Abstract
Although GABA type A receptors (GABAARs) in principal cells have been studied in detail, there is only limited information about GABAARs in interneurons. We have used the patch-clamp technique in acute rat hippocampal slices in combination with single-cell PCR to determine kinetic, pharmacological, and structural properties of dentate gyrus basket cell GABAARs. Application of 1 mm GABA (100 msec) to nucleated patches via a piezo-driven fast application device resulted in a current with a fast rise and a marked biexponential decay (time constants 2.4 and 61.8 msec). This decay could be attributed to strong receptor desensitization. Dose–response curves for the peak and the slow component yielded EC50 values of 139 and 24 μm, respectively. Zn2+ caused a marked blocking effect on both the peak and the slow component via a noncompetitive mechanism (IC50 values of 8 and 16 μm). This led to an acceleration of the slow component as well as a prolongation of recovery from desensitization. Zn2+ sensitivity was suggested to depend on the absence of γ-subunits in GABAARs. To test this hypothesis we performed single-cell reverse transcription PCR that revealed primarily the presence of α2-, β2-, β3-, γ1-, and γ2-subunit mRNAs. In addition, flunitrazepam increased the receptor affinity for its agonist, indicating the presence of functional benzodiazepine binding sites, i.e., γ-subunits. Thus, additional factors seem to co-determine the Zn2+ sensitivity of native GABAARs. The modulatory effects of Zn2+on GABAAR desensitization suggest direct influences on synaptic integration via changes in inhibition and shunting at GABAergic synapses.
- rat
- GABA receptor
- kinetics
- desensitization
- deactivation
- pharmacology
- benzodiazepines
- flunitrazepam
- Zn2+
- patch-clamp in situ
- nucleated patch
- fast application
- single-cell PCR
- subunits
- mRNA
Inhibitory GABAergic synapses of interneurons counterbalance the excitatory discharge pattern of many neuronal circuits in the CNS. The GABAergic interneurons of the hippocampus mediate both feedforward and feedback inhibition as well as disinhibitory inputs to the principal neurons (Freund and Buzsáki, 1996). Although much is known about kinetic (Edwards et al., 1990; Celentano and Wong, 1994; Jones and Westbrook, 1995; Draguhn and Heinemann, 1996), pharmacological (Westbrook and Mayer, 1987; Buhl et al., 1996; Martina et al., 1996), and structural (Laurie et al., 1992; Wisden et al., 1992; Fritschy et al., 1994; Fritschy and Möhler, 1995) properties of GABAA receptors (GABAARs) on hippocampal principal cells, only a few studies have dealt with interneuron GABAARs (Fritschy et al., 1994; Gao and Fritschy, 1994; Fritschy and Möhler, 1995).
The kinetic properties of GABAAR activation have been studied in a number of areas using either application techniques on excised patches or evaluation of IPSCs. The findings were contradictory with respect to the number of exponential functions necessary to describe the current decay (Edwards et al., 1990; Draguhn and Heinemann, 1996). Developmental regulation and cell type-specific and subunit-specific dissimilarities have been shown to account for the differences in kinetic parameters (Verdoorn et al., 1990; Puia et al., 1994; Draguhn and Heinemann, 1996; Tia et al., 1996). In the present study we have used a fast-application system to nucleated patches to minimize problems in space-clamp or drug exchange rate for the study of the kinetics and Zn2+ sensitivity of GABA-mediated currents.
GABAAR activity is modulated via several substances, such as benzodiazepines, barbiturates, and steroids (for review, see Kaila, 1994; Thompson, 1994). Another modulator, Zn2+(Smart, 1992), acts heterogeneously on GABAARs. Although the receptors of prepyriform cortical neurons are almost insensitive to this cation (Smart and Constanti, 1983), hippocampal pyramidal neurons exhibit a pronounced Zn2+ block (Westbrook and Mayer, 1987). In recombinant GABAARs, Zn2+ insensitivity has been correlated with the presence of γ-subunits (Draguhn et al., 1990; Smart et al., 1991) (but see White and Gurley, 1995). The effects of Zn2+ on neurotransmitter receptors might be of functional importance, because Zn2+ is present in synaptic vesicles, e.g., in mossy fiber boutons (Haug, 1967), and is released during synaptic activation. Thereafter, it may reach local concentrations of up to 300 μm (Assaf and Chung, 1984) and may modulate inhibitory and excitatory receptors (Westbrook and Mayer, 1987). Furthermore, synaptically released Zn2+ from sprouting mossy fibers (Laurberg and Zimmer, 1981) might play a role under pathophysiological conditions (Buhl et al., 1996).
