Stoichiometry of δ subunit containing GABAA receptors

Background and Purpose Although the stoichiometry of the major synaptic αβγ subunit-containing GABAA receptors has consensus support for 2α:2β:1γ, a clear view of the stoichiometry of extrasynaptic receptors containing δ subunits has remained elusive. Here we examine the subunit stoichiometry of recombinant α4β3δ receptors using a reporter mutation and a functional electrophysiological approach. Experimental Approach Using site-directed mutagenesis, we inserted a highly characterized 9′ serine to leucine mutation into the second transmembrane (M2) region of α4, β3 and δ subunits that increases receptor sensitivity to GABA. Whole-cell, GABA-activated currents were recorded from HEK-293 cells co-expressing different combinations of wild-type (WT) and/or mutant α4(L297S), β3(L284S) and δ(L288S) subunits. Key Results Recombinant receptors containing one or more mutant subunits showed increased GABA sensitivity relative to WT receptors by approximately fourfold, independent of the subunit class (α, β or δ) carrying the mutation. GABA dose–response curves of cells co-expressing WT subunits with their respective L9′S mutants exhibited multiple components, with the number of discernible components enabling a subunit stoichiometry of 2α, 2β and 1δ to be deduced for α4β3δ receptors. Varying the cDNA transfection ratio by 10-fold had no significant effect on the number of incorporated δ subunits. Conclusions and Implications Subunit stoichiometry is an important determinant of GABAA receptor function and pharmacology, and δ subunit-containing receptors are important mediators of tonic inhibition in several brain regions. Here we demonstrate a preferred subunit stoichiometry for α4β3δ receptors of 2α, 2β and 1δ.

The subunit composition of GABAA receptors is an important determinant of their functional properties as demonstrated by the type of α subunit and presence of the γ2 subunit affecting, for example, receptor sensitivity to benzodiazepines (Korpi et al., 2002). Given that orthosteric and many allosteric binding sites on GABAA receptors are interfacial (Sieghart et al., 2012), it becomes important to understand whether there are preferred subunit stoichiometries that will critically define the nature of these subunit interfaces and thus the receptor's response to ligand binding. Compared with αβγ receptors, we know least about δ subunitcontaining GABAA receptors, which play an important role mediating tonic inhibition in several brain regions (Brickley et al., 2001;Porcello et al., 2003;Farrant and Nusser, 2005;Santhakumar et al., 2010). To address this deficit, we have employed a pharmacological analysis, in combination with a reporter mutation, to better understand the structural properties of δ-containing receptors using heterologous expression systems (e.g. HEK-293 and Xenopus laevis oocytes). Previous reports note that some functional discrepancies have been observed for αβδ receptors, such as EC50 values for GABA and ethanol sensitivity (Wallner et al., 2003), which have been postulated to arise, in part, from differences in subunit stoichiometry (Borghese et al., 2006;Wagoner and Czajkowski, 2010). Although the stoichiometry of major synaptic αβγ GABAA receptor isoforms has broad consensus support for 2α:2β:1γ (Backus et al., 1993;Chang et al., 1996;Tretter et al., 1997), an unequivocal view of the stoichiometry for extrasynaptic δ-containing receptors remains elusive. Although atomic force microscopy of recombinant α4β3δ receptors has suggested a stoichiometry of 2α:2β:1δ (Barrera et al., 2008), biochemical analysis of recombinant α4β2δ receptors indicates that more than one δ can be incorporated into the receptor complex (Wagoner and Czajkowski, 2010). Moreover, it was recently demonstrated on the basis of using α1β3δ  and α6β3δ concatemers ) that more than one δ subunit can be incorporated into functional channels, although for the former subtype, a constrained conformation of 2α:2β:1δ most closely resembled the pharmacological profile of unconstrained recombinant α1β3δ receptors .
In this study, we have examined the subunit stoichiometry of functional recombinant α4β3δ receptors, utilizing polar substitutions of a highly conserved leucine residue within the second transmembrane region (M2) of GABAA receptors. This residue exchange acts as a reporter mutation causing a profound increase in agonist potency consequently displacing the agonist dose-response curve (Chang et al., 1996;Chang and Weiss, 1999), as also observed for nicotinic ACh receptors (nAChRs) (Filatov and White, 1995;Labarca et al., 1995) and 5-HT 3 receptors (Yakel et al., 1993). The extent of the curve shift is correlated with the number of polar substitutions per ion channel complex, and this has been used to deduce the subunit stoichiometry of recombinant α1β2γ2 GABAA receptors (Chang et al., 1996). By inserting this highly characterized 9′ serine to leucine (L9′S) mutation into α4, β3 and δ subunits, we derive a subunit stoichiometry of 2α:2β:1δ for functional α4β3δ GABAA receptors. Furthermore, our data indicate that for three different, but commonly used, cDNA transfection ratios, the number of incorporated δ subunits seemingly remains fixed at one.

