Mild in vitro trauma induces rapid Glur2 endocytosis, robustly augments calcium permeability and enhances susceptibility to secondary excitotoxic insult in cultured Purkinje cells

Abstract

Mild brain trauma results in a wide range of neurological symptoms that are not easily explained by the primary pathology. Purkinje neurons of the cerebellum are selectively vulnerable to brain trauma, including indirect remote trauma to the forebrain. This vulnerability manifests itself as a selective and delayed cell loss, for which the underlying mechanisms are poorly understood. Alterations to the surface expression of calcium impermeable AMPA receptors (GluR2-containing) may mediate post-traumatic calcium overload, and initiate biochemical cascades that ultimately cause progressive cell death. Our current study examined this hypothesis using an in vitro model of mild Purkinje trauma, delivered by an elastic stretch at 2.5–2.9 pounds per square inch (psi). This mild trauma alone did not increase cell loss as measured by propidium iodide (PI) uptake (at 20 h) compared to uninjured controls. However, there was a marked increase in cell loss, when cells following mild trauma, were exposed to 10 μM AMPA for 1 h compared to either mild trauma or AMPA exposure alone. Mild injury rendered Purkinje neurons significantly more permeable to AMPA-stimulated (4 μM) calcium influx at 15 min post-injury, including a sustained calcium plateau. This effect was eliminated by inhibiting protein kinase C-dependent GluR2 endocytosis with 2 µM Go6976 or blocking the calcium pore of GluR1/3 containing AMPARs with 500 nM 1-naphthylacetyl spermine (Naspm). Nifedipine (2 µM) eliminated the calcium plateau following mild injury but not the initial spike of Ca²⁺ increase. These results suggest that mild injuries resulted in a rapid AMPA receptor subtype switch (GluR2 was replaced by GluR1/3), which in turn resulted in an enhanced Ca²⁺ permeability. We further confirmed this by immunocytochemistry. Dendritic GluR2 co-localization with the pre-synaptic marker synaptophysin was markedly down-regulated at 15 min following mild stretch (P < 0.01), indicative of a rapid decrease in the synaptic expression of receptors containing this subunit. Carboxyfluorescence (CBF) assays revealed that mild stretch did not alter membrane integrity. Finally, we demonstrated that the combination of 500 nM Naspm and 5 nM Go6976 conferred a powerful neuroprotective effect on Purkinje cells by effectively eliminating the effects of mild stretch combined with AMPA in 95% of cells. These results represent a newly described mechanism rendering neurons susceptible to secondary injuries following trauma. Prevention of GluR2 endocytosis may be critical in the development of pharmacotherapies aimed at mild, seemingly inconsequential trauma, to avoid ensuing secondary damage.

GluR2, AMPA, stretch, traumatic brain injury, cerebellum, Purkinje cells, calcium, endocytosis

Introduction

In the developed world, traumatic brain injury (TBI) is a leading cause of morbidity and mortality, frequently resulting in debilitating cognitive, motor or autonomic CNS impairment (Luchter and Walz, 1995; Kersel et al., 2001; Blackman et al., 2004; Fujimoto et al., 2004). The majority of work implementing experimental models of neurotrauma has focused on hippocampal and cortical neuronal response to insult (for a review of neuroprotection and TBI, see Faden, 2002), with cerebellar injury not often considered when examining the anatomical distribution of damage to the brain following impact. Cerebellar damage should not be overlooked in trauma, as this area of the brain unquestionably plays a crucial role in motor coordination (reviewed by Middleton and Strick, 1998, 2000; Manni and Petrosini, 2004) and as evidenced more recently, higher neurological function (Middleton and Strick, 1994, 1998, 2001; Petrosini et al., 1998; Lalonde and Strazielle, 2003). Previous work from our laboratory (Park et al., 2006) has demonstrated a marked vulnerability of Purkinje cells to both mild and severe in vivo models of forebrain trauma. Posterior regions of the cerebellar vermis, the gyrus of the horizontal fissure and gyrus of lobules II and IV exhibit significant delayed Purkinje neuron cell death following cortical trauma (Park et al., 2006).

The formidable challenge in TBI research is that primary damage is essentially complete by the time any intervention is feasible. However, with adequate therapy, secondary vulnerability to subsequent insults or progressive degeneration over time could be prevented, particularly in mild trauma (Teasdale and Graham, 1998; Amar and Levy, 1999; Arundine et al., 2004). There is evidence from in vitro models of trauma to suggest that primary mechanical deformation to central cortical neurons in mild injuries affects the vulnerability of these cells to subsequent excitotoxic challenge (Arundine et al., 2003, 2004; Geddes-Klein et al., 2006). Despite the presence of preserved membrane integrity, viability and normal electrophysiological capacity, mild stretch injuries (also termed sublethal) can result in delayed overproduction of reactive oxygen species, mitochondrial dysfunction and increases in DNA fragmentation (Arundine et al., 2004; Geddes-Klein et al., 2006), all tell-tale signs of calcium overload in central neurons. As a result, further investigation is needed to examine alterations to intracellular signaling following mild trauma—and to elucidate how these changes contribute to enhanced calcium permeability and vulnerability to secondary excitotoxicity.

There is a plethora of evidence to suggest that adult Purkinje cells of the cerebellum do not express entirely functional NMDA receptors (i.e. receptors that open ion channels), attributed primarily to a developmental down-regulation of the NMDA glutamatergic response over the course of Purkinje cell development (Llano et al., 1988; Joels et al., 1989; Audinat et al., 1990). Though developing Purkinje neurons in slice cultures are thought to possess distinct, low-conductance, native NMDA receptors early in development (Momiyama et al., 1996), the functional sensitivity of these channels to both glutamate and NMDA is transient. That is, the response of Purkinje neurons to NMDA is markedly diminished in primary cultures by 14 days in vitro (DIV) (Yuzaki et al., 1996), and entirely eliminated by post-natal day 12 in slice cultures (Momiyama et al., 1996). Therefore, we hypothesize that calcium-permeable AMPA receptors (i.e. AMPARs lacking the GluR2 subunit, but containing the GluR1/3 subunits) may play a role in mediating post-injury calcium overload.

AMPA receptors are heterologous and assembled from a pool of four subunits—namely, GluR1-4 (reviewed by Groc and Choquet, 2006). Receptors containing the GluR2 subunit are calcium impermeable, due to RNA editing of the M2 transmembrane domain of this subunit. The M2 domain of the GluR2 subunit undergoes a glutamine (Q) to a charged arginine (R) residue switch at the channel pore region (position 607) that inhibits the passage of calcium ions through the receptor (Hume et al., 1991; Burnashev et al., 1992; Greger et al., 2002). This charged arginine is not observed in the mRNA of GluR1, GluR3 or GluR4 subunits. Thus, only receptors lacking the GluR2 subunit (e.g. heteromers of GluR1/3) are calcium permeable.

