Abstract
A persistent inflammatory and oxidative stress is a hallmark of most chronic CNS pathologies (Alzheimer’s (ALS)) as well as the aging CNS orchestrated by the proinflammatory cytokines tumor necrosis factor alpha (TNFα) and interleukin-1 beta (IL-1β). Loss of the integrity and plasticity of neuronal morphology and connectivity comprises an early step in neuronal degeneration and ultimate decline of cognitive function. We examined in vitro whether TNFα or IL-1β impaired morphology and motility of growth cones in spinal cord neuron cultures. TNFα and IL-1β paralyzed growth cone motility and induced growth cone collapse in a dose-dependent manner reflected by complete attenuation of neurite outgrowth. Scavenging reactive oxygen species (ROS) or inhibiting NADPH oxidase activity rescued loss of neuronal motility and morphology. TNFα and IL-1β provoked rapid, NOX-mediated generation of ROS in advancing growth cones, which preceded paralysis of motility and collapse of morphology. Increases in ROS intermediates were accompanied by an aberrant, nonproductive reorganization of actin filaments. These findings suggest that NADPH oxidase serves as a pivotal source of oxidative stress in neurons and together with disruption of actin filament reorganization contributes to the progressive degeneration of neuronal morphology in the diseased or aging CNS.
1. Introduction
A spreading inflammatory reaction accompanied by oxidative stress is prevalent in most chronic CNS diseases and acute CNS trauma as well as in the aging CNS [1–3]. The proinflammatory cytokines TNFα (tumor necrosis factor α) and IL-1β (interleukin-1β) exert pleiotropic functions both in CNS development and CNS pathogenesis [4–6]. Persistent high-level expression of TNFα and IL-1β is important to the progressive degeneration of neuronal connectivity and loss of neuronal plasticity ultimately leading to cognitive decline. The role of TNFα and IL-1β as inducers of apoptotic events is well documented, whereas the recognition of their morphogenetic function is more recent [7]. TNFα and IL-1β released from microglia cells inhibited neurite outgrowth, reduced branching, and caused neurite retraction in cultures of Neuro2A cells or primary hippocampal neurons [8, 9]. The intricate pattern of neuronal connectivity innate to cognitive function rests as much on the integrity and stability of the axonal and dendritic architecture as on the plasticity of motile structures to maintain, form, or regenerate connections, which is intimately linked to the dynamic reorganization of the actin cytoskeleton [10, 11]. ROS intermediates greatly affect both the dynamics and organization of actin filaments during oxidative stress of physiological redox signaling [12, 13]. TNFα paralyzed actin filament reorganization in neuroblastoma cells due to oxidative damage, whereas physiological levels of ROS intermediates seem to be necessary for proper growth cone motility [14, 15]. TNFα and IL-1β potently stimulate NADPH oxidase (NOX) activities in neurons and glia cells often localized in coalescing lipid rafts [16, 17]. The members of the NADPH oxidase (NOX) family (NOX1–5, DUOX1/2), defined by the large membrane flavoprotein of the phagocyte NAPDH oxidase (NOX2), are ubiquitously expressed in all cell types and have emerged as principal ROS sources both in cellular signaling and disease progression in response to cytokines, growth factors, and hormones [18, 19]. Functional NOX requires an intricate assembly between two membrane proteins (a NOX isoform, ) and several cytosolic factors (, , ) under the regulation of the small GTPases Rac1 or Rac2 [20]. Rho GTPases harbor a dual role in cytokine signaling as regulators of both NOX assembly and the reorganization of actin filament structures [21–23]. In light of these reports, we examined whether ROS intermediates generated by NOX activities in neuronal growth cones are implicated in mediating the neurotoxic effects of TNFα or IL-1β on neurite outgrowth. A mechanistic understanding of the detrimental consequences of TNFα and IL-1β on neuronal connectivity in the CNS neurons is vital to intervene with progressive neurodegeneration in the aging or diseased CNS.
2. Materials and Methods
2.1. Reagents
Unless state otherwise, all reagents were purchased from Sigma (St. Louis, MO). Dulbecco’s Modified Eagle Medium (high glucose DMEM), Leibovitz’s L-15, and isopropyl β-D-thiogalactopyranoside were obtained from Invitrogen Corporation (Carlsbad, CA). Characterized fetal bovine serum (FBS) was from Hyclone (Logan, UT) and it was heat-inactivated according to the manufacturer. Laminin was from Roche Diagnostics Corporation (Indianapolis, IN). The redox sensitive fluorescent indicators carboxymethyl-2′,7′-dihydrodichlorofluorescein diacetate, 2′,7′-dihydrodichlorofluorescein diacetate, and Calcein Blue were obtained from Molecular Probes (Eugene, OR). Human recombinant tumor necrosis factor α (TNFα) and interleukin-1β (IL-1β), Diphenylene iodonium (DPI), N-acetyl-L-cysteine (NAC), and Manganese(III) tetrakis(4-benzoic acid) porphyrin chloride (MnTBAP) were purchased from Calbiochem (San Diego, CA). Polyclonal rabbit anti-TNF receptor I, polyclonal rabbit anti-TNF receptor II, polyclonal rabbit anti IL-1 receptor type I, polyclonal goat anti-NOX2, polyclonal goat anti- including respective blocking peptides were all from St. Cruz Biotechnology (St. Cruz, CA). A monoclonal mouse anti-p44/42 MAPK, a monoclonal mouse anti-phospho p44/42 MAPK, and a monoclonal mouse anti-phospho JNK were from Cell Signaling Technologies (Danvers, MA). Fluorescein or rhodamine-labeled goat anti-rabbit and goat anti-mouse IgG (resp.) were from Chemicon (Temecula, CA), whereas rhodamine-conjugated phalloidin was obtained from Cytoskeleton Inc. (Denver, CO). The polyclonal rabbit anti- was purchased from Abcam (Cambridge, MA).