Transient receptor expression in HEK-293 cells
HEK-293 cells were cultured in DMEM supplemented with 10% v/v fetal calf serum, 2 mM glutamine, 100 u·mL −1 penicillin G and 100 μg·mL −1 streptomycin, and were incubated at 37°C in humidified 95% air and 5% CO2. HEK cells were plated on poly-L-lysine-coated coverslips and transfected using a calcium phosphate protocol. Briefly, cDNA encoding wild-type (WT) and/or mutant α4, β3 and δ subunits, in transfection ratios of 1:1:1, 1:1:10 or 10:1:10, were mixed with 340 mM CaCl2, and an equal volume of HBSS (50 mM HEPES, 280 mM NaCl and 2.8 mM Na2HPO4, pH 7.2) to form a precipitate. The total amount of cDNA used for each transfection was 4 μg. The DNA-calcium phosphate suspension was applied to HEK cells, which were incubated overnight, and used for electrophysiology 16-24 h after transfection. For the co-expression experiments, WT subunits and their respective L9′S mutants were co-expressed in equal amounts, with an overall transfection ratio remaining at 10α:1β:10δ.
Whole-cell agonist-activated currents were recorded with an Axopatch 200B amplifier (Molecular Devices, Sunnyvale, CA, USA). Cells were voltage clamped between −20 and −60 mV, depending on peak current size. Whole-cell currents were filtered at 5 kHz (−36 dB), digitized at 50 kHz via a Digidata 1332A (Molecular Devices) and recorded to disk (Dell Pentium Dual Core Optiplex 960; Dell, Berks, UK). Series resistances were monitored throughout each experiment and deviations >20% resulted in the data being excluded from further analysis.

Data analysis
Currents (IGABA) activated by concentrations of GABA ([A]) were normalized to the maximal current caused by a saturating concentration of GABA (IMax,GABA). The normalized concentration-response curves were fitted with the following equation using a non-linear least-squares method: where EC50 is the GABA concentration inducing a halfmaximal current, n is the nH, i is the number of components where j = 1 -3.
For WT α4β3 and α4β3δ receptors, the level of inhibition exhibited by 1 μM Zn 2+ on GABA EC50 responses for each receptor subtype was also assessed.
The spontaneous channel activity exhibited by each mutant was calculated by expressing the outward current induced by the blocker picrotoxin (IPTX; 1 mM) as a percentage of the maximum current, defined as the sum of IMax,GABA and IPTX. No spontaneous activity (SA) was observed for WT α4β3δ receptors. The level of SA was quantified according to: All data are expressed as mean ± SEM. Where appropriate, statistical analyses were performed using an unpaired Student's t-test or a one-way ANOVA.

