Integrated HTS², UPLC-MS/MS, and network pharmacology identifies Bruceae Fructus waste as potential source of flavonoids and quassinoids for inhibiting breast cancer

2.1. Preparation and extraction of BF, BFS, and BFK

BF (NO. 20210621) was purchased from the Bozhou Traditional Chinese Medicine market in Anhui province, China. The ripe fruits of B. javanica were identified by Prof. Guangzhi Wang from Chengdu University of Traditional Chinese Medicine. To obtain BFS and BFK, we hammered BF and separated the kernels from the mixture to obtain BFK as per Chinese Pharmacopoeia instructions. Meanwhile, the shells were also gathered as BFS. They were all stored in dry places, at room temperature, and away from lights. For extraction, BF, BFS, and BFK were smashed and extracted in a 95% ethanol solution using a Soxhlet extractor under heat for 8 hrs . After being evaporated, the remaining extracts were stored at -30°C.

2.2. Cell culture

MDA-MB-231 and HCC1806 cells were obtained from Zhejiang Meisen Cell Technology Co., Ltd. (Hangzhou, China). The cells were cultured in a 5% CO₂ humidified incubator at 37°C in medium supplied with 10% fetal bovine serum (FBS) (ExCell) and 100 units/mL of streptomycin and penicillin (Cytiva). MDA-MB-231 cells were cultured in DMEM (Gibco), while HCC1806 cells were cultured in RPMI1640 medium (Gibco).

2.3. Cell viability assay

The cells were seeded in 96-well plates (3,000 cell/well for HCC1806 and 5,000 cell/well for MDA-MB-231) and cell viability was measured using the CCK-8 assay. After being cultured for 24 hrs, the cells were treated with several different concentrations of the ethanolic extract of BF, BFS, or BFK for another 48 hrs (3 biological replicates for each concentration). The CCK-8 regent (Bioground) was then added to each well and incubated for 1 hr at 37°C. The absorbance of each well at 450 nm was measured using Varioskan LUX multimode microplate reader (Thermo Fisher Scientific Inc., MA, USA).

2.4. Colony formation and wound healing assay

For colony formation assay, about 1,000 MDA-MB-231 or HCC1806 cells were seeded in 6-well plates. The ethanolic extracts of BF, BFS, or BFK (5, 10 μg/mL for MDA-MB-231; 2.5, 5 μg/mL for HCC1806) were added in the culture medium, and the cells were incubated for 2 weeks (3 biological replicates for each concentration). Then the cells were washed twice with phosphate buffered saline (PBS) and fixed with 4% PFA at room temperature. The fixed cells were stained with a crystal violet staining solution (Beyotime) and the colony numbers were counted via ImageJ 1.54f [27]. For the wound healing assay, about 10,000 MDA-MB-231 cells were seeded in 96-well plates. The wound was made using Incucyte Woundmaker Tool (Sartorius, Germany). After washing with PBS, medium free of FBS was added with or without extracts at concentrations of 10, 20, and 40 μg/mL (4 biological replicates for each concentration). Migration of cells was monitored and pictured in IncuCyte Live-Cell Analysis System (Sartorius, Germany). The size of closure area was analyzed in ImageJ.

2.5. Cell cycle

About a million of MDA-MB-231 or HCC1806 cells were seeded in 6 cm dishes. After being cultured for 24 hrs, the cells were incubated with ethanolic extracts of BF, BFS, or BFK (5, 10 μg/mL for MDA-MB-231; 10, 20 μg/mL for HCC1806) for another 24 hrs (3 biological replicates for each concentration). After digesting with 0.25% trypsin (Servicebio), the cells were collected and washed with cold PBS. Then the cells were fixed in 75% ethanolic solution at -30°C overnight. The fixed cells were washed with PBS, stained using a PI/RNase Staining Buffer (BD), and detected in a FACSVerse flow cytometer (BD, NJ, USA). ModFit LT 6.0 was adopted for analyzing the distributions of cell cycle.