2.2. Spinal Cord Neuron Cultures
Dissociated, low-density cultures of spinal cord (SC) neurons were established as described [24]. Briefly, spinal cord tissue was dissected from 7 day-old chick embryos (E7) and a single cell suspension obtained after enzymatic digestion (0.5 mg/mL trypsin, 2 mM EDTA, 10 min, 37°C) followed by trituration. After preplating (1 h, 37°C, 5% CO2 atmosphere, high glucose DMEM, pH 7.3/10% FBS), nonadherent cells (predominantly SC neurons) were collected (3 min, 200 ×) and resuspended in SC medium (high glucose DMEM pH 7.3, 10% FBS, 12 nM fluorodeoxyuridine, 30 nM uridine, and 1% N3 nutrient supplement) [25]. SC explants were prepared by pushing freshly dissected E7 spinal cord tissue through a wire-mesh (50–75 μm). Dissociated SC neurons (75,000 cells/mL) or SC explants (2 per culture) were plated onto glass cover slips (22 × 22 mm2, number 1, Carolina Biological Supply Company, Burlington, NC) mounted over a 1.5 cm hole drilled into the bottom of a 35 mm culture dish. Glass cover slips were treated with poly-D-lysine (100 μg/mL, borate buffer pH 8.4, 30 min, RT) prior to coating with laminin (5 μg/cm2, 30 min, RT). These SC neuron cultures contained less than 5% nonneuronal cells with no immunoreactivity against microglia markers.
2.3. Growth Cone Particle Preparations
Preparations of highly enriched growth cone particles were obtained from freshly dissected whole chick embryo brains (E10-12) [26, 27]. After homogenizing (Dounce homogenizer, ice-cold 5 mM HEPES pH 7.3, 1 mM MgCl2, 0.32 M Sucrose, 6–8 volumes per wet weight), homogenates were centrifuged (15 min, 1600 ×, 4°C). Resulting supernatants were overlaid onto 5 mM HEPES pH 7.3, 1 mM MgCl2, and 0.75 M sucrose and then centrifuged (150,000 ×, 1 h, 4°C), and material at the interface was collected. After dilution (6-7 times in low sucrose buffer), the suspension was overlaid onto MaxiDense (4-5 sample volumes per volume Maxidense) and centrifuged at 40,000 × (1 h, 4°C), and growth cone particles (GCPs) were collected on top of the MaxiDense cushion. GCPs were resuspended in Kreb’s buffer (145 mM NaCl, 5 mM KCl, 1.2 mM NaH2PO4 × H2O, 1.2 mM MgCl2, and 5 mM HEPES pH 7.3) and protein concentration determined (BCA assay, Thermo Scientific, Rockford, IL, USA).
2.4. Measurement of Neurite Lengths
Dissociated SC neurons were grown after the onset of neurite outgrowth indicated by the majority of cells extending at least one process longer than 2 cell body diameters. Cultures were incubated (1 h) with pharmacological agents, followed by bath application (6–8 h) of 100 ng/mL TNFα, 100 ng/mL IL-1β, or 10 μg/mL ovalbumin. After fixation (2% glutaraldehyde), the length of the longest neurite per neuron was measured of at least 50 randomly selected SC neurons only considering processes adhering to the following criteria: (i) emerging from an isolated cell body, (ii) longer than two cell diameters, and (iii) no contact to other neuronal processes or cell bodies. The distribution of neurite length in a population of SC neurons was obtained by plotting the percentage of neurons with neurites longer than a given length against neurite length [28]. As a characteristic for neurite outgrowth under a given condition, the neurite length reached by 50% of neurite-bearing SC neurons (NL50) was calculated as criteria for statistical significance [29]. Purified, recombinant -GST, -GST, or GST were introduced into freshly dissociated SC neurons (7–10 mg/mL, 500,000 cells, 200 μL 50 mM Tris-Cl pH 7.5, 100 mM NaCl, and 5 mM MgCl2) by trituration loading at the time of plating [26]. Following trituration, SC neurons were immediately transferred in SC medium and plated as described above. For each condition, measurements were performed in duplicate cultures from two independent dissections.
2.5. Measurement of Growth Cone Advance
Dissociated SC neurons (12 to 16 h in culture) were transferred into observation medium (Leibovitz’L15 without phenol red, pH 7.4, 5 mg/mL ovalbumin, 1% N3), overlaid with light mineral oil to avoid evaporation, and placed onto the microscope stage equilibrated at °C (Nikon TE2000 U). After 15-minute recovery, images of advancing growth cones (20x, phase contrast) were acquired at 3-minute time intervals for a 30 min time period (Coolsnapfx, Photometrics, Tuscon, AZ). Cytokines (100 ng/mL TNFα, 100 ng/mL IL-1β) or 10 μg/mL Ovalbumin were bath-applied at = 8 min. The extension of the growth cone/neurite boundary (μm) was measured as a function of time (min) (Metamorph Software, Meridian Instrument Co, Kent, WA). At least 20 growth cones were monitored for each condition in duplicate cultures from at least two different dissections.
2.6. Recombinant Mutant Rac1 Protein
Recombinant -GST or -GST protein was induced in E. coli DH5α with 0.5 mM isopropyl β-D-thiogalacto-pyranoside (4 h) [30]. After cell lysis (50 mM Tris-Cl pH 7.5, 50 mM NaCl, 5 mM MgCl2, and 1 mM dithiothreitol), a soluble protein fraction was obtained (20,000 ×, 20 min, 4°C) and subjected to affinity chromatography on glutathione-conjugated agarose. Bound proteins were eluted (5 mM glutathione in lysis buffer), concentrated to 5–7 mg/mL in 50 mM Tris-Cl pH 7.5, 100 mM NaCl, and 5 mM MgCl2 (Centrifugal Devices, Millipore, Bedford, MA), and stored in 100 μL aliquots at −80°C after snap-freezing. Protein preparations showed single bands on Coomassie blue-stained 10% SDS polyacrylamide gels.
2.7. Adenoviral Expression of Rac1 Mutants in SC Neurons
Recombinant, replication deficient adenovirus carrying genes for constitutively active Rac1 ( with N-terminal FLAG tag), dominant negative Rac1 ( with N-terminal FLAG tag), or lacZ were expressed in E7 chick SC neurons as described previously [31]. At the time of plating, dissociated SC neurons were infected with recombinant adenovirus at 200 moi (multiplicity of infection) in 300 μL SC medium. Cultures were replenished with 200 μL fresh SC medium after 12 h and grown for additional 48 h. At three days after infection, yields of neuronal infections generally exceeded 70% [31, 32]. Amplification of viral stocks was performed in 293 HEK cells and titers greater than 5 × 108 plaque forming units per mL were routinely obtained. Viral stocks were stored at –80°C.
