c-MYC activation impairs the NF-κB and the interferon response: Implications for the pathogenesis of Burkitt's lymphoma
Article first published online: 8 JAN 2007
Copyright © 2007 Wiley-Liss, Inc.
International Journal of Cancer
Volume 120, Issue 7, pages 1387–1395, 1 April 2007
How to Cite
Schlee, M., Hölzel, M., Bernard, S., Mailhammer, R., Schuhmacher, M., Reschke, J., Eick, D., Marinkovic, D., Wirth, T., Rosenwald, A., Staudt, L. M., Eilers, M., Baran-Marszak, F., Fagard, R., Feuillard, J., Laux, G. and Bornkamm, G. W. (2007), c-MYC activation impairs the NF-κB and the interferon response: Implications for the pathogenesis of Burkitt's lymphoma. Int. J. Cancer, 120: 1387–1395. doi: 10.1002/ijc.22372
- Issue published online: 30 JAN 2007
- Article first published online: 8 JAN 2007
- Manuscript Accepted: 14 AUG 2006
- Manuscript Received: 8 JUL 2006
- Wilhelm Sander Stiftung
- Deutsche Forschungsgemeinschaft. Grant Number: SFB455
- Deutsche Krebshilfe
- Fonds der Chemischen Industrie
- The Intramural Research Program of the NIH, Center for Cancer Research, National Cancer Institute
- Boehringer Ingelheim Stiftung
- Burkitt's lymphoma;
- antigen presentation
Deregulation of the proto-oncogene c-myc is a key event in the pathogenesis of many tumors. A paradigm is the activation of the c-myc gene by chromosomal translocations in Burkitt lymphoma (BL). Despite expression of a restricted set of Epstein–Barr viral (EBV) antigens, BL cells are not recognized by antigen-specific cytotoxic T cells (CTLs) because of their inability to process and present HLA class I-restricted antigens. In contrast, cells of EBV-driven posttransplant lymphoproliferative disease (PTLD) are recognized and rejected by EBV-specific CTLs. It is not known whether the poor immunogenicity of BL cells is due to nonexpression of viral antigens, overexpression of c-myc, or both. To understand the basis for immune recognition and escape, we have compared the mRNA expression profiles of BL and EBV-immortalized cells (as PTLD model). Among the genes expressed at low level in BL cells, we have identified many genes involved in the NF-κB and interferon response that play a pivotal role in antigen presentation and immune recognition. Using a cell line in which EBNA2 and c-myc can be regulated at will, we show that c-MYC negatively regulates STAT1, the central player linking the Type-I and Type-II interferon response. Switching off c-myc expression leads to STAT1 induction through a direct and indirect mechanism involving induction of Type-I interferons. c-MYC thus masks an interferon-inducing activity in these cells. Our findings imply that immune escape of tumor cells is not only a matter of in vivo selection but may be additionally promoted by activation of a cellular oncogene. © 2007 Wiley-Liss, Inc.
Burkitt lymphoma (BL) and posttransplant lymphoproliferative disease (PTLD) are two human B-cell malignancies associated with Epstein–Barr virus (EBV). Despite this important commonality, BL and PTLD are completely different clinical entities with different etiology, clinical appearance and treatment. PTLD is an EBV-driven polyclonal or monoclonal B-cell proliferation arising under severe immunosuppression after bone marrow and solid organ transplantation,1 and reflects the fact that EBV is a powerful virus driving B cells into continuous proliferation in the absence of a functional immune system. The expression pattern of viral antigens in PTLD cells is similar to that in EBV-immortalized lymphoblastoid cell lines (LCLs),2, 3 indicating that LCLs may be regarded as an in vitro equivalent of PTLD. Treatment of choice for PTLD is therefore reduction of immunosuppression or application of polyclonal EBV-specific T cells that have been expanded in vitro using autologous4 or partially HLA-matched LCLs5 as stimulator cells.
BL is a high-grade B-cell malignancy with particular epidemiological features. It occurs with high frequency in tropical areas of Africa and New Guinea and with lower frequency all over the world.6, 7 Regardless of its geographical origin, BL is invariably characterized by chromosomal translocations juxtaposing the c-myc oncogene to regulatory elements of one of the immunoglobulin loci, leading to deregulated c-myc expression. Virtually all BL cases from high incidence areas, but only about 15% of the cases from low incidence areas, harbor the viral genome. EBV is thus an important but not indispensable factor in the development of BL, and its contribution to the pathogenesis is still unknown. In EBV-positive BL, expression of viral antigens is restricted to EBNA1 that is essentially required for maintenance of the viral genome as an episome. All other viral antigens required for driving resting B cells into proliferation, including EBNA2 and LMP1, are not expressed in BL cells. In contrast to PTLD, BL is not characterized by breakdown of the immune system, and the T cells limiting EBV-induced proliferation in vivo are still in function.