There is a large body of information detailing the expression of AMPAR subunit mRNA in Purkinje neurons. Using both in situ hybridization (Keinenen et al., 1990; Sommer et al., 1990; Monyer et al., 1991) and single cell RT-PCR (Lambolez et al., 1992; Tempia et al., 1996) a number of groups indeed identify the presence of all four AMPA receptor (i.e. GluR1-4) subunits in Purkinje neurons. Some evidence suggests that the majority of AMPA receptors on Purkinje cells are GluR2 containing receptors that are Ca²⁺ non-permeable (Linden et al., 1993; Tempia et al., 1996), while other groups report large calcium accumulation and excitotoxicity after AMPAR stimulation (Brorson et al., 1992, 1994; Sorimachi, 1993). Interestingly, some groups have demonstrated that the calcium permeability of cells in the cerebellum can be altered either by synaptic activities or pathological stimuli that change the composition of the receptors (Brorson et al., 1994, 1995; Liu and Cull-Candy, 2000). Accordingly, we conjectured that excitotoxic input following mechanical trauma is capable of changing the composition of synaptic AMPA receptors, thereby increasing their calcium permeability and perpetuating ensuing damage.

In supporting our hypothesis, recent work from Liu et al. (2006) suggests that mild in vitro ischaemia can alter GluR2 trafficking immediately following insult. Namely, mild oxygen-glucose deprivation (OGD) results in rapid GluR2 internalization by associating GluR2 with protein interacting with C kinase 1 (PICK1) and perturbing GluR2/AMPA binding protein (ABP) interactions in a PKCα dependent process. Subsequently, these cells underwent GluR1/3 exocytosis, thereby inserting calcium permeable receptors into the plasma membrane, in place of the impermeable GluR2 containing receptors (Liu et al., 2006). Given the common pathophysiology between ischaemic and mechanical cell death (see Arundine and Tymianski, 2003 for a review of excitotoxic mechanisms of cell death in ischaemia and trauma), we sought to examine whether there are similar alterations to the surface expression of GluR2 subunit containing AMPARs following mild in vitro trauma, and whether this results in an increased permeability to extracellular calcium, conferring a heightened susceptibility to subsequent glutamatergic challenge.

Accordingly, in the present study, we examined the secondary vulnerability of mildly stretched Purkinje neurons to excitotoxic insult. We have investigated changes to the surface expression of GluR2-containing AMPA receptors immediately following the mild injury via immunocytochemistry, and any concurrent changes to calcium permeability (evidenced through ratiometric Fura-2 analysis), controlling for changes in membrane integrity with a carboxyflourescein assay. Moreover, we sought to identify any neuroprotective effects of inhibiting PKCα-dependent clathrin-mediated GluR2 endocytosis, or blocking the calcium permeability of newly inserted GluR1/3 containing receptors. It was the goal of this work to identify a mechanism responsible for enhancing secondary Purkinje cell vulnerability, thereby providing insight into why this area of the brain exhibits profound delayed cell loss with mild cortical insults.

Materials and Methods

Isolation and dissociation of Purkinje cell cultures

All procedures described here were approved by the Animal Care Committee, St Michael's Hospital and complied with regulations of Canadian Council on Animal Care. Purkinje cell culture was performed as previously described (Slemmer et al., 2004). Cultures were prepared on 6-well BioFlex culture plates (FlexCell, Hillsborough, NC, USA). The bottom of the plate wells was made of flexible Silastic™ silicone substrates of 0.02 in. thick with a growth area of 57.75 cm². Wells were coated overnight with poly-l-lysine (5 μg/ml; Sigma). Purkinje cells were isolated from embryonic day 16–17 Wistar rats (Charles River Laboratories, Wilmington, MA, USA). Pregnant animals were anaesthetized with isofluorane and sacrificed via decapitation. Embryos were surgically removed, isolated from the amniotic sac and decapitated. Embryo heads were placed in 20 ml 1 × Hank's Balanced Salt Solution (HBSS, Invitrogen Corp. Carlsbad, CA, USA). Brains were removed and placed in a separate dish containing 20 ml supplemented HBSS. Cerebella were dissected from whole brains using microdissection forceps, and then were incubated in 2 ml of 0.1% trypsin (Sigma–Aldrich, St Louis, MO, USA) at 33–35°C for 10 min, and placed in 2 ml HBSS again. Tissue was triturated by glass pipette 10–20 times, and seeding medium Dulbecco's modified eagle's medium (DMEM) /F-12 containing 10% Horse serum, Invitrogen) was added. Purkinje cells were centrifuged for 5 min. Cell counts were done by loading phosphate-buffered saline (PBS), Typan Blue (Sigma–Aldrich) and 50 μl of cell suspension into a haemocytometer. Cell suspension was seeded in a plating medium (neurobasal medium containing 2% B-27 supplement, 1% fetal bovine serum, 0.5 mM l-glutamine, 25 μM glutamic acid and 100 nM l-thyroxine, Invitrogen) at a density of 1 × 10⁶ cells/well. After 96 h of isolation, cells were fed with fresh maintenance medium (neurobasal medium containing 2% B-27 supplement, 0.5 mM l-Glutamine, 100 nM l-thyroxine, Invitrogen) containing 10 μM FDU (5 mM uridine, 5 mM (+)-5-Fluor-2′-deoxyuridine, Invitrogen) and left to incubate for 48 h. Cells were fed with maintenance medium every 3–4 days until stretch assays. We used the cells for experiments 11–14 days after isolation to stay consistent with previous in vitro stretch assays (Zhang et al., 1996; Goforth et al., 1999; Weber et al., 1999; Arundine et al., 2003; Lusardi et al., 2004). Immunocytochemistry confirmed the presence of strictly neuronal, Purkinje cell cultures (Fig. 1).

Fig. 1

Validation of Purkinje cell culture system via immunocytochemical labelling using glial fibrillary acidic protein (GFAP) (A), calbindin D-28K (B), MAP2 (C,D) as markers for astrocytes, Purkinje cells and neurons, respectively. Note <5% prevalence of GFAP positive cells in established cultures (E) and ∼95% co-localization of MAP2/calbindin D-28K (F) suggesting the presence of predominantly Purkinje neurons. Scale bars = 200 μm.

