What have mouse models taught us about cancer biology?

Colloque organisé par Hugues de Thé (Hôpital Saint Louis) et Marco Giovannini (INSERM) du 22 au 27 octobre 2007

Participants

Allan Balmain, Ali Bazarbachi, Anton Berns, Richard Fehon, Jacques Ghysdael, Marco Giovannini, Tyler Jacks, Dominique Lallemand, Daniel Louvard, Andrea I. McClatchey, Fatima Mechta-Grigoriou, Pier Paolo Pandolfi, Luis Parada, Terence Rabbitts, Sylvie Robine, Hugues de Thé

Résumé

Une avancée récente en biologie expérimentale est la possibilité de manipuler le génome de mammifères avec une précision spatio-temporelle inégalée. Parmi les applications de ces techniques, l’obtention chez l’animal de tumeurs grâce à la combinaison d’anomalies génétiques observées dans les tumeurs humaines a constitué un objectif important. Ceci pose des problèmes imparfaitement résolus de la nature des oncogènes altérés, de la cellule d’origine, de la séquence temporelle de ces altérations ou encore des modifications génétiques ou épigénétiques du stroma environnant la tumeur. Les plus grands succès de ce type d’approche ont été obtenus dans les leucémies ou les sarcomes, pour lesquels l’anomalie génétique est souvent unique et la cellule d’origine caractérisée. L’autre modèle qui a été longuement discuté est la neurofibromatose, tumeur neurale provenant de la perte bi-allélique d’un gène suppresseur de tumeur, dont la compréhension a été grandement facilitée par l’utilisation de modèles animaux.

Outre des connaissances très fondamentales sur les mécanismes de contrôle de la prolifération ou de la sénescence, l’utilisation de ces tumeurs expérimentales mimant la pathologie humaine à des fins de recherche de thérapeutique préclinique représente probablement l’un des enjeux majeurs de ces prochaines années. Plusieurs exemples de thérapeutique expérimentale ont été présentés. Certaines de ces thérapeutiques ont été transposées avec succès chez les malades, constituant une preuve de principe du bien fondé de ce type d’approche.

Compte-rendu

The meeting, that was held from October 22 to 27, 2007, gathered a number of experts from Europe and America to explore the contributions of genetic modifications in animals to the current understanding of tumour biology. A particularly important practical aspect of this field is to reach a consensus on whether current animal, and in particular mouse, models can predict the treatment sensitivity of spontaneously occurring human cancers. Several topics were covered, including fine spatio-temporal tuning of oncogene expression, oncogene cooperation, oncogene-triggered changes in cell biology and therapeutic intervention.
The symposium was closed by a general discussion, which reached the conclusion that when faithfully reflecting the spatio-temporal pattern of oncogene expression, animal models could be extremely useful to predict patient response and can be used to optimize patient treatments. It was stressed that active lobbying should be carried out to push for a greater use of pre-clinical data derived from genetically defined mouse models in drug registration procedures. A future meeting comprising government agencies could be organized to inform them on the potential use of the critical information derived from this type of models.

Individual contributions

Anton Berns reported on a high throughput in vivo screen in approximately 1000 mice using MuLV infection to identify new oncogenes and tumour suppressor genes. Analysis of the tumours yielded over 20.000 insertion sites, of which roughly 30% were clustered in almost 600 common insertion sites marking known and unknown proto-oncogenes, (haploinsufficient) tumour suppressor genes, and microRNAs. The screen allows to score for genotype-specific common insertion sites and co-occurrences of mutations. He also discussed models for mesothelioma induced by the conditional inactivation of Nf2 and p53, or Nf2 and Ink4a/Arf in the mesothelial lining. He showed that in vitro propagation of tumour-derived cell-lines irreversible changed the expression profile of these cells. Individual cell-lines carrying the same initiating lesions often showed a very diverse response to cytotoxic drugs, casting doubts about the utility of such cell lines to test for new drugs

Allan Balmain has performed an extensive characterization of the genetic changes that are involved in the development of chemically induced mouse skin and lung tumours from the initiation stage through benign lesions to malignant carcinomas. Activation of ras genes (H-ras in skin, and K-ras in lung) appears to plays a very important role at multiple stages of tumour initiation and progression. Interspecific crosses between inbred Mus musculus and outbred Mus spretus, have allowed him to identify at least fifteen loci showing significant linkage to tumour susceptibility or resistance. Genetically engineered models of ras activation have also been used to identify modifiers that are independent of carcinogen activation or metabolism. These genetic studies have identified interactions between wild type and mutant ras alleles, and have highlighted the importance of ras activation within epithelial stem cells as a major determinant of malignant potential. Studies or gene expression in normal tissues from interspecific backcross mice have revealed patterns of gene expression that are under genetic control, i.e. groups of genes that are linked to the inheritance of specific genomic regions from the musculus or spretus parents.

