Astrocytes potentiate GABAergic transmission in the thalamic reticular nucleus via endozepine signaling

Significance

Benzodiazepines are commonly prescribed to treat neuropsychiatric disorders, and produce clinical effects on sleep, anxiety, and seizures by augmenting synaptic inhibitory currents through GABA_A receptors. Our previous work has indicated that peptides of the diazepam-binding inhibitor family act as endogenous benzodiazepines (endozepines) in the thalamic reticular nucleus (nRT), where they mediate antiepileptic and other effects. Here we report that astrocytes are required for the actions of endozepines on GABAergic transmission in the nRT. Thus, astrocytes in nRT are specialized to contribute to a localized increase in efficacy of synaptic inhibition relevant to endogenous seizure control.

Abstract

Emerging evidence indicates that diazepam-binding inhibitor (DBI) mediates an endogenous benzodiazepine-mimicking (endozepine) effect on synaptic inhibition in the thalamic reticular nucleus (nRT). Here we demonstrate that DBI peptide colocalizes with both astrocytic and neuronal markers in mouse nRT, and investigate the role of astrocytic function in endozepine modulation in this nucleus by testing the effects of the gliotoxin fluorocitrate (FC) on synaptic inhibition and endozepine signaling in the nRT using patch-clamp recordings. FC treatment reduced the effective inhibitory charge of GABA_A receptor (GABA_AR)-mediated spontaneous inhibitory postsynaptic currents in WT mice, indicating that astrocytes enhance GABA_AR responses in the nRT. This effect was abolished by both a point mutation that inhibits classical benzodiazepine binding to GABA_ARs containing the α3 subunit (predominant in the nRT) and a chromosomal deletion that removes the Dbi gene. Thus, astrocytes are required for positive allosteric modulation via the α3 subunit benzodiazepine-binding site by DBI peptide family endozepines. Outside-out sniffer patches pulled from neurons in the adjacent ventrobasal nucleus, which does not contain endozepines, show a potentiated response to laser photostimulation of caged GABA when placed in the nRT. FC treatment blocked the nRT-dependent potentiation of this response, as did the benzodiazepine site antagonist flumazenil. When sniffer patches were placed in the ventrobasal nucleus, however, subsequent treatment with FC led to potentiation of the uncaged GABA response, suggesting nucleus-specific roles for thalamic astrocytes in regulating inhibition. Taken together, these results suggest that astrocytes are required for endozepine actions in the nRT, and as such can be positive modulators of synaptic inhibition.

Benzodiazepines are widely used in the treatment of neurologic disorders, such as epilepsy, anxiety, and sleep disturbances. The classical mode of benzodiazepine action is allosteric potentiation of ionotropic currents through type A receptors for the inhibitory neurotransmitter GABA benzodiazepine-binding sites on GABA_A receptors (GABA_ARs) (1). The identification of benzodiazepine-binding sites in the CNS (2, 3) led to the hypothesis that endogenous brain-derived benzodiazepine-mimicking substances (i.e., endozepines) may exist (4). The 10-kDa peptide diazepam-binding inhibitor (DBI) and associated cleavage products bind to GABA_AR benzodiazepine-binding sites (5–7), but physiological roles for such substances are not well defined.

DBI mRNA transcript and peptide immunoreactivity have been observed in various brain regions in both neurons (8) and glial cells (9–12). DBI-derived peptides, including triakontatetraneuropeptide (34 amino acids), octadecaneuropeptide (18 amino acids), and octapeptide (8 amino acids), are secreted by astrocytes (13–18). The release of endozepines by cultured glial cells can be inhibited by GABA or somatostatin (14, 15) and stimulated by beta-amyloid peptide or pituitary adenylate cyclase-activating polypeptide (16–18). DBI peptides also have been demonstrated to affect astrocytic function via actions on the mitochondrial benzodiazepine receptor, also known as 18-kDa translocator protein, which influences GABA receptor activation through modulation of neurosteroid production (19, 20). Endogenous regulation of endozepine actions by native glial cells and subsequent effects on synaptic inhibition remain to be demonstrated.

