Nf1 Regulates Hematopoietic Progenitor Cell Growth and Ras Signaling in Response to Multiple Cytokines

Heterozygous Nf1^+/− mice were mated to produce Nf1^−/− embryos. Pregnant Nf1^+/− females were killed by cervical dislocation on day 13.5 of gestation and the embryos were removed through an anterior abdominal incision. After individual embryos were isolated, each fetal liver was removed and transferred to a 10 × 35 mm dish in IMDM supplemented with 20% FCS. A single cell suspension was prepared by pushing the hepatic tissues through progressively smaller needles (16–27 gauge). This procedure yielded 1–1.5 × 10⁷ mononuclear cells.

Once the liver was removed, the remainder of the embryonic tissues were used for genotype analysis. Genomic DNA from fetal tissues were isolated as previously described (30). An assay based on the PCR was used to perform genotype analysis. The targeting vector used to disrupt the murine Nf1 gene truncates exon 31 and inserts a neomycin resistance gene (neo). The 3′ (reverse) primer is complimentary to a sequence in the 3′ region of exon 31 that is present in both wild-type and targeted Nf1 genes. Two 5′ primers were designed to distinguish disrupted and wild-type genes. The first is complimentary to neo and the second is complimentary to a wild-type exon 31 sequence. Amplification with the neo and 3′ Nf1 primer pair yields a 350-bp product, whereas amplification with the 5′ Nf1 and 3′ Nf1 pair gives a 230-bp fragment. The sequences for the PCR primers used in these studies are: 5′ Nf1 exon 31: CACCTTTGTTTGGAATAT, 3′ Nf1 exon 31: TTCAATACCTGCCCAAGG, and neo: ATTCGCCAATGACAAGAC.

Unfractionated fetal liver cells (1–2 × 10⁶) from Nf1^−/− or Nf1^+/+ littermates were transplanted into C57Bl/6J recipient mice conditioned with 1,100 rads of gamma irradiation from a Cs source (Gamma cell 40 Extractor; MDS Nordion Kanata, Ontario, Canada) with 700 rads given in the first dose and 400 rads given 4 h later. Recipients of Nf1^−/− cells consistently develop myeloproliferative hematopoiesis with large numbers of mature neutrophils and monocytes in the peripheral blood and prominent splenomegaly (27). Bone marrow and splenic cells were obtained from recipients 18–30 wk after transplantation. Hematopoietic cells were evaluated at death by PCR to confirm the detection of only donor-derived cells.

Low density cells were incubated with a PE-conjugated c-kit monoclonal antibody or a PE-conjugated Sca1⁺ monoclonal antibody whose corresponding Ly6A/E antigen is highly expressed on all primitive hematopoietic progenitors and cells in the stem cell compartment in C57Bl6 strain (45). Cells were incubated with 1 μg of c-kit or Sca1 antibody/10⁶ cells, (PharMingen, San Diego, CA), and the cells were then incubated on ice for 20 min, pelleted, washed, repelleted, and then suspended in FACS buffer. Cells were also simultaneously stained with antibodies to the following lineage markers including: anti-B220, anti-CD3ε, anti–Gr-1, and anti–Mac 1 all conjugated to FITC (PharMingen). They were then separated by FACS^® using a dual laser FACStar PLUS^® equipped with an argon and a krypton laser (Becton Dickinson, Mountain View, CA).

Recombinant GM-CSF, IL-3, and SCF were obtained from Peprotech (Rocky Hill, NJ); recombinant mIL-1α and M-CSF were purchased from R&D Research Laboratories (Minneapolis, MN). Purified hematopoietic cells were placed into culture in triplicate 35-mm plates at a final concentration of 5 × 10² cells per ml. Cells and recombinant growth factors were added to the methylcellulose medium, (for growth of CFU-GM) or agar (for growth of low proliferating potential [LPP] colony-forming cells [CFC] and high proliferating potential [HPP]-CFC) and the solution was thoroughly admixed before plating as previously described (30, 46, 47). Cells cultured for growth of CFU-GM were maintained at 37°C in a humidified 5% CO₂ incubator 95% O₂ and scored on day 10 of culture. Growth factors used for culture of HPP-CFC and LPP-CFC included SCF, IL-1α, CSF-1, and GM-CSF or IL-3. Cultures for growth of LPP-CFC were cultured in 8% CO₂, 5% O₂, and were scored on day 7 of culture, whereas those for HPP-CFC were scored on day 14 of culture. Colonies were scored by indirect microscopy.

