Standing at odds: mutated RAS and hematopoietic stem cells

Acute myeloid leukemia (AML) is the most common acute leukemia in adults and is characterized by the accumulation of myeloid leukemic blasts unable to complete differentiation. However, AML is a complex disease with variable outcomes and prognoses.1 Underlying these heterogeneous phenotypes is the fact that each sub-type of AML is defined by a different set of mutations and is controlled by a specific transcriptional and signaling network distinct to that of normal stem and progenitor cells.2 Genes mutated in AML are involved in gene regulation and include transcription factors, chromatin modifiers remodelers, splicing regulators, DNA methyltransferases and signaling regulators that control the activity of inducible transcription factors. The result is a profound deviation from the normal differentiation trajectory, with each AML sub-type taking a different path and establishing its own cellular identity.32 Most AML sub-types carry more than one mutation and, with the exception of MLL-translocations (which are a hallmark of pediatric AML4), for a number of sub-types it has been shown that the first oncogenic hit is not sufficient to cause overt leukemia. In AML patients, mutations in genes from different functional categories co-exist, and data from sequencing studies as well as mouse models support this notion.65

The t(8;21) translocation, occurring in 7% of adult de novo patients, is one of the most frequent cytogenetic aberrations in AML.7 This translocation fuses the DNA-binding domain coding region of the hematopoietic master regulator RUNX1 (AML1) to the Eight-twenty-One (ETO, RUNX1T1 or MTG8) gene, which encodes a nuclear co-repressor. The result is the formation of the AML1-ETO (alternatively named RUNX1-ETO) chimeric protein, which retains the ability to bind to RUNX1 binding motifs but lacks the transactivation domain of RUNX1.98 Germline expression of full-length AML1-ETO in mice causes embryonic lethality,1110 but conditional expression in hematopoietic stem cells (HSC) leads to an initial expansion of myeloid progenitor cells, including HSC and granulocyte-macrophage progenitors (GMP). Such expansion was also seen with AML1-ETO-trans-duced human cord blood-derived HSC in vitro.12 Fusion t(8;21) transcripts have been detected in utero and in postnatal blood samples13 and remain expressed at low levels in blood samples from t(8;21) AML patients in long-term remission.14 Furthermore, several AML1-ETO-expressing mouse models have failed to fully develop t(8;21) AML unless challenged by mutagenesis or aging,1815 indicating the necessity of additional secondary mutations. These findings suggest that this chromosomal rearrangement is the driver mutation establishing a pre-leukemic clone. This notion is supported by the finding that t(8;21) patients present with a number of different secondary mutations.19 The most prominent of these mutations occur in genes encoding for the growth factor receptors KIT and FLT3,2120 rendering these receptors chronically active, together with activating RAS mutations such as the K-RAS (G12D), all of which co-operate with AML1-ETO. Interestingly, K-RAS activating mutations occur late during leukemia development and are rarely detectable within the leukemic stem cell compartment21 (Figure 1A). In myelodysplastic syndromes (MDS), it was shown that the order of acquisition of mutations makes a profound difference on the disease phenotype,22 but whether the same stands true for t(8;21) AML is not known.

Figure 1.Co-expression of AML1-ETO and mutant RAS is incompatible with hematopoietic stem cell (HSC) maintenance. (A) In acute myeloid leukemia (AML) patients, HSC express AML1-ETO, but this is not sufficient to cause overt disease. AML leukemic blasts represent myeloid progenitors harboring the translocation that have undergone secondary oncogenic events, including mutations in signaling pathway genes such as the K-RAS^G12D/+. However, prior to Di Genua et al.,²³ the molecular explanation for the lack of mutations in signaling pathway genes in the pre-leukemic stem cells was unknown. HSC (purple cells), myeloid progenitors (purple cells with pink dots). (B) Di Genua et al.²³ generated conditional murine knock-in models expressing human AML1-ETO and K-RAS(G12D) individually or in combination: Aml1^ETO/+ (purple nuclei), K-Ras^G12D/+ (pink nuclei), and Aml1^ETO/+ K-Ras^G12D/+(pink and dotted nuclei), respectively, and performed competitive transplantation assays to compare wild-type and transgenic stem cell activity and measure global gene expression as indicated. (C) Model of step-wise oncogenesis in t(8;21) patients: the presence of t(8;21) in HSC results in a quiescent phenotype. Acquisition of K-RAS(G12D) occurs at later stages and results in increased proliferation. However, the double oncogenic event is not sufficient to develop overt AML, and the nature and order of acquisition of additional oncogenic events remains unknown.

