Hydroxychavicol (HC) was bought from Hangzhou Imaginechem Co. Ltd. (Hangzhou, China) and (−)-epigallocatechin-3-gallate (EGCG) was bought from Sigma-Aldrich (United States). CellTiter 96^® AQueous Non-Radioactive Cell Proliferation kit (Promega, United States), FITC Active Caspase-3 and Annexin V-FITC Apoptosis Detection Kit, BD CycleTEST™ PLUS DNA Reagent kit from BD Biosciences (United States) and QCMTM 24-well Cell Invasion/Migration Assay kit (ECM550 and ECM508) (Millipore, United States). All of the other chemicals utilised were of analytical grade.

The human glioblastoma cell line 1321N1 (Grade II) was bought from the European Collection of Cell Culture (ECACC), while the American Type Culture Collection (ATCC) supplied SW1783 (Grade III) and LN18 (Grade IV) cell lines (Manassas, United States). 1321N1 and LN18 were cultured in Dulbecco’s modified Eagle medium (DMEM) supplemented with penicillin/streptomycin, 10% or 5% foetal bovine serum (FBS) respectively, in a humidified incubator at 37°C in an atmosphere of 95% air and 5% CO2. SW1783 was grown in Leibovitz, 10% FBS, under a 100% air atmosphere. The medium was replaced three times a week, and the cells were passaged with accutase.

Fresh EGCG stock solutions were made in culture growth media, whereas HC stock solutions were prepared in 100% ethanol and kept at −20°C. Vehicle control was added with 0.1% ethanol.

Glioma cancer cell viability treated with combined EGCG+HC or EGCG/HC singly was determined using the Cell Proliferation Assay (Promega, United States), as previously described [11]. Cells were seeded at 1×10⁴ cells per well in 96-well plates. After a 24-h incubation period, the media was withdrawn and 100 μl of medium were added, which contains a range of concentration for EGCG (50, 100, 150, 200, 300 μg/ml) or HC (50, 100, 150, 200 μg/ml). EGCG+HC compounds were titrated to a range of concentrations (1, 10, 50, 100 μg/ml). The treatments were incubated for 24 h. The media was then carefully removed, replaced with new medium, and 20 μl of [3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxy-phenyl)-2-(4-sulfophenyl)-2-(4-sulfophenyl)-2H-tetrazolium, inner salt] (MTS) was added to each well and incubated for 2 h at 37°C. In a VersaMax ELISA microplate reader, absorbance was measured at 490 nm (Molecular Device, United States). At each concentration, the percentage of viable cells was estimated by dividing the absorbance (490 nm) of treated cells by that of control cells. The cell viability (%) versus concentrations graph was used to calculate the half maximum inhibitory concentration (IC50). All tests were carried out in three independent experiments.

Cells were seeded at a density of 5 × 10⁵ cells/dish in a 60 mm culture dish for all functional assays. EGCG was dissolved in culture medium, while HC was dissolved in ethanol and added to the culture media with stipulated concentration. Cells were collected after 24 h and washed twice with PBS. The assays were carried out as specified in the manufacturer’s procedure. In brief, for active caspase-3 assay, cells were fixed in BD Cytofix/Cytoperm solution, incubated on ice for 20 min, washed with BD Perm/Wash buffer, and then incubated for 30 min at 25°C with FITC rabbit anti-active caspase-3 antibody. For annexin V-propidium iodide staining, cells were resuspended in 1X binding buffer. Annexin-V FITC and propidium iodide (PI) were added and incubated in the dark for 15 min at 25°C. The BD FACSCanto™ flow cytometer and CellQuest Pro (IVD) software (Becton Dickenson, United States) were used to detect fluorescence from a population of 1 × 10⁵ cells. Three independent tests were carried out in triplicate.

