Mechanisms of HPV-induced carcinogenesis

Scientific Projects:

The primary aim of our division is the understanding of the molecular and immunological mechanisms leading to malignant transformation during long-term HPV-host interactions.
Our group contains 3 subprograms which cover mechanistic, aetiological/preventive and translational aspects:

PROJECT 1
THE NET - C-FOS REGULATORY CIRCUIT Lars Krüger, Matthias Sobel, Johanna De Castro Arce, Anna Hitschler. In collaboration with Bohdan Wasylyk (Institute de Genetique et de Biologie Moléculaire et Cellulaire, Illkirch, France).


SUBJECT AND GENERAL AIMS
AP-1 acts as a junction point for many regulatory pathways associated with proliferation, apoptosis, differentiation and neoplastic transformation. The decision which cellular or viral target genes were finally turned on/off is mainly determined by AP-1 composition. Although AP-1 was hitherto considered as a positive regulatory protein in all HPV types, our experiments demonstrated that AP-1 is also a central key element within an intracellular surveillance network negatively controlling HPV transcription and in turn cell proliferation. In fact, there is ample evidence that AP-1 composition determines the in vivo behaviour of HPV-positive cells in nude mice, their sensitivity against growth-inhibitory cytokines as well as their ability to express certain chemokines (Soto et al. 1999; Oncogene 18: 3187; Finzer et al. 2000; Oncogene, 19: 3235). While in non-malignant human cells AP-1 mainly consists of Jun family members heterodimerized with Fra-1, c-Fos was found to be almost completely absent. Conversely, in HPV16/18 positive cervical carcinoma cells or tumorigenic segregants derived from non-malignant somatic cell hybrids, increased amounts of c-Fos and low levels of Fra-1 could be discerned, resulting in a prevalent Jun/c-Fos dimerization pattern (Soto et al., Int. J. Cancer, 86: 811). The hypothesis that c-Fos plays a fundamental role in virus-induced carcinogenesis was mainly corroborated by the fact that ectopic expression of either v-fos (Pei et al. 1993; Virology 196: 855) or c-fos in immortalized HPV-positive cells induces a one step conversion towards malignancy by changing the Jun/Fra-1 ratio in favour to Jun/c-Fos (Soto et al. 1999; Oncogene 18: 3187).
It is therefore the aim of the present project to unravel the role of c-Fos regulation during HPV-induced carcinogenesis.

RESULTS
We have examined the c-fos expression levels in malignant cervical carcinoma cells, in HPV16-immortalized keratinocytes and in other normal human cells. Here, c-fos expression was found to be either low or even undetectable in asynchronous growing normal cells. Conversely, in malignant cells c-fos was constitutively expressed even after serum starvation. Half-life measurements of the corresponding mRNA excluded the possibility that c-fos expression was post-transcriptionally regulated. This supports the notion that the extent and the temporal range of c-fos transcription is apparently differentially controlled in malignant versus non-malignant HPV-positive cells, leading to the idea that inappropriate c-fos expression may contribute to phenotypic changes.
Transcription of c-fos expression in response to stimulation is mainly mediated by three major promoter elements: the sis-inducible element, the serum response element (SRE), the cAMP response element and the AP-1-like site termed c-fos AP-1 site. The majority of the signals enter the c-fos promoter via the SRE, which is constitutively occupied by a dimer of the serum response factor (SRF), forming a complex with members of the ternary complex factor (TCF) family (for review, see Buchwalter et al., 2004; Gene 324: 1). At present, three different members of the TCF family have been identified: Elk-1, Sap1 and Net, where the latter was cloned by the group of our French collaborator (Giovane at al., 1994; Genes Dev. 13: 1502). Using different reporter plasmids, we dissected the c-fos promoter and showed that the expression was mainly controlled by the SRE-motif, which was active in malignant cells, but repressed in their non-malignant counterparts. This suggested that c-fos expression was regulated at the level of initiation of transcription.
In the absence of active MAPK signaling, the TCF family members Net, and to a lesser extent Elk-1, have been implicated in repression of transcription (for review, see Buchwalter et al., 2004; Gene 324: 1). As shown by our collaborating French group, Net can be also converted to an activator by Ras and MAP kinase induced phosphorylation of its activation domain. Notably, the tumour suppressor p53, known to be quantitatively diminished in HPV-positive cervical carcinoma cells, inhibits MAP kinase induced activation of Net (Nakade et al., 2004; Mol. Cell. Biol. 24: 1132). In other words, p53 helps Net to maintain its repressor function.
By using RT-PCR, we found considerable differences in Net expression when malignant and non-malignant HPV-positive cells were compared. While Net was constitutively expressed in normal cells, the corresponding mRNA level was significantly diminished or almost undetectable in tumorigenic cervical carcinoma cells. Chromatin-immunoprecipitation (ChIP) assays revealed that Net directly binds to the SRE nucleoprotein complex of c-fos in non-tumorigenic cells, but not their malignant counterparts. Intriguingly, Net harbours a repression domain, which is not conserved in other TCFs, but is capable to suppress transcription by interacting with CtBP as co-repressor. This connection may provide a possible mechanism for inhibition, since CtBP is involved in the recruitment of histone deacetylase (HDAC) and in turn in silencing of the chromatin (Criqui-Filipe et al., 1999; EMBO J. 18: 3392).
In any case, ectopically expressed net cDNA resulted almost in a complete suppression of SRE- and c-fos-promoter-directed reporter constructs under transient transfection conditions. Conversely, delivery of siRNA against Net resulted in c-fos transcription unequivocally demonstrating a repressive effect of Net in non-tumorigenic human cells. These data indicate that loss of Net and constitutive c-fos expression appear to be a key event in the transformation of cervical cancer cells (van Riggelen et al. 2005; J. Biol. Chem. 280: 3286).


