Differential compounds of licorice before and after honey roasted and anti-arrhythmia mechanism via LC-MS/MS and network pharmacology analysis

Abstract Licorice is mainly used to treat cough in clinics. After being roasted with honey, licorice has a cardiac protective effect. Due to its different sources, the chemical components in licorice are also diverse. To distinguish the different compounds between three sources of raw licorice (RL) and between three sources of honey-roasted licorice (HRL), a UPLC-QTOF-MS/MS method was established. Eighteen batches of samples were analyzed which were derived from Glycyrrhiza uralensis Fisch., Glycyrrhiza inflata Bat., and Glycyrrhiza glabra L. In addition, the potential key pathways of HRL to anti-arrhythmia were explored by network pharmacology. Among the three sources of RL, 77 differential compounds were detected in the positive ion mode; and 62 differential compounds were detected in the negative ion mode, of which 21 differential compounds were detected in both ion modes. Twenty-nine differential compounds were identified before and after honey-roasted of G. uralensis Fisch., 51 and 17 differential compounds were identified in G. glabra L. and G. inflata Bat., receptively. The network pharmacological analysis predicted that Rap1, Ras, PI3K-Akt, and cAMP signaling pathway might be closely related to the anti-arrhythmia mechanism of HRL. This study provides a theoretical reference for the quality improvement and clinical application of licorice. Graphical Abstract


Introduction
Licorice is harvested from the dried roots and rhizomes of Glycyrrhiza uralensis Fisch., Glycyrrhiza inflata Bat., or Glycyrrhiza glabra L. in the legume family. It is named "Gancao" because of its special sweet taste. [1] Previous studies have shown that licorice contains triterpenoids, flavonoids, coumarins, polysaccharides, and alkaloids, [2,3] which have well pharmacology effects, such as anti-cancer, anti-oxidant, anti-inflammatory, anti-fibrotic, anti-bacterial and anti-viral. [4][5][6][7][8] There are three plant sources of licorice, and the compounds and pharmacological activities of licorice from different sources are varying. Song et al. [9] found that three sources of licorice showed significant biosynthetic preferences, especially in coumarins, chalcones, isoflavones, and flavones. Yang et al. [10] found that the flavonoid components, i.e., liquiritin, isoliquiritin, liquiritigenin, and isoliquiritigenin were the highest in G. uralensis Fisch., the second highest in G. inflata Bat., and lowest in G. glabra L. The difference in the content of the different sources components will lead to differences in pharmacological effects. Honey roasting is a traditional Chinese medicine (TCM) processing method. After roasting with honey, it can improve the effect of licorice in tonifying the spleen and stomach, benefiting the "qi" and restoring the pulse. Recent research has shown that heating and adding honey during processing can have some certain effects on the constituents of licorice. [11] The results of our previous study showed that the contents of liguiritigenin-7-O-D-apiosyl-4 0 -O-D-glucosid and isoliquiritin apioside increased after honey roasting. [12] However, there is no relevant research on the differences in chemical components of licorice before and after honeyroasted from different sources. Therefore, it is necessary to investigate the differences in the composition of licorice from different sources before and after honey roasted, and preliminarily explore the cardioprotective pathway of HRL.
Ultra-performance liquid chromatography quadrupole time-of-flight tandem mass spectrometry (UPLC-QTOF-MS/MS) has the advantages of high speed, high efficiency, and high resolution, and has been widely used in the identification and pharmacokinetic research of components in TCM and prescriptions in recent years. [13][14][15][16] Network pharmacology is an emerging discipline that elucidates the mechanism of disease and drug action from the holistic perspective of biological networks. [17] The network pharmacology method is used to screen drug active ingredients and predict their possible pathways, which can provide a theoretical reference for the development of new drugs in Chinese medicine.
In this experiment, three sources of RL and HRL were selected for component identification by UPLC-QTOF-MS/MS. Principal Component Analysis (PCA), t-test, and Orthogonal Partial Least Squares-Discriminant Analysis (OPLS-DA) multivariate statistical methods were used to preliminatively screen the different components between RL and HRL, combined with network pharmacological analysis to predict the possible pathways of anti-arrhythmia of roasted licorice. This study provides a theoretical reference for the quality improvement and clinical application of licorice. were purchased from CHENGDU HERBPURIFY Co., Ltd (China), with the purity more than 98%. Deionized water was prepared by the Milli-Q system.

Reagents and standards
Eighteen batches of raw and roasted licorice decoction pieces were collected, details are shown in Table 1. The RL decoction pieces are provided by China Medico Corporation. The HRL decoction pieces of G. inflata Bat. and G. glabra L. were prepared according to the honeyroasting method in the Chinese Pharmacopeia (2020 Edition). [1] All raw and roasted licorice decoction pieces were authenticated by Associate Professor Lihong Chen of Nanjing University of Traditional Chinese Medicine.

