Canine models for characterization, diagnosis, and treatment of human cancers using comparative canine-human transcriptomics

National Center For Advancing Translational Sciences

Investigators

Linked publications, trials & patents

Abstract

Cancer is a multifaceted disease. Its study requires interrogation of biology at multiple levels and bioinformatic analysis to analyze, integrate and interpret complex data streams produced by experiments. In recent years, researchers have invariably moved toward multi-omics studies of cancer. Our work here follows this trend, its impetus being to determine if transcriptomic biomarkers, discovered by analyses of RNA-Seq data from canine cancers, appear as expressed proteins in the tissue and blood of canines and humans. Previously, we analyzed canine transcriptomic data using gene co-expression clustering and discovered 372 genes relevant to lung adenocarcinoma, melanoma, osteosarcoma, B- and T-cell lymphoma in canines and humans. We ran proteomics experiments on canine tissue and blood and found that 262 of the 372 genes coded for expressed tissue proteins, and that 45 of these 262 tissue proteins were also expressed in blood. This work was recently published in PLOS Computational Biology. Recently, the 262 tissue-expressed proteins were grouped according to the cancer for which they were most highly expressed. For each of the proteins in the various groups we selected approved drugs known to alter the expression of said proteins in opposition to that expression level exhibited by each cancer. These drugs are potential therapeutics because they turn off genes that are turned on in the context of cancer or vice versa. We found 60 drugs altogether that had this effect on the proteins associated with each cancer. The 60 drugs were tested for effectiveness in killing canine cell lines associated with melanoma, lymphoma, osteosarcoma, and leukemia. 40 of the 60 drugs were effective in killing at least one cancer cell line. From the 40 active drugs, 30 two-drug combinations were assembled. Drug combinations were chosen such that they each down-regulate a different protein associated with each cancer. Combinations were chosen in this way because drugs that hit different targets are more likely to exhibit synergistic efficacy. Synergistic drugs can be administered together at concentrations far smaller than they would need to be if the drugs were used individually, while at the same time retaining efficacy in killing cancer cells. This significantly reduces chemotherapeutic side-effects and minimizes the chance for the cancer to develop drug resistance. The 30 two-drug combinations were tested in matrix screens against four canine cancer cell lines (120 matrix screens total). Out of the 120 matrix screens we identified 20 drug combinations that exhibited synergy in at least one of the canine cancer cell lines. The 20 active drug combinations will be tested in PDX mice. Those found to be active in the mice will be confirmed in clinical trials on patient dogs. Large-scale proteomics studies in canine will be performed to gather statistics confirming biomarker specificity among cancer types. Biomarkers and drug combinations found to be effective in these canine studies will inform the design of future clinical studies in humans.