INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

-Aminobutyric acid (GABA) is the most abundant, and the most prominent, inhibitory transmitter in the mammalian CNS. According to the results of electrophysiological and pharmacological studies, GABA's actions are mediated by at least three different types of receptors: GABA_B receptors that are coupled to Ca²⁺ or K⁺ channels by means of G proteins and intracellular second messengers (Bowery 1993); and GABA_A and GABA_C receptors that are ligand-gated Cl channels (Bormann and Feigenspan 1995; Sieghart 1995; Whiting et al. 1995). GABA_A receptors, the most thoroughly studied receptor type, are blocked by bicuculline, picrotoxin, and some bicyclic cage compounds such as t-butylbicyclophosphorothionate (TBPS). GABA_A receptors appear to be pentameric hetero-oligomers assembled from combinations of different subunits. Genes encoding a repertoire of 16 distinct mammalian GABA_A receptor subunit proteins (_1-6, _1-3, _1-3, , , and _1-2) have been cloned and sequenced from human, rat, and bovine brains (Davies et al. 1997; Whiting et al. 1995, 1997). Recombinant receptor investigations have provided data about the relationship between subunit composition and functional properties (Whiting et al. 1995).

Studies in cat, rat, chick, and frog dorsal root ganglion (DRG) neurons show that, on the basis of their responses to GABA_A receptor antagonists, DRG GABA receptors in these species have the properties of GABA_A receptors (Choi et al. 1981; Deschennes et al. 1976; Gallagher et al. 1978; Inoue and Akaike 1988). Our recent investigations of GABA receptors in cultured adult human DRG neurons, however, reveal that these particular neurons express Cl channels with pharmacological properties distinct from those of GABA_A, GABA_B, and GABA_C receptors found in other vertebrate neurons (Valeyev et al. 1996). To see if this was also the case in cultured human embryonic DRG neurons, we used patch-clamp techniques to explore the characteristics of GABA-activated currents. In contrast to adult human DRG neurons, our present results found that embryonic human DRG neurons express Cl-selective GABA_A receptors the properties of which are similar to those described for DRG neurons in other species. The current study is the first detailed account of the properties of GABA receptors in human neurons. A preliminary account of these data has appeared (Valeyev et al. 1997).

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Cell culture

Embryonic spinal cords with attached DRGs were obtained with written informed consent and human subjects committee approval in accordance with published guidelines from 5- to 8-wk-old human fetuses through a collaborative agreement with the Central Laboratory for Human Embryology, Department of Pediatrics, University of Washington. The tissue was shipped overnight in cold storage medium. Fetal DRGs were removed, incubated on a rotary shaker in 0.25% trypsin in Ca²⁺/Mg²⁺-free Hanks' balanced salt solution for 45 min at 37°C, and dissociated to a single-cell suspension by trituration. The cells (neurons, Schwann cells, and fibroblasts) were resuspended in culture medium consisting of Eagle's minimum essential medium with 10% heat-inactivated fetal bovine serum, nerve growth factor (50 ng/ml), and neurotrophin 3 (50 ng/ml). Cells were plated onto collagen-coated Aclar 33C dishes and maintained in a 6% CO₂ atmosphere at 37°. Nonneuronal cells were eliminated from the cultures by three treatments with fluorodeoxyuridine (5-10 M) starting with day 1 of the culture period.

