Emodin and DHA potently increase arsenic trioxide interferon-α–induced cell death of HTLV-I–transformed cells by generation of reactive oxygen species and inhibition of Akt and AP-1

Adult T-cell leukemia (ATL) is associated with a poor clinical prognosis, and treatment of patients using conventional chemotherapy has limited benefit given that HTLV-I cells are resistant to conventional anticancer, apoptosis-inducing agents.^1-3  Cyclophosphamide, adriamycin, vincristine, and prednisolone (CHOP) protocol has a reported 17% rate of remission but a predicted survival of less than 15% after 3 years.^4-6  Some success has been reported in treating ATL with a combination of zidovudine (AZT) and IFN-α.^7-10  However, this success is fleeting as many patients relapse, because response to therapy requires the presence of a transcriptionally active p53 gene.¹¹  Other treatments include bone marrow transplantation,^12-14  topoisomerase inhibitors,^15,16  anti-Tac monoclonal antibody,¹⁷  inhibitors of NF-kB^18,19  and proteasome inhibitors.²⁰ 

Arsenic trioxide in combination with IFN-α is very effective in treating APL^21-23  and showed promising results in the treatment of patients with ATL.^24-29  However, differences in sensitivity to arsenic, as well as the toxicity associated with its use, require a slow dose increase in vivo, and treatment may need to be discontinued in some patients with ATL. Thus, it is worthwhile to investigate the effectiveness of docosahexaenoic acid (DHA) a nontoxic ω-3 polyunsaturated fatty acid found in fish oil. HL-60, an acute myeloid leukemia cell line that is resistant to clinically relevant doses of arsenic, shows a 90% reduction in viability along with an increase in apoptosis, an increase in intracellular reactive oxygen species (ROSs) and lipid peroxidation, as well as an up-regulation of Bax when treated with a combination of arsenic and DHA.³⁰  Further, treatment with oleic acid, a monounsaturated fatty acid that is not susceptible to lipid peroxidation, did not enhance the effect of arsenic treatment.³¹  In breast cancer cells, the addition of DHA results in an increase in malondialdehyde, an end product of lipid peroxidation, which increases sensitivity to doxorubicin treatment through oxidative stress. Additionally, DHA failed to increase sensitivity to mitoxantrone, a drug that does not alter hydroperoxide levels, indicating that the addition of DHA is only effective in combination with drugs that induce oxidative stress.³²  This presents a rationale for use of DHA in combination with emodin, a naturally occurring anthraquinone that has been shown to generate intracellular ROSs in cancer cells.³³  Treatment with arsenic increases the concentration of intracellular ROSs,³⁴  and it has been shown that lower levels of ROSs promote apoptosis, whereas higher levels cause necrosis. An increase in ROSs has been tied to the activation of Bax, a proapoptotic member of the Bcl-2 family, at the mitochondrial membrane.³⁵  It has recently been shown that arsenic treatment induces activation of Bax and that the activation can be blocked by overexpression of Bcl-2 or ROS scavengers but not by pan-caspase inhibitors, indicating that ROSs, or changes in redox states, are involved in cellular death upstream of Bax activation.³⁶ 

In this study we found a potent synergistic effect of the combination of As/IFN-α with emodin and DHA in HTLV-I–infected cells. The mechanism underlying these effects was in part linked to ROS production which leads to an increase in Bax and a decrease in Bcl-2 expression. In addition, combinatorial use of emodin and DHA with As/IFN-α resulted in the inhibition of the AP-1 and Akt pathways in HTLV-I cells, as demonstrated by a decrease in JunD, JAB1, and Akt expression. Our results suggest that As/IFN-α in combination with ROS-inducing/sensitizing agents may have benefits in ATL therapy.

IFN-α 2 (Cell Sciences, Canton, MA) and arsenic trioxide (Sigma, St Louis, MO) were used at 100 U/mL and 1 μM, respectively, as previously described.¹  DHA (Sigma) was dissolved in ethanol and used at 10 to 50 μM. Emodin (Sigma) was dissolved in DMSO and used at 10 to 50 μM. Ascorbic acid (Sigma) was dissolved in sterile dH₂O and used at 5 mM.

C8166 and MT-2 are HTLV-I–transformed T-cell lines. Jurkat and CEM HTLV-I–negative T-cell lines were grown in RPMI-1640 (Invitrogen Carlsbad, CA) supplemented with 10% fetal bovine serum, gentamycin, and penicillin-streptomycin. 1185 and LAF, IL-2–dependent T-cell lines immortalized by HTLV-I, and peripheral blood mononuclear cells (PBMCs) were cultured in 20% serum and 40 U/mL IL-2.

