The A2780 human ovarian cancer cell line and the A2780/Taxol paclitaxel-resistant ovarian cancer cell line were purchased from Keygen Biotech (Jiangsu, China). The human non-small cell lung cancer cell line A549 and the paclitaxel-resistant cell line A549/Taxol were donated by Prof. Yang from Qiqihar Medical College. All the cells were cultured in RPMI-1640 cell culture medium (Solarbio Science and Technology Co., Ltd., Beijing, China) at 37°C and 5% CO₂. The culture medium was supplemented with 10% inactivated fetal bovine serum (Gibco, NY, United States) and 1% penicillin and streptomycin. The A2780/Taxol cells were cultured in medium containing 800 ng/mL paclitaxel and then switched to a non-drug medium 2 weeks before the experiment. An antibody against P-gp (0.1 μg/m; ab170904), NF-κB/p65 (1/2000, ab32536), p-NF-κB/p65 (1/1000, ab239882) and VEGF (1/1000, ab32152) were purchased from Abcam (Cambridge, United Kingdom); antibodies against PI3K (1/1000, #4292), Akt (1/1000, #9272), p-Akt (1/1000, #9271), α-Tubulin (1/2000, #2144), MMP-9 (1/1000, #3852), MRP1 (1/1000, ab233383), MRP2 (1/2000, ab172630), LRP (1/1000, #64099), lamin B (1/1000, #17416) and ß-actin (1/1000, #4970) were purchased from Cell Signaling Technology (CST, MA, United States).

Cajanol was extracted by our laboratory with purity greater than 98%. The structural formula is shown in Figure 1A. The extraction method of cajanol refers to the published extraction method with slightly modified (Liu et al., 2011). Specifically, the crushed pigeon pea root was impregnated with ethanol-water (80:20, V/V) solution for 24 h with a solid-liquid ratio of 1:10, repeated 3 times. The filtrate was concentrated using a rotary evaporator to obtain a crude extract. And the crude extract was extracted with ethyl acetate/distilled water (3/1, v/v), the ethyl acetate was separated and concentrated to obtain brown product. The brown product was separated by a resin column, water and ethanol were used as mobile phases, and the 50% ethanol eluted part was retained for further purification. Using a silica gel column with chloroform-methanol as the mobile phase, cajanol was separated from the chloroform-methanol (12:1, v/v) fraction. After crystallization and recrystallization, white crystals were obtained. The HPLC and negative ion mode mass spectra of Cajanol are shown in Supplementary Figure S1. All other reagents were purchased from Sigma (St. Louis, MO, United States).

Cells to be tested (100 µL; 5 × 10³ cells/well) were seeded in a 96-well plate and cultured overnight. Paclitaxel and cajanol were prepared as stock solutions in dimethyl sulfoxide (DMSO) and diluted (1000, 500, 250, 125, 62.5, 31.25, 15.63, 7.81, 3.91, 1.96 and 0.98 μM). Seeded cells were treated in quadruplicate with the diluted solutions for 72 h. MTT was added, and the cells were incubated for 4 h. The culture medium was then removed, and DMSO (150 µL) was added. The plate was placed in a shaker to fully dissolve the purple formazan crystals. The OD was measured at 570 nm, and the growth inhibition rate was calculated. All the experiments were repeated in triplicate. GraphPad Prism 7.0 (GraphPad Software, Inc., La Jolla, CA, United States) was used to calculate the IC₅₀ values. Fold resistance (FR) is calculated by dividing the IC₅₀ obtained in the resistant cancer cells (with or without the reversal compounds) by the IC₅₀ of the non-resistant, parental cancer cells (Hussein et al., 2017).

Medium containing 2, 4 or 8 µM cajanol was prepared and incubated with A2780/Taxol cells to determine the effect of cajanol on ABCB1, MMP-9, VEGF, Tubα1a and Tubβ3 mRNA expression. After 48 h of co-incubation, total RNA was isolated using a RNA Extraction Kit (Takara Bio Inc., Dalian, China), and total RNA reverse transcription was performed using a Reverse Transcription Kit (Takara). Real-time quantitative PCR primers were designed using Primer Premier 5.0 (Supplementary Table S1), and SYBR^® Premix Ex Taq™ II kit (Takara) was used for the RT-qPCR assay. β-actin was selected as the internal reference gene; the reaction conditions have been previously described (Huang et al., 2017). Relative gene expression was calculated by the 2^{−ΔΔ Ct} method.

