An Asymmetric Contribution to γ-Aminobutyric Type A Receptor Function of a Conserved Lysine within TM2–3 of α1, β2, and γ2 Subunits*

Mutations that impair the expression and/or function of γ-aminobutyric acid type A (GABAA) receptors can lead to epilepsy. The familial epilepsy γ2(K289M) mutation affects a basic residue conserved in the TM2–3 linker of most GABAA subunits. We investigated the effect on expression and function of the Lys → Met mutation in mouse α1(K278M), β2(K274M), and γ2(K289M) subunits. Compared with cells expressing wild-type and α1β2γ2(K289M) receptors, cells expressing α1(K278M)β2γ2 and α1β2(K274M)γ2 receptors exhibited reduced agonist-evoked current density and reduced GABA potency, with no change in single channel conductance. The low current density of α1β2(K274M)γ2 receptors coincided with reduced surface expression. By contrast the surface expression of α1(K278M)β2γ2 receptors was similar to wild-type and α1β2γ2(K289M) receptors suggesting that the α1(K278M) impairs function. In keeping with this interpretation GABA-activated channels mediated by α1(K278M)β2γ2 receptors had brief open times. To a lesser extent γ2(K289M) also reduced mean open time, whereas β2(K274M) had no effect. We used propofol as an alternative GABAA receptor agonist to test whether the functional deficits of mutant subunits were specific to GABA activation. Propofol was less potent as an activator of α1(K278M)β2γ2 receptors. By contrast, neither β2(K274M) nor γ2(K289M) affected the potency of propofol. The β2(K274M) construct was unique in that it reduced the efficacy of propofol activation relative to GABA. These data suggest that the α1 subunit Lys-278 residue plays a pivotal role in channel gating that is not dependent on occupancy of the GABA binding site. Moreover, the conserved TM2–3 loop lysine has an asymmetric function in different GABAA subunits.

The N-terminal ␥2 subunit Arg-43 residue is conserved across GABA A receptor subunits and systematic Arg 3 Gln mutation in ␣1, ␤2, and ␥2 uncovered a general role for the arginine in receptor assembly (17). Likewise the ␥2 subunit Lys-289 residue is conserved in the extracellular TM2-3 loops of GABA A and glycine receptors, suggesting that information can be revealed about its role by a similar comparative mutagenesis approach.
The mechanism by which ␥2(K289M) reduces channel function is controversial. Baulac and colleagues (5) reported reduced current amplitude when compared with wild-type receptors upon expression of ␣1␤2␥2(K289M) receptors in Xenopus oocytes. By contrast, Bianchi and colleagues (14) described faster GABA-evoked current deactivation without altered ␣1␤3␥2(K289M)-mediated peak current or activation rate when compared with wild-type receptors expressed in human embryonic kidney (HEK293) cells. Using the same cells, examining the rate of current activation following laser initiated release of caged GABA onto ␣1␤2␥2(K289M) and wild-type receptors, Ramakrishnan and Hess (28) concluded that the mutation reduced current activation rate. Homology modeling of the ␣1␤2␥2 receptor onto the structural model of the Torpedo marmorata nicotinic acetylcholine receptor (29) revealed an additional possibility: that the K289M mutation may reduce single channel conductance (30).
The TM2-3 region may participate in the transduction of GABA binding to channel gating by coming into close proximity with N-terminal residues. Indeed an electrostatic interaction between the lysine residue on the ␣1 subunit and acidic residues in loops 2 and 7 may be responsible for intramolecular transduction, coupling GABA binding to channel opening (23).
To address the role of Lys-289 and equivalent lysine residues in the most common (␣1␤2␥2) GABA A receptor, we explored the effects of ␣1(K278M), ␤2(K274M), or ␥2L(K289M) on receptor-surface expression and function. Each of the subunits exhibited distinct phenotypes when mutated, indicating an important but asymmetric contribution of this site to GABA A receptor function.

EXPERIMENTAL PROCEDURES
Cell Culture and Transfection-COS7 cells (ATCC CRL 1651) and HEK293 cells were maintained in Dulbecco's modified Eagle's medium (Invitrogen) supplemented with 10% fetal bovine serum, 2 mM glutamine, 1 mM sodium pyruvate, 100 g/ml streptomycin, and 100 units/ml penicillin in an atmosphere of 5% CO 2 . Exponentially growing cells were transfected by electroporation (400 V, infinity resistance, 125 mF, Bio-Rad Gene Electropulser II) in the case of COS7 cells and calcium phosphate precipitation, in the case of HEK293 cells (17). Cells were transfected with equimolar ratios of GABA A subunit cDNAs. Cells were analyzed 12-18 h and 24 -96 h after transfection for biochemical and electrophysiological experiments, respectively.
