Research on Saponin Active Ingredient of Stir-fried Dolichos lablab L. Kernel for Treatment of Type-2 Diabetes based on UHPLC-Q-Exactive Orbitrap MS and Network Pharmacology

: Background: Although the clinical effect of stir-fried Dolichos lablab L. kernel has been approved in modern traditional Chinese medicine, existing associated studies mainly focus on its clinical studies and chemical ingredients. However, there are few studies on pharmacodynamics material basis and molecular mechanism of stir-fried Dolichos lablab L. kernel in treatment of type-2 diabetes(T2DM), thus restricting the further development and utilization of stir-fried Dolichos lablab L. kernel. Methods : A qualitative analysis on saponin chemical ingredients of stir-fried Dolichos lablab L. kernel was performed using UHPLC-Q-Exactive Orbitrap MS. A total of 10 saponin ingredients were selected. Moreover, target screening, biological process and metabolism pathway analysis were accomplished by network pharmacology. Four key proteins (EGFR, IGF1, MAPK1 and PIK3R1) of type-2 diabetes were selected for molecular docking verification with saponin ingredients. Specifically, molecular dynamics simulation of ingredients which have strong bindings with proteins was conducted. Results : In this study, 16 saponin ingredients were identified from stir-fried Dolichos lablab L. kernel. There were 91 intersection targets and the KEGG pathway enrichment involved 20 relevant pathways. According to the molecular docking verification, saponin ingredients of stir-fried Dolichos lablab L. kernel can form stable binding with key protein targets. The molecular dynamics simulation further verifies stability and reasonability of the docking results. Conclusions : This study provides references to identification of efficient ingredients of stir-fried Dolichos lablab L. kernel, screening of quality markers and explanation of relevant action mechanism by


Background
Diabetes is directly related with pancreas which is an important component of the digestive system. Pancreas generates enzyme and hormones to help digestion of foods. Diabetes belongs to the "consumptive thirst" category in traditional Chinese medicine and it is closely related with spleen and stomach. In five elements, spleen and stomach of people are corresponding to the earth. They cover several digestive organs including pancreas of human body. According to the etiology and pathogenesis that diabetes is caused by Yin deficiency and manifested by dry heats, nitrifying spleen and Yin is the key of Yin nutrition. Radix Trichosanthis, Radix Ophiopogonis and Dolichos lablab L. which are characteristic of sweet and cool or sweet and cold are suggested as the basic medicines for symptoms of spleen and Yin deficiency. Among them, Dolichos lablab L. belongs to the spleen channel and it can invigorate spleen to eliminate dampness. Stir frying is equivalent to increase the strengthening effect of spleen. In traditional Chinese medicine, Dolichos lablab L. is a kind of white seeds of leguminosae plants, Dolichos L. Some study demonstrated that triterpenoid saponin has been separated from other similar plants of Dolichos L. Triterpenes and gluocosides of sterols can inhibit absorption of alcohol and sugar [1] . Although the clinical effect of stir-fried Dolichos lablab L. kernel has been approved in modern traditional Chinese medicine, existing associated studies mainly focus on its clinical studies and chemical ingredients. However, there are few studies on pharmacodynamics material basis and molecular mechanism of stir-fried Dolichos lablab L. kernel in treatment of type-2 diabetes(T2DM), thus restricting the further development and utilization of stir-fried Dolichos lablab L. kernel.
Hence, saponin chemical ingredients of stir-fried Dolichos lablab L. kernel were identified in the present study by using UHPLC-Q-Exactive Orbitrap MS technology. On this basis, active saponin ingredients related with T2DM were screened by network pharmacology. Moreover, action mechanism of screened active ingredients was interpreted. Results could provide references to screen quality markers. Molecular docking and molecular dynamics simulation were carried out based on structures of key proteins and saponin ingredients. Binding free energy was calculated and main thrust force for binding between proteins and saponin ingredients was analyzed. Besides, amino acids which play the key role in the binding between proteins and saponin ingredients were disclosed and the relevant binding mechanism was discussed. Research conclusions can provide theoretical references for development of more efficient saponin for treatment of T2DM.

Instruments and materials
Instruments and materials used in the experiment include the series connection of four-stage rodelectrostatic field orbital hydrazine high resolution mass spectrometer and the Ultimate 3000 ultrahigh performance liquid chromatography system (Seymour Fisher Technologies Inc, USA), BT25S electronic scales (Sartorius Company), KQ-300DB numerical control ultrasonic cleaner (Kunshan Ultrasonic Instrument Co., Ltd) and methyl alcohol (Sinopharm Group Chemical Reagent Co. LTD, batch No.: 20190910).

Preparation of solution samples
Stir-fried Dolichos lablab L. kernel powder (0.5g) was taken and 10% methyl alcohol (10ml) was added in for ultrasonic processing by 30min. Later, the mixture was filtered and the solution samples were gained.

