INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The physiological and pharmacological properties of GABA receptors are largely determined by their subunit composition. Studies in heterologous expression systems have shown channel activation kinetics to depend on the  subunit (Gingrich et al. 1995; Lavoie et al. 1997; Verdoorn et al. 1990) and desensitization rate to depend on the  subunit (Haas and Macdonald 1999). Allosteric modulation by many drugs, including benzodiazepines, anticonvulsants, alcohol, and general anesthetics, is also subunit-dependent (see reviews by Harrison et al. 2000; Whiting et al. 2000). Additionally, GABA receptor subunit composition is developmentally regulated. In cerebellar granule cells, for example, the postnatal loss of benzodiazepine sensitivity correlates with the switch from the ₁ to the ₆ isoform (Zhu et al. 1995). Even the different complex behavioral effects of GABA modulators such as benzodiazepines are mediated by different  subunits (McKernan et al. 2000; Rudolf et al. 1999). It seems clear that "designer" receptors with desired properties could be engineered by controlling subunit composition of living neurons. These receptors might be used to correct defective inhibitory circuitry or alternatively to enhance the responsiveness to GABA-modulatory drugs.

We have recently reported successful adenoviral-mediated transduction of the GABA_C receptor ₁ subunit in hippocampal neurons (Cheng et al. 2001). Virally transduced neurons were identified by a coexpressed green fluorescent protein (GFP) reporter; double immunocytochemistry proved co-localization of the ₁ protein and the reporter; Western blot verified the expected molecular masses; and electrophysiological and pharmacological studies under whole cell patch-clamp confirmed the presence of functional GABA_C receptors. Moreover, in a model of neuronal hyperexcitability induced by chronic blockade of glutamate receptors, GABA_C receptors abolished the hyperactivity and the consequent delayed neuronal death.

Surprisingly, mIPSCs mediated by the GABA_C receptors were not observed, thus implying the failure of these receptors to form functional synapses. There are several possible causes for this absence of GABA_C receptor-mediated synaptic activity. First, there could be a kinetic mismatch between the relatively brief presynaptic GABA release and the slow GABA_C receptor-mediated postsynaptic activation: the activation kinetics of GABA_C receptors are 40 times slower than GABA_A receptors (Amin and Weiss 1994). Therefore the time course of presynaptic GABA release may be too rapid to activate subsynaptic GABA_C receptors. Second, there could be a general failure of virally transduced proteins to be properly targeted to subcellular sites. And third, there could be a specific failure of the ₁ subunit to be targeted to the subsynaptic membrane. Because successful definition of synaptic properties requires control of the composition of the subsynaptic receptor, we further investigated the reason for the lack of GABA_C receptor-mediated synaptic events.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Hippocampal culture

One- to 2-day-old Sprague-Dawley rat pups deeply anesthetized with halothane were decapitated, and the hippocampi were dissected out in ice-cold Ca²⁺-Mg²⁺-free Hank's balanced salt solution (Gibco, Gaithersburg, MD). The tissue was enzymatically digested with papain and bovine serum albumin (1 mg/ml for each; Sigma, St. Louis, MO) for 20 min at 37°C. Cells were disaggregated by trituration and plated on Matrigel-coated 35-mm tissue culture plates (Becton Dickinson, Bedford, MA) in Neurobasal medium (Gibco) supplemented with 2 mM L-glutamine, 10% fetal calf serum (Hyclone, Logan, UT), 5% horse serum, and B-27 supplement (Gibco). After 2-3 days growth in a 95% O₂-5% CO₂ humidified incubator at 37°C, the dishes were treated with 10 µM cytosine arabinoside for 24 h to suppress the growth of glia. Thereafter the medium was switched to a Neurobasal containing 5% horse serum and changed every 2-4 days until used for experiments.

GABA subunit epitope-tagging, site-directed mutagenesis, and adenovirus creation

A modified pCI-neo vector with three copies of HA-epitope-tag (YPYDVPDYA) inserted between the Mlu and Not sites was created. pCMV5-HA3 (a gift of Dr. Bradford Berk, University of Rochester), containing three copies of HA-epitope-tag sequence was used as a PCR template to synthesize a short double-stranded DNA encoding a Mlu restriction enzyme site, three copies of HA, a stop codon, and a NotI restriction enzyme site. The primers used were: 3HA-For (GAATTC-ACGCGTTACCCATACGATGTTCCTGAC), 3HA-Rev (GTGCGGCCGCTCACTAGCACTGAGCAGC). The HA-tagged GABA ₂(mut) and ₁ subunits were created by inserting the mutant cDNA in frame into the Mlu restriction enzyme site of the 3HA-pCI-neo vector. Proper function of the tagged ₂(mut)-HA construct was verified by transiently transfecting HEK293 cells as a heteromeric combination with wt ₁ and ₂. Likewise, homomeric tagged-₁ expression and function were confirmed. Proper expression of the HA tag was verified by immuno-staining the transfected HEK293 cells with anti-HA antibody.

