Effects of chemotherapeutic drugs on the antioxidant capacity of human erythroleukemia cells with MDR phenotype

In this work, we identified that different chemotherapeutic drugs may select cells with different antioxidant capacities. For this, we evaluated the sensitivity of two multidrug-resistant (MDR) erythroleukemia cell lines: Lucena (resistant to vincristine, VCR) and FEPS (resistant to daunorubicin, DNR) derived from the same sensitive cell K562 (non-MDR) to hydrogen peroxide. In addition, we evaluated how the cell lines respond to the oxidizing agent in the absence of VCR/DNR. In absence of VCR, Lucena drastically decreases cell viability when exposed to hydrogen peroxide, while FEPS is not affected even without DNR. To analyze whether selection by different chemotherapeutic agents may generate altered energetic demands, we analyzed the production of reactive oxygen species (ROS) and the relative expression of the glucose transporter 1 gene (glut1). We observed that the selection through DNR apparently generates a higher energy demand than VCR. High levels of transcription factors genes expression (nrf2, hif-1α, and oct4) were kept even when the DNR is withdrawn from the FEPS culture for one month. Together, these results indicate that DNR selects cells with greater ability to express the major transcription factors related to the antioxidant defense system and the main extrusion pump (ABCB1) related to the MDR phenotype. Taking into account that the antioxidant capacity of tumor cells is closely related to resistance to multiple drugs, it is evident that endogenous antioxidant molecules may be targets for the development of new anticancer drugs.


Introduction
Chemotherapy continues to be one of the main ways to treat different types of cancers. However, it is also observed as a result of chemotherapy, low prognosis, metabolic dysfunctions, poor effectiveness of systemic action, cytotoxicity, and multidrug resistance (MDR) phenotype [1]. MDR is one of the significant causes of cancer treatment failure leading to changes in cellular metabolism and consequent redox status deregulation, which are related to tumor aggressiveness and unfavorable cancer prognosis [2].
The selective events of chemotherapeutic agents usually cause changes in classic tumor metabolism pathways. Among these altered pathways, the overexpression of key genes that regulate cell metabolism and deregulated glucose uptake are hot spots [1,3]. As byproducts of the metabolic processes, the reactive oxygen species (ROS) are constantly generated in a wide variety of changes to cellular metabolic responses [4] and oxidative stress, in turn, is characterized as a metabolic imbalance between the formation and detoxification of ROS.
Chemotherapy drugs such as daunorubicin (DNR) and vincristin (VCR) may alter the levels of transcription factors that regulate the levels of energy supply, extrusion pump related to the MDR phenotype (ABCB1), activate response elements that are responsible for increasing or decreasing levels of antioxidant gene expression. One of the 1 3 cellular responses to these alterations is overexpression of the transcription factor known as hypoxia-1 induction factor (HIF-1α) [3]. HIF-1α plays an important role in increasing one of the major indicator genes of tumor metabolism, the glucose transporter 1 (glut1) encoded by slc2a1 gene [4]. In this context, overexpression of glut1 and hif-1α have been observed in acute myeloid leukemia, being also related to the MDR phenotype and metabolic processes [5]. hif-1α also appears mutated in breast cancer [6].
Another alteration in response to different chemotherapeutic agents occurs through the activation of the transcription factor NRF2 (nuclear factor-erythroid-2-related factor 2), acting as an oxidative stress sensor, signaling and activating the promoter region of the antioxidant response elements (ARE) [7]. The antioxidant genes code for enzymes of stage II of cellular detoxification like glutathione-S-transferase (GSTM4), catalase (CAT) and superoxide dismutase (SOD1, SOD2 and SOD3) [8,9]. NRF2 has been associated to the MDR phenotype by regulating the expression of the multidrug-resistance-associated protein-1 gene (mrp1). Resistant cells from lung cancer (H69AR) have high levels of MRP1 in relation to the sensitive lineage (H69) [10]. NRF2 also presents a relationship with abcf2 gene (subgroup of ABC family) in ovarian cancer [11].
The transcription factor OCT4 is normally associated with pluripotency and is responsible for maintaining cells with undifferentiated embryonic characteristics. This transcription factor is related to the MDR phenotype and is already known to activate extrusion glycoproteins genes like abcb1 [12]. Other studies have reported the association of OCT4 with prostate cancer [13] and lung cancer [14]. Although there are many studies on the MDR phenotype, few relate the resistance to the antioxidant capacity from the selective effect of the chemotherapeutic agents.
In this context, we studied chronic myeloid leukemia (CML) cell lines selected and resistant by different selective agents. The major molecular feature of CML is a reciprocal translocation of chromosomes 9-22 (t (9:22) (q34; q11)), known as Philadelphia chromosome (Ph1). This mutation generates the fusion of two genes BCR and ABL, which fuse and give rise to oncoprotein BCR-ABL, a protein tyrosine kinase [15]. Rumjanek et al. (2001) [16] established the cell line Lucena-1, which was selected resistant from the K562 erythroleukemic lineage, through increasing doses of the VCR chemotherapy. The same research group also selected another resistant line from K562. The FEPS line was selected as resistant through DNR chemotherapy [17].
Although the parental origin is the same (K562), those lineages have phenotypic and genotypic differences that make them an excellent model for research on the MDR phenotype [18]. In this work, for the first time, we identified that different chemotherapeutic drugs (VCR and DNR) might select cells with various antioxidant capacities in human erythroleukemic lines with multidrug resistance phenotype (Lucena-1 and FEPS).

