Φ value analysis underscores strong functional and structural compactness of the GABA_A receptor

Significance

Understanding how γ-aminobutyric acid type A receptors (GABA_ARs) couple ligand binding to ion channel opening is critical for elucidating the mechanisms underlying inhibitory neurotransmission. Using single-channel recordings with Φ value analysis across multiple point mutations, this study reveals that GABA_AR gating is governed by highly synchronized domain movements. Unlike related nicotinic acetylcholine receptors, GABA_ARs display a narrow range of Φ values, indicating compact allosteric coupling and a globally coordinated gating process. These findings highlight the pivotal role of the extracellular–transmembrane interface and identify structural hotspots essential for function, dysfunction, and potential therapeutic targeting. Our work provides a dynamic, structure–function framework for GABA_AR activation, offering insights for rational drug design.

Abstract

γ-aminobutyric acid type A receptor (GABA_AR) is a pentameric ligand-gated ion channel that plays a crucial role in inhibition in the adult brain. Structural and electrophysiological studies have provided numerous insights into the receptor’s functioning but the complete molecular mechanism of GABA_AR action remains elusive. Herein, we used high-resolution single-channel recording to analyze point-mutated α₁β₂γ₂ receptors and applied so called Φ value (REFER, rate-equilibrium free energy relationship) analysis which allows to infer the order of domains engagement during activation, offering a complementary “dynamic” insight into the receptor’s function. As anticipated, point mutations at the orthosteric binding sites reduced GABA binding affinity, with the magnitude of this effect diminishing progressively with increasing distance from the binding site. On the contrary, mutations located all over the macromolecule’s structure, e.g., in “peripheral top position,” extracellular/transmembrane interface, and channel pore, affected the receptor with a clear tendency to influence entire gating including opening/closing, preactivation, and desensitization with no clear correlation with the distance to the channel gate. Interestingly, the calculated Φ values for the GABA_AR showed a relatively narrow range (0.42 to 0.81), suggesting that the conformational transitions are highly synchronized and coordinated by a global network of interactions. This prediction is also in agreement with observation that, typically, single GABA_AR residue mutation alters the kinetics of not just one but many (often all) conformational transitions. We thus propose a dictum reflecting modus operandi of the GABA_AR: Binding is local and gating is global. In conclusion, our analysis indicates that GABA_AR shows particularly strong functional compactness manifested as global gating mechanisms and high allostery.

The γ-aminobutyric acid type A receptor (GABA_AR) is a pentameric ligand-gated ion channel (pLGIC) that plays a key role in inhibitory neurotransmission in the adult brain. Its pentameric structure is built of subunits that are arranged pseudosymmetrically around a central pore (Fig. 1A). There are 19 types of GABA_AR subunits cloned thus far (1, 2): six α (α_1-6), three β (β_1-3), three γ (γ_1-3), three ρ (ρ_1-3), and δ, ε, π, θ subunits. The most common GABA_A receptors consist of two α, two β and γ or δ subunits [with α₁β₂γ₂ being most frequent (3, 4)]. Each subunit consists of three distinct domains (extracellular—ECD, transmembrane—TMD, and intracellular—ICD) with multiple loops, modulatory or ligand-binding sites, and interfaces, which, together with heterogeneity of subunit types, contribute to a remarkable structural and functional diversity of the GABA_ARs. They are crucial for maintaining neuronal inhibition, and its dysfunction is implicated in pathological conditions such as epilepsy, depression, schizophrenia, autism, and anxiety (for review see: refs. 5–9). Comprehensive understanding of the structure–function relationship of GABA_ARs is vital for uncovering their mechanisms of action and guiding the development of targeted therapeutics to modulate their activity.

Fig. 1.

GABA_AR structure and sequence. (A, Top) Extracellular top–down view of the GABA_AR, showing subunit arrangement (two α₁ and β₂ subunits, one γ₂ subunit), the location of the two GABA-binding sites (GABA BS) at the β₂/α₁ subunit interfaces and the central ion pore. (Bottom) Side view of the receptor depicting ECD and transmembrane TMD domains. Key structural features including the GABA BS, domain interface, gate, and ion pore are labeled. Subunits are colored in a same manner as in the top visualization. (B) Side view of the GABA_AR with the investigated residues depicted in the Van der Waals spherical representation. The residues are located in the principal (β₂, red) and complementary (α₁, blue) subunits in the ECD N-terminal region, GABA BS, domain interface, and TMD helices. (C) Alignment of the GABA_AR α₁, β₂, γ₂ and selected pLGICs isoforms. Investigated residues highlighted in red (β₂ subunit) and blue (α₁ subunit). β-strands (1-10) of the ECD marked in yellow boxes, α-helices (M1-M4) of the TMD in blue ones. Key loops of the GABA BS (A–F) and domain interface (loop 2, β10-M1 linker, and M2-M3 loop) marked in gray boxes. Within the M2 ion pore lining helix selected residues marked using x’ notation (9’–hydrophobic leucine ring forming the channel gate). These residues are also summarized in Table 1.

Advances in structural biology and electrophysiology have provided numerous insights into the receptor’s general architecture, respective binding sites, and the conformational changes underlying key functional transitions (for review see: refs. 2 and 10–17). Particularly important insight into protein functioning at the molecular level can be brought by Φ value analysis (also called REFER, rate-equilibrium free energy relationship) which is a kinetic technique allowing to infer the order of domain movements during protein activity. Thus, the REFER analysis may offer a complementary “dynamic” insight at the molecular level which, together with e.g., approaches determining static protein structures, could enrich description of its functioning. This kind of analysis may be used to characterize multiple processes starting from simple chemical reactions up to protein folding and conformational transitions (18–22). So far, in the investigations regarding the pLGICs family, it was applied mainly to the nicotinic acetylcholine receptor (nAChR, e.g., ref. 23). The authors of this work proposed five subsequent intermediate steps in the nAChR gating: click and hold (of agonist in the binding site), ECD twisting, TMD tilting, ion pore dilating, and wetting. They highlighted that the gating conformational change starts at domain interface and ligand binding sites. At least some features of this mechanism are confirmed in GABA_AR: Structural studies validated the twisting and tilting of respective domains during gating (24, 25) and electrophysiological investigations supported the presence of the intermediate state between binding (corresponding to “click and hold”) and channel opening (ion pore dilution and wetting): the flipped (or preactivated) state (e.g., refs. 26 and 27).

