Abstract
Neuroactive steroids modulate the function of γ-aminobutyric acid type A (GABAA) receptors in brain; this is the presumed basis of their action as anesthetics. In a previous study using the neuroactive steroid analog, (3α,5β)-6-azi-3-hydroxypregnan-20-one (6-AziP), as a photoaffinity-labeling reagent, we showed that voltage-dependent anion channel-1 (VDAC-1) was the predominant protein labeled in brain. Antisera to VDAC-1 were shown to coimmunoprecipitate GABAA receptors, suggesting a functional relationship between steroid binding to VDAC-1 and modulation of GABAA receptor function. This study examines the contribution of steroid binding to VDAC proteins to modulation of GABAA receptor function and anesthesia. Photolabeling of 35-kDa protein with [3H]6-AziP was reduced 85% in brain membranes prepared from VDAC-1-deficient mice but was unaffected by deficiency of VDAC-3. The photolabeled 35-kDa protein in membranes from VDAC-1-deficient mice was identified by two-dimensional electrophoresis and electrospray ionization-tandem mass spectrometry as VDAC-2. The absence of VDAC-1 or VDAC-3 had no effect on the ability of neuroactive steroids to modulate GABAA receptor function as evidenced by radioligand ([35S] t-butylbicyclophosphorothionate) binding or by electrophysiological studies. Electrophysiological studies also showed that neuroactive steroids modulate GABAA receptor function normally in VDAC-2-deficient fibroblasts transfected with α1β2γ2 GABAA receptor subunits. Finally, the neuroactive steroid pregnanolone [(3α,5β)-3-hydroxypregnan-20-one] produced anesthesia (loss of righting reflex) in VDAC-1- and VDAC-3-deficient mice, and there was no difference in the recovery time between the VDAC-deficient mice and wild-type controls. These data indicate that neuroactive steroid binding to VDAC-1, -2, or -3 is unlikely to mediate GABAA receptor modulation or anesthesia.
Certain endogenous pregnane steroids and their structural analogs are potent anesthetics in vertebrates (Selye, 1941; Atkinson et al., 1965). These neuroactive steroids are thought to produce anesthesia by modulating the function of γ-aminobutyric acid type A (GABAA) receptors in the central nervous system (Harrison and Simmonds, 1984; Majewska et al., 1986; Harrison et al., 1987). The strong correlation between the ability of neuroactive steroid analogs to modulate GABAA receptors and their ability to produce anesthesia strongly supports this hypothesis (Harrison et al., 1987). The actions of neuroactive steroids on GABAA receptors are likely to be mediated via binding to specific recognition sites on the GABAA receptor protein complex. This concept is supported by the enantioselectivity of neuroactive steroids both in their modulation of GABAA receptor function and in their actions as anesthetics (Wittmer et al., 1996; Covey et al., 2000).
Extensive molecular biological studies (generation of chimeric receptors, site-directed mutagenesis) have been unsuccessful in identifying candidate regions or binding sites for neuroactive steroids on the GABAA receptor protein. The failure of molecular biological approaches to identify neuroactive steroid binding sites could be explained in several ways. 1) A binding site may involve multiple noncontiguous portions of the protein, foiling chimeric strategies. 2) There may be multiple binding sites mediating discrete but functionally additive effects. The functional assays used to screen mutated receptors may thus lack sensitivity. 3) The binding sites may be on an accessory protein rather than on a GABAA receptor subunit. To avoid the aforementioned problems and to provide a direct and unbiased search for neuroactive steroid binding sites, we have previously used the neuroactive steroid analog, 6-AziP [(3α,5β)-6-azi-3-hydroxypregnan-20-one], as a photolabeling reagent. Several brain proteins were shown to be specifically labeled by 6-AziP. The most prominently labeled protein in brain was identified as voltage-dependent anion channel-1 (VDAC-1) (Darbandi-Tonkabon et al., 2003).
VDAC-2 and VDAC-3 were also identified as proteins that may be labeled by 6-AziP. Furthermore, it was found that antisera to VDAC-1 coimmunoprecipitated GABAA receptor subunits from a detergent lysate of rat brain, indicating a strong physical association between VDAC-1 and GABAA receptors. This is consistent with the earlier observation that VDAC-1 and/or VDAC-2 copurified with GABAA receptors in a multistep purification procedure from detergent-solubilized mammalian brain (Bureau et al., 1992).
