A Novel Resveratrol-Arsenic Trioxide Combination Treatment Synergistically Induces Apoptosis of Adriamycin-Selected Drug-Resistant Leukemia K562 Cells

Leukemia cells can develop resistance to apoptosis induced by chemotherapeutic agents. Concomitant multidrug resistance of cells remains the greatest clinical obstacle in the effective treatment of blood and solid tumors. Natural products have been identified that possess the capacity to modulate chemotherapeutic resistance and induce apopotosis. In this study, we generated adriamycin-resistant K562 leukemia (K562/RA) cells and compared the responses of sensitive and resistant leukemia cells to the natural products arsenic trioxide (ATO) and resveratrol (Rsv), with a view to determining whether Rsv potentiates the sensitivity of drug-resistant cells to ATO-induced apoptosis and the associated molecular mechanisms. Our results showed that resistance of K562/RA cells induced by adriamycin treatment was significantly higher (115.81-fold) than that of parental K562 cells. Simultaneously, K562/RA cells were cross-resistant to multiple agents, with the exception of ATO. Rsv enhanced the sensitivity of K562/RA cells to ATO and reduced the required dose of ATO as well as associated adverse reactions by promoting the proliferation inhibitory and apoptosis-inducing effects of ATO, which may be associated with reduced expression of the drug resistance genes mdr1/P-gp, mrp1/MRP1 and bcrp/BCRP, as well as the apoptotic inhibitory genes bcl-2, NF-κB and P53, and conversely, activation of caspase-3. Our collective findings indicate that ATO and Rsv synergistically enhance the sensitivity of drug-resistant leukemia cells to apoptosis.


Introduction
Long-term or repeated use of specific chemotherapy drugs can induce multidrug resistance of leukemia cells, even at therapeutic doses [1][2][3]. Reduced apoptosis-inducing activity of chemotherapy drugs is an important mechanism contributing to the development of multidrug resistance (MDR) in leukemia cells [4][5][6]. Identification of novel drugs that have no cross-resistance with conventional chemotherapy drugs and development of agents that enhance the apoptosis-inducing effect of currently available drugs are potentially effective measures to overcome MDR of leukemia. Arsenic Ivyspring International Publisher trioxide (ATO) has good efficacy for many leukemia and solid tumor types with fewer side-effects than other agents [7][8][9][10][11]. In acute promyelocytic leukemia patients who achieve remission after retinoic acid treatment, combination therapy with ATO has been shown to consolidate the efficacy of retinoic acid. ATO is thus considered the 'gold partner' of retinoic acid [12,13]. Although results obtained with ATO provide an optimistic perspective in the management of leukemia, the compound exerts cytotoxicity, which causes damage to human organs. The majority of patients treated with ATO (~75%) experience grade 1 or 2 liver dysfunction [12][13][14]. In addition, a number of patients in remission after treatment with a combination of retinoic acid and ATO experience relapse. These patients often carry the missense mutations A216V, L218P, A216T, S214L, L217F and S220G in the promyelocytic leukemia protein-retinoic acid receptor α (PML/RARα) domain, which induce arsenic resistance, and consequently, a significant increase in mortality [12][13][14]. In vitro studies have confirmed a marked apoptosis-inducing effect of ATO on leukemia cells, especially those that are drug-resistant. However, divalent, trivalent and pentavalent arsenic intermediates have been detected in saliva, urine and blood of ATO-treated patients in clinical trials. The toxic effects of accumulated arsenic on the human liver, kidney and cardiovascular system cannot be overlooked [14][15][16].
Resveratrol (Rsv) is a natural antioxidant that not only has anti-viral and immune modulation properties but also prevents mutation and tumor development [17,19]. Rsv is reported to significantly reduce the malondialdehyde and NO contents in rat serum, liver, spleen, lung and brain induced by ATO, suggestive of effective antiperoxidase activity [20,21]. Rsv has been shown to enhance apoptosis of HeLa, MCF-7 and NB40 cells induced by ATO, both in vitro and in vivo. Moreover, Rsv protects normal cells against ATO toxicity by promoting apoptosis in tumor cells, although the precise mechanisms are currently unclear [22]. In this study, we generated drug resistance in K562 leukemia cells using adriamycin (ADM) and compared the responses of sensitive and resistant cell groups to ATO and Rsv. Additionally, the effects of combined treatment with ATO and Rsv on drug-resistant leukemia cells and the associated molecular mechanisms were investigated.

