Abstract
Chronic ethanol exposure may induce neuroadaptive responses in N-methyl-d-aspartate (NMDA) receptors, which are thought to underlie a variety of alcohol-related brain disorders. Here, we demonstrate that hyperexcitability triggered by withdrawal from chronic ethanol exposure is associated with increases in both synaptic NMDA receptor expression and activation. Withdrawal from chronic ethanol exposure (75 mM ethanol, 5–9 days) elicited robust and prolonged epileptiform activity in CA1 pyramidal neurons from hippocampal explants, which was absolutely dependent upon NMDA receptor activation but independent of chronic inhibition of protein kinase A (PKA). Analysis of Sr2+-supported asynchronous NMDA receptor-mediated miniature excitatory postsynaptic currents (mEPSCs) was employed to assess changes in NMDA neurotransmission. After chronic exposure, ethanol withdrawal was associated with an increase in mEPSC amplitude 3.38-fold over that after withdrawal from acute ethanol exposure. Analysis of paired evoked α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid EPSCs and spontaneous mEPSCs indicated that withdrawal after chronic exposure was also associated with a selective increase in action potential evoked but not spontaneous transmitter release probability. Immunoblot analysis revealed significant increases in total NR1, NR2A, and NR2B subunit expression after chronic exposure and unaffected by PKA-inhibition manner. Confocal imaging studies indicate that increased NR1 subunit expression was associated with increased density of NR1 expression on dendrites in parallel with a selective increase in the size of NR1 puncta on dendritic spines. Therefore, neuroadaptation to chronic ethanol exposure in NMDA synaptic transmission is responsible for aberrant network excitability after withdrawal and results from changes in both postsynaptic function as well as presynaptic release.
Alterations in synaptic receptors probably underlie neuropathological sequelae due to chronic high level alcohol abuse, and such abuse is associated with a marked degree of abnormal neuronal activity culminating in autonomic, limbic, and cortical malfunctioning (Tabakoff and Hoffman, 1993). The NMDA receptor is a molecular target responsible for a variety of alcohol-related brain disorders due to its wide expression, documented role in seizures, neurotoxicity, and neurodegeneration (Nadler et al., 1990; Choi, 1994; Arundine and Tymianski, 2003) and its ethanol sensitivity (Hoffman et al., 1989; Lovinger et al., 1989). Alterations in NMDA receptor-mediated neurotransmission after chronic exposure are well documented. Receptor function is up-regulated (Chandler et al., 1993), and NMDA antagonists block hyperexcitability after withdrawal in vivo (Grant et al., 1990; Morrisett et al., 1990). Chronic ethanol administration in vivo increases the number of binding sites (Grant et al., 1990; Gulya et al., 1991; Snell et al., 1996) and induces region-specific increases in NR1/2A protein (Trevisan et al., 1994; Snell et al., 1996; Kalluri et al., 1998).
However, evidence suggests that in some models, changes in NMDA receptor function may occur in the absence of increased NMDA expression. Chandler reported increased NMDA-mediated NO synthesis in the absence of any change in [3H]MK-801 binding after chronic exposure in cortical neurons (Chandler et al., 1997) and also in cerebellar granule neurons (Cebere et al., 1999). These findings are suggestive of neuroadaptive pathways by which receptors are recruited to synaptic zones. Indeed, the synaptic localization of NR1 is increased by chronic exposure in cultured hippocampal neurons (Carpenter-Hyland et al., 2004).
Despite the large body of work on chronic ethanol modulation of NMDA function, definitive electrophysiological evidence identifying that alterations in NMDA-mediated synaptic transmission are necessary and sufficient for the induction of ethanol withdrawal hyperexcitability has not been forthcoming. Therefore, we sought to identify ethanol-induced synaptic alterations that give rise to aberrant neuronal activity after withdrawal in a physiologically intact neural network. We used an organotypic explant model (Muller et al., 1993) and had previously observed evidence of enhanced NMDA receptor function after chronic ethanol exposure and withdrawal (Thomas et al., 1998) and enhanced NMDA-mediated neurotoxicity after withdrawal (Thomas and Morrisett, 2000). Although these prior findings are consistent with the hypothesis that neuroadaptation of NMDA receptors is sufficient for seizure expression after withdrawal, they were primarily correlational and indirect. Therefore, the intent of the present study was 4-fold: to directly measure NMDA quantal synaptic responses before and after withdrawal to identify pre- or postsynaptic alterations in voltage clamp, to determine how receptor alterations temporally coincide with seizure expression, to determine whether WD seizures are dependent on NMDA receptor activation, and to relate these electrophysiological studies with biochemical and immunocytochemical measures of NMDA receptor quantity and synaptic localization. Thus, a mechanistic basis for NMDA receptor neuroadaptation in withdrawal hyperexcitability might be derived.
Materials and Methods
Explant Preparation. Hippocampal slices were prepared from 6- to 9-day-old Sprague-Dawley rat pups of both sexes using previously described techniques (Thomas et al., 1998; Thomas and Morrisett, 2000) and in accordance with National Institutes of Health and University of Texas Institutional Animal Use and Care Committee guidelines. In brief, brains were rapidly removed, cooled to 0 to 4°C for at least 3 min, and then 500-μm transverse sections were prepared and transferred to a holding chamber containing equilibration ACSF bubbled with 95% O2/5% CO2 and maintained at 32°C for 60 min before explanting. Equilibration ACSF consisted of 120 mM NaCl, 25 mM NaHCO3, 3.3 mM KCl, 1.2 mM NaH2PO4, 1.8 mM CaCl2, 2.4 mM MgSO4, and 10 mM dextrose. After equilibration, slices were transferred under sterile conditions onto Millicell-CM organotypic membrane inserts (generally, one to three slices per insert; Millipore, Bedford, MA) with media consisting of 75% minimal essential medium and 25% heat-inactivated horse serum supplemented with 3 mM glutamine (Invitrogen, Carlsbad, CA), 5.5 mg/ml glucose (Sigma Chemical Co., St. Louis, MO), and penicillin-streptomycin (100 U/ml and 100 μg/ml, respectively; Invitrogen). Explants were maintained in 5% CO2 at 37°C, and the media were changed 24 h after explanting and every other day thereafter.