The cells were prepared according to the manufacturers’ protocol. Cells were trypsinized, washed, and fixed in Buffer Solution at 4°C. After 10 min of incubation with trypsin buffer at 25°C, 200 μl of trypsin inhibitor and RNase buffer solution were added and incubated for another 10 min. In a dark room, 200 μl of propidium iodide stain solution were added to the mixture and incubated on ice for 10 min. The analysis were done using the BD FACSAria™ flow cytometer, FACScan and ModFit software (Becton Dickenson, United States).

The wound healing test was carried out as previously reported, with several modifications (Liang et al., 2007). After 24 h of plating in a 6-well plate, cells were scratched with a 200 μl sterile pipette tip, rinsed three times with PBS, and incubated with the treatments for another 24 h. The cells were rinsed twice with PBS before being examined and photographed using a Nikon Eclipse TS100 phase-contrast microscope. Using the NIS-Elements imaging programme, the migration percentage was estimated by comparing cells which migrated into scratched regions to 0 h cells.

The assays were prepared according to the manufacturers’ protocol (Millipore, United States). Both invasion and migration kits utilized an 8 μm pore size polycarbonated membrane insert. A thin layer of ECMatrix™ were pre-coated on the insert for invasion kit which function to seal the membrane pores and block the non-invasive cell migration. Briefly, 300 μl of cells re-suspended in a serum-free medium was added to the upper chamber, while the bottom well was filled with 500 μl of complete culture medium or treatment. Following 24 h incubation, unmigrated cells were removed from the upper chamber, and the migration insert containing migrated cells were transferred into a clean well containing 400 μl of staining buffer. After 20 min incubation at 25°C, the inserts were rinsed in water, and unmigrated cells were removed from the inside of the insert with a cotton-tipped swab. After drying, the stained inserts were placed to a clean well containing 200 μl of Extraction Buffer for 15 min at 25°C. 100 μl of the mixture solution was pipetted a 96-well plate and the absorbance was measured at 560 nm.

Following a 24 h treatment with combined EGCG+HC and EGCG/HC singly, approximately 400 cells were plated in a 21 cm² culture dish. The cells were grown in a complete media for 12 days and the medium were replaced every 3 days. On day-12, the colonies formed were washed with PBS, and then fixed for 30 min with a mixture of crystal violet solution and methanol (1:1). The plate were rinsed three times with distilled water to remove any excess staining. ChemiDocTM MP (Biorad, United States) was used to capture the images of the stained plates, and Cell Counter v0.2.1 (http://nghiaho.com/?page id=1011) was used to count the colonies. Each treatment was carried out in triplicate.

Isobologram analysis based on the Chou-Talalay technique (Chou and Talalay, 1984; Zhao et al., 2004) was used to determine the interaction between the two treatments, with the output represented as combination indexes (CI). The CI between two compounds A and B is as follows: CI = d 1 ( D m ) 1 + d 2 ( D m ) 2 Where; CI: combination index d ₁: the IC₅₀ of combination dose for compound 1 d ₂: the IC₅₀ of combination dose for compound 2 (D _m)₁: the IC₅₀ dose for compound 1 (D _m)₂: the IC₅₀ dose for compound 2.

The magnitude of synergism/antagonism were measured using CI values. CI values between 0.9 and 0.85 indicate mild synergy, those between 0.7 and 0.3 indicate strong synergistic interactions between the treatments. A near additive effect is shown by CI values ranging from 0.9 to 1.10. SPSS 16.0 software was used to analyse the two-tailed Student’s t-test for comparison with vehicle control only (cell viability with single treatments), or two-way ANOVA for multiple comparisons of apoptosis, cell cycle, migration/invasion, and colony formation tests where p < 0.05 were considered statistically significant. The data were presented as mean ± standard deviation (SD).

