INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Ring A-reduced steroid metabolites of pregnanolone such as allopregnanolone (5-pregnane-3-ol-20-one) have become well established as endogenous modulators and activators of GABA_A receptor/Cl channels expressed by neurons throughout the adult vertebrate CNS (for recent review, see Rupprecht and Holsboer 1999). Extracellular application and 3-substitution of the A ring in the steroid nucleus are prerequisites for most of the interactions with GABA_A receptor/Cl channels reported thus far, since neither intracellular applications nor 3-substituted metabolites applied extracellularly are agonistic. The extracellular site at which steroid metabolites bind with stereospecific requirements is commonly thought to be associated closely with, if not an integral component of, the GABA_A receptor/Cl channel complex. This is supported by steroid modulation of GABA_A receptor/Cl channels characterized in excised outside-out patches (Twyman and Macdonald 1992). Molecular analyses of structure-function relationships emerging in studies of steroid metabolites on recombinant GABA_A receptor/Cl channels have identified a transmembrane domain in  or  subunits as a possible site (Rick et al. 1998). More recently, 3-substituted steroid metabolite modulation of transient GABAergic Cl currents recorded at synapses in neurons in hypothalamic slices from adult rats has been demonstrated to involve G protein-coupled second messenger phosphorylation pathways (Fáncsik et al. 2000), thus revealing a complex, more indirect interaction.

In the adult, allopregnanolone and the 3  isomer isopregnanolone are synthesized from progesterone via the sequential activities of 5-reductase and 3- or 3-hydroxysteroid dehydrogenase, respectively (for review, see Robel and Baulieu 1995). Recently, it has been reported that both allopregnanolone and isopregnanolone are also synthesized from progesterone in CNS tissues throughout the embryonic and early postnatal period of development in the rat (Pomata et al. 2000). These results imply that steroid metabolites may also play roles during CNS development. In this regard, allopregnanolone has been reported to affect neurite outgrowth and fillopodial extension in cultured embryonic rat hippocampal neurons, causing them to retract or regress after less than a 1-h exposure (Brinton 1994). In addition, exogenous allopregnanolone can depolarize cortical plate neurons in a bicuculline-sensitive manner and activate Ca²⁺ entry, which in turn can restore neurite outgrowth in neurons whose GABA synthesis and subsequent GABAergic signal-dependent neurite formation have been blocked (Maric et al. 2001).

Previous electrophysiological studies of cultured embryonic rat hippocampal neurons have demonstrated that allopregnanolone and the clinical anesthetic alfaxalone interact with GABA_A receptor/Cl channels, potentiating Cl current responses to GABA, prolonging GABAergic Cl transients, and inducing a stable macroscopic current via activation of GABA_A receptor/Cl channels (Harrison et al. 1987a,b; Valeyev et al. 1995). Allopregnanolone's effects at GABA_A receptor/Cl channels were initially characterized as "barbiturate-like" since their pharmacological activities at GABA_A receptor/Cl channels superficially resembled those of anesthetic barbiturates (Majewska et al. 1986). However, patch-clamp analysis of GABA_A receptor/Cl channels expressed by embryonic spinal neurons revealed that steroid metabolites modulated both the frequency of GABA-activated Cl channel activity and the kinetics of opened channels, while pentobarbital affected only the latter property (Twyman and Macdonald 1992). Furthermore, steroid-sensitive embryonic hippocampal neurons that do not respond to pentobarbital have been recorded and vice versa (Valeyev et al. 1995). Together, these and other results on recombinant GABA_A receptors (Puia et al. 1990) suggest that the steroids and pentobarbital affected GABA_A receptor/Cl channels via different mechanisms.

We report here that 1) allopregnanolone elicited a triphasic current response in many, but not all cultured embryonic hippocampal neurons involving activation of GABA_A receptor/Cl channels, which was closely reproduced in nonneuronal cells expressing recombinant GABA_A receptor/Cl channels composed of only two subunits; 2) at least two phases involved pertussis toxin-sensitive G proteins; and 3) the delayed phase was mimicked by mastoparan, which can directly activate G proteins.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Dissociation and culture of postnatal rat hippocampal astrocytes and embryonic rat hippocampal neurons

The procedures for culturing embryonic rat hippocampal neurons in astrocyte-conditioned medium have been detailed previously (Liu et al. 1996, 1997). Briefly, 3-day-old rat neonates were used to prepare hippocampal astrocytes in 75-cm² flasks. Serum-free conditioned medium was generated by washing the culture flasks twice and then incubating them with 12 ml of MEM (GIBCO, Grand Island, NY) containing 109 µM putrescine, 0.04 µM progesterone, 0.06 µM sodium selenite, 0.03 µM T₃, 0.12 µM corticosterone, and 1.67 µM insulin, 0.001% albumin and 0.02% transferrin (N3 components) (Romijn et al. 1984) for 24 h. The conditioned media were collected and usually used the same day. Occasionally, harvested media were frozen at approximately 70°C and used later.

To prepare hippocampal neurons, gestational day 19 rat embryos were obtained by caesarian section from pregnant mothers, which had been anesthetized with CO₂ and killed by cervical dislocation. Embryos were quickly decapitated with surgical scissors and the hippocampal tissue was dissected, minced into small pieces, transferred into 5 ml Earle's Balanced Salt Solution (EBSS) containing 20 U/ml papain, 0.01% DNase (both from Boehringer Mannhein Co., Indianapolis, IN), 0.5 mM EDTA, and 1 mM L-cysteine and rocked in an incubator for 35-40 min at 37°C. Single neurons, obtained by triturating the tissue with a Pasteur pipette, were resuspended in EBSS with 1 mg/ml trypsin inhibitor (TI) and 1 mg/ml bovine serum albumin (BSA) and layered over 5 ml of EBSS with 10 mg/ml TI and 10 mg/ml BSA in a 15-ml plastic centrifuge tube. The gradient was spun at approximately 80g for 5 min and the cell pellet was resuspended in astrocyte-conditioned medium and plated at a density of 3.5-4 × 10⁵ cells per dish in 35-mm plastic culture dishes precoated with low-molecular-weight (53 kDa) poly-D-lysine (PDL, Sigma, St. Louis, MO). The cultures were kept at 37°C in a humidified atmosphere containing 10% CO₂. Astrocyte-conditioned culture medium was used without change. All animal procedures were done in accordance with the Guide for the Care and Use of Laboratory Animals.

