Mutations in the NF1 tumor suppressor gene cause neurofibromatosis type I (NF1), a disease characterized by the formation of cutaneous neurofibromas infiltrated with a high density of degranulating mast cells. A hallmark of cell lines generated from NF1 patients or Nf1-deficient mice is their propensity to hyperproliferate. Neurofibromin, the protein encoded by NF1, negatively regulates p21ras activity by accelerating the conversion of Ras-GTP to Ras-GDP. However, identification of alterations in specific p21ras effector pathways that control proliferation in NF1-deficient cells is incomplete and critical for understanding disease pathogenesis. Recent studies have suggested that the proliferative effects of p21ras may depend on signaling outputs from the small Rho GTPases, Rac and Rho, but the physiologic importance of these interactions in an animal disease model has not been established. Using a genetic intercross between Nf1+/− and Rac2−/− mice, we now provide genetic evidence to support a biochemical model where hyperactivation of the extracellular signal–regulated kinase (ERK) via the hematopoietic-specific Rho GTPase, Rac2, directly contributes to the hyperproliferation of Nf1-deficient mast cells in vitro and in vivo. Further, we demonstrate that Rac2 functions as mediator of cross-talk between phosphoinositide 3-kinase (PI-3K) and the classical p21ras-Raf-Mek-ERK pathway to confer a distinct proliferative advantage to Nf1+/− mast cells. Thus, these studies identify Rac2 as a novel mediator of cross-talk between PI-3K and the p21ras-ERK pathway which functions to alter the cellular phenotype of a cell lineage involved in the pathologic complications of a common genetic disease.
Mutations in the NF1 tumor suppressor gene cause neurofibromatosis type I (NF1), a pandemic, autosomal dominant disorder with an incidence of 1 in 3,000. Neurofibromin, the protein encoded by NF1, functions as a GTPase-activating protein (GAP) for p21ras by accelerating the conversion of active p21ras-GTP to inactive p21ras-GDP 1,2. Cutaneous neurofibromas are slow-growing tumors that are pathognomonic for NF1, and mast cells have been implicated in the pathogenesis of neurofibroma formation as well as other cutaneous tumors 3,4,5,6,7. Recently, we have shown that Nf1+/− mice have increased numbers of peritoneal mast cells compared with wild-type littermates 8. In addition, bone marrow–derived mast cells (BMMCs) cultured from Nf1+/− mice demonstrate increased proliferation relative to wild-type cells in response to stem cell factor (SCF), a known mitogen for mast cells 8. However, although these studies show that alterations in p21ras activity can alter mast cell fates in vitro and in vivo, little is known about the contribution of specific p21ras effector pathways to the hyperproliferation of Nf1+/− mast cells or other Nf1-deficient cell types.
Although a critical role for p21ras in controlling cellular proliferation has been clearly established, recent studies have focused on identifying specific p21ras effector pathways important for cell cycle progression. Several studies have suggested that the proliferative effects of p21ras may depend on signaling inputs from the small Rho GTPases, Rac and Rho 9. Like p21ras, the small Rho GTPases exist in an inactive (GDP-bound) and active (GTP-bound) conformation, and interact with target proteins in their GTP-bound state 9. However, although these studies have implicated cross-talk between p21ras and the Rho GTPases in controlling cellular proliferation, the mechanisms underlying these interactions are still relatively poorly defined 9. In addition, the physiological importance of signaling mechanisms that link p21ras and the Rho GTPases and its contribution to disease pathogenesis have not been established.
Previous studies have emphasized that loss of neurofibromin results in hyperactivation of the classical p21ras-Raf-Mek-extracellular signal–regulated kinase (ERK) signaling pathway in Nf1-deficient cells leading to hyperproliferation 10,11. However, the contribution of Rho GTPases to this signaling cascade and to the Nf1 cellular phenotype have not been previously investigated in primary cells and in an animal disease model. Using Nf1+/− mice and BMMCs, we now provide pharmacologic, genetic, and biochemical data to support a model where hyperactivation of ERK mediated through the hematopoietic-specific Rho GTPase, Rac 2, directly contributes to the hyperproliferation of Nf1-deficient mast cells in vitro and in vivo. In addition, we demonstrate that ERK activation in response to SCF in mast cells is dependent on cross-cascade signals from phosphoinositide 3-kinase (PI-3K) converging on the p21ras-Raf-Mek-ERK pathway. Importantly, we identify Rac2 as a novel mediator of cross-talk between these two pathways and provide evidence that hyperactivation of Rac2 can cooperate with p21ras to alter the function of a cell lineage implicated in the pathogenesis of a common genetic disease.
Materials And Methods
Nf1+/− mice were obtained from Tyler Jacks at the Massachusetts Institute of Technology (Cambridge, MA) in a C57BL/6.129 background and backcrossed for 13 generations into the C57BL/6J strain. Rac2−/− mice were backcrossed into a C57BL/6J as described previously 12. These studies were conducted with a protocol approved by the Indiana University Laboratory Animal Research Center. The Nf1 and Rac2 allele were genotyped as described previously 12,13,14. Multiple F0 founders were used to generate the four Nf1 and Rac2 genotypes used in these experiments as outlined. F0: Nf1+/−; +/+ × +/+; Rac2−/−. F1: Nf1+/−; Rac2+/− × +/+; Rac2+/−. F2: Nf1+/−; Rac2−/−, +/+; Rac2−/−, Nf1+/−; +/+, +/+; +/+.
