Inducible mouse models

Our laboratory has made a significant investment in developing new mouse models for a variety of tumors, using Cre/Lox mediated switching of tumor suppressor genes and oncogenes. Both transgenesis and somatic gene transfer are employed to express (inducible) Cre recombinase in a controlled fashion. The methodology enables us to switch multiple oncogenes and/or tumor suppressor genes within cells in vivo at a defined time. This permits the induction of highly specific tumors within a narrow time window, and allows us to correlate specific genetic defects with phenotypic characteristics. We have induced a range of tumors, including melanomas, lung tumors, squamous cell carcinomas, and mesotheliomas. The general picture that perspires from these studies is that the mouse models show closer resemblance to the human tumors when an increasing number of mutations are shared. We are also incorporating sensitive in vivo imaging techniques to follow tumor growth and metastatic spread in real time in animals.

  Retroviral insertional mutagenesis

Besides focusing on the function of specific oncogenes and tumor suppressor genes known to be involved in the tumors mentioned above, we utilize mouse models to identify new oncogenes and tumor suppressor genes. Slow transforming retroviruses, such as the Moloney murine leukemia virus (M-MuLV), induce tumors upon infection of a host after a relatively long latency period. These retroviruses can transform the infected host cells through the accidental insertion of their proviruses into the host genome in the vicinity of genes that can confer growth advantages to cells. This means that the proviral insertions found in tumors induced by retroviral insertional mutagenesis mark genes contributing to the tumorigenic process. Since cancer is a complex multistep process, the proviral insertions in one clone of tumor cells also represent oncogenic events that co-operate in tumorigenesis. Novel advances, such as the launch of the complete mouse genome, high-throughput isolation of proviral flanking sequences, and genetically modified animals have revolutionized proviral tagging into an elegant and efficient approach to identify signaling pathways that collaborate in cancer. This project is executed together with Maarten van Lohuizen in our department. We expect to identify another 5000 insertion sites in the next couple of years. This should bring us close to the identification of all the relevant components of the oncogenic signaling pathways we focus on.

Hunting down gene function

By focussing on some of the genes that were found in these retroviral tagging screens, such as the Pim and Frat families, in more detail, we are trying to understand how proto-oncogenes confer their oncogenic potential. Therefore we try to understand the normal function of these genes on the one hand by studying the phenotypic abnormalities in mice lacking one or more of these proto-oncogenes, and by characterizing the signaling pathways in which these genes act. The latter experiments are performed in cell culture systems using biochemical methods.
 

Mouse models for cancer

The mouse is used as a model organism for establishing the role of oncogenes and tumor suppressor genes in tumor development. Using Cre/Lox mediated switching taking advantage of somatic gene transfer methods, expression of multiple oncogenes and tumor suppressor genes can be regulated in a tissue-specific and spatial-temporal fashion. This permits a more accurate modeling of tumorigenesis as it occurs in man, and therefore provides the opportunity for establishing relevant genotypephenotype correlations. These models also provide an excellent experimental tool to test prevention and intervention strategies especially when combined with sensitive in vivo imaging techniques. Finally, these models permit us to identify new oncogenes and tumor suppressor genes involved in tumor progression using a variety of techniques, such as array CGH, expression profiling and proviral insertional mutagenesis.

Functional analysis of oncogenes and tumor suppressor genes
 
We utilize mice carrying combinations of different oncogene and conditional tumor suppressor gene alleles to model a range of tumors. Our current focus is on several lung cancers subtypes and mesotheliomas. To achieve (sporadic) activation of oncogenes and inactivation of tumor suppressor genes we use Adeno-Cre or Lentivirus-mediated gene transfer to switch the conditional oncogenes and tumor suppressor gene alleles. Subsequently, tumor initiation and progression is monitored in longitudinal studies in which noninvasive imaging techniques are used.