These data suggested the possibility that neuroactive steroid modulation of GABAA receptor function is mediated by binding of the steroids to a voltage-dependent anion channel that is directly associated with the receptor protein. To test this hypothesis, we have examined the effects of neuroactive steroids in mice lacking either VDAC-1 or VDAC-3 and in VDAC-2-deficient cells transfected with GABAA receptor subunits.
Materials and Methods
Membrane Preparation. Mouse brains were prepared freshly after sacrificing the mice under deep halothane anesthesia. Cerebella and brain stem were trimmed from the brains and the cerebral hemispheres were used to prepare membranes with minor modification of previously described methods (Hawkinson et al., 1996). Briefly, brains were immersed in ice-cold 0.32 M sucrose (10 ml/g) and homogenized using a Teflon pestle in a motor-driven homogenizer. The homogenate was centrifuged for 10 min at 1500g and the pellet was discarded. The supernatant was centrifuged for 30 min at 10,000g to obtain the P2 pellet, which was washed three times with 50 mM potassium phosphate/200 mM NaCl, pH 7.4. The pellet was resuspended in 50 mM potassium phosphate/200 mM NaCl, pH 7.4, and re-collected by centrifugation for 20 min at 10,000g. The final pellet was resuspended using a Teflon homogenizer and stored at –80°C.
Transfection of Mouse Embryonic Fibroblasts with α1FLAGβ2γ2 Subunits of the GABAA Receptor. Mouse embryonic fibroblasts derived from embryonic day 7.5 chimeric embryos were treated with 0.2 mg/ml G418 (geneticin; Invitrogen, Carlsbad, CA) immediately after plating of the dispersed embryos and selected for 5 days to isolate VDAC-2 –/– cells (Cheng et al., 2003). Deficiency of VDAC-2 in mouse embryonic fibroblasts after G418 selection was confirmed by polymerase chain reaction genotyping and immunoblotting with a previously generated isoform-specific antibody (data not shown). Wild-type (VDAC-2 +/+) and VDAC-2 –/– embryonic mouse fibroblasts were maintained in Dulbecco's modified Eagle's medium high glucose plus pyruvate (Invitrogen) containing 10% fetal bovine serum (Hyclone, Logan, UT), and penicillin (100 U/ml) plus streptomycin (100 U/ml) in a humidified atmosphere containing 5% CO2. Transient transfections were performed using Effectene Transfection Reagent (QIAGEN, Valencia, CA), 300 μl of Enhancer and DNA-condensation buffer, 24 μl of Enhancer, and 50 μl of Effectene. The day before transfection, cells from nearly confluent cultures were split into 10-cm dishes at approximately 105 cells per dish. Three micrograms of DNA were used for transfection into each 10-cm dish (1 μg of rat α1FLAG, β2, and γ2, each). Cells were cultured in complete media with the addition of 2 mM sodium butyrate (Sigma-Aldrich, St Louis, MO), after transfection. Transfected cells were panned using the anti-FLAG M2 antibody (Sigma-Aldrich) as described previously (Chen et al., 1995). Anti-FLAG antibody was adsorbed to Dynal dynabeads (Dynal Biotech, Oslo, Norway) conjugated with goat anti-mouse antibody and applied to dishes containing panned cells (Ueno et al., 1997). Cells expressing the FLAG epitope were visualized by attached beads. Single-channel currents were recorded in the bead-identified cells and analyzed as described below. The analysis was carried out on single-channel clusters, which were defined and isolated as described previously (Steinbach and Akk, 2001).
Chemical Synthesis. [3H]6-Azi-pregnanolone ([3H]6-AziP) was prepared by a multistep synthesis from commercially available progesterone as previously reported. (Darbandi-Tonkabon et al., 2003).
