Malignant Cells Can Be Sensitized to Undergo Growth Inhibition and Apoptosis by Arsenic Trioxide Through Modulation of the Glutathione Redox System

ARSENIC TRIOXIDE (AS₂O₃), in a protocol originally developed in Harbin, China, has recently been confirmed to be an effective treatment for acute promyelocytic leukemia (APL) in patients who relapsed after chemotherapy and all-trans retinoic acid (tRA) treatment.1-4 The peak As₂O₃ plasma concentration in patients was 5 to 7 μmol/L and rapidly diminished to a more sustained level of 1 to 2 μmol/L, which is thought to be the therapeutic range in treating APL.3As₂O₃ (1 to 2 μmol/L) induces apoptosis in the t(15;17) APL cell line NB4 and APL cells in vitro and, to some extent, in vivo in patients without significant myelosuppression.5,6 As₂O₃-induced apoptosis in NB4 cells was associated with rapid degradation of PML/RAR-α protein,6 the oncogenic protein that is the product of the t(15;17) translocation characteristic of APL.7,8 The fact that, among a series of 37 patients, the two APL patients who failed to respond to As₂O₃lacked measurable PML/RAR-α message by polymerase chain reaction (PCR) suggests an important role for PML/RAR-α protein in As₂O₃-induced apoptosis. Additional clinical experience obtained in Harbin, China by Ting Dong Zhang (personal communication, 1996) showed As₂O₃ to be effective in patients with conventional chemotherapy-resistant lymphoproliferative and myeloproliferative disorders. Because As₂O₃ is reported to be clinically effective in patients previously treated with alkylating agents and anthracyclines,3,4 there should be lack of cross resistance between As₂O₃ and these chemotherapeutic agents.

Organic arsenic compounds are significantly more toxic than inorganic arsenic because of high binding affinity to vicinal SH group-containing proteins.9 Organic arsenicals such as melasporal, although more potent than As₂O₃ in inducing apoptosis in NB4 cells, are clinically too toxic to be used in the treatment of APL.10 Most previous studies that investigated the cellular and biochemical effects of arsenic were performed using concentrations greater than 5 μmol/L, often 50 μmol/L, and the relevance to therapeutic levels (1 to 2 μmol/L) remains to be determined. These high concentrations may initiate gene transcription by altering the phosphorylation state of signal transduction proteins such as tyrosine kinases.11 Arsenic effects phosphorylation by activating specific kinases, by inhibiting thiol-dependent phosphatases, or by interfering with phosphotransferase reactions.12 The protective effects of thiols, such as glutathione, cysteine,13 and dithiols, such as dithiotreitol, against the toxic effects of arsenic suggests that arsenic toxicity results from forming reversible bonds with the thiol groups of regulatory proteins.

The glutathione (GSH) redox system is known to modulate the growth-inhibitory effect of arsenicals.14-18 Until recently, the effect of this intracellular defense system has not been studied in arsenic-induced apoptosis. As⁵⁺, cadmium, thalium, selenium, zinc, or mercury did not induce apoptosis at 1 μmol/L in NB4 cells, suggesting a unique effect of As³⁺(as in As₂O₃) in this process.19The findings that arsenites (40 μmol/L) can induce apoptosis in hamster CHO cells,20 that hydrogen peroxide-resistant CHO cells are less responsive,21 and that catalase-deficient CHO cells are hypersensitive suggests an important role for hydrogen peroxide as a mediator of arsenic-induced apoptosis.21 

There is a need to understand better the cellular response to As₂O₃ at 1 to 2 μmol/L concentrations because it appears to be cancer-selective and therapeutically achievable. Accordingly, we investigated whether the GSH content and its modulation can be correlated with the sensitivity of As₂O₃in NB4 and other malignant cells. We found that the sensitivity to As₂O₃-induced apoptosis was inversely related to the intracellular GSH content and that pharmacologic modulation of intracellular GSH content modulates sensitivity to As₂O₃.

A 0.1% As₂O₃ solution was kindly supplied by Dr Ting Dong Zhang (Harbin, China). Buthionine sulfoxide (BSO), N-acetylcysteine (NAC), ascorbic acid, catalase, and lipoic acid were purchased from Sigma Chemical (St Louis, MO). All of the agents were dissolved in phosphate-buffered saline (PBS), except lipoic acid, which was dissolved in 100% ethanol at a stock solution of 0.1 mol/L. The final ethanol concentration in the medium was not greater than 0.1%.

