|Home | About | Journals | Submit | Contact Us | Français|
Children with the neurofibromatosis-1 (NF1) cancer predisposition syndrome exhibit numerous clinical problems that reflect defective central nervous system (CNS) neuronal function, including learning disabilities, attention deficit disorder, and seizures. These clinical features result from reduced NF1 protein (neurofibromin) expression in NF1+/− (NF1 heterozygosity) brain neurons. Previous studies have shown that mouse CNS neurons are sensitive to the effects of reduced Nf1 expression and exhibit shorter neurite lengths, smaller growth cone areas, and attenuated survival, reflecting attenuated neurofibromin cAMP regulation. In striking contrast, Nf1+/− peripheral nervous system (PNS) neurons are nearly indistinguishable from their wild-type counterparts, and complete neurofibromin loss leads to increased neurite lengths and survival in a RAS/Akt-dependent fashion. To gain insights into the differential responses of CNS and PNS neurons to reduced neurofibromin function, we designed a series of experiments to define the molecular mechanism(s) underlying the unique CNS neuronal sensitivity to Nf1 heterozygosity. First, Nf1 heterozygosity decreases cAMP levels in CNS, but not in PNS, neurons. Second, CNS neurons exhibit Nf1 gene-dependent increases in RAS pathway signaling, but no further decreases in cAMP levels were observed in Nf1−/− CNS neurons relative to their Nf1+/− counterparts. Third, neurofibromin regulates CNS neurite length and growth cone areas in a cAMP/PKA/Rho/ROCK-dependent manner in vitro and in vivo. Collectively, these findings establish cAMP/PKA/Rho/ROCK signaling as the responsible axis underlying abnormal Nf1+/− CNS neuronal morphology with important implications for future preclinical and clinical studies aimed at improving cognitive and behavioral deficits in mice and children with reduced brain neuronal NF1 gene expression.
Neurofibromatosis type 1 (NF1) is a common genetic disorder characterized by nervous system tumors (Listernick et al., 1997; Rosenfeld et al., 2010). While brain tumor formation is an important clinical feature of this condition, over 50% of children with NF1 have learning and behavioral deficits that most likely result from impaired neuronal function. These abnormalities in neuronal function are due to reduced, but not absent, neurofibromin expression in the brain (North et al., 1997; Rosser and Packer, 2003). In this regard, individuals with NF1 are born with one functional and one non-functional copy of the NF1 gene in every cell in their body (NF1+/− cells), such that their brains are composed of neurons with reduced neurofibromin levels. Similar to children with NF1, mice heterozygous for a germline Nf1 inactivating mutation (Nf1+/− mice) exhibit learning and behavioral deficits (Silva et al., 1997). These learning and memory deficits reflect reduced neurofibromin levels in neurons and not astrocytes (Cui et al., 2008), leading to impaired prefrontal and striatal inhibitory network function through activity-dependent GABA release (Costa et al., 2002; Shilyansky et al., 2010). In addition, we have also identified attention system abnormalities in Nf1 mutant mice that reflect reduced Nf1 gene expression in nigrostriatal dopaminergic neurons (Brown et al., 2010a).
Initial studies revealed that Nf1+/− peripheral nervous system (PNS) sensory ganglion neurons are nearly indistinguishable from their wild-type counterparts, whereas complete Nf1 loss in PNS neurons results in increased neurite lengths and neurotrophin-independent survival (Klesse and Parada, 1998; Romero et al., 2007; Vogel et al., 1995; Vogel and Parada, 1998). These deficits in Nf1-deficient PNS neurons result from loss of neurofibromin negative regulation of RAS/Akt signaling, and are restored by pharmacologic Akt inhibition. Similarly, Silva and colleagues showed that the spatial memory and learning deficits in Nf1+/− mice could be ameliorated by genetic reduction in RAS expression or by pharmacologic inhibition using the selective three-hydroxy-3-methylglutaryl coenzyme A (HMG-CoA) inhibitor, Lovastatin (Costa et al., 2002; Li et al., 2005). These findings suggest that neurofibromin primarily functions as a negative regulator of RAS in neurons.
