 |
The Journal of Neuroscience, September 14, 2005, 25(37):8439-8450; doi:10.1523/JNEUROSCI.1407-05.2005
 Previous Article | Next Article 
Cellular/Molecular
20-Oxo-5 -Pregnan-3 -yl Sulfate Is a Use-Dependent NMDA Receptor Inhibitor
Milos Petrovic,1
Miloslav Sedlacek,1
Martin Horak,1
Hana Chodounska,2 and
Ladislav Vyklick , Jr1
1Institute of Physiology, Academy of Sciences of the Czech Republic, 142 20 Prague 4, Czech Republic, and 2Institute of Organic Chemistry and Biochemistry, 166 10 Prague 2, Czech Republic
 |
Abstract
|
|---|
NMDA receptors are ligand-gated ion channels permeable to calcium and play a critical role in excitatory synaptic transmission, synaptic plasticity, and excitotoxicity. They are heteromeric complexes of NR1 combined with NR2A-D and/or NR3A-B subunits that are activated by glutamate and glycine and whose activity is modulated by allosteric modulators. In this study, patch-clamp recordings from human embryonic kidney 293 cells expressing NR1/NR2 receptors were used to study the molecular mechanism of the endogenous neurosteroid 20-oxo-5 -pregnan-3 -yl sulfate (35S) action at NMDA receptors. 35S was a twofold more potent inhibitor of responses mediated by NR1/NR2C-D receptors than those mediated by NR1/NR2A-B receptors. The structure of the extracellular loop between the third and fourth transmembrane domains of the NR2 subunit was found to be critical for the neurosteroid inhibitory effect. The degree of 35S-induced inhibition of responses to glutamate was voltage independent, with recovery lasting several seconds. In contrast, application of 35S in the absence of agonist had no effect on the subsequent response to glutamate made in the absence of the neurosteroid. A kinetic model was developed to explain the use-dependent action of 35S at NMDA receptors. In accordance with the model, 35S was a less potent inhibitor of NMDA receptor-mediated EPSCs and responses induced by a short application of 1 mM glutamate than of those induced by a long application of glutamate.
These results suggest that 35S is a use-dependent but voltage-independent inhibitor of NMDA receptors, with more potent action at tonically than at phasically activated receptors. This may be important in the treatment of excitotoxicity-induced neurodegeneration.
Key words: neurosteroids; NMDA receptor; inhibition; patch-clamp recording; recombinant receptors; culture
|
Introduction
|
|---|
NMDA receptors, a subtype of ionotropic glutamatergic receptors, mediate a slow component of EPSCs and participate in different forms of synaptic plasticity. However, their overactivation can induce excitotoxicity that may lead to symptoms characteristic of severe diseases of the nervous system in man (Doble, 1999 ; Sattler et al., 2000); (Bliss and Collingridge, 1993; Liu et al., 2004). Structurally, NMDA receptors are heteromeric complexes that consist of different combinations of three subunit types: NR1, NR2, and/or NR3 (Dingledine et al., 1999). Receptors composed of NR1 and NR2 subunits require two coagonists for activation: glycine, which binds to NR1, and glutamate, which binds to NR2 (Kuryatov et al., 1994; Laurie and Seeburg, 1994).
NMDA receptor activity can be influenced by exogenous and endogenous ligands, including neurosteroids. These compounds are of steroid origin, having a direct nongenomic effect on neuron excitability (Paul and Purdy, 1992). They are both synthesized de novo in the CNS and modified to neuroactive compounds from circulating precursors and can reach a high local concentration (Mellon and Griffin, 2002; Kawato et al., 2003). It has been well established that neurosteroids exert a positive, negative, or combined effect on NMDA receptors (Wu et al., 1991; Bowlby, 1993; Park-Chung et al., 1994; Horak et al., 2004). Pregnenolone sulfate (PS), an endogenously occurring neurosteroid, acts in a subunit-dependent manner as a combined positive and negative modulator of NMDA receptors (Malayev et al., 2002; Horak et al., 2004). The mechanism of action of PS associated with a positive effect is disuse dependent (i.e., NMDA receptor affinity for PS is decreased during receptor activation) and involves an increase in the peak probability of the NMDA receptor opening (Po). In contrast, 20-oxo-5-pregnan-3-yl sulfate [pregnanolone sulfate (35S)], also a naturally occurring neurosteroid that differs from PS in a single unsaturated bond, has an inhibitory action at NMDA receptors (Park-Chung et al., 1994; Abdrachmanova et al., 2001). Although the molecular mechanism of 35S action at NMDA receptors is only partially understood, its synthetic analog 35 hemisuccinate ester has been shown to have a neuroprotective effect, both in vitro (hippocampal neurons) and in vivo (spinal cord ischemia model), thereby indicating its potential therapeutic use (Weaver et al., 1998; Lapchak, 2004).
In the present study, we examined the mechanism of action of 35S using whole-cell and single-channel recordings from neurons and human embryonic kidney 293 (HEK293) cells transfected with NR1 and NR2A-D subunits of the NMDA receptor. Our results show that 35S is a use-dependent but voltage-independent inhibitor that exerts its effect by reducing the peak Po of NMDA receptor channels. Its inhibitory action is weaker for responses mediated by NMDA receptors activated by synaptically released glutamate than those tonically activated by the agonist.
|
Materials and Methods
|
|---|
All experiments were performed in accordance with the European Communities Council Directive (86/609/EEC) and with the approval of the Institutional Animal Care and Use Committee.