PROJECT 2:
CROSS-TALK OF RAR-β WITH THE EPIGENETIC MACHINERY IN SILENCING HPV EXPRESSION IN CERVICAL CARCINOMA CELLS Johanna De Castro Arce, Elisabeth Schwarz. In cooperation with: Ubaldo Soto (Loma Linda University California, USA), Renske Steenbergen (Department of Pathology, Unit of Molecular Pathology, VU University Medical Center, Amsterdam, The Netherlands).

SUBJECT AND GENERAL AIMS
As shown for many malignant cells, including HPV-positive cervical cancer cells, the retinoic acid receptor-beta(RAR-β), the most potent RAR involved in suppression of tumor-related phenotypes (Pavan et al. 2006; Curr. Med. Chem. 13: 3553), is epigenetically silenced via de novo methylation and chromatin remodelling (Berger & Daxenichler, 2002; J. Steroid Biochem. Mol. Biol. 80: 1). In other words, the absence of RARb provides a selective advantage during multi-step progression to cervical cancer, since the gene is obviously not compatible with unscheduled cell proliferation (Inanova et al., 2002; BMC Cancer 2: 4). Conversely, re-introduction of RARb in malignant cells interferes with anchorage-independent growth and diminishes tumor formation upon heterotransplantation into immunocompromized animals (Geisen et al., 2000; Int. J. Cancer 85: 289). Both RARs and RXRs possess trans-repressive function on the activator protein 1 (AP-1) in a ligand-dependent manner. In the presence of ligands, RARs or RAR/RXR can negatively affect AP-1 either by a direct interaction with Jun/Fos family members (Suzukawa & Colburn 2002; Oncogene 21: 2181) or by disrupting Jun-Fos dimerization (Zhou et al., 1999; Mol. Endocrinol. 13: 276). Retinoic receptors can also compete with the recruitment of transcriptional co-activators (Kamei et al., 1996; Cell 85: 403) or inhibit the c-Jun N-terminal kinase, in turn preventing phosphorylation-dependent activation of c-Jun (Dedieu & Lefebvre 2006; Cell Signal. 18: 889).
We have previously unravelled a mechanism which explains how RARb can also diminish AP-1 activity even in the absence of any ligand. Ectopic re-expression of RARb abrogates AP-1 binding by a post-translational mechanism where c-Jun is selectively degraded by the proteasome and not substituted in the AP-1 complex by other Jun family members. This uncovers a cross-talk between AP-1 and non-liganded RARb and explains, how AP-1 is turned off in the absence of an exogenous stimulus, but can be activated even in the presence of RARb by the mitogen-activated protein (MAP) kinase pathway upon stimulation (De-Castro Arce et al. 2004; J. Biol. Chem. 279: 45408).
To follow up this functional interplay, the present study was aimed to investigate further consequences of ectopic RARb expression in cervical carcinoma cells.