Despite the substantial progress observed in the recent years, the complete mechanism of action of the GABA_AR is not yet resolved. The aim of our study was to identify and map the key structural elements within the GABA_AR that facilitate the transition from the agonist-bound state to the fully open ion pore state. We sought to place these transitions within a chronological framework to provide a comprehensive understanding of the activation process. Furthermore, we explored potential differences in the localization of receptor residues involved in ligand binding and gating, assessing whether these processes are spatially and functionally interconnected. To achieve these goals, we combined the high temporal resolution single channel recordings of currents mediated by wild type (WT) and multiple single point mutant GABA_ARs in the α₁β₂γ₂ subunit assembly with kinetic modeling and Φ value analysis. As expected, mutations at the orthosteric binding sites weakened GABA binding but this effect decreased with the distance from the binding site. On the contrary, mutations located all over the receptor’s structure, e.g., in “peripheral top position,” ECD/TMD interface, and channel pore had tendency to influence entire gating including opening/closing, preactivation, and desensitization. Interestingly, the Φ values for the GABA_AR showed a relatively narrow range (0.42 to 0.81), suggesting that the conformational transitions reflect movements that are highly synchronized and coordinated by a global network of interactions. In conclusion, our analysis indicates that GABA_AR shows particularly strong functional compactness manifested as global gating mechanisms and high allostery.

Results

Prominent Effects of Mutations on the Receptor Function.

To clarify the role of specific structural regions of the GABA_AR in its function, several variants of single-residue mutation were selected for detailed electrophysiological investigations. The properties of these mutated receptors were compared to those of the WT receptor, enabling an assessment of the roles of individual residues and their surrounding environments. We focused on the α₁β₂γ₂ GABA_AR receptors which, as already mentioned, represent the most prevalent isoform in the brain (3, 4). A general depiction of this receptor subtype is presented in Fig. 1A. The structural localization of the residues selected for mutations within the protein is illustrated in Fig. 1B, mutations are also summarized in SI Appendix, Table S1. For sequence alignments with other pLGICs, refer to Fig. 1C. In the N-terminal region, mutations were introduced at α₁F14[A, C, R] and β₂F31[A, C, D]. Within ECD, modifications targeted the agonist-binding site, specifically on the principal side at loops A and C, including β₂E153[A, C, K], β₂E155[C, L, S, Q], and β₂F200[C, I, Y]. On the complementary side, mutations were introduced at loops G and D, at α₁F45[C, G, K, L] and α₁F64[A, C, G, L]. At the interface between the ECD and TMD, within loop 2, the following residues were mutated: β₂V53[A, E, H, K], β₂E52[K, Q], α₁H55[A, C, E, K], and α₁D54[K, N]. The β10-M1 linker was also examined, with mutations at β₂R216[N, D] and α₁R220[D, N]. Mutations were also introduced in the M2-M3 loop at β₂P273 [A, E, H, K] and α₁P277[A, E, H, K]. Within the TMD’s ion pore, the following residues were modified: β₂T256[A, D, K], β₂L259[A, D, S, T], β₂H267[A, K], β₂E270[C, F, K, Q], α₁T260[A, D, K], and α₁L263[A, D, S, T]. Finally, at the bottom of the TMD’s transmembrane helices, mutations were introduced at β₂L296[G, K, V], α₁G258[L, N, V], and α₁L300[G, K, V]. This systematic approach allowed for a comprehensive understanding of how individual residues contribute to GABA_AR function, particularly at key structural interfaces. The effects of these mutations can be classified into distinct categories, including alterations in the receptor’s dose–response relationship resulting primarily from modifications of agonist binding, modifications in response kinetics at saturating [GABA] (gating), a marked increase in spontaneous activity, and complete loss of function.

Alterations in Agonist Binding and Receptor Gating.

Mutations in the GABA_AR binding site (Fig. 2A) significantly increased EC50 values and required much higher GABA concentrations for current saturation, indicating a direct disruption of ligand binding, particularly evident for β₂E155 and several other binding site residues (SI Appendix, Table S2). In contrast, mutations outside the agonist binding site, including those in the N-terminal region, domain interfaces, TMD helices, and ion pore, had minimal impact on dose–response relationships and resulted in EC50 values similar to the WT (for more detailed explanation see SI Appendix).

Fig. 2.

Mutation effects on the ligand binding and current kinetics. (A) Zoomed-in cartoon and stick representation of the GABA BS (marked with an orange box on the schematic view of GABA_AR on the Left), with the GABA molecule (Van der Waals spherical representation) bound between key aromatic and charged residues. On the primary β₂ subunit side, residues β₂E153, β₂E155 (loop B), β₂F200, and β₂Y205 (loop C) from loops B and C are positioned to form electrostatic and hydrophobic interactions with GABA. On the complementary α₁ subunit side, residues α₁F45 and α₁F64 from loops G and D contribute to hydrophobic interactions that stabilize GABA binding. These intersubunit contacts underscore the cooperative architecture of the ligand-binding site and its sensitivity to mutations affecting neurotransmitter recognition and receptor activation. Mutation effects of various GABA_AR residues on agonist binding are presented in SI Appendix, Table S2. (B) Zoomed-in cartoon and stick representation of the TMD α-helices (marked with an orange box on the schematic view of GABA_AR on the Left), showing the selected pore lining residues. 9’ residues β₂L296, α₁L263, and γ₂L274 are symmetrically forming the channel hydrophobic gate. Residues β₂E270 and β₂H267 are also lining the ion pore, but are located at the very top part of the M2 helix. (C) Exemplary single-channel current traces (recorded at saturating GABA concentration) mediated by WT α₁β₂γ₂ GABA_AR and respective β₂E270 residue mutants. (D) Histograms showing frequency densities of apparent open (blue) and shut (red) periods in the WT α₁β₂γ₂ GABA_AR and respective β₂E270 residue mutants activity. Dashed lines depict fitted exponential functions used to estimate mean times of the respective events (presented in SI Appendix, Table S3). (E) Kinetic model used to describe the activity of GABA_AR at saturating agonist concentration. The respective states are A₂R (agonists bound at both binding sites), A₂F (receptor in flipped/preactive state), A₂O and A₂O (two open states), and A₂D and A₂D (two desensitized states). The model includes two open and four shut states, corresponding to the double-exponential distribution of open times and the quadruple-exponential distribution of shut times. Values of the model rates for WT α₁β₂γ₂ GABA_AR and β₂E270 are presented in SI Appendix, Table S4.

