Multiple roles for the first transmembrane domain of GABA_A receptor subunits in neurosteroid modulation and spontaneous channel activity

Abstract

Neurosteroids exert potent physiological effects by allosterically modulating synaptic and extrasynaptic GABA_A receptors. Some endogenous neurosteroids, such as 3α, 21-dihydroxy-5β-pregnan-20-one (5α, 3α-THDOC), potentiate GABA_A receptor function by interacting with a binding pocket defined by conserved residues in the first and fourth transmembrane (TM) domains of α subunits. Others, such as pregnenolone sulfate (PS), inhibit GABA_A receptor function through as-yet unidentified binding sites. Here we investigate the mechanisms of PS inhibition of mammalian GABA_A receptors, based on studies of PS inhibition of the UNC-49 GABA receptor, a GABA_A-like receptor from Caenorhabditis elegans. In UNC-49, a 19 residue segment of TM1 can be mutated to increase or decrease PS sensitivity over a 20-fold range. Surprisingly, substituting these UNC-49 sequences into mammalian α₁, β₂, and γ₂ subunits did not produce the corresponding effects on PS sensitivity of the resulting chimeric receptors. Therefore, it is unlikely that a conserved PS binding pocket is formed at this site. However we observed several interesting unexpected effects. First, chimeric γ2 subunits caused increased efficacy of 5α, 3α-THDOC potentiation; second, spontaneous gating of α₆β₂δ receptors was blocked by PS, and reduced by chimeric β₂ subunits; and third, direct activation of α₆β₂δ receptors by 5α, 3α-THDOC was reduced by chimeric β₂ subunits. These results reveal novel roles for non-α subunits in neurosteroid modulation and direct activation, and show that the β subunit TM1 domain is important for spontaneous activity of extrasynaptic GABA_A receptors.

Introduction

Neurosteroids are endogenous steroid molecules in the brain that can allosterically potentiate or inhibit GABA_A receptor function [2, 8]. Both synaptic (eg. α₁β₂γ₂) and extrasynaptic (eg. α₆β₂δ) GABA_A receptors can be modulated by neurosteroids; extrasynaptic GABA_A receptors may be of particular importance as targets of neurosteroid modulation [13, 21]. Enhancing neurosteroids, such as 5α, 3α-THDOC, play a role in depression, seizure susceptibility, ethanol action, and many other nervous system functions [15, 16]. Inhibitory neurosteroids, such as PS, are proconvulsive [11, 24], and enhance learning and memory [22], although some of these effects may be due to modulation of other channels such as N-methyl-D-aspartate receptors [26] and voltage-gated calcium channels [5]. Further investigation into the structural basis of neurosteroid potentiation and inhibition of GABA_A receptors will be important to better understand the physiological basis of neurosteroid effects on behavior, and to exploit their therapeutic potential.

Recent studies have identified GABA_A receptor residues important for positive and negative modulation by neurosteroids. A putative binding site for neurosteroid potentiation is formed by residues Q241 in TM1 and N407 in TM4 [10]. A separate binding site for direct activation by higher concentrations of neurosteroids is formed by residues T236 in the α₁ subunit TM1 and Y284 in the β₂ subunit TM3 [10]. Inhibitory neurosteroids appear to act at distinct sites [6, 19, 27], and a residue within the TM2 pore-forming domain plays an important role [1]. Studies of the GABA_A receptor homolog of Caenorhabditis elegans, UNC-49, also identified several residues near the enhancing neurosteroid site that were important for PS inhibition [23], suggesting the possibility that positive and negative neurosteroid modulation may have some shared structural foundation. The purpose of this study was to determine whether this part of the TM1 domain controls PS sensitivity of the mammalian GABA_A receptor as well. We swapped a segment of the C. elegans GABA receptor TM1 domain that correlates with PS insensitivity into mammalian α, β, and γ GABA_A receptor subunits, which are normally PS sensitive, and tested whether PS sensitivity decreased. The results provide several novel insights into the structural basis of neurosteroid action and, unexpectedly, implicate the β₂ subunit TM1 domain as a determinant of spontaneous openings in extrasynaptic GABA_A receptors.

