Cancer research and therapy in the 21st century

Symposium organized by Peter Howley, Arnold Levine and Moshe Yaniv, from 25 to 30 of July 2001.



Bursaux Elisabeth (journaliste, Le Monde, Paris), Clevers Hans (University Hospital, Utrecht, Pays-Вas), Dejean Anne (Institut Pasteur, Paris), Delattre Olivier (Institut Curie, Paris), Harlow Edward (Harvard Medical School, Charlestown, USA), Herrlich Peter (Institute of Toxiсology and Genetics, Karlsruhe, Allemagne), Howley Peter M. (Harvard Medical School, Boston, USA), Kimchi Аdi (Weizmann Institute of Science, Rehovot, Israël), De Lange Titia (The Rockefeller University,  New York, USA), Levine Arnold J. (The Rockefeller University, New York, USA), Mccormick Frank (University of California, San Francisco, USA), Mechta-Grigoriou Fatima (Institut Pasteur, Paris), Oren Moshe (Weizmann Institute of Science, Rehovot, Israël), Orth Gerard (Institut Pasteur, Paris), Ponder Bruce A. (University of Cambridge, Grande-Bretagne), Pouyssegur Jacques (CNRS, Nice), Varmus Harold (Memorial Sloan-Kettering Cancer Center,New York, USA), Versteege Isabella (Institut Curie, Paris), Yaniv Moshe (Institut Pasteur, Paris). 


The meeting on the subject of “Cancer research and thеrapy in the 21st century” was held at Les Treilles, France, on July 25-30 2001. Eighteen speakers covered wide areas of research in cell nuclear architecture (M. Yaniv, A. Dejean, E. Harlow, T. de Lange, O. Delattre,

  1. Versteege), apoptotic mechanisms (A. Levine, M. Oren, Kimсhi), signaling pathways, tumor models and thеrapy (H. Clevers, P. Herrick,
  2. Pouysségur, E Mechta-Grigoriou, H. Varmus and E McCormick) and viruses and human genetics (G. Orth, P. Howley, B. Ponder). There was, of course, considerable overlap between these subjects and a continual focus upon how tumor models, signal transduction pathways, nuclear structures, viral gene products and the genetic contributions of the human host could lead us to cancer prevention and therapies.

The meeting was held at a most interesting and critical time in the history of cancer research. The decade of the 1970’s saw the discovery of oncogenes and the realization that mutations in our own proto­oncogenes contributed to cancer in humans. The 1980’s contributed the concept that tumor suppressor genes help prevent cancers and form the basis for inherited predispositions to some cancers. The 1990’s showed the role that DNA repair genes play, uncovered the critical role of telomerase and telomere shortening and resulted in the assembly of portions of signal transduction pathways. With a new understanding of targets and their fonctions, rational approaches to cancer therapy begin with two classes of agents: 1) antibodies directed against either В-сell dеterminants, Нer2/Neu, or the EGF receptor and 2) drugs directed against oncogene еnzymаtic activities (Gleevac, Bcr-Abl).

But it is already clear that much remains to be done in both preven­tion and therapy. Developing new approaches to drug discovery (chemi­cal genetics, high throughput structures of proteins, novel animal models that better reflect human cancers, etc.), inmunological responses to can­cers and validation of drug targets are ail high priorities. The participants of the meeting formed a real consensus around these concepts.



Cell nuclear architecture

Edward Harlow (Harvard Medical School)

Spatial and temporal organization of DNA synthesis

In primary cells in culture early in S-phase, pulse labelling with BrdU for 15 minutes followed by antibody detection of the labelled DNA in the cell nucleus shows 4-6 discrete sites where DNA is synthesized. Later in S-phase, there are many more sites, which become distributed throughout the nucleus and in vеry late S-phase, the DNA synthetic sites overlay heterochomatin. These sites co-localize with DNA synthetic enzymes and replication fork proteins (orc’s, mcm’s, etc.). This pattern is reproducible in primary cells but not in permanent сell lines. Most inter­esting, the Rb proteins co-localize with the discrete sites of replication early in S-phase and these sites form in the nucleus at GO-G1 with Rb­HDAC in place. Thus, it appears that transcriptional repression (GO-G1) and activation (restriction point) is linked to replication. The regulation of this pattern of lосalized sites changes with high cell density, very low serum or, with cellular senescence. Cellular signaling mау be involved because medium from densely packed cells put upon sparce cell layers can alter the pattern of replication sites in the nucleus.