Preparation of roasted licorice
Refer to the Chinese Pharmacopeia (2020 Edition) 0213 in the general rules of concoction for the preparation of honey roasted method.

Preparation of licorice aqueous extract
Dried licorice samples were boiled with 10-fold water for 60 min, and then the filtrate was collected. The residue was added 8-fold water and boiled for 60 min. Following, the filtrate was collected again. Finally, the above two filtrates were mixed to obtain the licorice aqueous extraction.

Preparation of the sample solution
The aqueous extract of licorice was filtered with gradient methanol through LC-C 18 SPE solid phase extraction column, and the filtrate was collected to obtain a 5 mg/mL solution. The filtrate was diluted to 500 lg/mL again, and the solution was centrifuged at 13,000 r for 10 min. Next, the supernatant was passed through a 0.22 mm organic filter membrane and stored at 4 C for analysis.

Mass spectrometry conditions
The

Components identification
Based on the previous research reports on the chemical compositions of licorice, a library of licorice chemical compositions was established, which included the names, molecular formula, CAS number, relative molecular mass, and class. The mol structures of the target compounds in PubChem [18] websites using the CAS numbers of the chemical components were downloaded. A library of mol files for secondary fragment ion matching was created. The original data collected by Analyst TF 1.6 were imported into PeakView 1.2 software, and the XIC files were created in different modes using the new session of the XIC Manager module. The primary mass numbers and isotopes extracted from the total ion flow diagram of each compound were compared, and the mass-to-nucleus ratio error range within ±10 ppm and retention time error range within 2 min were used as the initial screening criteria. The molecular structures of the compounds were also matched with the secondary ion fragments, and the components with a match >75% were finally identified.

Differential compounds analysis
The original data of each group collected by Analyst TF 1.6 were imported into MarkerView1.2.1 software for t-test analysis. Then the analysis results were exported in text form and imported into SIMCA 14.1 software for PCA and OPLS-DA analysis. The components that met both t-test analysis p < 0.05 and OPLS-DA analysis VIP > 1 were the differential compounds.

Network pharmacology analysis
Licorice components-targets prediction and mapping of protein interaction network (PPI) with arrhythmia diseases By reviewing the research, a total of seven pharmaceutical components, which can exert anti-arrhythmia or cardioprotective effects through antioxidant, antiapoptotic, and antifibrotic, were reported, namely isoliquiritigenin, glycyrrhetic acid, licochalcone A, liquiritigenin, liquiritin, glycyrrhizic acid, and isoliquiritin (GC1-7). [19][20][21][22][23][24][25][26][27] The canonical SMILES numbers of the components were found in the PubChem database, and the numbers were imported into the Swiss Target Prediction database [28] to obtain the corresponding target information of the components. The Gene Cards [29] database was searched by entering "Arrhythmia" to obtain arrhythmia-related targets. The gene name of the active ingredient of licorice and the gene name of the arrhythmia target were input in the Draw Veen Diagram [30] to draw a Veen diagram and obtain the common key targets. Then the key targets were obtained by the String [31] website to map the protein interaction network of disease and drug common targets. Constructing components-target-disease network map The above pharmaceutical ingredients, targets, and arrhythmia were imported into Cytoscape3.6.0 software for network construction. The nodes of different shapes were used to represent the pharmaceutical ingredients, key targets, and diseases, and connecting lines were used to indicate the relationship between different ingredients.

GO and KEGG analysis
Key targets were imported into the Metascape [32] website for GO and KEGG pathway enrichment analysis, respectively.
The key nodes as well as signaling pathways, which are involved in the processes, were explored.

Components identification
Two hundred and twenty-five components were identified from all samples, of which 165 were identified in both positive and negative ion modes. Thirty specific components were identified in positive ion modes and another 30 specific components were identified in negative ion modes. There are 11 components that were compared with the standards. The total ion flow diagrams of RL and HRL in positive and negative ion modes for the three sources are shown in Figure S1. The differential compounds information of three sources of licorice is shown in Table S1.

