-1 monocytes to a macrophage phenotype (data not shown).

To determine whether any effect observed due to treatment with COB-187 in the subsequent cell-based assays could be due to nonspecific toxicity, the effect of 5-h COB-187 treatment on the metabolic activity of the THP-1 macrophages was determined using an MTS assay. As shown in Fig. 3, COB-187 had no significant effect on the metabolic activity of the THP-1 macrophages at COB-187 concentrations at or below 200 µM. Note that the subsequent THP-1 macrophage assays performed in the remainder of this study were conducted at COB-187 concentrations ≤50 µM.

Fig. 3.

COB-187 has a limited, if any, effect on the metabolic activity of the PMA-differentiated THP-1 cells. THP-1 macrophages were treated with various concentrations of COB-187 for 5 h and the effect of the compound on metabolic activity was determined using an MTS assay. Results are the average of three separate experiments where the value at each dose was determined in triplicate. Error bars represent the SE. Results were analyzed using a one-way ANOVA coupled with post hoc Dunnett’s test. Note that all values are relative to 0.1% DMSO control.

COB-187 reduces the phosphorylation of canonical GSK-3 substrates.

One of the hallmarks of cellular GSK-3 inhibition is reduction in the phosphorylation state of GSK-3 substrates. For example, GSK-3 constitutively phosphorylates its canonical substrates (3, 31, 35, 50, 60) glycogen synthase, and β-catenin at Ser641 and Ser33/37/Thr41, respectively. Phosphorylation of glycogen synthase at Ser641 reduces its activity while phosphorylation of β-catenin at Ser33/37/Thr41 initiates β-catenin ubiquitination and degradation by the proteasome. Thus inhibition of GSK-3 activity would be indicated by decreased phosphorylation of glycogen synthase at Ser641, decreased phosphorylation of β-catenin at Ser33/37/Thr41, as well as increased β-catenin accumulation (1, 11, 19, 42). The effect of COB-187 on phosphorylation and total expression levels of glycogen synthase and β-catenin in THP-1 macrophages were investigated using Western blot analysis, and the results are presented in Fig. 4, A and B, respectively. As illustrated in Fig. 4A, no significant changes were observed in the level of total glycogen synthase, while glycogen synthase phosphorylation at Ser641 was moderately reduced after treatment of THP-1 macrophages with ≥25 µM COB-187 relative to the DMSO control. Western blot analysis of β-catenin (Fig. 4B) revealed a significant dose-dependent increase in total β-catenin levels and a reduction in β-catenin phosphorylation at Ser33/37/Thr41 after treatment of THP-1 macrophages with COB-187 relative to the DMSO control. Thus it appears that COB-187 inhibits GSK-3 phosphorylation of β-catenin and, albeit to a lesser extent, glycogen synthase. As a complement to the THP-1 studies, the effect of COB-187 on phosphorylation and total expression levels of β-catenin was investigated in another macrophage cell line, namely the murine RAW 264.7 cell line. No significant effect on the metabolic activity of the RAW macrophages was observed at COB-187 concentrations ≤25 µM (data not shown); Thus treatments on RAW 264.7 cells were performed with COB-187 concentrations ≤25 µM. As shown in Fig. 5, treatment of RAW 264.7 macrophages with COB-187 resulted in a significant increase in total β-catenin levels at concentrations ≥10 µM as well as a dose-dependent reduction in β-catenin phosphorylation at Ser33/37/Thr41 relative to the DMSO control.

Fig. 4.

