Pharmacological characterisation of murine α4β1δ GABAA receptors expressed in Xenopus oocytes

Background GABAA receptor subunit composition has a profound effect on the receptor’s physiological and pharmacological properties. The receptor β subunit is widely recognised for its importance in receptor assembly, trafficking and post-translational modifications, but its influence on extrasynaptic GABAA receptor function is less well understood. Here, we examine the pharmacological properties of a potentially native extrasynaptic GABAA receptor that incorporates the β1 subunit, specifically composed of α4β1δ and α4β1 subunits. Results GABA activated concentration-dependent responses at α4β1δ and α4β1 receptors with EC50 values in the nanomolar to micromolar range, respectively. The divalent cations Zn2+ and Cu2+, and the β1-selective inhibitor salicylidine salicylhydrazide (SCS), inhibited GABA-activated currents at α4β1δ receptors. Surprisingly the α4β1 receptor demonstrated biphasic sensitivity to Zn2+ inhibition that may reflect variable subunit stoichiometries with differing sensitivity to Zn2+. The neurosteroid tetrahydro-deoxycorticosterone (THDOC) significantly increased GABA-initiated responses in concentrations above 30 nM for α4β1δ receptors. Conclusions With this study we report the first pharmacological characterisation of various GABAA receptor ligands acting at murine α4β1δ GABAA receptors, thereby improving our understanding of the molecular pharmacology of this receptor isoform. This study highlights some notable differences in the pharmacology of murine and human α4β1δ receptors. We consider the likelihood that the α4β1δ receptor may play a role as an extrasynaptic GABAA receptor in the nervous system.

Receptors incorporating α1, α2 or α3 subunits most commonly assemble with β and γ subunits, and are generally located at the synapse where they mediate phasic inhibition; whereas α4 and α6-containing receptors assemble with β and δ subunits, and are found extrasynaptically, mediating tonic inhibition [13]. The δ subunit is found almost exclusively in the extrasynaptic domain and forms α4βδ or α6βδ receptors in the cortex and cerebellum, respectively [2]. However, it has also been suggested that δ can co-assemble with α1 subunits in hippocampal interneurons [14,15].
The existence of α4β2/3δ receptors in brain regions such as the hippocampal dentate gyrus is well established, but there is a paucity of data which characterises the potential native expression of the α4β1δ receptor [2]. This is largely due to the limited number of β1-subtype-selective compounds available [16], and the fact that no β1 subunit knock-out mouse has been generated. As a consequence, most previous studies investigating α4βδ receptors have focused on those subtypes containing the β2 or β3 subunits as these are confirmed to be native receptor combinations [2,17,18]. However, co-regulation of the expression levels of α4, β1 and δ subunits in, for example, the periaqueductal grey matter throughout the female oestrous cycle has been reported, suggesting that α4β1δ receptors may perform a hitherto overlooked important physiological role [19]. Also, experimental evidence for the co-expression of α4, β1 and δ subunits at extrasynaptic sites exists for pyramidal hippocampal neurons [20]. Furthermore, recent studies have emphasized the importance of β1-containing receptors in relation to sleep control [21] and ataxia [22], and polymorphic mutants of the β1 subunit are thought to be important in determining the extent of alcohol consumption [23]. Therefore, since α4, β1 and δ subunits are coexpressed in several brain regions [24], the possibility exists that these subunits can co-assemble into native functional receptors, most likely at extrasynaptic sites, raising the prospect that such a receptor subtype may be important for neuronal physiology.
Few studies have investigated the properties of either human [25][26][27] or murine [23,28] recombinant α4β1δ receptors, and none of these have fully addressed the characterisation of classic GABA A receptor ligands. In these studies, interesting pharmacological differences between murine and human α4β1δ receptors are manifest: human receptors expressed in Xenopus oocytes [25] are constitutively active and display a GABA EC 50 in the mid-nanomolar range whereas the rat/ mouse receptor expressed in oocytes or HEK293 cells display low-micromolar sensitivity towards activation by GABA [23,28]. To address this apparent speciesdependent pharmacology and to address the effects of commonly used GABA receptor ligands, we report the pharmacological profiles for several GABA A receptor ligands at recombinant murine α4β1 and α4β1δ receptors expressed in Xenopus laevis oocytes.