NB4 t(15;17) (obtained from Dr M. Lanotte),22 HL-60 cells (American Type Culture Collection [ATCC], Rockville, MD), and t(14;18) B-cell lymphoma su-DHL-4 cell lines, which overexpress Bcl-2 (obtained from Dr M. Cleary),23were cultured in RPMI-1640 medium, supplemented with 100 U/mL penicillin, 100 μg/mL streptomycin, 1 mmol/L L-glutamine, and 10% fetal bovine serum. Human breast cancer cell lines T47D and MDA-MB-468 were obtained from ATCC and cultured as previously reported.24 RM5.21 human fibroblasts from normal breast tissue were isolated from reduction mammaplasty specimen and human embryo fibroblasts (HEF) were cultured as described.24 Cell viability was determined by trypan-blue exclusion. The cells were in logarithmic growth when seeded at 1 × 10⁵ cells/mL for studies performed in duplicate and repeated at least three times.

Apoptotic cells stained with acridine orange (AO) and ethidium bromide (EB) were assessed by fluorescence microscopy. Briefly, 1 μL of stock solution containing 100 μg/mL AO and 100 μg/mL EB was added to 25 μL of cell suspension. Total cells, as well as apoptotic cells that showed nuclear shrinkage, blebing, and apoptotic bodies, were counted. DNA fragmentation analysis was performed as described previously.25 

Intracellular GSH contents were measured using Glutathione Assay Kit (Calbiochem, San Diego, CA). In brief, 5 × 10⁶ cells were homogenized in 5% metaphosphoric acid using a Teflon pestle (Racine, WI). Particulate matter was separated by centrifugation at 4,000g. Supernatant was used for GSH measurement according to the manufacturer’s instruction, while the pellet was dissolved in 1 mol/L NaOH and analyzed for protein by Bio-Rad protein assay (Bio-Rad Laboratories, Hercules, CA). The GSH content was expressed as nanomoles per milligram protein.

Protein extracts (50 μg) prepared with RIPA lysis buffer (50 mmol/L Tris-HCl, 150 mmol/L NaCl, 0.1% sodium dodecyl sulfate [SDS], 1% NP-40, 0.5% sodium deoxycholate, 1 mmol/L phenylmethylsulfonyl fluoride [PMSF], 100 μmol/L leupeptin, and 2 μg/mL aprotinin, pH 8.0) were separated through an 8% or 12% SDS-polyacrylamide gel and transferred to nitrocellulose membranes. The membranes were stained with 0.2% Ponceau red to assure equal protein loading and transfer. After blocking with 10% nonfat milk, the membrane was incubated with polyclonal antibody to PARP (Boehringer Mannheim, Indianapolis, IN), and monoclonal antibodies to Cpp32 and Bcl-2 (Oncogene Research Products, Cambridge, MA). The immunocomplex was visualized by chemoluminescence (ECL kit; Amersham, Buckinghamshire, UK).

Mononuclear cells (MNC) from blood and/or bone marrow from normal adults collected into syringes containing 50 U heparin were prepared following dilution 1:1 with α medium, layered onto an equal volume of Ficoll-Paque, and centrifuged at 450g for 30 minutes at 18°C. For methylcellulose cultures, MNC were cultured according to methods previously described.26 Each milliliter of culture contained 300,000 MNC, 0.8% methylcellulose in alpha medium, 30% fetal bovine serum, 1% bovine serum albumin (BSA; COHN fraction IV), 0.1 mmol/L 2-mercaptoethanol, 0.1 U penicillin, and 0.1 μg/mL streptomycin. Erythroid cultures contained erythropoietin (2 U/mL). Myeloid cultures contained granulocyte-macrophage colony-stimulating factor (GM-CSF; 2 ng/mL) and interleukin-3 (IL-3; 50 U/mL). Cultures were performed in triplicate in 0.3-mL vol in Nunc four-well dishes (Intermed, Roskilde, Denmark) and incubated at 37°C in 4% CO₂ with or without As₂O₃, ascorbic acid, or the combination. Colonies derived from colony-forming units—erythrocyte (CFU-E) were counted on day 7 and those from burst-forming units—erythrocyte (BFU-E) and CFU-GM on day 13 using a Stereozoom dissecting microscope (Bausch & Lomb, Rochester, NY).

