INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

GABA_A receptors (GABA_A-Rs) mediate fast synaptic inhibition throughout the mammalian brain (McCormick 1992). Like other members of the ligand-gated ion channel superfamily of receptors, GABA_A-Rs are presumably pentameric structures consisting of multiple subunits (reviewed by Barnard et al. 1998; Chang et al. 1996; Tretter et al. 1997). Twenty subunits have been identified in mammalian tissues, including six , four , three , one , one , one , three , and one  subunit (for reviews see Barnard et al. 1998; Sieghart et al. 1999). Allowing for only , , and  subunits in a functional receptor, more than 10,000 configurations are possible, but in reality the final number is much smaller (McKernan and Whiting 1996).

The  subunit contributes to numerous properties of the GABA_A receptor, including agonist efficacy (Ebert et al. 1994; Krasowski et al. 1997), benzodiazepine pharmacology (Hadingham et al. 1993; Pritchett et al. 1989; Pritchett and Seeburg 1990; Wafford et al. 1993; Wisden et al. 1991), and the kinetics of activation and deactivation (Gingrich et al. 1995; Lavoie et al. 1997; Tia et al. 1996; Vicini et al. 2001). Although the pharmacological profiles of multiple receptor configurations have been extensively studied using heterologous expression systems (reviewed by Sieghart 1995), the contribution of specific GABA_A receptor subtypes, and of individual subunit species, to synaptic physiology is largely unknown.

The 1 subunit is the most common  isoform in the mammalian CNS, contributing to approximately 40 to 65% of the total number of GABA_A-Rs in the brain (McKernan and Whiting 1996; Sur et al. 2001). The 1 subunit has been investigated using gene targeting techniques, and the study of 1(H101R) knock-in mutant mice suggests that the 1 subunit contributes to the amnestic, sedative, and anticonvulsant properties of benzodiazepines but not to the anxiolysis, myorelaxation, or motor impairment caused by those drugs (McKernan et al. 2000; Rudolph et al. 1999). This evidence, in turn, suggests an important role for the 1 subunit in inhibitory synaptic transmission.

To date, specific knowledge of synaptic 1 subunit function is restricted to the cerebellum. In the cerebellum, the level of 1 subunit expression appears to increase between postnatal day 11 (P11) and P35, as evidenced by the fact that miniature inhibitory postsynaptic currents (mIPSCs) recorded in neurons obtained from wild-type mice decayed more rapidly in the P35 age group than in the P11 group and that "switch" was not observed in neurons obtained from 1 subunit knock-out (1 KO) mice (Vicini et al. 2001).

The degree to which the GABA_A receptor 1 subunit is incorporated into functional synaptic receptors in the hippocampus is unknown. In the adult rat, mRNA and GABA_A receptor protein corresponding to the 1, 2, and 5 subunits can be readily detected in the hippocampal formation, the 4 protein is predominantly detected in the molecular layer of the dentate gyrus, and the 3 protein is barely detectable (Fritschy and Möhler 1995; Laurie et al. 1992; Pirker et al. 2000; Sperk et al. 1997; Wisden et al. 1992). In the CA1 subfield, Sperk et al. (1997) were unable to detect any of the  subunit proteins on the somata of pyramidal neurons, and only 1 subunit protein could be detected on the somata of CA1 interneurons (see also Gao and Fritschy 1994). More recent data, however, have demonstrated that parvalbumin-containing basket cells form axosomatic synapses with pyramidal cells that are rich in the 1 subunit (Klausberger et al. 2002). In the dendritic field layers of CA1 pyramidal cells (strata radiatum and oriens), the relative subunit protein densities appeared to be 1 > 5 > 2 4 ~ 3, and 6 was undetectable.