In Vitro stretch protocol

Prior to stretch, the culture medium was replaced with 2 ml HEPES buffered Saline (concentrations in mM: 121 NaCl, 5 KCl, 20 glucose, 10 HEPES acid, 7 HEPES–Na salt, 3 NaHCO₃, 1 Na-pyruvate, 1.8 CaCl₂ and 0.01 glycine, adjusted to pH 7.4 with NaOH). Purkinje cell monolayer cultures were subjected to rapid stretch-induced injury as established by Ellis et al. (1995), using a cell injury controller II (Custom Design and Fabrication, Virginia Commonwealth University, Richmond, VA, USA). Our system is equipped with a feedback system that regulated the maximum pressure exerted on the plate bottom, minimizing the variation between trials. In brief, a controlled pulse of nitrogen gas induced a physiologically rapid deformation of the aforementioned BioFlex's Silastic™ bottom resulting in a quantifiable biaxial stretch of the Purkinje cells adhered to the Silastic™ surface, without detachment. Transient pulse duration was set to 50 ms and the stretch magnitudes ranged in pressure from 2.5–7.5 pounds per square inch (psi), representative of what would likely occur in humans after rotational acceleration/deceleration injury (Slemmer et al., 2002).

Secondary AMPA toxicity, determination of cell death and Go6976/NASPM neuroprotection assays

In order to establish the in vitro model, dose–response (injury–cell death) experiments were performed on the culture cells. Purkinje cells underwent varying levels of stretch (2.5–7.5 psi) in 2 ml HEPES buffer as described earlier. Wells were subsequently loaded with 10 μg/ml PI (warmed in 37°C water bath). The quantitative measurements of PI fluorescence were used as a determination of the prevalence of cell death using a Victor³V multiwell plate fluorescence scanner (PerkinElmer, Wellesley, MA, USA) controlled by Workout software (Dazdaq, Finland). All parameters including the size and number of scanning area, the duration of scanning, etc. were kept constant by using the same protocol for all groups. A second dye, fluorescein diacetate (FDA) was used as a marker of healthy, viable cells, as observed by us and others (McKinney et al., 1996; Weber et al., 1999; Pike et al., 2000; Zhao et al., 2000). It has been reported that damaged membranes lose their capacity to retain FDA, and thus will not fluoresce (Slemmer et al., 2002). In brief, immediately following stretch, baseline PI and FDA fluorescence readings were taken, cells were incubated at 37°C in the absence of CO₂ and a subsequent reading was taken 20 h later. Cell death along the continuum of mechanical deformation was normalized to unstretched wells exposed to 1 mM glutamate for 1 h (Glu). This exposure routinely produced nearly 100% cell death in our observations, and that of others (Bruno et al., 1994; David et al., 1996; Sattler et al., 1997; Arundine et al., 2003), and thus PI fluorescence for each condition was normalized to these wells. Cell death was calculated according to the formula: Fraction dead = F₂₀ − F₀/F_20GLU − F_0GLU, where F₂₀ = PI fluorescence 20 h post-stretch, F₀ = initial PI fluorescence, F_20GLU = PI fluorescence of cells 20 h post-exposure of 1 mM Glu for 1 h, F_0GLU = initial PI fluorescence of 1 mM Glu exposed wells. Cells exposed to 1 mM Glu were identical cultures from the same dissection.

To examine the vulnerability of mildly stretched cells to secondary AMPA toxicity, cells were subjected to mild stretch injury (2.5–2.9 psi), immediately followed by a challenge of 10 μM, or 20 μM AMPA for 60 min at 37°C (no CO₂). Wells were washed three times with HEPES buffer at the end of the incubation to remove AMPA, and PI readings were taken immediately and at 20 h post-AMPA challenge.

In our investigation into the neuroprotective effects of the PKCα inhibitor 12-(2-Cyanoethyl)-6,7,12,13-tetrahydro-13-methyl-5-oxo-5H-indolo(2,3-a)pyrrolo(3,4-c)-carbazole (Go6976, Calbiochem), or the GluR1/3 calcium pore blocker 1-naphthylacetyl spermine, (Naspm, Sigma) against secondary AMPA excitotoxicity of 20 μM, varying concentrations of drugs were loaded into the wells 30 min prior to mild stretch. Naspm concentrations tested were 100 nM, 300 nM and 500 nM, and Go6976 concentrations ranged from 1 nM to 500 nM. Drugs were diluted and dissolved in HEPES buffered saline, sterile filtered and warmed in 37°C water bath prior to bath application. All the toxicity assays were done in duplicate, and three or more dissections were used for each tested treatment.

[Ca2+] measurement

The membrane permeable calcium sensitive fluorescent dye, Fura-2AM (Molecular Probes, Eugene, OR, USA), was used to determine AMPA-induced alterations in intracellular calcium concentration in both mildly stretched neurons and matched controls. Purkinje cells were incubated with 5 µM Fura-2 AM for 40 min at 37°C, and then washed three times with HEPES buffered saline. Circular selections of membranes (0.75 in. diametre) were then removed from the well using a hole cutter, and placed in HEPES buffer. Cells were excited alternately at 340 and 380 nm at 1 s intervals, and an image from each excitation wavelength was captured using a high performance cooled CCD camera (Sensicam, Cooke, Eugene, OR, USA). In both conditions (15 min post-mild stretch and control) 20 pre-stimulation images were recorded (i.e. 20 s of baseline data was taken). AMPA of 4 µM was then applied via pipette pulse to the surrounding buffer solution, and calcium was monitored continuously for another 20–30 epochs. Glutamate of 50 μM was subsequently applied, to measure any immediate changes to calcium permeability in response. No changes in the responses to glutamate were observed in our initial five experiments. Given the sharp initial spike and subsequent plateau in intracellular free calcium in response to AMPA, it is likely that the ionic gradient would not favour influx of Ca²⁺ in response to glutamate immediately following AMPA pulse. Thus, glutamate stimulation was discontinued, with only AMPA induced calcium changes post-processed to calculate changes in permeability. Regions of interest were selected using Slidebook 4.0 software (Intelligent Imaging Innovations, Denver, CO, USA). All cells within the field of view were analysed for each experiment, with an average of 12 cells per reading (±2) and a range from 10 to 15. Three to four experimental wells from separate Purkinje cell culture were used for each experimental condition. The role of GluR2 endocytosis, and contribution of L-type voltage gated calcium channels was investigated through the application of various inhibitors, all loaded with Fura-2 AM (i.e. 40 min prior to stimulation). Protein kinase C inhibition (specifically PKCα) was accomplished with 2 μM Go6976 (Calbiochem). GluR1/3 containing AMPA receptors were blocked with the application of 2 μM Naspm trihidrochloride (Sigma–Aldrich), and voltage-gated calcium channels were antagonized with 3 μM Nifedipine (Sigma). Go6976 and Naspm were diluted to their final concentrations in sterile HEPES buffer; Nifedipine was first dissolved in 10–20 mM DMSO and further diluted to its final concentration in sterile HEPES buffer.