Materials
GABA (Sigma, Dorset, UK) and zinc chloride (BDH Biochemical, Poole, UK) solutions were prepared from 1 M stocks (in water), whereas 1 mM picrotoxin (Sigma) was dissolved in extracellular Krebs solution containing 0.05% v/v dimethyl sulfoxide (Sigma). Drugs were applied via a U-tube application system (Mortensen and Smart, 2007).
With regard to channel gating αβδ m, αmβδ and αβmδ, but not αβδ receptors, exhibited spontaneous activation in the absence of exogenously applied GABA, which was blocked by the chloride channel blocker, picrotoxin (1 mM; Figure 2B). Expressed as a proportion of the total GABA-activated plus spontaneous current [IPTX/(IPTX + IMax,GABA; Figure 2B inset)], the levels of spontaneous receptor activation for αmβδ and αβδm receptors were 21.9 ± 5.3 and 16 ± 1% respectively. The highest level of spontaneous activation was exhibited by αβmδ-expressing cells (76.6 ± 6.5%) relative to αβδm (P < 0.001) and αmβδ (P < 0.05) receptors (non-parametric ANOVA -Kruskal-Wallis test). The increased degree of spontaneous receptor activation observed for the β mutant is likely to reflect the predominant role this subunit plays in stabilizing open-shut GABA channel conformation(s). It is also noteworthy that β homomers can form spontaneously opening ion channels (Krishek et al., 1996;Davies et al., 1997;Wooltorton et al., 1997;Cestari et al., 2000) unlike their α, γ or δ subunit counterparts.
Both the SA and distinctive deactivation profiles exhibited by mutant subunit-expressing cells confirmed that each mutant was efficiently co-assembled into functional αβδ receptors.

L9′S mutations in α4, β3 and δ subunits increase GABA sensitivity
By compiling GABA dose-response curves for cells expressing the subunit combinations, αβδ, αβδm, αmβδ, αβmδ and αmβδm, those containing δm, αm or βm subunits exhibited an increased GABA sensitivity compared with WT αβδ receptors, manifest by a leftward shift in the curves for mutant receptorexpressing cells ( Figure 3A). Whereas WT αβδ receptors had a GABA EC50 of 1.91 ± 0.47 μM, single mutant subunitcontaining receptors possessed lower EC50 values of 0.46 ± 0.11 μM (αβδm), 0.12 ± 0.03 μM (αmβδ) and 0.11 ± 0.04 μM (αβmδ ; Table 1). Notably, the nH obtained for αβmδ receptors (0.63 ± 0.06; Table 1) was significantly lower than those obtained for αβδm (1.41 ± 0.12) and αmβδ (1.18 ± 0.19) receptors (Welch's t-test: P = 0.02 and P = 0.03 respectively). Although we cannot simply transpose the change in nH to any physical attribute, it is possible that the L9′S mutations altered the gating kinetics of the receptor, which may account for the changes in nH. Moreover, the effects on gating kinetics may be highly dependent on the subunit in which the mutation is inserted.

Figure 1
Primary sequence alignment of the second transmembrane region (M2) of α4, β3 and δ subunits. Prime notation (red) denotes the amino acids comprising the ion channel pore. The conserved hydrophobic 9′ leucine residues are indicated and boxed for α4, β3 and δ subunits, with their numbering in the mature subunit proteins.
The αβδm receptors exhibited an approximately fourfold (4.15) increase in GABA sensitivity, relative to αβδ receptors. The increased sensitivity of αβδm was not attributable to a large population of αβ receptors being expressed as the EC50 for αβ receptors (1.01 ± 0.13 μM; data not shown) is similar to that for αβδ receptors (1.91 ± 0.47 μM; Welch's t-test: P = 0.13). By comparison, the GABA sensitivities of αmβδ and αβmδ receptors were higher compared with δm-containing receptors, causing shifts of 16 (15.91) and 17 (17.36)-fold, respectively, in the GABA EC50 (Figure 3).
If we assume each subunit mutation has an equivalent effect on GABA potency, independent of the subunit class in which the L9′S mutation is inserted, and if we further assume that the receptor contains two α subunits, then we would expect the shift in EC50 by the 9′ mutation to be approximately the square of the change due to a single α subunit. Thus a shift of 15.91 predicts a single α subunit would cause a 3.99-fold change in EC50. Similarly for the β subunit, a 17.36-fold shift indicates each β subunit (if two copies are present in the receptor) should cause a 4.17-fold change.
These are equivalent to that caused by the δ subunit (4.15) and the findings therefore suggest that relative to the δ subunit, twice the number of α4 and β3 subunits are likely to exist in each receptor complex. Thus, because GABAA receptors are assumed to form pentameric complexes, αmβδ, αβmδ and αβδm probably contained 2αm, 2βm and 1δm subunits.