Recording solutions

DRG neurons were bathed in a solution of (in mM) 140 NaCl, 5 CsCl, 2 CaCl₂, 1 MgCl₂, 5 N-2-hydroxy-ethylpiperazine-N'2-ethane-sulfonic acid (HEPES), and 10 D-glucose, titrated to pH 7.35 with NaOH. Osmolarity was adjusted with sucrose, if needed, to 310 mosmol/kg. Patch pipettes for whole cell recording were filled with a solution of (in mM) 130 CsCl, 2 MgCl₂, 0.1 CaCl₂, 1.1 ethylene glycol-bis(-aminoethyl ether)-N,N,N',N'-tetraacetic acid (EGTA), 5 ATP, and 10 HEPES, buffered to pH 7.15 with CsOH. Osmolarity was adjusted, if needed, with sucrose to 290 mosmol/kg. Extracellular pipette solution for cell-attached recording consisted of (in mM) 120 NaCl, 20 TEA-Cl, 5 KCl, 5 4-aminopyridine, 0.1 CaCl₂, 10 MgCl₂, 10 glucose, and 10 HEPES, titrated to pH 7.35 with NaOH. Osmolarity was adjusted with sucrose, if needed, to 310-320 mosmol/kg.

Whole cell recording

Within 5-7 days after plating, whole cell patch-clamp techniques were used to record from DRG neurons maintained at 21-23°C. Eagle's medium was replaced with the extracellular solution. Before breaking the patch with suction, seal resistances in the gigaohm range were obtained with thin glass pipettes (with filament, 1.5 mm OD; WPI, Sarasota, FL) that had been pulled by a Flaming Brown micropipette puller (Sutter Instrument, San Rafael, CA). Whole cell currents were monitored via an Ag/AgCl wire with an Axopatch 200A amplifier (Axon Instruments, Foster City, CA) in the resistive-headstage mode, displayed on a Gould chart recorder (Gould, Cleveland, OH), and recorded on tape for off-line analysis with pClamp software (Version 6.02, Axon Instruments). Series resistance was <10 M and was 50-70% compensated. The voltage error was always <5 mV.

The amplitudes of the peak currents evoked by various concentrations of GABA were measured relative to the peak responses induced by 10 µM GABA. The data were fitted (KaleidaGraph, Synergy Software, Reading, PA) with the logistic equation
where [GABA] is the GABA concentration, I_max is the maximum current amplitude, I is the peak current at a given concentration of GABA, EC₅₀ is the concentration of GABA yielding a current half I_max, and n is the slope factor corresponding to the Hill coefficient.

Fluctuation analysis of GABA-activated membrane currents

Established fluctuation analysis techniques were employed to derive the elementary properties of GABA-activated Cl channels (Cull-Candy 1986; Serafini et al. 1998). Membrane currents were recorded at a low gain as a DC signal, amplified, filtered, and stored on tape. For analysis, data were played back, high-pass filtered at 0.1 Hz, low-pass filtered at 1 kHz (8-pole Butterworth, 3 dB, Model 9002, Frequency Devices, Haverhill, MA), and digitized at 2 kHz (LAB PC acquisition board, National Instrument, Austin, TX). Data were analyzed by either Strathclyde Electrophysiological Software SPAN 3.0 (courtesy of Dr. J. Dempster, University of Strathclyde, Glasgow, UK) or KaleidaGraph. Baseline power spectra of current variance were subtracted from signals obtained during the plateau phase of GABA responses that usually lasted 60 s (30-90 s). The main time constants governing kinetic behavior were determined from the resulting difference spectra fitted with Lorentzian functions of the form
where f is the frequency, S( f ) is the power spectral density at frequency f, S_i is the spectral density at zero frequency asymptote, and f_ci is the corner frequency. Because multiple Lorentzian components often are needed to fit the spectra of GABA-induced current noise (Cull-Candy and Usowicz 1989), the membrane current variance needs to be fitted with 1/f, single-, double-, or triple-Lorentzian functions. Assuming that the Lorentzian curves describe the kinetics of populations of two-state (open-closed) ion channels, the average duration of the open conducting state () of these channels can be estimated from the equation:  = (2f_c)¹ (Neher and Stevens 1977).

The variance, ², of agonist-induced current fluctuations was calculated as
where N is the total number of points in the record, i(k) are the individual data points, and i is mean value of the current record. The apparent average single-channel conductance  was estimated from the relationship between variance of the noise and the mean amplitude of whole cell current change using the equation
where ² is the agonist-induced current variance, I is the agonist-induced mean membrane current change, and V_D is the driving force (difference between the potential at which the membrane was clamped and the equilibrium potential for Cl ions) (Neher and Stevens 1977). In all responses evoked by 10 µM GABA, the mean current and the variance were constant during the drug application.

