Autophagy Is a Critical Mechanism for the Induction of the Antileukemic Effects of Arsenic Trioxide*

Arsenic trioxide (As2O3) exhibits potent antitumor effects in vitro and in vivo, but the precise mechanisms by which it generates such responses are not well understood. We provide evidence that As2O3 is a potent inducer of autophagy in leukemia cells. Such induction of autophagy by As2O3 appears to require activation of the MEK/ERK pathway but not the AKT/mammalian target of rapamycin or JNK pathways. In efforts to understand the functional relevance of arsenic-induced autophagy, we found that pharmacological inhibitors of autophagy or molecular targeting of beclin 1 or Atg7 results in reversal of the suppressive effects of As2O3 on leukemic cell lines and primary leukemic progenitors from acute myelogenous leukemia patients. Altogether, our data provide direct evidence that autophagic cell death is critical for the generation of the effects of As2O3 on acute myelogenous leukemia cells and raise the potential of modulation of elements of the autophagic machinery as an approach to enhance the antitumor properties of As2O3 and possibly other heavy metal derivatives.


Arsenic trioxide (As 2 O 3 ) exhibits potent antitumor effects in vitro and in vivo
, but the precise mechanisms by which it generates such responses are not well understood. We provide evidence that As 2 O 3 is a potent inducer of autophagy in leukemia cells. Such induction of autophagy by As 2 O 3 appears to require activation of the MEK/ERK pathway but not the AKT/mammalian target of rapamycin or JNK pathways. In efforts to understand the functional relevance of arsenic-induced autophagy, we found that pharmacological inhibitors of autophagy or molecular targeting of beclin 1 or Atg7 results in reversal of the suppressive effects of As 2 O 3 on leukemic cell lines and primary leukemic progenitors from acute myelogenous leukemia patients. Altogether, our data provide direct evidence that autophagic cell death is critical for the generation of the effects of As 2 O 3 on acute myelogenous leukemia cells and raise the potential of modulation of elements of the autophagic machinery as an approach to enhance the antitumor properties of As 2 O 3 and possibly other heavy metal derivatives.
Arsenic trioxide (As 2 O 3 ) 2 is a metalloid that exhibits potent antineoplastic effects in vitro and in vivo (1)(2)(3). This arsenic derivative induces apoptosis and suppresses the growth of various types of malignant cells of diverse origin in vitro (1)(2)(3). Different mechanisms by which As 2 O 3 promotes cell death of target cells have been extensively studied and described (reviewed in Refs. 1-3). As 2 O 3 -dependent generation of reactive oxygen species leads to activation of pro-apoptotic pathways in different types of cells (1)(2)(3). In addition, there is evidence for mechanisms involving As 2 O 3 -dependent cell typespecific targeting of malignant cells with distinct molecular abnormalities (reviewed in Ref. 3). For instance, there is evidence for As 2 O 3 -dependent specific targeting of cells expressing AML1/MDS1/EVI1 involving degradation of the abnormal fusion protein (4), evidence for arsenic-dependent BCR-ABL ubiquitination and proteasomal degradation (5), and eradica-tion of leukemia-initiating cells in acute promyelocytic leukemia via As 2 O 3 -inducible PML-RAR␣ degradation (6).
Despite extensive research over the years, the precise mechanisms of action of arsenic trioxide in malignant cells are not well understood. In particular, the specific cellular events that account for differential sensitivity of malignant cells to As 2 O 3 remain to be precisely defined. Notably, arsenic trioxide treatment induces responses in patients with acute promyelocytic leukemia in vivo (7,8), and it is an agent approved by the United States Food and Drug Administration for the treatment of this leukemia. The unusually high sensitivity of acute promyelocytic leukemia cells to the effects of arsenic trioxide likely reflects the requirement of lower As 2 O 3 concentrations for induction of leukemic cell differentiation seen in these cells versus apoptotic cell death (3), but other mechanisms may be involved as well.
