The ability of endogenous Nras oncogenes to initiate leukemia is codon-dependent

Dear Editor

The mammalian Ras proteins, H-, N-, and K-Ras, belong to the small GTPase superfamily. They cycle between a GTP-bound active form and a GDP-bound inactive form (1, 2). The conversion from the GTP-bound form to the GDP-bound form is mediated by the intrinsic GTPase activity of Ras proteins and is greatly accelerated by Ras-associated GAPs (GTPase-activating proteins). Canonical, cancer-associated Ras mutations at codons G12, G13, or Q61 severely compromise the hydrolysis of GTP to GDP, leading to the accumulation of Ras-GTP and hyperacivation of Ras downstream signaling. Mutations in the KRAS and NRAS genes (not in HRAS) are frequently identified in myeloid disorders (15-60%), including acute myeloid leukemia (AML), atypical chronic myeloid leukemia, chronic myelomonocytic leukemia (CMML), and juvenile myelomonocytic leukemia (JMML) (3). In these diseases, the frequency of NRAS mutations is significantly higher than that of KRAS mutations.

Although mutations at codons G12, G13, and Q61 of NRAS are all perceived oncogenic equivalents, they display distinct biochemical and functional properties. NRAS G12 and G13 mutations affect the intrinsic GTPase activity of Nras by preventing proper position of the Ras-GAP arginine finger within the catalytic site, while Q61 mutations lead to more severe impairment of Nras intrinsic GTPase function (1). Therefore, G12 and G13 mutations are less activating than Q61 mutations. In human cancers with NRAS mutations, G12 mutations are prevalent in myeloid leukemia, while Q61 mutations are predominant in plasma cell myeloma and melanoma (Fig. S1A). In hematopoietic malignancies, ~70% of NRAS G12 mutations are identified in patients with myeloid diseases but these mutations are rare in those with plasma cell myeloma (Fig. S1B). In contrast, the incidence of Q61 mutations is equal in patients with myeloid leukemias or plasma cell myeloma (~40% each, Fig. S1B). Of note, the acquisition of two copies of oncogenic NRAS G12D or G13D alleles is associated with JMML/CMML progression in humans and mice (COSMIC database and (4)), indicating that incremental activation of Ras signaling is a pathological mechanism contributing to JMML/CMML development. In a recent study of mutation-specific Nras oncogenicity, only an endogenous Nras Q61R mutation could promote spontaneous melanoma formation in mice (5). However, the ability of endogenous Nras G12D and Q61R mutants to drive leukemogenesis has never been evaluated in a side-by-side comparison.

We and others previously showed that expression of one or two copies of oncogenic Nras G12D (Nras^G12D/+ and Nras^G12D/G12D) in the hematopoietic compartment induces a chronic or an acute myeloproliferative neoplasm (MPN) respectively, closely resembling human JMML/CMML (6-9). Here, we used the same interferon-inducible Mx1-Cre to generate Nras^Q61R/+ and Nras^Q61R/Q61R mice (see details in Supplementary Methods). To our surprise, Nras^Q61R/Q61R mice were not found at the time of weaning, likely due to the leaky expression of Nras^Q61R on the paternal parent side, which was more detrimental than that of the equivalent Nras^G12D allele (Table S1). Therefore, our entire study focused on the comparison among Nras^G12D/+, Nras^Q61R/+, and Nras^G12D/G12D animals.