Bone Marrow Mast Cell Culture and Proliferation Assays.
Six to nine BMMCs from each murine genotype were generated and used for proliferation and biochemistry assays. All experiments were conducted using at least three lines from each genotype. BMMCs were cultured as described previously 15 with minor modifications and homogeneity of BMMCs was determined by Giemsa staining. Aliquots of cells were also stained with alcian blue and safranin to confirm that they were mast cells. Furthermore, FACS® analysis revealed similar forward and side light scatter characteristics and the same percentage of c-kit+ expression in BMMCs of all murine genotypes (data not shown). Proliferation assays were performed as described previously 8. BMMCs from each genotype were deprived of growth factors for 24 h, and 3 × 105 cells were plated in 24-well dishes in 1 ml RPMI containing 1% glutamine (BioWhittaker), 1.5% HEPES (BioWhittaker), 2% penicillin/streptomycin (BioWhittaker), and 25–100 ng/ml recombinant murine SCF (rmSCF; PeproTech) or no growth factors as indicated in a 37°C, 5% CO2, humidified incubator. In some experiments, LY294002 (Sigma-Aldrich), PD98059 (New England Biolabs, Inc.), or DMSO (Sigma-Aldrich) was added to the cultures 15 min to 1 h before addition of rmSCF. After 72 h, cells were counted using a hemocytometer. Cell viability was determined by a trypan blue exclusion assay. Assays were performed in triplicate.
For thymidine incorporation assays, BMMCs were deprived of growth factors for 24 h and 105 cells were plated in 96-well dishes in 200 ml of RPMI containing 1% glutamine, 10% fetal bovine serum (Hyclone Laboratories), and 25–100 ng/ml of rmSCF or no growth factors, as indicated, in a 37°C, 5% CO2, humidified incubator. Cells were cultured for 48 h, and tritiated thymidine (New Life Science Products, Inc.) was added to cultures 6 h before harvest. In some experiments, LY294002, PD98059, or DMSO was added to the cultures 15 min to 1 h before addition of rmSCF. Cells were harvested on glass fiber filters (Packard Instrument Co.) and β emission was measured. Assays were performed in triplicate.
Raf-1, Mek, ERK, Pak1, and Akt In Vitro Kinase Assays.
Activation of Raf-1, Mek, ERK, Pak1, and Akt kinases was determined by depriving BMMCs of serum and growth factors for 18–24 h and followed by stimulation with 10–100 ng/ml rmSCF for various amounts of time. Cells were washed twice with PBS containing 1 mM sodium orthovanadate and lysed in nonionic lysis buffer (20 mM Tris/HCl, 137 mM NaCl, 1 mM EGTA, 1% Triton X-100, 10% glycerol, 1.5 mM MgCl2, and complete protease inhibitors (Amersham Pharmacia Biotech), as described previously 16. The protein lysates were equalized for protein concentration using the bicinchoninic acid (BCA) assay (Pierce Chemical Co.), and equal loading of protein in these assays was confirmed by Western blot. ERK and Pak1 kinase immunoprecipitations were carried out with an anti-ERK-2 (C-14) antibody (Santa Cruz Biotechnology, Inc.) and an anti-Pak1(C-19) antibody (Santa Cruz Biotechnology, Inc.), respectively. Raf-1 and Mek kinase immunoprecipitations were carried out with an anti-Raf-1 (C-12) antibody (Santa Cruz Biotechnology, Inc.) and an anti-Mek-1 (C-18) antibody (Santa Cruz Biotechnology, Inc.), respectively. Akt kinase immunoprecipitations were carried out with an anti-Akt antibody (Cell Signaling Technology). The kinase immune complex assay was performed as described previously 16 and the following proteins were used as phosphorylation substrates: Elk-1 fusion protein (New England Biolabs, Inc.) for ERK kinase, histone 2B (Roche) for Akt kinase, myelin basic protein (Sigma-Aldrich) for Pak1 kinase, and recombinant Mek1 fused with GST at the NH2 terminus (Upstate Biotechnology) for Raf-1 kinase. For Mek, Raf-1, and ERK, kinase reactions were resolved on 10% Tris-glycine gels (Novex). For Pak1 and Akt, kinase reactions were resolved on 4–20% gradient Tris-glycine gels (Novex), and transferred to polyvinylidine difluoride (PVDF) filters. Gels and PVDF filters were dried and subjected to autoradiography. Densitometry of individual bands was conducted using NIH Image software.
Ras and Rac Activation Assays.
BMMCs were deprived of serum and growth factors for 18–24 h and stimulated at various amounts of time with 10–100 ng/ml rmSCF. Ras and Rac activation was subsequently determined using Ras and Rac activation assay kits (Upstate Biotechnology) according to the manufacturer's protocol and as described previously 17.
Immunoprecipitation and Western Blotting.