Given the predominant role that GABA_A-Rs play in mediating inhibitory synaptic transmission in the CNS, and their critical relevance to mediating the effects of a host of therapeutic compounds, including benzodiazepines and general anesthetics (Franks and Lieb 1997; Jones and Harrison 1993; Mihic et al. 1997; Olsen 1998), it is important to further define and understand the molecular nature of functional GABA_A receptors. Using WT and 1 KO mice, we examined mIPSC frequency, amplitude, decay time, and zolpidem modulation in CA1 interneurons and pyramidal cells in order to evaluate the contribution of the GABA_A-R 1 subunit to native receptors mediating fast inhibitory synaptic transmission in the CA1 region of the hippocampus.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Mutant mouse production

Production and initial characterization of GABA_A-R 1 KO mice have been described (Vicini et al. 2001). Mice that were heterozygous for the floxed, unrecombined allele (^F) and the floxed, recombined allele () were interbred to produce control homozygous floxed unrecombined mice (1^F/F), heterozygous mice (1^F/), and homozygous null allele mice (1^/ or 1 KO). For the purposes of this paper, WT, or 1^+/+, is defined as the floxed, unrecombined targeted locus without neo/TK (F), as illustrated in Vicini et al. (2001). These mice were of the F_5-6 generations and the genetic background consisted of a mixture of C57B1/6J, Strain 129/Sv/SvJ, and FVB/N. It is important to note that the genetic background of all mice is identical, including the 1 gene and flanking DNA. Since the 1 allele was floxed in Strain 129/Sv/SvJ embryonic stem cells (Nagy et al. 1990), the 1 gene and flanking DNA is Strain 129Sv/SvJ derived in all mice. Mice were genotyped by Southern blot analysis of tail clip DNA as described (Vicini et al. 2001). Breeding and genotype analysis occurred at the University of Pittsburgh and, subsequently, 3- to 4-wk-old animals were shipped to Weill Medical College, Cornell University, for electrophysiological analysis.

Electrophysiology

In accordance with institutional and federal guidelines, standard hippocampal sections were prepared from P23-P58 mice. Briefly, mice were anesthetized with isoflurane and decapitated; the brain was rapidly removed and placed in ice-cold (2-4°C) Krebs solution saturated with 95% O₂-5% CO₂. The brain was blocked, and three to four 350- to 400-µm coronal sections were obtained using a microslicer (DTK, Kyoto, Japan). Slices were maintained in saturated Krebs solution at 37°C for 30-60 min prior to use and then kept at room temperature (22-24°C). Ascorbic acid (1 mM) was present in the Krebs solution during the dissection and recovery period but was not present during drug application or data acquisition.

An individual slice was transferred to the recording chamber and held in place by nylon threads attached to a platinum frame and continuously superfused with O₂-CO₂ saturated Krebs solution. CA1 pyramidal cells and CA1 interneurons in the stratum radiatum and s. lacunosum-moleculare were visually identified using a Zeiss Axioskop FS microscope fitted with DIC-IR optics. Whole-cell patch clamp recordings were performed under voltage-clamp using either an Axopatch 1D or Axopatch 200A (Axon Instruments, Union City, CA) amplifier. Cells were voltage clamped at 60 mV after correcting for liquid junction potential and compensating for capacitance and series resistance. Access resistance was monitored using a 5-mV test pulse throughout the recording period; cells were included for analysis if the series resistance was less than 25 M and rejected if resistance changed by more than 25% during the experiment. Data were acquired at 10 kHz using pClamp 8 (Axon Instruments) and filtered at 2 kHz.

The Krebs solution contained 117 mM NaCl, 3.6 mM KCl, 2.5 mM CaCl₂, 1.2 mM MgCl₂, 25 mM NaHCO₃, 1.2 mM NaH₂PO₄, 11 mM D-glucose, 5 × 10⁴ TTX, and 300-305 mOsm (adjusted with sucrose). Recording electrodes were made of borosilicate glass and had a resistance of 3-5 M when filled with intracellular solution containing (in mM) 130 CH₃SO₃Cs, 8.3 CH₃SO₃Na, 1.7 NaCl, 1 CaCl₂, 10 EGTA, 2 Mg₂-ATP, 0.3 Na-GTP, 10 HEPES (pH 7.2 with CsOH and 295 mOsm with sucrose). All compounds were obtained from Sigma (St. Louis, MO) except for TTX, which was obtained from Alomone Labs (Jerusalem, Israel). Zolpidem was prepared as a 5-mM stock solution in 100% ethanol and serially diluted to 30 and 300 nM in Krebs solution; the sequence of drug application was Krebs solution, Krebs containing: TTX, TTX + 30 nM zolpidem, and TTX + 300 nM zolpidem. The sequence of application was always the same, and only one cell per slice was tested. In a number of experiments, bicuculline methiodide (20 µM) was used. Drugs were preapplied for 5 min prior to data acquisition.

