Context-Dependent Modulation of GABAAR-Mediated Tonic Currents

Tonic GABA currents mediated by high-affinity extrasynaptic GABAA receptors, are increasingly recognized as important regulators of cell and neuronal network excitability. Dysfunctional GABAA receptor signaling that results in modified tonic GABA currents is associated with a number of neurological disorders. Consequently, developing compounds to selectively modulate the activity of extrasynaptic GABAA receptors underlying tonic inhibition is likely to prove therapeutically useful. Here, we examine the GABAA receptor subtype selectivity of the weak partial agonist, 5-(4-piperidyl)isoxazol-3-ol (4-PIOL), as a potential mechanism for modulating extrasynaptic GABAA receptor-mediated tonic currents. By using recombinant GABAA receptors expressed in HEK293 cells, and native GABAA receptors of cerebellar granule cells, hippocampal neurons, and thalamic relay neurons, 4-PIOL evidently displayed differential agonist and antagonist-type profiles, depending on the extrasynaptic GABAA receptor isoforms targeted. For neurons, this resulted in differential modulation of GABA tonic currents, depending on the cell type studied, their respective GABAA receptor subunit compositions, and critically, on the ambient GABA levels. Unexpectedly, 4-PIOL revealed a significant population of relatively low-affinity γ2 subunit-containing GABAA receptors in the thalamus, which can contribute to tonic inhibition under specific conditions when GABA levels are raised. Together, these data indicate that partial agonists, such as 4-PIOL, may be useful for modulating GABAA receptor-mediated tonic currents, but the direction and extent of this modulation is strongly dependent on relative expression levels of different extrasynaptic GABAA receptor subtypes, and on the ambient GABA levels. SIGNIFICANCE STATEMENT A background level of inhibition (tonic) is important in the brain for controlling neuronal excitability. Increased levels of tonic inhibition are associated with some neurological disorders but there are no specific ligands capable of selectively reducing tonic inhibition. Here we explore the use of a GABA partial agonist as a selective chemical tool in three different brain regions. We discover that the activity of a partial agonist is heavily dependent upon the GABAA receptor subunit composition underpinning tonic inhibition, and on the ambient levels of GABA in the brain.

As an alternative to GABA A receptor antagonists, we hypothesized that low-efficacy partial agonists may act as "functional antagonists", given their ability to compete with GABA for the orthosteric binding site, and their reduced ability to activate GABA A receptors. Moreover, low-efficacy partial agonists may be less likely to induce convulsions, or unwanted side effects . Here the subtype-selective profile of the low-efficacy partial agonist, 5-(4-piperidyl)isoxazol-3-ol (4-PIOL; Kristiansen et al., 1991;Frølund et al., 1995;Mortensen et al., 2002Mortensen et al., , 2004 was assessed on recombinant and neuronal GABA A receptors. The activity profile of 4-PIOL on tonic and phasic currents varied between three selected brain regions. This revealed a strong dependence on GABA A receptor subunit composition, and on ambient GABA levels, and helped uncover a population of largely silent GABA A receptors in the thalamus that can contribute to tonic inhibition under specific conditions. Overall, partial agonists may be useful as therapeutic agents, but their effectiveness will critically depend on which GABA A receptor isoforms are present and the extent of their activation.

Materials and Methods
Transient receptor expression in HEK293 cells. Human embryonic kidney (HEK) 293 cells were cultured in DMEM, as previously described (Wooltorton et al., 1997). HEK cells were plated onto poly-L-lysine-coated coverslips and transfected using a calcium phosphate protocol. Briefly, GABA A receptor pRK5 cDNAs, with enhanced green fluorescent protein (eGFP) cDNA, were mixed with 340 mM CaCl 2 , and an equal volume of HEPES-buffered saline (50 mM HEPES, 280 mM NaCl and 2.8 mM Na 2 HPO 4 , pH 7.2). One microgram of each cDNA was used, and a total of 4 g cDNA was used for each transfection. The cDNA-calcium phosphate suspension was applied to cells, which were incubated overnight, and used for electrophysiology 18 -48 h after transfection.
