|Home | About | Journals | Submit | Contact Us | Français|
Differentiation therapy of cancer is being explored as a potential modality for treatment of myeloid leukemia, and derivatives of vitamin D are gaining prominence as agents for this form of therapy. Cyclooxygenase (COX) inhibitors have been reported to enhance 1,25-dihydroxyvitamin D3 (1,25D)-induced monocytic differentiation of promyeloblastic HL60 cells, but the mechanisms of this effect are not fully elucidated, and whether this potentiation can occur in other types of myeloid leukemia is not known. We found that combination treatment with 1,25D and non-specific COX inhibitors acetyl salicylic acid (ASA) or indomethacin can robustly potentiate differentiation of other types of human leukemia cells, i.e., U937, THP-1, and that ASA +/- 1,25D is effective in primary AML cultures. Increased cell differentiation is paralleled by arrest of the cells in the G1 phase of the cell cycle, and by increased phosphorylation of Raf1 and p90RSK1 proteins. However, there is no evidence that this increase in phosphorylation of Raf1 is transmitted through the ERK module of the MAPK signaling cascade. Transfection of small interfering (si) RNA to Raf1 decreased differentiation of U937 cells induced by a combination of ASA or indomethacin with 1,25D. However, phosphorylation levels of ERK1/2, though not of p90RSK, were increased when P-Raf1 levels were decreased by the siRNA, suggesting that in this system the ERK module does not function in the conventional manner. Identification of the strong antiproliferative activity of ASA/1,25D combinations associated with monocytic differentiation has implications for cancer chemoprevention in individuals who have a predisposition to myeloid leukemia.
Patients with acute myeloid leukemia (AML) are currently successfully treated using the available cytotoxic, anti-neoplastic agents in only a minority of cases, indicating the critical need for new therapeutic approaches. Among these, derivatives of vitamin D (deltanoids) have shown promise in pre-clinical studies (reviewed in refs. 1–5), but their use in the clinic has been limited by the possibility of life threatening hypercalcemia, induced by the classical action of vitamin D.6,7
To reduce this danger, combinations of low concentrations of deltanoids with other compounds are being sought which increase the ability of deltanoids to induce differentiation of neoplastic cells, but have no effects on systemic calcium homeostasis.8-10 Among these, several nonsteroidal anti-inflammatory agents have been reported to synergize with vitamin D3 to increase differentiation of HL60 cells, a frequently used in vitro model of AML (reviewed in refs. 11 and 12). In these studies, it was suggested that an enzyme of the aldoketoreductase family was the intracellular target downregulated by compounds that increase the activity of 1,25-dihydroxyvitamin D3 (1,25D), and evidence was provided that an inhibition of NFkB activity was also a factor. However, the participation of signaling cascades that control intracellular pathways responsible for the synergy between anti-inflammatory agents and 1,25D is still unclear.
In this study we investigated if the Raf-initiated MAPK pathway, already well established to participate in initiating 1,25D-induced monocytic differentiation, is also important for the enhanced differentiation when 1,25D is combined with inhibitors of cyclooxygenases (COXs).
These compounds have been used for GI cancer chemoprevention, and are approved for human use,13-15 so if effective in potentiating the anti-leukemic action of 1,25D, could be introduced to therapeutic regimens. The translational significance of this study was further increased by demonstrating that the COX-inhibitor plus 1,25D differentiation synergy is not limited to HL60 cells, but is evident to some degree in other human myeloid leukemia cells. Mechanistically, we found that while the COX inhibitors channel the differentiation-enhancing signals through Raf-1, a nodal point in 1,25D signaling of monocytic differentiation,16,17 the signaling cascade induced by COX-inhibitor +1,25D combinations does not include the classical MEK/ERK module.
