Deficiency of β Common Receptor Moderately Attenuates the Progression of Myeloproliferative Neoplasm in NrasG12D/+ Mice*

Background: GM-CSF signaling is important in establishing and maintaining juvenile/chronic myelomonocytic leukemia (JMML/CMML). Results: The common β chain of GM-CSF receptor (βc) is dispensable for the function of CMML-initiating cells, but βc−/− prolongs the survival of CMML-bearing mice. Conclusion: βc−/− slows down CMML progression but does not abrogate its initiation. Significance: Inhibiting GM-CSF signaling might alleviate JMML/CMML symptoms but would not eradicate the disease. Activating Ras signaling is a major driver in juvenile and the myeloproliferative variant of chronic myelomonocytic leukemia (JMML/MP-CMML). Numerous studies suggest that GM-CSF signaling plays a central role in establishing and maintaining JMML/MP-CMML phenotypes in human and mouse. However, it remains elusive how GM-CSF signaling impacts on JMML/MP-CMML initiation and progression. Here, we investigate this issue in a well characterized MP-CMML model induced by endogenous NrasG12D/+ mutation. In this model, NrasG12D/+ hematopoietic stem cells (HSCs) are required to initiate and maintain CMML phenotypes and serve as CMML-initiating cells. We show that the common β chain of the GM-CSF receptor (βc) is dispensable for NrasG12D/+ HSC function; loss of βc does not affect the expansion, increased self-renewal, or myeloid differentiation bias in NrasG12D/+ HSCs. Therefore, βc−/− does not abrogate CMML in NrasG12D/+ mice. However, βc deficiency indeed significantly reduces NrasG12D/+-induced splenomegaly and spontaneous colony formation and prolongs the survival of CMML-bearing mice, suggesting that GM-CSF signaling plays an important role in promoting CMML progression. Together, our results suggest that inhibiting GM-CSF signaling in JMML/MP-CMML patients might alleviate disease symptoms but would not eradicate the disease.

ture was identified in a subpopulation of monocytic cells defined as CD33 ϩ CD14 ϩ CD34 Ϫ CD38 lo cells (21). This subset of cells can be used to monitor disease status at diagnosis, remission, relapse, and malignant transformation to an acute phase. Consistent with this notion, we found that recipients transplanted with Nras G12D/ϩ cells acquire hypersensitivity to GM-CSF (both the Jak2/Stat5 and the Ras/Raf/MEK/ERK pathways are hyperactivated upon GM-CSF stimulation) during MP-CMML progression, which in some cases is attributed to uniparental disomy of the oncogenic Nras allele (10). Shortterm inhibition of GM-CSF not only induces remission of JMML in engrafted immunodeficient mice (22) but also achieves transient clinical response in an end-stage JMML patient (23).
The long-term effects of inhibiting GM-CSF on JMML are evaluated in the Nf1 Ϫ/Ϫ model. Four of six Gmcsf Ϫ/Ϫ mice transplanted with Nf1 Ϫ/Ϫ ; Gmcsf Ϫ/Ϫ fetal liver cells develop JMML-like phenotypes with prolonged latency (24), which is postulated to be due to the residual activity of pre-formed GM-CSF receptor in the absence of GM-CSF (25). Subsequently, a study using the Mx1-Cre transgene to inactivate a conditional Nf1 allele in ␤c Ϫ/Ϫ hematopoietic cells shows that the severity of JMML-like phenotypes is reduced but not abrogated, whereas mice transplanted with Nf1 Ϫ/Ϫ ; ␤c Ϫ/Ϫ stem cells do not develop significant JMML over 1 year (26). Therefore, it remains elusive how GM-CSF signaling impacts on JMML/ MP-CMML initiation and progression.
We addressed this question using the well characterized MP-CMML model induced by oncogenic Nras (10 -12, 27, 28). In this model, acute expression of oncogenic Nras from its endogenous locus leads to expanded hematopoietic stem cell (HSC) compartment, increased HSC self-renewal, and myeloid differentiation bias in mutant HSCs. The myeloid progenitor (MP) compartment is expanded without significant hyperactivation of GM-CSF signaling or hyperproliferation in these cells, suggesting that the increased number of MPs is largely due to the increased myeloid differentiation potential of Nras G12D/ϩ HSCs. Consistent with this notion, ϳ95% of recipients transplanted with Nras G12D/ϩ bone marrow cells develop MP-CMML-like phenotypes after a long latency. Nras G12D/ϩ HSCs are required to initiate and maintain the disease phenotypes and thus serve as CMML-initiating cells, whereas MPs acquire secondary genetic hits to gain hypersensitivity to GM-CSF and push monocytosis to develop in vivo.