DNA Constructions-Murine ␣1, ␤2, and ␥2L subunit cDNAs containing the myc or FLAG tag (between amino acids 4 and 5 of the mature polypeptide) have been described previously (31) and shown to be functionally silent with respect to receptor pharmacology and physiology. The mutant expression constructs ␣1(K278M) Myc , ␤2(K274M) Myc and ␥2L(K289M) Myc were generated by PCR. The fidelity of the final expression constructs was verified by DNA sequencing.
Antibodies-The 9E10 antibody was obtained from 9E10 hybridoma cells (32) and used directly as supernatant without purification. Antibodies to the FLAG epitope were purchased from Sigma. The secondary antibodies, goat anti-mouse Alexa Fluor 568 and goat anti-mouse Alexa Fluor 488, were purchased from Molecular Probes (UK), and goat antimouse horseradish peroxidase from Amersham Biosciences.
Immunofluorescence-COS7 cells were fixed in 3% paraformaldehyde (in PBS) and washed twice in 50 mM NH 4 Cl (in PBS) and blocked (10% fetal bovine serum, 0.5% bovine serum albumin in PBS) for 30 min. Subsequent washes and antibody dilutions were performed in PBS containing 10% fetal bovine serum and 0.5% bovine serum albumin. Following surface labeling, cells were permeabilized by the addition of 0.5% Triton X-100 (10 min), and the immunofluorescence protocol was repeated from the NH 4 Cl step. Cells were examined using a wide-field imaging system (Improvision).
Quantification of Cell-Surface Expression-COS7 cells were plated into 96-well dishes. Eight transfections were used per dish (12 wells per transfection, with nine determinants for each condition). Cells were fixed in 3% paraformaldehyde (in PBS). Cell-surface detection was performed in the absence of detergent, and total expression levels were determined following Triton X-100 (0.5%, 15 min) treatment. Cells were washed twice in 50 mM NH 4 Cl (in PBS) and blocked (10% fetal bovine serum, 0.5% bovine serum albumin in PBS) for 1 h. Subsequent washes were performed in block. Receptor expression was determined using an horseradish peroxidase-conjugated secondary antibody and assayed using 3,3Ј,5,5Ј-tetramethylenebenzidine (Sigma) as the substrate, with detection at 450 nm after 30 min, following the addition of 0.5 M H 2 SO 4 . The reaction rate was determined to remain linear for up to 1 h.
Electrophysiology-The whole cell patch clamp technique was used to record GABA-activated currents from HEK293 cells voltage-clamped at Ϫ60 mV. GABA (100 M) was applied by local pressure ejection from low resistance micropipettes (33). In experiments investigating the modulation of GABA-evoked currents by bath applied flunitrazepam, GABA was applied for 1 s at ϳEC 10 concentrations. Data for concentration-response relationships were recorded by applying GABA or propofol for 4 s. The recording chamber was continuously perfused (5 ml/min) with an extracellular solution comprised of (in mM) NaCl, 140; KCl, 4.7; MgCl 2 , 1.2; CaCl 2 , 2.5; glucose, 10; and HEPES-NaOH, 10 (pH 7.4). The electrode solution contained (in mM): CsCl, 140; MgCl 2 , 2.0; EGTA, 11; ATP (Mg 2ϩ salt) 3; and HEPES-CsOH, 10 (pH 7.4). Junction potentials were nulled with an open electrode in the recording chamber prior to each experiment. The liquid junction potential was trivial (ϳ2 mV), and its inappropriate compensation was ignored. Experiments were performed at room temperature (20 -24°C). Macroscopic GABA-evoked currents were monitored by an Axopatch-200B amplifier, low pass filtered with a cut-off frequency of 2 KHz, and then recorded and digitized using a Digidata 1320A interface (Axon Instruments, Union City, CA) for acquisition at 10 kHz onto the hard drive of a personal computer. Currents were averaged and measured using pCLAMP 8.0 software (Axon Instruments).
