Somatic activation of a conditional Kras^G12D allele causes ineffective erythropoiesis in vivo

Anemia with histopathologic and laboratory evidence of ineffective erythropoiesis is a hallmark of myelodysplastic syndromes (MDSs) and of juvenile and chronic myelomonocytic leukemia (JMML and CMML), which are classified as MDS/myeloproliferative disorder (MDS/MPD) overlap diseases.^1,2  In addition to monocytic myeloproliferation, JMML and CMML are characterized by anemia with splenomegaly, extramedullary hematopoiesis, and variable degrees of erythroid dysplasia.^3-6 

Deregulated signaling pathways play a fundamental role in the pathogenesis of myeloid malignancies.⁷  Mutations in NRAS or KRAS2 are among the most common genetic lesions found in acute myeloid leukemia (AML), MDS, and MPD. Ras proteins are signal switch molecules that regulate growth and differentiation by cycling between an active guanosine triphosphate (GTP)–bound state (Ras-GTP) and an inactive guanosine diphosphate (GDP)–bound state. Oncogenic Ras proteins accumulate in the GTP-bound conformation due to defective intrinsic GTP hydrolysis and resistance to GTPase activating proteins (GAPs).^8-10  In addition to oncogenic RAS mutations, Ras signaling is deregulated in myeloid malignancies by alternative genetic mechanisms that include the BCR-ABL fusion, PTPN11 mutations, and NF1 inactivation.^11-13 

Expressing oncogenic RAS alleles with retroviral vectors inhibits differentiation of erythroid progenitors in vitro, suggesting that hyperactive Ras has a direct pathogenic role in the anemia that occurs in MDS and in MDS/MPD overlap disorders.^14,15  However, retroviral transduction requires ex vivo manipulation of primary cells and results in supraphysiologic Ras protein levels that could contribute to these phenotypes.¹⁶  Somatic activation of oncogenic Kras^G12D results in a fatal monocytic MPD in Mx1-Cre, Kras^G12D mice that is associated with anemia and splenomegaly with extramedullary hematopoiesis.^17,18  Here we show that expressing Kras^G12D from its endogenous promoter induces ineffective erythropoiesis in vivo, with an apparent block in differentiation at the proerythroblast stage.

Mice were maintained in the sterile UCSF animal care facility and were fed pelleted chow and acidified water ad libitum. The experimental procedures were approved by the UCSF Committee on Animal Research. Breeding, genotyping, polyinosinic-polycytidylic acid (pIpC) injection and hematologic studies were performed as described.¹⁷ 

Serum erythropoietin (EPO) was measured by enzyme-linked immunosorbent assay (Quantikine MEP00, R&D Systems, Minneapolis, MN). Staining for CD71, TER119, 5′-bromodeoxyuridine (BrdU), and with annexin V and 7-amino-actinomycin D (7-AAD) were performed using commercial fluorochrome-conjugated reagents following manufacturer's instructions (BD Pharmingen, San Diego, CA). Mice were given intraperitoneal injections of BrdU (150 mg/kg) 1 hour prior to harvesting tissue for analysis.

Erythroid differentiation assays were performed according to published methodologies¹⁴  with the addition of a 4-day preincubation in IL-11 (100 ng/mL), SCF (100 ng/mL), Flt-3 (50 ng/mL), IL-6 (20 ng/mL), and IFN-β (400 U/mL; all from R&D Systems) to induce Mx1-Cre expression. Progenitor colonies were enumerated in methylcellulose medium (M3234; StemCell Technologies, Vancouver, BC, Canada) supplemented with recombinant human EPO (R&D Systems) and counted after incubation for 2 to 3 days (erythroid colony-forming units [CFU-Es]) or 7 days (erythroid burst-forming units [BFU-Es]). Photomicrographs were taken using a TMS-F inverted microscope fitted with a 10×/0.25 objective, Ph3 annulus, and Coolpix 5000 digital camera (Nikon, Melville, NY).

Mx1-Cre, Kras^G12D mice develop fatal monocytic MPD on both F1 C57BL/6 × 129Sv/Jae and inbred C57BL/6 backgrounds (Braun et al¹⁷  and data not shown). Although myeloid proliferation is modestly attenuated in C57BL/6 mice, there is no difference in time of onset or severity of anemia in these strains. Blood smears show polychromasia and reticulocytosis, with hypochromia and anisopoikilocytosis in moribund animals. Serum EPO levels are elevated in proportion to anemia, consistent with the histopathologic features suggestive of ineffective erythropoiesis (Figure 1A).

We enumerated erythroid progenitor colonies from Mx1-Cre, Kras^G12D bone marrow and spleen cells in methylcellulose medium. Surprisingly, although erythroid lineage cells were extremely rare in cytologic preparations of Mx1-Cre, Kras^G12D marrow (data not shown), we detected normal numbers of BFU-E and CFU-E progenitors (Figure 1B-C). BFU-Es from Mx1-Cre, Kras^G12D marrow formed abnormally large colonies that were characterized by both EPO-independent growth and hypersensitivity to EPO (Figure 1D-F). Furthermore, the enlarged spleens from Mx1-Cre, Kras^G12D mice contained abundant BFU-Es and CFU-Es (Figure 1B-C), with similarly enlarged BFU-E colonies. Thus, the erythroid progenitor compartment is massively expanded in Mx1-Cre, Kras^G12D mice despite severe anemia. Together with the elevated serum EPO levels and severe reduction in differentiated erythroid precursors, these data support the idea that impaired terminal erythroid development contributes to anemia. These data also underscore the similarity of MPD in Mx1-Cre, Kras^G12D mice to JMML, which is characterized by BFU-Es that exhibit a degree of spontaneous growth as well as growth factor hypersensitivity.^6,19 