Because information about the subunits constituting the GABAARs of interneurons is scarce, we have combined electrophysiological techniques with single-cell reverse transcription (RT)-PCR experiments to correlate functional and structural properties of basket cell GABAARs. Particularly, we were interested in the expression of β- and γ-subunits, which have been shown to be important for receptor desensitization, benzodiazepine responsiveness, and Zn2+ sensitivity in recombinant cells (Pritchett et al., 1989; Draguhn et al., 1990; Verdoorn et al., 1990).
MATERIALS AND METHODS
Brain slice preparation and basket cell identification. Transverse slices (300-μm-thick) of the hippocampus were cut from the brains of 15- to 20-d-old Wistar rats using a Vibratome (Campden, Loughborough, England). Dentate gyrus basket cells were visualized by infrared differential interference contrast (IR-DIC) videomicroscopy (Stuart et al., 1993) using a Newvicon camera (C2400; Hamamatsu, Hamamatsu City, Japan) and an infrared filter (RG9; Schott, Mainz, Germany) mounted on an upright microscope (Axioskop FS; Zeiss, Oberkochen, Germany). They were identified by their pyramidal morphology, by their location at the border between the granule cell layer and the hilus, and by their high-frequency firing of action potentials generated after sustained membrane depolarization in the current-clamp mode (Koh et al., 1995).
Patch-clamp recording and fast drug application. Patch pipettes were pulled from borosilicate glass tubing (2.0 mm outer diameter, 0.5 mm wall thickness; Hilgenberg, Malsfeld, Germany). When filled with internal solution, they had a resistance of 2.0–2.5 MΩ. Only neurons with resting potentials negative to −60 mV were used. To obtain nucleated patches (Sather et al., 1992), negative pressure (100–200 mbar) was applied during the withdrawal of the patch pipette.
Functional properties of GABAARs were investigated using fast application of agonists and modulators (Colquhoun et al., 1992;Jonas, 1995) to nucleated patches. The double-barrel application pipette was made from theta glass tubing (2 mm outer diameter, 0.3 mm wall thickness, 0.12 mm septum; Hilgenberg), and the piezo-electric element used was a PI-245.50 (Physik Instrumente, Waldbronn, Germany) driven by a P-270 high-voltage amplifier. The perfusion rate was 50–70 μl min−1. The exchange time (20–80%), measured with an open patch pipette during a change between Na+-rich and 10%-Na+-rich solution, was 150 μsec. Fast application experiments were started as soon as possible after patch excision (1–2 min after access to the cell interior was obtained). Agonist pulses were applied every 5–8 sec. If modulating substances were applied together with GABA, these were also included in the control solution of the application tool. After completion of the experiment, the patch was blown off and the zero-current potential was measured; it never exceeded 3 mV.
Membrane currents were recorded using an Axopatch 200A amplifier (Axon Instruments, Foster City, CA). Currents were filtered at 5 kHz using the internal four-pole low-pass Bessel filter of the amplifier. Data were digitized and stored on-line using a CED 1401plus interface (CED, Cambridge, England) connected to a personal computer. The sampling frequency was twice the filter frequency. All recordings were made at room temperature (20–24°C). The traces shown represent averages from two to four sweeps (recovery from desensitization) or averages from 15–30 sweeps (all other traces).
Analysis. The decay time constants of the GABAAR-mediated current were determined by least-squares fit of the decay phase after the peak current. The desensitization time course was evaluated using a fitting interval of 100 msec for 100 msec pulses; the residual current at that time might reflect incomplete desensitization. The fitting interval of the deactivation time course (1 msec pulses), in contrast, was extended to the return of the current to baseline. The value of the interpolated reversal potential (Vrev) and chord conductance ratios (gΔV1/gΔV2;g+60mV/g−60mV) were calculated using the values of the fittedI–V curve at Vrev + ΔV1, 2.
Values of the dose–response curves for the peak and the slow component were obtained as follows. For the peak, the maximal amplitude of the current response was used, whereas the amplitude of the slow component was the sum of the current described by the decay time constant τ2 and the residual current at the end of the fitting interval. If currents induced by low GABA concentrations could be well fitted by a monoexponential function, the peak amplitude of this current was used for the construction of the dose–response curve of the slow component. The values from different patches were normalized to the value obtained with 1 mm GABA, plotted semilogarithmically, and fitted by the Hill equation: where I is the amplitude of the current,Imax is the peak of saturating GABA current,c is the concentration of GABA, K is the half-maximum concentration, and n is the Hill coefficient. The dose–response curves were normalized toImax.
All numerical values denote mean ± SEM. Statistical significance was assessed by one-way ANOVA at the 0.01 significance level.