RESULTS
Using HPV-18 positive HeLa cells re-encoding RAR-β2 under the control of a heterologous promoter as a model system, evidence is provided that AP-1 is the only factor involved in the transcriptional regulation of the HPV18 URR which is quantitatively reduced in a ligand-independent manner. Re-expression of RARb was accompanied by a strong selective down-regulation of the viral E6/E7 oncogenes. Decreased oncogene expression was followed by a re-induction of cell cycle inhibitory proteins such as p53, p21CIP1 and p27KIP as well as by a cessation of cellular growth.
As shown by methylation-specific PCR and chromatin studies, reduced transcriptional activity as a consequence of selective AP-1 decay selectively targets the HPV18 upstream regulatory region (URR) for epigenetic modification. This is consistent with the notion that continuous availability of transcription factors can protect CpG islands within the URR from DNA methyltransferase activity (Han et al., Mol. Cell. Biol. 21: 3416). Conversely, deficiency of activating proteins (e.g. AP-1 upon RAR restoration), followed by a reduced transcriptional activity, enhances the probability that regulatory regions become targets for de novo DNA methylation and chromatin remodelling towards a heterochromatic state.
We propose a model where HPV18 gene silencing upon ectopic re-expression of RAR is initially triggered by c-Jun down-regulation. This leads to a reduced transcriptional activity, which in turn labels the viral URR for subsequent epigenetic modification. De novo DNA methylation and following heterochromatinization (or vice versa) are then no longer compensated by the availability of AP-1 and therefore a functional viral enhanceosome (Bouallaga et al. 2000; EMBO Rep. 1 (5): 422) can not be established. This mechanism is otherwise counterbalanced by active viral transcription in malignant cells, since RAR-β itself becomes inactivated during cervical carcinogenesis. Hence, our study shows that the temporal co-existence of a potential repressor and viral oncoproteins is mutual exclusive (De-Castro Arce et al. 2007; J. Biol. Chem., 2007).



DISSECTION OF THE INTERFERON/CHEMOKINE PATHWAY IN NORMAL AND MALIGNANT HPV-POSITIVE CELLS. Bladimiro Rincon Orozco, Rainer Zawatzky. In cooperation with Bohdan Wasylyk (Insitute de Genetique et de Biologie Moléculaire et Cellulaire, Illkirch, France).

Subject and general aims
At what time and to which extend tumor formation takes place is mainly dependent on the immunological status of the respective patient. Hence, tumor appearance during HPV-induced carcinogenesis can be regarded in part as the result of an immunological "escape" process during which certain inter-and intracellular surveillance mechanisms are either functionally ablated or where virus-positive cells are not longer susceptible to immunological control. Beside disturbance of a functional T-cell surveillance, dysregulation of chemokine expression may represent another important event during the multi-step progression to cervical cancer (for review, see Rösl et al., 1999; In: Chemokines and Cancer. Edited by: B.J. Rollins. Humana Press). Chemokines such as the Monocyte-Chemoattractant-Protein-1 (MCP-1) are responsible for the recruitment and activation of T-cells, NK cells and macrophages and are considered as the first line of defence against generalized virus infections (for review, Guidotti & Chisari. 2000; Virology 273: 221). In fact, spontaneous regression of benign warts is accompanied by a strong infiltration of mononuclear cells, where papilloma shrinkage directly correlates with high TNF-* expression in surrounding macrophages. Hence, TNF-* may not only represent a key regulatory cytokine in regression of benign tumors (Hagari et al., 1995; J. Invest. Dermatol. 104: 526), but could also play a pivotal role in the immunological control of dysplastic cervical lesions.
Indeed, TNF-α seems to be a key cytokine with pleiotropic functions, since it can suppress viral transcription and induces MCP-1 as well as interferon-β (IFN-β) gene expression, but only in non-malignant cells (Rösl et al., 1994; J. Virol. 68: 2142; Bachmann et al., 2002; J. Virol. 76: 280; Finzer et al., 2000; Oncogene 19: 3235). Considering this situation in context of the transforming potential of "high-risk" HPVs, it has been recently reported that the interferon pathway seems to be also targeted by viral oncoproteins (for review, see Goodbourn et al., 2000; J. Gen. Virol. 81: 2341). However, our data clearly suggest that the disturbance of the TNF-* mediated IFN-β expression and the loss of an immediate antiviral response is an additional event during multi-step progression to cervical cancer, being more the consequence of the in vivo phenotype of the host cell rather than a direct effect of E6/E7 oncoprotein expression (Bachmann et al., 2002; J. Virol. 76: 280).
In the present research project we try to dissect the type I interferon/chemokine signaling pathway in malignant and non-malignant HPV-positive cells.

RESULTS
Interferon-β signaling in malignant and non-malignant cells
Using somatic cell hybrids as model system, we demonstrated that the conversion to a malignant phenotype was accompanied by the loss of interferon-(IFN)-β synthesis. IFN-β is inducing the "delayed interferon response" and is responsible for signal amplification and finally for the actual antiviral/growth-inhibitory properties of type I interferons. Conversely, all cells independent from the in vivo phenotype could be protected against EMCV lysis when either IFN-α/β or IFN-γ (type II IFN) was exogenously supplemented. Moreover, infection with Sendai virus or Newcastle disease virus readily induces phosphorylation of IRF-3 and IFN-α expression, despite the presence of the viral oncoproteins. These data clearly indicate that only TNF-β induced IFN signaling is perturbed in cervical carcinoma cells, but not the antiviral state in general (Bachmann et al., 2007; Int. J. Cancer 120, 2119).


