A Network Pharmacology Analysis of Cytotoxic Triterpenes Isolated from Euphorbia abyssinica Latex Supported by Drug-likeness and ADMET Studies.

Euphorbia plants have been identified as potential sources of antitumor lead compounds. The current study aimed to isolate and identify the main active constituents of Euphorbia abyssinica latex followed by a cytotoxic evaluation. A network pharmacology approach was employed to predict the underlying mechanism. Finally, drug-likeness and ADMET studies were conducted for active compounds. The phytochemical investigation of the latex of E. abyssinica resulted in the isolation of two triterpenes, 3-acetyloxy-(3α)-urs-12-en-28-oic methyl ester (1) and lup-20(29)-en-3α,23-diol (2). The dichloromethane extract displayed potent cytotoxic activity against the MCF7 cell line with an IC50 value of 4.27 ± 0.12 μg/mL but weak activity against HepG2 and HeLa cell lines (IC50 = 20.47 ± 1.17 and 26.73 ± 2.99 μg/mL, respectively) compared to doxorubicin. Compound 1 showed an encouraging cytotoxic effect against MCF7 with IC50 = 4.20 ± 0.20 μg/mL, followed by compound 2 (IC50 = 5.8 ± 0.35 μg/mL). The network analysis revealed that the two isolated compounds are linked to 68 targets of human nature, among which 51 genes are linked to breast carcinomas and 5 targets (AR, CYP19A1, EGFR, PGR, and PTGS2) might be the top therapeutic targets of isolated compounds on breast cancer. Furthermore, the gene-enrichment analysis revealed that E. abyssinica could play a role in the treatment of breast cancer by striking 51 potential targets via mainly three signaling pathways: P13K-AKT, Wnt, and VEGF. Therefore, isolated triterpenes could be considered effective antitumor agents for breast cancer by elucidating their candidate target to alleviate breast cancer and related signaling pathways of the targets.

Introduction

Cancer is a serious public health issue and one of the leading causes of death and morbidity worldwide. It is distinguished by uncontrolled cell proliferation and multiplication.[1] Every year, an increasing number of people are diagnosed with cancer. GLOBOCAN (2020) statistics show that there are projected 19.3 million new cancer cases and 10.0 million cancer-related deaths. In 2040, the worldwide cancer burden is predicted to reach 28.4 million cases, a 47% increase from 2020.[2] To date, there have been some commonly used strategies including surgery, radiation, immunotherapy, hormones, and chemotherapy in cancer treatment. However, these treatments have been associated with serious side effects and multidrug resistance.[3] In the cancer battle, more effective new anticancer medications with fewer side effects are required. The development of lead therapeutic compounds relies on natural resources, notably medicinal plants. Many natural compounds of plant origin, such as vinblastine, vincristine, and paclitaxel, are now in clinical usage as anticancer treatments. They are isolated from plants or formed via semisynthesis of the isolates.[4]

Desert candle (Euphorbia abyssinica) is a succulent tree belonging to the family Euphorbiaceae. It is abundant in Ethiopia, Sudan, Nigeria, and Somalia.[5] Traditionally, it was used to treat intestinal worms, diarrhea, head fungal infections, external injuries, venereal diseases, and neck cancer.[6−8] Different E. abyssinica stem bark extracts are reported to have antifungal attributes.[9] Similarly, antibacterial and antifungal properties of plant latex were reported.[5] The root extract showed antimalarial potential in mice against Plasmodium berghei infection.[10] In the treatment of skin diseases, an ointment made from plant latex showed antibacterial and antiparasitic properties.[10] Previous chemical investigation in E. abyssinica latex showed the presence of 8(R)-hydroxy-dec-3(E)-en-oic acid, lupeol, oleanolic acid, β-sitosterol, and β-sitosterol-3-O-glucoside.[5] Metabolite profiling of E. abyssinica fruit and aerial part methanolic extracts using UPLC-MS revealed that they were a rich source of diverse compounds. Thirty-nine compounds including flavonoids and phenolic compounds as main metabolites were identified in both extracts.[11]

