Structural Link between γ-Aminobutyric Acid Type A (GABAA) Receptor Agonist Binding Site and Inner β-Sheet Governs Channel Activation and Allosteric Drug Modulation*

Background: Structural elements and protein movements underlying GABAA receptor activation are not completely resolved. Results: Glycine insertions in the extracellular β4-β5 linker decrease GABA activation, invert antagonist efficacy, and reduce allosteric modulation. Conclusion: β4-β5 linker is critical for mediating actions of GABAA receptor orthosteric and allosteric ligands. Significance: Detailed structural-functional understanding of GABAA receptor activation is key to understanding its modulation in diseased and healthy states. Rapid opening and closing of pentameric ligand-gated ion channels (pLGICs) regulate information flow throughout the brain. For pLGICs, it is postulated that neurotransmitter-induced movements in the extracellular inner β-sheet trigger channel activation. Homology modeling reveals that the β4-β5 linker physically connects the neurotransmitter binding site to the inner β-sheet. Inserting 1, 2, 4, and 8 glycines in this region of the GABAA receptor β-subunit progressively decreases GABA activation and converts the competitive antagonist SR-95531 into a partial agonist, demonstrating that this linker is a key element whose length and flexibility are optimized for efficient signal propagation. Insertions in the α- and γ-subunits have little effect on GABA or SR-95531 actions, suggesting that asymmetric motions in the extracellular domain power pLGIC gating. The effects of insertions on allosteric modulator actions, pentobarbital, and benzodiazepines, have different subunit dependences, indicating that modulator-induced signaling is distinct from agonist gating.

Background: Structural elements and protein movements underlying GABA A receptor activation are not completely resolved. Results: Glycine insertions in the extracellular ␤4-␤5 linker decrease GABA activation, invert antagonist efficacy, and reduce allosteric modulation. Conclusion: ␤4-␤5 linker is critical for mediating actions of GABA A receptor orthosteric and allosteric ligands. Significance: Detailed structural-functional understanding of GABA A receptor activation is key to understanding its modulation in diseased and healthy states.
Rapid opening and closing of pentameric ligand-gated ion channels (pLGICs) regulate information flow throughout the brain. For pLGICs, it is postulated that neurotransmitter-induced movements in the extracellular inner ␤-sheet trigger channel activation. Homology modeling reveals that the ␤4-␤5 linker physically connects the neurotransmitter binding site to the inner ␤-sheet. Inserting 1, 2, 4, and 8 glycines in this region of the GABA A receptor ␤-subunit progressively decreases GABA activation and converts the competitive antagonist SR-95531 into a partial agonist, demonstrating that this linker is a key element whose length and flexibility are optimized for efficient signal propagation. Insertions in the ␣and ␥-subunits have little effect on GABA or SR-95531 actions, suggesting that asymmetric motions in the extracellular domain power pLGIC gating. The effects of insertions on allosteric modulator actions, pentobarbital, and benzodiazepines, have different subunit dependences, indicating that modulator-induced signaling is distinct from agonist gating.
Electrochemical signaling in the CNS depends on ligandgated ion channels (LGICs). 2 These proteins couple the binding of neurotransmitter to the rapid opening of an integral ionconducting pore. The "Cys-loop" LGIC family of receptors comprises pentameric proteins (pLGICs) that include nicotinic acetylcholine receptors (nAChRs), glycine receptors (GlyRs), GABA type A receptors (GABA A Rs), and serotonin type-3 receptors. Although a structural picture of pLGICs is rapidly emerging from the 4 Å resolution cryo-EM structure of the Torpedo nAChR (1), the crystal structures of the extracellular binding domain of the nAChR ␣-subunit (2) liganded and unliganded acetylcholine-binding proteins (AChBP), which are homologs of the extracellular binding domain (3,4), the crystal structures of full-length prokaryotic pLGIC homologs from Erwinia chrysanthemi (ELIC) and Gloeobacter violaceus (GLIC) (5)(6)(7), as well as the recent crystal structure of a related invertebrate pLGIC (8), our understanding of the structural elements and protein movements that couple neurotransmitter binding to channel gating is still under debate. For receptors in this superfamily, the neurotransmitter binding site is located in the extracellular N-terminal domain between adjacent subunits formed by at least six noncontiguous protein regions (loops A-F), whereas the channel gate is located 50 Å away in the trans-membrane region (9). Neurotransmitter binding is believed to trigger structural movements at the binding site that are propagated as a conformational wave to the channel gate (10). The secondary structure of the extracellular domain is predominantly composed of 10 ␤-strands arranged in two sheets, inner and outer, that form a ␤-sandwich (see Fig. 1). Three flexible loops (␤4-␤5, ␤6-␤7, and ␤8-␤9 linkers) link the inner ␤-sheet to the outer sheet.