Single-channel recordings

Cell-attached patches were used for recording single-channel currents. Conductance measurements on cell-attached patches have shortcomings: among them, determination of the resting membrane potential and interpretation of current-voltage relationships produced by unequal intracellular and extracellular concentrations of Cl ions. Nonetheless, we used cell-attached patches because this configuration eliminates problems associated with patch excision, such as rundown of channel activity, alterations in receptor-channel kinetics, and activation of previously inactive Cl channels (Covarrubias and Steinbach 1990; Uchida and Yang 1995).

For single-channel recordings from cell-attached patches, pipette tips were coated with silicone elastomer (Sylgard 184; Dow Corning, Midland, MI) to reduce capacitance and to enhance the signal-to-noise ratio. In extracellular medium, the range of tip resistances was 5-8 M. After obtaining a gigaohm seal, an offset of 0.1 to 0.2 pA was observed in ~50% of the experiments, possibly reflecting slight damage to the seal. The lag time between the contact of pipette to cell and the recording of channel activity was ~10-15 s.

Current signals (Axopath 200A, Axon Instruments), in resistive-headstage mode, were filtered (low-pass 3 kHz; 8-pole Bessel filter, 3 dB; Frequency Devices, model 9002), digitized at 20 kHz (A/C VCR Adapter, Model PCM 4/8, Medical System, Greenvale, NY), and stored on tape. For analysis, recordings were replayed, low-pass filtered at 1 kHz through an 8-pole Bessel filter, and digitized (LAB PC acquisition board) at a sampling rate of 200 µs per point.

Thirty- to 60-s epochs were analyzed off-line using the routines available in pClamp (Version 6.02) software (FETCHAN and pSTAT, Axon Instruments). Data blocks containing artifacts were removed. Each opening was examined directly and, in case of drift or artifacts, the baseline level and the opening channel step amplitude were corrected before storing the data. Openings and closings were discerned using a 50% threshold-crossing algorithm for event detection. Only threshold crossings with durations greater than the dead time (t_d) of the system were measured. The following equation was used to calculate dead time: t_d = 0.179/f_c, where f_c is the effective bandwith of the recording system determined from the relationship
where f₁ and f₂ are the cutoff frequencies of the low- and high-pass filters, respectively, in series (Colquhoun and Sigworth 1995). f_c was 970 Hz. The corresponding rise time was 340 µs. It was calculated using the following equation: t_r = t_d/0.538. Patches were used only if, at the effective bandwith of 970 Hz, the 50% threshold for the subconductance level was 3.5 times the baseline root mean square noise. The rate of false events was maintained at one to two orders of magnitude lower than the rate of channel events (Colquhoun and Sigworth 1995). For amplitude analysis, events shorter than twice the rise time of the system were disregarded.

Dwell-time analysis was restricted to runs of single-channel current where detectable openings to conductance states other than the main conductance state were rare. Openings to other than the main conductance level together with adjacent closed intervals and events shorter than the rise time were excluded (Colquhoun and Sigworth 1995).

Single-channel amplitudes were calculated at patch potentials where the main conductance levels could be differentiated into two separate Gaussian peaks (+60 to 60 mV). Amplitude distributions of all events (D_A) were calculated with one or the sum of two Gaussians
where  is the event amplitude and _i, µ_i, and _i are the area, the mean, and the standard deviation of each Gaussian component.

I-V plots were constructed from the peaks of Gaussian-fitted amplitude histograms constructed at various holding potentials. The reversal potential was estimated by measuring the intersection of the I-V plot with the voltage axis. The chord conductance, , was estimated by the formula  = (I_A  I_B)V, where I_A and I_B are the current values of opposite polarity closest to the reversal potential.

Unless otherwise stated, all data are expressed as means ± SE.

