Antitumor effect of guava leaves on lung cancer: A network pharmacology study

2.1 Materials

The databases used in this study included TCMSP, PubChem, PharmMapper, STRING, UniProt, GeneCards, Venny 2.1.0, KOBAS 3.0, SystemsDock, DisGeNET, and CbioPortal. The software used included ChemDraw Office 2010 (PerkinElmer), Cytoscape 3.7.1 (Cytoscape), and FunRich 3.1. 3 (http://www.funrich.org/download).

2.2 Methods

3.1 Collection of active ingredients and their chemical structures

The 2D or 3D structures of all compounds were searched in the PubChem database, and the files in the sdf format were saved. Compounds not included in the database were identified by chemical structure, which were then drawn using the ChemDraw Ultra 12.0 software, and the sdf file was saved (Fig. 1). The structures of the compounds were searched by PubChem, and their molecular formulas and accession numbers in the database (PubChem CID) were retrieved. The relevant information of the active constituents of Guava leaves is listed in Table 1. In total, 66 active components of guava leaves, including 17 triterpenoids such as ursane and oleanane pentacyclic triterpenes, 19 sesquiterpenoids, and 30 flavonoids.

Close

Fig. 1 The structural formula of six compounds, which were found but not included in the PubChem database. T2: 3β-O-trans-p-coumaroylmaslinicacid; T15: psidiumoic acid; F12: guavinoside C; F24: quercetin-3-O-(6″-feruloyl); S1: Diguajadial; S2: Guadial A.

Export to PPT

Table 1 Molecular information of guava leaf constituents. The first column is the compound name, the second column is the molecular type of the compound, the third column is the number of the constituent, with triterpenoids numbered from T1 to T17, flavonoids from F1 to F30, and sesquiterpenoids from S1 to S19. The molecular formula, the molecular structure, the PubChen ID for compound retrieved from the database and the Pharmmapper JOB ID (the search number of targets in the Pharmmapper database) were listed in the columns four to seven.