Thus, it is clear that spatial control of replication sites in the nucleus is tightly regulated and this regulation is altered as normal cells develop into permanent cell form or cancer cells. The genes and their products that control these processes will now peed to be elucidated.

Moshe Yaniv (Pasteur Institute)

Chromatin remodeling, cell viability, and cancer

Chromatin remodeling si as to activate or repress gene transcription involves covalent modification of histones (acetF1ation, phosphorylation, methylation) and the action of chromatin remodeling machines com­posed of multi-protein complexes. A number of such complexes with overlapping biochemical functions co-exist in cells. The SNF-SWI com­plex contains 10-12 proteins including SNF5 and either one of the two Brahma related proteins, Brm and Brg-1, which are DNA dependant, ATPase-helicases related to the Drosophila Brahma and the yeast SNF2/SW12 gene. This complex facilitates the binding of transcription factors to DNA and the initiation of transcription. The helicase activity is essential for hormone receptor (GR and RAR) activated transcription. Brm and Brg-1 are mutually exclusive in the SNF complex and are regu­lated differently. Brg-1 expression is constitutive while Brm synthesis is suppressed by Ras transformation. Introduction of a Brm expressing plasmid into a Ras transformed mouse embryo fibroblast reversed the transformed phenotype.

SNF-5 is a maternally transmitted gene product and is zygotically expressed by the two cell stages. The mouse knockout of SNF5 is lethal around implantation with derived blastocyst cells undergoing apoptosis. Heterozygous mice (40%) develop tumors in the spinal chord, intracra­nial region or subcutaneous regions which may be of neural crest origin. The tumors are ail reduced to homozygosity for the SNF5 gene, con­firming its nature as a tumor suppressor gene.

Isabella Versteege (Сиriе Institute)

SNF5 is a tumor suppressor gene for rhabdoid tumors in humans and mice

Rhabdoid tumors in humans occur in infants (18 months or younger) and are very aggressive (80% lethality). They appear in the kidnеу or brain and the origin of the cell involved is unclear but it can differentiate into neural, epithelial, or muscle cells. Such tumors contar chromosome 22g11.2-1 deletions, which result in the deletion of SNF5 on chromo­some 22. The tumors have both copies of the SNF5 gene lost by muta­tion or reduction to homozygosity. About 10% of the cases of rhabdoid tumors have a mutation in one SNF5 gene in germ cells, which appears to have occurred early in spermatogenesis or oogenesis (the parents are SNF5 wild type). Five such families have been identified. Other tumors including choroid plexis carcinomas (4/6) contained SNF5 mutations but this was not observed in choroid plexis papillomas (0/6).

When the SNF5 cDNA is reintroduced into rhabdoid tumor cells (SNF5-/- cells) the cells arrest in G1. Co-transfection of cyclin D, E, or E1A genes overcomes this arrest in G1. Ail these products act upon Rb. When the SNF5 gene is put into Rb negative cells it does not arrest those cells in G1. These data suggest a mode! where the SNF/SW1 complex must act upon Rb-HDAC to mediate G1 repression. This mechanism is specific to rhabdoid cells; SNF5 does not arrest mouse embryo fibro­blasts in culture in G1.