To determine the potentiation of anti-proliferative and differentiation-inducing activities of representative, relatively specific or non-specific cyclooxygenase inhibitors, we chose three myeloid leukemia cells lines, promonocytic U937, promyeloblastic HL60, and monocytic THP-1, which are frequently used as in vitro models of the human disease (reviewed in refs. 19 and 20). This panel of COX inhibitors consisted of ASA, a relatively non-specific COX inhibitor, though with greater activity toward the COX-1 isoform,21 DuP-697 (5-bromo-2-(4-fluorophenyl-3-(4-methylsufonyl) phenylthiophene), a predominantly COX-2 inhibitor,22 FR 122047 (1-[[4,5-bis(4-methoxyphenyl)-2-thiazolyl]carbonyl]-4-methyl-piperazine, monohydrochloride hydrate), a selective COX-1 inhibitor23 and INDO (1-(chlorobenzoyl)-5-methoxy-2-methyl-3-indoleacetic acid),24 a rather non-specific inhibitor, though with greater activity against COX-1. For a positive control of enhancement of 1,25D-induced differentiation we used SB202190, an inhibitor of p38MAP kinase with a known ability to increase the JNK pathway activity in myeloid leukemia cells.10,25 To ensure that these compounds do not cause significant cell death in the experiments, as previously reported for ASA,26 the sub-toxic levels of each COX inhibitor were determined in preliminary experiments (not shown). Figure 1A shows that in all three cell lines the relatively nonspecific COX inhibitors were more potent proliferation inhibitors than the more specific COX inhibitors DuP-697 and FR-122047. When combined with 1,25D, the antiproliferative effects were more pronounced, though there was considerable variability, and only ASA/1,25D combination showed a consistent and significant (p < 0.05) combinatorial effect on cell proliferation (Fig. 1A). Importantly, these effects were observed without a major effect on cell viability, measured by two independent methods (Fig. 1B).
Inhibition of cell proliferation by ASA and INDO in the absence of evidence of cell death suggested that these compounds induced an arrest or slowing of the cell cycle progression. When analyzed by flow cytometry of propidium iodide-stained cells, we noted that ASA added alone or with 1,25D, a known regulator of the cell cycle,27 induced G1 arrest, as shown by the increased G1 to S phase ratio, in two of the three cell lines studied (Fig. 2). At subtoxic concentrations used in these experiments, INDO combined with 1,25D had only a minimal and inconsistent effect on G1 to S phase transition, suggesting that the observed inhibition of cell proliferation (Fig. 1) was due to a general retardation of cell cycle progression, with only a slightly greater effect on G1 cell cycle controls. There were no observable changes in cell cycle parameters in cells treated with the more specific COX inhibitors DuP-697 or FR-122047, or with the p38MAPK inhibitor SB202,190 (SB202), used here as a positive control of differentiation enhancement, alone or in the presence of 1,25D. Thus, it appears that inhibition by non-specific inhibitors of both principal isoforms of COX enzymes is necessary for the efficient retardation of the cell cycle progression.
When exposed to low concentrations of 1,25D, myeloid leukemia cells undergo phenotypic differentiation, shown by expression of surface markers e.g., CD11b and CD14,28,29 as well as of cytoplasmic enzymes e.g., monocytic esterase known as “non-specific esterase” (NSE), the extent of which depends on the time of exposure and the subtype of leukemia cells (Fig. 3). This differentiation can be potentiated by simultaneous addition of other compounds,4,10 including SB202,10 as also illustrated in Figure 3. It was previously reported that in HL60 cells ASA and INDO also have this synergy with 1,25D, but it was not clear if the more specific COX inhibitors are more, or less, effective potentiators of 1,25D-induced differentiation.11,30 We now show that ASA can be an effective potentiator of 1,25D-induced differentiation not only in HL60 but also in U937, THP-1 cells (Fig. 3) and in primary AML cells (Fig. 4), while INDO is more restricted in its range of activity, potentiating the 1,25D effects only in HL60 and THP-1 cells from among the cell types studied here (Fig. 3). The specific COX-2 inhibitor DuP-697 also had a consistent, though less marked, differentiation potentiating effect (Fig. 3), and this was also seen in a second patient sample (not shown), while FR-122047 showed no significant potentiation of 1,25D-induced differentiation (Fig. 3). Thus, ASA appeared to be the most promising differentiation-enhancing agent, and the U937 cell line appears to be the most closely representative of the ex vivo leukemic cells available for this study.
