Discovery of potential targets of Triptolide through inverse docking in ovarian cancer cells

Drug preparation

TPL (purity 99.0%, Beijing beihua hengxin biotechnology Co., Ltd., Beijing, China) and cisplatinum (P4394, Sigma-Aldrich Co. LLC, St. Louis, MO, USA) were dissolved in dimethyl sulfoxide (DMSO) to a stock concentration of 500 nM. The solution was stored at −20 °C and diluted in a cell culture medium.

Cell culture

Human ovarian cancer cell line SKOV-3 (ATCC: HTB-77) were cultured on cell plates at 37 °C, in 5% CO₂, in RPMI-1640 (HyClone) supplemented with 10% fetal bovine serum, 100 units/ml penicillin and 0.1 mg/ml streptomycin.

MTT assay

SKOV-3 cells were plated in the 96-well plates (5 × 10³/well). RPMI-1640 culture medium was used as the blank control group. A final concentration of 0, 3, 6, 12, 24, 48, 96,192 and 384 nM of TPL and 0, 1.25, 2.5, 5, 10, 20, 40, 80 and 160 μM of cisplatinum (used as the positive control) were added to the wells for 24, 48 and 72 h. Five duplicates were set for each concentration. MTT (20 μl, 5 mg/ml, Sigma, St. Louis, MO, USA) was added to each well at 37 °C for 4 h. Then 150 μL of DMSO was added to each well after the liquid supernatant was removed. The enzyme-linked immunosorbent assay reader (TECAN, SPARK 10M; Mannedorf, Switzerland) was used to obtain the results at 570 nm. The IC₅₀ values of TPL and cisplatinum were calculated by SPSS Software (IBM Inc., New York, USA, version 21.0).

Apoptosis assay

After harvested and washed twice SKOV-3 cells were stained with 2 μM JC-1 for 15 min at 37 °C. A flow cytometer using 480 nm excitation and 585 nm emission filters was applied to analyze the stained wells. Cell suspension mixed with 10 μM Carbonyl cyanide m-chlorophenyl hydrazine (CCCP) was set as the control. The cell mixture was incubated for 15 min at 25 °C in the dark followed by fluorescence-activated cell sorting (FACS) cater-plus flow cytometry (FACSAria III, BD CellQuest pro).

Inverse docking

Classical biological methods to find the antitumor targets of natural products usually consume manpower and material resources. New approaches such as inverse docking are developed to efficiently fish the potential targets of natural products like resveratrol (Kores et al., 2019), cassiar in alkaloids (Negi et al., 2018) and curcumin (Furlan, Konc & Bren, 2018). To identify the potential targets of TPL, we developed an inverse docking protocol (Bren, Fuchs & Oostenbrink, 2014) by using a GOLD program (Genetic Optimization for Ligand Docking, version 3.5) (Jones et al., 1997) in combination with Discovery Studio (DS, BIOVIA^TM, Rueil-Malmaison, France, version 4.0).

TPL was subjected to DS as the template ligand. The target library was built based on 2,250 protein/ligand complexes extracted from PDB (supplied by BIOVIA^TM, updated to 2015). Each protein was standardized with DS Prepare Protein protocol where missed loops were inserted, protonation was performed for pH of 7.4 with the “Protein Ionization” method, and CHARMm forcefield was used for calculations while native ligands, as well as water, were deleted. Then, the pockets where the native ligands placed were defined as the binding sites. The radius of the docking-grid sphere was set to 1.5 nm to accommodate the entire TPL molecule as long as at least one of its atoms is placed less than 0.6 nm away from the binding sites. The inverse docking protocol applied semi-flexible docking based on a genetic algorithm to save the CPU time and acquire accurate docking results (Xue et al., 2013). The algorithm sets the ligand and binding sites flexible to generate 255 poses while the other region of the protein rigid. The inverse docking search protocol could successfully dock TPL into a candidate protein before all possible positions and orientations of 255 poses were exhausted. At most five poses were stored from every docking simulation, except if the best three docking poses had root-mean-square deviations (RMSD) smaller than 0.15 nm. The X ray_Score calculated by GoldScore and Ludi_score was used to score and rank the poses of inverse docking (see in Supplemental File).

2D gel electrophoresis

SKOV-3 cells treated with TPL were washed twice with ice-cold PBS and resuspended in lysis buffer for 30 min to collect the cell suspension. After incubated for 15 min at 25 °C, samples were mixed with N. N-dimethylacrylamide for alkylation and DTT for centrifugation. Protein samples were separated by 2D gel electrophoresis using the first dimension, isoelectric focusing (IEF) and the second dimension, non-equilibrium pH gradient electrophoresis (NEPHGE). The first dimension was performed with a nonlinear pH 3–10 range at 25 °C in IEF buffer (0.5% Pharmalyte 3–10 NL, 2% (w/v) ASB, 15 mM DTT, 2 M thiourea, 6 M urea and bromophenol blue) containing 0.16 mg labeled proteins. After IEF, the immobiline strip with separated proteins was equilibrated for 15 min using equilibration buffer (Thermo Fisher Scientific, Waltham, MA, USA). Separated protein samples were then further separated on 4–12% polyacrylamide gel. After labeling with Cy3- or Cy5- fluorescent dye, gels were scanned on a 2920 2D-Master Imager (Bio-Rad, Hercules, CA, USA) according to manufacturer’s instruction. Protein spots were migrated and detected automatically, then quantified by calculating the fluorescence intensities using DeCyder Differential In-Gel Analysis Software (Bio-Rad, Hercules, CA, USA).