After 24 h of glioma cells treatment, total RNA was using TRI Reagent^® (Molecular Research Center, United States). RNase-free DNase treatment was used to remove contaminating DNA (QIAGEN), and RNA was purified using the RNeasy Mini Kit (QIAGEN). The quality of the RNA was then determined using an Agilent Bioanalyzer 2100 (Santa Clara, United States) and all readings have a minimum RIN score of 9.5. The Qubit 2.0 Fluorometer was used to measure the amount of RNA (Life Technologies, United States). The cDNA libraries were prepared according to the protocol outlined in the TruSeq RNA sample Preparation Kit v-2 (Illumina, San Diego, United States). The description of the procedure may be found elsewhere (Abdul Rahman et al., 2019). A total of 10–12 samples per lane were multiplexed and sequenced on an Illumina HiSeq 2500 using the paired-end cluster generation kit (Illumina).

The analysis of RNA-seq data was previously described elsewhere (Abdul Rahman et al., 2019). For quality control, RNA-seq raw data was trimmed at a PHRED score of <Q25, with a read length of at least 33 bp, and read quality was evaluated for each sample using FastQC. A total of 28–47 million paired end reads were aligned and mapped against the Human Genome version 37, GRCh37/hg19. Significant differences in gene/transcript expression were determined for pairwise comparisons between two sets of samples using the Empirical analysis of DGE (Robinson and Smyth, 2008) by Negative Binomial distribution followed by Bonferroni multiple testing correction (MTC) and Benjamini–Hochberg false discovery rate (FDR) using the CLC Genomics Workbench (7.0.6 version), and by Cuffdiff 2.0 in Tuxedo Suite pipeline (Trapnell et al., 2012). For each comparison, the intersection of the significant genes discovered using both algorithms were deemed differentially expressed.

Pathway Studio (Ariadne Genomics, United States) was used for network analysis, which included Gene Significant Enrichment Analysis (GSEA), Fisher Exact Test (FET), and subnetwork analysis of RNA-seq data, using the mean of RPKM values, and the gene expression log ratio of treatment to control cells were calculated. DAVID Bioinformatics Resources 6.7 was also used for gene ontology annotation. In the subnetwork analysis, genes were deemed key regulators of a network if they controlled five or more gene targets. These networks give a global picture of potentially important, interacting partners of genes that have undergone significant alterations. For the assessments of alternative splicing expression, two types of analyses were performed using the Partek^® Genomics Suite (Partek Inc., United States) and the Tuxedo Suite (Tophat and Cuffdiff 2.0). Alternative splicing entities expressed from both platforms were overlapped (Supplementary Tables S1–S3). The expression of alternative splicing events in both analyses was then overlapped to the list of transcripts expression at p ≤ 0.05 with FC ≥ 1.5, to get the list of alternative splicing events with a significant differential transcripts expression (Supplementary Tables S1–S3).

The genes were chosen for validation because they either demonstrated the most significant increase or reduction in response to EGCG+HC therapy, are a central gene, or are important to cancer formation. 300 ng of total RNA were used for cDNA synthesis using the iScript™ cDNA Synthesis kit (BioRad, United States). TaqMan^® Gene Expression Assays and TaqMan^® Fast Advanced Master Mix (Applied Biosystem, United States) were used for the amplification of genes and transcripts. cDNA synthesis reactions were performed on Veriti^® Thermal Cycler (Applied Biosystems) and qPCR were performed on CFX96 Touch™ C1000 Touch Thermal Cycler (BioRad, United States). The data were normalized to the expression of housekeeping genes TATA box binding protein (TBP) or glucuronidase, beta (GUSB) and analyzed using the standard 2^-∆∆CT method.

We have previously reported the effect EGCG and HC singly on glioma cell viability (Abdul Rahman et al., 2014;Rahman et al., 2014). The optimum IC50 doses for EGCG, HC and its combination for each cell lines were obtained by performing an initial dose response curve (Figure 1). The growth of 1321N1, SW1783 and LN18 cells was inhibited with the inhibitory concentration at 50% cell death (IC50) values for EGCG ranging from 82 to 302 µg/ml (Rahman et al., 2014) and HC (Abdul Rahman et al., 2014) with values of IC50 between 75–119 µg/ml (Table 1). The cytotoxicity induced by EGCG and HC was found to be dose dependent with 58–84% and 90–95% inhibition respectively achieved after 24 h of treatment (Figure 1).