Tyler Jacks focuses on the use of gene targeting in the mouse to study several aspects of lung cancer development and its response to therapy, using strains carrying Cre-LoxP-regulatable mutant alleles of the K-ras oncogene and the p53 tumour suppressor gene. Dr. Jacks described recent work on the development of a new allele of oncogenic allele of K-ras, which can be activated by the Flp recombinase. This allele will be useful in separating the different mutation temporally, by controlling K-ras mutation with Flp and other mutations using the Cre-LoxP system. The Jacks lab has observed that single copy expression of mutant EGFR is not sufficient to induce lung tumours in the mouse, suggesting that levels of expression of the mutant allele are critical to initiate tumour development. K-ras-induced lung tumours are responsive to cisplatin treatment, exhibiting cell cycle withdrawal, apoptosis and overall tumour shrinkage. However, treated mice do not survive longer than untreated controls and develop resistant tumours, which have a distinct gene expression profiles and exhibit frequent genomic changes.

Andi McClatchey’s overarching goal is to understand how cell membrane organization is utilized during morphogenesis and deregulated in tumourigenesis. This stems from a longstanding interest in delineating the molecular pathogenesis of familial neurofibromatosis type 2 (NF2). The NF2-encoded tumour suppressor, Merlin, is related to the ERM proteins (Ezrin, Radixin and Moesin), which provide regulated linkage between membrane proteins and the cortical cytoskeleton. Her studies of primary Nf2-/- cells revealed that Merlin normally mediates contact-dependent inhibition of proliferation by coordinating the stabilization of cell:cell contacts with silencing of Epidermal Growth Factor Receptor (EGFR) signaling, linking Merlin to a well-known oncogene. Mechanistic studies revealed that upon cell:cell contact Merlin associates with EGFR via the PDZ-domain-containing adapter NHE-RF1, altering its membrane distribution. This complemented her discovery thatEzrin is essential for apical localization of NHE-RF1 in the mouse intestinal epithelium. Thus Merlin/ERMs may organize distinct membrane domains by spatiotemporally controlling NHE-RF-associated receptors. This work highlights the heterogeneous, context-dependent nature of membrane receptor distribution/signaling.

Marco Giovannini reported that because of the increasing imbalance between the number of new agents and NF2 patients, it is essential to develop predictive preclinical models of NF2 tumours. Peculiarities in Schwann cell/NF2 biology and the lack of tumourigenic NF2 schwannoma/meningioma cell lines have hindered the development of a panel of cell lines. By the recent development of tumourigenic Schwann cell lines from genetically modified mouse (GEM) tumours and tissues, his lab has overcome a primary obstacle in establishing pertinent NF2 in vitro models. The ability of these cell lines to grow as xenografts has enabled studies of the growth and treatment response of engrafted tumours to specific therapeutic agents. Characteristics of GEM mouse models suitable for inclusion in a preclinical testing program include: (a) high penetrance of tumour development; (b) early initiation and time to progression; and (c) ability to be bred in large numbers. NF2 mouse models models are not optimal for testing against all agents because they are resource-intensive when used for testing. Recently however, he found that the latency and the penetrance of Schwann cell tumours in a transgenic mouse model of NF2in the FVB/N inbred mouse background could be significantly altered by the introduction of different genetic backgrounds.

Rick Fehon studies the functions of Merlin, the Drosophila homologue of the Neurofibromatosis 2 tumour suppressor gene, in controlling epithelial proliferation. He has shown previously that Merlin functions together with another tumour suppressor, Expanded, to regulate both proliferation and differentiation in developing imaginal epithelia. His recent studies show that in the absence of Merlin and Expanded transmembrane receptors that control proliferation and cell fate specification, such as Notch, the EGF receptor, Patched and Smoothened all accumulate to abnormally high levels at the plasma membrane. As a result output of the corresponding signaling pathways also can be upregulated. He proposes that Merlin and Expanded control proliferation by regulating the abundance, localization and turnover of cell surface receptors, and that misregulation of these processes may be a key component of tumourigenesis.