Hippocampal neurons were not directly exposed to high levels of growth factors contained in 20% serum in vivo, and strong evidence suggests that such exposure alters normal patterns of dendritic sprouting and synaptic transmission (Tyler and Pozzo-Miller, 2003). Therefore, we reasoned such high levels of serum could alter neuronal network dynamics during ethanol exposure and withdrawal. For this reason, explants were weaned from serum-containing media over 3 sequential days starting on DIV 7. On the 1st day of weaning, minimal essential medium was replaced with Neurobasal media (Invitrogen), the serum level was decreased from 25 to 10%, and 2% B-27 serum-free supplement (Invitrogen) was added. On day 2, serum was reduced again to 5%. On day 3 and at all subsequent feedings, explants were fed with serum-free media containing 2% B-27. Concentrations of glutamine, glucose, and pen/strep were unchanged in serum-free media.
Chronic Ethanol Exposure. One day after serum weaning, explants were fed with media supplemented with 95% ethanol (Sigma) to a concentration of 35 or 75 mM and transferred to airtight plastic containers. To compensate for the equilibration of ethanol added to media with the airspace in the sealed chamber, 100 or 200 ml of double-distilled H2O containing 90 mM ethanol was added to the bottom of each chamber. This additional ethanol was empirically determined to equilibrate with the ethanol added directly to the culture media to maintain a stable concentration of ethanol in the media at 35 or 75 mM. Ethanol concentration was monitored at every media exchange via enzymatic assay (Pointe Scientific, Lincoln Park, MI). For patch-clamping experiments, explants were maintained in ethanol for ≥5 days before recording; maximal exposure duration was typically 9 to 10 days. The minimal threshold for the reliable generation of WD seizures was empirically determined to be 4 days of chronic exposure to ethanol. For Western blotting and immunocytochemistry, all explants were maintained in ethanol for 7 days before WD and subsequent harvesting.
WD Seizure Recording. Tight-seal whole-cell patch-clamp recordings were made at 30 to 32°C from CA1 pyramidal neurons. Recording electrodes (2–2.5 MΩ) were made from thin-walled borosilicate glass (TW150F-4; WPI, Sarasota, FL). The intracellular solution used for current-clamp recording contained 135 mM KMeSO4, 12 mM NaCl, 0.5 mM EGTA, 10 mM HEPES, 2 mM Mg2+-ATP, and 0.3 mM Tris-GTP, 280–290 mOsm, pH 7.2. Recording ACSF was identical to equilibration buffer except that the concentrations of CaCl2 and MgSO4 were changed to 2.0 and 0.9 mM, respectively. Explants remained attached to membrane inserts during recording, and care was taken to maintain the continuity of ethanol exposure until experimenter-imposed WD; all recordings were performed on an upright Olympus BX50WI microscope (Leeds Inst., Dallas, TX).
WD seizures were recorded in the absence of any pharmacological agents in current-clamp mode with no current injection. After at least 5 min of baseline recording, ethanol was washed from the recording chamber for a period of 7 to 8 min. If seizure activity did not occur within 10 min of the end of this washout period, a single 100-μs synaptic stimulus was delivered to the Schaeffer collateral pathway from a monopolar tungsten electrode (A-M Systems, Carlsborg, WA). As we have previously reported, hippocampal explants maintained in 75 mM ethanol for ≥5 days showed no gross anatomical or electrophysiological signs of neurotoxicity (Thomas and Morrisett, 2000). Cells were discarded if the Vm at break-in was <–50 mV or exhibited considerable variability, consistent with electrophysiological hyperexcitability. Of a total 26 cells in which stable break-in was achieved during the WD seizure component of the study, only two were discarded based on this criterion. In addition, extended recordings were made from ethanol-naive explants for comparison over similar recording durations and conditions. These experiments confirmed that control explants are electrophysiologically quiescent even with prolonged recording (see Results) as we have previously reported (Thomas et al., 1998). To test the effects of H-89 (10 μM; Sigma Chemical Co.) on WD hyperexcitability, explants were treated with ethanol for 1 to 2 days, then H-89 was introduced via the culture media for an additional 3 days before recording to reduce H-89 exposure time due to the tendency of chronic exposure to PKA inhibitors to promote apoptosis in some systems.
Miniature Excitatory Postsynaptic Current Recording. For voltage-clamp recording of NMDA and α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptor-mediated miniature excitatory postsynaptic currents (mEPSCs), the intracellular solution contained 135 mM CsF, 12 mM NaCl, 0.5 mM EGTA, 10 mM HEPES, 2 mM Mg2+-ATP, 0.3 mM Tris-GTP, and 20 mM QX-314 (Alomone Labs, Jerusalem, Israel), 280–290 mOsm, pH 7.2. For NMDA mEPSC recording, CaCl2 was substituted with 4 mM SrCl2, and MgSO4 was reduced to 0.6 mM to enhance NMDA mEPSC current amplitude as we have previously reported (Hendricson et al., 2004). Sr2+-supported NMDA mEPSCs were recorded in the presence of the AMPA/kainic acid antagonist, 6,7-dinitroquinoxaline-2,3(1H,4H)-dione (DNQX; 10 μM; Sigma Chemical Co.). Blockade of GABAA currents with picrotoxin (25 μM; Sigma) led to intermittent but persistent high-amplitude (>200 pA) contaminating currents, most probably arising from partial loss of voltage clamp due to increased input resistance following GABAA channel blockade (in our hands, this effect is unique to explants). For this reason, voltage-clamp recordings of NMDA mEPSCs were made in the absence of GABAA antagonists at membrane voltages that approximated the Cl– reversal potential (generally, –70 to –80 mV) to negate currents through GABAA channels. The pharmacological identity of currents arising under these conditions was confirmed by application of dl-2-amino-5-phosphonopentanoic acid [(DL)-APV; 100 μM; Tocris-Cookson, Ellisville, MO] after the conclusion of post-WD recording.