Materials and Methods

Sequences used were as follows: NP000797 (human α₁), NP032096 (rat β₂), NP000807.2 (human γ_2S), NM000815 (human δ) NM021841 (rat α₆), AAD42383 (C. elegans UNC-49B.1), AAD42387 (C. elegans UNC-49C). Sequences differences between rat and human are minor: outside of the large intracellular loop between TM3 and TM4, β₂ subunits are identical and α₆ subunits differ at 6 positions (5 in the extracellular N and C termini, and 1 in TM4). The TM3-TM4 loop, which is not known to affect receptor pharmacology, shows greater divergence (β₂ subunits differ by 1 residue, α₆ subunits differ by 22). Subunit cDNAs were cloned into a pcDNA3.1(-)-based vector (Invitrogen, Carlsbad CA), linearized, and transcribed using the T7 mMessage mMachine kit (Applied Biosystems, Foster City CA). For α₁β₂γ₂ receptors, subunit mRNAs were injected at 0.1 μg/μl (α₁ and β₂) and 0.05 μg/μl (γ₂). For α₆β₂δ receptors, subunit mRNAs were injected at 1.0 μg/μl (α₆ and β₂), and 0.5 μg/μl (δ) using a Nanoliter Injector (50.6 nl; World Precision Instruments, Sarasota FL). α₁β₂γ₂ receptors expressed well, while, α₆β₂δ receptor generally expressed poorly, with many oocytes showing little or no GABA response. Electrophysiology was performed according to standard protocols, as previously described [23]. Cells were voltage clamped at −60 mV. GABA concentration-response curves and 5α, 3α-THDOC potentiation curves were fit with the equation:

$$I=I_{\text{max}}/\{1+{10}^{(log\text{EC}50-[\text{agonist}])\ast n}\}.$$

PS concentration-response curves and 5α, 3α-THDOC inhibition curves were fit using the equation:

$$I/I_{\mathit{\text{max}}}=1/\{{([\text{inh}]/{\text{IC}}_{\text{50}})}^n+1\}$$

where I is current at a given ligand concentration, I_max is current at saturation (for GABA and 5α, 3α-THDOC curves), in the absence of inhibitor (for PS curves), or at the peak of the bell-shaped curve (for 5α, 3α-THDOC inhibition), EC₅₀ and IC₅₀ are the ligand concentrations required to produce half-maximal effect, and n is the slope coefficient. For a subset of the neurosteroid concentration-response curves, it was necessary to constrain the value of n to 1, to achieve a fit that passed through the data points. Linear regression was used where only two data points were available. For each new batch of oocytes, we verified that the previously-determined GABA EC₅₀ concentration produced half of the current produced by saturating GABA. All drugs were obtained from Sigma-Aldrich (St. Louis, MO). Error presented is SEM. Statistical comparisons were performed using the Student's T test unless otherwise indicated.

Results

Structure and function of mutant GABA_A receptors

PS sensitivity of the UNC-49B isoform of the C. elegans GABA receptor can be increased 20 fold by swapping in a 19 amino acid segment from another C. elegans isoform (UNC-49C). Therefore, UNC-49B residues within this segment are associated with PS insensitivity [23]. This segment also contains the residue necessary for neurosteroid potentiation of mammalian GABA_A receptors [10] (Fig. 1A). To test how this segment affects neurosteroid inhibition and potentiation of mammalian receptors, we swapped the UNC-49B residues into mammalian GABA_A receptor subunits to create chimeric subunits (designated by ‘X’, i.e. α_1×) and co-expressed them with wild-type subunits in different combinations. All subunit combinations were functional except those in which the α₁ and β₂ subunits were both chimeric. Effects on GABA sensitivity (Fig. 1B) were statistically insignificant (P > 0.05) except for substitution of the γ₂ subunit, which reduced GABA sensitivity 2.3 fold compared to wild type.

Figure 1.