GENETIC REDUNDANCY IN HUMAN CERVICAL CARCINOMA CELLS: IDENTIFICATION OF CELLS WITH “NORMAL” PROPERTIES Anastasia Bachmann, Rainer Zawatzky

Subject and general aims
Despite the assumption that cancer represents a homogeneous accumulation of descendant cells, initially deriving from a single progenitor (Going, 2000; J. Pathol 200: 1), current hypotheses consider a tumor as a mixed cell population, inherently controlled by a three-dimensional communication network of individual constituents with different phenotypes (Bissell & Radisky, 2001; Nat. Rev. Cancer 1: 46). Consequently, a tumor stands as a synonym for a chaotic, non-linear dynamic system, controlled by a magnitude of genetic and epigenetic alterations, arising randomly, variably and unpredictably in response to external selection mechanisms (Calin et al., 2003; Med. Hypotheses 60: 258; Gatenby & Vincent, 2003; Cancer Res. 63: 6212). Such a heterogeneity facilitates cell survival even under hazardous conditions (Stelling et al., 2004; Cell 118, 675). To get insight in such processes (which may also explain the apprearance of chemotherapeutic resistant cells) is to our opinion one of the most important challenge in cancer research.

In a previous study we could demonstrate that the antiviral effect of TNF-β, which is mediated by IFN-α gene activation, is disturbed in all HPV-positive cervical carcinoma cells investigated so far. The reason for this failure is the lack of interferon-regulatory factor (IRF)-1 and p48 (IRF-9) expression, two key regulators, tightly involved in the transcriptional control of the intermediate and the delayed interferon response. IFN-* expression, however, can be re-established in non-tumorigenic somatic cell hybrids, suggesting that the antiviral effect of TNF-* is a recessive trait, which can be complemented only by genetic means (Bachmann et al., 2002; J. Virol. 76: 280).
On the other hand, considering cancer cells as a highly adaptive system (Schneider & Kulez-Martin, 2004; Carcinogenesis 25: 2033), we have developed a strategy to identify singular cells on the basis of the appearance of a particular phenotype. In this context, we asked whether TNF-β mediated IFN- signaling was irreversibly disturbed in all cells or whether it is possible to recover clones with a functional pathway within a bulk population of non-responsive tumorigenic HPV-positive cells.
To understand how such reversions towards “normality” emerged and how those cells were maintained (or regained) in a cell culture system established since several decades may provide novel insights in the development of human cancer and in spontaneous remissions of certain tumors.

RESULTS
Although it is generally assumed that cancer arises from a singular cell, a tumor must be considered as a dynamic and emergent biological structure, whose organizing principle is determined by genetic and epigenetic modifications, occurring variably in response to microenvironmental selection conditions. Regarding cancer as a non-linear system (Coffey 1998; Nat. Med. 4: 882), which may, even in the absence of an apparent selection pressure, fluctuate between different “metastable” phenotypes, we demonstrated that TNF- mediated IFN- induction is not irreversibly disturbed. Using the IFN- sensitive Encephalomyocarditis virus (EMCV) as a tool to monitor antiviral activity in long-term established malignant HeLa cells, rare IFN-* expressing clones were rescued from a population of non-responsive and EMCV-sensitive cells. Calculations revealed a frequency of 3 x 10–4 to which resistant clones can be obtained. The frequency to which such phenotypes appeared is higher than spontaneous mutation rates, indicating that neither adaptive nor compensatory mutations were responsible (Petterson et al., 2005; Genetics 169: 1105). Antiviral activity was mediated by the re-expression of IRF-1 and p48 (IRF-9), both key regulatory molecules previously found to be suppressed in cervical carcinoma cells. Inoculating selected clones into immunocompromised animals, a reduced or even an absence of tumorigenicity of initially highly malignant cells could be discerned. These data indicate that both the absence of interferon signaling and the ability to form tumors were reversed in a minority of cells (Bachmann et al., 2007; Int. J. Cancer 120, 2119).
Notably, these results can be interpreted as paradigm for the existence of innate genetic redundancy mechanisms (Krakauer & Plotkin, 2002; Proc. Natl. Acad. Sci. U.S.A 99: 1405; Wagner, 2005; Bioessays 27: 176), where a particular phenotype persists and can be isolated without application of drugs generally changing the epigenetic context. In fact, genetic redundancy may account for heterogeneity within a population of tumor cells, allowing either the re-expression or substitution of alternative trans-regulatory genes under defined selective constraints. Redundancy guarantees a kind of dynamic diversity whereby a specific function for a defined biological response can be replaced or compensated (Kitano 2004; Nat. Rev. Genet. 5: 826-837).
It is therefore conceivable that various “differentiation” states are somehow “frozen” within a cell population, enabling the fittest descendant clones to stay alive under appropriate selection conditions (Pardal et al., 2003; Nat. Rev. Cancer 3: 895). Such a complexity and plasticity in a sense of an adaptive, non-linear system may be an explanation why cells with biological properties described here finally appear.