BMMCs were lysed in nonionic lysis buffer as described previously 16. Cleared lysates were normalized for protein content using the BCA assay (Pierce Chemical Co.) and proteins were immunoprecipitated with protein A sepharose beads (Amersham Pharmacia Biotech) coupled with polyclonal antibodies for Raf-1 and Mek (Santa Cruz Biotechnology, Inc.). Immunoprecipitates were washed three times in lysis buffer, resuspended in sample buffer, boiled for 5 min, subjected to SDS-PAGE, and transferred to nitrocellulose. The membranes were blocked in TBS-Tween containing 5% BSA overnight. Membranes were incubated for 1–3 h at room temperature with the following antibodies: anti-phospho-Mek 1/2 (1:1,000; New England Biolabs, Inc.), anti-phospho-Raf-1 338 (1:5,000; Upstate Biotechnology), anti-phospho-Mek 298 (gift from Dr. M. Marshall), and anti-p21ras (Upstate Biotechnology). Secondary antibodies used were either anti–rat, anti–rabbit, or anti–mouse Ig horseradish peroxidase conjugated (Transduction Laboratories). Proteins were visualized by enhanced chemiluminescence (Santa Cruz Biotechnology, Inc.).
Studies In Vivo.
Adult Nf1+/−, +/+Rac2−/−, Nf1+/−Rac2−/−, and wild-type mice received a continuous infusion of various doses of rmSCF or vehicle (PBS) from microosmotic pumps (Alzet) placed under the dorsal back skin. Osmotic pumps were surgically placed under light avertin anesthesia. rmSCF or vehicle was released over 7 d at a rate of 0.5 μl/hour, and osmotic pumps were surgically removed on day 7 after sacrifice by cervical dislocation. To accurately identify cutaneous sections for quantitating changes in mast cell numbers in response to rmSCF, the dorsal skin was stained with a drop of India ink at the point of exit of rmSCF from the osmotic pump before removal of the pump. 3-cm sections of skin marked with India ink were removed, fixed in buffered formalin, and processed in paraffin-embedded sections. Specimens were stained with hematoxylin and eosin to assess routine histology and with Giemsa to identify mast cells. Cutaneous mast cells were quantitated in a blinded fashion by counting 2-mm2 sections in proximity to the India ink stain.
For studies evaluating basal numbers of mast cells in the Nf1 and Rac intercrossed genotypes, 1-cm sections of ears were removed, fixed in buffered formalin, and processed in paraffin-embedded sections. Specimens were stained with hematoxylin and eosin to assess routine histology and Giemsa to identify mast cells. Cutaneous mast cells from four to five mice per experimental group were quantitated.
Nf1+/− Mast Cells Have Increased p21ras Activity and Hyperactivation of Akt after Stimulation with SCF.
SCF binding to its receptor, c-kit, causes a rapid increase in p21ras activity in primary BMMCs, and neurofibromin functions as a GAP for p21ras 1,2,15. However, it remains unclear whether heterozygosity of Nf1 alters p21ras activity in primary cells. To investigate whether heterozygosity at the Nf1 allele alters p21ras activity in mast cells, Nf1+/− and wild-type BMMCs were stimulated with SCF and assayed for changes in active p21ras-GTP levels. After stimulation of BMMCs, levels of GTP-bound p21ras in cellular lysates were determined by precipitating the active GTPase with a GST fusion of the p21ras binding domain of Raf-1 kinase in an effector pull-down assay. Even though Nf1+/− mast cells had detectable, but reduced, levels of neurofibromin as determined by Western blot (data not shown), Nf1+/− mast cells had higher basal and SCF-stimulated p21ras-GTP levels compared with wild-type cells (Fig. 1 A). Importantly, preincubation of both mast cell genotypes with PI-3K inhibitors, wortmannin or LY294002, did not alter p21ras activity (data not shown). Thus, these data position p21ras either in parallel or upstream of PI-3K in mast cells and demonstrate that heterozygosity at Nf1 alters p21ras activity in primary BMMCs.
Direct binding of the activated c-kit receptor tyrosine kinase to the SH2 domain of the p85α regulatory subunit of type IA PI-3K is important for kinase activation after stimulation by its ligand, SCF 15. Importantly, in immortalized cell lines p21ras-GTP has previously been shown to bind to specific p110 catalytic subunits of type IA PI-3K to augment kinase activity 18,19,20. To determine the effect of increased p21ras-GTP levels on PI-3K activity in Nf1+/− mast cells, we compared the activation of the serine/threonine protein kinase B (PKB), also known as Akt, in Nf1+/− and wild-type BMMCs after stimulation with SCF. Activation of Akt is PI-3K dependent and provides a sensitive measure for PI-3K activity 21. Nf1+/− BMMCs demonstrated greater basal and SCF-stimulated Akt activity compared with wild-type cells over a range of SCF concentrations from 10 ng/ml (Fig. 1 B) to 100 ng/ml (data not shown). In addition, inhibitors of PI-3K inhibited Akt activity in both mast cell genotypes (Fig. 1 B), confirming that p21ras contribution to Akt activation is dependent on activation of PI-3K. Taken together, these data demonstrate that increased p21ras activity in Nf1+/− BMMCs after SCF stimulation is biochemically linked to increased activation of the PI-3K-Akt kinase pathway.