The thalamic reticular nucleus (nRT) is a brain structure in which endozepines regulate synaptic inhibition and suppress absence seizure activity (21). The nRT acts as an anatomical and functional gate between the thalamus and cortex, receiving excitatory inputs from corticothalamic and thalamocortical axons and powerfully inhibiting thalamocortical relay cells in the dorsal thalamus (22, 23). Intra-nRT inhibition arises from recurrent collateral connections between nRT cells, and modulation of intra-nRT inhibition can bidirectionally regulate the thalamic oscillatory activity that is a hallmark of absence seizures. Specifically, reductions in intra-nRT inhibition drive hypersynchronous epileptiform oscillations between the nRT and the adjacent ventrobasal nucleus (VB) (24), whereas gains in intra-nRT inhibitory strength suppress epileptiform oscillatory duration and power (25).

Our recent work suggests that DBI-derived endozepines in the nRT act in the latter mode, by potentiating inhibitory GABA_AR-mediated currents (21). These endozepine actions are absent in α3(H126R) mutant mice, in which classical benzodiazepine binding via α3-containing GABA_ARs is abolished, and in nm1054 mutant mice, which lack the Dbi gene (21, 26, 27). Which cells in the nRT (glia, neurons, or both) are primarily involved in nRT endozepine actions remains unclear.

Here we tested the effects of fluorocitrate (FC), a glia-selective metabolism inhibitor that blocks the aconitase enzyme, which has been shown to be effective in reducing astrocytic function both in vivo and in vitro (28–30), on synaptic inhibition and allosteric modulation of GABAergic currents in the thalamus. Our findings indicate that astrocytes are required for modulation of synaptic inhibition by DBI-derived endozepines in the nRT.

Results

DBI Peptide is Expressed in Both Astrocytes and Neurons in Mouse nRT.

DBI peptide is widely expressed in the mouse thalamus (21), and previous studies have shown that DBI or its processing products are expressed in the nRT in rats (31). To examine the cellular localization of DBI peptide in mouse nRT, we used double-labeled fluorescent immunohistochemistry for DBI and either the astrocytic marker GFAP or the neuronal marker NeuN, which confirmed the presence of DBI peptide in both astrocytes and neurons in the nRT (Fig. S1). In our experiments, we focused on elucidating the astrocytic contribution to endozepine signaling in the nRT.

FC Selectively Impairs Astrocytic Function in the nRT.

We investigated the glial contribution to endozepine signaling in the nRT using the gliotoxin FC (28, 29). The glia-selective impairment in the nRT induced by FC treatment was confirmed using sulforhodamine 101 (SR101), a red fluorescent dye that is selectively taken up by astrocytes (32, 33). FC treatment (100 μM for 30 min) reduced the capacity of astrocytes to incorporate the SR101 dye (Fig. 1A), but did not affect GFAP immunoreactivity (Fig. 1B). Thus, it appears that FC treatment selectively alters the functional capacity of astrocytes in the nRT while astrocytic marker expression is maintained.

Fig. 1.

Fluorocitrate selectively alters astrocytic function in the nRT. (A) Fluorescence images illustrating uptake of the SR101 dye in the nRT in control slices (Left) and after FC treatment (Right). (B) Merged image of SR101 staining (red) with immunolabeling of GFAP (green) and NeuN (blue) showing colocalization of SR101 and GFAP (yellow, arrows) in the same slices depicted in A. (Scale bars: 40 μm.)

The intrinsic membrane properties and cellular excitability of nRT neurons, as assessed by voltage–current (V–I) analysis in current-clamp recordings, were not altered by FC treatment (Table S1). Cell membrane capacitance, recorded in the voltage-clamp configuration, also was unaffected (control, 55.90 ± 4.47 pF, n = 11 cells; FC, 51.96 ± 4.70 pF, n = 10 cells; P > 0.5). Thus, the functional effects of FC in the nRT appear to be selective for astrocytes.

