Introduction
The persistent and worrisome hallmark of acute promyelocytic (M3) leukemia (APL) is the high risk of severe, often fatal, bleeding complications [
1‐
6]. Pathogenesis of the coagulopathy is complex and includes an insufficient production of platelets, as well as disseminated intravascular coagulation (DIC) [
2,
6‐
9], caused, at least in part, by tissue factor (TF) expressed on the leukemia cells and on leukemia cell-derived microparticles expressing TF and procoagulant phosphatidylserine on their surface [
10‐
14]. Fibrinolysis, mediated by t-PA bound to annexin 2 on the leukemia cell, is another important factor contributing to hemorrhagic complications [
15].
Treatment of APL patients with all-
trans retinoic acid (ATRA) or arsenic trioxide (ATO) leads, over a period of 1 to 3 weeks, to normalization of plasma concentrations of D-dimers and thrombin–antithrombin complexes [
7,
8,
16,
17] and of TF mRNA in patient-derived bone marrow cells [
8,
16,
18]. Studies performed with cultured bone marrow cells from APL patients revealed that exposure to ATRA reduced cell-associated procoagulant activity [
19]. Experiments using NB4 cells, an APL cell line that presents the characteristic 15;17 chromosomal translocation, showed that exposure to ATRA or ATO resulted in a reduction of TF mRNA and antigen [
18,
20‐
22] as well as of TF activity [
12]. However, as therapy by ATRA or ATO (mostly) leads to APL cell apoptosis and thus generation of microparticles [
10,
23], it is possible that ATRA-mediated differentiation of APL cells leads to a transient increase in procoagulant activities, despite its downregulating effect on TF mRNA [
13]. A further factor that has to be taken into account is the production by APL cells of proinflammatory cytokines, such as TNF and IL-1β [
24,
25]. This may be of clinical relevance because these cytokines are capable, among other properties, of increasing TF production in monocytes and endothelial cells and, considering that NB4 cells express TNF receptor 1 [
26], could also contribute to TF production by APL cells.
In the present study, we used NB4 cells to investigate in more detail the time course of the effects of ATRA and ATO on TF activity and on expression of the proinflammatory cytokines TNF and IL-1β. In addition, we investigated to what extent TF production by NB4 cells depends on TNF and IL-1β they also produce and whether it is affected by interfering with the inflammatory signaling intermediates p38, jun kinase, and NF-κB. We observed that exposure of NB4 cells to ATRA led within 1 h to a reduction of TF mRNA and to a reduction of TNF mRNA but only after 6 h. Exposure to ATO also induced a reduction of TF and TNF mRNA, which was detectable only after 3 and 6 h, respectively. Both ATRA and ATO increased IL-1β mRNA several fold. A partial reduction in TF antigen and TF activity was evident only after 24 h of ATRA or ATO treatment. Inhibition of TNF and, to a lesser extent, of IL-1β only partially reduced TF mRNA. Inhibition of p38 reduced TF mRNA but strongly increased TNF and IL-1β mRNA, while inhibition of JNK had no effect on TF and TNF mRNA but reduced IL-1β mRNA. Inhibition of NF-κB reduced TF and TNF mRNA in NB4 cells with more than 50% within 1 h but also reduced cell survival with a half-life of approximately 6 h.
Discussion
Human APL cells produce both TF and the inflammatory cytokines TNF and IL-1β. This study was undertaken to investigate the effect of ATRA and ATO on the expression of these proteins and to better understand the contribution of the inflammatory cytokines TNF and IL-1β, as well as the signaling intermediates p38, JNK, and NF-κB to TF expression by APL cells. For this, we used the NB4 human APL cell line, which has the 15;17 chromosomal translocation leading to the production of the PML-RARα fusion protein [
36], that is characteristic for the majority of acute promyelocytic leukemia cases [
37]. NB4 cells were chosen for this study, because the behavior of these cells was shown, in many studies, to be comparable to that of patient-derived primary APL cells [
12,
18,
34,
38‐
44]. We also analyzed the effect of ATRA and ATO in HL-60 acute myeloid leukemia cells which lacks the 15;17 translocation [
45]. Our results in NB4 cells show that ATRA and ATO reduce TF and TNF mRNA but increase IL-1β mRNA; that at the NB4 cell concentrations used (10
6 cells/ml), TNF and to a lesser extent IL-1β partially contribute to TF expression; and that NF-κB and p38 contribute to TF expression. Our results that ATRA reduces TF mRNA and activity in NB4 cells are in agreement with previous results using this cell line [
18,
20‐
22] and with results obtained with bone marrow cells from APL patients at different times after initiation of ATRA treatment [
7,
19,
27]. ATRA and ATO both have similar effects on TF mRNA. However, ATRA reduces by 90% already within 1 h, whereas the effect of ATO on TF mRNA is evident only after 6 h.