The incorporation of phosphate into myelin basic protein (MBP) (Sigma Chemical Co., St. Louis MO) was used as a measure of MAP kinase activity as previously described (48–51). Cell pellets were lysed in lysis buffer (10 mM K₂HPO₄, 1 mM EDTA, 5 mM EGTA, 10 mM MgCl₂, 1 mM Na₃ V0₄, 50 mM β-glycerophosphate, 1% NP-40, 10 μg/ml leupeptin, 1 μg/ml pepstatin, and 100 μg/ml PMSF) at 4°C, disrupted in 1 ml of Dounce (Kontes, Vineland, NJ), iced for 30 min, and clarified by ultracentrifugation at 100,000 g for 30 min at 4°C. Total protein was determined using a BCA protein assay kit (Pierce, Rockford, IL). The MBP reaction conditions consisted of 10 μl of cell lysate with 5 μg of MBP substrate and was incubated with 20 mM Tris-Hcl pH 7.25, 10 mM MgCl₂, 500 μM unlabeled ATP, 20 μg BSA, and 5 μCi γ-[³²P]ATP in a total reaction volume of 30 μl, at 30°C × 15′. Three replicates of each reaction were evaluated in each experiment. The reactions were stopped by the addition of 7.5 ml of 5× sample buffer and were then boiled for 2 min. The kinase reaction mixtures were resolved on 12.5% SDS-PAGE gels. Gels were then stained in a 0.2% Coomassie brilliant blue (Boehringer Mannheim, Indianapolis, IN) solution, dried on Whatman paper, and exposed to Kodak x-OMAT AR film (Eastman-Kodak Co., Rochester, NY) at −20°C. In addition to autoradiography, radiolabeled phosphate in MBP gel bands was also determined by liquid scintillation counting of the MBP bands. The cpm value was then used to calculate MAP kinase activity in pmol phosphate incorporated into MBP/min/μg protein in the cell lysate.

Purified populations of c-kit⁺ cells isolated from the bone marrows of mice transplanted with Nf1^−/− or Nf1^+/+ cells were treated with the MAP kinase/ERK (extracellular signal–regulated) kinase (MEK) inhibitor PD 98059 (New England Biolabs, Beverly, MA) to a final concentration of 50 μM as previously described (52). All cultures were maintained at a density of 2 × 10⁵ cells/ml in IMDM supplemented with 20% FCS. Cells were cultured in 100 ng/ml rmSCF and 50 U/ml rmIL-3, which is the maximum stimulating activity of each respective cytokine.

Previous studies in JCML bone marrow samples have implicated cytokine production by accessory cells as contributing to the hypersensitive pattern of myeloid progenitor growth (31, 32). We therefore isolated highly pure (95– 99%) Sca1⁺lin^−/dim cells derived from day 13.5 fetal livers by flow cytometric cell sorting and assayed the growth of CFU-GM in methylcellulose medium supplemented with varying concentrations of either GM-CSF or IL-3. A representative FACS^® analysis of cells before and after cell sorting is shown in Fig. 1. The results of four independent experiments indicate that CFU-GM derived from Nf1^−/− fetal liver Sca1⁺lin^−/dim cells retain a hypersensitive pattern of colony formation in response to GM-CSF but not IL-3 (Fig. 2, a and b). These data suggest that the alteration of growth responsiveness to GM-CSF seen in Nf1^−/− cells is intrinsic to the progenitor compartment.