Synaptic stimuli were delivered to Schaeffer collateral fibers via monopolar tungsten electrodes (WPI). Constant-current pulses (100-μs duration, 10–40 μA amplitude) were applied through a stimulus isolation unit driven by an analog stimulator. Asynchronous events occurred within 300 to 500 ms of the synaptic stimulus and for several hundred milliseconds thereafter similar to those previously described in acute hippocampal slices (Hendricson et al., 2004). In a typical experiment, 20 to 25 stimuli were delivered at 30-s intervals in the presence of ethanol and repeated immediately following ethanol washout.
Non-NMDA mEPSCs, mediated largely by the activation of AMPA receptors, were pharmacologically isolated via the inclusion of (DL)-APV (100 μM) and were recorded in the presence of tetrodotoxin (TTX; 1 μM; Alomone Labs) to block action potential conduction. Contaminating GABAA currents were eliminated by recording at hyperpolarized membrane voltages as described above; this condition had the additional effect of enhancing AMPAR mEPSC amplitude. Analyses of NMDAR and AMPAR mEPSCs included several hundred events per cell.
Electrophysiological Data Acquisition and Analysis. Recordings were made using an Axopatch 200B amplifier (Axon Instruments, Foster City, CA), filtered at 1 kHz, digitized at 10 to 20 kHz via an Axon Digidata 1200 interface board, and stored on hard media for offline analysis under a Windows XP environment. Recordings in which access resistance changed more than 10% during the course of the experiment were discarded. Miniature EPSCs were detected using MiniAnalysis software (version 5.6; Jaejin Software, Leania, NJ) and minimal amplitude thresholds (typically, 5–7 pA for NMDA mEPSCs, 10–15 pA for AMPA mEPSCs, and 30–40 mV for spikes) were held constant within cells. In WD seizure experiments, high-frequency bursts were detected using the burst analysis component of MiniAnalysis. Bursts were defined as a group of three or more events in which the maximal interspike interval between any two events did not exceed 100 ms. The interpolated fit of burst intensity in Fig. 2 was generated using Origin version 7.5 software (OriginLab Corp., Northampton, MA). WD seizures were recorded in 10-min epochs; the mean spike frequency for a given cell was calculated from the average of these values. The statistical significance of changes in mean event amplitude, frequency, or burst frequency was analyzed throughout using via Student's t test (p values < 0.05 were considered statistically significant). Data are presented throughout as mean ± S.E.M.
Western Blot Analysis. After ethanol treatment and/or WD, explants were harvested and immediately snap-frozen in 1% SDS homogenization buffer. After subsequent thawing and sonication, protein concentration was determined via commercial assay (BioRad, Hercules, CA). Proteins were loaded at 10 μg into 7.5% (NR2B) or 12% (NR1) gels for separation via SDS-polyacrylamide gel electrophoresis, then transferred to nitrocellulose (100 V for 40 min). To verify equal loading of protein, membranes were stained with Ponceau S solution (Sigma), then washed twice with double-distilled H2O. After washing, membranes were blocked for 25 min in either phosphate-buffered saline (PBS) containing 3% nonfat dry milk (for NR2B) or PBS with 0.1% Tween 20 and 5% nonfat dry milk (TPBSM for NR1) then incubated overnight at 4°C in primary antibody against NR2B (2 μg/ml in 3% PBS containing 3% nonfat dry milk) or NR1 (0.8 μg/ml in 5% PBS with 0.1% Tween 20 and 5% nonfat dry milk; Upstate Biotechnology, Lake Placid, NY). After washing, membranes were incubated with horseradish peroxidase-conjugated goat anti-rabbit IgG (1:6000 for NR2B and 1:4000 for NR1; Bio-Rad) for 1.5 h, then repeatedly rinsed. Peroxidase activity was detected using the Super Signal Chemiluminescent kit (Pierce, Rockford, IL). Each ethanol-treated explant was paired with an ethanol-naive explant harvested at identical time points, and comparative data were derived from these matched controls for all experimental conditions. Each individual determination (or n) was made from a sample derived from two to three pooled explants treated concurrently, and each pool of explants for immunocytochemical analysis was prepared from two to three pups from a single litter. Most blots were run containing two separate samples pooled from control explants to control for variability in basal NR subunit expression between preparations; the average optical density value between these control samples was then used as the normalization value for that blot.
Expression of eGFP and NR1 Immunocytochemistry. The Sindbis virus used to induce the expression of eGFP in hippocampal neurons was generated by cleavage of the dopamine D1 receptor moiety from the Sindbis virus previously described using the restriction enzymes Xbal and MluI (New England Biolabs, Beverly, MA) (Diaz et al., 2004). Explants were injected at three to four sites in s pyramidale with eGFP-Sindbis via positive pressure ejection of concentrated viral stock (20–30 nl) from a patch pipette mounted on a microdispenser (VWR, Suwanee, GA). Explants were then rinsed twice with sterile PBS and placed in ethanol incubation chambers for 5 days (optimal expression of GFP was empirically determined to occur 3–5 days post injection; approximately one in four explants were discarded from subsequent analysis due to insufficient GFP fluorescence).