Hippocampal cultures. Primary dissociated hippocampal cultures were prepared from 1- to 2-d-old postnatal rats using established methods (for details, see Mayer et al., 1989). Rats were decapitated, and the hippocampi were dissected. Trypsin digestion, followed by mechanical dissociation, was used to prepare a cell suspension. Single cells were plated at a density of 100,000 cells/cm2 onto a confluent glial feeder layer prepared 2 weeks earlier. Neuronal cultures were maintained in a medium composed of MEM, 10% horse serum, and a nutrient supplement consisting of transferrin, insulin, selenium, corticosterone, triiodothyronine, and progesterone. A metabolic inhibitor, 5-fluoro-2'-deoxyuridine, was used to suppress cell division.
Transfection and maintenance of HEK293 cells. HEK293 cells (ATTC number CRL1573; American Type Culture Collection, Rockville, MD) were cultured in Opti-mem I (Invitrogen, Gaithersburg, MD) with 5% fetal bovine serum at 37°C. The day before transfection, cells were plated in 24-well plates (2 x 105 cells per well) in 0.5 ml of their normal growth medium and became confluent on the day of transfection. Equal amounts (0.3 µg) of cDNAs coding for NR1, NR2, and green fluorescent protein (pQBI 25; Takara, Tokyo, Japan) were mixed with 2 µl of Lipofectamine 2000 (Invitrogen). DNA–Lipofectamine complexes were added to HEK293 cells for 5 h. The cells were then trypsinized, resuspended in Opti-mem I containing 1% fetal bovine serum supplemented with 20 mM MgCl2, 1 mM D, L-2-amino-5-phosphonopentanoic acid, and 3 mM kynurenic acid, and plated on 31 mm poly-L-lysine-coated glass coverslips.
The following genes encoding NMDA receptor subunits were used: NR1–1a (GenBank accession number U08261
[GenBank]
) (Hollmann et al., 1993), NR1–1b (GenBank accession number U08263
[GenBank]
) (Hollmann et al., 1993), NR2A (GenBank accession number D13211
[GenBank]
) (Ishii et al., 1993), NR2B (GenBank accession number M91562
[GenBank]
) (Monyer et al., 1992), NR2C (GenBank accession number M91563
[GenBank]
) (Monyer et al., 1992), and NR2D (GenBank accession number L31611
[GenBank]
) (Monyer et al., 1994). NR2A3C4A (NR2AM1-S632CV644-M834AL824-V1464) chimera was a generous gift from Dr. G. Westbrook (Oregon Health and Science University, Vollum Institute, Portland, OR) (Krupp et al., 1998; Vissel et al., 2002).
Animals and slice preparation. Neocortical rat slices were prepared from juvenile Wistar rats of either sex (postnatal day 13–14). The brains were rapidly removed, hemisected, and transferred to cold (4°C) artificial CSF (ACSF) of the following composition (in mM): 130 NaCl, 2.5 KCl, 28 NaHCO3, 1.25 NaH2PO4, 1 MgCl2, 2 CaCl2, and 25 glucose (gassed with 95% O2/5% CO2 to a nominal pH of 7.4). Brain slices 250 µm thick were cut using a vibratome (DTK-1000; Dosaka, Kyoto, Japan) and incubated in the ACSF at a temperature of 36°C for 45 min. Slices were maintained at room temperature (6–8 h) before being placed in a recording chamber, in which they were continually perfused with ACSF at a rate of 3 ml/min.
Recording from cultured cells and drug application. Experiments were performed 24–48 h after the end of HEK293 transfection; neurons maintained in culture for 5–8 d were used. Whole-cell voltage-clamp recordings were made with a patch-clamp amplifier (Axopatch 1D; Axon Instruments, Union City, CA) after capacitance and series resistance (<10 M ) compensation of 80–90% (Hamill et al., 1981; Stuart et al., 1993). Agonist-induced responses were low-pass filtered at 1 kHz with an eightpole Bessel filter (Frequency Devices, Haverhill, MA), digitally sampled at 5 kHz, and analyzed using pClamp software version 9 (Axon Instruments). Patch pipettes (3–4 M) pulled from borosilicate glass were filled with an intracellular solution containing the following (in mM): 125 gluconic acid, 15 CsCl, 5 EGTA, 10 HEPES, 3 MgCl2, 0.5 CaCl2, and 2 ATP-Mg salt, pH adjusted to 7.2 with CsOH.
Pipettes for outside-out recording were coated with Sylgard (Dow Corning, Brussels, Belgium) and had a resistance of 5–10 M. Single-channel currents were filtered at 2 kHz with an eight-pole Bessel filter, sampled at 20 kHz, and analyzed using pClamp software. Openings were defined as a transition from closed to open state using a 50% threshold criterion. Openings briefer than 255 µs (1.5x filter rise time) were excluded from the analysis (Colquhoun and Sakmann, 1985).
Mean current (MI) was calculated as follows:
 | (1) |
where Ai and ti are the amplitude and duration of individual NMDA receptor channel openings, and d is duration of data sample (usually 15–30 s). Relative effect (RE) of neurosteroid on NMDA-induced receptor channel activity was defined as follows:
 | (2) |
where MISteroid and MIControl are mean currents induced by NMDA receptor channel openings in the presence of neurosteroid and control extracellular solution.