2.6. Apoptosis

Annexin V-fluorescein isothiocyanate (FITC)/ Propidium iodide (PI) Apoptosis Detection Kit (Vazyme) was applied to detect apoptosis in cells. Briefly, a million of MDA-MB-231 or HCC1806 cells were seeded in 6 cm dishes for 24 hrs and then treated with ethanolic extracts of BF, BFS, or BFK (10, 20 μg/mL) for 24 hrs (3 biological replicates for each concentration). The cells were gathered after digesting with 0.25% trypsin without ethylenediaminetetraacetic acid (EDTA) (Servicebio), then washed twice with cold PBS, and resuspended with 100 μL binding buffer. The cells were stained with Annexin V-FITC and PI staining solution for 10 min at room temperature. Additional 400 μL binding buffer was added before detection in FACSVerse flow cytometer (BD, NJ, USA). The data was analyzed in FlowJo 10.6.2.

2.7. HTS² assay

About 3,000 MDA-MB-231 cells were seeded in each well of 384-well plates. After incubation for 24 hrs, the cells were treated with ethanolic extracts of BF, BFS, or BFK at two concentrations (15 and 25 μg/mL, based on CCK-8 assay in 2.3) for another 24 hrs (3 biological replicates for each concentration). For choosing the concentrations of BF, BFK, and BFS in HTS², we first conducted a CCK-8 assay to test their impact on the cell viability of MDA-MB-231. IC50 values of BF, BFK, and BFS were 12.45, 11.53, and 17.67 μg/mL separately (perturbed for 48 hrs). But the perturbation time in HTS² was 24 hrs . At last, we chose 15 and 25 μg/mL for HTS² assay. Then, the cells were lysed and transferred to the HTS² platform for establishing cDNA library of transcripts as described [20]. The library was sequenced using NovaSeq 6000 (Illumina), and the mRNA level of the target gene was quantified as described [28]. The differentially expressed genes (DEGs) were calculated using DESeq2 [29] and the cutoff was set as |fold change| > 1.5 and P < 0.05. Additionally, the DEGs were imported to the ConnectivityMap (CMAP, https://clue.io/) [30] to query possible molecular mechanisms of functions.

2.8. Gene enrichment analysis

Gene ontology (GO) enrichment analysis, Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis, and Gene Set Enrichment Analysis (GSEA) were conducted in R 4.2.3 [31] using packages including clusterProfiler 4.6.2 [32], fgsea 1.24.0 [33] and tidyverse 2.0.0 [34].

2.9. UPLC-MS/MS analysis

Compounds were separated and identified using the Vanquish UPLC equipped with Q Exactive Orbitrap HRMS (Thermo Fisher Scientific Inc., MA, USA). Detailed chromatographic separation was conducted using Accucore^TM C₁₈ column (3 mm × 100 mm, 2.6 μm, Thermo Fisher Scientific Inc., MA, USA) at a temperature of 30°C. The mobile phase consisted of eluent A (0.1% formic acid aqueous solution, v/v) and eluent B (0.1% formic acid acetonitrile, v/v), and gradient elution was carried out as follows: 0-40 mins, 5%-95% B, 40-45 mins, 95% B, post time 5 mins. The flow rate was 0.3 mL/min and injection volume was 3 μL. The detection was under both positive and negative iron mode with HESI as an ionization source. Other conditions as follows: scan range, 100-1,500 m/z; spray voltage, 3,500 V (positive iron mode), 3,000 V (negative iron mode); capillary temperature, 320°C; probe heater temperature, 350°C; sheath gas, 35 arb; aux gas, 10 arb; scan mode, full MS/dd-MS²; resolution, 70,000 (full MS), 17,500 (dd-MS²); stepped collision energies were 20, 40, 60 eV. Raw data was processed using Compound Discoverer 3.3 (Thermo Fisher Scientific Inc., MA, USA) with self-built analysis template for identifying unknown compounds. As for validation strategy for UPLC-MS/MS compound identification, we compared MS/MS data with that in databases such as mzCloud, mzVault, Chemspider, and local chemicals libraries. We set filter criteria of MS/MS data (mzCloud Best Match or mzVault Best Match > 80) to obtain possible compounds. To confirm these compounds more robustly, we next searched MS/MS data of each compound in databases like SCIFinder and Spectral Database for Organic Compounds (SDBS). Only compounds with correct molecular ion peak and reasonable fragment ion peak were thought to be the compounds in BF, BFK or BFS. And the list was verified next with reported data.

