Increased c-Jun-NH₂-Kinase Signaling in Neurofibromatosis-1 Heterozygous Microglia Drives Microglia Activation and Promotes Optic Glioma Proliferation

The tumor microenvironment is considered to play an important role in tumor formation and progression by providing both negative and positive signals that influence tumor growth (1). In this regard, brain tumor formation and growth in children, similar to normal brain development, may also be dictated in part by the presence of spatially and developmentally regulated cues from the surrounding stroma. Using Neurofibromatosis type 1 (NF1) as a model system to study the contribution of the tumor microenvironment to cancer formation and growth, we and others have shown that benign nervous system tumor formation in Nf1 genetically engineered mice (GEM) also requires signals derived from the microenvironment. Elegant studies by Parada and colleagues first showed that plexiform neurofibroma (peripheral nerve sheath tumor) development in Nf1 GEM required both Nf1 inactivation in Schwann cell precursors (tumor cells) and reduced Nf1 expression (Nf1 heterozygosity; Nf1+/−; ref. 2). In these experiments, wild-type mice lacking Nf1 gene expression in Schwann cell precursors did not develop plexiform neurofibromas, whereas Nf1+/− mice lacking Nf1 gene expression in Schwann cell precursors formed plexiform neurofibromas. Within these benign tumors, there was pronounced infiltration of Nf1+/− mast cells. Further studies by Clapp and colleagues (3) showed that these Nf1+/− mast cells secrete elevated levels of transforming growth factor-β, which results in increased fibroblast proliferation and collagen synthesis, hallmarks of human neurofibromas. Based on these results, it was postulated that Nf1+/− mast cells in the neurofibroma microenvironment are responsible for generating the stromal conditions necessary for peripheral nerve tumor formation and growth.

In the central nervous system, we have shown that Nf1+/− microglia are important cellular determinants of optic glioma (low-grade brain tumor) growth. In these experiments, Nf1 inactivation in glial cells is not sufficient for gliomagenesis; whereas nearly 100% of Nf1+/− mice lacking glial Nf1 gene expression (Nf1+/−^GFAPCKO mice) develop low-grade gliomas involving the prechiasmatic optic nerves and chiasm by 10 to 12 weeks of age (4). Careful examination of these regions of the mouse optic nerve before obvious glioma formation (ages <6 weeks) showed the presence of monocyte-like cells, termed microglia (5). These CD68-positive/CD45R-negative microglia were detected in both NF1-associated and sporadic low-grade pediatric gliomas (6–8), and likely represented resident, rather than bone marrow-derived, microglia.

Microglia are characterized by two main morphologic states: Amoeboid microglia have large ovoid shaped cell bodies with few processes (9) and are thought to be activated microglia (10); in contrast, resting microglia have numerous ramified processes. The majority of the microglia observed in Nf1+/−^GFAPCKO mouse optic gliomas likely are activated microglia by virtue of their amoeboid shape. Previously, we showed that inactivation of these activated Nf1+/− microglia using minocycline resulted in decreased glial cell proliferation within the optic gliomas in vivo (11). This observation suggested that Nf1+/− microglia represent logical cellular targets for stromal-directed antitumor therapy.

The purpose of the current study was to determine the effect of Nf1 heterozygosity on microglia proliferation, motility, and activation as well as to identify the intracellular signaling pathway(s) responsible for neurofibromin regulation of these cellular properties. In this report, we show that Nf1+/− microglia exhibit increased proliferation in vitro and in vivo, and express transcripts associated with lipopolysaccharide (LPS)-induced activated microglia. We further show that neurofibromin regulates microglia proliferation, motility, and activation in a c-Jun NH₂-kinase (JNK)-dependent manner involving mixed lineage kinase (MLK) and Rac1 signaling. Lastly, we show that inhibition of JNK signaling in vivo reduces optic glioma proliferation. Collectively, these results reveal a critical role for Nf1+/− microglia in promoting optic glioma growth and implicate the JNK signaling pathway as a novel target for stroma-based brain tumor therapy.