Drug application

Drug delivery was accomplished with pressure-ejection pipettes with 2- to 3-µm tip diameters that were positioned within 20-50 µm of the soma under study and resulted in concentration changes of the solution bathing the neuron within <50 ms of the ejection artifact in the record (estimated from the rate of onset of the current change produced by 10 µM GABA applications). In cell-attached experiments, GABA was included in the pipette solution at a concentration of 10 µM. This concentration of GABA consistently provided a frequency of Cl-channel openings sufficient for quantitative study of amplitude and kinetics.

All agonists and antagonists were either dissolved in extracellular medium or else prepared in ethanol and then diluted with extracellular solution at the time of an experiment. The final ethanol concentration was never >0.2%, a concentration that, in control experiments, had no effect on GABA-activated whole cell currents when applied to single neurons by pressure-ejection pipettes.

Materials

GABA was purchased from Calbiochem (San Diego, CA); alphaxalone, baclofen, diazepam, cis-4-aminocrotonic acid (CACA), and TBPS were obtained from RBI (Natick, MA). All other compounds were purchased from Sigma (St. Louis, MO).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Cultured embryonic human DRG neurons varied between 20 and 30 µm in diameter. Larger neurons were not seen. They were ovoid in shape and possessed neurites of varying lengths.

Whole cell currents

GABA-activated inward currents (I_GABA) produced in response to micromolar concentrations of the amino acid could be demonstrated in virtually all neurons held at negative holding membrane potentials (Fig. 1A). Clearly, the amplitude of the responses was dependent on the concentration of GABA applied. Sustained application of high concentrations of GABA caused desensitization of I_GABA with a decline of the response during the time course of the agonist application (Fig. 1A). In lower concentrations, the amino acid did not generate currents that showed discernible desensitization (Fig. 1A).

View larger version (14K): [in this window] [in a new window]   Fig. 1. GABA-activated whole cell currents from cultured, embryonic dorsal root ganglion (DRG) neurons. A: effects of GABA applied in concentrations of 5, 10, and 50 µM. Whole cell recordings were carried out at a holding potential of 60 mV. Downward deflections represent inward currents. In this, and in subsequent figures, horizontal bars represent the duration of application of GABA and drugs. B: plot of normalized amplitude of peak current vs. GABA concentration. All GABA responses were normalized to the peak current induced by 10 µM GABA. Each data point represents the mean ± SE of the results of 3-5 independent recordings. Error bars not shown were smaller than the size of the symbols. Solid line is the best fit obtained by nonlinear regression analysis to the equation described in the text. From the logistic fit to the data points, GABA activated currents with an EC50 of 111 ± 20 µM and a Hill slope of 1.7 ± 0.2.

Threshold current responses were elicited with GABA concentrations of 2.5 µM. They saturated at a concentration of 1.0 mM. Figure 1B shows the concentration-response curve of the peak Cl currents induced by GABA. Construction of log concentration-response curves with least square fitting of the points gave an EC₅₀ value of 111 ± 20 µM and a Hill coefficient of 1.7 ± 0.2.

Pharmacology of GABA-induced currents

In the absence of GABA, inward currents also were elicited by applications of the neurosteroid anesthetic alphaxalone (51.2 ± 4.4 pA, 1.0 µM, n = 12) and the barbiturate pentobarbital (16 ± 5.8 pA, 50 µM, n = 12; Fig. 2). More than 90% of neurons were directly activated by alphaxalone and pentobarbital. (For the direct effects of alphaxalone on embryonic human DRG neurons see the companion paper.)

View larger version (18K): [in this window] [in a new window]   Fig. 2. Comparison of the actions of GABA (10 µM), pentobarbital (50 µM) and alphaxalone (1.0 µM) in cultured human embryonic DRG neurons. Neurons were voltage clamped at 60 mV.