Treatment of THP-1 macrophages with COB-187 results in the reduction of phosphorylation of canonical GSK-3 substrates. A: treatment of THP-1 macrophages with COB-187 for 5 h did not significantly affect the level of total glycogen synthase (GS) but treatment with ≥25 µM COB-187 did moderately reduce the phosphorylation of glycogen synthase at Ser641 (top: images from Western blot analysis; middle: quantification of total glycogen synthase relative to β-actin; bottom: quantification of Ser641 phosphorylated glycogen synthase relative to total glycogen synthase). B: Western blot analysis of lysate from THP-1 macrophages treated with COB-187 revealed a dose-dependent reduction in β-catenin phosphorylation at Ser33/37/Thr41 and consequent increase in β-catenin accumulation (top: images from Western blot analysis; middle: quantification of total β-catenin relative to β-actin; bottom: quantification of Ser33/37/Thr41 phosphorylated β-catenin relative to total β-catenin). Western blot images are representative of three separate experiments. Quantitative analysis of results from three separate experiments were averaged and were analyzed using a one-way ANOVA coupled with post hoc Dunnett’s test and presented as bar charts. Error bars represent the SE for the three separate experiments. *Results with P < 0.05 were considered significantly different relative to 0.1% DMSO control. β-Actin was used as a loading control.

Fig. 5.

Treatment of RAW 264.7 macrophages with COB-187 results in increases in the protein level of β-catenin. Western blot analysis of lysates isolated from RAW 264.7 cells treated with COB-187 revealed a dose-dependent reduction in β-catenin phosphorylation at Ser33/37/Thr41 and consequent increase in β-catenin accumulation [left: images from Western blot analysis; right: quantification of total β-catenin relative to β-actin (top) and quantification of Ser33/37/Thr41 phosphorylated β-catenin relative to total β-catenin (bottom)]. Western blot images are representative of three separate experiments. Quantitative analysis of results from three separate experiments were averaged and were analyzed using a one-way ANOVA coupled with post hoc Dunnett’s test and presented as bar charts. Error bars represent the SE for the three separate experiments. *Results with P < 0.05 were significantly different relative to 0.1% DMSO control. β-Actin was used as a loading control.

COB-187 enhances β-catenin localization to the perinuclear and nuclear region.

Another hallmark of GSK-3 inhibition, and consequent β-catenin stabilization, is accumulation of β-catenin in the nucleus (1, 11, 19, 42). Thus the effect of COB-187 on β-catenin localization was evaluated using immunocytochemistry. As shown is Figs. 6 and 7, confocal microscopic analysis of THP-1 and RAW 264.7 macrophages, respectively, subsequent to immunostaining, revealed an increase in total β-catenin as well as accumulation of β-catenin in the perinuclear and nuclear regions with increasing concentration of COB-187. Translocation of accumulated β-catenin from the cytoplasm to the nucleus is more pronounced in RAW 264.7 cells (Fig. 7) relative to THP-1 macrophages (Fig. 6). These findings compliment the findings in the section above and further support that COB-187 inhibits phosphorylation of β-catenin.

Fig. 6.

Treatment of THP-1 macrophages with COB-187 results in an apparent dose-dependent translocation of β-catenin to the perinuclear and nuclear region. The effect of COB-187 on β-catenin localization in THP-1 macrophages was determined using immunocytochemistry. Each image, acquired by confocal microscopy, shows nuclear staining with DAPI (blue) and β-catenin staining (green). Bottom left panel of each set of data shows the merged composite image. Results are representative of two separate experiments. Nuclear and perinuclear accumulation is indicated by brightening of the blue area and accumulation of a green signal (halo) around the nucleus, respectively, in the merged images with increasing concentration of COB-187. Scale bars = 10 µm. THP-1 macrophages were treated for 5 h with COB-187 before immunocytochemical analysis. Isotype control image is THP-1 macrophages treated with an isotype control for β-catenin (image is representative of what was observed for all COB-187/DMSO/medium treatment groups).

Fig. 7.

