Distinct roles of synaptic and extrasynaptic GABAAreceptors in striatal inhibition dynamics

Striatonigral and striatopallidal projecting medium spiny neurons (MSNs) express dopamine D1 (D1+) and D2 receptors (D2+), respectively. Both classes receive extensive GABAergic input via expression of synaptic, perisynaptic, and extrasynaptic GABAA receptors. The activation patterns of different presynaptic GABAergic neurons produce transient and sustained GABAA receptor-mediated conductance that fulfill distinct physiological roles. We performed single and dual whole cell recordings from striatal neurons in mice expressing fluorescent proteins in interneurons and MSNs. We report specific inhibitory dynamics produced by distinct activation patterns of presynaptic GABAergic neurons as source of synaptic, perisynaptic, and extrasynaptic inhibition. Synaptic GABAA receptors in MSNs contain the α2, γ2, and a β subunit. In addition, there is evidence for the developmental increase of the α1 subunit that contributes to faster inhibitory post-synaptic current (IPSC). Tonic GABAergic currents in MSNs from adult mice are carried by extrasynaptic receptors containing the α4 and δ subunit, while in younger mice this current is mediated by receptors that contain the α5 subunit. Both forms of tonic currents are differentially expressed in D1+ and D2+ MSNs. This study extends these findings by relating presynaptic activation with pharmacological analysis of inhibitory conductance in mice where the β3 subunit is conditionally removed in fluorescently labeled D2+ MSNs and in mice with global deletion of the δ subunit. Our results show that responses to low doses of gaboxadol (2 μM), a GABAA receptor agonist with preference to δ subunit, are abolished in the δ but not the β3 subunit knock out mice. This suggests that the β3 subunit is not a component of the adult extrasynaptic receptor pool, in contrast to what has been shown for tonic current in young mice. Deletion of the β3 subunit from D2+ MSNs however, removed slow spontaneous IPSCs, implicating its role in mediating synaptic input from striatal neurogliaform interneurons.

GABAergic inhibition is mediated by a combination of fast, phasic inhibition and slow, tonic inhibition (Farrant and Nusser, 2005;Glykys and Mody, 2007). Synaptically released GABA mediates phasic inhibition through low-affinity GABA A receptors and is effective in generating neuronal rhythmic activity and synchronicity (Cobb et al., 1995). GABA A receptor mediated tonic current can be both agonist-dependent and independent (Wlodarczyk et al., 2013). GABA spillover from vesicular and non-vesicular sources activates high-affinity extrasynaptic receptors to increase input conductance, affecting neuronal excitability (Brickley et al., 1996;Walker and Semyanov, 2008). In addition perisynaptic GABA receptor mediates an intermediate form of inhibition that is mediated by the neurogliaform subtype of neuropeptide Y positive interneurons (NGF-NPY; Capogna and Pearce, 2011;English et al., 2011). The subunit composition of GABA A receptors that mediate striatal tonic inhibition are of crucial interest in understanding the underlying factors that control MSN output and neuronal excitability. Our previous study used pharmacology (Janssen et al., 2009) and Cre-lox genetics (Janssen et al., 2011) to identify the β3 subunit of GABA A receptor as an important regulator of both striatal D1+ and D2+ MSN tonic current. Here we further extend those findings relating presynaptic activation with pharmacological analysis of inhibitory conductance in mice where the β3 subunit is conditionally removed in fluorescently labeled D2+ MSNs and in mice with global deletion of the δ subunit. We also investigate the role of two classes of presynaptic GABAergic interneurons (Gittis and Kreitzer, 2012), the fast spiking parvalbumin expressing (FS-PV) and NGF-NPY, in distinct GABAergic conductances.

ANIMALS
The use of several strains of transgenic mice examined in this study to genetically identify rare striatal interneurons and have Frontiers in Neural Circuits www.frontiersin.org been described previously (Luo et al., 2013).These include the BAC-npy-eGFP and parv-Cre; rosa26-tdTom. Conditional and neuron specific GABA A β3 subunit knockout (KO) mice and appropriate control mice were produced as reported in Janssen et al., 2011, by crossing floxed β3 mice (β3 f /f ) (Jackson Labs # 008310; Ferguson et al., 2007) to transgenic mice that express Cre recombinase under the drd2 promoter (GENSAT, ER44; Gong et al., 2007). These Cre-dependent β3 subunit KO mice were subsequently crossed to "floxed-stop" rosa26-tdTom mice (Jackson Labs # 007914; Madisen et al., 2010); a second Cre-dependent line to allow visual identification of neurons lacking the β3 receptor subunit restricted to D2+ neurons. We designate this line as drd2-Cre;β3 f /f ;Rosa tdTom . Control mice used in these studies included drd2-Cre;β3 f /+ ;Rosa tdTom and Cre-negative β3 f /f ;Rosa tdTom . We did not detect any statistical differences between these two genotypes and combined them for control experiments. We employed commercially available genotyping of tail biopsies of littermates via Transnetyx, Inc. (Cordova, TN, USA) before conducting experiments. We also used global GABA A δ-receptor subunit knockout mice previously described (Mihalek et al., 1999;Janssen et al., 2009).