The selective GABA_B receptor agonist baclofen (100 µM) did not produce any currents when the patch pipettes contained 130 mM KCl (n = 7). Nor did the conformationally restricted GABA analog and GABA_C receptor agonist CACA (100 µM) induce a detectable response (n = 7). Embryonic human DRG neurons also failed to respond to the putative amino acid neurotransmitters glycine (100 µM, n = 7) and taurine (100 µM, n = 7; not shown).

GABA-activated currents were depressed substantially in a reversible manner by blockers of the GABA_A receptor/Cl channel complex. As seen in Fig. 3, bicuculline (100 µM, 92.0 ± 5.0% of peak amplitudes of control GABA-responses, n = 6), picrotoxin (100 µM, 96.0 ± 2.0%, n = 6), and TBPS (100 µM, 96.0 ± 2.0%, n = 6) markedly reduced GABA responses.

View larger version (20K): [in this window] [in a new window]   Fig. 3. Block of GABA-activated currents by bicuculline, picrotoxin, and t-butylbicyclophosphorothionate (TBPS). Left: whole cell inward currents activated by GABA (5.0 µM). Right: currents activated by GABA in the presence of bicuculline methochloride (BCC, 100 µM), picrotoxin (PTX, 100 µM), and TBPS (100 µM). Neurons voltage clamped at 60 mV.

I-V relationship of GABA-induced currents

Because activation of GABA_A receptors typically opens channels that are selective for anions, it was not unexpected that the current generated by GABA in human embryonic DRG neurons would be carried mainly by Cl ions. When Cl was the major intracellular anion, ([Cl]_o/[Cl]_i = 151 mM/134 mM), I_GABA reversed around 0 mV (3.0 ± 3.0 mV, n = 6), close to the Cl equilibrium potential calculated by the Nernst equation (Fig. 4). Moreover, the reversal potential shifted according to the Nernst relationship for a Cl-selective current when part of the intrapipette chloride was replaced isotonically by the relatively impermeant anion fluoride. Reduction of the [Cl]_i to 120 mM (by isotonic substitution of CsF for CsCl) shifted the reversal potential to 16.6 ± 4.0 mV (n = 6) (expected reversal potential according to the Nernst equation: 13.0 mV). It is unlikely that I_GABA represented a nonspecific cation current because such currents are insensitive to changes in anion concentrations (Colquhoun et al. 1981; Yellen 1982).

View larger version (12K): [in this window] [in a new window]   Fig. 4. Current-voltage (I-V) relationship for GABA-activated whole cell currents. Peak amplitudes of currents activated by GABA (50 µM) are plotted as a function of membrane potential. IGABA reversed polarity at 3.0 ± 3.0 mV ([Cl]o/[Cl]i = 151 mM/134 mM). Outward rectification seen at positive membrane potentials. Each point represents the mean of values obtained in 6 neurons. Vertical lines: SE.

The whole cell I-V relationship for the peak amplitudes of GABA-activated currents demonstrated a small outward rectification (i.e., outward currents were larger than inward currents at equivalent holding potentials; Fig. 4). These data are compatible with previous reports that demonstrate outward rectification in whole cell recordings of I_GABA (Bormann et al. 1987; Curmi et al. 1993; Peters et al. 1989; Weiss et al. 1988).

Fluctuation analysis of GABA-evoked current noise

The increase in current noise on application of low concentrations of GABA lent itself well to fluctuation analysis (Fig. 5A). As illustrated in Fig. 5B, spectral density plots were derived from fluctuations that occurred during the plateau phase (60 s) of I_GABA (GABA concentration, 10 µM). These were best fitted by the sum of three Lorentzian components with time constants corresponding to mean channel opening times of 161 ± 7.2 ms for the long-lasting, 35 ± 3.3 ms for the intermediate-lasting, and 2.0 ± 0.6 ms for the short-lasting components (n = 16). Triple exponential functions like these derived from fluctuation analysis of the GABA-activated Cl currents in embryonic human DRG neurons previously have been reported in embryonic rat hippocampal, spinal cord, and olfactory neurons (Liu et al. 1996; Serafini et al. 1995).

