1. ¹Department of Genetics, University of North Carolina, Chapel Hill, NC, USA
2. ²The Carolina Genome Sciences Center, University of North Carolina, Chapel Hill, NC, USA
3. ³Department of Pathology and Laboratory Medicine, University of North Carolina, Chapel Hill, NC, USA
4. ⁴Gene Expression and Disease Unit, Department of Developmental Biology, Pasteur Institute, Paris, France

Correspondence: Dr SJ Bultman, Department of Genetics, University of North Carolina, 103 Mason Farm Road, 4340B MBRB, Chapel Hill, NC 27599-7264, USA. E-mail: Scott_Bultman@med.unc.edu

⁵Current address: Novartis Institutes for Biomedical Research, Cambridge, MA 02139, USA.

Received 1 May 2007; Revised 7 June 2007; Accepted 11 June 2007; Published online 16 July 2007.

Abstract

Mammalian SWI/SNF-related complexes have been implicated in cancer based on some of the subunits physically interacting with retinoblastoma (RB) and other proteins involved in carcinogenesis. Additionally, several subunits are mutated or not expressed in tumor-derived cell lines. Strong evidence for a role in tumorigenesis in vivo, however, has been limited to SNF5 mutations that result primarily in malignant rhabdoid tumors (MRTs) in humans and MRTs as well as other sarcomas in mice. We previously generated a null mutation of the Brg1 catalytic subunit in the mouse and reported that homozygotes die during embryogenesis. Here, we demonstrate that Brg1 heterozygotes are susceptible to mammary tumors that are fundamentally different than Snf5 tumors. First, mammary tumors are carcinomas not sarcomas. Second, Brg1^+/- tumors arise because of haploinsufficiency rather than loss of heterozygosity. Third, Brg1^+/- tumors exhibit genomic instability but not polyploidy based on array comparative genomic hybridization results. We monitored Brg1^+/-, Brm^-/- double-mutant mice but did not observe any tumors resembling those from Snf5 mutants, indicating that the Brg1^+/- and Snf5^+/- tumor phenotypes do not differ simply because Brg1 has a closely related paralog whereas Snf5 does not. These findings demonstrate that BRG1 and SNF5 are not functionally equivalent but protect against cancer in different ways. We also demonstrate that Brg1^+/- mammary tumors have relatively heterogeneous gene expression profiles with similarities and differences compared to other mouse models of breast cancer. The Brg1^+/- expression profiles are not particularly similar to mammary tumors from Wap-T121 transgenic line where RB is perturbed. We were also unable to detect a genetic interaction between the Brg1^+/- and Rb^+/- tumor phenotypes. These latter findings do not support a BRG1–RB interaction in vivo.