In this issue of Haematologica, Di Genua et al.23 uncovered the mechanistic explanation for the absence of signaling pathway mutations, such as K-RAS activating mutations, within the t(8;21) pre-leukemic HSC compartment by analyzing hematopoietic development and gene expression in conditional murine knock-in models expressing human AML1-ETO and K-RAS(G12D) individually or in combination, resulting in the generation of an Aml1^ETO/+ (AM), a K-Ras^G12D/+ (KM), and a double-targeted Aml1^ETO/+; K-Ras^G12D/+ (AKM) mouse lines. Prior studies had shown that mice expressing K-Ras(G12D) from its endogenous locus develop MDS but not AML.24 When combined with AML1-ETO in human hematopoietic cells, K-RAS(G12D) promoted leukemic transformation in murine transplantation models.25 However, here expression levels were likely to be non-physiological and it had not been defined whether the presence of mutated RAS was compatible with the maintenance of AML1-ETO-expressing HSC. Di Genua et al. addressed these issues by conditionally introducing both mutant genes from the respective genomic loci in mice in order to reproduce physiological oncogene expression levels. They then subsequently performed competitive transplantation assays to be able to directly compare stem cell activity in wild-type and transgenic cells (Figure 1B).

AML1-ETO expression on its own affected platelet, B-and T-cell development, and led to an increased number of functional HSC together with enhanced myeloid reconstitution in secondary recipients. Expression of mutant K-RAS on its own resulted in a myeloproliferative phenotype, led to a lack of HSC expansion, decreased engraftment and diminished in vitro re-plating potential, which, importantly, was seen regardless of the presence of AML1-ETO. Interestingly, co-expression of both oncoproteins led to a milder myeloproliferative phenotype but not overt AML, as seen in patients, and only aggravated the defect in platelet development. The authors suggest that the competitive advantage observed in AML1-ETO-expressing HSC was due to an enhanced self-renewal capacity. To further elucidate the mechanisms behind the functional impairment of K-Ras(G12D)-expressing HSC, they compared gene expression in HSC from genetically modified and control cells using transplantation assays. The presence of mutant K-RAS conferred increased cell cycle activity as well as upregulation of the expression of checkpoint associated genes, such as E2F, Myc and G2M-associated genes, as compared to HSC harboring the AML1-ETO transgene only. Moreover, the presence of mutated K-RAS protein resulted in the downregulation of gene expression signature associated with self-renewal activity and acquisition of a GMP-associated transcriptional signature. Overall, the transcriptional profiles of K-RAS(G12D)-expressing HSC resembled those of myeloid progenitors. Gene expression changes in double mutant cells were distinct from those of cells carrying the individual mutations. For example, two genes, Gja1 and Gzmb, were up-regulated in the double mutant HSC and the authors suggest that they regulate the p53 pathway and oxidative phosphorylation, respectively, both increasing cell death and apoptosis.

Taken together, Di Genua et al.23 show that expression of a mutant K-RAS is not compatible with a pre-leukemic state of murine AML1-ETO-expressing HSC. Although AML1-ETO alone confers a competitive advantage to HSC, the additional presence of K-RAS(G12D) results in loss of HSC function, most likely by exhaustion. The study shows that this phenomenon is explained by an increase in cell cycle activity leading to a loss of quiescence in HSC co-expressing both mutations. Therefore, they hypothesize that acquisition of mutant K-RAS in t(8;21) AML occurs at the myeloid progenitor stage rather than within pre-leukemic HSC compartment harboring the t(8;21) translocation. Overall, this study demonstrates that the order of appearance of each class of mutations is also relevant for leukemic transformation in AML.

Additional questions arise from this and previous work (Figure 1C). Firstly, Di Genua et al. show that, even in the presence of K-RAS(G12D), AML1-ETO is unable to cause an overt leukemic phenotype in vivo. The same was found in xenotransplantation experiments with retrovirally transduced human CD34 cord blood cells expressing AML1-ETO together with N-RAS(G12D).26 This notion ties in with the observation of Cabezas-Wallscheid et al.16 that the development of AML in an AML1-ETO mouse model is a slow process, probably requiring multiple mutations or epigenetic reprogramming events, which were not examined in their study. Moreover, it has been previously shown that expression of AML1-ETO in an inducible transgenic mouse model leads to a block in differentiation, but not to an enhanced proliferation.17 Di Genua et al.23 add to this result that AML1-ETO-expressing pre-leukemic stem cells have a quiescent phenotype, leading to an accumulation of cells with high re-plating activity that are capable of going into cycle in a transplantation setting. Given that the double oncogenic event studied here is not enough to cause AML, the question that is now apparent is which additional events cause AML in patients. This work demonstrates that, even with a type of AML that has been studied in fine molecular detail for decades, the players of selection and clonal evolution in patients still have additional cards up their sleeve.