Bioimaging analysis of nuclear factor-κB activity in Philadelphia chromosome-positive acute lymphoblastic leukemia cells reveals its synergistic upregulation by tumor necrosis factor-α-stimulated changes to the microenvironment
Division of Molecular Therapy, Advanced Clinical Research Center, Institute of Medical Science, The University of Tokyo, Tokyo, Japan
Graduate Institute of Medicine, College of Medicine, Kaohsiung Medical University, Kaohsiung
National Institute of Cancer Research, National Health Research Institutes, Tainan, Taiwan
To gain an insight into the microenvironmental regulation of nuclear factor (NF)-κB activity in the progression of leukemia, we established a bioluminescent imaging model of Philadelphia chromosome-positive acute lymphoblastic leukemia (Ph+ALL) cells transduced with a NF-κB/luciferase (Luc) reporter and cocultured with murine stromal cells and cytokines. Stromal cells alone did not augment Luc activity, taken as an index of NF-κB, but Luc activity was synergistically upregulated by the combination of stromal cells and tumor necrosis factor (TNF)-α. Dehydroxymethylepoxyquinomicin (DHMEQ), a specific inhibitor of NF-κB DNA binding, rapidly induced the apoptosis of Ph+ALL cells, indicating that NF-κB is necessary for the growth and survival of these cells. However, the DHMEQ-induced suppression of NF-κB activity and the apoptosis of leukemia cells were attenuated by the presence of stromal cells and TNF-α. In NOD-SCID mice transplanted with NF-κB/Luc reporter-containing Ph+ALL cell lines and monitored periodically during the progression of the leukemia, murine TNF-α was significantly expressed in lesions in which the leukemia cells emitted a significant NF-κB signal. These results support the notion that TNF-α also triggers microenvironmental upregulation of NF-κB activity in vivo. Collectively, the results indicated that TNF-α-stimulated microenvironment may contribute to the survival and progression of Ph+ALL cells through the synergistic upregulation of NF-κB activity. (Cancer Sci 2011; 102: 2014–2021)
There are five members in the nuclear factor (NF)-κB family, namely RelA (p65), c-Rel, RelB, p50 (and its precursor p105) and p52 (and its precursor p100). These proteins are assembled as homo- or heterodimers bound to IκB family proteins and retained in the cytoplasm in unstimulated cells.(1,2) Responding to extracellular stimuli, NF-κB signaling is activated and plays a crucial role in the regulation of innate and adaptive immune responses.(3,4) More recently, a significant role for NF-κB signaling in cancer development and progression has been demonstrated.(5,6) Activation of NF-κB has been noted in a variety of hematopoietic malignancies.(7–9) In addition, NF-κB is absolutely required for the growth and survival of certain hematologic malignancies.(10–12)
Development of Philadelphia chromosome-positive acute lymphoblastic leukemia (Ph+ALL) is essentially attributed to genomic recombination between the ABL gene and BCR gene, which results in the formation of p190 or p210 Bcr-Abl tyrosine kinase.(13,14) Approximately 20–30% of adult ALL is Ph+ALL, which is the most refractory subtype.(15–17) Despite progress in targeted molecular therapy,(18–21) clinical outcomes are not as promising as expected, mainly due to acquired drug resistance.(22,23) Although a series of Abl kinase domain mutations play a crucial role in acquired drug resistance, it is well documented that a close interaction with stromal cells, the so-called microenvironment, can protect leukemia cells from spontaneous and/or drug-induced apoptosis through diverse mechanisms,(24–26) one of which may be due to the activation of the NF-κB signal in leukemia cells. Constitutively active NF-κB has been detected in primary blast cells and cell lines derived from Ph+ALL,(27) and is required for Bcr–Abl-mediated transformation of bone marrow cells and interleukin (IL)-3-dependent cell lines, as well as tumorigenicity in nude mice.(28,29) However, the role of NF-κB in the progression of Ph+ALL is not yet well understood. In addition, microenvironmental regulation of NF-κB activity in Ph+ALL has not been clarified. Therefore, in the present study, we transduced Ph+ALL cells with an NF-κB/luciferase (κB/Luc) reporter construct and established a bioluminescence imaging model for in vitro and in vivo analysis. The aim of the present study was to elucidate the role of cytokines and cellular interactions in the regulation of NF-κB activity in Ph+ALL cells during the progression of leukemia.