Increasing GABA sensitivity with the number of co-assembled L9′S mutant subunits
For heteromeric muscle nAChRs, each 9′ polar substitution within the ion channel confers an additional ∼10-fold

Figure 3
L9′S mutations in α4, β3 and δ subunits increase GABA sensitivity. (A) GABA concentration-response curves for αβδ, αβδm, αmβδ, αβmδ and αmβδm receptors. Data represent the mean ± SEM from n = 5-9. Each data set was fitted to the Hill equation (continuous lines) using a non-linear least-squares method. Parameters from these fits are provided in Table 1. (B) Relationship between the number of mutant receptors incorporated into the receptor pentamer (assuming a stoichiometry of 2α:2β:1δ) and the average GABA EC50 values.
increase in agonist sensitivity (Filatov and White, 1995;Labarca et al., 1995). However, such a linear relationship has not been observed for recombinant α1β2γ2S GABAA receptors (Chang and Weiss, 1999), where mutations in α, β and γ subunits contributed unequally to the increased GABA sensitivity, thus prohibiting an estimate of receptor stoichiometry (Chang and Weiss, 1999). By contrast, for our α, β and δ subunit receptors, the shifts appeared more consistent, with each mutation contributing an approximately fourfold increase in GABA sensitivity. We therefore investigated the relationship between GABA potency and the number of mutant substitutions in δ-containing GABAA receptors with αmβδm-expressing cells.
Based on our predictions, αmβδm receptors would be expected to contain three mutant subunits (i.e. two αs and one δ), and thus exhibit even greater sensitivity to GABA, than either αmβδ (double mutant) or αβδm (single mutant) receptors.
For αmβδm receptors an EC50 of 0.08 ± 0.02 μM (Table 1) was determined, which equates to a 23.87-fold increase in GABA sensitivity, which is greater than that for αmβδ and αβmδ receptors and approximates to a threefold shift (2.88) per mutant subunit. Based on the double mutant receptors, we predicted an approximately fourfold shift per subunit, and thus for three mutant subunits, we might have expected a 64-fold increase in GABA sensitivity. The discrepancy between the predicted and actual shift observed for αmβδm could arise from the δm subunit being absent from αmβδm receptor-expressing cells, leaving cell surface receptors mainly composed of αβ receptors containing just two mutant αm subunits. However, this seemed unlikely given that for αmβδmexpressing cells, there was clear evidence of cell surface δm-GFP fluorescence. Furthermore, αmβδm-expressing cells exhibited a level of SA (49.4 ± 8.4%) that was comparable with the combined spontaneous activities of αmβδ (21.9 ± 5.3%) and αβδm receptors (15.7 ± 1.3%; Figure 2B). Taken together, these data suggest that both αm and δm subunits were efficiently incorporated into functional αmβδm receptors.
Our predicted shift of 64-fold for α mβδm receptors was predicated on the assumption that each additional mutant subunit within the receptor complex acts independently. Although this holds for receptors with two mutant subunits, it is conceivable that with three such substitutions, some degree of interaction between adjacent mutant subunits might cause deviations from the predicted curve shifts.
cDNA transfection ratio has no effect on α4β3δ receptor stoichiometry Recently, using an imaging approach based on inserting bungarotoxin binding sites into GABAA receptor subunits that could be subsequently labelled by fluorescent bungarotoxin, it was demonstrated that the number of δ subunits incorporated into the α4β2δ receptor complex could vary with the cDNA transfection ratio (Wagoner and Czajkowski, 2010). We therefore investigated this possibility using our functional approach with HEK cells transfected with one of the following three commonly used α : β : δ cDNA ratios -1:1:1, 1:1:10 or 10:1:10 (Borghese et al., 2006;Stórustovu and Ebert, 2006;Barrera et al., 2008;Hoestgaard-Jensen et al., 2010).
Overall, altering the transfection ratio had no significant effect on the relative GABA EC50 shifts between δand δm-expressing cells ( Figure 4B), suggesting that at least for these three transfection ratios, the number of δ subunits incorporated into α4β3δ receptors remains relatively constant.