However, neurofibromin has also been implicated as a positive regulator of cyclic AMP (cAMP) levels, such that Nf1-deficient embryonic CNS neurons (Tong et al., 2002) and Nf1-deficient astrocytes (Dasgupta et al., 2003; Warrington et al., 2007) have reduced cAMP levels. In these studies, neurofibromin loss was associated with decreased basal cAMP levels, and cAMP elevation using pharmacologic approaches reduced cell death in Nf1-deficient astrocytes in vitro (Warrington et al., 2007) and attenuated optic glioma growth in vivo (Warrington et al., 2010). Similarly, we have recently shown that Nf1+/− CNS neurons have decreased neurite lengths, growth cone diameters, and survival resulting from reduced neurofibromin-mediated cAMP generation (Brown et al., 2010b). These results raise the possibility that some biological properties of Nf1+/− neurons reflect defective cAMP regulation. A critical role of cAMP activity in the CNS is supported by numerous studies demonstrating that cAMP levels and downstream signaling through PKA and Epac regulate the formation of hippocampal synapses (Yamamoto et al., 2005; Tominaga-Yoshina et al., 2002), long-term potentiation (Gelinas et al., 2008; Ster et al., 2009; Abel et al., 1997; Qi et al., 1996), and learning and memory (Abel et al., 1997; Wang et al., 2011; Ouyang et al., 2008; Li et al., 2011).
The purpose of the current study was to define the molecular mechanism underlying neurofibromin cAMP regulation of CNS neuronal function, specifically neurite length and growth cone area – two morphological abnormalities resulting from reduced Nf1 gene expression in CNS neurons. First, we show that Nf1 heterozygosity reduces cAMP levels only in CNS, but not in PNS, neurons, and that there is no additional effect of complete Nf1 loss on cAMP levels in CNS neurons. Second, we demonstrate that neurofibromin regulates neurite length and growth cone area in a cAMP/PKA-dependent manner. Third, these abnormal Nf1+/− CNS neuronal phenotypes result from impaired cAMP/PKA control of Rho-associated kinase (ROCK) activity and myosin light chain (MLC) phosphorylation in vitro and in vivo. Collectively, these findings demonstrate differential effects of Nf1 gene dose on neuronal function and cAMP generation unique to CNS versus PNS neurons, and establish cAMP/PKA/ROCK pathway signaling as an important target for therapies aimed at correcting the CNS neuronal dysfunction in children with NF1.
All chemicals were purchased from Sigma unless otherwise indicated: Tuj-1 (1:1000 dilution; Covance), phospho-ROCK-Tyr249 (1:50 dilution; Abcam), and phospho-MLC Ser-19 (1:1500 dilution; gift from Robert Wysolmerski, West Virginia University School of Medicine; (Brown et al., 2009)). Cyclic AMP analogs, 8-bromoadenosine-3′,5′-cyclic monosphosphate (Br-cAMP, 20μM), 2′-O-methyladenosine-3′,5′-cyclic monophosphate (Me-cAMP, 20μM), and N6-phenyladenosine-3′,5′-cyclic monophosphate (Phe-cAMP, 20μM), were purchased from Biolog Life Science Institute. DDA (100μM) and CN01 (1unit/ml; Cytoskeleton) were dissolved in 10% DMSO and water, respectively. All drug treatments were performed for entire culture period in vitro, with the exception of CN01, which was added 10 min before fixation.