Sequence, GABA sensitivity, and PS inhibition of GABA_A receptors

(A) Amino acid sequences of GABA_A receptor subunits and UNC-49 subunits. The swapped TM1 region is indicated by a bar above the residues, the α subunit glutamine necessary for neurosteroid potentiation is indicated with an arrow, and the UNC-49C residues that can be swapped into UNC-49B to alter PS sensitivity are indicated with triangles (black indicates residues that increase UNC-49B PS inhibition, white indicates a residue that reduces PS inhibition of UNC-49B). (B) GABA concentration-response curves for wild-type and chimeric α₁β₂γ₂ receptors (left) and α₆β₂δ receptors (right; n=4 for α₁β₂γ₂ and α₆β_2×δ, n=3 for all others). (C) Representative traces of wild-type and chimeric α₁β₂γ₂ GABA_A receptors exposed to EC₅₀ GABA alone, or EC₅₀ GABA plus 10 μM PS. Bars above traces indicate duration of drug exposure. (D) PS concentration-response curves for wild-type and chimeric α₁β₂γ₂ receptors. (E) Plot of PS IC₅₀ values for all combinations of wild-type and chimeric α₁β₂γ₂ receptors (* P < 0.05 Student's T test, # P < 0.05 Mann-Whitney test, compared to wild-type).

Neurosteroid sensitivity of α₁β₂γ₂ receptors

To examine modulation by inhibitory and enhancing neurosteroids, PS was applied to oocytes with EC₅₀ GABA, and 5α, 3α-THDOC was applied to oocytes with EC₂₀ GABA. Neurosteroids were pre-applied for 20s to equilibrate binding sites, and to determine whether neurosteroids exerted direct GABA-independent effects. Surprisingly, substituting the UNC-49B TM1 segment caused no reduction of PS sensitivity in the mammalian GABA_A receptors in any combination, but instead significantly increased PS sensitivity when substituted into the α1 subunit (P<0.05 for α_1×β₂γ₂ and P=0.05 for α_1×β₂γ_2× compared to wild type; Fig. 1D, E). No direct effects of PS were observed for wild-type or chimeric receptors (Fig. 1C). These results suggest that the importance of the TM1 domain for PS inhibition is shared between C. elegans and mammals, but there are clearly substantial mechanistic differences.

THDOC modulation was also affected by the TM1 substitutions. THDOC produced little direct activation (Fig. 2A), and its modulation of GABA-evoked currents was bell-shaped, with maximal potentiation at 1 μM (Fig. 2B). As expected, α₁ chimeras showed markedly reduced 5α, 3α-THDOC potentiation at all concentrations (Fig. 2B), because the substituted segment contains a tryptophan residue in place of Q241, already known to correlate with reduced sensitivity [10]. None of the other TM1 substitutions significantly altered the potency of THDOC potentiation or inhibition, or the efficacy of THDOC inhibition compared to wild type (P > 0.05). However, receptors containing chimeric γ₂ subunits showed significantly increased efficacy of 5α, 3α-THDOC potentiation (P < 0.05). This effect was not observed when the β₂ subunits were also chimeric, suggesting that β₂ and γ₂ subunits together influence neurosteroid potentiation (Fig. 2C).

Figure 2.

5α, 3α-THDOC modulation of α₁β₂γ₂ GABA_A receptors

A) Representative traces of wild-type and chimeric α₁β₂γ₂ GABA_A receptors exposed to EC₂₀ GABA alone, or EC₂₀ GABA plus 1 μM 5α, 3α-THDOC. (B) Concentration-response plots for 5α, 3α-THDOC potentiation of EC₂₀ GABA currents. The bell-shaped concentration-response curves are separated into the rising phase (top plot) and the falling phase (lower plot). The 1.0 μM data point is shown in both plots. (C) Comparison of potentiation by 1 μM 5α, 3α-THDOC of all wild-type and chimeric combinations (n = 6 for each data point, * P < 0.05, # P = 0.05 Mann-Whitney test, compared to wild-type).

Neurosteroid sensitivity of α₆β₂δ receptors

We also tested the effects of TM1 substitutions in the β₂ subunits when expressed with wild-type α₆ and δ subunits. Pre-application of PS to wild-type α₆β₂δ, in the absence of GABA, caused large outward currents (Fig. 3A), consistent with blockade of a chloride leak current. A previous study reported that α₆β₂δ receptors expressed in Xenopus oocytes have a high rate of spontaneous opening, producing a constitutive chloride current that could be inhibited by bicucculine, picrotoxinin, furosemide, and Zn²⁺ [7]. The outward currents that we observed are consistent with PS inhibition of spontaneous activity. Regardless of the leak current, subsequent application of EC₅₀ GABA to the wild-type or mutant receptors resulted in an inward current (Fig. 3A), and this current was equally sensitive to PS inhibition in the wild-type and chimeric receptors (Fig. 3B).