Comparing the pathogenesis of BL and PTLD, it is a crucial question why BL cells are not recognized by cytotoxic T cell (CTL) in vivo. Several mechanisms may account for the invisibility of BL cells. Clearly, recognition of BL cells by T cells is impaired by the low expression of adhesion and costimulatory molecules, activation markers, as well as HLA class I and II molecules.8 As a consequence, BL cells are not or only poorly recognized by allogeneic T cells in a mixed lymphocyte reaction.9
In addition, intracellular processing and presentation of antigens is functionally impaired in BL cells.10, 11, 12, 13, 14 EBNA1 has long been regarded as a completely nonimmunogenic protein that escapes proteasomal degradation and immune recognition due to the presence of a glycine–alanine repeat.15 However, this escape is neither complete nor the only cause for this phenomenon, as EBNA1-specific CD8-positive T cells do recognize EBNA1 in LCLs, but notably not in BL cells.16, 17 Importantly, even BL tumor cells that express the highly immunogenic EBV antigens EBNA3A, -3B, and -3C in vivo are not recognized by antigen-specific CTL clones.18 Likewise, expression of foreign antigens in BL cells by recombinant vaccinia virus does not lead to T-cell killing by antigen-specific CTLs, whereas CTL killing can be restored by addition of the respective peptide.10, 11, 12, 13 In strong contrast, LCLs and PTLD cells are efficiently recognized and eliminated by CTLs in vitro and in vivo. It has remained an open question whether the poor immunogenicity of BL cells as compared to LCLs and PTLD is caused by the absence of EBNA2 and LMP1, the overexpression of c-myc, or both.
To gain insight into the changes imposed by different EBV genes and the cellular oncogene c-myc, we have established in vitro model systems that recapitulate important features of PTLD and BL cells in vitro. To this end, we have generated a conditionally EBV-immortalized cell line (EREB2) that expresses a hormone-regulatable EBNA2 protein,19 and 2 derivative cell lines that overexpress a constitutively active (A1)20 and a tetracycline-regulatable c-myc gene (P493), respectively.21 The c-myc overexpressing cells can proliferate in the absence of estrogen and adopt the growth pattern and phenotype of BL cells.20, 22, 23 Like BL cells, these cells have downregulated the inducible components of the immune proteasome.14 Most importantly, foreign antigens expressed in these cells, upon vaccinia virus infection, are not recognized in these cells by antigen-specific HLA-matched CTL clones, whereas pulsing with peptide at least partially restored killing.23
To better understand the molecular mechanisms determining (non)immunogenicity of EBV-immortalized cells, BL cells, and the c-myc overexpressing cells, respectively, we have compared their mRNA expression profiles. We show here that many of the differentially expressed genes are regulated by NF-κB and the IFN system. These genes play an important role in antigen presentation and T cell recognition, and are expressed at high level in EBV-immortalized cells and at low level in BL and in c-myc-driven cell lines. We provide evidence for the first time that c-MYC actively downregulates transcription of key players of the interferon response, including STAT1. c-MYC acts at least at 2 levels, directly on the STAT1 promoter and indirectly by impairment of induction of type I interferons.
Material and methods
Cell culture conditions and treatment
Cell lines for microarray analysis: EREB2 cells were growth-arrested (estrogen-deprived for 60 hr) and restimulated for the indicated times after addition of estrogen. P493 cells were growth-arrested by addition of 0.1 μg/ml tetracycline (Sigma, Taufkirchen, Germany) for 60 hr and restimulated by washing out tetracycline. For c-myc dose titration experiments, cells were kept for 48 hr in medium with the indicated tetracycline concentrations. Interferon activity was determined by treating 3 × 106 BL41 cells with 1 ml of supernatant from the indicated cell line for 1 hr.