BRAIN SLICE PREPARATION
Male and female mice (post-natal day 14-18) were sacrificed by decapitation in agreement with the guidelines of the AMVA Panel on Euthanasia and the Georgetown University ACUC. The whole brain was removed and placed in an ice-cold slicing solution containing (in mM): NaCl (85), KCl (2.5), CaCl 2 (1), MgCl 2 (4), NaH 2 PO 4 (1), NaHCO 3 (25), glucose (25) All recordings presented here were performed at room temperature, 22-24 • C. Voltage-clamp recordings were achieved using the whole-cell configuration of the patch-clamp technique at a holding potential of −70 mV using the Multiclamp 700B amplifier (Molecular Device Co., Sunnyvale, CA, USA). Synaptically coupled interneurons were subjected to holding current values which hyperpolarized neurons to a resting membrane potential of ∼−70 mV in current-clamp mode. Access resistance was monitored periodically during recordings and experiments with >20% change were discarded. Membrane potential measurements were not corrected for liquid junction potential errors.
Recordings were filtered at 2 kHz with a low-pass Bessel filter and digitized at 5-10 kHz using a personal computer equipped with Digidata 1322A data acquisition board and pCLAMP10 software (both from Molecular Devices). Data analysis, curve fitting, and Figure preparation were performed with Clampfit10 software (Molecular Devices). Spontaneous or miniature inhibitory postsynaptic currents (sIPSCs and mIPSCs) were identified using a semi-automated threshold-based mini detection software (Mini Analysis, Synaptosoft Inc., Fort Lee, NJ, USA) and were visually confirmed. IPSC averages were based on more than 50 nonoverlapping events, and decay kinetics were determined with averaged IPSC traces using double exponential curve fittings and reported as weighted time constants (T w ): where t n is the decay time constant for a particular component of the curve and A x is the peak amplitude of the corresponding component. Similarly, decay time constants of unitary evoked IPSCs in synaptically coupled pairs were determined from the average of N evoked responses. mIPSCs were isolated with TTX (0.5 μM) and NBQX (5 μM) while NBQX was not included in sIPSC measurements to not perturb the network activity after removing rapidly decaying sEPSCs as previously described (Forcelli et al., 2012).
THIP mediated inward currents were primarily measured with an all-points histogram that measured the mean holding current 10 s before and during drug application (Luo et al., 2013). Holding current changes during high frequency stimulation of synaptically connected pairs were assessed during depolarization of presynaptic neurons and low pass filtering of the post-synaptic current recording (2-5 Hz).

Frontiers in Neural Circuits
www.frontiersin.org

STATISTICS
Box and whisker plots were generated for more accurate representation of critical data. The whiskers include the minimum and maximum values, while the box outlines the 25th and 75th percentile of data points. The median value is represented by a bar inside the box. Other summary plots display the mean and error bars denote the SEM for a given data set. Statistical significance was determined using the paired twotailed Student's t-test to compare pre-drug conditions with recordings made under drug conditions (etomidate) of the same cell population. Kolmogorov-Smirnov tests were conducted on all data sets to check for normal distribution and appropriate statistics were chosen dependent upon these results. Significance criteria were set at p ≤ 0.05, and all values in text and figures are expressed as mean ± SEM.