CULTURE OF CELL LINES. WSS-1 cells (CRL-2029, American Type Cell Collection, Manassas, VA) stably expressing rat 1 and 2 GABA_A receptor subunits were cultured as previously described (Wong et al. 1992). The methods to culture Chinese hamster ovary (CHO) cells (CCL-61, American Type Cell Collection) and to transfect CHO cells with combinations of rat 1 and 2 or 1 and 3 GABA_A receptor subunits were described in a previous publication (Valeyev et al. 1998).

CURRENT RECORDING AND ANALYSIS. All recordings were made at room temperature (22-25°C) on an inverted microscope (Nikon). Before recordings, dishes were removed from the incubator and the culture medium was completely replaced with Tyrode's solution containing (in mM) 145 NaCl, 5.4 KCl, 1.8 CaCl₂, 0.8 MgCl₂, 10 Glucose, 10 HEPES-NaOH (pH 7.4 and 310 mOsm). Cells were recorded either under static bath conditions or they were continuously superfused with a perfusion system comprised of a locally made perfusion controller and miniature electric solenoid valves (The Lee Co., Essex, CT) that allows fast switching (<200 ms complete solution exchange time) among different solutions (Liu et al. 1999). The perfusion rate (~ 0.3-0.5 ml/min) was controlled by the air pressure applied to the solution reservoir. Standard patch-clamp recordings (Hamill et al. 1981) were made with pipettes pulled in three stages from 1.5 mm OD glass capillary tubes (WPI, Sarasota, FL) with a computer-controlled pipette puller (BB-CH-PC, Mecanex SA, Switzerland). These pipettes had a resistance of 3-5 M when filled with an internal solution composed of (in mM) 145 CsCl, 2 MgCl₂, 0.1 CaCl₂, 1.1 EGTA, 5 HEPES, 5 ATP(potassium salt), 5 phosphocreatine (pH 7.2 and 290 mOsm). Whole-cell currents were recorded with a L/M EPC-7 patch-clamp amplifier (Medical Systems Corp., Greenvale, NY) at a gain of 5 mV/pA. Series resistance was compensated for more than 70%. Current signals were digitized with a Digidata 1200 B (Axon Instruments, Foster City, CA) and sampled with AxoScope 7.0 (Axon Instruments) on a Pentium-based personal computer. Current signals were also stored on a videocassette recorder and a VR-100 digital recorder (Instrutech, New York) for off-line digitization and analysis. For fluctuation analysis of GABA- and steroid-induced currents, membrane currents were high-pass filtered at 0.1 Hz and low-pass filtered at 1 kHz and then properly amplified to allow computer-assisted analysis using Strachclyde Electrophysiological Software (Dr. John Dempster, University of Strathclyde, Glasgow, Scotland). Fourier-transformation of the steady-state whole cell current generates power spectra that were fitted with the following Lorentzian function
where S(f) is the total power of the current, S(0)₁ is the value of S of the first Lorentzian function when frequency is infinitesimal, f is the frequency, f_c1 is the corner frequency (the frequency at which the power decayed to one-half of its maximum) of the first Lorentzian function. S(0)₂, f_c2, S(0)_n, and f_cn have the same meanings for the second and n^th Lorentzian functions. The time constants (approximately equal to the mean open times) were derived by the following equation
The relative contribution of each Lorentzian function to the total power were calculated as

STATISTICAL TESTS. Data are shown as means ± SE. Two-tailed t-tests were used to assess significance. Differences were considered significant at P < 0.05 or P < 0.01.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Previous electrophysiological studies of embryonic rat hippocampal neurons differentiating in culture have revealed quite widespread effects of allopregnanolone and the structurally related anesthetic alfaxalone on GABA_A receptor/Cl channels. All of the cultured hippocampal neurons included in this study expressed bicuculline-sensitive Cl conductance responses to GABA and fluctuation analyses of GABA-activated Cl currents revealed estimated unitary properties of the underlying channels, which were consistent with previous results. In addition, the great majority of the embryonic neurons also responded to allopregnanolone, thus facilitating study of its interaction with GABA_A receptor/Cl channels emerging during the early phases of hippocampal neuron differentiation.

Transient and persistent effects of allopregnanolone on GABA_A receptor/Cl channels expressed by embryonic hippocampal neurons differentiating in culture

Addition of allopregnanolone to the bathing medium altered the electrical properties of the majority of the 50 neurons recorded in this study. Slow diffusion of allopregnanolone into the static bath to generate a final level of 1 µM led, with approximately 5- to 10-s delay, to a progressive increase in baseline current recorded at negative potentials together with more intensified microscopic fluctuations (Fig. 1, A1; n = 3). The gradual increase in visible membrane current variance closely paralleled the increase in macroscopic current as progressively more ion channels became activated. The current response and superimposed fluctuations were well sustained in most of the recorded neurons and in stable recordings could be consistently maintained as long as allopregnanolone was present. Following superperfusion and wash out of the steroid, the baseline electrical properties of most neurons recovered to control values within 15-30 s with continuous medium exchange (Fig. 1, A1). The rest of the neurons recorded in this study were continuously perfused to characterize allopregnanolone's effects under differing experimental conditions.