We next examined the effects of adding a low concentration of murine SCF (10 ng/ml) in combination with either GM-CSF or IL-3 on Sca1⁺lin^−/dim cells on colony formation. CFU-GM colony growth under these conditions is shown in Fig. 2, c and d, and the absolute numbers of progenitor colonies enumerated at maximal concentrations of GM-CSF or IL-3 singly and in combination with SCF are shown in Table 1. As with GM-CSF alone, the dose– response curve for CFU-GM colony formation from Nf1^−/− cells was significantly left-shifted relative to Nf1^+/+ cells in cultures stimulated with GM-CSF + SCF (Fig. 2,c). The hypersensitive pattern of growth seen in Nf1^−/− cells was more pronounced in parallel cultures stimulated with GM-CSF and SCF than in plates containing GM-CSF alone, and there was an increase in the absolute number of progenitors enumerated (Table 1). Surprisingly, the addition of SCF to cultures stimulated with IL-3 induced a hypersensitive pattern of growth in Nf1^−/− cells (Fig. 2,d) that was not seen with IL-3 alone (Fig. 2,b). Although SCF is relatively inefficient at inducing colony formation in the absence of other growth factors (Table 1), we also detected a significant left shift in the dose–response curve of Nf1^−/− versus Nf1^+/+ cells cultured in the presence of SCF alone (Fig. 2 e). Taken together, these data suggest that loss of Nf1 deregulates the growth of hematopoietic progenitors in response to SCF, which acts on more immature populations of progenitors than do IL-3 or GM-CSF, and demonstrate that SCF can synergize with either factor to induce a hypersensitive pattern of in vitro growth in Nf1^−/− progenitors.

One potential explanation for the hypersensitive growth seen in response to SCF alone or in combination with IL-3 is that SCF stimulation induces autocrine production of GM-CSF by Nf1^−/− progenitor cells. To test this hypothesis, an anti–GM-CSF neutralizing antibody was added to the methylcellulose cultures stimulated with a combination of IL-3 and SCF. In pilot experiments, a concentration of anti–GM-CSF antibody that completely blocked CFU-GM colony formation in response to 10 U/ml of GM-CSF (maximum stimulating activity) was established and used in subsequent experiments (data not shown). Fig. 3 summarizes the results of three independent experiments examining the number of CFU-GM colonies seen when either anti–GM-CSF or an isotype control antibody was added to cultures containing IL-3 + SCF. These data indicate that the abnormal pattern of CFU-GM colony formation observed in Nf1^−/− Sca1⁺lin^−/dim cells is maintained in the presence of anti–GM-CSF antibody, and therefore we conclude that hypersensitive growth in the presence of IL-3 + SCF is not due to autocrine GM-CSF production by Nf1^−/− cells.

The experiments described above assayed the growth of colonies derived from CFU-GM, a relatively differentiated, lineage-restricted population of myeloid progenitors. However, data from cultures stimulated with SCF suggested that neurofibromin might also regulate the growth of more primitive hematopoietic compartments. To investigate this possibility, two-layer agar cultures were established to assay the growth of primitive progenitors with significant replating potential indicative of a primitive stem/progenitor cell (46) and referred to as HPP-CFC. Fetal liver cells from Nf1^−/−, Nf1^+/−, and Nf1^+/+ embryos were cultured in two different combinations of four growth factors at saturating activities to enumerate HPP-CFC. The frequency of HPP-CFC observed in cultures were then multiplied by the number of low density cells per organ to calculate the absolute number of HPP-CFC per fetal liver. The results shown in Fig. 4 demonstrate that significantly higher numbers of HPP-CFC are present from the Nf1^−/− fetal liver cells. The data suggest that the loss of Nf1 results in an expansion of HPP-CFC in vivo in the fetal liver and that neurofibromin is important in modulating the growth of primitive myeloid progenitor compartments as well as CFU-GM progenitors.

