Background
Arsenic trioxide (As
2O
3) shows impressive efficacy in the treatment of patients with acute promyelocytic leukemia (APL) [
1‐
4]. As
2O
3 induces clinical remission in APL patients by multiple mechanisms [
5]. As
2O
3 promotes cell differentiation at low concentrations [
6,
7], whereas it induces apoptosis at higher concentrations [
8]. The high sensitivity of the APL cell line NB4 to As
2O
3-induced cytotoxicity is associated to its low content of reduced glutathione (GSH) and increased production of reactive oxygen species (ROS) [
9‐
11]. Although most studies have been focused on the APL, As
2O
3 could be beneficial against various hematopoietic malignancies and solid tumors [
12]. Moreover, we have shown that As
2O
3 also possesses immunomodulatory properties, and might be a therapeutic agent for autoimmune diseases. Indeed, As
2O
3 selectively eliminates the pathogenic B220-expressing double negative (DN) CD4
-CD8
- T cells that accumulate in autoimmune MRL/
lpr mice due to the
lpr mutation of the death receptor Fas [
13,
14].
In normal murine and human T cells, CD4
+ and CD8
+ effector T cells massively induce the expression of transmembrane tyrosine phosphatase B220 before undergoing apoptosis by the Fas/Fas ligand (FasL) pathway [
15,
16]. In Fas-deficient mice and patients, CD4
+ and CD8
+ effector T cells also express the B220 molecules at their surface, but then they downregulate their CD4 or CD8 molecules while maintaining B220 plasma membrane expression. B220 (or CD45RABC) is one of the five isoforms of the transmembrane tyrosine phosphatase CD45 found on lymphocytes. CD45 isoforms are generated by cell-type and activation-state specific alternative splicing of exons 4/A, 5/B, and 6/C encoding domains at the NH
2-terminus. Naive T cells express high molecular weight CD45 isoforms (CD45RA or CD45RB) containing the A domain in humans or the B domain in mice whereas effector/memory T cells expressed the low molecular weight isoform CD45RO lacking extracellular domains A, B and C. All CD45 isoforms share the same intracellular region, which contains two phosphatase domains. Although the function of each isoform remains unknown, it is well established that CD45 phosphatase activity is crucial for lymphocyte development, and antigen and cytokine receptor signaling [
17‐
19]. CD45 might also regulate apoptosis of T and B lymphocytes [
20‐
22].
In this study, we found that murine (EL-4, BW5147, L1210) and human (Jurkat, CD45-deficient Jurkat variant, HPB-ALL) leukemic T-cell lines dramatically differed in their sensitivity to As
2O
3-induced cell death. In contrast with previous findings in APL cell line NB4 [
9,
10], these differences in As
2O
3 sensitivity are independent of intracellular GSH content and O
2- production. Unexpectedly, we found that As
2O
3 differently induced B220 cell surface expression in the leukemic T-cell lines in a dose- and time-dependent manner. Moreover, the levels of B220 expression correlated with the sensitivity of these T-cell lines to As
2O
3. Induction of B220 membrane expression by As
2O
3 treatment is reminiscent of that observed on antigen-activated normal T-cell blasts before undergoing apoptosis [
15,
16]. Therefore, the leukemic T-cell lines were activated with calcium ionophore A23187, which triggers both cell activation and cell death. Calcium ionophore A23187 also induced B220 expression and cell death, but with reverse efficiencies in the leukemic T-cell lines compared to As
2O
3. In addition, T-cell lines treated with A23187 most probably died by an activation-induced cell death mechanism since the T-cell activation marker CD69 is expressed before B220 expression and cell death. In contrast, CD69 was not detected on As
2O
3-treated cells, indicating that B220 expression occurs independently of leukemic T-cell activation. Surprisingly, we found that B220 is expressed constitutively on L1210 T cells. L1210 cells were highly sensitive to A23187 treatment whereas they were highly resistant to As
2O
3 cytotoxicity, indicating that the constitutive high-level expression of B220 did not favor cell death triggered by As
2O
3. B220 induction on the T-cell lines after treatment with As
2O
3 or calcium ionophore A23187 strictly correlates with sensitivity to cell death, emphasizing the role of B220 as a proapoptotic factor. However, our data indicate that As
2O
3 and A23187 trigger B220 induction and cell death through different upstream signaling pathways. Different signaling pathways, such as the c-Jun NH
2-terminal kinase, have been implicated as mediators of the cytotoxic effects of As
2O
3 in the APL cell line NB4 [
23]. Here, we show that high induction of B220 expression on leukemic T cells is a determining factor leading to As
2O
3-triggered cell death. Thus we hypothesize that B220 might play a checkpoint role in death pathways.
