A natural DYRK1A inhibitor as a potential stimulator for β‐cell proliferation in diabetes

Dear Editor, Dual-specificity tyrosine-phosphorylation-regulated kinase 1A (DYRK1A) was demonstrated as a promising therapeutic target for diabetes for its influence on pancreatic β-cell mass and proliferation.1–8 In this study, we identified desmethylbellidifolin (DMB) as a novel and potent DYRK1A inhibitor. It was found to stimulate proliferations of β-cell both in vitro and in vivo via targeting DYRK1A. Twelve xanthone compounds (1-12; Figure S1A) from Swertia species were selected bymolecular docking for further testing (Table S1). Among them, DMB has the lowest Kd and IC50 values. The equilibrium dissociation constant (Kd) of DMB (compound 1, 5.11 ± 0.33 μM) with DYRK1A (Figure 1A) wasmuch smaller than that of harmine (81.7± 12.8 μM) (Figure 1B,C), suggesting a strong binding (Table S1, Figure S1B). Enzymatic activities assay further confirmed that DMB strongly inhibited DYRK1A, with lower IC50 values than harmine (Table S1). The crystal structure of DMB-DYRK1A complex was solved at 2.7 Å (PDB ID 6LN1, Table S2). DMB was bound with DYRK1A at the same binding pocket as harmine (Figures 1D, S1C), adopted an extended conformation and occupied the whole flat-shaped pocket (Figures 1D, S1D). Hydrogen bonds were formed between 3-hydroxyl and Leu241, also 8-hydroxyl and Lys188. Besides, there were hydrophobic interactions betweenDMBwith Phe170, Leu294, Val306, and Phe238 (Figure 1E,F). Interestingly, the conformation of DMB in crystal structure was almost completely overlapping with that predicted by docking (Figure 1F). In crystal structure, DMB rotated approximately 10 degree toward the direction of Phe170 with 1hydroxyl oxygen shifted 0.958 Å, impairing docking predicted hydrogen bondingwithGlu239 (Figure 1G,H). However, the shorter bond lengths of other two hydrogen bonds, and the closer distance to the hydrophobic environment of Phe170, favored the real conformation of DMB in the crystal structure (Figure 1F, Table S3).