Data analysis

Off-line analysis was performed using MiniAnalysis 5.5 (Synaptosoft, Decatur, GA), SigmaPlot 6.0 (SPSS, Chicago, IL), and Prism 3 (GraphPad, San Diego, CA). Using the random selection function in MiniAnalysis, ensemble mIPSCs were created by randomly selecting a total of 40-50 individual mIPSCs from each recording condition (in 9 instances, less than 50 events were analyzed due to the low frequency of suitable mIPSCs, and the lowest number of events averaged was 27) and aligning them to the 50% rise time; the 90-10% decay phase was fit to two exponents with the least-squares simplex method; overlapping events were excluded from the ensemble. The weighted decay time constant (_w) was calculated as _w = (A₁ × ₁ + A₂ × ₂)/(A₁ + A₂), where ₁ and ₂ are the time constants of the first and second exponential functions, respectively, and A₁ and A₂ are the current amplitudes measured at time, t, equal to ₁ and ₂, respectively (Banks et al. 1998; Vicini et al. 2001). Data are presented as mean ± SE unless otherwise indicated. Statistical significance was determined using paired and unpaired two-way t-tests within cell groups as appropriate, Kolmogorov-Smirnov test for comparing cumulative probabilities, and one-way ANOVA with Tukey's post hoc test for comparisons between interneurons and pyramidal cells; significance was assumed for P < 0.05. To minimize experimental bias, recordings were performed without knowledge of the genotype, and the genotype was not revealed until after the data were analyzed.

Peak-scaled nonstationary noise analysis (Brickley et al. 1999; De Koninck and Mody 1994; Perrais and Ropert 1999; Silver et al. 1996; Traynelis et al. 1993; Yoshimura et al. 1999) was performed using MiniAnalysis 5.5 (Synaptosoft). Validity of this approach requires that the current decay time is stable over the course of the recording and that there is no correlation between mIPSC amplitude and decay time; plots of decay time and peak amplitude were created for each group of cells tested and no correlation was observed.

For each data subset obtained as described above, mIPSCs were reexamined and only those within 25% of the mean amplitude were included in the noise analysis. The unitary current (i) and total number of channels (N) were estimated by fitting the following equation

(1)

where ² is the variance at given time, t, of the mIPSC, i is the unitary current, I is the current at given time t, N is the total number of channels, p is the probability of channel opening at the peak of the mIPSC, and is the variance in the baseline noise measured before the peak. It should be noted that the algorithm forces p to approach unity, and consequently, p is not reported. Fitting was performed without including the offset and after subtracting the baseline variance.

An estimate of single channel conductance () was calculated as

(2)

where i is the calculated unitary current, V_hold is the amplifier holding potential, and E_rev is the measured mIPSC reversal potential.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

mIPSC characteristics

CA1 interneurons and CA1 pyramidal cells in the stratum radiatum and s. lacunosum-moleculare were visually identified by their relatively bright somata and proximal dendrites. We selected as candidate interneurons those cells that were at least 10 µm away from the border between the s. pyramidale and s. radiatum. The majority of a sampling of neurons recorded within s. pyramidale displayed action potential accommodation during depolarizing current injection that is characteristic of CA1 pyramidal neurons (Madison and Nicoll 1984; Thompson et al. 1985).

mIPSCs recorded in neurons from WT mice were compared with those recorded in neurons from 1 KO mice. mIPSC frequency in WT interneurons and pyramidal cells was essentially identical (2.5 ± 0.5 and 2.4 ± 0.5 Hz, respectively). In neurons from 1 KO mice there was a 50-60% decrease in mIPSC frequency, and this decrease was significant for both cell groups (Table 1). Mean mIPSC amplitude was decreased by about 20% in neurons from 1 KO mice compared to WT littermates, while the mean mIPSC decay time increased by ~59% in 1 KO interneurons (Fig. 1, A-H, Table 1) and by ~44% in 1 KO pyramidal cells (Fig. 2, A-H, Table 1).