Euphorbia plant latexes were considered toxic substances because they could cause irritation of mucous membranes in humans, especially the nose and mouth.[12] However, several previous studies revealed the cytotoxic activity of Euphorbia plants.[13−16] Despite the plethora of chemical and biological investigations on the genus Euphorbia, there is little work on E. abyssinica, and no comprehensive study of its cytotoxic activity exists. Therefore, the aim of this study was to assess the cytotoxic effect of the latex of E. abyssinica, conduct a phytochemical investigation, and isolate the active component(s). Additionally, network pharmacology analysis was established to clarify the potential mechanism of isolated compounds against breast cancer. To assess these compounds’ drug-likeness, selected computed molecular parameters pertaining to absorption, distribution, metabolism, excretion, and toxicity (ADMET) of phytochemicals were studied and compared to those of well-known drugs. Furthermore, different scores were utilized to assess the possible interaction of the examined compounds with one of the most important proteins involved with breast cancer, the progesterone receptor.

Results and Discussion

Structural Elucidation of Isolated Compounds

The fractionation of the DCM extract of E. abyssinica latex resulted in the separation of two compounds. The structures of these compounds were fully elucidated based on spectroscopic data (Figures S1–S6) and in reference to published data as 3-acetyloxy-(3α)-urs-12-en-28-oic methyl ester (1)[17,18] and lup-20(29)-en-3α,23-diol (2)[17,19,20] (Figure ). Both compounds are reported for the first time from E. abyssinica.

Figure 1

Chemical structures of isolated compounds from E. abyssinica latex. (1) 3-acetyloxy-(3α)-urs-12-en-28-oic methyl ester; (2) lup-20(29)-en-3α,23-diol.

Cytotoxic Activity

Natural products have long been considered crucial in cancer therapy. Natural products are not only effective anticancer medications, but they are also useful lead compounds for the production of new targeted chemotherapeutic regimens.[21] The cytotoxic activity of the DCM extract of E. abyssinica latex was tested against three cancer cell lines, namely, human breast cancer (MCF7), human hepatocellular carcinoma (HepG2), and human cervical cancer (HeLa) cell lines, and the noncancerous human normal melanocyte (HFB4) cell line using SRB assay. As depicted in Table , the DCM extract of the latex exhibited remarkable cytotoxic activity against the MCF7 cell line with IC50 = 4.27 ± 0.12 μg/mL and weak activity against HepG2 and HeLa cell lines (IC50 = 20.47 ± 1.17 and 26.73 ± 2.99 μg/mL, respectively) compared to standard doxorubicin (IC50 = 3.27 ± 0.12, 4.80 ± 0.20, and 4.13 ± 0.42 μg/mL against MCF7, HepG2, and HeLa cell lines, respectively). The lack of existing therapy specificity and toxicity to noncancerous cells contributes to the search for new cancer medications from plants to mitigate these disease groups. Although many molecules have in vitro anticancer activity, few can exhibit anticancer activity in clinical studies without killing normal cells. When compared to doxorubicin (IC50 = 3.87 ± 0.23 μg/mL), the DCM extract of the latex showed probable selectivity against cancer cells rather than normal cells (IC50 = 23.67 ± 3.3 μg/mL on HFB4).

Table 1

In Vitro Cytotoxic Activities of the Dichloromethane Extract and Isolated Compounds from E. abyssinica Latexa

	IC50 (μg/mL)
extract/compound	MCF7	HepG2	HeLa	HBF4
dichloromethane extract	4.27 ± 0.12	20.47 ± 1.17	26.73 ± 2.99	3.87 ± 0.23
compound 1	4.20 ± 0.20	-	-	-
compound 2	5.8 ± 0.35	-	-	-
doxorubicin	3.27 ± 0.12	4.80 ± 0.20	4.13 ± 0.42	3.87 ± 0.23

All data are presented as mean ± S.D.

The promising results of cytotoxic activity of the DCM extract on breast cancer cell lines encourage us to evaluate the anticancer activities of isolated compounds against the MCF7 cell line. As shown in Table , 3-acetyloxy-(3α)-urs-12-en-28-oic methyl ester exerted a potent cytotoxic effect against the MCF7 cell line with IC50 = 4.20 ± 0.20 μg/mL, followed by lup-20(29)-en-3α,23-diol (IC50 = 5.80 ± 0.35 μg/mL).

