Activating mutations in the small GTPase N-RAS occur in about 10% of acute myeloid leukemia (AML) cases. Active N-RAS is thought to drive the disease and is therefore a potential target for drug development. There have been numerous unsuccessful efforts to target RAS itself with small molecules, and blocking post-translational modifications of RAS proteins, such as through inhibition of farnesyl transferase, has similarly not proven useful. In addition, efficacy of targeting critical downstream effectors has been limited by the complexity of RAS signaling, such as redundancy of signaling pathways and feedback mechanisms. While targeting RAS is challenging, it was our hypothesis that inhibiting the right combination of downstream pathways in a particular lineage with small molecules could be effective. Initially, we created a Ba/F3 cell line that was completely dependent on oncogenic N-RAS-G12D for growth and survival. Growth was suppressed >99% by shRNA for N-RAS, but could be rescued entirely by interleukin-3 (IL-3), which does not require N-RAS signaling in these cells. Using this cell line, we performed a high-throughput chemical screen with a large library of multi-targeted kinase inhibitors. The lead compound (NRAS1) showed a 70-fold difference in the EC50 for growth inhibition between BaF3-NRAS G12D cells cultured in the absence (0.01μM) or presence (0.77μM) of IL-3. Importantly, this compound showed selectivity towards several leukemia cell lines that were shown to be dependent on mutant N-RAS by shRNA compared to cells expressing wild-type N-RAS (p=0.02). Also, in a xenotransplant model using NRAS-G12D+ OCI-AML3 cells, this compound significantly reduced tumor burden (P=0.005) and prolonged survival (P=0.002) compared to controls. Next, we sought to identify the targets of NRAS1, Interestingly, the compound did not suppress MEK or ERK, which are classical targets of RAS signaling in epithelial cells. NRAS1 profoundly reduced AKT and RPS6 phosphorylation. Kinase selectivity profiling of this compound (1μM) in OCI-AML3 cells (EC50: 0.3μM) identified 13 major binding partners with more than 85% efficacy. The targets consisted mainly of SRC family proteins (SRC, FGR, and LYN etc.) and MAPK family proteins (MAP4K2, 3, 5, and p38 etc.) and others (ZAK and BTK etc), but not MEK and ERK, and AKT was not detected in this assay. In preliminary studies, most of these target kinases were knocked-down by shRNA and, as expected, no single kinase was found to be responsible for mediating growth inhibition. Using a phospho-antibody microarray, the most significantly de-phosphorylated kinases were p38, AKT and SRC, which supports our preliminary findings. To validate the significance of these results, we treated Ba/F3-N-RAS cells with combinations of kinase inhibitors. Combining the AKT inhibitor MK2206 and Dasatinib (SRC family inhibitor) revealed marked synergy, while neither had activity individually. Also, the combination of MK2206 and a cleaner SRC family inhibitor, AZD0530, also synergized, although to lesser extent. In both examples, however, the inhibition of N-RAS transformed cells by NRAS1 proved superior, suggesting that one or more additional targets are required for inhibition of NRAS signaling. To identify additional critical targets of our compound we generated several derivatives with different potency. In particular, one less potent analog of NRAS1 (analog 6, 1% EC50 of original compound) showed a loss of binding activity towards the MAP4K family of proteins, especially MAP4K2. Observed synergy between the selective MAP4K2 inhibitor NG25 and selective inhibitors of MK2206 and Dasatinib in Ba/F3-NRAS G12D cells further points toward MAP4K2 as being of additional significance for oncogenic RAS signaling. Together with the previous data, we propose AKT and MAP4K2 as critical targets of NRAS1. In conclusion, we have identified a novel and selective kinase inhibitor of the N-RAS signaling pathway by chemical screen using Ba/F3-N-RAS G12D cells. By combination of signaling study, kinase selectivity profiling and phosphoproteomics, the main functional targets were found to be AKT, and MAP4K2, and additional functional targets will be elucidated. Our approach also could be applied for other type of oncogenes, and it could help to find therapeutic compound and also help to decipher signaling mechanisms of the oncogenes which are thus far undruggable.
No relevant conflicts of interest to declare.
Asterisk with author names denotes non-ASH members.