Discussion
In APL-derived NB4 cells, As
2O
3 triggers apoptosis at concentrations of 0.5 to 2 μM [
8]. The cytotoxic properties of As
2O
3 are not restricted to APL, and As
2O
3 induces apoptosis in various types of hematopoietic and solid tumors [
12]. However, hematologic tumor cells vary considerably in their sensitivity to As
2O
3. To gain insight into the mechanisms underlying the As
2O
3 sensitivity of malignant T cells, we firstly selected two human (Jurkat and HPB-ALL) and three murine (EL-4, BW5147 and L1210) T-cell lines for their marked differences in sensitivity to As
2O
3 cytotoxicity over a large range of doses (i.e. 1 μM to 20 μM). Thus, 50% of EL-4 cells are killed at a clinically relevant concentration of about 1–2 μM As
2O
3, whereas concentrations of ~2.5 μM As
2O
3 for BW5147, ~6 μM As
2O
3 for Jurkat and ~12–15 μM As
2O
3 for L1210 and HPB-ALL T-cell lines are required to kill approximately 50% of the cells. Using this tumor panel, we have shown that: 1) differences in the intracellular levels of GSH and O
2- are not sufficient to account for their differences in As
2O
3 sensitivity; 2) transmembrane tyrosine phosphatase B220 is induced on EL-4, BW5147, Jurkat and HPB-ALL T-cell lines, but not APL-derived NB4 cells, upon treatment with As
2O
3 in a dose and time dependent manner; 3) the degree of B220 induction on the T-cell lines is strongly correlated with the sensitivity to As
2O
3 cytotoxicity; 4) surprisingly, B220 is constitutively expressed on the L1210 T-cell line; 5) membrane-bound HSP70, known to induce antitumor immunity, is upregulated by As
2O
3 in parallel with B220 induction; 6) initiator caspases 8 and 9 are activated by As
2O
3 in the T-cell lines where this activation parallels B220 induction, but not in L1210 cells; 7) As
2O
3 represses nuclear translocation of NF-κB p50 in a dose dependent manner; 8) FasL upregulation by As
2O
3 is found on Fas-negative EL-4 cells only, indicating that caspase 8 activation is most probably independent of the Fas/FasL pathway [
34]. However, the absence of FasL on some T-cell lines (either human or murine) might be due to the rapid shedding of FasL by protease activities induced by As
2O
3 or calcium ionophore A23187, as we have reported for TNF-α and CD62L upon treatment with Ionomycin or ATP [
35].
Contradictory data have been published on the efficacy of As
2O
3 treatment against tumors belonging to the lymphoid lineages. Evidence for a pro-apoptotic effect of As
2O
3 against human malignant T- and B-cell lines [
36], cutaneous T-cell lymphoma [
37] or HTLV-I-associated adult T-cell leukemia [
38] was provided by in vitro studies. However, it has been shown, in a multi-institution phase II study, that As
2O
3 exhibits limited efficacy against lymphoid malignancies, even though the patients received ascorbic acid in addition to As
2O
3[
39]. The authors of this clinical study expected that agents such as ascorbic acid, which depletes intracellular GSH, could potentiate arsenic-induced apoptosis, since it has been described previously [
10] that the sensitivity of cells to As
2O
3 are inversely correlated with their intracellular GSH content. As
2O
3 interacts with sulfhydryl (SH) groups of biologically active molecules. The binding of As
2O
3 to the SH group of GSH could cause a drastic decrease in the capacity to scavenge ROS in cells with a low basal GSH content, resulting in overproduction of intracellular ROS that could trigger cell apoptosis. However, our present study shows that EL-4 and BW5147 T-cell lines, which were the most sensitive to As
2O
3-induced cytotoxicity among the cells lines studied (IC
50 of approximately 1 to 2.5 μM) had the highest baseline GSH content (Figure
2). Likewise, L1210 and HPB-ALL T cells that were significantly more resistant to As
2O
3 cytotoxicity (IC
50 of approximately 12 to 15 μM) had significantly lower baseline GSH content than EL-4 and BW5147 cells (Figure
2). In the presence of As
2O
3, intracellular GSH content was differently modulated in the T-cell lines, but without any correlation with their As
2O
3 sensitivity. Thus, GSH level in the highly sensitive NB4 cell line was unaffected by the dose of 1 μM As
2O
3, whereas it was markedly decreased in the BW5147 T-cell line and also, although to a lesser extent, in the resistant Jurkat cells (Figure
2). Because intracellular GSH content did not appear to be a key factor in determining As