After the induction of oncogenic Nras expression using polyinosinic-polycytidylic acid (pI-pC) injections (Fig. S2A), we evaluated the levels of GTP-bound Ras isoforms in total bone marrow cells (Fig. 1A). Consistent with our previous results (6), Nras-GTP level moderately but significantly increased in Nras^G12D/+ cells compared to that in control cells (Fig. 1A). This increase was slightly higher in Nras^Q61R/+ cells and maximal in Nras^G12D/G12D cells. In contrast to what is observed in Kras^G12D/+ cells, wherein levels of GTP bound wild-type (WT) H- and N-Ras are significantly elevated (10), no increase in WT Hras and Kras activation was observed in Nras^G12D/+, Nras^Q61R/+, and Nras^G12D/G12D cells (Fig. 1A). Consistent with this incremental increase of Nras-GTP levels, Nras^G12D/+, Nras^Q61R/+, and Nras^G12D/G12D mice displayed incrementally more severe MPN phenotypes in general, as measured in spleen weight and morphology (Fig. 1B and S2B), white blood cell count (Fig. S3), and the abundance of monocytes (Mac1⁺ Gr1^-) and neutrophils (Mac1⁺ Gr1⁺) within various hematopoietic tissues (Fig. S2C). We closely monitored Nras^Q61R/+ mice and found that they exhibited an intermediate survival rate between Nras^G12D/+ and Nras^G12D/G12D mice (Fig. 1C). Detailed evaluation of moribund mice revealed that 100% of them died with an advanced MPN (Fig. S4) and ~10% simultaneously developed acute T-cell lymphoblastic leukemia/lymphoma (T-ALL) (Fig. 1D). In contrast, 100% of Nras^G12D/+ mice developed MPN and/or histiocytic sarcoma without T-ALL, while 100% of Nras^G12D/G12D mice developed MPN and ~30% simultaneously developed T-ALL (Fig. 1D). These results indicate that Nras^Q61R/+ has an intermediate leukemogenic activity between Nras^G12D/+ and Nras^G12D/G12D.

Figure 1. Somatic activation of Nras^Q61R/+ leads to intermediate leukemia phenotypes between Nras^G12D/+ and Nras^G12D/G12D.

(A, B) Six-seven weeks old control (C), Nras^G12D/+ (G/+), Nras^Q61R/+ (Q/+), and Nras^G12D/G12D (G/G) mice were injected with pI-pC and sacrificed on Day 5 for analysis as described in Methods. (A) Whole cell lysates were extracted from total bone marrow cells and assayed for expression levels of different Ras isoforms. Ras-GTP was affinity purified from WCL using a GST fusion with the Ras binding domain of Raf (Raf RBD) immobilized on agarose beads. Different Ras isoforms and total Ras bound with GTP were analyzed using Western blot analyses. The results were quantified using ImageJ software. The ratios of Ras-GTP/Ras in control cells are arbitrarily set at 1. (B) Quantification of spleen weight. (C) Kaplan-Meier survival curves were plotted against days after 1^st pI-pC injection. P values were determined using the Log-rank test. (D) Disease incidences of moribund mice. The results are presented as mean ± SD. * P<0.05; *** P<0.001.

We further investigated the underlying cellular mechanisms contributing to the phenotypes observed in Nras^G12D/+, Nras^Q61R/+, and Nras^G12D/G12D mice. Compared to control mice, the numbers of Lin^- c-Kit⁺ Sca1⁺ (LSK) cells and myeloid progenitors (MPs) were generally increased in the bone marrows of all three groups of oncogenic Nras mice and incrementally increased in their spleens (Fig. S5A and 2A), correlating with their respective spleen size. Consistent with our previous results (4, 11), the common lymphoid progenitor (CLP) compartment was only expanded in Nras^Q61R/+ and Nras^G12D/G12D mice but not in Nras^G12D/+ mice (Fig. S5B). This observation may explain why TALL was developed in Nras^Q61R/+ and Nras^G12D/G12D mice, but not in Nras^G12D/+ animals. Cell cycle analysis revealed a significant increase in the proliferation of Nras^G12D/+, Nras^Q61R/+, and Nras^G12D/G12D LSK cells (Fig. S5C), but increased proliferation was only significant in Nras^Q61R/+ and Nras^G12D/G12D MPs but not in Nras^G12D/+ MPs (Fig. 2B). Colony forming assays demonstrated that in the absence of cytokines or in the presence of IL-3/mGM-CSF, Nras^G12D/+, Nras^Q61R/+, and Nras^G12D/G12D bone marrow cells formed incrementally increased number and size of colonies (Fig. 2C and S6). Together, the MP phenotypes in Nras^G12D/+, Nras^Q61R/+, and Nras^G12D/G12D mice are highly consistent with the increasing severity of MPN phenotypes found in these mice.

Figure 2. Nras^Q61R/+ displays intermediate capability to promote proliferation, colony formation, and cytokine signaling in myeloid progenitors.