Nf1+/+ and Nf1+/− hippocampal neuronal cultures were generated from embryonic day 13.5 mice derived from C57 timed-pregnant females. Hippocampal neuronal cultures (“CNS neurons”) were prepared as previously described (Clarris et al., 1994) using Hibernate-E for dissection media. In brief, hippocampi were dissociated in HBSS containing 1% papain (Worthington Biochemicals, Lakewood, NJ) and 5 U/mL DNase (Gibco), transferred to a solution containing 1% ovomucoid (Worthington Biochemicals), and plated in neural basal + B27 and 2mM L-glutamine for 3 days. Neuronal culture purity was >90% as assessed by Tuj-1 and GFAP dual labeling. Retinal explant neuronal cultures were prepared with the RGC layer facing down and contacting the substrate (Hansen et al., 2004). RGCs were cultured in neurobasal media supplemented with B27, selenium, putrescine, triiodothyronine, transferrin, progesterone, pyruvate (1 mM), glutamine (2 mM), ciliary neurotrophic factor (CNTF; 10 ng/ml), brain-derived neurotrophic factor (BDNF; 50 ng/ml) and insulin (5 μg/ml) at 37°C and 10% CO2. DRG neuronal cultures (“PNS neurons”) were generated as previously reported (Brown et al. 2009). DRG neurons dissociated in 0.02% trypsin/EDTA were grown in C10-2 medium supplemented with nerve growth factor (NGF; 50ng/ml) for 48 hours.
Western blots were performed as previously described (Dasgupta and Gutmann, 2005; Hegedus et al., 2007) using the primary antibodies listed in Table 1. Appropriate HRP-conjugated secondary antibodies (Cell Signaling) were used for detection by enhanced chemiluminescence (New England Biolabs). Each experiment was performed with samples from at least three independent experiments.
Cell cultures were processed as previously described (Brown and Bridgman, 2009). Images were acquired using an Olympus confocal or an inverted microscope equipped with a Cooke Sensicam camera. Fixed cultures and paraffin-embedded tissues were imaged by phase confocal microscopy using identical laser intensities and post-image processing and scored for growth cone areas and neurite lengths using NIH Image-J software.
Growth cone areas and neurite lengths were determined as previously described (Brown, et al., 2010b). Briefly, growth cone areas were measured using Image-J, starting at the neck and tracing around the growth cone. Neurite lengths were also measured using Image-J, and only neurons with intact growth cones were included. Neurites were traced starting at the outer edge of the cell body, so that the cell body size would not affect measurements. Fluorescence intensities of the retina’s retinal ganglion layer were measured as a mean grey scale value using the Image-J software. Phospho-MLC intensity was calculated by normalizing the p-MLC fluorescence intensity to total neuronal tubulin (Tuj-1 staining) to adjust for changes in growth cone thickness (Brown et al., 2009). Investigators were blinded to the genotype of the neurons being analyzed.
Dissected forebrains, hippocampal neurons, or DRG neurons snap-frozen in liquid nitrogen were triturated in ice-cold 5% trichloroacetic acid (10μl per mg tissue) and centrifuged at 1000×g for 10min at 4°C. Supernatants supplemented with equal volumes of 0.1M HCl were extracted with water-saturated ether thrice prior to desiccation in a vacuum centrifuge. cAMP levels were quantitated using a cyclic AMP enzyme immunoassay kit (Assay Designs).
3-month-old Nf1+/− mice were treated with Rolipram in their drinking water (5mg/kg/day via gavage) for 2 weeks. Following perfusion, the eyeballs were placed in 1% agar in PBS and oriented with the retro-orbital optic nerve horizontal to the cutting surface. 5μm paraffin sections were generated for antibody labeling.
Student’s T-tests were used to determine significant changes in growth cone areas and neurite lengths. The mean and standard error of the mean were used for all graphs. All in vitro experiments were performed at least three times in a blinded fashion with identical results.
Our previous studies demonstrated that reduced neurofibromin expression conferred by Nf1 heterozygosity resulted in shorter neurite lengths and smaller growth cone areas in several different populations of CNS neurons (retinal ganglion cells, hippocampal neurons, and dopaminergic striatal neurons) without any significant effect on PNS (DRG) neurons (Brown et al., 2010b). These CNS neuronal abnormalities reflect attenuated neurofibromin positive regulation of intracellular cAMP levels in Nf1+/− neurons. To further explore the molecular basis for this distinctive response of CNS neurons to Nf1 heterozygosity, we performed several experiments examining the impact of neurofibromin reduction on its downstream signaling pathways in CNS compared to PNS neurons (Fig. 1a).