View larger version (23K): [in this window] [in a new window]   Fig. 1. Characteristics of miniature inhibitory postsynaptic currents (mIPSCs) in interneurons. A: representative sweeps showing mIPSCs recorded from a CA1 hippocampal interneuron obtained from a wild-type (WT) mouse. B: continuous sweeps showing mIPSCs recorded in a CA1 hippocampal interneuron from an 1 subunit knock-out (1 KO) mouse. C: superimposed averaged mIPSC traces from the neurons in A and B. Note the small reduction in the 1 KO mIPSC averaged amplitude. D: scaled superimposed averaged mIPSC traces from the neurons in A and B. Note the increase in the decay time of the averaged 1 KO mIPSC trace. E: bar graph showing the averaged mIPSC amplitude for both WT (n = 19) and 1 KO (n = 10) interneurons. F: bar graph showing the averaged w for WT and 1 KO interneurons. * P < 0.05 (unpaired t-test) compared to WT. G: cumulative probability curves for mIPSC amplitude in WT interneurons and 1 KO interneurons; the median amplitudes were 23.4 and 19.8 pA, respectively, and the populations were significantly different (P < 0.001). Total number of events analyzed was 927 in WT interneurons and 462 in 1 KO interneurons. H: cumulative probability curves for mIPSC decay times in WT interneurons and 1 KO interneurons; the median ws were 20.2 and 34.5 ms, respectively, and the populations were significantly different (P < 001).

View larger version (25K): [in this window] [in a new window]   Fig. 2. Characteristics of mIPSCs in pyramidal cells. A: representative sweeps showing mIPSCs recorded from a CA1 hippocampal pyramidal cell obtained from a WT mouse. B: continuous sweeps showing mIPSCs recorded in a CA1 hippocampal pyramidal cell from an 1 KO mouse. C: superimposed averaged mIPSC traces from the neurons in A and B. Note the large reduction in the 1 KO mIPSC amplitude. D: scaled superimposed averaged mIPSC traces from the neurons in A and B. Note the increase in the decay time of the averaged 1 KO mIPSC trace. E: bar graph showing the averaged mIPSC amplitude for both WT (n = 14) and 1 KO (n = 15) pyramidal cells. F: bar graph showing the averaged w for WT and 1 KO pyramidal cells. * P < 0.05 compared to WT. G: cumulative probability curves for mIPSC amplitude in WT pyramidal cells and 1 KO pyramidal cells; the median amplitudes were 21.0 and 17.2 pA, respectively, and the populations were significantly different (P < 0.001). Total number of events analyzed was 665 in WT and 647 in 1 KO pyramidal cells. H: cumulative probability curves for mIPSC decay times in WT and 1 KO pyramidal cells; median decay times were 23.2 and 32.2 ms, respectively, and populations were again significantly different (P < 0.001).

The observed mIPSCs were always outward and were blocked by 20 µM bicuculline, indicating that they were GABA_A-R mediated (Fig. 3). The measured reversal potential of the average mIPSC was 95 mV. The I-V curve shows outward rectification that results from the large chloride gradient that was present between the intra- and extracellular solutions as predicted by the Goldman-Hodgkin-Katz current equation (Hille 2001).

View larger version (25K): [in this window] [in a new window]   Fig. 3. mIPSC reversal potential and block by bicuculline. A: individual traces of mIPSCs in a single interneuron recorded at different holding potentials; the holding potential is indicated to the left of each trace. B: plot of average mIPSC amplitude versus membrane holding potential. Each point represents 1-4 interneurons, and 3-120 mIPSCs were averaged from each cell; near the calculated Nernst reversal potential for chloride (90 mV), very few events were detectable. The data were fit (solid line) using a polynomial equation of the form y = y0 + ax + bx2 + cx3. C: bicuculline (20 µM) completed blocked mIPSCs, confirming that they were mediated by GABAA-Rs.

Effect of zolpidem on interneurons

We next examined the effect of zolpidem on mIPSCs. Zolpidem, an imidazopyridine that is structurally unrelated to benzodiazepines but binds to the benzodiazepine site on the GABA_A-R, has a high affinity for GABA_A-Rs containing the 1 subunit (Pritchett and Seeburg 1990). In functional studies of recombinant receptors, zolpidem potentiates GABA-evoked whole-cell currents generated by 1-containing GABA_A-Rs at nanomolar concentrations (Table 2).