Triterpenes (C30) are a diverse and important class of natural compounds found in a wide range of plants. Pentacyclic triterpenoids are well known for their various biological activities, including cytotoxic potential.[22] Our results are consistent with a previous study where several ursolic acid derivatives and lupane triterpenoids showed significant cytotoxic and growth inhibitory effects against different cancer cell lines.[20,23,24]

Computational Study

Compound–Common Target Network Analysis

A total of 68 human target genes (Homo sapiens) were obtained (Table S1). To obtain prospective targets of the two isolated compounds acting on breast cancer, we integrated the E. abyssinica-isolated compounds’ predicted targets in a compound–common target network and excluded those replicate genes (Figure ). Finally, a total of 51 genes were identified related to breast cancer. The two isolated active compounds of E. abyssinica latex are as follows: 3-acetyloxy-(3α)-urs-12-en-28-oic methyl ester (1) interacts with 46 targets, while lup-20(29)-en-3α,23-diol (2) interacts with 12 targets, only 5 targets (AR, CYP19A1, EGFR, PGR, and PTGS2) of which might be the top therapeutic targets of isolated compounds on breast cancer (Figure ).

Figure 2

Compound–common target network of isolated compounds from E. abyssinica latex and breast cancer. Yellow oval shapes represent the isolated compounds from E. abyssinica latex; green triangles represent common targets of isolated compounds and breast cancer. Edges represent interactions between isolated compounds and common targets.

Protein–Protein Interaction (PPI) Data

The interactions among resulting target proteins were performed using the online complex protein database STRING, and the results were visualized and organized by the Cytoscape software. The PPI network was adjusted at low confidence (0.150), and it was built with 81 nodes and 379 edges with an average node degree of 9.36 and an average clustering coefficient of 0.583. The PPI enrichment was considered significant at P < 1.0 × 10–16. The PPI network was presented with three clusters (Figure ).

Figure 3

The PPI network represents common targets of isolated compounds of E. abyssinica in three clusters. Line edges represent interaction between targets, and dotted lines represent interactions between clusters. The network was maintained using the STRING database and analyzed by the Cytoscape software.

Common Target–Breast Cancer Network Analysis

In this network, targeted genes were interacting with different cancers and melanomas (Table S2). To further find the target of E. abyssinica latex-isolated compounds on breast cancers, we built a compound–common target network filtered to different types of breast cancers and melanomas affecting both males and females. The network connects 51 hub targets, and all breast cancer types were introduced in a network (Figure ). Five genes (AR, CYP19A1, EGFR, PGR, and PTGS2) have interactions with most breast cancer types (>10 interactions for each gene). These hit genes represent the higher affinity genes toward different types of breast cancers.

Figure 4

Common target–breast cancer network analysis: blue rectangle shapes represent types of breast cancers (Diseases IDs), and red circles represent common targets (UniProt IDs).

Common Target–Invasive Breast Cancer Network Analysis

MCF7 is a breast cancer cell line representative of breast invasive carcinoma, so a network analysis was established to configure the relationship between targeted genes and invasive types of breast cancers; a total of 11 hit genes were correlated to four types of breast invasive carcinoma types (Figure ) and the top hit genes (AR, CYP19A1, EGFR, PGR, and PTGS2) appeared to be the most prominent genes to the invasive types of breast carcinoma (Table S3).

Figure 5

Network correlation between target genes and invasive types of breast cancers: green triangles represent the invasive types of breast cancers, and pink circles represent target genes.

Compound–Common Target–Breast Cancer Pharmacology Network

The combination and merging of the plant–compound, compound–target, and target–breast cancer networks formed the complete pharmacology network that correlates the isolated compounds from E. abyssinica latex, 3-acetyloxy-(3α)-urs-12-en-28-oic methyl ester (1) and lup-20(29)-en-3α,23-diol (2), to the target proteins and to breast cancers as described in Figure . Both triterpenoids isolated from E. abyssinica had good pharmacological effects that could play a critical role in the treatment of breast cancer.