[35S]TBPS Binding. [35S]t-Butylbicyclophosphorothionate ([35S]TBPS) binding assays were performed using previously described methods (Hawkinson et al., 1994; Covey et al., 2000) with modification. Briefly, aliquots of membrane suspension (0.05 mg/ml final protein concentration in assay) were incubated with 5 μM GABA (Sigma-Aldrich), 4 nM [35S]TBPS (60–100 Ci/mmol; PerkinElmer Life and Analytical Sciences, Boston, MA) and 5-μl aliquots of steroid in Me2SO4 solution (final steroid concentrations ranged from 1 nM to 10 μM), in a total volume of 1 ml of 200 mM NaCl, 50 mM potassium phosphate buffer, pH 7.4. Control binding was defined as binding observed in the presence of 0.5% Me2SO4 and the absence of steroid; all assays contained 0.5% Me2SO4. Nonspecific binding was defined as binding observed in the presence of 200 μM picrotoxin and ranged from 12.4 to 32.6% of total binding. Assay tubes were incubated for 2 h at room temperature. A Brandel cell harvester (Brandel Inc., Gaithersburg, MD) was used for filtration of the assay tubes through Whatman glass fiber (GF/C) filter paper. Filter paper was rinsed with 4 ml of ice-cold buffer three times and dissolved in 4 ml of ScintiVerse II (Fisher Scientific Co.,, Pittsburgh, PA). Radioactivity bound to the filters was measured by liquid scintillation spectrometry and data were fit to the Hill equation using Sigma Plot (SPSS Inc., Chicago, IL), where binding was normalized to control binding, B is radioligand bound, [C] is steroid concentration, IC50 is the half-maximal inhibitor concentration, and n is the Hill coefficient. Each data point was determined in triplicate.
In Vivo Anesthesia Studies. The ability of pregnanolone and allopregnanolone to produce anesthesia was measured in 8-week-old male mice weighing approximately 20 g. Steroids were made up as a stock solution of 100 mg/dl in 4.4% ethanol and 8% Cremophor EL (Sigma-Aldrich). The solution was administered at a dose of 5 mg/kg by tail vein injection. Ten seconds after injection, the mice were placed in a supine position on a heating pad at 37°C. Sleep time was defined as the time from the moment mice displayed loss of righting reflex until they were able to right themselves. Mice were only included in results if they recovered fully without observable neurological deficits.
Electrophysiology. The hippocampal cultures were prepared from wild-type and VDAC-1 –/– mice, ages postembryonic day 1 to 3, as described previously (Mennerick et al., 1995), with slight modifications. Mouse culture medium was supplemented with insulin/transferrin/selenium (Sigma-Aldrich). At day in vitro 4, cultures were fed with a half medium exchange using Neurobasal medium plus B27 supplement (Invitrogen). The electrophysiological experiments were carried out 4 to 6 days after plating. The single-channel data were analyzed as described previously (Steinbach and Akk, 2001). In brief, the currents were recorded in the cell-attached and inside-out patch configurations at pipette potentials of +50 to +100 mV. The receptors were activated by 10 μM GABA in the absence and presence of 1 μM (3α,5α,17β)-3-hydroxyandrostane-17-carbonitrile (ACN). The channel open times were determined from the single-channel currents using the QuB suite (http://www.qub.buffalo.edu) (Qin et al., 1996).
Whole-cell, patch-clamp recordings were obtained at a holding potential of –70 mV. The extracellular bath contained 140 mM NaCl, 4 mM KCl, 2 mM CaCl2, 1 mM MgCl2, 10 mM glucose, 10 mM HEPES, pH 7.25. NBQX (1 μM), 50 μM d-APV, and 0.5 mM tetrodotoxin also were included in the bath to block spontaneous excitation. The whole-cell pipette solution contained 130 mM CsCl, 4 mM NaCl, 0.5 mM CaCl2, 5 mM EGTA, pH 7.25. Drug applications were achieved with a gravity-fed multibarrel pipette with a common tip placed within 500 μm of the recorded cell. For data analysis, peak GABA responses were measured in the presence and absence of steroid. Statistical comparisons between genotypes were made by unpaired, two-tailed t test.
Electrophoresis, Western Blot, and Gel Slicing. Polyacrylamide gel electrophoresis was performed using 10% polyacrylamide gels, under reducing conditions (Laemmli, 1970). After electrophoresis, the gels were either stained, sliced, or used for Western blot. Gels were silver stained (Hatzimanikatis et al., 1999) using modified ammoniacal silver stain (Amersham Biosciences Inc., Piscataway, NJ) or were stained with Coomassie Blue using a Novex Colloidal Coomassie G-250 Kit (Invitrogen).