The lower IC50 values of EGCG (50/100 μg/ml; compared to the IC50 values of EGCG obtained, ranging from 82 to 302 μg/ml (Table 1)) were titrated on a range of HC concentrations (1–100 μg/ml). Results obtained showed lower IC50 values ranging from 10 to 25 μg/ml for HC when combined with EGCG (Table 2), compared to HC treatment alone on glioma cells (Table 1). Moreover, the combination dose of EGCG and HC were observed to induce morphological changes in the glioma cells by microscopic examination (Figures 2A,B).

The suppression of cell viability was shown to be stronger in the combination of EGCG+HC at lower concentrations than either component alone (Figure 1C, Table 2). An isobologram plot was done for synergism analysis (Supplementary Figure S1) and to determine the combination index (CI). Synergistic interactions were seen in all glioma cells treated with combined sub-effective concentrations of EGCG and HC, as shown in Table 2, with CI = 0.88 for 1321N1, CI = 0.54 for SW1783, and CI = 0.43 for LN18 cells. The combined EGCG+HC treatment for 1321N1 is 50 μg/ml EGCG + 20 μg/ml HC, SW1783 is 100 μg/ml EGCG + 25 μg/ml HC and LN18 is 50 μg/ml EGCG + 10 μg/ml HC.

The synergy effect of combined EGCG+HC on 1321N1 and SW1783 amount to multiplication of caspase-3 activation of 52 and 57% respectively than by either EGCG or HC alone (Figure 3). Induction of active caspase-3 in LN18 EGCG+HC treated cells was observed to be the lowest (9.4%) compared to both 1321N1 and SW1783 (Figure 3).

Figure 4B shows that the percentage of both early (29.4%) and late (8.3%) apoptotic cells for EGCG+HC treatment in 1321N1 cells increased significantly compared to vehicle control and EGCG, but no changes were observed when compared to HC treatment alone. An increase of early (17.1%) and late (32.5%) apoptosis in LN18 treated with combined EGCG+HC was observed when compared to vehicle control and HC alone, but no changes were observed for early apoptosis when compared to EGCG treatment alone (Figure 4D). While for SW1783, a significant increase was seen in the percentage of late apoptosis (64%) for cells treated with EGCG + HC compared to vehicle control, ECGC (32%) or HC (14.6%) treatment alone (Figure 4C).

Figure 5 shows that no cytotoxicity was observed on normal cells (foreskin fibroblasts cells and WRL68) for combined EGCG+HC treatment using IC50 doses of 1321N1 (50 μg/ml EGCG + 20 μg/ml HC) and LN18 (50 μg/ml EGCG + 10 μg/ml HC) obtained from the MTS assay data. However, a significant reduction of cell proliferation was seen on normal cells using the combination dose of SW1783 (100 μg/ml EGCG + 25 μg/ml HC) (Figure 5). For this reason, the dose of 100 μg/ml EGCG + 25 μg/ml HC on SW1783 was excluded for further testing.

The proportion of cells in G0/G1 phase reduced (35% for 1321N1 and 30.4% for LN18, respectively) in EGCG+HC treated cells when compared to vehicle control, while S phase was enhanced in 1321N1 (50.4%) and LN18 (49.9%) compared to the vehicle control and EGCG alone (Figure 6). The G2M phase was marginally reduced (14.6%) compared to the vehicle control (Figure 6A).