In HL-60 cells, which lack the 15;17 chromosomal translocation, ATRA acted more slowly on TF mRNA and a maximal of 35% inhibition was obtained after 3 h, whereas ATO increased TF mRNA up to twofold. This implies that the effect of ATRA and ATO on TF expression is different in NB4 and HL-60 cells.
In NB4 cells, both ATRA and ATO induce the degradation of the PML-RARα fusion protein [
32], which is capable of stimulating an increased TF expression [
18]. Our finding that Zn
2+ stimulation of U937-PR9 cells strongly increases TF and TNF mRNA (Fig.
2) and that ATO treatment of these Zn
2+ stimulated cells reduced TF mRNA is compatible with the hypothesis that ATRA and ATO reduced TF and TNF via the degradation of PML-RARα. However, the increase in IL-1β also induced by Zn
2+ stimulation of U937-PR9 cells (Fig.
2) contrasts with the fact that neither ATRA nor ATO reduced this cytokine. In addition, the rapidity of the effect of ATRA on TF expression rather suggests a primary action that is independent of PML-RARα degradation. Indeed, one hypothesis for APL pathogenesis suggests that PML/RARα recruits co-repressors and histone deacetylases to their putative target genes at physiological concentrations of ATRA [
46,
47]. A primary effect of ATRA at pharmacological concentrations is to change the PML-RARα fusion protein from a repressor into a transcriptional activator [
48]. It is therefore likely that the effect of ATRA on TF reduction is not direct but rather related to upregulation of a protein that negatively regulates TF expression. The rapid ATRA-induced reduction in TF mRNA, with a half-life of less than 1 h, is compatible with the short TF mRNA half-life previously described for TNF-stimulated endothelial cells [
49]. A plausible mechanism for the effect of ATRA would therefore be an increase in TF mRNA turnover rather than a mechanism dependent on PML-RARα degradation; the latter may be relevant after ATO treatment. The rapid turnover of TF mRNA may involve an AU-rich element (ARE) in the TF mRNA molecule, which is known to recruit RNA-degrading enzymes [
50]. Indeed, ARE interacting proteins such as tristetraprolin and PARP14 exert selective posttranscriptional control of TF in macrophages [
51,
52].
We complemented our analysis on TF mRNA with assays of procoagulant activity of NB4 cells and of NB4 cell-derived microparticles. Several lines of evidence point to a role of microparticles emitted by leukemic cells in the disturbances of hemostasis [
13,
14,
53]. These microparticles may exert a direct or an indirect procoagulant effect. Indeed, Fang et al. [
54] have shown that vascular endothelial cells may uptake NB4-derived microparticles and microparticle-derived TF may make the endothelial cells procoagulant. Coagulation studies were not only the chromogenic factor Xa generation assay but also the CAT assay [
35,
55]. The latter is thought to be more relevant to the in vivo phenomena, because it takes into account all factors that contribute to the pro- and anticoagulant activities of NB4 cells and microparticles [
56]. The effect of ATRA on TF protein and activity is delayed with respect to its rapid effect on TF mRNA and is only evident at 24 h. Even then, significant procoagulant activity is still detected and this information is quite relevant in the context of reducing hemorrhagic complication in the early treatment phase [
57]. In agreement with previous reports [
58,
59], we found with the CAT assay that cells and microparticles bring both TF and an appropriate surface, since a substantial inhibition was obtained in presence of anti-TF antibody and since we did not add extra phospholipids. Of note, inhibition of thrombin generation as assessed with the CAT assay appeared to be less than that measured with the artificial Xa generation assay. This underscores the importance to assay TF activity under more physiological conditions.