Chemicals and solutions. Slices were superfused continuously with a physiological extracellular solution containing (in mm): 125 NaCl, 25 NaHCO3, 2.5 KCl, 1.25 NaH2PO4, 2 CaCl2, 1 MgCl2, 25 glucose, bubbled with 95% O2and 5% CO2. The HEPES-buffered Na+-rich external solution used for fast application contained (in mm): 135 NaCl, 5.4 KCl, 1.8 CaCl2, 1 MgCl2, 5 HEPES, pH adjusted to 7.2 with NaOH. The internal solution used in fast application experiments contained (in mm): 140 KCl, 10 EGTA, 2 MgCl2, 2 Na2ATP, 10 HEPES, pH adjusted to 7.3 with KOH. The intracellular solution used in the PCR experiments contained (in mm): 140 KCl, 5 EGTA, 3 MgCl2, 5 HEPES, pH adjusted to 7.3 with KOH.
All drugs and chemicals were from Sigma (St. Louis, MO) or Merck (Darmstadt, Germany). Stock solutions of 1 m GABA, 300 mm ZnCl2, and 10 mm(−)bicuculline methiodide were prepared in HEPES-buffered Na+-rich external solution; 1 mmflunitrazepam was prepared in DMSO. Serial dilutions provided the final concentrations given below.
Single-cell RT-PCR. Patch pipettes used for the RT-PCR experiments had tip outer diameters of ∼2–3 μm, corresponding to resistances of 0.8–1.4 MΩ when filled with intracellular solution. The glass tubing for the pipettes was heated before use (250°C, 4 hr), and the intracellular solution was autoclaved; the silver wire connected to the patch electrode was chlorided before each experiment. Cytoplasm and nucleus were harvested into the patch pipette under visual control. Cells were used only when the seal remained intact until the very end of the aspiration and were rejected whenever debris remained attached to the outside of the patch pipette. Controls were performed by applying positive pressure to the pipettes, advancing them into the tissue, and using their contents for RT-PCR. The contents of the patch pipettes were expelled into reaction tubes using a valve-controlled pressure system (N2, 4 bar). Subsequently, reverse transcription was performed (Monyer and Jonas, 1995).
Coamplification of GABAAR-subunits was performed by nested hot-start PCR using primer sets for the α-, β-, and γ-subunits. The α6-subunit was not tested because of its selective localization in cerebellar granule cells. The sequences of the primers for the first round of PCR were α5′, 5′-TGGAC(TC)CC(AT)GA(TC)AC(ACT)TT(TC)TT-3′; α3′, 5′GC(AGTC)AT(GA)AACCA(GA)TCCATGGC-3′; β5′, 5′-CTGGATGA(GA)CAAAACTGCAC-3′; β3′, 5′-AC(AG)AA(CG)AC(GA)AA(AG)CA(AC)CCCAT-3′; γ5′, 5′-T(TG)AA(CT)AGCAA(CT)ATGGTGGG-3′; γ3′, 5′-TTGATCCA(AG)AA(AGT)GA(CT)ACCCAGG-3′.
In certain experiments primer pairs for the specific amplification of the α1-subunit were used. For this reaction the above mentioned 3′ primer and the following 5′ primer were used for the first amplification: α1spec., 5′-GGACAGCCCTCCCAAGATGAAC-3′.
The cycling parameters for the first amplification were 94°C for 5 min, 35 cycles (94°C, 30 sec; 53°C, 30 sec; 72°C, 40 sec), and 72°C for 10 min. One microliter of the first-round PCR was reamplified in a second PCR by using a nested 5′ primer and the same 3′ primer: α5′n, 5′-AA(AG)TTTGG(GAC)AG(CT)TATGC(TCA)TA-3′; β5′n, 5′-GATGACAT(TC)GAATTTTACTGG-3′; γ5′n, 5′-CACTGGAT(AC)AC(AGC)AC(GTA)CCCAA-3′. For the specific amplification of the α1-subunit, the α5′-primer was used as the nested primer. The second PCR was performed according to the same program. Both PCR reactions contained a 0.4 mm concentration of each primer, 1.5 mmMgCl2, 0.6 mm dNTPs, and 2.5 U ofTaq DNA polymerase in the buffer supplied by the manufacturer (Life Technologies, Eggenstein, Germany). As previously described (Geiger et al., 1995), PCR conditions and different primer pairs were tested extensively to guarantee that under conditions used in these experiments GABAAR-subunits were amplified with a comparable efficiency. The 3′ and 5′ primer locations span several introns, thus preventing amplification of genomic DNA.