Figure 6

Total merged networking of E. abyssinica-isolated compound–targeted gene–targeted breast cancer disease in code forms: the pink oval shape represents E. abyssinica latex, yellow oval shapes represent isolated compound names, blue oval shapes represent target genes (UniProt gene IDs), and purple arrows represent interacted breast cancer types (Disease IDs).

Target Gene–Pathway Network

To establish target gene–pathway networking, we formed the network by all the hit genes targeted by the two isolated compounds interacting with different specified types of breast cancers using the KEGG database and ShinyGO database. The pathway analysis constructed by the KEGG diagram focuses on three signaling pathways: P13K–AKT, Wnt, and VEGF signaling pathways. As described by the KEGG-labeled diagram (Figure ), the main responsible genes for the anticancer pathways are AR, F2, and EGFR in a direct way.

Figure 7

Diagrammatic description of pathways and genes involved in cancer treatment: red rectangles represent the hit genes involved in the cancer treatment. The diagram was formed by the KEGG (Kyoto Encyclopedia of Genes and Genomes) database.

By GO terms, regulation of cell population proliferation (GO: 0042127), positive regulation of intracellular signal transduction (GO: 19025330), and positive regulation of signal transduction (GO: 0009967) are among the top pathways in the biological process category (Figure A) correlated with cancer. Top pathways in the cellular component category (Figure B) are the extracellular space pathway (GO: 0005783), endoplasmic reticulum pathway (GO: 0005783), and nuclear outer membrane–endoplasmic reticulum membrane network (GO: 0042175).

Figure 8

Target GO (Gene Ontology) chart analysis. (A) Biological process; (B) cellular component.

To explore the relationship of the two isolated compounds from E. abyssinica latex, common targets, and involved pathways, we constructed a compound–common target network. This network revealed that 3-acetyloxy-(3α)-urs-12-en-28-oic methyl ester (1) and lup-20(29)-en-3α,23-diol (2) isolated from E. abyssinica latex retained multiple targets and multiple pathways toward the mutagenesis of breast cancer formation. Five targets (AR, CYP19A1, EGFR, PGR, and PTGS2) might be the top therapeutic targets of isolated compounds on breast cancer by filtering these results and decreasing the scope of breast cancer types to invasive carcinoma types, as represented types for the MCF7 cell line, the same top target genes (AR, CYP19A1, EGFR, PGR, and PTGS2), seemed to possess maximum binding to the invasive types. Furthermore, the gene-enrichment analysis revealed that E. abyssinica could play a role in the treatment of breast cancer by striking 51 potential targets via mainly three signaling pathways: P13K–AKT, Wnt, and VEGF. The KEGG database enrichment genes proved the direct link of specific genes to cancer pathways, and these genes are AR, F2, and EGFR. These results assume that the potential anticancer activity of 3-acetyloxy-(3α)-urs-12-en-28-oic methyl ester (1) and lup-20(29)-en-3α,23-diol (2) isolated from E. abyssinica latex is due to the interaction of isolated compounds with target genes (AR, F2, and EGFR) through P13K–AKT, Wnt, and VEGF signaling pathways. Future research is required to analyze the isolated compounds from E. abyssinica latex and to validate the current results using individual biological models.

Drug-likeness and ADMET Properties

To assess these compounds’ drug-likeness, selected computed molecular descriptors pertaining to absorption, distribution, metabolism, excretion, and toxicity (ADMET) of phytochemicals were studied and compared to those of known drugs. Based on the obtained results, the two studied compounds demonstrated satisfactory values of predicted ADMET properties. In particular, the two compounds showed molCACO2 and molPAMPA values close to −5, which indicates high permeability (Table ).

Table 2

Prediction of Drug-likeness and ADMET Properties of Isolated Compounds

molecule	MolPAINS	MolCACO2	molPAMPA	BBB-Score	molLD50	Tox-Score	molHERG
compound 1	0.07	–4.79	–4.82	3.6	1.96	0	0.00
compound 2	0.04	–4.87	–5.13	4	2.07	0	0.01

To evaluate the potential toxicity of studied compounds, molLD50, ToxClass, and Tox-Score properties were calculated. Based on the obtained results, none of the compounds showed potential toxicity. All of the tested compounds have a molPAINS value of less than 0.5, which indicates that they have low “promiscuity” levels and do not tend to nonspecifically interact with the proteome. According to the obtained blood–brain barrier scores, none of the compounds received a score greater than 4, which indicates a low probability of their passing through the blood–brain barrier. Predicted values of the molHERG score indicate a low probability of interaction of selected compounds with hERG. The data greatly support the ability of the studied compounds to act as drugs.