Results
GABA concentration response relationships for α4β1 and α4β1δ receptors GABA concentration response curves were generated for α4β1 and α4β1δ receptors to evaluate GABA potency at these receptors. Oocytes were challenged with increasing concentrations of GABA ( Figure 1A). Depending upon expression levels, maximal currents to GABA ranged from 100 to 3000 nA for cells expressing α4β1 receptors and from 300 to 3000 nA for cells expressing α4β1δ receptors. The holding currents were between 0 and -40 nA and the receptors were not constitutively active. The GABA EC 50 values were 0.89 μM for α4β1δ receptors and 2.7 μM for α4β1 receptors ( Figure 1B, Table 1). GABA concentration-response data were fitted as monophasic curves, with Hill coefficients of 1.1 and 0.87 for α4β1δ and α4β1 receptors, respectively (Table 1).

Inhibition of GABA responses by divalent cations
Zn 2+ has been shown to be a useful pharmacological tool to distinguish αβ receptors from αβδ/γ receptors in recombinant expression systems, as δ/γ subunit-containing GABA A receptors are less sensitive to inhibition by Zn 2+ compared to their αβ counterparts [29,30]. The increased potency at αβ receptors is attributed to the incorporation of a third β subunit that contains an important histidine residue for coordinating Zn 2+ ions at the top of channel lining M2 region [30,31].
We evaluated Zn 2+ -inhibition of the GABA response for both α4β1 and α4β1δ receptor isoforms to investigate whether Zn 2+ can distinguish between these receptor subtypes. Oocytes were pre-incubated for 1 min with various concentrations of Zn 2+ followed by co-application of Zn 2+ and GABA at an EC 75 (3 μM for α4β1δ and 8 μM for α4β1). As expected, Zn 2+ exhibited a higher potency at α4β1 than α4β1δ receptors. For α4β1δ receptors, Zn 2+ inhibited the GABA response with an IC 50 of 3.3 μM (Figure 2A,B, Table 1). By contrast, for α4β1 receptors, biphasic inhibitory behaviour was observed with a high potency component (IC 50A = 0.13 nM) accounting for 25% of the total inhibition, and a more dominant lower potency component (IC 50B = 56 nM) accounting for the remaining (~75%) inhibition ( Figure 2A,B, Table 1).
Cu 2+ is another well-established inhibitor of GABA currents at γ subunit-containing GABA A receptors [32], but its inhibitory profile at δ subunit-containing receptors has only recently been explored [33]. This revealed that Cu 2+ was far more potent as an inhibitor of steady-state GABA responses at α6β3δ receptors (IC 50 = 65 nM) compared to α1β2γ2 receptors (IC 50 = 85 μM) expressed in human embryonic kidney 293 (HEK293) cells [33]. To further investigate the inhibitory selectivity of Cu 2+ at α4β1δ receptors, we co-applied increasing concentrations of Cu 2+ with GABA (EC 75 : 3 μM) for 70 s, and measured the steadystate responses, reproducing the experimental conditions used by McGee et al. [33] to determine the IC 50 values. This response was taken as the fraction of the peak response of a previously-applied GABA concentration, and normalized to the steady-state response of GABA EC 75 in the absence Cu 2+ ( Figure 2C). An almost complete inhibition of the α4β1δ GABA-activated response was observed with increasing concentrations of Cu 2+ (IC 50 = 0.82 μM) ( Figure 2D, Table 1).