View larger version (19K): [in this window] [in a new window]   Fig. 5. Fluctuation analysis of GABA-activated Cl currents. A, top: low-gain DC record; bottom: high-gain AC record of IGABA (GABA, 10 µM) with the neuron clamped at 60 mV. B: power spectral density plot of membrane current fluctuations during GABA application (after subtraction of baseline fluctuations). Spectrum was best fitted by the sum of three Lorentzians. , corner frequencies corresponding to estimated apparent mean open-time constants for GABA-activated Cl channels of 1 = 175, 2 = 26.5, and 3 = 2.3 ms.

The elementary conductance estimated by the relationship between mean amplitude of whole cell current and variance was 22.6 ± 4.8 pS (n = 16). This conductance for GABA activation of Cl currents compares favorably with values previously reported for GABA-gated channels evaluated by means of fluctuation analysis in other cultured embryonic mammalian preparations (Jackson et al. 1982; Serafini et al. 1995; Smart 1992).

Single GABA-activated channel currents in cell-attached patches

Without GABA in the intrapipette solution, no spontaneous channel activity was recorded (n = 3). When GABA (10 µM) was included in the intrapipette saline, a characteristic set of channel currents developed in a third of 21 cell-attached patches; Fig. 6 is representative. As is usual for GABA currents, patches responded to GABA either with brief single-channel openings uninterrupted by brief closures or with more complex events consisting of long bursts of single-channel currents (Macdonald et al. 1989). Direct transitions between at least two different conductance states (a main-state predominant level and a subconductance level) without an interposed closing were discernible in the records. Subconductance levels did not occur in the absence of main-state activity. That the majority of patches exposed to GABA exhibited multiple-channel activity composed of equal and integral amplitudes suggests the presence of more than one channel in the patch.

View larger version (47K): [in this window] [in a new window]   Fig. 6. GABA-activated single-channel currents. Current records of GABA-gated single channels from a cell-attached patch recorded at different patch potentials. Pipette solution contained GABA (10 µM). Numbers (left) of each trace indicate the patch potential (Vp) shift of patch membrane potential with respect to neuron resting membrane potential indicated in mV. - - -, note presence of 2 amplitude levels.

The distribution of amplitudes of single-channel currents was best fitted with the sum of two Gaussian functions. The Gaussian fits to amplitude histograms showed peaks that averaged 5.9 ± 0.6 pA and 3.2 ± 0.5 pA at a patch potential (V_p) of +40 mV (n = 7 patches; Fig. 7). Single-channel amplitudes were only calculated at holding potentials where both the main conductance and subconductance levels could be differentiated as two separate Gaussian peaks. Plotting current as a function of V_p, these two current amplitudes corresponded to two-chord conductance levels, a main level of 30.0 ± 2.0 pS and a subconductance level of 18.6 ± 2.2 pS (n = 7). These values fall well within the range recorded in patches from cultured embryonic mammalian neurons which typically is between 20 and 30 pS (Allen and Albuquerque 1987; Geetha and Hess 1992; Liu et al. 1996; Macdonald et al. 1989; Smith et al. 1989).

View larger version (15K): [in this window] [in a new window]   Fig. 7. Distribution of GABA-activated elementary channel amplitudes in on-cell recordings. Data refer to single-channel recordings obtained with GABA (10 µM) in recording pipette. Histogram fitted best by the sum of 2 Gaussian curves representing a main and a subconductance level. Mean current levels (Vp = +40 mV) were 0.32 ± 0.04 and 0.5 ± 0.06 pA. Histogram derived from data partly illustrated in Fig. 5.