Anne Dejean (Pasteur Institute)

SUMO: a covalent link between the PML nuclear bodies and the chromatin compartment

Acute Promyelocytic Leukemia (PML) is characterized by the presence of a t (15;17) translocation that produces a fusion protein cor­posed of PML and RAR alpha proteins. Treatment of these leukemias with retinoic acid (RA) leads to a morphological coupled remission by inducing the differentiation of leukemic cells providing the first example of differentiation therapy. The PML protein loсalizes in discrete nuclear domains called PML bodies which are disrupted in an RA reversible man­ner in PML cells. These bodies have been shown to recruit a number of proteins including SP100, p53 and СВЕ PML can undergo covalent modification by the ubiquitin-related SUMO-1 modifier that is both responsible for targeting PML to the nuclear body and formation of these nuclear structures. Arsenic, another potent therapeutic agent for APL, induces the degradation of the PML-RAR alpha protein and subsequent restoration of the PML nuclear bodies. SUMO modification is involved in subcellular compartmentalization and protein-protein interactions (also with RAN-GAP-1, IkB, c-jun, р53) and SUMO can antagonize ubiquination in some of these targets.

The HDAC’s (class I-1, 2, 3 and class II — 4, 5, 7, 8) are sumoylated (class I SUMO and Ub, class II — SUMO only) in vitro and in vivo and an HDAC mutant that cannot be sumoylatеd (К559 R) has a reduced transcriptional repression and deacetylation activity.

The PML proteins have a number of splice variant isoforms and only the PML-4 isoform is able to induce cellular senescence. Isoforms 1, 3 and 4 are able to recruit р53 and CBP into PML bodies so that alone can­not account for the senescence phenotype.

Titia de Lange (The Rockefeller University)

Telomeres and the maintenance of genomic integгity

When human fibroblast cells divide in the absence of telomerase, the telomeres shorten and at a critical length, this signals to Rb andp53 which results in cellular senescence. In В or T cells, short telomeres result in apoptosis. Telomeres are associated with at least two proteins called TRF-1 andTRF-2. TRF-1 is a telomere sequence specific DNA binding protein. It is a dimer that in turn binds to Tin-2 and tankyrasе. Tankyrase is an enzyme that poly ADP-ribosylates TRF-1 which then no longer binds to telomeres and permits telomerase to lengthen the telomeres. TRF-2 binds to the junction of the T-hop where the end of the telomere loops back and base pairs. TRF-2 stabilizes that structure and binds to other proteins: hRAP-1, and NBS-1, which in turn binds to Mre-1 and Rad-50.

A dominant negative TRF-2 protein produces telomeres that behave like shortened telomeres or naked telomeres. End-to-end chromosome fusions occur (dicentric) and bridges between chromosomes at anaphase are observed. The TRF-2 dominant negative mutant plus SV40 T-anti­gen induces cell growth (p53 and Rb are inactivated) but an unstable karyotype is observed in these cells. The over-expression of wild type TRF-2 when the cells are in crisis reduces the level of chromosomal fusions. Clearly, telomere lengths and chromosome end protection by proteins plays a critical role in genomic stability.

Olivier Delattre (Curie Institute)

Pediatric cancer, front bench to bedside

Different kinds of childhood tumors each show distinct patterns of genomic instability in a cell type or tissue type specific manner. In Ewing sarcoma, there is an 11:22 chromosome translocation which produces a fusion protein from the Ews-Fli-1 genes. FI-1 is an Ets family transcrip­tion factor found at an integration site for the Friend leukemia virus in erythroleukemic cells. The Fli-1 gene expressed in cells rives them from G-1 to S phase and increased levels of Myc and Cyclin Dl are seen along with decreased levels of p57Kip-1. A CGH analysis reveals entire chro­mosome gains or losses suggesting that a kinetochore defect leads to mis­segregation of chromosomes.

In retinoblastoma, similar entire chromosome gains or losses are seen. By contrast, liposarcomas amplify loci or delete loci in chromosomes. 1DM-2 gene amplifications are common.