Single Channel Recording-Single channel currents recorded from cell-attached and outside-out patches were low-pass filtered at 2 and 1 KHz, respectively (digitized at 10 KHz). Data were acquired as described previously (34). GABA was either applied to outside-out patches at 1 M or 1 mM; there was no significant difference in the observed single channel conductances. GABA (1 mM) was applied to cell-attached patches through the recording electrode, which contained extracellular solution. Patches were voltage-clamped using electrode potentials provided in the figure legends. Sections of digitized data in which unitary events predominated were selected for analysis and were leak subtracted using Clampfit for the creation of all-points amplitude histograms and event lists using Fetchan (pCLAMP 8.0, Axon Instruments).
Analysis of Whole Cell Data-Graphs of GABA concentration-response relationships were fitted using the Hill equation as described previously (33). For fitting propofol concentration-response relationships (normalized to maximum GABA-evoked current) the Hill equation was modified as in Equation 1.
In this equation the whole cell current amplitude activated by propofol (I Prop ) is normalized to that activated by 10 mM GABA (I GMax ). I PMax is the maximum amplitude of the propofol activated current relative to I GMax . EC 50 is the concentration of propofol required to activate half of the maximum I Prop , and H is the slope factor of the concentrationresponse relationship. Current density measurements were calculated from each cell by dividing the peak GABA-or propofol-activated current amplitude (measured in picoamps (pA)) by the cell's capacitance (measured in picofarads (pF)).
Analysis of Single Channel Data-All-points amplitude histograms for single channel recordings were fitted with multiple Gaussians (least squares minimization) to amplitude histograms using the Simplex method within pSTAT (pCLAMP 8.0). The amplitude of the single channel current recorded from each patch was determined from the difference between the mean current amplitudes determined from the Gaussians fitted to the closed-and unitary open-state currents. Single channel conductances are reported as the chord conductance is the holding potential, and E rev is the mean reversal potential of GABA-evoked single channels derived from linear fits to current-voltage relationships. In several outside-out patch recordings two unitary conductances were evident. Our analysis of single channels was restricted to the main state.
To quantify channel open times event lists were generated from single channel data obtained from cell-attached patches. Fetchan (pCLAMP 8.0) was used to create event lists of unitary events recorded from cell-attached patches using the 50% threshold detection method (35), having determined baseline and open current amplitudes from all points amplitude histograms. Event lists were analyzed using pSTAT yielding values of mean channel open time. Event lists were also pooled from multiple patches to obtain a representative sample of events (Ն4 patches). All such data were included in the open time histogram (10 bins per decade) plotted with a square root ordinate and logarithmic abscissa (36). The maximum likelihood method was used to fit the sum of three exponentials to open time histograms (pSTAT), omitting those events from the fit that were briefer than 0.15 ms that were compromised by the system dead-time (37).
Statistics-All data are expressed as the arithmetic mean Ϯ S.E. Unless otherwise stated, statistical analysis involved analysis of variance (ANOVA) with the posthoc Tukey's test.

Conservation of a TM2-3 Loop Lysine between GABA A and Glycine
Receptor Subunits-A comparison of the primary sequence of GABA A receptor ␣, ␤, and ␥ subunits within the extracellular loop, between the transmembrane domains TM2-3, reveals a high level of conservation between subunits of the same class ( Fig. 1). A consensus sequence R-LPK-Y exists between all ␣␤␥ subunits, with the proline residue being conserved in all members of the Cys-loop receptor superfamily, including receptors for acetylcholine, 5-hydroxytryptamine, and glycine. Interestingly, the lysine (Lys-289) residue, associated with epilepsy when mutated in the ␥2 subunit (5), is conserved in ␣ and ␤ subunits (Arg in ␥3, ␦, and ⑀, Asn in , and His in ) as well as in glycine receptors, where it has been implicated in hyperekplexia (19). Intriguingly, this lysine residue present in the ␣1 (Lys-278) and ␤2 (Lys-274) subunits has been implicated in the gating of GABA A receptors and may play a conserved and essential role in all subunits in receptor function (23,24). We undertook, therefore, to examine the effect of the mutation in ␣1(K278M), ␤2(K274M), and ␥2L(K289M) subunits.