For the identification of parvalbumin (PV)-positive cells, the primers parv.ex3, 5′-CTGCAGACTCCTTCGACCAC-3′, and parv.ex4, 5′-CTTCAACCCCAATCTTGCCG-3′ located in exons 3 and 4, respectively, were used. PCR conditions were as follows: 94°C for 5 min, 40 cycles (94°C, 30 sec; 51°C, 30 sec; 72°C, 30 sec), and 72°C for 10 min. For the second amplification a nested 3′ primer (parv.ex4n, 5′-GCGGCCAGAAGCGTCTTTG-3′) was used.
Analysis of PCR products by Southern blot. After gel electrophoresis, PCR products were denatured by NaOH and transferred to Hybond N+ membranes (Amersham, Braunschweig, Germany). The membranes were hybridized with radiolabeled GABAAR-subunit-specific oligonucleotide probes (α1-SB, 5′-TGAGCGGGCTGGCTCCCTTG-3′; α2-SB, 5′-AGAGTCAGAAGCATTGTAAG-3′; α3-SB, 5′-AGATTTGTTCTTCCCAAGAG-3′; α4-SB, 5′-TGACTTCTCAGGGCCTTTGG-3′; α5-SB, 5′-AGACTTGGTGGAACCATTGG-3′; β1-SB, 5′-TTGGACACCATCTTGTAGTC-3′; β2-SB, 5′-ATGACAATGCAGTCACGGGA-3′; β3-SB, 5′-CTGGAGACCAGACGGTGCTC-3′; γ1-SB, 5′-CGGAGATTGTGTGAGAGAT-3′; γ2-SB, 5′-GTAGTGAAGACAACTTCTGG-3′; γ3-SB, 5′-CTGCAGATGTAGTCACGAT-3′) overnight at 37°C and were subsequently washed with 0.5× SSC at 55°C. Standard procedures for probe labeling, hybridization, and posthybridization washing were used (Sambrook et al., 1989).
RESULTS
Deactivation and desensitization kinetics
For the kinetic experiments we used a standard concentration of 1 mm GABA, a value that seems to reflect the concentration in the cleft after vesicle fusion (>500 μm) (Jones and Westbrook, 1995). GABA was applied to the nucleated patches for 1 or 100 msec at a holding potential of −60 mV (Fig.1A,B, Table1). The responses to both pulse protocols were characterized by a fast rise (rise time 20–80% of the peak in the range of 295–775 μsec) (Table 1). The current decay could be best fitted by the sum of two exponential functions, described by their decay time constants, τ1 and τ2, and relative amplitudes, A1 and A2 (Table 1). The amplitude fraction of A1 was obtained by dividing A1 by (A1 + A2) (Table 1). In 11 patches both short and long GABA pulses were applied, and comparison of the above-mentioned parameters gave no significant differences between the two pulse protocols (p> 0.1). The peak current amplitude was variable for 1 as well as 100 msec pulses (range, 169–4772 and 198–5177 pA, respectively). The residual current at the end of 100 msec pulses of 1 mm GABA had an amplitude of 12.4 ± 3.6% of the peak amplitude. In most experiments, no differences between deactivation and desensitization kinetics were seen (current decay did not change its time constant after the end of the GABA pulse) (Fig. 1B).
Application of two 1 msec pulses of 1 mm GABA with variable interpulse intervals resulted in paired-pulse depression reflecting the degree of desensitization. The second pulse gained full amplitude after an interpulse interval of ∼5 sec (100.5 ± 3.2%) and was half-maximal after ∼1 sec (50.4 ± 2.8%; n = 8) (Fig. 1C).
The GABAAR-mediated nature of the currents under study was verified by their reversal potential under symmetrical chloride and their sensitivity to the competitive antagonist bicuculline. Application of 1 mm GABA for 100 msec at holding potentials between −80 and 80 mV resulted in an inwardly rectifying dose–response curve (not shown). The reversal potential (Vrev) was 4.9 ± 1.2 mV, and the average chord conductance ratio at Vrev + ΔV1,2 (gΔV1/gΔV2;g+60mV/g−60mV) was 0.68 ± 0.03 (n = 8). Although τ2 and the amplitude ratio were voltage-independent, τ1 showed a significant (p < 0.01) decrease at more depolarized potentials (ratio τ1at 60 mV divided by τ1 at −60 mV). Coapplication of bicuculline blocked a 1 msec pulse of 1 mm GABA with an IC50 of ∼300 nm (n = 2), whereas 1 mm GABA with bicuculline concentrations as high as 30 μm for 100 msec resulted in the unbinding of the competitive antagonist (cf. Clements, 1996) and a marked GABA-mediated current (not shown).