ETIOLOGY AND PREVENTION

MASTOMYS COUCHA: A UNIQUE IN VIVO MODEL TO STUDY PAPILLOMAVIRUS INDUCED CARCINOGENESIS Julia Nafz, Kai Schäfer, Myriam Ohnesorge, Dorothea Muschik, Sen Feng Chen (in cooperation with: Martin Müller and Tim Waterboer, Division of Lutz Gissmann, ATV; Ingo Nindl, Eggerth Stockfleth, Dept. Dermatology; Charité Berlin and Edward Geissler, University of Regensburg, Germany).

Subject and general aims
Extensive studies have approved a causal relationship between high risk human papillomaviruses (HPVs) and the development of cervical cancer. Additionally, particular cutaneous HPVs have transforming properties in vivo and are involved in skin carcinogenesis of immunosuppressed and/or genetically predisposed patients (for review, see Pfister 2003; J. Natl. Cancer Inst. Mono. 52). However, the pathogenic nature of host-virus interactions of most papillomaviruses (PVs) is still poorly understood, mainly hampered by the absence of suitable model systems. The dependency on epithelial differentiation and their strict species specificity are the main hindrance of studying the natural viral life cycle and putative events leading to malignant transformation. During the last years, this disadvantage has been overcome by the use of organotypic cultures which mimic the in vivo differentiation process under tissue culture conditions (Lambert et al., 2005; Meth. Mol. Med. 119: 141). Also, transgenic mice helped to understand the oncogenic potential of singular PV genes, targeting viral gene expression to specific tissues and monitoring their pathological outcome (Riley et al. 2003; Cancer Res. 63: 4862; Schaper et al. 2005; Cancer Res 65, 1394). However, the field of model systems in their natural context is still underdeveloped, but would provide an in toto approach to explore different aspects of carcinogenesis in a complex system of immunocompetent animals.
In this project, we used the soft furred multimammate rat Mastomys coucha (previously assigned to the species Mastomys natalensis) as a small laboratory animal to study papillomavirus-induced carcinogenesis in molecular and immunological terms. The animal colony at the German Cancer Research Center (DKFZ) is the only one worldwide which is latently infected with the Mastomys natalensis papillomavirus (MnPV) (Amtmann et al. 1984; Nature 308, 291). As reported earlier, these animals spontaneously develop multiple benign skin tumours such as papillomas and keratoacanthomas with MnPV as aetiological agent of these lesions (Müller & Gissmann 1978 J. Gen. Virol. 41, 315). MnPV DNA persists episomally without any evidence of integration (Amtmann et al. 1984; Nature 308, 291). MnPV-induced tumours never regress and only proceed after topical application of carcinogens and tumour promotors to squamous cell carcinomas. Moreover, we could recently confirm the oncogenic potential of MnPV in transgenic mice encoding the E6 oncoprotein under the control of the cytokeratin 14 promoter (Helfrich et al., 2004; J. Virol. 78, 4797).
The aims of this project are to study virus-host interactions in its natural host. This context allows analysing
(1) Viral transmission within the colony and viral spread in individual animals
(2) MnPV/McPV2 activation during ageing and under immunosuppressive conditions
(3) molecular and immunological processes during MnPV/McPV2-induced skin carcinogenesis
(4) the role of co-carcinogens in MnPV/McPV2 activation
(5) therapeutic concepts for curing and preventing MnPV/McPV2 -induced lesions (see attached: a short summary of our recently founded cooperation unit between the DKFZ, the Charité and the European SCOP (Skin Care in Organ transplant Patients)–network)
(6) screening for novel papillomaviruses in our unique colony