Materials and Methods
Cell lines. The p190-expressing Ph+ALL cell lines Sup-B15, OM9;22, KOPN-30, and KOPN-72 (gift from Dr Oyashiki, Tokyo Medical College, Tokyo, Japan) were maintained in our laboratory as described previously.(30) The p210-expressing IMS-PhL1 cell line was established in our laboratory (Division of Molecular Therapy, Institute of Medical Science, The University of Tokyo, Tokyo, Japan) from a Ph+ALL patient, as reported previously.(31) The culture of the murine bone marrow stromal cell line HESS-5 was as described elsewhere.(32) The leukemia cell lines were cultured in RPMI-1640 containing 10% FBS. When IMS-PhL1 and Sup-B15 cells were seeded onto an HESS5 monolayer, the culture medium was changed to α-minimum essential medium (α-MEM) containing 10% FBS.
Reagents. All cytokines, namely tumor necrosis factor (TNF)-α, IL-3, Flt3-ligand (Flt3L), stem cell factor (SCF), granulocyte–macrophage colony-stimulating factor (GM-CSF), IL-1α, and interferon (IFN)-γ, were purchased from PeproTech (Rocky Hill, NJ, USA). Fibronectin was purchased from Invitrogen (Carlsbad, CA, USA) and imatinib was obtained from Novartis Pharmaceuticals (East Hanover, NJ, USA). Dehydroxymethylepoxyquinomicin (DHMEQ), a potent inhibitor of the DNA binding of the NF-κB complex,(33) was dissolved in 100% of DMSO prior to use in the experiments. The final maximal concentration of DMSO that the cells were exposed to was 500 fold dilution of pure DMSO (0.2%).
Construction and production of lentiviral vectors. A self-inactivating lentiviral vector encoding humanized Renilla reniformis (R.reniformis) green fluorescent protein (GFP) driven by a CMV promoter (HIV-CMV/hrGFP) was produced as described previously.(30) To construct NF-κB reporter vectors, this expression cassette was replaced by a firefly luciferase gene downstream of three copies of NF-κB-responsive elements linked to a basal TATA promoter (HIV-NF-κB/Luc). The NF-κB specificity of these constructs was confirmed by comparing reporter activity with a mutant enhancer-containing reporter (HIV-mTA/Luc) in various cell lines and with various stimuli. In addition, HIV-EF1α/Luc and Venus were prepared as constitutively expressing vectors. Viral supernatant and high titer stocks were prepared as described previously.(30) The functional titers of these vectors was determined by flow cytometry of infected HeLa cells and exceeded 1 × 109 units/mL.
Transduction of leukemia cells and analysis of NF-κB activity. For viral transduction of leukemia cells, 2 × 105 cells were pelleted and incubated with viral supernatant at a multiplicity of infection (MOI) of 10 in 1.5-mL Eppendorf tubes. After incubation for 2 h at 37°C in 5% CO2, infected cells were seeded onto an HESS-5 monolayer and cultured for 1 week. Single cell sorting was performed using a BD FACSAria (Becton Dickinson, Franklin Lakes, NJ, USA) and the resulting cell clones were cocultured with HESS5 cells in 96-multiwell dishes for 3 weeks. A number of reporter vector-transduced clones were selected by screening TNF-α-induced Luc activity on an ARVO MX multilabel counter (PerkinElmer, Waltham, MA, USA) and used in further experiments. Selected clone-derived cells were cultured in 6-cm dishes and subjected to in vitro bioluminescence imaging (BLI) analysis with a cooled charge-coupled device (CCD) camera system (IVIS Imaging System 100; Xenogen, Alameda, CA, USA).(34)
Apoptosis assay. The IMS-PhL1 cells were cultured in 24-multiwell plates and treated with 10 μg/mL of DHMEQ for 0, 24 and 48 h. Treated cells were collected, labeled with annexin V–phycoerythrin (PE) and 7-aminoactinomycin D (7-AAD), and then analyzed on a BD FACSCalibur (Becton Dickinson).