2.8. Quantitative ROS Imaging in SC Neuron Cell Bodies
Dissociated SC neurons were loaded with 10 μM 2′,7′-dihydrodichlorofluorescein diacetate (DCF) or 5 μM dihyroethidium (DMEM/10% FBS) for 30 min (37°C, 5% CO2 atmosphere), washed, and allowed to recover (15 min, DMEM/10% FBS). Cultures were switched to observation medium, overlaid with light mineral oil, transferred to the heated microscope stage (Zeiss Axiovert125S), and allowed to adapt for 15 min. Images of dissociated SC neurons (random fields of view) were acquired under phase contrast (20x Plan-Apo objective) and FITC illumination (Ex 465–495 nm, DM 505, Em 515–555) using a Peltier cooled CCD camera (Sensys, Photometrics, Tuscon, AZ) after the recovery phase (defined as basal condition), after pharmacological treatments (30 min), and after cytokine exposure (100 ng/mL TNFα or 100 ng/mL IL-1β). Hydrogen peroxide was added to all cultures following treatments to ensure proper loading. As our criteria for SC neurons, we analyzed only cells displaying a large round cell body and one neuronal process at least longer than three cell diameters. Maximum DCF fluorescence intensity per neuronal cell body was determined on a pixel-by-pixel basis following background subtraction (average background of all images at = 0 min for each condition) followed by erosion (2 pixels) and overlay with the original image (Zeiss imaging analysis software KS 300). All values were normalized to the average DCF fluorescence intensity in control (initial conditions). No morphological changes of neuronal cell bodies were detectable in our assay conditions indicated by constant cell body areas. ROS measurements were performed in duplicate cultures obtained from 3 to 5 independent dissections. At least 50 SC neurons were measured in duplicate cultures of three independent dissections.
2.9. Quantitative ROS Imaging in Advancing SC Neuron Growth Cones
SC explants were incubated (30 min, DMEM/10% FBS) with 20 μM DCF (5′-(and 6′)chloromethyl-dichlorodihydrofluorescein diacetate) and 4 μM Calcein Blue (CB), an oxidation-inert fluorescence indicator. After recovery (15 min, DMEM/10% FBS), cultures were switched to observation medium and overlaid with light mineral oil. Images of advancing growth cones were acquired under phase contrast (40x Plan-Apo), FITC illumination (DCF fluorescence), and DAPI illumination (CB fluorescence) (Coolsnapfx) before (pre-stimulus images at = 0, 2, 4 min) and after (after stimulus images, 2 min time intervals, 16 min time period) bath application of 100 ng/mL TNFα, 100 ng/mL IL-1β, or 10 μg/mL ovalbumin. As our positive loading control, following the observation period, all growth cones were exposed to 100 μM hydrogen peroxide. For image analysis, DCF and CB fluorescence intensities ( and ) were integrated (pixel-by-pixel basis) over the growth cone area for each growth cone observed at each time point ( and with = time interval) after background subtraction ( = 0 image) and the ratio of integrated DCF and CB fluorescence intensities for each growth cone at each time point was calculated ( with = time interval). Next, the average ratio of integrated DCF and CB fluorescence intensities of all growth cones observed at = 0 min (pre-stimulus) was calculated () followed by normalization of all ratios for at = 0 min (, = growth cone 1, 2, to for each condition). At least 15 growth cones were analyzed per condition from three different dissections to provide statistical significance (*).
2.10. ROS Quantification in Growth Cone Particle Preparations
Freshly prepared GCPs (100 μg in Kreb’s buffer) were loaded with 20 μM DCF (30 min, 4°C) in the presence of pharmacological reagents (10 μM DPI, 500 μM NAC), washed (14,000 ×), allowed to recover (10 min, 4°C) with pharmacological reagents present, and then exposed to 100 or 200 ng/mL TNFα (45 min, 4°C). For lysis, GCPs were resuspended in 2% SDS, 10 mM Tris-Cl pH 7.5, 10 mM NaF, 5 mM dithiothreitol, and 2 mM EGTA; sonicated; and cleared by centrifugation (14,000 ×, 5 min). Total DCF fluorescence intensity was measured (100 μL aliquots, black 96 well plates) using a Beckman Coulter Multimode DTX 880 microplate reader (495 nm excitation filter, 525 emission filter). All data were adjusted to total soluble protein concentration (BCA assay) and normalized to control condition to account for unspecific fluorescence and/or autofluorescence artifact. For all conditions, measurements were obtained in duplicates from three different GCP preparations.
2.11. Indirect Immunocytochemistry
Dissociated SC neurons were fixed (4% paraformaldehyde, 10 mM MES pH 6.1, 138 mM KCl, 3 mM MgCl2, 2 mM EGTA, 15 min, RT) followed by three washes with TBS (20 mM Tris-Cl pH 7.5, 150 mM NaCl). After blocking (30 min), cultures were incubated (2 h, RT) with primary antibodies (4 μg/mL, 2 mg/mL BSA in TBS) against cytokine receptors or MAP kinases, rinsed with TBS (3 times, 15 min each), and incubated with respective secondary antibodies. Cultures were transferred to 60% glycerol/PBS. Images were acquired under FITC illumination on a Radiance 2000 confocal microscope (Bio Rad).
2.12. Actin Filament Quantification
To visualize actin filaments, SC neurons were fixed and permeabilized (0.5% Triton X-100, 15 min). After rinsing (0.1% Triton X-100 in TBS), cultures were incubated (20 min) with rhodamine-conjugated phalloidin (1 : 10 in 1% Triton X-100 in TBS, Cytoskeleton Inc., Denver, CO), washed, and stored in 60% glycerol (4°C) until inspection. Images were acquired (40x oil, Plan Fluor) using a Zeiss LSM 510 confocal microscope equipped with a HeNe laser and an Argon laser. For each condition, 60 randomly selected SC neuron growth cones were scored for the presence of at least one large lamellipodia-like structure (three dissections, ) and the percentage of responding growth cones determined.