Stimulation with a variety of growth factors including SCF, GM-CSF, or IL-3 activates the Ras-Raf-MAP kinase cascade in cultured cell lines (5–9). We hypothesized that if neurofibromin is critical to regulate Ras signaling in response to these cytokines in primary hematopoietic progenitors, then loss of Nf1 function might result in an augmented MAP kinase activity after stimulation. To test this possibility, c-kit⁺ cells were isolated from the bone marrows of mice previously irradiated and then reconstituted with fetal liver cells from day 13.5 gestation Nf1^−/− and Nf1^+/+ littermates. In pilot experiments, we determined that the c-kit⁺ cells isolated from transplant recipients had a similar hyperresponsiveness to GM-CSF in clonogenic assays as did Nf1^−/− fetal liver cells (data not shown). After isolation using fluorescence-activated cell sorting, c-kit⁺ cells were maintained in liquid cultures overnight without exogenous growth factors. The cells were then stimulated with either SCF, IL-3, or GM-CSF alone or with a combination of SCF + IL-3. MAP kinase activity was measured in unstimulated cells as well as at 5 and 60 min after the addition of growth factor(s). Three independent experiments yielded similar results; data from one of these is shown in Fig. 5. A threefold elevation in MAP kinase activity was observed in unstimulated c-kit⁺ cells isolated from the Nf1^−/− recipients. Kinase activity increased significantly in Nf1^−/− cells after exposure to IL-3, SCF, or SCF + IL-3, and persisted above baseline at 60 min. There was also an increase in MAP kinase activity at 60 min after stimulation with GM-CSF that approached statistical significance. In contrast, c-kit⁺ cells from the Nf1^+/+ recipients had a much less pronounced increase in MAP kinase activity at 5 min after stimulation, and the kinase returned to baseline by 60 min. We conclude that absence of neurofibromin in c-kit⁺ hematopoietic progenitors results in a constitutive activation of Ras-MAP kinase signaling that can be further stimulated in response to SCF, IL-3, and GM-CSF.

To confirm an association between the clonogenic and biochemical data presented above, c-kit⁺ cells were again isolated from transplant recipients. MAP kinase activity was determined on aliquots of Nf1^+/+ and Nf1^−/− cells after overnight incubation in the presence or absence of an inhibitor of MEK. The remaining cells were either immediately cultured in methylcellulose to score the growth of myeloid progenitor colonies or were placed into liquid culture containing SCF and IL-3 together with MEK or the vehicle control for 48 h before plating. Biochemical data from one of three representative experiments are shown in Fig. 6 and the colony numbers are presented in Fig. 7. Addition of the MEK inhibitor reduced the MAP kinase activity of unstimulated Nf1^−/− cells to wild-type levels and blocked kinase activation in response to SCF and IL-3. Similarly, addition of the MEK inhibitor to liquid cultures of c-kit⁺ cells for 48 h before plating in methylcellulose markedly decreased the growth of differentiated cells, differentiated progenitor cells (LPP-CFC), and primitive hematopoietic progenitors (HPP-CFC) as compared with cultures containing the vehicle only. A reduced biochemical and cellular response to the MEK inhibitor was noted in the Nf1^+/+ hematopoietic cells. Taken together, these data implicate deregulated signaling through the Ras-MEK-MAP kinase in the aberrant pattern of cytokine-induced hematopoietic progenitor colony growth seen in Nf1^−/− cells.

Ras signaling plays a central role in normal myelopoiesis and is deregulated by at least three distinct genetic mechanisms in myeloid leukemia (for review see references 20, 53, 54). Ligand-induced activation of many hematopoietic growth factor receptors transiently increases the Ras-GTP levels, which in turn influences the survival, proliferation, and differentiation of myeloid-lineage cells. In leukemic cells, acquired mutations of NRAS or KRAS, formation of a chimeric protein as a result of BCR-ABL translocations, and inactivation of NF1 all deregulate Ras signaling. Lines of knockout mice provide novel model systems for ascertaining how loss of function of specific genes influences complex biologic processes such as hematopoiesis and leukemogenesis. Although inactivation of the murine homologues of many human tumor suppressor genes results in a spectrum of neoplasms that differs significantly from the respective human cancers, Nf1 mice spontaneously develop a type of leukemia that closely resembles JCML and Nf1^−/− cells display a similar pattern of GM-CSF hypersensitivity in myeloid colony-forming assays.

A distinct advantage of this murine model is the ability to reconstitute irradiated mice only lacking Nf1 and isolate highly purified populations of phenotypically defined cells to characterize their in vitro colony growth and intracellular signaling.