Neuroblastoma type I contains few chromosome abnormalities (some triploidy) and is commonly well differentiated. These tumors have a good prognosis. Neuroblastoma type 2 have Мус amplifications and chromosome 1p:7q translocation (where the reciprocal chromosome is never observed). These tumors are not differentiated and have a poor prognosis. The absence of a reciprocal translocation suggests a break and replication mode! for formation of the translocation.

Apoptotic mechanisms

Arnold J. Levine (The Rockefeller University)

The regulation of p53 mediated apoptosis

The wnt-inducable secreted protein (WISP-1) is encoded by a gene regulated by the Wnt-beta catenin — APСsignal transduction pathwaу. It is one member of a family (CCN-family) of six genes which regulate сell growth, tissue repair, and a subset of which can contribute to tumor сell growth. In mice transgenic for the MMTV-promoter-wnt oncogene, breast tumors arise where the epithelial cells produce Wnt and the adja­cent stromal cells produce WISP-1. In some tumors the carcinoma cells gain the ability to sуnthesize WISP-1. In human colon cancers W1SP-1 is made by the stromal cells (about 60% of the time) while other colon can­cers sуnthesize WISP-1. The gene for WISP-1 is amplified in some colon cancer сell lines. When cells are treated with WISP-1 protein the АKТ protein kinase is activated, and this activation is blocked by antibodies to WISP-1. When these cells are exposed to DNA damaging agents such as UV light or etoposide, wild type p53 is activated, but unlike cells rot treated with W1SP-1, the level of p53 mediated apoptosis is much reduced in W1SP-1 treated cells. In these cells, р53 levels rise and p53 regulated transcription appears normal. Caspase 3 is not cleaved and acti­vated so that АKT must act at the level of mitochondrial release of cytochrome C, its interaction with Apaf-1 or the cleavage of caspase 9. WISP-1, like р53 mutation or the over-expression of 1DM-2 in cancer cells, can block р53 functions and promote tumorgenesis. Antibodies directed against WISP-1 maу be useful in cancer therapу.

Moshe Oren (Weizmann Instituto)

P53, beta catenin and ARP – a paradigm of p53 inactivation in cancer

Deregulation of beta catenin levels and signaling is an early event in several cancers via APС mutations (colon) or beta catenin mutations (some colon, melanoma, gastric). Very frequently, p53 mutations are encountered along with APС mutations suggesting the existence of p53 activating stress signals generated by the preceding oncogene activation.

To test the hypothеsis that the Wnt-beta catenin-APC pathway results in a stress signal that activates р53, resulting in selection against it, sev­eras experiments were performed. The expression of excess beta catenin in cells was shown to result in the stabilization and activation of p53 tran­scriptional асtivity. In primarу mouse fibroblasts, beta catenin over-expres­sion lead to increases in Arf levels and the Arf gene was essential for beta catenin mediated р53 responses as shown by using Arf-null (-/-) fibroblasts. Primary fibroblasts over-expressing beta-catenin showed a senescence phеnotype and this was not observed in cells that lacked Arf or р53 (from -/- mice).

While beta-catenin plus Ras did not contribute to cellular transforma­tion in normal MEF’s, beta-catenin plus Ras transformed MEF’s that were null for p53 or Arf. In a similar manner, Arf levels were reduced in human colorectal cells in culture when axin was over-expressed (reduces beta-catenin levels) or a dominant negative TCF-4 construct was used (blocks beta-catenin mediated transcription).

These data help to explain why Arf promoter silencing (via methуla­tion) occurs in up to 30% of colorectal cancers with prior APС mutations and р53 mutations occur in 60-70% of such cancers.