For gel slicing, the gels were cut in vertical columns and sliced in 1-mm horizontal slices using a DE 113 Manual Gel Slicer (Hoeffer, San Francisco, CA). Slices were digested with 4 ml of tissue solubilizer consisting of 3a20 (Research Products International Corp., Mt. Prospect, IL) and TS-2 (ratio 9:1; Research Products International Corp.) for 24 h, and the radioactivity in each slice was determined by scintillation spectrometry.
For Western blotting, proteins from the SDS-PAGE gels were transferred onto multistack membranes (Kodak Biomax Multi-Blot kit, Catalog Number 193 4439) using a Bio-Rad (Hercules, CA) wet transfer system. For immunoblotting with the anti-VDAC-1 monoclonal antibody (Oncogene Research Products, EMD Biosciences, San Diego, CA), the membranes were blocked with 10% dried milk for 15 min at room temperature and incubated for 30 min with anti-VDAC-1 antibody (2 μg/ml) diluted in 10% dried milk. The membranes were washed three times with Tris buffer saline-Tween 20 and then incubated for 30 min with peroxidase-conjugated anti-mouse IgG (1:1000). Immunoreactive bands were visualized using the ECL-plus Western blotting detection system (Amersham Biosciences Inc.). Immunoblotting with the anti-VDAC-2 and -3 monoclonal antibody (generated in the Craigen Laboratory, Baylor College of Medicine, Houston, TX) was performed as described above, with the exception that peroxidase-conjugated anti-chicken IgG (1:1000) was used as secondary antibody.
Photolabeling. Photolabeling was performed as previously described (Darbandi-Tonkabon et al., 2003). Briefly, mouse brain membranes were placed in a quartz cuvette in buffer (50 mM potassium phosphate buffer, pH 7.4, 150 mM NaCl, 5 μM GABA) at a concentration of 400 μg of membrane protein/ml and preincubated with [3H]6-AziP for 90 min at 4°C in the dark. The cuvette was then placed in a photoreactor at a distance of 8 cm from the source. [The photoreactor uses a 450-W Hanovia medium pressure mercury lamp (Hanovia Ltd., Slough, Berkshire, UK) as the light source. The lamp is filtered through a 1.5-cm-thick saturated copper sulfate solution. This filter absorbs all light of wavelength <315 nm (Katzenellenbogen et al., 1974).] The samples were routinely irradiated for 3 min while continuously cooled to 4°C. After irradiation, the membranes were harvested by centrifugation and solubilized in SDS-sample buffer (312.5 mM Tris HCl, 5% SDS, 0.5 M dl-dithiothreitol, 50% glycerol, and 0.1% bromphenol blue) and analyzed by electrophoresis on a 10% SDS-polyacrylamide gel. The gels were sliced and radioactivity was measured in each slice.
Autoradiography. For autoradiography, SDS gels were fixed for 30 min in isopropanol/water/acetic acid (25:65:10) at room temperature and then dried under vacuum. The dried gels were placed in cassettes and exposed to 3H-sensitive ultra-film (Kodak Biomax light film) at –70°C for periods ranging from 5 days to 2 weeks.
Two-Dimensional Electrophoresis. Two-dimensional electrophoresis was performed as previously described (Darbandi-Tonkabon et al., 2003). The final gels were either processed for autoradiography or stained with silver stain or Coomassie blue. The nonradioactive gels were stained with Coomassie Blue. The gels with tritiated samples were silver-stained and all stained spots in the area identified in the original autoradiography were excised, digested, and analyzed by scintillation spectrometry. The spots on the Coomassie Blue-stained gels corresponding to spots containing maximum radioactivity were manually excised. The excised samples were prepared for mass spectrometry by in-gel proteolytic digestion with Promega sequencing grade modified trypsin (Jimenez et al., 2000). The recovered tryptic peptides were analyzed on a Micromass Q-TOF Micro electrospray mass spectrometer utilizing a Waters Capillary HPLC (Waters, Milford, MA) with a nano-spray emitter. Data were searched using Mascot software by Matrix Science Ltd. (London, UK). Analyses were performed at the Protein and Nucleic Acid Chemistry Laboratories at Washington University School of Medicine (St. Louis, MO).