Cells and culture procedures
The K562, Lucena-1 and FEPS cell lines were provided of Immunology Laboratory at the Medical Biochemistry Institute Leopoldo de Meis of the Federal University of Rio de Janeiro (Brazil). The cells were were cultured in the Cell Culture Laboratory of the Federal University of Rio Grande at 37 • C in 5% CO 2 in disposable plastic flasks containing RPMI 1640 (Gibco, Brazil), medium supplemented with sodium bicarbonate (2.0 g/L) (Vetec, Brazil), l-glutamine (0.3 g/L) (Vetec, Brazil), Hepes (25 mM) (Acros, Belgium), 10% fetal bovine serum (Gibco, Brazil), 1% antibiotic (penicillin, 100 U/mL and streptomycin 100 mg/mL) (Gibco, Brazil) and antimycotic (amphotericin B 0.25 mg/mL-Sigma, Brazil). For maneitance of MDR phenotype, Lucena-1 cell line was maintained with 60 nM of VCR (Sigma, Brazil) and FEPS cell line with 300 ng/ml of DNR (Sigma, Brazil).

Vincristine (VCR) and daunorubicin (DNR) treatment
The cell lines were cultured in the presence of the chemotherapeutic -Lucena-1 cells with VCR (60 nM) and FEPS cells with DNR (300 ng/ml) or 30 days without the respective chemotherapeutic agents. Subsequently, the cells of each treatment (with or without the chemotherapeutic drug) were used for the analyses of cell viability, reactive oxygen species levels and gene expression. K562 cells were used as control. All experiments were performed with six samples from each cell line.

Evaluation of Reactive oxygen species (ROS) levels
K562, Lucena-1 with or without VCR, and FEPS cells with or without DNR (2,5 × 10 5 cell/mL) were washed with PBS and incubated for 30 min at 37 °C with the fluorogenic compound 2′,7′-dichlorofluorescin diacetate (H 2 DCF-DA) at a final concentration of 40 ⎧M. After loading with H 2 DCF-DA, the cells were washed with PBS and then resuspended in fresh PBS. Five replicates of 160 ⎧L aliquots of each sample were placed into an ELISA plate, and the fluorescence intensity was determined during 90 min at 37 °C, using a fluorometer (FilterMax F5), with excitation and emission wavelengths of 485 and 520 nm, respectively. The ROS levels were expressed in terms of fluorescence area, after fitting fluorescence data to a second-order polynomial and integrating between 0 and 90 min to obtain the area.

Total RNA extraction and cDNA synthesis
Total RNA of K562, Lucena-1 with or without VCR, and FEPS cells with or without DNR (2 × 10 6 cell/mL) was extracted by Trizol reagent method, according to manufacturer's instructions (Invitrogen, Brazil) from six samples of each treatment. RNA was quantified by BioDrop µLite (England) and its integrity was confirmed by electrophoresis in agarose gel (1%) stained with 0.5 µg/mL of ethidium bromide. The RNA was then treated with Dnase1 (Invitrogen) and 2 µg of total RNA, following the manufacturer's instructions for reverse transcriptase (RT High capacity, Applied Biosystems, Brazil).