In addition to above-described alterations in the dose–response relationship, most considered mutations had significant effects on the kinetics of GABA_AR receptor responses under both rapid GABA application conditions (macroscopic measurements in whole-cell and excised patch modes) and steady-state conditions with constant saturating agonist concentration (single-channel recordings). Trying to classify the effects of considered mutations, two important observations were made. As already mentioned, GABA sensitivity (EC50) was affected by mutations at the binding sites or close to it. In contrast, nearly all investigated mutations located all over the receptor structure (including orthosteric binding site) resulted in significant alterations in the receptor gating. An example of this kind of behavior is the residue β₂E270, located in the upper part (20’ position) of the ion pore (Fig. 2B). Mutations to cysteine, lysine, glutamine, and phenylalanine caused significant alterations in the single-channel current kinetics, as illustrated by the representative traces in Fig. 2C. The most pronounced changes were observed in the β₂E270C mutation, where almost all characteristics of the shut and open dwell times dramatically differed from those in the WT. Smaller effects were seen in β₂E27Q and β₂E270F, with the smallest impact observed in β₂E270K (Fig. 2D and SI Appendix, Table S3). A common pattern among β₂E270[C, Q, F] was the alteration in the proportions of the 1st, 2nd, and 4th component of the shut states dwell times. Kinetic modeling using the complete receptor gating scheme (Fig. 2E) revealed that β₂E270C mutation affected all transitions (preactivation, opening/closing, desensitization), whereas in the case of β₂E270[Q, F] the effect is limited to transition to the open state (SI Appendix, Table S4). The unexpected similarity of the charge-reversal mutant β₂E270K and the WT may be explained by interaction of this residue with β₂H267 (17’ residue, one helix turn below β₂E270, Fig. 2B) which may alter its protonation due to interaction with β₂E270 mutation compensating thereby the mutation effect. Other examples of single residue mutations significantly altering the receptor gating may be found in our previous reports, summarized in SI Appendix, Table S1.

Interestingly, only a few mutations resulted in a complete or near-complete loss of the receptor function. This was observed for residues α₁R220[D, N] and β₂R216[N, D] in the β10-M1 linker, as well as β₂E52 [K, Q] and α₁D54[K, N] in loop 2 (Fig. 3A). All these residues play critical roles in the domain interface region—while the β10-M1 linker serves as the only covalent connection between the ECD and TMD, loop 2 establishes multiple noncovalent interactions with the TMD. Notably, β₂E52 and α₁D54 form strong electrostatic interactions with β₂R216 and α₁R220, respectively (Fig. 3A), highlighting the functional importance of this specific region of domain interface. To determine whether β10-M1 linker residues affect receptor expression rather than gating, immunofluorescence staining experiments were conducted, confirming that the mutated receptors were successfully expressed on the cell membrane (Fig. 3B).

Fig. 3.

Specific effects of the mutations at the domain interface and ion pore. (A) Zoomed-in cartoon and stick representation of the domain interface (marked with an orange box on the schematic view of GABA_AR on the Left). Symmetrically, in both β₂ and α₁ subunits loop 2 (connecting β1 and β2 strands in ECD, residues β₂V53 and α₁H55) forms noncovalent interaction with M2-M3 loop (TMD, residues β₂ P273 and α₁P277) whereas β10-M1 linker directly connects ECD and TMD forming the domain interface. In addition, linker residues (β₂R216 and α₁R220) form electrostatic interaction with loop 2 (β₂E52 and α₁D54). (B) Confocal microscopy images showing the expression of the WT α₁β₂γ₂ GABA_AR and respective β10-M1 linker residue mutants on the surface of HEK293 cells (blue – cell nucleus, green – membrane, cyan or magenta – β₂ and α₁ GABA_AR subunits respectively). (C) Zoomed-in cartoon and stick representation of the ion pore (marked with an orange box on the schematic view of GABA_AR on the Left). (D) Exemplary current traces (recorded using agonist-free solution in single-channel configuration) mediated by WT α₁β₂γ₂ GABA_AR and respective ion pore residue mutants.

In contrast to the interface region, where some mutations led to a loss of function, several mutations in residues within the ion pore of TMD resulted in a significant level of spontaneous activity. This effect was particularly evident in mutations affecting residues that form the channel gate (9’ position): β₂L259[A, S, T] and α₁L263[A, S, T], as well as one helical turn below (6’ position): β₂T256[A, D] and α₁T260A. Notably, all mutations that increased spontaneous ion pore opening in the absence of the agonist involved substitutions to either smaller amino acids, such as alanine, which disrupt the steric barrier maintaining the closed state, or more hydrophilic residues, such as serine or threonine, which reduce the effectiveness of the hydrophobic gating mechanism. For structural visualization, see Fig. 3C, and exemplary current traces are presented in Fig. 3D.

Φ Value Analysis.

The key findings on the role of specific residues in GABA_AR function are as follows: Binding is predominantly determined by local binding site residues, with minor or any influence from other regions of the protein. In contrast, nearly all investigated residues across the receptor impacted either all or at least one step in the receptor gating mechanism, including transitions to the flipped, open, or desensitized states. Additionally, certain residues are essential for function—mutations in the interdomain interface can lead to loss of function, while modifications in the ion pore can induce spontaneous activity.

To further elucidate the role of specific GABA_AR structural regions in the gating process—defined as the transition from the fully bound state to the open state—we performed Φ value analysis also called REFER. The Φ value is estimated for each residue separately and ranges from 1 to 0. Values close to 1 indicate that given residues move at the beginning of the gating transitions, close to 0 – at the end of it. Unlike the previously presented kinetic modeling, as described in detail in Materials and Methods, this analysis employed a simplified two-state model (Fig. 4A) consisting of only the shut, doubly bound state (C) and the open, doubly bound state (O). In relation to the model shown in Fig. 2E, the C state encompasses both A₂R and A₂F states, while the O state includes both A₂O and A₂O’ states. Desensitized states were excluded from this model and long shut events corresponding to desensitization were omitted from analysis by setting appropriate critical time using the Colquhoun and Sakmann equation (Eq. 1) (28). Further methodological details regarding Φ value analysis are provided in the Materials and Methods section.