2O
3 sensitivity of these five leukemia T-cell lines, we further explored the evolution of O
2- production upon treatment with As
2O
3. The five T-cell lines showed a great heterogeneity in levels of intrinsic O
2- production (Figure
2), which, interestingly, were 2 to 8 times higher than in APL-derived NB4 cells. As expected, the levels of O
2- were significantly increased in APL-derived NB4 cells in the presence of As
2O
3. In T-cell lines, the situation was more complex. While the levels of O
2- increased significantly, in Jurkat, L1210, and to a lesser extent HPB-ALL T cells treated with As
2O
3, they remained unchanged in EL-4 and BW5147 T cells. Therefore, the changes of O
2- production upon treatment with As
2O
3 are not correlated with the differences in As
2O
3 sensitivity that we observed among the five T-cell lines. In contrast to other tumor cells [
5], neither the intracellular GSH content nor the production of O
2- had any decisive effect on As
2O
3-induced apoptosis of the five T-cell lines studied. Therefore, our findings suggest that the depletion of leukemic T cells in their intracellular GSH content by ascorbic or butyric acid are not necessarily relevant to potentiate the cytotoxic effect of As
2O
3. Furthermore, it has been reported that the expression level of aquaglyceroporin (AQP)9, a transmembrane protein that controls arsenic transport, correlated positively with As
2O
3-induced cytotoxicity in myeloid and lymphoid leukemia cell lines [
40]. In contrast, the overexpression of AQP9 in melanoma cells significantly increased the resistance to arsenite-induced apoptosis [
41]. These reports prompted us to determine the levels of AQP9 mRNA by semi-quantitative RT-PCR in EL-4, BW5147, L1210, Jurkat and HPB-ALL T-cell lines, and in APL-derived NB4 cells. However, we did not find any correlation between the expression levels of AQP9 mRNA and the As
2O
3 sensitivity in the T-cell lines (unpublished data). Therefore, our data strongly suggest the existence of additional factors determining the sensitivity of T cells to As
2O
3 cytotoxicity. Herein, we have hypothesized that phosphatase B220 could be such a factor, since we have shown previously that treatment with As
2O
3 of autoimmune MRL/
lpr mice selectively eliminates pathogenic B220-expressing T cells in vivo [
13,
14].
Normal effector T cells entering apoptosis after repeated activation by their antigen express the tyrosine phosphatase B220 on their surface [
15,
16,
42], suggesting a role for B220 induction in the transition from activation to apoptosis. On the other hand, B220 is the isoform of CD45 predominantly expressed on pathogenic DN T cells from patients and mice with a deficiency in the death receptor Fas, or its ligand FasL. FasL-deficient mice (
gld mutation) with only one functional CD45 allele (
gld/
gld, CD45+/-) display a strong reduction in the pathogenic DN T-cell population [
43], suggesting that CD45 is an important regulator of T-cell apoptosis or a survival factor for T cells. Whether and how B220 expression on T cells regulates signaling for death or survival remains unknown. The induction of B220 and FasL (in EL-4 cells) as well as the activation of initiator caspase-8 in As
2O
3-treated leukemic T-cell lines were reminiscent of apoptosis in normal effector T cells triggered by repeated antigenic stimulation. Therefore, leukemic T-cell lines were activated with calcium ionophore A23187, a well-known trigger for T-cell activation and death. Importantly, we found that As
2O
3 and calcium ionophore A23187 have opposite efficiencies on the T-cell lines, evidenced by the induction of B220, activation marker CD69 and membrane-bound HSP70 expression, and cell death (data summarized in Table
1). While both B220 expression and cell death were massively induced in EL-4 cells after treatment with As
2O
3, they were only slightly induced after treatment with calcium ionophore A23187. The complete reverse situation was observed in HPB-ALL cells, indicating that A23187 and As
2O
3 had opposite effects on B220 expression and cell death on the same leukemic T-cell panel. Moreover, treatment with A23187, but not As
2O
3, induced the activation marker CD69 on T-cell lines before B220 expression and cell death, indicating that A23187, but not As
2O
3, kills the T-cell lines by an activation-induced cell death mechanism.