Calcium analysis

Normalized ratiometric values (340 nm/380 nm) representing the changes of intracellular calcium were calculated by taking the average ratio at each time point (i.e. 1–60, for each of the 12 cells, with the 20th epoch representing stimulation induced changes). Baseline calcium was then averaged as readings 1 through 19. Each epoch was normalized to the calculated average baseline and plotted. This was repeated for three to four wells in all conditions, with representative traces of the average normalized readings across all trials.

Detections of alterations to plasma membrane permeability

Plasma membrane permeability following mild mechanical stretch was assessed by evaluating uptake of the ordinarily impermeant fluorescent molecule, CBF (MW = 380 Da, radius = 0.5 nm; Sigma). We adopted this technique (established by Geddes-Klein et al., 2003, 2006) for use in stretch-induced alterations to cell permeability. The technique however, has also previously been implemented to detect permeability changes in electroporated cells (Bartoletti et al., 1989; Gift and Weaver, 2000). Immediately prior to injury, cells were treated with 100 μM CBF, and nuclei were stained with Hoechst 33 342 (20 μg/ml; Molecular Probes, Eugene, OR, USA). Purkinje cells were stretched in the presence of CBF and incubated at 37°C, 5% CO₂ for 10 min to maximize diffusion of the dye into cells (Geddes-Klein et al., 2006). Cells were then rinsed with buffer to ensure the removal of extracellular CBF. Sections of membranes were detached (0.75 in.), placed in HEPES buffer, and fluorescent images were taken from five different areas per section of membrane. Cells positively stained with CBF were later counted and normalized to the total number of Hoescht-positive nuclei. This procedure was repeated 3–4 times in each condition, across separate Purkinje cell isolations.

Immunocytochemistry and quantification of synaptic GluR2

In order to label total GluR2 expression, cells were fixed at 15 min post 2.9 psi in 4% paraformaldehyde for 10 min, washed 3 times for 5 min each with PBS, and blocked with 10% normal goat serum at room temperature for 1 h. Primary polyclonal rabbit anti-GluR2 (Chemicon, Temecula, CA, USA) antibodies (1 : 100) and primary monoclonal mouse anti-synaptophysin antibodies (1 : 300) were added and cells were incubated overnight at 4°C. Concurrently with primary antibody application, membranes were permeabilized with 0.3% Triton ×-100 (Chemicon) for 10 min. This allowed for all GluR2 expression to be accounted for, as unpermeabilized membranes would only have labelled surface receptor subunits. Cells were washed with PBS thoroughly, and secondary antibodies (anti-rabbit, Alexa Fluor 488-conjugated, 1 : 1000, and anti-mouse, 1 : 1000, Cy3, Invitrogen) were diluted and applied with 3% normal goat serum. Cells were incubated at room temperature for 1 h in the dark. To minimize non-specific reactivity, cells were washed thoroughly again, membranes were cut and removed and mounted on slides.

During our data acquisition and analysis, neurons were selected randomly under brightfield optics, and the investigator was blind to the treatment condition.

Florescence images were captured using a confocal microscope (Nikon Eclipse E100 equipped with a Radiance 2100, laser scanning system, Bio-Rad, Hercules, CA, USA) controlled by a software, Lasersharp 2000. To reduce the chance of photo bleaching, we used epi-fluorescence optics to locate and observe the area of interest. All parameters for capturing in each channel were kept constant among images (capture speed, laser intensity, number of passes and optical filtering). We used a 100× oil lens (digitally zoomed to 2.5× for dendritic processes). As an assessment of synaptic localization of GluR2 subunit, GluR2 and synaptophysin images of the same field were merged, and two to three dendrites per cell were captured and analyzed for yellow puncta. For each condition, we analysed fluorescent signal in regions of interest by taking the total area occupied by yellow puncta. Co-localized yellow masks were generated separately for each cell using Image Pro Plus 4.0 analysis software, and total area (in pixels) corresponding to each mask was calculated. Total puncta area was analysed for every 50 μm length of process, with dendrites not exceeding 2 μm in width. Identical procedures were followed to quantify total GluR2 and synaptophysin expression. We analysed 10 cells per well in each condition, and repeated this 3–4 times (across separate cultures) in both control and mildly stretched neurons.

Statistical analysis

Fura-2 data and toxicity data are presented as means ± standard error of the mean (SEM). Immunohistochemical, carboxyfluorescein, are all presented as means + SEM. Unpaired two-tailed Students t-tests or one-way analysis of variance (ANOVA) were used for statistical analysis.

Results

Stretch injury characterization and secondary excitotoxic vulnerability

To establish the in vitro model for Purkinje cell injury, we first validated the cell culture system using immunocytochemistry. We used MAP2, GFAP and calbindin D-28K as markers for neurons, astrocytes and Purkinje neurons, respectively. We found that in our optimized cultures, there were <5% GFAP positive cells. More than 95% overlapping of MAP2 and calbindin 28 k positive staining were found in the culture, suggesting the majority of the cells in our culture were Purkinje neurons (Fig. 1).

To validate our stretch paradigm as an experimental model of delayed Purkinje cell death, we examined cell death at 20 h post-injury along the continuum of stretch amplitudes as assessed by PI uptake. A robust gradient was observed in PI uptake from magnitudes ranging between 3.5 and 7.5 psi (e.g. cell death averaged 23.2 ± 2.45%, n = 3 for 3.5 psi versus 70.1 ± 6.5%, n = 3 for 7.5 psi, Fig. 2A). Each magnitude tested resulted in significantly greater PI uptake than the lower pressure tested (P < 0.05–0.001, Fig. 2A). However, stretch at 2.5 psi did not alter PI uptake relative to controls (P > 0.05). Both conditions averaged approximately 11.5 (±0.85% for control, ±1.73% for 2.5 psi, n = 3 for both conditions) of the PI uptake relative to wells treated with 1 mM Glu (Fig. 2A). Mildly stretched neurons stained brightly with FDA, whereas severely injured neurons did not (data not quantified, see Fig. 2B). This would suggest insult at 2.5 psi is rather mild and does not confer delayed cell death on its own.