Fig. 4.

Φ value analysis. (A) Simplified two state kinetic model and schematic view of the plot used to assess the Φ value of respective GABA_AR residues. (B) Log–log plots showing the assessment of the Φ values for the respective residues. Data points correspond to the forward rate and equilibrium constant estimated for each receptor type (mutation). The Φ value (presented on the plots and in Table 1) is the slope of the fitted linear function. Structural localization of the residues is presented in Fig. 1B. (C) Proposed structural compartmentalization of the GABA_AR activation timeline pathway. Upon ligand binding receptor undergoes conformation transitions starting at the ECD N-terminal propagating through binding site and domain interface area up to the ion pore. The binding process is determined mostly by local binding site residues, whereas gating process is shaped by the whole macromolecule. In addition, regions of specific mutation effects like loss and gain of function are marked.

The results, presented as log–log plots of the gating equilibrium constant and opening forward rate for respective residues, are shown in Fig. 4B. Certain residues or mutants were excluded from this analysis due to either a high level of spontaneous activity, minimal differences relative to the WT or extremely high EC50 values (making it impossible to saturate the response), making accurate Φ value assessment unfeasible. In the case of the two residues: β₂E153 and α₁H55 the error of the Φ values estimation was relatively high, because of the similarity of the mutants to the WT. The calculated values of the slope of presented plots, that is Φ values, are presented on the plots in Fig. 4B and in the Table 1. Surprisingly, all obtained values range from 0.42 to 0.81, suggesting that the sequential movement of the respective receptor regions occurs in a tightly coordinated manner throughout the gating process. Notably, this finding qualitatively differs from results on nAChR obtained by Auerbach’s group where the smallest Φ value was 0.26 and the largest were close to 0.95 (29), whereas in the newer report (23), smallest value of 0.06 was presented. Possible interpretations for this difference are presented in Discussion.

Table 1.

Summary of the Φ-values assessed for the investigated residues

residue	Φ	residue	Φ	location
β2F31	0.81 ± 0.05	α1F14	0.72 ± 0.07	ECD N-terminal
β2F200	0.67 ± 0.02	α1F64	0.63 ± 0.05	binding site
β2E153	0.63 ± 0.18	α1F45	0.65 ± 0.04
β2V53	0.65 ± 0.03	α1H55	0.68 ± 0.48	domain interface
β2P273	0.56 ± 0.05	α1P277	0.64 ± 0.04
β2L296	0.50 ± 0.07	α1L300	0.56 ± 0.10	TMD ion
β2E270	0.42 ± 0.04	α1G258	0.54 ± 0.07	pore & outer helices
β2H267	0.54 ± 0.02

Plots used to determine those values are presented in Fig. 4B. Residues are row-aligned by their analogous function in primary and complementary subunits.

Table 1 presents the Φ values organized by residues with functionally analogous roles in the primary (β₂) and complementary (α₁) subunits. The highest Φ values (> 0.7) were observed for residues located at the top of the receptor, specifically β₂F31 and α₁F31 (Figs. 1 and 4B). Although these residues are not part of the binding site, they are positioned above it in a highly mobile, solvent-exposed extracellular region. The slightly higher Φ value of β₂F31 suggests that movement may initiate at the principal subunit site. Residues forming the agonist binding site (β₂F200, β₂E153, α₁F64, α₁F45) and the domain interface (β₂V53, β₂P273, α₁H55, α₁P277) exhibited intermediate Φ values (0.6 to 0.7, Figs. 1 and 4B). The similarity in Φ values between binding site and domain interface residues indicates that upon ligand binding, a larger portion of the ECD and its interface with the TMD moves in a coordinated, rather than sequential, manner. The absence of differences between the principal and complementary subunits further suggests their synchronized movement along the transition pathway. The lowest Φ values (<0.6) were observed for TMD residues (β₂E270, β₂L296, α₁G258, and α₁L300, Figs. 1 and 4B), which constitute the final structural component in the GABA_AR transition to the open state. It is likely that residues directly forming the gate of the ion pore would exhibit even lower Φ values; however, due to their high level of spontaneous activity, they were not included in this analysis. The Φ value analysis suggests a sequential gating mechanism in GABA_ARs, where early conformational changes occur in ECD top residues, followed by coordinated movement of the binding site and domain interface, and concluding with transitions in the TMD leading to channel opening (Fig. 4B and Table 1). Importantly, residues that play similar functions in β₂ and α₁ subunit (Table 1) show very similar Φ values indicating a concerted movement of these subunits rather than their sequential involvement in gating transitions.

Discussion

In this study, the effects of single-residue mutations on the function of the α₁β₂γ₂ GABA_AR were systematically investigated. Through detailed electrophysiological analyses, the functional consequences of mutations across key structural regions, including, ECD (N-terminal and orthosteric binding site in particular) and TMD (ion pore and outer helices) domains, and their interface were assessed. Critical insights into the roles of specific residues in receptor function were implicated, revealing distinct categories of effects including dose–response shifts, alterations in gating kinetics, spontaneous receptor activity, and complete loss of function (Fig. 4C).

Agonist Binding.