We chose a genetic approach to stably delete GM-CSF signaling in Nras G12D/ϩ mice. The GM-CSF receptor shares a common ␤ subunit (␤c) with IL-3 and IL-5 receptors (29). The GM-CSFR␣ and IL-5 receptor ␣ subunits only pair with ␤c, whereas IL-3 receptor ␣ subunit forms heterodimers with both ␤c and an IL-3-specific ␤ subunit in mouse (30). Therefore, ␤c Ϫ/Ϫ bone marrow cells respond to IL-3 but not GM-CSF. Here, we report that Nras G12D/ϩ ; ␤c Ϫ/Ϫ HSCs are very similar as Nras G12D/ϩ HSCs; both of them show comparably increased numbers, increased self-renewal, and increased myeloid differentiation bias resulting in an expanded MP compartment. These results suggest that ␤c is dispensable for Nras G12D/ϩ HSCs, the CMML-initiating cells. Therefore, deletion of ␤c does not abrogate CMML formation in recipients transplanted with Nras G12D/ϩ cells. However, loss of ␤c-mediated GM-CSF signaling indeed attenuates CMML progression as demonstrated by reducing splenomegaly, abolishing spontaneous colony formation, and prolonging the survival of recipients with Nras G12D/ϩ cells. Our data suggest that inhibiting GM-CSF signaling in JMML/MP-CMML patients might provide transient symptomatic improvements but would not eradicate the disease.
To induce Cre expression, 5-7-week-old mice were injected intraperitoneally with 100 g of polyinosinic-polycytidylic acid (pI-pC; GE Healthcare) every other day for two doses. The day of first pI-pC injection was defined as Day 1. All animal experiments were conducted in accordance with the Guide for the Care and Use of Laboratory Animals (National Institutes of Health) and approved by an Animal Care and Use Committee at the University of Wisconsin-Madison. The program is accredited by the Association for Assessment and Accreditation of Laboratory Animal Care.
Flow Cytometric Analysis of Hematopoietic Tissues-For lineage analysis of peripheral blood, bone marrow, and spleen, flow cytometric analyses were performed as described previously (9). HSCs, multipotent progenitors, Lin Ϫ Sca1 ϩ cKit ϩ (LSK), and MPs in bone marrow and spleen were analyzed as described previously (12,27). Stained cells were analyzed on a FACSCalibur or LSRII (BD Biosciences).
Colony Assay-A total of 5 ϫ 10 4 bone marrow cells were plated in duplicate in semi-solid medium MethoCult M3234 (StemCell Technologies) supplemented with mouse GM-CSF or IL-3 (PeproTech, Rocky Hill, NJ) according to the manufa-cturer's protocol. The colonies were counted after 7-10 days in culture.
EdU Incorporation-EdU (Invitrogen) was administered as a single dose of 1 mg injected intraperitoneally. EdU incorporation was measured 16 h later using the Click-It EdU Pacific Blue flow kit (Invitrogen) as described previously (31). Briefly, Sca1 ϩ cells were enriched using an AutoMACS (Miltenyi). Enriched cells were first stained with FITC-conjugated antibodies against CD41, CD48, B220, TER119, Gr1, and allophycocyanin-conjugated CD150. After Click-It reaction, cells were then stained with phycoerythrin-conjugated c-Kit and PerCP Cy5.5-Sca1. The stained cells were analyzed on an LSRII (BD Biosciences).