View larger version (32K): [in this window] [in a new window]   Fig. 1. Allopregnanolone induces a triphasic current response in embryonic rat hippocampal neurons. E19 hippocampal neurons were cultured for 1 day, then recorded with CsCl-filled patch pipettes in the whole cell mode and clamped at 80 mV unless otherwise noted. A1: allopregnanolone (3, 5) diffused into the static bath (arrow) to generate a final concentration of 1 µM gradually induces a steady inwardly directed macroscopic current response of several hundred pA superimposed with microscopic fluctuations, which rapidly disappears following superfusion with saline (perfusion). A2: continued perfusion of the same neuron reduces the baseline current variance, as is evident from the thickness of the trace compared with that shown in Al. Application of 1 µM 3, 5 from a closely positioned pipette triggers a triphasic current response, which consists of an immediate transient (approximately 50pA) that rapidly decays within seconds and is followed first by a steady low-amplitude signal (~10pA) and then after approximately 30 s by a high-amplitude response (~approximately 400pA) similar to the response in A1. B1-3. Close application of 3, 5 to another cell while clamping the potential at 0 mV (~ECl) does not lead to a current response until the cell is repolarized to 80 mV to generate the necessary driving force. A second application of 3, 5 to the same cell immediately induces a steady low-amplitude current together with more intensified fluctuations, which is followed by the large amplitude response. Following continuous washing, fluctuations persist on the baseline and a third application of 3, 5 along with 50 µM bicuculline attenuates the fluctuations and shifts the baseline current in a positive direction by approximately 10pA. Removal of bicuculline while continuing to apply 3, 5 leads to an immediate low-amplitude current that relaxes and is followed after a delay by the large amplitude response. Transient co-application of bicuculline attenuates the large amplitude response. C1-2. 3, 5 elicits a triphasic current response, which is completely blocked by perfusing the cells with bicuculline and then co-applying bicuculline with 3, 5. D1-3. 3, 5 induces the delayed large-amplitude current response in another cell without either the transient or steady low-amplitude phases. Recovery during continuous washing occurs exponentially ( = 25 s). Co-application of 3, 5 with 50 µM bicuculline noticeably shifts the baseline to less negative values and on termination of both substances immediately leads to a substantial "off response", which decays exponentially ( = 16 s). Co-application reproduces these effects on the baseline, while continued application of 3, 5 after terminating bicuculline leads immediately to an off response of similar amplitude and decay and, after a delay, to the high-amplitude response, which decays exponentially on termination of allopregnanolone ( = 26 s).

After recovery to the baseline current levels initially recorded, rapid superfusion of allopregnanolone (to the same neuron) from a closely positioned pipette (Fig. 1, A2) revealed that the delayed and progressive changes in electrical properties were not due to the time course of steroid metabolite diffusion in the static bath before reaching the recorded neuron. Rather, when applied from a nearby pipette, three distinct phases of steroid metabolite-triggered changes in membrane properties were detected to varying degrees in the population studied. In about half of the neurons tested, rapid application of allopregnanolone immediately triggered an inwardly directed transient current response of relatively low amplitude (approximately 10-50 pA), which decayed within seconds (with a mean time constant of 1.81 ± 0.15 s, means ± SE; n = 5) to a lower, but detectable level (~5-10 pA) that was well maintained (Fig. 1, A2). The steady low-amplitude current, which exhibited more variance than the baseline signal recorded in superfused neurons under control conditions at negative potentials (as can be seen in the relatively thicker current trace), was then followed in approximately 20-40 s by a progressive increase in inwardly directed macroscopic current that typically plateaued in the approximately 200-500 pA range (309 ± 24 pA; n = 136) and was well maintained. The latter recapitulated the characteristics of the delayed current following diffusion of allopregnanolone in the static bath (Fig. 1, A1). Fast superfusion with the bathing saline led to a recovery of baseline properties in tens of seconds (Fig. 1, A2) in many (n = 21), but not all neurons. In about one-half of the neurons studied that responded to allopregnanolone with a delayed current, at least several minutes of constant superfusion at 10-fold bath volume exchanges every minute were required to recover baseline properties, indicating that some effects of the steroid persisted long after the metabolite was likely to be eliminated from the extracellular saline.

An example of four experiments showing the persistent effects of the steroid accumulating during prolonged exposure is illustrated in Fig. 1B. Before the first application of allopregnanolone, the baseline current signal exhibited relatively low levels of membrane current variance characteristic of continuously superfused embryonic hippocampal neurons (Fig. 1, B1). Depolarizing the membrane potential to 0 mV (approximate equilibrium potential for Cl under these conditions) reduced membrane current variance to its lowest levels, indicating that a component of the membrane current signal recorded at negative potentials with Cl-filled patch pipettes likely involves constitutive Cl channel activation. Application of allopregnanolone had little effect on the current signal recorded at 0 mV but, on returning to 80 mV, immediately resulted in a baseline signal of more than 200 pA, which was superimposed with microscopic fluctuations (Fig. 1, B1) similar to those recorded in other cells (e.g., Fig. 1A). Thus neither a negative membrane potential nor a driving force acting on Cl channels was necessary for the delayed effects of allopregnanolone on membrane current. Terminating the allopregnanolone application and thoroughly washing the neuron over approximately 20-30 s led to a substantial recovery of the baseline properties (Fig. 1, B1 and B2).

