Combined effect of all-trans retinoic acid and arsenic trioxide in acute promyelocytic leukemia cells in vitro and in vivo

Acute promyelocytic leukemia (APL) is a specific type of acute myeloid leukemia characterized by the t(15;17) translocation that fuses the PML gene on chromosome 15 to the retinoic acid receptor α (RARα) gene on chromosome 17 to form the fusion gene and leukemogenic protein PML-RARα.1,2 The specific sensitivity of APL cells to all-trans retinoic acid (tRA)-induced differentiation has been exploited to achieve a 90% remission rate with a projected 60% to 70% cure of patients with APL when combined with chemotherapy.3-7 However, the disease may relapse with resistance to further tRA and chemotherapy treatment.8 The recent discovery that treatment with As₂O₃induces durable remission in APL patients in relapse after tRA or chemotherapy has provided a novel therapy for APL patients.9-11 However, the reported chronic toxicities and carcinogenicity of As₂O₃ has hampered its acceptance as a first-choice drug,12 even though limited side effects were found in relapsed APL patients successfully treated with As₂O₃.10,11 Consideration of the facts that the toxicity of As₂O₃ is dose dependent and reversible and that small amounts of As₂O₃ have been used in traditional Chinese medicine suggests that it may be possible to use As₂O₃ as a first-choice drug if low concentrations prove effective.

Arsenic trioxide induces both apoptosis and partial differentiation in APL cells in vitro and in vivo.11,13-15 More than 0.5 μM As₂O₃ is required for induction of H₂O₂-mediated apoptosis in NB4 cells derived from t(15;17) APL.16 The H₂O₂concentration in cells is regulated mainly by selenium-dependent glutathione peroxidase (Gpx) activity. Because standard tissue culture conditions are deficient in selenium, the H₂O₂-removing activity of Gpx is low.17 By increasing selenium to physiologic concentration in the tissue culture medium the activity of Gpx increases; consequently, the H₂O₂ content is lowered and the apoptotic activity of As₂O₃ (at low concentration) is diminished.16 In contrast to NB4 cells, APL cells in primary culture are less sensitive and variable to 1 μM As₂O₃-induced apoptosis.11,13,14,18 These observations suggest that differentiation-induced terminal cell division rather than direct apoptosis may play a major role in As₂O₃-induced remission in patients with APL.

Patients with APL expressing PML-RARα by reverse transcriptase-polymerase chain reaction (RT-PCR) are responsive to As₂O₃, whereas patients with undetectable PML-RARα are not.19 This observation, along with the fact that As₂O₃ at concentrations as low as 0.1 μM degrades PML-RARα,13 suggests that this degradative process contributes to the therapeutic effect of As₂O₃. Because PML-RARα protein has been shown to function as a dominant-negative transcriptional repressor of RARα, As₂O₃-induced degradation of PML-RARα may allow physiologic tRA concentrations to trigger APL cell differentiation in vivo. tRA and As₂O₃ appear to degrade PML-RARα through different pathways,20-22suggesting that the combination may be synergistic therapeutically. Because controversial effects have been reported on tRA and As₂O₃ interactions,14,23 the present studies were developed to define their combined effects on apoptosis and differentiation in tRA-sensitive and tRA-resistant APL cells at the molecular and cellular level.

The tRA was purchased from Sigma Chemical Co (St. Louis, MO). As₂O₃ (0.1%) was provided by Dr Ting-tong Zhang (Harbin, China). Anti-RARα (F region) antibody RPα(F) was kindly provided by Dr P Chambon (Strasbourg, France).