Mouse specimens. Mice were used in accordance with established Animal Studies Protocols at the Washington University School of Medicine. Nf1^flox/flox; Nf1^flox/mut; GFAP-Cre (Nf1+/−^GFAPCKO), Nf1+/+ (wild-type), and Nf1+/− mice were all maintained on an inbred C57Bl/6 background.

Primary microglia cultures. Primary murine microglia were isolated from mixed glial cultures as described previously (11). Mixed glial cultures grown in 6-well plates were trypsinized with 0.05% trypsin-EDTA to remove contaminating astrocytes. Microglia cells that attached to the wells were recovered by trypsinization with 0.25% trypsin and vigorous pipetting before replating in glia-conditioned medium.

Primary astrocyte cultures. Murine cortical astroglial cultures, containing >95% glial fibrillary acidic protein (GFAP)-immunoreactive cells (astrocytes), were generated from postnatal day 1 to 2 Nf1^flox/flox pups as previously described (12, 13).

Thymidine incorporation. Microglia cell proliferation was measured by ³H-thymidine incorporation as described previously (11). Briefly, microglia were seeded into 24-well plates, grown for 2 d, serum starved for 24 h, and then pulsed with 1μCi tritiated thymidine (Amersham) per mL for 18 h. Microglia were washed with 0.1M PBS and solubilized in 500 μL of 0.2 mol/L NaOH. Counts (cpm) were then determined in a scintillation counter.

Boyden chamber assay. Sterile 8-μm transwell chambers (BD Pharmingen) were coated with Matrigel (Sigma). The filters were hydrated with 500 μL serum-free DMEM at room temperature for 2 h before use. The lower chambers of a 24-well plate were filled with 500 μL DMEM with 10% fetal bovine serum. Two hundred microliters of serum-free DMEM containing 2 × 10⁴ microglia cells were then added to the Matrigel-coated transwell chambers and incubated at 37°C in a 5% CO₂ humidified atmosphere for 3 h. At the end of incubation period, the cells on the upper surface of the filter were completely removed with a cotton swab. The filters were fixed in ice-cold methanol and stained with H&E. The cells on the lower surface of the filter were counted under a light microscope with a magnification of ×200 and photographed using an inverted Eclipse TE300 (Nikon) microscope equipped with an optical camera (Optronics). Triplicate samples were used for each experiment, and the data were expressed as the average cell number of five representative fields.

Facial nerve axotomy. F1 offspring from mating between wild-type mice and Nf1+/− mice were used, and facial nerve axotomy (FNA) was performed as previously described (14, 15). Briefly, pups were anesthetized on postnatal day 1 to 1.5 (P1–1.5) by hypothermia, and the right facial nerve was transected as it exited the stylomastoid foramen. The contralateral side served as an internal control. Mice were killed at 5 d after axotomy with an overdose of Nembutal and fixed by intracardiac perfusion with 4% paraformaldehyde. Brains were removed and processed for paraffin embedding and immunohistochemistry. The number of CD68+ cells in the region of both facial nerve nuclei was determined in three consecutive sections for each mouse by an observer blinded to the mouse genotype.

Quantitative reverse transcription-PCR. Real-time PCR (quantitative PCR) was performed as described previously using SYBR Green detection according to manufacturer's instructions (11). Total RNA was extracted from untreated microglia or microglia treated with LPS (200 μg/mL) or SP600125 (10 μmol/L) using Trizol. cDNA was synthesized from total RNA samples using the Omniscript kit (Qiagen). Amplification was performed in 96-well plates in a real-time sequence detection system instrument (ABIPRISM 9700HT; Applied Biosystems). SDS system software was used to convert the fluorescent data into cycle threshold (CT) measurements. The ΔΔCT method was used to calculate changes in fold expression relative to Nf1+/+ microglia using β-actin as an internal control.