Autophagy is a cell death mechanism distinct from apoptosis, also defined as type II programmed cell death, involving autophagosomic/lysosomal degradation of cellular components (9). There is emerging evidence that autophagy plays an important role in the regulation of malignant cell survival (9). This process was first linked to human cancers by the demonstration that beclin 1, an integral component of the autophagic machinery, has tumor suppressor activity (10). Such initial studies raised the possibility that decreased expression of autophagy proteins may promote disease progression in breast cancer and possibly other tumors (10). Although the relationship and coordination of autophagic cell death and apoptosis in the control of tumorigenesis and tumor cell survival are still not well understood, there is emerging evidence that autophagy plays roles in the regulation of antineoplastic responses in BCR-ABL-expressing leukemias (11,12), as well as in malignant melanoma cells targeted using certain immunotherapeutic approaches (13).
In this study, we provide evidence that arsenic trioxide induces autophagy in acute myelogenous leukemia (AML) cells. Our data demonstrate that As 2 O 3 is a potent inducer of autophagy, and such induction is mediated by engagement of the MEK/ERK pathway. Importantly, our findings establish that pharmacological or molecular targeting of proteins involved in arsenic-mediated autophagy results in partial reversal of the antileukemic effects of As 2 O 3 on primary hematopoietic precursors from AML patients, establishing a critical role for autophagy in the generation of As 2 O 3 -dependent antileukemic responses.
Cell Lysis and Immunoblotting-Cells were incubated for the indicated times with the indicated concentrations of As 2 O 3 and subsequently lysed in phosphorylation lysis buffer (15,16). Immunoblotting using an ECL methodology was performed as in our previous studies (15)(16)(17). In the experiments in which pharmacological inhibitors of ERK1/2 and JNK were used, the cells were pretreated for 60 min with the indicated concentrations of the inhibitors prior to the addition of As 2 O 3 to the cultures.
siRNA-mediated Knockdown of Beclin 1 or Atg7 in Human Leukemic Cells-siRNA knockdowns were performed using TransIT-TKO as recommended by Mirus Bio LLC (Madison, WI) or by using Nucleofector kits from Amaxa Biosystems (Gaithersburg, MD). Cells were transfected with beclin 1-or Atg7-specific siRNA pools from Dharmacon RNAi Technologies (Lafayette, CO). siRNAs against Mek1, Mek2, and a different siRNA against beclin 1 (beclin 1 siRNA-2) were obtained from Ambion, Inc (Foster City, CA). To assess the effects of the siRNA-mediated knockdown of beclin 1 or Atg7, cells were cultured in a methylcellulose assay system as in previous studies (17).
Assays of Primary Hematopoietic Progenitors from AML Patients-Peripheral blood from patients with AML was collected after obtaining informed consent approved by the Institutional Review Board of Northwestern University. The effects of As 2 O 3 on leukemic progenitor colony formation were assessed by clonogenic assays in methylcellulose, as described previously (15,18).
Acridine Orange Staining and Flow Cytometric Analysis-Formation of acidic vesicular organelles (AVOs), a morphological characteristic of autophagy (19), was quantitated by acridine orange staining (19). Acridine orange (0.5 mg/ml) (Invitrogen) was added 15 min prior to collection, and after washing with phosphate-buffered saline, cells were analyzed by flow cytometry.
Immunofluorescence-Human leukemic cells were transfected with either an empty vector or a GFP-LC3-containing plasmid from Addgene 11546 (Cambridge, MA). The EGFP-LC3 plasmid was created by Karla Kirkegaard (Stanford Uni-versity) (20). Cells were fixed with 3% paraformaldehyde and then stained with DAPI solution. For immunofluorescence staining, GFP-LC3 transfected cells, a polyclonal GFP (FL) antibody by Santa Cruz Biotechnology, Inc., was used. Fluorescence was detected using a Nikon Eclipse C1si confocal microscope system.