First, we show that Nf1 heterozygosity has similar effects on RAS signaling in CNS and PNS neurons. Consistent with the function of neurofibromin as a RAS-GAP molecule, reduced neurofibromin expression in both CNS (hippocampal) and PNS (DRG) neurons results in increased ERK1/2 activation, as measured using phospho-specific ERK1/2 antibodies (Fig. 1b). In these studies, Nf1+/− CNS neurons exhibited a 1.9-fold increase in ERK1/2 activation relative to their wild-type (WT) counterparts, whereas Nf1+/− PNS neurons had 2.3-fold more activated ERK1/2 than WT PNS neurons. Second, we demonstrate that only Nf1+/− CNS neurons exhibit reductions in intracellular cAMP levels compared to their WT counterparts (Fig. 1c). Direct measurement of cAMP levels revealed a 23% reduction in intracellular cAMP in Nf1+/− CNS (hippocampal) neurons relative to WT neurons (P=.003; n=8), whereas no differences in cAMP levels were observed between Nf1+/− and WT DRG neurons. It is worth noting that the baseline cAMP levels in PNS neurons are similar to those found in Nf1+/− CNS neurons.
As an additional measure of cAMP signaling in CNS and PNS neurons, we used phospho-PKA (p-PKA) substrate antibodies as an indirect read-out of cAMP-mediated PKA activation (Hatakeyama et al., 2004; Lei et al., 2007; Schmitt and Stork, 2002). In total cell lysates from CNS and PNS neurons, we detected four prominent bands consistently (100, 70, 40, and 32 kDa, respectively) on Western blot (Fig. 1d). While the identities of these protein species are largely unknown, a 32 kDa band has been identified as DARPP-32 by co-immunoprecipitation (data not shown), they serve as surrogate markers of PKA activity. Consistent with the reduced cAMP levels in Nf1+/− CNS neurons, all four p-PKA substrate proteins exhibited reduced expression in Nf1+/− hippocampal neurons relative to their WT counterparts (Fig. 1d). In contrast, three of the four p-PKA substrate proteins were increased in Nf1+/− PNS (DRG) neurons. These observations demonstrate that the unique effect of Nf1 heterozygosity on CNS neurons reflects decreased cAMP production.
Third, we next sought to determine whether lowering cAMP levels in PNS neurons could recapitulate the growth cone and neurite length defects observed in Nf1+/− CNS neurons. For these experiments, we used 2′,5′-Dideoxyadenosine (DDA), an inhibitor of adenylyl cyclase function, to reduce cAMP levels in WT and Nf1+/− DRG neurons. No significant changes in either DRG neurite length (Fig. 2a) or growth cone area (Fig. 2b) were observed in WT or Nf1+/− PNS neurons following DDA treatment while this dose of 100μM was sufficient to reduce hippocampal neuron growth cone spreading and neurite outgrowth.
Collectively, these results establish cAMP as a selective target of neurofibromin regulation of neurite length and growth cone area in CNS, but not PNS, neurons.
Previous studies in PNS sensory neuronal populations (DRG, nodose, and sympathetic neurons) demonstrated Nf1 gene dose-dependent increases in RAS pathway activation, such that only total neurofibromin loss resulted in PNS neuronal abnormalities (Klesse and Parada, 1998; Vogel et al., 1995; Vogel and Parada, 1998). In these reports, Nf1+/− PNS neurons were indistinguishable from their WT counterparts (Fig. 2; Brown et al., 2010b), whereas Nf1−/− PNS neurons exhibited increased neurite outgrowth and relative neurotrophin-independent survival (Brown et al., 2010b; Romero et al., 2007). To determine whether these Nf1 gene dose effects on RAS pathway signaling and cAMP levels were similarly observed in CNS neurons, we measured ERK1/2 activity using phospho-specific ERK1/2 antibodies and cAMP levels by immunoassay. Whereas CNS neurons show Nf1 gene dose-dependent elevations in ERK1/2 activity (Fig. 3a), we observed no additional decrease in cAMP levels in Nf1− deficient CNS neurons relative to their Nf1+/− counterparts (Fig. 3b). The failure to further reduce cAMP levels following total neurofibromin loss is in striking contrast to the effects of Nf1 expression on RAS signaling, and supports the hypothesis that reduced Nf1 gene expression in CNS neurons is sufficient to maximally attenuate cAMP levels and impair neuronal function.