GABA dose–response curve and predesensitization
Different GABA concentrations (range 10 μm to 10 mm) were applied for 100 msec to the same nucleated patch. Low concentrations resulted in current responses with small amplitudes and amplitude fractions, as well as long rise times and decay constants (Fig. 2A). With GABA concentrations ≤100 μm, only a single decay time constant could be detected. The concentration dependence of the peak and the slow component are shown in Figure 2B and Table 1. The decay constant obtained for GABA concentrations ≤100 μm is similar to τ2 of the biexponential decay found with saturating concentrations (Table 1). Thus, the peak amplitude was used for the construction of the dose–response curve of the slow component under low concentration conditions. Peak and slow component displayed different EC50 values and Hill coefficients, likely reflecting different affinities for GABA (Fig.2B, Table 1).
In predesensitization experiments, low GABA concentrations (range, 10 nm to 10 μm) were applied to the nucleated patch before a 100 msec test pulse of 1 mm GABA (Fig.2C). Plotting the current elicited by this test pulse against the preapplied GABA concentration yielded half-maximal predesensitizing concentrations of 642 and 629 nm, respectively, for peak and slow component (Fig. 2D, Table 1). These values were not significantly different (one-way ANOVA;p > 0.5)
Zn2+ block
Coapplication of 30 μm Zn2+ with 1 mm GABA (100 msec; n = 11) reduced the amplitude of the peak to 27.2 ± 3.9% and that of the slow component to 34.1 ± 4.4% of the control values, respectively. The decay time constants τ1 and τ2, in contrast, behaved completely differently in the presence of Zn2+: whereas τ1 was unchanged (122.9 ± 9.9% of control), τ2 was markedly reduced (33.7 ± 1.8% of control), resulting in the disappearance of the GABA-mediated current within 30 msec after the beginning of the pulse (Figs. 3A,C,4A).
Zn2+ in concentrations between 100 nmand 1 mm was coapplied with 1 mm GABA for 100 msec (Fig. 3A). To assess the magnitude of block, a 1 mm GABA test pulse was first applied without Zn2+, and the following responses were normalized to this one. A dose–response curve for the peak and the slow component yielded IC50 values of 8 ± 3 and 16 ± 7 μm, respectively. They were not significantly different (p > 0.1) (Fig. 3B, Table 1). To test the mechanism of Zn2+ block, different GABA concentrations (in the range of 10 μm to 10 mm) were coapplied with the almost saturating Zn2+ concentration of 100 μm and normalized to the response to 1 mm GABA alone (Fig.3C). The dose–response curve of the peak (Fig.3D) showed an unchanged EC50 for GABA (167 ± 23 vs 139 ± 29 μm under control conditions) (Table 1), whereas the Imax was reduced to 19% of the control. These data indicate a noncompetitive blocking mechanism.
To further investigate the possibility of a direct channel block by Zn2+, 1 mm GABA was applied at holding potentials of −60 and 60 mV, either alone or together with 30 μm Zn2+. The inward rectification of the current–voltage relationship found in control conditions was present also in the presence of Zn2+. The fraction of the currents blocked by Zn2+ was the same at both holding potentials (ratio block −60 mV/60 mV = 1.06 ± 0.02,n = 5; mean ± SEM; not shown). Therefore, the Zn2+ block is voltage-insensitive as shown previously by Mayer and Vyklicky (1989) and Draguhn et al. (1990), and these data do not suggest the possibility of a voltage-sensitive channel block.
Subsequently, we tested whether the blocking effect of Zn2+ on τ2 was related to a change of the recovery from desensitization. Therefore, the paired-pulse paradigm from Figure 1C was used to estimate the degree of desensitization. GABA at 1 mm (n = 4), 1 mm plus 30 μm Zn2+(n = 4), and 30 μm (n = 3) was applied for 20 msec each, with increasing interpulse intervals (Fig. 4A). Application of 30 μm GABA for 1 msec resulted in very small currents; therefore longer pulses were used. Although the recovery from desensitization was complete after an interpulse interval of ∼5 sec for each of the three experiments (97.3 ± 0.6% of the first pulse amplitude for 1 mm GABA, 95.3 ± 4.3% for 1 mm GABA plus 30 μm Zn2+, and 96.0 ± 0.6% for 30 μm GABA), the second pulse evoked half-maximal amplitude after different times (Fig.4B). With a 1 sec interpulse interval, 1 mm GABA reached 53.3 ± 1.6% of the test response (identical to the 50.4 ± 2.8% found in Fig. 1C), 1 mm GABA plus 30 μm Zn2+reached 39.0 ± 3.5%, and 30 μm GABA reached 59.3 ± 1.2%. Thus, coapplication of Zn2+slowed the recovery from desensitization (Fig. 4B). This could be attributable to a change of the rate constants in favor of the desensitized states or a disadvantage of the open states. Paired-pulse application of 30 μm GABA, in contrast, showed a faster recovery from desensitization (Fig.4B). The differences for 1 mm GABA plus 30 μm Zn2+, and 30 μmGABA, respectively, in comparison to 1 mm GABA were statistically significant for interpulse intervals between 50 and 700 msec (one-way ANOVA; p < 0.01). The biexponential course of the recovery curve suggests the presence of two mechanisms for the recovery from desensitization. (Figs. 1C,4B).