2.13. Plasma Membrane Translocation of
Freshly prepared GCPs (100 μg per sample) were treated with methyl-β-cyclodextrin (0.1%) or buffer (30 min, 4°C) and then exposed to 200 ng/mL TNFα (1 h, 4°C), whereas cultures of E10 forebrain neurons were subjected to 200 nM PMA (1 h, 37°C). GCPs (centrifugation) or forebrain neurons (scraping) were transferred into a 0.33 M sucrose buffer (20 mM Tris-HCL pH 8.0, 2 mM EDTA, 0.5 EGTA, 2 mM AEBSF, and 25 μg/mL Leupeptin), lysed by sonication, and centrifuged (25,000 ×, 15 min) to obtain an enriched plasma membrane fraction (pellet). Pellets were resuspended in sucrose buffer containing 1% Triton X-100 and 0.01% saponin (4°C, 15 min) or in 20 mM Tris-HCL pH 8.0, 2 mM EDTA, 0.5 mM EGTA, and 2 mM AEBSF with regard to gel electrophoresis or ELISA analysis, respectively. Suspensions were centrifuged (25,000 ×, 15 min, 4°C), membrane protein was recovered as the supernatant, and total protein concentration was determined. For ELISA, aliquots of plasma membrane protein (20 μg/mL, TBS/1% Triton-X-100) were incubated (12 h, 25°C) in 96-well high protein absorbent plates (Falcon). After blocking with BSA (5% w/v, TBS, 1% Trition-X-100, 1 h), wells were incubated (overnight, 4°C) with a polyclonal rabbit anti- (3 μg/mL), rinsed (TBS/1% Triton-X-100), and incubated with a goat anti-rabbit IgG antibody conjugated to HRP (1 : 2000, 45 min, 25°C). After three consecutive washes, wells were supplemented with 100 μL TMB (tetramethylbenzidine) and absorbance at 620 nm was measured 10 min later (Beckman Coulter Multimode DTX 880 microplate reader. All values were adjusted to total membrane protein and normalized to control conditions (duplicates, three independent experiments).
2.14. Gel Electrophoresis and Western Blotting
SC tissue, SC neuron cultures, or GDPs were solubilized in 2% SDS, 10 mM Tris-Cl pH 7.5, 10 mM NaF, 5 mM dithiothreitol, and 2 mM EGTA and sonicated, and a total soluble protein fraction was obtained by centrifugation (supernatant, 14,000 ×, 5 min). Fractions of total plasma membrane proteins were obtained as described above. Samples containing 20 μg of total soluble protein in Laemmli buffer were subjected to 10% SDS gel electrophoresis followed by western blotting onto PVDF membranes (Bio-Rad, Hercules, CA) [33, 34]. Membranes were blocked (2% nonfat dry milk, TBS pH 8.5, overnight, 4°C), washed (TBS/0.05% Tween 20/2% BSA), and incubated with polyclonal antibodies against cytokine receptor, MAP kinase, phospho MAP kinase, NOX2, , or (1 h, 4 μg/mL, in TBS/0.05% Tween 20/2% BSA). Membranes were washed (TBS/0.05% Tween 20/2% BSA) and incubated (1 h, RT) with the respective secondary, alkaline phosphatase-conjugated IgG (1 : 10,000, Sigma, St. Louis, MO) followed by colorimetric development (NBT/BCIP one step, Thermo Scientific, Rockford, IL). The specificity of antibodies was verified in lysates of HeLa cells, RAW264 cells, and SH-SY5Y cells (including blocking peptides).
2.15. Statistical Analysis
One-way ANOVA analysis and a Kruskal-Wallis test were employed for comparisons among multiple conditions. Dunnett’s -test was used when comparing the means of multiple conditions with a single control. All statistical values are given as SEM with a significance of unless indicated otherwise. Measurements were obtained in duplicates from 3 to 5 separate experiments unless stated otherwise.
3. Results
3.1. TNFα and IL-1β Paralyze Growth Cone Motility and Induce Growth Cone Collapse
Recent reports detailed that TNFα exerts morphogenetic functions (rearrangements of the actin cytoskeleton) without induction of apoptosis [7]. We examined whether TNFα or IL-1β has the potency to alter motility of neuronal growth cones, an actin-cytoskeleton driven mechanism. Live-video, phase contrast microscopy of advancing SC neuron growth cones revealed that an acute exposure to TNFα or IL-1β paralyzed growth cone motility and also caused the collapse of growth cone morphology accompanied by neurite beading and retraction compared to control (ovalbumin) (Figure 1(a)). To demonstrate a direct effect of cytokines on growth cones, we utilized polystyrene beads (4 μm in diameter) covalently coated with TNFα, IL-1β, or ovalbumin (control) to restrict contact between neurons and cytokines exclusively to advancing growth cones [35]. Growth cones encountering cytokine-coated beads displayed a series of stereotypic changes in behavior ultimately resulting in growth cone collapse (Figure 1(b)). After establishing long-lasting adhesion contacts with filopodia, growth cones entered a phase characterized by highly mobile, lamellipodia-like structures, which were nonproductive for advance, followed by a collapse of morphology (TNFα: %, and IL-1β: %, of observed growth cones, resp.). None of these growth cone responses occurred upon contact with ovalbumin-coated beads. Cytokine receptors were expressed on SC neurons and their processes (Figure 1(c)). Whole SC tissue extracts (western blotting) exhibited immunoreactivity against TNF receptor 1 (TNF-R1, apparent MW = 48 kDa), TNF receptor 2 (TNF-R2, apparent MW = 70 kDa), and IL-1β receptor type 1 (IL-1R, apparent MW = 76 kDa) in accordance with previous reports [36]. Expression of TNF-R1 and IL-1R was found on neuronal cell bodies, neurites, as well as growth cones; however, TNF-R2 expression was predominantly localized to neuronal cell bodies with virtually no expression on neurites or growth cones.
(a)
(b)
(c)
Next, we quantified growth cones advance (the spatial displacement of the growth cone/neurite boundary over time) in SC explant cultures upon acute exposure to TNFα, IL-1β, or ovalbumin (Ov, control) (Figure 2). Both TNFα and IL-1β caused a rapid, dose-dependent paralysis of growth cone advance and even neurite retraction within less than 10 minutes of addition compared to Ov (Figures 2(a) and 2(b)). Acute bath application of 100 ng/mL TNFα or IL-1β reduced the percentage of advancing growth cones by % (, ) compared to bath application of 10 μg/mL Ov. In contrast, a considerable fraction of growth cones resumed advance following bath application of 50 ng/mL TNFα or IL-1β (TNFα: 42 ± 14%, , and IL-1β: 30 ± 8% , ) however at much slower growth rates ( μm/h and 21 ± 5 μm/h, resp.) compared to growth rates prior to cytokines application (TNFα: 79 ± 8 μm/h, and IL-1β: 89 ± 9 μm/h, ). As expected, application of 10 μg/mL Ov had no effect on growth cone advance (before application: 108 ± 10 μm/h and after application: 113 ± 6 μm/h). According to reports and our own findings, we tested whether ROS intermediates were implicated in cytokine-mediated growth cone paralysis and collapse. Scavenging ROS with 10 μM MnTBAP or inhibition of NADPH oxidase (NOX) activities with 2 μM DPI both protected growth cone advance upon exposure to 100 ng/mL TNFα (%, and %, of advancing growth cones, resp.) albeit with decreased growth rates ( μm/h, 17% reduction and μm/h, 26% reduction, resp.) compared to 10 μg/mL Ov (Figure 2(c)). Growth cone advance was not affected by an addition of MnTBAP ( μm/h) or DPI ( μm/h) in the presence of ovalbumin during the period of observation. Growth cone advance in the presence of 100 ng/mL IL-1β was also rescued by 10 μM MnTBAP or 2 μM DPI (77 ± 12%, and 69 ± 6%, , resp.) (Figure 2(d)). Exogenous addition of ROS to growth cones (100 μM hydrogen peroxide) also reduced the percentage of advancing growth cones (%, , ).
