Synthesis and biological evaluation of flavonoid-based IP6K2 inhibitors

Abstract Inositol polyphosphates (IPs) are a group of inositol metabolites that act as secondary messengers for external signalling cues. They play various physiological roles such as insulin release, telomere length maintenance, cell metabolism, and aging. Inositol hexakisphosphate kinase 2 (IP6K2) is a key enzyme that produces 5-diphosphoinositol 1,2,3,4,6-pentakisphosphate (5-IP7), which influences the early stages of glucose-induced exocytosis. Therefore, regulation of IP6Ks may serve as a promising strategy for treating diseases such as diabetes and obesity. In this study, we designed, synthesised, and evaluated flavonoid-based compounds as new inhibitors of IP6K2. Structure-activity relationship studies identified compound 20s as the most potent IP6K2 inhibitor with an IC50 value of 0.55 μM, making it 5-fold more potent than quercetin, the reported flavonoid-based IP6K2 inhibitor. Compound 20s showed higher inhibitory potency against IP6K2 than IP6K1 and IP6K3. Compound 20s can be utilised as a hit compound for further structural modifications of IP6K2 inhibitors.


Results and discussion
Previously, we reported flavonoid analogs as potential inhibitors of thymic stromal lymphopoietin (TSLP), an alarmin cytokine involved in allergic immune responses. 33 First, we screened the IP6K2-inhibitory activities of representative compounds (2a-2f, Figure 2) from an in-house compound library at concentrations of 10 and 50 mM using an in vitro ADP-Glo assay. We focussed on determining the effect of 5,7-di-OH substitution on IP6K2 inhibition compared to that of 6,7-di-OH substitution at the A ring, which is not common in natural flavonoids. Compounds (2a, 2c, and 2e) substituted with 6,7-di-OH were found to inhibit IP6K2 more strongly than the corresponding compounds (2b, 2d, and 2f) with 5,7-di-OH.
On the basis of preliminary studies, we designed and synthesised flavonoid analogs by replacing the -OH group of the A ring with the -F group to investigate the necessity of the hydroxyl group for IP6K2 inhibition. Scheme 1 describes the synthesis of fluoro-substituted flavonoid derivatives with various functional groups on the B ring. Commercial acetophenones 3-5 were used as starting materials for the synthesis of monofluoro-substituted compounds. In the case of difluoro-substituted compounds, starting materials 6 and 7 were prepared from 3,4-difluorophenol and 3,5-difluorophenol, respectively, by applying a reported synthetic procedure. 34,35 The reactions of compounds 3-7 and appropriate aldehydes with barium hydroxide in methanol (or ethanol) at 50 C for 1-17 h afforded chalcone compounds 8a-8k (44-95% yield) via the Claisen-Schmidt condensation reaction. The intramolecular cyclisation of compounds 8a-8k with iodine (I 2 ) at 110 C for 6-24 h generated compounds 9a-9k (47-97% yield). Compounds substituted with a methoxy group or an ethyl ester were reacted with BBr 3 to yield OH-substituted compounds (10b, 10d, 10f, and 10h) and COOH-substituted compound (10i), respectively. According to the results of the ADP-Glo assay, compounds (9a-10h) substituted with the -F group in the A ring showed considerably weaker IP6K2-inhibitory activity than the corresponding compounds with the -OH group at 10 and 50 lM concentrations (See Supporting Information Figures 1 & 2). Therefore, we maintained the -OH substituent in the A ring for further structural modification of the flavonoid-based IP6K2 inhibitors.
Next, we focussed on the structural modification of the B ring while maintaining the dihydroxyl substituents of the A ring at either the 5,7-or 6,7-positions. We introduced a polar carboxylic acid substituent at the meta-and para-positions of the B ring to generate potential ionic or hydrogen-bonding interactions. Scheme 2 describes the synthesis of flavonoid derivatives (16a-16d) with a carboxylic acid functional group in the B ring. The synthetic strategy for 16a-16d, which was similar to that for the F-substituted compounds (Scheme 1), included Claisen-Schmidt condensation and intramolecular cyclisation starting from methoxy-substituted acetophenones. The global dealkylation step was achieved using boron tribromide in dichloromethane at 50 C, affording the final compounds 16a-16d (50-87% yield).
The ADP-Glo kinase assay was conducted using the synthesised compounds 16a-16d (Table 1). Carboxylic-acid substitution at the meta position of the B-ring (16b and 16d) led to superior IP6K2 inhibition compared to para substitution (16a and 16c). Consistent with the results of the primary screening with 2a-2f, compounds 16a-16b with 6,7-dihydroxyl groups exhibited a stronger inhibitory effect on IP6K2 than the corresponding compounds 16c-16d with 5,7-dihydroxyl groups. Compounds 16a and 16b were as potent as quercetin, which possesses an -OH group at the 3-position of the C-ring.
To increase IP6K2-inhibitory activity, we introduced a hydroxyl group at the 3-position of the C-ring. Scheme 3 describes the synthesis of compounds with an -OH group at the 3-position of the C ring. The reaction of 11 with the appropriate benzaldehydes under basic conditions (NaOMe in MeOH/THF) provided high yields of chalcone compounds (18a-18s and 21a-21c). The cyclized compounds (19a-19q and 22a-22c) were obtained via the Algar-Flynn-Oyamada (AFO) reaction of chalcone compounds with hydrogen peroxide (35% aq.) and sodium hydroxide (3 M aq.) in ethanol (or methanol) at 40 C for 16-20 h. In the case of methylester compounds 19r and 19s, sodium methoxide was applied in the AFO reaction instead of sodium hydroxide. Reaction of the cyclized compounds with boron tribromide in dichloromethane at 50 C for 6-16 h afforded compounds (20a-20s and 23a-23c) with substitution at the 3-OH of the C ring. The chemical structures and purities of the final compounds were confirmed using 1 H NMR, LC/MS, and HPLC. In-house flavonoid compounds evaluated using ADP-Glo assay.