The NB4 t(15;17) cells were obtained from Dr M. Lanotte (Paris, France)24 and tRA-resistant NB4 derivatives, R4 and MR-2, were developed as reported previously.25HL-60/Res cells were obtained from Dr R. Gallagher (Bronx, NY).26 The cells were cultured in RPMI-1640 medium supplemented with 100 U/mL penicillin, 100 μg/mL streptomycin, 1 mMl-glutamine, and 10% fetal bovine serum. Cells in logarithmic growth were seeded at 1 × 10⁵ cells/mL for studies done in duplicate and repeated at least 3 times. The nitroblue tetrazolium (NBT) reduction assay was performed as previously reported.27 

Protein extracts (50 μg) prepared with RIPA lysis buffer (50 mM Tris-HCl, 150 mM NaCl, 0.1% sodium dodecyl sulfate [SDS], 1% NP-40, 0.5% sodium deoxycholate, 1 mM PMSF, 100 mM leupeptin, and 2 μg/mL aprotinin, pH 8.0) were separated on an 8% SDS-polyacrylamide gel and transferred to nitrocellulose membranes. The membranes were stained with 0.2% Ponceau S red to ensure equal protein loading and transfer. After blocking with 5% nonfat milk, the membranes were incubated with anti-RARα antibody (Dr. P Chambon). The immunocomplex was visualized by enhanced chemiluminescence (ECL kit, Amersham, Parsipanny, NJ).15 

Total RNA was isolated with messenger RNA (mRNA) isolation kit (Gentra, Minneapolis, MN) from 10⁶ cells. Twenty micrograms RNA was sized fractionated on a 1.2% agarose-2.2 M formaldehyde gel, transferred to hybrid-N+ membrane (Amersham) in 20 × standard sodium citrate (SSC), and UV-crosslinked (Stratalinker; Stratagene, La Jolla, CA). Whole length complementary DNA (cDNA) of human myeloperoxidase (MPO),28 RARβ, tissue transglutaminase II (TTGII),29 RIG-E,30Bfl-1/A1,31 and a commercial GAPDH cDNA probe (Ambion, Austin, TX)32 were used for probing. The probes were labeled with a ³²P dCTP by random priming to a specific activity of 0.5 to 1 × 10⁹ cpm/μg. The membranes were prehybridized for 4 hours at 42°C in 50% formamide, 6 × sodium chloride, sodium phosphate, EDTA (SSPE), 5 × Denhardt reagent, and 0.2 mg/mL salmon sperm DNA, and hybridized with radiolabeled probe. Membranes were then washed twice in 6 × SSC, 0.1% SDS, followed by a stringent wash with 0.2 × SSC, 0.1% SDS at 65°C.32 

Female severe combined immunodeficient (SCID) mice, 4 to 8 weeks old, were bred and maintained under pathogen-free conditions. The NB4 SCID mouse ascites model was established as in our previous report.33 SCID mice were inoculated with 1 × 10⁶ NB4 cells at passage less than 12. Mice injected with NB4 cells were allowed to develop a tumor burden for 7 days before treatment with various drug regimens. As₂O₃ was diluted appropriately with phosphate-buffered saline (PBS), pH 7.4, and injected intraperitoneally in tumor-bearing animals. Control (untreated) animals received PBS. The main end point of this study was survival, with each treatment group containing 10 to 20 tumor-bearing animals. The survival of all mice was followed, and a median survival time (MST) was calculated. A percent increase over the life span of control animals was calculated by dividing the MST of a treatment group by the MST of control group. Statistical analysis was performed by a log-rank test and the Wilcoxon test at the 5% significance level.

Studies were designed to clarify the observation that As₂O₃ treatment induces apoptosis with minimal differentiation of NB4 cells, whereas it induces partial differentiation but not apoptosis in primary cultures of APL cells. Treatment of APL cells in primary culture for 6 days with 0.1 to 1 μM As₂O₃ induced 20% to 30% differentiation (Figure 1) and less than 10% apoptosis (as measured by fluorescence microscopic morphologic change, data not shown). Under standard culture conditions, As₂O₃ (0.5 μM) did not induce differentiation in NB4 cells (Figure 2B). As₂O₃ slightly enhanced tRA-induced differentiation when physiologic concentrations of tRA were used (1-10 nM) but decreased tRA-induced differentiation when pharmacologic tRA concentrations were used (0.1-1 μM) (Figure 2B). There was no significant effect of As₂O₃ on cell number (Figure 2A) or cell viability (> 90% cell viability was observed in all groups). However, the differentiation was less than when tRA was used at therapeutic doses (1 μM). To create in vitro conditions more relevant to in vivo for evaluating the effect of As₂O₃ on NB4 cells, NB4 cells were precultured in the presence of 100 nM selenium (physiologic concentration) to induce Gpx activity.16 In NB4 cells grown in the presence of selenium for several passages, As₂O₃treatment (1 μM), which by itself did not induce significant apoptosis16 or differentiation (Figure3), enhanced physiologic (10 nM) tRA-induced differentiation 2-fold (Figure 3).