Cell proliferation assay (WST1 assay). Samples (in triplicate) of 5 × 104 leukemia cells were incubated with serial twofold dilutions of TNF-α (from 100 ng/mL to 0) and DHMEQ (from 10 μg/mL to 0) in 96-multiwell plates. Cultures were maintained at 37°C in 5% CO2 for 48 h. Then, 10 μL of Tetracolor1 solution (water-soluble tetrazolium salt; Seikagaku, Tokyo, Japan) was added to each well and cells were incubated for a 1 h. Absorbance was measured at 450 nm using an ARVO MX multilabel counter (PerkinElmer).
Other methods including electrophoretic mobility shift analysis, animal study, and polymerase chain reaction analysis are provided in Data S1.
NF-κB activity is constitutive and inducible in Ph+ALL cells. After a single cell sorting of two Ph+ALL cell lines infected with a NF-κB/Luc lentivirus, we obtained a number of clones suitable for use in an NF-κB reporter assay. Data from representative clones are reported below.
First, we examined the steady state NF-κB activity of IMS-PhL1 cells. Basal level NF-κB activity was low in IMS-PhL1 cells, but 1.6-fold higher than background levels in HIV-mTA/ Luc-transduced cells in three independent assays. Among a panel of cytokines, only TNF-α strongly induced NF-κB activity in IMS-PhL1 cells (Fig. 1A). Similar results were obtained using the IVIS system (Fig. 1B). The inducibility of NF-κB activity by TNF-α was no less than 20-fold, with maximal induction observed at 12–24 h. However, in Ph-negative NALM6 cells, TNF-α failed to significantly induce NF-κB activity (data not shown). Supporting these observations, significant NF-κB binding activity was detected in IMS-PhL1 cells in the electrophoretic mobility shift assay (EMSA; Fig. 1C). In a gel supershift assay, p50 and p65 were identified as the main components of NF-κB in IMS-PhL1 cells. In addition to, there was another high molecular weight complex observed in IMS-PhL1 cells that has not been reported previously. We have not yet identified the exact components of this complex, which is specific to NF-κB because 10-fold competitor probes easily canceled this band (Fig. S1). When IMS-PhL1 cells were treated with 100 ng/mL of TNF-α for 6 h, the binding activities of NF-κB that were composed mainly of p50 and p65 were significantly enhanced. We also found constitutive NF-κB binding activity in which the major components were p65 and/or p50 in another four other Ph+ALL cell lines (Fig. 1C). Because TNF-α usually activates apoptotic signaling, Ph+ALL cell lines were treated with TNF-α for 48 h, but this had no effect on their survival (Fig. 1D). The expression of TNF-α receptors was confirmed in IMS-PhL1 and Sup-B15 cells by FACS (Fig. S2).
Nuclear factor-κB activity is critical for the survival of Ph+ALL cells. In the present study, DHMEQ was used to inhibit NF-κB activity. It has been shown that DHMEQ specifically inhibits the DNA binding of NF-κB and does not affect TNF-α-induced phosphorylation and degradation of IκB.(33,35) When IMS-PhL1 cells were treated with 10 μg/mL of DHMEQ, constitutive and TNF-α-induced NF-κB binding activity in IMS-PhL1 cells were substantially and completely suppressed after 6 and 12 h treatment, respectively, as determined by the EMSA (Fig. 2A). Without DHMEQ, Luc activity, as an index of NF-κB (NF-κB signal), was transiently induced more than 20-fold over basal levels 12–24 h after TNF-α stimulation, followed by a gradual decrease until 48 h. In contrast, the TNF-α-induced NF-κB signal was constantly negligible throughout the duration of 10 μg/mL of DHMEQ treatment (Fig. 2B). When five Ph+ALL cell lines were treated with graded doses of DHMEQ and were subjected to the water-soluble tetrazolium salt (WST1) assay after 48 h of culture, a similar dose-dependent growth inhibition was seen (Fig. 2C). Following labeling DHMEQ-treated IMS-PhL1 cells with PE-conjugated annexin V and 7-AAD, the proportion of annexin V+ apoptotic cells increased in a time-dependent manner (Fig. 2D). These results indicate that NF-κB activity is critical for the survival of Ph+ALL cells.