This study addressed two questions: (a) what is the role of neurofibromin in modulating the cytokine responsiveness and growth of both primitive myeloid progenitors with multilineage and replating potential, and of more differentiated lineage-restricted myeloid progenitors?, and (b) how does inactivation of Nf1 affect constitutive and cytokine-stimulated intracellular signaling through the Ras-Raf-MAP kinase pathway in these primary hematopoietic cells? We used antibodies to purify immature hematopoietic cells and focused on the SCF/c-kit signaling pathway because the binding of SCF to c-kit increases Ras-GTP levels and activates Raf-1 and MAP kinase activity in myeloid cell lines (7). In addition, the presence of multiple hematopoietic defects in lines of mice with mutations affecting either c-kit (dominant white spotting, W) or SCF (Steel) (34–37, 39, 55, 56), as well as numerous in vitro studies showing a synergistic effect between SCF and other cytokines in promoting the survival and proliferation of primitive hematopoietic progenitor cells, indicate that the SCF/c-kit signaling plays an important role in early hematopoiesis (38–41).

An intriguing finding demonstrated here was that SCF alone or in combination with IL-3 induced a hypersensitive pattern of CFU-GM colony growth in Nf1^−/− cells. Though some reports using JCML cells have observed hypersensitive colony growth in response to only GM-CSF, others have inferred the involvement of other cytokines in the pathogenesis of the myeloproliferative disease (31–34, 57). Our results address an interesting paradox: since the GM-CSF and IL-3 receptors share a common β signaling chain that has been implicated in Ras activation, it has been perplexing why cultured Nf1^−/− cells are hypersensitive to GM-CSF but not IL-3 in progenitor colony-forming assays. One explanation that has been suggested previously is that p120 GAP, which is abundant and active in Nf1^−/− hematopoietic cells (27, 30), is able to downregulate Ras-GTP when it associates in a complex with the IL-3 receptor and other proteins, but not when Ras associates with the GM-CSF receptor (27, 30). Biochemical evidence that MAP kinase is hyperactivated in c-kit⁺ Nf1^−/− cells in response to IL-3, or SCF (Fig. 5), and cell culture data showing that SCF alone and in combination with IL-3 or GM-CSF induce a hypersensitive pattern of in vitro growth in Nf1^−/− cells (Fig. 2), suggest that aberrant intracellular signaling occurs in response to IL-3, GM-CSF, and SCF. However, the biochemical abnormality is not translated to hypersensitive colony formation in response to IL-3 only. It is possible that SCF and GM-CSF activate downstream signaling molecules important for terminal differentiation that are not induced by IL-3. For example, focal adhesion kinase is a recently identified kinase involved in the terminal differentiation of myeloid cells that is activated in response to GM-CSF (58) and SCF (59) but not IL-3. Nf1-deficient hematopoietic cells and cell lines should prove useful for further characterizing signaling pathways that are differentially activated and regulated in response to GM-CSF and IL-3. Our studies of purified Sca1⁺lin^−/dim cells suggest that the hypersensitive pattern of colony formation is intrinsic to the progenitors and is not dependent on accessory cells as suggested by some (31–33), but not all (21), previous studies in JCML cells.

Cultures established from primary d13.5 Nf1^−/− fetal liver cells also demonstrated an increase in the numbers of primitive progenitor (HPP-CFC) cells in the fetal liver. In fact the expansion in this compartment was greater than was noted for more differentiated progenitors (LPP-CFC) scored in simultaneous experiments (our unpublished results). As the HPP-CFC and LPP-CFC assays were performed in the presence of saturating concentrations of hematopoietic growth factors, our data indicate that inactivation of Nf1 is not only associated with a leftward shift in the dose–response curves for colony formation in response to exogenous growth factors but also with an absolute increase in the numbers of clonogenic progenitors in vivo. Preliminary data indicate that this increase in progenitors is associated with a relative resistance of Nf1^−/− cells to apoptosis upon withdrawal of hematopoietic growth factors (Zhang, Y.-Y., and D.W. Clapp, unpublished results).

