Structure Identification and Anti-Cancer Pharmacological Prediction of Triterpenes from Ganoderma lucidum

Ganoderma triterpenes (GTs) are the major secondary metabolites of Ganoderma lucidum, which is a popularly used traditional Chinese medicine for complementary cancer therapy. In the present study, systematic isolation, and in silico pharmacological prediction are implemented to discover potential anti-cancer active GTs from G. lucidum. Nineteen GTs, three steroids, one cerebroside, and one thymidine were isolated from G. lucidum. Six GTs were first isolated from the fruiting bodies of G. lucidum, including 3β,7β,15β-trihydroxy-11,23-dioxo-lanost-8,16-dien-26-oic acid methyl ester (1), 3β,7β,15β-trihydroxy-11,23-dioxo-lanost-8,16-dien-26-oic acid (2), 3β,7β,15α,28-tetrahydroxy-11,23-dioxo-lanost-8,16-dien-26-oic acid (3), ganotropic acid (4), 26-nor-11,23-dioxo-5α-lanost-8-en-3β,7β,15α,25-tetrol (5) and (3β,7α)-dihydroxy-lanosta-8,24-dien- 11-one (6). (4E,8E)-N-d-2′-hydroxypalmitoyl-l-O-β-d-glucopyranosyl-9-methyl-4,8-spingodienine (7), and stigmasta-7,22-dien-3β,5α,6α-triol (8) were first reported from the genus Ganodema. By using reverse pharmacophoric profiling of the six GTs, thirty potential anti-cancer therapeutic targets were identified and utilized to construct their ingredient-target interaction network. Then nineteen high frequency targets of GTs were selected from thirty potential targets to construct a protein interaction network (PIN). In order to cluster the pharmacological activity of GTs, twelve function modules were identified by molecular complex detection (MCODE) and gene ontology (GO) enrichment analysis. The results indicated that anti-cancer effect of GTs might be related to histone acetylation and interphase of mitotic cell cycle by regulating general control non-derepressible 5 (GCN5) and cyclin-dependent kinase-2 (CDK2), respectively. This research mode of extraction, isolation, pharmacological prediction, and PIN analysis might be beneficial to rapidly predict and discover pharmacological activities of novel compounds.

Compound 1 was obtained as a yellow oily solid, yielded a positive reaction to 10% H 2 SO 4 -EtOH reagent. 1  Compound 2 was obtained as a yellow oily solid, yielded a positive reaction to 10% H 2 SO 4 -EtOH reagent. The structure was the same as that of compound 1, except for the substituent of C-26, with a carboxy group on it. 1  Compound 3 was obtained as a yellow oily solid, yielded a positive reaction to 10% H 2 SO 4 -EtOH reagent. The structure was the same as that of compound 1, except for its more than a hydroxyl of C-28 and opposite stereoconfiguration of chiral C-15. Compound 4 was obtained as a white amorphous power, yielded a positive reaction to 10% H 2 SO 4 -EtOH reagent. The structure was possessed two oxygenic five-membered rings system in the side chain of lanostane skeleton. 1  Compound 5 was obtained as a white solid, yielded a positive reaction to 10% H 2 SO 4 -EtOH reagent. The structure was the same as that of compound 1, except for its the degradation of the carboxyl group and the attachment of a hydroxyl group to C-25 and opposite stereoconfiguration of chiral C-15, with its more than two H atoms of C-16 and C-17. 1