(a)
(b)
(c)
(d)
(e)
(f)
Both cytokines also elicited a dose-dependent growth cone collapse quantified 30 min after addition as shown for TNFα (25 ng/mL: 16 ± 3%, ; 50 ng/mL: 59 ± 4%, *, ; 100 ng/mL: 71 ± 8%, , ) compared to 10 μg/mL Ov (%, ) (Figure 2(e)). Moreover, scavenging ROS (10 μM MnTBAP) or inhibiting NOX (2 μM DPI) (%, and %, , resp., **) protected growth cone morphology from the presence of 100 ng/mL TNFα as opposed to the absence of pharmacological agents (%, , *) (Figure 2(f)). Similarly, MnTBAP (31 ± 3%, ) and DPI (44 ± 1%, ) also negated IL-1β-induced growth cone collapse (69 ± 8%, , ). The percentage of collapsed growth cone remained unaltered by the presence of 10 μM MnTBAP (%, ), yet it did significantly increase with 2 μM DPI (%, ). Taken together, these data revealed that TNFα and IL-1β caused rapid degeneration of growth cone morphology and complete loss of motility through a redox-dependent mechanism. The rapid response of growth cones to cytokines together with the expression of cytokines receptors on growth cones strongly suggested a direct mechanism as opposed to indirect effects through neurodegenerative processes originating in neuronal cell bodies.
3.2. A Cytokine-Activated NADPH Oxidase Generates ROS in Growth Cones and Cell Bodies of SC Neurons
To determine the generation of intracellular ROS in advancing growth cones exposed to cytokines, we employed quantitative ratiometric fluorescence analysis utilizing the oxidation-sensitive fluorescent indicator 2′,7′-dihydrodichlorofluorescein (DCF, 10 μM) and the redox inert fluorescence indicator Calcein Blue (CB, 4 μM). Ratiometric analysis distinguishes changes in DCF fluorescence due to ROS formation from dilution/concentration effects simply due to rapid changes in growth cone morphology. Bath application of 100 ng/mL TNFα or 100 ng/mL IL-1β elicited a rise in the relative maximum DCF/CB fluorescence ratio indicative of the formation of ROS (Figure 3). As shown in Figure 3(a) (false colored DCF/CB ratio images), ROS formation was sustained and preceded loss of morphology of the advancing growth cone reflected by atrophy and beading of filopodia, condensation of growth cone body, and retraction. Quantitative analysis demonstrated a significant and sustained ROS formation in advancing growth cones upon exposure to TNFα (100 ng/mL, ), IL-1β (100 ng/mL, ), or 100 μM hydrogen peroxide (positive control, ) compared to 10 μg/mL ovalbumin (negative control, ) (Figure 3(b)). Neither loading with DCF ( μm/h, ) nor loading with CB ( μm/h, ) at the concentrations used affected growth cone advance per se compared to unloaded control ( μm/h).
(a)
(b)
To examine NOX-dependent ROS production in neuronal growth cones, we utilized freshly isolated growth cone particle preparations (GCPs) obtained from chick E10 forebrain neurons, which are capable of showing complex responses to extrinsic stimuli [37–39]. As shown in Figure 4(a), addition of 100 ng/mL TNFα to laminin-adherent GCPs resulted in MAP kinase and JNK activation, which was corroborated in cultured forebrain neurons. No basal MAP kinase activity (phosphor MAP kinase) was detectable in GCPs under control conditions possibly due a lack of growth factors (serum free plating conditions). Indicative for the formation of a functional NOX multiprotein complex, the NOX activator PMA (200 ng/mL) significantly increased the relative plasma membrane association of in chick forebrain neurons (Figure 4(b)). Moreover, native gel electrophoresis and NBT staining revealed several NADPH oxidoreductase activities in GCPs, one colocalizing with NOX2 immunoreactivity. The large membrane subunit NOX2 and the cytosolic subunits and were also detected in lysates of SC neurons (Figure 4(b)). Addition of TNFα (100 ng/mL) to DCF-loaded GCPs stimulated a significant, dose-dependent increase in ROS formation, which was negated by the NOX inhibitor DPI (5 μM) or the ROS scavenger NAC (1 mM) (Figure 4(c)). The small spatial dimensions of plated GCPs prevented ROS quantification utilizing microscopic imaging. Therefore, we resorted to a protocol involving lysis as previously established [14]. Moreover, plasma membrane association of in GCPs significantly increased in response to 100 ng/mL TNFα, which was abolished by the lipid raft disrupter methyl-β-cyclodextrin, suggesting that the assembly of NOX occurred predominantly in lipid rafts (Figure 4(d)). The insignificant increase of in plasma membrane fractions from methyl-β-cyclodextrin-treated cells could result from unspecific adsorption due to the dramatic change in membrane lipid composition. TNFα-coated polystyrene beads or bath application of 100 ng/mL TNFα to DCF-loaded SC neurons stimulated a transient increase in DCF fluorescence intensity in neuronal cell bodies (Figures 5(a) and 5(b)). Cytokine-stimulated ROS formation was dose-dependent with significant stimulation at concentrations higher than 50 ng/mL and saturation at 100 ng/mL (data not shown). Maximum ROS formation (, ) in SC neuron cell bodies occurred min following bath application of 100 ng/mL TNFα lasting for min ( cells per time interval) (Figure 5(c)). 