Gene expression-Real-time quantitative PCR
The gene expression analysis was performed using a Realtime PCR System 7500 (Applied Biosystems) through the SYBR Green PCR Master Mix kit (Applied Biosystems). Specific primers were designed for each gene analyzed using the Primer-Blast tool from the NCBI website (http:// www. ncbi. nlm. nih. gov). The analyzed genes (Table 1) were from antioxidant system (cat, gstm4, sod1, sod2 and sod3), glucose transporter (glut1), member of ABC family (abcb1) and transcription factors (nrf2, hif-1α and oct4). The ef1α and b2m genes were used as internal controls for data normalization. geNorm applet (Vandesompele et al., 2002) was used to validate the reference genes through a normalization factor, which was calculated as the geometric mean of the expression values of the reference genes. Relative expression levels of the target genes were calculated by dividing the expression value of the target gene by the normalization factor.

Statistical analysis
All experiments were repeated at least three times. Statistical significance between groups was determined by one-way ANOVA, followed by Tukey's post hoc test. The results are expressed as mean ± S.E.M. P values < 0.05 were considered statistically significant.

Results
Cell Viability Assay Figure 1 shows cell viability for the three cell lines studied here, K562, Lucena with and without chemotherapy (VCR), and FEPS with and without chemotherapy (DNR). The first bar indicates the control (light gray bar). In contrast, the other ones show cell viability after 24 h exposure to H2O2 at concentrations of 20 (gray bar) and 60 mM (dark gray bar), respectively, to all cell lines.
K562 showed a significant decrease in cell viability for concentrations of 20 and 60 mM compared to the control. Although there is a strong trend, both treatments showed no significant difference.
The Lucena cell line treated with chemotherapy VCR shows a drop in cell viability only at a concentration of 60 mM. However, compared to Lucena without chemotherapy, both concentrations are cytotoxic to the cell, with a significant drop in cell viability. FEPS with and without chemotherapy DNR does not show a significant difference and remains resistant to H2O2 for all treatments.

Intracellular levels of ROS and relative gene expression of glucose transporter 1 (glut1)
In the absence of chemotherapeutics, both MDR cell lines decreased ROS levels. However, this decrease was more pronounced in FEPS cells (Fig. 2a). For glut1 expression, we observed that FEPS had a higher level of expression in relation to the other cell lines (K562 and Lucena-1), independently of DNR presence (Fig. 2b). For glut1 expression, we did not observe alterations in the absence of the chemotherapy drugs (Fig. 2b).

Transcriptions factors expression
VCR induces the expression of nrf2, oct4 and hif-1α in relation to K562. The same does not happen with DNR, being only hif-1α increased by this. In the absence of VCR, only hif-1α remains high while the other two genes return to the parental cell line-level. Differently, in the DNR absence all genes remain in high expression level (Fig. 3).

Antioxidant genes and extrusion pump expression
All genes analysed (cat, gstm4, sod1, sod2, sod3 and abcb1) showed a consistent pattern of expression. VCR induces all six genes in high level, while the absence of this chemotherapeutic reduces significantly their expression to the parental cell line-level or below. In contrast, DNR induced almost all genes except sod1 and sod2. Besides, its absence was able to decrease only abcb1 expression but even so at levels higher than parental cell line (Fig. 4).