HSP70 is overexpressed in various cancer cells [
44]. HSP70 inhibits apoptosis by modulating multiple events within apoptotic pathways, which might promote cancer development [
45,
46]. However, a tumor-specific plasma membrane form of HSP70 has been described [
44,
47], which facilitates tumor rejection by the immune system [
27,
47‐
49]. In the present study, we found a strong upregulation of membrane-bound HSP70 by As
2O
3 treatment, but not by the calcium ionophore A23187. This upregulation of HSP70 by As
2O
3 strictly paralleled the induction of B220 on EL-4, BW5147, Jurkat and HPB-ALL T-cell lines (Table
1). Consequently, As
2O
3-sensitive EL-4 cells expressed both high levels of HSP70 and B220, whereas low expression levels of HSP70 and B220 were found on As
2O
3-resistant HPB-ALL cells. Likewise, constitutive expression of B220 on As
2O
3-resistant L1210 cells was associated with low expression levels of membrane-bound HSP70. In vivo, the direct cytotoxic effects of As
2O
3 could be amplified by the upregulation of membrane-bound HSP70 on tumor cells, which might facilitate tumor immune rejection.
CD45 is known to positively regulate antigen-receptor signaling during activation of normal T and B cells via dephosphorylation of src kinases, and to negatively regulate cytokine receptor signaling via dephosphorylation of JAK kinases [
17,
18]. In this study, we show that the modulation of B220 cell surface expression plays an important role in determining the sensitivity of leukemic T cells to As
2O
3 and calcium ionophore A23187 cytotoxicity. In addition, we found that As
2O
3 treatment represses nuclear translocation of NF-κB p50 in a dose dependent manner. Moreover nuclear translocation of NF-κB p50 was increased in CD45-deficient Jurkat T cell line (clone J45.01) after treatment with As
2O
3, suggesting a link between B220 and NF-κB signaling pathways.
In conclusion, on a panel of mouse and human leukemic T-cell lines, we have presented evidence for a tight correlation between the induction of B220 membrane expression and their sensitivity to cell death induced by As2O3 or A23187. Our data strongly support the hypothesis that B220 plays a checkpoint role in death pathways. This could provide additional tools to potentiate As2O3 therapy against leukemic T cells.
Materials and methods
Reagents
Arsenic trioxide (As2O3), phorbol 12-myristate 13-acetate (PMA) and 4′,6-diamidino-2-phenylindole (DAPI) were purchased from Sigma-Aldrich (St. Louis, MO), and Calcium ionophore A23187 was from Calbiochem (EMD Biosciences Inc, San Diego, CA). As2O3 was dissolved in 1 M NaOH, and stored as a 330 mM stock solution, which was further diluted to 5 mM with phosphate-buffered saline (PBS).
Cell culture and cell treatment
Leukemic cell lines used in this study included mouse T-cell lines EL-4, BW5147 and L1210, human T-cell lines HPB-ALL, Jurkat (clone E6-1) and CD45 deficient variant of the E6-1 clone of Jurkat (clone J45. 01) (European Collection of Cell Cultures), and human acute promyelocytic leukemia cell line NB4. Leukemic T cell lines (EL-4, BW5147, L1210, Jurkat (clone E6-1), HPB-ALL), and APL derived cell line NB4 were kindly provided by Dr Colette Kanellopoulos-Langevin (Centre for Inflammation Research, INSERM, Hôpital Bichat, Paris, France) and Dr Jacqueline Robert-Lézénès (Inserm U940, Hôpital Saint-Louis, Paris, France), respectively. All cells were grown in RPMI 1640 containing Glutamax (Invitrogen, Cergy Pontoise, France) and supplemented with 10% heat-inactivated fetal calf serum, 50 U/ml penicillin and 50 μg/ml streptomycin at 37°C in a humidified 5% CO2 atmosphere. This culture medium will be referred to as complete medium. To avoid possible effects of cell density on cell growth and survival, cells were maintained at less than 5 × 105 cells/ml with daily adjusting cell density through the addition of fresh medium. Cell viability was estimated by the 4% Trypan-blue dye exclusion assay.
Leukemic T cells were seeded in 12-well plates at a density of 1 × 105 cells /ml and incubated in complete medium alone or in the presence of different concentrations of As2O3 or calcium ionophore A23187 at 37°C for 12, 24 and 48 h depending on the experiment.