Zolpidem (30 nM) had no appreciable effect on mIPSC amplitude. At a concentration of 300 nM, zolpidem significantly increased mIPSC amplitude in WT interneurons compared to KO interneurons (Fig. 4A, Table 3). Zolpidem (300 nM) also significantly increased mIPSC frequency in interneurons from WT but not KO mice.

View larger version (18K): [in this window] [in a new window]   Fig. 4. Zolpidem modulation of mIPSC properties in interneurons and pyramidal cells. A and B: effect of zolpidem on amplitude and decay time in interneurons. A: zolpidem (30 nM) did not significantly change mIPSC amplitude in interneurons from either WT or 1 KO mice. Zolpidem (300 nM), however, significantly increased the average mIPSC amplitude in WT but not 1 KO interneurons (* P < 0.05 compared to WT, unpaired t-test). B: zolpidem increased w in both WT and 1 KO interneurons. Both concentrations of zolpidem significantly prolonged w in WT interneurons while only 300 nM zolpidem prolonged w in interneurons from 1 KO mice (§ P < 0.05 compared to control, one-sample t-test). C and D: effect of zolpidem on amplitude and decay time in pyramidal cells. C: zolpidem (30 nM) did not significantly change mIPSC amplitude in pyramidal cells. In contrast, 300 nM zolpidem significantly increased mIPSC amplitude in WT pyramidal cells compared to baseline, but the difference was not significant compared to the effect in pyramidal cells from 1 KO mice. D: zolpidem, at 30 and 300 nM, increased w in pyramidal cells from both WT and 1 KO mice compared to baseline. Although zolpidem (at both concentrations) increased w to a greater extent in WT pyramidal cells than in 1 KO pyramidal cells, the difference between those two groups was not significant. For 300 nM zolpidem, the observed increase in w in WT pyramidal cells was significantly less than the increase observed in WT interneurons (P < 0.05, 1-way AVOVA).

At a concentration of 30 nM, zolpidem increased the current decay time in WT interneurons by 30%. In the presence of 300 nM zolpidem, the current decay time in WT interneurons increased by ~93%, and this is significantly greater than the ~41% increase seen in interneurons from KO mice (Fig. 4B, Table 3).

Effect of zolpidem on pyramidal cells

As was the case for interneurons, 30 nM zolpidem had no appreciable effect on pyramidal cell mIPSC amplitude. Zolpidem (300 nM) increased mIPSC amplitude in WT, but not 1 KO pyramidal cells (Fig. 4C, Table 3). Zolpidem (300 nM) also increased mIPSC frequency in WT pyramidal cells but not 1 KO pyramidal cells.

In contrast to the clear effect on the mIPSC decay time observed in interneurons, zolpidem had less of an effect on the mIPSC decay time measured in 1 KO pyramidal cells. Zolpidem, at 30 and 300 nM, increased the current decay time in both WT and 1 KO pyramidal cells, but the difference between the two groups was not significant (Fig. 4D, Table 3).

Nonstationary fluctuation analysis

To further define the contribution of the 1 subunit to synaptically localized receptors, peak-scaled nonstationary fluctuation ("noise") analysis was performed. In Fig. 5, data are presented from a single CA1 interneuron obtained from either a WT (Fig. 5, A and B) or 1 KO (Fig. 5C) mouse. Note the large increase in peak current in the presence of zolpidem in the WT interneuron (Fig. 5B); this effect was markedly reduced in the 1 KO interneuron (Fig. 5C). The single channel conductance () estimate was the same in all cell groups (Table 4), and the pooled unitary conductance for all WT control mIPSCs was 23.7 ± 0.9 pS (1,437 mIPSCs from 54 cells). Zolpidem had no effect on unitary conductance.