The Role of Lysine at the Homologous Position in the ␣1 Subunit on Transport to the Cell Surface-To determine, qualitatively, the ability of the ␣1(K278M) subunit to access the cell surface, we examined its cellular distribution when expressed in COS7 cells. COS7 cells were used in these studies due to their clear definition of intracellular compartments (38). The existence of surface receptors was determined in the absence of detergent using anti-Myc antibodies and Alexa Fluor 488 secondary antibodies. Following permeabilization, cells were re-probed as above, using Alexa Fluor 568 secondary antibodies. As observed previously for wild-type ␣1 Myc (31), the ␣1(K278M) Myc subunit could not access the cell surface when expressed alone (data not shown). When ␣1 Myc was co-expressed with the ␤2 subunit, there was robust cellsurface staining ( Fig. 2A, upper right panel). Likewise, the co-expression of ␣1(K278M) Myc with the ␤2 subunit produced robust cell-surface labeling ( Fig. 2A, ␣*␤, lower right panel) as well as strong intracellular labeling. Identical results were observed when ␣1(K278M) Myc was coexpressed with ␤2 and ␥2L FLAG subunits and immunofluorescence was performed via the FLAG epitope on the ␥2L subunit (Fig. 2B). Because the ␥2L subunit cannot access the surface in the absence of either the ␣1 or ␤2 subunits (31), its surface expression is a faithful indicator of the presence of ␣1␤2␥2L receptors. From hereafter '␥2Ј refers to the ␥2L subunit, which was used throughout this study.
To quantify our observations, we used the cell ELISA technique to compare surface and total expression levels. Cell-surface expression (in the absence of detergent) of the ␣1(K278M) Myc is presented as a percentage of total (in the presence of detergent) levels and normalized to wild-type controls (␣1 Myc ) performed in parallel. Using this approach (Fig. 2C), the cell-surface level for ␣1(K278M) Myc ␤2 receptors was 66 Ϯ 11%, compared with the normalized wild-type ␣1 Myc ␤2 receptor level of 100 Ϯ 11%. Similarly, the cell-surface level for ␣1(K278M) Myc ␤2␥2 FLAG receptors was determined (via FLAG epitope) to be 87 Ϯ 34%, compared with the normalized wild-type ␣1 Myc ␤2␥2 FLAG receptors at 100 Ϯ 13%. Thus, the presence of K278M in the ␣1 subunit does not have a major impact on biogenesis or the surface transport of ␣1(K278M)␤2␥2 receptors.
The Role of Lysine at the Homologous Position in the ␤2 Subunit on Transport to the Cell Surface-To determine, qualitatively, the ability of the ␤2(K274M) subunit to access the cell surface, we examined its cellular distribution when expressed in COS7 cells. As observed previously for wild-type ␤2 Myc (31), the ␤2(K274M) Myc subunit could not access the cell surface when expressed alone (data not shown). When ␤2 Myc was co-expressed with the ␣1 subunit, there was robust cell-surface staining (Fig. 3A, upper right panel). In contrast, when ␤2(K274M) Myc was co-expressed with the ␣1 subunit there was no cell-surface labeling (Fig. 3A, ␣␤*; lower right panel). Instead, strong intracellular labeling within the endoplasmic reticulum was observed, as evidenced by the characteristic reticular pattern, typically observed in COS7 cells (38). We obtained identical results when co-expressing ␤2(K274M) Myc with the ␣1 and ␥2 FLAG subunits, performing immunofluorescence via the FLAG epitope on the ␥2 subunit to detect the presence of all three subunits (Fig. 3B). Quantification of these findings, using cell ELISA (Fig. 3C), revealed the cell-surface level for ␣1␤2(K274M) Myc receptors to be 4.2 Ϯ 18.9%, compared with the normalized wild-type ␣1␤2 Myc receptor at 100 Ϯ 11%. Similarly, the cell-surface level of the ␣1␤2(K274M) Myc ␥2 FLAG receptor was determined (via FLAG epitope) to be 12.7 Ϯ 12.6%, compared with the normalized wild-type ␣1 Myc ␤2␥2 FLAG receptor level of 100 Ϯ 31%. Thus, the presence of K274M in the ␤2 subunit perturbs receptor distribution, preventing the significant expression of cell-surface ␣1␤2(K274M)␥2 receptors.
The Role of Lysine in the ␥2 Subunit on Transport to the Cell Surface-To determine, qualitatively, the ability of the ␥2(K289M) to access the cell surface, we examined its cellular distribution when expressed in COS7 cells. As observed previously for wild-type ␥2 Myc (31), the ␥2(K289M) Myc subunit could not access the cell surface when expressed alone, or in the presence of ␣1 or ␤2 subunits (data not shown). Co-expression of ␥2 Myc with ␣1 and ␤2 subunits produced robust cell-surface staining (Fig. 4A, upper right panel). Likewise, coexpression of ␥2(K289M) Myc with ␣1 and ␤2 subunits also produced robust cell-surface labeling (Fig. 4A, ␣␤␥*; lower right panel) as well as strong intracellular labeling. To quantify these observations, we used the cell ELISA technique to compare surface and total expression levels. Cell-surface expression (in the absence of detergent) of the ␣1␤2␥2(K289M) Myc receptor is presented as a percentage of total (in the presence of detergent) levels and normalized to wild-type controls (␣1␤2␥2 Myc ) performed in parallel. Using this approach (Fig. 4C), the cell-surface level for ␣1␤2␥2(K289M) Myc receptors was 108 Ϯ 26%, compared with the normalized wild-type ␣1␤2␥2 Myc receptor level of 100 Ϯ 31%. Thus, the presence of the epilepsy mutation K289M in ␥2 did not perturb the biogenesis or the surface transport of ␣1␤2␥2(K289M) receptors.