Explants were rapidly fixed (in the presence of ethanol for the chronic exposure group) while remaining attached to Millipore membranes by immersion in 4% paraformaldehyde/0.1 M sucrose PBS for 60 min. After fixation, explants were gently removed from the membrane inserts and incubated overnight at 4°C in 0.1 M glycine-supplemented PBS, permeabilized overnight in PBS supplemented with 0.2% Triton X-100 (Bio-Rad) for tissue. After blocking for 1 h in 7% goat serum, a primary antibody against NR1 (goat anti-mouse, 1:1000; Chemicon, Temecula, CA) was applied for 48 h at 4°C. Following 4 × 15-min rinses, an ALEXA-647 secondary antibody (goat anti-mouse, 1:500; Invitrogen) was applied for 1 h, and the slide was then rinsed, mounted, and the glass coverslip was sealed with Flouromount G (Southern Biotechnology Associates, Birmingham, AL).
Confocal Microscopy. Explants were imaged on an Olympus IX70 upright confocal laser scan head (Olympus FVX; Leeds Inst, Irving, TX) equipped with dual 488-/633-nm laser emission. After identification of regions of GFP-expressing CA1 pyramidal neurons under mercury lamp illumination at 20× magnification, a 60× oil immersion objective was engaged for fine resolution of discrete regions of dendritic fluorescence in the stratum radiatum layer of CA1. Dual laser line scans at standardized PMT gain and laser power settings were acquired with 7× digital magnification in Z-series format with 0.5-μm steps between planes. The ensuing Z-stacks, consisting of five to eight optical sections, were exported to a computer for offline analysis using the Metamorph analysis platform (version 6.2.6; Molecular Devices, Sunnyvale, CA). Care was taken to construct thin projection sections of dendrite to avoid spines in and out of the plane of focus. Regions of dendrite were selected for examination based on their spine density and intensity of GFP fluorescence such that spines were readily recognizable as conforming to accepted standards for hippocampal pyramidal neuron spine morphology and size.
Image Analysis. Image analysis was automated to the greatest degree possible using predefined operations in Metamorph. Acquired data stacks were repartitioned into green and red channels, then compressed into projection images. The green (GFP) image was thresholded such that all pixels above a minimal grayscale intensity were given equal weight; a transparent suprathreshold image outline was then generated from this threshold template and superimposed on the red channel (NR1) image. The red image was thresholded in a similar manner at a grayscale intensity level that was held constant across all images and treatment conditions. Image acquisition was standardized to yield segments of dendrite approximately 40 to 60 μm in length.
Measurements of the density and intensity of punctate NR1 immunoreactivity associated with dendrites were made using automated morphometry operations in Metamorph. Identification of spine- and nonspine-associated puncta was performed by a blinded observer to ethanol treatment condition; spines were identified on the transparent overlay using standardized size/shape criteria, and suprathreshold NR1 puncta were manually designated as spine/nonspine. Generally, a spine was defined as any protuberance that extended beyond the perimeter of the dendritic shaft more than 0.5 μm. In calculating the percentage of spine-associated NR1 puncta, the manually determined number of spine-associated puncta was divided by the total number of puncta within each segment of dendrite as determined by automated morphometry analysis. In calculating the percentage of NR1-labeled spines, the number of spines per segment of dendrite associated with ≥one NR1 puncta was divided by the total number of spines within each segment. For comparison of the size of spine- and nonspine-associated NR1 puncta, automated area measurements were made of all spine-associated puncta and a minimal 20 randomly selected nonspine-associated puncta per image segment.
Because each explant contained multiple infected neurons, several Z-series were taken from different neurons within each explant, and the mean values derived from these multiple determinations within each explant were compared with corresponding values from the opposite treatment condition by Student's t test. In total, 110 separate dendritic segments (40–60 μm) from 71 neurons from 21 separate explants (10 chronic ethanol, 11 sham) were processed identically and in parallel for this analysis.
Results
Ethanol Withdrawal Seizures in Hippocampal Explants. CA1 pyramidal neurons in explants chronically exposed to ethanol (35 or 75 mM, ≥5 DIV) recorded in the absence of any synaptic blockers and under current-clamp conditions exhibited stable membrane potentials, very limited spiking behavior, and were essentially quiescent aside from intermittent inhibitory postsynaptic potentials (Fig. 1A). The mean Vm at break-in was –62.5 ± 1.3 mV, and the mean action potential discharge rate was 0.02 ± 0.02 Hz (n = 25). Wash to ethanol-free after chronic exposure to 75 mM ethanol consistently elicited robust WD seizures exhibiting classic tonic-phasic discharge characteristics (Ayala et al., 1973; Thomas et al., 1998). Seizure onset after 75 mM ethanol withdrawal was associated with a paroxysmal depolarizing shift (PDS) in membrane voltage that was typically associated with a period of tonic high-frequency action potential discharge (Fig. 1B). Mean seizure duration (defined as the interval between PDS and the cessation of phasic discharge) was 15.2 ± 2.6 min after withdrawal, and recurrent ictal shifts in such cells were common; 17 individual seizures (defined as a PDS followed by a progression from tonic to phasic firing) were recorded from 10 cells in the 75 mM WD seizure group. Upon seizure initiation, the mean spike frequency increased 295-fold after withdrawal from 75 mM ethanol, from 0.02 ± 0.02 to 5.9 ± 0.8 Hz (Fig. 1C, p < 0.001, n = 10). The average per cell duration of WD seizure recording was 25.0 ± 5.7 min. After withdrawal from chronic exposure to 35 mM ethanol, spike frequency was enhanced (Fig. 1C, p < 0.03, n = 6) but primarily consisted of interictal type firing; thus, much less robust seizure phenomena were observed after withdrawal from 35 mM than that seen after withdrawal from 75 mM ethanol exposure.