Compound Name	Molecular type	Number	Molecular Formula	Molecular structure	PubChem CID	Pharmmapper JOB ID
2α-hydroxyoleanolic acid	Triterpenoid	T1	C30H48O4		15,560,128	190,417,034,302
3β-O-trans-p-coumaroylmaslinicacid	Triterpenoid	T2	C39H54O5		/	190,417,034,505
asiatic acid	Triterpenoid	T3	C30H48O5		119,034	190,417,034,630
corosolic acid	Triterpenoid	T4	C30H48O4		6,918,774	190,417,034,747
goreishic acid I	Triterpenoid	T5	C30H46O4		3,081,756	190,318,032,131
guajanoic acid	Triterpenoid	T6	C32H50O6		101,211,343	190,417,035,053
guavacoumaric acid	Triterpenoid	T7	C39H54O7		101,211,344	190,417,035,342
guavanoic acid	Triterpenoid	T8	C32H50O6		101,211,343	190,417,035,431
ilelatifol D	Triterpenoid	T9	C30H46O4		102,572,108	190,318,031,159
isoneriucoumaric acid	Triterpenoid	T10	C39H54O6		10,100,394	190,417,035,723
jacoumaric acid	Triterpenoid	T11	C39H54O6		11,700,083	190,417,035,813
obtusinin	Triterpenoid	T12	C15H18O6		3,604,942	190,318,030,756
obtusol (3β, 27-dihydroxy-urs-12-ene)	Triterpenoid	T13	C30H50O2		15,895,316	190,417,040,026
oleanolic acid	Triterpenoid	T14	C30H48O3		10,494	190,417,040,126
psidiumoic acid	Triterpenoid	T15	C32H50O5		/	190,417,040,228
ursolic acid	Triterpenoid	T16	C30H48O3		64,945	190,417,040,330
uvoal	Triterpenoid	T17	C30H50O2		92,802	190,417,040,432
Apigenin	Flavonoid	F1	C15H10O5		5,280,443	190,414,093,403
Avicularin	Flavonoid	F2	C20H18O11		5,490,064	190,414,093,603
Biochanin	Flavonoid	F3	C16H12O5		5,280,373	190,403,023,437
Daidzein	Flavonoid	F4	C15H10O4		5,281,708	190,414,094,000
demthoxymatteucinol	Flavonoid	F5	C17H16O4		180,550	190,414,094,140
formononetin	Flavonoid	F6	C16H12O4		5,280,378	190,414,094,323
Genistein	Flavonoid	F7	C15H10O5		5,280,961	190,414,094,458
Genistin	Flavonoid	F8	C21H20O10		5,281,377	190,414,094,600
Glycitin/daidzin	Flavonoid	F9	C22H22O10		187,808	190,414,094,811
guaijaverin	Flavonoid	F10	C20H18O11		5,481,224	190,414,095,056
guavaric A	Flavonoid	F11	C32H50O6		101,211,343	190,414,095,209
guavinoside C	Flavonoid	F12	C27H22O15		/	190,414,095,324
hyperin	Flavonoid	F13	C21H20O12		133,568,467	190,417,041,614
isoquercetin(Isoquercitrin)	Flavonoid	F14	C21H20O12		5,280,804	190,414,095,613
kaempferol	Flavonoid	F15	C15H10O6		5,280,863	190,414,095,757
kaempferol-3-glucoside	Flavonoid	F16	C21H20O11		5,282,102	190,414,095,928
Leucocyanidin	Flavonoid	F17	C15H14O7		71,629	190,414,100,124
morin-3-O-α-L-lyxopyranoside	Flavonoid	F18	C20H18O11		10,455,578	190,414,100,422
myricetin	Flavonoid	F19	C15H10O8		5,281,672	190,414,100,546
Ononin	Flavonoid	F20	C22H22O9		442,813	190,414,100,718
prunetin	Flavonoid	F21	C16H12O5		5,281,804	190,414,100,847
quercetin	Flavonoid	F22	C15H10O7		5,280,343	190,414,101,024
quercetin3-O-β-D-xylopyranoside	Flavonoid	F23	C20H18O11		5,320,861	190,414,101,158
quercetin-3-O-(6″-feruloyl) -β-D-galactopyranoside	Flavonoid	F24	C31H28O15		/	190,414,101,321
quercetin-3-O-gentiobioside	Flavonoid	F25	C27H30O17		13,915,963	190,414,101,504
quercetin-3-O-β-D-glucuronide	Flavonoid	F26	C21H18O13		13,258,914	190,414,101,626
quercitrin	Flavonoid	F27	C21H20O11		5,280,459	190,414,101,724
reynoutrin	Flavonoid	F28	C20H18O11		5,320,863	190,414,101,824
rutin	Flavonoid	F29	C27H30O16		5,280,805	190,414,101,941
xanthone	Flavonoid	F30	C13H8O2		7020	190,414,102,035
Diguajadial	Sesquiterpenoids	S1	C60H66O9		/	190,321,031,034
Guadial A	Sesquiterpenoids	S2	C25H26O5		/	190,321,032,624
Guadial B	Sesquiterpenoids	S3	C25H26O5		122,377,745	190,319,051,501
Guadial C	Sesquiterpenoids	S4	C25H26O5		122,377,746	190,319,051,610
Guajadial	Sesquiterpenoids	S5	C30H34O5		101,447,677	190,318,025,632
Guajadial B	Sesquiterpenoids	S6	C30H34O5		137,346,032	190,319,050,244
Guajadial C	Sesquiterpenoids	S7	C30H34O5		134,714,902	190,319,050,403
Guajadial D	Sesquiterpenoids	S8	C30H34O5		134,714,901	190,319,050,732
Guajadial E	Sesquiterpenoids	S9	C30H34O5		134,714,904	190,319,050,851
Guajadial F	Sesquiterpenoids	S10	C30H34O5		134,714,903	190,319,050,938
Guapsidial A	Sesquiterpenoids	S11	C29H32O5		122,377,744	190,319,051,859
Guajadial	Sesquiterpenoids	S12	C30H34O5		46,197,930	190,320,031,551
Psidial A	Sesquiterpenoids	S13	C30H36O6		45,104,960	190,318,025,129
Psidial B	Sesquiterpenoids	S14	C30H36O6		45,104,961	190,318,025,249
Psidial C	Sesquiterpenoids	S15	C30H34O5		49,844,493	190,318,025,408
Psiguadial A	Sesquiterpenoids	S16	C30H34O5		49,844,493	190,318,025,837
Psiguadial B	Sesquiterpenoids	S17	C30H34O5		102,052,649	190,318,025,954
Psiguadial C	Sesquiterpenoids	S18	C30H34O6		122,224,646	190,319,051,306
Psiguadial D	Sesquiterpenoids	S19	C30H34O5		77,984,632	190,321,030,934