The high Zn2+ sensitivity of the basket cell GABAARs suggests the possibility that they may not contain γ-subunits (Draguhn et al., 1990; Smart et al., 1991). To check the expression of γ-subunits (i.e., functional benzodiazepine binding sites) (Pritchett et al., 1989) we used flunitrazepam as a pharmacological tool. For this purpose, GABA dose–response curves were obtained in the presence of 1 μm flunitrazepam and compared with the curves shown in Figure 2B. The only significant effect of the coapplication of flunitrazepam was that even responses to GABA concentrations as low as 100 μm showed two decay time constants (Fig.5A, Table 1). All other kinetic parameters were unchanged when application of 100 μm and 1 mm GABA with and without 1 μm flunitrazepam were compared (one-way ANOVA;p > 0.05). This suggests an increase in the affinity of the GABAARs to the agonist that was also reflected in a shift to the left of the dose–response curve for peak and slow component (Fig. 5B, Table 1). The potentiation of GABA in the presence of flunitrazepam was statistically significant for 10 μm and 30 μm GABA (one-way ANOVA;p < 0.01), but not for higher GABA concentrations.
Single-cell expression of GABAAR-subunit mRNA
To determine directly the molecular basis of the previous findings, we used single-cell RT-PCR (Monyer and Jonas, 1995) to detect GABAAR-subunit mRNAs. Reverse transcription of single-cell mRNA was followed by PCR set-up to coamplify the subunits belonging to one subunit family. For this purpose we designed degenerate primers for the α-, β-, and γ-subfamilies, because the divergence at the nucleotide level in the conserved regions did not permit the design of optimal primers for the concomitant amplification of all subunits. All primers were tested extensively on total brain cDNA and on plasmids containing different subunit cDNAs to exclude preferential amplification of subunits. Because the multiplex PCR resulted primarily in the amplification of the β-subunit family, the expression profile of the three subfamilies had to be tested on different sets of basket cells. Cells without any PCR product were rejected, and the cells in which PCR products were obtained (n = 14 cells for the α-subunits; n = 12 for the β-subunits; andn = 12 for the γ-subunits) were analyzed by Southern blot hybridization, using probes specific for each GABAAR-subunit. This resulted predominantly in the expression of the α2-, β2-, β3-, γ1-, and γ2-subunit mRNAs (Table 2). All other subunits were expressed less frequently. For the α-subunits, the percentage of cells in which subunit mRNA was detected was 21% for α1, 86% for α2, 14% for α3, 50% for α4, and 36% for α5. For the β- and γ-subunits the values were 42% for β1, 83% for β2, 100% for β3, 83% for γ1, 100% for γ2, and 58% for γ3. This high abundance of γ-subunits fitted well with the strong modulation of basket cell GABAARs by benzodiazepines.
The α1-subunit protein has been reported to be strongly expressed in parvalbumin (PV)-positive basket cells (Gao and Fritschy, 1994). We detected this subunit in a subset of the basket cells (21%) tested in this study after amplification with the degenerate primers. To confirm these results we designed a specific primer for the α1-subunit, and the PCR analysis of an additional 18 basket cells resulted in the detection of the α1-subunit in 38% of the cells. To test the percentage of PV-positive basket cells, PV primers were designed, and single-cell PCR resulted in the expression of this gene in 22% (n = 2 of 9 tested) of the cells. Hence, these data suggest that the α1-subunit is expressed in PV-positive basket cells.
DISCUSSION
Kinetic properties of GABAA receptors
The present study describes the kinetics of GABAAR-mediated currents in nucleated patches of dentate basket cells and their modulation by Zn2+. In other studies, native GABAARs show multiexponential (Edwards et al., 1990; Celentano and Wong, 1994; Maconochie et al., 1994; Jones and Westbrook, 1995; Galarreta and Hestrin, 1997; Mellor and Randall, 1997b) as well as monoexponential (Otis and Mody, 1992; Puia et al., 1994; Soltesz and Mody, 1994; Draguhn and Heinemann, 1996) decay. In recombinant GABAARs, fast application of saturating GABA concentrations always results in a current response with a biexponential decay. The ratio and time course of decay, in contrast, are related to the presence of different GABAAR-subunits (α1 vs α3, Verdoorn, 1994; Gingrich et al., 1995; β2 vs γ2, Verdoorn et al., 1990). This heterogeneous impression is intensified by the fact that decay kinetics seem to depend on the recording configuration or differences between extrasynaptic versus synaptic receptors (Purkinje cell whole-cell recordings vs fast application to outside-out patches) (Puia et al., 1994). Thus, conditions of improved clamp control and solution exchange might reveal additional information about the kinetic properties.