The murine ₂ subunit (a gift of Dr. David Burt, University of Maryland) was used for oligonucleotide-mediated site-directed mutagenesis. The oligonucleotides used to replace threonine 257 with phenylalanine were: T246F-For (TTAGGAATTTTCACT-GTCCTAACAATGACC), T246F-Rev (TAGGACAGTGAAAATCCTTAATGCAACCCG). The mismatch base pairs are indicated in bold. Successful mutagenesis and absence of unintended mutation was confirmed by automatic sequencing of the final clone.

The recombinant E1-, E3-deleted replication-deficient human adenovirus type-5 was created through homologous recombination between the pXCR shuttle vector and pBHG10 parent vector (Bett et al. 1994). The shuttle vector was modified to contain two expression cassettes, both driven by the RSV promoter followed by a multiple cloning site and a poly-adenylation sequence. The GFP cDNA (Clontech, Palo Alto, CA) was subcloned into the first cassette and the human GABA ₁ subunit cDNA (a gift of Dr. Gary Cutting, Johns Hopkins University), the GABA ₂(mut) subunit, or the HA-tagged subunit cDNA into the second. Therefore the two proteins, the GFP and one or the other of the GABA subunits, were expressed independently. The pBHG10 plasmids and the shuttle vector containing the transgene were cotransfected into HEK293 cells using Lipofectamine Plus following the manufacturer's recommended protocol (Gibco). Lytic plaques were isolated and expanded, and the presence of the transgene and the absence of E1 gene confirmed by PCR. High titer adenovirus, twice purified by CsCl gradient centrifugation, was stored as a 10% glycerol suspension at 80°C. The titer of each adenovirus preparation was determined by counting GFP-positive plaques formed in a virus-transduced confluent HEK293 monolayer overlaid with 0.5% low melting agarose. The concentration of virus used for transduction of cultured cells is reported as plaque forming units (pfu)/ml rather than pfu/cell since the number of viable cells in a culture dish was unknown. The pBHG10 and pXCR plasmids were purchased from Microbix (Toronto).

Electrophysiology

Patch electrodes were pulled from 1.2 mm OD borosilicate capillary glass (WPI, Sarasota, FL) and fire polished. Typical electrodes had a resistance of 5-10 M when filled with intracellular solutions. For voltage-clamp experiments, the intracellular solution consisted of (in mM) 140 CsCl, 4 NaCl, 2 MgCl₂, 10 K-EGTA, and 10 HEPES; for current-clamp recordings of action potentials, 140 mM K-gluconate replaced CsCl. Solutions were titrated to pH 7.3 with CsOH or KOH, and supplemented with 2 mM Mg-ATP. The external solution contained (in mM) 140 NaCl, 2.8 KCl, 1 MgCl₂, 3 CaCl₂ 10 HEPES, 10 glucose, and was titrated to pH 7.4 with NaOH. Recordings were made using an AxoPatch 200A amplifier (Axon Instruments, Foster City, CA). A typical access resistance of ~15 M in the whole cell mode of patch clamp was compensated by 75%. Voltage measurement errors due to uncompensated series resistance were <3 mV. The cell input capacitance was approximated by directly reading off the capacitance compensation dial of the amplifier. Recorded membrane currents were filtered at 5 kHz, digitized using Clampex v8.0 and analyzed with Clampfit v6.0 (Axon Instruments). A syringe pump delivered external solutions at 15 ml/h through orifices of a  tube mounted on a piezoelectric transducer (Burleigh Instruments, Fishers, NY). Command steps at 120-s intervals rapidly moved the perfusion ports, exposing the cell to eit the control or the drug solution. The perfusion device allowed exchange of solution in ~15 ms (10-90% rise time) for the whole cell recording configuration. All experiments were performed at room temperature (20-25 C°). Aqueous solutions of GABA, bicuculline methiodide, 6-cyano-7-nitroquinoxalene-2,3-dione (CNQX), and TTX, and a DMSO solution of picrotoxin prepared as 1,000× stock were stored as frozen aliquots at 20 C^o and freshly diluted on the day of the experiments. Drugs were purchased from Sigma Chemicals.

The natural inhibitory postsynaptic current (IPSC) was approximated by a 2-ms pulse of 1 mM GABA applied to an excised outside-out patch (i.e., simulated-IPSC). The experiments were performed essentially as described above for whole cell patch clamp but with a different drug application device (Polytec Optronics, Costa Mesa, CA). Furthermore the septum of the  tube was carefully etched to a minimal thickness with hydrofluoric acid to reduce flow turbulence. With this drug applicator, a 10-90% rise time of < 150 µs was attained at the open tip of the recording pipette. Bicuculline was preapplied prior to GABA application for the simulated-IPSC experiment because of potential kinetic competition between the antagonist and the agonist at this time resolution.