Based on the cell cycle and differentiation data described above, we selected U937 cells for mechanistic studies, since primary patient samples do not provide sufficient material for such studies. Previous studies have linked the Raf-MEK-ERK-RSK pathway, or its variants, to 1,25D-induced differentiation of human leukemia cells.17,31,32 We therefore investigated if potentiation of differentiation by COX inhibitors parallels enhancement of the activity of this pathway, as shown by the increased phosphorylation of individual kinases in this signaling cascade. Indeed, there were higher levels of P-Raf1, P-MEK1 and P-p90RSK1 in U937 cells treated with 1,25D together with ASA, and to a lesser extent in cells treated with 1,25D and INDO or DuP-697, compounds which showed only a minor enhancement of differentiation, and no increase in P-Raf1 was detected in cells treated with FR-122047 (Fig. 5), an inefficient potentiator of differentiation (Fig. 3). A similar increase in P-Raf1 levels was noted when HL60 cells were treated with ASA and INDO (data not shown), and although this increase could not be demonstrated in THP-1 cells (not shown), this was most likely due to the high constitutive levels of P-Raf1 in these cells which exhibit phenotypic markers of early monocytic differentiation.33
In contrast to the clear correlation between differentiation and P-Raf-1 levels, there were actual decreases in P-ERK 1/2 levels in all cell types studied when treated with combinations of 1,25D and COX inhibitors (Fig. 5 and data not shown). This indicates that in this cell setting, activated Raf1 does not signal differentiation through the conventional MAP kinase signaling cascade that includes ERK kinases.
To determine if Raf1 signaling is required for the potentiation of 1,25D-induced differentiation by ASA we transfected U937 cells with Raf1 siRNA and observed the effects on differentiation and MAP kinase activation. As expected, there was a marked reduction in the expression of markers of differentiation, compared to cells that were untransfected or transfected with control siRNA as summarized in Figure 6A, and illustrated in Figure 6B. This inhibition of differentiation was accompanied by a marked reduction in the level of P-Raf1, attesting to the efficient targeting of Raf1 expression by the siRNA, as well as by a reduction in P-p90RSK1 level, a potential downstream target of Raf1 (Fig. 7A). However, the activation of p90RSK1 by 1,25D/ASA was unlikely to have been transmitted by ERK, as the levels of P-ERK 1/2 slightly increased in cells transfected with siRaf1 as compared to cells transfected with control siRNA (Fig. 7A), and more marked effects of siRaf1 on P-ERK 1/2 were observed when the differentiation was induced by the combination of 1,25D and INDO (Fig. 7B). This suggests that P-Raf1 signaling represses, rather than activates, the ERK signaling in COX inhibitor/1,25D-induced differentiation of U937 cells, as previously proposed for differentiation signaling by a prolonged exposure of HL60 cells to 1,25D, used as a single agent.17
There are two complementary aspects of this report; one translational, the other mechanistic. On the practical side, we asked if combination of low concentrations of 1,25D with COX inhibitors is likely to be useful in the prevention or treatment of human myeloid leukemia, and if so, would selective COX-1 or COX-2 inhibitors be more effective than the non-selective inhibitors previously reported to synergize with 1,25D in HL60 cells.11,30 We found that leukemia cell lines other than HL60 also exhibit enhanced differentiation, and arrest in G1 phase of the cell cycle, when exposed to combinations of 1,25D and non-specific COX inhibitors. Of note, ex vivo leukemic cell samples freshly obtained from two patients, also respond to these combinations, particularly ASA/1,25D (Fig. 4, and data not shown). While a more extensive series of patients needs to be accrued, this finding points to a possible success in future translational studies. The greater synergy obtained with the nonspecific inhibitor ASA than with other compounds tested suggests that both the constitutive COX-1 and COX-2 enzymes need to be inhibited to enhance the differentiating action of 1,25D, although a minor effect also was seen with the specific COX-2 inhibitor DuP-697 (Fig. 3). Specific COX-2 inhibitors have recently been reported to have serious side effects when administered to patients, especially on the cardiovascular system, such as myocardial infarctions, strokes, and hypertension.34,35 Thus, even though ASA at high dosage can cause gastrointestinal ulcers and renal toxicity,36 ASA would appear to be a COX inhibitor of choice to combine with 1,25D for prevention of myeloid leukemia in high risk populations, as it is both effective in the present studies, and safe in moderate doses.