Hyperproliferation of Nf1+/− Mast Cells in Response to SCF Is Mediated through Cross-Cascade Activation of ERK from PI-3K.
Proliferation of wild-type BMMCs in response to SCF is dependent on PI-3K activation 15. Although we have previously shown that Nf1+/− mast cells have increased proliferation in response to SCF, identification of p21ras effector pathways responsible for this phenotype remain unclear. To test whether SCF-induced hyperproliferation of Nf1+/− BMMCs is mediated through hyperactivation of PI-3K, we performed proliferation assays with Nf1+/− and wild-type BMMCs in the presence or absence of the PI-3K inhibitor, LY294002. Nf1+/− and wild-type BMMCs were cultured in 10% fetal calf serum without growth factors for 24 h before the addition of SCF. Cell numbers were determined at the time SCF was added (day 0) and after 72 h in culture. Nf1+/− BMMCs showed a greater proliferative response to SCF compared with wild-type cells, as observed previously 8, and the proliferation of both mast cell genotypes was completely inhibited in the presence of LY294002 (Fig. 2 A). Similar results were obtained when mast cell proliferation was assessed using thymidine incorporation assays to measure DNA synthesis (data not shown). In addition, consistent differences in proliferation were observed between the genotypes over a range of SCF concentrations (25–100 ng/ml, data not shown). Thus, increased SCF-induced proliferation of Nf1+/− mast cells is mediated through hyperactivation of the p21ras-PI-3K pathway.
While hyperactivation of the p21ras-PI-3K signaling pathway provides a simple model to account for the phenotype of Nf1+/− BMMCs, we next determined which specific downstream effectors of PI-3K were responsible for the hyperproliferation of Nf1+/− mast cells. Recent studies in immortalized cell lines have shown that signals from PI-3K converge on the p21ras-Raf-Mek-ERK pathway to activate ERK and promote transcription of genes that control mitogenesis 22,23,24,25. In addition, we have previously demonstrated that Nf1+/− BMMCs have elevated ERK activity relative to wild-type cells after stimulation with SCF which is consistent with hyperactivation of p21ras 8. To test whether ERK activity contributes to either wild-type or Nf1+/− BMMC proliferation, we performed proliferation assays as described previously with both mast cell genotypes stimulated with SCF or medium alone, in the presence or absence of the Mek inhibitor, PD98059. The Mek inhibitor completely blocked ERK activation in both mast cell genotypes after stimulation with SCF (data not shown). In multiple independent experiments (n = 6), addition of PD98059 reduced the proliferative response of wild-type BMMCs by ∼10–15% and that of Nf1+/− BMMCs by 40–50% (Fig. 2 A). Importantly, whereas PD98059 caused a minimal reduction in wild-type mast cell proliferation, addition of PD98059 reduced the SCF-induced hyperproliferation of Nf1+/− mast cells to that of wild-type levels, establishing the importance of hyperactivation of ERK to the Nf1+/− proliferative phenotype (Fig. 2 A). Similar results were obtained when mast cell proliferation was assessed using thymidine incorporation assays (data not shown). Thus, these pharmacologic data suggest that hyperactivation of ERK confers a distinct proliferative advantage to Nf1+/− BMMCs in response to SCF.
Given that PI-3K inhibitors significantly block the proliferation of both mast cell genotypes and Mek inhibitors partially inhibit proliferation, we hypothesized that PI-3K is biochemically linked to ERK activation in BMMCs. To test this hypothesis, we stimulated both mast cell genotypes with SCF in the presence or absence of PI-3K inhibitors and measured ERK activity. Similar to previous studies 8, Nf1+/− BMMCs demonstrated a greater increase in ERK activity from baseline compared with wild-type cells (Fig. 2 B). However, ERK activity was completely inhibited in both genotypes in the presence of PI-3K inhibitors (Fig. 2 B). Thus, these data demonstrate that ERK activation in mast cells, in response to SCF, is dependent on biochemical signals generated through PI-3K converging on the p21ras-Raf-Mek-ERK pathway.
Cross-Cascade Activation of ERK from PI-3K Occurs at the Level of Mek and Raf-1 Kinase in Nf1+/− and Wild-Type Mast Cells.
In other experimental systems, cross-cascade activation of ERK from PI-3K occurs at the level of either Mek or Raf-1 kinase 22,23,24,25,26. To test whether cross-talk between PI-3K and the p21ras-Raf-Mek-ERK pathway occurs at the level of Mek, we stimulated Nf1+/− and wild-type BMMCs with SCF in the presence or absence of a PI-3K inhibitor and measured Mek activity. Nf1+/− BMMCs demonstrated a greater increase in Mek activity from baseline compared with wild-type cells, and Mek activity was inhibited in the presence of PI-3K inhibitors in both mast cell genotypes (Fig. 3 A).