One model of receptor activation based on nAChR structural studies suggests that agonist binding in the extracellular domain induces a clockwise rotation of the extracellular inner ␤-sheet in two out of five subunits, which triggers movements in the trans-membrane helices that result in channel gating (11). Comparison of the recent ELIC and GLIC bacterial channel structures (closed and open, respectively) suggests that channel activation is accompanied by an anti-clockwise concerted twist of each extracellular ␤-sandwich domain (6,7,12). In homomeric glycine receptors, receptor activation is believed to occur via a reorganization of the extracellular ␤-sandwich hydrophobic core and the negative subunit interface loops (13).
In the GABA A R, the ␤-subunit ␤4-␤5 linker links the loop A region of the GABA binding site to the inner ␤-sheet (see Fig. 1) and is in an ideal position to propagate initial binding site movements to gating movements in the channel domain. Consistent with this idea, molecular dynamic simulations of the nAChR identified motions in the ␤4-␤5 linker that were correlated with motions in loops 2, 7, and M2-M3 near the extracellular mouth of the channel (14), and mutations in the nAChR ␣-subunit ␤4-␤5 linker alter channel gating (15,16). Based on the crystal structures of the nAChR ␣-subunit extracellular domain, AChBP, GLIC, ELIC, and the invertebrate glutamate-activated chloride channel (2,3,(5)(6)(7)(8), this linker region spans each subunit and is relatively unstructured. Here, we inserted glycine residues in the ␤4-␤5 linkers of the GABA A R ␣1-, ␤2-, and ␥2-subunits to alter their length and flexibility and examined the effects on GABA-mediated channel activation and pentobarbital (PB)-mediated channel activation and on benzodiazepine (BZD) modulation of GABA responses. The absence of a ␤-carbon allows glycine to access energetically prohibited protein dihedral angles (17).
Glycine insertions in the ␤-subunit (␤Gly) 3 reduced GABA binding and GABA-induced channel gating, and surprisingly, converted the competitive antagonist SR-95531 into a partial agonist, demonstrating that this linker is a key structural element whose length and flexibility are optimized for transducing GABA binding to efficient channel gating. Insertions in the ␣and ␥-subunits had little to no effect on GABA or SR-95531 actions, supporting the idea that asymmetric subunit motions in the extracellular domain help power GABAmediated gating. Moreover, the effects of the glycine insertions on PB activation and BZD modulation were much smaller and had different subunit dependences, indicating that the structural mechanisms underlying GABA, PB, and BZD actions are distinct.
Oocyte Electrophysiology-Oocytes were voltage-clamped at Ϫ80 mV in a 200-l chamber and continuously perfused with ND96 at 10 ml/min, and data were acquired as described previously (20). Stock solutions of 1 M GABA, 10 mM SR-95531 (Sigma-Aldrich), and 10 mM flurazepam (FZ) (Research Biochemicals, Natick, CA), were prepared in ND96, stored at Ϫ20°C, and thawed once before use. Stock solutions of 30 or 50 mM PB (Research Biochemicals, Natick, CA) were prepared fresh on the day of the experiment.