Fig. 2

(A) Injury magnitude-propidium iodide (PI) uptake dose–response curve for stretch pressures ranging from 2.5 to 7.5 psi. The percentage of cell death was normalized to 1 mM glutamate exposed cells at 20 h post-stretch. Note the absence of increased cell death at 20 h in mildly stretched neurons (2.5 psi) as compared to control wells. (B) Representative FDA (green fluorescence, marker of viability) and PI (red fluorescence, marker of cell death) micrographs of mildly injured (B1, B2) and severely injured (B3, B4) Purkinje neurons. Scale Bars = 200 µm. (C) Despite the absence of cell death and retention of FDA, mildly injured Purkinje neurons are rendered increasingly vulnerable to secondary excitotoxic insult of either 10 µM AMPA or 20 µM AMPA immediately following injury, suggesting mild injury confers a propensity for delayed cell death when coupled with excitatory amino-acid (EAA) stimulation. *p < .05, **p <.01, ***p < .001.

Relative to both control neurons and 2.5 psi treated neurons, application of 10 μM AMPA to uninjured cells resulted in significant uptake of PI at 20 h (20.2 ± 2.85%, n = 3 versus 11.9 ± 0.85%, n = 3, P < 0.01, Fig. 2C). However, combining mild stretch with 10 μM AMPA induced markedly higher PI uptake relative to cells treated with AMPA alone (32.6 ± 2.32%, n = 3 for the combination versus 20.2 ± 2.85%, n = 3 for 10 μM AMPA alone, P < 0.01). Similarly, when exposed to 20 μM AMPA, the mildly injured cells had significantly greater PI uptake than non-injured cells (38.8 ± 1.71%, n = 3 for the combination versus 30.1 ± 2.4%, n = 3 for 20 μM AMPA alone, P < 0.05, Fig. 2C). Subsequent experiments were designed to identify the mechanisms underlying the increased vulnerability of Purkinje cells to AMPA toxicity following mild injury.

Calcium influx in response to AMPA stimulation following mild stretch

We next sought to measure the acute cellular calcium response to application of 4 μM AMPA both in control Purkinje cells, and those exposed to 2.5–2.9 psi, 15 min following injury. It should be noted that previous studies have determined that FlexCell Silastic™ membranes contribute no background fluorescence in acetoxymethyl ester (Fura-2AM) imaging experiments (Lusardi et al., 2004). Prior to AMPA application, we observed no detectable changes in free cytosolic calcium in response to control pulse of HEPES buffer in either mildly injured or control Purkinje cells (data not shown), demonstrating that the responses we later observed were in fact, ionotropic cellular responses, and not artifacts of the pipette pulse. Interestingly, we also observed no difference in the Fura-2 AM ratios of baseline emission at 340/380 nm between stretched and control neurons (respectively: 0.21 ± 0.04, n = 3 separate cultures and 36 total cells; 0.19 ± 0.04, n = 3 separate cultures and 33 total cells, P = 0.34) indicating that the mild mechanical deformation itself had no effect on cytosolic free Ca²⁺.

We did, however, observe remarkably different responses to application of 4 μM AMPA between mildly injured and control Purkinje cultures. AMPA application in control cells induced a small transient rise in cytosolic Ca²⁺, a response that quickly returned to baseline (Fig. 3A). Control cells appeared to demonstrate intact extrusion capacities, as evidenced by the transient nature of the increase in free calcium. Intracellular calcium, when normalized across all trials, increased on average to approximately 152.1 (±7.3%, n = 3) of baseline in response to 4 μM AMPA. Conversely, in mildly injured neurons 15 min post-insult, an identical application of 4 μM AMPA induced markedly greater Ca²⁺ influx, calculated to equal approximately 331.8 (±47.6%, n = 3) of baseline values (P < 0.001 as compared to the response in non-injured control cells, Fig. 3B). Interestingly, the acute calcium response of mildly injured neurons in response to AMPA also contained a plateau that was observed following the initial rise in free Ca²⁺ (Fig. 3B). This plateau (30th epoch used for statistical analysis) was steady at approximately 240.4 (±54.3%, n = 3) of baseline calcium levels (P < 0.01, Fig. 3D). The residual high Ca²⁺ levels are indicative of a stage of Ca²⁺ overload, which reflects either an impaired ability of the cells to extrude free calcium or an excessive amount of Ca²⁺ loading that exceeds the cell's extruding capacity. Either way, the overload of Ca²⁺ most likely was responsible for the lethality observed in the above 2.5 psi plus AMPA toxicity experiments.

Fig. 3

Mild stretch robustly augments AMPA induced calcium influx as evidenced by normalized Purkinje cell Fura-2AM ratios (340 nm/380 nm) following application of 4 µM AMPA. (A) Calcium influx is moderate, and quickly returns to baseline values in control cells. In the absence of mild stretch, exposure of the cells to the PKCα inhibitor Go6976, 1-napthylacetyl spermine (Naspm) or Nifedipine, has relatively benign effects on calcium influx (P > 0.05). (B) Fura-2AM ratios immediately following mild stretch (2.5–2.9 psi) in response to identical AMPA stimulation are markedly higher. Immediately following injury (∼15 min), calcium influx in response to AMPA is significantly greater relative to the response of uninjured controls (P < 0.001) and induces a plateau at ∼240% of baseline values. Baseline calcium did not differ between injured and control neurons, indicating no significant increase in calcium uptake following mild trauma on its own. Nifedipine eliminates the calcium plateau induced by 4 µM AMPA following mild stretch, while Naspm and Go6976 markedly decrease the immediate calcium response (P < 0.01). Error bars represent SEM, but are omitted in 3A and in drug treated conditions in 3B for clarity of the graphs. Please refer to the text for SEM values.

We sought to examine the contribution of L-type calcium channels to the AMPA stimulated increase in free Ca²⁺. The calcium plateau was Nifedipine-sensitive, suggesting that the initial depolarization caused by AMPA in mildly stretched neurons was large enough to activate voltage-gated calcium channels, in turn sustaining free intracellular Ca²⁺ at a high level after the initial influx through AMPA receptors. Nifedipine treated cells responded with a statistically significant initial increase in Ca²⁺ (∼262.1 ± 19.1%, n = 3, P < 0.01), however this increase rapidly returned to baseline levels in injured Purkinje neurons (i.e. no plateau was seen, Fig. 3B). While the initial large Ca²⁺ influx appears to reflect ionotropic AMPA channel activity, the increase in intracellular calcium to ∼ 330% of baseline, as well as the plateau, appear to both be the result of L-type channel activation.