Drug solutions. For HEK293 cells, GABA was applied alone, or in combination with other drugs using a Y-tube application system (Mortensen and Smart, 2007). The 10 -90% solution exchange times of the application system were within 20 -30 ms as measured in open pipette tip recordings. GABA and bicuculline were obtained from Sigma-Aldrich, THIP and DS2 were purchased from Tocris Biosciences, and diazepam was sourced from Roche. Drug solutions were either prepared from water or DMSO concentrated stocks, or dissolved directly into the extracellular medium, depending on the final concentration. Drug solutions were corrected for pH before use.
Data acquisition. Whole-cell currents were recorded using an Axopatch 200B amplifier. Currents were filtered at 5 kHz, digitized at 50 kHz via a Digidata 1332A (Molecular Devices), and recorded to disk (Dell Pentium Dual Core-Optiplex 960). Series resistances were monitored throughout each experiment and deviations Ͼ20% resulted in the data being excluded from further analysis.
Analysis of GABA concentration-response curves. GABA-activated currents (I GABA ) were normalized to the maximal current evoked by a saturating concentration of GABA (I Max,GABA ). The normalized concentration-response curves were fitted using a modified Hill equation (Eq. 1), using a least-squares method.
EC 50 is the concentration of GABA, [GABA], which produced 50% of the maximal response, and n H , is the Hill coefficient. I Min,GABA is the minimum "plateau" response induced in the presence of 4-PIOL. For the GABA concentration-response curve constructed in the absence of 4-PIOL, I Min,GABA is zero. The parameters obtained from individual curve-fittings were collated and the data were expressed as the mean Ϯ SEM.
Neuronal whole-cell electrophysiology. Thalamic relay neurons of the dorsal lateral geniculate nucleus (dLGN), CA1 pyramidal neurons in the hippocampus, and CGCs were visualized using infrared differential interference contrast optics and a Basler scA750 -60fm camera. Cells were perfused with aCSF or Krebs, supplemented with 2 mM kynurenic acid, or a mixture of 20 M D-AP5 (Tocris Bioscience) and 10 M CNQX (Abcam) to block glutamatergic currents. Recordings were made using patch pipettes (2-4 M⍀) filled with a solution containing the following (in mM): 140 CsCl, 2 NaCl, 10 HEPES, 5 EGTA, 2 MgCl 2 , 0.5 CaCl 2 , 2 Na-ATP, and 2 QX-314 bromide. pH was adjusted using 1 M CsOH. In a subset of recordings, thalamic slices were preincubated for 30 min, in aCSF containing the GABA transporter (GAT) inhibitors, NNC-711 (10 M; Abcam) and SNAP-5114 (20 M; Tocris Bioscience). Both compounds were also present throughout the recordings. A saturating concentration of bicuculline (20 M; Ueno et al., 1997) was bath-applied at the end of all electrophysiological recordings, to confirm that all synaptic events were GABAergic, and to reveal GABA-mediated tonic currents. All recordings were performed at room temperature.
For tonic currents, the average holding current for a 30 s epoch in each drug condition was measured using WinEDR software (v3.1; John Dempster, University of Strathclyde, Glasgow, UK). Changes in holding current were calculated by subtracting the average holding current after drug application, from the average holding current before drug application. In addition, the root mean square (rms) baseline noise was measured over a 30 s epoch, sampled every 100 ms. Because sIPSCs increase rms baseline noise, Microsoft Excel was used to calculate a threshold for eliminating contaminated 100 ms epochs. A running threshold (routinely the median) was calculated at 5 s time intervals, over a 30 s recording period, and any rms value greater than the calculated threshold, was automatically excluded from further analysis. Effective thresholding was validated by manually analyzing a small section of each recording (ϳ10 s), and manually eliminating 100 ms epochs contaminated by synaptic currents.