Concentration Response and Data Analysis-Concentration response analyses were performed as described previously (21). For GABA concentration response experiments, each test concentration was preceded by a low nondesensitizing concentration to correct for the drift in I GABA over the course of the experiment. Currents induced by each test concentration were normalized to the corresponding nondesensitizing concentration before curve fitting.
Concentration responses for PB direct activation of wildtype (WT) and mutant receptors were measured using two methods. In the first method, each test PB concentration was preceded by a low nondesensitizing PB concentration to correct for any drift in PB-induced current during the experiment. Responses to each test PB concentration were then normalized to its corresponding low nondesensitizing concentration, prior to curve fitting. In the second method, the PB-induced currents were not normalized to a low, nondesensitizing PB concentration. The curve fits and calculated values obtained using the two methods were not different, and thus, the data were pooled for statistical analysis. At high micromolar concentrations and above, PB blocks GABA A R current responses. The relief of channel block upon drug wash yields a characteristic tail current. For PB concentration-response curves, PB current amplitudes at high micromolar concentrations were measured using the tail currents.
SR-95531 IC 50 experiments were performed as described previously (22). Oocytes expressing WT or mutant receptors were challenged with EC 50 GABA concentration (except for experiments involving ␤Gly2, -4, and -8, which were performed at 30 mM GABA) followed by co-application of the same concentration of GABA and a test concentration of SR-95531. GABA-induced currents in the presence of increasing SR-95531 concentrations were then normalized to the GABA response in the absence of SR-95531 before curve fitting.
FZ potentiation experiments were performed at EC 4 -8 GABA. Potentiation is defined as ((I GABAϩBZD /I GABA ) Ϫ 1), where I GABAϩBZD is the GABA-mediated current in the presence of flurazepam and I GABA is the GABA-mediated current in the absence of flurazepam. Because GABA EC 50 values could not be reliably measured for ␤Gly2-, -4-, and -8-containing receptors, they were not tested for flurazepam potentiation.
Nonlinear regression analysis for GABA, PB, FZ, and SR-95531 concentration response experiments was performed using the GraphPad (San Diego, CA) Prism 4 software. GABA and PB concentration responses were fit to the following equation: where I is the peak response (including the tail current for PB) to a given GABA or PB concentration, I max is the current amplitude (including the tail current for PB) of the maximal GABA-or PB-evoked current, EC 50 is the concentration of GABA or PB that produces a half-maximal response, [A] is the agonist concentration, and n is the Hill coefficient. For SR-95531 competition experiments, inhibition was calculated as I GABAϩSR-95531 / I GABA . Data were fit to the following equation: 50 is the concentration of antagonist that blocks half of I GABA ; [Ant] is the concentra-  Table 1. Right panel, representative current traces from oocytes expressing ␣Gly2␤␥, ␣␤Gly2␥, and ␣␤␥Gly2 receptors elicited by increasing concentrations of GABA (M or mM). tion of the antagonist; and n is the Hill coefficient. K I values were calculated using the Cheng-Prusoff/Chou equation (23,24): is the concentration of GABA used, and EC 50 is the concentration of GABA that elicits a half-maximal response.

Current Rise Time and Maximal Current Amplitude Ratio
Analysis-10Ϫ90% apparent rise times for GABA-induced currents at saturating GABA concentrations (10 mM for WT and 100 mM for ␤Gly1) were calculated using a built-in feature in the WinWCP data acquisition software (provided by J. Dempster, Univ. of Strathclyde, Glasgow, UK). Data from n Ͼ 3 oocytes from at least two different batches were pooled for statistical analysis.
For maximum PB versus maximum GABA-induced current ratio determinations, each oocyte (WT or ␤Gly1) was first challenged with maximum GABA concentration (10 mM for WT and 100 mM for ␤Gly1) and then allowed to recover completely by washing sufficiently to remove any trace remnants of GABA from previous exposure, which was followed by application of maximum PB concentration (10 mM for WT and 30 mM for ␤Gly1). GABA-and PB-induced currents at saturating concentrations were then measured, and ratios (maximum PB/maximum GABA) were determined by dividing the maximum PB current amplitude by maximum GABA current amplitude for each experiment. Ratiometric data from n Ͼ 3 oocytes from at least two different batches were pooled for statistical analysis. PB current amplitudes at saturating concentrations were measured using tail currents.