Since Raf-1 functions as the immediate upstream effector of Mek by phosphorylating Mek Ser217/221 27, we stimulated Nf1+/− and wild-type mast cells with SCF in the presence or absence of a PI-3K inhibitor and measured Raf-1 kinase activity and phosphorylation of these residues by immunoblot. Whereas wild-type BMMCs demonstrated minimal Raf-1 kinase activation after stimulation with 10 ng/ml of SCF, Nf1+/− BMMCs had a significant increase in Raf-1 kinase activity from baseline in multiple independent experiments (n = 5) which was completely inhibited in the presence of the PI-3K inhibitor (Fig. 3 B). In addition, Nf1+/− BMMCs had increased phosphorylation of Mek Ser217/221 which correlates with increased Raf-1 activity in these cells (Fig. 3 C). Importantly, phosphorylation of these residues was also inhibited by a PI-3K inhibitor in both mast cell genotypes (Fig. 3 C).
Taken together, as PI-3K inhibitors do not alter p21ras activity (data not shown), these data suggest a biochemical model where PI-3K activation of ERK occurs at the level of either Raf-1 alone or at the level of both Raf-1 and Mek in BMMCs after stimulation with SCF. Importantly, increased Raf-1 and Mek activity observed in Nf1+/− mast cells is consistent with the hypothesis that the hyperproliferative phenotype of Nf1+/− BMMCs is directly related to increased cross-talk from PI-3K to the p21ras-Raf-Mek-ERK pathway.
Heterozygosity of Nf1 Results in Higher Basal and SCF-stimulated Rac and Pak1 Activity in BMMCs.
Given the importance of cross-talk between the PI-3K and the p21ras-Raf-Mek-ERK pathways to the hyperproliferation of Nf1+/− BMMCs, we performed experiments to identify proteins responsible for cross-talk between the two pathways. Studies in hematopoietic cells have positioned the small Rho GTPase, Rac, downstream of PI-3K, and some experiments in immortalized cell lines have suggested that Rac1 may function as a mediator of cross-talk between PI-3K and the p21ras-Raf-Mek-ERK pathways 23,28,29,30,31,32. Importantly, Rac1 and Rac2 activation have also been linked to mast cell proliferation but their contribution to ERK activation in mast cells has not been studied 33,34. Given these prior experiments and our current observations, we measured total Rac activity in both mast cell genotypes in response to SCF. After stimulation of BMMCs, levels of active, GTP-bound Rac in cellular lysates were determined by precipitating the active GTPase with a GST fusion of p21 activated kinase in an effector pull-down assay. Nf1+/− mast cells had higher basal and SCF-stimulated Rac-GTP levels compared with wild-type cells, providing evidence that p21ras is biochemically linked to Rac activation. (Fig. 4 A).
Pak1 is a well characterized downstream effector of Rac, and the catalytic activity of Pak1 is regulated in part by binding of Rac-GTP to the highly conserved NH2 terminus, known as the Cdc42/Rac interactive binding (CRIB) domain 35,36. Importantly, Pak1 has been shown to mediate cross-talk between PI-3K and the p21ras-Raf-Mek-ERK pathway in immortalized cell lines by phosphorylating specific Raf-1 and Mek residues 22,25. To test whether heterozygosity of Nf1 also altered Pak1 activity in BMMCs, we measured Pak1 activity in both mast cell genotypes in response to SCF. Whereas wild-type mast cells displayed minimal Pak1 activation, Nf1+/− BMMCs had dramatically increased basal Pak1 activity which increased modestly after stimulation with SCF, which is consistent with increased Rac activity (Fig. 4 B). Thus, these studies suggest that hyperactivation of the Rac and potentially the Pak1 pathway confer a distinct proliferative advantage to Nf1+/− BMMCs.
Hyperactivation of Rac2 Directly Contributes to the Hyperproliferation and Increased ERK Activity in Nf1+/− BMMCs in Response to SCF.
Rac2 is a hematopoietic specific Rho GTPase, and Rac2−/− mice have multiple mast cell defects secondary to diminished Rac activity 12,34. Importantly, other studies have recently positioned different isoforms of Rac downstream of PI3-K in hematopoietic cells (for a review, see reference 37). To test our hypothesis that hyperactivation of Rac2 pathway is directly responsible for the hyperproliferation of Nf1+/− BMMCs and to potentially identify which Rac protein may be involved in this signaling pathway, we intercrossed Rac2−/− and Nf1+/− mice, established BMMCs from the four F2 progeny (see Materials and Methods), and performed proliferation, Pak1, and ERK activity assays. To initially determine whether Pak1 was located downstream of Rac2 in Nf1+/− and wild-type mast cells, BMMCs harvested from the mice of the four F2 genotypes were stimulated with SCF and Pak1 activity was measured. Consistent with our initial observations, Nf1+/− mast cells had higher basal and SCF induced Pak1 activity compared with wild-type cells (Fig. 5 A). However, Nf1+/+ Rac2−/− and Nf1+/−Rac2−/− BMMCs had minimal Pak1 activation at baseline and in response to SCF (Fig. 5 A). Thus, these studies identify Pak1 as a downstream effector of Rac2 in both mast cell genotypes.