Discussion
This work identified that different chemotherapeutic drugs might select cells with different antioxidant capacities. At first, we compared the sensibility of the two resistant cell lines (MDR) against an oxidant agent (H 2 O 2 ). The cell lines used were selected by different chemotherapeutics. Lucena-1 was became MDR using VCR and FEPS using DNR. Usually, the cell lines are maintained in the presence of these respective chemotherapeutics for preserving the MDR phenotype. So we compared the sensitivity of these cell lines maintained in the presence of VCR or DNR, and in the absence of them (for one month). Our results show that cell lines have different sensitivities when exposed to H 2 O 2 . Lucena-1 and FEPS were resistants to both concentrations of H 2 O 2 but after the period without VCR, viability of Lucena cells is drastically reduced (69%) (Fig. 1), while FEPS cells maintain high viability (100%) even in the absence of DNR. These data suggest that the selective effect of DNR generates a population of cells resistant to chemotherapy with increased antioxidant capacity that is more persistent. This antioxidant capacity does not depend on the presence of the DNR because it is maintained even when the cells are cultured for a month without this chemotherapeutic agent.
The increased antioxidant capacity observed for FEPS cells may be related to the need for increased cell energy to cope with DNR. This chemotherapeutic belongs to a family of anthracyclines drugs and is used to treat some types of cancers, including myeloid and lymphoid leukemias [20]. Two main mechanisms are proposed for DNR's effect in tumor cells. The first is the intercalation of DNR to the double helix and the interruption of DNA repair mediated through topoisomerase II. The second mechanism is the generation of oxidative stress. DNR is oxidized to semiquinone, an extremely unstable metabolite, and the semiquinone is converted back into DNR in a process that releases reactive oxygen species [21]. Thus, DNR appears to be more stressful for cells than the VCR, which has as its main mechanism of action the binding to the tubulin protein, which affects chromosome separation during the metaphase and induces apoptosis [22].
The results obtained from the exposure of the Lucena-1 and FEPS cells to the oxidizing agent suggest that chemotherapeutic agents act differently on oxidative metabolism. So, what is the mechanism behind it? Was the energy expenditure of the two cell lines selected by different chemotherapeutic drugs the same? To answer those questions, we evaluated the levels of ROS and the relative expression of Fig. 3 Relative gene expression of transcriptions factors genes nrf2 (a), hif-1α (b) and oct4 (c) were analyzed in K562 (non-MDR), Lucena (with or without VCR) and FEPS (with or without DNR). The data are expressed as the mean ± standard error. The different letters indicate a significant difference for each cell line (p < 0.05) glut1. Taking into account that ROS are byproducts generated during the production of ATP in the respiratory chain that occurs in the mitochondria, we can consider this parameter as an indirect measure of energy demand caused by exposure to DNR. On the other hand, glut1 is a glucose transporter that has great importance in entering glucose into the cell to supply the energetic demand.
The results obtained in the ROS analyses in the two resistant cell lines used in this study pointed to a different energy demand caused by exposure to the two chemotherapeutics: VCR and DNR. The withdrawal of VCR in Lucena-1 cell line caused a 33% decrease in ROS production, whereas the withdrawal of DNR in FEPS cell cultures caused a decrease of approximately 90% in this same parameter (Fig. 2a). glut1 expression did not demonstrate significant changes in Lucena-1 cells when compared to the K562 control. However, the expression of this transporter was increased in FEPS cells, with the peculiarity that this expression remained high even with DNR removal (Fig. 2b). This fact suggests that DNR-resistant cells appear to have been selected in order to keep this key gene constitutively high. The question that arises is whether this transcriptional feature extends to other genes related to the antioxidant defense system and the MDR phenotype.
In terms of understanding the transcriptional mechanisms that govern the expression pattern of a given gene, the first step is to analyze the transcription factors. In the case of glut1, the main transcription factor that regulates its expression is HIF-1α. Fig. 3b shows relative expression of hif-1α, where it is evident that the levels of expression of this transcription factor are higher in FEPS, being maintained this way even in the absence of DNR. Fig. 3a, c show the same pattern of constitutive expression for two other important transcription factors: nrf2 (associated with control of expression of genes from the antioxidant defense system), and oct4 (associated with stem tumor cells and the MDR phenotype).
Taken together, these results demonstrate that DNR selects cells with constitutively active transcription factors, resulting in altered control of target genes expression. As glut1 is controlled by HIF-1α, NRF2 controls cat, gstm4, sod1, sod2, sod3 and extrusion pump abcb1. The extrusion pump is also controlled by OCT4. The results shown in Fig. 4 corroborate the hypothesis that constitutively expressed transcription factors alter the expression of target genes. All six genes analyzed maintain their high expression even in the absence of DNR, which can be explained by the maintenance of nrf2 expression in this experimental condition. The opposite is observed in the Lucena cells, where the withdrawal of the VCR causes a drastic fall in all analyzed genes.
In conclusion we identified for the first time that different chemotherapeutic VCR and DNR may select cells with different antioxidant capacities. It seems that selection with DNR favors cells with constitutive expression of the main transcription factors that control genes related to the energy supply, antioxidant defense system and MDR phenotype. This topic seems to be an interesting research target for the advancement in the knowledge of the mechanisms involved in the development of resistance to chemotherapeutic drugs, which is a great barrier in the treatment of most types of cancers.