There has been sustained interest in using IFNs and other immunomodulators to treat benign and premalignant HPV-associated lesions; however, inconsistent clinical outcomes have been observed (19
), and the mechanistic bases for these inconsistencies are not yet understood.
In the present study, we addressed the role of the HPV16 E5 oncoprotein, which is expressed only before viral integration but is often lost as lesions progress toward malignancy, in type I IFN signaling. Our data demonstrate that E5, per se, is able to induce IFN-β by a mechanism involving IRF-1. An IFN gene expression signature is similarly induced by E5.
E5, together with E6 and E7, is considered an HPV oncoprotein (52
). The role of E6 and E7 in HPV-mediated cell transformation has been well established (15
), and both proteins have been implicated in IFN host response inhibition (4
). Similarly, the complete HPV31 genome has been shown to suppress some interferon-inducible genes in keratinocytes (8
). Conversely, the role of E5 in HPV-mediated carcinogenesis is still poorly understood, and its impact on the IFN system has not been yet defined. Interestingly, while E6 and E7 are expressed throughout the course of the disease and are required for transformed phenotype maintenance, E5 gene is often lost as lesions progress toward malignancy (47
). This suggests that E5 may play a critical role in the early phases of cell transformation before virus integration. Recently, in a proposed model of HPV16-induced cervical carcinoma, the loss of regulatory episomes appears to be a crucial step for selection of integrants and cancer progression. This event has been associated with a transient antiviral state inducible by type I IFN (21
). Our results, showing that HPV16 E5 stimulates IFN-β and ISGs, make this protein a good candidate as the trigger to the transition phase from premalignant to malignant lesions when episomal viral forms are lost and expression of integrants is deregulated.
Here we have also demonstrated that the increase in E5-mediated IFN-β expression is due to transcriptional activation. The induction of type I IFN is mainly regulated at the transcriptional level, wherein IRFs play central roles. Specifically, IRF-3 and IRF-7 have been implicated as main regulators of IFN gene transcription, with essential and distinct roles, and are responsible for prompt synthesis in response to infections. Indeed, IRF-3 is constitutively expressed in a variety of cell types, whereas IRF-7 is expressed at high levels only in plasmacytoid dendritic cells and is upregulated by IFN, lipopolysaccharide, and viral infection in most cell types. IRF-3 and IRF-7 are both activated by phosphorylation and then translocate into the nucleus, where they stimulate expression of target genes (22
). We found that the E5 protein did not activate either IRF-3 or IRF-7. Given the paucity of cell lines suitable for de novo
HPV infection, it is difficult to study different aspects of virus replication and IFN stimulation. We therefore cannot rule out that in the setting of HPV natural infection, some viral replication intermediates may signal through cytosolic receptors, as recently demonstrated for Epstein-Barr virus (1
), and induce IFN by activating IRF-3 and/or IRF-7.
Nevertheless, our data indicate that, in the absence of this classical activation pathway, E5 yet stimulates IFN-β gene transcription by means of IRF-1. On the other hand, the possibility that stimulation of IFN-β expression occurs independently from IRF-1 in E5-expressing cells and that IRF-1 stimulation is a consequence of IFN-β production was ruled out by IFN type I neutralization experiments that showed that the addition of IFN-α/β-neutralizing antibodies did not abolish IRF-1 stimulation induced by the E5 protein.
The casual relationship between IRF-1 stimulation and induction of IFN-β was definitely demonstrated in the experiments with IRF-1 silencing. Cells expressing specific IRF-1-targeting siRNA, in which the E5-mediated IRF-1 stimulation was inhibited by 80%, no longer produced IFN-β upon E5 expression. Nevertheless, since it is known that IFN-β in turn induces IRF-1, we can thus foresee that, in E5-expressing cells, an autocrine positive loop between IRF-1 and IFN-β does exist that leads to a substantial stimulation of an IFN signature.
At variance with IRF-3 and IRF-7, which are activated by phosphorylation, the different levels of IRF-1 expression are, indeed, one of the factors that dictate how it functions. Most cell types do not express IRF-1 at detectable levels, but its expression is rapidly induced at the transcriptional level following virus infection or exposure to various inducers, including IFNs and proinflammatory cytokines. We have shown that stimulation of IRF-1 expression by HPV16 E5 occurs through both the STAT1 and NF-κB consensus sequences on the IRF-1
gene promoter. In agreement with these results, it has been reported that HPV16 E5 activates NF-κB (27
). Conversely, so far no data on the modulation of STAT1 activity by HPV16 E5 have been reported. It will, therefore, be interesting to study the interference of this protein with signaling pathways triggered by IFN receptor engagement.