Dominique Lallemand has shown that Nf2 inactivation in Schawn cells results in the accumulation of several proteins at the plasma membrane due to an increase of protein transport from the Golgi apparatus to the plasma membrane. Importantly, he observed increased levels of the ErbB familly of growth factor receptors, known to be essential for Schwann cell proliferation. This translates into the activation of downstreampromitogenic signaling pathways, such as AKT and MAPK, resulting in higher proliferation and immortalization of Nf2-deficient Schwann cells. He concludes that accumulation of ErbB receptors and stimulation of the AKT signaling pathway constitutes the major initiating event in schwannoma development in humans. He also showed that Ezrin, a member of the ERM family of protein is capable of inhibiting Schwann cell proliferation. By creating chimeric proteins between merlin and Ezrin, he identified different domains of the two proteins that are essential for their growth inhibition activity. Finally, he found that the inactivation of merlin doesn’t result in a growth advantage in cells in wich the expression of the ERMs has been suppressed suggesting that merlin requires the ERMs to inhibit proliferation.

Luis F. Parada reported on NF1, a genetically heritable disease that involves the transmission or de novo appearance of a null mutation in one of the two NF1 alleles. Over the lifetime of the individual, usually in development or in childhood, loss of heterozygosity results in the manifestation of tumours in the peripheral nervous system called neurofibromas. Neurofibromas are complex and unique tumours in that the contain many abnormal appearing cell types and in that spontaneous non-NF1 neurofibromas are exceedingly rare. In recognition that all cells in an NF1 individual are heterozygous while the tumour causing cells are homozygous null, they have developed mouse models to investigate these relationships. Genetic studies demonstrated that: 1) the Schwann cell is the tumour cell of origin that undergoes LOH; 2) Heterozygosity is required for tumour formation. Mice with wild-type cells but nullizygous Schwann cells do not develop tumours. In collaboration with D.W. Clapp and colleagues, adoptive transfer experiments have been performed that indicate a critical role for heterozygous mast cells in the tumourigenic intitiation. Such insights, only obtainable using mouse models, provide a platform for clinical trials aimed at blocking mast cells activity in NF1 patients.

Daniel Louvard discussed that fact that cancer cells become metastatic by acquiring a motile and invasive phenotype. This step requires remodeling of the actin cytoskeleton and the expression of exploratory, sensory organelles known as filopodia. Aberrant ß-catenin-TCF target gene activation plays a major role in colorectal cancer development. They identified fascin1, a key component of filopodia, as a target of ß-catenin-TCF signaling in colorectal cancer cells. Fascin1 mRNA and protein expression were increased in primary cancers in a stage-dependent manner. Fascin1 was exclusively localized at the invasive front of tumours also displaying nuclear ß-catenin. Forced expression of fascin1 in colorectal cancer cells increased their migration and invasion in cell cultures and caused cell dissemination and metastasis in vivo, whereas suppression of fascin1 expression by small interfering RNA reduces cell invasion. He proposed that transient up-regulation of fascin1 in colorectal cancer promotes the acquisition of migratory and invasive phenotypes that lead to metastasis. Genes involved in cell migration and invasion, such as fascin1, could serve as novel targets for metastasis prevention.

Pier Paolo Pandolfi discussed approaches towards the development of a “pro-senescence” therapy for cancer chemoprevention and treatment, which aims at utilizing for therapy a built-in failsafe mechanism in mammalian cells triggered in response to an aberrant oncogenic stimulus. The irreversible cell growth arrest, termed “cellular senescence”, is in fact emerging as a powerful intrinsic tumour suppressive mechanism, which is critical to restrict initiation and progression of early cancerous lesions in humans. Pandolfi and colleagues have recently reported in murine models for prostate cancer that complete inactivation of the tumour suppressor Pten elicits such a senescence response, which significantly opposes tumorigenesis in vivo through activation of a p53 activating pathway. They went on characterizing this response from a molecular standpoint and presented data whereby this failsafe mechanism is profoundly different from the classic oncogene induced senescence response. These findings have profound implications regarding the ability to target for therapy the cancer-initiating cell and, in particular, the quiescent cancer-initiating cell. Dr. Pandolfi presented at the meeting novel ways by which these concepts can be tested in preclinical trails in animal models and translated in formal clinical trials in human beings for the treatments and prevention of human cancer.