One of the most prominent findings was the significant impact of mutations on the receptor’s dose–response relationship. Mutations introduced in the agonist-binding site, particularly at β₂E155, β₂F200, α₁F64, and α₁F45 (Figs. 1, 2B, and 4C and SI Appendix, Table S2; for more detailed analysis see refs. 26 and 30–33), led to pronounced increases in EC50 values, indicating reduced receptor sensitivity to GABA. These changes were attributed to direct impairment of the ligand binding process. Notably, mutation of the β₂E155 and α₁F64 residues, in addition to increasing the EC50 value, markedly enhanced the spontaneous activity of the receptor (26, 34, 35) indicating a complex mechanism of interference of this mutation with the orthosteric binding site and its surrounding areas. Induction of spontaneous activity upon mutating the orthosteric binding site may be interpreted as mutations of these residues may somehow “mimic” the presence of the agonist leading to channel opening. This would be similar to effects of the mutations of the modulatory binding sites “mimicking” the presence and effect of the modulator—as for example β₂M286W and α₁M236W mutations potentiate agonist-evoked currents similarly to general anesthetics (36, 37). Notably, in contrast to β₂E155, positioned nearby the β₂E153 residue had a much weaker effect on the EC50 value, further confirming that GABA affinity is determined locally at the orthosteric binding site. Indeed, on the contrary to discussed here mutations at the binding sites, mutations at other structural regions exhibited relatively minor effects on EC50 but influenced receptor efficacy through gating modifications. In addition, mutations of the β₂Y205 on the loop C, but not other binding site structures, caused the complete loss of function suggesting that this loop is crucial also in the gating process (33). In line with this finding, mutation of other loop C residue β₂F200, as expected, strongly increased EC50 but in addition to that markedly affected also all gating characteristics (33). Interestingly, all investigated binding site mutations caused the increase of the EC50 value, but none of them led to reduction of this value indicating the optimal structure and functioning of the orthosteric binding site. Whether the agonist binding can be somehow directly upregulated is not sure – for example PAMs like benzodiazepines (BDZs), according to some data can enhance GABA binding via long range allosteric interactions (38–40). However, more recent studies provided extensive body of evidence indicating that BDZs modulate GABA_ARs by affecting rather their gating [desensitization (41), efficacy (42, 43), preactivation (flipping), (11, 44, 45), for review see work by Goldschen-Ohm (11)].

Receptor Gating.

The most intriguing observation made by our group in previous and in the present study is that mutations of residues located at many different loci all over the GABA_AR macromolecule affect the receptor gating. Moreover, most typically, mutations tend to affect not any specific feature but rather all of the gating characteristics including opening/closing, preactivation, and desensitization. As already mentioned, mutation of the β₂F200 residue known to play a crucial role in agonist binding and being “strategically” located for this purpose at loop C, was found to strongly affect all characteristics of the GABA_AR gating (33). Gielen and coworkers (46) proposed that regulation of the desensitization process is localized at the so-called desensitization gate, at the second and third transmembrane segment of GABA_AR. We have investigated point mutation in this localization (β₂G254V, α₁G258V, α₁L300V, and β₂L296V) and reported that these substitutions, besides affecting desensitization, also markedly altered openings/closings and, to a smaller extent, also preactivation (47). Also, studied in the present work, mutations of the residue β₂E270 were found to affect multiple gating transitions, resulting in varying degrees of impact on receptor currents (Fig. 2 B–D and SI Appendix, Tables S3 and S4). Most surprisingly, we found that mutations of β₂F31 and α₁F14 residues, located very “peripherally” at the “top” β/α subunits interface of the receptor (far “above” the orthosteric binding site), significantly affected receptor gating, especially preactivation and opening, and to smaller extent also desensitization (48). Similar pattern of effects on the receptor gating was found by us in other localizations within the GABA_AR macromolecule including also, ECD orthosteric binding site [β₂E155, β₂E153, β₂F200, α₁F64, and α₁F45, (26, 30–33)], ECD/TMD interface region [β₂V53, β₂P273, α₁H55, α₁P277, (49–52)] and the mentioned above TMD region proposed as “desensitization gate” [β₂L296, α₁G258 α₁L300, (47), Figs. 1 and 4C]. The described above observations point to a general rule that whereas binding to the orthosteric binding site is largely a local phenomenon, the receptor gating appears to be a global process presumably relying on a broad network of interactions between respective structural regions. In particular, it seems that there are no specialized “gates” that would have exclusive control on specific gating features such as opening/closing, preactivation, or desensitization. Clearly, there are residues or local loops that play a crucial role in specific gating processes (e.g., “activation gate” at 9’ residue critically affects opening or the 2’ residue in the transmembrane region strongly affecting desensitization), but at the same time specific gating features are also regulated more globally by a variety of molecular interactions encompassing large portions of the macromolecule. This conclusion is strongly supported by our kinetic analysis of mutated receptors but as we discuss below, a particularly strong support for this concept of “global gating of GABA_AR” came from the Φ value analysis.

A plausible prediction from these investigations is that the proposed global gating mechanism in the GABA_AR is expected to render this receptor particularly susceptible to pharmacological modulation. Indeed, it can be predicted that wherever we “touch” the GABA_AR structure, with e.g., a pharmacological modulator, we have good chances to strongly alter the receptor functioning by activating, blocking, or modulating its gating. Notably, in line with this prediction GABA_AR is indeed exceptionally prone to modulation by multiple compounds having its binding sites scattered within receptor’s structure—for example BDZs in the ECD’s subunit interface and in TMD general anesthetics in the upper part of the TMD, neurosteroids in multiple sites in the TMD (for review of the modulatory binding sites see refs. 2, 10, 11, and 53). This variety of the modulatory binding sites and significant effects of many residues’ mutations on receptor function underscores its allosteric character.

Loss of Function and Spontaneous Activity.

Notably, mutations at the β10-M1 linker and loop 2 (β₂R216, α₁R220, β₂E52, α₁D54) resulted in complete or near-complete loss of function, emphasizing the crucial role of these residues in maintaining receptor integrity at the ECD/TMD interface (Figs. 3 A and B and 4C). In addition to the loop C mutations (33) that was the second region of the receptor in which mutations caused such severe effects. These findings suggest that within the binding site loop C, and β10-M1 linker and loop 2 in the domain interface are critical for the transduction of the activating signal (Fig. 4C). Conversely, increased spontaneous receptor activity was observed in certain mutations within the ion pore. Specifically, substitutions at residues (β₂L259, α₁L263, β₂T256, α₁T260) that either reduce steric hindrance or alter the hydrophobic environment of the pore facilitated channel opening in the absence of an agonist (Figs. 3 C and D and 4C). Apart from the ion pore, spontaneous activity was also upregulated by some binding site mutations [β₂E155 and α₁F64, (26, 34, 35)]. Thus, these two regions—binding site and ion pore—are distinguished by being prone to modifications inducing spontaneous activity. However, the mechanisms behind this behavior may substantially differ—in the case of the binding site mutations may mimic the presence of the agonist shifting the open/shut state equilibrium whereas in the pore—these mutations may directly disrupt the mechanism of ion flow (see work by Kumari et al. (54) for detailed investigation on ion passage through pore in other pLGIC). Our investigations presented in this report add one more possibility that vast areas of the receptor macromolecule are cross linked by a “mean field” of interactions resulting in high susceptibility of the receptor to alter its global conformational transitions in response to some local alterations of the receptor structure or to interactions with pharmacological agents acting on specific binding sites.

