Master regulators of tumorigenesis and drug sensitivity in prostate cancers.


Cory Abate-Shen, Michael Shen, Andrea Califano


In Progress


We have recently demonstrated that unbiased analysis of a genome-wide molecular-interaction network (interactome), inferred from a large collection of molecular profiles of GBM patients, could elucidate the genetic determinants of the most aggressive subtype of the disease (Carro 2009). Such networks, assembled from naturally occurring phenotypic variability in patients, while highly informative on disease etiology fail to capture the tumor response to specific therapeutic agents. Similarly, while extensive molecular profiles produced by chemical perturbation of cell lines have been assembled (Lamb 2006), these fail to capture compound activity in vivo.

To address these issues the Abate-Shen, Shen, and Califano labs were recently funded by the Mouse Model of Human Cancer Consortium (MMHCC) to assemble and interrogate a prostate cancer interactome by perturbing several genetically engineered mouse models with perturbagens targeting key pathways in the disease. Such a mouse Prostate Cancer interactome (mPCi) will provide valuable information on drug-related pathway activity in vivo and would be ideally complemented by an equivalent in vivo human Prostate Cancer interactome (hPCi). This combination would provide tremendous insight into compounds whose pathways activity is conserved in both organisms, allowing a truly informed use of the mouse as a model to test compounds and biomarkers for translation into humans. Unfortunately, performing these studies in human subjects in not viable because of the complexity of the protocols and because many of the probes that can perturb relevant pathways may be too toxic to be used in human subjects.

This project will couple the expertise in prostate cancer models from the Abate-Shen and Shen labs with the interactome assembly and analysis tools developed in the Califano lab to create, validate, and interrogate the first human interactome obtained from in vivo pharmacological perturbations of prostate cancer xenografts in a murine host. While far from perfect, this effort will produce the first molecular-interaction model of human prostate cancer representing the biology of these tumors in a 3D, in vivo context, under pharmacological stimulation with tumor-specific compounds. The same set of perturbagens used for the mPCi assembly will be used thus allowing (a) direct comparison of conserved and divergent regulatory mechanisms in human and murine tumors and (b) better characterization of the tumor response to agents affecting established molecular targets in prostate cancer, such as the androgen receptor, PI3K, AKT, and MYC, among others. We will use the hPCi both in isolation and in conjunction with the mPCi to elucidate key driving mechanisms and events in prostate cancer progression and to identify mechanism-based druggable targets and associated biomarkers for therapeutic intervention. If successful, this paradigm will allow identification of both convergent and divergent regulatory mechanisms in the two organisms, allowing informed use of the mouse as a model to validate drug-targets and biomarkers for translation into human clinical applications.


We have identified and obtained human xenograft lines for the generation of the perturbation-based hPCi, in collaboration with Dr. Bob Vessella (University of Washington). We have selected 6 xenograft lines that have been generated in the Vessella lab, corresponding to LuCaP 23.1, 73, 77, 78, 81, and 147. These xenografts have been maintained continuously in immunodeficient mice, maintaining the histological and molecular heterogeneity and properties of the parental tumors, and have been evaluated for their castration-resistance, expression of the Nkx3.1 homeobox gene, and expression of androgen receptor. These lines have been successfully transferred from the Vessella lab, and have been maintained by serial subcutaneous passaging in NOG immunodeficient mice in the Shen lab mouse colony.

In a pilot study, we have examined whether the drug treatment regimens that have been previously utilized for the construction of the mouse prostate cancer interactome (funded by the MMHCC) can be used for the perturbation approaches using human prostate cancer xenografts. For this purpose, we have used two different xenograft lines, LuCaP 73 and LuCaP 147, and have treated these after growth to 50 or 70-80 mm3 in size, using vehicle control as well as two small molecule perturbagens, MK2206 and PD0325901. 8 mice were used in total, carrying between 2 to 4 total grafts, for a total of 21 samples. The xenograft-bearing mice were treated for 5 days, followed by examination of the dissected xenografts for their size, weight, and histology, as well as processing for RNA-seq analysis of their transcriptomes. This pilot study has been completed, and indicates that the experimental protocols developed for this project will be successful.

We are now using these 6 LuCaP lines in drug treatment regimens that have been previously utilized for the construction of the MMHCC-funded mouse prostate cancer interactome. Specifically, we are treating NOG immunodeficient mice carrying these grafts with 13 drugs as perturbagens (Testosterone, Calcitriol, MDV3100, MK2206, LY294002, Rapamycin, Imatinib, Sorafenib, PD325901, Dasatinib, BAY 11-7082, WP1066, Docetaxel), as well as a vehicle control . These drugs correspond to compounds that target major signaling pathways known or strongly suspected to be altered in prostate cancer, or represent standard-of-care treatments for human patients. In addition, the treated mice will either be hormonally intact, or will have been castrated to induce androgen-deprivation, the primary treatment for human prostate cancer. In combination, this will result in 168 total expression profiles to be generated: 6 xenograft lines x 14 conditions (13 perturbagens + 1 vehicle control) x 2 (intact and castrated).


Carro MS, Lim WK, Alvarez MJ, Bollo RJ, Zhao X, Snyder EY, Sulman EP, Anne SL, Doetsch F, Colman H, Lasorella A, Aldape K, Califano A, Iavarone A. The transcriptional network for mesenchymal transformation of brain tumours. Nature. 2009.

Lamb J, Crawford ED, Peck D, Modell JW, Blat IC, Wrobel MJ, et al. The Connectivity Map: using gene-expression signatures to connect small molecules, genes, and disease. Science. 2006;313(5795):1929-35.