Desensitization is defined as the current decay in the presence of agonist, whereas the decay after agonist removal is named deactivation. The apparent lack of differences between current responses to 1 or 100 msec GABA pulses in the present study suggests that the fast application of high agonist concentrations forces the GABAAR population into a bound and desensitized state (unbinding becomes slower than reopening) (for kinetic model, see Jones and Westbrook, 1995). Thus, the channels seem to desensitize very strongly if GABA is present for a long time, and they hold onto GABA quite tightly after a short pulse.
The time constants of GABAAR current decay found in basket cells (2.4 and 61.8 msec) are faster than those described in the majority of the previous studies. Only Edwards et al. (1990) (2.2 and 54.4 msec for miniature IPSCs on dentate granule cells) and Galaretta and Hestrin (1997) (5.1 and 62.4 msec for fast application of 1 mm GABA to outside-out patches of neocortical pyramidal neurons) describe similar values. In contrast to these studies, however, we find that the amplitude of the fast component is larger than that of the slow one. Using a paired-pulse protocol, we studied the recovery from desensitization of the two components found with high GABA concentrations. The strong desensitization seen as a large A1/(A1 + A2) fraction is also reflected in the long time needed for complete recovery from desensitization (compare Jones and Westbrook, 1995; Galarreta and Hestrin, 1997; Mellor and Randall, 1997b). Because of the slow desensitization kinetics observed previously, desensitization was thought to be of minor importance for IPSCs. The present results, in agreement with recent reports (Celentano and Wong, 1994; Jones and Westbrook, 1995, 1996; Galarreta and Hestrin, 1997), suggest on the other hand that desensitization might play an important role in determining the IPSC shape.
In basket cells, rise, amplitude, decay, and amplitude fractions are markedly concentration-dependent. An amplification of the current as well as an acceleration of current rise and decay with increasing agonist concentrations is also described by other groups (Celentano and Wong, 1994; Maconochie et al., 1994; Gingrich et al., 1995; Jones and Westbrook, 1995) (but see Galarreta and Hestrin, 1997). At low concentrations, ligand-receptor association is slow and desensitization is in concurrence with activation, resulting in slow rise and decay. At high concentrations, association is fast and activation precedes desensitization. This leads to larger responses with additional transient components. The dose–response curves for peak and slow component yield different affinities for the two components. This has also been observed by Celentano and Wong (1994): the fastest desensitization component had the lowest affinity. The activation of the fast component with its low affinity may be attributable to the use of a fast application system in the present study (also see Galarreta and Hestrin, 1997). The EC50 for the second component is in agreement with previously reported values for the fast component [native receptors: range, 20.0–48.7 μm (Jones and Westbrook, 1995; Mellor and Randall, 1997b); recombinant receptors: 7 and 80 μm (Gingrich et al., 1995)]. Differences in affinities and kinetics therefore may reflect the differences in the speed of agonist perfusion systems and their ability to drive the receptors into a markedly desensitized state.
To further test the possibility of two receptor populations with different affinities, predesensitization experiments were performed (also see Celentano and Wong, 1994). Preapplication of low GABA doses resulted in an equal block of the peak and the slow component (IC50 values of 642 and 629 nm, respectively), suggesting the presence of a single receptor population. The half-maximal predesensitizing GABA concentrations are in the range of the GABA concentrations that have been measured in the hippocampal extracellular space (200–800 nm) (Lerma et al., 1986;Tossman et al., 1986) and the cerebrospinal fluid (66–177 nm) (Perry et al., 1982; Schaaf et al., 1985). Hence, it is possible that GABAARs in vivo can be desensitized partially by tonic GABA, depending on the efficiency of GABA uptake mechanisms at the synapse.
The Zn2+ block of native interneuron GABAA receptors and its functional implications
Zinc ions induce a complex pattern of blockade and potentiation on glutamate and GABAA receptors. At micromolar concentrations they potentiate AMPA receptor-mediated currents, whereas they block NMDA receptor and GABAAR responses (Westbrook and Mayer, 1987; Rassendren et al., 1990; but see Hollmann et al., 1993); in millimolar concentrations they block AMPA receptors (Rassendren et al., 1990).