Figure 3.

Inhibition of α₆β₂δ receptors by PS and other inhibitors

(A) Representative traces of wild-type α₆β₂δ and α₆β_2×δ receptors exposed to EC₅₀ GABA or EC₅₀ GABA plus 10 μM PS. The wild-type, but not the chimera receptor showed large outward currents during preapplication. (B) Concentration-response curve for PS inhibition of GABA-evoked currents from α₆β₂δ (n=6) and α₆β_2×δ (n=5) receptors. (C) Magnitude of PS-evoked outward current relative to the EC₅₀ GABA current for selected subunit combinations (α₁-containing receptors are also shown for comparison; n ≥ 5 for each data point). (D) Traces from oocytes expressing wild-type α₆β₂δ receptors exposed to a series of inhibitors: 10 μM PS, 10 μM picrotoxinin, 500 μM furosemide, and 100 μM Zn²⁺. Each row represents an independent cell; EC₅₀ GABA traces are shown for comparison. (E) Same as D, for α₆β_2×δ receptor-expressing cells.

Although PS inhibition of GABA-evoked currents was unaffected, the β₂ subunit TM1 substitution nearly eliminated the apparent PS-evoked outward current (Fig. 3A, C; P<0.05, Mann-Whitney test). We considered two possible explanations: β₂ TM1 substitution might eliminate PS sensitivity of the spontaneous current, or it might eliminate the spontaneous current itself. We therefore tested other inhibitors of α₆β₂δ receptor spontaneous currents: picrotoxinin, furosemide, and Zn²⁺ [7]. Like PS, the apparent outward currents produced by these inhibitors were eliminated in α₆β_2×δ receptors (Fig. 3D, E). Since these compounds are structurally dissimilar and act through different mechanisms, these results strongly suggest that the β₂ TM1 substitution eliminates the spontaneous current produced by the α₆β₂δ receptors.

Finally we compared 5α, 3α-THDOC modulation of wild-type and chimeric α₆β₂δ receptors. Potentiation and direct activation were observed (Fig. 4A-C). The concentration-response curves for both effects were strongly bell shaped for the α₆β_2×δ chimera, and tended toward bell shaped for the wild-type receptor, with maximal effects observed in the 1-3 μM range (Fig. 4B, C). At 3 and 10 μM, THDOC was inhibitory, reducing GABA-evoked currents of the chimeric receptor below the levels of GABA alone (Fig. 4C). TM1 substitution in the β₂ subunit reduced both direct activation and potentiation at higher 5α, 3α-THDOC concentrations (above 0.3 μM for direct activation, and above 1.0 μM for potentiation). Potentiation of GABA-evoked currents at neurosteroid concentrations below 3.0 μM was unaffected. The efficacy of inhibition at 10 μM 5α, 3α-THDOC was also increased (Fig. 4B, C). These data suggest that the β₂ subunit substitution either increases sensitivity to 5α, 3α-THDOC inhibition or reduces sensitivity to 5α, 3α-THDOC potentiation and direct activation. In either case, they demonstrate that the β₂ subunit TM1 domain is important for the actions of high concentrations of 5α, 3α-THDOC.

Figure 4.

5α, 3α-THDOC direct activation and potentiation of α₆β₂δ receptors

(A) Wild-type α₆β₂δ (left) or chimeric α₆β_2×δ (right) receptors were exposed to EC₂₀ GABA or EC₂₀ GABA plus 1 μM 5α, 3α-THDOC. Concentration-response plots of 5α, 3α-THDOC direct activation (as a proportion of EC₂₀ GABA current; B) and potentiation of EC₂₀ GABA currents (C). As in Fig. 2B, the rising and falling phases of the bell-shaped curves are plotted separately (upper and lower traces, respectively). Statistical significance (wild-type vs. chimera; P < 0.05) is indicated by * (Student's T test) or # (Mann-Whitney test); n = 9 for wild-type, n=3 for chimera.