The results described above in combination with our previous studies (Brown et al., 2010b; Hegedus et al., 2007) establish cAMP signaling as the critical pathway underlying the Nf1+/− CNS neuron growth cone and neurite length defects. However, it is not known whether cAMP exerts its primary effect on CNS neuronal function through the Epac or PKA pathway. To determine which cAMP downstream signaling pathway is responsible for the reduced growth cone areas and neurite lengths observed in Nf1+/− CNS neurons, RGC neurons were treated with one of three cAMP analogs: Br-cAMP (which activates both PKA and Epac), Me-cAMP (which activates only Epac), or Phe-cAMP (which activates only PKA) (Kang et al., 2006; Kloss et al., 2004; Kroeber et al., 2000). Following treatment, growth cone areas and neurite lengths were measured. We found that only the cAMP analogs which activate the PKA pathway (Br-cAMP and Phe-cAMP) restore Nf1+/− RGC neuron growth cone areas to WT levels (Fig. 4a). Similar effects on growth cone areas were seen following cAMP analog treatment in Nf1+/− hippocampal neuronal cultures (data not shown). In addition, we found that cAMP analogs that activate the PKA pathway also rescue the reduced neurite length defects in Nf1+/− hippocampal neurons (Fig. 4b), further establishing cAMP-PKA as the main signaling pathway responsible for the defects in growth cone and neurite length conferred by reduced neurofibromin expression in CNS neurons.
To determine whether reduced cAMP-PKA signaling was also observed in vivo, retinal preparations from 3-month-old mice were immunostained with antibodies that recognize phosphorylated PKA (p-PKA) substrates. Consistent with the in vitro findings, Nf1+/− mice had reduced p-PKA substrate immunostaining in retinal neurons. Following treatment with Rolipram, a phosphodiesterase-4 inhibitor which elevates cAMP levels (Sup Fig. 1a), p-PKA substrate immunostaining in Nf1+/− mouse retinal neurons was increased (Fig. 4c). In contrast, there was no effect of activated KRAS expression on retinal neuronal p-PKA substrate immunostaining using a neuroglial progenitor BLBP-Cre mouse strain (Sup Fig. 1b) (Hegedus et al., 2007). Together, these data demonstrate that cAMP-PKA signaling is responsible for the Nf1+/− CNS neuron growth cone and neurite length defects.
Careful inspection of the growth cones in Nf1+/− CNS neurons demonstrated relatively preserved microtubule architecture, with strong neck and central region Tuj1 staining and finger-like projections extending into filopodia (Fig. 5a). In contrast, the actin cytoskeleton, as revealed by phalloidin staining, had lost much of its defined peripheral region and lacked a defined actin meshwork with long actin bundles. The presence of actin cytoskeletal defects in Nf1+/− CNS neurons prompted us to examine neurofibromin regulation of Rho activation, using ROCK activation as a surrogate marker of Rho activity. Initial studies employed Rho activation assays, but the limited protein obtained from these cultures precluded accurate measurements of Rho activity (Brown JA, unpublished results, 2010). Western blot analysis of phosphorylated (activated) ROCK (p-ROCK) normalized to total ROCK levels revealed decreased ROCK activity in Nf1+/− hippocampal neurons relative to their WT counterparts (P=.0001; N=8, Fig. 5b). No significant change in ROCK activity was observed in Nf1+/− DRG neurons compared to WT DRG neurons.