Lymphochip and Affymetrix microarray analysis
Western blot analysis
Western blot analyses were performed as described.28 Briefly, the blots were blocked for 30 min in TBST-MLK (Tris-buffered saline, 0.1% Tween 20.5% dried skimmed milk), incubated with specific antibodies in TBST/5% BSA for 1 hr, washed 3 times with TBST, incubated with the corresponding horseradish peroxidase-conjugated secondary antibodies for 1.5 hr followed by 3 washes in TBST. The blots were rinsed 2–5 times with water after each step. Bands were visualized by enhanced chemiluminescence on ECL films (Amersham, Freiburg, Germany). Antibodies against the following proteins were used: phospho (tyrosine 701)-STAT1, phospho (serine 727)-STAT1, STAT1 (9H2) (New England Biolabs, Frankfurt, Germany); STAT2, phospho (tyrosine 689)-STAT2 (Upstate, Dundee, UK); STAT1α, Myc, IRF3 (Santa Cruz, Heidelberg, Germany); IRF9 (BD Transduction Laboratories, Heidelberg, Germany). Secondary horseradish peroxidase-coupled antibodies were anti-rabbit IgG (Biorad, München, Germany) and anti-mouse IgG (Promega, Mannheim, Germany).
Real-time PCR analysis
The conditions of the quantitative PCR analysis are given as supplementary information.
Transient reporter assays and plasmids
Vero or HeLa (3 × 105) cells were transfected in 6-well plates with Lipofectamine (Invitrogen) or PolyFect (Qiagen, Hilden, Germany), using 0.5 μg luciferase reporter constructs, 1 or 3 μg of c-Myc or control expression plasmids, and 0.5 μg or 1 μg of CH110 plasmid (Pharmacia, Freiburg, Germany) expressing β-galactosidase, respectively. HeLa and Vero cells were treated for 24 hr with 150 and 100 U/ml IFN-β, respectively. Luciferase activity was normalized to β-galactosidase activity. Details of the cloning of the 5538 bp STAT1-promoter-luciferase construct are given as supplementary information.
Per immunoprecipitation, 2 × 107 P493 cells incubated for 72 hr with and subsequently for 8 hr without tetracycline were subjected to chromatin immunoprecipitations as described.29 DNA precipitated with the respective antibody was quantified by real time PCR using the following primers: Stat1 fwd 5′-atctgtcctctgcctggattctc-3′, Stat1 rev 5′-gcagcccagtcccgtgt-3′, control region fwd 5′-agcaatcaggctggagttgc-3′, and control region rev 5′-cagtggtatcctcagacaggcc-3′.
mRNA expression profiling
To identify genes differentially expressed in EBV-immortalized cells and BL cells, we have applied the lymphochip27 for expression profiling of 3 EBV-positive group I BL lines (Elijah, BL29 and Akata) and 3 cell lines immortalized by EBV-B95.8 (LCL1.11, 1.13 and 1.25). In addition, the EREB2 cell line19 was used for the investigation of gene expression changes caused by activation of EBV latent genes. The c-myc overexpressing EREB2-derivative cell lines A1 and P493 enabled us to examine the direct impact of c-MYC on the mRNA expression pattern (Fig. 1). The expression level of all genes was normalized to that of arrested EREB2 cells, shown in black as middle lane in Figure 2. As we were particularly interested in genes involved in antigen presentation that are modulated in their expression by viral gene products and/or c-MYC, we focused our attention on genes that are expressed at significantly different levels in EBV-immortalized cells as compared to BL cells and cells driven into proliferation by overexpression of c-myc (A1 and P493 cells).
Differential expression of NF-κB target genes in EBV- and c-myc driven cells
Many of the genes expressed at high level in EBV-immortalized cells and at low level in c-myc driven cells are NF-κB target genes30 (Fig. 2a). This group of genes encodes proteins involved in NF-κB and AP-1 signaling, chemokines and chemokine receptors, molecules involved in cell adhesion and activation, and antiapoptotic genes. Their differential expression may be caused by either the reported activation of NF-κB by LMP131, 32 or inversely by the inhibition of NF-κB by c-MYC, which occurs by at least 2 independent mechanisms.33, 34 For some of these genes (e.g. Bcl-2) evidence has indeed been presented that they are negative target genes of c-MYC.35 Also, it has been reported recently that virtually all components of the NF-κB signaling pathway are downregulated in B cells from Eμ-myc transgenic mice.36
IFN response genes are expressed at low level in BL cells and in c-myc driven cells
A second group of differentially expressed genes was defined by its particular expression pattern. These genes were expressed at highest level in arrested EREB2 cells, at very low level in BL, A1 and P493 cells, and at intermediate level in EBV-immortalized cells (Fig. 2b). Notably, the majority of genes of this group represents mediators of the IFN response or is known to be regulated by interferons, such as STAT1, STAT2 and IRF9. Other genes involved in B-cell receptor signaling, the antiproliferative BTG1 gene, cathepsin B and glutathione-S-transferase M1/M2 showed a similar expression pattern. Comparison of our lymphochip data with that described by de Veer et al.37 and Der et al.38 and in the ISG-database (interferon-stimulated genes) revealed that the cluster defined on the lymphochip comprises primarily genes strongly induced by IFNα.