We next examined the effect of the equivalent Lys 3 Met mutation in the ␣1 and ␤2 subunits. Unlike the ␥2(K289M) mutant, both the ␣1(K278M) and ␤2(K274M) mutants caused a significant rightward shift in the GABA concentration-response relationships of ␣1(K278M)␤2␥2 and ␣1␤2(K274M)␥2 receptors, respectively (Fig. 5A). The EC 50 values were 42 Ϯ 6 M (n ϭ 5) for the ␣1(K278M)␤2␥2 receptor and 55 Ϯ 3 M (n ϭ 5) for the ␣1␤2(K274M)␥2 mutant receptor. In both cases the mutant subunits caused a substantial decline in the current density, once again the ␤2(K274M) mutant had the most deleterious effect (Fig. 5B). Neither the ␣1(K278M) nor the ␤2(K274M) mutant subunit altered the potentiation of GABA-evoked currents by flunitrazepam (100 nM) (Fig. 5C).  (30). Such a location could imply a role for the basic residue in ion conduction in which case its replacement by the uncharged methionine would be expected to cause a reduction in single channel conductance. Assuming a stoichiometry of 2␣:2␤:1␥ one would expect such an effect to be greatest when the Lys 3 Met mutation is present in either ␣ or ␤ subunits (39). We tested this hypothesis by recording single GABAactivated channels from outside-out patches excised from cells expressing ␣1␤2␥2, ␣1(K278M)␤2␥2, ␣1␤2(K274M)␥2, and ␣1␤2␥2(K289M) receptors. In all cases there was a linear relationship between single channel current amplitude and voltage, with no difference in their equilibrium potentials (Fig. 6E ). In all cases GABA-activated single channels with similar amplitudes were recorded from patches at Ϫ60 mV (Fig. 6,  A-D). The chord conductances for unitary events mediated by ␣1␤2␥2, ␣1(K278M)␤2␥2, ␣1␤2(K274M)␥2, and ␣1␤2␥2(K289M) receptors   (Fig. 7). This approach ensures steady-state exposure of receptors to a saturating concentration of GABA. Stretches of data were analyzed in which unitary events occurred Ͼ90% of the time. Three methods for comparing channel open time of wild-type and mutant receptors revealed that both ␣1(K278M) and ␥2(K289M) made openings briefer, whereas ␤2(K274M) had no effect (Fig. 7). The open time histogram generated from events mediated by wild-type ␣1␤2␥2 receptors, when fitted with the sum of three exponentials (14,37), revealed three time constants (1)(2)(3) representing open times of brief, medium duration, and long lasting events (Fig. 7A). The average open time of ␣1␤2␥2 receptors was 3.9 Ϯ 0.3 ms (n ϭ 4) corresponding to the midpoint of the cumulative distribution curve (Fig. 7, E and F ) a less conventional method of displaying the data that nevertheless provides a convenient comparison of the full data range (35). By contrast to wild-type receptors, ␣1(K278M)␤2␥2 receptors exhibited brief openings with only a minor contribution of long lasting events represented by the 3 component (Fig. 7B). The mean open time of ␣1(K278M)␤2␥2 receptors (1.4 Ϯ 0.4 ms, n ϭ 7) was significantly briefer ( p Ͻ 0.001) than the mean open times of both wild-type ␣1␤2␥2 receptors and mutant ␣1␤2(K274M)␥2 receptors (3.5 Ϯ 0.4 ms, n ϭ 5) (Fig. 7F ). The open times of ␣1␤2(K274M)␥2 receptors were similar to those of wildtype receptors suggesting that the conserved lysine in the ␤2 subunit does not participate in gating kinetics. By contrast, but in agreement with a previous report (14), ␣1␤2␥2(K289M) receptors have disrupted gating kinetics (Fig. 7D) (23). If this is the case, such mutations may not affect the ability of anesthetics, such as propofol, to directly activate the GABA A receptor (40) through sites distinct from the GABA binding site (41). We examined the ability of propofol relative to GABA to activate ␣1␤2␥2, ␣1(K278M)␤2␥2, ␣1␤2(K274M)␥2, and ␣1␤2␥2(K289M) receptors. Propofol activates GABA A receptors at concentrations between 3 and 100 M. The blocking effect of propofol (33,40,41) reduces the amplitude of the propofol (300 M)-activated current and is associated with a pronounced surge current upon cessation