Synergistic upregulation of NF-κB activity in Ph+ALL cells by TNF-α and stromal cells. When IMS-PhL1 cells were seeded onto an HESS5 cell monolayer and cocultured in the absence of TNF-α, the NF-κB activity of the IMS-PhL1 cells increased by only 1.3-fold. In contrast, when cells were cocultured in the presence of 100 ng/mL of TNF-α for 24 h, NF-κB activity was upregulated to 62.6-fold over basal levels. This upregulation appears to be the result of a synergistic effect between stromal cells and TNF-α, because alone TNF-α alone induced a 17.1-fold increase in NF-κB activity in IMS-PhL1 cells (Fig. 3A). These data were confirmed with the IVIS system (Fig. 3B), and similar results were obtained using Sup-B15 cells (data not shown). To elucidate the requirement of direct cell contact for these effects, IMS-PhL1 and HESS5 cells were cocultured with or without direct cell contact. Without direct cell contact, the synergistic upregulation of the NF-κB signal by TNF-α and HESS5 was inhibited. The Luc activity of IMS-PhL1 cells that were not adherent to an HESS5 layer was comparable to that of IMS-PhL1 cells cultured in the absence of HESS5 cells (Fig. 3C). Accordingly, direct cellular interactions are a prerequisite for synergistic upregulation of NF-κB activity by TNF-α and HESS5 cells.
To investigate whether the upregulation of NF-κB activity affects drug response, EF1α/Luc-transduced IMS-PhL1 cells were used to evaluate viable cell mass with or without an HESS5 layer. Although the same number of IMS-PhL1 cells was used under both conditions, Luc activity was significantly reduced in the presence of an HESS5 layer (Fig. 3D). This is because most of the IMS-PhL1 cells migrated beneath the HESS5 layer over time, and exhibited reduced cell growth, compared with IMS-PhL1 cells cultured under HESS5-free conditions.(31) As shown in Figure 3(D), the growth of IMS-PhL1 cells was significantly suppressed by optimal doses of imatinib (1 and 10 μM), regardless of the presence of HESS5 cells and/or TNF-α. The IMS-PhL1 cells were significantly resistant to a suboptimal dose of imatinib (0.1 μM), suggesting that the upregulation of NF-κB activity by TNF-α and stromal cells may confer imatinib resistance to Ph+ALL cells.
Stromal cells attenuate DHMEQ inhibition of NF-κB activity in Ph+ALL cells. When IMS-PhL1 cells were treated with 10 μg/mL of DHMEQ in the presence of HESS5 cells and 100 ng/mL of TNF-α for 48 h, the NF-κB signal was abolished after 6 h, but partially restored as early as 12 h (Fig. 4A). Compatible with this observation, the EMSA showed that the NF-κB binding activity of IMS-PhL1 cells with or without HESS5 cells was reduced by DHMEQ at 6 h, but was restored at 12 h only in DHMEQ-treated IMS-PhL1 cells cultured in the presence of HESS5 cells (Fig. 4B). Together, the data indicate that stromal cells attenuate the effects of DHMEQ on NF-κB activity in Ph+ALL cells.