Pharmacophoric Profiling of GTs
Reverse pharmacophoric profiling of six isolated GTs and two positive GTs (GT-A and GT-D) was implemented to explore the pharmacological activity of GTs. In total, the reverse profiling results of eight GTs contained 30 human targets, and targets from the other creatures were eliminated because of less correlation with cancer. Top 20 interaction information of GTs and targets were showed in Table 1, based on the hit number of SBP models and fit values. Compound 1, 2, 3, and 5 and two positive GTs could match with CDK2 pharmacophore model (PDB: 1KE5) with fit value more than 0.5, which indicated they might have better biological action to CDK2. The superior mapping result of compound 3 and the best model was showed in Figure 1. In general, the structural framework of the eight GTs is oxygenated lanostane, wherein the chemical structure of compound 1, 2, 3 and 5 are more similar to that of GA-A and GA-D, because the branch chain of substituent of C-17 is a straight chain, including carboxyl, hydroxyl, or methyl ester, which might be the common structural basis of CDK2 activities, while, the branch chain of compound 4 had two oxygenated quaternary carbons (C-23 and C-17) instead of the carbonyl (C-23) and methyl (C-17) in compound 3, and the branch chain of compound 6 had also no carbonyl or hydroxyl (C-23), which may be the cause of poor biological action to CDK2.
In order to observe the biological effect of GTs visually, ingredient-target network of GTs was constructed in Figure 2. According to the classes of corresponding diseases and the number of interactive compounds for each target, 19 high frequency targets were obtained, which could interact with more than one compound. As shown in the top of network, eight targets interacted with GA-A and GA-D could also be hit by the other six GTs, including CYP2C9, CDK2, PPAR γ, GCN5, MEK1, PTP1B, COX1, and COX2. It indicated that GTs might have common pharmacological activities related to cancer, cardiovascular diseases and metabolic syndrome. The 11 specific high-frequency targets of six isolated GTs were shown in the bottom of network, which included the cancer-related targets (CDK5, c-Met, Pim-1, Chk1, and Syk), metabolism-related targets (FXR, CYP3A4, 11beta-HSD1, FGFR1, PDE5A) and immune-related targets of Lck. These 11 specific targets of six isolated GTs might suggest specific pharmacological activities of GTs based on this extraction process. Other non-high-frequency targets might indicate miscellaneous pharmacological activities of GTs, which needed to analyze each ingredient-target interaction of a single GT. A brief description of high frequency targets was given in Table 2. * Hit is the hit number of SBP models from one target.    * Hit is the hit number of SBP models from one target.

Protein Interaction Network Analysis of GTs
In this study, 19 high frequency targets were used to construct PIN ( Figure 3) for further predicting the anti-cancer activity of GTs. The generated network contained 185 nodes (proteins) and 733 edges (relations). In order to cluster the pharmacological activity of GTs, 12 modules ( Figure 4) were identified from PIN by MCODE algorithm. Functional enrichment analysis by BinGO method was shown in Table 3 to annotate potential biological function of the proteins in each module. The results showed that GTs played an essential role in pharmacodynamics with the biological processes, such as regulation of the cell division cycle, lipid metabolic process, protein acetylation, apoptotic signaling pathway, and so on. Half of the modules were related to cancer, including module 1, 3, 6, 7, and 11.   Histone acetylation (HAT) of module 1 was closely related to cancer, including KAT2A, TAF10, TAF9, TADA2A, TADA3, and so on. KAT2A is the gene name of GCN5, which is a typical transcriptional cofactor and requirement for HAT [34]. GCN5 could form multiple complexes by TAFs, including TAF9 [35,36], TADA2A, TADA3 [37], which is quite essential to cancer and HAT. TAFs are important to regulate the cellular differentiation and the speed of cancer cell migration [37]. TAFs are also the important part of transcription factor complex IID (TFIID), which could regulate and control cellular totipotency and differentiation in cancerous and normal cells. Therefore, GTs might regulate HAT by acting GCN5, which was beneficial to inhibit tumor proliferation and metastasis.
Module 3 showed anti-cancer activity including CCNE1, CCNB1, RBL1, RB1 and CDK2. Cyclindependent kinase-2 (CDK2) is an important member in protein kinase family, which could regulate  Histone acetylation (HAT) of module 1 was closely related to cancer, including KAT2A, TAF10, TAF9, TADA2A, TADA3, and so on. KAT2A is the gene name of GCN5, which is a typical transcriptional cofactor and requirement for HAT [34]. GCN5 could form multiple complexes by TAFs, including TAF9 [35,36], TADA2A, TADA3 [37], which is quite essential to cancer and HAT. TAFs are important to regulate the cellular differentiation and the speed of cancer cell migration [37]. TAFs are also the important part of transcription factor complex IID (TFIID), which could regulate and control cellular totipotency and differentiation in cancerous and normal cells. Therefore, GTs might regulate HAT by acting GCN5, which was beneficial to inhibit tumor proliferation and metastasis.
Module 3 showed anti-cancer activity including CCNE1, CCNB1, RBL1, RB1 and CDK2. Cyclin-dependent kinase-2 (CDK2) is an important member in protein kinase family, which could regulate division cycle in eukaryotic cell [38]. Previous study had demonstrated that overexpressing CDK2 would lead to abnormal regulation in cell cycle, which was the important feature for overexpression of cancer cells. CDK2 could also combine with cyclin (gene name: CCN) and phosphorylate downstream targets RB1 and RBL1 [39,40]. Therefore, CDK2 inhibitors are essential for anti-cancer treatment. Based on the pharmacophoric profiling of GTs, GTs are possible to inhibit CDK2 activity for downregulating DNA replication and mitotic cell cycle.
Module 6 and module 7 indicated that GTs might play an important role in anti-tumor effect by regulating receptor tyrosine kinase signaling pathway, especially fibroblast growth factor receptor (FGFR) signaling pathway. Module 11 showed that GTs might inhibit the negative regulation of apoptosis by activating the serine/threonine protein kinase (Pim1), which could promote the apoptosis of tumor cells.