3.2 Screening of potential targets and acquisition of disease genes

A total of 115 potential targets for four triterpenoids and 303 potential targets for the 19 flavonoids were obtained from the TCMSP database (Table 2). Molecular information on guava leaves, including Lipinski's “five-law” parameters, namely relative molecular mass (MV), octanol–water partition coefficient (AlogP), possible hydrogen bond donor number (Hdon), possible hydrogen, number of bond receptors (Hacc), number of bonds allowed to rotate freely (RBN), oral bioavailability (OB), and drug-like degree (DL) were obtained. The targets of the active ingredients of guava leaves were obtained from the PharmMapper database, and targets with z'-score ≥ 1 were screened as potential targets of guava leaf active ingredients. From the definition in the database website, “Fit Score” and “z'-score” are scores generated by the metric's Fit score, which is a pre-calculated library score matrix, and a large positive z'-score represents the target-to-query combination. A total of 198 target genes for triterpenoids, 215 target genes for sesquiterpenoids, and 302 target genes for flavonoids were retrieved from the PharmMapper database. A total of 246 potential targets for triterpenoids, 215 potential targets for sesquiterpenoids, and 535 potential targets for flavonoids were obtained from TCMSP and Pharmmapper. The names of all potential targets were imported into the STRING database and Uniprot database, and converted into Gene Symbols.

In the Genecards database, with “Lung Cancer” as the key word, a total of 20,649 results were associated with lung cancer, and the top 500 targets with “Score” values ranged in descending order.

The genes associated with the three types of compounds were compared with the lung cancer genes in the Venny 2.1.0 database, and a chart of gene interaction was obtained (Fig. 2A). A total of 153 genes were obtained from lung cancer, and the target genes of guava leaf constituents against lung cancer were obtained. The total number of genes was 16.4%. Among them, there were 4 (0.4%) cross-reactive genes in three sputum and lung cancer; 68 (7.3%) cross-genes between flavonoids and lung cancer; 3 (0.3%) genes between triterpenoids, sesquiterpenes, and lung cancer; 9 (1%) cross-linking genes for scorpion, flavonoids, and lung cancer; 27 (2.9%) cross-linking genes for triterpenoids, flavonoids, and lung cancer; and 42 (4.5%) cross-linking genes for triterpenoids, sesquiterpenes, flavonoids, and lung cancer.

Close

Fig. 2 Drug-disease interaction network analysis (A) Screening of guava leaf-lung cancer common gene. (B) Topological analysis of drug-disease interactive gene network.