Before 75 mM ethanol WD, all neurons (16 of 17) exhibited stable membrane characteristics at break-in and exhibited WD seizures after ethanol washout. Of these cells, 12 underwent a spontaneous PDS followed by ictal spiking within 5 min of postethanol washout; the remaining five cells underwent epileptogenesis after a single 100-μs constant-current stimulus to the Schaeffer collateral pathway delivered 5 min after ethanol completion of the ethanol washout period. In contrast, in 10 sham ethanol-treated explants, no evidence of hyperexcitability was observed even with prolonged recording; the mean AP discharge rate was 0.04 ± 0.04 Hz over the 10 min immediately after break in (Fig. 1C) and 0.05 ± 0.03 Hz after 30 min of recording in which synaptic stimuli were delivered at regular intervals to test for the spontaneous development of network hyperexcitability (data not shown). Therefore, in comparison with sham CA1 neurons, those neurons that had undergone WD displayed approximately 200-fold increase in spontaneous firing rate (Fig. 1C).
The progression of high-frequency discharge patterns through 30 min of 75 mM WD seizure recording was significant and resembled patterns of ictal discharges in vivo (Fig. 2). The mean amplitude of PDS-associated depolarization was 32.3 ± 1.5 mV. Typically following a PDS, the membrane voltage gradually repolarized to within 5 mV of pre-PDS levels, with a mean latency of 4.0 ± 0.7 min. Membrane repolarization was typically associated with the onset of phasic action potential bursts. These phasic bursts were brief (mean burst duration, 202.9 ± 27.2 ms), rapidly regenerating (mean interburst interval, 3.7 ± 1.6 s), and high frequency (mean intraburst spike frequency, 30.6 ± 2.3 Hz).
NMDA Receptor Activation Is Required for WD Seizure Expression. To test the hypothesis that WD epileptogenesis is dependent upon NMDA receptor activation, we applied the NMDAR antagonist (DL)-APV (100 μM) throughout pre-WD baseline recording as well as during ethanol washout and subsequent recording (Fig. 3). Single synaptic stimulation to CA1 neurons from normal ethanol-naive explants elicited only small excitatory postsynaptic potential complexes (Fig. 3A1); however, as described above, such stimulation following WD elicited a PDS with recurrent suprathreshold spikes (Fig. 3A2). However, when ethanol was withdrawn in the presence of (DL)-APV, PDS-spike complexes were completely absent in all cells tested (Fig. 3B1). Synaptic stimulation after subsequent removal of (DL)-APV from the recording solution elicited WD seizures in all cells tested (Fig. 3B2). The mean spike frequency of these seizures (4.0 ± 1.9 Hz, n = 4) was markedly elevated relative to the spike frequency of cells withdrawn in the presence of (DL)-APV (Fig. 3C; p < 0.05) but did not differ significantly from the mean spike frequency of cells withdrawn in the absence of APV (mean pre-WD spike frequency, 0.001 ± 0.001 Hz; mean post-WD frequency, 0.003 ± 0.001 Hz; n = 13, p = 0.2).
Protein Kinase A Is Not Required for WD Seizure Expression. A recent report in primary neuronal cultures prompted us to hypothesize that WD seizures may be due to PKA-dependent alterations in the synaptic localization of NMDA receptors (Carpenter-Hyland et al., 2004). To test for PKA dependence of WD seizure generation, we chronically exposed explants to a combination of 75 mM ethanol and the PKA inhibitor H-89 (10 μM). Chronic treatment with H-89 alone did not elicit seizures (Fig. 4A1). Chronic cotreatment of explants with ethanol in combination with H-89 had no effect on the initiation of WD seizures (Fig. 4, A2 and A3, n = 3). To verify that PKA inhibition was produced by H-89 application and maintained during chronic H-89 application, we monitored spike frequency adaptation of action potentials in response to a prolonged depolarization since PKA-dependent regulation of this process is well documented (Pedarzani and Storm, 1995). We first monitored adaptation in CA1 neurons from control explants before and after acute treatment with H-89, which was associated with a marked enhancement of spike adaptation (Fig. 4, B1 and B2). This acute effect of H-89 to enhance spike frequency adaptation persisted in explants chronically exposed to H-89 and ethanol (Fig. 4B3). H-89 cotreatment did significantly reduce seizure duration and intensity after ethanol WD. Mean action potential frequency in the H-89/WD was 2.5 ± 0.3 versus 5.9 ± 0.8 Hz in the WD alone group (Fig. 4C; p < 0.001). Mean seizure duration in H-89/WD explants was 6.3 ± 0.9 min as compared with 15.2 ± 2.6 min in the WD group (p < 0.01, data not shown). Mean high-frequency burst intensity was 20.6 ± 1.3 versus 30.6 ± 2.3 Hz in the WD alone group (p < 0.001, data not shown).
Withdrawal from Chronic Ethanol Exposure Markedly Enhances NMDA mEPSCs. To monitor changes in the strength of NMDA synaptic transmission associated with withdrawal from chronic ethanol exposure in the absence of network hyperexcitability, we studied Sr2+-supported NMDA mEPSCs recorded in the presence of the AMPA/KA antagonist, DNQX. We have previously reported Sr2+ substitution for analysis of acute ethanol effects on NMDA mEP-SCs in hippocampus as well as nucleus accumbens (Hendricson et al., 2004; Zhang et al., 2005). This technique promotes the analysis of quantal NMDA receptor-mediated synaptic transmission because the synaptic stimulation promotes event frequency while still permitting mEPSC analysis since those events arising in the asynchronous phase of Sr2+-supported release are nonaction potential dependent.