Logarithmic cells (4 × 10⁵/mL density) were seeded in a 12-well plate. Positive control verapamil (8 µM) and appropriate concentrations of cajanol were added, and the cells were incubated for 2 h at 37°C. Then, rhodamine-123 at a final concentration of (1 μg/mL) was added, and the cells were incubated at 37°C, 5% CO₂ for 1 h. After the reaction was completed, the supernatant was discarded after centrifugation at 4°C, and the cells were washed twice with pre-chilled PBS buffer (4°C), terminating the rhodamine efflux. The cells were then re-suspended in cold PBS and the intracellular drug fluorescence was determined by flow cytometry (Beckman Coulter, Fullerton, CA, United States), with an excitation wavelength of 466 nm and an emission wavelength of 535 nm. The results were analyzed by Expo32ADC software (Beckman Coulter).

To determine whether cajanol affects the expression of P-gp through the PI3K/Akt signaling pathway, we measured the expression of P-gp, t-Akt, p-Akt, NF-κB/p65 and p-NF-κB/p65 in A2780/Taxol cells. The A2780/Taxol cells were treated with cajanol (2, 4, or 8 μM) for 48 h before the total cell protein was extracted. Cell lysis buffer for Western and IP (P0013, Beyotime Biotechnology, Shanghai, China) is selected as the whole cell protein extraction lysate. The nucleoprotein separation method is strictly implemented in accordance with the instructions. Specifically, the cells are scraped off with a cell scraper, and the cell pellet is collected after centrifugation. Add 200 µl of cytoplasmic protein extraction reagent A supplemented with PMSF for every 20 µl of cell pellet. Vortex for 5 s to completely suspend and disperse the cell pellet. After ice bath for 10 min, add 10 µl of cytoplasmic protein extraction reagent B and Vortex for 5 s. After the sample was ice bathed for 1 min, Vortex again for 5 s, and then placed in a 4°C centrifuge at 12,000 g for 5 min. Immediately draw the supernatant into a pre-cooled plastic tube, which is the cytoplasmic protein extracted. Equal amounts of proteins were separated by SDS-PAGE and transferred to a polyvinylidene fluoride membrane. After blocking, the membrane-bound proteins were probed with the relevant primary antibodies. The membranes were washed and then incubated with the secondary antibodies; finally antibody-bound proteins were detected using enhanced chemiluminescence reagents.

Six-week-old healthy female pure BABL/c nude mice (weighing 18–20 g) were purchased from Vital River Experimental Animal Co. Ltd. (Beijing, China). The mice were placed in a thermostatted laminar flow box and kept in a specific pathogen-free environment with a 12 h/12 h light/dark cycle. The animal experiments and use of tumor cells were approved by the Animal Care Welfare Committee of the First hospital of Qiqihar according to the ethical guidelines of the animals (scientific procedures) act 1986 amendment regulations (SI 2012/3039); Approval number: QAEC20190043.

A2780/Taxol cells were cultured in RPMI-1640 culture medium containing 10% fetal bovine serum. Logarithmic growth phase cells were collected to prepare a suspension with a concentration of living cells of 1×10⁵/mL. Live cells accounted for more than 95% of the population. The cell suspension was inoculated subcutaneously in the right armpit near the back of each nude mouse (approximately 10⁵ cells). After 1 week of inoculation, 16 mice with successful tumor formation were randomly divided into four groups, ensuring adequate numbers for statistical analysis: normal saline group, Taxol group (0.5 mM/kg), cajanol group (2 mM/kg) and paclitaxel (0.5 mM/kg) + cajanol (2 mM/kg) combined group. The agents were administered through the caudal vein on the 1st, 8th and 15th day after successful tumor implantation. The body weights and condition of the mice and growth of the transplanted tumors were observed every 3 days for 24 days. The tumor volume was calculated with the following formula: V = (length × width²)/2. When the tumor diameter was over 2 cm or the ulcer area was greater than 4 mm, the mice were euthanized by cervical dislocation after inducing a coma by inhalation of 2.5% isoflurane vapor. Remaining mice were euthanized after 24 days. The tumor tissue was collected for immunohistochemistry, fluorescence quantitative PCR and western blot analyses.