In vivo imaging of NF-κB activity during Ph+ALL progression and its correlation with TNF-α expression. In the present study, BLI analysis was used to monitor the NF-κB activity of IMS-PhL1 and Sup-B15 cells during their progression in NOD-SCID mice; EF1α/Luc-transduced Ph+ALL cells were used as a control to estimate the whole leukemic cell mass in transplanted mice. In the case of IMS-PhL1 cells, the EF1α/Luc signal, representing the leukemia burden, could be detected as early as 3–4 days after transplantation, propagated mainly across the bone marrow and spleen, and it increased in intensity rapidly over time. Conversely, a rather weak NF-κB signal appeared as late as 3 weeks after transplantation mainly in the hepatic region and later in the bone marrow (Fig. 5A). Repeated experiments showed similar results. After injection of 500 ng of TNF-α into mice transplanted with IMS-PhL1-NF-κB/Luc cells, the NF-κB signal increased prominently in intensity and was propagated systemically within 6 h, before decreasing gradually to basal levels by 48 h (Fig. 5B). When Sup-B15-NF-κB/Luc cells were transplanted into NOD-SCID mice, the mode and pattern of NF-κB activation differed from that seen following the transplantation of IMS-PhL1 cells. Specifically, following transplantation of Sup-B15-NF-κB/Luc cells, the NF-κB signal was detected first in the bone marrow nearly 6 weeks after transplantation, and later in the liver and spleen (Fig. 5C).
In mice transplanted with IMS-PhL1 cells (both NF-κB/Luc and EF1α/Luc), there was significant infiltration of leukemia cells into the bone marrow and spleen. Leukemic infiltration into the liver was less than that into the bone marrow and spleen, but prominent infiltration was observed into the hepatic perivascular area. Quantitative reverse transcript (QR)-PCR analysis of murine TNF-α mRNA revealed that murine TNF-α expression in diseased liver was 7.8- and 8.2-fold higher than control following transplantation of IMS-PhL1 and Sup-B15 cells, respectively, suggesting that leukemic infiltration results in an approximate eightfold induction of TNF-α expression in the liver. These data were normalized against murine GAPDH mRNA copy numbers, representing murine tissue mass, in each sample (Fig. 5D) and, after adjustment, the results were compatible with those of the immunohistochemical detection of murine TNF-α (Fig. 5E). Furthermore, murine TNF-α expression in Sup-B15- and IMS-PhL1-infiltrated bone marrow was increased 2.3 and 1.4-fold, respectively, compared with that in control mice. Conversely, the expression of murine TNF-α in Sup-B15- but not IMS-PhL1-infiltrated spleen was 1.4-fold higher than control (Fig. 5D). These results indicate that murine TNF-α is significantly expressed in the lesions in which the NF-κB signal in leukemia cells was detectable in vivo.
In the present study, we analyzed endogenous NF-κB activity in intact Ph+ALL cells and investigated the effects of stromal cells, cytokines, and therapeutic agents. A number of explanations have been proposed for the refractoriness of Ph+ALL, including endogenous NF-κB activation(27–29) and microenvironmental support.(24–26) Herein, we provide some evidence that these two factors are linked by TNF-α.
In the present study, we used FACS-sorted clones that stably harbored an NF-κB/Luc reporter. Using this reporter assay and EMSA, we verified that NF-κB is constitutively active in leukemia cells expressing p210 or p190.(27) In a canonical pathway, activation of NF-κB-dependent gene expression is generally regulated by its nuclear translocation via phosphorylation, ubiquitination, and proteasomal degradation by IκB kinase (IKK). The mechanism by which Bcr-Abl activates NF-κB remains contentious. It has been reported that Bcr-Abl increases the DNA-binding activity of NF-κB in the IL-3-dependent myeloid cell lines DA1 and 32D(28,29) and that Bcr-Abl activated NF-κB-dependent gene expression in the proB cell line Ba/F3 without its nuclear translocation.(28) Nonetheless, significant nuclear accumulation of NF-κB has been demonstrated in primary Ph+ALL cells(27) and cell lines in the present study, suggesting the presence of a canonical pathway for NF-κB activation by Bcr-Abl.