Lymphocytes were isolated from normal adult blood and³H-thymidine incorporation was measured in phytohemagglutinin (PHA)-activated lymphocytes according to the reported method.27 

BDF1 female mice were obtained from Charles River Laboratories (Wilmington, MA). All procedures confirmed to the National Institutes of Health (NIH) guidelines for the care and use of laboratory animals. Transplantable P388D1 cells were obtained from ATCC and carried intraperitoneally in BDF₁ mice. For experiments, 0.1 mL containing 2 × 10⁶ cells obtained from ascites after 7 days of transplantation were inoculated intraperitoneally. Mice were randomly divided into four groups each with five mice. After 24 hours, each group was given saline, As₂O₃ (5 mg/kg), and ascorbic acid (500 mg/kg) alone or in combination intraperitoneally every other day for seven times. The percentage increase in lifespan over control (ILS) was calculated as follows: ILS% = T/C% minus 100, where T is the test mean survival time, and C is the control mean survival time. Pairedt-test was used to determine significance.

The 50% inhibitory concentration (IC₅₀) after 3 days of As₂O₃ treatment for NB4 cells was 0.5 μmol/L, for su-DHL-4 cells 1.5 μmol/L, and for HL-60 cells greater than 3 μmol/L (Fig 1A). As₂O₃ 1.2 μmol/L induced apoptosis at day 3 in 50% of the NB4 cells, whereas 1.8 μmol/L was required in su-DHL-4 cells (Fig 1B), while even 5 μmol/L As₂O₃ was insufficient to induce 50% apoptosis in HL-60 cells. The GSH content was three times greater in HL-60 cells than in NB4 cells, whereas in su-DHL-4 cells GSH content fell in between these two values (Fig 1C).

Increasing the GSH content by NAC treatment28 in NB4 cells (Table 1) prevented 2 μmol/L As₂O₃ induction of apoptosis (Fig2A). Similarly, lipoic acid completely blocked As₂O₃-induced apoptosis, perhaps by binding As₂O₃ to its vicinal thiol groups and increasing GSH content.29 Both NAC and lipoic acid treatment inhibited As₂O₃-induced activation of CPP32 and apoptosis. However, NAC did not and lipoic acid only partially inhibited the degradation of PML/RAR-α protein (Fig 2B, lanes 4 and 6). Thus, the degradation of PML/RAR-α alone was not sufficient for the As₂O₃ induction of apoptosis (Fig 2B). In contrast, BSO is known to decrease intracellular GSH levels by inhibition of γ-glutamylcysteine synthase activity.30 During a 24-hour treatment, BSO alone reduced the GSH level to approximately 10 nmol/mg protein in all three cell lines (Table 1), while it did not affect cell growth or apoptosis (Fig3). Treatment with BSO for 4 hours followed by 1 μmol/L As₂O₃ for 12 hours sensitized the three cell lines to As₂O₃ since 75% of the NB4 cells, 70% of the su-DHL-4 cells, and 55% of the HL-60 cells underwent apoptosis (Fig 3). Thus, leukemic cell lines devoid of PML/RAR-α can be made as sensitive to As₂O₃ as NB4 cells.

In HL-60 and su-DHL-4 cells, ascorbic acid acts as an oxidizing agent decreasing the GSH content of the cells and synergizing with the growth-inhibitory and apoptotic effect of As₂O₃(Fig 4 and Table 1). The apoptotic morphology and DNA ladder, as well as Cpp32 activation, in su-DHL-4 cells treated with 1 μmol/L As₂O₃ and 62.5 μmol/L ascorbic acid are shown in Fig 5. In contrast to ascorbic acid, dehydroascorbic acid did not potentiate the effect of As₂O₃ (Table2). It is plausible that the autooxidation of ascorbic acid that results in the formation of H₂O₂31 is responsible for the synergistic effect obtained by ascorbic acid. This suggestion is substantiated by our findings that (1) H₂O₂synergized the effect of As₂O₃ (Table 2), and (2) catalase abolished the synergistic effects obtained by ascorbic acid.

The synergistic growth-inhibitory effect of As₂O₃ and ascorbic acid was observed not only in cell lines, but also in primary cultures of chronic lymphocytic leukemia (CLL) cells (Table 3). The data in Table 3 show that a 5-day exposure of the cells to 1 μmol/L As₂O₃ or 125 μmol/L ascorbic acid alone had minimal effect on cell growth, but the combination of the two agents reduced the viable cell number by approximately 60% (Table3).