Inhibition by the β1 subunit-selective blocker SCS
The salicylic acid derivative, SCS, is one of only a few compounds known to selectively target β1 subunit-containing GABA A receptors over β2and β3-containing receptors [34]. SCS is reported to inhibit GABA-activated responses of α2β1γ2, α2β1γ1 and α2β1γ1θ receptors, but the pharmacological profile of SCS at α4 or δ-containing receptors, has not been addressed. To examine the interaction of SCS with α4β1δ receptors, we evaluated the inhibition of the GABA EC 20 response following pre-application of the antagonist. In this instance, a low EC 20 concentration was chosen to enable a direct comparison between our study and that obtained previously [34]. SCS inhibited the GABA response with an IC 50 of 4.2 nM, but did not cause complete inhibition of the GABA current, attaining a steady-state level of inhibition at 67.2 ± 3.2% of control ( Figure 2E, Table 1).

Neurosteroid potentiation at α4β1δ receptors
Neurosteroids are potent modulators at most, if not all, GABA A receptor isoforms [35]. Some studies report that neurosteroids are more potent at recombinant δ subunit-containing receptors compared to their γ subunit-containing counterparts [17,36,37]. Accordingly, we evaluated the sensitivity of α4β1δ receptors to neurosteroidinduced potentiation. Increasing concentrations of THDOC were co-applied with an approximately EC 7 GABA concentration. A significant increase in the GABA peakresponse was recorded with THDOC concentrations of 30 and 100 nM (P < 0.05, one-way analysis of variance (ANOVA) with Dunnett's post test, Figure 2F).

Discussion
In this study we investigated the pharmacological profiles of several GABA A receptor ligands at murine α4β1δ receptors. The GABA potency of 0.89 μM (EC 50 ) at α4β1δ receptors determined in this study is similar to that previously reported for the rat receptor (2.02 μM) expressed in Xenopus oocytes [28], and the mouse/rat receptor (1.99 μM) expressed in HEK293 cells [23]. However, a much higher GABA potency (EC 50 of 24 nM) has been reported for the human α4β1δ receptor expressed in Xenopus oocytes [25]. The GABA potency for human α4β1 (0.72 μM) is reported to be somewhat lower [25] than the murine combination studied here. Furthermore, constitutive activity in the absence of GABA was reported for the human recombinant α4β1δ receptor [25], but this was not evident in our studies with the murine equivalent. From these observations it is possible that differences in constitutive activity may explain the considerable differences in GABA potency reported between species observed by others and in this study. However, several other factors may also influence the observed potency shift, such as differences in relative subunit expression levels which may significantly impact on the final receptor stoichiometry. In particular, δcontaining receptors are prone to such variations in stoichiometry. This is evident from studies in which variations in the amount of RNA injected into Xenopus oocytes apparently yield different receptor populations [38]; likewise, concatemeric receptors also demonstrate that alternative assemblies of functional δ-containing receptors are possible [12,39]; and from reports that both high and low affinity conformations of the related α4β3δ receptor apparently exist [25,40].
In this study we report the first characterisation of specific GABA antagonists at α4β1δ receptors. The approximately 60-fold potency shift in the IC 50 values for Zn 2+ from low micromolar to mid nanomolar, when comparing α4β1δ and α4β1, is consistent with prior observations made for α4β3δ and α4β3 receptors [29]. Our determination of monophasic Zn 2+ potency suggests it is highly unlikely that there is any contamination with binary α4β1 receptors in our α4β1δ-expressing oocytes. Also, the inability of Zn 2+ to inhibit the response below the holding current baseline corroborates the absence of any constitutive activity caused by homomeric β1 receptors [41].
It has previously been suggested that the stoichiometry of binary GABA A receptors containing only α and β subunits is dependent on the type of α and β subunit present, with a 3α:2β stoichiometry predominating for α1β2 receptors, but a 2α:3β stoichiometry preferred for α4β2 receptors [11]. The biphasic inhibitory behaviour of Zn 2+ in our study would suggest the presence of a mixed GABA A receptor population, perhaps with both 3α:2β and 2α:3β stoichiometries expressed in the oocyte. This conclusion is rationalized by the fact that an additional β subunit will contribute another H267 residue, known to be critical for Zn 2+ inhibition [31], which may underpin the higher sensitivity component to Zn 2+ in the biphasic inhibition curves.
This study also evaluated the pharmacology of additional compounds at α4β1δ known to interact with GABA A receptors. We have established that the pharmacological profile of the β1-selective inhibitor SCS, at α4β1δ receptors, is in agreement with the values reported for α2β1γ2 receptors when expressed in Xenopus oocytes (IC 50 = 4.36 nM and maximum inhibition of 33% [34], suggesting that α4 and δ subunits do not strongly influence the molecular interaction with SCS. Our data evaluating the inhibition of desensitized GABA responses by Cu 2+ suggest that the maximal inhibition of Cu 2+ is greater at α4β1δ (98% inhibition at 10 μM: expressed in oocytes) than at α6β3δ receptors (expressed in HEK293) where the maximal inhibition of the steady-state response was reported to be~68% [33]. Although this difference may conceivably be due to the different expression systems, it is equally likely that it reflects some subunit-selectivity (α4 > α6) for Cu 2+ .
As hypothesized, THDOC concentrations at 30 nM and 100 nM were able to potentiate low (EC 7 ) GABA concentrations, suggesting that α4β1δ receptors are likely to be modulated by physiologically relevant concentrations of neurosteroids, as akin to many other GABA A receptor subtypes [35]. This is unsurprising given that the highly conserved neurosteroid binding site for potentiating GABA responses is contained within the receptor α subunit.