2.10. Gathering chemical information of BF and prediction of potential targets

To collect compounds isolated from BF, we searched in electronic databases including SciFinder, PubMed, Wiley, Springer, CNKI, and Google Scholar using keywords “Brucea javanica,” “Bruceae Fructus,” and “yadanzi.” The names of compounds were recorded, and the structures were drew using ChemDraw 19.0. Using SMILES of each compound, we predicted their targets in SwissTargetPrediction (http://www.swisstargetprediction.ch/) [35].

2.11. Collection of breast cancer targets and generation of network

We searched for targets of breast cancer in public websites such as OMIM (https://www.omim.org/, updated February 23rd, 2024) [36], TTD (https://db.idrblab.net/ttd/) [37], PHARMGKB (https://www.pharmgkb.org/) [38], and Drugbank (https://go.drugbank.com/) [39]. Then we picked overlapped genes with that in which 2.9 as BF’ targets breast cancer. The overlapped genes were also searched in STRING (https://cn.string-db.org/, version 12.0) [40] to get the hub genes, and were visualized in Cytoscape 3.10.0.

2.12. Molecular docking

The crystal structures of proteins in this study were obtained from the RCSB Protein Data Bank (RCSB PDB, https://www.rcsb.org/) [41]. And molecular docking was conducted using Schrodinger’s Maestro (Schrödinger Release, 2019–1). In brief, the protein was prepared using Protein Preparation module. Using LigPrep module, the possible 3D structures of compounds were prepared for docking. According to the known active binding sites, Glide was constructed for next molecular docking calculations and simulations [42].

2.13. Statistical analysis

Data in this study is presented as mean ± SEM. Comparation between two groups involved two-tailed unpaired Student’s t-test, and one way ANOVA tests followed by Tukey’s multiple comparisons were used for multiple groups. The significance level was set at *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

3.1. Comparing different pharmacological activities of BF, BFK, and BFS

3.1.2. Effects on cell cycle and apoptosis of TNBC cells

Considering BF, BFK, and BFS could inhibit the growth, proliferation, and migration of TNBC cells at relatively low concentrations, we explored whether it was relevant to cell cycle and apoptosis. With analysis of DNA content, we found that BF arrests the cell cycle of MDA-MB-231 cells at the G0/G1 phase, and HCC1806 at the S stage (Figure 2a). The activity of inducing apoptosis was also observed in MDA-MB-231 and HCC1806 (Figure 2b). Both BFK and BFS exhibited similar effects of arresting the cell cycle and inducing apoptosis.

Close

Figure 2.

The ethanolic extracts of BFK, BF, and BFS induce cell cycle arrest and apoptosis of MDA-MB-231 and HCC1806 cells. (a) DNA content analysis in MDA-MB-231 and HCC1806 using flow cytometry. (b) Apoptosis analysis in MDA-MB-231 and HCC1806 using flow cytometry. BF: Bruceae Fructus, BFK: Kernels of Bruceae Fructus, BFS: Shell of Bruceae Fructus. DMSO: Dimethyl sulfoxide.

Export to PPT

3.2. HTS² analysis

3.2.1. BFS regulates similar number of DEGs in MDA-MB-231 with BF or BFK

Using dimethyl sulfoxide (DMSO) (negative control) and JQ1 (positive control), we treated MDA-MB-231 cells with two concentrations of ethanolic extracts of BF, BFK, and BFS. The correlation between each pair of samples in the DMSO or JQ1 group was found to be greater than 0.94, indicating excellent repeatability within the experimental groups (Figure A1a, b). Samples in DMSO and JQ1 group were separated into two different clusters using uniform manifold approximation and projection (UMAP) analysis (Figure A1c), suggesting the reliability of HTS² assay.