Immunohistochemistry was performed on paraformaldehyde-fixed paraffin-embedded sections. The detailed staining procedure has been described previously (Xu et al., 2017).

Statistical analyses were carried out using GraphPad Prism 7.0 (GraphPad Software, La Jolla, CA). One-way ANOVA was performed between groups. For ANOVA, the observed variance was partitioned into components according to different explanatory variables. *p < 0.05 was considered to be significant.

The paclitaxel MTT IC₅₀ values after treatment of A2780 and A2780/Taxol cells for 72 h were 1.23 ± 0.10 µM and 35.85 ± 1.23 µM, respectively (Table 1and Figure 2). The IC₅₀ values in A2780/Taxol cells for paclitaxel treatment combined with 2, 4, 8 or 16 μM cajanol were 25.67 ± 0.94 µM, 16.25 ± 0.54 µM, 6.54 ± 0.37 µM and 6.05 ± 0.33 µM, respectively. When cajanol concentration was 0 µM, 2 µM, 4 µM, 8 µM or 16 μM, the fold resistance of A2780/Taxol cells on A2780 cells were 29.15, 22.52, 14.51, 5.64 and 5.50, respectively. For A549/Taxol cells, cajanol also showed a good drug resistance reversal effect. The IC₅₀ of paclitaxel on A549 cells and A549/Taxol cells were 7.52 ± 0.46 µM and 289.34 ± 11.46 µM, respectively. After treatment with 2 µM, 4 µM, 8 µM or 16 μM cajanol, the IC₅₀ of paclitaxel for A549 cells was almost unchanged, but the IC₅₀ for A549/Taxol cells decreased to 189.43 ± 10.87 µM, 68.95 ± 4.87 µM, 27.9 ± 1.23 µM and 23.76 ± 1.12 µM, respectively. And the corresponding fold resistance is also reduced from 38.48 to 3.15. The results showed that co-treatment of cajanol with paclitaxel substantially inhibits A2780/Taxol cells and that the IC₅₀ value is decreased by more than 4-fold at a cajanol concentration of 8 µM. These results suggest that cajanol effectively reverses the resistance of A2780/Taxol cells to paclitaxel and that the effect is concentration dependent at levels up to 8 µM.

In order to study the mechanism by which cajanol restores the sensitivity to paclitaxel of A2780/Taxol cells, the expression of resistance-related genes in A2780/Taxol cells was determined after treatment with different concentrations of cajanol (2, 4 or 8 μM). The expression of ABCB1 gene was significantly down-regulated by treatment with cajanol in a concentration-dependent manner (Figure 1B). When the concentration of cajanol reached 8 μM, the expression of ABCB1 gene was about 10% of the control group. The expression levels of VEGF, MMP-9, Tubα1a and Tubβ3 did not change. Further, western blots were used to determine the protein expressions of P-gp, VEGF, MMP-9, α-Tubulin and βIII-Tubulin in A2780/Taxol cells after cajanol treatment, and the results were consistent with the results of fluorescence quantitative PCR (Figures 1C,D). The expression levels of VEGF, MMP-9, α-Tubulin and βIII-Tubulin were not significantly changed, while expression levels of P-gp were significantly reduced. These results suggest that cajanol can reverse paclitaxel sensitivity by inhibiting this MDR protein in A2780/Taxol cells.

In order to further determine whether cajanol has the same inhibitory effect on other MDR proteins, we determined the expression of MRP1, MRP2 and LRP proteins in A2780/Taxol cells treated with cajanol. Western blot results showed that the expression levels of MRP1, MRP2 and LRP were not significantly changed (Figures 1C,D). These results all indicate that cajanol specifically reduces expression of ABCB1 and P-gp.

