Abstract 2238: Modeling local network connectivity to identify druggable enzyme targets.
Karin Jensen, Christian B. Moyer, Kevin A. Janes · Cancer Research · 2013
Abstract While targeted therapies have provided extended lifespans to some cancer patients, the majority of cancers do not have these “golden bullet” therapies despite their increasing need. To address this, large-scale drug screening efforts are conducted to identify novel therapeutic targets in various cancers. While these screening efforts identify important nodes in the cancer cell signaling network, they do not necessarily reflect the “druggability” of those nodes. We predict that the local molecular connections surrounding a drug target will be predictive of inhibitor efficacy and should be considered in drug screening efforts and prioritizing drugs for clinical development. Towards this end, we have used computational modeling to survey the behavior of a three-tiered enzyme cascade configured with all possible one- and two-feedback interactions. To characterize the effect of local network topology on pathway inhibition, we compared inhibition by RNAi vs. small-molecules in the computational models. Our modeling predicts that certain topologies will be more sensitive to RNAi than small-molecules, and that other topologies will show the opposite behavior. We next experimentally tested one network wiring that our modeling predicts will be more potently inhibited by a small-molecule inhibitor than RNAi inhibition, where the third enzyme in the cascade negatively feeds back on the first. This topology is found in the Raf-MEK1/2-ERK1/2 pathway where ERK1/2 can hyperphosphorylate and inactivate Raf, creating a negative feedback loop. To test our model hypothesis, we compared the potency of shMEK constructs relative to equivalent inhibition by a small-molecule inhibitor of MEK1/2, U0126. We analyzed pathway output (pERK1/2) over time in response to pathway activation. Our data suggests that pERK1/2 responses are both qualitatively and quantitatively different for cells inhibited by shMEK compared to U0126. In U0126-treated cells, pERK1/2 activation after stimulation is delayed and transient compared to shMEK1/2 cells, which show reduced pERK1/2 compared to controls but no differences in signaling kinetics. We further investigated the phenotypic response from this difference in pERK1/2 signaling and found that upon stimulation the shMEK-inhibited cells proliferated whereas the U0126-inhibited cells did not. These results support our model prediction that small-molecule inhibition of MEK1/2 will more potently inhibit pathway output compared to RNAi. While global loss-of-function screens can identify new targets for therapy, it is critical to determine whether drugging such targets will yield the same outcome as eliminating them. Our work demonstrates the need to consider local network topology in drug screening and target identification for drug development. Prioritizing targets for drug development in this way will reduce the cost of drug development and allow more effective therapies to reach patients in the clinic. Citation Format: Karin J. Jensen, Christian B. Moyer, Kevin A. Janes. Modeling local network connectivity to identify druggable enzyme targets. [abstract]. In: Proceedings of the 104th Annual Meeting of the American Association for Cancer Research; 2013 Apr 6-10; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2013;73(8 Suppl):Abstract nr 2238. doi:10.1158/1538-7445.AM2013-2238