Immunocytochemistry. Microglia were seeded onto 10-mm coverslips and grown for 2 d. Microglia were washed in 0.1M PBS and fixed with 4% formaldehyde at room temperature for 15 min. Microglia were then washed and permeabilized with 0.1% Triton X-100 in PBS for 30 min. After additional washes, cells were blocked in 5% goat serum for 30 min followed by incubation with primary antibody (CD68; Serotec; 1:200 dilution or phospho-c-Jun-Ser73; Abcam; 1:100 dilution) for 1 h at 37°C and the appropriate Alexa Fluor—tagged secondary antibody (Molecular; Probes; 1:1000 dilution) for 30 min at 37°C. 4′,6-Diamidino-2-phenylindole (DAPI) was used for counterstaining. The number of cells with nuclear phospho-c-Jun staining was quantified as a percentage of the total number of DAPI+ nuclei in five randomly chosen high-power fields. Positive controls for these experiments included wild-type microglia treated with 200 μg/mL LPS, whereas negative controls for immunostaining involved omission of the primary phospho-c-Jun antibody.

Immunohistochemistry. For CD68, GFAP, and Ki67 staining of optic nerves and brain, mice were perfused transcardially with ice-cold PBS and fixed with 4% paraformaldehyde. The optic nerves and brains were dissected, fixed, and processed for paraffin embedding and sectioning. Five-micrometer paraffin-embedded sections of mouse optic nerves and brains were used for analysis. The sections were deparaffinized followed by antigen retrieval in citrate buffer. After the washing and blocking steps, brain sections were incubated overnight with CD68 (1:100 dilution), GFAP (1:200), or Ki67 (1:200 dilution) antibodies followed by incubation with biotinylated secondary antibodies (1:200) at room temperature for 1 h. Immunoreactivity was visualized with the Vectastain ABC System and 3,3-diaminobenzidine (Vector Laboratories). All sections were photographed with a digital camera (Optronics) attached to an inverted microscope (Nikon).

Jnk inhibitor (SP600125) treatment in vivo. SP600125 (Calbiochem) was dissolved in DMSO and administered at a dosage of 50 mg/kg consecutively for 7 d. Mice were divided into two groups: one received i.p. injections of SP600125 and the other received injections of vehicle (DMSO). There were 10 animals in each group. Seven days after the final injection, mice were euthanized and perfused with normal saline followed by 4% paraformaldehyde to prepare paraffin-embedded optic nerve sections. Proliferation was measured using Ki67 staining and scored by an observer blinded to the treatment group, as previously reported (16).

Western blot analysis. Western blots were performed as described previously (11). JNK and phospho-JNK (Thr183/Tyr185), S6 and phospho-S6 (Ser240/244), AKT and phospho-AKT (Ser473), extracellular signal-regulated kinase (ERK)1/2 and phospho-ERK1/2 (Thr202/Tyr204), p38-MAPK and phospho-p38-MAPK (Thr180/Tyr182), S6 and phospho-S6 (Ser240/244), MLK3 and phospho-MLK3 (Thr277/Ser281), and MKK4 and phospho-MKK4 (Ser257/Thr261) antibodies were used at 1:1,000 dilutions (Cell Signaling Technology). Appropriate horseradish peroxidase–tagged secondary antibodies (Cell Signaling Technology) were used for detection by chemiluminescence after exposure to Biomax film (Amersham Biosciences). Films were scanned for analysis with Gel-Pro Analyzer 4.0 (Media Cybernetics).

Statistical analysis. Statistical analyses were performed using GraphPad Prism 4.0 software. Student's two-tailed t test was used with significance set at a P value of <0.05. All in vitro experiments were performed at least thrice with similar results.