Cell culture and incubation
Human leukemia K562 cells were purchased from American Type Culture Collection (ATCC, Manassas, USA). Cells were maintained in RPMI 1640 medium (Gibco; Thermo Fisher Scientific, Inc. Waltham, MA, USA) supplemented with 10% fetal calf serum (Hyclone; GE Healthcare Life Sciences, Logan, UT, USA) and cultivated at 37°C in a 5% CO2 incubator.

Generation of ADM-resistant cells
ADM resistance in human leukemia K562 cells was induced by long-term exposure to continuous stepwise increments of ADM. Cells were cryopreserved and recovered after 2 to 3 months of induction to prevent mutations until they were stably able to tolerate 16 μM ADM. This procedure was used to generate a drug-resistant leukemia subline, designated K562/RA.

In vitro drug sensitivity analysis
A total of 0.8×10 5 target cells/ml were plated in 96-well plates and cultured at 37°C with the specified concentrations of chemotherapeutic agents for 24-72 h. Absorbance was quantified using a Powerwave X plate reader (Omega Bio-Tek, Inc., Norcross, GA, USA). The MTT assay (in which each well was incubated with 5 mg/ml MTT for 4 h at 37°C, followed by overnight incubation with 100 µl 10% SDS at 37°C) was employed for determination of the half-maximal inhibitory concentration (IC50) values for ADM and cytotoxicity assays.

Morphological features of apoptotic cells
Target cells were collected and stained with Giemsa-Wright stain (Solarbio, Beijing, China) according to the manufacturer's instructions. Morphologic changes were analyzed under an AX80 optical microscope (Olympus, Tokyo, Japan).

Statistical analysis
All data are presented as means ± standard deviation. Multiple comparisons among the three groups were performed using one-way analysis of variance (ANOVA). Intragroup comparisons were conducted using the paired Student's t test. Error bars represent standard deviation, and P<0.05 indicates statistically significant differences between groups.

Drug resistance and cross-resistance of K562
During the generation of ADM-induced resistance in K562 cells, tolerance of cells to ADM increased gradually with increasing concentrations and treatment durations. When cells could tolerate 16 µM ADM, IC 50 values at 48 and 72 h for resistant cells were 84.2 and 94.2 times higher than that for parental cells, respectively ( Figure 1A). Notably, these cells additionally developed resistance to other chemotherapeutic drugs including pirarubicin, daunorubicin, 5-FU, etoposide, vincristine and paclitaxel ( Figure 1B), indicating that long-term and repeated stimulation with ADM induces acquisition of MDR in K562 cells. Interestingly, however, no significant cell resistance to ATO was evident ( Figure  1C). K562 cells that tolerated 16 µM ADM were designated K562/RA.

Expression of drug-resistant genes and proteins in K562 cells during ADM induction of resistance
Initially, we observed extremely low expression of mdr1/P-gp, mrp1/MRP and bcrp/ BCRP in K562 cells. During the development of ADM-induced resistance of K562 cells, expression of mdr1, mrp1 and bcrp genes increased with the tolerable ADM concentration (Figure 2A). In K562 cells able to tolerate 16 µM ADM, expression of mdr1, mrp1 and brcp was 132.5-fold, 2.85-fold and 8.87-fold higher than that in parental cells, respectively. Flow cytometry (FCM) additionally showed an increase in the positive rate (PR) of P-gp protein expression and mean fluorescence indensity (MFI) with higher tolerable ADM concentrations. At a tolerable ADM concentration of 16 µM, nearly 100% cells displayed positive expression of P-gp ( Figure 2B). Western blot data confirmed increased expression of P-gp, BCRP and MRP1 proteins in K562/RA cells. Notably, P-gp and MRP1 were almost undetectable in parental K562 but significantly upregulated in K562/RA cells ( Figure 2C).