RESULTS
A) First of all, we systematically analysed different skin sections of tumour-free and tumour-bearing animals differing in age, as well as various organs for virus presence and MnPV-specific gene expression. We have monitored the prevalence of MnPV infections in different skin areas and various organs. We could show that hair follicle cells may serve as reservoir for viral DNA replication and that the amount of MnPV in normal skin may be considered as a predictor for the development of skin tumours in follow-up studies. Notably, MnPV infection is not restricted to the skin but can also be detected in inner organs. This implies that MnPV has a broader tissue specificity than other PVs investigated so far. Since the blood and the lymphatic system were also found to be virus-positive, a hematogenic propagation of MnPV must be assumed. However, MnPV is not transmitted through the germ-line since fetuses lack viral DNA despite infection of the mother. Moreover, in situ hybridisation studies demonstrate for the first time that MnPV can also latently persists in the brain. Hence, Mastomys coucha is not only useful to study papillomavirus-induced skin carcinogenesis, but may also serve as a model to identify additional, still unknown target cells of PV infections and the potential pathological impact (Nafz et al., 2007; J. Gen. Virol., in press).
B) Remarkably, these animals also developed warty-like lesions at anogenital regions (vulva, anus, penis), but the biopsies surprisingly revealed low incidence rates or even a complete absence of MnPV genomes. This raised the question whether a different virus might be aetiologically involved. Taking advantage of the modified rolling circle amplification (RCA) method (Rector et al., 2004; J. Virol. 78: 4933), we have analysed those lesions for the presence of additional papillomavirus types. Here, we identified and characterised a novel virus, referred as Mastomys coucha papillomavirus 2 (McPV2). McPV2 has a similar tropism as previously described for MnPV (Nafz et al., 2007; submitted), but the latter is apparently far abundant in our colony. We have analysed the distribution and transcription pattern within its natural host. To date, this is the first animal papillomavirus found in anogenital lesions.

TRANSLATIONAL ASPECTS

BYPASSING HPV TRANSFORMATION BY HISTONE DEACETYLASE INHIBITORS Katalin Darvas, Handan Karaduman. In cooperation with Prof. Peter Krammer; DKFZ Heidelberg.

Subject and Aims
Although the final outgrowth of cancer requires a coordinate change of gene expression, this process is apparently still controlled by a superimposed mechanism referred as epigenetics. Both de novo DNA methylation and histone modification act functionally together in a dynamic, cooperative and reversible fashion, leading to the conceptional idea that cancer can also be considered as an epigenetic disease. Hence, an integrated approach to understand such changes in gene expression during tumour progression is of fundamental importance, because epigenetic alterations are potentially reversible. In fact, chromatin remodelling as an epigenetic event has supplied a plethora of potential targets for intervention in cancer. Here, in particular histone deacetylase (HDAC) inhibitors (HDIs) and their potential utilization in cancer and anti-metastatic therapy have received much attention, mostly due to their growth inhibitory, differentiation inducing and anti-invasive properties (Kopelovich et al., 2003; J. Natl. Cancer Inst. 23: 1747). In the current project we have investigated the effect of HDIs in the context of HPV-induced carcinogenesis. Here, we could demonstrate that HDIs can arrest cells at G1 to S transition by entirely circumventing viral oncogene function (Finzer et al. 2001; Oncogene 20: 4768; .Finzer et al., 2002; Virology 304: 265). While HPV expression was still ongoing, viral oncoprotein function became completely neutralized by increased expression of the cyclin-dependent kinase (cdk) inhibitors p21CIP1 and p27KIP1 and concomitant suppression of cdk2 activity. Moreover, we noticed that HDIs can induce an intrinsic form of apoptosis by triggering an E7-dependent degradation of pRB, while the pRB binding partner E2F remained unaffected. Intracellular accumulation of “free” E2F-1 in turn strongly induced the expression of the pro-apoptotic isoforms of the p53 homologue p73, providing a functional link between the presence of free E2F-1 during growth arrest and the subsequent generation of apoptotic cells (Finzer et al., 2004; Oncogene 23: 4807). Disturbing the interaction between E7 and pRB by treating the cells with the serine-protease inhibitor TLCK (Stöppler et al. 1996; Virology 217: 542), pRB degradation and p73 induction is prevented (Karaduman et al, 2007; in preparation).
The present project is aimed at studying the molecular and therapeutic effects of HDIs on growth inhibition and eradication of HPV-positive cells during viral induced carcinogenesis.