Statistical Analysis-LogEC 50 values for GABA, PB, FZ, potentiation, and LogK I values for SR-95531 concentration responses were analyzed using one-way ANOVA followed by a post hoc Dunnett's test to determine the level of significance between wild-type (WT) and mutant receptors at an ␣-level of 0.05. The Dunnett's test compares group means and is used to identify samples whose means are significantly different from the mean of a reference group, in our case the WT sample. 10 -90% rise times in response to maximal GABA concentration and ratios of maximum PB versus maximum GABA current amplitudes for oocytes expressing WT and mutant GABA A Rs were analyzed using unpaired two-tailed Student's t test. All data reported are mean Ϯ S.E. unless noted otherwise.
Structural Modeling-Homology modeling was performed as described previously (20). Briefly, we modeled the GABA A R extracellular domain after the AChBP (3)    27 Subsequently, the two structures were docked, and global energy minimizations were undertaken followed by examining the protein for gross structural distortions. The GABA A R model images were developed using PyMOL (Schrödinger, LLC, New York).

RESULTS
Effects of Glycine Insertions on GABA Actions-Gly1, Gly2, Gly4, or Gly8 were inserted after position Lys-103 in the ␤-subunit (␤K103) and after aligned positions in the ␣-(␣K105) and ␥-(␥K118) (Fig. 1, A and B) subunits to evaluate how increasing the length and flexibility of the linker regions in the ␣-, ␤-, and ␥-subunits would affect GABA A R function. Oocytes were injected with mutant and wild-type ␣-, ␤-, and ␥-subunit cRNAs to form ␣␤␥ receptors and functionally characterized using two-electrode voltage clamp. All of the mutant subunits assembled into functional receptors that responded to GABA. In general, the ␣Gly and ␥Gly insertions had minimal effects on GABA EC 50 values (Ͻ4-fold) as compared with WT (14.1 Ϯ 2.2 M), except for ␣Gly4 that had an 11-fold increase in GABA EC 50 (Fig. 2, Table 1). The ␣Gly and ␥Gly insertions had no effects on the Hill slopes for GABA activation. The maximal GABA-activated currents elicited from receptors containing the ␣Gly or ␥Gly insertions ranged from 3.5 to 11 A and did not significantly differ from WT (9.8 Ϯ 1 A; n ϭ 6).
GABA EC 50 values for ␤Gly2, -4, or -8 could only be estimated because the high concentrations of GABA (Ͼ300 mM) needed to reach maximal current responses changed the extracellular solution osmolarity and could not be used. The maximal GABA current amplitudes for ␤Gly1 receptors were significantly smaller (1.28 Ϯ 0.1 A; n ϭ 14) as compared with WT receptors (9.8 Ϯ 1 A; n ϭ 6). To determine whether ␤Gly1 was affecting GABA efficacy and/or receptor expression, we measured and compared currents induced by a saturating GABA concentration with those induced by a saturating PB concentration in the same oocyte. For WT receptors, saturating concentrations of PB and GABA elicited currents similar in magnitude (I PB max / I GABA max ratio ϭ 1.03 Ϯ 0.07, Fig. 3A). In oocytes expressing ␤Gly1 receptors, the currents elicited by saturating concentrations of GABA were 4-fold smaller than currents induced by saturating concentrations of PB, indicating a reduction in GABA efficacy (Fig. 3A). We also measured apparent 10 -90% current rise times at saturating GABA concentrations for ␤Gly1 and WT receptors. Although current onset is limited by the slow solution-exchange times when recording from oocytes (300 ms), differences in apparent current rise times between WT and mutant receptors would imply changes in channel gating. GABA apparent rise times for ␤Gly1 receptors were significantly slower than WT receptors (0.35 Ϯ 0.02 s for WT versus 1.5 Ϯ 0.2 s for ␤Gly1, Fig. 3B), suggesting that insertion of a single glycine in the ␤4-␤5 linker of the ␤-subunit reduced channel opening.