Conclusions
In summary, we have extended the pharmacological profile of the α4β1δ receptor, a subtype that has been studied surprisingly little compared to its β2and β3containing counterparts, despite its presence in many significant brain areas. We report a high GABA potency (EC 50 of 0.89 μM), a weaker sensitivity to Zn 2+ compared to α4β1 receptors, and a high sensitivity to SCS and THDOC.
The observed discrepancy in potency compared to the human ortholog highlights the need for caution when making comparisons of pharmacological parameters across species for this particular GABA receptor subtype. For example, in addition to the different GABA potencies, there are notable differences in constitutive activity for this receptor isoform. Whether these differences translate into the native behaviour of this receptor remains to be determined. Finally, the different expression pattern of the β1 subunit compared to β2 and β3 subunits [24] suggests that β1-containing subtypes may offer novel therapeutic targets and opportunities, and emphasizes a greater need for subtype-selective ligands and/or genetically modified mice to better understand the native function and pharmacology of this GABA A receptor isoform. Indeed recent findings suggest a clinically relevant role for β1-containing extrasynaptic GABA A receptors in the nucleus accumbens mediating addictive behaviours related to excessive alcohol intake [23].

Electrophysiological recordings
Recordings were performed at room temperature 2-4 days (α4β1) or 3-6 days (α4β1δ) post injection using two-electrode voltage-clamp (Axoclamp 2B amplifier, Digidata 1322A interface and pClamp 8 (Molecular Devices, Wokingham, UK)). Electrodes were filled with 3 M KCl and had resistances of 0.5 to 3 MΩ. The oocytes were clamped at a holding potential of -60 mV and were perfused with buffer (100 mM NaCl, 2 mM KCl, 2 mM CaCl 2 , 1 mM MgCl 2 , 5 mM HEPES, pH adjusted to 7.4 with NaOH) throughout the recordings. All currents were adjusted when necessary for voltage drop using equation 1: Where I measured is the observed current, V h is the holding potential (mV), V rev is the Cl − reversal potential for oocytes (usually~-30 mV), and ΔV is the change in the holding potential (mV) due to loss of clamp during at the peak of large GABA-evoked currents. Drugs were applied using a fast application system coupled to a small bath volume (~1 ml), enabling drug onset times of~10 ms.

Generation of concentration response curves
Agonist concentration response curves were generated by measuring peak responses normalized to a standard concentration of GABA applied between each or every second application (typically an EC 10-20 concentration). To pool data from different oocytes, the data were fitted and normalized to the calculated peak of the curve using GraphPad Prism (version 6.0) according to equation 2;