Results
Photolabeling
Photolabeling with [3H]6-AziP was performed in brain membranes from VDAC-1- and VDAC-3-deficient (VDAC –/–) and strain-controlled (VDAC +/+) mice to confirm that VDAC-1 was the major photolabeled protein in brain and to determine whether other VDAC isoforms were also photolabeled. Brain membranes from the mice were photolabeled with 10 μM [3H]6-AziP and analyzed by SDS-PAGE with gel slicing. In VDAC-3 –/– and VDAC-3 +/+ (C57 Black), radioactivity was incorporated into two major protein bands, one at approximately 35 kDa and one at approximately 60 kDa. There was no discernible difference in the labeling of either peak between the VDAC-3 –/– and the VDAC-3 +/+ membranes, indicating that VDAC-3 protein either is not labeled by [3H]6-AziP or represents a negligible component of the labeled protein (Fig. 1A). In contrast, brain membranes from VDAC-1 +/+ (CD 1) mice showed a single major peak of radiolabel incorporation at 35 kDa (Fig. 1B). A minor peak of radiolabel incorporation was consistently observed at 60 kDa; the size of this peak varied between experiments, and the apparent difference in the size of the 60-kDa peak between Figs. 1A (C57 Black) and 1B (CD 1) does not represent a consistent finding. The 35-kDa peak was reduced by 85% (area under the curve) in VDAC-1 –/– membranes, with a small residual peak of radiolabel incorporation remaining at 35 kDa. These data confirm that VDAC-1 is the major protein labeled by [3H]6-AziP in brain. The data also show that a 35-kDa protein other than VDAC-1 is photolabeled.
Identification of Radiolabeled Proteins in VDAC-1-Deficient Mice. Membranes prepared from the brains of VDAC-1 –/– and VDAC-1 +/+ mice were labeled with 10 μM [3H]6-AziP; labeled proteins were separated by two-dimensional electrophoresis and visualized by autoradiography. The resultant autoradiograms (Fig. 2) show labeled spots at 35, 60, and approximately 18 kDa in the VDAC +/+ membranes. Several of the 35-kDa spots and the 18-kDa spots are absent in the VDAC –/– membranes, indicating that these spots represent VDAC-1 (with various post-translational modifications) and proteolytic fragments of VDAC-1. The spots at 18 kDa are only present in the VDAC +/+ gels and are therefore likely to represent proteolytic fragments of VDAC-1 that contain a 6-AziP binding site. This is consistent with the earlier observation that VDAC-1 antisera immunoprecipitate a [3H]6-AziP-labeled protein band of 18 kDa from rat brain (Darbandi-Tonkabon et al., 2003).
Three prominent radiolabeled spots were observed at 35 kDa in membranes from VDAC-1 –/– mice; two of these spots were more prominently labeled than the third. Silver staining of two-dimensional gels from the VDAC-1 –/– animals showed a single silver-stained spot (circled in Fig. 2) that corresponded to one of these radiolabeled spots. This spot was excised and analyzed by electrospray ionization-tandem mass spectrometry. Three peptides were identified (LTLSALVDGK, YQLDPTASISAK, and LTFDTTFSPNTGK), which showed sequence identity with VDAC-2 (gi|10720224). This provides a definitive identification of the protein as VDAC-2. Mass spectrometric analysis did not identify any other proteins in the excised spot.
Immunoblotting. To confirm the identity of the radiolabeled 35-kDa spots in two-dimensional gels from photolabeled VDAC-1 –/– brain, immunoblots were performed using antisera to VDAC-1, -2, and -3 on two-dimensional gels of brain membranes from VDAC-1 +/+ and VDAC-1 –/– mice. As shown in Fig. 3, VDAC-1 and VDAC-2 both have molecular weights of 35,000, and both are represented by multiple spots. The VDAC-2 spots have distinctive pI values from the VDAC-1 spots. The VDAC-2-immunostained spots correspond well to the radiolabeled spots observed in the autoradiograms of [3H]6-AziP-labeled VDAC-1 –/– brain (Fig. 2); this is consistent with the identification of VDAC-2 as the other photolabeled 35-kDa protein in brain. Immunoblotting with VDAC-3 antisera indicates that VDAC-3 is minimally expressed in mouse brain. Immunoblots prepared from VDAC-1 –/– animals confirm the absence of VDAC-1 in brain membranes.