Effects of Glycine Insertions on Gabazine (SR-95531) Actions-
We also examined the effect the glycine insertions had on the ability of the competitive antagonist SR-95531 to inhibit GABA-activated currents. ␣Gly and ␥Gly insertions caused less than 3-fold changes in SR-95531 K I as compared with WT (76 Ϯ 11 nM) (Fig. 4, Table 1). Larger increases in SR-95531 K I were observed for the ␤Gly insertions. Insertion of a single glycine resulted in an ϳ10-fold increase in SR-95531 K I (Table 1). Because the GABA EC 50 values were too right-shifted to be precisely determined for ␤Gly2, -4, and -8 receptors (Table 1), we approximated the -fold changes in SR-95531 K I using the Cheng-Prusoff equation (K I ϭ IC 50 /(1 ϩ ([A]/EC 50 )), where A is the concentration of GABA used in the experiment, EC 50 is the GABA concentration that elicits half-maximal response, and IC 50 is the concentration of SR-95531 that inhibits 50% of the GABA-induced current. The equation predicts that at [A] Ͻ Ͻ EC 50 , K I approaches IC 50 , and at [A] ϭ EC 50 , K I ϭ (IC 50 )/2. The SR-95531 inhibition experiments for ␤Gly2, -4, and -8 used 30 mM GABA, which from curve-fitting estimates is below the GABA EC 50 (Table 1). Thus, for ␤Gly2 receptors, we estimate that the SR-95531 K I is between 3750 nM (using 30 mM as an upper limit for GABA EC 50 ) and 7500 nM (assuming that 30 mM is Ͻ Ͻ EC 50 ), an ϳ50 -100-fold increase in K I as compared with WT. The estimated -fold increases in SR-95531 K I for ␤Gly4 and -8 receptors were even larger (100 -400-fold).  Table 3. Right panel, representative GABA (EC 4 -8 ) current responses from oocytes expressing ␣Gly2␤␥, ␣␤Gly1␥ and ␣␤␥Gly2 receptors elicited in the absence and presence of increasing concentrations of FZ (M). Note that FZ potentiation of GABA responses was not measured for ␤Gly2-, -4-, and -8-containing receptors (see "Results").  FEBRUARY 24, 2012 • VOLUME 287 • NUMBER 9
Effects of Glycine Insertions on PB Actions-PB is an allosteric modulator of the GABA A R that binds at a site distinct from GABA (25). At high concentrations, PB can directly open the channel. The single channel conductances of GABA A Rs activated by PB and GABA are similar (26,27), suggesting that the receptor open-state channel structures induced by their binding are similar (28). To test whether disrupting the linker regions in the ␣-, ␤-, and ␥-subunits affected PB activation, we measured PB concentration responses from wild-type and mutant receptors. Glycine insertions in both the ␤-subunits and the ␥-subunits significantly increased PB EC 50 values by ϳ10-fold as compared with WT receptors (231 Ϯ 10 M) (Fig.  5, Table 2). The ␣Gly insertions altered PB EC 50 Յ 3-fold (Fig. 5, Table 2). The mean maximal currents elicited by PB for all the mutant receptors ranged from 3.3 to 12.4 A and did not significantly differ from WT (9.1 Ϯ 1.3 A; n ϭ 12), indicating that none of the glycine insertions in any of the subunits altered GABA A R surface expression.