Impairment of Astrocytic Function Decreases Spontaneous Inhibitory Postsynaptic Current Duration in WT nRT, an Effect Blocked by both α3(H126R) and nm1054 Mutations.

If nRT astrocytes are required for endozepines to augment synaptic GABA responses, then the gliotoxin FC should decrease inhibitory postsynaptic current (IPSC) duration in the nRT. We recorded spontaneous IPSCs (sIPSCs) to test this hypothesis. In slices from WT C57BL/6 mice, FC-treated cells (n = 22) exhibited a decreased sIPSC duration (P < 0.001) compared with control cells (n = 13) (Fig. 2 A, B, D, and E). FC treatment did not affect sIPSC amplitude (P > 0.15), but did reduce sIPSC frequency (control, 2.14 ± 0.13 Hz; FC, 1.70 ± 0.13 Hz; P < 0.05) (Fig. 2 F–G), suggesting that presynaptic mechanisms also contribute to the effects of FC. The changes in sIPSC duration were absent in slices from α3(H126R) mice (control, n = 12 cells; FC-treated, n = 10 cells; P > 0.3) (Fig. 2 C–E), but the FC-induced reduction in sIPSC frequency was maintained (control, 2.02 ± 0.24 Hz; FC, 1.33 ± 0.19 Hz; P < 0.05) (Fig. 2G), indicating that presynaptic effects of FC are endozepine-independent. There were no differences in input (R_in, F = 0.86, P > 0.4) or series resistance (R_s, F = 0.24, P > 0.8) across these groups, demonstrating that these differences are not related to variations in recording quality. These experiments suggest that astrocytes are required for positive allosteric modulation of neuronal GABA_ARs in the nRT, and that these modulators are likely endozepines that act via the α3 subunit benzodiazepine-binding site.

Fig. 2.

Fluorocitrate reduces sIPSC duration in the nRT, and this effect is blocked by the α3(H126R) mutation. (A) Representative continuous traces of sIPSCs recorded in nRT cells from WT C57BL/6 mice in control conditions (Left) and after FC treatment (Right). (B) Averaged sIPSCs from nRT cells from WT mice in control conditions (black trace) and after FC treatment (gray trace), normalized to peak amplitude. (C) Averaged sIPSCs from nRT cells from α3(H126R) mice in control conditions (dark-blue trace) and after FC treatment (light-blue trace), normalized to peak amplitude. (D) Mean ± SEM for sIPSC half-width in WT and α3(H126R) mice. (E) Probability distributions comparing sIPSC half-width in WT and α3(H126R) mice in control and FC-treated conditions (n = 1,000–2,200 events/group). (F and G) Mean ± SEM for sIPSC amplitude (F) and frequency (G) in WT and α3(H126R) mice. *P < 0.05, ***P < 0.001 vs. WT con (D, E, and G) or α3(H126R) con (G).

To determine whether FC alters nRT sIPSC duration by preventing the actions of DBI-derived peptides, we examined the effect of FC treatment in slices from nm1054 homozygous mutants, in which the Dbi gene is deleted (25), and from WT littermates on the 129S6/SvEvTac background. Staining for DBI, GFAP, and NeuN demonstrated a lack of DBI colocalization in either neurons or astrocytes in nm1054 mutant nRT (Fig. S1). As seen in C57BL/6 mice [WT strain for the α3(H126R) mutants], FC treatment resulted in decreased sIPSC duration in nRT neurons from nm1054-related WT mice (control, n = 11 cells; FC-treated, n = 10 cells; P < 0.05) (Fig. 3 A, C, and D); however, in nm1054 mutants, FC had no effect on sIPSC duration (control, n = 10 cells; FC-treated, n = 11 cells; P > 0.4) (Fig. 3 B–D). There were no between-group differences in R_in (F = 0.67, P > 0.5) or R_s (F = 1.6, P > 0.2). Thus, the FC-induced reduction in sIPSC duration reflects the removal of a source of DBI-derived endozepine peptides.