Rsv enhanced the proliferation inhibitory activity of ATO on K562/RA cells
After 24-72 h treatment with 0.5-8 μM ATO, proliferation rates of both K562/RA and parental K562 cells were significantly inhibited. The inhibitory rate was positively correlated with ATO concentration and treatment duration. Overall sensitivities of K562/RA and K562 cells to ATO were not significantly different ( Figure 3A). After 24-72 h treatment with 10-80 μM Rsv, proliferation activity of K562/RA and K562 cells was also inhibited, with no significant differences in tolerance of the two cell types to Rsv ( Figure 3B). Upon a combination treatment of 2 μM ATO and 20 or 40 μM Rsv (2 μM ATO + 20 μM Rsv and 2 μM ATO + 40 μM Rsv), the inhibitory rates in K562/RA cells were 67.7% and 72.1% after 24 h, 76.7% and 88.6% after 48 h and 93.4% and 99.1% at 72 h, respectively, compared to 28.8% and 39.4% after 24 h, 43% and 59.3% after 48 h and 65.2% and 77.7% at 72 h, respectively, in K562 cells. These data demonstrate that the efficacy of combination therapy with ATO and Rsv is significantly higher in K562/RA than parental K562 cells ( Figure 3C), which suggests that Rsv enhances the suppressive effect of ATO in resistant cells.

Morphological changes of apoptosis
We observed no significant morphological changes in K562 or K562/RA cells following treatment with 20 μM Rsv. After 24-h treatment with 2 μM ATO or 2 μM ATO plus 20 μM Rsv, morphological changes characteristic of apoptosis, such as chromatin condensation and cell shrinkage, were observed, which were more pronounced in cells treated with the drug combination ( Figure 4A).

Caspase-3 activation
FCM analysis showed that after 24-h treatment with 2 μM ATO alone, the positive rate of caspase-3 activation in K562 and K562/RA groups increased from 2.13% and 2.26% in untreated control cells to 8.27% and 8.40%, respectively. After treatment with 40 μM Rsv alone, the positive rate increased to 22.19% and 16.78% while combined treatment with ATO and Rsv induced an increase in the rate of caspase-3 activation to 48.7% and 81.93%, respectively. Our results clearly indicate that both ATO and Rsv enhance caspase-3 activity in resistant and sensitive K562 cell types. Moreover, treatment with the ATO-Rsv combination promoted caspase-3 activity more significantly than either ATO or Rsv alone. This effect was more prominent in drug-resistant K562/RA cells ( Figure 4B). The ATO and Rsv combination induced a significant increase in cleaved caspase-3 expression, particularly in K562/RA cells. These findings suggest that Rsv significantly enhances the apoptosis-inducing effect of ATO in drug-resistant cells ( Figure 4C).

BAX/BCL-2 expression
Treatment with 2 μM ATO or 20 or 40 μM Rsv alone led to significant upregulation of BAX protein in K562 and K562/RA cells. Concomitantly, expression of BCL-2 was weakly downregulated. The BAX/BCL-2 ratios in K562 cells treated with 2 μM ATO, 20 μM Rsv and 40 μM Rsv were 2.67-, 1.39-and 1.32-fold higher than that of the control group, respectively ( Figure 5A-D). After combination treatment with 2 μM ATO and 20 μM or 40 μM Rsv, the BAX/BCL-2 ratio was 2.42-and 7.7-fold higher than that of the control group for the K562 cell line. For the K562/RA cell line, the ratio was 20.27-and 97.33-fold higher in the treatment groups, respectively ( Figure 5E). Our results suggest that Rsv enhances the apoptosis-inducing effect of ATO via concomitantly upregulating BAX and downregulating BCL-2. This synergistic apoptosis-promoting activity is more pronounced in drug-resistant K562/RA cells.