Mini analysis

Miniature GABA-mediated synaptic events were recorded from neurons held at 60 mV in the presence of 10 µM CNQX and 2.8 mM Mg²⁺ to block glutamatergic currents and 1 µM TTX to block action-potential-dependent synaptic events. Continuous data were stored on VHS tape and digitized off-line with a DigiData-1200A A/D converter controlled with Clampex V8.0 (Axon Instruments). Data were sampled every 0.2 ms and filtered at 2 kHz. Miniature events were analyzed using Minianalysis (Synaptosoft, Leonia, NJ). The program finds a local minimum (since the events are downward going) within a data window. The peak amplitude is calculated by taking the amplitude at the local minimum minus the average baseline current level at the beginning of the data window. Because the peak-to-peak noise level of the data was typically <8 pA, the peak amplitude threshold of detection was set at 10 pA. Events were further selected by a rise time criterion of <5 ms to assure analysis of events only originating from proximal release sites with minimal cable distortion. We routinely scrolled through the detected events and visually rejected any compound events or spuriously detected events. The events were selected for exponential fitting only if no other events occurred within 200 ms of the peak. Decay time constants were calculated by fitting a biexponential function. Rise times were estimated as the time required for the current to reach 10-90% of the baseline-to-peak amplitude. Kolmogorov-Smirnov two-sample test (DeGroot 1975) was used for comparing two cumulative frequency distributions of event amplitudes or decay time constants and the Student t-test for comparison of means. Significance was set at P < 0.001 for the Kolmogorov-Smirnov test and P < 0.05 for the Student t-test since the former statistical test appeared more sensitive to noise in the data requiring a more stringent statistical criterion for defining significance. Values are noted as means ± SE.

Immunocytochemistry and confocal microscope imaging

Cells grown on glass coverslips were fixed in 4% paraformaldelyde in 0.1 M phosphate buffer for 10 min and permeabilized in phosphate-buffered saline containing 0.2% Triton X-100 (PBST). The cells were blocked in PBST with 10% serum for 10 min, and all subsequent reactions were carried out in PBST with 2% serum. Primary antibodies used for immunostaining were as follows: mouse anti-HA (Babco, Richmond, CA, 1:500), rabbit anti-synaptophysin (Zymed, South San Francisco, CA, 1:500), mouse anti-MAP1B (Sigma, 1:1000), mouse anti-GABA_{A 2/3} (Chemicon, 1:500), mouse anti-GFP (Boehringer Mannheim, Indianapolis, IN, 1:200). Fluorescent secondary antibodies were all obtained from Jackson Immunoresearch Laboratory (West Grove, PA) and used at 1:500.

The digital images were acquired on a Leica TCS NT confocal microscope using the TCS-NT software (Leica, Deerfield, IL). A ×63, 1.32 NA, oil-immersion objective lens was used to visualize individual cells and neurites. The GFP (488 nm) and rhodamine (543 nm) laser gain was set at 70 and 90%, respectively, with the pinhole aperture set at 0.70-1.2 µm. Optical serial sections (z series) of 0.2-0.3 µm were taken through the cells where each planar image was an accumulation of eight time scans, and 12 to 15 planar images were reconstructed to yield the final projection images.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Expression of functional GABA_C receptors and picrotoxin-resistant GABAA receptors

Recombinant adenoviruses were constructed to deliver a GFP reporter molecule and either a GABA_{C 1} subunit or a GABA_A receptor ₂ subunit with a picrotoxin resistance mutation [₂(mut)]. We also constructed a control virus, Ad(GFP), designed solely to express the GFP reporter protein. In transduced hippocampal neurons, GFP fluorescence could be detected in the somata after 24 h and in neurites after 2-6 days as reported previously (Cheng et al. 2001). At higher viral titers, more neurons exhibited GFP fluorescence; however, nonneuronal cells were also transduced and gross morphological abnormalities (such as swelling of the neurons and stunned neurite outgrowth) occurred. Because a viral titer of 1-5 × 10⁵ pfu/35 mm dish had no apparent effect on the neuronal morphology while transducing ~10-15% of the neurons, this titer was used in all subsequent experiments. Neurons transduced with the adenovirus Ad(GFP), Ad(₁/GFP), or Ad(₂(mut)/GFP) are referred to as GFP, ₁/GFP, or ₂(mut)/GFP neurons.

To confirm the functional expression of the ₁ and ₂(mut) receptor subunits, the electrophysiological properties of GFP, ₁/GFP, and ₂(mut)/GFP neurons were examined with whole cell patch clamp. In all neurons examined, GABA (30 µM), applied to transduced neurons by a rapid perfusion device (Fig. 1, left), evoked an inward-going current, consistent with activation of endogenous GABA_A receptors. In GFP neurons, co-applications of either bicuculline, a GABA_A receptor antagonist, or picrotoxin, a nonspecific chloride ionophore blocker, almost completely inhibited the GABA-evoked current (Fig. 1, top).

View larger version (12K): [in this window] [in a new window]   Fig. 1. Effects of bicuculline and picrotoxin on 1/GFP and 2(mut)/GFP neurons. The 3 panels represent whole cell currents in response to 30 µM GABA (left), 30 µM GABA with 100 µM bicuculline (middle), or 30 µM GABA with 100 µM picrotoxin (right) from Ad(GFP) (top), Ad(1/GFP) (middle), or Ad(2(mut)/GFP) (bottom) transduced hippocampal neurons. Recordings were obtained 2 days after viral transduction. The bars above the 1st row of traces denote the duration of drug application.