We hypothesized that the endocytosis of GluR2-containing AMPARs was responsible for the initial increase in AMPA stimulated Ca²⁺ permeability, and sought to disrupt the cascade that leads to the internalization of GluR2 subunits. We were successful in reducing the calcium response of injured Purkinje cells to AMPA stimulation by inhibiting PKC-dependent clathrin mediated endocytosis of GluR2 with the use of 2 μM Go6976 (calcium response on average = 156.4 of baseline ±13.9%, n = 4, P < 0.01 compared to 2.5 psi) or by inhibiting the Ca²⁺ pore of GluR1/3 containing AMPA receptors with 2 μM Naspm (calcium response on average = 148.4 of baseline ±21.3%, n = 3, P < 0.01 compared to 2.5 psi). The Ca²⁺ influx returned to that of control cells in both conditions (Fig. 3B). Each drug had no measurable effects when used to treat control cells (2 μM Go6976 = 142.8 ± 11.6%, n = 3, P > 0.05 compared to control, 2 μM Naspm = 189.1 ± 17.2%, n = 3, P > 0.05 compared to control, 3 μM Nifedipine = 158.5 ± 12.2%, n = 3, P > 0.05 compared to control, Fig. 3A).

Immunocytochemistry—Mild stretch reduces co-localization of GluR2 containing AMPA receptors with the pre-synaptic marker synaptophysin

To examine the impact of mild mechanical deformation on synaptic localization of GluR2, we juxtaposed dendritic immunocytochemical labelling of GluR2 with the pre-synaptic marker synaptophysin. Synaptophysin is currently the most widely used marker of nerve terminals and has been used extensively to characterize glutamate subunit expression as synaptic (Burette et al., 2001; Ogoshi and Weiss, 2003; Liu et al., 2006). In control neurons, a large proportion of dendritic GluR2 clusters (Fig. 4A, green fluorescence) co-localized to synaptic sites with synaptophysin (Fig. 4B, red fluorescence), resulting in large yellow puncta (Fig. 4C). In contrast, as assessed 15 min following mild mechanical stretch, injury significantly reduced the total overlying yellow area, and presumably the number of dendritic GluR2-containing AMPARs (Figs 4F and 5B, n = 10 cells, repeated over three cultures, P < 0.01). Permeabilizing the membrane with 0.3% Triton ×-100 allowed us to quantify total GluR2 expression (including endocytosed protein) and total synaptophysin expression. There were no significant differences between total GluR2 protein expression (P > 0.05, Fig. 5A), or total synaptophysin expression (data not shown), suggesting mild insult does not effect GluR2 mRNA levels or transcriptional regulation of GluR2, but rather favours the hypothesis that the protein is internalized from synapses. Highly consistent with recent work examining GluR2 endocytosis following mild ischaemia (Liu et al., 2006), on average, ∼53.4% of control GluR2 was synaptic (±10.1%), versus 15.3% (±11.6%) in mildly stretched Purkinje cells (P < 0.001, Fig. 5C).

Fig. 4

Immunocytochemical labelling of control and injured neurons for GluR2 (A & D respectively), synaptophysin (B & E respectively), and the overlay of the two (C and F respectively). Mild stretch decreases dendritic GluR2 expression at synaptic sites in Purkinje cells. Representative images show the juxtaposition of GluR2 and the pre-synaptic marker synaptophysin in control and mildly injured neurons 15 min post-stretch. Higher-magnification images in Fig. 5 correspond to the boxed areas in the lower-magnification images seen here. Scale bars = 10 µm.

Fig. 5

(A) High-magnification images of dendritic processes of mildly injured (A1) and control (A2) Purkinje neurons. Mild injury significantly decreases the area occupied by yellow puncta on dendritic processes of Purkinje neurons 15 min following insult (Scale bar = 2 µm, P < 0.01). (B) Immunocytochemical quantification of GluR2/Synaptophysin dendritic co-localization. While total GluR2 expression is unchanged following injury, there is a robust decrease in the percentage of subunits that co-localize with synaptophysin. (C) Synaptic GluR2 expression expressed as a percentage of total GluR2 in control conditions and 15 min post 2.5 psi. **p <.01, ***p < .001.

Mild mechanical stretch does not alter membrane integrity as assessed by CBF uptake

There were no observable changes to neuronal morphology immediately following mild stretch, as evidenced through the differential interference contrast (DIC) micrographs in Fig. 6B. However, it is possible that stretched cells may exhibit enhanced calcium permeability in response to AMPA stimulation as a result of changes to membrane permeability. We sought to verify that free cytosolic Ca²⁺ was not rising as a result of entry through non-specific holes or tears in the membrane. Applications of CBF enabled us to image uptake of an ordinarily impermeable fluorescent molecule and as a result determine immediate, post-stretch alterations to plasma membrane permeability in mildly stretched Purkinje cells. Representative CBF micrographs are shown in Fig. 6C for control cultures (Fig. 6C1, C2 and C3), mildly stretched cultures (Fig. 6C4, C5 and C6, 2.5 psi) and severely stretched cultures (Fig. 6C7, C8 and C9, 6.5 psi). We observed almost no CBF uptake (denoted by bright green staining) in both control and mildly stretched cultures (quantified at ∼6.5 ± 1.31% and 5.6 ± 1.91% CBF positive neurons, respectively, n = 3 for both conditions, Fig. 6A). CBF uptake was significantly higher in severely stretched cultures (∼33.6 ± 4.03%, n = 3, P < 0.001, Fig. 6A). Thus, CBF uptake was a function of the pressure exerted on cultures, and did not increase in mildly stretched neurons relative to controls. It should be noted, however, that changes to permeability that would have occurred more than 10 min post-stretch would not have been accounted for. However, recent work by Geddes-Klein et al. (2006) suggests that plasma membrane permeability changes are transient and repaired rapidly following stretch.

Fig. 6

Mild injury does not increase non-specific Purkinje cell membrane permeability. (A) The percentage of carboxyfluorescein (CBF) positive cells normalized to Hoescht positive nuclei does not increase in mildly stretched Purkinje neurons (5%) relative to control cultures (6%), suggesting the preservation of membrane integrity. Severe injury does however, significantly increase CBF uptake (P < 0.001). Accordingly, no changes to Purkinje cell morphology are evident following insult at 2.9 psi (B1) relative to controls (B2). Scale bar = 20 µm. (C) Representative micrographs of CBF uptake and Hoescht staining in control, mildly injured and severely injured Purkinje cells. Calcium influx in response to AMPA following mild stretch does not appear to be the result of non-specific membrane holes or tears from mechanical deformation. Scale bar = 100 µm. ***p < .001.

Neuroprotection of Purkinje cells by Go6976 and Naspm

Finally, we tested the neuroprotective effects of the PKCα inhibitor Go6976 and the GluR1/3 calcium pore blocker Naspm against the secondary AMPA toxicity (20 μM) we observed initially following mild injury. We hypothesized that inhibiting PKC dependent clathrin-mediated GluR2 endocytosis with Go6976, or blocking the Ca²⁺ permeability of exocytosed GluR1/3 containing AMPARs would rescue cells otherwise destined to die from excitotoxicity. Fraction dead in Fig. 7A was calculated identically to initial toxicity protocols (see Materials and Methods section).