100 ng/mL IL-1β also increased ROS formation (, ) within min upon application and lasted for min ( cells per time interval). Scavenging ROS with 2 mM NAC or 10 μM MnTBAP both suppressed TNFα-stimulated ROS formation (; and ; , resp.) to levels indistinguishable from control (10 μg/mL ovalbumin, , ) in contrast to TNFα alone (, , ) (Figure 5(d)). Inhibiting NOX activity (5 μM DPI) also negated ROS formation (; , ) upon exposure to TNFα. Similar data were obtained upon acute addition of 100 ng/mL IL-1β to SC neurons (data not shown). As our positive control, 200 μM hydrogen peroxide greatly increased relative maximum DCF fluorescence intensity (; , ), which was abolished in the presence of 10 μM MnTBAP (; ) (Figure 5(d)). Depleting Rac1 activity using adenoviral-mediated expression of FLAG-tagged, dominant-negative completely suppressed TNFα or IL-1β-stimulated ROS formation (, and , , resp.) as opposed to lacZ expression (TNFα: , , , and IL-1β: , , , resp.) (Figure 6(a)) [31]. In contrast, expression of constitutively active in SC neurons was sufficient to increase ROS formation (, , ) compared to 200 μM hydrogen peroxide (, , ) (Figures 6(b) and 6(c)). ROS scavenging with 2 mM NAC or 10 μM MnTBAP as well as NOX inhibition with 10 μM DPI abolished -stimulated ROS formation (, ; , ; and , , resp.). Similar results were obtained using purified recombinant GST chimeras of and transfected into E7 SC neurons by trituration loading (data not shown). Expressing alone (, ) or lacZ (, ), our control, had no effect on basal levels of ROS (Figure 6(c)). However addition of 2 mM NAC or 10 μM DPI to -expressing SC neurons both reduced ROS formation significantly below basal levels (, and , , resp., ). These findings revealed that TNFα and IL-1β stimulate a functional assembly of a Rac1-regulated NOX2 complex in the plasma membrane of neuronal growth cones resulting in a transient increase in ROS formation.
(a)
(b)
(c)
(d)
(a)
(b)
(c)
(d)
(a)
(b)
(c)
3.3. TNFα and IL-1β Attenuate Neurite Outgrowth of SC Neurons in a Redox-Dependent Manner
Next, we examined whether the redox-dependent impairment of growth cone advance is of consequence for neurite outgrowth. Bath application of 100 ng/mL TNFα or IL-1β attenuated neurite outgrowth in SC neuron cultures in a dose-dependent manner indicated by a shift in the distribution to shorter neurite lengths compared to 10 μg/mL ovalbumin (Ov), our control (Figures 7(a) and 7(b), Table 1). Scavenging ROS (10 μM MnTBAP) rescued neurite outgrowth in the presence of TNFα and IL-1β, whereas the NOX inhibitor (2 μM DPI) provided only partial protection of neurite outgrowth (Figures 7(c) and 7(d)). Notably, neurite outgrowth of SC neurons on laminin exhibited inherent redox dependence even in the absence of TNFα or IL-1β (Figures 7(e) and 7(f)). In accordance with previous reports, increasing concentrations of the ROS scavenger N-acetyl-L-cysteine (2 mM NAC, 50% reduction) or 20 μM MnTBAP (10% reduction) inhibited neurite outgrowth compared to control, whereas 10 μM MnTBAP was ineffective [15]. However, we measured a significant increase in neurite outgrowth with 5 μM MnTBAP (25% increase).
(a)
(b)
(c)
(d)
(e)
(f)
To assess the role of Rac1 on neurite outgrowth, (constitutively active Rac1) or (dominant negative Rac1) was introduced as purified recombinant GST chimera into freshly dissected SC neurons by trituration loading [26] (Figure 8). Depletion of Rac1 activity (-GST) in SC neurons protected neurite outgrowth upon addition of 100 ng/mL TNFα (NL50 = μm, , ) or 100 ng/mL IL-1β (NL50 = μm, , ) compared to GST-loaded SC neurons (NL50 = μm, and NL50 = μm, ), respectively (Figure 8(a)). However, introduction of -GST diminished neurite outgrowth even in the absence of adverse stimuli (NL50 = μm, , ) compared to GST-loaded SC neurons (NL50 = μm, ) in accordance with previous findings [26, 40]. Constitutive activation of Rac1 (-GST) in SC neurons also resulted in reduction of neurite length (NL50 = μm, , ) compared to GST-loaded SC neurons (NL50 = μm, ) (Figure 8(b)). Interestingly, addition of 5 μM MnTBAP restored neurite outgrowth of -GST-loaded SC neurons (NL50 = , , ) to levels indistinguishable from GST-loaded SC neurons (NL50 = μm, , ). This finding implied a role for Rac1 as a regulator of TNFα or IL-1β-stimulated oxidative stress. These findings provided evidence that attenuation of neurite outgrowth in the presence of TNFα or IL-1β required a persistent, Rac1-regulated formation of ROS. Moreover, neurite outgrowth on laminin exhibited an innate redox-sensitivity.
(a)
(b)
3.4. Cytokines Elicit a Redox-Dependent Reorganization of Actin Filaments in Neuronal Growth Cones
Earlier studies in neuroblastoma cells revealed oxidative damage to actin due to TNFα exposure [14]. We found that growth cones responded rapidly to cytokine exposure with changes in morphology prior to collapse suggesting reorganization of actin filaments. We quantified the percentage of growth cones responding to cytokines with at least one distinct actin filament-rich, lamellipodial structure as a function of both time and pharmacological treatment (Figure 9). Growth cones exhibited a rapid, yet transient increase in actin filament-rich structures within 15 min upon exposure to 100 ng/mL TNFα (65 ± 5%, , ) or 100 ng/mL IL-1β (84 ± 5%, , ) compared to control (23 ± 5%, , *) and persisted up to 30 min (TNFα: 67 ± 5%, , , and IL-1β: 58 ± 5%, , ) (Figures 9(a) and 9(d)). However, the percentage of growth cones with actin filament-rich structures greatly subsided 45 min after addition (TNFα: 40 ± 4%, , , and IL-1β: 37 ± 5%, , ) compared to the initial response. Inhibiting NOX activity (5 μM DPI) greatly diminished the percentage of responding growth cones upon bath application of cytokines (DPI + TNFα: 32 ± 6%, , and DPI + IL-1β: 49 ± 6%, , ) compared to absence of DPI (TNFα: 60 ± 6%, , , and IL-1β: 76 ± 6%, , ) measured 15 min after addition of cytokines (Figures 9(b) and 9(e)). The percentage of responding growth cones was not altered by DPI treatment alone (18 ± 6%, ) compared to control (23 ± 6%, ). Similarly, scavenging ROS with 20 μM MnTBAP also negated the formation of actin filament-rich structures in growth cones exposed to cytokines (Mn + TNFα: 24 ± 8%, , and Mn + IL-1β: 18 ± 8%, ) to levels indistinguishable from control (18 ± 4%, ) compared to cytokines alone (TNFα: 66 ± 8%, , , and IL-1β: 84 ± 8%, , ) (Figures 9(c) and 9(f)). A presence of MnTBAP did not alter the percentage of growth cones with actin filament-rich structures (%, ). Taken together, these data suggested that TNFα and IL-1β stimulated a redox-dependent, yet transient reorganization of actin filaments in neuronal growth cones prior to collapse of morphology.