In contrast, 20/22 fluorescent ₁/GFP neurons, after GABA_A receptor blockade by bicuculline (Fig. 1, middle, center), continued to demonstrate GABA-evoked current, but reduced 60-70% in amplitude and with the slower activation, desensitization and deactivation kinetics typical of GABA_C receptors (which are not affected by bicuculline). Picrotoxin completely inhibited current in all ₁/GFP neurons (Fig. 1, middle, right). The GABA_C receptor appears to have been successfully expressed.

₂(mut)/GFP-neurons, showed the opposite pattern. GABA-evoked current was nearly completely inhibited by bicuculline (Fig. 1, bottom, center), but only partially inhibited by 100 µM picrotoxin (38 ± 12% of control, n = 8; Fig. 1, bottom, right). The picrotoxin-resistant component had the same kinetics as the control current. This implies that the ₂(mut) subunit, sensitive to bicuculline but resistant to picrotoxin, has been properly incorporated into the GABA_A receptors of the ₂(mut)/GFP neurons.

Absence of bicuculline-resistant GABA_C receptor-mediated mIPSCs

We next examined whether the virally transduced GABA_C receptors were forming functional synapses by searching for GABA_C-receptor-mediated IPSCs. To reduce competing signals, we blocked glutamatergic receptors with CNQX and action potentials with Mg²⁺ and TTX. In GFP neurons held at 60 mV, inward-going mIPSCs were readily observed. The frequency of mIPSCs ranged from 0.5 to 4.6 Hz, with a mean amplitude of 41.8 ± 1.3 pA (Fig. 2, A and D-F). Multiexponential fits to the events revealed a fast rise time (t_rise = 2.9 ± 0.2 ms) and a biexponential decay time course (_dec1 = 7.0 ± 0.3 and _dec2 = 38.2 ± 1.9 ms; all mean values, n = 8). The GFP-neuron mIPSCs were reversibly blocked by bicuculline (10 µM), suggesting that these events were mediated by subsynaptic GABA_A receptors exclusively.

View larger version (23K): [in this window] [in a new window]   Fig. 2. Absence of bicuculline-resistant miniature inhibitory postsynaptic currents (mIPSCs) in 1/GFP neurons. The 3 panels represent whole cell voltage-clamp recording in 10 µM 6-cyano-7-nitroquinoxalene-2,3-dione (CNQX) and 1 µM TTX and 2.8 mM Mg2+ (left), with 10 µM bicuculline (middle), or with 100 µM picrotoxin (right) from Ad(GFP) (A)- or Ad(1/GFP) (B)-transduced hippocampal neurons. C: representative examples of mIPSCs from Ad(GFP)- or Ad(1/GFP)-transduced neurons with a superimposed biexponential decay fit. Cumulative histograms of peak event amplitude (D), fast decay dec1 (E), and slow decay dec2 (F) for an Ad(GFP) transduced neuron () and Ad(1/GFP) transduced neuron (- - -). See Table 1 for a summary.

In ₁/GFP neurons, mIPSCs recorded in control solution (without bicuculline) were quantitatively indistinguishable from those obtained in control GFP neurons (Fig. 2, B and D-F; 37.9 ± 1.5 pA, _rise = 3.2 ± 0.3 ms, _dec1= 6.5 ± 0.3, and _dec2 = 33.8 ± 1.4 ms, all mean values, n = 10, P > 0.5 by 2-sample Student's t-test and no difference in cumulative event distribution by Kolmogorov-Smirnov 2-sample test). Thus neither Ad(GFP) nor Ad(₁/GFP) viral infection affected GABAergic mIPSCs (see Table 1); both presynaptic GABA release and postsynaptic GABA_A receptors were seemingly unchanged. When bicuculline (10 µM), which does not block GABA_C receptors, was added to the ₁/GFP neurons, the mIPSCs completely disappeared. Thus no functional subsynaptic GABA_C receptors could be identified (despite robust expression of this receptor on the cell soma in n = 24 neurons from 8 separate transductions).

Ruthenium red fails to unmask bicuculline resistant receptor-mediated mIPSCs

In an attempt to reduce a possible presynaptic-postsynaptic kinetic match, we enhanced presynaptic release of GABA. Ruthenium red (RR) has been proposed to increase quantal transmitter release through calcium-independent direct binding to the presynaptic membrane (Trudeau et al. 1996) and has been shown to increase the frequency of mIPSCs in neonatal rat hippocampal neurons (Sciancalepore et al. 1998). We reasoned that if kinetic mismatch prevented activation of subsynaptic GABA_C receptors, RR enhancement of the presynaptic GABA release might reveal GABA_C receptor-mediated bicuculline-resistant mIPSCs.

