A preclinical platform for assessing anti-tumor effects and systemic toxicities of cancer drug targets

Anti-cancer drug development campaigns often fail due to an incomplete understanding of the therapeutic index differentiating the efficacy of the agent against the cancer and its on-target toxicities to the host. To address this issue, we established a versatile preclinical platform in which genetically defined cancers are produced using somatic tissue engineering in transgenic mice harboring a Doxycycline-inducible short hairpin RNA against the target of interest. In this system, target inhibition is achieved by addition of doxycycline, enabling simultaneous assessment of efficacy and toxicity in the same animal. As proof-of-concept, we focused on CDK9 — a cancer target whose clinical development has been hampered by compounds with poorly understood target specificity and unacceptable toxicities. We systematically compared phenotypes produced by genetic Cdk9 inhibition to those achieved using a recently developed highly specific small molecule CDK9 inhibitor and found that both perturbations led to robust anti-tumor responses. Remarkably, non-toxic levels of CDK9 inhibition could achieve significant treatment efficacy, and dose-dependent toxicities produced by prolonged CDK9 suppression were largely reversible upon Cdk9 restoration or drug withdrawal. Overall, these results validate a versatile in vivo target validation platform that can be employed for rapid triaging of therapeutic targets and lend support to efforts aimed at advancing CDK9 inhibitors for cancer therapy. For RNA-seq analysis of the transcriptional profiles of Myc;p53-/- liver tumors expressing shCdk9 or shRenilla (control) induced by doxycycline, as well as normal liver tissues with doxycycline-induced shRenilla expression, total RNA was extracted from bulk tissue using the RNeasy Mini Kit (Qiagen). Purified polyA mRNA was subsequently fragmented, and first and second strand cDNA synthesis performed using standard Illumina mRNA TruSeq library preparation protocols. Double stranded cDNA was subsequently processed for TruSeq dual-index Illumina library generation.