Cell proliferation was measured using the Cell Proliferation Kit II (XTT; Roche, Mannheim, Germany) according to manufacturer's instructions. Cells (10⁵/mL) were treated 48 to 72 hours in a 96-well plate and then incubated with tetrazolium salt XTT for 4 hours. Proliferation was quantified by measuring cleavage of XTT to an orange formazan dye using an enzyme-linked immunoabsorbent assay (ELISA) reader at 450 nm.

Cells (2 × 10⁶) were treated 24 to 72 hours then washed with PBS. After resuspension, cells were fixed on ice with 80% EtOH for 30 minutes. Cells were washed with PBS and treated with RNase for 15 minutes at 37°C, followed by staining with 50 μg/mL propidium iodide for 15 minutes at room temperature. Cells were then washed and resuspended in PBS and analyzed by flow cytometry.

Production of ROSs in cells was measured using 2′,7′-dichlorofluorescin diacetate (DCFH-DA; Sigma), an oxygen-sensitive fluorescent probe. Cells (5 × 10⁵) were incubated for 12 hours, as indicated, with arsenic and IFN-α with or without ascorbic acid, then DHA and emodin were added for another 12 hours. DCFH-DA was added at 5-nM final concentration to each sample and incubated at 37°C for 15 minutes. Cells were then collected and washed with cold PBS. Cells were analyzed by flow cytometry after resuspension in 500 μL PBS.

Cells (5 × 10⁵) were treated as described in “Measurement of intracellular ROS production” for the measurement of ROS production, then collected and washed with cold PBS. Cells were then stained with annexin V/propidium iodide using the Vybrant Apoptosis Assay Kit no. 2 (Molecular Probes, Eugene, OR) according to the manufacturer's instructions.

Cells (1 × 10⁶) were treated for 24 hours, then changes in the mitochondrial membrane potential (ΔΨm) were measured using the ApoAlert Mitochondrial Membrane Sensor Kit (Clontech, Mountain View, CA) according to the manufacturer's instructions.

Cells were washed with PBS containing phosphatase inhibitors and lysed in RIPA buffer with phosphatase and protease inhibitors (Complete Cocktail; Roche) after 24 to 48 hours of treatment. Protein concentration was determined with the Bio-Rad protein assay (Bio-Rad, Hercules, CA). Proteins were separated by 10% SDS–polyacrylamide gel electrophoresis and transferred to Immobilon PVDF membranes (Millipore, Billerica, MA), visualization was done with the Supersignal West Dura enhanced chemiluminescence system (Pierce, Rockford, IL). Spot densitometry quantification of expression was performed as previously reported³⁷  and was expressed as a percentage of the appropriate control sample referred as 100%. The study protocol was approved by the institutional review board at the University of Kansas Medical Center.

Cellular proliferation was assessed using the Cell Proliferation Kit II XTT, according to the manufacturer's instructions. Proliferation was quantified by measuring cleavage of XTT to an orange formazan dye using an ELISA reader at 450 nm, as previously reported.²⁶  Results are expressed as the percentage of control-treated cells and represent the mean values of at least 2 independent experiments performed in triplicate. Our results indicated a very strong synergy between As/IFN-α and DHA/emodin at all time points and doses used in the HTLV-I–infected cells C8166 and MT-2, as well as the IL-2–dependent 1185 and LAF cell lines. This was in contrast to those effects not seen in the control PBMCs (Figure 1A-D) or with buffer controls (data not shown). Consistent with previous reports, MT-2 cells were more resistant to As/IFN-α–induced cell death when compared with C8166, LAF, or 1185 cells. However, all HTLV-I cell lines appeared to be equally and highly sensitive to the combination of As₂O₃/IFN-α with emodin and DHA (AIDE) (Figure 1A).

The effect of emodin and DHA was also tested in the absence of arsenic trioxide and IFNα. As reported in Figure 1B, no effect on cell growth was detected for C8166 cells when cultured in the presence of emodin and DHA at a concentration of 10 μmol/L, which confirms the synergistic action of emodin and DHA with arsenic trioxide and IFN-α (Figure 1A). Because of arsenic trioxide toxicity, a major limitation to its use as a therapeutic agent for patients with ATL is the time required to achieve the clinically active dose. Because emodin and DHA showed a strong synergy with As/IFN-α, we tested whether we could lower the dose of arsenic while still preventing proliferation of tumor cells. The combination of emodin and DHA allowed up to a 100-fold decrease in arsenic concentrations with little loss in efficacy against tumor-cell proliferation (Figure 1E), whereas control cell lines Jurkat and CEM were not significantly affected (Figure 1F).