Application of 10 µM RR to both GFP and ₁/GFP neurons markedly increased the frequency of GABAergic mIPSCs (Fig. 3, a mean increase of 151%, range 127-204%, and 139%, range 134-144%, for GFP and 1-GFP neurons, respectively). The effect of RR was rapid in onset, and the increased mIPSC frequency became stable within 5 min of RR application. The amplitude, rise time, and decay constants of the observed mIPSCs did not change in the presence of RR in either GFP or ₁/GFP neurons (data not shown) in agreement with an earlier observation on nontransduced hippocampal neurons (Sciancalepore et al. 1998). Notably, because co-application of bicuculline in the presence of RR still completely blocked mIPSCs in ₁/GFP neurons, RR did not uncover evidence for functional GABA_C synapses (Fig. 4).

View larger version (25K): [in this window] [in a new window]   Fig. 3. Ruthenium red enhances mIPSC frequency but fails to unmask GABAC receptor mediated mIPSCs. A: the traces are whole cell voltage-clamp recordings in 10 µM CNQX and 1 µM TTX and 2.8 mM Mg2+ (top), with 10 µM ruthenium red alone (middle), or with 10 µM bicuculline (bottom) from Ad(GFP) (left) or Ad(1/GFP) (right) transduced hippocampal neurons. B: summary of mIPSC frequency for Ad(GFP) () and Ad(1/GFP) () under the designated conditions. *, significant effect of RR (P < 0.01, paired t-test, n = 6).

View larger version (24K): [in this window] [in a new window]   Fig. 4. Rapid applications of GABA to outside-out patches predict observable GABAC receptor-mediated IPSCs. A: current responses of an excised patch from Ad(GFP) neurons elicited by a 2-ms pulse application of 1 mM GABA (left) or the same with 100 µM bicuculline (right). The bicuculline application preceded the GABA application by >100 ms. B: the same as in the preceding text except from a Ad(1/GFP) neuron. Recordings were obtained 2 days after viral transduction.

Simulated IPSC predicts a readily observable GABA_C receptor-mediated synaptic current

Because the precise effect of RR on presynaptic release is unknown, lack of bicuculline-resistant mIPSCs after RR cannot completely eliminate the possibility of a presynaptic/postsynaptic kinetic mismatch. Therefore by using ultra-rapid perfusion of excised outside-out patches to simulate synaptic GABA transients, we directly determined whether a brief GABA pulse could activate virally introduced GABA_C receptors.

In excised patches from GFP-neurons, the current evoked by a pulse of GABA application (2 ms, 1 mM) showed a rapid rate of rise followed by a biexponential decay time course, with kinetic parameters in agreement with a prior report using hippocampal neurons (Jones and Westbrook 1995). This simulated IPSC was blocked by bicuculline, consistent with GABA_A receptor mediation (Fig. 4A).

In patches from ₁/GFP neurons, the same GABA pulse evoked a two-component response: a bicuculline-sensitive rapid phase, identical to the GFP-neuron response, followed by a bicuculline-resistant slow phase (Fig. 4B). These kinetic and pharmacological properties are consistent with simulated IPSCs mediated by GABA_A and GABA_C receptors successively. The fast phase was selectively blocked by bicuculline while the isolated slow phase remained intact. The time course of the GABA_C receptor-mediated current could be simulated with a kinetic model previously described for this receptor (Chang and Weiss 1999) with a modification to include a blocked state distal to the open state (data not shown). Thus despite slow activation, a short GABA transient is sufficient to activate, with characteristic time course and pharmacology, virally introduced GABA_C receptors.

PTX-resistant mIPSCs after viral-transduction with GABA_A 2(mut) subunit

To rule out a general mistargeting of most or all virally transduced proteins, we transduced neurons with a GABA_{A 2} subunit containing a Thr257Phe point mutation. This point mutant, when incorporated with ₁ and ₂ subunits into a heteromeric GABA_A receptor, has been previously shown to confer picrotoxin resistance (Gurley et al. 1995). If the₂(mut) subunit targets correctly, synapses incorporating it should be identified through the presence of PTXN-resistant mIPSCs.

Two days after viral transduction, bicuculline-sensitive, PTXN-resistant mIPSCs were observed in 11 of 12 ₂(mut)/GFP neurons recorded (Fig. 5A). The mIPSC frequency in the presence of PTXN was only ~10% of control, suggesting that the ₂(mut) subunit was incorporated into a small proportion of active synapses. Furthermore, in the presence of PTXN, the mIPSC amplitude distribution was shifted with a significant decrease in the mean event amplitude (36.8 ± 1.0 vs. 26.0 ± 1.8 pA, P < 0.01, n = 8). Both the fast and slow mean mIPSC decay time constants also decreased from _dec1 = 5.3 ± 0.3 to 3.3 ± 0.3 ms and _dec2 = 26.8 ± 1.5 to 17.0 ± 1.5 ms (P < 0.01, n = 8 for both) with the addition of PTXN (Fig. 5, D-F). This conclusion was supported by comparisons of the cumulative event distributions by the Kolmogorov-Smirnov two-sample test. The percentage of fast-to-slow components remained unchanged.