Treatment of RAW 264.7 cells with COB-187 results in an apparent dose-dependent translocation of β-catenin to the nucleus. The effect of COB-187 on β-catenin localization in RAW 264.7 cells was determined using immunocytochemistry. Each image, acquired by confocal microscopy, shows nuclear staining with DAPI (blue) and β-catenin staining (green). Bottom left panel of each set of data shows the merged composite image. Results are representative of two separate experiments. Nuclear accumulation is indicated by brightening of the blue area in the merged images with increasing concentration of COB-187. Scale bars = 10 µm. RAW 264.7 cells were treated for 5 h with COB-187 before immunocytochemical analysis. Isotype control image is RAW 264.7 cells treated with an isotype control for β-catenin (image is representative of what was observed for all COB-187/DMSO/medium treatment groups).

Treatment of THP-1 macrophages and RAW 264.7 cells with COB-187 for 5 h does not increase the mRNA level of β-catenin.

To determine if the increase in the level of β-catenin observed in the Western blot (Figs. 4B and 5) and immunocytochemistry (Figs. 6 and 7) could be due to a change in the level of de novo β-catenin production, RT-qPCR was performed on mRNA isolated from THP-1 macrophages and RAW 264.7 cells treated with varying concentrations of COB-187. As shown in Fig. 8, A and B, treatment of THP-1 macrophages and RAW 264.7 cells with COB-187 for 5 h did not cause a significant increase in the level of β-catenin mRNA, relative to the DMSO control, suggesting that the increase in β-catenin observed in Figs. 4B, 5, 6, and 7 is not due to a marked increase in de novo generation of β-catenin.

Fig. 8.

Treatment with COB-187 does not change the β-catenin mRNA level in THP-1 macrophages and RAW 264.7. A and B: THP-1 macrophages (A) and RAW 264.7 cells (B) were treated with varying concentrations of COB-187 or 0.1% DMSO for 5 h. Subsequently, the β-catenin mRNA levels were determined via RT-qPCR. No significant changes in β-catenin mRNA levels were observed. ACTB and Hprt1 were used as housekeeping genes for THP-1 macrophages and RAW 264.7 cells, respectively. Results are the average of three independent experiments performed in triplicate. Error bars represent the SE. Results were analyzed using a one-way ANOVA coupled with post hoc Games-Howell test.

Treatment of THP-1 macrophages with COB-187 for 5 h increases the expression of a Wnt target gene.

In the nucleus, translocated β-catenin binds to the lymphoid enhancer factor/T-cell factor (LEF/TCF) transcription factor, increases its transcriptional activity, and induces the expression of Wnt target genes (e.g., cyclin D1) by displacing corepressors and recruiting additional coactivators (18, 25, 55). To further investigate the effect of COB-187 on GSK-3 inhibition, the effect of COB-187 on cyclin D1 was evaluated using RT-qPCR on mRNA isolated from THP-1 macrophages treated with varying concentrations of COB-187. As illustrated in Fig. 9, treatment of THP-1 macrophages with ≥25µM COB-187 for 5 h significantly increased cyclin D1 mRNA levels suggesting an increase in nuclear β-catenin in response to treatment with COB-187.

Fig. 9.

Treatment with COB-187 significantly increases the expression of the Wnt target gene cyclin D1. THP-1 macrophages were treated with varying concentrations of COB-187 or 0.1% DMSO for 5 h. Subsequently, the cyclin D1 mRNA levels were determined via RT-qPCR. Cyclin D1 mRNA levels were significantly increased in response to the treatment with ≥25 µM COB-187. ACTB was used as a housekeeping gene. Results are the average of three independent experiments performed in triplicate. Error bars represent the SE. Results were analyzed using a one-way ANOVA coupled with Dunnett’s test. *Results with P < 0.05 were significantly different relative to 0.1% DMSO control.

COB-187 reduces the phosphorylation of tau in HEK293 cells transfected with the tau gene.