A high-affinity block of GABAAR-mediated currents by Zn2+ was correlated with the absence of γ-subunits (IC50 = 560 nm for a α1β2-subunit combination and >100 μm for a α1β2γ2-subunit combination) (Draguhn et al., 1990). In contrast, White and Gurley (1995) showed IC50 values between 1.2 and 9 μm for γ2-containing recombinant receptors. In native receptors, the IC50 values show a wide range [5.8–300 μm (Martina et al., 1996; Hollrigel and Soltesz, 1997)], which was attributed to the different abundance of γ-subunits. These results, together with the IC50 values obtained in our study (8 and 16 μm), made the single-cell RT-PCR study of GABAAR mRNAs mandatory. The PCR showed the presence of the γ2-subunit mRNA in all cells investigated. To confirm the presence of the functional γ-subunit protein in somatic patches, we tested the effect of flunitrazepam. It shifted the GABA dose–response curve to the left and confirms thereby the existence of functional benzodiazepine binding sites pointing to the incorporation of γ-subunits in the GABAAR molecules (Pritchett et al., 1989). Thus, native basket cell GABAARs are Zn2+-sensitive and contain γ-subunits.
Zn2+ exerts its block via two effects: an acceleration of the slow decay component and a prolongation of the recovery from desensitization. The slow component is correlated with the reopening of the GABAAR pore (Jones and Westbrook, 1995). A reduction of these reopenings and a simultaneous amplification of the desensitization via changes in the corresponding rate constants would result in the described block pattern. The dramatic effects of Zn2+ on the transferred charge (Figs.3A,C, 4A) (see also Mellor and Randall, 1997a) may have enormous implications on the GABA-mediated IPSCs. If granule cells show just a little activity, Zn2+ from their glutamatergic mossy fiber terminals would have only minute or no effects on the postsynaptic basket cell GABAARs. High granule cell activity, in contrast, may lead to local Zn2+ concentrations of up to 300 μm(Assaf and Chung, 1984). These zinc ions may spill over to neighboring GABAergic synapses on the basket cell and may disinhibit them. This mechanism could secure the feedback inhibition of the granule cells via their presynaptic basket cells during high activity states.
Structure function relationship of basket cell GABAARs
The most prominently expressed GABAAR-subunit mRNAs in dentate basket cells are α2, β3, γ2. Only one immunocytochemistry study deals with the structural components of putative dentate gyrus basket cell GABAARs (Fritschy and Möhler, 1995) showing a prevalence of α1-, β2,3-, and γ2-GABAAR-subunits for this cell type and other hilar interneurons. In addition to immunohistochemistry, single-cell PCR provides the possibility for describing the structural components of the GABAARs of a characterized interneuron population.
Gao and Fritschy (1994) found an expression of the α1-subunit in a subset of putative basket cells, namely the PV-positive cells. This contrasts with the dominance of α2 in hippocampal principal cells (Fritschy and Möhler, 1995). In identified basket cells we found a subpopulation of α1-subunit mRNA-containing cells (21% and 38% using the degenerate primer set and α1spec, respectively). Because of the lack of detailed data on the abundance of PV-positive basket cells (Ribak et al., 1990; Freund and Buzsáki, 1996), single-cell PCR was performed to assess the percentage of PV mRNA-containing basket cells. With this approach, 22% of the harvested cells contained PV mRNA. Thus, these data indicate the presence of at least two distinct subpopulations of basket cells and are in accordance with previous studies demonstrating the presence of α1 in PV-positive basket cells. Despite this heterogeneity, the basket cells studied here did not differ with respect to electrophysiological properties.
In addition to providing the main inhibition in the CNS, basket cells and other GABAergic interneurons play a critical role in synchronizing the activity of large ensembles of principal cells. A comprehensive molecular and functional characterization of identified interneurons will ultimately lead to a better understanding of the importance of the large diversity of GABAergic cell types and their significance in the control of network activity.
Footnotes
This work was supported by the Deutsche Forschungsgemeinschaft (Grant Be 1859/1-1 to T.B., SFB505/C5 to U.K., and Mo 432/3-1 to H.M.). C.S. was supported by the graduate program of Molecular and Cellular Neurobiology of the University of Heidelberg. We thank U. Amtmann, D. Haschke, J. Ihrmer, B. Plessow-Freudenberg, and S. Roth for excellent technical assistance, and Drs. H. Backus, J. Bischofberger, P. Jonas, M. Jones, and M. Martina for helpful discussions and careful reading of this manuscript.
Correspondence should be addressed to Dr. Thomas Berger, Institute of Physiology, University of Bern, Bühlplatz 5, CH-3012 Bern, Switzerland.