Peptide mass fingerprinting

Protein spots excised from the gel were destained with trypsin (Sigma Aldrich, St. Louis, MO, USA) then incubated in MilliQ-H₂O for 10 min to cut into small pieces. Peptides were extracted twice from supernatants of the disposed of gel pieces with 50% (v/v) acetonitrile and 0.1% (v/v) trifluoroacetic acid. After pooled and dried, extracts were analyzed using a TofSpec-2E mass spectrometer (Micromass, Manchester, UK). Peptide mass fingerprinting maps were acquired and screened in the Swiss-Prot database performed by the Mascot Software.

Western blotting

Cells were washed twice and lysed in ice-cold RIPA lysis buffer for 30 min to collect the cell suspension. After separated with SDS-PAGE, cell lysate aliquots (40 μg) were blotted onto a polyvinylidene difluoride membrane and incubated with primary antibody, rabbit polyclonal anti-Annexin A5 (diluted 1:300; Abcam), anti-ATP synthase (diluted 1:300; Abcam), anti-β-tubulin (diluted 1:300; Abcam) and anti-HSP90 (diluted 1:300; Abcam) overnight at 4 °C. Membranes were washed and incubated with goat anti-rabbit IgG horseradish-peroxidase (HRP) pre-adsorption secondary antibody (1:2,000; ab7090, Abcam, Cambridge, UK). Bands were visualized by an enhanced chemiluminescence system (ECL, Abcam, Cambridge, UK), then quantified using the manufacturer protocol (Bio-Rad, Hercules, CA, USA). The PowerPacTM HC electrophoresis instrument and film transfer instrument (Bio-Rad, Hercules, CA, USA) was used in this section.

RNA isolation and quantitative real-time PCR

Total RNA was extracted from SKOV-3 cells treated with TPL (0, 10, 20, 40 nM for 48 h) using the RNA extraction kit (Invitrogen, Life Technologies, Carlsbad, CA, USA). cDNA was synthesized from 2 μg of total RNA using Superscript reverse transcriptase (Life Technologies, Carlsbad, CA, USA). The primers used for quantitative real-time PCR (qRT-PCR) were as follows: for HSP90, sense primer (5′-TTAAGGTACTACACATCTGCCTCT-3′) and antisense primer (5′-TGCTTTCGGAGACGTTCCACAA-3′); for Annexin A5, sense primer (5′-CAGTCTAGGTGCAGCTGCCG-3′) and antisense primer (5′-GGTGAAGCAGGACCAGACTGT-3′); for ATP synthase, sense primer (5′-TCTTTGCTGGTGTTGGTGAA-3′) and antisense primer (5′-TGAGCTCATCCATACCCAAA-3′); for β-tubulin, sense primer (5′-TGCATTGACAACGAGGC-3′) and antisense primer (5′-CTGTCTTGACATTGTTG-3′). The housekeeping gene glyceraldehyde phosphate dehydrogenase (GAPDH) was used for normalization (sense: 5′-TGATGACATCAAGAAGGTGGTGAAG-3′; and antisense: 5′-TCCTTGGAGGCCATGTG GGCCAT-3′). PCR products were analyzed by 1.2% agarose gel electrophoresis with ethidium bromide for UV light transilluminator visualization using LightCycler^® 96 qRT-PCR instrument (Roche Life Science, Penzberg, Germany).

Statistical analyses

Statistical analysis was performed by using one-way analysis of variance, followed by Dunnett’s Multiple Comparison T-test using SPSS Software (IBM Inc., New York, USA, version 21.0) when appropriate. All data were presented as mean ± standard deviation (SD). The difference was considered statistically significant at *p < 0.05 and **p < 0.01.

Evaluation of the anti-cancer effect of TPL on SKOV-3 cell line

The efficacy of TPL on ovarian cancer and its target genes have not been deeply studied. To test the ability of TPL to inhibit the survival of ovarian cancer cell in vitro, cell viability assay using the SKOV-3 cell line was performed. Compared to cisplatinum, TPL exhibits a better potency with IC₅₀ values of the nanomolar level. Significant suppression of proliferation in SKOV-3 cells is observed after treatment with TPL at the concentrations of 3, 6, 12, 24, 48, 96,192 and 384 nM respectively (Fig. 1). When the TPL dose increased, a gradual decrease in cell viability is observed either cells are treated for 24, 48 or 72 h. At the concentration of 3 and 6 nM, little significant time dependance of cell treatment for 24, 48 or 72 h was observed. Along with the increasing concentration, the time dependance becomes clear among 24, 48 and 72 h. It can also be seen in Table 1, TPL exhibits the IC₅₀ value of 70.3 ± 1.17 nM and 31.7 ± 1.23 nM on 24 and 72 h. Meanwhile, cisplatinum exhibits larger IC₅₀ value of 33.0 ± 1.77 μM and 9.6 ± 0.74 μM on 24 and 72 h. All the data show that TPL inhibits ovarian cancer cell proliferation more potently than cisplatinum in a dose-dependent manner in SKOV-3 cell lines.