Six-seven weeks old control (C), Nras^G12D/+ (G/+), Nras^Q61R/+ (Q/+), and Nras^G12D/G12D (G/G) mice were injected with pI-pC and sacrificed on Day 5 for analysis as described in Methods. (A) Quantitative analysis of common myeloid progenitors (CMPs), granulocyte-monocyte progenitors (GMPs), and megakaryocyte-erythrocyte progenitors (MEPs) in hind limb bone marrow [BM(H.L.)] and spleen (SP). (B) Cell cycle analysis of myeloid progenitor cells using Ki67 and DAPI staining. (C) Bone marrow cells were cultured in methylcellulose-based medium M3234 supplemented with purified mGM-CSF or mIL-3 as described in Methods. The colonies were counted after 7 days in culture. (D, E) Whole bone marrow cells were serum- and cytokine-starved for 2 hours at 37°C. Levels of p-ERK1/2 were measured using phospho-flow cytometry (D). Cells were then stimulated with different concentration of mGM-CSF or mIL-3 for 10 minutes at 37°C. Levels of p-ERK1/2 were measured using phospho-flow cytometry (E). Non-neutrophil bone marrow cells were gated for data analysis. Myeloid progenitors are enriched in Lin^-/low c-Kit⁺ cells, whereas myeloid precursors are enriched in Lin^-/low c-Kit^- cells. To quantify the activation of ERK1/2, median intensities of p-ERK1/2 at different GM-CSF or IL-3 concentrations are compared to their respective control cells at 0 ng/ml, which is arbitrarily set at 1. Representative plots from one experiment are shown. The results are presented as mean ± SD. Asterisks of different colors represent P values when comparing corresponding group with control. * P<0.05, ** P<0.01 and *** P<0.001.

To determine whether the expansion of the MP compartment is associated with the hyperactivation of cytokine signaling, we performed phos-flow analysis in Lin^- c-Kit⁺ cells (enriched for MPs) and Lin^- c-Kit^- cells (enriched for myeloid precursors) from Nras^G12D/+, Nras^Q61R/+, and Nras^G12D/G12D mice. Total bone marrow cells were first serum- and cytokine-deprived for 2 hours before fixation and permeabilization to measure phospho-ERK1/2 (p-ERK1/2) and p-STAT5 levels (Fig. 2D and S7A). As expected, the ERK pathway was hyperactivated in both populations of cells from Nras^G12D/+, Nras^Q61R/+, and Nras^G12D/G12D mice, while pSTAT5 levels in these cells were comparable to those in control cells. Upon cytokine stimulation, both Lin^- c-Kit⁺ and Lin^- c-Kit^- cells from Nras^G12D/+, Nras^Q61R/+, and Nras^G12D/G12D mice showed significant hyperactivation of ERK1/2 in response to GM-CSF but no or marginal hyperactivation in response to IL-3 (Fig. 2E). Interestingly, p-STAT5 level was moderately increased in Lin^- c-Kit^- cells from all 3 genotypes but not in Lin^- c-Kit⁺ cells following GM-CSF or IL-3 stimulation (Fig. S7B).

Interestingly, despite the more potent leukemogenic activity of the Nras^Q61R allele, NRAS Q61 mutations are less frequent than G12 mutations in hematopoietic malignancies (Fig. S1). One possible explanation of this observation is that NRAS G12 mutations could be initiating events in leukemogenesis or acquired during leukemia progression, while Q61 mutations are more likely to serve as secondary events during leukemia development. This possibility is supported by the observation that NRAS G12 mutations are identified in both chronic and acute stages of hematopoietic malignancies, while Q61 mutations are enriched in acute or more advanced stage of these diseases (3).

In summary, our data indicate that endogenous Nras^Q61R/+ induces an increase of Nras-GTP and cytokine-evoked signaling, which is intermediate between Nras^G12D/+ and Nras^G12D/G12D. Therefore, the leukemogenic activity of Nras^Q61R/+ is stronger than Nras^G12D/+ but weaker than Nras^G12D/G12D. The simplicity of breeding and the intermediate leukemogenic activity of the Nras^Q61R/+ model make it ideal to test whether additional genetic mutations that might accelerate or alleviate its leukemia phenotypes. In addition, given the prevalence of NRAS Q61 mutations in certain hematopoietic malignancies (e.g. plasma cell myeloma), the conditional Nras^Q61R allele, when combined with appropriate transgenic Cre lines, could be used to physiologically mimic these specific diseases in vivo.

Sincerely,