One of the primary actin cytoskeleton targets regulated by ROCK is non-muscle myosin through phosphorylation of its regulatory light chain. ROCK increases myosin light chain (MLC) phosphorylation either directly or by inhibiting myosin light chain phosphatase (Amano et al., 1998). Following phosphorylation, MLC binds tightly to actin filaments to increase myosin motor activity (Yamakita et al., 1994). Consistent with the decrease in ROCK activity observed in Nf1+/− CNS neurons, we detected decreased MLC phosphorylation in Nf1+/− CNS neurons relative to their WT counterparts (P=.008; N=8, Fig. 5c). As above, little change in MLC phosphorylation was observed in Nf1+/− DRG neurons compared to WT DRG neurons.
Next, to demonstrate that attenuated Rho signaling is responsible for the growth cone and neurite length defects observed in Nf1+/− CNS neurons, Nf1+/− and WT hippocampal neurons were treated with the CN01 Rho activator (Fig. 6a). Whereas pharmacologic Rho activation slightly decreased growth cone areas in WT CNS neurons (21% decrease; P=.003, N=30), treatment of Nf1+/− hippocampal neurons with CN01 restored growth cone areas to WT levels (Fig. 6b). These findings demonstrate that neurofibromin regulates actin cytoskeleton function in CNS neurons through Rho-mediated signaling.
To determine whether defective cAMP/PKA signaling is responsible for the reduced ROCK and MLC activity observed in Nf1+/− CNS neurons, we performed two additional experiments. First, using cAMP analogs that preferentially activate the cAMP-regulated PKA and Epac pathways, we show that only cAMP analogs which activate the PKA pathway (Phe-cAMP) significantly increase phospho-MLC immunostaining intensity in Nf1+/− hippocampal neurons in vitro (Fig. 7a). No effect of Me-cAMP (Epac activator) on Nf1+/− hippocampal neuron phospho-MLC immunostaining intensity was observed. Second, Nf1+/− mice were treated with Rolipram to elevate cAMP levels in vivo, and phospho-ROCK and phospho-MLC levels in retinal preparations were assessed by fluorescence immunostaining. Whereas Nf1+/− retinal neurons exhibit dramatically attenuated phospho-ROCK and phospho-MLC immunostaining relative to their WT counterparts, Rolipram treatment of Nf1+/− mice dramatically increased ROCK (Fig. 7b) and MLC (Fig. 7c) activity levels in vivo. Collectively, these results establish reduced cAMP/PKA/Rho signaling as the primary defect responsible for the neuron growth cone and neurite length abnormalities unique to Nf1+/− CNS neurons.
Children with the neurofibromatosis-1 (NF1) tumor predisposition syndrome exhibit numerous clinical features that likely reflect a key role for neurofibromin in CNS neuronal function. In this regard, affected individuals are prone to developmental delays, seizures, learning disabilities, and behavioral problems, including attention deficit disorder (Eliason, 1986; Hyman et al., 2006; North et al., 1995; Ozonoff, 1999). These neuronal defects result from reduced, but not absent, NF1 gene expression, as the brains of these children are composed of NF1+/− cell types, including neurons.
In several different populations of CNS neurons (hippocampal, retinal ganglion and striatal dopaminergic neurons), Nf1 heterozygosity results in cell autonomous reductions in neurite lengths, growth cone areas, and survival. Biochemical analyses showed that these particular CNS neuronal defects result from impaired neurofibromin cAMP regulation (Brown et al., 2010b). In contrast, Nf1+/− PNS neurons are nearly indistinguishable from their wild-type counterparts. Moreover, complete Nf1 inactivation has the polar opposite effect on peripheral sensory neurons, leading to increased neurite lengths and relative neurotrophin-independent survival (Romero et al., 2007; Vogel et al., 1995). In addition, unlike their CNS counterparts, these enhancements in neuronal function are a consequence of absent neurofibromin negative regulation of RAS/Akt signaling (Romero et al., 2007). To gain insights into the differential responses of CNS and PNS neurons to reduced neurofibromin function, experiments in this report were designed to define the molecular mechanism(s) underlying the unique CNS neuronal sensitivity to Nf1 heterozygosity.