Supplementary Figure 1

Download

Next, we visualized the gene regulations in MDA-MB-231 cells treated with different extracts using volcano plots (Figure 3a). A Venn plot revealed a total of 240 DEGs identified in the BFK group, 182 in the BF group, and 181 in the BFS group at a concentration of 15 μg/mL. At a concentration of 25 μg/mL, 470 DEGs were identified in the BFK group, 446 DEGs in the BF group, and 437 DEGs in the BFS group (Figure. 3b). With respect to gene perturbation, BFS exhibited similar pattern of gene regulation with BF or BFK in MDA-MB-231 cells.

Close

Figure 3.

HTS² technology illustrated mechanisms of BF, BFK, and BFS treating breast cancer. (a) Volcano plot profiling DEGs and (b) Venn diagrams of DEGs in MDA-MB-231 treated with ethanolic extracts of BF, BFK, and BFS. (c) GO enrichment analysis and (d) KEGG pathway enrichment analysis of DEGs in MDA-MB-231 treated with ethanolic extracts of BF, BFK, and BFS. BF: Bruceae Fructus, BFK: Kernels of Bruceae Fructus, BFS: Shell of Bruceae Fructus, GO: Gene ontology, KEGG : Kyoto encyclopedia of genes and genomes, HTS²: High throughput screening, DEGs: Differentially expressed genes.

Export to PPT

3.2.3 Results of GSEA

Cell cycle and apoptosis were enriched in GO and KEGG enrichment analysis. Previous experiments have validated the ability of BF, BFK, and BFS to arrest the cell cycle and induce apoptosis in TNBC cells. We performed GSEA using HTS² data to illustrate how they influence these pathways at molecular level. As a result, most genes in “KEGG CELL CYCLE” were downregulated in MDA-MB-231 treated with extracts of BF, BFK, and BFS. Some genes encoding cell cycle checkpoint proteins like CDC6, CDK1, CCNB1, and CCNB2 were downregulated. In Figure 4 SKP2, a regulator of p21 and p27 [43], was also downregulated (Figure 4a). Additionally, most genes in “HALLMARK P53 PATHWAY” were upregulated, indicating activation of the pathway after treatment (Figure 4c). The activation of p53 might result in the cell-cycle arrest of TNBC cells [44,45]. Genes in “HALLMARK APOPTOSIS” were upregulated after treatment with extracts of BF, BFK, and BFS. For example, in the BFK group, CASP7, an activator of apoptosis [46], was upregulated. The increasing expression of TXNIP, DDIT3, and GADD45A suggested the accumulation of reactive oxygen species (ROS) and DNA damage [47-49]. The upregulations of these genes were also observed in the BF and BFS groups (Figure 4b).

Close

Figure 4.

BF, BFS, and BFK inhibited the pathway of (a) “KEGG CELL CYCLE”, and activated pathways of (b) “HALLMARK APOPTOSIS” and (c) “HALLMARK P53 PATHWAY”. The lower heatmap shows an expression of the core gene in the pathway in the form of log2 (count + 1). BF: Bruceae Fructus, BFK: Kernels of Bruceae Fructus, BFS: Shell of Bruceae Fructus, NES: Normalized enrichment score, KEGG: Kyoto encyclopedia of genes and genomes.

Export to PPT

3.3. The chemical composition of BF, BFS, or BFK varies

BFK is traditionally used in clinical practice, unlike BF or BFS. Therefore, we compared the differences in chemical composition among them using UPLC-MS/MS (Figure 5a, b). We identified a total of 60 ingredients in BF, with 31 discovered in negative mode and 29 in positive mode. Quassinoids such as bruceine H were detected only in BFK, while yadanziolide B was only found in BFS. Although peak areas were smaller in BFS than in BFK, compounds like bruceolide, brusatol, bruceine A, and bruceine D were present in both parts of BF. Additionally, organic acids and flavones such as caffeic acid, isovanillic acid, pimelic acid, ferulic acid, quercetin, quercetin-3β-D-glucoside, kaempferol, and luteolin were more readily found in BFS. The MS/MS data and peak areas of each compound have been presented in Table 1. These findings suggest that BFS contains lower content of quassinoids, but more flavones compared to BFK.

Close

Figure 5.