Nf1 heterozygosity leads to increased microglia proliferation and motility in vitro and in vivo. Previous experiments from our laboratory showed that Nf1+/− microglia exhibited increased proliferation in response to either serum or colony stimulating factor-1 (CSF-1) in vitro (11). This increase in proliferation, as assessed by thymidine incorporation, ranged between 2- and 3-fold (Fig. 1A). To extend these findings, we examined the effect of Nf1 heterozygosity on motility, another property of activated microglia. Using a modified Boyden chamber assay, Nf1+/− microglia exhibited a 2-fold increase in motility relative to wild-type microglia (Fig. 1B). This increase was not the result of increased proliferation in these experiments but rather an increase in the number of microglia traversing the membrane.

Next, we sought to determine whether the increase in Nf1+/− microglia proliferation and motility was also observed in vivo. For these experiments, we used a commonly used injury model, FNA (17), on postnatal day 1 to 1.5 Nf1+/− and wild-type mice (8 mice per genotype). Five days postaxotomy, mice were euthanized and the number of CD68+ microglia, and GFAP-immunoreactive astrocytes were examined by immunohistochemistry (Fig. 1C). In both Nf1+/− and wild-type mice, we observed an increase in the number of reactive GFAP+ cells (astrocytes). However, in the Nf1+/− mice, we observed a 2-fold increase in the number of CD68+ microglia in the region of the facial nucleus after axotomy compared with the uninjured contralateral side (Fig. 1D). Collectively, these results show that Nf1 heterozygosity leads to increased microglia proliferation and motility in vitro, and results in greater numbers of microglia in response to injury in vivo.

Nf1+/− microglia share expression profiles with activated microglia. Because Nf1+/− microglia exhibit biological properties common to bacterial LPS-treated activated microglia (e.g., increased motility, augmented response to facial axotomy), we sought to determine whether Nf1+/− microglia increase their production of proinflammatory cytokines, similar to activated microglia. First, we used quantitative PCR (qPCR) to measure the effect of LPS treatment on the expression of four previously described proinflammatory cytokines and transcripts increased in activated microglia: cyclo-oxygenase-2 (cox2), interleukin-6 (IL-6), inducible nitric oxide synthase (iNOS), and tumor necrosis factor-α (TNF-α; refs. 18, 19). In response to LPS treatment, the relative expression of cox2, il6, inos, and tnf-α mRNA were increased relative to untreated wild-type microglia (Fig. 2A). Next, we examined the relative expression of cox2, il6, inos, and tnf-α mRNA in Nf1+/− microglia compared with wild-type microglia (n = 3 independent cultures per genotype). Although the pattern of overexpression was similar to that observed in LPS-treated microglia, Nf1+/− microglia generally had lower levels of increased expression (Fig. 2B). These results suggest that Nf1+/− microglia exhibit an expression profile shared with activated microglia, and raise the possibility that Nf1 heterozygosity may promote the generation of a specialized microglial phenotype that facilitates tumor formation and growth.

Nf1+/− microglia exhibit increased JNK pathway activation. Previous studies on Nf1 growth regulation have shown that hematopoietic cell proliferation and motility is regulated by the Ras effectors, Akt and MAPK (ERK). In this regard, Nf1+/− mast cell proliferation and motility reflects increased Akt and ERK activation in vitro (20). To identify the deregulated signaling pathway in Nf1+/− microglia, we examined the activation status of Akt, ERK, p38-MAPK, and ribosomal S6 using activation-specific phospho-antibodies. To our surprise, Nf1+/− microglia had no reproducible changes in the activation of these neurofibromin-regulated signaling molecules (<2.0-fold increased phosphorylation relative to wild-type microglia after normalization; Fig. 3A). In contrast to other Nf1+/− and Nf1−/− cell types, Nf1+/− microglia exhibited a 5.0-fold increase in JNK activation compared with wild-type microglia. Because JNK phosphorylates and activates c-Jun, resulting in its nuclear translocation, we examined the nuclear localization of phospho-c-Jun by immunocytochemistry in Nf1+/− and wild-type microglia. Whereas 20% of the wild-type microglia had nuclear phospho-c-Jun staining, 80% to 85% of the Nf1+/− microglia exhibited phospho-c-Jun nuclear staining (Fig. 3B). To establish that this nuclear localization pattern resulted from JNK activation, we used a JNK inhibitor (SP600125; 10 μmol/L) to show that JNK inhibition blocked phospho-c-Jun nuclear localization in both Nf1+/− and wild-type microglia. We chose this dose based on its ability to inhibit phospho-c-Jun nuclear immunoreactivity (data not shown; Fig. 3) and JNK activation by Western blot (Supplementary Fig. S1). Furthermore, the relative specificity of SP600125 was established using Nf1−/− cells in which only JNK phosphorylation was inhibited without any significant effects on MAPK (ERK), S6 or Akt activation—the major Ras downstream signaling pathways regulated by neurofibromin (Supplementary Fig. S1).