Effects of Glycine Insertions on FZ Actions-BZDs modulate GABA responses by binding at a site formed at the interface between the extracellular N-terminal regions of the ␣and ␥-subunits (29 -31) (Fig. 1, A and B). At subsaturating concentrations of GABA, positive BZD modulators increase GABAinduced current. To examine whether the linker regions were involved in mediating BZD-positive allosteric modulation, we tested the effects the Gly insertions had on the ability of the BZD-positive modulator FZ to potentiate GABA currents. For WT GABA A Rs, FZ maximally potentiated EC 4 -8 GABA-induced current with a potentiation value of 2.49 Ϯ 0.18 and an EC 50 of 468 Ϯ 94 nM (Fig. 6, Table 3). All of the ␣Gly and the ␤Gly1 insertions reduced FZ potentiation of EC 4 -8 GABA-mediated current responses (50 -60%) without changing the FZ EC 50 (Fig. 6, Table 3), suggesting that these regions are important for mediating BZD efficacy. Because the magnitude of the maximal BZD potentiation of I GABA measured is highly dependent upon the effective GABA concentration being applied (32), FZ potentiation of GABA currents for ␤Gly2-, -4-, and -8-containing receptors were not measured because their GABA EC 50 values could not be precisely determined. The ␥Gly insertions not only reduced FZ potentiation of GABA currents (40 -50% for ␥Gly4 and ␥Gly8) but also increased FZ EC 50 values ϳ4-fold (Fig. 6, Table 3).

DISCUSSION
Here, by altering the length and flexibility of the ␤-subunit ␤4-␤5 linker, we demonstrate that this linker plays a critical role in mediating agonist-induced GABA A R functional responses. Moreover, agonists and allosteric modulators react uniquely toward insertions in the linker in the ␣-, ␤-, and ␥-subunits of the GABA A R, suggesting that the allosteric trajectories underlying their actions are distinct. Overall, our data identify the ␤4-␤5 linker as a key structural element in the GABA A R that shapes the energetic landscape associated with channel activation and drug modulation.
␤-Subunit Linker Region Couples GABA Binding and Gating-Glycine insertions in the ␤-subunit linker region, which physically connects the GABA binding site (loop A) to the extracellular domain inner ␤-sheet, increased GABA EC 50 more than 10,000-fold, whereas insertions in the ␣and ␥-subunits had little to no effect on GABA EC 50 (Fig. 7A, Table 1). These data are consistent with structural and molecular dynamic studies (1,3,4,(33)(34)(35) in the pLGIC family that indicate asymmetric subunit motions in the extracellular domain help power agonist-mediated gating. Recently, photochemical cleavage of the ␣-subunit GABA A R linker was shown to disrupt GABA activation (36). In this study, the cleavage site was located near the ␤/␣ interface close to loop E of the GABA binding site (at ␣M113, see Fig. 1). We speculate that proteolysis at this site likely altered the structure of the GABA binding site, which resulted in the loss of GABA-mediated functional responses observed. Here, the ␣-subunit glycine insertions are located near the ␣/␥ and ␣/␤ interfaces (at non-GABA binding site interfaces). Surprisingly, inserting even up to 8 glycine residues in the ␣and ␥-subunits was tolerated, indicating that the length and/or flexibility of the linker in these subunits is not critical for GABA-mediated current responses. In the crystal structures of the nAChR ␣-subunit extracellular domain, AChBP, GLIC, ELIC, and the invertebrate glutamate-activated chloride channel (2, 3, 5-8), the linker region is relatively unstructured and located on the surface of each subunit facing the extracellular channel vestibule, which likely allows the glycine insertions to be accommodated without large perturbations in the overall folding and structure of the ␤-sandwich cores of the subunits. Unstructured loop regions may impart flexibility that is essential for protein function. If the linker in the ␤-subunit is involved in propagating GABA-triggered ligand binding site movements to the channel, one would predict that mutations in this region would affect ligand binding and channel gating. To evaluate whether the glycine insertions in the ␤-subunit altered the GABA binding site structure, we examined the ability of the competitive antagonist SR-95531 to inhibit GABA-gated current. ␤Gly insertions significantly increased SR-95531 IC 50 (Fig. 7B, Table 1), suggesting that the orthosteric binding pocket was altered and that the shifts in GABA EC 50 observed with the glycine insertions are in part due to a change in GABA microscopic binding affinity. The proximity of the ␤Gly insertions to loop A of the GABA binding site (37) and the Ͼ10,000-fold changes in GABA EC 50 measured also suggest that, at least in part, the insertions alter GABA binding (25). In support of this idea, in the related nAChR, deletion of residues in the ␤4-␤5 linker in the ␣-subunit altered microscopic acetylcholine binding as well as inhibited channel gating (15).