Components cleavage pattern
Triterpenoids and flavonoids are abundant chemical components in licorice. [9] In this study, the cleavage patterns of triterpenoids and flavonoids in licorice were briefly analyzed using glycyrrhizic acid and isoliquiritin apioside as examples, respectively.
Take glycyrrhizic acid as an example to briefly analyze the mass spectrometric cleavage pattern of triterpenoid components in licorice: [M þ H] þ excimer ion peak m/z 823 was formed in positive ion mode. First, one molecule of the glucuronide group was lost to get fragment ion m/z 647. Second, the fragment ion m/z 471 for glycyrrhetinic acid was formed by losing a molecule of glucuronide group again. Last, one molecule of H 2 O continued to be lost to get fragment ion m/z 453 for glycyrrhetinic acid, which was identified as glycyrrhizic acid by comparison with the reference substance. The possible mass spectrometric cleavage pathway is shown in Figure 1A. The mass spectrometric cleavage pattern of flavonoid components in licorice was briefly analyzed by taking isoliquiritin apioside as an example: the [M þ H] þ excimer ion peak m/z 551 was formed in the positive ion mode. First, one molecule of apigenin group was lost to get fragment ion m/z 419 for isoliquiritin. Second, the fragment in m/z 257 for isoliquiritigenin was formed by losing a molecule of glucose group. Last, one molecule of OC 6 H 5 and H 2 O continued to be lost to get fragment ion m/z 147 or lost one molecule of OC 8 H 8 to get fragment ion m/z 139, and the possible mass spectrometric cleavage pathways are shown in Figure 1B.

Differential compounds analysis
PCA results showed that RL from different sources was clustered into one category in positive and negative ion modes, indicating that there were significant differences in the types or contents of components in RL from different sources ( Figure 2). Further OPLS-DA analysis in supervised mode (Figure 3) was performed to predict the VIP value of the data. The components with VIP > 1 and t-test's p < 0.05 were screened and matched with the information of compounds identified in PeakView1.2 (Tables S2 and S3). A total of 77 different compounds were identified in the positive ion mode and 62 in the negative ion mode, mostly triterpenoids, flavonoids, and a few coumarins. There were 21 common differential compounds of three sources in both positive and negative ion modes. There are glycyrrhizic acid, ononin, glycyroside, glycyrrhetic acid 3-o-glucuronide, isoononin, licorice saponin G2, licorice saponin H2, pinocembrin, uralsaponin A, uralsaponin B, uralsaponin N, uralsaponin U, (18b,20a)-glycyrrhizic acid, 18a-glycylrrhizin,  glucoside, among which glycyroside, glycyrrhizoside G2, and glycyrrhizoside H2 were not identified in the samples of G. glabra L., and all the components were found in G. inflata Bat. except licuraside, which was the highest in G. uralensis Fisch.
Between RL and HRL, PCA results showed that they were clustered into two category in positive and negative ion modes, indicating that compounds of RL were significantly changed after honey-roasted ( Figure 4). Further OPLS-DA analysis in supervised mode ( Figure 5) was performed to predict the VIP value of the data. A total of 29 differential compounds were identified in G. uralensis Fisch. RL and HRL were compared in positive and negative ion modes, 51 in G. glabra L. and 17 in G. inflata Bat (Tables  S4-S6). Most of these components were triterpenoids and flavonoids. In G. uralensis Fisch. sources, glycyrrhizic acid, licochalcone D, licorice-saponin A3, and uralsaponin F are the special differential compounds. Isoliquiritin apioside, liguiritigenin-7-O-D-apiosyl-4 0 -O-D-glucoside, licorice saponin A3, 6 00 -O-acetylliquiritin, isovestitol, licoagrochalcone A, licoisoflavone B, morachalcone A, phaseollinisoflavan, erythrinin C, gancaonin M, and parvisoflavones-A are the special differential compounds of G. glabra L. sources. And licochalcone A, glycyroside, liquiritigenin, pinocembrin were special differential compound in G. inflata Bat. sources.
According to the relevant research, t-value <0 in the t-test indicated that the content of components was increased in the concoction; t-value >0 indicated that the content of components was decreased. [33] The statistical results indicated that there was a change in the constituents in the three sources of licorice after honey roasted, which may be the reason why the raw and roasted licorice samples were significantly clustered into two classes in the PCA clustering analysis.

Network pharmacology analysis
Venn diagram of the seven pharmaceutical compounds targets and arrhythmia targets were made and 51 common targets as the key targets between the seven pharmaceutical compounds and arrhythmia were screened. The key targets were imported into String website to draw PPI network diagram, after hiding the isolated targets, there were 99 nodes and 320 relationships ( Figure S2). Cytoscape3.6.0 software was used to draw the pharmacophores-disease-targets map, in which GC1 (isoliquiritigenin), GC2 (glycyrrhetic acid) and GC4 (liquiritigenin) were more closely related to each target ( Figure 6). The common targets were imported into Metascape online analysis platform for GO and KEGG enrichment analysis, in which 956 results were obtained for GO analysis, including 832 for biological process (BP), 52 for cellular component (CC), and 72 for molecular function (MF). One hundred and twenty-six results were obtained for KEGG analysis, and the top 10 signaling pathways were Proteoglycans in cancer, Rap1 signaling pathway, Lipid and atherosclerosis, Ras signaling pathway, cAMP signaling pathway, Pathways in cancer, MicroRNAs in cancer, Alzheimer disease, PI3K-Akt signaling pathway and Pathways of neurodegeneration-multiple diseases (Figure 7). The Rap1 signaling pathway, Ras signaling pathway, PI3K-Akt signaling pathway, and cAMP signaling pathway may be closely related to the anti-arrhythmic effects of licorice components.