Synaptic current analysis. The sIPSC frequency was determined using MiniAnalysis software (Synaptosoft). During 4-PIOL application, a significant increase in rms current noise was observed that might mask smaller sIPSCs, thereby introducing a bias toward larger events, compared with control. To limit the bias, only the largest hundred amplitude events from each condition were compared. The average decay kinetics for sIPSCs for each cell was determined by fitting uncontaminated events (Ͼ50 events for each condition) with either a mono-or bi-exponential decay function. To combine data obtained for mono-or bi-exponentially fitted events, decay times were transformed to a weighted decay time, w , according to the following: where 1 and 2 represent decay time constants, and A1 and A2 are the relative amplitude contributions of 1 and 2 . For mono-exponential decaying events, A2 and 2 are zero. The mean sIPSC frequency, amplitude, 10 -90% rise time, and w were calculated for each cell.

Functional properties of synaptic-type GABA A receptors activated by a partial agonist
To evaluate the functional profile of 4-PIOL at synaptic-type GABA A receptors, peak whole-cell GABA currents were recorded from ␣1␤3␥2-expressing HEK293 cells, in the absence or presence of 10, 100, or 1000 M 4-PIOL ( Fig. 1A). At these concentrations, 4-PIOL induced a rightward shift in the GABA concentration-response curve, with a discernible crossover with the control GABA curve (Fig. 1B)   ␣1␤3␥2-expressing cells revealed a small agonist response, particularly with 100 or 1000 M 4-PIOL (Fig. 1A). This agonist activity appeared on the concentration-response curves by an elevated minimum response ( Fig. 1 B, C) with 100 and 1000 M 4-PIOL inducing agonist currents that were 5.5 Ϯ 3.2% and 6.8 Ϯ 1.3% of the maximum GABA response (Fig.  1C). Crucially for this synaptic GABA A receptor subtype, neither 10 nor 100 M 4-PIOL inhibited the maximum responses to higher, synaptic concentrations of GABA (percentage control: 98.8 Ϯ 1.2% and 98.7 Ϯ 1.3% for 10 and 100 M 4-PIOL at 1 mM GABA; Fig. 1D), indicating that at these concentrations of 4-PIOL, the partial agonist is potentially capable of  All data are presented as mean Ϯ SEM (n ϭ 5-6). Control and 4-PIOL data were compared using a paired t test.
inhibiting responses to predicted extrasynaptic concentrations of GABA, without depressing synaptic GABA currents.
Overall, these data indicate that 10 M 4-PIOL may potently inhibit ␣4␤2␦-mediated tonic currents, without affecting ␣1␤3␥2-mediated phasic currents. Moreover, 10 M 4-PIOL is expected to produce only a modest effect on ␣6␤2␦-mediated tonic currents (eg, in the cerebellum), whereas any inhibition of extrasynaptic ␥2-containing receptors, although minimal, will be strongly dependent on the ambient GABA concentration in neuronal preparations. Although our earlier studies suggested that ␣␤ receptors represent only a modest component of the extrasynaptic GABA A receptor population in hippocampal pyramidal cells (ϳ10%; Mortensen and Smart, 2006), their contribution to the tonic current will be inhibited by 10 M 4-PIOL.
Examining GABA A receptor-mediated tonic and phasic inhibition with a partial agonist As our observations indicated that 4-PIOL may have differential effects on synaptic and extrasynaptic GABA A receptors, we chose to explore the functional profile of 4-PIOL on tonic and phasic GABA currents in neurons. Initially, whole-cell recordings were    (Table 1), respectively. A saturating concentration of bicuculline (20 M) was applied at the end of the recordings to confirm that all synaptic events were GABAergic, and to measure the amplitudes of GABA A receptor-mediated tonic currents. As expected, for both cell types, bicuculline abolished the sIPSCs and induced outward shifts in the membrane holding current, indicative of GABA tonic currents ( Fig. 3 A, C). For CGCs and hippocampal neurons, the basal tonic current amplitudes were as follows: 22.8 Ϯ 8.5 and 42.8 Ϯ 6.3 pA ( Fig. 3E; Table 2).