of application (Fig. 8A). The blocking effect complicates attempts to determine the EC 50 of propofol (Fig. 8, B and  C). Surge currents were negligible following the application of 100 M propofol suggesting that there is minimal blockade with this concentration. Therefore we fitted data points between 3 and 100 M propofol with the modified Hill equation to obtain estimates of EC 50 values and the efficacy of propofol relative to GABA as an activator of current. We excluded the final data point obtained with 300 M propofol due to blockade associated with this concentration. Using this approach the EC 50 value for propofol as an activator of ␣1␤2␥2 receptors was 32 Ϯ 3 M (slope factor ϭ 2.0 Ϯ 0.3) (Fig. 8B). The apparent maximum efficacy of propofol, relative to GABA (10 mM), was 56 Ϯ 3%. The mutant ␣1(K278M) subunit reduced the apparent potency of propofol, shifting the concentration-response relationship to the right (Fig. 8B) without altering the maximum current amplitude activated by propofol (100 M) relative to GABA (10 mM). Assuming an unaltered maximum efficacy of propofol relative to GABA (56%) the fit to the propofol concentration-response relationship provided an estimate of the EC 50 of propofol as an agonist of the ␣1(K278M)␤2␥2 receptor of 68 Ϯ 2 M (slope factor ϭ 3.7 Ϯ 0.2).
We determined the propofol current density by expressing peak propofol-activated current amplitudes as a function of cell capacitance (Fig. 8D). The pattern of propofol current densities for wild-type and mutant receptors resembles that seen for GABA current densities (Fig. 5B).

DISCUSSION
We examined the impact of replacing a TM2-3 loop lysine residue by methionine in ␣1(K278M), ␤2(K274M), and ␥2(K289M) subunits on GABA A receptor surface expression and function. The ␥2(K289M) mutation is associated with hereditary epilepsy characterized by febrile seizures (5). By introducing the Lys 3 Met mutation into each of the three major GABA A subunits expressed in the brain we examined whether the role of the homologous lysine was similar at each position in the heteropentamer.
We assayed surface expression of epitope-tagged receptors qualitatively and quantitatively using epifluorescence microscopy and ELISA approaches, respectively. Surface expression was compared with functional receptor expression by measuring GABA-activated current density. We used outside-out and cell-attached patch recording configurations to evaluate the effects of the Lys 3 Met mutation on GABA- activated single channel conductance and open times, respectively. Finally we examined whether deficits in channel function associated with mutant subunits were specific to GABA activation or generalized to activation by propofol, a GABA A receptor agonist that acts through a site distinct from that of GABA (41).
Using these varied approaches we found that the conserved lysine within the extracellular TM2-3 loop plays an asymmetric role in GABA A receptor function and expression that is dependent on the subunit in which it is located.
The ␣1(K278M) and ␥2(K289M) constructs had no effect on the level of receptor surface expression, whereas ␤2(K274M) caused a marked reduction in cell-surface receptors. A recent report demonstrates that ␥2S(K289M) has a temperature-dependent effect on GABA A receptor expression. Elevating the temperature of HEK293 cells from 37 to 40°C reduced cell-surface expression of ␣1␤2␥2S receptors, an effect that wasmorepronouncedinthepresenceof␥2S(K289M).Thistemperaturedependent effect may participate in the deficit in inhibitory signaling underlying febrile seizures in individuals harboring the mutation (18). Our functional studies were performed at room temperature, and cellsurface expression assays were performed on cells cultured at 37°C prior to fixation. Furthermore it is worth noting that we used the ␥2L subunit variant, which may have different trafficking properties compared with those of ␥2S.