Both tRA and As₂O₃ degrade PML-RARα, whereas RARα levels may be decreased by tRA and not As₂O₃ treatment. As₂O₃degrades PML-RARα without an accumulation of an intermediate, whereas tRA treatment results in the accumulation of a truncated product termed ΔPML-RARα (Figure 4A). The cleavage product is not detectable in NB4 cells treated with low concentrations of tRA alone or when combined with As₂O₃ but is formed following therapeutic (pharmacologic) concentrations of tRA, even in the presence of As₂O₃. As₂O₃ partially inhibits tRA induction of Bfl-1/A1, RIG-E, RARβ, TTGII, and the down-regulation of MPO expression, which are known to be modulated by tRA in NB4 cells (Figure4B). The inhibition by As₂O₃ could be abrogated by increasing the concentration of tRA so that at a concentration of 1 μM the expression of these genes was only minimally diminished (Figure 4B).

MR-2 and R-4 are NB4 subclones that are resistant to tRA-induced differentiation.25 Treatment with tRA at pharmacologic concentrations in these cells does not induce differentiation as determined by NBT reduction or the expression of retinoic acid responsive genes and does not cleave PML-RARα (Figure5). On the other hand, As₂O₃ completely degrades PML-RARα in these cells but does not induce cell differentiation or the expression of retinoic acid target genes. However, combining As₂O₃ with tRA induced differentiation and the expression of TTGII and Bfl-1/A1 in the resistant cells as in the parental sensitive NB4 cells. These data support the concept that tRA resistance can be at least partially overcome by As₂O₃-induced degradation of PML-RARα, diminishing the dominant effect of PML-RARα over RARα. Interestingly, RARα levels appear to be decreased in the presence of tRA or tRA with or without As₂O₃, whereas As₂O₃ had a minimal effect (Figure 5C). In contrast, As₂O₃ does not overcome tRA differentiation resistance in HL-60/Res cells with mutated RARα.

The induction of differentiation in APL cells treated with combined As₂O₃ and tRA is dose dependent and may be antagonistic or additive (Figure 2). Similar considerations are necessary when evaluating the effect of the drug combination on the induction of apoptosis. As₂O₃-induced apoptosis in NB4 cells is diminished by pretreatment with tRA and is time related (the longer the pretreatment with tRA the less apoptosis is induced by As₂O₃ treatment) (Figure6A). This antagonism is not demonstrated in the tRA-resistant R4 cell line. One possible explanation is that Bfl-1/A1 expression, a member of the Bcl-2 family, is induced by tRA in NB4 cells despite a fall in Bcl-2 protein level and may account for the inhibition of As₂O₃-induced apoptosis (Figure6B,C). In contrast, Bfl-1 is not induced by tRA in R4 cells. Thus, despite maintenance of Bcl-2 protein, apoptosis is significant following As₂O₃ treatment. Thus, the induction of Bfl-1 may contribute to a greater extent than Bcl-2 to the inhibitory effect of tRA on As₂O₃-induced apoptosis.

The in vivo effect on survival in SCID mice bearing NB4 cells following treatment with tRA or As₂O₃ given alone, sequentially, or simultaneously is shown in Table1. As₂O₃ (8 mg/kg) or tRA (10 mg/kg), administered every other day for a total of 8 courses, increased the life span of the host to a similar extent (35%-39%). Sequential treatment of SCID mice with either agent first for 8 courses followed by the other for an additional 8 courses resulted in an additive outcome (70%-80% increase in life span), regardless of the sequence of treatment. However, combined treatment with As₂O₃ and tRA caused toxicity with weight loss and early treatment-associated death.

