A quantitative systems pharmacology analysis of KRAS G12C covalent inhibitors

Abstract

The KRAS oncogene is the most common, activating, oncogenic mutation in human cancer. KRAS has proven difficult to target effectively. Two different strategies have recently been described for covalently targeting the most common activating KRAS mutant in lung cancer, KRAS G12C. Previously, we have developed a computational model of the processes that regulate Ras activation and this model has proven useful for understanding the complex behaviors of Ras signaling. Here, we use this model to perform a computational systems pharmacology analysis of KRAS G12C targeted covalent inhibitors. After updating our model to include Ras protein turnover, we verified the validity of our model for problems in this domain by comparing model behaviors with experimental behaviors. The model naturally reproduces previous experimental data, including several experimental observations that were interpreted as being contrary to conventional wisdom. Overall, this suggests that our model describes the Ras system well, including those areas where conventional wisdom struggles. We then used the model to investigate possible strategies to improve the ability of KRAS G12C inhibitors to inhibit Ras pathway signaling. We identify one, as of yet unexplored mechanism, that, if optimized, could improve the effectiveness of one class of KRAS inhibitor. We also simulated resistance to targeted therapies and found that resistance promoting mutations may reverse which class of KRAS G12C inhibitor inhibits the system better, suggesting that there may be value to pursuing both types of KRAS G12C inhibitors. Overall, this work demonstrates that systems biology approaches can provide insights that inform the drug development process.

Inhibition of both mutant and wild-type RAS-GTP in KRAS G12C colorectal cancer through cotreatment with G12C and EGFR inhibitors

“…For mutant RAS-GTP, the predicted dose response for KRAS G12C inhibition showed increasing levels of synergy for increasing levels of coincident EGFR inhibition (Figure 7B). This is consistent with previous empirical results (12,13,23,24) and with our original KRAS G12C modeling (21). Additionally, the model suggests that there will also be considerable levels of synergy for WT RAS-GTP ( Figure 7C).…”

“…Part of our study involved the use of a mathematical model. Mathematical models have played a role in multiple studies of KRAS G12C inhibitors (21,23,38,39). Here, we use the model to evaluate the contribution of two biochemical defects of KRAS G12C that were here observed, impaired binding to NF1 and to CRAF, on the observed sensitivity of KRAS G12C cancer cells to EGFR inhibition.…”

“…Previous uses of mathematical models in G12C inhibitors include evaluating whether observed behaviors are or are not consistent with in vitro data, leading to the realization that there must be a contribution of cellular GEFs to mutant signaling (23). Other uses include evaluating how properties of inhibitors and sensitivity to GEF loading are likely to impact loading kinetics (21,38,39). In all of these models, the ability of mathematical models to relate biochemical processes to cellular observables has been particularly valuable, as inference without a mathematical model can sometimes be challenging for biological networks.…”

“…Whether this sensitivity is explained by the partial reduction in impaired NF1 binding relative to KRAS G13D, and/or by reduced binding to RAS effectors, and whether zygosity modulates, are features that we can investigate with our mathematical model. Previously, we performed a computational analysis of KRAS G12C and its response to G12C covalent inhibitors (21). That study utilized published biochemical rate constants for KRAS G12C (22); however, the data available at that time did not characterize the interaction between KRAS G12C and NF1, and we thus originally modeled KRAS G12C to bind to NF1 equivalently to KRAS WT and KRAS G12V.…”

“…We applied our computational model to the problem to determine which behavior(s) would be logically consistent with the available biochemical and biophysical data. To simulate the effects of covalent KRAS G12C inhibitors, we utilize a form of our computational model that includes RAS production and degradation (21) as these processes are essential for understanding the effects of covalent inhibition. We utilized the updated parameters for KRAS G12C demonstrated in Figure 4, where we now assume an order of magnitude reduction in binding to NF1 to approximate the observed impairment in binding to NF1.…”