To control for ethanol exposure, we compared WD from chronic ethanol treatment with WD from acute (sham) exposure to ethanol (Fig. 5, A1 and A2, respectively). The mean amplitudes of NMDA mEPSCs before WD from chronic exposure (6.0 ± 0.7 pA, n = 9) and after acute exposure to ethanol for 25 to 30 min (7.2 ± 1.0 pA, n = 9) did not differ significantly (p = 0.3, data not shown). However, after ethanol washout, both chronic and acutely exposed groups of neurons displayed marked enhancement of both NMDA mEPSC amplitude and frequency (reduced interevent interval). Accordingly, we observed marked differences in the cumulative amplitude and frequency distributions of NMDA mEPSCs after WD from chronic exposure in comparison with those after acute ethanol WD (Fig. 5, B1 and B2, respectively).
Withdrawal from chronic ethanol exposure was associated with the most notable effects on NMDA mEPSC amplitude (Fig. 5C1). After ethanol washout, the mean mEPSC amplitude in chronic ethanol treatment (CET) explants increased by 163.3 ± 36.1% to 15.8 ± 1.6 pA. In comparison, an identical maneuver in acute ethanol (sham) explants elicited a significantly smaller 48.6 ± 25.1% increase to 10.7 ± 2.1 pA (Fig. 5C2; p < 0.005), corresponding to the removal of the acute inhibitory effect of ethanol on NMDAR currents (Hendricson et al., 2004). We have also previously demonstrated that ethanol modulates NMDA mEPSC frequency in hippocampus and nucleus accumbens (Hendricson et al., 2004; Zhang et al., 2005). To test whether WD from chronic ethanol exposure is associated with alterations in transmitter release probability, we also analyzed WD-associated changes in NMDA mEPSC interevent intervals (IEIs) in the same group of neurons as above. WD from chronic exposure was associated with a 71.8 ± 7.2% decrease in mean IEI. It is noteworthy that in sham ethanol explants, an identical maneuver was associated with a significantly smaller 29.0 ± 13.1% decrease in mean IEI (Fig. 5C2; p < 0.02 n = 9). Finally, to confirm the pharmacological identity of the currents under investigation, following recording of post-WD mEPSCs, all neurons cells were subsequently treated with ACSF containing (DL)-APV (100 μM), which completely antagonized the expression of NMDA mEPSCs (Fig. 5, A1 and C1/2).
AMPA-Mediated Synaptic Transmission and Chronic Ethanol Exposure. The finding that NMDA mEPSC frequency is enhanced after WD suggests that chronic ethanol exposure modifies glutamate release. To further investigate this possibility, we examined the effects of chronic exposure on action potential-dependent transmitter release using paired evoked AMPA EPSCs (Fig. 6A1). After ethanol washout, the baseline paired-pulse facilitation exhibited by paired AMPA EPSCs synaptically evoked at a 75-ms interstimulus interval decreased 38.1 ± 4% (Fig. 6A2; p < 0.001, n = 5). Taken together, the two preceding findings, an increase in the frequency of asynchronous NMDA mEPSCs evoked via synaptic stimulation and a decrease in the paired-pulse facilitation of AMPA EPSCs, suggests that chronic exposure affects the release of transmitter initiated by the propagation of an action potential into the presynaptic terminal. We next analyzed AMPA mEPSCs recorded in the presence of TTX to test whether basal, unstimulated glutamate release was altered after WD from chronic exposure (Fig. 6B1). Before ethanol washout, mean mEPSC amplitude was 14.2 ± 1.0 pA, and mean event frequency was 0.8 ± 0.2 Hz (n = 7). These values did not change significantly after ethanol washout, even with extended monitoring to 20 min post-WD (Fig. 6B2).
NMDA Receptor Immunoblot Analysis. The enhancement of NMDA mEPSC amplitude suggests that an increase in NMDA receptor protein may result from chronic exposure and thereby underlie WD seizure expression. Therefore, we analyzed NR1, NR2A, and NR2B subunit expression using immunoblot analysis (Fig. 7, A1/2 and B). All subunits examined were significantly elevated by chronic exposure to 75 mM ethanol by 18 to 31% over that detected in control sham ethanol-exposed explants (Fig. 7B; p < 0.001 by analysis of variance, F2,54 = 36.94, n = 5–6 samples of two to three explants pooled for each condition, post hoc tests; Games-Howell and Dunnett C). Chronic exposure to 35 mM ethanol did not seem to affect NR1 or NR2A/B subunit expression. However, chronic coincubation of explants with 75 mM ethanol and the PKA antagonist, H-89 (as in Fig. 4), produced a similar enhancement of NR subunit expression seen after exposure to 75 mM ethanol alone.
Chronic Ethanol Exposure Increases NR1 Containing Synaptic Puncta. To examine the relationship between ethanol-induced changes in the expression of NMDA subunits and the extent of delivery of these proteins to synaptic and extrasynaptic sites on CA1 pyramidal cell dendrites, we expressed eGFP in explant neurons via infection with a Sindbis-eGFP viral vector. Explants were then treated chronically with ethanol via the standard protocol, then fixed and NR1 immunolabeled with an ALEXA-647-conjugated secondary antibody. Dendritic areas of eGFP and ALEXA-647-NR1 fluorescence were detected using confocal microscopy with simultaneous 488-/633-nm laser excitation, respectively, thresholded by an automated imaging process, and NR1 fluorescent puncta were defined by a blinded observer as residing on the dendritic spine or shaft.