Wave of Conformational Transitions.

Finally, Φ value analysis provided additional mechanistic insights into the receptor gating process. The results suggest a sequential gating mechanism, beginning with early conformational changes in N-terminal extracellular residues (represented in our study by β₂F31, α₁F14), followed by coordinated movements in the binding site (β₂F200, β₂E153, α₁F64, α₁F45) and domain interface (β₂V53, β₂P273, α₁H55, α₁P277), and concluding with transitions in the TMD that ultimately lead to channel opening (β₂E270, β₂L296, α₁G258, α₁L300) (Figs. 1 and 4 and Table 1). Highest Φ values of the N-terminal residues (β₂F31, α₁F14) suggest that upon ligand binding, the conformational gating transition starts in this region. We speculate, that when the ligand is bound and the receptor equilibrium is shifted toward open state, the conformational transition would start at the most mobile region of the protein—and MD study showed, that indeed the N-terminal region shows biggest fluctuations even at the steady state (roughly 1.5 Å) (55). Also, our more detailed study on this area of the GABA_AR underlined the importance of the intersubunit interactions: Introduction of single mutations separately (β₂F31C or α₁F14C) strongly altered the receptor kinetics, but double mutant (β₂F31Cα₁F14C) largely recovered the WT phenotype which was disrupted again by dithiothreitol, which breaks disulfide bridges (48). Analysis of the available GABA_AR experimental structures also shows that depending on the receptor state the distance between these two residues changes by ~2 Å indicating their importance in the gating transition. In addition, this phenylalanine pair is present only at the β/α interface above the agonist binding site and is not conserved at other subunit interfaces. Interestingly, this area in the GlyR is a binding site for allosteric positive modulators (56). Similar Φ values of the binding site and domain interface residues indicate synchronized motion of the ECD from the orthosteric site up to the top part of the TMD. Notably, all considered here residues in this area are shaping the gating process of the receptor, with prominent roles of e.g., loop C capping the binding site [residue β₂F200, (33, 55)] or loop D transferring the gating signal through rigid β-sheet formation on the complementary subunit side [residue α₁F64, (31)]. In the domain interface an important role is played by M2-M3 loop and loop 2. Relative motion of these two segments was shown to be an important step in the receptor activation especially in the principal subunit side [residues β₂V53, β₂P273, α₁H55, α₁P277, (49–52)]. After the domain interface the gating transition reaches the helices of the TMD which undergo the conformational transition leading to the opening of the ion pore gate. Interestingly, no clear differences were observed between functionally analogous residues in the primary and complementary subunits (Table 1). This suggests that the movements of the respective subunits are well synchronized, with all subunits moving jointly rather than in a sequential, “subunit-by-subunit” manner. This scenario is expected because even though the domains are not covalently linked, they form a broad network of interactions. Additionally, the agonist orthosteric binding site and multiple positive allosteric modulator binding sites are located at subunit interfaces along the receptor axis, further reinforcing this noncovalent interaction network. The observed synchronization of subunits aligns with proposed mechanisms of pLGIC activation, in which the extracellular domain undergoes a twisting movement (anticlockwise when viewed from the top) during gating, leading to tightening of the subunit interfaces (12, 57). An important question for future research is whether the movement of the γ₂ subunit—which binds only allosteric modulators—is synchronized with that of the α₁ and β₂ subunits. Interestingly, in the case of the nAChR β₁ subunit (structurally corresponding to the GABA_AR γ₂ subunit), residues in the ECD had low energetic contribution to the gating process, whereas residues in the transmembrane domain exhibited Φ values comparable to those of corresponding residues in other subunits (58).

The calculated Φ values, are presented in the Table 1 and as already mentioned in Results, they show a relatively narrow range from 0.42 to 0.81 which is markedly smaller than that obtained for nAChR by Auerbach’s group (0.26 or 0.06 to 0.95) (23, 59). An intuitive explanation of this difference is that events related to the conformational transitions in GABA_ARs are considerably more synchronized and tightly coordinated by a global network of interactions than in the case of nAChR. This scenario would also predict that in GABA_AR distinct conformations of the receptor (as postulated e.g., in Fig. 2E) may rely on a common structural background to a larger extent than in nAChR. This prediction agrees with our observation that, typically, mutation of a single residue in GABA_AR markedly alters the kinetics of not just one, but many conformational transitions (often all of them). Similarly, as already mentioned, some commonly investigated PAM as e.g., BDZs with a binding site at the interface between α and γ subunits (bearing similarity to the orthosteric binding site) appear to affect different conformational transitions suggesting that they share at least some part of the molecular mechanisms of activation. Taking thus into account the evidence coming from kinetic analysis of different mutants, pharmacological studies, and now the present Φ value analysis for the GABA_AR, we propose that this receptor is characterized by a particularly high functional compactness manifested by the feature that very local structural alterations give rise to a global impact manifested by altered kinetic features of the receptor.

In an attempt to explain these striking differences between Φ values estimation for GABA_AR and nAChR, we must admit that the kinetic analysis of the latter has an important advantage showing the conductance nearly twice as large as in the case of GABA_AR. In our hands, even when applying a very useful and clever procedure implemented by David Colquhoun of traces idealization by visually inspected time-course event fitting our best achievement for resolution was 40 to 50 μs with the range for the whole analysis 40 to 90 μs. This represented the best recording quality achievable despite additional precautions, including Sylgard coating of pipettes, minimizing the solution volume in the recording dish, and using the shortest possible glass capillaries. This difference in single channel conductance and resulting limits in resolution meant that, in particular, we were constrained to apply stronger filtering (Materials and Methods) than Auerbach’s group [e.g., 12 kHz filtering for rate estimation, (60)]. Thus, it is likely that the resolution in our studies was smaller than in previous studies on nAChR and that the values of the α and β rate constants could be underestimated. However, it is hard to precisely determine how large this underestimation was as a part of effect of stronger filtering was compensated by the time-course event fitting procedure. Notably, larger filtering could lower the value of β (exit from short lived shut event) and thereby the highest Φ values, but we would expect that the impact of insufficient resolution on the lower limit of Φ values would be smaller, but it was actually unexpectedly high: 0.42. It is noteworthy that in the case of the muscle-type nicotinic receptor the opening rate constant is much larger than for GABA_AR underscoring differences in gating mechanisms of the two receptors. The lower resolution than in Auerbach's study on nAChR and the narrow range of Φ values for GABA_AR might result in some experimental imprecision, especially when confronting Φ values within proposed domains where these differences are minimal. In particular, our observation that conformational changes at the top peripheral domain take place before changes inside the transmitter binding site might be to some extent affected by these conditions.