For both types of cultured neurons, we used the ␦-subunit selective agonist, THIP, to assess the presence of ␦-containing GABA A receptors. Application of 1 M THIP evoked significant inward currents, confirming the surface expression of extrasynaptic ␦-GABA A receptors in the cultured cell preparations  To characterize the effects of 4-PIOL under more physiological conditions, we recorded from CGCs, hippocampal CA1 pyramidal neurons and thalamic relay neurons of the dorsal lateral geniculate nucleus in acute brain slices. These three cell types were selected because their GABA A receptor-mediated tonic currents are thought to be mediated largely by ␣6␤␦, ␣5␤␥, and ␣4␤␦ GABA A receptors, respectively (Jones et al., 1997;Brickley et al., 2001;Caraiscos et al., 2004;Cope et al., 2005;Glykys et al., 2008).
Applying 4-PIOL to dLGN slices significantly enhanced tonic currents by 57.4 Ϯ 3.3 pA (Fig. 4 F, H ), and notably reduced both the frequency (percentage control: 13.3 Ϯ 2.2%; p ϭ 0.001) and amplitude of sIPSCs (percentage control: 71.8 Ϯ 2.7%; p ϭ 0.01; Table 3), relative to synaptic events measured in control aCSF.  Because of the low frequency and amplitude of sIPSCs in the presence of 4-PIOL, and the increase in rms baseline noise induced by 4-PIOL (13.2 Ϯ 1.6 pA), no detailed analysis of sIPSC decay or rise times was performed for dLGN relay neurons.

Probing the identity of GABA A receptors in dLGN relay neurons
According to our recombinant expression studies, 10 M 4-PIOL showed no discernible agonist activity at ␣4␤␦ receptors, and was predicted to reduce GABA-mediated tonic currents at these receptors, assuming that the ambient GABA concentration in slices is ϳ 0.1-1 M GABA (Fig. 2E). Because tonic currents in dLGN relay neurons are thought to be mediated by ␣4␤␦ receptors (Cope et al., 2005;Bright et al., 2007;Nani et al., 2013;Ye et al., 2013), the finding that 4-PIOL showed agonist behavior, by enhancing tonic currents in dLGN relay neurons, was unexpected. To verify the GABAergic origin of this current, we used bicuculline. Current responses to 10 M 4-PIOL were abolished in the presence of coapplied bicuculline (percentage control 4-PIOL response: 2.2 Ϯ 0.5; Fig. 5A), indicating that 4-PIOL was exclusively activating GABA A receptors to induce an inward current.
If 4-PIOL is acting on ␦-containing receptors in dLGN relay neurons, we might also expect 4-PIOL (10 M) to reduce the steady-state THIP (1 M) current in dLGN relay neurons. As expected, THIP significantly enhanced the dLGN tonic current by 96.6 Ϯ 12.6 pA, confirming the functional expression of ␦ subunit-containing receptors. However, coapplication of 4-PIOL with THIP generated a further inward current (58.6 Ϯ 6.6 pA; Fig. 5C), with a mean magnitude that was similar to the control 4-PIOL current (57.4 Ϯ 3.3 pA; p ϭ 0.67). These data indicate that THIP and 4-PIOL may not be competing for the same ␦ subunit-containing receptors in dLGN relay neurons.
Together, these data indicate that although ␦ subunitcontaining receptors are expressed in dLGN relay neurons, as confirmed by THIP and DS2 modulation of basal tonic currents, the 4-PIOL current appears to be largely mediated by ␥2containing receptors, with little, or no, contribution from ␦ subunit-containing receptors.