We next analyzed cell-cycle progression following treatment of HTLV-I cells with As/IFN-α (AI) or AIDE.

Results revealed a marked G₀/G₁ arrest of tumor cells cultured in the presence of emodin and DHA along with As/IFN-α (Figure 2A). In HTLV-I–infected cells the retinoblastoma protein (Rb) is constitutively phosphorylated, thereby allowing tumor cells to progress to S phase.^38,39  In agreement with cell-cycle results presented in Figure 2A, we found that Rb accumulated in its hypophosphorylated form in 1185 and C8166 cells (Figure 2B). This effect was more pronounced in cells treated with As/IFN-α with emodin and DHA than those treated with As/IFN alone.

Changes in mitochondrial membrane potential, ΔΨm, are a critical step in cells undergoing apoptosis, regardless of the death signal. Therefore, ΔΨm was measured in C8166 cells (Figure 3A). As previously reported As₂/IFN-α–treated cells that entered apoptosis showed an altered mitochondrial membrane permeability that led to ΔΨm collapse. Consistent with the increased ROS generation, mitochondrial membrane permeability collapse was much more pronounced when DHA and emodin were added to As/IFN-α (Figure 3A). By contrast, control C8166 cells treated with buffer did not show any change in ΔΨm.

ΔΨm collapse is usually associated with the activation of caspases. Caspase-3 is a critical downstream protease in the caspase cascade, and poly-ADP-ribose polymerase (PARP) is 1 substrate of caspase-3 protease activity that is associated with apoptosis. Several studies have reported that As/IFN-α–mediated apoptosis of HTLV-I–infected cells is partially a caspase-dependent process and partially caspase independent. To gain insight into the mechanism used by emodin and DHA, cells were analyzed by flow cytometry using double staining (annexin V/PI). Apoptotic cells were scored as annexin V⁺/PI⁻ versus necrotic dead cells, which were PI⁺. Results indicated that DHA and emodin did not significantly increase the apoptotic activity of As/IFN-α but, instead, triggered a potent necrotic-cell death signal, resulting in nearly 90% cell death after 48 hours of treatment (Figure 3B). Similar results were observed with lower doses of DHA and emodin and longer incubation times (data not shown).

These results were further confirmed by Western blot analysis of procaspase 3 and PARP (Figure 3C). Our results indicate that DHA and emodin are not associated with cleavage of the PARP, because the 85-kDa fragment representing the cleaved form was not detected after treatment of HTLV-I cells by As/IFN-α with DHA and emodin.

Together, with the flow cytometry data, this suggests that the combination of As/IFN-α with emodin and DHA is a potent inducer of necrotic-cell death in HTLV-I– infected cells (Figure 3B-C). Spot densitometry quantification analyses of the expression of several proteins implicated in survival pathways revealed a decrease in Bcl-2 expression of 37% and 68% following treatment with AI and AIDE, respectively (data not shown). Concomitantly, we found an increase in Bax expression, 200% and 150%, following treatment with AI and AIDE, respectively (data not shown). In contrast to a previous report, the proapoptotic cleaved form of Bcl-2 was not detected in our assay (data not shown). This is possibly due to differences in experimental procedures because we did not use purified mitochondria in our study. Other markers, such as Bcl-xL, were not significantly affected in our experimental conditions (data not shown).

Because previous studies have shown that lower levels of ROSs promote apoptosis and higher intracellular levels of ROSs are associated with necrosis, we tested whether emodin and DHA treatment may increase the concentration of intracellular ROSs in HTLV-I cells following treatment with As/IFN-α and DHA/emodin. ROS production was not changed in As/IFN-α–treated C8166 cells, but ROS production significantly increased when DHA and emodin were added with As/IFN-α. PBMCs used as control were not affected under the same experimental conditions (Figure 4A).