Collectively, these findings shed light on how individual residues contribute to GABA_AR function and highlight the intricate interplay between different structural domains in receptor activation. We believe that this study may also have broad implications for pharmacological modulation strategies targeting GABA_AR in various neurological and psychiatric disorders.

Materials and Methods

Cell Culture and Transfection.

Human Embryonic Kidney (HEK293) cells, obtained from the European Collection of Authenticated Cell Cultures, were used in this study. The cells were cultured in Dulbecco’s Modified Eagle’s Medium supplemented with 10% fetal bovine serum and 1% penicillin/streptomycin, all purchased from Thermo Fisher Scientific, USA. The HEK293 cells were maintained at a constant temperature of 37 °C in a humidified atmosphere containing 5% CO₂ in Nunc flasks.

Prior to transfection, the cells were detached from the flasks and replated onto coverslips coated with 1 μg/mL poly-D-lysine (Sigma, USA). Transient transfections were conducted using FuGENE HD (Promega, USA) with a FuGENE: DNA ratio of 3:1. Plasmids used for transfection contained complementary DNAs encoding subunits of GABA_AR receptors, specifically: gene ID 29705 (α₁ subunit), 25451 (β₂), and 29709 (γ₂). Additionally, plasmids containing a human cluster of differentiation 4 (CD4) or enhanced green fluorescent protein (EGFP) were included as markers for successful transfection. All plasmids utilized cytomegalovirus promoters. For WT GABA_AR receptors, the DNA encoding the α₁, β₂, and γ₂ subunits was used in a ratio of 0.5:0.5:1.5 μg, ensuring an excess of γ₂ subunit to minimize the expression of α₁β₂ receptors. An additional 0.5 μg of CD4 or EGFP plasmid was included to aid in identifying transfected cells. For mutated receptors, the amount of plasmid coding the mutated subunit was increased to 1.5 μg. The specific mutations investigated included α₁β₂F31[A,D]β₂γ₂, α₁F14[A,R]β₂γ₂, α₁β₂E153[A,C,K]γ₂, α₁H55Cβ₂γ₂, α₁β₂P273Hγ₂, α₁β₂E52[K,Q]γ₂, α₁D54[K,N]β₂γ₂, α₁β₂R216[D,N]γ₂, α₁R220[D,N]β₂γ₂, α₁β₂H267[A,K]γ₂, α₁β₂E270[C,F,K,Q]γ₂, α₁β₂T256[A,D,K]γ₂, α₁T260[A,D,K]β₂γ₂, α₁β₂L259[A,D,S,T]γ₂, α₁β₂L296[G,K]γ₂, α₁L263[A,D,S,T]β₂γ₂, α₁L300[G,K]β₂γ₂ and α₁G258[L,N]β₂γ₂ (SI Appendix, Table S1). Twenty-four hours after transfection, the culture medium was replaced with fresh medium.

For experiments involving EGFP cotransfection, the identification of successfully transfected cells was carried out using an inverted microscope (Leica DMi8, Wetzlar, Germany) equipped with a 470 nm fluorescent illuminator (CoolLED, Andover, UK). In cases where CD4 was used as a transfection marker, Dynabeads CD4 magnetic binding beads (Invitrogen, USA) were utilized. The beads were added to the cell culture medium 1 h before electrophysiological recordings.

Immunofluorescence Staining.

To verify the expression and presence in the cell membrane of the receptors with mutations causing the loss of function the immunofluorescence staining was performed. HEK293 cells were transfected with plasmids encoding α₁β₂γ₂ GABA_AR along with EGFP, as described above. Cells were fixed with 2% paraformaldehyde, washed with phosphate-buffered saline and blocked with 5% bovine serum albumin to reduce nonspecific binding. Primary antibodies against the α₁ (rabbit) and β_2/β3 (mouse) subunits were applied, followed by Alexa Fluor 633-conjugated secondary antibodies. Nuclei were stained with DAPI. Confocal images were acquired to assess subunit expression and localization.

Electrophysiological Recordings and Analysis.

Single-channel recordings were performed in the cell-attached configuration at a holding potential of 100 mV. Signals were amplified using an Axopatch 200B amplifier (Molecular Devices, Sunnyvale, CA, USA), filtered at 10 kHz with a built-in low-pass Bessel filter, and digitized at a sampling rate of 100 kHz using a Digidata 1550B acquisition system and Clampex 10.7 software (Molecular Devices, Sunnyvale, CA, USA). Thick-walled borosilicate glass pipettes (outer diameter: 1.5 mm; inner diameter: 0.87 mm; Hilgenberg, Malsfeld, Germany) were fabricated using a P-1000 horizontal puller (Sutter Instruments, Novato, CA, USA) to achieve a resistance range of 8 to 12 MΩ. To minimize noise, the pipettes were coated with Sylgard 184 (Dow Corning, Auburn, MI, USA) and subsequently fire-polished using a microforge.

The external (and intrapipette) solution consisted of (in mM): 102.7 NaCl, 20 Na-gluconate, 2 KCl, 2 CaCl₂, 1.2 MgCl₂, 10 HEPES, 20 TEA-Cl, 14 D-(+)-glucose, and 15 sucrose (Carl Roth, Karlsruhe, Germany). These components were dissolved in deionized water, and the pH was adjusted to 7.4 using NaOH. To further reduce noise, the volume of external solution in the dish was minimized to 0.9 to 1 mL in a 35 mm diameter dish. Only patches with a seal resistance exceeding 10 GΩ were included in the analysis. Unless specified otherwise, all chemicals were obtained from Merck (Darmstadt, Germany).