Adi Kimchi (Weizmann Institute)

DAP-kinase: a pro-apoptotic gene that functions as a tumor suppressor A genetic selection was еmрlоуеd using the introduction of anti-sense cDNA libraries into cells, exposing these cells to a killer cytokine and selecting cells that survived apoptotic death. In this way, six novel pro­apoptotic genes were identified, one of which, DAP-kinase, possesses tumor suppressor activity. DAP kinase is a Cа++/calmodulin regulated serine/threonine kinase localized on actin microfiliments. It contains eight ankyrin repeats and a death domain required for its function. DAP­kinase is regulated by two auto-inhibitory domains: 1) Auto-phosphory­lation of the calmodulin domain reduces calmodulin binding and blocks the active site of the kinase. This site is dephosphorylated when cells are exposed to pro-apoptotic signals such as TNFά. 2) The serine-rich C-ter­minus of DAP-kinase is phosphorylated and this blocks kinase activity.

DAP-kinase activation triggers two caspase independent cell processes that contribute to apoptosis: 1) membrane blebbing due to DAP-kinase phosphorylation of the myosin light chair, 2) the formation of auto­phagosomes which engulf cellular organelles.

In primary fibroblasts, a DAP-kinase c-DNA activates p53 in a Arf­dependant mariner. The oncogenes E2F-1 and c-myс can activate DAP­kinase and a p53 mediated apoptosis can result from this. Thus, DAP­kinase can play a role in detection of oncogene activation. Consistent with this DAP-kinase expression is lost, via DNA methylation in B-cell lymphomas, non small cell lung carcinomas, head and neck tumors, and thуroid carcinomas. A few cases of LOI at the DAPK locus have been reported in colon cancers. Thus, DAP-kinase appears to be a good can­didate for a tumor suppressor gene.

Signaling pathways, tumor models, and cancer therapy

Hans Clevers (University Hospital, Utrecht)

TCF’s in development and cancer

The TCF family is composed of four proteins (TCF-1, LEF, TCF-3, TCF-4) that bind to a 12bp sequence of bases in the minor groove of DNA and bend the DNA. The TCF’s interact with beta-catenin to acti­vate transcription and with Groucho to inactivate transcription. In the mouse, TCF’s are essential for thуmoсytе progenetor differentiation into T-cells. TCF-4 knockout mice have no intestinal crypt cells which develop via signaling of the Wnt-beta-cateninTCF-4 pathway.

Colon tumors have been divided into two phenotypes: Min, with microsatellite instability which can have beta-catenin or axin mutations in the Wnt signal transduction pathway and CEN, with APС mutation and extensive chromosomal instability.

Wild-typе APС binds to microtubles and at the kinetochore of the chromosome through the EB-1 protein that binds to the C-terminus of APС. Colon cancer with APС mutations fails to have EB-1 or APС at the kinetocores. This results in chromosome instability (CIN) and activates the mitotic checkpoint for apoptosis. Bcf-2 rescues this checkpoint medi­ated apoptosis and it is possible that this is a p53-mediatéd event which in turn would put selection pressure for р53 mutations. CIN cancers contain APС andр53 mutations while MIN cancers contain beta-catenin, axin, or conductin mutations and, more rarely, р53 mutations.

Virtually all colon cancer cell fines have a constitutively active Wnt­beta-cateninTCF pathway. The introduction of a dominant negative TCF-4 into such cells shuts off about 30-40 genes at the level of tran­scription and turns on several hundred genes that participate in differen­tiation of the colon crypt cell type to mature terminal cells. Introduction of this dominant negative TCF into colon cells results in differentiation and cessation of cell division. Among the genes up regulated during this process is p21 which would stop cell division and permit differentiation.

Drugs that block the beta-cateninTCF-4 interaction site and so turn off the Wnt-beta-cateninTCF pathway like the dominant-negative TCF­4 are being developed.