To assess the role of Rac2 activation on the proliferation of Nf1+/− mast cells, BMMCs harvested from mice of the four F2 genotypes were cultured in 10% fetal calf serum without added growth factors for 24 h before addition of SCF. Cell numbers were determined at the time SCF was added (day 0) and after 72 h in culture. As shown previously 8,34, Nf1+/−Rac2+/+ BMMCs showed a greater proliferative response to SCF after 72 h in culture compared with wild-type cells, whereas Nf1+/+Rac2−/− cells showed a slight decrease in proliferation (Fig. 5 B). However, Nf1+/−Rac2−/− BMMCs had a 40–50% reduction in proliferation compared with Nf1+/−Rac2+/+ BMMCs in response to SCF (Fig. 5 B). Importantly, the proliferative responses of Nf1+/−Rac2−/− BMMCs were similar to wild-type cells in multiple independent experiments (n = 6; Fig. 5 B). Thus, these experiments provide direct genetic evidence that the hyperproliferative phenotype of Nf1+/− Rac2+/+ mast cells is mediated through activation of the specific Rac2 isoform.
The reduction in proliferation of Nf1+/−Rac2−/− BMMCs relative to Nf1+/−Rac2+/+ cells was similar to the reduction in proliferation observed with the addition of a Mek inhibitor (and subsequent inhibition of ERK activity) to the cultures of Nf1+/−Rac2+/+ mast cells. Given the similar results of these two experiments, we hypothesized that Rac2 was biochemically linked to ERK activation after stimulation of mast cells with SCF. To test this hypothesis, we compared ERK activity in the four F2 mast cell genotypes after stimulation with SCF. Consistent with prior observations, Nf1+/−Rac2+/+ mast cells had higher ERK activity compared with wild-type cells (Fig. 5 C). However, Nf1+/+Rac2−/− BMMCs had a 50% reduction in ERK activity compared with wild-type cells, providing genetic evidence that Rac2 is directly linked to ERK activation (Fig. 5 C). Consistent with the hypothesized role of hyperactivation of ERK to the proliferative phenotype of Nf1+/−Rac2+/+ mast cells, Nf1+/−Rac2−/− mast cells had equivalent levels of ERK activity compared with wild-type cells in response to SCF (Fig. 5 C). Thus, these data provide further genetic confirmation that Rac2-mediated hyperactivation of ERK directly contributes to the hyperproliferation of Nf1+/− mast cells in response to SCF.
Increased Accumulation of Mast Cells in Nf1+/− Mice in Response to Local Cutaneous Administration of SCF Is Mediated through Rac2 In Vivo.
Local cutaneous injection of SCF into wild-type mice results in a dramatic accumulation of mast cells at the site of injection 38,39. Using 5-bromo-2′-deoxyuridine (Brdu) labeling in vivo, Tsai et al. demonstrated that accumulation of mast cells in this experimental model was closely linked to local cellular proliferation 39. To test whether heterozygosity at Nf1 alters accumulation of mast cells to SCF in vivo, slow release microosmotic pumps were filled with a range of concentrations of SCF or vehicle and placed into the dorsal back skin of either wild-type or Nf1+/− mice. 7 d after placement, pumps were removed and cutaneous skin biopsies were taken from the site of pump insertion and stained with Giemsa stain to identify mast cells (Fig. 6a and Fig. b). Consistent with our in vitro experiments, Nf1+/− mice demonstrated greater accumulation of mast cells at the site of SCF release compared with wild-type mice over a wide range of SCF doses (Fig. 6 C). Interestingly, Nf1+/− mice also had a greater percentage of degranulating mast cells at this dose (28 vs. 8%) compared with wild-type mice at the site of SCF release.
To investigate whether the increased accumulation of mast cells in Nf1+/− mice was mediated through Rac2, we repeated this experiment using the F2 progeny generated from the Rac2−/− and Nf1+/− genetic intercross. The SCF dose chosen for these experiments was 2 μg/kg/day, as the greatest difference in mast cell accumulation between wild-type and Nf1+/− mice was observed at this dose (Fig. 6 C). Consistent with our in vitro proliferation results, the accumulation of mast cells in Nf1+/−Rac2−/− mice was greatly reduced compared with Nf1+/−Rac2+/+ mice (Fig. 6 D). Importantly, no differences in mast cell accumulation were observed between wild-type and Nf1+/−Rac2−/− mice (Fig. 6 D). To determine whether loss of Rac2 alters endogenous cutaneous mast cell numbers in Nf1+/− mice, histologic sections from the ears of the 4 F2 genotypes were evaluated after giemsa staining (n = 4–5 mice/group). Cutaneous sections from the ear have been found in previous studies to be a sensitive site to detect genotypic differences in mast cell numbers 40. Interestingly, mice that were doubly mutant at Nf1 and Rac2 had significantly lower numbers of mast cells in the ear compared with mice that were heterozygous at Nf1 alone (Table). Thus, cumulatively these studies demonstrate that the biochemical mechanisms identified in vitro for the Nf1+/−Rac2+/+ mast cell phenotype are operative in a more physiologic experimental system.
Although p21ras signaling outputs were originally described as operating in parallel with activation of PI-3K pathways, recent studies have suggested that receptor tyrosine kinase–mediated activation of ERK is potentially PI-3K dependent 9,37. However, the majority of these studies were conducted in immortalized cell lines, and a clear example of the physiolgic relevance of cross-talk between the p21ras-ERK and PI-3K pathways in an animal disease model has not been established previously. Multiple lineages of neurofibromin-deficient cells have a propensity to hyperproliferate in response to various growth factors which directly relates to the wide diversity of clinical manifestations observed in NF1 41. Although neurofibromin has been shown to function as a GAP for p21ras to control cellular proliferation, lack of identification of alterations in specific p21ras effectors which control mitogenesis has limited our understanding of disease pathogenesis. In these experiments, we show that the hyperproliferation of Nf1+/− mast cells is directly related to increased cross-talk between the PI-3K and the p21ras-Raf–ERK pathway and identify Rac2 as a principal mediator of this aberrant signaling pathway (Fig. 7).