It has been reported that tRA induces differentiation of APL cell lines and APL blasts in vivo and in vitro,3,34 whereas As₂O₃ induces apoptosis in both tRA-sensitive and tRA-resistant APL cell lines in vitro.13,14 However, we and others have shown that As₂O₃ induces NB4 cell apoptosis through a redox-mediated pathway16 that is PML-RARα independent15,35 and is diminished by either increased serum,23 selenium supplementation,16 or addition of granulocyte-macrophage colony-stimulating factor.36 Therefore, the dramatic in vitro induction of apoptosis by As₂O₃ may be minimized by physiologic growth and antiapoptotic factors that exist in vivo. Indeed, although As₂O₃ is capable of inducing apoptosis in primary APL cell cultures,13,14recent reports indicate that primary cultured human APL cells18 and transgenic mice APL cells37 are less sensitive to 1 μM As₂O₃-induced apoptosis than serially cultured NB4 cells. These considerations prompt the question of which is the main mechanism underlying As₂O₃-induced remission in APL patients. We report here, in agreement with the results of others,18,37that primary cultured APL cells undergo partial differentiation in response to 0.1 to 1 μM As₂O₃ (Figure 1). Although As₂O₃ does not induce differentiation in NB4 cells (Figure 2), even in presence of selenium (Figure 3), our results show that it enhanced differentiation induction by near-physiologic tRA concentrations (1 to 10 nM tRA) (Figures 2and 3). Note that in Figure, 3 As₂O₃ was tested at 1 μM, which normally induces NB4 cell apoptosis, but with the addition of selenium this concentration caused differentiation (Figure3) instead of apoptosis (data not shown). Taken together, these findings suggest that the As₂O₃ potentiation of physiologic tRA-induced differentiation may account in large part for the induction of APL remission by As₂O₃. Such potentiating effect of As₂O₃ may be explained by As₂O₃-induced PML-RARα degradation, as previously suggested by others.38 

We found that As₂O₃ potentiated NB4 cell differentiation by 1 to 10 nM tRA but suppressed differentiation by 0.1 to 10 μM tRA (Figure 2; differentiation was scored using the NBT reduction assay). What might be the basis for this biphasic effect of As₂O₃? As₂O₃ induced extensive PML-RARα degradation at all concentrations of tRA tested (Figure 4A, right), indicating that the explanation lies elsewhere. When we monitored the tRA induction of differentiation target genes,39 we found that the tRA dose dependence for gene induction (in absence or presence of As₂O₃, Figure 4) did not agree well with the NBT reduction score. For instance, in contrast to a dose-dependent increase in NBT reduction (Figure 2), the extent of gene modulation by 10 nM and 1 μM tRA was about the same (Figure 4). In addition, As₂O₃ inhibited gene induction by 10 nM tRA but not 1 μM tRA (Figure 4). We do not understand the basis for this discrepancy between the 2 differentiation assays. However, as explained below, we have a hypothetical explanation for the observed relationship between tRA dose and target gene modulation.

The tRA treatment of NB4 cells results in partial PML-RARα proteolysis, with formation of a stable ∼90-kd product, which we refer to as ΔPML-RARα (Figure 4). This is in marked contrast to As₂O₃ treatment, which results in extensive PML-RARα proteolysis without accumulation of ΔPML- RARα (Figure4). On the basis of its recognition by an anti-C terminal RARα antibody and its apparent size, we estimate that ΔPML-RARα may lack the RING finger and B-boxes of the PML moiety, while retaining the coiled-coil domain. PML, through its RING finger, interacts with key regulatory proteins targeting them to nuclear bodies, and these events are disrupted by PML-RARα,40,41 suggesting that the predicted loss of the RING finger may allow the resumption of normal PML function (from the remaining allele). The coiled coil domain, which we predict is retained in ΔPML-RARα, has been shown to be essential for the PML-RARα-mediated enhancement of tRA-induced differentiation in hematopoietic cell lines.42 Therefore, we hypothesize that ΔPML-RARα does not act as a dominant-negative receptor and that it functions instead as wild-type RARα, contributing positively to tRA-induced cell differentiation. This is a testable hypothesis that offers an explanation for the observed effects of tRA on differentiation target gene expression. Thus, Figure 4 shows that there is a marked correlation between the amount of ΔPML-RARα formed and the extent of target gene modulation in response to different tRA doses (in absence or presence of As₂O₃), suggesting that ΔPML-RARα plays a positive role in target gene modulation. Moreover, Figure 5 demonstrates that in tRA-resistant R4 and MR-2 cells, ΔPML-RARα was not formed and cell differentiation (NB4 reduction and target gene induction) was not achieved. As₂O₃ treatment could partially overcome tRA resistance in R4 and MR-2 cells, presumably due to the induction of extensive PML-RARα breakdown (Figure 5). In these instances, differentiation by 0.1 μM tRA was attained without a detectable increase in ΔPML-RARα formation.