STAT1 is a key mediator of interferon responses, as it is involved in IFNα/β (type I) as well as IFNγ (type II) signaling.39 Upon IFNγ stimulation, STAT1 is phosphorylated at tyrosine 701 (Y701) and serine 727 (S727), homodimerises through reciprocal Y701P-SH2-interactions and accumulates in the nucleus to activate genes carrying IFNγ-activated sequences. Upon IFNα/β (IFN type I) stimulation, STAT1 and STAT2 are phosphorylated at tyrosines and form, together with IRF9, an active IFN-stimulated gene factor 3 (ISGF3) complex that accumulates in the nucleus and activates promoters carrying IFNα/β-stimulated response elements. Owing to their central role in interferon signaling, we further focused on the analysis of STAT1 and the components of the ISGF3 complex.
Low expression and activation of ISGF3 complex components in c-myc driven cells
Western blot analyses were performed to verify the particular expression pattern of STAT1, STAT2 and IRF9, and to analyze the degree of STAT1 and STAT2 phosphorylation. STAT1 protein was indeed found to be expressed at very low level in EBV-positive and EBV-negative BL lines and A1 cells, and at much higher level in LCLs (Fig. 3). STAT1 is phosphorylated at tyrosine 701 and at serine 727 in LCLs, thus allowing homodimerisation or heterodimerisation with STAT2, nuclear accumulation and transcriptional activation.
The other components of the ISGF3 complex, STAT2 and IRF9, are also detected in LCLs, and STAT2 is phosphorylated on tyrosine, indicating that the ISGF3 complex is indeed active in these cells.28, 40 LCL1.11 appears to be exceptional in that the interferon pathway was found to be activated to a lesser extent in these cells as compared to other LCLs (Fig. 3, see also Fig. 2b, 4b).
In accordance with the mRNA expression data, STAT2 expression was less variable in BLs and LCLs when compared to STAT1 and IRF9. These data indicate that expression and activation of components of the ISGF3 complex is significantly lower in cells proliferating due to high c-myc expression than in EBV-immortalized cells.
c-myc and STAT1 RNA follow a reciprocal expression pattern in P493 and EREB2 cells
The experiments described in Figures 2b and 3 suggested that genes of the IFN response might be regulated by c-MYC. As this had not been reported previously, we carefully reevaluated our finding in the conditional cell lines P493 and EREB2. In 4 independent experiments, P493 cells were arrested and restimulated by tetracycline-mediated modulation of c-myc levels, and the c-myc and STAT1 mRNA levels were evaluated using Affymetrix chips.
These experiments clearly showed that STAT1 RNA is downregulated by a factor of 2.5 when c-myc expression is induced (Fig. 4a). Likewise, reciprocal expression of c-myc and STAT1 was observed in arrested and restimulated EREB2 cells using the lymphochip (Fig. 4b). As the c-myc gene is a direct target of EBNA2,41 its activation causes a significant increase in c-myc expression that reaches a transient peak after 3–8 hr. This dramatic change in c-myc expression was accompanied by a reciprocal decrease of STAT1 expression (Fig. 4b).
For an independent verification of these findings, Northern blot analyses of more detailed time-course experiments were performed (Fig. 4c, left panel). As before, EBNA2 activation in EREB2 cells caused a rapid and strong c-myc induction that peaked transiently after 6–8 hr and was mirrored by STAT1 expression in a reciprocal fashion with slight delay. At later time points, EBNA2 also exerts a positive effect on STAT1 expression through its viral target gene LMP1.28, 40
In P493 cells, downregulation of c-myc was also accompanied by a reciprocal slight increase in STAT1 expression during the first 6–8 hr. Unexpectedly, there was a second and much more pronounced increase in STAT1 expression that peaked after 24 hr (Fig. 4c, right panel), and clearly exceeded the decrease observed when the c-myc gene is switched on in P493 cells (Fig. 4a). Northern blot analysis confirmed this quantitative difference: STAT1 decreased slightly upon c-myc induction and increased to a much higher extent when c-myc was switched off (Fig. 4d). We will come back to this issue later.