Biochemical data from c-kit positive hematopoietic cells demonstrating constitutive activation of MAP kinase signaling and a prolonged hyperactivation of MAP kinase activity in response to stimulation with cytokines suggests that deregulation of the Ras/MAP kinase pathway accounts for the perturbations of in vitro colony growth. Indeed, reducing the MAP kinase activity with an inhibitor of MEK was associated with a profound, specific reduction in colony formation from Nf1^−/− c-kit⁺ cells. The elevated and sustained levels of MAP kinase that we measured in response to SCF, IL-3, or GM-CSF in highly progenitor-enriched primary cells are consistent with the previous observation of a sustained elevation of Ras-GTP levels after GM-CSF stimulation in Myb immortalized Nf1^−/− fetal hematopoietic cell lines (27). Similarly, primary bone marrow cells from children with NF1 and leukemia showed an elevated percentage of Ras-GTP (30). Although the biochemical data presented here, the predilection to myeloid leukemia seen in children with NF1 and in Nf1^+/− mice, and the myeloid disorder that invariably develops in the recipients of Nf1^−/− fetal liver cells provide compelling evidence that neurofibromin is essential to regulate Ras output in myeloid lineage cells, future studies are indicated to further delineate how neurofibromin modulates Ras in response to cytokines such as SCF. In particular, p120 GAP accounts for most of the in vivo and in vitro GTPase activating protein activity in bone marrow cells (30, 60, 61). It is possible that p120 GAP is sequestered away from growth factor receptor–Ras complexes in activated myeloid cells or is inactivated by a more direct mechanism. In this respect, the recent cloning of the gene encoding p62^Dok (Dok) is of interest as this protein is phosphorylated in leukemic cells that carry the BCR-ABL translocation and phosphorylated Dok selectively associates with p120 GAP through Src homology 2 (SH2) domains. Dok is rapidly phosphorylated in hematopoietic cells in response to SCF (62). If the binding of phosphorylated p62^Dok inhibits the function of p120 GAP, cells that have been stimulated with SCF may require neurofibromin to properly regulate Ras-GTP (neurofibromin does not contain SH2 domains [for review see references 4, 63, 64], and is unlikely to interact with Dok). Additional biochemical data are required to examine if SCF-induced phosphorylation of p62^Dok affects the function of p120 GAP in normal and Nf1-deficient myeloid lineage cells.

In summary, we have shown that the loss of Nf1 results in constitutive activation of Ras signaling in primitive and lineage-restricted hematopoietic progenitors. This is associated with hyperresponsiveness to multiple growth factors in biochemical and progenitor colony-forming assays, as well as in expansion of myeloid compartments in Nf1^−/−embryos and in adult recipients reconstituted with Nf1^−/− fetal liver cells. Taken together, these data strongly suggest that the size and growth factor sensitivity of cells within these multiple hematopoietic compartments are regulated by neurofibromin through its ability to effect the activation state of the Ras/MAP kinase cascade. These data have implications for the use of GM-CSF antagonists to treat JCML. On the one hand, our data showing that inactivation of Nf1 results in aberrant signaling in response to other growth factors provide evidence that GM-CSF–specific inhibitors might not show efficacy. However, mice that overexpress GM-CSF develop a JCML-like disorder (65), and in one recent study xenograft mice transplanted with JCML cells and treated with antagonists to GM-CSF had a significant reduction in JCML cells as compared with control mice treated with an isoimmune nonspecific antibody (66). Thus, it may be that the loss of Nf1 results in aberrant Ras activation in multiple myeloid compartments, the importance of GM-CSF on terminal differentiation may be pivotal to acquisition of the clinical disease, and it remains possible that the clinical disorder will be abrogated by blocking the GM-CSF receptor. Nf1 mice provide a valuable model system to test basic aspects of myeloid growth control and to evaluate novel therapeutics directed against hyperactive Ras.

We thank Patricia Fox for secretarial support.

This work was supported by Public Health Service grant P5ODK49218 R29 CA-74177-01 (to D.W. Clapp) and American Cancer Society grant 1874SC (to D.W. Clapp and K. Shannon). Tyler Jacks is an investigator of the Howard Hughes Medical Institute.

CFC

colony-forming cells

GAP

GTPase-activating protein

HPP

high proliferating potential

JCML

juvenile chronic myelogenous leukemia

LPP

low proliferating potential

MAP

mitogen-activated protein

MBP

myelin basic protein

MEK

MAP kinase/extracellular signal-regulated kinase

NF1

neurofibromatosis type 1

SCF

stem cell factor