Effect of 6-AziP on [35S]TBPS Binding. Neuroactive steroids are known to allosterically inhibit binding of the caged convulsant [35S]TBPS to the picrotoxin binding site of GABAA receptors (Majewska et al., 1986). Figure 4A shows that pregnanolone completely inhibits [35S]TBPS binding in both VDAC-1 –/– and VDAC-1 +/+ mouse brain membranes; the pregnanolone inhibition curves for the VDAC-1 –/– and +/+ membranes were virtually identical. Allopregnanolone [(3α,5α)-3-hydroxypregnan-20-one] also completely inhibited [35S]TBPS binding in VDAC-1 –/– and +/+ mouse membranes (Fig. 4C). The allopregnanolone inhibition curves were virtually identical in VDAC-1 –/– and VDAC-1 +/+ mouse brain membranes. Pregnanolone (Fig. 4B) and allopregnanolone (Fig. 4D) also completely inhibited [35S]TBPS binding in VDAC-3 –/– and VDAC-3 +/+ mouse brain membranes. The presence or absence of VDAC-3 had no discernible influence on the steroid inhibition curves. These results suggest that neither VDAC-1 nor VDAC-3 is required for neuroactive steroid (pregnanolone or allopregnanolone) modulation of GABAA receptors in mouse brain.
Electrophysiology
We examined the effect of ACN on single-channel currents elicited by 10 μM GABA. Sample currents from wild-type and VDAC-1 –/– mice, recorded in the absence and presence of 1 μM steroid, are shown in Fig. 5A. At 10 μM GABA, the currents consisted of isolated openings and gating bursts. No single-channel clusters were evident. The mean open durations were 3.6 ± 1.3 ms (n = 5) in wild-type neurons and 5.4 ± 1.2 ms (n = 4) in VDAC-1-deficient mice. Exposure to ACN resulted in an increase in the mean open duration of the channel as seen in our previous observations with recombinant GABAA receptors (Akk and Steinbach, 2003). The mean open duration in the presence of 1 μM ACN was 7.2 ± 1.5 ms (n = 4) in wild-type and 10.2 ± 3.5 ms (n = 3) in VDAC-1 –/– neurons.
The similarity in the effect of steroid on the channel mean open time strongly suggests that GABA responses in VDAC-1-deficient cells can be modulated by neuroactive steroids. However, it is possible that components of single-channel behavior not explicitly examined in our open time analysis are differentially affected by steroids in the two genotypes and are reflected in whole-cell current potentiation. We thus examined steroid potentiation of whole-cell currents from cultured hippocampal neurons. At low concentrations of GABA (2 μM) and steroid (200 nM), to avoid saturation effects, we found no significant difference between wild-type and VDAC-1 –/– neurons with regard to potentiation by either 5α-reduced or 5β-reduced steroids (Fig. 5, C and D). These results thus confirm the lack of difference in steroid effect on channel activity by ACN and extend the results to other 5α-reduced steroids and to 5β-reduced steroids.
Electrophysiological Studies in VDAC-2-Deficient Fibroblasts. We next examined the effect of ACN on rat α1FLAGβ2γ2 GABAA receptors expressed in VDAC-2-deficient mouse fibroblasts. Sample single-channel clusters from cells lacking VDAC-2, and from control, wild-type fibroblasts, are shown in Fig. 6A. The clusters were elicited by 50 μM GABA. This concentration corresponds to approximately EC50 of the dose-response curve for such receptors expressed in HEK 293 cells (Steinbach and Akk, 2001). The coapplication of ACN along with GABA resulted in significant changes in the receptor activity and cluster kinetic properties. First, the mean open duration increased in the presence of GABA and 1 μM ACN (Fig. 6B). It has been shown previously that the increase in the channel mean open duration, in the presence of ACN, is caused predominantly by an increase in the duration of the longest-lived component of the three intracluster open time classes (Akk and Steinbach, 2003). These data agree with the results obtained from the VDAC-2 –/– and wild-type fibroblasts expressing the α1FLAGβ2γ2 GABAA receptor (Fig. 6C). In addition to changes in the channel open times, the intracluster closed times are affected in the presence of ACN. The rates of entry into all intracluster closed states are reduced as a result of the increase in the mean open duration. However, the rate of entry into an activation-related component, CTβ, is selectively reduced to a much greater extent in the presence of ACN (Fig. 6D). The occurrence of CTβ reflects the channel closing rate, which is reduced by ACN. The changes seen in the relative frequency of the CTβ component of the intracluster closed times, as well as changes in the open time durations observed upon exposure to ACN, are similar to ones described previously for α1FLAGβ2γ2 GABAA receptors expressed in HEK 293 cells (Akk and Steinbach, 2003), or for receptors expressed in wild-type mouse fibroblasts (Fig. 6). Hence the data suggest that the presence of VDAC-2 is not a requirement for positive modulation of GABAA receptor currents by ACN.