Extracellular solution (ECS) contained the following (in mM): 160 NaCl, 2.5 KCl, 10 HEPES, 10 glucose, and 0.2 CaCl2, pH adjusted to 7.3 with NaOH. In some experiments, the concentration of CaCl2 was increased to 2 mM, and 1 mM MgCl2 was added as indicated. Single-channel recordings were made using ECS containing 0.7 mM CaCl2 and 0.2 mM EDTA. NMDA was used as an agonist to activate native NMDA receptors, recombinant NMDA receptors when fast deactivation of response was desirable (see Fig. 2 B), and receptors in outside-out patches. Glycine (10 µM) (except in experiments aimed to determine peak Po, when it was 30 µM), an NMDA receptor coagonist, was present in the control and test solutions. TTX (0.5 µM) and bicuculline methochloride (5 µM) were used in experiments on cultured hippocampal neurons. 35S solutions were made from freshly prepared 20 mM stock in dimethylsulfoxide (DMSO). Control experiments were performed in extracellular solution containing DMSO at the same concentration that was present in solutions containing steroids.
A microprocessor-controlled multibarrel fast perfusion system, with a time constant of solution exchange around cells of  10 ms, was used to apply test and control solutions (Vyklicky et al., 1990). To study the deactivation kinetics of glutamate responses and the peak Po, small lifted cells were used. A brief application of glutamate (6 and 20 ms as indicated) was made by a stepping motor-controlled movement of double-barreled theta tubings (Hilgenberg, Malsfeld, Germany), and 10–90% rise time was 0.58 ± 0.12 ms (n = 5).
Peak Po was calculated according to Jahr (1992) and Chen et al. (1999) as follows:
 | (3) |
Parameters a and b are defined as follows:
 | (4) |
where QPostMK is the total charge transfer of the response induced by a brief application of glutamate recorded after a single application of glutamate made in the presence of 20 µM (+)-5-methyl-10,11-dihydro-5H-dibenzo[a,d]cyclohepten-5,10-imine maleate (MK-801), and QControl is the total charge transfer of control glutamate-induced response; and
 | (5) |
where QMK20 is the charge transfer that occurs within 20 ms after the beginning of the glutamate response made in the presence of 20 µM MK-801, and QMK is the total charge transfer of the response induced by glutamate made in the presence of MK-801.
Experiments were performed at 23–26°C at a holding potential of –60 mV, except when indicated otherwise.
Recording from slices. Neocortical neurons were visualized in slices using a combination of near-infrared and differential interference contrast microscopy (Axioskop-FS; Zeiss, Jena, Germany). Pyramidal neurons in the layer II/III were identified by their position in the cortical slices, triangular somatic shape, and bases of apical dendrites. Whole-cell patch-clamp recordings (Hamill et al., 1981; Stuart et al., 1993) were made using an Axopatch 200B amplifier (Axon Instruments) at room temperature (23–26°C). To elicit EPSCs, a second pipette filled with extracellular solution and positioned in layer I 50–100 µm from the postsynaptic neuron was used. Current pulses of constant amplitude (10–30 µA) and duration (0.5 ms) were applied at a 30 s intervals. EPSCs were recorded at a holding potential of +40 mV with no correction for the liquid junction potential (+9 mV). The responses were filtered at 2 kHz with an eight-pole Bessel filter, sampled at 5 or 10 kHz, and analyzed using pClamp software. NMDA receptor EPSCs were recorded in the presence of control and test solutions. A solution containing neurosteroid was prepared from ACSF bubbled with O2/CO2 for at least 30 min, to which an appropriate amount of 20 mM stock 35S in DMSO was added. This solution was kept in contact with the carbogen atmosphere without bubbling to avoid neurosteroid precipitation. Control experiments were performed in ACSF containing DMSO at the same concentration that was present in test solution containing the steroid. Both the control and test solutions were routinely supplied with CNQX (5 µM), 10 µM bicuculline methochloride, 1 µM strychnine, and 10 µM glycine.
The drugs were purchased from Sigma (St. Louis, MO) or Tocris Cookson (Avonmouth, UK). Neurosteroid, 35S, was synthesized by H.C. according to a previously published method (Arnostova et al., 1992), and its purity (>98%) was repeatedly tested using nuclear magnetic resonance, thin-layer chromatography, and elemental analyses.
Statistics. Results are presented as mean ± SD, with n equal to the number of cells studied; statistical comparison of groups was performed using Student's t test or a one-way ANOVA. p < 0.05 was used to determine the significance.
Kinetic modeling. To describe the effects of 35S on NMDA receptor channels, we have adopted the kinetic model of NR1/NR2B receptor function proposed recently by Banke and Traynelis (2003). Simulations and fitting were performed using Gepasi software (version 3.21) (Mendes, 1993, 1997; Mendes and Kell, 1998).
|
Results
|
|---|
35S inhibition is influenced by the NMDA receptor subunit composition
To investigate the influence of the composition of the NMDA receptor on the effect of 35S, cDNAs encoding for the NR1 and NR2 subunits were cotransfected into HEK293 cells. Fig. 1A shows control responses to the fast application of glutamate (1 mM) in HEK293 cells transfected with NR1–1a and NR2B subunits recorded in the presence of 10 µM glycine, no added Mg2+, and 0.2 mM Ca2+. The responses to a coapplication of 30 µM 35S and glutamate made after the onset of the response to glutamate were inhibited, but less in NR1–1a/NR2B receptors than in NR1–1a/NR2C receptors. The results of the dose–response analysis indicate significantly (approximately twofold) lower potency of 35S to inhibit NR1–1a/NR2A-B receptors and native receptors in cultured hippocampal neurons than to inhibit NR1–1a/NR2C-D receptors (Table 1). No significant differences were found in the mean values of the Hill coefficient (h), which varied from 1.2 to 1.5. These result agree well with those published recently by Malayev et al. (2002).