JNK inhibition reduces Nf1+/− microglia proliferation, motility, and proinflammatory cytokine production. Next, to determine whether the abnormal phenotypes observed in Nf1+/− microglia resulted from increased JNK activation, we treated Nf1+/− and wild-type microglia with the JNK inhibitor SP600125. In these experiments, SP600125 treatment reduced the proliferation (Fig. 4A) and motility (Fig. 4B) observed in Nf1+/− microglia to levels comparable with wild-type microglia. No significant effect of SP600125 treatment on wild-type microglia proliferation or motility was observed.

To determine whether the increased levels of cox2, il6, inos, and tnf-α mRNA expression found in Nf1+/− microglia also resulted from increased JNK pathway activation, we used qPCR to measure the levels of cox2, il6, inos, and tnf-α mRNA in Nf1+/− microglia before and after SP600125 treatment. In these experiments, SP600125 treatment decreased the levels of cox2, il6, inos, and tnf-α mRNA expression in Nf1+/− microglia relative to wild-type microglia (n = 3 independent cultures per genotype; Fig. 4C). Taken together, these data suggest that JNK pathway activation, as a result of reduced neurofibromin expression, is likely responsible for the abnormal cellular phenotypes observed in Nf1+/− microglia.

JNK hyperactivation in Nf1+/− microglia reflects increased Rac1/MLK3/MKK4 signaling. Previous studies on JNK activation have implicated deregulated MLK signaling (21, 22). Consistent with these observations, we found that Nf1+/− microglia exhibit increased MLK3 and Jun-kinase-kinase-4 (SAPK/ERK kinase or MKK4) activation relative to wild-type microglia using phospho-specific antibodies by Western blot (Fig. 5A). One of the important regulators of MLK3/MKK4 activation is Rac1 (23). Due to the large numbers of microglia required for direct Rac1 activation assays, we were unable to perform biochemical assays. Instead, we used an established Rac1 inhibitor to determine whether Rac1 regulates JNK pathway activity in Nf1+/− microglia. In these experiments, Rac1 inhibition (NSC23766; 10 μmol/L) resulted in decreased phospho-c-Jun nuclear localization in both Nf1+/− and wild-type microglia (Fig. 5B). These findings suggest that JNK hyperactivation in Nf1+/− microglia likely results from deregulated Rac1/MLK3/MKK4 pathway activation.

SP600125 treatment reduces optic glioma proliferation in vivo. We previously showed that inactivation of microglia with minocycline in vivo resulted in decreased Nf1−/− optic glioma astrocyte proliferation (11). Using an Nf1 GEM model of optic glioma in which Nf1+/− mice lack neurofibromin expression in GFAP+ cells (glia), we showed that the increased proliferation in these tumors reflects astrocyte, and not microglia, proliferation (16). Using this Nf1 optic glioma GEM model, we sought to determine whether pharmacologic inhibition of Nf1+/− microglia JNK-mediated activation in vivo would result in decreased optic glioma proliferation. First, we examined JNK activation in Nf1−/− astrocytes to ensure that the primary effect of SP600125 treatment in the Nf1 optic glioma GEM model resulted from inhibition of Nf1+/− microglia, and not Nf1−/− astrocyte, JNK activity. No reproducible changes in JNK activation were observed in Nf1−/− astrocytes relative to wild-type astrocytes as assessed by Western blotting using activation-specific phospho-antibodies (n = 5 independently generated paired Nf1−/− and wild-type astrocyte cultures). In these experiments, the fold increase in JNK activation in Nf1−/− astrocytes relative to wild-type astrocytes ranged between 0.50 and 2.0 (mean, 0.9-fold increase) when normalized to total JNK levels by scanning densitometry (Fig. 6A).