The mechanistic rationale for combining a COX inhibitor with 1,25D in cancer therapy may be provided by the reported inhibition of the prostaglandin pathway by 1,25D37,38 and its involvement in 1,25D signaling,39 so that COX inhibitors can enhance this effect by blocking COX enzyme activity. However, how 1,25D can initiate signals that impinge on the transcriptional machinery involved in the regulation of the prostaglandin pathway is not known. We, and others, have previously provided extensive evidence that MAPK pathways are involved in the control of transcription factors, such as C/EBPs, which upregulate programs required for monocytic differentiation,40 and that Raf1 is a critical nodal control point for 1,25D-induced differentiation of human leukemia cells, consistent with its activation under all conditions tested to date that lead to 1,25D-induced differentiation.16,17,41,42 We have also shown that the MEK/ERK module, classically downstream from Raf1, is modulated in a biphasic manner following exposure of the cells to 1,25D.31 It was proposed that the initial exposure to 1,25D directly upregulates the expression of a scaffold protein KSR1 through a VDRE element in its promoter,41 and this increases the efficiency of growth factor and cytokine signals generated at the cell surface. Thus, KSR1, and perhaps also the more recently discovered hKSR2,43 provide a platform that binds Raf1, MEK1 and ERK1/2, and facilitates the phosphorylation of MEK1 by Raf1.44,45 However, it is also known that an excess of a scaffold protein actually inhibits the reactions that it facilitates when at lower concentrations, as Raf1, MEK1 and ERK may bind to different KSR1 molecules when these are numerous, thus impeding phosphorylation of ERK1/2.46 In the 1,25D differentiation system this then leads to reduced P-ERK 1/2 levels, consistent with the results obtained here (Fig. 5), and offering an explanation for the G1 block observed when the cells are exposed to ASA/1,25D combinations (Fig. 2). Thus, it appears that COX inhibitors act on the MAPK pathway upstream of Raf1, perhaps by inhibition of EGFR signaling by growth factors, as shown for an ASA derivative,47 while downstream of Raf1 the signals are similar, though more intense, to the signals provided by 1,25D when it is administered as a single agent. Future studies of the anti-proliferative and differentiation-inducing effects of COX inhibitor/1,25D combinations on leukemic cells should therefore be principally directed to the upstream molecular events that regulate the expression and activation of Raf1.
In summary, this report adds to the growing evidence that combinations of 1,25D with COX inhibitors can retard emergence and progression of malignant clones in various organs, such as the prostate38 and breast.48 The studies presented have also suggested that the paradigm of the Raf-MEK-ERK-RSK cascade in mammalian cells, which was principally based on work with fibroblasts (reviewed in refs. 49 and 50), may require revision when other cell types are considered.
1,25-Dihydroxyvitamin D3 (1,25D) was provided by Dr. Milan Uskokovic, Bioxell, Nutley, NJ. Phospho-Raf-1 (Ser 338), phospho-MEK1 (Ser217/221), Phospho-p44/42 MAPK (Thr202/Tyr204), and Phospho-p90RSK (Ser380) antibodies were purchased from Cell Signaling (Beverly, MA). The antibodies against Raf-1 (c-12, rabbit polyclonal), MEK-1 (c-18, rabbit polyclonal), ERK1/2 (k-23, rabbit polyclonal), and p90RSK (c-21, rabbit polyclonal) were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). COX inhibitors Acetyl Salicylic Acid (ASA) and Indomethacin (INDO) were purchased from Sigma (Sigma-Aldrich Saint Louis, MA), while DuP-697 and FR-122047 were purchased from Cayman Chemical Company. SB202190 was purchased from Calbiochem (San Diego, CA).
U937 Cells, HL60-G cells 18 and THP-1 cells were cultured in suspension at 37°C in RPMI 1640 medium (Mediatech, Herndon, VA) supplemented with 10% iron-enriched bovine calf serum (Hyclone, Logan, UT). The cultures were routinely checked for mycoplasma expression by routine microbiological methods. For experiments, the cells were suspended at 1.5–4.0 × 105 cells/ml (U937- at 4.0 × 105; HL60 and THP- at 1.5 × 105), in fresh medium and incubated for 48 hours, with or without 10 nM 1,25D plus the indicated compounds. Cell numbers were determined by hemocytometer counts, and viability by trypan blue (0.4%) exclusion (TB Viability), as well as from scatter plots on Flow Cytometry analysis (Flow Viability). Fresh peripheral blood specimens were obtained, following informed consent, from patients with newly diagnosed acute myelogenous leukemia, utilizing a consent document approved by the IRB of the University of Medicine and Dentistry of New Jersey. Cells were separated by Ficoll-hypaque, washed twice in sterile phosphate buffered saline, and placed into cell culture for experiments described below.