Structural identification
According to the principle that the error between testing value and theoretical value is smaller than 5ppm and nitrogen laws, the expected compounds were gained through high-resolution extraction ion chromatography by combining the Compound Discover version 3.0. MS2 data of the expected compounds was collected through the PRM mode. Finally, saponin candidates were determined according to fragment ions, retention time and references.  Intersection between compound targets and disease targets was selected, which covered 91 genes.

Protein interaction network
The 91 treatment genes were input into the STRING database (https://string-db.org/) and species was limited within human. To increase reliability of results, the lowest interaction score was limited at high confidence (0.700) and results were input into cytoscape3.8.0 for visualization and analysis. The network was analyzed by cytoscape plug-in and individual node parameters in the network were calculated. In this study, 22 key genes were screened by using means of Degree, BetweennessCentrality and ClosenessCentrality as the thresholds. These 22 genes might be key targets for traditional Chinese medicine treatment of type-2 diabetes.

Active ingredients-target network
Compounds corresponding to 91 therapeutic targets were identified one by one. Besides, an active ingredients-therapeutic target network was built up and input into cytoscape3.8.0 for visualization and analysis. The network was analyzed by cytoscape plug-in and individual node parameters in the network were calculated. GO and KEGG enrichment analyses of 91 therapeutic genes were finished using clusterprofiler package of Rstudio. Moreover, results were visualized by ggplot2 and put in a descending order according to pvalue. GO shows top 10 items of BP, CC and MP, respectively. KEGG shows top 20 significant pathways.

Molecular docking
Molecular docking experiment between active ingredients and key targets was carried out based on AutoDock Vina_1.1.2 to verify the interaction activity between ingredients and key targets.
AutoDock Vina applied the semi-flexible molecular docking. In other words, pharmacologic molecules are flexible and rigidity of proteins is kept constant in the docking process, and docking results are evaluated by a semi-experience free energy function [2] .

Molecular dynamics simulation and ligand channels
The orbit of molecular dynamics simulation was finished by gromacs 2019-3. Graphs of root-meansquare deviation (RMSD), rotating rate, mean-square deviation (MSD), total energy, root-mean-square fluctuation (RMSF) of ligands and proteins were drawn. Clustering analysis was calculated. All potential channels for binding between ligands and proteins were explored using caver 2.0. Parameters were set as follows: minimum Probe radius =0.9, shell radius =3, and maximum shell depth in the surface region =4.

Structural identification based on UHPLC-Q-Exactive Orbitrap MS
Ingredients of stir-fried Dolichos lablab L. kernel were identified based on the UHPLC-Q-Exactive Orbitrap MS technology through abovementioned sample processing and analysis conditions, and then analyzed. Firstly, samples were injected into UHPLC-Q-Exactive Orbitrap MS and high-resolution mass data were acquired through MS full scan. Secondly, molecular formula of saponin was predicted by setting parameters of metabolic working flow for Compound Discover, and then confirmed by MS full scan and high-resolution extraction ion chromatography (HREIC). In this way, an ion list was generated. Finally, fragment ions were gained using UHPLC-Q-Exactive Orbitrap MS through parallel reaction monitoring mode (the above constructed ion list). Finally, saponin candidates were determined according to fragment ions, retention time and references. In this study, 16 saponin ingredients were identified from stir-fried Dolichos lablab L. kernel under the negative ion mode.
Results are shown in Table 1.

Construction and analysis of PPI network
Based on PPI relationship, the interaction relations among therapeutic targets of T2DM based on stir-fried Dolichos lablab L. kernel were acquired from the STRING database in order to understand its action mechanism better. A total of 22 key genes were screened by using means of Degree, BetweennessCentrality and ClosenessCentrality as thresholds (Table 2).These targets are hub nodes of the network and they are core targets for treatment of T2DM. Four key proteins of T2DM (EGFR, IGF1, MAPK1 and PIK3R1) were chosen for molecular docking verification.

GO bioprocess analysis and KEGG metabolic pathway analysis
GO enrichment analyzes functional distribution of key targets and key targets are in a descending sequence according to pvalue. GO shows top 10 items of BP, CC and MF, respectively (Fig.1). KEGG shows the top 20 significant pathways. Additionally, the bubble graph is drawn (Fig.2). PI3K-AKT signaling pathway (Fig.3) and EGFR tyrosine kinase inhibitor resistance (Fig.4) process.

Building of an active ingredients-targets-pathway network
The Saponin active ingredients-targets-signal pathway network was built using Cytoscape3.8.0 (Fig.5). It can be seen from Fig.5 that there are 101 nodes (10 compounds and 91 targets) and 593 sides. The V-shaped nodes represent compounds, while the round nodes represent genes. Colors of nodes from light to dark represent the increasing values of node degree. The therapeutic mechanism of stir-fried Dolichos lablab L. kernelsaponin ingredients to T2DM is related with inflammatory response, autoimmune damage and several other morbidity theories. This study interpreted the therapeutic mechanism of stir-fried Dolichos lablab L. kernelsaponin ingredients to T2DM from different perspectives.