UPLC-MS/MS chromatograms of ethanolic extracts of BF, BFS, and BFK in (a) positive and (b) negative modes. BF: Bruceae Fructus, BFK: Kernels of Bruceae Fructus, BFS: Shell of Bruceae Fructus, UPLC: Ultra performance liquid chromatography, MS: Mass spectrometry.

Export to PPT

3.4. Results of network pharmacology

3.4.2. Collection of breast cancer targets and PPI network analysis

We collected therapeutic targets of breast cancer from OMIM, TTD, PHARMGKB, and Drugbank (Table A2). Ultimately, a total of 114 targets overlapped with those identified previously (in section 3.4.1) and were considered therapeutic targets (Figure 6a). Subsequently, we constructed a compound-target network (Figure 6b), which revealed that 62 quassinoids exerted significant anti-cancer effects. Analysis of the protein-protein interaction (PPI) network (Figure 6c) and topological analysis identified hub genes including TP53, SRC, PIK3CA, PIK3CB, PIK3R1, ESR1, AKT1, PIK3CD, CTNNB1, and CYP3A4 (Table 2). Given the higher abundance of quassinoids and lower presence of flavones in BFK compared to BFS, we compared the different targets of these compounds. PPI network analysis (Figure 6d, e) revealed that quassinoids influenced hub genes such as EGFR, SRC, AKT1, PIK3CA, and PIK3CB, while flavones acted on TP53, EGFR, SRC, AKT1, and ESR1. Together, BF may exert anti-breast cancer effects through modulation of these hub genes.

Supplementary Table 2

Download

Close

Figure 6.

Network pharmacology illustrating mechanisms of BF treating breast cancer. (a) Venn diagram of targets of BF and breast cancer. (b) Compound-target network was built using 114 overlapped genes. The round nodes are chemicals in BF while the square ones are therapeutical targets of breast cancer. (c) PPI network of 114 overlapped targets, (d) targets of quassinoids, and (e) targets of flavones. The size or color shade of the node is positively associated with the degree value. (f) GO enrichment analysis of 114 targets of BF treating breast cancer. (g) KEGG pathway enrichment analysis of 114 targets of BF treating breast cancer. BP: Biological process; CC: Cellular component; MF: Molecular function, GO: Gene ontology, KEGG : Kyoto encyclopedia of genes and genomes.

Export to PPT

Table 2. The topological analysis of the PPI network.

GENE	Degree centrality	GENE	Betweenness centrality	GENE	Closeness centrality
TP53	27	TP53	0.314625726	TP53	0.461139896
SRC	23	ESR1	0.20892383	ESR1	0.458762887
PIK3CA	21	CYP19A1	0.19645871	MAPK8	0.419811321
PIK3CB	18	SRC	0.134042024	AKT1	0.41588785
PIK3R1	18	MAPK8	0.107051319	CTNNB1	0.413953488
ESR1	18	GSTP1	0.084028189	PIK3CA	0.412037037
AKT1	18	CYP3A4	0.083869476	EGFR	0.402714932
PIK3CD	17	AKT1	0.082982932	SRC	0.395555556
CTNNB1	16	CASP3	0.072349842	ERBB2	0.39380531
CYP3A4	16	PTGS2	0.070442016	CYP19A1	0.388646288
EGFR	15	PIK3CA	0.069218826	CCND1	0.388646288
CYP1A1	15	CYP1A1	0.064888167	BCL2	0.381974249
IGF1R	13	CTNNB1	0.045637558	MDM2	0.38034188
ERBB2	12	HIF1A	0.045168344	CASP3	0.378723404
CCND1	12	TOP2A	0.039542082	HIF1A	0.378723404
CYP1A2	12	UGT2B7	0.035694459	ESR2	0.377118644
UGT2B7	11	EGFR	0.03368072	PIK3R1	0.375527426
CYP2A6	10	CCND1	0.031433278	AR	0.375527426
CYP2C9	10	PIK3R1	0.03136549	PTGS2	0.372384937
BCL2	10	CYP2C9	0.031137693	IGF1R	0.367768595

PPI: Protein protein interaction.