JOURNAL OF ENZYME INHIBITION AND MEDICINAL CHEMISTRY
For accurate quantification of the IP6K2-inhibitory activity, the IC 50 values of the synthesised compounds 20a-20s and 23a-23c were determined by performing an ADP-Glo kinase assay in a dose-dependent manner ( Table 2). We confirmed the positional effect of OH groups in the A ring on IP6K2 inhibition by comparing compound 20a with quercetin. The IC 50 value of compound 20a was 1.77 mM, while that of quercetin was 3.31 mM under the same experimental condition, consistently indicating that 6,7-substitution is preferred to 5,7-substitution in this series. The IC 50 value of quercetin was reported as 0.70 mM. 24 However, under our experimental conditions, it was 3.31 mM. The difference could be due to the different experimental conditions including the IP6K2 enzyme concentration and reaction temperature.
We examined the effects of substituents at the meta-or paraposition of the B ring on IP6K2 inhibition. Compound 20b, with no substituents on the B-ring, was synthesised as a control compound. Compounds 20c-20e with a -CH 3 group and compounds 20k-20l with a -Cl group were less potent than compound 20b. This result suggests that hydrophobic substituents on the B ring decrease IP6K2 inhibition. In addition, compounds 20f-20h with the -F group and the CF 3 -substituted compounds 20i-20j were less potent than 20b. In particular, the disubstituted compounds (e.g. 20e and 20h) completely lost their inhibitory activity against IP6K2. This may be due to the increased volume of the hydrophobic substituents. This negative effect of increased volume due to lipophilic substituents was confirmed for compounds (20m-20o) with bulky and hydrophobic substituents, such as the -Ph and -OPh groups.
In contrast, compound 20q, with a hydrophilic -OH group at the meta position in the B ring (20q), strongly inhibited IP6K2 with an IC 50 value of 2.96 lM and showed a 10-fold greater inhibition than compound 20p, with the -OH group at the para position. Consistent with the result for analogs with a carboxylic acid substituent 16a-16d, this result implies that the introduction of a hydrophilic substituent, particularly at the m-position of the B ring, contributes to increased IP6K2 inhibition. As expected, the substitution of the carboxylic acid group in the B ring resulted in an increase in IP6K2 inhibition. Compounds 20r and 20s showed strong inhibitory activity against IP6K2. In particular, the carboxylic compound 20s was the most potent with an IC 50 value of 0.55 mM, showing 5-fold more potent IP6K2 inhibition than quercetin (Table 2 and Figure 3). Compound 20s has a carboxylic acid moiety at the meta position in the B ring, which is approximately 7-fold more potent than compound 20r with the -COOH group at the para position. We additionally introduced heteroaromatic rings such as pyrazole (23b) and Scheme 1. Synthesis of fluoro-substituted flavonoid analogs with a variation of the B ring. Reagents and conditions: (i) (a) Acetyl chloride, pyridine, CH 2 Cl 2 , rt, 30 min, (b) AlCl 3 , 150 C, 10 min; (ii) appropriate aldehydes, Ba(OH) 2 , MeOH (or EtOH), 50 C, 1-17 h; (iii) I 2 , DMSO, 110 C, 6-24 h; (iv) BBr 3 , CH 2 Cl 2 , 50 C, 14-18 h. thiophene (23c). The pyrazole compound 23b exhibited an IC 50 value of 1.79 mM, rendering it 2-fold more potent than quercetin.
To figure out the selectivity for IP6K2 over IP6K1 and IP6K3, we decided to evaluate the inhibition rate of quercetin and the most potent compound 20s. The IC 50 value and the ratio of IP6K2 and other IP6Ks were shown in Table 3 (See Supporting Information Figure 3). Compound 20s showed IC 50 value of 2.87 mM against IP6K1 resulting 5.22-fold IC 50 ratio of IP6K1/IP6K2 and IC 50 value of 3.56 mM against IP6K3, resulting 6.47-fold IC 50 ratio of IP6K3/ IP6K2. When the final concentration of ATP and IP6 was adjusted to 10 lM, compound 20s showed 4.65-fold and 2.32-fold IC 50 ratio of IP6K1/IP6K2 and IP6K3/IP6K2, respectively (See Supporting Information Figure 4). This result represented that compound 20s had selectivity against IP6K2 rather than IP6K1 or IP6K3 compared with quercetin. Overall, quercetin is a pan-IP6K inhibitor while compound 20s is a more selective IP6K2 inhibitor. Next, we determined the membrane permeability of 20s by an in vitro PAMPA permeability assay. 36 Compound 20s exhibited very low permeability with an apparent permeability coefficient of <1.8 nm/s while quercetin did with 5.4 nm/s (See Supporting Information Table 1). We further examined whether 20s can inhibit IP7 synthesis in human colorectal cell line (HCT116) after treatment of the compound (10 lM) for 6 h since IP6K2 has been reported as a major enzyme for cellular IP7 synthesis among three isoforms of IP6Ks in HCT116. 37 However, 20s failed to reduce cellular IP7 levels (See Supporting Information Figure 5), suggesting its low cell permeability.

