|Home | About | Journals | Submit | Contact Us | Français|
Cross talk between p53 and interferon-regulated pathways is implicated in the induction of gene expression by biologic and genotoxic stresses. We demonstrate that the interferon-stimulated gene ISG15 is induced by p53 and that p53 is required for optimal gene induction by double-stranded RNA (dsRNA), but not interferon. Interestingly, virus induces ISG15 in the absence of p53, suggesting that virus and dsRNA employ distinct signaling pathways.
To promote host survival, cells respond to viral challenge by activating both protective and cell-death pathways. Interferon (IFN) is induced in virus-infected cells and functions in a paracrine manner to exert a protective effect on surrounding cells. IFN can also induce apoptosis in an antiviral strategy that sacrifices the infected cell to prevent viral spread (15, 37). Independent of IFN, virus infection or double-stranded RNA (dsRNA) directly activates a subset of interferon-stimulated genes (ISGs) (11, 29). Virus- or dsRNA-induced gene expression occurs through a pathway distinct from the IFN-activated Jak-Stat pathway (3, 12). The interferon regulatory factors (IRFs) 1 and 3 appear to play central roles in virus- or dsRNA-induced gene expression, targeting interferon-stimulated response element (ISRE)-like elements in the promoters of responsive genes (26). In addition, the dsRNA-activated protein kinase, PKR, phosphorylates the NF-κB inhibitor, IκB (21), leading to activation of NF-κB, which is required for the induction of IFN-β and some ISGs by virus and dsRNA (8, 22, 41). By bypassing the requirement for IFN induction, virus-induced gene products are thought to confer immediate protection to the infected cell (11).
The tumor suppressor p53 can effect a protective or a suicidal response to genotoxic stress (1, 2). An accumulation of evidence for cross talk between p53 and the IFN system has implicated p53 in the host response to viral challenge. For example, the transcription factor IRF1, which both is induced by IFN and functions in the regulation of IFN and ISGs (26), cooperates with p53 in the induction of WAF-1 and is essential for radiation-induced cell-cycle arrest (36). The IFN-stimulated gene ISG15 was identified in screens for p53 and radiation-induced genes (17, 27) and thus resembles WAF-1 in being regulated by both IFN and p53 (32). Finally, the IFN-regulated PKR functions in ISG induction by dsRNA (22, 41), interacts with p53, and enhances its transactivating activity (9, 10). Taken together, these studies suggest a role for p53 in the regulation of IFN- or virus-induced genes; however, direct evidence of a p53-dependent step is lacking.
ISG15 is independently induced in response to IFN and virus (3, 29) and was identified in a screen for p53-induced genes (17, 27); therefore, we first examined the direct regulation of ISG15 by p53. ISG15 expression was measured by Western blot in HeLa cells stably transfected with a temperature-sensitive p53 mutant (7). No ISG15 protein was detected in the presence of mutant p53; however, ISG15 expression was induced within 6 h of the shift to 32°C, at which temperature p53 adopts a wild-type conformation (Fig. (Fig.1A).1A). ISG15 expression continued to increase through 24 h at 32°C and reached a level approximately equal to that induced by IFN. ISG15 induction was not due to the change in temperature, as parental HeLa cells did not express ISG15 when cultured at 37 or 32°C. ISG15 induction appeared to be a primary response to p53, as cycloheximide treatment did not prevent the increase in ISG15 mRNA following the temperature shift (Fig. (Fig.1B).1B). Protein synthesis inhibition reduced the basal levels of ISG15 mRNA; however, the relative induction by p53 was equivalent (threefold) in the presence or absence of cycloheximide. Rehybridization of the blot in Fig. Fig.1B1B with probes for other IFN- or dsRNA-responsive genes revealed marginal (IRF1) or no (ISG43) induction, indicating the response to p53 alone is unique to ISG15 in this cell line (data not shown).