STAT1 is a negative target gene of c-MYC
As STAT1 expression was highest in quiescent EREB2 cells, we next examined whether enforced overexpression of c-MYC can overcome the upregulation of STAT1 that is observed when EBNA2 is inactivated by estrogen withdrawal. To this end, estrogen-treated proliferating EREB2 cells were transfected with a novel tetracycline-regulatable bicistronic tet-on vector42 that encodes on one hand c-MYC or luciferase as a control, and on the other hand a truncated NGF receptor that allows isolation of transfected cells. Twenty-four hours after transfection, EBNA2 was inactivated by estrogen withdrawal and at the same time c-myc expression induced by addition of doxycycline. Transfected cells were harvested after additional 48 hr and analyzed by Western blotting for STAT1 and c-MYC. As shown in Figure 4e, STAT1 protein levels were significantly reduced in c-MYC overexpressing cells, indicating that the increase in STAT1 expression observed in EREB2 cells upon estrogen withdrawal is caused by c-MYC downregulation.
To also demonstrate a causal relationship between c-MYC and STAT1 expression in P493 cells, a dose–response experiment was performed. Addition of increasing concentrations of tetracycline to proliferating P493 cells led to a gradual decrease in c-MYC and a reciprocal increase in STAT1 expression (Fig. 4f). The reciprocal relationship between c-MYC and STAT1 expression is highly significant (Spearman-rank correlation MYC-STAT1: r = −0.95).
Direct and indirect regulation of STAT1 transcription by c-MYC
The c-MYC switch-on and switch-off experiments in Figures 4a–d show an apparent discrepancy in the order of magnitude of STAT1 induction and repression, respectively, depending on whether c-MYC has been switched off or on. Technically, the main difference between the 2 approaches is that c-myc is switched off by addition of tetracycline to P493 cells, while switch-on experiments require successive washing steps in medium lacking (or, as control, containing) tetracycline. Conceptually, the stronger increase in STAT1 expression upon addition of tetracycline may thus be caused by factors secreted into the medium that would be removed by the washing steps.
The ISGF3 complex is known to be induced and activated by type I interferons. To ask whether type I IFNs are induced upon c-myc inhibition, proliferating P493 cells were treated with tetracycline, and IFNβ and IFNα2 expression was monitored by quantitative RT-PCR. As shown in Figure 5a, the amount of IFNβ mRNA indeed started to increase gradually after about 4 hr, reached a maximum after 24 hr of 25- and 5-fold induction for IFNβ and IFNα2, respectively, and rapidly declined thereafter. IFNγ was not induced under the conditions tested (data not shown).
To monitor secretion of IFN by the cells, supernatants were harvested at the various time points, added to BL41 cells, and phosphorylated tyrosine 701 in STAT1 was detected. As shown in Figure 5b, IFN activity accumulated in the supernatants of tetracycline-treated P493 cells concomitantly with STAT1, Y-701 phosphorylated STAT1 and STAT2 (Fig. 5c). These data indicate that c-MYC negatively regulates expression of IFNβ mRNA and of at least some members of the IFNα family in P493 cells.
The question whether an interferon-independent mechanism also contributes to STAT1 regulation by c-MYC was addressed by 3 types of independent experiments. First, we cloned the STAT1 promoter (including the first exon, first intron and start of the second exon corresponding to the 5′-untranslated region of STAT1 mRNA) in front of a luciferase gene to study regulation of the STAT1 promoter by c-MYC in HeLa and Vero cells. As Vero cells lack endogenous type I interferon genes,24 interpretation is not confused by the presence of endogenous IFNα and -β. c-MYC clearly inhibits the activation of the STAT1 promoter by IFNβ in HeLa (Fig. 6a) as well as in Vero cells (Fig. 6b). To assess whether c-MYC directly binds to the STAT1 promoter, we have performed chromatin immunoprecipitation experiments in P493 (Fig. 6c) as well as HeLa cells (data not shown). The amount of STAT1 promoter DNA bound to c-MYC was by a factor of 6 (HeLa) to 13 (P493) higher than that bound to control antibodies. Miz-1, a protein known to mediate repression of the genes encoding the CDK inhibitors p15INK4b and p21 by c-MYC,29 was not detectable to a significant extent at the STAT1 promoter (data not shown).