However, closer inspection of the recovering baseline current signal revealed a persistent increase in membrane current variance following each allopregnanolone application despite continued perfusion in normal saline. This can be seen by comparing the thicknesses of the three consecutive baseline current traces before (Fig. 1, B1) and after allopregnanolone applications (Fig. 1, B2 and B3). At 80 mV, a second application of allopregnanolone to the same neuron immediately triggered a low-amplitude inwardly directed current (approximately 10-20 pA), which was superimposed with detectably more microscopic fluctuations (Fig. 1, B2). The low-amplitude current signal remained quite steady for approximately 20 s, similar to that illustrated in Fig. 1, A2, which had emerged after the rapid decay of the initial 100-pA transient. After approximately 20 s, there was a progressive increase in inwardly directed baseline current, reaching a plateau of similar amplitude (~ 200 pA) as recorded intially (Fig. 1, B1). After extensive washing, the baseline current signal remained superimposed with microscopic fluctuations. These fluctuations largely disappeared and the macroscopic baseline current shifted to less negative values (by approximately 10 pA) when 50 µM bicuculline was coapplied with allopregnanolone (Fig. 1, B3). The absolute baseline current and associated variance values recorded in the presence of bicuculline were close to those initially measured before the first application of allopregnanolone (Fig. 1, B1). Removal of bicuculline (after ~30 s of coapplication with the steroid) while continuing to apply allopregnanolone led to an immediate negative shift in baseline current level along with noticeable fluctuations. This low-amplitude response (~20 pA) slowly relaxed over approximately 20 s, leaving a residual sustained signal and fluctuations, which were similar to those illustrated in Figs. 1, A2 and B2. This again was followed by a progressive increase in inwardly directed baseline current together with intensified fluctuations identical to those occurring previously in the same cell. Inclusion of bicuculline together with allopregnanolone during the plateau phase immediately blocked most of the delayed current response and associated fluctuations (Fig. 1, B3). Recovery in drug-free saline revealed the slow decay of the delayed current response as it relaxed toward control levels.

Thus applications of bicuculline during responses to allopregnanolone rapidly and reversibly eliminated most of the effects of allopregnanolone on the baseline current signal, including those remaining from previous applications, implicating persistent activation of GABA_A receptor/Cl channels after the steroid was ostensibly cleared from the continuously perfusing saline. Furthermore, the transient, sustained, and delayed phases of the allopregnanolone-induced current responses (Fig. 1, C1) were completely eliminated when bicuculline was applied before, during, and after application of allopregnanolone (Fig. 1, C2). Similar results were obtained with picrotoxin (not shown), another antagonist of GABA at GABA_A receptor/Cl channels. These pharmacological results indicate that GABA_A receptor/Cl channel activation likely underlies all phases of the allopregnanolone-induced triphasic current response.

Some cells did not exhibit such reproducible allopregnanolone-induced large-amplitude currents, but instead the amplitude of the delayed response progressively diminished with repeated 60-s applications (not shown). In other cells, repeated applications of allopregnanolone shortened the delay before induction of the high-amplitude response. This variability in the allopregnanolone-induced currents was not due to changes in available GABA_A receptor/Cl channels, since responses to GABA remained quite consistent in time course and exhibited quite constant amplitudes.

Interestingly, coapplication of both bicuculline and allopregnanolone for approximately 60 s, which blocked spontaneous activation of GABA_A receptor/Cl channels and shifted the baseline to less negative values, followed by removal of bicuculline but not allopregnanolone immediately resulted in a large-amplitude (approximately 100-200 pA), slowly decaying current response (Fig. 1, D3) that was subsequently followed by the delayed large-amplitude response similar to that recorded in the same cell in response to allopregnanolone without bicuculline (Fig. 1, D1). Surprisingly, the slowly decaying "off response," which emerged immediately on removal of bicuculline, did not require the presence of allopregnanolone, since it also occurred following coincident termination of both bicuculline and allopregnanolone (Fig. 1, D2). The off response occurring after application of bicuculline (and allopregnanolone) was terminated was greater in amplitude and much slower in decay (  16 s) than the initial transient response to allopregnanolone (Fig. 1, A2), which consistently lasted several seconds. Thus the off response was not clearly related to the initial transient. The exponential decay of the off response was faster (~16 s) than that calculated for exponential relaxation after termination of the delayed response to allopregnanolone, which was about 25-26 s, even when both response amplitudes were similar (Fig. 1, D3). Presumably, the off response reflects the rate of unblocking of bicuculline from GABA_A receptor/Cl channels, which have in some way been affected by the steroid.

The off response emerging after terminating both bicuculline and allopregnanolone reveals that overt activation of GABA_A receptor/Cl channels was not required for the underlying effects of the steroid metabolite to become apparent. This is consistent with the emergence of delayed current responses to allopregnanolone in the absence either of an initial transient (Figs. 1, A1, B2, and D1) or a steady low-amplitude signal at 80 mV (Fig. 1, D1). Thus the delayed and progressive increase in macroscopic current reflecting recruitment of active GABA_A receptor/Cl channels did not require either of the earlier phases, although the latter also involved GABA_A receptor/Cl channel activity.

Drugs effective at GABA_A receptor/Cl channels modulate different components of the triphasic current response to allopregnanolone

The antagonistic effects of bicuculline and picrotoxin, which are thought to interact in competitive and noncompetitive ways with GABA activation of GABA_A receptor/Cl channels, were compared directly on the delayed current phase induced by allopregnanolone. Coapplication of either bicuculline or picrotoxin with allopregnanolone during this phase rapidly attenuated the macroscopic current and associated microscopic fluctuations (Fig. 2, A1 and A4). Recovery from the bicuculline block was rapid and exponential (  0.7 s), while recovery from the picrotoxin block was relatively slow and exponential (  16 s) (Fig. 2, A2 and A4). These results indicate that the blocking effects of the classic antagonists of GABA at GABA_A receptor/Cl channels also likely involve different pharmacological mechanisms when the channels are activated by allopregnanolone.