View larger version (19K): [in this window] [in a new window]   Fig. 5. Peak-scaled nonstationary fluctuation analysis of mIPSCs. A: the average waveform of 22 mIPSCs (solid line) is shown scaled to the peak of an individual mIPSC (dotted line) recorded in an interneuron from a WT mouse. B: plots of variance against mean current for the neuron in A. Closed circles are the average pooled variance for control mIPSCs while open circles are the average pooled variance of mIPSCs recorded in the presence of zolpidem. Solid line is the fitted parabola under control conditions. The following parameters were derived from the fitted curve: unitary current (i) = 0.659 pA, maximum current (I) = 25.8 pA, and the number of channels open at the peak (Np) = 37.8. Unitary conductance () was calculated using an electromotive driving force of 35 mV based on the membrane holding potential and the measured average reversal potential; for this cell,  = 18.8 pS. Dashed line is the fitted parabola for mIPSCs recorded in the presence of 300 nM zolpidem from the same cell. Derived values were i = 0.812 pA, I = 45.1 pA, Np = 55.9, and  = 23.2 pS. Note the large increase in both I and Np. C: plots of variance against mean current for a neuron obtained from an 1 KO mouse. Note the change in scale relative to B. Solid line is the fitted parabola for mIPSCs recorded under control condition while dashed line is the fitted parabola for mIPSCs recorded in presence of 300 nM zolpidem. For the control parabola, the derived values were i = 0.842 pA, I = 13 pA, Np = 14.2, and  = 28.1 pS; while in the presence of zolpidem, the values were i = 0.884 pA, I = 19.8 pA, Np = 21.4, and  = 18.7 pS. Note the relatively small increase in I and Np in this cell compared to the changes observed in the WT interneuron.

Estimates of the number of channels open at the peak of the mIPSC were also obtained using noise analysis. In 1 KO interneurons, there was a significant decrease in the number of open channels at the peak of the mIPSC (Table 4). In 1 KO pyramidal cells, there was nearly a 20% decrease in the number of open receptors compared to WT; this was in excellent agreement with the ~18% reduction seen in the mean mIPSC amplitude in 1 KO compared to WT pyramidal cells.

Consistent with its effects on mIPSC amplitude, 300 nM zolpidem significantly increased the number of open channels in neurons from WT mice (Fig. 5B, Table 4). In WT interneurons, 300 nM zolpidem increased the number of open channels by ~36%; this is in excellent agreement with the ~34% zolpidem-induced increase in the mean mIPSC amplitude seen in these cells (Table 3). In WT pyramidal cells, zolpidem again significantly increased the number of open channels.

In 1 KO interneurons, 300 nM zolpidem did not alter the number of open receptors, and this agrees with the lack of a zolpidem effect on mIPSC amplitude in these cells (Fig. 4, Table 3). In 1 KO pyramidal cells, however, 300 nM zolpidem produced a significant increase in the number of open channels (Table 4).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Summary of observations

We examined mIPSCs in CA1 hippocampal interneurons and pyramidal cells in WT and 1 KO mice. mIPSCs recorded from interneurons and pyramidal cells obtained from 1 KO mice were detected less frequently, were smaller in amplitude, and were significantly slower in the decay phase compared to mIPSCs recorded in neurons from WT mice. The 1-selective ligand zolpidem increased mIPSC amplitude and prolonged the current decay time in both WT interneurons and WT pyramidal cells but had a markedly reduced effect on those parameters in neurons from 1 KO mice. Nonstationary noise analysis indicated that 1) neurons from WT animals had more synaptically localized GABA_A-Rs than neurons from 1 KO animals and 2) application of 300 nM zolpidem led to a large increase in the number of open channels in neurons from WT animals without altering estimates of single channel conductance.

Kinetics of 1 containing GABA_A-Rs

1 subunit expression influences the decay time of synaptic currents mediated by GABA_A receptors. Synaptic GABA_A receptor-mediated currents recorded in neurons from WT mice decayed more rapidly than those recorded in neurons from 1 KO mice. This suggests that the loss of the 1 subunit in synaptically localized receptors leads to a population of synaptic receptors with slower deactivation kinetics, such as those primarily containing the 2 subunit (Okada et al. 2000). This is consistent with results obtained from transfected HEK cells, in which GABA receptors consisting of 112 subunits had deactivation kinetics that were ~6.5 times faster than receptors consisting of 212 (Lavoie et al. 1997).