View larger version (24K): [in this window] [in a new window]   Fig. 5. Properties of picrotoxin-resistant mIPSCS in GABAA-2(mut) neurons. Whole cell mIPSC recording from Ad(GFP) (A) or Ad(2(mut)/GFP) (B) transduced neurons. The 3 panels are recordings in control (left), with 10 µM bicuculline (middle), or with 100 µM picrotoxin (right). C: representative examples of mIPSCs with a biexponential decay kinetics from Ad(2(mut)/GFP) transduced neurons without or with 100 µM picrotoxin. Cumulative histogram of peak event amplitude (D), fast decay dec1 (E), and slow decay dec2 (F) from an Ad(2(mut)/GFP)-transduced neuron without picrotoxin () or in the presence of 100 mM picrotoxin (- - -). See Table 1 for a summary.

The existence of PTXN-resistant mIPSCs after ₂(mut)/GFP-transduction indicates GABA receptor subunit targeting to the subsynaptic membrane may be highly specific in nature, and it shows that the viral transduction process per se does not hinder subcellular targeting.

₂(mut) subunit colocalizes with presynaptic markers; the ₁ subunit does not

It is possible that the ₁ subunits are correctly targeted to the subsynaptic region but are nevertheless nonfunctional. To directly determine through immunohistochemical means the subcellular locations of the virally transduced subunits and to distinguish them from the endogenous ₂ subunits, we created a virus expressing HA-epitope-tagged ₂(mut) and ₁ subunits. Although no endogenous ₁ proteins exist in hippocampal neurons, epitope-tagging this subunit allowed detection of both subunits with the same anti-HA antibody.

Confocal imaging of endogenous ₂/₃ subunits revealed punctate immunoreactivity with 20-30% overlap with synaptophysin-immunoreactivity (Fig. 6A). This was expected because not all synaptic GABA_A receptors contain these subunits and the nonspecific presynaptic marker synaptophysin marks all synapses, including glutamatergic synapses. Examination of ₂(mut)/GFP-neurons 2 days after viral transduction revealed strong HA immunoreactivity in the cell bodies and bright punctate signals distributed along the neurites (Fig. 6B). When compared with the same cell probed with the synaptophysin, the signal overlap was greater. Similarly, the HA hot-spots correlated well with the specific presynaptic GABAergic marker GAD (data not shown). It seems clear that the₂(mut) subunits localize to the subsynaptic receptors.

View larger version (61K): [in this window] [in a new window]   Fig. 6. 2 but not 1 subunit-immunoreactivity colocalizes with extra-somatic synaptophysin-immunoreactivity. A: nontransduced control hippocampal culture probed with anti-synaptophysin antibody (Ai), anti-GABAA-2/3 antibody (Aii), and the 2 images superimposed (Aiii). B: 2(mut)-HA-transduced neuron probed with anti-synaptophysin antibody (Bi), anti-HA antibody (Bii), and both images overlaid (Biii). Note the high degree of colocalization of the 2 immunofluorescent signals. C: 1-HA transduced neuron probed with anti-synaptophysin antibody (Ci), anti-HA antibody (Cii), and both images overlaid (Ciii). Note the presence of anti-HA signal almost exclusively in the cell soma. Images for A-C are 8 z-series images captured at a 0.2-µm interval with a confocal microscope collapsed into a single projection image. D: a phase contrast and immunofluorescence image pair of a hippocampal culture probed with anti-MAP1B antibody. Bar = 26 µm for A-C and 100 µm for D.

In contrast, HA-immunoreactive ₁/GFP neurons were found largely in the cell soma and did not overlay with the synaptophysin signal (Fig. 6C). The ₁ subunit, which is not expressed in native hippocampal neurons, may possess different targeting machinery from other GABA receptor subunits. HA immunoreactivity was not seen in nontransduced GFP-negative neurons from the same culture dish. A Western blot with tagged subunits of the expected molecular mass showed that the HA signal corresponds to the epitope-tagged receptor (data not shown). Thus the immunohistochemical data supports proper subsynaptic targeting of ₂(mut) subunits but not of ₁.

The microtubule-associated protein MAP1B has recently been shown to influence the clustering of exogenously expressed ₁ subunits. MAP1B immuno-staining of the neuron culture reveals robust expression of this protein (Fig. 6D), confirming an earlier report (Sato-Yoshitake et al. 1989) and ruling out deficiency of MAP1B as the cause of mistargeting of the ₁ subunits.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Viral transduction of hippocampal neurons with ₁ subunits results in the formation of functional GABA_C receptors on the neuronal soma, but the receptors fail to be targeted to the subsynaptic membrane and therefore fail to be incorporated into functional synapses. Identical transduction of the same neurons with the picrotoxin-resistant ₂ subunit mutant results in incorporation of this subunit into functional picrotoxin-resistant GABA_A receptors. We base this conclusion on several pieces of evidence: first, in ₁/GFP neurons, a bicuculline-resistant GABA_C receptor-mediated current was clearly present in whole cell recordings, but no bicuculline-resistant mIPSCs were observed even in the presence of the increased transmitter release caused by ruthenium red. Second, in ₂(mut)/GFP neurons, both picrotoxin-resistant whole cell currents and picrotoxin-resistant mIPSCs were observed. Third, brief pulses of GABA to excised, outside-out membrane patches from ₁/GFP neurons revealed clearly discernible GABA_A and GABA_C receptor-mediated current components, suggesting GABA_C receptors, if functionally present, should produce synaptic events. And fourth, epitope-tagged ₂(mut) subunits colocalized in neurites with synaptic components; epitope-tagged ₁ subunits did not.