View larger version (36K): [in this window] [in a new window]   Fig. 2. Differential effects of GABAA receptor modulators on the triphasic current response to allopregnanolone. E19 hippocampal neurons cultured for one day were clamped at 80 mV with CsCl-filled pipettes and current responses to 3, 5 were studied with well-established modulators of GABAA receptor/Cl channels. A1: 3, 5 induces a large-amplitude delayed current response, which is rapidly and reversibly blocked by transient exposure to coincident application of 50 µM bicuculline. A2: recovery from the blocked condition is exponential and rapid (<1 s). A3. Coincident application of 50 µM picrotoxin (PTX) blocks the steroid-induced current to the same extent as bicuculline. A4. Recovery is exponential and slow (approximately 17 s). B1: 3, 5 induces a triphasic current response the initial transient and delayed current phases of which are variably potentiated by 10 µM diazepam (DZP) and are slow to recover. B2: amplified trace shows that the transient phase is potentiated approximately fivefold without change in its relaxation and with little effect on the steady low-amplitude current phase. B3: 3, 5 induces a triphasic current response, the intermediate and delayed (but not the initial transient) phases of which are potentiated by 10 µM pentobarbital (PB) with rapid recovery. B4. The amplified trace shows that while the transient phase is not potentiated, the steady low-amplitude current is.

Two clinically relevant drugs (diazepam and pentobarbital) established to potentiate GABA activation of GABA_A receptor/Cl channels throughout the CNS were tested on the steroid-induced triphasic current responses. Diazepam markedly potentiated the initial rapidly decaying transient, which in three cells was increased four- to fivefold (Fig. 2, B1 and B2). The decay of the potentiated transient response paralleled that recorded before diazepam and was followed by a steady low-amplitude current (approximately 5-10 pA), which approximated that recorded before diazepam (Fig. 2, B2). In addition, the peak amplitude of the delayed phase was increased, but less than twofold. Full recovery required extensive washing. In contrast, pentobarbital had no apparent effect on the amplitude of the initial transient response but instead potentiated the subsequent low-amplitude phase, which was increased to approximately 20-75 pA and superimposed with the intensified fluctuations reflecting increased channel activity (Fig. 2, B3 and B4). Pentobarbital also increased the delayed phase of the steroid-induced response, again less than twofold, and recovery from these effects did not require such extensive washing.

These results provide further support for the notion that all three phases of the allopregnanolone-induced triphasic current response involve activation of GABA_A receptor/Cl channels, which exhibit similar sensitivities to antagonists of GABA at GABA_A receptors but varying sensitivities to modulatory substances known to potentiate the effects of GABA at GABA_A receptors.

Allopregnanolone concentration-response curve reveals different potencies for triggering transient and sustained phases

We carried out dose-response curves in four hippocampal neurons to investigate the concentration requirements for activating the different phases of the triphasic current response to allopregnanolone. Submicromolar levels generated a low-amplitude (approximately 10-40 pA) delayed current response in the absence of either the initial transient or the steady low-amplitude current composing the intermediate phase, both of which required supramicromolar levels for activation (Fig. 3).

View larger version (28K): [in this window] [in a new window]   Fig. 3. Different concentration requirements for activation of different phases of the triphasic current response to allopregnanolone. E19 hippocampal neurons cultured for one day were clamped at 80 mV and then exposed to increasing concentrations of 3, 5 with extensive washing between each application. A. Representative effects of 3, 5 show that the lowest concentrations tested (50 and 500 nM) induce responses of low-amplitude after a delay without triggering the initial transient or the immediate steady low-amplitude phase. The latter phases emerge at [3, 5]  1 µM as the delayed phase increases in amplitude. Note change in calibration bar. B. Normalized plot of results obtained in four neurons, which have been fitted by the Hill equation (see text). Hill coefficients for activating each phase are similar (approximately 3), but the potency (or KD value) is five-fold greater (or the KD lower) for evoking the delayed phase compared with the earlier phases.

The amplitudes of the three phases to the current response evoked by allopregnanolone increased in a sigmoidal manner with allopregnanolone concentration and were well fitted from the following Hill equation
where I is the current amplitude, I_MAX is the maximum current, [allo] is the concentration of allopregnanolone, K_d is the dissociation constant of allopregnanolone with its putative receptors/binding sites, and n is the Hill coefficient. While the apparent Hill coefficients were similar (3.2, 2.6, and 2.7 for the initial transient, low-amplitude steady signal, and delayed high-amplitude current response, respectively), the K_d value for the delayed current (1.0 µM) was significantly lower than those for the transient and low-amplitude steady current responses, which were similar (5.1 and 5.2 µM, respectively). These results demonstrate different concentration requirements for activating the three phases of the allopregnanolone-induced current response involving GABA_A receptor/Cl channels, with the slowly emerging delayed phase activated at the lower concentrations.

Cl ions mediate the allopregnanolone-induced sustained current responses

The pharmacological results with known antagonists and modulators of GABA at GABA_A receptor/Cl channels strongly suggest that the triphasic current response to allopregnanolone involves Cl conductance mechanisms. We focused on the voltage sensitivities and ion dependencies of the sustained phases in the allopregnanolone-induced current response. The delayed phase of the response could be elicited at both negative and positive potentials (Fig. 4A). One-second ramp commands from 80 to +40 mV under control conditions and during the delayed phase were used to construct corresponding current-voltage (I-V) plots, which, when subtracted, revealed the I-V relations of the steroid-induced signal (Fig. 4B). In three experiments, the latter reversed polarity at approximately 0 mV (E_Cl), which is consistent with the delayed current response exclusively involving GABA_A receptor/Cl channels. When the Cl gradient was altered using K⁺ aspartate instead of CsCl in the pipette, allopregnanolone evoked a sustained outward current response at 0 mV (Fig. 4C). These results indicate that neither Cs⁺- or Cl -filled cells, which were used in most of the experiments, are prerequisite to the steroid activation of GABA_A receptor/Cl channels, which can be induced at positive membrane potentials. Furthermore, replacement of extracellular Na⁺ and Ca²⁺ did not eliminate either the immediate low-amplitude or the delayed high-amplitude phases of the current response to allopregnanolone (Fig. 4D).