Co-expressing WT and mutant subunits confirms α4β3δ receptor stoichiometry
Our deductions so far are based on the assumption that each subunit mutation has an equivalent effect on GABA potency, regardless of the subunit in which the L9′S mutation is inserted. Although this holds for our α4β3δ receptors, some deviation may occur when the number of mutant subunits per receptor is increased, as noted for α1β2γ2S GABAA receptors (Chang et al., 1996) and heteromeric nAChRs (Labarca et al., 1995). To overcome this methodological limitation, Chang et al. (1996) proposed an alternative approach for deducing subunit stoichiometry that does not rely on the relative EC 50 shifts induced by different classes of mutant subunits, but instead relied on co-expressing WT and L9′S mutants to generate multiple populations of receptors expressed in the same cells (Chang et al., 1996). In principle, the co-expression of WT subunits with their respective L9′S mutants (e.g. α and αm) should introduce discrete and discernible components into the GABA doseresponse curve of expressing cells. For example, assuming there are two α subunits per receptor pentamer, these components would represent distinct GABAA receptors of ααβδ, αmαmβδ and ααmβδ and its equivalent, αmαβδ. Thus, the GABA sensitivity that is attributable to an individual receptor population would be observed by an inflection in the doseresponse curve. This analysis was used to infer the subunit stoichiometry for α1β2γ2S GABAA receptors (Chang et al., 1996).
Taking a similar approach for α4β3δ receptors, we generated GABA dose-response curves for cells co-transfected with β, δ and equal amounts of α and αm cDNAs. The ααmβδ GABA dose-response curve exhibited three discernible components, which were described by the sum of three Hill equations ( Figure 5A). The first and third components accounted for 34.0 ± 3.3 and 17.5 ± 2.7% of the total receptor population, with GABA EC50 values of 0.023 ± 0.005 and 5.72 ± 0.90 μM respectively (Table 2). These EC50 values are in close proximity to those for αmβδ (0.46 ± 0.11 μM) and αβδ (1.9 ± 0.47 μM) receptors, suggesting these two components to the doseresponse curves of ααmβδ ( Figure 5A) are attributable to αmβδ and αβδ receptors. Moreover, the appearance of an intermediary component with an EC50 of 0.30 ± 0.03 μM (48.7 ± 5.6%) suggested the expression of receptor pentamers containing one WT and one mutant subunit (i.e. ααmβδ or αmαβδ). Similar to ααmβδ-expressing cells, the GABA doseresponse curves of αββmδ-expressing cells revealed three distinct components ( Figure 5B). The first component (24.7 ± 3.5%) had an EC50 of 0.03 ± 0.01 μM, corresponding to the αβmδ receptor population. Approximately 31.5 ± 2.6% of receptors exhibited an EC50 of 4.68 ± 0.74 μM attributable to αβδ receptors. Again, the appearance of an intermediary component with an EC50 of 0.31 ± 0.05 μM (43.9 ± 3.5%) was indicative of a third receptor population containing both β subtypes (i.e. β and βm).
By contrast, the mean GABA dose-response curve for αβδδm receptors did not exhibit obvious multiple components ( Figure 5C). However, detailed analysis of individual doseresponse curves revealed that for most cells sampled (7/10), two components were discerned ( Figure 5C inset). For those cells exhibiting two components, the majority of receptors (75.8 ± 3.0%) exhibited a GABA EC50 of 0.21 ± 0.01 μM, whereas 24.2 ± 2.8% of receptors exhibited an EC50 of 2.14 ± 0.48 μM. The GABA sensitivities of these two components are similar to the observed EC50s for αβδm and αβδ receptors respectively (Table 2). Although the remaining cells (3/10) did not overtly display multiple components, their GABA sensitivities were intermediary to those of αβδand αβδmexpressing cells. The absence of a third component suggested that αβδδm-expressing cells exhibit only two receptor populations, αβδ and αβδm, and thus it follows that each receptor complex is most likely to contain only one δ subunit.
Collectively, these data demonstrate that αβδ receptors are most likely to contain two αs, two βs and one δ subunit.