3.5. PI3K/Akt signaling pathway was inhibited in MDA-MB-231

Since the PI3K/Akt signaling pathway was enriched significantly in network pharmacology, we checked the gene expression in this pathway. As shown in Figure 7a, most genes were downregulated including PIK3CB, PIK3CD, PIK3R2, PIK3R3, AKT1, and AKT2, while upstream regulators like ITGB1, ITGB4, ITGA6, ITGA2, and ITGAV were downregulated as well. We also observed that BF, BFK, and BFS groups had different DEGs at high concentrations, though they shared some of them (Figure 3b). To explore the effects of different DEGs, these genes were analyzed in CMAP. As a result, the common gene regulation profile of the three groups was similar to proteasome inhibitor, NFκB inhibitor, HDAC inhibitor, topoisomerase inhibitor, PI3K inhibitor, MTOR inhibitor, and CDK inhibitor (Figure 7b). The work pattern of DEGs regulated only in BF or BFK group was the same as common DEGs (Figure 7c, d). However, profiling of DEGs only in the BFS group behaved more like a protein synthesis inhibitor (Figure 7e). The above indicated that BF, BFK, and BFS all inhibited the PI3K/Akt signaling pathway.

Close

Figure 7.

BF, BFK, and BFS inhibited the PI3K/Akt signaling pathway. (a) DEGs in PI3K/Akt signaling pathway. L means the low concentration set at 15 μg/mL, while H means the high concentration at 25 μg/mL. Top 20 Mode of action analysis (MoA) terms analyzed in CMAP using common DEGs or (b) particular DEGs in groups treated with ethanolic extracts of (c) BF, (d) BFK, and (e) BFS at the concentration of 25 μg/mL. BF: Bruceae Fructus, BFK: Kernels of Bruceae Fructus, BFS: Shell of Bruceae Fructus, DEGs: Differentially expressed genes, PI3K/Akt: Phosphoinositide 3-kinase/protein kinase B, CMAP: Connectivity map, MoA: Mode of action.

Export to PPT

3.6. Quassinoids and flavones in BF bind to targets in PI3K/Akt pathway

According to the results of HTS² and network pharmacology analysis, the PI3K/Akt signaling pathway might be key in treating breast cancer. The PI3K/Akt signaling pathway is thought to promote cell growth, survival, and cell cycle progression [50]. However, it is aberrantly activated in various cancers, which results in bad diagnoses and prognoses [51]. To explore what compounds affect the PI3K/Akt pathway, we chose several key proteins of this pathway to conduct molecular docking. Crystal structures of PI3Kα (PDB ID: 6PYS), PI3Kγ (PDB ID: 4ANV), AKT1 (PDB ID: 7NH4), and PDK1 (PDB ID: 3QC4) were downloaded from RCSB PDB, and 3D structures of compounds in BF were prepared.

With a threshold of gscore < -6, we identified 112 compounds probably binding to PI3Kα, 92 to PI3Kγ, 117 to AKT1, and 77 to PDK1 (Table A3). Except for flavones like rutin and isovitexin, several quassinoids could bind to these targets. Yadanzigan binds to PI3Kα (docking gscore: -10.637), PI3Kγ (docking gscore: -10.133), or AKT1 (docking gscore: -11.645) by hydrogen-bonding with amino acid residues of the protein. In the interaction with PI3Kα, the hydroxyl group at C1, C11, C14, C4’, and C6’ of yadanzigan interacted with Ser774, Gln859, Lys802, and Asp933 respectively, while the ketone group at C16 worked with Thr856 (Figure 8a). With PI3Kγ, the hydroxyl group at C1, C1’, C12, and C14 formed hydrogen-bonds with Ala805, Ser806, and Asp954, and that at C6’ with Lys802 and Val803 (Figure 8b). As for AKT1, the hydroxyl group at C1, C11, C12, C14, and C3’ interacted with Lys276, Asn279, Glu278, Leu156, TYR18, and Asp274 (Figure 8c). And glycosyl of bruceoside B (docking gscore: -8.595) inserted into the PDK1 pocket by hydrogen-bonding with Ser160 and Ala162 (Figure 8d). Together, our analysis suggested that quassinoids and flavones in BF might bind to key targets in the PI3K/Akt signaling pathway.