The primary role of NF-κB in Bcr-Abl-induced leukemia also remains to be determined, because following IL-3 withdrawal apoptosis was induced in DA1/Bcr-Abl cells treated with antisense oligonucleotides to p65/RelA,(29) but not in 32D/Bcr-Abl cells transduced with a super-repressor form of IκBα.(28) In the present study, all the Ph+ALL cell lines were quite sensitive to DHMEQ and were rendered apoptotic following treatment, supporting a crucial role for NF-κB in their survival. Collectively, the function of NF-κB may depend on the cell context within the experimental setting.
Tumor necrosis factor-α has been detected in malignant and normal stromal cells from cancer biopsy specimens(36) and it is a major mediator of cancer-related inflammation when chronically produced in the tumor microenvironment.(37) Not only do TNF-α signals activate NF-κB, but they also induce apoptosis. It should be noted that NF-κB activation by TNF-α, ionizing irradiation, or chemotherapeutic drugs is absolutely necessary to escape the induction of apoptosis by these agents.(38,39) One of the peculiar findings of the present study is a synergistic upregulation of NF-κB activity in Ph+ALL cells by the combination of TNF-α and HESS5 cells. This stromal-mediated NF-κB activation requires direct cellular interaction and underlies the inhibitory effects of DHMEQ. In a similar situation, binding of myeloma cells to bone marrow stroma triggered NF-κB-dependent transcription and secretion of IL-6, which mediates the survival of tumor cells.(40) Considering that NF-κB activation of HESS5 is not inhibited by DHMEQ (data not shown), TNF-α-inducible surface molecule(s) on HESS5 may stimulate a certain signaling pathway for activation of NF-κB in Ph+ALL cells. Partial restoration of NF-κB DNA binding activity during treatment suggests either relative DHMEQ insufficiency or a DHMEQ-unrelated pathway for nuclear NF-κB activation. The present coculture system appears to simulate the in vivo tumor microenvironment, where macrophage-derived TNF-α may trigger inflammation and the subsequent cytokine network.
In vivo BLI analysis was suitable to ascertain when and where NF-κB in Ph+ALL cells was activated during their propagation in NOD-SCID mice. In mice transplanted with Ph+ALL cell lines, the highest NF-κB signal was observed in the bone marrow of Sup-B15 leukemic mice. Unexpectedly, the first and highest NF-κB signal was captured in the liver and no significant signal could be obtained from the spleen in IMS-PhL1-transplanted leukemic mice. This distribution of the NF-κB signal is roughly correlated with the inducibility of TNF-α expression. Hepatic infiltration was less prominent than infiltration in the bone marrow and spleen, but TNF-α expression was highest in the liver. In IMS-PhL1-transplanted leukemic mice, the spleen showed a significant infiltration, but negligible expression of TNF-α. By analogy with a previous report that TNF-α produced by tissue macrophages promotes inflammation-associated liver cancers,(41) the initial source of TNF-α in the liver and other organs may be tissue macrophages in response to leukemic infiltration. These results support the idea that TNF-α also triggers microenvironmental upregulation of NF-κB activity in vivo. From a therapeutic point of view, imatinib has a beneficial effect through inhibiting TNF-α production by macrophages in vitro.(42) Nevertheless, NF-κB upregulation in Ph+ALL cells was not prevented by imatinib alone when it was used at suboptimal concentrations, subsequently resulting in imatinib resistance, as shown in Figure 3(D). In a recent report, it was shown that TNF-α induced activation-induced cytidine deaminase (AICD) in human cholangiocarcinoma cells.(43) Adapting the same scenario to the cholangiocarcinoma cells as that in Ph+ALL cells, the induction or upregulation of AICD by TNF-α may lead to mutations in certain genes, especially in the kinase domain of the BCR–ABL gene as reported previously.(44) In conclusion, TNF-α-stimulated changes to the microenvironment may contribute to the progression of Ph+ALL via the synergistic upregulation of NF-κB activity.
This work was supported by the Japan Society for the Promotion of Science (SK) in Japan.