Frank McCormick (UCSF Cancer Center)

Integration of signaling pathways and implications for therapy

Three approaches are presently being еmployed to selectively kill can­cer cells: 1) small molecules, 2) antibodies, 3) viruses or viral vectors used for gene thеrapy. Some approaches target the inflammatory cells that invade a cancer which can secrete proteases or gelatinase B which, in turn, are used for tumor spread and angiogenesis. Ideally, one would like an agent that targets a protein in the tumor that is nonessential in normal cells (H-ras, N-ras, abl) or is absent in normal cells (telomerase). In the ras pathway, inhibitors of the Raf kinase (Onyx-Bayer) and MEK (Park­Davis-Pfizer) are being developed. Inhibitors of the EGF receptor (ImClone, Pfizer, Astro-Xen) are also in clinical trials. Inhibitors of the Ras-РI3K AKТ pathway are in earlier development. Ras up regulates 1DM-2 gene transcription via an Ets site which, in turn, down regulates p53. One would predict then that a MEK inhibitor (in the Ras pathway) along with a DNA-damaging agent (activates р53) would kill a р53 wild type cancer cell. In normal cells, Ras activation tins on both Arf and 1DM-2 and Arf blocks 1DM-2 activity. Cancer cells with a mutant-Arf and a wild tyре р53 would clearly be at risk with this therapy while wild type cells would be protected by the Arf protein.

Jacques Pouysségur (CNRS, Nice)

Signaling tumor angiogenesis via МАР kinases and hypoxia

Phosphorylation (activation) of p42 and p44 CAP kinases (ERK1 & 2) results in the entry of these proteins into the nucleus in G1 so as to promote entry into S-phase. They must also then leave the nucleus (via shuttling) or be dephosphorylated by a MAPK phosphotase (MKP-1 or МКР-2) for entry into S-phase. MAPK can phosphorylate MKP and this lengthens its half-life from 30 minutes to 120 minutes. In vascular endothelial cells MAPK signaling is essential for cell division when the cells are at a monolayer and fresh serum is added. MAPK activity also protects against apoptosis when the cells are removed from the substra­tum. This is also true in fibroblasts or MDCK cells and this anti-apop­totic effect is mediated by CAP-K and AKТ kinase.

Ras signaling increases VEGF-A levels and the promotor of this gene shows increased activity after raf activation. The VEGF-A promotor ele-ment can be gel shifted by AP-1 and SP-1. MAPK phosphorylates Sp-1 at two sites on the protein and thus increases its transcriptional activity, and the SP-1 DNA binding sequence is essential for VEGF transcription.

The VEGF promoter also has a 1IF-1 binding site (IRE). 1IF-1 is a basic helix-loop helix heterodimer (alpha and beta sub-units). HIF-1a m-RNA is made constitutively but the protein has a very short half-life which is regulated by an E3 ligase called VIL protein complex which ubiquitinates HIF-1alpha. A proline hydroxylase modifies the pruines in 1IF-1 alpha (HIF-lbeta’s half-life is not regulated) and this is the signal for VIL binding and degradation of 1IF-1 alpha in the nucleus or cyto­plasm. One might well expect that mutations in the proline hydroxylase gene in cancers would supply more VEGF-A and could be selected for. The MAPK phosphorylates 1IF-1 alpha and the activation of Raf in cells leads to transcriptional activation of VEGF via the IRE site in the gene.

Thus, the МАР-К pathway plays a central role in cell division and, сoordinately, angiogenesis.

Peter Herrlich (Institute of Toxicology and Genetics, Forschungszentrunm, Karlsruhe)

A protein complex involved in the regulation of invasive behavior

A variety of metastatic cells express Met, a receptor tyrosine kinase

that can act as an oncogene. Нepatocyte growth factor (HGF) or scatter factor is the ligand for the normal Met receptor but a specific splice vari­ent of CD-44, termed CD44V6, is required as a growth co-promoter for the activation of this receptor. The activated (phosphorylated) receptor binds Grb and activates the Ras pathway. The сytoplasmic tail of CD-44 binds ezrin and organizes cortical actif filaments needed for both cell division and invasion. The Met-CD44 ezrin complex is disrupted if cells contact high molecular weight hyaluronic acid dampening the cell sig­naling and cortical actif formation.