Conclusion
In the present study, new flavonoid-based IP6K2 inhibitors were designed, synthesised, and evaluated in vitro. Systemic structureactivity relationship studies for IP6K2 inhibition revealed the substituent preference for each ring in the flavonoid backbone as follows: -OH groups at the 6-and 7-positions in the A ring, -COOH group at the meta position of the B ring, and -OH group at the 3-position of the C ring. Furthermore, hydrophilic substituents such as -OH and -COOH in the B ring imparted stronger IP6K2-inhibitory activity than hydrophobic ones, including -CH 3 , -Cl, CF 3 , -Ph, and -OPh. Additionally, compounds with a substituent at the meta position of the B ring were more effective than the corresponding compounds with a substituent at the para position. Among the compounds synthesised, compound 20s was the most potent inhibitor with an IC 50 value of 0.55 mM against IP6K2, which renders it 5-fold more potent than quercetin. In addition, compound 20s showed higher inhibitory potency against IP6K2 than IP6K1 and IP6K3. The molecular docking study showed that compound 20s formed additional hydrogen bonds with Gln260 and Asp383 compared to quercetin. Although compound 20s showed low membrane permeability, its physicochemical properties can be improved by a prodrug approach. Overall, compound 20s has the potential to be used as a hit compound for the structural optimisation of flavonoid-based IP6K2 inhibitors and as a tool compound for studying IP6K2 signalling pathways.