View larger version (27K): [in this window] [in a new window]   Fig. 4. Ionic determinants of the delayed current induced by allopregnanolone. A: neuron was recorded with a CsCl-filled pipette with ECl  0 mV. Delayed current responses to 3, 5, which have opposite polarities, are evoked at negative and positive holding potentials. B: 1-s voltage ramp commands were applied before and during a delayed current response to 3, 5. The subtracted current-voltage relationship is linear, reverses polarity at 0 mV (~ECl), and gives a slope conductance of approximately 50nS. C: 3, 5 induces a delayed current response at 0 mV when the patch pipette contains K aspartate instead of CsCl, consistent with the shift in ECl from approximately 0 mV to negative potentials. D: 3, 5 induces an immediate low-amplitude steady current followed by the delayed phase in a neuron bathed in Na+- and Ca2+-free saline and clamped at a negative potential using a CsCl-filled pipette.

Together with the pharmacological results, these findings lead us to conclude that the sustained phases involve primarily, if not exclusively, Cl ion conductance mechanisms, most likely associated with GABA_A receptor/Cl channel activity, and both can be elicited independently of membrane potential or extracellular cations.

Unitary properties inferred for Cl channels activated by allopregnanolone are similar to those estimated for GABA but the relative proportions of short and long burst-length durations are complementary

We analyzed the microscopic fluctuations associated with the sustained and delayed phases of the current response to allopregnanolone using spectral techniques and compared these results to those calculated for GABA-induced current responses. Fluctuation analyses of sustained and stable current responses permitted estimates of the properties of the population(s) of activated receptor/channels underlying the macroscopic signals. Spectra calculated for the current responses were all well fitted by two Lorentzian terms, reflecting two exponentially distributed burst-length durations for the activated Cl channels (Fig. 5). Although there was considerable variability in absolute values estimated for the exponentially distributed burst-length durations, _SHORT and _LONG, on average, the pair of burst-length durations calculated for allopregnanolone and GABA were not statistically different. _LONG values were 49.8 ± 4.3 ms (GABA, n = 14; mean ± SE) and 51.1 ± 2.9 ms (allopregnanolone, n = 11), while _SHORT values were 2.8 ± 0.3 ms (GABA) and 3.2 ± 0.4 ms (allopregnanolone) (P > 0.05). The sustained low-amplitude current signal evoked by the steroid that preceded the high-amplitude signal also involved both short and long burst-length durations whose values were similar to those associated with the high-amplitude signal (Fig. 5B). However, the relative contributions of the two kinetic components to the Cl current responses induced by GABA and allopregnanolone were complementary. Most of the power in the delayed current response to allopregnanolone (78.8 ± 3.3%) was carried by Cl channels with short burst-length durations, while most of the power in the current evoked by GABA (77.2 ± 2.7%) was conveyed by Cl channel activity with long burst-length durations. The low-amplitude phase of the steroid-induced response also involved predominantly Cl channels with short burst-length durations (Fig. 5B). There were no significant differences in the estimated unitary conductances of the Cl channels underlying each of the currents: 18 ± 4 pS (allopregnanolone low- and high-amplitude currents) and 19 pS ± 5 pS (GABA).

View larger version (29K): [in this window] [in a new window]   Fig. 5. Different proportions of Cl channels with similar kinetics underlie current responses to GABA and allopregnanolone. Spectral analyses of Cl current responses to GABA and allopregnanolone were carried out with the cell clamped at 80 mV to compare estimates of the underlying unitary properties associated with the activated GABAA receptor/Cl channels (see METHODS for details). Representative spectra of current responses elicited on the same neuron show that each can be fitted by two Lorentzian components (identified by downward arrows reflecting corner frequencies), which represent exponentially-distributed burst-length durations of channel activity. A: baseline current variance exhibits a monotonic 1/f spectrum whose power declines directly with increasing frequency. Most of the power in the GABA-evoked current (74%) is conveyed via Cl channel activity with long burst-length durations (LONG = 83 ms). B: spectra of the low-amplitude current induced by 35, which exhibits <10% of the power in the high-amplitude signal, demonstrate similar LONG and SHORT values. Most of the power in both steroid-induced currents (76%) is associated with Cl channel activity having short burst-length durations (SHORT = 2.2-3.6 ms).

These results show that the estimated unitary Cl channel properties underlying the intermediate and delayed phases of allopregnanolone-induced current responses closely resemble those activated by exogenous GABA, although the relative proportions of short- and long-lasting openings are quite different. Allopregnanolone activates primarily short-lasting channel activity, while GABA predominantly opens channels with long-lasting burst-length durations.

Steroidal activation of GABA_A receptor/Cl channels requires 3 substitution and can be triggered in nonneuronal cells transfected with ₁ and ₂ subunit transcripts

The structure-activity relationship underlying the steroidal activation of GABA_A receptor/Cl channels was studied using steroids with different substitutions. Only Ring A-reduced steroid metabolites with 3-substitutions triggered the three phases described for allopregnanolone (Fig. 6A). Inclusion of 1 µM allopregnanolone in the patch pipette to introduce the steroid into the cell did not induce any changes in baseline current (n = 2 cells; not shown). The time course of the response to the clinically useful anesthetic steroid alfaxalone (whose A ring is also 3 substituted) was different from those induced by naturally occurring 3-substituted metabolites. The macroscopic current response increased in an approximately linear rather than sigmoidal manner, while the relaxation was quite rapid unlike the natural compounds. These differences were not pursued further. 5- and 5- reduced steroids with 3-substitution were entirely ineffective, as were progesterone and 5-pregnanolone without 3-substitution (Fig. 6B). Furthermore, 3-substituted steroid metabolites did not block the effects of 3-substituted ones when the former were coapplied at 10 times higher concentration (not shown). The requirement for 3-substitution of the A ring and an extracellular site for activation are consistent with previous studies on embryonic rat hippocampal neurons (Harrison et al. 1987b).