Numerous studies have shown that active p21ras-GTP can bind to multiple isoforms of the p110 catalytic subunits of type IA PI-3K to specifically augment kinase activity 18,19,20. However, in neurofibromin-deficient cells, most attention has been focused on modulation of the classical p21ras-Raf-Mek-ERK pathway to augment cellular proliferation. In primary mast cells, activation of PI-3K after c-kit binding to SCF is critical for controlling proliferation 15,33. Our study shows that Nf1+/− mast cells have a twofold increase in proliferation in response to SCF compared with wild-type cells, and proliferation was completely reduced with PI-3K inhibitors, confirming the importance of PI-3K activation to the Nf1+/− cellular phenotype. Importantly, placement of cutaneous microosmotic pumps which released a continuous infusion of SCF resulted in a greater accumulation of mast cells at the site of insertion in Nf1+/− mice compared with wild-type mice. Others have previously shown that in vivo mast cell expansion in response to local injection of SCF occurs secondary to local proliferation 38,39. Thus, our in vivo observations argue that the biochemical mechanisms identified in vitro for the Nf1+/− cell phenotype are valid and biologically operative in a more physiologic system. As Nf1+/− mast cells have increased p21ras activity in response to SCF and a PI-3K inhibitor completely blocks proliferation, these cell biology studies argue that hyperactivation of the p21ras-PI-3K pathway in Nf1+/− mast cells altered mast cell proliferation in vivo and in vitro.
While hyperactivation of PI-3K provides a simple model for the increased proliferation of Nf1+/− mast cells, we provide several lines of pharmacologic, genetic, and biochemical data to demonstrate that hyperactivation of the Rac2-ERK signaling pathway is directly responsible for the Nf1+/− cellular phenotype. First, we have previously reported that Nf1+/− mast cells display increased ERK activation compared with wild-type mast cells after stimulation with SCF, which is consistent with hyperactivation of p21ras 8. However, we unexpectedly found that activation of ERK is dependent on signals from PI-3K and that hyperactivation of ERK was directly linked to the proliferative phenotype of Nf1+/− mast cells. PI-3K inhibitors completely blocked the activation of ERK in both mast cell genotypes. The physiological relevance of cross-talk between PI-3K and the p21ras-Raf-Mek-ERK pathway in Nf1+/− cells is supported by the effect of a specific Mek inhibitor on the SCF-stimulated proliferation of these cells. A Mek inhibitor reduces the proliferation of wild-type cells modestly (10–15%) in our studies. However, preincubation of Nf1+/− mast cells with a Mek inhibitor reduces proliferation much more significantly (40–50%), and importantly, diminishes the proliferative response to wild-type levels. Thus, in contrast to wild-type cells, hyperactivation of ERK from PI-3K provides a unique signal to confer a distinct proliferative advantage to Nf1+/− cells.
One interesting observation from these studies is that constitutive activation of multiple upstream effectors of ERK including p21ras and Rac does not consistently lead to constitutive activation of ERK. One potential explanation for this observation is that mitogen-activated protein kinase phosphatase (MKP), a phosphatase that specifically binds and dephosphorylates ERK, may be elevated in Nf1+/− cells 42,43. Recently, Brondello et al. have argued that activation of ERK regulates MKP-1 protein expression through both an increase in the rate of transcription and in the rate of proteasome-mediated degradation of MKP-1 42,43. Further, these authors argue that upregulation of MKP-1 in response to activation of ERK may provide an important negative feedback mechanism to limit inappropriate or constitutive activation of ERK. Given the importance of ERK activity in regulating proliferation and potentially other functions in Nf1+/− mast cells, investigation of the basal activity of this phosphatase could be important in future studies. In addition, the regulation of ERK activity in vivo is likely to be far more complex. It also remains possible that constitutive activation of upstream effectors of ERK may circumvent normal feedback mechanisms and provide a subthreshold level for activation of signaling pathways necessary for proliferation. In this context, it is intriguing to note that administration of low dosages (2 μg/kg/day) of exogenous recombinant SCF to Nf1+/− mice results in profound increases in mast cell numbers while having no significant effect on wild-type controls.