Fig. 3.

Effects of fluorocitrate in the nRT are absent in nm1054 mutant mice. (A) Averaged sIPSCs from nRT cells from WT littermates of nm1054 mice in control conditions (black trace) and after FC treatment (gray trace), normalized to peak amplitude. (B) Averaged sIPSCs from nRT cells from nm1054 mice in control conditions (red trace) and after FC treatment (pink trace), normalized to peak amplitude. (C) Mean ± SEM for sIPSC half-width in WT and nm1054 mice. (D) Probability distributions comparing sIPSC half-width in WT and nm1054 mice in control and FC-treated conditions (n = 972–1,036 events/group). *P < 0.05; **P < 0.01; ***P < 0.001 vs. WT control (C and D).

FC Alters IPSC Duration in VB Neurons through Impairment of Astrocytic GABA Transporter Function.

Our previous results indicate that endozepines are released in the nRT, but not in the VB (19). To examine whether the effects of FC are similarly nucleus-specific, we recorded sIPSCs in VB neurons under control conditions (n = 14) and after FC treatment (n = 8) (Fig. S2 A and B). FC treatment shifted the probability distribution for sIPSC half-width toward longer events (Fig. S2C, P < 0.001), but neither sIPSC amplitude nor sIPSC frequency were affected (P > 0.8), suggesting that the effects of FC in VB neurons are mainly postsynaptic.

We hypothesized that impairment of GABA uptake by astrocytic GABA transporters (GATs) (34–36) may underlie the difference in sIPSC duration. We thus tested the combined effects of the GAT-1 antagonist 1,2,5,6-tetrahydro-1-[2-[[(diphenylmethylene)amino]oxy]ethyl]-3-pyridinecarboxylic acid hydrochloride (NNC-711; 4 μM) and the GAT-3 antagonist 1-[2-[Tris(4-methoxyphenyl)methoxy]ethyl]-(S)-3-piperidinecarboxylic acid (SNAP-5114; 10 μM). These two GAT subtypes, the primary GATs expressed in the thalamus, appear to be expressed exclusively on astrocytes (37, 38). Consistent with this hypothesis, GAT antagonists increased sIPSC duration under control (i.e., intact glial function) conditions (n = 11; P < 0.001) (Fig. S2 B and C), in a manner not different from that seen after FC treatment (P > 0.2). There were no differences in either R_in (F = 0.28, P > 0.7) or R_s (F = 1.49, P > 0.2) across groups. These results suggest that, in contrast to the inhibition-reducing effects of FC in the nRT, GABAergic inhibition in the VB is enhanced after FC treatment, and these effects largely reflect altered GAT function.

GAT Blockade Alone Does Not Alter sIPSCs or the Response to Benzodiazepine-Binding Site Antagonism in the nRT.

The opposing effects of FC on sIPSC duration in the nRT and VB could reflect differences in allosteric receptor modulation and/or differences in GAT uptake between these two nuclei. Thus, we tested the effect of GAT blockade alone on sIPSCs in the nRT. In contrast to the effects of GAT blockade in the VB, GAT blockade did not affect sIPSC duration in nRT neurons (control, n = 12; GAT blockade, n = 10; P > 0.2) (Fig. S3 A and B). In addition, GAT blockade did not affect the response to the benzodiazepine-binding site antagonist flumazenil (FLZ, 1 μM) (Fig. S3C), indicating that endozepine modulation of sIPSCs in the nRT remains intact when GAT function is impaired. Thus, the differing effects of FC on sIPSCs in the nRT and VB reflect nucleus-specific contributions of endozepine modulation and GAT-mediated uptake on sIPSC duration.

FC Treatment Blocks nRT Potentiation of VB Membrane Patch Responses to GABA Uncaging.