Based on the molScore values, compound 2 showed strong potential to interact with the progesterone receptor (Table ). Based on the molpKd values, the two compounds demonstrated potential interaction with the progesterone receptor at submicromolar values.

Table 3

Prediction of Potential Interaction of Studied Compounds with the Progesterone Receptor by the Kernel Regression Chemical Fingerprint Classification (KCC) Model

In summary, we report here the cytotoxic effect of the dichloromethane extract of the latex of E. abyssinica and show that the extract displayed the best cytotoxic properties against the human breast cancer (MCF7) cell line. Two triterpenes, 3-acetyloxy-(3α)-urs-12-en-28-oic methyl ester (1) and lup-20(29)-en-3α,23-diol (2), were isolated from the dichloromethane extract. Both compounds showed promising cytotoxic effects against the MCF7 cell line. The approach of the pharmacology network was used essentially to elaborate the mechanistic role of the isolated compounds from E. abyssinica as anticancer agents targeting human breast cancers, identify the target genes that take part in the antitumor potential, and configure the possible pathways. The findings suggested that the activity of the two isolated compounds could provide support mostly in the development of a therapeutic strategy to introduce compound 1 and compound 2 as effective antitumor agents for breast cancer by providing a definitive mechanistic description of the possible target genes and pathways. Additionally, this promising cytotoxic activity supported by drug-likeness and ADMET investigations motivates the notion of integrating these compounds as candidates for the development of new naturally occurring anticancer drugs.

Materials and Methods

General Experimental Procedure

The 1H NMR and 13C NMR spectra were recorded in chloroform-d (CDCl3) using a Bruker AMX 400 MHz spectrometer with standard pulse sequences at 400 and 100 MHz, respectively, with tetramethylsilane (TMS) as an internal standard. Silica gel 60 (70–230 mesh, Merck, Sigma-Aldrich) was used for column chromatography. Thin-layer chromatography (TLC) was conducted on precoated silica gel 60 F254 plates (0.25 mm in thickness, Merck). Dichloromethane (DCM) was purchased from El-Nasr Company for Pharmaceuticals and Chemicals (Egypt). Doxorubicin, a reference cytotoxic drug, was purchased from Sigma-Aldrich Chemicals Co., St. Louis, MO, USA.

Plant Material

Samples of E. abyssinica used in this study were collected from Helal Cactus botanical garden, El Mansoria, Cairo, Egypt, in March 2019. Identification of the plant material was kindly verified by Mrs. Therese Labib, Head of the Taxonomists at El-Orman Botanic Garden. The voucher specimen (no. 4.3.2019) is kept in the herbarium of the Department of Pharmacognosy, Faculty of Pharmacy, Cairo University.

Extraction, Isolation, and Characterization

Fresh latex (750 g) was mixed with distilled water (2 L) and extracted with dichloromethane (2 L × 3). The dichloromethane extract (DCM) was evaporated under reduced pressure, yielding 100 g of residue. Fifty grams of DCM extract was chromatographed on a silica gel column (15 × 12 cm) eluted with n-hexane/DCM (100:0 to 0:100) and then DCM/ethyl acetate (100:0 to 0:100), yielding two major fractions. Fraction 1 (4 g) was chromatographed on a silica gel column and eluted with DCM/ethyl acetate (9:1 v/v) to give compound 1 (25 mg). On a silica gel column, fraction 2 (5 g) was chromatographed and eluted with DCM/ethyl acetate (8:2 v/v) to yield compound 2 (3 g).