To demonstrate that ROSs were involved in HTLV-I–cell death, we used ascorbic acid, a well-known ROS scavenger. Treatment with ascorbic acid effectively reduced the generation of ROSs in C8166 cells treated with As/IFN-α and emodin and DHA (data not shown). In parallel experiments the presence of ascorbic acid efficiently prevented the death of HTLV-I cells treated with As/IFN-α and emodin and DHA as demonstrated by propidium iodide and FACS analysis (Figure 4B; compare left and right panels). Generation of ROSs has been linked to increased Bax expression, and ROS scavengers can prevent cell death by reducing Bax protein expression. In fact, when we performed Western blot analysis in HTLV-I cells treated with As/IFN-α and emodin and DHA in the absence or in the presence of ascorbic acid, we found that the presence of ascorbic acid reduced Bax expression by 30% to 35% (data not shown).

Recent studies suggest that DHA inhibits activator protein 1 (AP-1) activity. AP-1 plays a critical role in HTLV-I oncogenesis, is expressed at high levels in ATL patient samples, and may be required for maintenance of the malignant phenotype.⁴⁰  Among AP-1's components, c-Jun is a potent transcriptional activator, and its activity is regulated through phosphorylation by Jun N-terminal kinase (JNK), a member of the MAP kinase family.⁵⁸  AP-1 becomes activated through phosphorylation of Jun family transcription factors by members of the c-Jun NH₂-terminal kinase/stress-activated protein kinase (JNK/SAPK) family. Previously, constitutive JNK activation has been reported in HTLV-I–infected T cells,^41,42  suggesting that HTLV-I–induced JNK activation may contribute to the activation of AP-1 in T cells. However, in our experiments only a small decrease in phosphorylation of JNK2 or JNK3 was detected in HTLV-I cells treated with As/IFN-α with emodin and DHA (data not shown). Previous studies found that JunD is the major component of the AP-1–DNA complex in HTLV-1–infected T-cell lines.^43-45  Others have reported that the Jun-activating binding protein (JAB1) selectively potentiates transactivation by JunD.⁴⁶  Interestingly, the expression of both JunD and JAB1, in cells treated with As/IFN with DHA and emodin, was reduced (Figure 5A). Because recent studies suggest that Akt, an essential survival pathway, is activated by HTLV-I,^47,48  we also tested expression of Akt in HTLV-I cells treated with As/IFN-α and with emodin and DHA. Although As/IFN-α had no significant effect on Akt expression, the addition of emodin and DHA strongly down-regulated Akt expression by 25% to 50% (Figure 5B).

Spot densitometry quantification indicated that a longer incubation time or a higher dose was required to achieve inhibition levels comparable to those observed on Akt, and that JunD was more affected than JAB1 (Figure 5C). However, proliferation assays demonstrated that HTLV-I–transformed cells are more susceptible to As/IFN-α and with emodin and DHA than normal PBMCs (Figure 5D). Antiproliferative effects were more pronounced when emodin and DHA were used at 25 and 50 μmol/L, suggesting that the inhibition of both AP-1 and Akt may have an additive effect on preventing growth of HTLV-I–transformed cells.

Disruption of cellular death pathways can contribute to a number of diseases, including cancer. It is now well established that anticancer agents induce apoptosis or necrosis and that disruption of cellular death pathways can reduce treatment sensitivity. In vitro and in vivo HTLV-1–transformed cells are highly resistant to most death-inducing agents. Although the basis for this resistance is not fully understood, it has been reported that HTLV-I–infected cells overexpress antiapoptotic factors such as Bcl-2, Bcl-xL, survivin, and I-309.^49-52 

Studies suggest a broad spectrum of antileukemic activity for As₂O₃ because of its ability to increase intracellular ROS production.⁵³  In As₂O₃-sensitive cells, such as the APL-derived cell line NB-4, increased intracellular levels of ROSs lead to collapse of the mitochondrial membrane potential, activation of the caspases, and apoptosis. However, cells with intracellular antioxidant defense mechanisms, such as the acute myeloid leukemia (AML)–derived cell line HL-60, are resistant to As₂O₃. To render As₂O₃-resistant tumor cells more susceptible to As₂O₃-mediated cell death, ways to enhance the effects of As₂O₃ have been investigated. Importantly, agents modulating the GSH redox system as well as agents increasing intracellular ROS concentrations have been shown to increase As₂O₃ efficacy in cells otherwise resistant to As₂O₃.⁵⁴  It is well known that clinically achievable concentrations of As₂O₃ (that is, 1-2 μM) induce insufficient amounts of ROSs to induce apoptosis of HL-60 cells in vitro. However, recent studies have shown an enhancing effect of DHA on As₂O₃-mediated cytotoxicity in some As₂O₃-resistant cell lines, indicating that the combined effect of As₂O₃ and DHA is not restricted to AML-derived HL-60 cells but may also be observed in other As₂O₃-resistant hematologic malignancies, including Burkitt lymphoma and hairy-cell leukemia. Recent clinical studies have shown some antileukemic effect of As/IFN-α, suggesting that ATL cells may respond by apoptosis or cell-cycle arrest in vivo. Differences in sensitivity to arsenic, as well as the toxicity associated with its use, make it worthwhile to investigate drug combinations that could enhance the actions of arsenic and reduce the level necessary for treatment. In this study, we found a synergistic effect of the combination of As/IFN-α with emodin and DHA in HTLV-I–infected cells. IL-2–dependent or IL-2–independent HTLV-I–infected cells, as well as freshly isolated patient samples, were all highly sensitive to this drug combination. The rationale for this drug combination was because emodin has been shown to reduce invasiveness of tumor cells⁵⁵  and to enhance As-mediated ROS production.^56,57  In addition, DHA was added to this combination because DHA sensitizes tumor cells to ROS-inducing agents (As and emodin in our case).