We recently demonstrated that outside-out membrane “sniffer patches” pulled from VB neurons exhibit a prolonged response to focal laser photolysis of caged GABA when placed in the nRT compared with the VB (21, 39). This methodology has been termed the sniffer patch laser uncaging response (SPLURgE) (39). This effect is reduced by both benzodiazepine-binding site antagonism and the nm1054 mutation, indicating that endozepines underlie a major part of this potentiation. To determine whether astrocyte-derived endozepines are responsible for this effect, we tested the response of sniffer patches to GABA uncaging when placed in nRT or VB of slices from C57BL/6 mice either under control conditions or after FC treatment. Under control conditions, SPLURgE duration was significantly prolonged when patches were placed in the nRT compared with the VB (P < 0.001; Fig. 4 and Fig. S4), confirming our previous results (21). In FC-treated slices, although an overall increase in SPLURgE duration in the VB was seen (P < 0.05, Fig. 4B), only the late decay portion of the response (i.e., the 90–10% decay time) was enhanced by placement in the nRT (half-width, P > 0.6; Fig. 4; 90–10% decay time, P < 0.05, Fig. S4).

Fig. 4.

Combined blockade of GATs and benzodiazepine-binding sites mimics the effects of fluorocitrate on nRT potentiation of VB membrane patch responses to GABA uncaging. (A) Responses to GABA uncaging averaged across all outside-out patches pulled from VB neurons and recorded after placement in either VB or the nRT in control conditions (black traces), after FC treatment (gray traces), GAT blockade (green traces) or combined GAT blockade plus FLZ (orange traces). Currents are normalized to the peak amplitude. All traces are shown to the same scale. (B) Mean ± SEM for half-width of uncaging responses recorded under control conditions (black bars), after FC treatment (gray bars), in the presence of GAT antagonists (green bars), or in the presence of GAT antagonists and FLZ (orange bars). Each bar represents between six and nine patches. (C) Mean ± SEM for ratio of values obtained when patches placed in the nRT compared with averaged value for patches placed in VB for each respective group. hν, 1-ms UV laser stimulus. *P < 0.05; ^###P < 0.001 vs. respective group values for patches placed in VB.

A Combination of GAT Blockade and FLZ Mimics the Effects of Fluorocitrate on Sniffer Responses.

To determine whether the increased SPLURgE duration produced by FC could be attributed to GAT disruption, we directly tested the effects of GAT blockade on the responses. GAT blockade increased the SPLURgE duration of patches placed in the VB to the same degree as that seen after FC treatment (P < 0.05 compared with control; Fig. 4B and Fig. S4A), suggesting that the FC-increased duration of response could be attributed entirely to astrocytic GABA transport blockade. Under these conditions, however, potentiation of the SPLURgE by placement in the nRT was preserved (difference in VB vs. nRT response, both after GAT blockade; P < 0.001; Fig. 4 B and C and Fig. S4) to the same degree as that in control conditions (Fig. 4C and Fig. S4B), indicating that GAT blockade does not interfere with the endozepine-dependent enhancement of the SPLURgE. Furthermore, treatment with FLZ in the presence of the GAT antagonists prevented nRT-dependent potentiation while preserving the prolongation of responses in the VB (Fig. 4 B and C and Fig. S4). In addition, similar to FC, combined GAT blockade and FLZ did not significantly affect SPLURgE half-width, but prolonged the 90–10% decay time (Fig. 4B and Fig. S4A). This result confirms and extends our previous finding in nm1054-related WT mice that the combination of GAT antagonists and FLZ is sufficient to block nRT-dependent potentiation of the uncaging response (21).

Taken together, these results suggest that the effects of FC on the SPLURgEs involve a combination of impaired GAT function, largely restricted to the VB, and loss of endozepine signaling in the nRT. Thus, glial mediation of endozepine action is primarily responsible for the nRT-dependent potentiation of uncaging responses.