3-Acetyloxy-(3α)-urs-12-en-28-oic methyl ester (1). White needles; EI-MS (70 eV: m/z 512 (M+, C33H52O4), 452, 497, 437, 262, 249, 203, 189, 133, 119, 69 (Figure S1). 1H NMR (CDCl3, 400 MHz): δ 5.22 (1H, t, J = 3.7, 3.4 Hz, H12), 4.47 (1H, dd, J = 6.2, 9.6 Hz, H3), 3.58 (3H, s, CH3), 2.21 (1H, d, J = 11.2 Hz, H18), 2.02 (3H, s, CH3), 1.61 (1H, d, J = 9.8 Hz, H16a), 1.05 (3H, s, H27), 0.92 (3H, s, H23), 0.91 (3H, d, J = 4.5 Hz, H29), 0.84 (3H, d, J = 4.4 Hz, H30), 0.83 (3H, s, H25), 0.83 (3H, s, H26), 0.72 (3H, s, H24) (Figure S2). 13C NMR (CDCl3, 100 MHz): δ 178.06 (C28), 170.89 (COOCH3), 138.11 (C13), 125.47 (C12), 80.93 (C3), 55.30 (C5), 52.87 (C18), 51.43 (COOCH3), 48.08 (C17), 47.48 (C9), 41.98 (C14), 39.50 (C8), 39.03 (C20), 38.87 (C19), 38.29 (C1), 37.68 (C4), 36.87 (C10), 36.63 (C22), 32.88 (C7), 30.64 (C21), 28.07 (C15, C23), 24.22 (C16), 23.56 (C2, 29), 23.31 (C27), 21.30 (C30), 21.18 (OCOCH3), 18.20 (C6), 17.06 (C11), 16.90 (C26), 16.72 (C24), 15.49 (C25) (Figure S3).

Lup-20(29)-en-3α,23-diol (2). White needles; EI-MS: m/z 512, 442 (M+, C30H50O2), 427, 411, 393, 93, 68 (Figure S4). 1H NMR (CDCl3, 400 MHz): δ 4.65 (1H, d, J = 1.8 Hz, H29a), 4.54 (1H, d, J = 1.8, H29b), 3.69 (1H, d, J = 10.4 Hz, H23), 3.59 (1H, dd, J = 7.2, 9.2 Hz, H3), 3.39 (1H, d, J = 10.4 Hz, H23), 2.32 (1H, ddd, J = 5.6, 10.6, 11.2 Hz, H19), 1.65 (3H, s, H30), 1.00 (3H, s, H26), 0.91 (3H, s, H24), 0.84 (6H, s, H25, 27), 0.75 (3H, s, H28) (Figure S5). 13C NMR (CDCl3, 100 MHz): δ 11.23 (C24), 14.57 (C27), 15.99 (C25), 16.45 (C26), 18.00 (C28), 18.46 (C6), 19.30 (C30), 20.93 (C11), 25.10 (C12), 27.02 (C2), 27.44 (C15), 29.84 (C21), 34.05 (C7), 35.56 (C1, 16), 37.06 (C13), 38.01 (C10), 39.99 (C22), 40.82 (C8), 42.85 (C4, 14), 43.00 (C17), 47.98 (C18), 48.28 (C19), 49.93 (C9), 50.43 (C5), 72.09 (C23), 77.02 (C3), 109.35 (C29), 150.95 (C20) (Figure S6).

Evaluation of Cytotoxic Activity

Cell Cultures

The American Type Culture Collection (ATCC, Minnesota, USA) provided the human cell lines MCF7 for breast cancer, HepG2 for hepatocellular carcinoma, HeLa for cervical cancer, and HBF4 for noncancerous human normal melanocytes. Cells were maintained in DMEM (Dulbecco’s modified Eagle’s media) enriched with streptomycin (100 μg/mL), penicillin (100 units/mL), and 10% heat-inactivated fetal bovine serum (FBS) in a humidified, 5% CO2 atmosphere at 37 °C. The cell lines were conserved by serial subculturing at the National Cancer Institute, Cairo, Egypt.