We have shown recently that PS, a neurosteroid structurally similar to 35S, has a positive and negative effect on NMDA receptors and further that the structure of the extracellular M3–M4 loop of the NR2 subunit is critical for the expression of both effects (Malayev et al., 2002; Horak et al., 2004, 2005). To explore the molecular basis of 35S action at NR2 subunits, which may be similar to the inhibitory effect of PS, we used a chimera (NR2A3C4A) in which the extracellular M3–M4 loop of the NR2A subunit, which is less sensitive to the inhibitory effect of 35S, was exchanged for the homologous region of the NR2C subunit, which is more sensitive to the inhibitory effect of 35S. Analysis of the dose–response effect of 35S on chimeric receptors (NR1–1a/NR2A3C4A) indicated a significant increase in the neurosteroid potency compared with the effects of 35S at NR1–1a/NR2A receptors (Table 1). These results stress the importance of the M3–M4 loop of the NR2 subunit for the 35S effect; however, they also indicate that there may be other structural and/or functional characteristics of the receptor that affect the neurosteroid effect. Next, experiments were performed to test the role of the NR1 subunit. NR1 subunits exist in eight splice variants differing by the presence (denoted b) or absence (denoted a) of a 21 amino acid insert in the extracellular domain of the N terminus (Hollmann et al., 1993). The potency of 35Sto inhibit glutamate responses mediated by NR1–1b/NR2B receptors was not significantly different from its potency to inhibit responses mediated by NR1–1a/NR2B receptors.
View larger version (23K):
[in this window]
[in a new window]
|
Figure 2. 35S is a use-dependent inhibitor of NMDA receptor channels. Examples of records obtained from HEK293 cells transfected with NR1–1a and NR2B. A1, Response to coapplication of 300 µM 35S and 1 mM glutamate (Glu) made after the onset of the response to the agonist was inhibited by 95.3% and recovered from the inhibition on a slow timescale. A2, The onset of the response to application of 1 mM glutamate made immediately after preapplication of 300µM 35S for 37 s was rapid, similar to the control glutamate response from A1. A3, To illustrate the difference in the time course of responses to glutamate made after coapplication of 35S and glutamate (b; boxed area in A1) and that made after preapplication of 35S (c; boxed area in A2), both responses are shown superimposed on an expanded time-scale. A4, The rise time of the responses to glutamate made after preapplication of 35S (c; boxed area shown in A2) was similar to that of control response to 1 mM glutamate (a; boxed area shown in A1). B1, Five consecutive responses to 100 µM NMDA, coapplication of 35S (300 µM), and recovery after varying durations (0, 0.3, 0.45, 0.6, and 1 s) of the neurosteroid washout in the absence of agonist (Wash) are shown. B2, The recovery from inhibition is shown on an expanded timescale. Note that the application of control extracellular solution immediately after coapplication of neurosteroid and NMDA resulted in an after-response (indicated by asterisk). C1, Analysis of the rate of 35S unbinding from NMDA receptors in the absence of the agonist. Data points (open circles) are averaged values of the normalized amplitude of the quick phase of the response to agonist reapplication plotted versus duration of the neurosteroid washout in the absence of agonist. Error bars represent SD; n = 6. Mean time course of the recovery of NMDA-induced responses made immediately after coapplication of 35S and NMDA (neurosteroid dissociation from liganded receptors) is represented by dots (scatter plot; traces recorded in 6 cells were digitally averaged). C2, Model 1 and model 2 (for values of the rate constants, see diagram of NMDA receptor states in Fig. 8 B, Table 2) were used to simulate the response to 1 mM glutamate made immediately after coapplication of 300 µM 35S and 1mM glutamate. Recovery from inhibition in the continuous presence of glutamate (lines) and recovery from inhibition after varying durations (0–5 s) of the neurosteroid washout in the absence of agonist (circles and triangles for model 1 and model 2, respectively).
|
|
35S is a use-dependent inhibitor
We have shown previously that PS is a disuse-dependent NMDA receptor modulator with action that depends on the relative timing of the application of agonist and neurosteroid (Malayev et al., 2002; Horak et al., 2004). To test whether 35S has an action similar to PS at NMDA receptors, two protocols of 35S and glutamate applications were used: (1) a coapplication of 35S and glutamate made after the onset of the response to glutamate and (2) a sequential application of glutamate after preapplication of 35S for 37 s. Figure 2A1 shows responses of NR1–1a/NR2B receptors to a coapplication of 35S (300 µM) with glutamate made after the onset of the response. The neurosteroid induced almost complete inhibition (96.8 ± 1.1%; n = 5) of responses to a saturating concentration of glutamate (1 mM; three orders of magnitude higher than the glutamate EC50). After cessation of the coapplication of 35S, the response to glutamate recovered on a slow timescale, with complex kinetics, characterized by a fast and a slow time course of recovery (10–90% recovery time was 2.7 ± 0.2 s; n = 5). Similar results in terms of the degree of 35S inhibition and the time course of recovery from the inhibition were observed when NMDA (100 µM; threefold higher concentration than the NMDA EC50) was used to study the neurosteroid-induced inhibition in cultured hippocampal neurons.
Subsequent experiments were designed to test the effect of 35S on subsequent responses to glutamate or NMDA when 35S is preapplied in the absence of NMDA receptor agonist. As can be seen from records in Figure 2A, the response to 1 mM glutamate made after a preapplication of 300 µM 35S was significantly faster than that made after coapplication of 35S and glutamate and was not significantly different from that induced by the control response to 1 mM glutamate (made without the neurosteroid preapplication). There were no significant differences in 10–90% rise time between NR1–1a/NR2B receptor responses activated by a low-affinity agonist (500 µM NMDA; 12 ± 2 ms; n = 6) and those activated after a preapplication of 35S (300 µM) (13 ± 3 ms; n = 6). These results show that the inhibitory action of 35S at NMDA receptors depends on the receptor activation, and they indicate a neurosteroid use-dependent mechanism of action on this type of glutamate receptors.