As₂O₃ (1 μmol/L) inhibited cell viability in two human breast cancer cells (Table 4). MDA-MB-468, a breast cancer cell line with low glutathione-S-transferase (GST) activity, was remarkably sensitive to 1 μmol/L As₂O₃ (97% inhibition of growth), while T47D with higher GST activity was less sensitive to As₂O₃ (Table 4). The sensitivity to As₂O₃ was inversely proportional to GSH levels in both breast carcinoma cell lines. Ascorbic acid increased T47D sensitivity to As₂O₃. In contrast, normal human breast fibroblasts or HEF were more resistant to As₂O₃ alone or in combination with ascorbic acid.

Human bone marrow or peripheral blood MNC grown in methylcellulose were treated with As₂O₃ and ascorbic acid alone or in combination for up to 2 weeks. A 1 μmol/L quantity of As₂O₃ inhibited CFU-E cells by approximately 60%, but had minimal effect on CFU-GM or BFU-E colony formation. As₂O₃ significantly inhibited colony-forming ability at concentrations greater than 2 μmol/L (Fig6A). Treatment with ascorbic acid did not inhibit colony formation at concentrations less than 500 μmol/L (Fig6B). Ascorbic acid did not enhance As₂O₃inhibition of CFU-GM or BFU-E, but at high concentration (500 μmol/L) enhanced As₂O₃ inhibition of CFU-E (Fig 6C). Similarly, the mitogenic response to PHA by normal lymphocytes as measured by ³H-thymidine incorporation was not significantly affected by As₂O₃ and the effect was not augmented by cotreatment with up to 125 μmol/L ascorbic acid (Fig 7). Thus, it seems that malignant cells are more sensitive to the combined treatment of ascorbic acid and As₂O₃ than normal cells.

In vivo studies to evaluate the effect of ascorbic acid and As₂O₃ in the treatment of lymphoma were initiated. Mouse lymphoma P388D₁ cells, which demonstrate in vitro ascorbic acid enhancement of As₂O₃growth inhibition, were implanted (2 × 10⁶ cells) into the peritoneum of BDF1 mice. On the second day As₂O₃ (5 mg/kg) and ascorbic acid (500 mg/kg) alone or in combination was given every other day for seven times. The combination treatment improved survival time, with an ILS of 40%, whereas single-agent treatment at these nontoxic concentrations did not influence survival (Table 5). The significantly prolonged survival was without additive toxicity as compared with As₂O₃ or ascorbic acid treatment alone.

The therapeutic efficacy of As₂O₃ in APL2-4 prompted our investigations to elucidate the mechanism of action of As₂O₃ in APL derived NB4 cells. Our data indicate that the modulation of the redox system, and particularly the GSH content, determines the sensitivity of the cells toward As₂O₃. We found that 1 to 2 μmol/L As₂O₃, the therapeutically effective concentrations of As₂O₃ which induce remission in APL with minimal toxicity,3 induces apoptosis within 3 days in NB4 cells while other leukemic cells are less sensitive to As₂O₃ (Fig 1). The effect of As₂O₃ is associated with the morphologic changes characteristic of apoptosis, with activation of CPP32, cleavage of PARP, and fragmentation of DNA (Figs 2 and 5). However, degradation of Bcl-2 previously reported during As₂O₃-induced apoptosis of NB4 cells5 was not observed in su-DHL-4 cells even in the presence of ascorbic acid (Fig 5C) and may be cell-type specific.

We investigated whether PML/RAR-α, the oncogenic protein of APL, is responsible for the exquisite sensitivity of NB4 toward As₂O₃. We have confirmed6 that As₂O₃ at concentrations less than 0.5 μmol/L for 3 days of treatment induces degradation of PML/RAR-α in NB4 cells without inducing apoptosis. In the presence of NAC or lipoic acid, As₂O₃, even at high concentration, does not induce apoptosis, but it is still effective in inducing degradation of the PML/RAR-α protein (Fig 2). That lipoic acid, a vicinal SH group-containing compound that may directly bind arsenic,9was more effective (Fig 2, lane 4) than NAC (Fig 2, lane 6) in preventing PML/RAR-α degradation suggests that lipoic acid competes with PML/RAR-α in binding to As₂O₃ and may prevent As₂O₃-induced structural changes in PML/RAR-α. The high sensitivity to As₂O₃degradation of the nucleoplasmic fraction of PML and PML/RAR-α, but not other proteins in the nuclear dense body,32 may be due to their high content of vicinal cysteines in ring fingers and B-box motifs7 to which As₂O₃ may bind with high affinity.9 