Discussion

A growing body of evidence demonstrates that astrocytes affect and modulate neuronal function in a wide variety of ways throughout the lifespan (40). Using the gliotoxin FC to selectively impair astrocytic function, we have shown that astrocytes are required for the modulation of synaptic inhibition by DBI-derived endozepines in the nRT. These effects are lost in both α3(H126R) and nm1054 mutant mice, which harbor deficits in benzodiazepine binding and DBI expression, respectively. We also demonstrate that astrocytic modulation of the response of thalamic GABA_ARs to GABA reflects a combination of at least two effects, GABA uptake via GATs and endozepine action. These findings demonstrate a unique mechanism by which astrocytes can modulate fast synaptic inhibition, and have implications for understanding the role of neuromodulation of thalamic synaptic inhibition in the normal functions of sleep and sensory processing as well as in pathological states, such as absence seizures.

The results presented here provide further insight into the epileptogenic effects of FC. Intravenous, intracranial, or i.p. injections of the FC precursor fluoroacetate can cause spike-and-wave discharges (SWDs) and absence seizures in various species, including mice, cats, dogs, and rabbits (41–44), with varying latencies and dose dependencies. With respect to the present study, the observation that injection of fluoroacetate in cats was found to lead to paroxysmal (40 Hz) activity in the nRT shortly before the development of synchronized SWD cortical activity (41), is particularly intriguing. Our present results suggest a cellular mechanism for this effect in the nRT: a decrease in synaptic inhibition as reflected in decreased sIPSC duration, owing to a loss of endozepine signaling, and reduced sIPSC frequency. The development of SWD activity and absence seizures as a consequence of FC treatment is thus consistent with the hypothesized role for intra-nRT inhibition in decreasing thalamocortical oscillatory circuit activity (24), although other nonendozepine actions, such as reduced glutamate uptake and recycling (45), also could contribute to seizures, as could effects of astrocytic disruption in regions outside of the thalamus.

Whereas sIPSC duration was decreased by FC treatment in WT nRT, sIPSCs in VB neurons could be enhanced, and this effect was mimicked by GAT blockade. Although the impaired GAT function as a result of FC treatment may possibly mask a loss of endozepine signaling in this nucleus, this is unlikely given our previous findings that FLZ alone does not affect VB sIPSCs (21), as well as the results of our SPLURgE experiments demonstrating that SPLURgEs obtained in the VB in the presence of GAT antagonists do not differ from those obtained after the addition of FLZ. These results thus provide further evidence that endozepines apparently are not expressed in the VB (21). Furthermore, potentiated sIPSCs in VB cells would be expected to enhance postinhibitory rebound responses in these cells (46), suggesting another mechanism that may contribute to FC-induced SWD.

The differing effects of FC and GAT blockade on synaptic inhibition in the nRT and VB suggest that the mechanisms of astrocytic modulation of GABAergic function likely vary widely across different brain areas. Although the modulation of GABA_B responses via GAT activity is well established (35, 47–49), investigations into GABA_AR-mediated IPSCs have yielded inconsistent findings. Tonic inhibition and evoked currents appear to be more affected by GAT activity compared with spontaneous synaptic currents (34, 36, 47, 50, 51), perhaps reflecting contributions of GABA receptors at extrasynaptic locations tightly regulated by GATs (35). The results presented here, showing the apparently stronger GAT activity in the VB than in the nRT, suggest that differential GAT activity may underlie at least some of these apparent discrepancies.

In the hippocampus, astrocytes can indirectly increase synaptic inhibition by releasing glutamate, which activates ionotropic glutamate receptors on interneurons, leading to increased GABA release (52). Conversely, pathological activation of astrocytosis in the hippocampus also has been shown to decrease synaptic inhibition as a result of glutamate/glutamine cycle disruption (53), which may explain the reductions in sIPSC frequency in the nRT after FC treatment seen in the present study. Astrocytes themselves also may be a source of GABA, as demonstrated in the cerebellum, the olfactory bulb, and cultured hippocampal cells (54–56), and chelating astrocytic calcium in the barrel cortex alters the excitability of nearby neurons in a manner similar to that of combined GABA_A and GABA_B receptor blockade (57). As demonstrated in the present study, endozepine signaling adds to this growing list of the mechanisms by which astrocytes can modulate fast synaptic inhibition.