We examined bone marrow cells and splenocytes from anemic Mx1-Cre, Kras^G12D mice by flow cytometry to investigate later stages in erythroid differentiation (Figure 2A). These studies revealed normal numbers of early erythroid cells in bone marrow (CD71^hiTER119^–/low, corresponding to proerythroblasts and some CFU-Es²⁰ ), but a paucity of all the more mature TER119^hi populations (regions II-IV, Figure 2A, and Table S1, which is available on the Blood website; see the Supplemental Table link at the top of the online article). This suggests an inefficient transition from TER119^– to TER119^hi stages of erythropoiesis. Supporting this idea, spleens of Mx1-Cre, Kras^G12D mice also contained an excess of immature CD71^hiTER119^–/low cells. However, these spleens also harbored large numbers of TER119^hi erythroblasts, which suggests that the splenic microenvironment partially mitigates the erythropoietic defects caused by hyperactive Ras. The abundance of TER119^hi erythroblasts in spleens of Mx1-Cre, Kras^G12D mice also indicates that inefficient production of CD71⁺TER119^hi cells from CFU-Es does not fully explain the anemia. This is consistent with studies of human MDS, which imply multifactorial etiologies of anemia.¹  The overrepresentation of immature erythroid cells was recapitulated in recipients given transplants of Mx1-Cre, Kras^G12D bone marrow (n = 2) or splenocytes (n = 3), verifying that this phenotype is intrinsic to the hematopoietic compartment. Finally, similar cellular distributions were noted when Mx1-Cre, Kras^G12D fetal liver cells were induced to undergo erythroid differentiation in vitro (Figure 2B), demonstrating a significant cell intrinsic component to the differentiation block.

We gave Mx1-Cre, Kras^G12D mice injections of BrdU to determine the percentage of cells in the S phase of the cell division cycle. The CD71^hiTER119⁺ fraction demonstrated similarly high rates of proliferation in vivo in wild-type and Mx1-Cre, Kras^G12D mice (Figure 2C). Some previous studies of transduced erythroid cells showed that expressing oncogenic Ras proteins increases the percentage of apoptotic cells,¹⁵  whereas others did not.¹⁴  To address this question, we assayed freshly harvested bone marrows and spleens for annexin V staining. Although increased annexin V staining was observed occasionally in CD71^hiTER119⁺ cells from moribund Mx1-Cre, Kras^G12D mice, some mice with severe anemia demonstrated wild-type levels of annexin V staining in all erythroid populations (data not shown). These data indicate that excessive apoptosis may occur in erythroblasts of some animals but does not, in general, explain the anemia in Mx1-Cre, Kras^G12D mice.

In summary, we find that K-Ras^G12D causes dyserythropoiesis in vivo in a genetically and phenotypically accurate model of JMML and CMML. This analysis of primary erythroid cells that express oncogenic Kras from its endogenous promoter in the context of an intact organism extend previous in vitro studies. Our data showing impaired differentiation during terminal erythroid maturation are generally consistent with results from retroviral systems; however, the finding that Mx1-Cre, Kras^G12D mice sustain erythropoiesis more effectively in spleen than in bone marrow also demonstrates a role for the microenvironment in modulating this effect. Because terminal maturation of erythroid cells is highly dependent on their association with supportive macrophages,²¹  we speculate that anemia in Mx1-Cre, Kras^G12D mice may derive, in part, from effects of hyperactive Ras on macrophage differentiation or function or both.

The Nf1 tumor suppressor gene encodes a GAP that attenuates Ras signaling. Ablating a conditional Nf1 mutant allele induces MPD in Mx1-Cre, Nf1^flox/flox mice that is qualitatively similar to the disorder in Mx1-Cre, Kras^G12D mice, but more indolent.²²  Similarly, expressing a conditional mutant Ptpn11 allele results a mild MPD.²³  Interestingly, although splenic erythropoiesis is prominent in both strains, neither develops anemia. Nf1 inactivation and mutant Ptpn11 expression both result in hyperactive Ras signaling; however, Ras-GTP levels are highest in Mx1-Cre, Kras^G12D hematopoietic cells. These data, as well as analyses of germline activated KRAS2 alleles,²⁴  support the idea that effects on erythroid differentiation vary with the degree and duration of Ras activation.

We find that K-Ras^G12D causes anemia, at least in part, through inhibitory effects on erythroid differentiation. Our data support the hypothesis that hyperactive Ras directly contributes to ineffective erythropoiesis and anemia in some patients with MDS and MDS/MPD overlap disorders, in whom transfusion dependence is a major morbidity.²⁵  In vitro data implicate aberrant MEK activity downstream of Ras-GTP in the erythroid differentiation arrest induced by oncogenic Ras.²⁶  Pharmacologic inhibitors of MEK, which are being developed as potential cancer agents, might therefore ameliorate the anemia seen in JMML and CMML, and in some patients with MDS. Mx1-Cre, Kras^G12D mice are a robust preclinical platform for testing this new therapeutic strategy.

T.E.J. is an investigator of the Howard Hughes Medical Institute.