Main

Human cancers progress in fits and starts, when a few cells acquire spontaneous mutations that change gene expression patterns and sometimes allow those cells to escape tumor-suppression pathways. Any given protein may have several roles in these processes, but what role that protein has when is difficult to uncover. As Memorial Sloan-Kettering Cancer Center researcher Yi-Chieh Nancy Du describes, “many somatic gene mutations and the altered expression of many genes have been discovered in cancers, but it is not easy to distinguish genes important for tumorigenesis from 'passenger' events.” To answer those questions, researchers need to target established tumor cells at different stages. Now, Du and her colleagues in Harold Varmus's laboratory describe a system that makes tumor cells susceptible to infection by an avian retrovirus, allowing researchers to manipulate gene expression in the tumor at defined time points.

There are several ways to induce transgene expression, including by Cre-lox–based recombination and under tetracycline- or estrogen-dependent regulation. These techniques, however, require the genomic insertion and genetic transmission of the construct, and each new transgenic line is costly in terms of money and time. Additionally, these methods affect most cells in the induced tumor and may not accurately emulate human cancer progression.

When looking for other approaches, Du and colleagues proposed to introduce the genes of interest into the developing tumors. Previous work on the avian leukosis virus had shown that ectopic expression of the TVA receptor makes a mammalian cell susceptible to avian viral infection. If a tumor cell expresses this receptor, infection with the avian retrovirus can introduce new DNA into the cell, allowing researchers to manipulate gene expression in vivo. As a beneficial side-effect of low infection rates, few cells would receive the gene of interest, which more closely mimics the spontaneous mutations that cause the progression of human cancers.

To develop this method, the researchers turned to a well-known tumor model in which SV40 T antigen is expressed specifically in pancreatic β cells. T antigen blocks two tumor suppressor pathways, causing a reproducible tumor progression in pancreatic islets that mimics that of human cancers. In this model, they could co-express the T antigen with the TVA receptor for the avian viral vector, inducing virus-susceptible tumors. Researchers then have a choice of two ways to analyze phenotypes: in vivo infection of tumors for physiological studies or in vitro cell lines from these tumors for molecular and biochemical studies.

Du and colleagues found that, although infection with a control virus had no effect on tumor progression, genes could be transmitted to the tumor cells and dramatically influenced tumor behavior. For instance, expression of a dominant-negative E-cadherin (a cell adhesion molecule) in these tumors caused similar phenotypes as coexpression of T antigen and dominant-negative E-cadherin in these pancreatic islet tumors. Opportunely, the researchers used this system to characterize a previously unknown role of the anti-apoptotic protein Bcl-xL in the cytoskeletal rearrangements that affect metastasis.

This system works particularly well for tumors that have stereotyped progression. As Du explains, “knowing the timing of development of hyperplastic lesions is important for introducing these avian viruses” at the desired stage of tumorigenesis. Once these variables are known, however, “this method has the flexibility to deliver a combination of the avian viruses encoding different genes simultaneously or sequentially to study their interactions in tumorigenesis,” she says. So using viruses to deliver genes may not just save money and time over traditional transgenic techniques, but also make it possible to manipulate tumors in vivo in unprecedented ways.