In Vivo Anesthesia Studies
The ability of pregnanolone to produce anesthesia was studied in VDAC-1- and VDAC-3-deficient mice and in littermate strain-controlled animals. There was no significant difference in the ability of pregnanolone (5 mg/kg) to cause anesthesia (loss of righting reflex) or in the duration of its anesthetic effect between VDAC-1 –/– and VDAC-1 +/+ mice (Fig. 7A1) or between VDAC-3 –/– and VDAC-3 +/+ mice (Fig. 7B). The ability of allopregnanolone to produce anesthesia was also tested in VDAC-1 –/– and +/+ mice. There was no significant difference in the ability of allopregnanolone to produce anesthesia or in the duration of the anesthetic effect between VDAC-1 –/– and +/+ mice (Fig. 7A2). These data indicate that neither VDAC-1 nor VDAC-3 is necessary for the anesthetic actions of neuroactive steroids.
Discussion
In previous work we have shown that the neuroactive steroid analog photolabeling reagent, 6-AziP, modulates GABAA receptor function, and that the brain protein most prominently photolabeled by [3H]6-AziP is VDAC-1 (Darbandi-Tonkabon et al., 2003). Provocatively, VDAC-1 has also been shown to associate with GABAA receptors, as evidenced by copurification (Bureau et al., 1992) and by coimmunoprecipitation (Darbandi-Tonkabon et al., 2003) of the two proteins. In the current study, VDAC-1- and -3-deficient mice were used to confirm that VDAC-1 is the major VDAC isoform photolabeled by [3H]6-AziP and to show that VDAC-2, but not VDAC-3, is also labeled. The observed labeling pattern reflects the relative abundance of the three isoforms in mouse brain (Fig. 3) and is not indicative of isoform-specific labeling. The data presented in this article provide strong evidence that VDAC isoforms do not mediate the GABAergic or anesthetic effects of the neuroactive steroids. The electrophysiological, binding, and behavioral data in VDAC-deficient mice show that neither VDAC-1 nor VDAC-3 is responsible for steroid modulation of GABAA receptors or anesthesia. Similarly, electrophysiological data in VDAC-2-deficient fibroblasts indicate that VDAC-2 is not required for neuroactive steroid modulation of GABAA receptor currents (Fig. 6). It is, however, conceivable that VDAC-1 and VDAC-2 can both mediate neuroactive steroid modulation of GABAA receptors and can substitute one for the other. The absence of VDAC-2-deficient animals precludes looking at the double knockout and definitively eliminating this possibility.
The current data do not identify the functional significance of specific neuroactive steroid interactions with VDAC isoforms. In addition to their actions as GABAA receptor modulators and anesthetics, neuroactive steroids are thought to have other biological activity. At the behavioral level, neuroactive steroids have neuroprotective effects (Claudio and Stefano, 2000). At the cellular level, the steroids have been shown to influence the growth and differentiation of myelin-specific proteins in oligodendrocytes and to promote myelination of neurites in tissue culture (Jung-Testas et al., 1999); they are also thought to modulate apoptosis via a caspase-3-dependent pathway (Cascio et al., 2002). The VDAC family of proteins are pore-forming proteins (Colombini et al., 1996) that enable permeability to the outer mitochondrial membrane (Rostovtseva and Colombini, 1996), and are thought to be important in various aspects of mitochondrial function (Rostovtseva and Colombini, 1997). There is also substantial evidence that one or more VDAC isoforms may play a central role in apoptosis (Shimizu et al., 2000, 2001). Finally, VDAC-1-deficient mice have been shown to have a deficit in both spatial learning and hippocampal long-term potentiation (Weeber et al., 2002). Interactions of neuroactive steroids with VDAC isoforms could thus be modulators of mitochondrial function, synaptic plasticity, or regulators of apoptosis. A role in preventing apoptosis could, in part, explain the neuroprotective actions of the neuroactive steroids. With regard to all of these possibilities, it will be important to examine the actions of neuroactive steroids on the electrophysiological properties of VDAC-1 and -2 as well as on other VDAC actions.