In a third approach, we wished to exclude a role of EBV in the observed regulation of STAT1 by c-myc, as EBV might predispose the cells to interferon production. We therefore made use of a conditional murine (i.e. EBV-negative) B-cell lymphoma induced by a tetracycline-regulated c-myc gene.25 In a derivative cell line, the inhibition of c-MYC indeed led to induction of STAT1 protein (Fig. 6d), although to a lesser extent than in P493 cells. Differently to P493 cells, though, supernatants of tetracycline-treated murine lymphoma cells did not harbor STAT1 tyrosine 701 phosphorylating IFN activity (data not shown). We conclude that c-MYC not only suppresses induction of interferon(s) in P493 cells, it also acts directly at the STAT1 promoter and downregulates STAT1 expression in the absence of IFN.
We have performed an expression profiling approach to identify genes that may contribute to the nonimmunogenicity of BL cells. Our data show that c-MYC activation may directly promote immune escape by at least 2 distinct mechanisms: the downregulation of many genes involved in the NF-κB response, and a remarkable and multifaceted impairment of the cellular interferon response. STAT1, the central player in type I and type II IFN signaling, is negatively regulated by c-MYC via direct inhibition of transcription as well as via an indirect mechanism that involves impairment of type I interferon production.
These data provide a perfect explanation for the previously described defect in antigen processing and presentation in BL cells and c-myc overexpressing cells that abolishes their recognition by HLA class I restricted CTLs, even upon overexpression of the respective antigens.10, 11, 12, 23 This phenotype includes downregulation of adhesion and costimulatory molecules, of HLA class I and class II, of inducible components of the immune proteasome, as well as of the peptide transporters TAP1 and TAP2.10, 11, 12, 13, 14, 16, 22, 23 In all previous studies, the question has remained open whether c-myc overexpression renders cells nonimmunogenic and thus actively contributes to immune escape, or whether nonimmunogenicity of BL and c-myc overexpressing cells is the default phenotype that is passively imposed when expression of the EBV antigens EBNA2 and LMP1 is switched off. Our report now provides an answer to this question and stresses the importance of cellular proto-oncogene activation for immune recognition and escape. Our data imply that immune escape of tumor cells is not only a matter of in vivo selection, but may be additionally promoted or supported by activation of a cellular oncogene such as c-myc.
Our observation that c-MYC downregulates NF-κB responsive genes is in agreement with 2 previous reports describing negative regulation of NF-κB by c-MYC.33, 34 It also supports the notion that the nfkb1 gene is dispensable for c-myc-induced B-cell lymphomagenesis, and that all components of NF-κB signaling are downregulated in B cells from Eμ-myc transgenic mice.36
The finding that many IFN-regulated genes are subject to negative regulation by c-MYC is novel and important in view of the central role of the IFN system in regulation of immune responses. Many genes involved in cell adhesion, T cell activation and antigen presentation as well as in the production of cytokines and chemokines are coregulated by NF-κB and type I and/or type II interferons, indicating that both systems are tightly coupled.43 As c-MYC downregulates the NF-κB as well as the IFN response, c-MYC may be regarded as a global negative modulator of immune function.
It is an important question by which mechanism c-MYC downregulates the NF-κB and IFN responses. Analysis of the STAT1 promoter in an IFNα and IFNβ negative cell line provided evidence that c-MYC downregulates STAT1 expression directly. This is underlined by the fact that the c-MYC protein is bound to the STAT1 promoter. On the other hand, type I interferons secreted by arrested P493 cells contribute to the strong increase in STAT1 mRNA by signaling through the type I IFN receptor. In a B-cell line established from a murine B-cell lymphoma that expresses a tetracycline-regulated c-myc gene, switching off c-myc expression also led to upregulation of STAT1, although at a lesser extent than in P493 cells and without upregulation and secretion of interferons. By switching off c-myc expression in P493 cells, a positive feedback loop is apparently initiated that leads to production and secretion of interferons and concomitant strong upregulation of STAT1. The data imply that P493 cells harbor an intrinsic interferon inducing activity, presumably through the presence of EBV or an EBV gene product, that is repressed by c-MYC and unmasked when c-myc expression is shut off. It will be important to unravel the mechanism how EBV induces interferon production in P493 cells and how this interferon inducing activity is suppressed by c-MYC. It will be most important to see whether EBV-positive BLs harbor a similar interferon-inducing activity that is suppressed by c-MYC.