Given these findings, we explored which GABA A receptor isoforms underpin tonic currents in dLGN relay neurons. Since tonic currents in these cells are thought to be mediated by ␦ subunit-containing receptors (Cope et al., 2005;Bright et al., 2007;Nani et al., 2013;Ye et al., 2013), we investigated whether there was a correlation between currents induced by a ␦-selective concentration of THIP (1 M) and by bicuculline, for individual dLGN relay neurons. 4-PIOL and bicuculline were individually applied (Fig. 8A), and to account for variation in cell size, the holding currents were normalized to whole-cell capacitance (pF). A scatter plot comparing current densities revealed a positive correlation ( Fig. 8B; r ϭ 0.61; p ϭ 0.02). Thus, cells with a larger THIP-induced current also displayed larger GABA A receptormediated tonic currents. These data indicate that a higher expression of ␦ subunit-containing receptors may underlie the larger tonic currents, although other factors, such as the ambient GABA concentration, will also be important.
If, as our previous data indicates, THIP and 4-PIOL are acting at potentially different GABA A receptors (Fig. 5), no positive correlation would be expected between the THIP and 4-PIOL responses in dLGN relay neurons. Indeed, a scatterplot of THIP and 4-PIOL currents revealed no significant correlation (Pearson's correlation coefficient, r ϭ Ϫ0.31; p ϭ 0.28), supporting the notion that 4-PIOL was activating a distinct receptor population from the ␦-containing receptors activated by THIP (Fig. 8C).
Because 4-PIOL and THIP are likely to be acting on different GABA A receptors in dLGN relay neurons, it was intriguing to explore whether the ␥2-containing receptors that mediate the 4-PIOL current, also contribute to dLGN tonic currents. A scatterplot comparing 4-PIOL and bicuculline currents recorded from individual dLGN relay neurons showed poor correlation ( Fig. 8D; r ϭ 0.32; p ϭ 0.17), indicating that the receptors that mediate the 4-PIOL current, are unlikely to contribute substantially to basal GABA A receptor-mediated tonic currents in dLGN relay neurons.

Modulation of tonic currents depends on ambient GABA levels
For recombinant ␣1␤3␥2 receptors, 4-PIOL enhanced the steady-state GABA current when the GABA concentration was low (ϳ0.1 M GABA), but produced a small inhibition when the ambient GABA level was raised to 1 M (Fig. 2D). In dLGN relay neurons, the robust 4-PIOL enhancement of baseline tonic currents (mainly via ␥2 subunit-containing receptors) indicates that the ambient GABA levels in this slice may be low (Ͻ1 M).
To determine whether 4-PIOL could switch from acting as an agonist (at low ambient GABA levels), to acting as an antagonist (at higher ambient GABA levels) in the native environment of dLGN relay neurons, GABA levels were raised in slices by inhibiting GABA uptake. Because GABA uptake in the thalamus is mediated by the GABA transporters, GAT1 and GAT3 (De Biasi et al., 1998), slices were preincubated (for 30 min) and subsequently recorded in aCSF supplemented with the GAT1 inhibitor, 10 M NNC-711 (Borden et al., 1994) and the GAT2/3 inhibitor, 20 M SNAP-5114 (Borden, 1996). Following a period of control recording (in the presence of the GAT inhibitors), 10 M 4-PIOL was applied to dLGN relay neurons and subsequently washed out (Fig. 9A), before application of bicuculline to measure the GABA-mediated tonic current. As expected, the tonic current was significantly larger in the presence of the GAT blockers compared with control aCSF (132 Ϯ 19.5 and 23.8 Ϯ 2.1 pA, respectively; Fig. 9B), even when these currents were normalized to cell capacitance (0.7 Ϯ 0.1 pA/pF and 0.1 Ϯ 0.01 pA/pF, respectively; p Ͻ 0.0001). These data are consistent with elevated ambient GABA levels, in GAT-blocked slices, increasing the activation of extrasynaptic GABA A receptors.
