Target protein-oriented isolation of Hes1 dimer inhibitors using protein based methods

Natural products isolation using protein based methods is an attractive for obtaining bioactive compounds. To discover neural stem cell (NSC) differentiation activators, we isolated eight inhibitors of Hes1 dimer formation from Psidium guajava using the Hes1-Hes1 interaction fluorescent plate assay and one inhibitor from Terminalia chebula using the Hes1-immobilized beads method. Of the isolated compounds, gallic acid (8) and 4-O-(4”-O-galloyl-α-L-rhamnopyranosyl)ellagic acid (11) showed potent Hes1 dimer formation inhibitory activity, with IC50 values of 10.3 and 2.53 μM, respectively. Compound 11 accelerated the differentiation activity of C17.2 NSC cells dose dependently, increasing the number of neurons with a 125% increase (5 μM) compared to the control.


Results and Discussion
It is essential to inhibit undesired PPIs in drug development, and bioactivity-guided isolation using the inhibitory activity of PPIs is an attractive method to obtain effective inhibitors. However, currently there are only a few examples of such approaches using PPI assay systems 22,23 . We previously developed a Hes1-Hes1 interaction fluorescent plate assay and reported the screening results of our natural products compound library ( Fig. 2A) 24 . Glutathione-S-transferase (GST) fused rat Hes1  protein was expressed in Escherichia coli and purified rat Hes1 protein was immobilized on the bottom of 96 well plates. We prevented GST-GST interactions, which would result in false positives, by preparing GST-free Hes1 protein by GST cleavage with Turbo3C protease. After  www.nature.com/scientificreports www.nature.com/scientificreports/ labeling Hes1 with Cy3, this florescent Hes1 protein was added to the wells of the above plate and incubated for 24 h at 4 °C. Hes1-Hes1 interaction was successfully detected as Cy3 fluorescence intensity. Using this assay system, we screened our 118 plant extract library and identified the MeOH extract of Psidium guajava leaves to contain naturally occurring compounds that inhibit Hes1 dimer formation. The MeOH extract (29.9 g) was fractionated using Diaion HP-20 with a MeOH-acetone solvent system to afford fractions 1A to 1C. Active fraction 1A (27.2 g) was suspended in 10% aq. MeOH and partitioned with hexane, EtOAc and BuOH to obtain hexane (1.1 g), EtOAc (5.7 g), BuOH (4.7 g) and aqueous (18.9 g) soluble fractions. Part of the active BuOH soluble fraction was subjected to ODS column chromatography and reverse-phase HPLC. Activity-guided separation yielded ten compounds (1-10; Fig. 3). The isolated compounds were identified as morin (1) 25 , isoquercitrin (2) 26 , methyl gallate (3) 27 , (+)-catechin (4) 28,29 , dihydrophaseic acid (5) 30 , quercetin (6) 26,31 , avicularin (7) 32,33 , gallic acid (8) 34 , protocatechuic acid (9) 35 and 4-hydroxybenzoic acid (10) 36 based on comparisons of their spectral data with spectra in the literature. The Hes1-Hes1 interaction inhibitory activities of the isolated compounds were examined (Fig. 4) and 3, 7, 8 and 9 produced moderate inhibition (IC 50 12.7, 26.5, 10.3 and 23.8 μM). The most potent inhibitor was gallic acid (8). Commercially available gallic acid also exhibited comparable inhibition (IC 50 8.9 μM). Inhibition by the gallic acid derivatives 3, 8, 9 and 10 showed that the number of phenolic hydroxyl groups affects inhibitory activity, with activity decreasing as the number of phenolic hydroxyl groups decrease.
We recently developed another protein-based screening method, the target protein oriented natural products isolation method (TPO-NAPI) using protein beads (Fig. 2B). Agalloside, inohanamine, α-mangostine, BE-14106, isomicromonolactam, staurosporin and linarin were isolated as Hes1 binding compounds using the TPO-NAPI method 15,17 . Rat Hes1 (1-95) containing basic and helix-loop-helix domains was immobilized because the helix-loop-helix domain is known to be important for Hes1-Hes1 interaction; therefore, utilizing this domain in the beads method would likely be effective for screening Hes1 dimer inhibitors. GST-Hes1 immobilized beads were prepared by mixing freshly prepared GST-Hes1 protein with glutathione Sepharose 4B beads. GST-only beads were prepared as a control. After incubating the beads with plant MeOH extracts at 4 °C for 2 h, bound compounds were eluted by adding EtOH and heating at 100 °C for 3 min, then the eluted compounds were analyzed by HPLC. Of the 105 plant MeOH extracts screened using this method, the Bangladesh plant Terminalia chebula was found to contain a Hes1 binding compound. The MeOH extract (64.6 g) of Terminalia chebula bark was partitioned with hexane, EtOAc and BuOH to obtain hexane (1.5 g), EtOAc (3.6 g), BuOH (42.6 g), and aqueous (20.5 g) soluble fractions. The EtOAc fraction contained the target peak and was subjected to silica gel column chromatography to give eight fractions (1A-H). Fraction 1D contained the target peak and was separated www.nature.com/scientificreports www.nature.com/scientificreports/ by ODS column chromatography and reverse-phase HPLC to give compound 11 (0.4 mg). Compound 11 was identified as 4-O-(4″-O-galloyl-α-L-rhamnopyranosyl)ellagic acid by comparison of its NMR data with reported spectral data 37 . Next, we evaluated the Hes1 dimer inhibitory activity of compound 11 (Fig. 4) and found that it exhibited the most potent inhibition of Hes1 dimer formation of the compounds tested, with an IC 50 value of 2.53 μM. This is the strongest Hes1 dimer inhibitor reported to date. The ability of 11 to accelerate the differentiation of C17.2 mouse neural stem cells was evaluated (Fig. 5). C17.2 cells (2 × 10 5 cells/mL) were seeded in a poly-L-lysine-coated 24 well plates and incubated for 24 h, then the cells were treated with DMSO (control), 100 μM valproic acid (VPA), 5 μM retinoic acid (RA) (positive controls) or 11 (1 or 5 μM) for 6 days. A confocal microscope was used to obtain images of differentiated neural cells after immunostaining class III β-tubulin (Tuj1) in neurons, and glial fibrillary acidic protein (GFAP) in astrocytes and nuclei (TO-PRO-3). The number of neurons increased following the addition of VPA, RA or compound 11, suggesting that these compounds enhanced the number of C17.2 neurons. In contrast, the control dish containing cells treated with DMSO contained small number of neurons. The differentiated neurons with 11 were 33.6% (1 µM) and 45.4% (5 µM), which www.nature.com/scientificreports www.nature.com/scientificreports/ are 67% and 125% increase compared to those of control. These results suggested that compound 11 accelerates the differentiation of NSCs to neurons by inhibiting Hes1 dimer formation.
We also predicted the interaction between the HLH domain of Hes1 protein and compound 11 (4-O-(4″-O-galloyl-α-L-rhamnopyranosyl)ellagic acid) by performing in silico docking analysis of compound 11 with the HLH domain of Hes1. As shown in Fig. 6A,B, the galloyl site of compound 11 might interact with the loop region of Hes1, aiding the formation of Hes1(Arg46 of helix region)-Hes1(Glu76 of loop region) and preventing mutual recognition by Hes1 molecules. On the other hand, the ellagic acid site of compound 11 might bind with the helix region of Hes1, which consists of Ile50, Leu54 and Leu81, preventing hydrophobic core formation in the Hes1 dimer. Orange shows the hydrophobic region in Hes1 (Fig. 6C). Moreover, hydrogen bond formation between the ellagic acid site of compound 11 with Lys77 might obstruct Hes1(loop region)-Hes1(loop region) formation. Blue shows the hydrophilic region. As shown in Fig. 6C, the interaction of galloyl moiety with the hydrophilic region seems to be important. Therefore, the decrease of inhibition with decrease of number of phenolic hydroxyl groups in galloyl group would be reasonable. In addition, the α-L-rhamnopyranosyl unit appears to be an efficient linker, enabling tight interaction between compound 11 with the Hes1 monomer via its galloyl and ellagic acid sites.