A variety of studies suggest that GSK-3 is one of the important protein kinases involved in Alzheimer’s diseases where its activity leads to hyperphosphorylation of microtubule-associated protein tau at different serine and threonine residues [e.g., Ser202 (14), Thr181 (14), and Ser396/Ser404 (PHF-1) (4, 30)] as well as formation of neurofibrillary tangles (30, 47). Thus a possible therapeutic approach for Alzheimer’s disease, and other neurodegenerative diseases, could be inhibition of GSK-3 activity and consequent reduction in the phosphorylation state of tau (9, 20, 31). HEK293 cells are a commonly used cell model to study tau pathology (24) but require tau overexpression via transfection with a tau expression vector, since the endogenous tau protein expression in HEK293 cells is low. Thus we investigated the effect of COB-187 on phosphorylation and total expression levels of tau in tau transfected HEK293 cells. Note that COB-187 had no significant effect on the metabolic activity of the HEK293 at concentrations ≤50 µM (data not shown). Therefore, concentrations ≤50 µM were used on the HEK293 cells. As illustrated in Fig. 10, transfection with the tau expression vector led to significant production of tau protein in HEK293 cells relative to the nontransfected cells. Treatment of transfected HEK293 cells with COB-187 resulted in no significant changes in the level of total tau. In contrast, phosphorylation of tau at Ser396 was significantly reduced after treatment of the cells with ≥25 µM COB-187 relative to the DMSO control. A significant reduction in phosphorylation of tau at Ser404 was also observed but only at 50 µM COB-187.

Fig. 10.

COB-187 reduces the phosphorylation of tau in HEK293 cells transfected with tau expression vector. Western blot analysis of lysates isolated from HEK293 cells transfected with tau expression vector revealed distinct production of tau protein relative to the nontransfected HEK293 cells. Treatment of transfected HEK293 cells with COB-187 for 5 h did not significantly affect the level of total tau. Treatment with ≥25 µM COB-187 did reduce the phosphorylation of tau at Ser396. A significant reduction in phosphorylation of tau at Ser404 was also observed at 50 µM COB-187. Top: images from Western blot analysis. Bottom: quantification of total tau relative to β-actin (left), Ser396 (middle), and Ser404 (right) phosphorylated tau relative to total tau. Western blot images are representative of three separate experiments. Quantitative analysis of results from three separate experiments were averaged and were analyzed using a one-way ANOVA coupled with post hoc Dunnett’s test and presented as bar charts. Error bars represent the SE for the three separate experiments. *Results with P < 0.05 were significantly different relative to 0.1% DMSO control. β-Actin was used as a loading control.

DISCUSSION

In this study, COB-187 and COB-152 were introduced as two novel, potent, and highly selective inhibitors of GSK-3. To our knowledge, these compounds have not previously been described, and their chemical structures (Fig. 1) are clearly distinct from commercially available and previously screened GSK-3 inhibitors (2, 5, 17, 28, 32).

COB-187 was compared with the clinically advanced heterocyclic thiadiazolidinone (Tideglusib) in an extensive molecular screen consisting of 414 kinase assays representing 404 unique kinases. COB-152, a close isomer of COB-187, was also investigated in a robust but less extensive screen (Supplemental Table S1; see https://doi.org/10.6084/m9.figshare.9546998). The results revealed that COB-187 and COB-152 are significantly more selective relative to Tideglusib (Table 2). Indeed, excluding GSK-3α and GSK-3β, COB-187 and COB-152 were observed to inhibit only 3 kinases at ≥40% while Tideglusib inhibited 50. Our findings for Tideglusib, i.e., that it inhibits numerous kinases in addition to GSK-3, is in agreement with the Domínguez et al. (16) report that Tideglusib inhibits a multitude of kinases. The exceptional selectivity of COB-187 and COB-152 for GSK-3 suggests that these novel compounds are likely not ATP-competitive inhibitors since ATP competitive inhibitors tend to be promiscuous due to the similarity of ATP binding sites between kinases (2, 5). That said, we cannot rule out the possibility that COB-187 and COB-152 bind to a region of the GSK-3 ATP binding site that is unique to GSK-3. We are currently pursuing kinetic and modeling studies to elucidate the mechanism of action of COB-187 and COB-152 inhibition of GSK-3. As shown in Fig. 2, Tideglusib exhibited a more significant inhibitory effect on GSK-3β relative to GSK-3α (specifically a 1.8-fold increase in IC₅₀ for Tideglusib on GSK-3α compared with GSK-3β). This result was expected from Domínguez et al. (16) that reported Tideglusib was a more effective inhibitor of GSK-3β than GSK-3α. We also found that COB-187 and COB-152 are more potent than Tideglusib, with the IC₅₀ for Tideglusib ranging from 3.8-fold to >41-fold higher than the IC₅₀s for COB-152 and COB-187 (Fig. 2). Interestingly, COB-187 appeared to exhibit greater potency than COB-152 (Fig. 2). Therefore, we chose to investigate the inhibitory effect of COB-187 on the activity of GSK-3 in cellular assays.