Unaltered surface expression of ␣1␤2␥2(K289M) receptors compared with wild-type receptors coincided with unaltered GABA-evoked current density. By contrast, despite having no effect on surface expression, ␣1(K278M) reduced GABA-evoked current density. Two mechanisms contribute to current density: expression levels of cell-surface receptors and GABA A receptor function. Because there was no difference in the surface expression of ␣1(K278M)␤2␥2 receptors compared with wild-type receptors, reduced current density must result from a functional deficit. Consistent with this assertion GABA had a substantially reduced apparent potency as an activator of ␣1(K278M)␤2␥2 receptors compared with wild-type receptors; the GABA concentration-response relationship was shifted to the right, reflecting a 4-fold increase in the GABA EC 50 . Such a reduction in the potency of GABA could be caused by reduced binding affinity, impaired transduction of GABA-binding to channel activation, or both (42). Analysis of single channels activated in cell-attached patches identified a deficit in channel function caused by ␣1(K278M): the durations of channel openings were substantially reduced compared with those of wild-type receptors. This functional deficit presumably reduces the efficacy of GABA leading to a diminished current density.
In contrast to wild-type ␣1␤2␥2, ␣1␤2␥2(K289M), and ␣1(K278M)-␤2␥2 receptors, ␣1␤2(K274M)␥2 receptors had substantially reduced surface expression, suggesting that the ␤2 TM2-3 region may be important in receptor biogenesis and/or transport. The consequence of reduced surface expression of ␣1␤2(K274M)␥2 receptors is a dramatic reduction in current density. GABA-activated channels mediated by ␣1␤2(K274M)␥2 receptors were indistinguishable from those mediated by wild-type receptors in terms of their open times, suggesting a lack of a role for the ␤2 Lys-274 in gating by GABA. However, the GABA concentration-response relationship was shifted to the right by ␤2(K274M) consistent with a reduced GABA potency. It seems likely that this shift is caused by a reduction in GABA affinity for the mutant receptor.
Replacement of the conserved TM2-3 loop lysine by either alanine (43) or aspartate (23,24) in either the ␣1 or the ␤2 subunit also increases the GABA EC 50 . In the case of the human ␣1(K279D) subunit (equivalent to mouse ␣1(K278)) the deficit in receptor function could be res-cued by simultaneously introducing either D57K or D149K mutations into the subunit's N-terminal domain loops 2 and 7, respectively (23). These data support the hypothesis that, upon activation of wild-type receptors by GABA, there is an interaction between acidic residues in the N-terminal domain of the ␣1 subunit and the lysine in the TM2-3 loop. This mechanism appears to be unique to the ␣1 subunit of the GABA A receptor, because similar simultaneous charge reversals in the ␤2 subunit fail to recover the reduced GABA potency induced by the ␤2(K274D) mutation (24). This is an intriguing difference that again demonstrates asymmetry in the role of the conserved TM2-3 lysine in GABA A receptor function. The failure of charge reversal to recover the potency of activation of ␣1␤2(K274D)␥2 receptors for GABA may also point to a role of ␤2(K274) in GABA binding affinity rather than receptor activation.
The ␥2(K289M) mutant reduced GABA-gated channel open time albeit to a lesser extent than did ␣1(K278M). The attenuation of open time induced by ␥2(K289M) was not sufficient to significantly impact current density. Our data examining ␣1␤2␥2(K289M) receptors in cell-attached patches agree with those previously reported for ␣1␤3␥2(K289M) receptors in outside-out patches (14) suggesting that decreased inhibition underlying epilepsy associated with the mutation could be caused in part by briefer channel openings.
Interestingly, none of the mutant subunits significantly altered potentiation by flunitrazepam. Similarly there was no difference in the potentiation by a maximal concentration of diazepam of GABA-evoked currents mediated by wild-type ␣1␤3␥2 and mutant ␣1␤3␥2(K289M) receptors (14). These data suggest that the conserved lysine is not responsible for transducing the potentiating effects of benzodiazepines. However, this needs to be tested over a broader range of benzodiazepine concentrations in future studies.
Febrile seizures, the most common of childhood seizures, are linked to several environmental factors and a host of mutations, several of which affect genes encoding ion channels (44). A more complete understanding of the molecular mechanisms underlying each ion channel deficit associated with familial febrile seizures may lead to the development of individualized pharmacotherapies. The observation that ␣1␤2␥2(K289M) and ␣1␤3␥2(K289M) receptors (14) have reduced GABA-evoked channel open times, compared with their respective wild-type receptors, suggests that drugs that prolong open time may help rectify the functional deficit. Benzodiazepines increase the amplitude of sub-maximal GABA-evoked currents by increasing the frequency of GABA-gated channel bursts (45). This contrasts with the mechanism of enhancement by barbiturates and other general anesthetics that predominantly increase channel burst duration (45,46). It is possible that the anesthetic potentiation of GABA-evoked responses mediated by mutant ␣1(K278M)␤2␥2 and ␣1␤2␥2(K289M) receptors will be enhanced through "correction" of the functional deficit of brief open time. This hypothesis will be tested in future single channel studies.