4. Discussion
We demonstrated that long-term exposure of SC neurons to the proinflammatory cytokines TNFα and IL-1β provoked the loss of growth cone motility and the subsequent degeneration of growth cone morphology (collapse). These changes in growth cone advance and behavior translated into the impairment of neurite outgrowth and disruption of process architecture of SC neurons, which was rescued either by scavenging ROS, inhibiting NOX activity, or depleting Rac1 activity despite a presence of TNFα or IL-1β. Both cytokines stimulated the formation of ROS intermediates and a transient phase of actin filament organization in advancing growth cones. Importantly, ROS intermediates and actin filament reorganization preceded paralysis of growth cone motility and collapse of morphology implying a causative action of this signaling mechanism. Taking into account that exhaustive ROS scavenging (excess of antioxidants) also impaired neurite outgrowth, it is feasible that productive growth cone motility could demand an optimal level of ROS intermediates. We propose that a redox rheostat under the regulation of the Rac1 shapes growth cone motility and hence neurite outgrowth, in response to many extrinsic stimuli other than cytokines (Figure 10).
Prolonged bath application of TNFα or IL-1β (6–8 h) to dissociated SC neurons or SC explants stunted neurite outgrowth in a dose-dependent manner (Figure 7, Table 1) in accordance with previous findings in hippocampal neurons or neuroblastoma cells [8, 9]. In lieu of the well-documented apoptotic potency of TNFα and IL-1β, it was imperative to determine whether inhibition of neurite outgrowth had its origin in neuronal cell bodies (apoptosis, necrosis) or directly in the distal compartment of growth cones. Foremost, cytokines were applied to SC neurons after the onset of neurite outgrowth with the majority of processes longer than two cell diameters. Since the number of neurites per neuron (a measure of neurite initiation) remained unaltered and no measureable neuronal cell death occurred over the time period of cytokine exposure (6–8 h) as determined by a live/dead fluorescence assay (calcein green/propidium iodide counterstaining), neither a decrease in neurite initiation nor a decrease in neuronal cell death could account for the observed reduction in neurite outgrowth. Significant neuronal cell death (>15%) was however apparent 24 hours after cytokine addition and increased to over 40% after 2 days. Evidence for direct action of cytokines on growth cone motility and morphology was obtained by live-video phase microscopy (Figures 1 and 2). Acute exposure to TNFα or IL-1β (bath applied) impaired growth cone motility in a dose-dependent manner within 10 to 15 minutes upon application. Initial paralysis of growth cone advance was followed by degeneration of morphology and subsequent collapse. Restricting cytokine exposure exclusively to advancing growth cones in our bead assay corroborated these findings. In addition, bead assays revealed a transient increase in lamellipodia-like structures or in the vicinity of neuron-bead contacts, which were nonproductive for growth cone advance. This transient increase in actin filaments (Figure 9) was reminiscent of our findings in SH-SY5Y neuroblastoma, yet in the case of primary neurons it revealed a physiological consequence [14]. Direct effects of cytokines on growth cones were further supported by the expression of IL-R1 (IL-1β receptor) and TNF-R1 (high affinity p55 TNFα receptor) on the entire neuronal surface including cell bodies and growth cones (Figure 1). Together these findings provided strong evidence that TNFα and IL-1β directly impaired growth cone motility and morphology as the underlying cause for the attenuation of neurite outgrowth over longer time periods of exposure.
Initial evidence for a role of ROS intermediates was suggested by studies demonstrating that antioxidants (ROS scavenging) or DPI (inhibiting NOX-like activities) largely rescued neurite outgrowth, growth cone motility, growth cone morphology, and lastly actin filament organization in the presence of cytokines. Interestingly, neurite outgrowth of SC neurons on laminin exhibited innate redox dependence. Whereas moderate MnTBAP concentrations (5 μM) significantly increased neurite outgrowth, concentration higher than 10 μM stunted neurite outgrowth corroborating studies in Aplysia neurons [15]. It is plausible that moderate MnTBAP concentration (5 μM) scavenges excessive amounts of superoxide derived from mitochondrial respiration and as byproduct of several enzymatic reactions. Under basal culture conditions, this excessive superoxide might compromise optimal growth cone motility (actin dynamics, tubulin dynamics) and hence the observed enhancement of neurite outgrowth by moderate concentrations of scavenger [41–43]. In contrast, higher MnTBAP concentrations could impair vital redox-dependent mechanism necessary for proper neurite outgrowth [44]. Using the oxidation-sensitive fluorescence indicator DCF, we demonstrated substantial yet transient generation of ROS (likely superoxide) in growth cones and cell bodies of SC neurons upon exposure to TNFα and IL-1β (Figures 3 and 5). Similar results were obtained using the superoxide-specific fluorescence indicator dihydroethidium, yet investigations suffered from considerable neurotoxicity of dihydroethidium (data not shown). Previous studies in chick cortical neurons and human SH-SY5Y neuroblastoma cells demonstrated superoxide production in response to TNFα utilizing dihydroethidium oxidation or a SOD-inhibitable cytochrome C oxidation as demonstrated [14, 45]. A significant contribution from nonneuronal cells (microglia) was unlikely since (i) SC neuron cultures contained less than 3% nonneuronal cells, (ii) analyzed cells all exhibited distinct neuronal characteristics (i.e., round cell body, one process longer than 2 cell diameters), and (iii) microglia invade and populate the developing CNS at late embryonic or even postnatal stages [46]. Ratiometric imaging demonstrated significant ROS production in advancing growth cones within 3 to 5 min upon exposure to inflammatory stress as opposed to changes in motility and morphology observed 10 to 15 min following exposure. Several findings strongly suggested a NOX2-like activity as the source of ROS in SC neurons, in particular superoxide [14, 47]. SC neurons exhibited immunoreactivity against NOX2 and the