The crystal structure of DMB-DYRK1A complex was solved at 2.7 Å (PDB ID 6LN1, Table S2). DMB was bound with DYRK1A at the same binding pocket as harmine ( Figures 1D, S1C), adopted an extended conformation and occupied the whole flat-shaped pocket (Figures 1D, S1D). Hydrogen bonds were formed between 3-hydroxyl and Leu241, also 8-hydroxyl and Lys188. Besides, there were hydrophobic interactions between DMB with Phe170, Leu294, Val306, and Phe238 ( Figure 1E,F). Interestingly, the conformation of DMB in crystal structure was almost completely overlapping with that predicted by docking ( Figure 1F). In crystal structure, DMB rotated approximately 10 degree toward the direction of Phe170 with 1hydroxyl oxygen shifted 0.958 Å, impairing docking predicted hydrogen bonding with Glu239 ( Figure 1G,H). However, the shorter bond lengths of other two hydrogen bonds, and the closer distance to the hydrophobic environment of Phe170, favored the real conformation of DMB in the crystal structure ( Figure 1F, Table S3).
This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. Compared to cytoplasmic NFATc1 in INS-1 cells, DMB treatment induced increase in nuclear NFATc1, which indicated that DMB may stimulate translocation of NFATc1 from the cytoplasm to the nucleus (Figure 2A). NFATc proteins were translocated into the nucleus to alter gene expression, ultimately activating transcription of insulinrelated genes such as Ccnd1, Ccnd2, Ccnd3, and CDK4. 9 RT-PCR revealed that levels of Ccnd1, Ccnd2, and Ccnd3 mRNA were dose-dependently upregulated after DMB treatment ( Figure 2B). By treatment with DMB, expression levels of cell-cycle inhibitors (eg, p15 INK4 , p16 INK4 , and p57 CIP2 ) were lower, whereas expression levels of relevant cyclins and CDKs (eg, cyclin A, CDK1) remained unchanged ( Figure 2C). E2F1 expression was increased but not in dose-dependence, and immunofluorescence results further confirmed the increase of E2F1, suggesting that DMB could promote cell cycles of INS-1 cells ( Figure 2D).
Gene expression profiling revealed that DMB strongly up-regulated CDKL4, GPR116, and INHBE, and downregulated GPR180, IRS1, DCN, LTBP1, and SP1 (Table  S4). These genes are involved in β-cell proliferation and blood glucose regulation. Gene expression levels of CDKL4, IRS1, and GPR116 were confirmed by RT-PCR (Figures S4A  pathway was involved in β-cell proliferation ( Figure S4B). Previous research also revealed combined inhibition of DYRK1A and TGF-β signaling generates further synergistic increases in β-cell proliferation. 10 DMB treatment led to a reduction in SMAD3 phosphorylation and a concomitant increase in SMAD3 abundance. Furthermore, PDX1 expression in INS-1 cells was increased, while FOXO1 expression was reduced after DMB treatment ( Figure 2F). DMB-induced inhibition of DYRK1A activated proliferation of β-cells that yielded double-labeled Ki67 + and insulin + nuclei ( Figure S5A). DMB also induced p-histone-H3, a marker of cell-cycle transition. The proliferationrelated indicators, p-P38 and p-erk, were increased at the protein levels in β-cells ( Figure S5B). β-cell death or DNA proliferation-related factors (Ccnd1, Ccnd2, Ccnd3, and CDK4) after DMB administration compared with control (n = 3, * P < .05, ** P < .01, ***P < .001). (C) Representative Western blot of key cell cycle molecules in INS-1 cells treated with DMB. Protein levels were quantified using grey value analyses by Image J software in the right. (n = 3, *P < .05, **P < .01, ***P < .001) (D) Representative images of immunofluorescence in INS-1 cells for E2F1 in response to DMB treatment (Magnification 400×, the scale bar indicates 50 μm.) (E) Relative gene expression levels determined by qPCR after treatment with DMSO or DMB for 72 hours (n = 3, *P < .05, **P < .01, ***P < .001). (F) Western blot analysis of PDK1, FOXO1, TGF-β2, SMAD3, and phospho-SMAD3 after treatment with DMSO or DMB for 72 hours. Protein levels were quantified using grey value analyses by Image J software in the right (n = 3, *P < .05, **P < .01, ***P < .001) F I G U R E 3 Effects of DMB on markers of β-cells proliferation and glycemic control in db/db mice in vivo. (A) Blood glucose concentrations measured weekly for 6 weeks. (B) Body weight increased steadily in each administration group. There was no significant damage induced by DMB was not observed by p-γ-H2AX labeling ( Figure S6).
The β-cell mass and size were higher in DMB-and harmine-treated db/db mice groups, insulin/EdU doublepositive cells were a little higher in DMB-treated group ( Figure S7A,B). The number of insulin/Ki67 doublepositive cells was approximately four-fold higher increased after DMB treatment ( Figure S7C,D).
For groups of db/db mice treated with DMB, harmine, and metformin, the 6-hour fasting blood glucose levels and symptoms of diabetes were ameliorated (Figures 3A,  S8, S9), with no significant changes of body weight (Figures 3B). The glucagon and insulin were gradually restored to normal levels after DMB treatment ( Figure 3C).
The blood glucose levels in all treated mice was decreased from 30 minutes after drug administration, and the areas under the curves (AUCs) of the DMB-treated groups were reduced ( Figure 3D,E). The serum insulin levels of treated group were increased at the sixth week after drug administration ( Figure 3F). Furthermore, DMB partially ameliorated dyslipidaemia and antioxidative stress in db/db mice ( Figure S10, Table S5). After treated with DMB, expressions of increased PDX1 and insulin, and reduced FOXO1 were consistent with the enhanced antioxidative capacity in serum ( Figure 3G).
In conclusion, DMB from Swertia bimaculata was found to be a novel and potent DYRK1A inhibitor. The crystal structure of DYRK1A-DMB complex revealed that hydrogen bonding and hydrophobic interactions enabled the shape-driven binding mode of DMB onto DYRK1A. Finally, we elucidated additional β-cell replication pathways, including Ca 2+ /CaN/NFAT/DYRK1A and TGFβ signaling pathways, which were mediated by DMBdependent DYRK1A inhibition. Compared to these known DYRK1A inhibitors, DMB exhibits better druggability, with low toxicity and broad availability. Next, DMB as a stimulator for β-cell proliferation will be combined with other classes of drugs to achieve therapeutic effect on diabetes.

A U T H O R C O N T R I B U T I O N S
LXC, HL and JCC designed the research and wrote the manuscript. MZZ, QZZ, and CLZ performed the research and acquired the data. CRW, KYY, ZRS, QQW and CL participated the experiments. All authors made substantial contributions to the analysis and interpretation of data.
All authors were involved in drafting the manuscript and all approved the final version. HL is responsible for the integrity of the work as a whole.

C O N S E N T F O R P U B L I C AT I O N
All the authors consent for publication.

C O N F L I C T O F I N T E R E S T
The authors declare no conflict of interest.

D ATA AVA I L A B I L I T Y S TAT E M E N T
The datasets used or analyzed in this study are available from the corresponding author on reasonable request.