RESULTS
In initial studies, we sought to determine whether treatment of cells with arsenic trioxide results in induction of autophagy of leukemic cells. For that purpose, levels of expression of the microtubule-associated protein 1 light chain 3 (LC3)-II, a marker for the presence of completed autophagosomes (9), were determined. Treatment of KT1 cells with As 2 O 3 resulted in strong up-regulation of LC3-II that was clearly detectable within 8 h of treatment of cells with As 2 O 3 and was still present after 48 h of treatment (Fig. 1A). Such an up-regulation of LC3-II was not detectable after very short incubations (10 -120 min) of the cells with As 2 O 3 ( Fig. 1B), suggesting that induction of autophagy occurs after several hours of exposure to As 2 O 3 . We also examined whether there is induction of autophagic flux (21) in arsenic-treated cells. In experiments employing the lysosomal specific protease inhibitors, pepstatin A and E64d, we noticed an increase in LC3-II levels, suggesting that the inhibitors interfere with LC3-II lysosome-dependent degradation (Fig. 1C). To definitively establish induction of autophagic flux, the effects of As 2 O 3 treatment on the expression of p62/ SQSTM1, an LC3-interacting protein degraded in the autolysosomes (22), was examined. Treatment of KT1 cells with As 2 O 3 resulted in a time-dependent decrease of detectable p62/ SQSTM1 protein levels, with maximum suppression reaching a minimum after 48 h of As 2 O 3 treatment (Fig. 1D). Similar results were obtained when the U937 acute myelomonocytic leukemia cell line was studied (Fig. 1E). Induction of autophagy by arsenic trioxide was also demonstrated by monitoring leukemic cells transfected with an LC3-GFP-expressing vector, for the presence of punctate LC3-GFP, using confocal microscopy ( Fig. 1F, right lower panel).
There has been previous evidence that different MAPK cascades are activated during treatment of cells with As 2 O 3 , including p38 MAPK (23), JNK (24), and ERK (25). Although several studies have previously examined the role of these As 2 O 3 -induced MAPK pathways in the regulation of apoptosis, the precise roles of As 2 O 3 -dependent MAPK signals in the induction of autophagy remain to be defined. Treatment of KT1 cells with As 2 O 3 resulted in induction of ERK1/2 phosphorylation/activation that reached maximal levels after 8 -16 h of treatment ( Fig. 2A). When the effects of MEK/ERK or JNK pharmacological inhibitors on the induction of As 2 O 3 -induced autophagy were determined, we found that the MEK inhibitor UO126 blocked up-regulation of LC3-II levels, whereas the JNK inhibitor SP600125 did not (Fig. 2B). Such effects were also seen in confocal microscopy experiments for the presence of punctate LC3-GFP (Fig. 2C) and were further confirmed in studies examining the effects of U0126 or S600125 on arsenic-dependent suppression of p62/SQSTM1 protein levels (data not shown). Consistent with this, when the upstream kinases of ERK, MEK1/2, were knocked down using specific siRNAs, up-regulation of expression of autophagy markers was also reversed (Fig. 2D). In parallel studies, we sought to determine the potential involvement of the AKT/mTOR pathway in the process. We compared As 2 O 3 -dependent autophagy in Akt1/2 double knock-out MEFs and parental MEFs. Although there was some increase in LC3-II levels in AKT knock-out MEFs, As 2 O 3 -inducible LC3-II expression was not enhanced further in the absence of Akt1/2 (Fig. 2E). Consistent with these findings, treatment of KT1 cells with rapamycin did not result in enhanced expression of LC3-II compared with cells treated in the absence of rapamycin (Fig. 2F). In other studies, when the effects of the reducing agents DTT (Fig. 2, G and H) or N-acetylcysteine (Fig. 2H) on arsenic-dependent autophagy in leukemic cells were determined, we found that there were no significant effects on As 2 O 3 -inducible LC3-II expression (Fig. 2G). To further establish the role of the MEK/ERK pathway in the induction of arsenic-dependent autophagy, experiments were performed in which the formation of AVOs was assessed by staining of KT1 cells with acridine orange. These studies demonstrated reversal of autophagic cell death in response to inhibition of MEK/ERK, but not JNK, activity (Fig. 3).