As seen in Fig. 8, the current amplitudes of the main conductance state and of the subconductance state both depended on changes in V_p. In contrast to what one finds when whole cell currents are recorded, the I-V relationship for the main conductance state activated by GABA was linear with a reversal when the patch potential (V_p) was 50 mV (the resting membrane potential was 55 ± 5.3 mV, n = 7, determined by breaking the membrane patch after the single-channel measurements were made). The subconductance level also had a linear I-V relationship with a reversal potential identical to that of the main conductance state. The lack of rectification of single GABA-activated Cl channels is similar to the findings of previous investigations (Allen and Albuquerque 1987; Bormann et al. 1987; Curmi et al. 1993; Fatima-Shad and Barry 1992; Gray and Johnston 1985; Hamill et al. 1983; Smith et al. 1989; Weiss et al. 1988). The linearity of I-V plots of single-channel activity indicates that the putative outward rectification observed in I-V plots of GABA-gated whole cell currents was not, in fact, caused by rectification of individual GABA-gated Cl channels.

View larger version (10K): [in this window] [in a new window]   Fig. 8. I-V relationship of GABA-activated single-channel currents. Mean amplitudes of main conductance and subconductance levels of the GABA-induced Cl currents partly illustrated in Fig. 5 plotted against Vp. Each point represents the mean of the best-fit Gaussian distribution of open-channel current amplitudes for both main conductance and subconductance levels at each holding potential. Values were obtained from 6 neurons. Vertical lines: SE; filled circle, main conductance levels; filled square, subconductance levels.

The presence of depolarizing, GABA-activated Cl channels is consistent with findings that GABA depolarizes both DRG neurons and primary afferent terminals (Davidoff and Hackman 1985).

Single-channel kinetics

The gating mechanisms of the main conductance state activated by GABA was examined in cell-attached patches. Open dwell-time histograms showed that a best fit was obtained with triple exponential functions with decay time constants of 1.17 ± 1.01, 25.1 ± 9.7, and 52.3 ± 26.4 ms (n = 7 patches, GABA 10 µM, V_p = 30 mV; Fig. 9A). Analysis of the distributions of closed times revealed that the best fit occurred with two exponential functions with decay time constants of 0.6 ± 0.01 and 640 ± 54.0 ms (n = 7 patches, GABA 10 µM, V_p = 30 mV; Fig. 9B). GABA receptors in a number of locations exhibit complex kinetic behavior such that the channel open- and closed-time distributions require fitting by more than one exponential (Bormann and Clapham 1985; Liu et al. 1996; Macdonald et al. 1989; Mistry and Hablitz 1990; Smart 1992; Taleb et al. 1987; Vicini et al. 1987; Weiss 1988; Weiss and Magleby 1989).

View larger version (23K): [in this window] [in a new window]   Fig. 9. Dwell-time distribution of single channel currents evoked by GABA. A: distribution of open dwell times determined for single-channel openings of currents evoked by GABA (10 µM) at Vp = 30 mV. Distribution of open intervals was best fitted with the sum of 3 exponential components. Histograms were constructed from data partly illustrated in Fig. 5. Calculated curves were superimposed over the histogram. B: closed duration properties of single channel GABA-activated currents. Distribution was best fitted by the sum of 2 exponential functions. Fitting equation was: F(t)=Pn exp[T  n  exp(t  n)], where Pn is the proportion of each of the terms, , the time constant for each term, and T, the interval of the frequency of opening. Curve was fit by the method of Simplex-LSQ. Weighing: function (2). Number of iterations was 5,000. Vp = 30 mV.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Cultured embryonic human DRG neurons appear to express functional GABA_A receptors with properties similar to those of GABA_A receptors in other locations. For example, the GABA receptor channel on DRG neurons is selective for Cl ions as are all GABA_A receptors. In addition, the receptors were blocked by antagonists (the competitive GABA_A receptor antagonist bicuculline and by the noncompetitive GABA_A receptor blockers picrotoxin and TBPS) that block responses mediated by GABA_A receptors in other locations and in other species. The EC₅₀ obtained in the present experiments is within the wide range (5.5-150 µM) reported for EC₅₀ values for cultured embryonic neurons (Kristiansen et al. 1995; Orser et al. 1994; Rho et al. 1996; Sah 1995). In addition, in a number of neuronal systems, Hill coefficients derived from GABA concentration-response curves range between 1.5 and 2.0, reflecting the presence of at least two GABA binding sites at the GABA receptor (Choi and Fischbach 1981; Randle and Renaud 1987; Suzuki et al. 1990). Furthermore the GABA channels in DRG neurons demonstrated complex kinetic behavior such that the channel open- and closed-time distributions required fitting by more than one exponential. The latter property has been described in GABA receptors from a number of locations (Bormann and Clapham 1985; Liu et al. 1996; Macdonald et al. 1989; Mistry and Hablitz 1990; Smart 1992; Taleb et al. 1987; Vicini et al. 1987; Weiss 1988; Weiss and Magleby 1989). Wide variability in the durations of open times of GABA-activated Cl channels is reported (Macdonald et al. 1989; Sakmann et al. 1983). This may occur because the contributions to power spectra of GABA-induced fluctuations often are dominated by lower frequency, long-duration events that approximate the burst length duration rather than reflect the open-time distribution (Colquhoun and Hawkes 1977; Jackson et al. 1982). In addition, fluctuation studies are limited by the frequency response and often cannot resolve frequent brief openings or closings. Moreover with single-channel recordings, many patches contain multiple channels with multiple superimposed events. Superimposed events complicate kinetic analysis. Nevertheless the values of the fast time constants obtained for GABA-gated channels in this investigation were similar to the fast time constants in patches from other cultured embryonic mammalian neurons (Allen and Albuquerque 1987; Liu et al. 1996; Macdonald et al. 1989).