Coapplication of 10 M 4-PIOL with GAT inhibitors enhanced the tonic current by 53.0 Ϯ 10.1 pA (Fig. 9A,C). This 4-PIOLinduced current was similar to that observed in control aCSF (57.4 Ϯ 3.3 pA; Fig. 9C; p ϭ 0.47), and indicates that under control condi- tions, 4-PIOL was not a substrate for GAT1-3. Although the increased tonic current indicated that the ambient GABA level in the slice was raised, it may still be too low to alter the response profile for 4-PIOL. To further increase and normalize ambient GABA levels, 1 and 3 M GABA were individually applied to GAT-inhibited slices, followed by coapplication with 10 M 4-PIOL (Fig. 9A,C). Both 1 and 3 M GABA enhanced the baseline tonic current by 139 Ϯ 21.8 and 301 Ϯ 35.2 pA, respectively (Fig. 9C). Coapplication of 10 M 4-PIOL with 1 M GABA also elicited an inward current (Fig. 9A), although the resultant 4-PIOL current was significantly smaller than that observed in control aCSF (29.6 Ϯ 11.5 pA; Fig. 9C). By contrast, coapplying 10 M 4-PIOL with 3 M GABA produced an outward current (27.8 Ϯ 9.5.1 pA; Fig. 9A). Thus, as observed for recombinant ␣1␤3␥2 receptors, 4-PIOL exhibited a dominant agonist profile at low GABA concentrations (Յ1 M), but produced a small inhibition of dLGN tonic currents when the ambient GABA concentration was increased.
To determine whether ambient GABA levels also influence the 4-PIOL current in cultured CGCs and hippocampal neurons, low concentrations of GABA (0.3 and 1 M) were preapplied until steady-state currents were achieved and subsequently, 4-PIOL was coapplied (Fig. 10 A, B). Similar to its effects on endogenous CGC tonic currents, coapplication of 4-PIOL produced no significant shift in the holding current, even when the preapplied GABA concentration was raised to 1 M (Fig. 10A).

Discussion
GABA A receptor-mediated tonic inhibition is an important regulator of cell and network excitability (Mann and Mody, 2010), and its dysfunction is associated with several pathophysiological states Brickley and Mody, 2012). Selectively modulating the activity of extrasynaptic GABA A receptors may therefore be therapeutically useful for the treatment of such disorders. Given the absence of suitable subtype-selective antagonists, we took a different approach by investigating the selectivity profile of the weak partial agonist, 4-PIOL (Mortensen et al., 2002(Mortensen et al., , 2004, to determine whether it could be used as a selective modulator of GABA-mediated tonic currents.

Partial agonist modulation of GABA synaptic currents
Our receptor expression studies predicted that at high GABA (synaptic) concentrations, 4-PIOL should not affect ␣1␤3␥2mediated currents. However, 4-PIOL variably reduced sIPSC amplitudes and frequencies in many of our neuronal preparations. Using TTX in relay neurons revealed that 4-PIOL reduced mIPSCs to a lesser extent than sIPSCs, which is in accord with both presynaptic and postsynaptic effects. The reduced sIPSC frequency is indicative of a presynaptic action, with 4-PIOL activating extrasynaptic ␥2-GABA A receptors, reducing interneuron excitability and lowering GABA release (Axmacher and Draguhn, 2004). This is supported by the bidirectional 4-PIOL responses of CA1 neurons. Cells with inward 4-PIOL currents show reduced sIPSC frequencies, consistent with inhibition of presynaptic in- (1 or 3 M) in the presence of GAT inhibitors. Slices were incubated in aCSF supplemented with GAT inhibitors for at least 30 min before electrophysiological recordings. All recordings were performed at room temperature. B, Bar graph of BIC current in control aCSF (black; n ϭ 30), or in the presence of GAT inhibitors (white; n ϭ 25). C, Bar graph of the mean 4-PIOL current measured in aCSF (black; n ϭ 42), ϩ GAT inhibitors (n ϭ 8), or ϩ GAT inhibitors with 1 M (n ϭ 11) or 3 M GABA (n ϭ 4). Data are presented as mean Ϯ SEM. **p Ͻ 0.01, ***p Ͻ 0.001, unpaired t tests.