RESULTS
Pre-treatment with HDIs renders resistant cervical cancer cells sensitive to death-ligand (type I) induced apoptosis. Already 5 hours incubation with TNF- or TRAIL is sufficient to eradicate about 60 % of pre-treated cells, which are normally completely refractory against TNF-/TRAIL even after long-term application. This was due to a selective transcriptional suppression of c-FLIP, which normally prevents the recruitment of caspase-8 to the death-inducing signalling complex. When c-FLIP is reduced, “type I” or receptor-mediated apoptosis can be favoured (Krueger et al., 2001; Mol. Cell. Biol. 21: 8247). Therefore, ectopic expression of either the short or long splicing variant of c-FLIP in HPV-positive cervical carcinoma cells under the control of a non-HDAC responsive promoter completely abrogates their sensitisation. One major way how c-FLIP is regulated is via which is also controlled by reversible acetylation (Chen et al., 2001, Science 293: 1653). However, using siRNA knock-out approaches, an involvement of NFB can be excluded.
It should be noted that the synergistic effect of a combined HDI/TNF- treatment in eradication of cervical cancer cells is not observed in HPV-negative cells, despite c-FLIP was also found to be down-regulated. Here, no pRB degradation and (therefore ?) no sensitation to HDI/TNF- induced apoptosis could be discerned (Darvas et al., to be submitted)

INTERFERENCE WITH THE ENERGY METABOLISM: AN ALTERNATIVE THERAPEUTIC CONCEPT TO TREAT CERVICAL CANCER Julia Nafz, Johanna DeCastro Arce, Verena Fleig. In cooperation with Dr. Sibylle Mazurek, Institute of Biochemistry and Endocrinology, Giessen, Germany.

Subject and general aims
A hallmark of many tumours is their high glycolytic rate, strongly supporting Warburg´s dictum of a relationship between a glycolytic shift and malignant progression. This has a quite obvious selective advantage. Cells which shift to anaerobic glycolysis not only survive a hypoxic microenvironment, but are also resistant to acid-induced toxicity, caused by secretion of lactic acid during incomplete glucose metabolism. Likewise, chronic acidification can even promote invasion of formerly premalignant cells because adjacent normal populations, which are sensitive to acidosis, are destroyed (for review, see Gatenby & Gillies, 2004; Nat. Rev. Cancer 4, 891).
Since HPV-positive cells also depend on glycolysis for energy production (Vrbacky et al., 2003; Physiol. Res. 52, 545), they are much stronger dependent on that pathway than normal cells. A glycolytic phenotype, which guarantees constant energy supply even when oxygen levels decreased, is apparently also an essential prerequisite to maintain transcriptional activity of the viral oncoproteins E6 and E7 (Maehama et al., 1998; Int. J. Cancer 76, 639). In fact, this relationship seems to be an evolutionary well-designed circuit between sustained HPV transcription and oncoprotein functionality, since E7 itself is capable to reprogram the metabolism of its host cell (Zwerschke et al., 1999; Proc. Natl. Acad. Sci. 96, 1261). Accordingly, for a beneficial virus-host interaction it is reasonable that viral transcription, which continuously requires a high metabolic state for keeping up cell proliferation, is down-regulated by a negative feedback loop when the host cell is sensing low energy conditions.
The general aim is therefore to analyze the influence of nutrient depletion and metabolic stress on HPV-transcription and proliferation of cervical carcinoma versus non-malignant HPV-positive cells. Consequently, development of drugs which selectively interfere with the energy metabolism are of potential clinical relevance.

RESULTS
In this project we focussed our interest on the metabolism of cancer cells and studied the effects of cellular energy depletion in the context of HPV-induced pathogenesis. Based on our previous work (Maehama et al., 1998; Int. J. Cancer 76, 639), we could recently show that cervical carcinoma cells are susceptible to undergo either growth arrest or apoptosis under conditions of metabolic stress induced by 5-aminoimidazole-4-carboxamide-1--D-ribofuranoside (AICAR), a known activator of the AMP-activated protein kinase (AMPK). AMPK represents a central metabolic stress-sensing protein, which attracted increasing attention in cancer research due to its pivotal role in controlling energy consumption and cell proliferation. AMPK is a multi-enzyme complex, which switches off, once activated, many ATP-utilizing processes to sustain energy homeostasis (Luo et al., 2005; Trends Pharmacol. Sci. 26, 69). Our results revealed that AICAR treatment of cervical carcinoma cells led to a reduced binding affinity of the transcription factor AP-1 and in turn to a selective suppression of viral gene expression on the level of initiation of transcription. Promoter exchange experiments identifies the HPV upstream regulatory region as a novel target for energy sensing enzymes such as AMPK.
Moreover, the effect of AICAR on proliferation and survival was dependent on p53 activation and in particular on the presence of LKB1 (Nafz et al. 2007; Biochemical J. 403, 501). LKB1 is an important master serine/threonine kinase, which phosphorylates AMPK and in turn monitors intracellular energy-deprivation. LKB1 is lost in the Peutz-Jeghers syndrome, resulting in gastrointestinale polyposis and enhanced colon cancer risk (Bardeesy et al., 2002; Nature 419, 162). Sporadic mutations/inactivation in the LKB1 gene have been also identified in other cancer entities such as breast, pancreas, lung, prostate, ovary (Marignani, 2005; J. Clin. Pathol. 58: 15) and cervical carcinoma cells. This suggests that depending on the cellular context and the microenvironmental conditons, LKB1 may also act as a tumor suppressor gene.
Taking advantage of HPV18 positive non-tumorigenic and LKB1 expressing somatic cell hybrids, which loose LKB1 expression after conversion to a malignant phenotype, we could show that expression of LKB1 protects cells from apoptosis induced by agents which modulate the ATP:AMP ratio. This may open new strategies in the treatment of cervical cancer, using the absence of LKB1 expression as a predictor for therapeutical success (Nafz et al. 2007; Biochemical J. 403, 501).