Differences in the time course of NMDA receptor responses made after preapplication and coapplication of 35S indicate that the 35S binding kinetics differs substantially for agonist-liganded and non-agonist-liganded receptors. Next, experiments were performed that aimed to determine the time course of 35S unbinding from NMDA receptors after agonist washout. In these experiments, neurosteroid washout of varying durations (0.3–5 s) in the absence of NMDA receptor agonist was introduced after a coapplication of 100 µM NMDA and 300 µM 35S. The NR1–1a/NR2B receptor response after reapplication of NMDA demonstrated both quickly and slowly rising phases (Fig. 2B2). It is tempting to conclude that the presence of a quick phase reflects the proportion of receptors that unbound the neurosteroid in the absence of agonist. Figure 2C1 shows the results of experiments in which the rate of recovery from 35S inhibition of agonist-activated receptors (continuous line) was compared with recovery after varying durations of NMDA washout (dots). Although detailed comparison is complicated by desensitization, the rate of 35S unbinding follows a similar time course regardless of NMDA receptor activation. This implies that the rate of neurosteroid unbinding (in contrast to the neurosteroid binding) is similar for activated and nonactivated receptors.
Figure 3A (trace a) shows that the application of control extracellular solution immediately after coapplication of 300 µM 35S and 1 mM glutamate resulted in an after-response that was 26.2 ± 6.4% (n = 6) of the control response to glutamate. After coapplication of 1 mM glutamate and 300 µM 35S, the occurrence of the after-response was delayed if application of extracellular solution containing 300 µM 35S was interposed for 5 and 10 s before application of control extracellular solution (Fig. 3A, traces b, c). The graph of the amplitude of the after-response as a function of duration of 35S application (Fig. 3B) indicates that dissociation of glutamate from the NR1–1a/NR2B receptor was decelerated by the neurosteroid. These results suggest the existence of neurosteroid-induced conformational state(s) of the NMDA receptor channel complex that prevent dissociation of the glutamate from the neurosteroid-modulated NMDA receptor. If this hypothesis is true, then 35S dissociation will allow an agonist-bound NMDA receptor channel to open before agonist dissociation, resulting in a macroscopic after-response.
Use-dependent inhibition may be produced by the binding of a charged compound inside the ion channel pore, resulting in a voltage-dependent action. Because positively charged compounds produce a voltage-dependent block of NMDA receptor channels, it is not surprising that the effect of a negatively charged compound, such as 35S, is voltage independent at responses induced in NR1–1a/NR2B receptors by 100 µM NMDA at holding potentials of –80 to +60 mV (Fig. 4). This agrees well with conclusions drawn from studies of 35S action at native NMDA receptors (Park-Chung et al., 1994; Abdrachmanova et al., 2001).
35S decreases peak Po of NMDA receptors
Next, experiments were performed that aimed to determine the mechanism by which 35S induces inhibition of NMDA receptor responses. Several mechanisms were considered. The degree of inhibition induced by 30 µM 35S was not dependent on a glutamate concentration of 43.1 ± 2.0 and 42.0 ± 1.3% (n = 5) for 1 mM and 1 µM glutamate, respectively (these agonist concentrations are, respectively, three orders of magnitude higher than and equal to the glutamate EC50). In addition, responses to 1 mM glutamate made in the absence or continuous presence of 30 µM 35S show a similar degree of macroscopic desensitization. This argues against neurosteroid-induced change in the agonist binding and desensitization as the principal mechanisms by which the NMDA receptor channel activity is inhibited.
Following the protocol used by Jahr (1992) and Chen et al. (1999), an open NMDA receptor channel blocker, MK-801, was used as an experimental tool to determine the peak Po of NR1–1a/NR2B receptor channels. Peak Po in control cells was determined in three steps, as described by Jahr (1992) and Chen et al. (1999) and used in our recent experiments (Horak et al., 2004). After a cell was lifted, consecutive 20 ms applications of agonist (1 mM glutamate with 30 µM glycine) were recorded. These were repeated at 20 s intervals for stabilization of peak current amplitude in at least three consecutive records. In the second phase, a single response was evoked in the presence of 20 µM MK-801 in both the control and agonist solutions (Fig. 5A). Subsequent applications of glutamate made in the absence of MK-801 elicited currents with smaller amplitudes but similar time courses compared with the pre-MK-801 responses (Fig. 5A). After washout of MK-801, the charge transfer by the response to glutamate was reduced by 84% compared with the control response made before the MK-801 application. To estimate the peak Po for the peak current, we followed the analysis used by Jahr (1992) and Chen et al. (1999) and measured the fraction of the total charge transfer that occurred during a 20 ms application of glutamate in the presence of 20 µM MK-801; this was 11.7 ± 3.5% (n = 6). The peak Po [calculated as the product of the relative decrease of the charge transfer of the response to glutamate after MK-801 application and the relative charge transfer that occurred within 20 ms after the beginning of glutamate application in the presence of MK-801 (Eq. 3)] was calculated to be 9.9 ± 2.5% (n = 6), which is similar to that reported for the same receptor channel composition by Chen et al. (1999).