Clusters of single-channel openings were visually identified and manually selected for further analysis. In addition to the initial 10 kHz analog low-pass Bessel filtering provided by the Axopatch 200B amplifier (Molecular Devices, USA), digital 8-pole Bessel filter was applied using Clampfit software to achieve a signal-to-noise ratio of 15:1. The filtered signal was subsequently down-sampled to maintain a 10:1 ratio between the sampling frequency and the final filter cut-off frequency (f_c). The cut-off frequency (f_c) was determined using the equation, 1/f_c = 1/f_a + 1/f_d where f_a is the analog filter cut-off frequency (10 kHz) and f_d is the digital filter cut-off frequency.

Recordings were idealized using SCAN software (DCProgs, http://www.onemol.org.uk/), kindly provided by David Colquhoun. Idealization of selected traces was achieved by time-course fitting and this procedure was visually inspected for each single-channel event. In our hands, this analysis, although laborious, allowed to achieve the time resolution for closed and open events of 40 to 90 μs, which was markedly better (by at least several tens of microseconds) than in the case of classic threshold detection algorithms. This aspect was particularly important for GABA_ARs which have relatively small conductance (compared to e.g., acetylcholine receptors) yielding single-channel currents with a relatively low amplitude. The resulting idealization files (.scn) were used to generate open and shut time distributions in EKDIST software (DCProgs) and for further modeling using the HJCFIT software (described below).

Published Data.

Several GABA_AR mutants were already investigated by our group in the recent years. Results obtained in these investigations were analyzed again using the Φ value analysis. The methodological details of electrophysiological recordings of currents mediated by these mutants are presented in respective literature summarized in SI Appendix, Table S1. Novel mutants are also presented in SI Appendix, Table S1.

Kinetic Modeling and Φ Value Analysis.

Kinetic modeling was conducted using the HJCFIT maximum likelihood method (DCProgs) on idealized single-channel current traces obtained using the SCAN software. Calculations were done using Python scripts (61). For each trace the analysis started with fitting (with HJCFIT) obtained idealized data (.scn) with a full kinetic model such as in Fig. 2E. To show how the fitting was conducted for the full kinetic model (Fig. 2E) we present here such an analysis for WT receptors and for β₂E270 residue mutants (Fig. 2 B–D and SI Appendix, Table S3). Briefly, distributions of open and closed times were constructed and fitted with sums of exponential functions (Fig. 2D, typically two open and four or less shut components) and rate constants of the model were optimized to reproduce these distributions (SI Appendix, Table S4).

For Φ value analysis, we have used a respective approach [similar to proposed by work of Auerbach’s group (59, 62)] to assess the opening (β) and closing (α) rates of a simplified two-state model (Fig. 4A). All mutations causing qualitative changes in the receptor activity like spontaneous activity, extreme shifts of the dose–response curve, total loss of function or minimal activity were excluded from this analysis. Based on the results obtained for the full model (Fig. 2E, four shut and two open states) and keeping in mind that the longest shut components correspond to the desensitized transitions [similarly to long shut events in nAChR (63)] we have restricted our idealized data to two open states and two shortest shut states, excluding thus closed states corresponding to desensitized states. This selection of shut states was done by determining the critical time (t_crit) between the second and the third shut component in the shut state distribution by using the Colquhoun–Sakmann (CS) equation (28):

$$e^{-t_{\mathit{crit}}/{\tau }_f}=1-e^{-t_{\mathit{crit}}/{\tau }_s},$$ [1]

where τ_f and τ_s represent the mean durations of fast (second component) and slow (third component) shut events, respectively, and t_crit is the critical time threshold used to distinguish between them [equal proportions of misclassifications of short and long events, formula 3 in work by Colquhoun and Sakmann (31)]. The first and the second components of the shut dwell-time distribution (Fig. 2D) correspond to sojourns in the agonist-bound shut or flipped state (Fig. 2E), i.e., the two major shut states occurring in bursts. Since Eq. 1 does not have an analytical solution, it was solved numerically using a Python script. Typically, τ₂ and τ₃ for the shut states were markedly different and the determination of t_crit was straightforward and uncritical. This procedure enabled us to subtract the openings and closings that could be best fitted with two open and two closed components. The next step in the simplification procedure was to fit these data with a two-state model (Fig. 4A) i.e., with only one exponential function for open and shut states distributions. Typically, this simplification yielded reasonable fits as in most cases the differences between the two time constants in open and closed states distributions were not very large. The Φ value for a specific residue was determined by constructing log–log plots for the respective kinetic constants determined for the WT and mutants of a given residue. Namely, the log of the forward rate constant (β) for the wildtype and respective mutants was plotted on the y-axis, while the log of equilibrium constant (β/α) was plotted on the x-axis. The Φ values were obtained as the slope of these plots (Fig. 4B). Workflow of the Φ value analysis is also presented in the SI Appendix, Fig. S1.

Structure and Sequence Analysis.

All structural analyses and visualization were made using the α₁β₂γ₂ GABA_AR receptor structure in complex with GABA (PDB code 6X3Z, (24) in ChimeraX software (64). Sequences of the GABA_AR isoforms and other pLGICS were obtained from UniProt database (65). Sequences were initially aligned using T-Coffe (66) and manually curated with JalView (67). Visualization of the sequence alignment was also prepared in JalView.

Statistical Analysis.

The data were stored and analyzed using MS Excel for initial organization and Python scripts were used for advanced data analysis and visualization. Normality was assessed with the Shapiro–Wilk test, and homogeneity of variance was evaluated using Levene’s test if needed. In the case of the analysis of the mutation effects on channel kinetics, depending on the data distribution, either ANOVA or the Kruskal–Wallis H-test was employed to test significance. Post hoc comparisons within groups were made using Tukey’s test or Dunn’s test with Holm–Sidak correction. CS equation for t_crit was solved using the Newton–Raphson method and Φ values were estimated using linear regression using the ordinary least square method. All statistical analyses were performed using Python scripts with the Pandas, Numpy, and Scipy packages.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

The code used for analysis is available at GitHub Repository (https://github.com/michal2am/bioscripts/) (61).