Fig. 7

(A) Combination therapy of PKCα inhibition and blockade of calcium permeable GluR1/3 containing AMPA receptors confers profound neuroprotection against delayed cell death caused by 20 µM AMPA and 2.9 psi insult. Cultures were pre-treated for 30 min prior to stretch. To inhibit PKCα and block GluR1/3 containing receptors 5 nM Go6976 and 500 nM Naspm were used, respectively. Cell death in preparations where both agents were applied was reduced to levels of control cells. (B) The lethal effects of 20 µM AMPA + 2.9 psi are eliminated in ∼95% of Purkinje neurons treated with the aforementioned cocktail. Drug application was repeated 3–4 times per condition, with each trial done on an independent culture. *p < .05, **p <.01, ***p < .001.

Fig. 7A illustrates the neuroprotective effects of the various doses of Naspm, tested and the only neuroprotective dose of Go6976 (5 nM). We were successful at reducing cell death to control levels, by combining the most effective doses of both drugs (i.e. 500 nM Naspm and 5 nM Go6976, Fig. 7A). Naspm doses (100 nM, 300 nM, 500 nM) on their own corresponded on average (n = 3 trials for each dose) to 34.7 ± 4.4%, 28.4 ± 3.1%, 23.1 ± 5.2% cell death, respectively, of which the latter two were significant reductions from untreated injured cells (41 ± 4.9%, P < .05 for 300 nM Naspm, P < .01 for 500 nM Naspm, Fig 7A). Remarkably, cells treated with the combination of 500 nM Naspm and 5 nM Go6976 demonstrated on average only 16.7 cell death (±3.3% n = 3, P < 0.001, Fig. 7A), which was at the same level as in non-injured controls (16.43 ± 4.5%,)

As a more effective way to present this data we sought also to represent our findings as the percentage of cells saved relative to injury. Given that control cells exhibited on average 16.4% cell death at 20 h, (Figs 2A and C and 7A), the maximum possible percentage of surviving cells relative to stretched neurons was 83.6% of the entire well. It was, therefore, necessary to normalize the cytoprotective data relative to the total surviving cells under control conditions. The percentage of cells saved was calculated as:

Using the above formula, on an average we were able to eliminate the lethal effects of mild stretch +20 μM AMPA in 95.2 (±3.3%) of cells by pre-treating them with 500 nM Naspm and 5 nM Go6976 (Fig. 7B). The percentages of cells saved under other treatments are summarized in Fig. 7B.

Discussion

The contribution of AMPA receptor trafficking to synaptic reorganization under pathological conditions is not well established, and represents an interesting hypothesis in the neurobiology of disease states. Through immunocytochemistry, we show that mild trauma, in the absence of conferring delayed cell death, reduces synaptic GluR2 as early as 15 min post-injury in cultured Purkinje neurons of the cerebellum, thereby robustly enhancing both calcium permeability and the susceptibility of these cells to secondary excitotoxic challenge. These changes occurred in the absence of any observable changes to morphology or damage to membrane integrity.

We demonstrate, here, as well the sensitivity of the calcium response of injured neurons to inhibitors of GluR1/3 containing AMPA receptors (Naspm), as well as Go6976, a PKCα inhibitor intended to disrupt clathrin-mediated GluR2 endocytosis. Interestingly, our data suggest that mild trauma renders Purkinje cells hyper-excitable when stimulated with exogenous transmitter, possibly allowing for the lethal, delayed activation of voltage gated calcium channels, inducing a Ca²⁺ plateau that overrides the extrusion capacities of these cells. We further identify the subsequent novel neuroprotective effects of both Go6976 and Naspm in our model of mild trauma coupled with 1 h of excitotoxic challenge, hypothesized to be related to the prevention of GluR2 subunit internalization after mild injury. Taken together, our findings implicate AMPAR trafficking in mediating post-trauma cerebellar neuropathology and identify potential targets of post-trauma therapeutics.

Relevance of cerebellar injury to forebrain TBI

There exists substantially less data on the pathophysiological changes to the cerebellum after forebrain TBI relative to cortical or hippocampal regions. However, clinical cases of forebrain TBI have noted diffuse cerebellar axonal injury, cerebellar atrophy and various degenerative changes as indicated by MRI (Gale et al., 1995; Soto-Ares et al., 2001; Gorrie et al., 2002) Accordingly, we propose that clinical manifestations of supratentorial TBI might be due, in part, to delayed changes at the cellular level in the cerebellum (including those that are unidentified by conventional imaging techniques), in addition to primary mechanical injury of the cerebrum. We have demonstrated experimentally that Purkinje cells display a marked delayed vulnerability to mild and severe forebrain head trauma in the rat (Park et al., 2006), and this data is supported by other groups who have noted a similar phenomenon (Fukuda et al., 1996; Mautes et al., 1996; Igarashi et al., 2007). Given that many studies have attributed higher-order cognitive processing independent of motor function to the cerebellum (Schmahmann, 1991; Joyal et al., 1996; Petrosini et al., 1998; Leggio et al., 2000), we propose that cerebellar injury or dysfunction after TBI could play an important role in the progression of neurologic and motor impairments, despite going unidentified during post-TBI diagnostics. Based on the data we have collected previously in our laboratory, and on the clinical findings of other groups, we suggest that the importance of cerebellar injury to the overall pathophysiology of supratentorial TBI needs to be addressed at both the anatomic and cellular levels. This study begins to address this need.

Calcium overload in Purkinje cells following mild trauma

Increases in intracellular calcium concentrations have been shown previously in a number of in vitro models of TBI (LaPlaca et al., 1997; Weber et al., 1999; Geddes and Cargill, 2001; Lusardi et al., 2004). However, our data are not suggestive of immediate trauma-induced changes to calcium concentrations, as baseline Fura-2AM ratios did not differ between mildly injured and control cells. Rather, our findings demonstrates a pathologically high calcium permeability in response to exogenous excitatory amino-acid (EAA) stimulation, and a possible form of pathological synaptic remodelling. With limited pre-synaptic neurotransmission in our in vitro preparation (i.e. the absence of climbing fibres, parallel fibres), exogenous AMPA-induced stimulation served to mimic in vivo afferents that would ordinarily connect to this population of cells in the intact brain.