Next, we treated 10 Nf1+/−^GFAPCKO mice each at ages 12 weeks with either SP600125 or vehicle (DMSO) for 7 days. JNK inhibition in vivo was assessed by measuring JNK activation in spleen lysates. Spleen was selected based on experiments examining the level of JNK activation in various organs of Nf1+/− mice relative to wild-type littermates. Of all the organs examined, including brain, the highest level of JNK activation was observed in Nf1+/− spleen (data not shown). For these experiments, JNK activation was examined in fresh spleen tissue before perfusion from mice in both the vehicle- and SP600125-treated groups. The Western blot results from two representative mice in each group are shown in Fig. 6B, with an average of 3.5-fold reduced JNK activity for treated mice relative to untreated controls.

Finally, examination of optic glioma proliferation was performed using Ki67 immunohistochemistry, based on previous experience with the Nf1+/−^GFAPCKO mouse optic glioma model (16). As shown in Fig. 6C, we observed a 2-fold reduction in optic glioma proliferation after SP600125 treatment compared with vehicle-treated mice. These findings show that inhibition of Nf1+/− microglia function through blockade of a deregulated signaling pathway specific to microglia is sufficient to reduce optic glioma growth in vivo.

Current therapies for brain tumors focus on targeting deregulated growth control pathways within the neoplastic cells. Although these therapies have great merit, they do not consider the critical relationship between the tumor cells and the tumor microenvironment. NF1 provides an excellent experimental system to define the complex relationship between stromal elements and tumor cells. Brain tumors contain numerous diverse cell types, including immune system cells (e.g., microglia), nonneoplastic “reactive” astrocytes, tumor stem cell-like cells (cancer stem cells), and blood vessels (endothelial cells). The recognition that blood vessels are important for tumor growth and for providing a supportive niche for cancer stem cells is widely appreciated and has already led to several preclinical and clinical studies aimed at targeting these stromal cells. In contrast, comparatively less is known about the role of microglia in brain tumor formation and growth (24, 25).

Because low-grade glioma formation requires cooperativity between Nf1+/− stromal cells and Nf1-deficient glia, we have exploited a well-characterized Nf1 GEM optic glioma model to define the relationship between microglia and neoplastic glial cells (astrocytes). We have recently shown that Nf1+/− microglia produce growth and survival factors that uniquely drive the expansion of Nf1−/− astrocytes in vitro. The elaboration of specific growth factors and cytokines from microglia is usually associated with a state of microglia activation—a process by which these immune system cells change their morphologic appearance. In this regard, treatment of Nf1 GEM with minocycline to inactivate microglia results in reduced tumor cell proliferation in vivo (11). These observations identify Nf1+/− microglia as critical cells in the tumor microenvironment and suggest that treatments that abrogate microglia activation might be suitable therapies for NF1-associated optic glioma. The purpose of the present study was to identify the biologically relevant signaling pathway deregulated in Nf1+/− microglia in an effort to determine whether pharmacologic approaches that specifically target microglia might represent viable approaches to the treatment of NF1-associated optic glioma.