Lung tumors
 
We focus on small cell lung cancer (SCLC), non-small cell lung cancer (NSCLC) and squamous cell carcinoma (SCC). When Rb and p53 are inactivated
specifically in lung, SCLC ensues. The marker profile of these tumors is closely resembling that of human SCLC. Even similar genomic aberrations are found such as the amplification of the L-Myc gene. These tumors are often heterogeneous consisting of different cell types that either grew as spheres in suspension or attached to substrate. Cloning of the suspended and attached cells from individual tumors showed a consistently different marker profile between these two cell types. Cells growing in suspension carries neuroendocrine markers whereas the adherent nonneuroendocrine cells showed more progenitor-like features. Interestingly, these phenotypically very diverse cell lines shared some highly distinct genetic aberrations indicating that they were derived from a common progenitor. K-ras introduction into neuroendocrine cells changed the phenotype into non-neuroendocrine characteristics. We wondered why these different cell types persisted in the tumor. To search for possible paracrine effects between these cells, in vitro co-culture and in vivo cografting experiments were conducted. In serum-free medium co-culture strongly promoted the proliferation of both cell types. In subcutaneous grafts such growth stimulation was not found. Grafts of each of the cell types gave rise to a local tumor, however, grafts of the mixture resulted in a high incidence of metastatic spread of the neuroendocrine cells indicating that the non-neuroendocrine cells in the graft empowered the neuroendocrine cells to metastasize. We are currently investigating which signaling events are responsible for this effect. To gain insight into the cell of origin of SCLC and NSCLC we have designed a series of cell-type specific Adeno-Cre viruses that enable us to switch oncogenes and tumor suppressor genes in distinct lung cell types in vivo. Using promoters specific for Clara cells, Alveolar type II cells, neuroendocrine cells and basal epithelial cells to drive Cre expression upon Adeno-Cre infection indicated that Cre driven from a neuroendocrine-specific promoter gave rise to SCLC with high efficiency, whereas Cre driven from the alveolar-specific promoter SPC showed a delayed onset of SCLC. Expression from a Clara cell-specific promoter gave mostly hyperplasias in the bronchial epithelial lining and resulted only rarely to SCLC. We are performing these experiments with additional promoters to further substantiate these findings.

Mesotheliomas
 
Earlier we have shown that inactivation of Nf2 and Ink4a/Arf or Nf2/p53/Ink4a by intrathoracic Adeno-Cre injection of compound conditional knockout mice gives rise to mostly sarcomatoid mesotheliomas and that Ink4a plays an important role in the aggressiveness of the tumor. Also germline Ink4b/Ink4a/Arf triple knockout mice show a high incidence of mesotheliomas. Furthermore, sporadic local inactivating these genes in combination with activation of mutant KiRas results in mesotheliomas with a pronounced epitheloid phenotype, closely resembling the corresponding predominant subtype found in human mesotheliomas. We have set up a primary tumor grafting model system that allows us to test various intervention strategies following subcutaneous and orthotopic grafting of primary tumor cells. These tumors appear highly refractory to most intervention protocols and show highly variable responses to drug combinations, indicating that additional alterations that have occurred in these tumors greatly influence the drug response profile of these tumors. Therefore, a more thorough analysis of activated signaling pathways in these tumors will be conducted to gain insight in the molecular basis of their drug response characteristics. Parallel to these experiments we have begun to establish cultures from mesothelioma specimen of human mesotheliomas. While propagation in mice will require more work, these cells grow effectively in vitro. We are planning to generate response profiles from these tumor cell lines to determine whether there are good matches between the human and mouse mesothelioma cell lines.

Ink4 proteins
 
We have produced Ink4-less mice. Mice lacking all the four Ink4a-d genes are viable. We are currently monitoring the spontaneous tumor incidence in
these mice, will determine the effects of additional germline inactivation of p19Arf, and establish mouse embryo fibroblast cell lines to look for altered characteristics of the cells in culture. This will teach us to what extent this tumor suppressor family plays a role in tumor suppression in the various tissues. In addition, this might reveal unexpected physiological functions of these proteins.

Functional mutagenesis

The cyclin-dependent kinase (CDK) inhibitors p15, p16, p21, and p27 are frequently deleted, silenced, or downregulated in many malignancies. We have used mice deficient for one or a combination of two CDK inhibitors to conduct a high-throughput murine leukemia virus (MuLV) insertional mutagenesis screen. We retrieved approximately 9000 retroviral insertions from 476 lymphomas and found hundreds of loci that are mutated significantly more frequently than expected by chance. Many of these are skewed toward a specific genetic context of predisposing germline and somatic mutations. We also found associations between these loci and gender, age of tumor onset, and with lymphocyte lineage (B or T cell). Comparison of retroviral insertion sites with single nucleotide polymorphisms associated with chronic lymphocytic leukemia reveals significant over- lap between these data sets. Together, these data highlight the importance of genetic context within large- scale mutation detection studies and show a novel use for insertional mutagenesis data in prioritization of disease-associated genes resulting from genome-wide association studies.