A QUANTITATIVE APPROACH TO IFN-/ AND TNF- SIGNALING IN HPV-TRANSFORMED CELLS Simone Hedtrich, Rainer Zawatzky (supported and member of the network “VIROQUANT”: Systems Biology of Virus-Cell Interactions)

Subject and general aims
Cellular signalling pathways handle an enormous range of computations. The fine-tuned balance of virus-host cell interaction permitting persistent or acute infection depends on the modulation of these pathways by viral proteins at multiple levels. To identify viral strategies interfering with anti-viral responses and to predict targets for intervention, we will use a systems biology approach to compare the induction of innate immune responses induced by IFN and viral escape mechanisms exerted by HPV to counteract the antiviral effects. HPV infection can promote the progression towards cervical cancer formation for which the 3 early genes E5, E6 and E7 play a major role.
Several studies in the past have shown that the oncoproteins E6 and E7 from HPV16 or 18 interact with transcription factors activated during IFN induction, i.e. IRF-1 and IRF-3 or during signaling at or downstream of the type I IFN receptor, i.e. Tyk2 and p48/IRF-9. Binding of HPV16 E6 to IRF-3 prevented its transactivation function which ultimately resulted in inhibition of IFN- induction in response to Sendai virus. In another study, E7 was reported to block – via physical interaction – IRF-1 mediated transactivation of the IFN- promoter in transiently transfected cells possibly involving the recruitment of histon-deacetylases (HDAC) to the promoter (Park et al. 2000; J. Biol. Chem. 275: 6764). In addition, E7 is also believed to inhibit the induction of IFN- inducible genes through binding to p48/IRF-9, preventing the formation of the ternary complex IFN-stimulated gene factor 3 (ISGF3) which additionally consists of a STAT1 and 2 heterodimer (Barnard & McMillan 1999; Virology 259: 305). Finally, evidence has been provided that the type I IFN-receptor associated tyrosine kinase Tyk2 is a target of HPV18 E6 (Li et al. 1999; Oncogene 18: 5727). Interaction of E6 and Tyk2 resulted in inhibition of STAT1 and 2 phosphorylation after binding of IFN- to its receptor (for review, see Goodbourn et al. 2000. J. Gen. Virol. 81: 2341-2364).

Summary of own previous work
In our work a highly sensitive assay was used to monitor the response of malignant and non-malignant HPV-positive cells derived from somatic cell hybrids to minute amounts of IFN- induced by treatment with TNF- We found that in HPV18 positive cervical cancer cells tumorigenicity correlated with a loss of IFN- induction following treatment with TNF-. We have speculated that this loss is implicated in the escape from immunological surveillance mechanisms resulting in tumor formation (Bachmann et al. 2002; J. Virol. 76: 280).
More precisely, our results have documented a high sensitivity to IFN in HPV positive cells expressing E6 and E7 oncoproteins as well as a normal inducibility for IFN synthesis using Newcastle Disease Virus or Sendai virus. These findings argue against functional ablation of IRF-1, IRF-9 or IRF-3 in these cells (Ronco et al. 1998; Genes Dev. 12: 2061). Furthermore, MxA and IFRG28 as known representatives of IFN-responsive genes showed a normal induction profile in these cells.
Our data are in agreement with previous studies (Alazawi et al. 2002; Cancer Res. 62: 6959; Pett et al. 2006; Proc. Natl. Acad. Sci. U. S. A 103: 3822). using the cervical keratinocyte cell line W12. These cells were generated from a cervical low-grade squamous intraepithelial lesion “naturally” infected with HPV16 and at low passage only contain episomal HPV16. During long-term culture, HPV16 integrates into the host cell genome. Integration is associated with episome loss, increased levels of the E7 oncogene and the endogenous activation of IFN-induced genes, including p48/IRF-9 and IRF-7.

to top