We show that neurofibromin negatively regulates RAS signaling in both CNS and PNS neurons. While increased ERK1/2 activation was observed in both Nf1+/− CNS and PNS neuronal populations, increased RAS signaling is unlikely to account for the neurite length and growth cone defects observed in Nf1+/− CNS neurons. Using mice in which Nf1 is inactivated in neural stem cells in vitro or in vivo, the reduced brain neurite lengths is only restored following treatments which elevate cAMP and is not recapitulated by forced expression of activated RAS or Akt alleles (Hegedus et al., 2007). Moreover, the reduced neurite length, growth cone area, and survival seen in Nf1+/− CNS neurons are only rescued by treatments that elevate cAMP levels (non-hydrolyzable cAMP analogs, forskolin, and Rolipram) in vitro and in vivo (Hegedus et al., 2007; Brown et al., 2010b), and not by expression of the neurofibromin RAS-GAP domain in vivo (Hegedus et al., 2008) or pharmacologic MEK or PI3K inhibition in vitro (Brown et al., 2010b; Hegedus et al., 2007; Hegedus et al., 2008), indicating a RAS-independent mechanism in Nf1+/− CNS neurons. Collectively, these observations support a model in which abnormal CNS neurite lengths, growth cone diameters, and survival following changes in neurofibromin expression reflects RAS-independent signaling.
We then demonstrate that Nf1 heterozygosity reduces cAMP levels only in CNS neurons. Further, we show that cAMP levels in CNS neurons are not reduced further upon total loss of neurofibromin expression. The fact that Nf1−/− hippocampal neurons have cAMP levels indistinguishable from their Nf1+/− neuronal counterparts suggests that complete Nf1 inactivation confers no further advantage to CNS neurons. This result contrasts with neurofibromin regulation of neuronal function in the PNS, where complete Nf1 loss is required to appreciate a significant effect on sensory neurite length and survival. It is likely that these differential effects of Nf1 gene expression on cAMP and RAS signaling reflect the manner in which neurofibromin functions in these distinct cell types. The notion that neurofibromin loss has cell type-specific effects on intracellular pathway signaling is underscored by several observations. While neurofibromin primarily regulates astrocyte and Schwann cell growth in a RAS/Akt/mTOR-dependent manner (Banerjee et al., 2010; Johannessen et al., 2005; Johansson et al., 2008; Sandsmark et al., 2007), RAS/MAPK appears to be the major neurofibromin growth control pathway in leukemic cells (Lauchle et al., 2009). Similarly, neuronal differentiation from neural stem cells (NSCs) is dependent on neurofibromin cAMP signaling, while astrocyte differentiation from NSCs requires RAS/Akt signaling (Hegedus et al., 2007).
While the mechanism underlying neurofibromin regulation of cAMP levels has not been fully elucidated, previous studies in Drosophila have suggested that de-regulated receptor tyrosine kinase function can activate adenylyl cyclase in a RAS-dependent manner (Hannan et al., 2006), and that learning impairments in Nf1 mutant flies reflect increased RAS activity (Ho et al., 2007). These intriguing observations raise the possibility that neurofibromin could influence adenylyl cyclase-mediated cAMP generation either directly or indirectly through other signaling intermediates. Future mechanism experiments will be required to define how neurofibromin controls cAMP generation in CNS neurons.