In the wound healing experiment, 1321N1 and LN18 cells treated with EGCG+HC had lower migratory potential, with only 10.9 and 14.8% of cells migrated, respectively, as compared to the vehicle control and cells treated with EGCG/HC alone (Figure 7). Transwell migration assays revealed a similar outcome, with less 1321N1 (34.8%) and LN18 (50.7%) EGCG+HC treated cells migrated across the membranes compared to the vehicle control and EGCG/HC alone (Figure 8A). A thin coating of ECM was utilised as an impediment to non-invasive cells in vitro in the transwell invasion experiment. As indicated in Figure 8B, EGCG+HC treated cells had lower percentage of 1321N1 and LN18 cell invasion (42.1 and 52.8% respectively) compared to vehicle control and EGCG/HC alone.

EGCG+HC treatment was more effective than vehicle control and EGCG/HC singly in preventing 1321N1 and LN18 cell colony formation with 10.5 and 11.8% of colonies survived, respectively (Figure 9).

Supplementary Table S1 illustrates a total of 2103 and 2442 differentially expressed genes (FDR p ≤ 0.05 and fold change (FC) ≥ 1.5) in 1321N1 and LN18, respectively, treated with EGCG+HC compared to controls. The list of the most significantly expressed genes with the highest FC is shown in Supplementary Tables S2A,B. Approximately 52.8% (1321N1) and 61.5% (LN18) genes were downregulated in EGCG+HC treated cells compared to controls. According to the hierarchical clustering analysis, all of the control and treatment groups were well separated and grouped based on their expression similarity (Supplementary Figure S2). The red colour represented elevated genes, whereas the green colour represented downregulated genes.

Approximately 3782 (1321N1) and 4793 (LN18) transcripts were differently expressed in EGCG+HC treated cells when compared to controls (FDR p ≤ 0.05 with fold change (FC) ≥ 1.5) (Supplementary Table S3). The list of the most significantly expressed genes with the highest FC is shown in Supplementary Tables S4A,B. The results revealed that around 27% of transcripts in both cell lines were selectively expressed, with no alterations seen in genes corresponding to the transcripts expressed (Supplementary Table S5). The Partek^® Genomics Suite and Tuxedo were used to evaluate the transcripts implicated in alternative splicing events (Supplementary Table S6). The alternative splicing expression generated from Partek ∩ Tuxedo ∩ transcript (FC ≥ 1.5) is provided in the Supplementary Tables S7A,B. Alternative splicing expression of DYNLL1 (downregulated) and DDX39B (upregulated) transcript were altered in 1321N1, whereas RBMX (downregulated) and SEC31A (downregulated) transcript were altered in LN18 treated with EGCG+HC.

Table 3 (GSEA)summarised the biological pathways and the number of genes implicated (control vs treatment). The most significant biological pathway that is upregulated in EGCG+HC treated cells is the endoplasmic reticulum (ER) unfolded protein response (UPR), followed by the activation of the inflammatory response pathway. EGCG+HC downregulated pathways such as mitotic cell cycle control, telomere maintenance, and DNA repair. FET analysis (gene dataset (p < 0.05, FDR)) reveals that the apoptotic process, axon guidance and cell cycle arrest were the most substantially enriched BP in EGCG+HC treated groups (Table 4). The selective effects of EGCG+HC on various gene targets in each cell line are most likely responsible for the differences in cellular response seen between the glioma cell lines. Subnetwork studies of gene lists, for example, revealed that major regulator genes such as MYC, TGFB1, EGR1, and KLF4 were present in 1321N1 cell lines treated with EGCG+HC (p < 0.05, FDR). E2F4, MTOR, E2F1, and BRCA were among the LN18-specific central regulator genes (Table 5).

qPCR was used to validate 14 genes from lists of biological processes provided by GSEA, FET, and analysis of gene regulatory subnetworks, as well as three transcripts implicated in alternative splicing events in LN18. Additionally, 13 genes were validated which included three transcripts implicated in alternative splicing events in 1321N1. All the genes and conditions tested were parallel with RNA-seq results (Figure 10).