Cytotoxic Assay

The in vitro cytotoxic activity was assessed using the sulforhodamine-B colorimetric assay (SRB).[25] Briefly, in a 150 μL fresh medium, cells were seeded at a concentration of 3 × 103 cell/well in 96-well microtiter plates and left for 24 h to adhere to the plates. Different concentrations of DCM extract, compound (1 and 2), and standard doxorubicin (0, 5, 12.5, 25, and 50 g/mL) were applied, and three wells were used for each drug concentration. After 48 h of incubation, the cells were fixed for 1 h at 4 °C with 50 μL of cold trichloroacetic acid at a 10% final concentration. The plates were rinsed with distilled water (automatic washer Tecan, Germany), stained for 30 min at room temperature with 50 μL of 0.4% SRB/1% acetic acid, then washed with 1% acetic acid, and air-dried. The dye was dissolved in 100 μL of 10 M tris base (pH 10.5) per well. Each well’s optical density (O.D.) was measured using an ELISA microplate reader (Sunrise Tecan reader, Germany) at 570 nm. The following formula was used to compute the surviving fraction:

Statistical Analysis

The tests were conducted in triplicate, and IC50 (50% inhibitory concentration) was calculated using GraphPad Prism 5 (GraphPad Software Inc.) and expressed as μg/mL.

Compound Targets for E. abyssinica

The PubChem database (http://pubchem.ncbi.nlm.nih.gov/) was used as an online chemical database[26] to obtain the 3D molecular structure of isolated compounds. The 3D molecular structure files were imported into the Binding DB database (https://www.bindingdb.org/bind/index.jsp),[27] which is a drug-target identification online tool. After combining the duplicate data, we selected target genes with normalized fit scores greater than 0.7 as targets for E. abyssinica-isolated compounds (Table S1).

Breast Cancer Targets

The NCBI Gene database (http://www.ncbi.nlm.nih.gov/gene/),[28] the Online UniProtKB/Swiss-Prot database (https://www.uniprot.org/help/uniprotkb),[29] and the DrugBank database (http://www.drugbank.ca/)[30] were all used to obtain breast cancer-related genes.

Protein–Protein Interaction (PPI) Data

The PPI data were obtained from the online databases DisGeNET (https://www.disgenet.org/search)[31] and STRING (https://string-db.org/cgi/network?taskId=bIDN4htc9NBY&sessionId=bZWvNlZHMn9h)[32] as reliable databases for predicting protein–protein interactions. The target proteins were chosen with the human species “H. sapiens” and a confidence score greater than 0.4. DisGeNET and STRING were used to identify the proteins that interacted with the identified targets of E. abyssinica-isolated compounds and breast cancer directly or indirectly.

Network Construction

The obtained networks were visualized and restructured using the Cytoscape network analysis software,[33] version 3.9.0 (a software platform that visualizes complex networks and integrates the results).

Gene Ontology and Pathway Analysis

For all potential targets, the KEGG (Kyoto Encyclopedia of Genes)-enrichment analyses (https://www.genome.jp/kegg)[34] and Gene Ontology (ShinyGO) database (http://bioinformatics.sdstate.edu/go/)[35] were employed to investigate the biological process, cellular component, and associated pathways.

ICM-PRO[36] (www.molsoft.com) was used for the calculation of physicochemical descriptors and prediction of the ADMET properties and drug-likeness of isolated compounds (absorption, distribution, metabolism, excretion, and toxicity). Several scores, including “Blood–Brain Barrier Score” (if score is >4, then it indicates that the chemical can pass the BBB), “ToxScore” (if score is ≥1, then it indicates likely toxicity based on substructure match), “MolLD50” that predicts LD50 in mg/kg (a value of <0 indicates 1 mg/kg toxicity, and a value of 2 indicates 100 mg/kg toxicity), “MolCACO2” that predicts CACO-2 permeability LogP (a value > −5 indicates high permeability), “MolPAMPA” that predicts PAMPA permeability (a value > −5 indicates high permeability), “MolPAINS” for pan-assay interference compounds (a value > 0.5 indicates high probability of being a PAINS compound), and “MolHERG” that indicates interaction with hERG protein (values ≥ 0.5 indicate potential interaction), were used for the calculation of ADMET parameters and evaluation of the potency of isolated compounds.

The Kernel Regression Chemical Fingerprint Classification (KCC) model, which is a hybrid 2D QSAR/fingerprint machine learning model of MolScreen, was used for evaluation of potential interaction of studied compounds with one of the most key proteins associated with breast cancer, the progesterone receptor. Several scores were used to evaluate potential interaction: “molScore” (score > 3 indicates a binder) and “molpKd” (score > 6 indicates interaction at submicromolar values).