Molecular docking
Molecular docking experiment between 10 saponin active ingredients and the screened key protein targets related with T2DM was carried out using AutoDock Vina_1.1.2. AutoDock Vina evaluates the bonding strength between micromolecules and proteins mainly through affinity, that is, the fitted ΔG value after calculation. If ΔG value is lower than 0, the ligands can bind with receptors spontaneously. Moreover, the lower ΔG value brings the higher affinity and the active ingredients are easier to bind with receptors [12] . Results are listed in Table 3. Graphs are drawn for each protein and the best docked compound in Table 3 (Fig.6). Clearly, lablaboside_E enters into the active pockets of EGFR completely and forms one or two hydrogen bonds with active site residues Thr830, Cys773,

Molecular dynamics simulation and ligand channels
According to molecular docking results, molecular dynamics simulation on Chikusetsusaponin Iva which has relatively binding with the key protein PIK3R1 was carried out, which further verified stability and reasonability of docking results. The trajectory of protein rotating value is shown in Fig.7.
This trajectory is used to study the mean distance of each protein atom to the center, thus illustrating closeness of each corresponding proteins. The trajectory of protein MSD is used to calculate the error between each ligand and protein. The total energy trajectory graph can be used to evaluate stability of ligand-protein compound.
The mean potential energy, kinetic energy and total energy which are needed for ligand-protein binding were calculated as -2923388.5KJ/mol, 552577.6KJ/mol and -2370811KJ/mol, respectively. Based on these parameters, binding stability of ligand-protein compounds was compared.
Representative structure of the ligand-protein compounds was discussed through clustering analysis. There are 7 classes of typical structures of Chikusetsusaponin Iva-protein compounds (Fig.8). (a) (b) Fig.9 Three-dimensional simulation of ligand channels

Discussions
Four key proteins, including EGFR, IGF1, MAPK1 and PIK3R1, were chosen. Chu et al. [13] discovered that systematic insulin-like growth factor-1(IGF1-1) therapy can reverse the hyperpathia phenomena of mice with diabetic peripheral neuropathy and improve their mobility. Mauvaris Jarvis [14] pointed that PI3 kinase plays the crucial role in metabolism of insulin. Moreover, the levels of p85α and AS53 could be decreased by hybrid damages to PIK3R1 genes, thus enabling to increase insulin sensitivity and lower fasting and postprandial blood glucose levels. As result, morbidity of diabetes which is related with genetic insulin resistance is decreased significantly. MAPK3 and MAPK1 are two crucial MAPKs in MAPK/ERK cascading. MAPK/ERK pathway plays an important role in diabetes and its complications. It is reported that blocking EGFR can inhibit infiltration and oxidative stress of kidney immune cells and strengthen autophagy activity of pancreas, finally improving diabetic nephropathy [15] . blood glucose [1] . The blood glucose lowering mechanism might be related with improving immunological functions of patients with diabetes, adjusting combination of insulin and receptors, and increasing sensitivity of body to insulin.
According to metabolic pathway analysis, EGFR signal pathway is the pathway with a relatively high enrichment factor. Pathways which have relatively more enrichment targets include PI3K-Akt signal pathway, MAPK signal pathway, Ras signal pathway, Rap1 signal pathway, and pathways of proteoglycans in cancer. Some studies demonstrate that therapeutic pathways of T2DM based on saponin of stir-fried Dolichos lablab L. kernel include cancer pathways like proteoglycans in cancer.
According to a large-scaled epidemiological investigation, morbidity of cancers at some specific positions in patients with T2DM is significantly higher compared to that of other patients [16] . In addition, cancer mortality of patients with T2DM is relatively higher. The mechanism can be interpreted by that the effect of hyperinsulinemia on microenvironment of cancer cells and signal transduction in cells of diabetic patients is in favor of tumor growth [17] . It can be seen from molecular docking experiment that all compounds can enter into the active sites of targets and form 3-14 hydrogen bonds with active site residues. Moreover, Van der Waals' force exists universally between compounds and active sites. According to affinity, all compounds form strong binding with targets, indicating that compounds can treat T2DM by adjusting activity of these targets.

Conclusion
Based on acquisition of accurate chemical ingredients, active ingredients of stir-fried Dolichos lablab L. kernel for treatment of T2DM and relevant therapeutic mechanism are discussed in this study by combining UHPLC-Q-Exactive Orbitrap MS analysis and network pharmacology. Research conclusions lay a solid foundation for identification of effective ingredients, screening of mass markers and interpretation of action mechanism of stir-fried Dolichos lablab L. kernel.

Availability of data and materials
The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request Abbreviations T2DM: type-2 diabetes