It is certainly possible that the injured Purkinje cells display more depolarized membrane potentials upon a whole cell clamp, and we are currently investigating this possibility. However, because we have not done the electrophysiology in this study, we do not know if the injured cells are more depolarized immediately after the insult. We have seen in vivo that Purkinje neurons in slice preparations taken from injured animals have more depolarized resting potentials after injury relative to control neurons (Ai et al., 2007), but we do not know if this transfers to our primary culture/stretch injury preparation. From the data we have collected, we can suggest that the calcium response of these injured cells is, in part, mediated by the endocytosis of GluR2-containing receptors. The pharmacological manipulations we have employed inhibit the kinase responsible for GluR2 serine 880 phosphorylation, and it is well established that this modification is essential to the internalization of these receptors via PICK1 (reviewed by Groc and Choquet, 2006, in Cell and Tissue Research). Given that PKCa blockade significantly blunts the calcium response of the cultured cells to AMPA, we believe the phosphorylation (and therefore the internalization) of GluR2 plays a substantial role in the ionotropic response we are seeing. As well as evidenced by our immunostaining, there is in fact a high level of somatic GluR2 immunoreactivity, due to the large numbers of GluR2-containing extrasynaptic AMPARs present in Purkinje cells (Momiyama et al., 2003). Thus, a large contingent of the AMPA receptors being activated in our calcium experiments might be extrasynaptic. However, it is unlikely that this would influence the applicability of our findings to an intact brain preparation (where these receptors might receive considerably less activation), due to the evidence that extrasynaptic and synaptic AMPA receptors possess very similar electrophysiological properties [i.e. linear current–voltage relationships, as well as low single-channel conductances (Momiyama et al., 2003)].

This study, therefore, identifies a dysfunction of Purkinje cell post-synaptic responsivity following mild trauma. It is possible that our model of mild injury did not induce delayed cell death in Purkinje cells because of the absence of pre-synaptic fibres and associated EAA spillage that is responsible for excitotoxicity following traumatic brain injury (reviewed by Arundine and Tymianski, 2004). It should be noted that distal cortical trauma has been shown to result in mild cerebellar impact, (Fukuda et al., 1996; Mautes et al., 1996; Park et al., 2006). Our results suggest that this mild impact may induce Purkinje cell GluR2 endocytosis, and allow for EAA spillage from forebrain afferents to cause dangerously high levels of Purkinje cell cytosolic [Ca²⁺]. Forebrain trauma might induce intracellular modifications in areas as distal as the cerebellum, thereby priming this area of the brain for indirect delayed cell death mediated by excess [Ca²⁺].

Excitotoxic transmission from forebrain afferents to cerebellum following trauma

Indeed, we have previously demonstrated pathophysiological changes in the cerebellum in an animal model of forebrain TBI (fluid percussion injury) (Park et al., 2006), but the intracellular modifications responsible for these changes were not fully addressed. We observed increased calbindin expression in surviving cells, suggesting a neuroprotective role for calcium buffering capacities, but did not examine the possibility of forebrain-cerebellar synaptic remodelling. Coupling these in vitro findings to what we observed in animal models suggests that the cerebellum is highly vulnerable to even mild trauma induced by an impact that is primarily cortical. Thus, cerebellar synaptic modifications could play an important role in the progression of neurological and motor impairments in patients who suffer forebrain trauma. Our results suggest that preventing Purkinje cell GluR2 internalization through the use of PKCα inhibitors or blockade of GluR1/3 containing AMPARs may prevent delayed cerebellar neuronal death in animal models of cortical impact.

Pathological synaptic remodelling as a recurring theme

Previous work from our laboratory has shown other insult-induced changes to in vitro synaptic transmission. We recently demonstrated a form of pre-synaptic pathological LTP at CA3-CA1 synapses following a few minutes of anoxia and aglycaemia in hippocampal slices, attributed primarily to hyperexcitability of afferent fibres following ischaemia (Ai and Baker, 2006). However, any post-synaptic modifications that may have occurred were not accounted for. Very interestingly, recently published work from Liu et al. (2006) showed immediate GluR2 endocytosis in cultured hippocampal cells following brief, mild OGD. The results from our slice preparation (Ai and Baker, 2006) and the work presented here are highly consistent with Liu et al.'s observations, perhaps identifying a post-synaptic modification that perpetuates excitotoxic transmission in both ischaemia and trauma.

Our study demonstrates mechanically-induced modifications to neuronal AMPA responsivity and possibly highlights a form of pathological synaptic remodelling. There is a growing body of evidence that suggests Ca²⁺ permeable AMPA channels might be crucial contributors to neuronal injury in stroke, trauma and amyotrophic lateral sclerosis, particularly in cell populations that lack functional NMDA receptors (Brorson et al., 1994, 1995; Weiss and Sensi, 2000; Tanaka et al., 2005; Kwak and Weiss, 2006). In contributing to this evidence, our findings implicate AMPAR trafficking in mild cerebellar trauma, and identify concomitant changes to calcium permeability that may underlie delayed cerebellar cell death that worsens symptoms in patients who have suffered traumatic cortical impact. It is likely that the complexity and dispersion of neuronal damage following brain impact is vast, and is reflected in this vulnerability of Purkinje cells to secondary excitotoxic challenge following mild insult.

Conclusions

The present study shows that Purkinje cells of the cerebellum are left increasingly vulnerable to excitotoxic injury following mild mechanical deformation. While the nature of the insult does not induce structural or morphological abnormalities, we demonstrate changes to intracellular receptor endocytosis as early as 15 min post-injury; namely, that of the Ca²⁺-impermeable AMPA receptor subunit GluR2. Additionally, it appears as though L-type calcium channels trigger a lethal calcium plateau in response to AMPA stimulation only following mild deformations, and not in control conditions. Abolishing this plateau, or circumventing lethal calcium influx in response to AMPA through PKCα inhibition or blocking the calcium pore of GluR1/3 containing receptors confers a profound mode of neuroprotection in TBI research. The rapid alterations to AMPA receptor trafficking and pathological synaptic remodelling demonstrated here may underlie the susceptibility of cells of the cerebellum to delayed neuronal death that we and others have previously observed with cortical trauma (Fukuda et al., 1996; Mautes et al., 1996; Park et al., 2006).

Acknowledgements

Funding to pay the Open Access publication charges for this article was provided by the Cara Phelan Endowment, St. Michael's Hospital, Toronto.

Abbreviations:

Abbreviations:
 
• ABP

  AMPA binding protein

 
• CBF

  carboxyfluorescence

 
• EAA

  excitatory amino acid

 
• FDA

  fluorescein diacetate

 
• HBSS

  Hank's balanced salt solution

 
• Naspm

  1-naphthylacetyl spermine

 
• OGD

  oxygen–glucose deprivation

 
• PI

  propidium iodide

 
• PICK1

  protein interacting with C kinase 1

 
• psi

  per square inch

 
• TBI

  traumatic brain injury

Author notes

* These authors contributed equally to this work.