The current decay time constants reported here in hippocampal neurons in WT mice are slower than the weighted decay time constant reported in cerebellar granule cells (_w ~ 10 ms; Vicini et al. 2001) but are similar to the unweighted decay time constant in laterodorsal thalamic neurons ( ~ 16 ms; Okada et al. 2000) and CA1 pyramidal cells ( ~ 18 ms; Banks and Pearce 2000). Synaptic receptors in all three cell groups have now been shown to contain the 1 subunit. Differences in the mIPSC decay time constant recorded in neurons from separate anatomic regions may reflect underlying differences in the degree to which the 1 subunit is incorporated into synaptic GABA_A-Rs; variations in the temperature at which the recordings were made are unlikely to be the cause of those differences (see, for example, Thompson et al. 1985) as all recordings were made at room temperature.

Zolpidem modulation of -containing GABA_A-Rs

Zolpidem increased mIPSC amplitude and current decay time in WT interneurons and pyramidal cells. Receptors containing the 1 subunit have a binding affinity for [³H]zolpidem that is one to three orders of magnitude greater than that for other  subunit-containing receptors (Pritchett and Seeburg 1990). The differences in affinity are reflected in differences in the functional EC₅₀ for zolpidem at different receptor configurations (Table 2), (KA Wafford, unpublished observations). The pronounced effect of zolpidem on mIPSC amplitude and current decay time recorded in neurons from WT mice is most likely mediated by synaptically localized receptors containing the 1 subunit, although 2 subunit-containing receptors may also have contributed to the observed effects, especially at 300 nM. The persistent effects of zolpidem seen in neurons from 1 KO mice were likely mediated by receptors containing the 2 subunit, rather than the 3 subunit given the differences in the EC₅₀ of zolpidem for heteroligomers containing those subunits (Wafford et al. 1993; KA Wafford, unpublished observations).

Baseline mIPSC frequency in neurons from WT mice was ~2.5 Hz, and zolpidem increased both frequency and amplitude in those cells. In contrast, zolpidem had no effect on either mIPSC frequency or amplitude in neurons from 1 KO mice. The observed increase in mIPSC frequency recorded in WT neurons is likely to reflect an increase in event detection (rather than an increase in quantal release) as a consequence of increased mean mIPSC amplitude.

In neurons from 1 KO mice, mIPSC frequency was significantly decreased compared to that recorded in WT mice. Cumulative probability plots demonstrated a significant decrease in mIPSC amplitude in 1 KO interneurons and pyramidal cells (compared to WT). In addition, noise analysis demonstrated a significant decrease in the number of receptors open at the peak of the mIPSC in cells from 1 KO compared to WT mice. We suggest that a loss in the number of receptors at the synapse in 1 KO mice accounts for smaller mIPSCs, leading to decreased event detection and an apparent decrease in mIPSC frequency.

Noise analysis and synaptic GABA_A-Rs

Nonstationary fluctuation analysis revealed a number of differences and similarities between mIPSCs recorded in WT interneurons and pyramidal cells. Synapses on interneurons appeared to contain slightly more GABA_A-Rs contributing to the peak of the mIPSC than did synapses on WT pyramidal cells (~37 vs. ~28, respectively), and this parallels the larger mIPSC amplitudes in interneurons.

There was a marked decrease in open channel number in neurons from 1 KO mice. The loss of open channels is of the same magnitude as the decrease in mIPSC amplitude (Table 1). Interestingly, the decrease in the number of synaptic receptors is less than the total loss of hippocampal GABA_A receptors in mice lacking the 1 subunit (reportedly on the order of 53%; Sur et al. 2001).

Estimates of single channel conductance in all WT and 1 KO neurons were the same, indicating that channel conductance was independent of 1 subunit expression (see also Verdoorn 1990). The pooled conductance for all cells was ~24 pS and is similar to that reported in dentate granule cells (De Koninck and Mody 1994), internal cerebellar granule cells (Brickley et al. 1999), cortical neurons (Perrais and Ropert 1999), and cerebellar stellate cells (Nusser et al. 1997).