A hallmark of cellular GSK-3 inhibition is a reduction in the phosphorylation of GSK-3 substrates. Thus we investigated the effects of COB-187 on the phosphorylation state of the classic GSK-3 substrates, glycogen synthase, β-catenin, and tau. As shown in Fig. 4A, treatment of THP-1 macrophages with ≥25 µM COB-187 moderately reduced the level of glycogen synthase phosphorylation at Ser641. This result is somewhat in agreement with Martin et al. (35) that showed abrogation of glycogen synthase phosphorylation at Ser641 in freshly isolated monocytes after treatment with the GSK-3 inhibitor SB216763 at a markedly higher, compared with what we used for COB-187, concentration of 10 mM. In a more dramatic fashion, COB-187 diminished phosphorylation of β-catenin (Fig. 4B and Fig. 5). In general, GSK-3 can be free in cells or present in a protein complex that is a key component of its ability to phosphorylate certain substrates (25). For β-catenin, GSK-3 is incorporated into the β-catenin destruction complex containing the scaffold protein Axin and adenomatous polyposis coli (APC), in addition to β-catenin. This complex brings β-catenin into close proximity of GSK-3 and results in phosphorylation of β-catenin at Ser33/37/Thr41 and consequent ubiquitination and degradation of β-catenin by the proteasome. Inhibition of GSK-3 activity results in dephosphorylation of β-catenin (at Ser33/37/Thr41), leading to β-catenin stabilization and accumulation (1, 11, 19, 42). Indeed, previous studies have reported that GSK-3 inhibitors including Tideglusib (8), 6-BIO (35, 50), LiCL, Azakenpaullone, SB216763 (35), LY2090314 (3), and L807mts (31) reduce the phosphorylation of β-catenin (Ser33/37/Thr41) and consequently cause an increase in the levels of β-catenin. For COB-187, we found a dose-dependent decrease in the phosphorylation of β-catenin at Ser33/37/Thr41 and an increase in the level of total β-catenin for THP-1 macrophages and RAW 264.7 cells treated with COB-187 (Figs. 4B and 5, respectively), a finding consistent with the hypothesis that COB-187 is a GSK-3 inhibitor.