Close

Figure 8.

Representative molecular docking results of breast cancer targets. (a) The patterns of yadanzigan binding to PI3Kα, (b) PI3Kγ, (c) AKT1, and (d) bruceoside B binding to PDK1.

Export to PPT

3.7 Discussion

Traditional Chinese medicine (TCM) has been used in China for more than 2000 years. Although animals, insects, or minerals can be the sources of TCM, most TCM uses plants as sources. Long-term clinical practice has developed the principles of certain positions in a plant exerting particular therapeutical effects. For example, peels and seeds of Trichosanthis Fructus (“Gualou” in Chinese) are always used separately. The seeds can loosen the bowels and relieve constipation, while the pericarps prefer to clear heat, resolve sputum, and promote the flow of qi to soothe chest oppression according to the TCM theory [52]. As a result, non-medicinal parts of plants are usually abandoned in the process of actual production. It is no doubt a waste of great amounts of plant biomass. Incineration, stacking, and burying are traditional methods to dispose the waste, meanwhile ethanol and biogas preparation, biofuel and bioenergy production, paper making, and bioactive components extraction are also ways of proper utilization of the waste in industry [53,54]. However, potential medicinal values should be assessed first in terms of TCM residues to fulfill sustainable practices in pharmaceutical development. This study practiced a strategy that data mining and UPLC-MS/MS reveal the phytochemical profiles together with network pharmacology, HTS², and experimental validation exploring pharmacological activities. BFS was proven to exert an anti-breast cancer effect by inhibiting proliferation and migration, arresting the cell cycle, and inducing apoptosis. The bioactive compounds are mainly flavones and quassinoids, which target the PI3K/Akt pathway. These compounds could be candidates for exploring PI3K/Akt inhibitors in treating breast cancer. Regarding other agro-industrial residues or underutilized medicinal plants, the integrative strategy could begin by identifying their chemical composition using UPLC-MS/MS. Based on traditional uses or areas of interest, potential disease models and cell models could then be selected. The gene expression profiles of cells treated with extracts from these residues or plants could be obtained through the HTS² assay. Ultimately, this approach may lead to the discovery of potential compounds targeting specific diseases.

B. javanica is also called “Kusum” in Malaysia, “Kuwalot” in Indonesia, “balaniog” in Philippines, and “ratchadat” in Thailand. It is used for treating malaria [55,56]. In Cambodia, the roots are anthelmintic and can be used in the treatment of malaria and dysentery [57]. And the leaves have effects on acne [58]. However, there are no relevant products in use, manufactured with these parts. The life span of B. javanica is only a few years. Thus, it is worth studying the usage of the roots, stems, barks, and leaves without worrying about the production of BF. Perhaps, the strategy proposed in this study could help discover the economic value of these parts. Certainly, it would provide new insights into discovering proper utilization of other agro-industrial residues. By now, over one hundred quassinoids of great pharmacological activities such as anti-cancer, anti-inflammatory, and anti-malaria have been isolated from this plant [12,13,15,59]. But none have been approved for clinical uses because of severe side effects [60-62]. In the future, the chemicals isolations from this plant, or structure modification of quassinoids should be carried out.

With a data mining approach to profiling the phytochemical basis of BF, about 167 isolated compounds were collected, of which most were quassinoids and flavones. The results of network pharmacology and molecular docking also suggested that quassinoids and flavones were the main active ingredients targeting the PI3K/Akt pathway. PI3K is a critical protein mediating tumor cell growth, proliferation, and metabolism [63]. The PI3K/Akt signaling pathway is thought to promote cell growth, survival, and cell cycle progression [50]. However, it is aberrantly activated due to multiple genomic alterations of PIK3CA and AKT across various cancers leading to bad diagnosis and prognosis [51,64]. PIK3CA mutations have been identified in approximately 10% of TNBC cases. By now, only three PI3K/Akt inhibitors alpelisib, capivasertib, and everolimus have been approved by FDA for breast cancer treatment. Therapeutic resistance of PI3K/AKT inhibitors is associated with receptor tyrosine kinase reactivation, acquired mutation and/or amplification of PIK3CA or PIK3CB, and PI3K reactivation [50]. As Figure 7(a) exhibits, most genes in the PI3K/Akt pathway were downregulated by BF, BFK, and BFS, which illustrates that the PI3K/Akt signaling pathway was inhibited more effectively.