P53 expression
Treatment with 2 µM ATO, 20 µM or 40 µM Rsv alone had no significant effects on P53 levels. However, expression of P53 was markedly decreased in K562 and K562/RA cells after combined treatment with ATO and Rsv ( Figure 6A-D). In particular, co-treatment with 2 µM ATO and 20 µM or 40 µM Rsv induced a more significant decrease in P53 expression in K562/RA cells ( Figure 6A-D).

NF-κB expression
Activated NF-κB inhibits apoptosis. Treatment with 2 µM ATO led to significant downregulation of NF-κB expression in K562/RA but exerted a weak effect on K562 cells ( Figure 6A and B). Expression of NF-κB was not affected upon treatment with 20 or 40 µM Rsv. However, 2 µM ATO combined with 20 or 40 µM Rsv induced downregulation of NF-κB in both K562 and K562/RA cell lines. In K562 cells, the combination treatment suppressed NF-κB expression to 63.03% and 20.61% relative to the control group, while in K562/RA cells, expression decreased to only 4.02% and 1.75% that of the control group ( Figure 6C and D), suggesting that ATO and Rsv act synergistically to induce apoptosis in cells by inhibiting the NF-κB signaling pathway, particularly in drug-resistant leukemia cells.

Rsv enhances the inhibitory effect of ATO on expression of resistance-related genes in K562/RA cells
Expression of mdr1, bcrp and mrp1 was high in ADM-induced resistant K562/RA cells but nearly undetectable in parental K562 cells (Figure 2). ATO (2 μM) significantly inhibited expression of these genes in K562/RA cells (21%, 33% and 36%, compared to the control group, respectively). Rsv alone (20 μM or 40 μM) had no or a weak inhibitory effect on the genes responsible for drug resistance, while combined treatment with 2 μM ATO and 20 or 40 μM Rsv synergistically downregulated the expression of drug-resistant genes in K562/RA cells. Levels of mdr1 were only 15% and 5% while those of bcrp were 30% and 25% relative to the control group, respectively, and 0.4% for mrp1, as detected with real time RT-PCR ( Figure 7A and B). Western blot results were in line with RT-PCR findings ( Figure 7C and D), further confirming that ATO combined with Rsv exerts a stronger apoptotic effect on K562/RA cells. The underlying mechanisms may be associated with downregulation of resistance-related genes through synergistic effects of ATO and Rsv.