Binding and postsynaptic receptor occupancy

Zolpidem has been used to assess the degree to which postsynaptic GABA_A-Rs are saturated following quantal release of transmitter (Hájos et al. 2000; Perrais and Ropert 1999). We observed that 300 nM zolpidem increased the average mIPSC amplitude by ~34% in WT interneurons and by ~17% in WT pyramidal cells, and the increase was substantially less in neurons from 1 KO mice. The zolpidem-induced increase in mIPSC amplitude reported here likely reflects an increase in the number of open receptors, as zolpidem does not appear to change the probability of channel opening nor does it alter conductance (De Koninck and Mody 1994; Perrais and Ropert 1999; this study). In 1 KO pyramidal cells, 300 nM zolpidem increased the number of open channels and increased mIPSC amplitude (compared to baseline), but this was not the case for 1 KO interneurons (Figs. 3A and 4A, Table 4).

Zolpidem has also been used to assess receptor affinity for GABA (Perrais and Ropert 1999). The  subunit controls the apparent affinity of the receptor for GABA, with the relative affinities in heterologous expression systems being 3 < 1 < 5  6  2 (Ebert et al. 1994; Gingrich et al. 1995; Krasowski et al. 1997; Levitan et al. 1988; Verdoorn 1994). Zolpidem (30 nM) significantly increased the mIPSC current decay time, not amplitude, in WT interneurons and pyramidal cells; this suggests that zolpidem may increase GABA binding to a small number of receptors with a high affinity for GABA, such as those containing the 2 subunit, without opening additional receptors.

Subunit contribution to synaptically localized GABA_A receptors

Our data indicate that CA1 hippocampal interneurons and pyramidal cells express populations of synaptically localized GABA_A receptors that contain the 1 subunit. This conclusion is consistent with results obtained using immunogold labeling (Nusser et al. 1996). Interneurons had nearly twofold greater increases in mIPSC amplitude and current decay time in response to 300 nM zolpidem than did WT pyramidal cells, suggesting that WT interneurons contain a greater percentage of receptors expressing the 1 subunit than WT pyramidal cells. This view is supported by immunohistochemical data demonstrating that interneurons had a greater density of synaptic 1 protein than did pyramidal cells (Klausberger et al. 2002).

The 2 subunit is likely to be included in hippocampal GABA_A-Rs based on high levels of mRNA expression (Laurie et al. 1992; Wisden et al. 1992), protein detection (Fritschy and Möhler 1995; Pirker et al. 2000; Sperk et al. 1997), and localization to a subset of synapses on pyramidal cell somata and dendrites and most synapses on pyramidal cell axon initial segments (Nusser et al. 1996; see also Brünig et al. 2002).

Hippocampal neurons may express synaptic receptors containing 3 (Brünig et al. 2002) and 4 (Banks et al. 1998; but see Thomson et al. 2000) subunits despite limited expression in this region (Pirker et al. 2000; Sperk et al. 1997). Hippocampal neurons also express GABA_A receptors containing the 5 subunit; some of this protein may be incorporated into synaptic receptors (Pawelzik et al. 1999; Collinson et al. 2002), but most 5 subunit-containing GABA_A-Rs appear to be extrasynaptic (Brünig et al. 2002).

In 1 KO mice, 2, 3, and/or 5 subunits are again possible candidates for inclusion in the postsynaptic receptor. Given that 300 nM zolpidem prolonged the current decay time even in KO mice (Figs. 3 and 4), it is unlikely that the postsynaptic receptor in 1-null mice contains the 5 subunit in significant amounts since it has a K_i for zolpidem in excess of 15 µM (Pritchett and Seeburg 1990) and is insensitive to zolpidem at the concentrations used in this study. Excluding the 5 subunit, that leaves the 2 and 3 subunits as the most likely  subunits to be expressed in synaptically localized GABA_A-Rs in the hippocampus in KO mice.

Conclusions

CA1 interneurons and pyramidal cells from WT mice express GABA_A-Rs containing the 1 subunit; these receptors are not saturated following quantal release of transmitter at room temperature. In 1 KO mice, the synaptic receptor population likely contains a significant proportion of 22/32 and/or 32/32 heteroligomers. Anatomic data have indicated that the 1 subunit is incorporated into synaptic GABA_A-Rs expressed by CA1 pyramidal cells (Klausberger et al. 2002; Nusser et al. 1996); the physiological data now confirm a contribution of the 1 subunit to synaptically localized GABA_A-Rs expressed by both hippocampal pyramidal cells and interneurons.