Although As₂O₃ induction of apoptosis may be diminished by conditions existing in vivo, apoptotic cells have been found in APL patients (whether apoptosis is a first event or follows As₂O₃-induced differentiation is not known).11 Because As₂O₃ is used to treat patients with APL who have been or will be treated with tRA, it is important to evaluate As₂O₃-induced apoptosis after pretreatment with tRA in sensitive or resistant APL cells. tRA pretreatment for 48 hours decreased As₂O₃-induced apoptosis in tRA-sensitive NB4 cells, but not in tRA-resistant R4 cells (Figure 6). Consistent with previous reports,43,44 we found that Bcl-2 was degraded by tRA in NB4 cells but not in R4 cells (Figure 6B). The degradation of Bcl-2 induced by tRA would be expected to sensitize NB4 cells to As₂O₃-induced apoptosis; however, as noted above, tRA was inhibitory (Figure 6A). Thus, other antiapoptotic genes modulated by tRA may compensate for the tRA-induced Bcl-2 degradation and protect against cell death. Several antiapoptotic genes identified by differential display, including Bfl-1 and Dad1, are induced by tRA in NB4 cells.39,45 We found that Bfl-1/A1 was rapidly and highly induced by tRA in NB4 cells, but not in R4 cells (Figure 6C). Bfl-1/A1 transfected cells have been shown to be resistant to several apoptotic inducers.31,46,47 Thus, tRA induction of Bfl-1 may in part account for the tRA inhibition of As₂O₃-induced apoptosis in NB4 cells.

Studies using the SCID mouse NB4 cell ascites model (Table 1) suggest an improved survival when the combination of tRA and As₂O₃ is used sequentially as compared to either agent alone. These data are not conclusive because the sequential treatments were longer in duration than when individual agents were used; however, they confirm the results obtained in transgenic mice models of APL.37 Sequential treatment where tRA is used prior to As₂O₃ would be predicted to be antagonistic according to the data in Figure 6A. However, at the concentrations used, the in vivo studies did not demonstrate this antagonism. There was toxicity and decrease in survival when tRA and As₂O₃ were used simultaneously at maximal therapeutic concentrations. This may be due to the high concentration of As₂O₃ and tRA used as compared to other in vivo animal studies.37 Therefore, we suggest that reducing the concentrations of tRA and As₂O₃ may allow therapeutic efficacy with tolerable toxicity. Note that, in vitro, differentiation induction by the combination of low tRA and low As₂O₃ (10 nM and 0.5 μM, respectively) was optimal, that is, could not be improved by increasing either the tRA or the As₂O₃concentration (Figures 2 and 3). Clinical studies using these in vitro predicted dose modifications when using combination tRA and As₂O₃ are required to determine efficacy in the treatment of tRA-resistant patients in relapse or toxicity in de novo patients.

We thank the following investigators for sharing their reagents: Pierre Chambon (anti-RARα antibody and mRARβ plasmid); H. Phillip Koeffler (mMPO plasmid); Peter Davies (TTGII plasmid); G. Chinnadurai (Bfl-1/A1 plasmid). We also appreciate the critical reading by Drs Geoge Acs and Rafael Mira-y-Lopez.