Glioma-associated microglia in both Nf1 GEM optic gliomas and their human counterparts are resident microglia that populate the brain during embryonic development. In this regard, these CD68+/CD45− monocyte-like cells do not derive from the bone marrow. Consistent with their proposed origins, reconstitution experiments using green fluorescent protein (GFP)-expressing Nf1+/− bone marrow cells injected into irradiated mice lacking Nf1 expression in glial cells did not result in glioma formation, despite the appearance of GFP+ bone marrow–derived cells in the recipient brains.¹

Careful examination of Nf1 GEM glioma-associated microglia reveals that the majority of these cells have an amoeboid appearance, consistent with an activated phenotype (11). Previous in vitro studies to characterize activated microglia have used bacterial LPS treatment, which results in increased microglia proliferation and motility (26, 27). In addition, LPS-treated microglia cultures increase their expression of numerous proinflammatory molecules, including IL-6, cox2, iNOS, and TNFα (18, 19). To better characterize the functional properties of Nf1+/− microglia, we performed several experiments. First, we showed that Nf1+/− microglia exhibit increased proliferation and motility relative to wild-type microglia in vitro. Second, we observed a marked increase in the number of microglia in Nf1+/− mice compared with wild-type mice in response to facial axotomy in vivo, a known stimulus for microglia recruitment (28). Third, Nf1+/− microglia have increased RNA expression of il6, cox2, inos, and tnfα compared with wild-type microglia. These data argue that Nf1+/− microglia harbor properties shared with activated microglia.

Activated microglia represent a specialized subtype of microglia, similar to what has been reported for macrophages (29). In this regard, researchers have identified distinct classes of specialized macrophages that play unique roles in tumorigenesis. Type 1 macrophages (M1) are regarded as potent effector cells that kill tumor cells and produce proinflammatory cytokines, whereas type 2 macrophages (M2) are composed of at least three subtypes that fine tune the inflammatory response, scavenge cellular debris, and promote angiogenesis and tissue remodeling (30). These M2 macrophages may be critical for promoting tumor progression (31, 32). Microarray analysis of Nf1+/− microglia showed increased expression of some transcripts associated with M1 macrophages as well as others associated with M2 macrophages.¹ The use of Nf1+/− microglia as a model system to study the class of brain microglia most capable of promoting tumorigenesis may provide unique insights into potential strategies aimed at modulating the balance between microglia-mediated tumor elimination and tumor progression.

To gain insights into the deregulated signaling pathway responsible for conferring the activated phenotype on Nf1+/− microglia, we examined known neurofibromin-controlled intracellular cascades. Unlike other Nf1+/− or Nf1−/− cell types, we observed no reproducible changes in MAPK (ERK), Akt, or S6 activation. However, Nf1+/− microglia show a significant increase in JNK pathway activity relative to wild-type microglia, as evidenced by increased MLK3, MKK4, and JNK activity as well as the translocation of phosphorylated c-Jun into the nucleus. To our knowledge, this is the first report of neurofibromin regulation of JNK pathway signaling. The hyperactivation of JNK pathway molecules is consistent with previous reports describing increased MKK4 and JNK activity after LPS treatment of microglia (33, 34). We further showed that c-Jun translocation was modulated by the MLK3/MKK4 regulator, Rac1. Although mast cell proliferation and motility is regulated by Akt/MAPK crosstalk, it is interesting that this signaling is mediated by Rac2 (20). In an effort to provide additional support for Rac1/MLK/MKK4 regulation of JNK activation, we attempted to transduce Nf1+/− microglia with murine stem cell virus, lentivirus, and adenovirus. Unfortunately, low transduction efficiencies precluded further studies aimed at replacing the NF1-GTPase activating protein domain or silencing components of the JNK pathway using short hairpin RNA interference. Studies are currently under way to develop efficient methods for gene delivery to primary mouse microglia in vitro.