Exposure of NB4 cells to ATRA resulted in a reduction of TNF mRNA, as well as an increase in IL-1β mRNA. There are previous reports on the absence of an effect of ATRA on TNF mRNA [
24,
25]. It remains to be established whether the difference in effect of ATRA is due to differences in timing of TNF mRNA measurements (up to 24 h versus 2 to 4 days) or in ATRA concentration (0.5 versus 10 μM). On the other hand, the ATRA-induced increase in IL-1β expression is in agreement with previous reports [
24,
25]. The effect of ATRA in reducing TNF mRNA appears to be specific for NB4 cells, because ATRA treatment of HL60 acute myeloid leukemia cells had no effect on TNF mRNA (this study) or increased TNF mRNA [
24]. TNF is known to increase TF expression in different cell types. The ATRA-induced reduction of TNF mRNA occurs later than that of TF mRNA (several hours versus 30 min). Therefore, the decrease in TNF expression cannot explain the effect of ATRA on TF mRNA.
However, TNF does contribute, at least partially, to TF expression in NB4 cells because inhibition of TNF by using adalimumab consistently reduced in TF mRNA by 25% and reduced the procoagulant activity of NB4 cells and NB4 cell-derived microparticles. The contribution of IL-1β to TF mRNA was less than that of TNF, and we could not detect an effect of IL-1β inhibition on CAT activity expressed by NB4 cells and NB4 cell-derived microparticles. In APL patients, the contribution of TNF or of IL-1β may be higher because NB4 cell densities used in our experiments are much lower than the densities of leukemic cells encountered in patient’s blood and bone marrow. We were unable to study the contribution of TNF or of IL-1β at higher cell concentrations because, in our in vitro cell culture system, these are limited to approximately 10
6 NB4 cells/ml due to acidification and exhaustion of the cell culture medium within 24 h. Leukemia cell-derived cytokines may also be relevant for thrombo-hemorrhagic events in APL patients, because they increase TF expression by endothelial cells [
60‐
62] and monocytes [
63].
We observed that inhibition of the signaling intermediates p38 or of NF-κB that act downstream of TNF or IL-1β had a more pronounced effect (38% and 62% inhibition of TF mRNA, respectively), than selective inhibition of TNF or IL-1β alone (28% and 20% inhibition of TF mRNA, respectively), whereas inhibition of jun kinase had no effect. However, it has to be stressed that inhibition of p38 results in a strong increase in TNF expression (sixfold) and an even more pronounced increase in IL-1β expression (26-fold). Thus, it appears that key elements of the signaling pathways of inflammatory cytokines are also involved in the regulation of their production. An unexpected finding of our study was that inhibition of NF-κB induced cell death within a few hours. The resulting liberation of TF-bearing microparticles may explain the increase in procoagulant activity of NB4 cell-derived microparticles, as measured with the CAT assay, despite an initial reduction in TF mRNA.
The effect of NF-κB inhibition should stimulate more detailed studies on the importance of NF-κB for APL cell survival, the mechanisms by which NF-κB inhibition induces cell death and whether NF-κB inhibition can be used as adjunct therapy in APL [
64,
65]. The consequences of potential effects of NF-κB inhibition on APL cell survival should be carefully evaluated, as the apoptotic cells and the cell-derived microparticles express phosphatidylserine at their surface and may thereby exacerbate thrombo-hemorrhagic complications, despite an initial reduction in TF mRNA, but not yet in TF protein.
In conclusion, ATRA and ATO reduce expression by NB4 APL cells of both TF and TNF and inhibition of TNF or IL-1β or of the inflammatory signaling intermediates p38 or NF-κB also reduces TF.
Thus, TNF and IL-1β have a regulatory function on TF expression by NB4 APL cells, but the effect of ATRA and ATO on TF can only partially be accounted for their impact on these cytokines. The results presented here should stimulate further in vitro and in vivo preclinical studies to determine to what extent inhibition of inflammatory cytokines or inflammatory signaling intermediates can be used for the further development of adjunct therapies of APL.