The percentage of total dendritic area exhibiting NR1 immunofluorescence was employed as a measure of ethanol-induced changes in total dendritic NR1 content. Chronic ethanol exposure produced a 331.3 ± 49.3% increase in total dendritic area exhibiting NR1 immunoreactivity (including both spine- and shaft-associated NR1; Fig. 8, C1; 3.2 ± 4.4% versus 13.8 ± 3.3%, p < 0.02; n = 10). In the same context, the density of punctate NR1 immunoreactivity per unit dendritic area also increased by 200 ± 33% (Fig. 8C2; 1.0 ± 0.4 puncta/1 E3 pixel area units versus 3.0 ± 0.8 puncta/1 E3 pixel area units, p < 0.001). To test the extent to which chronic ethanol exposure induced preferential distribution of NMDA receptor subunits to synaptic sites, patterns of NR1 localization to dendritic spines were compared in CET versus sham explants. The percentage of spines that were labeled with ≥1 NR1 puncta did not change after CET (Fig. 8C3; 43.1 ± 8.6% in sham slices versus 47.2 ± 11.9% in CET slices, p = 0.8). However, the percentage of spine area containing associated NR1 puncta was significantly greater after chronic exposure relative to sham controls (Fig. 8C4; 53.0 ± 6.8% in sham explants versus 109.3 ± 23.8% after chronic exposure, p < 0.04). The mean area of nonspine-associated puncta did not increase significantly (p = 0.1; data not shown).
Discussion
Electrophysiological Analyses. We observed electrographic seizures in CA1 pyramidal neurons of hippocampal explants after withdrawal from chronic high-level ethanol exposure. All neurons that underwent chronic exposure displayed hyperexcitability after withdrawal spontaneously or after a single stimulus. The morphology of these events was reminiscent of epileptiform activity seen in hippocampal and other limbic structures because they were robust, long-lasting with tonic-clonic burst activity (Goddard et al., 1969). The concentration-response relationship observed for hyperexcitability following 35 and 75 mM ethanol exposure is consistent with previous reports (Thomas et al., 1998) (Thomas and Morrisett, 2000). When ethanol withdrawal was performed when NMDA receptors were blocked, seizures did not occur, indicating that withdrawal hyperexcitability seems dependent upon NMDA receptor activation. Thus, NMDA receptor activation may be necessary for seizure expression, yet we cannot determine whether this alone is sufficient for the expression of this pathology. Thus, NMDA receptors may be required for network activity and NMDA receptor activity may drive network firing that also contributes to seizures.
The chronic ethanol-induced enhancement of NMDA receptor function may be dependent upon activation of protein kinase A (Carpenter-Hyland et al., 2004) and, thus, may underlie withdrawal hyperexcitability. Therefore, explants were treated with the PKA antagonist H-89; however, this treatment did not prevent WD seizures. The activity of this compound was confirmed by measuring spike frequency adaptation (Pedarzani and Storm, 1995). WD seizures did display slightly reduced intensity and duration after H-89 cotreatment, and yet an increase in adaptation persisted after chronic H-89 treatment. Therefore, attenuation in spiking could result from H-89 directly and may be responsible for the reduction in seizure intensity and duration. These data suggest that blockade of PKA does not modify the onset of seizures.
To assess the synaptic activation of NMDA receptors, we measured quantal asynchronous NMDA mEPSCs in the presence of Sr2+ evoked in the presence of Mg2+ (0.6 mM) (Morrisett et al., 1991; Zhang et al., 2005). We previously reported that acute exposure to ethanol has dose-related effects on the frequency, amplitude, and decay kinetics of Sr2+-supported NMDA mEPSCs, indicating distinct presynaptic and postsynaptic actions of ethanol on NMDA synaptic transmission (Hendricson et al., 2004). Before ethanol withdrawal from 75 mM exposure, there were no differences in mEPSC frequency or amplitude between acute ethanol-exposed explants and those chronically exposed. After withdrawal from chronic exposure (versus after acute exposure), enhancement in both these measures was markedly enhanced; thus, the increase in mEPSC amplitude after chronic exposure was 3-fold greater than that observed after acute exposure, and NMDA mEPSC frequency exhibited a 2-fold enhancement after withdrawal. These differences following WD from chronic exposure versus those we reported in acute slices indicate that a robust synaptic neuroadaptive response by NMDA receptors occurs during chronic ethanol exposure (Hendricson et al., 2004). This adaptation has functional consequences on NMDA synaptic transmission involving both presynaptic and postsynaptic components. In addition, this increase in current amplitudes was independent of WD hyperexcitability because it occurred under voltage-clamp conditions and in the presence of DNQX, which has a potent antiepileptic effect via blockade of fast excitatory neurotransmission (Fountain et al., 1998). Finally, the temporal occurrence of these alterations in NMDA mEPSCs following WD was immediate upon ethanol washout and in concordance with the expression of withdrawal hyperexcitability.
Presynaptic function after chronic exposure was further analyzed and also in light of recent reports from our lab suggesting that ethanol exerts an acute inhibitory effect on evoked release (Hendricson et al., 2004; Maldve et al., 2004) while sparing unstimulated release (Hendricson et al., 2003). We examined the effects of WD on synaptic events arising from two modes of transmitter release: action potential independent (i.e., AMPA mEPSCs recorded in TTX) and action potential evoked (i.e., macroscopic AMPA EPSCs). These experiments indicate that a selective enhancement of spike-dependent release after WD from chronic exposure occurred and implicate ethanol-sensitive components of spike-dependent release in the pathophysiology of WD. It is interesting to note that voltage-gated calcium channel antagonists reduce behavioral WD manifestations after chronic exposure (Little et al., 1986; Wu et al., 1987; Whittington and Little, 1991), and ethanol is also well known to directly block VGCCs (Harris and Hood, 1980; Hendricson et al., 2003); in particular, N and P/Q-type channels that are known to subserve transmitter release in multiple brain preparations (Woodward et al., 1990; Wang et al., 1991; Maldve et al., 2004).