Apoptotic changes in TPL-treated ovarian cancer cell

As the increased doses of TPL, we also detected a simultaneous increase in both LR fraction (early apoptosis) and LL fraction (regarded as necrotic) subpopulations. As is shown in Fig. 2A, the degree of apoptosis is quantitatively expressed as the control (DMSO) presented 13.20% population after 24 h treatment. The apoptosis rate of SKOV-3 cell rises after the treatment of TPL. Both the low and high concentrations of TPL significantly promote the apoptosis of SKOV-3 cells with the early apoptosis rate of 26.25% and 40.73% (Fig. 2B). In contrast to low concentration, a high concentration of TPL slightly enhances the apoptosis of SKOV-3 cells and is higher than those treated with 10 μM CCCP. A dose-dependent increase of early apoptosis can also be seen from the results after TPL treatment of SKOV-3 cells. The results of the apoptosis assay indicate that TPL inhibits SKOV-3 cells by inhibiting certain targets rather than by killing cytotoxic SKOV-3 cells.

Inverse docking

Interestingly, TPL is more active than cisplatinum. To find potential target genes of TPL, an inverse docking protocol in silico was designed based on GOLD. With ranked X ray_Score values calculated by the GOLD scoring protocol, the top 19 poses with the X ray_Score values larger than 150 were selected (see in Supplemental Files). Among them, seven proteins (5lnz, 4bqg, 2qf6, 2xab, 2brc, 2xab, 2yki) belong to HSP90, six proteins (5j2t, 5iyz, 4o2a, 4drx, 4o2b, 4tuy) belonged to β-Tubulin, three proteins (2xo2, 2h0m, 1yii) belong to Annexin A5 and three proteins (5dn6, 3m4y, 2wpd) belong to ATP synthase respectively (Fig. 3A). HSP90 protein has the highest docking scores compared with the other three proteins. As in Fig. 4B, two hydrogen bonds between TPL and HSP90, Ser52 and Gly135 ensure the correct binding pose of TPL, as well as hydrophobic interactions, stabilize the binding mode. However, the amount of HSP90 proteins are not significantly different from other proteins. Therefore, further biological experiments are necessary to identify the potential targets for TPL.

Differential gene expressions in TPL-treated SKOV-3 cell

To further determine the main targets of TPL on ovarian cancer cells, a 2D-DIGE assay was utilized to examine global changes in gene expression in SKOV-3 cells after treatment with 10nM TPL for 48 h. Total proteins from SKOV-3 cell were minimally labeled with two different fluorescent dyes, mixed and separated by isoelectric focusing on a 2D gel on SDS-PAGE. The identical proteins migrating to the same 2D spots were detected and quantified based on their fluorescence intensities to compare their expression by the DeCyder program.

The difference of spots’ density between normal and TPL-treated SKOV-3 cells discloses the changes of expression in eighteen protein spots (Fig. 4). Among 18 protein spots, the density between TPL and control is larger than two in eight spots. They are chosen to acquire finger printers of the peptide mass via MS/MS spectrum. According to the database searching results, the eight proteins are verified as Annexin A5, Tropomyosin (alpha-3 chain), ATP synthase (subunit beta), POTE ankyrin (domain family member E), Rho (GDP-dissociation inhibitor), Actin (cytoplasmic), Tubulin (β-4B chain) and Heat shock protein 90-alpha (HSP90). The representative genes found to be up- and down-regulated by 10 nM of TPL treatment are summarized in Table 2, respectively. Their known or proposed functions are analyzed by Gene Ontology. These eight proteins are mostly involved in maintaining the cellular structure, signal transduction and transport.

Effects of TPL on the expression of novel proteins

To further elucidate the mechanism of TPL in ovarian cancer and define the main target gene, western blotting (WB) and qRT-PCR analysis were performed on the four proteins to detect their differential expression. The results (shown in Fig. 5 and Table 3) reveal that TPL markedly up-regulates the levels of Annexin A5 and ATP synthase proteins as well as down-regulates the levels of β-Tubulin and HSP90 in ovarian cancer cell lines. WB results show that the expression of Annexin A5 and ATP synthase increases in a dose-dependent manner after TPL treatment for 24 h. According to qRT-PCR assay results, TPL stimulates the expression of Annexin A5 at 40 nM, especially more dramatically than ATP synthase (6.34 ± 0.07 fold vs. 4.08 ± 0.08 fold). On the other hand, TPL inhibited β-Tubulin and HSP90 significantly more than 50% at the concentration of 20 and 40 nM. When cells are treated with 40 nM TPL. The relative expression of β-Tubulin mRNA decreases from 1 to 0.11 ± 0.12 fold.