Discussion
The changes in the content of licorice after the processing may be the reason for the different use of licorice in raw and roasted forms. The results showed that glabridin was the characteristic component of G. glabra L., which was only identified in G. glabra Bat. and only detected in the negative ion mode. [34] The content of glabridin increased after honey roasted, which was consistent with previous reports. [35] The overall content of isoliquiritin apioside, isoliquiritin, isoliquiritigenin, and liquiritigenin in HRL from the three origins showed an increasing trend, which presumably due to the decomposition or interconversion of the components within licorice caused by high temperature. For example, glucoisoliquiritin apioside can be converted into isoliquiritin apioside to increase its content. Isoliquiritin apioside hydrolyzes one molecule of apioside to obtain isoliquiritin at high temperature conditions. Isoliquiritin continues to decompose to obtain isoliquiritigenin. Liquiritin hydrolyzes loss of one molecule of glucose to obtain liquiritigenin. [11] Previous research showed that isoliquiritin, glycyrrhizic acid, and liquiritigenin have anti-arrhythmia and cardioprotective effects. [23,24,27] The efficacy of HRL in the treatment of "palpitation and pulse generation" may be related to the increased content of these anti-arrhythmic components. With the help of high resolution mass spectrometry, 225 compounds were screened from all licorice samples. It is worth noting that mass spectrometry cannot distinguish isomers in the absence of standards. Other methods are still needed to help distinguish the isomers in our results.
Rap1 is a small GTPase protein with high homology to Ras, which is activated upon binding to GTP, a process regulated by EPAC (cAMP directly activated exchange protein). Rap1-GTP inhibits the production of mitochondrial ROS and reduces susceptibility to early arrhythmia after cardiac depolarization. [36] Similarly, the Ras family, in which guanosine triphosphate binds to G proteins, is one of the major components of signaling endo-transduction required for normal cardiac growth. There are three main Ras genes in humans, H-Ras, K-Ras, and N-Ras, among which H-Ras mainly regulates cardiomyocyte size, while K-Ras regulates cardiomyocyte proliferation and N-Ras is least studied in cardiomyocytes and less relevant to cardiac defects. [37] In cardiovascular disease, PI3K is activated in PI3K-Akt signaling pathway to produce PIP3 that phosphorylates Akt, which activates or inhibits downstream target proteins, thereby regulating cell proliferation, differentiation, apoptosis, and migration. Researches have shown that enhanced PI3K activity can delay or prevent the progression of heart disease, [38] and reduced PI3K activation increases susceptibility to atrial fibrillation. [39] cAMP (cyclic adenosine monophosphate)is a second intracellular messenger that activates PKA (phosphokinase A) to phosphorylate downstream target proteins, thereby affecting cellular metabolism and cellular behavior. Moreira et al. [40] found that ACMs (atrial myocytes) secrete calcitonin as a first messenger that acts on calcitonin receptors on the membranes of ACFs (atrial fibroblasts), which can increase intracellular cAMP levels, thereby inhibiting the secretion of BMP1 (bone forming protein 1) and reducing myocardial fibrosis to decrease AF susceptibility. It is worth noting that the signaling pathways do not act individually but have many cross-linkages, for instance, cAMP not only activates PKA but also affects EPAC, thereby regulating the Rap1 signaling pathway. [36] Chinese medicine is characterized by multiple components, multiple pathways, and multiple targets, so comprehensive consideration should be given to the research of therapeutic effects of HRL on cardiac arrhythmias while paying attention to the components whose contents changed significantly after being processed, which may be closely related to its antiarrhythmic and cardioprotective effects.

Conclusion
The components of TCM are complex, so the identification of components in TCM is needed for quality improvement and clinical applications. UPLC-QTOF-MS/MS has become a key complex components identification technology of multi-component profiles, especially in TCM. In this study, UPLC-QTOF-MS/MS was established to identify the components of RL and HRL, and combined with multivariate statistical methods to screen the differential compounds between different sources of RL and between RL and HRL in the same sources. The possible signaling pathways of licorice in anti-arrhythmia were predicted by network pharmacology. This method system is not only suitable for the identification and potential signaling pathways prediction of licorice but also for other varieties of TCM.

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

Funding
This work was supported by National Key R&D Program of China under Grant (Nos. 2018YFC1706500 and 2018YFC1707000).