Experimental section
General All chemicals and solvents used in the reaction were purchased from Sigma-Aldrich, TCI, and Acros and were used without further purification.  0.1% trifluoroacetic acid); for method H, mobile phase was acetonitrile and water (20:80, v/v, 0.1% trifluoroacetic acid); for method I, mobile phase was acetonitrile and water (15:85, v/v, 0.1% trifluoroacetic acid); for method J, mobile phase was acetonitrile and water (12:88, v/v, 0.1% trifluoroacetic acid). All compounds were eluted with a flow rate of 1 mL/min and monitored at a UV detector (220 nm or 254 nm). The purity of all tested compounds was >95%.

Chemical synthesis
General Procedure A for the synthesis of compounds 8a-8n and 14a-14d To a stirred solution of substituted acetophenone (3-7 and 11-13) in methanol (MeOH) or ethanol (EtOH) was added barium hydroxide (2.0 eq) and appropriate aldehyde (1.2 eq) at room temperature. The reaction mixture was stirred under argon at   40-50 C until complete conversion monitored by TLC analysis (typically 1-20 h), quenched with acetic acid, and extracted with EtOAc and H 2 O. The organic layer was washed with saturated aqueous NaHCO 3 , dried over MgSO 4 , filtered, and concentrated under reduced pressure. The crude residue was purified by column chromatography on a silica gel to furnish compounds 8a-8n and 14a-14d.

General Procedure D for compounds 18a-18s
To a stirred solution of substituted acetophenone compound in tetrahydrofuran (THF) was added appropriate aldehyde (1.2 eq or 1.5 eq) at room temperature. The reaction mixture was added sodium methoxide (5.4 M, 1.2 eq) in methanol solution at 0 C and stirred for 5 min. The reaction mixture was warmed to room temperature and stirred at room temperature until complete conversion monitored by TLC analysis (typically 12-16 h), quenched with acetic acid, and extracted with EtOAc and H 2 O. The organic layer was washed with saturated aqueous NaHCO 3 , dried over MgSO 4 , filtered, and concentrated under reduced pressure. The crude residue was purified by column chromatography on a silica gel to furnish compounds 18a-18t.

In silico docking studies
Ligand preparation and optimisation: All ligands were generated as 2D and 3D structure by ChemDraw Ultra (ver. 12.0.2) and Chem3D Pro (ver. 11.0.1), respectively. Ligand preparation and optimisation was followed by "Sanitize" preparation protocol in SYBYL-X 2.1.1 (Tripos Inc., St Louis) to clean up of the structures involving filling valences, standardising, removing duplicates and producing only one molecule per input structure. The group of ligands was saved as .sdf file.
Protein preparation: For IP6K2, the protein model of human IP6K2 (UniProt: Q9UHH9) was downloaded from AlphaFold2 structure database (https://alphafold.ebi.ac.uk) as PDB and FASTA format. SYBYL-X 2.1.1 program was employed for protein preparation including conflicted side chains of amino acid residues fixation. Hydrogen atoms were added under the application of AMBER7 FF99 Force Field setting for IP6K2, respectively. Minimisation process was performed by POWELL method, and initial optimisation option was set to None for IP6K2. Termination gradient and max iteration were set 0.05 kcal/(mol Ã Å) and 100 times.
Docking and scoring function studies: The docking studies of all prepared ligands were performed by Surflex-Dock GeomX module in SYBYL-X 2.1.1. For CYP3A4, docking was guided by the Surflex-Dock protomol and docking site was defined by the "Ligand" method with the reported ligand quercetin. Surflex-Dock protomol set to "Residues" method with selected amino acids (Lys42, Leu206, Glu207, Asn208, Leu209, Thr210, Val 218, Leu219, Asp220, Leu221, Lys222, Asp383; radius setting: 2.2; Those amino acids were selected based on the active site of EhIP6KA.) was used to guide docking site for IP6K2 homology model. Two factors related with a generation of Protomol are Bloat(Å) and Threshold were set to 0.5 and 0, respectively. Other parameters were applied with its default settings in all runs.