Our findings regarding open and closed dwell-times suggest that GABA-induced Cl channels in embryonic human DRG neurons are homogeneous but have complex channel kinetics with at least two or three open states and two closed states. However, we do not have sufficient data to allow us to discard alternative hypotheses that GABA may open a single class of Cl channels when the receptor has bound either one or two GABA molecules or that GABA may activate a nonhomogeneous population of two-state (open-closed) Cl channels with different open and closed times (cf. Macdonald et al. 1989; Weiss 1988; Weiss and Magleby 1989).

Differences between adult and embryonic human DRG neurons

Our present results with GABA receptor blockers differ significantly from those we reported recently for GABA receptors in cultured adult human DRG neurons (Valeyev et al. 1996). The GABA-currents in adult DRG cells were unaffected by the concentrations of bicuculline and picrotoxin that antagonized GABA responses in embryonic DRG neurons in the present experiments. Several explanations might account for this. First, in various species the properties of neurotransmitter-gated receptors differ between adult and embryonic neurons. In particular, according to their developmental stages, GABA_A receptors have varying pharmacological and biophysical properties (Serafini et al. 1995; Smart 1992; Strata and Cherubini 1994). And because molecular biological studies have shown that levels of mRNA encoding particular GABA_A receptor subunits change during rat brain development (Gambarana et al. 1990; Laurie et al. 1992; Ma et al. 1993; Poulter et al. 1992, 1993), these differences in pharmacological and biophysical properties presumably reflect dissimilarities in subunit composition and stoichiometry. Because in situ hybridization, polymerase chain reaction (PCR) studies, and immunohistochemical investigations that use subunit isoform-specific monoclonal antibodies have demonstrated several GABA receptor subunit mRNAs (e.g., ₂, ₃, ₂) in rat DRG neurons (Alvarez et al. 1996; Furuyama et al. 1992; Ma et al. 1993; Persohn et al. 1991; Wu et al. 1993), we presume that developmental changes similar to those in CNS neurons also occur in DRG neurons. Such developmental differences in pharmacological properties indeed might reflect differences in the expression of GABA receptor subunits in adult and embryonic DRG neurons.

We conclude that the actions of GABA on cultured human embryonic DRG neurons are mediated through the activation of GABA_A receptors. The properties of GABA-activated Cl channels on these neurons are similar to those found in the CNS of human and other mammalian species but differ from those of human adult DRG neurons.