RESULTS
To investigate the role of synaptic, perisynaptic and extrasynaptic GABA A receptors in striatal neurons we performed dual whole cell recordings from MSNs and two subtypes of GABAergic interneurons, the NGF-NPY and FS-PV, in mice expressing fluorescent markers (Luo et al., 2013). Single presynaptic action potentials produced reliable IPSCs when neurons were synaptically coupled ( Figure 1A). The rise and decay time of evoked IPSCs from these two interneuron classes differed significantly ( Figure 1B). We then characterized the change in holding current ( I hold ) in post-synaptic MSNs in response to increasing action potential frequency of presynaptic interneurons ( Figure 1C). Figure 1D summarizes these data and illustrates the occurrence of tonic current in MSNs resulting from presynaptic activation.
Tonic GABAergic currents in MSNs from adult mice are carried by extrasynaptic receptors containing the α4 and δ subunit (Santhakumar et al., 2010), while in younger mice this current is mediated by receptors that contain the α5 and β3 subunit (Ade et al., 2008;Janssen et al., 2009). These studies have shown that deletion of the δ subunit removes GABAergic tonic current from adult MSNs while deletion of the β3 subunit removes tonic current from young MSNs. Although α4 and δ subunit mediated tonic current has been found in younger MSNs, it does not differ between D1+ and D2+ subtypes (Janssen et al., 2009). It is the differential expression of the β3 subunit that has been suggested to account for the observed difference in tonic current between D1+ and D2+ subtypes.
To investigate if deletion of the β3 subunit also affects tonic currents mediated by α4 and δ subunit, we used THIP (gaboxadol), a δ subunit-containing GABA A receptor superagonist (Brown et al., 2002). As shown in examples in Figure 2A and the summary data in Figure 2B, response to low doses of gaboxadol (2 μM) are abolished in δ but not β3 subunit knock out mice. This suggests that, in contrast to what was previously shown for tonic current in young mice (Janssen et al., 2009), the β3 subunit is not a component of δ subunit-containing extrasynaptic receptor pool in adult mice (Santhakumar et al., 2010).
Removing the β3 subunit of GABA A receptors from MSNs reduced whole cell current mediated by the general anesthetic etomidate, a β3 subunit preferring modulator (Janssen et al., 2011). Here we extend these results by investigating the action of this general anesthetic on sIPSCs recorded from fluorescently identified D2+ MSNs where the β3 subunit is conditionally deleted. We used a concentration of etomidate (0.5 μM) that does not produce a sustained tonic current ( Figure 3A). As shown in the example traces in Figure 3A and the summary data in Figure 3B, etomidate (0.5 μM) doubled the decay time of sIPSCs in control D2+ MSNs but not in D2+ MSNs with the β3 subunit removed. sIPSCs from MSNs in δ subunit −/− mice were also significantly prolonged by etomidate.
Our previous results revealed that mIPSCs in unidentified MSNs from mice where the β3 subunit was conditionally deleted in D2+ MSNs were significantly faster than in control littermates (Janssen et al., 2011). We refined these results by recording from fluorescent D2+ MSNs and non fluorescence (D2−) MSNs in newly developed mice where the β3 subunit is conditionally deleted in D2+ MSNs that also express red fluorescence (drd2-Cre;β3 f /f ;Rosa tdTom ). Individual mIPSCs recorded in voltage clamp from control and β3 subunit KO MSNs (Figure 4A) are averaged and fitted with double exponential decay to show the Frontiers in Neural Circuits www.frontiersin.org Note that at 10 μM THIP begins to activate non δ-containing receptors * p < 0.05. One-way ANOVA for repeated measures followed by Dunn's post hoc test for the 2 μM THIP cell groups while the Mann-Whitney test was used for the 10 μM THIP groups. difference in decay kinetics ( Figure 4B). As seen in the summary results ( Figure 4C), in D2+ MSNs with the β3 subunit removed, mIPSCs decay faster compared to MSNs in control mice and also D2− MSNs in KO mice. Hentschke et al. (2009) showed that deletion of the β3 subunit in hippocampal CA1 pyramidal neurons removes slow sIPSCs. Therefore we speculate that another contributing factor to the change in average mIPSC kinetics in D2+ MSNs with deletion of the β3 subunit could be the lack of slow IPSCs mediated by presynaptic NGF-NPY interneurons, as shown in the examples in Figure 4D. To investigate this, we looked for the occurrence of slow sIPSCs in several D2+ MSNs in control and in mice with β3 subunit conditional deletion. Spontaneous slow IPSCs occurred very infrequently (<1 event/5 min), in 12 % (6/51) of D2+ MSNs from 10 control mice, 13 % (4/31) of MSNs from 5 δ subunit −/− mice, but they were not observed in 58 D2+ MSNs from 13 mice with the β3 subunit deletion. The differences observed were statistically significant for D2+ MSNs in mice with β3 subunit conditional deletion compared to wild type or δ subunit −/− mice (P = 0.009 and 0.013; Fisher's Exact Test). This is supported by pair recordings from putative NGF-NPY interneurons and D2+ MSNs with (Figure 4E left) and without the β3 subunit (Figure 4E right). The NGF-NPY cells were identified in Frontiers in Neural Circuits www.frontiersin.org the β3 subunit conditional deletion mice by soma size and characteristic firing pattern (Figure 4E inset, Ibáñez-Sandoval et al., 2011 andLuo et al., 2013). As illustrated a burst of action potentials in the NGF-NPY neurons elicited a large slow IPSC only in the D2+ control neuron and not in β3 subunit lacking D2+ MSN.
A lack of slow synaptic responses were observed in four additional NGF-NPY interneuron and D2+ MSN pairs in β3 subunit conditional deletion mice. To further investigate this we recorded from D2+ MSNs during local application of extracellular solution with 12 mM K + or 50 μM 4-aminopyridine to increase synaptic activity. Under these conditions we observed occasional bursts of action potentials in NGF-NPY neurons and large slow IPSCs in D2+ MSNs. In addition, slow IPSCs occurred in 60 % (9/15) of D2+ MSNs from 3 control mice, 75 % (3/4) of MSNs from 1 δ subunit −/− mouse, but they were observed only in 1 out of 10 D2+ MSNs from 2 mice with the β3 subunit deletion (P = 0.018 and 0.041; Fisher's Exact Test). Taken together these data suggest that the role of β3 subunits in mediating synaptic input from neurogliaform neurons as seen in the hippocampus is likely to hold for the striatum as well. But given the rare occurrence of slow IPSCs, their presence would not cause significant changes in the average mIPSC kinetics. Thus other factors may be responsible for the faster mIPSCs in MSNs from mice with β3 subunit conditional deletion.