Aliquots of 1 × 106 cells were harvested, washed with 1 × PBS, and stained with MY4-RD1 (anti-CD14) and M01-FITC (anti-CD11b) antibodies from Beckman Coulter (Miami, FL). Two color parameter analysis was performed using a Beckman Coulter EPICS XL Flow Cytometer. The procedures were performed as described before.8 Isotypic mouse Ig G1 was used to set threshold parameters. Monocytic differentiation was also monitored by cytochemical determination of the activity of nonspecific esterase (NSE), as described previously.8
Cells (1 × 106) were washed twice with 1 × PBS and fixed in 75% ethanol at -20°C for at least 24 h. The fixed cells were spun down to remove fixative buffer and washed twice with 1 × PBS, then the cell pellets were resuspended in 1 ml propidium iodide solution (PI, 10 μg/mL; RNase A, 10 μg/ml, Sigma) at 37°C for 30 minutes. The cells were analyzed using an Epics XL-MCL instrument (Coulter, Fullerton, CA), and cell cycle distribution was determined by Multicycle Software Program (Phoenix Flow System, San Diego, CA). The G1/S ratios were calculated and served to monitor the block of G1 to S phase progression.
The Raf-1 siRNA duplex was designed to target human Raf1 sequence 5′-AAU GUC CAC AUG GUC AGC ACC-3′ (913–933) (Dharmacon RNA Technologies). Generic scrambled siRNA was purchased as control (Dharmacon RNA Technologies). siRNAs (5 μM) were transfected into U937 cells using Amaxa Nucleofector (Gaithersburg, MD) according to the manufacturer's protocol. The cells were allowed to recover for 48–72 hours in 6 cm wells in RPMI 1640 medium, supplemented with 10% bovine calf serum (BCS), and checked for the efficiency of transfection by the levels of Raf1 protein, total or phosphorylated. Transfected cells were seeded at a concentration of 3 × 105 cells/mL and exposed to the specified agents for the indicated times.
Western blotting was performed using whole cell extracts. The cells were harvested and washed twice with ice-cold 1× PBS. The washed cell pellets were solubilized with a lysis buffer (20 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1% Triton X-100, 2.5 mM sodium pyrophosphate, 1 mM β-glycerophosphate, 1 mM Na3VO4, 1 mM PMSF, 1 μg/ml leupeptin, 1 μg/ml aprotinin), followed by centrifugation at 16, 000 g for 15 min. The protein concentrations of the extracts were determined by using Bio-Rad (Hercules, CA) protein assay kit and then incubated in (2:1) 3× SDS sample buffer (150 mM Tris-HCl, pH 6.8, 30% glycerol, 3% SDS, 1.5 mg/ml bromophenol blue dye, and 100 mM dithiothreitol). Forty μg of whole cell extracts were separated on 10% SDS-PAGE gel and transferred to nitrocellulose membranes. The membranes were blocked with 5% milk in TBS/0.01% Tween 20 for 1 hour, subsequently blotted with primary antibodies for 1 hour, then the membranes were blotted with a horseradish-linked secondary antibody for 1 hour. The protein bands were visualized with a chemiluminescence assay system (Amersham). The optical density (OD) of each band was measured using an image quantitator (Molecular Dynamics, Sunnyvale, CA).
All experiments were repeated for a minimum of three times. Significance of differences between mean values was assessed by two-tailed Student's t test. All computations were performed with an IBM-compatible personal computer using Microsoft EXCEL.
We thank Dr. Milan Uskokovic, BioXell, Nutley, NJ, for the gift of 1,25-dihydroxy-vitamin D3, we also thank Dr. Michael Danilenko and Dr. Ewa Marcinkowska for their comments on the manuscript, and Mrs. Vivienne Lowe for her excellent secretarial work in the preparation of this manuscript.
This work was supported by NIH grant R01-44722 from the National Cancer Institute.