The same protocol was used to estimate the peak Po of NR1–1a/NR2B receptor channels affected by 35S. Responses to a 20 ms application of 1 mM glutamate made in the continuous presence of 35S (100 µM) were recorded before and after (Fig. 5B) a single response to the application of 35S and glutamate made in the presence of 20 µM MK-801 in the control and agonist solution (Fig. 5B). The analysis showed that, after washout of MK-801, the charge transfer by the response to glutamate recorded in the continuous presence of 35S was reduced by 75% compared with the response made before the application of MK-801. In the presence of 20 µM MK-801, 5.9 ± 0.8% (n = 6) of the charge transfer occurred within 20 ms after the beginning of the glutamate application (Fig. 5B). The estimated value of the peak Po of the NR1–1a/NR2B receptor channels affected by 35S (100 µM) was significantly decreased, to 4.5 ± 0.6% (n = 6). An independent set of experiments was performed to determine the degree of 35S (100 µM) inhibition of responses to a 20 ms application of 1 mM glutamate in lifted HEK293 cells. The peak amplitude of responses in the presence of 100 µM 35S was inhibited by 59 ± 5% (n = 5), similar to the decrease of the peak Po (by 55%).
We also tested the effect of 35S (100 µM) on the single-channel activity induced by the application of NMDA (10 µM) in outside-out patches isolated from HEK293 cells expressing NR1–1a/NR2B receptors and recorded in the presence of 0.5 mM Ca2+ (0.7 mM CaCl2 and 0.2 mM EDTA) and Mg2+-free extracellular solution at –95 mV. Figure 6 shows an example of the inhibitory effect of 35S on the NMDA-induced single-channel activity. The relative degree of inhibition [expressed as the ratio of mean current induced by NMDA receptor channel openings recorded in the presence of 35S to the control activity (Eqs. 1, 2)] was 61.9 ± 14.4% (n = 6), which is similar to the effect of the neurosteroid on the whole-cell NMDA receptor responses. In the presence of 35S (100 µM), the frequency of NMDA receptor channel openings was significantly reduced, by 54.2 ± 11.2% (n = 6) relative to the control, and the mean open time of NMDA receptor channel openings was significantly shortened from  = 3.02 ± 0.39 ms (n = 6) recorded in the control to = 2.73 ± 0.43 ms (n = 6) in the presence of 35S. In contrast, no significant difference was observed in the distribution of NMDA receptor channel amplitudes recorded in the control (5.2 ± 0.2 pA; n = 6) and the presence of 35S (5.3 ± 0.2 pA; n = 6). The results of the single-channel analysis are compatible with 35S-induced diminution of the peak Po of NMDA receptor channels and indicate that the neurosteroid affects mainly the rate constant of channel opening rather than a channel closing.
View larger version (31K):
[in this window]
[in a new window]
|
Figure 7. The effect of 35S on NMDA receptor-mediated EPSC.A, Example of NMDA receptor-mediated EPSC (4 consecutive responses were averaged) evoked in a cortical lamina II/III pyramidal neurons voltage clamped at +40 mV before (Control) and 23–25 min after the start of 100µM 35S application. On the right, both responses are displayed filtered at 100 Hz, normalized to the peak amplitude of the control response, and overlaid to show the difference in the deactivation time course. B, Plot of the amplitude of NMDA receptor-mediated EPSC as a function of time of recording. Filled bar indicates duration of 100 µM 35S application (same neuron as in A1). C, Examples of the records obtained from HEK293 cells expressing NR1–1a/NR2B receptors. Control response to the fast application of 1 mM glutamate (Control) and that induced in the presence of 35S (100 µM) are shown. Above, a 6 ms application of glutamate is indicated by an arrow. Filtered at 100 Hz, superimposed and normalized responses with respect to the peak amplitude of the control response are shown on the right. Note the accelerated initial component and decelerated late component of the deactivation of glutamate responses recorded in the presence of 35S. D, Model 1 and model 2 (for values of the rate constants, see diagram of NMDA receptor states in Fig. 8 B, Table 2) were used to simulate the response to 6 ms application of 1 mM glutamate (Control) and that in the presence of 100 µM 35S. On the left, the simulated responses are shown superimposed. On the right, they are shown superimposed and normalized with respect to the peak amplitude of the control response. E, Summary of the effect of 35S (30–300 µM) on the amplitude of NMDA receptor-mediated EPSC and responses induced by short (6 ms) and long (5 s) application of 1 mM glutamate in HEK293 cells expressing NR1–1a/NR2B receptors. Predicted inhibition of responses to 6 ms application of 1 mM glutamate by 30–300 µM 35S using model 1 and model 2 is indicated by open and filled arrowheads, respectively. Error bars represent mean ± SD, and the number of cells is indicated in parentheses. For any neurosteroid concentration, there is a statistically significant difference (*p < 0.001) between inhibition of responses induced by long and short coapplications of glutamate and neurosteroid, as well as long coapplications of glutamate and neurosteroid and NMDA receptor-mediated EPSCs.