View larger version (23K): [in this window] [in a new window]   Fig. 6. Steroid-induced Cl current responses in hippocampal neurons require 3 substitution and triphasic responses can be triggered in transfected nonneuronal cells expressing 1 and 2 GABAA receptor subunits. A and B: hippocampal neurons were clamped at 80 mV and micromolar concentrations of steroids with different substitutions on the A Ring or different structures were applied. Only steroid metabolites with 3 substitutions of the A Ring are effective in triggering triphasic current responses. A current response with faster on and off rates is also triggered by alphaxalone (alfax), which contains a 3-substituted A Ring. C: 3, 5 triggers a triphasic response in a cell transfected with 1 and 2, subunit transcripts, but not in those transfected with either 1 and 1 or 1 and 3 subunits, which do respond to GABA (see Fig. 7B). The delayed component of the triphasic current is blocked by coincident application of 50 µM bicuculline.

We used nonneuronal (human embryonic kidney and Chinese hamster ovary) cells transfected with one of three combinations of two GABA_A receptor subunit transcripts (_{1 2}, _{1 2}, or _{1 3}) to study the subunit requirements for evoking the triphasic current response to allopregnanolone. All transfected cells tested responded to GABA (not shown, but see Fig. 7, B and C), indicating widespread functional expressions of the heteromeric constructs. Allopregnanolone triggered the same triphasic current response as recorded in embryonic hippocampal neurons in cells transfected with ₁ and ₂ subunits but not in cells transfected with either ₁ and ₂ or ₁ and ₃ subunits (Fig. 6C). Coincident application of bicuculline during the delayed phase of the current response rapidly and reversibly depressed it, closely paralleling its effects on delayed current responses in hippocampal neurons (e.g., Figs. 1, B3 and 2, A1). Application of bicuculline before, during, and after allopregnanolone eliminated all of the triphasic current response (not shown).

View larger version (20K): [in this window] [in a new window]   Fig. 7. Allopregnanolone potentiates GABA-induced Cl currents independently of the triphasic current response. Cells were clamped at 80 mV with CsCl-filled pipettes and brief pulses of 5 µM GABA were applied before during and after exposure to 3, 5. A: amplitudes of current responses to repeated applications of GABA in a hippocampal (HPC) neuron (downward signals) show that there is a delay before the potentiating effects of 3, 5 become evident. Further potentiation occurs as the delayed current phase evolves and then disappears with recovery. B: 3, 5 potentiates current responses to GABA in nonneuronal cells transfected with either 1 and 2 or 1 and 3 subunit transcripts, neither of which exhibit the triphasic current response.

These results demonstrate that 1) steroidal activation of GABA_A receptor/Cl channels in embryonic hippocampal neurons requires 3-substitution of a Ring A-reduced metabolite and 2) in the complete absence of GABA, the triphasic current response can be closely mimicked in nonneuronal cells by activating GABA_A receptor/Cl channels composed only of ₁ and ₂ subunits. Thus the triphasic response to allopregnanolone involves direct and/or indirect activation of GABA_A receptor/Cl channels rather than modulation of GABA_A receptor/Cl channels activated by GABA.

Allopregnanolone potentiation of Cl current responses to GABA occurs independently of the triphasic current response

Brief pulses of GABA were applied before, during, and after allopregnanolone to compare possible potentiating effects of the steroidal metabolite with the triphasic current response. The potentiating effects of allopregnanolone on the Cl currents evoked by GABA in embryonic hippocampal neurons were not immediate but required at least a 5- to 10-s delay to become evident (Fig. 7A). Multifold potentiation of the GABA-induced Cl current emerged coincident with the delayed phase. Recovery from potentiation required more than 25 s. Thus the time course in the potentiating effects of allopregnanolone on pharmacological responses to GABA largely paralleled the delayed phase. However, the potentiating effects could be triggered independently of the triphasic response, since potentiation of GABA-induced Cl currents by allopregnanolone occurred in cells transfected with either 1 and 2 (Fig. 7B) or 1 and 3 (Fig. 7C), which did not exhibit the triphasic response to allopregnanolone.

We also examined the effects of allopregnanolone on spontaneous bicuculline-sensitive GABAergic Cl transients, which appeared in neurons after approximately 3-4 days in culture. Although the frequency of transients was consistently too low to compile amplitude histograms, allopregnanolone clearly increased the time course of transient decay. We detected three types of transient decay following uninterrupted rising phases of transients, which are characteristic of unitary all-or-none signals: short (<10 ms), long (>50 ms), and biphasic with both short and long decays (not shown). Allopregnanolone led to longer-lasting transient decays after a delay, which coincided with the delayed phase of the current response to allopregnanolone (Fig. 8, A and B). Although the effects of allopregnanolone were not studied in detail, the longer-lasting transients were most clearly prolonged. Quantitative analysis showed that the long-lasting decay of the transients remained monoexponential and, after a delay, increased severalfold in the presence of allopregnanolone (Fig. 8B). These effects on GABAergic transients gradually disappeared with extensive washing coincident with the recovery of the baseline current.

View larger version (25K): [in this window] [in a new window]   Fig. 8. Allopregnanolone increases the time constant of GABAergic transient decay. Embryonic hippocampal neurons cultured for three days were clamped at 80 mV with CsCl-filled pipettes. Downward-going events of variable amplitude are bicuculline-sensitive GABAergic transients involving GABAA receptor/Cl channels. A: spontaneous GABAergic transients with fast and slow components to their decay (a) are altered after a delay (b and c), with the slower component becoming dominant. B: after a delay, coinciding with the progressive appearance of the high-amplitude current response, 3, 5 increases the monoexponential decay of GABAergic transients from 70 ms to 312 ms.