Owing to the complexity of the IFN system and the involvement of positive feedback loops, it is premature to provide a comprehensive view of how and at which levels c-MYC precisely acts. It is obvious that this complex network can only be unraveled by genetic means. The most important question raised by our findings concerns the possible biological importance of the interplay between c-MYC, STAT1 and the IFN system during normal growth control and tumorigenesis. The antagonism of c-MYC and the IFN system is not restricted to our finding that c-MYC inhibits induction of IFNβ and STAT1. STAT1 is not only regulated by c-MYC, but is itself a negative regulator of c-myc expression.44 c-MYC and STAT1 thus regulate each other at a transcriptional level in a negative feedback loop, favoring that either the c-MYC- or the STAT1-program is active.
Remarkably, besides the reciprocal negative transcriptional regulation of c-MYC and STAT1, there is an antagonism at still another level. IFNγ has been reported to relieve the v-myc-mediated block in TPA-induced differentiation of U937 cells, without affecting the level of v-myc protein expression,45 indicating that IFNγ is able to antagonize and overcome the action of the v-myc oncogene. Likewise, the STAT1 target gene IRF1 was shown to inhibit proliferation of a number of mammalian cell lines,46 to overcome c-myc- and FosB-induced transformation of mouse embryonic fibroblasts47 and to revert the phenotype and tumorigenicity of c-myc + mutated c-Ha-ras transformed cells, without affecting the level of c-myc expression in these cells.48 It has remained an open question, though, whether IRF1 antagonizes the action of c-myc, or the activated c-Ha-ras gene or both in this experimental system.
The pronounced antagonism of c-myc and the interferon system at more than 1 level of regulation suggests that the IFN system may play an important role in physiological and malignant growth control. The first genetic evidence for tumor suppressor activity of an IRF was provided by Holtschke et al., who showed that ICSBP−/− (IRF8−/−) mice develop a chronic myeloid leukemia-like syndrome.49 Remarkably, IRF1−/− but not IRF1+/+ cells, can be transformed in vitro by a mutated ras oncogene.50 In vivo, loss of IRF1 increases the susceptibility to tumor development of mice with a mutated c-Ha-ras transgene or p53 nullizygosity,51 and STAT1 deletion accelerates tumor development in p53−/− mice.52 Takaoka et al. have recently provided evidence that the p53 and the interferon system are closely interconnected,53 and we have shown that fludarabine-induced, p53-mediated cell cycle arrest and apoptosis are dependent on STAT1α in EBV-immortalized cells.54 Additional evidence has been presented linking STAT1 to BRCA1, ATM and DNA damage response.55, 56, 57
In addition to its well known role as an antiviral defense system and its central role in the regulation of the immune response, it is therefore important to reinforce the role of STAT1 and the IFN system in physiological growth control, including DNA repair, cell cycle arrest, senescence and apoptosis, and to comprehensively address the importance of perturbations of the interferon system during tumorigenesis.
We are grateful to Mr. J. Darnell, Mr. J. Hiscott, Mr. J. Pagano, Mr. T. Fujita, Mr. C. Horvath, Mr. T. Taniguchi and Mr. P. Staeheli for IRF and STAT expression and reporter constructs, and Mr. K. Conzelmann for antibodies. We thank Mr. Hans-Jörg Hauser, Mr. Thomas Decker and Mr. Klaus Conzelmann for reagents and many helpful discussions. M.S. received a fellowship from the Boehringer Ingelheim Stiftung. We are most grateful to Mrs. Berit Jungnickel for critically reading the manuscript.
- 6Burkitt's lymphoma, a human cancer model for the study of the multistep development of cancer: proposal of a new scenario. In: Klein G, ed. Advances in viral oncology, vol. 7. New York: Raven, 1987. 173–206., .
- 11Restoration of endogenous antigen processing in Burkitt's lymphoma cells by Epstein–Barr virus latent membrane protein-1: coordinate up-regulation of peptide transporters and HLA-class I antigen expression. Eur J Immunol 1995; 25: 1374–84., , , , , , , , , , , et al.