Export to PPT

3.3 Topological analysis of target protein networks and gene assignment

The 153 targets related to lung cancer and identified as targets for guava leaf constituents were introduced into the STRING database to obtain a protein interaction network. Topological analysis of the target protein network was performed in Cytoscape. The topological analysis results are shown in Fig. 2B. The color depth and size of the nodes represent the strength of interaction between the gene and other genes. The interaction network had a total of 153 number of nodes. The number of edges associated with the target protein was 443, and the average node degree was 5.79, with p value (PPI enrichment p-value) < 1.0e−16. The target genes with a degree greater than or equal to 10 in the guava leaf constituent and lung cancer cross-linking network topology analysis were sequentially introduced into the DisGeNET database to obtain the protein type corresponding to the target. The results showed that the types of these proteins were nucleic acid binding, transcription factor, calcium-binding protein, kinase, transferase, signaling molecule, enzyme modulator, receptor, hydrolase, protease, and transfer/carrier protein (Table 3). The network topological analysis of genes associated with lung cancer in guava leaves genes with a score of 10 or higher and their genes are also shown in Table 3.

3.4 Cytoscape network interaction analysis

There is a genetic interaction between drugs and diseases. This is a characteristic of multi-component and multi-target network pharmacology analysis. Through Cytoscape network visualization analysis, a “Guava leaf-compound class-active ingredient-gene” interaction network was obtained. The compound and target information on nodes ≥20 are listed in Table 4, where the Average Shortest Path Length is the average shortest path, Closeness Centrality is the center proximity, and Radiality is the radial degree.

As shown in Fig. 3 and Table 4, the three compounds in guava leaves that interacted strongly with lung cancer were quercetin, genistein, and apigenin. Moreover, the top ten targets were vitamin D3 receptor (VDR), cyclin-dependent kinase 2 (CDK2), dual specificity mitogen-activated protein kinase 1 (MAP2K1), cyclin-dependent kinase 6 (CDK6), hepatocyte growth factor receptor (MET), tyrosine-protein kinase ABL1 (ABL1), peroxisome proliferator-activated receptor gamma (PPARG), interleukin-2 (IL2), epidermal growth factor receptor (EGFR), and progesterone receptor (PGR).

Close

Fig. 3 Interactive network of ‘guava leaf - compound class – active compound – gene’.

Export to PPT

The network diagram of “Guava leaf-compound class-active constituents-gene” generated by Cytoscape is shown in Fig. 3, in which the green rectangle shape represents the traditional Chinese medicine guava leaf, and the three pale blue diamond shapes represent sesquiterpene, flavonoids, and triterpenoids; pink ovals represent compounds and yellow hexagons represent genes for the treatment of lung cancer with guava leaves. In the interaction network, the target information corresponding to the compound having a degree value of >20 and the corresponding gene are listed in Table 5.

3.5 GO and KEGG enrichment analysis

Through ClueGO analysis, a total of 1430 GO biological processes were obtained, with 7536 interactions between the biological processes; 35 molecular functions, with 28 interactions between the molecular functions; and 203 GO cell components, with 470 interactions between the cell components. KEGG analysis was performed by KOBAS 3.0, and 217 pathway enrichment results were obtained. GO analysis results of the interacting genes between guava leaves and lung cancer are shown in Table 6.

The top 10 enriched biological processes are shown in Fig. 4A. The enrichment results included apoptosis, programmed cell death, regulation of programmed cell death, regulation of cell death, regulation of apoptotic process, regulation of cell population proliferation, response to oxygen-containing compound, negative regulation of programmed cell death, and negative regulation of cell death. The top 10 enriched cellular components are shown in Fig. 4B. The enrichment results included platelet alpha granule lumen, platelet alpha granules, vesicle lumen, cytoplasmic vesicle lumen, secretory granule lumen, cyclin-dependent protein kinase holoenzyme complex, membrane raft, membrane microdomain, membrane region, and protein kinase complex. The top 10 enriched molecular function terms are shown in Fig. 4C. The enrichment results included protein kinase activity, positive regulation of catalytic activity, phosphotransferase activity, alcohol group as acceptor, phosphokinase activity, kinase activity, protein serine/threonine kinase activity, regulation of transferase activity, regulation of kinase activity, regulation of protein serine/threonine kinase activity, regulation of protein kinase activity, and positive regulation of kinase activity.