An additional unresolved question is why VDAC-1, and possibly VDAC-2, associates with the GABAA receptor. This may be an interaction that occurs in a detergent lysate but may not happen in an intact cell. VDAC-1 is a very abundant protein that may nonspecifically adhere to detergent-solubilized hydrophobic proteins such as the GABAA receptor. Alternatively, interactions of VDAC-1 with ligand-gated ion channels may serve a role in localizing mitochondria to energy-requiring postsynaptic nerve endings.
Finally, the protein that binds neuroactive steroids and mediates their actions on GABAA receptors remains to be identified. In addition to the VDAC isoforms, 6-AziP labels a number of other proteins in a lysate of brain membrane (Fig. 2) (Darbandi-Tonkabon et al., 2003). The VDAC isoforms were studied first, because they are most prominently labeled; this is largely a reflection of their abundance in brain. There are also prominently labeled proteins at 58 kDa (pI = 4–5) that could mediate the anesthetic and GABAergic actions of the steroids. These proteins are being identified and studied and will be the subject of a subsequent report. It is also important to note that two-dimensional autoradiograms of [3H]6-AziP-labeled mouse brain (Fig. 2) show a single faintly labeled spot at pI = 8 to 9 and molecular weight of 50,000 to 60,000. This is the predicted pI and mass of most GABAA receptor subunits and suggests the possibility that 6-AziP also labels a specific GABAA receptor subunit. This spot is not visualized by silver staining of the two-dimensional gels and has thus not been identified by mass spectrometry; we are pursuing isolation and identification of this spot. It is important to reiterate that quantitative immunoprecipitation of [3H]6-AziP-photolabeled, epitope-tagged GABAA receptors (α1β1FLAG) from HEK 293 cells does not yield any radiolabeled GABAA receptor; GABA-elicited currents in these cells are, however, potentiated by neuroactive steroids (data not shown).
This presents the possibility that there is selectivity to labeling such that 6-AziP binds to a site on the GABAA receptor but does not covalently photolabel the site. To address this possibility, we are in the formative stages of preparing alternative neuroactive steroid analog photolabeling reagents in which the photolabeling moiety is located in different regions of the molecule or in which there is a different chemical mechanism of labeling. Finally, there are several faintly labeled proteins on autoradiograms of two-dimensional gels (best seen after prolonged exposure of the film to the gel). If the efficiency of photolabeling is modest (<10%) and the protein of interest interacts with GABAA receptors at 1:1 stoichiometry, faint labeling would be predicted. These proteins also need to be identified.
Acknowledgments
We thank Zong-Jin Cai for technical assistance and Dr. Charles Zorumski for helpful discussions.
Footnotes
-
This work was supported in part by National Institutes of Health Grants P01-GM47969 (to J.H.S., D.F.C., and A.S.E.), AA12952 (to S.J.M.), and R01 GM055713 (to W.J.C.), by a grant from the Klingenstein Foundation (to S.J.M.) and by the Alcoholic Beverage Medical Research Foundation (to G.A.).
-
DOI: 10.1124/jpet.103.058123.
-
ABBREVIATIONS: 6-AziP, (3α,5β)-6-azi-3-hydroxypregnan-20-one; VDAC, voltage-dependent anion channel; G418, geneticin; TBPS, t-butylbicyclophosphorothionate; ACN, (3α,5α)-3-hydroxyandrostan-20-carbonitrile; PAGE, polyacrylamide gel electrophoresis; CTβ, activation-related closing time; 3α5αP, allopregnanolone [(3α,5α)-3-hydroxypregnan-20-one]; 3α5βP, pregnanolone [(3α,5β)-3-hydroxypregnan-20-one].
- Received August 23, 2003.
- Accepted October 30, 2003.
- The American Society for Pharmacology and Experimental Therapeutics