Complete Blood Count and Histopathology-Complete blood count analysis was performed using a Hemavet 950FS (Drew FIGURE 1. Loss of ␤c decreases Nras G12D/؉ -induced splenomegaly. Mice were treated with pI-pC and euthanized on Day 12 for analysis as described under "Materials and Methods." A, schematic illustration of the strategy to generate experimental mice. B, genotyping analysis of genomic DNA to detect different alleles in representative control, Nras LSLG12D/ϩ , and Nras LSLG12D/ϩ ; Mx1-Cre; ␤c Ϫ/Ϫ mice. C, the ratio of spleen weight to body weight (BW) of control, Nras G12D/ϩ , and Nras LSLG12D/ϩ ; ␤c Ϫ/Ϫ mice. D, complete blood count was performed on peripheral blood samples collected from control, Nras G12D/ϩ , and Nras G12D/ϩ ; ␤c Ϫ/Ϫ mice. The range and median of the data are shown. E, flow cytometry analysis of bone marrow (BM), peripheral blood (PB), and spleen (SP) cells using myeloid lineage-specific markers. Data are presented as mean Ϯ S.D. *, p Ͻ 0.05; **, p Ͻ 0.01. Scientific). Mouse tissues were fixed in 10% neutral buffered formalin (Sigma-Aldrich) and further processed at the University of Wisconsin Carbone Comprehensive Cancer Center (UWCCC) Histology Lab.
Immunohistochemistry-Immunohistochemistry was performed on the Ventana Discovery XT biomarker platform. Sections were pretreated through antigen retrieval using CC2, a proprietary citrate buffer (Ventana, 760-107). The following steps were carried out with Reaction Buffer (Ventana 950-300) rinses between them. Primary antibody pan-cytokeratin (Thermo Scientific, PIPA521985) was diluted 1:100 using Renoir Red antibody diluent (BioCare Medical, PD904H), applied to sections, and incubated for 28 min at 37°C. Omni-Map anti-rabbit HRP biotin-free conjugate (Ventana, 760-4311) was applied for 12 min at 37°C followed by detection using ChromoMap DAB kit (Ventana, 760-159) for 8 min. Tissues were counterstained with CAT hematoxylin (BioCare Medical, CATHE-M) and coverslipped. All these steps were performed by the Department of Pathology's Translational Research Initiatives in Pathology, part of the UWCCC Translational Science BioCore.

Results
Deletion of ␤c Decreases Oncogenic Nras-induced Splenomegaly-To investigate whether deletion of ␤c affects the hematopoietic phenotypes induced by oncogenic Nras, we generated Mx1-Cre, Nras LSL G12D/ϩ ; Mx1-Cre and Nras LSL G12D/ϩ ; Mx1-Cre; ␤c Ϫ/Ϫ mice (Fig. 1, A and B). At 5-7 weeks of age, these mice were administrated with pI-pC, which stimulates endogenous interferon production and induces Cre expression in various tissues but predominantly the hematopoietic tissues from the interferon-inducible promoter Mx1 (32). The Cre recombinase subsequently removed the stop cassette and induced oncogenic Nras expression from its endogenous locus. The day of the first pI-pC injection is defined as Day 1. After two rounds of pI-pC injection, all mice were sacrificed on Day 12. We refer to the pI-pC-treated compound mice as Nras G12D/ϩ and Nras G12D/ϩ ; ␤c Ϫ/Ϫ mice, respectively, and pI-pC-treated Mx1-Cre mice as control mice throughout this study. 3 After acute induction of oncogenic Nras expression in hematopoietic tissues, both Nras G12D/ϩ and Nras G12D/ϩ ; ␤c Ϫ/Ϫ mice were grossly normal, with unremarkable white blood cell counts, hematocrit, and platelet counts (Fig. 1D) and normal myeloid differentiation in bone marrow and peripheral blood (Fig. 1E). However, Nras G12D/ϩ mice showed moderately but significantly enlarged spleen when compared with control mice (Fig. 1C). Flow cytometric analysis demonstrated that the percentages of granulocytes (Mac ϩ Gr1 ϩ ) and monocytes (Mac1 ϩ Gr1 Ϫ ) were significantly increased in Nras G12D/ϩ spleens when compared with those in control spleens (Fig. 1E). Noticeably, the splenomegaly and percentage of splenic monocytes were significantly reduced in Nras G12D/ϩ ; ␤c Ϫ/Ϫ mice when compared with those in Nras G12D/ϩ mice (Fig. 1, C and E). These results indicate that deletion of ␤c decreases oncogenic Nrasinduced monocytic cell expansion in spleen, which might contribute to the reduced splenomegaly in Nras G12D/ϩ ; ␤c Ϫ/Ϫ mice.
suggest that deletion of ␤c does not affect increased self-renewal of Nras G12D/ϩ HSCs. Together, our results demonstrate that ␤c is dispensable for Nras G12D/ϩ HSCs.