The neurofibromatosis-2 (NF-2) gene is a tumor suppressor. The absence of the NF-2 gene product results predominantly in schwannomas of the eighth cranial nerve. The NF-2 gene product is called schwan­nomin or merlin and expression of it in tumors suppresses tumor growth and the formation of agar colonies. Merlin binds to CD-44 displacing ezrin with a loss of tumorigenic potential. Phosphorylated merlin fails to bind to CD-44 so protein modification regulates merlin-ezrin binding at the same site on CD-44.

Fatima Mechta-Gгigогiou (Pasteur Institute)

Role of AP-1 in embryonic angiogenesis and tumorigenesis

The AP-1 complex plays a central role enabling cells to integrate mul­tiple extracellular signals and initiate the appropriate genetic response.

AP-1 is composed of dimers of Jun (c-Jun, Jung, JunD) and Fos (c-Fos, FosB, Fra-1, Fra-2) proteins. We attempt to unravel AP-1 еоmplеxity by dissecting the specific functions of the Jun proteins using genetic and bio­chemical approaches (overexpression, conditional alleles and gene inacti­vation).

To examine a potential genetic interaction between c-jun and junD, we crossed mice heterozygous for both mutations. White c-jun embryos die at dау 13.5 during development, deletion of the junD gene does not affect embryonic development. Disruption of both genes resulted in embryonic lethality around daу 11, suggesting essential and redundant functions for c-jun and junD during embryogenesis. Although the double mutation did not prevent the formation of primary vascula­ture, deletion of both c-jun andjunD resulted in a complex set of vascu­lar abnormalities, including an extreme enlargement of the dorsal aorta, and major cardiac malformations. The heart of c-jun ” junD ” embryos was almost devoid of trabeculae and the myocardium is abnormally thin. Differentiation of smooth muscle cells is severelу affected by the double mutation whereas expression of specific endothelial differentiated mark-ers seems normal. Nevertheless, the endothelial cells present an erratic orientation and remain collapsed with the myocardium. These data sug­gest that c-Jun and JunD are involved in the reciprocal signalling between endothelial cells and the surrounding mesenchyme required for cardiac morphogenesis and establishment of a mature vasculature.

We have previously established that the balance between different Jun proteins controls progression through the cell cycle, cell growth and sur­vival. JunD negatively regulates Ras signalling and its overexpression partiallу inhibits Ras-mediated transformation by acting on cell cyсlе. JunD misexpression inhibits also growth of blood vessels in tumors induced by Ras. JunD regulates negatively the expression of potent vascular growth factors and decreases the neo-vascularisation of tumors.

Harold Varmus (Меmorial Sloan-Kettering Cancer Center)

Mouse models for two common human cancers: lung and ovary

In order to introduce oncogenes into specific cell or tissue types the gene for the avian leukosis virus receptor was introduced into mice as a transgene controlled by a ubiquitous tissue enhancer, beta actif or an epithelioid cell specific enhancer element. The ovarian cells from such miсе can then be infected by an avian leukosis virus containing an onco­gene (up to 2.5Kb in size) and the cells either inserted back roto nude or isogenic mice into the tissue of origin or subсutaneously. The onco­genes employed were с-Мус, K ras, Akt and Her2/neu and these were crossed into a p53 wild type or knockout background. At 4 weeks after infection, no oncogenes alone produced ovarian tumors in a p53 wild type background and even combinations of с-Мyс and Ras, K-ras and Akt or Mус and Akt failed to produce tumors except in a р53-/- back­ground where tumors were produced. Clеarly, the target cell producing tumors was an epithelial cell. This model comes closer to human cancer paradigms, which require multiple oncogenes and a tumor suppressor gene for tumorgenesis and the use of tumor tissue inserted into an organ of origin.