Altogether, our data established that, beyond apoptosis, arsenic trioxide induces autophagy of leukemia cells, raising the possibility that autophagic cell death plays a role in the generation of the suppressive effects of As 2 O 3 on leukemic hematopoiesis. To address this, experiments were performed to determine the effects of pharmacological or molecular targeting of elements of the autophagic machinery on the generation of the inhibitory effects of As 2 O 3 on leukemic progenitor cell growth.
In initial experiments, the effects of chloroquine, an antimalarial drug known to inhibit autophagy (26), were examined. KT1 cells were treated with As 2 O 3 in the presence or absence of chloroquine, and leukemic CFU-L colony formation was assessed in clonogenic assays in methylcellulose. Treatment with chloroquine partially reversed the suppressive effects of As 2 O 3 on KT1-derived CFU-L growth (Fig. 4A), suggesting involvement of autophagic cell death in the process.    Two key members of the family of autophagy-related proteins (ATG) are beclin 1, part of the class III PI3K complex, and Atg7, an E1 ubiquitin-like enzyme (26,27). Both of these proteins play key roles in the formation of the autophagosome and induction of autophagy (26,27). To determine their roles in the induction of As 2 O 3 -dependent antileukemic responses, beclin 1-or Atg7-specific siRNAs were used to knock down the corresponding proteins (Fig. 4B). As expected, siRNA-mediated knockdown of Atg7 resulted in decreased expression of LC3-II (Fig. 4C). Consistent with the findings of experiments using chloroquine, the suppressive effects of As 2 O 3 on leukemic progenitor colony formation by U937 cells in which beclin 1 or Atg7 was knocked down using specific siRNAs were confirmed (Fig. 4, D and E). Similar results were seen in experiments in which the effects of As 2 O 3 on leukemic progenitor colony formation derived from KT1 cells transfected with siRNAs against beclin 1 or Atg7 were examined (Fig. 4, F-H).
To further determine the significance of arsenic-induced autophagy in a more pathophysiologically relevant system, studies were performed to assess its role in the suppressive effects of As 2 O 3 on primitive leukemic progenitors from patients with AML. Treatment with As 2 O 3 suppressed the growth of primary CFU-L progenitors from AML patients in clonogenic assays in methylcellulose, and such effects were partially reversed by chloroquine (Fig. 5A). Moreover, siRNA-mediated knockdown of beclin 1 or Atg7 also resulted in reversal of the suppressive effects of As 2 O 3 on primary leukemic progenitors (Fig. 5, B and C), strongly suggesting that autophagy is an important mechanism by which As 2 O 3 generates antileukemic responses on primitive AML progenitors in vitro.

DISCUSSION
There is emerging evidence that autophagy, a caspase-independent cell death process, plays critical roles in the generation of antineoplastic responses and mediates caspase-independent malignant cell death (26 -28). Depending on the cellular context and/or initiating stimulus, autophagy may act as a protective mechanism for malignant cells or exhibit opposing effects and promote generation of antineoplastic responses. For instance, there is evidence that autophagy promotes resistance of breast cancer cells to the monoclonal anti-HER2 antibody trastuzumab (29). Similarly, hypoxia-induced autophagy leads to chemoresistance of hepatocellular carcinoma cells (30), whereas inhibition of autophagy appears to enhance the effects of BCR-ABL kinase inhibitors on Ph(ϩ) leukemia cells, including primary chronic myelogenous leukemia stem cells (31). On the other hand, there is evidence that defects in autophagic responses are associated with tumorigenesis, whereas various proteins that induce autophagy are known to exhibit tumor suppressor activities (reviewed in Ref. 26). Notably, autophagic cell death has been implicated as a mechanism by which certain antineoplastic agents generate their antitumor activities. Previous work has established that autophagy mediates cell death of acute lymphoblastic leukemia cells by dexamethasone (32,33), promotes differentiation of K562 erythroleukemia cells (34), promotes cell death by histone deacetylase inhibitors in chondrosarcoma cell lines (35), and may constitute a key mechanism by which rapamycin and other mTOR inhibitors promote generation of antitumor responses (36,37).