Biochemical and Imaging Studies. Biochemical mechanisms for chronic ethanol-induced alterations were measured via Western blot and immunocytochemical analysis. Western blot analysis of NR1/2 expression demonstrated that production of NMDA proteins was significantly elevated after 75 mM chronic exposure and before withdrawal. These data support and extend our earlier in situ hybridization study (Thomas et al., 1998) and provide evidence that increases in NMDA receptor subunit expression precede WD hyperexcitability. Chandler et al. (1997) have made the important observation in cultured neurons that enhancement of NMDA receptor function are separate from any apparent increase in receptor protein level. Spanagel and colleagues reported increases in hippocampal NR1 splice variant protein levels independent of mRNA changes (Winkler et al., 1999). Valenzuela and colleagues performed a thorough analysis of hippocampal glutamate receptor protein expression after a chronic in vivo alcohol exposure paradigm that produced approximately 10 to 14 g of ethanol/kg/day intake (Ferreira et al., 2001). This level of intake was sufficient to induce withdrawal hyperexcitability, but no changes in total ionotropic Glu receptor levels were observed when animals were sacrificed at the peak of ethanol intake (∼20–50 mM BAL). We contend that the reduced level of ethanol exposure attained via voluntary intake in this study was most probably insufficient to produce changes in total levels of NMDA receptor expression. Taken together, these studies and our findings suggest that ethanol exposure above 50 mM is required to observe increases total NR1/2 subunit expression. Such a contention is supported by the observation that total NR subunit levels were not altered after chronic exposure to 35 mM ethanol. Likewise, marginal hyperexcitability was observed after exposure to this lower concentration of ethanol; thus, we suggest that the synaptic alterations of NMDA receptors probably play a critical role in the expression of ethanol withdrawal hyperexcitability. Finally, we did not observe significant differences in NR1 subunit expression after exposure of explants to 75 mM ethanol in the presence of the PKA antagonist, H-89; thus, in an intact explant preparation, the enhancement in NMDA subunit expression, as well as withdrawal hyperexcitability, due to high-level ethanol exposure seems to be independent of PKA activation.
Analysis of chronic ethanol effects on dendritic NR1 immunofluorescence revealed increased density of NR1 puncta, in good agreement with the present Western blot data. Although chronic exposure did not increase the overall percentage of NR1-labeled spines on CA1 dendrites, the mean area of spine-associated NR1 puncta was significantly increased, as was the total dendritic area that stained for NR1. Conversely, chronic ethanol treatment did not seem to significantly increase the size of individual extraspine NR1 puncta. This finding is indicative of chronic ethanol-induced recruitment of NMDA receptors specifically to dendritic spines with a generalized increase in NMDA subunit expression across the dendritic structure. These data are similar to those of Chandler and colleagues, who noted an increase in synaptic NMDA receptors and not the size of extraspine NMDA receptors after several days of incubation of dissociated hippocampal neurons in 50 mM ethanol (Carpenter-Hyland et al., 2004). We both also observed an increase in the size of NR1-associated synaptic puncta, with no change in the size of those extrasynaptic puncta. Furthermore, the total amount of extraspine NR1 was significantly increased in our hands after exposure to 75 mM; as discussed above, there seems to be a significant dose relationship in the effects of ethanol from below 50 to 75 mM ethanol on the extraspine and spine localization of NR1 subunits.
Summary. The odds ratio for unprovoked seizures for humans roughly triples with every 50 g/day increase in ethanol intake; for instance, the risk of a seizure in the general population who average 50 to 100 g/day intake is approximately 3× that of nondrinkers; at 100 to 200 g/day, the risk is 8× that of nondrinkers, and above 200 g/day, the risk is increased 20-fold over nondrinkers (Ng et al., 1988). In an urban population of men, the prevalence of epilepsy was 7.5% in alcohol-dependent patients, whereby it was only 0.6% in the general population (Hillbom et al., 2003). Taken together, these clinical reports and demographics demonstrate that chronic high level abuse and withdrawal continue to have prevalent and deleterious effects on neural function. Here, we posit that one molecular mechanism underlying such neuropathology after high-level chronic exposure involves a substantial enhancement of the function and distribution of NMDA receptors to synaptic sites with a more generalized increase in receptor density corresponding with increased subunit protein levels. Presynaptic alterations in spike-dependent release further amplify network excitability after ethanol withdrawal. Thus, a model of neuroadaptive mechanisms underlying ethanol withdrawal hyperexcitability should encompass both presynaptic and postsynaptic modes of ethanol action.
Acknowledgments
We thank Douglas Lee, Paula Guerrero, Yunqing Han, Sandra Nguyen, and Mohammed Moridnia for excellent technical assistance.
Footnotes
-
This work was supported by National Institute on Alcohol Abuse and Alcoholism Grants R01AA09230 and U01AA13497 (to R.A.M.) and F32AA014068 (to A.W.H.) and by University of Texas-Austin Bruce Jones Fellowships (to J.W.T. and T.A.Z.).
-
Article, publication date, and citation information can be found at http://jpet.aspetjournals.org.
-
doi:10.1124/jpet.106.111419.
-
ABBREVIATIONS: NMDA, N-methyl-d-aspartate; MK-801, dizocilpine maleate; WD, withdrawal; H-89, N-[2-(p-bromocinnamylamino)-ethyl]-5-isoquinolinesulfonamide dihydrochloride; AMPA, α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid; mEPSC, miniature excitatory postsynaptic current; QX-314, N-(2,6-dimethylphenylcarbamoylmethyl)triethyl ammonium chloride; DNQX, 6,7-dinitroquinoxaline-2,3(1H,4H)-dione; (DL)-APV, dl-2-amino-5-phosphonopentanoic acid; TTX, tetrodotoxin; PBS, phosphate-buffered saline; eGFP, enhanced green fluorescent protein; GFP, green fluorescent protein; PDS, paroxysmal depolarizing shift; CET, chronic ethanol treatment; IEI, interevent interval; ACSF, artificial cerebrospinal fluid; DIV, days in vitro; AMPAR, α-amino-3-hydroxy-5-methylisoxazole-4-propionic acid.
- Received July 21, 2006.
- Accepted January 16, 2007.
- The American Society for Pharmacology and Experimental Therapeutics