The ␤Gly insertions also affected channel gating. When the currents elicited from maximum PB versus maximum GABA were compared in the same oocyte, the GABA currents elicited from ␤Gly1-containing receptors were ϳ4 -5-fold smaller than the PB currents (Fig. 3A), indicating that ␤Gly1 transformed GABA into a partial agonist and decreased GABA efficacy. Moreover, ␤Gly1 increased the 10 -90% apparent current rise times for maximum GABA (Fig. 3B), suggesting that ␤Gly1 alters channel gating by decreasing the channel opening rate. In the GABA A R, nAChR, and serotonin type-3 receptor, mutating residues in the ␤4-␤5 linker near loop A increase unliganded channel opening (15,16,37,38), indicating that this region influences channel gating. Moreover, agonist-mediated movements in the nAChR ␣-subunit ␤4-␤5 linker have been observed (39). Thus, although a detailed understanding of the effects of the glycine insertions on GABA binding and channel gating will require higher resolution kinetic studies, taken together, the data provide strong evidence that the ␤-subunit ␤4-␤5 linker is an important structural element involved in transducing GABA binding to channel gating.
A surprising observation seen with the ␤Gly2-8 insertions was that the competitive antagonist SR-95531 was converted into a weak partial agonist and elicited ionic currents in the absence of GABA (Fig. 4). Because glycine does not have a ␤-carbon, it can adopt a large variety of conformations and impart a regional flexibility that may lead to a widening of the GABA binding pocket. This widening may allow the relatively bulkier SR-95531 to promote local movements in the GABA  FEBRUARY 24, 2012 • VOLUME 287 • NUMBER 9 binding pocket that trigger and stabilize an open channel protein conformation. Moreover, the binding site expansion would hinder GABA, a smaller molecule, from optimally establishing the necessary structural contacts and triggering the local movements required (i.e. binding site contraction) for receptor activation and would thus convert it into a partial agonist. Agonist affinity for the GABA binding site is linearly correlated to ligand length (40) consistent with the idea that altering the binding site size would alter GABA-site ligand actions.

Structural Determinants of GABA A R Activation
In the scenario that GABA A R activation involves a 10°clockwise rotation of the inner ␤-sheet, as is suggested for nAChR (1), inserting glycine(s) in the ␤-subunit ␤4-␤5 linker could act much like a shock absorber and dampen the strength of the signal being transmitted to the inner ␤-strands. Alternatively, the glycine insertions may affect local positioning of critical residue(s) important for GABA A R activation. In GABA A R homology models, a potential salt bridge between residues ␤K102 (loop A), which is nearby the insertion point of the glycines, and ␤D56 in the ␤-subunit (loop 2) is observed. Mutation of either of these residues alters GABA-mediated current responses, suggesting that this salt bridge is important for linking the binding site to loop 2 in the gating interface (41). Consistent with this idea, recent work from Cadugan and Auerbach (16) has shown that ␣A96 (loop A) and ␣I49 (loop 2) in the mouse nAChR, which are aligned to ␤K102 and ␤D56 of the GABA A R, respectively, are energetically coupled. GABA A R Activation by PB Has Different Structural Requirements than GABA-Previous studies have shown that the structural elements underlying GABA A R activation by GABA and PB, which bind at different sites (25), are different (36,42,43). The differential effects of the glycine insertions on GABA and PB activation provide additional support for this idea. Glycine insertions in both the ␤-subunit and the ␥-subunit linkers increased PB EC 50 (Fig. 7C, Table 2), whereas insertions in the ␣-subunit were well tolerated. Consistent with the ␣Gly insertions being well tolerated, photochemical cleavage of the ␤4-␤5 