A lung tumor model was obtained using the K ras oncogene regulated by the CCSP promotor which expressed K-ras in a doxycycline inductable fashion in type II alveola cells or pneumoсytes. Transgenic mice contain­ing this construct show cell division in the lung rapidly after the addition of doxycycline. Ву one month of drug treatment, hyperplasia was observed and by three months large tumors were present. In the long term, both adenocarcinomas and adenomas were observed. Removal of doxycycline treatment results in a rapid shrinking of the tumor or a com­plete reversion with apoptotic cells quite evident. In a p53-/- background the doxycycline induced tumors are more rapidly developed and more aggressive. In the р53-/- tumor removal of doxycycline still results in apoptosis and tumor shrinkage. In an INК4A/knockout mouse (р16 and ARF negative) tumors also arise more rapidly and were more aggressive andapoptosis resulted upon removal of doxycycline. Thus, р53 does not appear to mediate this apoptosis. These kinds of experiments help to val­idate K-ras activity as a target for tumor therapy and thеу begin to eluci­date the pathways that mау be involved in such therapy if an inhibitor of K-ras activity is found.

Human genetics, viruses and cancer

Вruсe Ponder (Cambridge University)

Low penetrance genes for breast cancer: how and why

  1. Ponder and his group are in the process of completing a case con­trolled study to determine the role of normal genetic variation between individuals for their risk of acquiring cancer. To date, high penetrance genes that are predisposed to cancer have been uncovered (BRCA-1, 2, р53, PTEN, Cdk2, Rb). The evidence for genetic predisposition in breast cancer is: 1) familial history of breast cancer is the single highest risk fac­tor for a relative acquiring breast cancer. Because sisters and mothers of the proband are at equal risk, such rare alleles are thought to be domi­nant. 2) Studies with monozygotic and dizygotic twins indicate that about 30% of breast cancers have a genetic component. There is about a two-fold greater probability of developing breast cancer if your monozy­gotic twin had it.
  2. Ponder is looking for genes that have rare alleles that provide increased odds ratios of 1.3-1.4. For this, he uses a case-control study with 2,000 individuals with breast cancer and matched controls. This should detect a 1% occurrence with 85% accuracy. Не uses SNP’s linked to 35 candidate genes. Five SNP’s per gene provides a haplotype cover­ing about 75Кb of DNA. Of the 35 genes studied, seven had some impact upon cancer frequencies. The genes are all recessive and provide odds ratios of 1.2-1.3. The p21 gene has a coding polymorism with a heterozygous frequenсy of 0.54 (homozygotes; .0029). Of 2,000 people studied 21 cases of cancer had this p21 allele while seven were found in the control group. This is an odds ratio of 4.0 but it is not clear what the funсtional impact of this polymorphism is in the p21 protein.

In a second study with 3,000 families that have a high risk of breast cancer (with Beca-1, 2 mutations eliminated) Ponder concluded that 10% of the population is at a risk for 46% of the breast cancers. If one could identify that 10% by allele assignments, this would be of great public health benefit.

Gérard Orth (Pasteur Institute) Genetic control of viral oncogenesis

Human papilloma virus type 5 (HPV-5) is wide-spread but causes the disease Epidermodysplasia verruciformis (EV) only rarely in people. HPV-5 dependent disease requires an autosomal recessive allele of (a) gene(s) which confer(s) a defect in cutaneous cell mediated immunoty. Such genes have been mapped to two loci on 17425 (EV-1) and 2p21-p24 (EV-2). A gene corresponding to the EV-1 allele was isolated by positional cloning. It encodes for a membrane protein with no homology to any known gene. This EV-1 locus overlaps a region previously mapped for susceptibility to psoriasis.

Control of persistence of the Shope papilloma virus in rabbits mapa to the МНС locus in these animals. Two related virus strains have very dif­ferent persistence levels in rabbits, depending upon the genetic back­ground of the rabbit and the sequence of the virus.

These experiments demonstrate that both — the host immune system and viral sequences — control persistance and tumor progression.