Close

Fig. 4 GO and KEGG analysis of ‘guava leaf – lung cancer’ interactive gene with P-values from small to large. (A) GO biological process analysis of ‘guava leaf – lung cancer’ interactive gene with P-values from small to large. (B) GO cell components analysis of ‘guava leaf – lung cancer’ interactive gene with P-values from small to large. (C) GO molecular function analysis of ‘guava leaf – lung cancer’ interactive gene with P-values from small to large. (D) KEGG analysis of ‘guava leaf – lung cancer’ cross-gene.

Export to PPT

KEGG enrichment analysis results of common genes between guava leaf constituent target and lung cancer related genes are shown in Fig. 4D, Table 7 and Table 8. The top 10 common enriched pathways are listed below: cancer pathway, hepatitis B, proteoglycans in cancer, prostate cancer, PI3K-Akt signaling pathway, AGE-RAGE signaling pathway in diabetic complications, endocrine resistance, pancreatic cancer, EGFR tyrosine kinase inhibitor resistance, and FoxO signaling pathway. The non-small cell lung cancer and small cell lung cancer pathways were also recorded.

One of these pathways is the PI3K-Akt signaling pathway, which has been shown to play an important regulatory role in tumor therapy. The PI3K-Akt signaling pathway is an important pathway for cell survival, metabolism, angiogenesis, apoptosis, proliferation and differentiation. The substrate is used to control key cellular processes, and many targets are involved in this pathway and AKT also plays an important role in the regulation of this pathway (Ebrahimi et al., 2017). In order to investigate the effectiveness of guava leaf constituents on lung cancer via the PI3K-Akt signaling pathway, we analyzed the expression (Fig. 5A) and the pathway activities (Fig. 5) of these genes in lung squamous cell carcinoma (LUSC) and lung adenocarcinoma (LUAD) samples from TCGA database in the GSCAlite online platform. The results indicated that the PI3K-Akt signaling pathway genes that could be potentially affected by the guava leaf constituents showed differential expression with some of them being downregulated (Fig. 5A, blue color dot) or upregulated (Fig. 5A, red color dot) in both LUSC and LUAD. In addition, we found that some of the target genes of the guava leaf constituents did not show differential expression in lung cancer (Fig. 5A). Furthermore, in pathway activity analysis, we found that these target genes were involved in the inhibition or activation of important pathways such as PI3K-Akt, apoptosis, cell cycle, EMT, DNA damage response, hormone AR, TSC/mTOR, RAS/MAPK, RTK and hormone ER pathways.

Close

Fig. 5 The expression and pathways activity analysis of guava leaf constituent target genes in the PI3K-Akt signaling pathway. (A) The mRNA expression profile of genes. The dots represent the fold change; the blue color indicates downregulation while the red color indicates upregulation. The size of the dot is proportional to the expression foldchange (FC). (B) Pathway activity analysis.

Export to PPT

3.6 Molecular docking

The previous drug-disease interaction gene network topological analysis showed that the 10 genes with the strongest interaction were STAT3, TP53, AKT1, JUN, IL6, VEGFA, SRC, TNF, HRAS, and CDKN1A, which are effective in the “Guava leaf-constituent category” network. The Cytoscape interaction network of component-genes showed that the five compounds with strong interaction between guava leaf constituents and lung cancer genes were quercetin, genistein, apigenin, ursolic acid, and daidzein. The five compounds and 10 target proteins were molecularly docked by online molecular docking on the systemsDock website. The score of the database system docking was between 0 and 10. The larger the docking score, the better the docking effect, and the greater the binding activity between the docking molecule and the target. A docking score greater than 4.25 indicates that a ligand has a certain binding activity with a receptor; a score greater than 5.0 indicates a better binding activity; a score greater than 7.0 indicates a strong binding activity (Hsin et al., 2016). The PDB ID of the 10 genes was molecularly docked with the five compounds. The docking scores of all compounds and proteins were greater than 4.25, indicating the presence of binding activity. The molecular docking results are shown in Table 9.