The role of the Jak/Stat pathway in the induction of ISG15 by IFN is well established (35), whereas the pathways leading to its induction by virus and dsRNA are less well resolved. To determine if gene induction by these agents is dependent on p53, ISG15 expression was examined by Northern blot analysis of RNA from mouse embryo fibroblasts (MEFs) derived from wild-type (WT) and p53 knockout (KO) mice (13). ISG15 mRNA was induced to similar levels in WT or KO cells following treatment with IFN-α, -β, or -γ (Fig. (Fig.2A).2A). In contrast, dsRNA induced ISG15 mRNA in WT but not in KO cells. To confirm the requirement of p53 for ISG15 induction by dsRNA in human cells, p53 null (HCT-116/379.2) and WT (HCT-116/40.16) human colorectal carcinoma cell lines were employed (6). Similar to the results in MEFs, ISG15 was induced by IFN independent of p53, whereas ISG15 induction by dsRNA was markedly reduced in p53 null cells (Fig. (Fig.2B).2B). A small increase of ISG15 signal was detected in dsRNA-treated p53 null cells, indicating that an attenuated response to dsRNA can occur in the absence of p53.
The cellular response to dsRNA is thought to mimic the effect of dsRNA produced in the course of virus infection. To determine if p53 is required for the induction of ISG15 by virus, p53 WT and null HCT-116 cells were infected with a picornavirus, encephalomyocarditis virus (EMCV), and the paramyxoviruses Newcastle disease virus (NDV) and Sendai virus (SV). Following a 1-h infection at a multiplicity of infection of 1.0, the cells were washed, and the RNA was isolated at 12 h postinfection. Surprisingly, NDV and SV induced ISG15 mRNA independent of p53 status; no induction was observed in EMCV-infected cells (Fig. (Fig.3).3). Densitometric analysis of a shorter exposure of the Northern blot in Fig. Fig.33 and of replicate blots revealed a slight (i.e, 1.5-fold compared to 4-fold for dsRNA treatment) reduction in ISG15 induction by virus in p53-deficient cells (data not shown). To determine if the dsRNA-induced expression of genes other than ISG15 required p53, we examined the expression of ISG43 (24) and IRF-1 in the p53 WT and null cells. Like ISG15, both mRNAs exhibited p53-dependent regulation by dsRNA, whereas induction by virus was largely unaffected by p53 status. The presence of p53, or treatment with dsRNA or virus, did not alter the expression of glyceraldehyde-3-phosphate dehydrogenase mRNA, indicating that the observed effects are specific for ISGs and that equal amounts of RNA are present in all samples (Fig. (Fig.22 and and3).3).
The ISRE and flanking sequence mediate the transcriptional induction of ISG15 by dsRNA (12). To determine if the requirement of p53 for optimal ISG15 induction is conferred through promoter elements, a fragment spanning −353 to +74 of the human ISG15 promoter and first exon was cloned upstream of a luciferase reporter gene. Sequence analysis of the human ISG15 promoter revealed a candidate p53-responsive element from −125 to −116 relative to the start of transcription, which is adjacent to the core ISRE located between −108 and −94 (28). The requirement of p53 for maximal induction of ISG15 was assessed by transfecting the ISG15 promoter-luciferase reporter construct (pGL3/ISG15-Luc) into the WT and p53-null HCT-116 cell lines and treating with IFN, SV, or dsRNA. Consistent with the Northern blot results, there were no differences in luciferase activity between the two cell lines following IFN-α treatment (Fig. (Fig.4).4). dsRNA treatment resulted in a 50% higher induction of luciferase activity in the p53-positive HCT-116 cells compared to the p53-null cells. SV induction of luciferase activity was 20% greater in p53-positive cells. This p53 dependence is greater than that observed for the induction of endogenous ISG15 (Fig. (Fig.3),3), suggesting that promoter elements outside those present in our reporter may function in virus-induced gene expression. These results suggest that p53 modulates dsRNA-induced ISG15 expression at the transcriptional level. The characterization of specific dsRNA-responsive promoter elements is required to dissect the mechanism by which p53 influences dsRNA-induced gene expression.