Next, we sought to determine whether JNK activation was responsible for the activated microglia phenotypes observed in Nf1+/− microglia. In these experiments, we showed that treatment of Nf1+/− microglia with the JNK inhibitor SP600125 reduced both the proliferation and motility of Nf1+/− microglia to wild-type levels. Moreover, SP600125 treatment also reduced the increased expression of il6, inos, cox2, and tnfα RNA transcripts in Nf1+/− microglia. These findings are consistent with numerous reports demonstrating that JNK inhibition reduces IL-6 (35), iNOS (36, 37), cox2 (33), and TNFα (38, 39) production in LPS-treated activated microglia. In total, these findings support the notion that Nf1+/− microglia are specialized microglia, similar to LPS-treated microglia, and identify JNK as the primary signaling pathway affected by reduced Nf1 expression in microglia. Importantly, unlike previous reports using microglia cell lines (40), we found no increased activation of p38-MAPK. Although we interpret this result to signify that JNK pathway deregulation due to Nf1 heterozygosity is sufficient to lead to microglia activation, SP600125 has been reported to inhibit other signaling kinases, including serum-regulated and glucocorticoid-regulated kinase, p70-ribosomal S6 kinase, AMP-dependent protein kinase, cyclin-dependent kinase-3, casein kinase-1γ, and dual-specificity tyrosine-regulated kinase 1A, at similar doses used to inhibit JNK activity (41). In this regard, future studies using other more selective JNK inhibitors or short hairpin RNAi silencing will be necessary to firmly establish JNK as the central regulator of Nf1+/− microglia function.

Finally, we sought to determine whether pharmacologic inhibition of JNK hyperactivation in Nf1+/− microglia present in the Nf1 GEM optic glioma microenvironment results in decreased tumor growth. In these experiments we found that SP600125 treatment resulted in greatly attenuated optic glioma proliferation. Because the proliferating cells in the GEM optic glioma tumors are GFAP+ glial cells (16), we examined Nf1−/− astrocytes for evidence of JNK hyperactivation. We found that JNK activity in Nf1−/− astrocytes was either unchanged or slightly lower than observed in wild-type microglia. The lack of JNK hyperactivation in Nf1−/− astrocytes provided a unique opportunity to primarily inhibit Nf1+/− microglia in the intact tumor. Although we cannot formally exclude the effect of JNK inhibition on other Nf1+/− stromal cells, such as endothelial cells, it is most likely that Nf1+/− microglia were the primary target of SP600125 inhibition in light of recent experiments demonstrating that Nf1+/− endothelial cell proliferation is dependent on MAP/ERK kinase/MAPK signaling (42).

The idea that the severity of neurologic disease may be attenuated by reducing microglia function is gaining considerable traction. For example, elegant experiments in mouse models of cranial nerve injury (facial axotomy) and central nervous system demyelinating disease (experimental allergic encephalomyelitis) have shown that microglia ablation reduced chemokine release and decreased disease intensity (43). Our observations that targeting microglial function using minocycline (11) or SP600125 treatment (current study) attenuates glioma growth in vivo extend these stroma-based therapies to brain tumors. Although it is not clear exactly how minocycline blocks microglia function, one study suggested that it might suppress JNK pathway signaling (44). It is also worth noting that treatment of Nf1+/− microglia with JNK inhibitors also decreased the expression of two previously identified Nf1+/− microglia-produced factors that increase Nf1−/− astrocyte growth, MGEA5 and CXCL12.¹ As previously reported, MGEA5 is a member of the hyaluronidase family of molecules that uniquely stimulates the proliferation of Nf1−/−, but not wild-type, astrocytes (11), whereas CXCL12 is a chemokine that increases the survival of Nf1−/−, but not wild-type, astrocytes (45).

In summary, the findings described in this report highlight the critical contribution of nonneoplastic cells in the tumor microenvironment to the growth of brain tumors. Future studies that specifically target microglia activation or deregulated microglia signaling pathways relevant to their function may serve as important adjuvant therapies to complement current treatments that focus on the neoplastic cells in the tumor.

No potential conflicts of interest were disclosed.

Grant support: Department of Defense (DAMD17-03-1-0215) and National Cancer Institute (U01-CA84314); to D.H. Gutmann. G.C. Daginakatte was supported by a nested postdoctoral fellowship from the Department of Defense.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We thank Danielle Schwartz and Ryan Emnett for expert technical assistance, and Dr. Sutapa Banerjee for experiments on JNK activation in Nf1−/− astrocytes.