Subsynaptic targeting of GABA receptors

Synaptic transmission mediated by GABA_C receptors is present in retinal bipolar cells of salamander (Lukasiewicz and Shields 1998) and rat (Euler and Wassle 1998). Immunohistochemical studies demonstrate punctate ₁ immunoreactivity clusters in the retina (Enz et al. 1996) localized to axon terminals of rod and cone bipolar cells. Furthermore, there is no colocalization of ₁ immunoreactivity with GABA_A or glycine receptor immunoreactivity, suggesting no coassembly of ₁ with the various GABA_A subunits (Enz et al. 1996; Koulen et al. 1998). Thus ₁ subunits are independently targeted to the subsynaptic membrane in the retina. Study of the relative current contributions GABA_A and GABA_C receptors in cone cells indicates that the ratio between these two receptors varies depending on cell morphology (Euler and Wassle 1998), in turn implying specific receptor targeting mechanisms.

Recent studies have begun to elucidate the mechanisms regulating GABA_A receptor subcellular trafficking (reviewed in Barnes 2000). The current view suggests a multilevel regulation beginning with receptor oligomerization followed by surface clustering. Subsequent targeting of the assembled oligomers to the membrane surface is relatively inefficient in nonneuronal baby hamster kidney cells (Gorrie et al. 1997) compared with neurons, where 85-90% of  1 oligomers have a surface localization (Calkin and Barnes 1994). In our experiments, both functional GABA_C and ₂(mut)-containing GABA_A receptors were abundantly expressed on the neuronal soma, confirming successful subunit oligomerization and somatic membrane surface targeting. Any difference between processing of the two receptors appears to involve the finer regulation of receptor targeting to the somatic or the subsynaptic membrane.

An early study using polarized Madin-Darby canine kidney (MDCK) cells indicated that the GABA_A subunit itself contains the subcellular targeting information (Perez-Valazquez and Angelides 1993). This observation was recently confirmed by Connolly et al. (1996). Again in MDCK cells, GABA_A receptors incorporating ₁₂ or ₁₃ complexes targeted to the basolateral membrane, while ₁₁-containing receptors showed a nonpolarized distribution.  subunits probably also contain synaptic targeting information. ₅-containing GABA_A receptors are mainly found on somatic and dendritic portions of hippocampal pyramidal neurons membranes, while ₂ subunits are localized to the axon initial segment (Nusser et al. 1996).

If the ₁ subunits follow this pattern, the full-length protein, sufficient for proper targeting in retinal neurons, should contain the subcellular targeting information. Either this is not the case or, alternatively, the translocation machinery of the hippocampal neurons fails to recognize the ₁ subunit. As with GABA_A subunits, ₁ targeting appears to involve interactions with cytoskeletal proteins. A yeast two-hybrid screening demonstrated specific ₁ subunit binding to the cytoskeletal protein MAP1B (Hanley et al. 1999); MAP1B does not bind GABA_A subunits or ₂ subunits. Co-immunoprecipitation experiments suggest that MAP1B serves as a linker protein between ₁ and the cytoskeleton through its actin/tubulin binding domain. Coexpression of MAP1B with ₁ in COS cells results in a more punctate surface localization of the ₁ protein (Hanley et al. 1999). The ₁-MAP1B interaction does more that simple receptor localization; in retinal bipolar cells, disruption of the interaction increases the sensitivity to GABA (Billups et al. 2000). The GABA receptor-associated protein (GABARap) has also been proposed to assemble neurotransmitter receptors through interaction with the cytoskeleton, and GABA_A receptor aggregation by GABARap has a similar functional consequence (Chen et al. 2000). MAP1B is abundantly expressed within the brain, particularly in developing neurons (Sato-Yoshitake et al. 1989). If it provides the subcellular targeting mechanism for ₁ subunits, then the virally transduced ₁ subunits should target properly. However, the presence of MAP1B in hippocampal neurons in itself was insufficient for proper subsynaptic targeting of the ₁ subunit protein.

Three possible explanations for the failure of GABA_C receptor targeting are suggested by the literature. First, hippocampal neurons may lack a yet-unidentified protein necessary for subsynaptic targeting of ₁. Second, the native GABA_C receptor may be composed of more than ₁ subunits. Third, there may have been inadequate posttranslational modification (e.g., phosphorylation) of the targeting machinery. The first possibility derives from recent data on ₂ subunit knock-out mice. Cortical neurons from these mice show loss of the synaptic clustering molecule gephyrin and of synaptic function (Essrich et al. 1998). The ₂ subunit and gephyrin are critical for postsynaptic clustering of GABA_A receptors. Synaptic targeting is suggested to be a two-step process: the  and  subunits determine extrasynaptic versus synaptic destination and the other subunits of the oligomeric receptor determine the finer segregation (Essrich et al. 1998). However, because gephyrin is present in both the retina (Sassoe-Pognetto et al. 1995) and the hippocampus (Craig et al. 1996), it is unlikely to be the putative ₁ targeting protein. Whether gephyrin influences synaptic localization of the GABA_C receptors in the retina has not been reported.