Flow cytometry and imaging flow cytometry
The expression levels of cell surface markers on untreated and As2O3- or A23187-treated leukemic T-cell lines was analyzed by flow cytometry using either fluorescein isothiocyanate (FITC)-, phycoerythrin (PE)-, allophycocyanin (APC)- or biotin-conjugated monoclonal antibodies (mAb): rat anti-mouse Thy-1.2/CD90.2 (clone 53–2.1), anti-CD45 (clone 2D1), anti-B220/CD45R (clone RA3-6B2), anti-mouse and human CD69 (clone H1.2 F3 and FN50), anti-mouse and human Fas (clone Jo2 and DX2), anti-mouse and human FasL (clone MFL3 and NOK-1) and anti-HSP70 (clone SMC-103A) (all from eBioscience, CliniSciences, Montrouge, France), and rat IgG2a, mouse IgG, mouse IgG1, and Armenian hamster IgG1 as the isotype control (eBiosciences). Use of mAb to mouse and human Fcγ receptor (PharMingen, BD Bioscience, San Jose, CA) avoided non-specific antibody binding.
The subcellular localization of B220/CD45R molecules was determined by imaging flow cytometry, following the protocol supplied by the manufacturer (Amnis Corp., Seattle, WA). Briefly, cells (1 × 106) stained with PE-conjugated anti-B220/CD45R mAbs and DAPI were run on an ImageStream apparatus (ImagoSeine, Institut Jacques Monod, CNRS-Université Paris Diderot, France). At least 10,000 images were collected per sample at 40× or 60× magnification, and analyzed using IDEAS image-analysis software (Amnis Corp.).
Cell proliferation, cell death and caspase activation assay
Total cell numbers in untreated and As2O3-treated groups were determined by flow cytometry by acquiring events for a fixed time period of 1 min. As2O3- and Calcium ionophore-induced cell death was analyzed by propidium iodide (PI) (Invitrogen) staining, and flow cytometry. Among PI+ cells, to discriminate between apoptotic and necrotic cells, the cells were stained using either FITC-conjugated Annexin V (PharMingen) or CaspaTag Caspase 8 or Caspase 9 In situ Assay Kit, Fluorescein according to the manufacturer's instructions (Chemicon, Temecula, CA). Annexin V staining and the levels of active caspase-8 and caspase-9 were measured by flow cytometry.
Analysis of reduced glutathione content and O2- production
GSH content and O2- production in T-cell lines treated or not with As2O3 were measured by flow cytometry using 100 nM CellTracker probe CMFDA and 5 μM DHE probes, respectively, following the manufacturer’s instructions (Molecular Probes, Eugene, OR).
B220/CD45R mRNA quantification by RT-PCR
Total RNA was extracted from 5 × 106 leukemic T cells treated or not with As2O3 using the RNeasy Plus Mini kit (Qiagen, Courtaboeuf, France) following the manufacturer’s instructions and was used to generate cDNA utilizing oligo(dT) primer and SuperScript II Reverse Transcriptase (Invitrogen). PCR were conducted for 25 cycles with the following primer pairs: B220/CD45R forward primer (5'-CAC ATA TCA TCC AGG TGT GTT ATC C-3') and reverse primer (5'-GTC CTC TCC CCT GGC ACA CCT G-3'); β-actin forward primer (5'-ATC GTG GGC CGC CCT AGG CAC-3') and reverse primer (5'-TGG CCT TAG GGT TCA GAG GGG C-3'). Semi-quantitative determination (ImageJ densitometric analysis software program) of B220 cDNA, present in each of the various samples, was normalized with respect to the concentration of internal control cDNA (β-actin) detected in the same sample, and B220/ β-actin cDNA ratios were calculated.
Quantitative measurement of NF-κB activation
Nuclear extracts were prepared from 8 × 106 leukemic T cells treated or not with As2O3. The protein from nuclear extracts was quantified by the Bradford method (Bio-Rad, France). An equal amount of nuclear extract (5 μg) was assayed for NF-κB p50 activity using a TransAm NFκB p50 Transcription Factor Assay Kit according to the manufacturer’s recommendations (ActiveMotif, Rixensart, Belgium).
Statistical analyses
Data are reported as fluorescence means ± SE. Significant differences between sample means were determined using the Student t test. Statistical significance was accepted at P ≤0.05.