Discussion
ATO has been widely used in the management of leukemia and shown to be effective in treating a variety of solid tumors, including lung, liver, ovarian, colon and esophageal cancer types [4,[7][8][9][10][11][12]. The mechanisms underlying the anti-tumor activity of ATO are complex. ATO can induce tumor cell differentiation, promote degradation of PML/RARα fusion protein, block specific intracellular signaling pathways, and trigger apoptosis by activating the intracellular caspase cascade, downregulating Bcl-2 expression and inhibiting NF-kB. The apoptosis-inducing effect of ATO on leukemia and cancer cells represents the key mechanism that contributes to its anti-tumor activity [4,[7][8][9][10][11][12]24]. In 1998, Rsv was initially reported to induce apoptosis of human leukemia HL-60 cells via the Fas-FasL pathway with no apparent side-effects on peripheral lymphocytes [25,26]. Subsequent studies confirmed that Rsv exerts apoptotic effects on a variety of tumor cells, including T47D human breast cancer, U87 glioma, K562 leukemia, lymphoma and prostate cancer cells, with no significant side-effects on normal cells [22,[25][26][27][28][29]. Notably, the compound has a strong killing effect on cancer stem cells. Additionally, previous experiments by our group have confirmed that Rsv induces apoptosis in leukemia K562 and stem cells.  Recent studies suggest that Rsv not only has a synergistic inhibitory effect with ATO on tumor cells, but also reduces the toxicity of ATO alone. Here, we combined a lower concentration of ATO with Rsv to determine the proliferation inhibitory and apoptosis-inducing effects on resistant K562/RAcells. ATO, in combination with Rsv, exerted a synergistic inhibitory effect on proliferation of both K562/RA and parental K562 cells, which was more pronounced in the resistant cell line. The cells underwent morphological changes typical of apoptosis, and the apoptotic rate was significantly increased. Although caspase-3 expression was not affected to a marked extent, its activity was remarkably enhanced, particularly in K562/RA cells, indicating that Rsv enhances the proliferation inhibitory and apoptotic effects of ATO on K562/RA cells.
Interactions between BCL-2 on the mitochondrial membrane and BAX in the cytoplasm affect the stability of the mitochondrial membrane and determine cell apoptosis [28]. ATO combined with Rsv suppressed the expression of BCL-2 and P53 and increased the expression of BAX, thereby increasing the BAX/BCL-2 ratio. In drug-resistant K562/RA cells, Rsv and ATO exerted a stronger combined effect on downregulation of P53, which could trigger changes in the expression of downstream genes, such as BCL-2 and Bax. Accordingly, the permeability of the mitochondrial membrane was increased and CytC released, further activating caspase cascade reactions, leading to cellular apoptosis. NF-κB is a transcription factor and the associated signaling pathway is inactive in normal cells but continuously activated in tumor cells. NF-κB activation leads to upregulation of the anti-apoptotic protein, BCL-2, which downregulates IκBα expression to further enhance activation of NF-κB via a feedback mechanism. The transcription factor plays a critical role in the proliferation and apoptosis of tumor cells. We propose that ATO and Rsv synergistically inhibit the expression and activation of NF-κB in K562/RA cells, thus suppressing its promotory effect on BCL-2 expression to induce cell apoptosis.
The membrane transporters P-gp, MRP1 and BCRP protect cells against injury induced by chemotherapeutic drugs and other exotoxins. However, these proteins also mediate the development of multidrug resistance in tumor cells.
Meanwhile, evidence has shown that P-gp participates in the regulation of cellular apoptosis. In our experiments, the levels of mdr1, bcrp and mrp1 genes in K562/RA cells were significantly higher than those of parental K562 cells. Rsv did not affect expression of mdr1, bcrp and mrp1 in K562/RA cells but significantly enhanced the inhibitory effects of ATO on gene expression. These results suggest that Rsv and ATO synergistically enhance apoptosis of K562/RA cells, which may be associated with inhibition of mdr1, bcrp and mrp1. Additional studies [30] have shown that the promoter of mdr1 contains the binding site of NF-κB. Binding of NF-κB activates the transcription of mdr1 and upregulates mdr1/P-gp. In tumor cells, mutated P53 can also activate the mdr1 promoter and upregulate the gene at the transcriptional level. Moreover, overexpression of BCL-2 promotes mdr1 expression. The collective findings support a complex regulatory network constituting apoptosis-related and drug-resistant genes that act in concert to modulate drug resistance and apoptosis of resistant leukemia cells.

Conclusions
Rsv enhances the apoptosis-inducing effect of ATO on both drug-resistant K562/RA and parental K562 cells and significantly suppresses expression of the resistance-related genes mdr1/P-gp, mrp1/MRP1 and bcrp/BCRP. Moreover, apoptosis induced by Rsv and ATO is associated with the Bcl-2/Bax and NF-κB pathways. Further research is warranted to elucidate the detailed mechanisms underlying the synergistic apoptotic effects of these drugs.