Loss of ␤c Abolishes Oncogenic Nras-induced Spontaneous Colony Formation-To test whether loss of ␤c affects MP expansion in Nras G12D/ϩ mice, we analyzed the MP (Lin Ϫ IL7R␣ Ϫ Sca1 Ϫ cKit ϩ ) compartment in control, Nras G12D/ϩ , and Nras G12D/ϩ ; ␤c Ϫ/Ϫ mice. The absolute numbers of MPs, including common myeloid progenitors, granulocyte-monocyte progenitors, and megakaryocyte-erythroid progenitors, in Nras G12D/ϩ and Nras G12D/ϩ ; ␤c Ϫ/Ϫ bone marrow were comparable with each other, and both numbers were significantly higher than those in control bone marrow (Fig. 4A). A similar trend was also observed in spleen (Fig. 4A). Our data indicate that deletion of ␤c does not affect oncogenic Nras-induced MP expansion.
We and others previously reported that bone marrow cells from Nras G12D/ϩ mice form a significant number of colonies in the absence of cytokines (10,11). To determine whether the spontaneous colony formation of Nras G12D/ϩ cells depends on ␤c-mediated signaling, we isolated bone marrow cells from control, Nras G12D/ϩ , and Nras G12D/ϩ ; ␤c Ϫ/Ϫ mice and plated them in semi-solid medium in the absence of cytokines. Consistent with previous studies (10, 11), Nras G12D/ϩ cells formed a significant number of colonies, whereas Nras G12D/ϩ ; ␤c Ϫ/Ϫ cells formed a much lower number of colonies (Fig. 4B). In the presence of mGM-CSF or mIL-3, Nras G12D/ϩ cells formed significantly more and bigger colonies than control cells (Fig. 4B). As expected, Nras G12D/ϩ ; ␤c Ϫ/Ϫ cells did not form a significant number of colonies in the presence of mGM-CSF but formed a comparable number and size of colonies as Nras G12D/ϩ cells in the presence of mIL-3, consistent with our signaling studies (Fig. 4B). Our results suggest that the spontaneous colony formation of Nras G12D/ϩ cells largely depends on ␤c-mediated GM-CSF signaling.
␤c Deficiency Significantly Delays the Progression of Oncogenic Nras-induced Leukemias in a Cell-autonomous Manner-To investigate whether ␤c deficiency attenuates oncogenic Nras-induced leukemogenesis in a hematopoietic cell-specific manner, we transplanted bone marrow cells from control, Nras G12D/ϩ , and Nras G12D/ϩ ; ␤c Ϫ/Ϫ mice together with competitor cells into lethally irradiated mice (Fig. 5). Consistent with our previous study (10), ϳ97% of recipients transplanted with Nras G12D/ϩ cells developed a CMML-like disease and ϳ7% developed acute T-cell lymphoblastic leukemia/lymphoma. Similarly, ϳ94% of recipients with Nras G12D/ϩ ; ␤c Ϫ/Ϫ cells developed a CMML-like disease and ϳ12% developed T-cell lymphoblastic leukemia/lymphoma (Fig. 5B). Some mice developed both diseases. However, recipients transplanted with Nras G12D/ϩ ; ␤c Ϫ/Ϫ cells survived significantly longer than those with Nras G12D/ϩ cells (median survival: 537 days versus 357 days) (Fig. 5A). Despite different survival curves, both groups of mice with CMML displayed similar disease phenotypes at the moribund stage, including markedly enlarged spleen (Fig. 5C) with significant extramedullary hematopoiesis (Fig. 5F), increased white blood cell counts and anemia (Fig.  5D), and a predominant expansion of granulocytes and/or monocytes in hematopoietic tissues (Fig. 5E). These results indicate that deletion of ␤c cannot abrogate oncogenic Nrasinduced CMML formation but it does significantly delay CMML progression.