Conclusion
In conclusion, several Hes1 binding natural products were isolated using two target protein-oriented isolation methods: a Hes1-Hes1 interaction fluorescent plate assay and a Hes1-immobilized beads method. Ten compounds were isolated from Psidium guajava, of which eight showed Hes1 dimer inhibitory activity. 4-O-(4″ -O-Galloyl-α-L-rhamnopyranosyl)ellagic acid (11) isolated using the Hes1 beads method showed potent Hes1 dimer formation inhibitory activity, with an IC 50 of 2.53 μM. Compound 11 accelerated the differentiation of C17.2 NSC cells into neurons dose dependently, increasing the number of neurons with a 125% increase (5 μM) compared to the control. There are few reports of Hes1 dimer inhibitors and thus these results will be useful for evaluating Hes1-related mechanisms in cells. We believe that protein-based isolation is an effective approach for identifying bioactive compounds.

Plant materials. Psidium guajava (Myrtaceae) and
In silico analysis. The initial 3D structures of 4-O-(4″-O-galloyl-α-L-rhamnopyranosyl)ellagic acid were constructed using Build in MOE (version 2016; Chemical Computing Group, Montreal, Canada) with standard geometric parameters. The ligand was then minimized using Energy Minimize in MOE with the Amber force field until the root-mean-square (rms) energy gradient was less than 0.001 kcal mol −1 Å −1 . The protein model was constructed based on the structure of HLH domain taken from the Protein Data Bank (PDB) (PDB code: 2MH3) 38 . Water molecules in the structure were removed. All hydrogen atoms were added and Amber all-atom charges were assigned for the whole protein. The molecular docking simulations were performed using the Induced Fit methods in MOE-Dock.