Discussion
Although the stoichiometry of synaptic α1β2γ2 subunitcontaining GABAA receptors has consensus support for 2α:2β:1γ (Backus et al., 1993;Chang et al., 1996;Tretter et al., 1997), the stoichiometry of extrasynaptic δ-containing receptors remains unclear and potentially variable with a dependence on experimental conditions. This may reflect a difference in the co-assembly properties of δ , with different α and β subunits Kaur et al., 2009;Wagoner and Czajkowski, 2010). We used an alternative approach to probe α4β3δ stoichiometry by introducing a wellcharacterized 9′ leucine-to-serine mutation into the M2 domains of α4, β3 and δ subunits. Each polar substitution increased the GABA sensitivity of mutant subunit-containing receptors (by approximately fourfold) in relative proportion with the number of mutant subunits assembled in the receptor. This, in conjunction with data derived from cells co-expressing mutant and respective WT subunits, revealed a relatively consistent subunit stoichiometry, by these methods, of 2α, 2β and 1δ.

Assumptions and limitations
Our deductions regarding α4β3δ GABAA receptor stoichiometry are predicated on the assumption that the L9′S mutations do not perturb the 'normal' subunit stoichiometry of these receptors. Because N-terminal motifs have been established as the key determinants of GABAA receptor subunit assembly (Connolly et al., 1996;Taylor et al., 1999;Klausberger et al., 2001), it seemed unlikely that a point mutation within the ion channel-lining M2 region would alter receptor subunit stoichiometry. However, it is intriguing that for most αβδδm-expressing cells, the component that was attributable to αβδm receptors was larger than that for αβδ receptors (∼75 and 24%, respectively), suggesting that δm might be more efficiently incorporated into functional receptors than δ.
Given the M2 location of the point mutation, a more likely explanation for the disproportionate percentage components is that the mutation may affect the gating kinetics of the receptor. For nAChRs (Filatov and White, 1995) and GABAA α1β3γ2L receptors (Bianchi and Macdonald, 2001), it has been demonstrated that 9′ mutant-containing receptors can exhibit altered single-channel conductances and/or open probabilities. This could cause the apparent percentage components of αβδ and αβδm to vary (Chang et al., 1996). Nevertheless, because our conclusions rely on the number of observable components in the dose-response curves and not on the relative contribution of each individual component, our conclusion that α4β3δ receptors contain only one δ subunit still remains valid.

Comparison with previous studies
To date, only two studies have investigated the subunit stoichiometry of unconstrained recombinant α4β2/3δ receptors. Although atomic force microscopy has revealed a subunit stoichiometry of 2α4:2β3:1δ (Barrera et al., 2008), the immunopurification of cell surface α4β2δ receptors suggested that by increasing the relative abundance of δ, more than one δ can be incorporated into the receptor complex (Wagoner and