IRF-1 is a pleiotropic transcription factor that is critical for cell defense against viral infections but also crucial for the development of both the innate and adaptive immune responses (5
). High IRF-1 expression levels are required for the induction of a set of target genes, including ISGs (20
). Notably, we have shown that E5 protein up-modulates IRF-1 expression at levels sufficient to stimulate ISG transcription. Indeed, in E5-expressing cells, IRF-1 upregulation in turn affects host gene expression mediated by IRF-E/ISRE.
Among the E5/IRF-1-stimulated genes, we identified PKR and caspase 8, specific targets of IRF-1 (28
). The role of PKR in the virus-induced antiviral response has been extensively demonstrated (16
); however, recent reports also demonstrate that PKR is constitutively active in a variety of tumors and is required for tumor maintenance and growth (25
Caspase 8 belongs to a family of proteases and plays a key role in apoptosis (14
), but several nonapoptotic roles have been described, including promotion of cell motility, adhesion, and migration of cancer cells (31
). Interestingly, caspase 8, together with caspase 10, has been also reported as an essential component that mediates NF-κB-dependent inflammatory responses in antiviral signaling (53
Interestingly, chronic inflammation has been identified as a cofactor for HPV-driven cervical carcinogenesis (7
), and IRF-1 is induced by and is an effector of inflammatory cytokines. This raises the possibility that IRF-1, as already reported for E5-induced NF-κΒ (27
), can mediate an inflammatory response during HPV infection. E5, through IRF-1 upregulation, can thus affect host gene expression on a more global setting.
Notably, we also demonstrated the functional relationship between E5 expression, IRF-1, and IFN signature stimulation in the unique model of HPV16-related cervical squamous carcinogenesis W12 cells (23
). This cell model has been extensively characterized, and it has been shown that it accurately resembles cervical neoplastic progression during long-term culture, with spontaneous transition from cells containing episomal HPV16 to a population containing only integrated HPV16 (2
). It will therefore be interesting to assess whether the events that we have reported here are also seen in patients, based on analysis of IRF-1 expression and function during both active viral replication and selection of integrants. This could in part explain the various clinical outcomes of IFN therapy of papillomatoses and the differential regulation of HPV expression observed in various cancer cell lines upon IFN treatment (9
). Similarly, studies on the effect of E5 protein on IFN expression in the natural context of the complete viral genome during a de novo
HPV infection would help to further elucidate its normal functions, as already reported for E5 activity on cell proliferation (40
From our results, we can hypothesize the following scenario: during HPV productive episomal infection, stimulation of IRF-1 and IFN-β by E5 or other unchecked factors leads to the establishment of an antiviral state that, as reported, may accelerate episomal clearance (21
). As soon as only integrated virus is transcriptionally active, E5 is no longer expressed, and the IFN–IRF-1–mediated response is shut down. Meanwhile, high expression levels of the E6 and E7 oncogenes lead to inhibition of expression and/or activity of IFN synthesis key factors and the IFN-induced signal transduction pathway.
Modulation of IRF-1 expression by viral proteins has been reported for different viruses and can result either in inhibition of its expression, as reported for HPV E7 (36
) and HCV core protein (10
), or in up-modulation, as we demonstrated in the setting of HIV-1 infection (41
) and here for HPV16 E5. The final output of this modulation, however, can be always regarded as a bright way of hijacking a key immune regulator to turn its activity to the virus's advantage.
In conclusion, we have presented data defining a critical role of the HPV16 E5 protein in IFN-β expression stimulation mediated by increased IRF-1 protein accumulation in human keratinocytes. An IFN gene expression signature was similarly induced. Our findings highlight an important function for the HR-HPV E5 oncogene and once more underline that great care should be taken during interferon treatment of HR-HPV-associated cervical lesions.
Inhibition of E5 and/or IRF-1 expression may also represent a new avenue for therapeutic interventions in early phases of infection, when activation of the innate immune response to protect may instead play an important role in disease progression.