mUNC Seq

Next Generation Sequencing and Murine Models

An advance taken by the PRU has been the use of GEMMs that are both driven by the somatic DNA mutations found in corresponding human tumors and are “credentialed” by a comparison of RNA expression to human tumors. For example, in breast cancer, we have used RNA expression analysis to compare human and murine tumors, and have identified faithful murine models of human luminal, basal-like, and Claudin-low subtypes. Thus, through the genomic testing of GEMMs versus human tumor data (much of from The Cancer Genome Atlas (TCGA) of which Dr. Perou is a leader), we can link each GEMM to its closest human counterpart.

This molecular characterization of GEMM tumors is currently performed using massively parallel “Next Generation” sequencing of tumor-derived DNA and RNA. For DNA analysis, two approaches are used prior to NextGen sequencing: either whole exome capture or targeted capture of a custom-set of 198 cancer-related genes. Most models are characterized using the 198 gene approach, whereas exome sequencing is reserved for discovery studies (e.g. to identify mutant genes associated with secondary resistance to therapy). Sequencing of both germline and tumor DNA is performed, but at a reduced ratio (1 normals per 4 tumors) given tumors occur in inbred mice. Sequencing is performed in the UNC High-Throughput Sequencing Facility on an Illumina HiSeq2000. We have developed a bioinformatics pipeline to rapidly analyze murine sequence to identify tumor-associated somatic genetic events, and this approach identifies point mutations, small insertion-deletions, and copy number changes in oncogenes and tumor suppressors genes. This approach can be used to characterize the tumor genomes of non-treated, responding and resistant tumors.

To-date we have sequenced over 150 murine tumors, which show significant intra-model heterogeneity in secondary events. We project an output of ~1000 tumors sequenced annually beginning next year. For RNA analysis, our transcriptional profiling in the MP1U has changed from DNA microarray-based analysis of RNA expression into NextGen cDNA-sequencing (RNASeq). We have considerable experience with this approach as the Perou Lab is the lead site for TCGA RNA expression profiling, where we have profiled >5,000 human tumors via Illumina HiSeq2000 RNASeq. We have also begun the analysis of GEMM tumors using this same protocol and analysis pipeline. RNASeq data can be analyzed using similar bioinformatic approaches as those employed for oligonucleotide array data (e.g. hierarchical clustering, principal components analysis, and supervised analysis) and the Perou lab is well versed in these approaches. RNASeq data can also be compared to DNA sequencing data to efficiently identify and confirm most somatic genetic driver events in tumor nucleic acid. Of note, sample bar-coding allows DNA/RNA from multiple different samples to be sequenced in a single HiSeq2000 lane, thereby reducing sequencing costs 4 to 10-fold.