Plant Material
The fruiting bodies of G. lucidum were collected from Hainan in February 2013. The samples were authenticated by Professor Chun-Sheng Liu (School of Chinese Materia Medica, Beijing University of Chinese Medicine, Beijing, China). A voucher specimen (No.130201) has been deposited at the 513 lab of School of Chinese Pharmacy, Beijing University of Chinese medicine.

Extraction
The dried fruiting bodies of G. lucidum (30 kg) were milled and soaked into 240 L 70% ethanol in the ratio of 1:8 (w/v). The mixtures were left for 12 h at room temperature, and then refluxed for three times, each time for two hours. The extract was filtered through a Whatman number 3 filter paper and concentrated with the rotary evaporator (Shanghai Yarong Biochemical Instrument Company, Shanghai, China) at 55˝C to obtain the crude 70% ethanol extract of G. lucidum. The crude extract was dissolved in water (12 L), and extracted with ethyl acetate (12 L) and n-butyl alcohol (12 L) for three times, respectively. The extract was filtered and concentrated, yielded an ethyl acetate fraction (463 g, dry weight) and an n-butyl alcohol fraction (160 g, dry weight).

Reverse Target Identification
Ligand profile module was utilized to reversely identify targets for six isolated and two known positive GTs in Discovery Studio 4.0 (DS, Accelrys Inc., San Diego, CA, USA). Previous researches indicated GA-A and GA-D have the clear anti-cancer activity [18,41]. Therefore, in this paper, GA-A and GA-D were selected to be the typical positive compounds for guiding the anti-cancer bioactive prediction of six isolated GTs. Eight compounds were first minimized in CHARMm force field with MMFF 94 partial charge. Diverse conformations of 8 GTs were constructed by the BEST method within 255 conformations, and the relative energy threshold was set to 20.0 kcal/mol.

Reverse Target Identification
Ligand profile module was utilized to reversely identify targets for six isolated and two known positive GTs in Discovery Studio 4.0 (DS, Accelrys Inc., San Diego, CA, USA). Previous researches indicated GA-A and GA-D have the clear anti-cancer activity [18,41]. Therefore, in this paper, GA-A and GA-D were selected to be the typical positive compounds for guiding the anti-cancer bioactive prediction of six isolated GTs. Eight compounds were first minimized in CHARMm force field with MMFF 94 partial charge. Diverse conformations of 8 GTs were constructed by the BEST method within 255 conformations, and the relative energy threshold was set to 20.0 kcal/mol.