DISCUSSION
Synaptic, perisynaptic, and extrasynaptic GABA A receptors have distinct roles in mediating key aspects of neuronal microcircuitry that have been extensively characterized in the cerebellum (Farrant and Nusser, 2005) and the thalamus (Jia et al., 2007). These roles include mediating fast feedforward and feedback inhibition, dendritic shunting, regulation of excitability (Semyanov et al., 2004), and generation of oscillatory activity (Cobb et al., 1995;Leresche et al., 2012). Striatal MSNs have crucial roles in controlling movement and are implicated in several neurological disorders (Graybiel et al., 1994). GABA is critical to striatal function as MSNs are GABAergic in nature, forming inhibitory collateral axons among themselves (Gittis and Kreitzer, 2012). Further, GABAergic interneurons make extensive synaptic connections within the entire nucleus. Yet the role of GABAergic inhibition in this crucial area is poorly understood. Our results extend recent important findings on the dynamics of GABAergic inhibition and GABA A receptors in striatal neurons. The power of mouse genetics has allowed in recent years the ability to identify distinct subtypes of GABAergic interneurons by expressing fluorescent markers (Tepper et al., 2010;Gittis and Kreitzer, 2012). We recently reported that NGF-NPY interneurons and the FS-PV interneurons have high synaptic connectivity with MSNs and exhibit nicotinic acetylcholine receptor mediated responses (Luo et al., 2013). In the present study, we refined those results and quantified the differences in both rise and decay time of evoked IPSCs in MSNs for presynaptic NGF-NPY and FS-PV interneurons. In Luo et al. (2013), we have also shown that low frequency (<5 Hz) activation of NGF-NPY can produce sustained GABAergic currents in MSNs. This is different from FS-PV interneurons that require higher presynaptic firing rate. We confirmed and better quantified those results in the present study. We also show that higher presynaptic firing of NGF-NPY interneuron will produce massive GABAergic conductance. However one should consider that our studies were performed at room temperature, which may differently regulate GABA transporter and thus synaptic and extrasynaptic GABA affecting the integration of IPSCs. The question that arises from these data is what GABA A receptors are activated by distinct presynaptic interneurons. Development of mice that allow fluorescent identification of both interneurons and MSNs with GABA A receptor subunit specific deletions are required to answer this question. Our results however, make a first contribution to this answer by showing that removal of the δ but not the β3 subunit affects the response of extrasynaptic receptor selective agonist. This is in contrast to Frontiers in Neural Circuits www.frontiersin.org the result that deletion of the β3 subunit but not the δ subunit affects the prolongation of synaptic decay by the general anesthetic etomidate. Taken together these data allow us to partially model the subunit composition of synaptic and extrasynaptic GABA A receptor in striatal MSNs, and propose that synaptic but not extrasynaptic GABA A receptor must contain the β3 subunit. In contrast, the δ subunit is restricted to extrasynaptic GABA A receptors as expected from current evidence in other brain areas. What is the role of δ and β3 subunits in perisynaptic receptors? Current models propose perisynaptic receptors as central mediators to the synaptic action of neurogliaform neurons in the hippocampus (Capogna and Pearce, 2011). Clearly more studies are needed to extend these findings to other brain regions. The data presented here, suggests that removal of the β3 subunit affect the synaptic action of neurogliaform neuron similarly in the striatum, by significantly reducing the occurrence of spontaneous slow IPSCs (Fisher's Exact Test). The infrequent finding of MSNs with slow sIPSCs suggests that the faster decay of synaptic currents associated with β3 subunit deletion in MSNs is not related to diminished activity at NPY-NGF and MSN synapses, but rather to enhanced expression of α1 subunit-containing synaptic receptors as we previously proposed (Janssen et al., 2011). The data presented show that this effect is indeed specifically due to the deletion of the β3 subunit. In summary, this work emphasizes that in order to better understand GABAergic control of striatal microcircuitry, we need to consider the activation mode and heterogeneity of presynaptic interneurons together with the presence of specific GABA A receptor subtypes with distinct function.