Terence Rabbitts developed macromolecular drugs, such as antibody fragments, in an attempt to disrupt critical protein-protein interactions. An example is using a single VH domain that binds to the switch region of RAS, thus competing PI3K and other RAS-affectors from their normal binding site on RAS. This completely inhibits cancer formation in cancer models. Specific chromosomal translocations typify human leukaemias and lymphomas and represent sites of oncogene activation and oncogene creation by gene fusion. We have been working on new methods to produce chromosomal translocation mimics to emulate the primary events found in human leukaemias. One conclusion is that MLL-fusion proteins have the ability to cause lineage reassignment in lymphoid cells, demonstrating the plasticity of this lineage and the instructive role of MLL-fusions in leukaemia.

Jacques Ghysdael presented a mouse model for the T-ALL associated with the t(9;12)(p24; p13) chromosomal translocation, leading to the expression of a TEL-JAK2 fusion protein. STAT5 activation is critical to both the emergence and maintenance of TEL-JAK2-induced leukemia. Genetic studies using mutant mice arrested at specific stages of T cell development show that (i) early T cell progenitors (pro-T and pre-T cells) are efficient targets for TEL-JAK2-induced transformation; (ii) rapid leukemia progression requires signals emanating from the pre-TCR. Finally, signaling cues that depend upon the in vivo tumour cell environment are shown to induce the activation of the calcineurin/NFAT signaling pathway in several mouse models of T- and B-cell leukemia. Treatment of TEL-JAK2 or ICN1 (an activated allele of NOTCH1) leukemic mice with the calcineurin inhibitors cyclosporine A or FK506, results in leukemia regression associated with tumour cells proliferation arrest and apoptosis, demonstrating that persistent calcineurin activation has a pro-oncogenic role in the maintenance of the leukemic phenotype in vivo.

Fatima Mechta-Grigoriou discussed the “free radical theory of aging” correlating ROS production to aged-related pathologies and premature death. In agreement with its protective role against oxidative stress, junD-deficient mice exhibit features of premature aging, develop cancers and die before the control littermates. Interestingly, junD-deficient mice also suffer from persistent hypoglycemia due to a chronic hyperinsulinemia. Since the Insulin/IGF-1 pathway modulates lifespan in various species, she hypothesized that constitutive hyperinsulinemia in junD-/- mice could result, at least in part in decreased longevity. A constitutive oxidative stress in the pancreas increases b-islet vascularisation induces insulin secretion and constitutive activation of insulin signalling. Anti-oxidant treatment completely rescues junD-/- phenotype including hypoglycemia, hyperinsulinemia and aging. Thus these studies establish the pro-angiogenic effect of ROS in systemic regulation of insulin that modulates lifespan.

Ali Bazarbachi studies adult T-cell leukemia/lymphoma (ATL), an aggressive malignancy of mature activated T-cell caused by human T cell lymphotropic virus type I (HTLV-I). The viral transactivator protein Tax is responsible for many aspects of the malignant transformation such as proliferation, resistance to apoptosis, genetic instability, tumour dissemination, immnune escape and chemotherapy resistance. HTLV-I Tax transgenic mice, accurately reproduce human ATL disease. We tested the following drugs using the in vivo ATL SCID model: zidovudine, interferon alpha (IFN), arsenic trioxide, the proteasome inhibitor bortezomib, and the corresponding combinations. Inhibition of NF-KB using bortezomib or arsenic alone almost doubled the mice survival but failed to eradicate ATL. The most relevant effect on the mice survival was obtained when we used the combination of arsenic trioxide and interferon alpha, which cured 60% of the mice. This combination is now tested in newly diagnosed ATL patients.

Hugues de Thé reported on APL, a rare leukemia caused by the expression of the PML/RARA oncogene. The later is a powerful transcriptional repressor that blocks myeloid differentiation and enhances the self-renewal of leukemia stem cells. APL is a monogenic disease and de Thé showed that agents that target PML/RARA, such as retinoic acid, arsenic or cAMP all promote leukemia regression. He showed that these agents differentially affect the bulk of leukemia cells and leukemia initiating cells (LIC). APL eradication results from the loss of LIC, which mirrors PML/RARA catabolism. The later is best achieved with the association of retinoic acid, arsenic and cAMP, which should soon yield clinical trials aimed at APL cure without any DNA-damaging chemotherapy. Induction of oncogene degradation could be a general therapeutic strategy in translocation-associated malignancies, that express a gain of function oncogene conferring stem-cell like properties to the malignant cells.