Next, we show that neurofibromin/cAMP regulation of CNS neuronal function operates through a PKA-dependent pathway. In this regard, the abnormal Nf1+/− CNS neuronal phenotypes reflect reduced PKA and Rho/ROCK signaling. This conclusion is supported by reduced PKA substrate phosphorylation in Nf1+/− CNS neurons in vitro and in vivo. In addition, using specific cAMP analogs, only those compounds that activate PKA, but not Epac, rescue the neurite length/growth cone area defects in Nf1+/− CNS neurons. We further show that Nf1+/− CNS neurons exhibit reduced ROCK phosphorylation in vitro and in vivo, indicative of reduced Rho activity. The ability of cAMP to activate Rho has been reported for other cell types (Yamauchi et al., 2008) where it reflects PKA signaling (Leemhuis et al., 2002; Lecuona et al., 2003). Consistent with these findings, we show that pharmacologic Rho activation, similar to cAMP treatment, rescues the Nf1+/− CNS neuronal defects in vitro. The importance of cAMP and PKA activity to regulate CNS neuronal function is supported by genetic and pharmacologic experiments in which this signaling axis regulates synaptogenesis (Yamamoto et al., 2005; Tominaga-Yoshina et al., 2002), long-term potentiation (LTP) (Qi et al., 1996; Abel et al., 1997), and hippocampal-dependent learning and memory (Ouyang et al., 2008). Moreover, ablation of adenylate cyclase-3 to reduce cAMP levels in mice impairs learning and memory (Wang et al., 2011), whereas Rolipram treatment to raise cAMP levels of mice with the phosphodiesterase-4 genetic disruption enhances LTP and memory (Gong et al., 2004; Li et al., 2011).
Finally, we provide a mechanistic connection between cAMP/PKA signaling and Rho-mediated actin cytoskeleton dynamics in Nf1+/− CNS neurons. In this report, we demonstrate that the disorganized actin cytoskeleton observed in Nf1+/− hippocampal neurons reflects reduced neurofibromin/cAMP-PKA regulation of myosin light chain (MLC) phosphorylation. Non-muscle myosin is a key driver of actin cytoskeleton dynamics (Yamakita et al., 1994), such that myosin bipolar filaments allow force to be applied to the actin bundles to facilitate neurite outgrowth and growth cone spreading (Bridgman et al., 2001; Brown et al., 2009; Rochlin et al., 1995; Tullio et al., 2001). The observation that Nf1+/− CNS neurons have reduced Rho-dependent MLC phosphorylation and actin cytoskeleton defects is consistent with the established role of PKA and Rho/ROCK in the regulation of actin-based neurite outgrowth (Aizawa et al., 2001). Growth cones represent the primary locomotor structure responsible for axonal path finding and synaptogenesis (Lankford et al., 1990) both during development and regeneration (Ledig et al., 1999; Mueller, 1999; Tanelian et al., 1997), suggesting that some of the learning disabilities in children with NF1 may result from defective neurofibromin/cAMP/PKA regulation of Rho-mediated MLC actin cytoskeleton dynamics. Future studies on the impact of Nf1 heterozygosity on CNS neuronal plasticity, synaptogenesis and path finding will likely reveal new insights into these common NF1-associated cognitive problems.
In summary, the studies in this report elucidate the mechanism by which cAMP regulates neurite lengths and growth cone areas in Nf1+/− CNS neurons. Moreover, we demonstrate differential effects of Nf1 gene dose on cAMP generation in CNS versus PNS neurons, and establish cAMP/PKA/Rho/ROCK pathway as the responsible axis for some aspects of CNS neuronal function. It is possible that other neuronal properties impaired by changes in Nf1 gene expression reflect de-regulated signaling through other pathways, including RAS (Costa et al., 2002). Understanding the relative contributions of both cAMP and RAS function in CNS neurons has important implications for future preclinical and clinical studies aimed at improving cognitive and behavioral deficits in mice and children with NF1 as well as reducing secondary neuronal damage resulting from brain tumor growth and treatment.
(A) Treatment for 2 weeks with Rolipram increases cAMP levels in Nf1+/− mice in vivo (P=.001, N=8). (B) KRAS mice demonstrate elevated levels of p-ERK compared to WT mice at E13.5 (P=.001, N=6).
This work was supported in part by a grant from the National Cancer Institute (U01-CA141549-01 to DHG) with Diversity Supplement funding (to JAB). This work was also supported by the Bakewell Neuroimaging Core and NIH Neuroscience Blueprint Interdisciplinary Center Core Grant P30 (NS057105 to Washington University).
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.