Deletion of ␤c in Nras G12D/ϩ Mice Promotes Hepatic Histiocytic Sarcomas with Atypical Morphology-Our previous results show that ϳ50% of primary Nras G12D/ϩ mice (7 out of 15) died with hepatic histiocytic sarcoma within a year after pI-pC injections (10). Although the median survival of Nras G12D/ϩ ; ␤c Ϫ/Ϫ mice was indistinguishable from that of Nras G12D/ϩ mice, the disease latency was prolonged (Fig. 6). Like Nras G12D/ϩ mice, most of the Nras G12D/ϩ ; ␤c Ϫ/Ϫ mice (6 out of 8) developed CMML-like phenotypes with increased monocytosis in peripheral blood (Fig. 6C). Two out of three Nras G12D/ϩ ; ␤c Ϫ/Ϫ mice also developed multiple hepatic tumor nodules with varying morphology but consisting of histiocytelike cells ranging from small and monotonous to large and multinucleated. Because the tumors also entrapped steatotic hepatocytes and residual sinusoidal endothelial channels (Fig.  7A), we initially considered whether this might represent a nonhematopoietic tumor type, but negative pan-keratin staining (Fig. 7B) and the overall morphologic features favor a histiocytic/monocytic derived neoplasm. We speculate that the morphologic differences between these tumors and those we have previously described in Nras G12D/ϩ mice (10) might be due to their inability to normally respond to GM-CSF-derived signal.

Discussion
In this study, we show that ␤c deficiency indeed abolishes GM-CSF signaling in Nras G12D/ϩ cells but IL-3 signaling is preserved (Fig. 2). Consequently, loss of ␤c does not affect Nras G12D/ϩ HSC functions (Fig. 3) and therefore does not abrogate CMML in Nras G12D/ϩ mice (Fig. 5). However, deletion of ␤c does significantly slow down the progression of CMML and prolong the survival of recipients transplanted with Nras G12D/ϩ cells (Fig. 5). Our findings are summarized in Fig. 8.
We previously reported that in the Nras G12D/ϩ -induced CMML model, mutant HSCs are required to initiate and maintain CMML-like phenotypes and serve as CMML-initiating cells (27). Consistent with an earlier report that ␤c is dispensable for normal HSCs (35), we found that ␤c is dispensable for Nras G12D/ϩ HSCs as well; loss of ␤c does not affect the expansion, increased proliferation and self-renewal, and myeloid differentiation bias in Nras G12D/ϩ HSCs (Figs. 3 and 4). We believe that ␤c is also dispensable for Nf1 Ϫ/Ϫ HSCs. Therefore, it is not surprising that the MP compartment remains expanded in   Nras G12D/ϩ ; ␤c Ϫ/Ϫ (Fig. 4) and Nf1 Ϫ/Ϫ ; ␤c Ϫ/Ϫ mice and that deletion of ␤c does not abrogate CMML in these animals (26).
Despite potential redundancy of other cytokine signaling (e.g. IL-3, G-CSF, and M-CSF) in the absence of ␤c-mediated GM-CSF signaling, ␤c deficiency indeed significantly reduces Nras G12D/ϩ -induced splenomegaly (Fig. 1C) and spontaneous colony formation (Fig. 4B) and prolongs the survival of CMML mice (Fig. 5A), suggesting that GM-CSF signaling plays an important role in promoting CMML progression. Our result is consistent with previous human and mouse studies (10,(21)(22)(23)36). However, in t(8;21)-induced acute myeloid leukemia (AML), GM-CSF is found to reduce the replating ability of RUNX1-ETO-expressing cells and therefore has a negative impact on leukemogenesis; expression of RUNX1-ETO in ␤c Ϫ/Ϫ cells leads to a high penetrance of AML (37). Therefore, hyperactive GM-CSF signaling potentially opposes AML formation by inhibiting transformation of MPs to AML-initiating cells. This might explain the absence of spontaneous AML in oncogenic Ras models and the low incidence of transformation to AML in JMML patients. Although we and others did not see AML genesis in Nras G12D/ϩ ; ␤c Ϫ/Ϫ and Nf1 Ϫ/Ϫ ; ␤c Ϫ/Ϫ mice, we could not rule out the possibility that long-term inhibition of GM-CSF signaling in JMML/CMML patients might increase their risk to develop AML.