ADP-Glo kinase assay
Inhibition activity of the synthesised compounds against IP6K2 was assessed using ADP-Glo TM Kinase Assay (Promega). Kinase reaction mixtures were first shaking-incubated with individual drugs (diluted in DMSO) for pre-incubation at RT for 15 min, and then ATP was added (final 10 mM) to initiate the reaction. 20 ng of recombinant human IP6K2 protein was used per 15 mL of IP6K2 reaction mixture (50 mM Tris-HCl, pH 6.8, 10 mM MgCl 2 , 2.5 mM DTT, 0.02% Triton X-100, 10 mM IP6). Kinase reaction was proceeded at 37 C for 30 min with 100 rpm shaking, and the amount of ADP produced was measured on white plates using Mithras LB940 plate reader (Berthold) according to the manufacturer's protocol. Each detection set was prepared on 96-well (25 mL kinase reaction) as duplicates. Resulting enzyme activity was calculated as following; Activity (%) ¼ ([ADP experimental ]/[ADP DMSO ])Â100. For IC 50 estimation of compounds, kinase assay was prepared on 384-well plates (20 ng of recombinant human IP6K2 protein was used per 5 mL kinase reaction mixture) as duplicates. The final concentrations for IP6K2, IP6, and ATP were 80 nM, 10 lM, and 10 lM, respectively. Kinase reaction was proceeded at 37 C for 40 min with 300 rpm shaking, and the amount of ADP produced was measured using Synergy Neo microplate reader (Biotek) according to the manufacturer's protocol. IP6K1 and IP6K3 assays were performed by Liao et al.'s conditions. 25 The final concentrations for IP6K1, IP6, and ATP were 60 nM, 100 lM, and 1 mM, respectively. The final concentrations for IP6K3, IP6, and ATP were 120 nM, 100 lM, and 1 mM, respectively. The mixed reaction plate was placed into a 37 C incubator and rotated on an orbital shaker at 300 rpm for 30 min. All statistical analyses including IC 50 estimation was performed using Prism 7 or Prism 9 (Graphpad Software).

Extractions and quantification of intracellular inositol phosphates by HPLC
For HPLC analysis, 2 Â 10 5 cells/60 mm dish of HCT116 were treated with 60 lCi [ 3 H] myo-inositol (NET1177001MC, PerkinElmer). After 3 days, soluble inositol polyphosphates from HCT116 (human colorectal cancer cell line) were extracted and analysed as previously described. 39 Intracellular inositol phosphates were extracted with acid extraction buffer (1M HClO 4 , 3 mM EDTA, and 0.1 mg/mL IP6), and neutralised with neutralisation buffer (1M K 2 CO 3 and 3 mM EDTA). The lysates were centrifuged for 10 min, and the soluble fraction was resolved by HPLC as described earlier. 39 The lipid pellet was lysed with 0.1% Triton X-100 in NaOH overnight. Each fraction was mixed with Ultima-Flo AP liquid scintillation cocktail (6013599, PerkinElmer), and radioactivity was counted in a scintillation counter. Quantity of inositol phosphates was presented as total counts/min (CPM) normalised by total lipid contents.

Disclosure statement
No potential conflict of interest was reported by the author(s).