Further insights can be gained by examining the localization of β-catenin and consequently expression of Wnt target genes (e.g., cyclin D1) (18, 25, 55). Specifically, Beurel et al. (7) pointed out that β-catenin nuclear levels increase after the β-catenin destruction complex is inactivated. β-Catenin nuclear translocation can be assessed by immunocytochemistry coupled with confocal microscopy. Thus we used immunocytochemistry to evaluate the effect of COB-187 on the total level of β-catenin and its translocation to the nucleus. This analysis revealed a dose dependent increase in total β-catenin for THP-1 macrophages treated with increasing levels of COB-187 (Fig. 6). The analysis also revealed an apparent increase in the perinuclear/nuclear localization of β-catenin with increasing concentrations of COB-187 treatment with the most pronounced effects at ≥25 µM (Fig. 6). Similar results were observed for RAW 264.7 cells treated with COB-187. As illustrated in Fig. 7, a dose-dependent increase in total β-catenin along with an apparent increase in the nuclear localization of it was observed in COB-187-treated RAW 264.7 cells. These results suggest that COB-187 disrupts the activity of the β-catenin protein destruction complex most likely due to inhibition of GSK-3 kinase activity. Activation of Wnt/β-catenin signaling and increase in expression of a Wnt target gene in THP-1 macrophages were evaluated via RT-qPCR. A significant increase in the expression of cyclin D1 was observed after treatment with ≥25 µM COB-187 (Fig. 9). It should be noted that the increase in total β-catenin was likely not due to an increase in de novo protein expression of β-catenin since treatment of THP-1 macrophages and RAW 264.7 cells with COB-187 did not cause an increase in β-catenin mRNA levels (Fig. 8, A and B). Further, other authors have suggested that stabilization of β-catenin requires inhibition of both GSK-3α and GSK-3β activity (12). This fact, combined with the Western blot, immunocytochemistry, and RT-qPCR analyses (Figs. 4B, 5, 6, 7, and 8) provided in this study, supports the hypothesis that COB-187 is a cellular inhibitor of GSK-3.

Tau is another well-known substrate of GSK-3 that gets hyperphosphorylated at different serine and threonine residues [e.g., Ser396/Ser404 (PHF-1) (4, 30)] in neurodegenerative diseases such as Alzheimer’s disease (30, 47). Previous studies have shown that inhibitors of GSK-3 [e.g., AR-A014418 (9), L807mts (31), and LiCl (20)] effectively inhibit the hyperphosphorylation of tau at Ser396 and/or Ser404, a critical step in the pathogenesis of Alzheimer’s disease. In this study, as illustrated in Fig. 10, treatment with ≥25 µM COB-187 markedly reduced the phosphorylation of tau at Ser396/Ser404 in transfected HEK293 cells while causing no significant changes in the level of total tau.

In conclusion, GSK-3 plays a key role in a plethora of physiological and pathophysiological processes and GSK-3′s elevated activity has been implicated in a host of diseases. Therefore, reagents that inhibit GSK-3 activity provide a means to investigate the role of GSK-3 in cellular physiology and pathophysiology and could become valuable therapeutics. Our group has generated novel small organic compounds that appear to be potent and highly selective inhibitors of GSK-3 activity. This finding provides a critical first step in defining a region of chemical design space that contains compounds that could help unravel the molecular mechanisms of GSK-3 cellular activity and that have the potential to become therapeutics for a variety of diseases caused by aberrant GSK-3 activity.

GRANTS

This work was supported by National Institute of General Medical Sciences Grant R15 GM110602-01A1 (to D. J. Goetz, K. D. McCall, and S. C. Bergmeier).

DISCLOSURES

Ohio University owns a patent on the novel compounds (COB 152/187/197) described in this paper. C. M. Orac, M. C. McMills, S. C. Bergmeier, and D. J. Goetz are the inventors. None of the other authors has any conflicts of interest, financial or otherwise, to disclose.

AUTHOR CONTRIBUTIONS

M.S.N., P.M.B., M.C.M., S.P.D., S.C.B., K.D.M., and D.J.G. conceived and designed research; M.S.N., P.M.B., M.C.C., D.G., C.C., and C.M.O. performed experiments; M.S.N., M.C.C., D.G., C.C., F.L.S., K.D.M., and D.J.G. analyzed data; M.S.N., C.M.O., M.C.M., F.L.S., S.P.D., K.D.M., and D.J.G. interpreted results of experiments; M.S.N. and S.C.B. prepared figures; M.S.N. and D.J.G. drafted manuscript; M.S.N., S.C.B., K.D.M., and D.J.G. edited and revised manuscript; M.S.N., P.M.B., M.C.C., D.G., C.C., C.M.O., M.C.M., F.L.S., S.P.D., S.C.B., K.D.M., and D.J.G. approved final version of manuscript.