KRAS is the most frequently mutated oncogene. The incidence of specifi c KRAS alleles varies between cancers from different sites, but it is unclear whether allelic selection results from biological selection for specifi c mutant KRAS proteins. We used a crossdisciplinary approach to compare KRAS G12D , a common mutant form, and KRAS A146T , a mutant that occurs only in selected cancers. Biochemical and structural studies demonstrated that KRAS A146T exhibits a marked extension of switch 1 away from the protein body and nucleotide binding site, which activates KRAS by promoting a high rate of intrinsic and guanine nucleotide exchange factorinduced nucleotide exchange. Using mice genetically engineered to express either allele, we found that KRAS G12D and KRAS A146T exhibit distinct tissue-specifi c effects on homeostasis that mirror mutational frequencies in human cancers. These tissue-specifi c phenotypes result from allele-specifi c signaling properties, demonstrating that context-dependent variations in signaling downstream of different KRAS mutants drive the KRAS mutational pattern seen in cancer. SIGNIFICANCE: Although epidemiologic and clinical studies have suggested allele-specifi c behaviors for KRAS , experimental evidence for allele-specifi c biological properties is limited. We combined structural biology, mass spectrometry, and mouse modeling to demonstrate that the selection for specifi c KRAS mutants in human cancers from different tissues is due to their distinct signaling properties.

Key Points N-Ras expression is essential for the proliferative advantage of acute myeloid leukemias with oncogenic NRAS/Nras mutations. Mitogen-activated protein kinase kinase inhibition prolongs survival in Nras-mutant AML by reducing proliferation, but fails to undergo apoptosis.

Chronic and juvenile myelomonocytic leukemias (CMML and JMML) are myelodysplastic/myeloproliferative neoplasia (MDS/MPN) overlap syndromes that respond poorly to conventional treatments. Aberrant Ras activation due to NRAS, KRAS, PTPN11, CBL, and NF1 mutations is common in CMML and JMML. However, no mechanism-based treatments currently exist for cancers with any of these mutations. An alternative therapeutic strategy involves targeting Ras-regulated effector pathways that are aberrantly activated in CMML and JMML, which include the Raf/MEK/ERK and phosphoinositide-3´-OH kinase (PI3K)/Akt cascades. Mx1-Cre, KrasD12 and Mx1-Cre, Nf1flox/− mice accurately model many aspects of CMML and JMML. Treating Mx1-Cre, KrasD12 mice with GDC-0941 (also referred to as pictilisib), an orally bioavailable inhibitor of class I PI3K isoforms, reduced leukocytosis, anemia, and splenomegaly while extending survival. However, GDC-0941 treatment attenuated activation of both PI3K/Akt and Raf/MEK/ERK pathways in primary hematopoietic cells, suggesting it could be acting through suppression of Raf/MEK/ERK signals. To interrogate the importance of the PI3K/Akt pathway specifically, we treated mice with the allosteric Akt inhibitor MK-2206. This compound had no effect on Raf/MEK/ERK signaling, yet it also induced robust hematologic responses in Kras and Nf1 mice with MPN. These data support investigating PI3K/Akt pathway inhibitors as a therapeutic strategy in JMML and CMML patients.