|
|
The effect of 35S on EPSCs
Next, experiments were performed that aimed at characterizing the effects of 35S on the amplitude and deactivation kinetics of NMDA receptor-mediated EPSCs in lamina II/III pyramidal neurons of the rat cortex. NMDA receptor-mediated EPSCs were evoked in neurons voltage clamped at +40 mV by focal electrical stimulation of the adjacent tissue. After an initial period in which control NMDA receptor-mediated EPSCs were recorded, the perfusion solution was exchanged for extracellular solution containing 35S for a period sufficient to produce a steady-state effect on the amplitude of the EPSCs (usually 15–25 min). Figure 7A shows a typical example of the effect of 100 µM 35Sonthe amplitude and time course of NMDA receptor-mediated EPSCs. In the presence of 35S, the amplitude of NMDA receptor-mediated EPSCs slowly decreased and reached 50% of the amplitude before the neurosteroid application. The effect was slowly reversible, presumably attributable to a slow diffusion of the neurosteroid from the slice (Fig. 7B). Taking into account the profound inhibitory effect of 35S on NMDA- and glutamate-induced whole-cell responses recorded from both neurons and HEK293 cells, it was surprising that the inhibitory effect of 35S on the amplitude of NMDA receptor-mediated EPSCs was significantly smaller at all neurosteroid concentrations tested (30–300 µM) (Fig. 7E). The difference in the relative degree of 35S-induced inhibition of steady-state agonist-evoked responses and NMDA receptor-mediated EPSCs was more pronounced at low neurosteroid concentrations (Fig. 7E).
The decay of NMDA receptor-mediated EPSCs recorded in the control extracellular solution was best fit by a double-exponential function with mean time constants 1 = 108 ± 36 ms (A1 = 46 ± 7%) and 2 = 369 ± 21 ms (n = 9). It has been shown previously that EPSCs in lamina II/III pyramidal neurons are mediated by NMDA receptors composed of NR2A and NR2B subunits with the relative density changing as a function of postnatal age (Flint et al., 1997). This fits well with the results of our experiments showing that NMDA receptor-mediated EPSCs in lamina II/III pyramidal neurons at postnatal day 14 were inhibited by 47 ± 4% (n = 4) when recorded in the presence of ifenprodil (10 µM; selective inhibitor of NR2B-containing NMDA receptors) (Williams, 1993). In the presence of 35S (100 or 300 µM), the deactivation of NMDA receptor-mediated EPSCs had complex kinetics characterized by an accelerated initial and decelerated late component (see overlaid responses in Fig. 7A). Similar effect of the neurosteroid on the deactivation time course of NMDA receptor-mediated EPSCs was observed in all 10 neurons analyzed.
Several mechanisms were considered that would account for the minor effect of the neurosteroid on the amplitude of NMDA receptor-mediated EPSCs, e.g., presynaptic action of the neurosteroid, neurosteroid breakdown, and the presence of endogenous compounds interfering with the neurosteroid action. To exclude mechanisms other than postsynaptic ones, an effect of 35S was studied on responses induced by short applications of glutamate in recombinant NR1–1a/NR2B receptors. Control responses to glutamate (1 mM; applied for 6 ms) were compared with those induced in the continuous presence of 35S (Fig. 7C). At 30–300 µM 35S, the degree of inhibition of responses induced by a brief application of glutamate in lifted HEK293 cell-expressing NR1–1a/NR2B receptors was not significantly different from neurosteroid-induced inhibition of NMDA receptor-mediated EPSCs; however, it was significantly different from a steady-state inhibition of responses to a long application of glutamate (Fig. 7E). These results suggest that a postsynaptic mechanism accounts for the difference in the 35S-induced inhibition of NMDA receptor-mediated EPSCs and responses to a long application of glutamate.
View larger version (17K):
[in this window]
[in a new window]
|
Figure 8. Model of NMDA receptor inhibition by 35S. A1, Diagram of NMDA receptor states; the model is based on a scheme suggested previously by Banke and Traynelis (2003). A model of gating scheme assumes that, after binding of two molecules of agonist (A), NMDA receptor (R) can reach an active state with the ion channel open (O) via either fast or slow conformational change (RA2f, RA2s). One desensitized state (D) was added, as predicted by the single-exponential time course of current response to prolonged glutamate application. A2, According to the use-dependent action of 35S at NMDA receptors, it is assumed that neurosteroid binds to agonist-induced conformational state(s): RA2, D, RA2f, RA2s, and/or O. B, Onset and offset of 35S-induced inhibition of response to glutamate were fit to model 1 (assuming neurosteroid binding to conformational states RA2f and O) or model 2 (assuming neurosteroid binding to conformational states RA2 and O). C, NR1–1a/NR2B receptor response to 1 mM glutamate, coapplication of glutamate with 300 µM 35S, and neurosteroid washout (duration of 35S and glutamate is indicated by filled and open bar, respectively) is represented by dots. The transient induced by glutamate was fitted to the seven-state kinetic model shown in A1. The following rate constants (k(X, Y) indicates rate constant that determines transition from conformational state X to Y) were fixed: k(R, RA = 19 µM–1s–1; k(RA, RA2) = 9.5 µM–1s–1; k(RA, R) = 29 s–1; k(RA2, RA) = 58 s–1; k(RA2, RA2f) = k(RA2 s, O) = 1557 s–1; k(RA2f, RA2) = k(O, RA2s) = 182 s–1; and k(RA2s, RA2) = k(O, RA2f) = 135 s–1. Only rate constants of receptor desensitization and channel opening were allowed to vary during fitting. Fitted rates were as follows: k(RA2, D) = 4.7 s–1; k(D, RA2) = 0.7 s–1; and k(RA2, RA2s) = k(RA2f, O) = 12.7 s–1. The optimally fitted transient from the kinetic model is superimposed (line a). The transient induced by neurosteroid washout was fit to models shown in B, and the rate of 35S unbinding from conformational states NRA2f and NO (model 1) and NRA2 and NO (model 2) was determined. Fitted rates were as follows: k(NRA2f, RA2f) = 55.3 s–1 and k(NO, O) = 0.75 s–1 (model 1); k(NRA2, RA2) = 0.75 s–1 and k(NO, O) = 9.8 s–1 (model 2). The optimally fitted transients using model 1 and model 2 are shown superimposed (line d, e). With rate constants determined in previous steps, the transient induced by neurosteroid application in the presence of glutamate was fit, and the rate constants of 35S binding to conformational states RA2f and O (model 1) and RA2 and O (model 2) were determined. Fitted rates were as follows: k(RA2f, NRA2f) = 0.29 µM–1 s–1 and k(O, NO) = 0.84 µM–1 s–1 (model 1); k(RA2, NRA2) = 0.86 µM–1 s–1 and k(O,NO) = 1.57 µM–1 s–1 (model 2). The optimally fitted transients are shown superimposed (line b, c). Inset shows boxed region of the response and both fits plotted on a logarithmic timescale. Note that both the onset and offset of the neurosteroid inhibition were equally well fit by both models; therefore, lines b, c and d, e overlap.