Our results indicate a potent antiproliferative effect on tumor cells, which was linked to the accumulation of hypophosphorylated Rb and G₀/G₁ arrest. Previous reports have shown that Tax can induce JNK activity and c-Jun phosphorylation, leading to the formation of Smad3/c-Jun complexes and inhibition of TGF-β signaling.⁴¹  In fact, in ATL and in normal T cells transduced by Tax, c-Jun is constitutively phosphorylated and acts as a repressor of TGF-β signaling. Thus, our results suggest that As/IFN-α in the presence of emodin and DHA may lead to reactivation of TGF-β, thereby preventing the growth of tumor cells. Further investigations in this area are warranted. We also found a significant reduction in Akt expression which may in part be responsible for increased cell death in treated cells.

Previous studies have shown that HL-60 cells are resistant to clinically relevant doses of arsenic but have 90% reduction in viability, increased intracellular ROSs, and up-regulation of Bax when treated with a combination of arsenic and DHA. Similarly, we readily detected increased ROS production in HTLV-I–infected cells, but not in healthy PBMCs, treated with As/IFN-α and emodin and DHA. To further demonstrate that ROS production played a role in tumor growth suppression and death of HTLV-I–expressing cells we measured cell death by FACS in the absence or presence of ascorbic acid, a ROS scavenger. Results clearly demonstrate the importance of ROSs, because significant cell death was prevented by ascorbic acid with concomitant down-regulation of Bax to normal endogenous levels. In human hematopoietic malignancies, continuous AP-1 DNA binding activity involving JunD has been found in cell lines of cutaneous T-cell lymphoma and patients with Sézary syndrome. Previously, it has been shown that HTLV-I–transformed cells and samples from patients with ATL have constitutive AP-1 and JNK activity and that these may play a role in the maintenance of the tumor growth potential. In fact, our data are in agreement with these observations, because we found an inhibition of the AP-1 pathway characterized by a dose-dependent decrease of JunD and JAB1. Inhibition of NF-kB pathway or survivin expression by As/IFN-α have also been reported by other groups and are likely to play a role in the suppression of tumor-cell proliferation by arsenic. However, they do not appear to be involved in the synergy described here between As/IFN-α, emodin, and DHA and, therefore, were not further investigated.

In conclusion, we have shown high synergy of As/IFN-α with emodin and DHA treatment with regard to the inhibition of proliferation and induction of cell death of HTLV-I–transformed T cells. In addition, the synergistic effectiveness of the combination treatment at pharmacologic doses was specific for HTLV-I–infected cells and had marginal effect on PBMCs isolated from healthy donors. The drug combination described here allows the As concentration to be decreased by 100-fold while preserving its antitumor activity, suggesting a broad application in the treatment of leukemias and lymphomas and warrants consideration for a phase 2 clinical study.

Contribution: M.B. performed experiments described in Figures 1 to 5; M.B. performed some experiments described in Figure 3C; C.N. is responsible for the design of the project and interpretation of the results.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Christophe Nicot, University of Kansas Medical Center, Department of Microbiology, Immunology, and Molecular Genetics, 3025 Wahl Hall West, 3901 Rainbow Blvd, Kansas City, KS 66160; e-mail: cnicot@kumc.edu.

This work was supported by the National Cancer Institute (grant RO1CA106258) (C.N.). The flow cytometry core is supported in part by National Institutes of Health grant PZO RR016443 from the National Center for Research Resources.