Virus and dsRNA induce an overlapping set of genes; however, an increasing body of evidence indicates that these agents employ distinct signaling pathways. For example, NDV but not dsRNA can induce IFN-α/β in IRF1−/− and PKR−/− MEFs (30, 41), pointing out the distinct requirements for gene induction by virus and dsRNA and demonstrating an essential role for these factors in dsRNA signaling. Virus or transfected dsRNA activates a latent dsRNA-activated factor (DRAF1) which is comprised, in part, of IRF3 and CREB-binding protein/p300 and is capable of binding the ISRE (39). In virus-infected cells, IRF3 is phosphorylated and translocates to the nucleus (25, 39, 42) where it is required for the induction of early-phase IFN-α/β genes (33) and may function in ISG induction (42). In contrast, dsRNA treatment does not lead to IRF3 phosphorylation (34); thus, IRF3 may serve distinct functions in virus and dsRNA signaling. A cell line defective for the induction of ISGs by dsRNA was competent to induce IRF3 nuclear translocation and 561 mRNA expression in response to virus, providing further evidence of separate dsRNA and viral signaling pathways (16, 23). Our studies indicate that depletion of p53 dramatically reduces dsRNA-induced gene expression, whereas virus-induced gene expression is affected to a lesser degree. This result is consistent with a model in which virus infection induces a dsRNA-mediated, p53-dependent signal and secondary p53-independent signal(s) to induce gene expression. Indeed, ISG expression can be induced by cytomegalovirus glycoprotein or by SV and NDV in the absence of viral transcription, demonstrating the existence of dsRNA-independent viral signals (5, 8).
Our studies indicate that p53 is an important mediator of dsRNA-induced gene expression; however, its relationship to other factors implicated in dsRNA signaling is not yet known. Activation of temperature-sensitive p53 induced ISG15 in the absence of exogenous stimuli (Fig. (Fig.1),1), and genotoxic agents that induce p53 also induce ISG15 (17; our unpublished data). The finding of a putative p53-responsive element in the ISG15 promoter suggests that p53 may directly interact with other dsRNA-activated transcription factors. Indeed, cross talk between p53 and PKR-, NF-κB-, and IRF1-regulated signaling has been reported (9, 31, 36). dsRNA-mediated induction of both ISG15 and IRF1 exhibited p53 dependence; however, IRF1 and ISG induction by dsRNA are thought to proceed through distinct NF-κB- and ISRE-responsive pathways, respectively. Studies in PKR-deficient MEFs implicate PKR in dsRNA induction of IRF1 through an NF-κB site in its promoter (22). In addition, dsRNA induced protein complexes bound to the IFN-γ-responsive elements of the IRF1 promoter in a PKR-dependent manner; this may relate to a role for PKR-stat1 interactions in regulating PKR signaling and stat1 DNA binding (40). In contrast, virus or dsRNA activation of dsRNA-activated factor and ISRE binding does not require PKR (39). Interestingly, the defect in a cell line deficient in the induction of ISGs by dsRNA has been determined to lie upstream of IRF1 and NF-κB activation, but downstream of PKR (23). This relatively early event in the dsRNA signaling may constitute a p53-sensitive step which is common to IRF1 and ISG induction. Indeed, the mutant cells are deficient in the induction of both NF-κB-regulated genes (e.g., IRF1 and IFN-β) and of ISRE-regulated genes (e.g., 561) by dsRNA (23). A complete understanding of how p53 functions in dsRNA signaling first requires a more comprehensive definition of promoter elements required for gene induction by dsRNA.
The identification of a p53-dependent response to dsRNA suggests that dsRNA may be an important signal in other p53-regulated responses. For example, dsRNA produced as a result of direct perturbations to cellular RNA by genotoxic agents (18) or as a secondary effect resulting from aberrant transcription of damaged DNA (38) may function as an intracellular stress sensor or signal. In light of the role of p53 in dsRNA-induced gene expression and increasing evidence that dsRNA may not be an ideal model for virus infection, we propose that dsRNA treatment may be a good model for genotoxic stress. The simultaneous production of dsRNA and activation of p53 in response to genotoxic stress may provide added selectivity and potency in the induction of target genes. IRF1, -3, and -7 are activated by both virus and genotoxins (19, 20, 36), consistent with a role for dsRNA as a common intermediate in stress response pathways. Interestingly, stress agents induce an amino terminus phosphorylation of IRF3 which does not result in nuclear translocation; the biologic function of this modification is not known (34). The extent to which dsRNA is formed in response to various genotoxins and the mechanisms by which a dsRNA signal is transmitted to p53 are areas of future investigation.
We thank Ernest C. Borden, The Cleveland Clinic Foundation, for the generous gift of ISG15 antibody; Bert Vogelstein, The Johns Hopkins University, for providing the HCT-116 cells; Paula Pitha, The Johns Hopkins University, for the NDV and Sendai virus; and Nancy Reich, SUNY, Stonybrook, for ISG15 genomic DNA clones. We are grateful to Carianne Judge and Mingjuan Liu for critical reading of the manuscript.
This work was supported by grant RPG-99-195-01-GMC to B.A.H. from the American Cancer Society.