Next, given the importance of the hyperactivation of ERK to the proliferative phenotype of Nf1+/− mast cells, we focused on identifying potential mediators of cross-talk between the PI-3K and p21ras-Raf-Mek-ERK pathways. Recent studies have suggested that the proliferative effects of p21ras are modulated at least in part through signaling inputs from the small Rho GTPase, Rac 9. p21-activated kinase1 (Pak1) has emerged as a molecule that can link p21ras and Rac signaling by converging on the p21ras-Raf-Mek-ERK pathway 22,23,25. ERK activation has been linked to the induction of cyclin D1 expression and progression through G1 of the cell cycle 44,45. Using a genetic intercross, we show that disruption of the hematopoietic-specific Rho GTPase, Rac2, in Nf1+/− mast cells restores the proliferative response and ERK activation of these cells to wild-type levels. Prior studies have shown that Rac1 activation of c-Jun NH2-terminal kinase (JNK) is critical for mast cell proliferation 33. However, these studies used Rac1 dominant negative constructs to transduce primary-derived mast cells, and recent data suggest that dominant negative Rac1 proteins can potentially inhibit different Rac proteins including Rac2 46. Our genetic data demonstrate that hyperactivation of the specific Rac2 protein alters the proliferation of Nf1+/− mast cells, and provides the first evidence in primary cells that p21ras and Rac2 are biochemically linked. Further, while hyperactivation of Rac2 may enhance cyclin D1 expression through increased ERK activation to promote proliferation, it is possible that Rac2 could also increase cyclin D1 accumulation through an ERK-independent mechanism. Alternatively, it is also possible that hyperactivation of the Rac2-ERK pathways promotes proliferation through a significantly different mechanism such as the transcription of genes which control autocrine production of growth factors. Though our studies cannot definitively discriminate between these possibilities, it is clear that hyperactivation of p21ras and Rac2 cooperate to alter the proliferation of neurofibromin-deficient mast cells in vitro and in vivo. Studies in Nf1-deficient cells are currently underway to evaluate how p21ras and Rac2 coordinate signals necessary for controlling cell cycle entry and progression and for measuring release of autocrine growth factors.
Using immortalized cell lines, Pak1 has been shown to increase signaling through ERK activation by phosphorylating specific residues on Raf-1 and Mek 22,24,25. Specifically, activated Pak1 can phosphorylate Mek1 at Ser298 to enhance the stable association between Raf-1 and Mek1 22, and Pak1 can also phosphorylate Ser338 in the Raf-1 catalytic domain to enhance Raf-1 kinase activity 25. In these prior studies the net effect of increased activation of Mek and Raf-1 by Pak1 was increased ERK activation in response to different stimuli. Our data suggest that similar pathways are hyperactive in Nf1-deficient mast cells and function to increase ERK activity and proliferation in these cells. Pak1 activity and phosphorylation levels of Ser338 on Raf-1 and Ser298 on Mek1 (data not shown) in response to SCF stimulation were greatly increased in Nf1+/− mast cells compared with wild-type cells. Though our studies cannot eliminate the possibility that other effectors may contribute to cross-talk between PI-3K and the p21ras-Raf-Mek-ERK pathway, our data strongly suggest that hyperactivation of the Rac2 pathway directly accounts for the hyperproliferation of mast cells (Fig. 7). The elevated levels of Pak1 activity, as well as studies where expression of a dominant negative Pak1 cDNA reduced ERK activity in Nf1+/− mast cells to wild-type levels (unpublished results), suggest that Pak1 may serve as the effector molecule of Rac2 to enhance ERK activity in Nf1+/− mast cells. These findings are especially intriguing, as placement of a dominant negative Pak1 or a dominant negative Rac1 has been shown to reverse the transformation of a neurofibrosarcoma cell line taken from an individual with NF1 28. However, as dominant negative constructs may potentially bind other upstream effectors of Pak1, future studies using Pak1 knockout mice will allow direct testing of this hypothesis. Given that different isoforms of Rac and Pak alter the biology of different neurofibromin-deficient cell lineages, further characterization of these pathways in Nf1+/− and Nf1−/− primary cells could provide insight into the early events leading to loss of heterozygosity and the subsequent malignant manifestations in NF1.
In summary, recent studies have emphasized an increasing appreciation of the complexity of interactions between signaling pathways which control specific cellular responses. Our data provide a clear example in an animal disease model of how aberrant signaling between two major pathways can specifically alter the phenotype of a cell lineage implicated in the pathogenesis of a common genetic disease. Further characterization of this pathway and identification of similar signaling mechanisms in other neurofibromin-deficient cells should provide biochemical insight into the complex pathogenesis of NF1.
We thank Dr. Tyler Jacks for generously providing us with Nf1 heterozygous mice. We also thank Dr. Kevin Shannon for numerous helpful scientific discussions throughout the course of these experiments and for discussions during preparation of this manuscript. Dr. David Gutmann kindly provided antisera to neurofibromin and also conducted Western blots of Nf1+/− mast cells. We thank Patricia Fox and Marsha Hippensteel for their help with manuscript preparation.
This work was supported by the National Institutes of Health grant R29 CA74177 (D.W. Clapp), National Institutes of Health grant R01 DK48605 (D.A. Williams), and by the National Institutes of Health Pediatric Scientist Development Program Grant K12-HD0050 (D.A. Ingram).
A. King's current address is SmithKline Beecham Pharmaceuticals, P.O. Box 1539, King of Prussia, PA 19406.
Abbreviations used in this paper: BMMC, bone marrow–derived mast cell; ERK, extracellular signal–regulated kinase; GAP, GTPase-activating protein; MKP, mitogen-activated protein kinase phosphatase; NF1, neurofibromatosis type I; PI-3K, phosphoinositide 3-kinase; rmSCF, recombinant murine SCF; SCF, stem cell factor.