GSH is the major autooxidant of the cells and functions to scavenge free radicals and to detoxify toxins and chemotherapeutic agents.33,34 GSH can bind arsenic from attacking its target by formation of a transient As(GS)₃ complex.18Many enzymes and factors modulate the level of GSH. It is known that the glutathione transferase GST-π, an enzyme involved in metabolic detoxification of a variety of xenobiotics, is increased in an arsenic-resistant CHO cell line.14,35 We also found that the MDA-MB-468, a breast cancer line with low GST activity, is sensitive to very low concentrations of As₂O₃(Table 4).36 Glutathione peroxidase is lower in another CHO subclone than in the parent cells and is less resistant to arsenite-induced genotoxicity.37 Drugs that inhibit the activity of GSH peroxidase increase the efficacy of the apoptotic activity of As₂O₃ in malignant lymphocytic cell lines (manuscript in preparation). Multidrug-resistant protein (MRP) is important in determining the sensitivity of many natural toxins, including sodium arsenite, and cotransports GSH and xenobiotics from the cell.38,39 The relative importance of adenosine triphosphate (ATP)-dependent membrane export by MRP and the GSH detoxification pathway in determining sensitivity to As₂O₃ remains to be determined. Meanwhile, using subtractive hybridization, it was found that metallothionine and thioredoxin reductase, two other detoxification system-related factors,40,41 are upregulated in arsenic treated NB4 cells (M. Mao, personal communication, 1997).

We have shown that compounds that lower the GSH content potentiate the apoptotic effect of As₂O₃, whereas protecting or increasing the GSH level protects the cells from the apoptotic effect of As₂O₃ (Figs 2 and 3). NB4 cells are protected from the apoptotic effect of As₂O₃ if treated with the antioxidant NAC, which preserves the GSH content (Fig2A). On the other hand, BSO, which inhibits the synthesis of glutamyl-cysteine, lowers the GSH content of NB4 cells and potentiates the apoptotic effect of As₂O₃. In BSO-treated cells, 1 μmol/L As₂O₃ induces apoptosis in 70% of the cells in only 12 hours of treatment, and less sensitive cell lines like su-DHL-4 and HL-60 become as sensitive as NB4 (Fig 3).

Ascorbic acid, which can act as a major antioxidant can protect cells from oxygen radicals formed by 4-HPR,42curcumin,43 and other agents44,45 that induce apoptosis. In contrast, ascorbic acid can induce apoptosis alone46 or in combination with agents in other cell systems.47 In HL-60 cells, dehydroascorbic acid but not ascorbic acid is transported via glucose transporters and is coupled to its intracellular reduction to ascorbic acid.48 In our studies, ascorbic acid in combination with As₂O₃ has a potentiating effect due to its capacity to undergo autooxidation resulting in the formation of H₂O₂ which potentiates the effect of As₂O₃ (Table 1). This assumption is fortified by our findings that in the presence of catalase, ascorbic acid does not enhance As₂O₃ and the combined treatment of As₂O₃ and H₂O₂ is synergistic for the induction of apoptosis (Table 2). Thus, the balance between GSH and H₂O₂ may determine the apoptotic effect of As₂O₃ in leukemic cells. However, ascorbic acid did not potentiate the effect of As₂O₃ on the colony formation capacity of normal hematopoietic cells (Figs 6 and 7). The synergistic effect of As₂O₃ with ascorbic acid manifested itself also in one epithelial mammary carcinoma line T47D, whereas normal breast or embryo fibroblasts were insensitive to As₂O₃and not sensitized by the addition of ascorbic acid (Table 4). This selective action of ascorbic acid may be partially due to the low concentration of catalase in some malignant cells.49,50Thus, it should be possible to use low concentration of As₂O₃ with ascorbic acid to selectively induce apoptosis in some malignant cells without causing severe side effect in normal tissues. We tested this possibility in an in vivo mouse model. We found that the combined effect of As₂O₃ and ascorbic acid increased the survival time of mice injected with P388D-lymphoma cells, whereas single-agent treatment did not influence survival (Table 5).

In summary the redox system of the cell and the capacity to eliminate reactive oxygen species determine the efficacy of As₂O₃. It is noteworthy that normal cells in general are more efficient in eliminating reactive oxygen species than malignant cells. Thus, it seems that malignant cells with low GSH levels or GST activity should be sensitive to As₂O₃ or the combined treatment of As₂O₃ and ascorbic acid.

We appreciate the technical assistance of Yelena Galperin for bone marrow colony assays and the guidance of Dr George Acs throughout these studies.