A heat map of the molecular docking is shown in Fig. 6A. The deeper the color, the better the docking effect. The effect of molecular docking is also shown by the heat map. Table 9 and Fig. 6A showed that the compound with the highest binding activity to the protein receptor STAT3 was genistein; that with the highest binding activity to the protein receptor TP53 was kaempferol; that with the highest binding activity to the protein receptor AKT1 was kaempferol; that with the highest binding activity to the protein receptor AKT1 was quercetin; that with the highest binding activity to the protein receptor JUN was daidzein; that with the highest binding activity to the protein receptor VEGFA was genistein; that with the highest binding activity to the protein receptor SRC was ursolic acid; that with the highest binding activity to the protein receptor TNF was ursolic acid; that with the highest binding activity to the protein receptor HRAS was ursolic acid; and that with the highest binding activity to the protein receptor CDKN1A was ursolic Acid.

Close

Fig. 6 Molecular docking. (A) A heat map showing the docking of receptor protein and ligand. (B) A diagram showing the genetic variation in lung squamous cell carcinoma.

Export to PPT

The 3D and 2D structures of their docking are shown in Table 10. In the 2D docking diagram of the third column, there is a specific case where the ligand binds to the receptor, and the surrounding red bar represents a bond-free protein residue within the ligand; the structure linked by the bat is a ligand compound and is subjected to the body protein. Purple represents a ligand compound; yellow represents a receptor protein. Red font represents the binding mode between the ligand and the receptor; the green dotted line represents a hydrogen bond. In these docking results, the docking activity of the receptor protein AKT1 and the ligand ursolic acid was 8.331, which was bound by Met 227 (A), Glu234 (A), and Thr291 (A); the receptor protein CDKN1A and the ligand compound ursolic acid, with docking activity of 8.148, and bound by a hydrogen bond with Asn247 (B) and Met246 (B), with bond lengths of 3.12 and 2.76, respectively, as well as with Cys242 (B), Met243 (A), Asn239 (B), Leu137 (B), Cys238 (B), Arg175 (B), and Arg174 (B), which directly bound.

Table 10 Molecular docking diagrams of the 3D and 2D structure.

Docking result of 3ALQ and ursolic acid
Docking result of 3TS8 and ursolic acid
Docking result of 5GMT and kaempferol
Docking result of 5TO1 and Daidzein
Docking result of 6BFT and Genistein
Docking result of 6D5W and ursolic acid
Docking result of 3CQW and kaempferol
Docking result of 1ALU and quercetin
Docking result of 2BDJ and ursolic acid
Docking result of 1BG1 and Daidzein

3.7 cBioPortal analysis

cBioPortal is a database used for tumor research. The expression of the genes STAT3, TP53, AKT1, IL6, and JUN in 501 patients with lung squamous cell carcinoma was studied. The expression of these genes in the selected lung cancer samples is shown in Fig. 6B. The first part is the expression of genes, including inframe mutation, missense mutation, truncating mutation, amplification, deep deletion, mRNA level upregulation (mRNA High), mRNA level downregulation (mRNA Low), and no alterations. The second part is a heat map of the gene mRNA levels.

According to the data in Fig. 6B, TP53 and AKT1 were highly expressed in these 501 cancer patients. TP53 was mainly expressed in patients with mutations (129 cases, 25.75%), increased copy number (1 case, 0.2%), mRNA level upregulation (6 cases, 1.2%), high protein level (5 cases, 1%), low protein level (8 cases, 1.8%), and multiple alterations (17 cases, 3.39%). AKT1 was expressed mainly in patients with mutations (1 case, 0.2%), increased copy number (5 cases, 1%), mRNA level upregulation (34 cases, 6.79%), mRNA level downregulation (34 cases, 6.79%), high protein level (2 cases, 0.4%), low protein level (15 cases, 2.99%), and multiple alterations (16 cases, 3.19%).