Sustained phases of allopregnanolone-induced current are pertussis toxin sensitive

The possibility that some phases of the allopregnanolone-induced current response involved relatively indirect rather than direct binding to and activation of GABA_A receptor/Cl channels prompted us to investigate the effects of pertussis toxin and mastoparan, which are known to inactivate and activate G proteins, respectively (Higashijima et al. 1990; Klinker et al. 1996). In four cells, brief exposure to pertussis toxin eliminated the steady low-amplitude current completely and markedly attenuated the delayed large-amplitude current in a reversible manner (Fig. 9A). Pertussis toxin did not affect Cl current responses to GABA (not shown). The initial transient induced by allopregnanolone was not investigated. Mastoparan only induced the delayed current phase in three cells exhibiting all three phases in response to allopregnanolone and this was consistently lower in amplitude (Fig. 9, B1 and B2). The mastoparan-induced delayed current response was sensitive to bicuculline (Fig. 9C). These results demonstrate the involvement of pertussis toxin-sensitive G protein-coupled pathways in the generation of both sustained phases of the current response to allopregnanolone.

View larger version (24K): [in this window] [in a new window]   Fig. 9. Sustained phases of the current response to allopregnanolone involve pertussis toxin-sensitive G proteins. Embryonic hippocampal neurons cultured for one day were clamped at 80 mV with CsCl-filled pipettes. A1: 3, 5 induces both the steady low-amplitude and delayed high-amplitude phases. A2: exposure to pertussis toxin (1 µg/ml) for 10 min and inclusion of the toxin in the saline containing 3, 5 eliminates the low-amplitude signal and markedly attenuates the delayed response. A3: both recover fully after a short period of washing. B1: 3, 5 induces a typical triphasic current response in another neuron. B2: mastoparan, which activates G proteins directly, evokes a delayed current of lower amplitude. C: mastoparan induces the delayed current response in another neuron and this is blocked in a reversible manner by 50 µM bicuculline.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Modulatory and direct affects of allopregnanolone on GABA_A receptor/Cl channels

Modulatory and direct affects of allopregnanolone on GABA_A receptor/Cl channels catalogued in a wide variety of well differentiated CNS preparations have demonstrated that 3 substitution of the A Ring and binding at an extracellular site are prerequisite to pharmacological activity (for recent review, see Rupprecht and Holsboer 1999). Similar structural requirements were found in the present study of steroid-mediated triphasic current responses. Transient and sustained components of bicuculline-sensitive Cl currents induced by 2 µM allopregnanolone have recently been reported in recordings of acutely dissociated juvenile rat medial preoptic neurons (Haage and Johansson 1999). Thus multiphasic activation of GABA_A receptor/Cl channels by steroid metabolites may persist beyond development and extend to neurons in other regions. Our results are consistent with the hypothesis that a steroid binding site exists at the extracellular domain of the GABA_A receptor/Cl channel (Paul and Purdy 1992; Lambert et al. 1995). In this regard, an allosteric interaction of the anesthetic steroid alphaxalone with the N-terminal region of the second transmembrane (TM) domains of ₂ and/or ₁ subunits has been identified in recombinant studies involving chimeric GABA-glycine receptor/Cl channels (Rick et al. 1998). Direct effects of 1 µM alphaxalone were not recorded in cells expressing 2 1-containing GABA_A receptors; however, a low-amplitude transient was induced in cells expressing chimeric receptors composed of GABA 1 and glycine 1, which were joined near the start of TM2 (E site) but not when they were connected near the start of TM3 (X site). Interestingly, minimal structural requirements for effects of steroid metabolites were initially demonstrated in the first recombinant studies using transfected cells (Puia et al. 1990).  subunits were also not required for steroidal potentiation of GABA's activation of GABA_A receptor/Cl channels in this early study while homomeric ₁ and heteromeric _{1 1} subunit receptor/Cl channels were both sensitive. More recent structure-activity studies of recombinant GABA_A receptors carried out in frog oocytes have revealed that the  subunit isoform determines the efficacy but not the potency of steroidal potentiation of GABA_A receptor activity, while specific  subunit isoforms, when present, determine both the maximal efficacy and the potency (Maitra and Reynolds 1999). Collectively, these results indicate that there are multiple sites for steroidal interaction with heteromeric GABA_A receptor/Cl channels composed of ,  and  subunits. However, the extracellular binding sites remain to be elucidated.

3-substituted steroid metabolite modulation of postsynaptic GABA_A receptor/Cl channels involves G proteins and PKC phosphorylation

The minimal structural requirements for the effects of steroid metabolites on GABA_A receptor/Cl channels do not provide clear support for a specific binding site(s) despite the stereospecificity in the steroid pharmacology. Recently, steroidal modulation of GABAergic Cl transients recorded in adult hypothalamic slices has been demonstrated to involve G protein-coupled phosphorylation involving protein kinase C (PKC) (Fancsik et al., 2000). The potentiating effects of allopregnanolone included phosphorylation via PKC activity since antagonists of the latter blocked the steroid's effects on GABAergic transients. But, inclusion of the PKC agonist phorbol 12- myristate 13- acetate in the pipette did not mimic the steroid pharmacology. Thus although it is uncertain how specific G proteins are coupled to PKC to transduce the steroidal effect on GABA_A receptor/Cl channels, the results imply that at least some of the steroid effects involve second messenger signal transduction pathways. Since allopregnanolone modulation of GABA_A receptor/Cl channels has been shown to occur in outside-out patches (Twyman and Macdonald 1992) the second messenger components appear to be closely co-localized with GABA_A receptors in the pipette-enclosed membrane.

3-substituted steroid metabolite modulation of voltage-dependent Ca²⁺ channels involves G proteins and PKC phosphorylation

Interestingly, 3-substituted steroid metabolites have been reported to modulate voltage-dependent Ca²⁺ channels via pertusis toxin-sensitive G proteins and PKC activation (ffrench-Mullen et al. 1994). The steroid effects required extracellular application and led to a reduction in a major component of voltage-activated Ca²⁺ current. Although 3-substituted metabolism were not evaluated in this study, Fancsik et al. (2000) found that allopregnanolone depressed the frequency of the spontaneous tetrodotoxin-sensitive GABAergic transients. In addition, Haage and Johansson (1999) reported that allopregnanolone altered the frequency of minia