In contrast to the nucleus-specific effects of GAT blockade alone on sIPSCs in the nRT and VB, SPLURgE duration was increased after GAT blockade in both nuclei. This finding likely reflects the differences in methodology and GABA_AR populations studied in each experiment, and suggests either that GABAergic currents mediated by nRT GABA_ARs are not as sensitive to GABA uptake as receptors in VB neurons, or that GABA uptake itself is less robust in the nRT than in the VB. We recently demonstrated that the 90–10% decay time SPLURgE parameter, representing the late decay kinetics of the responses, is more sensitive to changes in GAT function than the half-width, or early decay (39). This is consistent with the results presented here, in which FC did not significantly increase the SPLURgE half-width for patches placed in the nRT, but did increase the 90–10% decay time. The further enhancement of SPLURgE 90–10% decay time under GAT block + FLZ conditions compared with FC treatment may represent a small degree of uptake mediated by GATs located on neurons rather than on astrocytes in the nRT.

Our results suggest that either nRT astrocytes release DBI endozepines or that an unknown astrocytic signal is required for neuronal endozepine release or peptide cleavage. Further experiments are needed to elucidate the mechanism(s) underlying endozepine modulation in the nRT. Identifying neuronal and astrocytic sources of secreted DBI will have implications for our understanding of the mechanisms of DBI release, which are unconventional and characterized by exophagic secretion (58). Analysis of release of the DBI homolog acyl CoA-binding protein (ACBP) indicates that soluble NSF attachment protein receptor (SNARE)-dependent fusion of vesicles containing ACBP/DBI to the plasma membrane is required for secretion (59, 60). Although there is some controversy regarding astrocytic exocytosis mechanisms (61), astrocytes appear to express functional SNARE complex components (62–64). It will be interesting to determine whether SNARE-dependent exophagy underlies DBI release from astrocytes, and whether these mechanisms exhibit region-dependent specificity.

Both of the mutant mouse models used here exhibit greater SWD activity compared with their WT counterparts (21). The present study, which demonstrates a lack of astrocytic endozepine actions in these animals, suggests that the enhancement of astrocytic mechanisms related to endozepine modulation may provide a useful therapeutic avenue for epilepsy and other neurologic diseases.

Materials and Methods

WT, α3(H126R), and nm1054 mutant mice were bred and housed as described previously (21, 26, 65). FC stocks were prepared according to published procedures (28), and acute horizontal brain slices containing the nRT and VB were prepared following established methods (66). A subset of slices was treated with 100 μM FC for at least 30 min at room temperature. For SR101 staining, slices were incubated in 1 μM SR101 in oxygenated artificial cerebrospinal fluid (ACSF) for 20 min, washed in control ACSF for at least 10 min at 34 °C, and then fixed in 4% paraformaldehyde and processed for immunocytochemistry. In sIPSC and uncaging experiments, a CsCl-based isotonic chloride intracellular pipette solution was used, and ionotropic glutamatergic currents were blocked using kynurenic acid or d-(−)-2-amino-5-phosphonovaleric acid (APV) and 6,7-dinitroquinoxaline-2,3-dione (DNQX) in the extracellular bath solution.

Current-clamp recordings were performed using a K-gluconate pipette solution in ACSF containing kynurenic acid and picrotoxin. FLZ and the GAT antagonists NNC-711 and SNAP-5114 were bath-applied as indicated. Laser photolysis of caged GABA (100 μM) was achieved via 1-ms UV laser exposure. Experimental procedures are described in detail in SI Materials and Methods.