cytosolic subunits and . A presence of other NOX isoforms can however not be excluded, yet antibody quality against NOX1 or NOX3 was insufficient to produce a conclusive finding. TNFα stimulated a dose-dependent increase in ROS intermediates in freshly isolated growth cone particle preparations in conjunction with a presence of NADPH oxidoreductases activity. The phorbol ester PMA as well as TNFα induced translocation of the cytosolic subunit to plasma membranes. Parallel studies on the translocation of the cytosolic subunit to plasma membranes were hampered by the considerable variability of immunoreactivity. Expression of NOX-like activities was reported in both primary PNS and CNS neurons [48–51]. Lastly, the antioxidant MnTBAP inhibits the TNFα or IL-1 β-stimulated ROS formation. Also, pharmacological inhibition with DPI was effective in blocking TNFα or IL-1β-mediated responses of SC neurons including ROS formation and partial rescue of growth motility, neurite outgrowth, growth cone collapse, and actin filament reorganization (Figures 2, 4, 5, 7, and 9). Notably, all these experiments revealed considerable toxicity of DPI and thus, depending on the duration of DPI application, put restrictions on the usable concentrations of DPI. Mechanistically, DPI blocks single electron transport reactions such as NOX-mediated generation of superoxide. Nevertheless, numerous other enzymatic reactions are also compromised foremost electron transport of complex I in mitochondria, which could account for the observed toxicity. Lastly, depleting Rac1 activity negated cytokine-mediated ROS formation in SC neurons, whereas Rac1 overexpression alone was sufficient to increase ROS formation. The small GTPase Rac1 emerged as a pivotal regulator of cytokine-stimulated ROS formation in SC neurons in accordance with previous reports [22, 52, 53]. In our study, depleting Rac1 activity abolished ROS formation in the presence of TNFα or IL-1β yet without affecting basal ROS levels (Figures 6(b)–6(d)). In contrast, overexpression of constitutively active Rac1 alone was sufficient to stimulated ROS generation in SC neurons and also reduced neurite outgrowth, which was not further exacerbated upon addition of TNFα or IL-1β (data not shown). Effects of Rac1 depletion or overexpression were addressed by trituration loading of purified recombinant Rac1-GST mutant chimeras since adenoviral-mediated expression of Rac1 mutants required a 3-day expression period incompatible with neurite outgrowth analysis [26, 31]. Not unexpectedly, introduction of or resulted in a measurable reduction in neurite outgrowth on laminin even in the absence of cytokines [11, 26, 54, 55]. In support of our findings, numerous reports link Rac1 as a regulator of ROS intermediates in cellular signal transduction. Neurite extension in PC12 cells upon NGF stimulation encompasses increases in Rac1 activity and H2O2 formation [56]. A Rac1 mutant lacking residues 124–135 of the insert region blocking ROS generation disrupted membrane ruffling and mitogenesis in fibroblasts and caused a downregulation of RhoA [44, 57–59]. With respect to actin cytoskeleton dynamics, TNFα and IL-1β induced a transient phase of actin filament reorganization (increases in actin filament density) in neuronal growth cones, which was however nonproductive for motility. Interestingly, actin filament reorganization was not repeatable with another addition of cytokines even after wash-out and extensive recover time suggesting permanent or persistent damage to actin filament reorganization by cytokines. Irreversible oxidative damage (carbonylation) to actin was detectable in neuroblastoma cells exposed to TNFα [14]. These findings provided evidence that NOX activation in neuronal growth cones serves as source of ROS intermediates in response to cytokines and that ROS intermediates are likely causative for cytokine-mediated degeneration of neuronal growth cones.
Inflammatory and oxidative stress is a hallmark of neurodegeneration in most chronic, acute, and even some psychiatric CNS pathologies [3, 60, 61]. The proinflammatory cytokines TNFα and IL-1β are important to orchestrate inflammatory and oxidative stress in the disease and aging CNS although pleiotropic actions in the adult and developing CNS are known, hence establishing a complex and delicate balance between neuroprotection and neurotoxicity. Members of the Nox/Duox family have emerged as [46, 62] key sources of oxidative stress in aging and the progression of many pathologies beyond the CNS and have been recognized as key therapeutic targets [63–66]. Our findings provide support for a role of NOX activity in primary neurons as a principal source of oxidative stress triggered by TNFα or IL-1β [16]. Overabundance of ROS intermediates is likely to disrupt proper actin filament dynamics, which is vital for the plasticity and morphology of distal neuronal processes including neurite outgrowth, sprouting, and spine dynamics. Strategies to block NOX activities could thus proof beneficial to blunt inflammatory stress in the diseased and aging CNS and to halt cognitive decline [66, 67].
Abbreviations
| DCF: | 2′,7′-dichlorofluorescein |
| DMEM: | Dulbecco’s Modified Eagle Medium |
| DPI: | Diphenylene iodonium |
| FBS: | Fetal bovine serum |
| GST: | Glutathione-S-transferase |
| IL-1β: | Interleukin-1 beta |
| MnTBAP: | manganese(III) tetrakis(4-benzoic acid) porphyrin chloride |
| NAC: | N-acetyl-L-cysteine |
| ROS: | Reactive oxygen species |
| SC: | Spinal cord |
| TBS: | Tris-buffered saline |
| TNFα: | Tumor necrosis factor alpha. |
Conflict of Interests
The authors declare that there is no conflict of interests regarding the publication of this paper.
Acknowledgments
I would like to extend my appreciation to Drs. Richard Bridges and Mike Kavanaugh (University of Montana) and Dr. Paul Letourneau (University of Minnesota, Minneapolis) for many helpful discussions that were vital in shaping this research. Thanks to Shellie Stewart-Smeets for her contribution on actin filament experiments and Dianne Cabrie for technical assistance. I am indebted to Dr. Mark Wright (University of Alaska, Fairbanks) for his support with native gel electrophoresis and Dr. Kelly Auer (University of Alaska, Fairbanks) for her support with MAP kinase analysis. Support for this work was provided by the Christopher Reeve Paralysis Foundation (KA1-0004-2), NIH/NINDS grant 1U54NS41069, and COBRE The University of Montana.