Flow cytometry was used to assess the effect of cajanol on P-gp protein transport. Rhodamine-123 is transported by P-gp and can be detected by its emission of yellow-green fluorescence. By monitoring Rhodamine-123 accumulation in cells, we evaluated the transport capacity of P-gp protein in A2780/Taxol cells treated with cajanol at different concentrations (Figure 3). The intracellular fluorescence intensity of control cells A2780 (24.5%) was found to be significantly higher than that of untreated A2780/Taxol cells (3.7%). After treatment with 2, 4 or 8 μM cajanol, the fluorescence intensity in A2780/Taxol cells increased to 14.8, 18.5 and 21.3% respectively. The P-gp inhibitor verapamil also increased the fluorescence intensity of A2780/Taxol cells to 23.8% at 8 μM. This result further confirms the utility of cajanol in reversing drug resistance by inhibiting the expression of P-gp, thus reducing the efflux of paclitaxel.

The above results show that the expression of ABCB1 and P-gp in A2780/Taxol cells is inhibited by cajanol, suggesting that cajanol could inhibit P-gp transcription and/or translation. Since Akt and NF-κB signal transduction are highly correlated with P-gp expression, it is suggested that cajanol may regulate P-gp through the Akt/NF-κB/P-gp signaling pathway. The expression of each of PI3K, Akt and phosphorylated Akt in A2780/Taxol cells was detected by western blot. Notably, cajanol reduced PI3k and p-Akt in A2780/Taxol cells, but did not significantly affect the expression of Akt itself (Figures 4A,B). Further, we examined the effect of cajanol on the expression of NF-κB/p65 and phosphorylated NF-κB/p65 in cells and in the nucleus. Similar to Akt and P-Akt, down-regulation of p-NF-κB/p65 was observed in 2780/Taxol cells treated with cajanol. The expression levels of NF-κB/p65 and p-NF-κB/p65 in the nucleus were down-regulated with the increase of cajanol concentration. To verify the effect of inhibiting the PI3K/Akt/NF-κB pathway on P-gp protein expression, the PI3K inhibitor LY294002 was used to treat A2780/Taxol cells. Treatment with LY294002 had effects on the expression of Akt, p-Akt8, NF-κB/p65, p-NF-κB/p65 and P-gp similar to those found after exposure to 8 μM cajanol. These results fully demonstrate that cajanol can inhibit PI3K and phosphorylation of Akt in A2780/Taxol cells, thereby preventing the phosphorylation and nuclear translocation of NF-κB/p65, and ultimately inhibiting the transcription and translation of P-gp.

To verify whether cajanol can restore Taxol sensitivity in mice, we established a BABL/c nude mouse tumor model using A2780/Taxol cells with the following four groups: control, cajanol, paclitaxel and cajanol + paclitaxel, each group consisted of 4 nude mice that had been successfully modeled. After treatment for 24 days, the tumor volumes of the mice in the combined group were significantly smaller than those of the other three groups (Figures 5A–C); the final tumor volume of the cajanol + paclitaxel group was 182.4 ± 20.4 mm³, while tumors in the paclitaxel, cajanol and control groups had volumes of 758 ± 154 mm³, 680 ± 176 mm³ and 981 ± 215 respectively. After treatment, only the mice in the combined treatment group could maintain a body weight of approximately 25 g, while the mice in the other three groups weighed less than 20 g (Figure 5D). These results show that cajanol (2 mM/kg) combined with paclitaxel (0.5 mM/kg) in vivo has a significant inhibitory effect on paclitaxel-resistant cancer.

We also measured the expressions of ABCB1 mRNA and of P-gp protein in the tumor tissues. Western blots showed that cajanol inhibits the expression of P-gp in tumor tissues, similar to the results in vitro (Figure 5E). The immunohistochemistry results confirm that expression of P-gp in the cajanol treatment group was significantly inhibited (Figure 5F).