The second possibility is that in vivo GABA_C receptors are heteroligomers between ₁ and another yet-unidentified subunit. This unknown partner could carry the synaptic targeting information, and its absence would render our homoligomeric ₁ receptors incapable of proper targeting in the hippocampus. In fact, molecular cloning has identified three GABA_C receptor-forming subunits, ₁, ₂, and ₃ (reviewed in Bormann 2000). Reverse transcription PCR revealed both ₁- and ₂-mRNA in rod bipolar cells and in all brain regions, although with the highest level of expression in rod bipolar cells (Boue-Grabot et al. 1998). Heteromeric assembly of different -subunits resulting in receptors with novel kinetic and pharmacological properties does occur (Enz and Cutting 1999). Thus the possibility that correct targeting of the GABA_C receptor requires a heterologous assembly ₁ and non-₁ subunits cannot be ruled out.

The third explanation is that the components necessary for proper ₁-subunit targeting were present but somehow not functional in hippocampal neurons. For example, the yeast two-hybrid study clearly demonstrated interaction between ₁ and MAP1B. Whether this interaction is between native or phosphorylated proteins is not known. In neurons, phosphorylated MAP1B predominates in the axons and nonphosphorylated MAP1B in the cell body and dendrites (Sato-Yoshitake et al. 1989). Our anti-MAP1B immunohistochemistry documented the presence of MAP1B in hippocampal neurons and glial cells (Fig. 6D), but its phosphorylation status is unknown.

Virally transduced GABA_A receptor -subunit coassembles with endogenous subunits

In contrast to the failure of ₁ subunits to form subsynaptic GABA_C receptors, virally transduced GABA_A receptor  subunits properly oligomerized and formed functional subsynaptic GABA_A receptors. This conclusion is supported by the emergence of PTXN-resistant mIPSC and immunohistochemical localization of the epitope-tagged -subunit in the dendrites. In fact, coassembly of the  subunit with other subunits is probably necessary for the proper subsynaptic targeting of the oligomeric receptor. However, the virally overexpressed -subunit targeting may not be as stringent as under a more physiological condition indicated by the high degree of overlap between HA- and synaptophysin-immunoreactivity in ₂(mut)-HA neurons (Fig. 6B). Greater targeting fidelity might be restored by a controlled expression of the ₂ subunit through promoter regulation and a titrated expression of the subunit protein. In addition, we cannot specify the subunit composition of the ₂(mut)-containing receptors other than to say that at least one ₂(mut) was incorporated into each functional receptor with the PTXN-resistant phenotype. In vivo, 50% of receptor pentamers contain two different -subunits (Li and DeBlas 1997). Creation of tandem subunits would eliminate this ambiguity and allow for more precise control of the GABA_A receptor subunit composition.

Adenovirus-mediated GABA receptor engineering

Adenoviral vectors can transduce hippocampal neurons both in slice and in culture with high efficiency (Griesbaeck et al. 1997; Wilkenmeyer et al. 1996), and, at least in the short term, without affecting electrical excitability or synaptic transmission (Johns et al. 1999; Lissen et al. 1998; Smith et al. 1997). However, the low PTX-resistant mIPSC frequency observed in our study despite immunohistochemical evidence of proper subsynaptic targeting of the ₂(mut) subunit may be indicative of the first-generation adenovirus vector-induced cellular toxicity. For the GABA receptor, Semliki forest virus (an RNA virus) co-expression of epitope tagged-₁ and ₂ subunits resulted in immunofluorescence in the membrane surface of superior cervical ganglion neurons (Gorrie et al. 1997). Another recent report demonstrated successful targeting of the ₁ subunit after Sindbis virus (an RNA virus) transduction of thalamic organotypic culture (Okada et al. 2000). These vectors or a higher-generation adenovirus vector (Smith and Romero 1999) may be required for a long-term expression of receptor subunits without toxicity.

The possible applications of GABA receptor engineering are many. For example, a recent study of GABA_A receptor ₁-subunit knock-in mice demonstrated that sedative properties of benzodiazepines are mediated by the ₁ subunit while the anxiolytic effects are mediated by the ₂ subunit (McKernan et al. 2000; Rudolph et al. 1999). Direct viral overexpression of GABA receptors driven by the ₂ subunit promoter in the brain might mimic the anxiolytic effect of benzodiazepines without the sedative side effects. Diseases of focal neuronal hyperexcitability, such as medial temporal sclerosis, may be amenable to enhanced inhibition through a localized overexpression of GABA receptors. At the least, the ability to influence GABA receptor subunits in live neurons should shed light on the functional role of GABA receptor subunit heterogeneity and assist in tackling the fundamental questions regarding subcellular trafficking of these receptors.