Deficits in the function of GABA A receptors containing ␣1(K278M) or ␥2(K289M) may indicate a role for the conserved lysine residue in the transduction of GABA binding to channel gating. Indeed this residue appears ideally situated for such a mechanism. If so, is this intramolecular transduction mechanism specific to GABA binding or can it be generalized to agonists that act through different sites on the GABA A receptor? Several general anesthetics, including the intravenous agent propofol, directly activate GABA A receptors when applied in the mid to high micromolar concentration range in the absence of GABA (40). Propofol also potentiates GABA-evoked currents starting in the low micromolar range and at concentrations Ͼ100 M causes a concentration-dependent receptor blockade (33). Direct activation of GABA A receptors by propofol occurs through a distinct mechanism from that of GABA-evoked activation as demonstrated by the observation that propofol activates homomeric ␤3 receptors that are resistant to activation by GABA (41). Furthermore, activation of recombinant ␣1␤2␥2 receptors 4 by propofol is resistant to blockade by the competitive GABA antagonist SR95531 (gabazine). Despite having different sites on the GABA A receptor, activation by both propofol and GABA is adversely affected by the ␣1(K278M) mutation. The disruption of propofol activation by ␣1(K278M) demonstrates that intramolecular transduction of binding to gating by this residue is not specific to GABA, instead the ␣1(Lys-278) plays a role in gating by diverse agonists.
The subunit-dependent effect of the Lys 3 Met mutation on GABA A receptor activation by GABA and propofol is reminiscent of the asymmetric effects of equivalent mutations in glycine receptors (25,26,47). Incorporation of either the human startle disease ␣1(K276E) mutation or other non-conservative mutations at the same locus (including K276C) into recombinant homomeric ␣1 glycine receptors causes a substantial rightward shift in the glycine concentration-response relationship (25,47). By contrast, the equivalent Lys 3 Cys mutation in the ␤ subunit has no effect on the glycine concentration-response relationship (26). It appears that the functional roles of homologous lysines in the TM2-3 loops of subunits of both GABA A and glycine receptors are different.
A structural model of the ␣1␤2␥2 GABA A receptor, produced by homology modeling its amino acid sequences onto the 4-Å resolution cryoelectron microscopic structure of the Torpedo marmorata nicotinic acetylcholine receptor (29), suggests that the conserved TM2-3 lysine lies in the conduction pathway within the channel's outer vestibule (30). On the basis of such a model it was proposed that removal of the basic residue from the mutant ␥2(K289M) subunit may reduce the number of Cl Ϫ ions within the outer vestibule thus reducing single channel conductance. Such a deficit in GABA A receptor function may contribute to the epileptic phenotype. However, there was no difference in the conductance of GABA-activated channels mediated by wild-type ␣1␤2␥2 and ␣1␤2␥2(K289M) receptors in this study in agreement with a previous report (14). There is likely to be only one ␥2 subunit per receptor perhaps minimizing the contribution of the ␥2(K289) to single channel conductance. Assuming a stoichiometry of 2␣1:2␤2:1␥2 (39) a greater contribution of ␣1(K278) and ␤2(K274) residues to single channel conductance may be expected. However, neither the ␣1(K278M) nor the ␤2(K274M) mutations had any effect on single channel conductance.
In conclusion, the conserved TM2-3 loop lysine plays an asymmetric role in GABA A receptor membrane expression and function. In the ␥2 subunit, under conditions of unaltered cell-surface expression, the K289M mutation reduces channel open time. In the ␤2 subunit the K274M mutation impairs surface expression and reduces both the apparent potency of GABA and the efficacy of propofol as an agonist relative to GABA. In the ␣1 subunit the K278M mutation reduces the apparent potency of gating by both GABA and propofol without altering surface expression. Furthermore, the ␣1(K278M) mutation substantially reduced GABA-evoked channel open times.
Our data suggest that conserved TM2-3 lysine residues in the ␣1 and ␥2 subunits stabilize the open state of the receptor perhaps through a previously described electrostatic interaction with N-terminal acidic residues (23). The Lys-278 residue in the ␣1 subunit appears to be required for normal activation by both GABA and propofol suggesting that its putative role in initiating gating is not dependent on GABA binding.