Arsenic trioxide has important anti-leukemic effects in vitro and in vivo and is known to mediate its effects via its ability to induce apoptosis (1, 3). A key mechanism for such induction of apoptosis is the generation of reactive oxygen species, resulting in the induction of pro-apoptotic signals and engagement of the caspase cascade. As 2 O 3 -dependent targeting of the thioredoxin system that plays an important role in intracellular redox reactions appears to be key in such arsenic-dependent apoptosis (1,38). Recent evidence has also established that activation of the JNK/MAPK pathway is a cellular event essential for induction of apoptosis of acute promyelocytic leukemia cells (24).
Although much is known on the mechanisms of induction of apoptosis by arsenic trioxide, very little is known on the potential involvement of autophagy as a regulator of arsenic-dependent neoplastic cell death. Previously, there has been some evidence that beyond a caspase-dependent cell death, arsenic can also induce cell death under certain circumstances in a caspaseindependent manner (39). Interestingly, in the same study (39), it was shown that in bone marrow samples from myeloma patients, arsenic trioxide was inducing a caspase-independent mechanism in the majority of cases (39), suggesting an important role for caspase-independent events in the anti-myeloma effects of As 2 O 3 . Nevertheless, the precise mechanisms by which such events occur are not well understood.
In this study, we directly examined the ability of As 2 O 3 to induce autophagic cell death in acute leukemia cells and the functional consequences of such induction. Our data establish that As 2 O 3 is a potent inducer of autophagy in leukemic cells and that such induction is dependent on activation of the MEK/ ERK pathway but unrelated to engagement of JNK. In experiments using a pharmacological inhibitor of autophagy (chloroquine), we found that the suppressive effects of arsenic trioxide on primary leukemic progenitors are in part mediated via induction of autophagic cell death. Such a critical role for autophagy in the generation of the anti-leukemic effects of As 2 O 3 was further established by experiments in which key elements of the autophagic pathway, such beclin 1 and Atg7, were knocked down using specific siRNAs.
The role of autophagy in the induction of As 2 O 3 -dependent antineoplastic effects in different types of malignant cells remains to be precisely determined in future studies. It is of particular interest that in malignant glioma cells, arsenic induces autophagic cell death but not apoptosis (40). In that case, a novel mechanism involving up-regulation of mitochondrial cell death protein BNIP3 has been implicated (41). There has been also some previous evidence that in addition to apoptosis, arsenic trioxide induces autophagy in the human Tlymphocytic leukemia cell line Molt-4 (42), whereas another arsenic compound, sodium arsenite, promotes autophagy of human uroepithelial cells (43). Thus, it is possible that beyond AML, autophagy may be a key or alternative mechanism for the generation of the antitumor effects of As 2 O 3 in other malignant cell types as well.
Our findings that autophagy plays a critical role in the generation of the antileukemic effects of As 2 O 3 in primary hematopoietic progenitors from AML patients may prove to have important clinical-translational implications. There is emerging evidence that treatment of malignant cells with arsenic trioxide results in negative feedback activation of cellular pathways that counteract apoptosis. In particular, activation of the p38 MAPK and its effectors Msk1 and Mnk1 (15,16,23,44) result in generation of antiapoptotic responses in AML cells (44 -46), as well as in multiple myeloma cells (45). Similarly, arsenic-induced engagement of the AKT/mTOR pathway appears to exhibit negative regulatory effects on the generation of the suppressive effects of As 2 O 3 on leukemic hematopoiesis (18,46). Such studies have suggested that combinations of As 2 O 3 with inhibitors of MAPK or mTOR pathways may provide a novel approach to overcome leukemic cell resistance in certain cases. The findings of this study, implicating autophagy as a mechanism for the induction of the suppressive effects of arsenic on primitive leukemic progenitors, raise the potential that agents that promote autophagy could act as sensitizers of leukemic cells to the suppressive effects of As 2 O 3 . Small molecules that modulate autophagy in other systems have been previously identified (47), and efforts to design and develop similar small molecule compounds that could be used in combination with As 2 O 3 may prove important in the future in overcoming leukemic cell resistance in vitro and possibly in vivo.