␦-containing counterparts, and so may also be potentiating the 4-PIOL current at ␥2-GABA A receptors in relay neurons.
Overall, the simplest explanation for these data are that functional effects of 4-PIOL on tonic currents in dLGN relay and hippocampal neurons, are largely dominated by its actions on ␥2 subunit-containing receptors, though we cannot completely exclude a contribution from ␦ subunit-containing receptors. Although the significant presence of extrasynaptic ␥2 subunitcontaining receptors on dLGN relay neurons was unexpected, given that they are thought to accumulate at synaptic sites, immunohistochemical and functional studies indicate that a significant number of ␣1-␣3 subunits, which associate with ␥2 subunits, may also exist at extrasynaptic sites (Soltesz et al., 1990;Nusser et al., 1998;Mangan et al., 2005;Kasugai et al., 2010). Although their functional significance remains to be established in native systems, a recent study has proposed that tonically active ␣1␤3␥2 receptors might contribute to the clinical actions of positive allosteric modulators, such as etomidate and propofol (Li and Akk, 2015). Thus, the relative expression levels of extrasynaptic ␦and ␥2-containing receptors may be important in determining the functional effects of compounds, such as 4-PIOL, which can modulate both receptor isoforms under distinct conditions.

Low ambient GABA levels in neuronal preparations
Under our experimental conditions, the ambient GABA concentration in all three neuronal preparations was estimated to be significantly Ͻ1 M. In accord with these findings, although difficult to measure precisely in the structurally tortuous environment of the brain, microdialysis studies estimate that extracellular GABA concentrations in vivo range from 30 nM to 2.9 M (Glaeser and Hare, 1975;Lerma et al., 1986;de Groote and Linthorst, 2007;Wlodarczyk et al., 2013), whereas the activity of GABA transporters predicts that ambient GABA levels are within 0. 1-0.4 M (Attwell et al., 1993;Richerson and Wu, 2003;Wu et al., 2007).
Given the low ambient GABA levels that we estimate in slices, and the low GABA sensitivity of ␥2 subunit-containing GABA A receptors (Brown et al., 2002;Mortensen et al., 2010Mortensen et al., , 2011, it is unsurprising that the extrasynaptic population of ␥2-containing receptors, detected in dLGN relay neurons, did not significantly contribute to basal tonic currents under our experimental conditions. However, this does not discount the possibility that ␥2containing receptors may contribute to tonic currents when ambient GABA levels are significantly increased, for instance, during behavioral or pathophysiological disease states. Although applying diazepam to dLGN relay neurons enhanced dLGN tonic currents, demonstrating that ␥2-containing receptors can contribute to tonic currents, it is difficult to discount the possibility that diazepam may have increased the apparent affinity of these receptors for GABA (Gielen et al., 2012), thus recruiting a population of extrasynaptic ␥2-containing receptors that may be inactive under control conditions. The observation that ambient GABA levels strongly influence the functional profile of 4-PIOL is unsurprising given that both 4-PIOL and GABA act via the same binding site. Indeed, similar observations have been made for the agonist, THIP, whose enhancement of ␦-mediated tonic currents in CGCs was attenuated at higher ambient GABA concentrations (Houston et al., 2012). Thus, when evaluating the potential effects of therapeutic compounds on tonic currents, an important consideration is how this modulation will be affected by variable ambient GABA concentrations.

Therapeutic potential of low-efficacy agonists
Overall, the therapeutic potential of a low-efficacy partial agonist, such as 4-PIOL, has merit, but its action will critically depend on a number of factors, including which GABA A receptor isoforms are expressed, their expression levels on the cell surface, and ambient GABA levels. Given its enhancement of tonic currents at low ambient GABA concentrations, a weak partial agonist like 4-PIOL might be most useful for neurological conditions where an increased tonic inhibition is desirable, for example in Fragile X syndrome and sleep disorders (Brickley and Mody, 2012;Whissell et al., 2015).