Reverse Target Identification
Ligand profile module was utilized to reversely identify targets for six isolated and two known positive GTs in Discovery Studio 4.0 (DS, Accelrys Inc., San Diego, CA, USA). Previous researches indicated GA-A and GA-D have the clear anti-cancer activity [18,41]. Therefore, in this paper, GA-A and GA-D were selected to be the typical positive compounds for guiding the anti-cancer bioactive prediction of six isolated GTs. Eight compounds were first minimized in CHARMm force field with MMFF 94 partial charge. Diverse conformations of 8 GTs were constructed by the BEST method within 255 conformations, and the relative energy threshold was set to 20.0 kcal/mol. Eight GTs were in silico profiled by structure-based pharmacophore (SBP) models database called pharmaDB in DS [42][43][44]. In order to evaluate the predict activity of GTs, two indexes were chosen to judge the predicted results. The initial judgment criterion was the overlap degree of molecules and pharmacophore, which was represented by the fit value [45]. Fit value greater than 0.5 indicated better-quality conformational coverage between compounds and SBP, which suggested potential biological activity of compounds for corresponding target. The second judgment criterion was the hit number of SBP models, which could to some extent reveal the biological activity of compounds and importance of the target. As SBP models were constructed by the crystal structure in the RCSB protein data bank (PDB), one target might have more than one crystal structure, and meanwhile one crystal structure could produce several SBP models. Therefore, a single target could have multiple SBP models, which represented various binding modes between ligands and receptors [46]. According to the two indexes, the results of predicted targets and interaction were obtained with the fit value more than 0.5 and the rank of hit number of SBP models. Based on the result of reverse target identification, ingredient-target network of GTs was structured in Cytoscape 3.2.0 (Cytoscape Consortium, San Diego, CA, USA) [47]. The high-frequency targets and pharmacological action of GTs were predicted and validated by the analysis of network nodes and edges.

Network Construction and Analysis
The PPI information of high-frequency targets was derived from the online database of String 9.1 [48] which could provide experimental and predicted PPI information. Confidence scores could be provided from String, which represented the stable degree of interaction. In this paper, the PPIs of high-frequency targets, with a confidence score higher than 0.9, were collected to construct PIN by Cytoscape 3.2.0. Then, union calculation was implemented and duplicated edges of PPIs were removed by Advanced Network Merge [49]. Finally, the largest connected sub-graph was obtained as the PIN of GTs.
A functional module is the structural unit of PIN, which can represent cellular functional organization and extract meaningful information from the complex PIN. In this paper, functional modules with more than four sides were identified by the MCODE methods [50]. Based on the identified modules of GTs, GO enrichment analysis was used to predict possible biological processes of the identified modules using the BinGO method [51]. Wherein, p-value was utilized to represent the probability that a group of genes was from the same gene ontology, and less p-value indicated more probability.

Conclusions
G. lucidum is one of the most commonly used TCM for thousands of years. The discovery of therapeutic basis and action mechanism of G. lucidum was essential to clinical use of drugs from G. lucidum. In this paper, an innovated research mode of extraction, isolation, pharmacological prediction, and PIN analysis was presented to discover the novel compounds and predict the anti-cancer activity of G. lucidum. Twenty-four compounds were isolated and identified from the 70% ethanol extract of the fruiting bodies of G. lucidum, including nineteen triterpenes, three steroids, one cerebroside, and one thymidine. Six GTs (Compounds 1-6) were first reported from the species of G. lucidum. One cerebroside and one steroid were first isolated from the genus Ganodema. Then, pharmacophoric profiling of these six GTs was implemented to analyze their potential therapeutic targets. Nineteen high-frequency targets were obtained and mainly related to cancer and metabolic syndrome, especially CDK2 and PPAR γ. PIN and modules analysis of the 19 high frequency targets indicated that the cell division cycle regulating and protein acetylation might be the main anti-cancer mechanism of these six GTs. In this paper, molecular simulation and PIN analysis were directly utilized to research the potential activity of isolated compounds from the source of extraction of TCM. Actually, the results also need to be validated by wet experiments. However, compared to biochemistry and animals experiment, it might be a rapid, economical, and efficient method for discovering potential pharmacological activity of new isolated compounds.