Discussion

In this study, we manipulated the structure of the TM1 domains of GABA_A receptor α₁, β₂, and γ₂ subunits to test whether sequence requirements for PS inhibition of the C. elegans UNC-49B GABA receptor were conserved with mammalian GABA_A receptors. We substituted TM1 residues that caused PS insensitivity in UNC-49B into the mammalian subunits, and tested the PS sensitivity of the resulting chimeric receptors. Surprisingly, none of the TM1 substitutions caused the predicted reduction of PS sensitivity in any context, and in one case the mutation increased PS sensitivity. These results are unexpected because residues important for allosteric regulation generally play conserved roles throughout the family of ligand-gated chloride channels. Based on these results, we conclude that the residues identified in the C. elegans receptors are not forming a conserved PS binding pocket, and if they are playing another conserved role, their actions must be dependent upon diverged residues elsewhere in the receptor. One caveat to this interpretation is that in the C. elegans experiments, all five subunits of the pentamer were mutated [23], whereas in this study, either the α₁ or β₂ subunits needed to be wild type to preserve receptor functionality. However it seems unlikely that the α₁ and β₂ subunits could be functionally identical and redundant with respect to PS inhibition, considering that their sequences differ at 8 out of 19 residues within the substituted region. Therefore it is also possible that the neurosteroid sites in the C. elegans UNC-49 receptor are significantly diverged from the mammalian sites, and therefore may be valuable as novel targets in the development of new anthelminthics and possibly insecticides with limited toxicity to mammals.

Although this study did not establish a conserved basis for neurosteroid inhibition between mammalian and C. elegans GABA_A receptors, several new insights emerged into the action of enhancing neurosteroids. First, 5α, 3α-THDOC potentiation was increased by substituting the γ₂ subunit M1 domain, but that the increased potentiation was only seen when the β₂ M1 domain was wild-type. This result identifies minor roles for β₂ and γ₂ subunits in neurosteroid potentiation where previously, only the α₁ subunit was shown to be important [9, 10]. Second, chimeric β₂ subunits in α₆β_2×δ receptors conferred reduced sensitivity to 5α, 3α-THDOC direct activation and increased sensitivity to 5α, 3α-THDOC inhibition at concentrations above 1 μM, thus identifying a second site on the β₂ subunit that is important for neurosteroid action, at least in this subunit combination. Third, sensitivity to neurosteroid direct activation was also correlated with a high level of spontaneous activity, raising the possibility that neurosteroid-gated channel opening and spontaneous channel opening are mechanistically similar. Finally, we observed no parallel effects of our mutations on the actions of 5α, 3α-THDOC and PS. Consistent with earlier pharmacological studies [6, 19, 27], this finding provides structure-based evidence that the binding sites and mechanisms of enhancing pregnane neurosteroid inhibitory sulfated steroids are separate and distinct.

The most striking effect we observed was the decrease in spontaneous activity of α₆β₂δ receptors by substitution of the β₂ subunit TM1 domain. Spontaneous activity has been observed in several homomeric and heteromeric GABA_A receptors in recombinant studies, and is associated with particular subunits such as β₁ [20], ε [17], and α₆ [7]. Recent studies highlight its potential to regulate neuronal excitability in vivo as well [12]. Residues in the TM2 and TM3 domains are known to affect spontaneous GABA_A receptor gating [4, 14, 18]. Our results additionally demonstrate a role for the β₂ subunit TM1 domain. Physiologically, these findings are important because they identify another potential role for endogenous PS. α₆β₂δ receptors are expressed in cerebellar granule cells; their onset of expression during development correlates with motor learning in the cerebellum [25]. α₆β₂δ receptors are extrasynaptic, and are believed to generate tonic inhibition in proportion to the levels of extrasynaptic GABA that spills over from nearby GABA synapses [3]. The finding that spontaneous activity of α₆β₂δ receptors may also be inhibited by PS raises the possibility that inhibitory tone may be modulated by PS in both a GABA-dependent and GABA-independent manner. Considering its somewhat specific functional effect (i.e. eliminating spontaneous current in α₆β₂δ receptors without affecting GABA sensitivity appreciably), the chimeric β₂ subunit used in this study may prove useful for further study of structural and physiological aspects of spontaneous activity in extrasynaptic GABA_A receptors.