Abstract PR08: New tools for mapping genetic modifiers of cancer risk in the tumor microenvironment

Abstract

Abstract The majority of heritable breast cancer risk is unknown. One potential source of “missing heritability” is genetic modifiers in the tumor microenvironment (TME). Although genetic modifiers in the TME have long been suspected, they have rarely been studied and are largely unknown. Here, we used two new techniques: the Consomic Xenograft Model (CXM) and species-specific RNA-seq (SSRS) to map genetic modifiers in the TME. In CXM, human breast cancer xenografts are implanted in immunodeficient consomic rat strains and tracked for tumor progression. Because the rat strains vary by one chromosome (i.e., consomic), whereas the malignant tumor cells do not differ, any observed changes in tumor phenotypes are due to genetic modifiers in the TME and can be localized to one chromosome. The SSRS method uses probabilistic mapping of RNAseq reads to a joint human and rat transcriptome to assess differential expression (DE) in malignant (human) tumor cells and the nonmalignant (rat) TME. Validation of SSRS revealed >99.4% specificity in calling human or rat reads, which was significantly better than conventional RNA-seq. Using CXM, we found that BN-derived genetic variant(s) on rat chromosome 3 significantly reduced growth of MDA-MB-231-Luc (231Luc+) tumors by 49% (P<0.05) in the SS.BN3IL2Rγ CXM strain compared with parental SSIL2Rγ. This coincided with a 3.1-fold (P<0.001) decrease in blood vascular invasion by 231Luc+ tumor cells and 7.3-fold (P<0.05) lower metastatic burden in the lungs in SS.BN3IL2Rγ compared with SSIL2Rγ, despite a paradoxical 27% (P<0.05) increase in blood vessel density (BVD) in SS.BN3IL2Rγ rats. The tumor-associated blood vessels in SS.BN3IL2Rγ rats appeared collapsed and dysfunctional, possibly explaining the decreased tumor growth and metastasis, despite increased BVD. Lymphatic vasculature and lymphogenous metastasis were completely unaffected by the SS.BN3IL2Rγ background, suggesting that the causative variant(s) on BN rat chromosome 3 are vascular cell-type specific. We used SSRS to begin identifying the TME-specific mediators on rat chromosome 3 (RNO3) that inhibit growth and hematogenous metastasis of human 231Luc+ breast cancer xenografts implanted in the SS.BN3IL2Rγ. Compared with SSIL2Rγ tumors, we identified a network of 539 DE transcripts in the TME of SS.BN3IL2Rγ rats, of which 28% (150 genes) reside on RNO3, which was significantly higher (>4-fold; P<0.001) than any other rat chromosome. Moreover, a two-sample Kolmogorov-Smirnov test revealed that the difference in distributions of adjusted p-values for RN03 versus the rest of the genome was highly significantly higher for DE genes (P=3.152e-08) or DE transcripts (P=3.441e-16). Compared with other rat chromosomes, RNO3 also had by far the highest incidence of alternative isoform usage (91% of all instances). Pathway analysis of DE genes using DAVID revealed that the two most significant GO clusters were extracellular matrix (49 genes; P<10-20) and blood vessel development (43 genes; P<10-17), which recapitulated the vascular defects observed in the SS.BN3IL2Rγ tumors. Collectively, our data demonstrate that CXM and SSRS can be used to detect genetic modifiers in the TME. Citation Format: Michael J. Flister, Alexander Stoddard, Shirng-Wern Tsaih, Angela Lemke, Jozef Lazar, Howard Jacob. New tools for mapping genetic modifiers of cancer risk in the tumor microenvironment. [abstract]. In: Proceedings of the AACR Special Conference on Computational and Systems Biology of Cancer; Feb 8-11 2015; San Francisco, CA. Philadelphia (PA): AACR; Cancer Res 2015;75(22 Suppl 2):Abstract nr PR08.