Figure 5
Co-expression of WT and mutant L9′S α4, β3 and δ subunits. GABA concentration-response curves (black continuous lines) for (A) ααmβδ (n = 6), (B) αββmδ (n = 6) and (C) αβδδm (n = 10), fitted using a two or three component Hill equation (see Methods section). Also shown for each receptor class are the Hill equation fits for the GABA concentration-response curves of WT αβδ receptors (blue dashed lines) and their respective L9′S mutants (red dashed lines): αmβδ (A), αβmδ (B) and αβδm (C). For αβδδm, the GABA dose-response curves of 7/10 cells exhibited two discernible components. The inset is an example of GABA dose-response curve from a cell exhibiting two such components. The total amount of cDNA used for each transfection was 4 μg, and equal amounts of WT and their respective L9′S mutant cDNA were used. Data points are shown as mean ± SEM.
Czajkowski, 2010). Moreover, a study using receptor expression in oocytes reported that increasing relative amounts of δ cRNA increased the GABA EC50 and decreased the Hill slopes for α4β3δ GABA dose-response curves (You and Dunn, 2007).
Although a change in stoichiometry may account for altered receptor function, it is also equally plausible, from studies using concatemers and thus constrained subunit positions, that δ subunits may assume various locations within a functional receptor pentamer, and also potentially contribute to an agonist-binding site Sigel et al., 2009). Thus the previously described effects on receptor function may also have arisen from the rearrangement of subunits within the receptor. Indeed, for concatemeric α1β3δ receptors, those with an βαβαδ (anticlockwise) subunit arrangement appear to be ∼26-fold less sensitive to GABA than receptors with the βαβδα (anticlockwise) subunit arrangement , demonstrating the functional importance of subunit location within a receptor pentamer.
Our data indicate that, at least for three commonly used α : β : δ transfection ratios 1:1:1, 1:1:10 or 10:1:10 (Borghese et al., 2006;Stórustovu and Ebert, 2006;Barrera et al., 2008;Hoestgaard-Jensen et al., 2010), the number of incorporated δ subunits seemingly remains fixed at one. Moreover, we found no significant effect of altering cDNA transfection ratio on α4β3δ receptor function. In accordance with our findings, another oocyte study had demonstrated no significant effect of altering cRNA transfection ratio on the sensitivity of WT α4β3δ receptors to GABA or Zn 2+ (Borghese and Harris, 2007).
Although the discrepancy between our observations and those previously reported remain unclear, one difference may be the use of different expression systems. Alternatively, the use of different β isoforms may also give rise to these discrepancies. Given that β2 and β3 subunits have been demonstrated to have distinctive assembly properties (Taylor et al., 1999), this might have important implications for their oligomerization with δ subunits.

Subunit positioning
Although we demonstrate a stoichiometry of 2α:2β:1δ for α4β3δ receptors, our data give little indication of subunit arrangement, which could be an important determinant of αβδ receptor function Kaur et al., 2009). The subunit positional arrangement of α1βγ2 receptors is widely accepted to be βαβαγ (anticlockwise) (Baumann et al., 2001;Baur et al., 2006;Smart and Paoletti, 2012). Given the conflicting evidence regarding the number of incorporated δ subunits, it is unsurprising that the subunit arrangement of recombinant αβδ remains undefined. For α4β3δ receptors with a stoichiometry of 2α:2β:1δ, structural microscopic analysis has revealed a predominant βαβαδ anticlockwise arrangement (Barrera et al., 2008), suggesting δ can assume the position of the γ2 subunit in an αβγ receptor. However, in the same study, a minority of receptors (∼21%) were found to have an alternative βαβδα subunit arrangement, indicating more than one arrangement may be possible (Barrera et al., 2008). Indeed it has been recently demonstrated that δ can assume multiple positions when constrained within αβδ concatemers Kaur et al., 2009). Intriguingly, concatemeric α4β2δ receptors with the βαβαδ conformation (Shu et al., 2012) form functional receptors with similar pharmacological profiles to unconstrained recombinant α4β2δ receptors , whereas α1β3δ receptors formed from the alternative βαβδα anticlockwise arrangement exhibit similar GABA and Zn 2+ sensitivities to non-concatenated receptors . These findings suggest that the arrangement of recombinant and native δ-containing receptors is still open to question.
To conclude, we demonstrate that the subunit stoichiometry of heterologously expressed α4β3δ receptors is 2α:2β:1δ. This stoichiometry remains unchanged even with varying cDNA transfection ratios, which may reflect that this is the preferred, dominant subunit assembly for this important extrasynaptic GABAA receptor subtype that underlies tonic inhibition in some areas of the brain.