linker in the GABA A R ␣-subunit does not affect PB EC 50 or its maximum current amplitudes (36). Regardless of the number of glycine insertions in either the ␤-subunits or the ␥-subunits, PB EC 50 increased about 10-fold ( Table 2), suggesting that GABA A R activation by PB does not depend on length and/or flexibility of the ␤or ␥-subunit linkers. The differences in magnitude of the effects of the glycine insertions on GABA and PB activation as well as differences in the subunit specificity of the effects indicate that the allosteric structural pathways linking GABA binding and pentobarbital binding to GABA A R activation are different. We also examined the effects of introducing 2 glycine residues (Gly2) simultaneously in both the ␤-subunit and the ␥-subunit linkers (␣␤Gly2␥Gly2) on PB EC 50 (data not shown). We hypothesized that if the ␤and ␥-subunit linkers contributed independently to the PB activation cascade, then the dual subunit insertions would have additive effects on PB EC 50 and increase PB EC 50 by ϳ100-fold. The changes in PB EC 50 for ␣␤Gly2␥Gly2 were not additive and increased only about 17-fold (EC 50 ϭ 3994 Ϯ 269 M, n ϭ 5 versus 231 Ϯ 10 M, n ϭ 3 for WT), suggesting that the mutations in these linkers may affect a shared downstream element in the transduction pathway mediating PB activation of the receptor. Pre-vious studies (44 -46) have also identified the ␤-subunit as a key participant in PB activation. GABA A R Modulation by Flurazepam Has Different Structural Requirements than GABA-Although spatially distinct, the GABA and BZD binding sites allosterically communicate with each other (47). Glycine insertions in each of the GABA A R subunits decreased BZD modulation of GABA responses (Fig.  7, D and E), suggesting that allosteric coupling of BZD binding to modulation of GABA responses involves multiple subunits and transduction pathways. Insertions in the ␣-subunit linker decreased FZ maximal potentiation of I GABA without affecting FZ apparent affinity (EC 50 ) (Fig. 7, D and E, Table 3), suggesting that the ␣-subunit linker is critical for mediating BZD agonist efficacy. The ␣-subunit ␤4-␤5 linker spans the ␣-subunit and physically connects the BZD and GABA binding sites (Fig. 1, A  and B). We speculate that BZD-initiated movements in the BZD binding site are propagated directly through the ␣-subunit linker to the GABA binding site and that the ␣Gly insertions hinder this propagation. BZD modulation of GABA A R was also altered by ␥Gly insertions (Fig. 7, D and E), which increased BZD EC 50 as well as decreased BZD efficacy (Fig. 7, D and E, Table 3). The ␥Gly insertions are positioned near the ␥/␤ interface (a nonbinding site interface). Interestingly, mutations of Arg-43 at this interface alter BZD unbinding (48) consistent with structural perturbations at this interface having long range effects on BZD actions. ␤Gly1 also decreased BZD efficacy ( Table 3, Fig. 7D). Some studies suggest that BZD agonists increase unliganded GABA A R channel opening (49 -51). Because ␤Gly1 appears to stabilize a closed state of the GABA A R by slowing channel opening (Fig. 3B), one might expect that ␤Gly1 would decrease BZD efficacy. Overall, the glycine insertions in the ␤4-␤5 linkers of each of the subunits altered BZD actions, suggesting that BZD agonist-induced structural changes are transmitted via multiple pathways.
Linear free energy relationship analysis in the nAChR indicates that the ␤4-␤5 linker moves early in the activation process (15) and likely transfers the energy imparted by the agonist to loop 2 of the coupling interface that then leads to receptor activation. Thus, we envision that increasing the length and flexibility of this linker in the GABA A R ␤-subunit decreases GABA activation presumably by accruing an energetic barrier, indicating that the structural dynamics of this linker are optimized for efficient signal propagation within the native protein.