UPLC-MS/MS revealed that BFS contains more flavones like quercetin, quercetin-3β-D-glucoside, kaempferol, and luteolin than BFK. Quercetin is a common dietary flavone and has been proven to exert anti-breast cancer effects by inducing apoptosis and arresting the cell cycle [65]. Kaempferol suppresses the growth of breast cancer cells by upregulating the expression of p21 and bax [66]. In addition, it effectively decreases the incidence of breast precancerous lesions by activating the LKB1/AMPK pathway and inducing excessive mitochondrial fission and lethal mitophagy [67]. Another flavone, luteolin, could suppress the proliferation and metastasis of breast cancer cells by inactivating the AKT/mTOR pathway [68]. There are also other flavones exerting anti-TNBC effects. Sophoraflavanone G was reported to inhibit proliferation and metastasis by inactivating the EGFR/PI3K/Akt signaling pathway [69]. And both fisetin and chrysin could inhibit metastasis of TNBC [70,71]. Flavones like quercetin, kaempferol, and luteolin were seldom reported to induce toxicity or side effects. However, a high dose of quercetin might induce kidney-related issues, headaches, and stomach aches [72]. Although these flavones exert outstanding anti-cancer effects, their oral bioavailability is extremely poor, which limits their application in clinical settings [73,74]. Besides, quassinoids like brusatol, bruceine A, bruceine D, and bruceine H also exhibit outstanding anti-cancer effects in various types of cancers. The common dosage of these quassinoids in animal models is 1 mg/kg or 2 mg/kg (i.p or i.v). Although their anti-cancer effects are obvious, short t1/2 and MRT might increase times of doses and induce severe side effects [75,76]. Bruceantin was considered in clinical trials treating metastatic breast carcinoma and advanced malignant melanoma [60,62] but failed because of poor effects and severe side effects, including nausea, vomiting, mild hypotension, and fever. Their abilities of anti-breast cancer have seldom been evaluated. Approximately 97 quassinoids have been identified from BF, but the pharmacological activities of only several compounds were explored. Thus, the research of BF in the future could focus on these aspects.

TNBC is defined as a “cold tumor” characterized by a lack of effective immune cell infiltration within the tumor microenvironment. This feature poses a significant challenge to immunotherapy approaches, including PD-1/PD-L1 inhibitors. Only 10% of TNBC patients could response to checkpoint blockade immunotherapies like PD-1/PD-L1 inhibitors [77]. Scientists make efforts to turn “cold tumors” into “hot tumors” to treat breast cancer. Some inhibitors were identified to promote infiltration of effective CD8⁺ T cells and work better with immune checkpoint inhibitors [77,78]. PI3Kβ, encoded by PIK3CB, was found to promote immune escape in PTEN-null tumors [79]. In this study, BF treatment could decrease the expression of PIK3CB (Figure 7a), suggesting that BF might inhibit PI3Kβ and promote anti-tumor immune responses. Except for PI3K/Akt signaling pathway, cell cycle, and apoptosis, we identified other interesting pathways that might treat breast cancer. In this study, “IL−17 signaling pathway” was enriched. IL-17E, also known as IL-25, is a subtype in the IL-17 family. Studies have shown that IL-17E can induce apoptosis of breast cancer cells, and related treatment can reduce tumor growth [80,81]. Also, it can attract lymphocytes [82]. In addition, some immune pathways like “NOD-like receptor signaling pathway,” “T cell receptor signaling pathway,” “Toll-like receptor signaling pathway,” and “B cell receptor signaling pathway” were significantly enriched (Figure 3c and d), suggesting that BFS might communicate with immune cells in the immunosuppressive microenvironment. Thus, it is worth exploring whether BFS could attract more effective immune cells like T-cells or NK cells to the environment.