|
|
The deactivation of control responses mediated by NR1–1a/NR2B receptors after a brief application of glutamate was best fit by a double-exponential function with mean time constants 1 = 212 ± 23 ms (A1 = 41 ± 5%) and 2 = 860 ± 121 ms (n = 5). As can be seen from the superimposed responses shown in Figure 7C, 35S accelerated the initial component and decelerated the late component of the deactivation. The deactivation had complex kinetics, and, similarly to the deactivation of NMDA receptor-mediated EPSCs, we failed to fit it by an exponential function.
Model for 35S action at NMDA receptors
Despite considerable structural and biophysical research, the molecular mechanism of NMDA receptor channel activation is still poorly understood. Several kinetic models were proposed to account for the receptor macroscopic and microscopic properties (Lester and Jahr, 1992; Banke and Traynelis, 2003; Popescu and Auerbach, 2003). Figure 8A1 shows our working model used to infer on the molecular mechanism of 35S action at NMDA receptors. The model is based on a scheme suggested previously by Banke and Traynelis (2003) for the same NMDA receptor subunit combination used in our experiments (NR1–1a/NR2B).
Kinetic analysis of responses induced in lifted HEK293 cells by (1) fast application of 1 mM glutamate, (2) coapplication with 300 µM 35S, and (3) recovery of glutamate responses after the neurosteroid washout was performed in three consecutive steps (Fig. 8C). First, the model was fitted to the macroscopic current response induced by 1 mM glutamate (Fig. 8C, trace a), and the desensitization/resensitization rate constants k(RA2, D) (rate constant that determines transition from conformational state RA2 to D), k(D, RA2), and k(RA2, RA2S) = k(RA2F, O) were determined (Table 2). All other rate constants were the same as those set by Banke and Traynelis (2003) and are listed in the Figure 8 legend. Subsequently, concentrations of all conformational states were determined at the time of initiation of neurosteroid application (4 s after the start of glutamate application).
In all six cells analyzed, only one desensitized state was sufficient to fit the whole-cell responses induced by a saturating concentration of glutamate. This is predicted by a single-exponential time course of current response to a saturating concentration of glutamate. Attempts to use a model with two desensitized states proposed by Banke and Traynelis (2003) provided inconsistent data. It is possible that the increase in the desensitization of NMDA receptor channels after patch isolation from the cell (Tong and Jahr, 1994) is associated with the appearance of a new desensitized state (Banke and Traynelis, 2003) that is rare or nonexisting for receptors with the relatively intact microenvironment expected to occur in the initial states of whole-cell recording. Estimates of the rate constants k(RA2, RA2S) and k(RA2F, O) (values listed in Table 2) predict a peak Po of 6.3%, which agrees well with our experimental data (see above) and estimates made previously (Chen et al., 1999; Horak et al., 2004).
Figure 8A2 shows an overall scheme combining 15 models that were considered in assessing the molecular mechanism of 35S action at NR1–1a/NR2B receptors. The models assume that the neurosteroid (N) binds to single or multiple agonist-induced conformational state(s) of the NMDA receptor with the ion channel closed (RA2, RA2s, and RA2f), desensitized (D), and ion channel open (O). In the initial stages of the analysis, we considered a model assuming two equivalent and independent binding sites for 35S at the NMDA receptor. This assumption was vaguely supported by the results of our experiments indicating the importance of the NR2 subunit for the 35S inhibitory effect and by the conclusion that functional NMDA receptors assemble as heteromers of two NR1 and two NR2 subunits (Laube et al., 1998); it was further supported by a Hill slope for the 35S inhibition of >1. A tentative 11-state model was, however, too complicated and did not allow a reproducible determination of the neurosteroid binding and unbinding rate constants, in contrast to the one-to-one stoichiometry model that was sufficient to fit the data adequately (see below).
For each of the 15 models (two of them are shown in Fig. 8B), rate constant(s) of neurosteroid unbinding from respective conformational state(s) were set as free parameter(s), and the models were fit to the offset of 35S-induced inhibition of glutamate response (for an example, see Fig. 8C, lines d, e). The sum of the squares of residuals (SSR) determined for each model, and cell response (n = 6) was used to compare the goodness of the fits. In all six cells analyzed, the best fit was obtained using model 2, and therefore SSR obtained through this model was used to determine the relative value of SSR when other models were used (Table 3). Statistical comparison of models (Rao, 1973) indicated that all 10 models that assumed neurosteroid unbinding from combinations of two conformational states provided significantly (p < 0.05) better fit than the models that assumed neurosteroid unbinding from a single conformational state (ND, NO, NRA2s, NRA2f, and NRA2) (Table 3). The conformational state NO stands for agonist-activated NMDA receptor with the neurosteroid bou |
Top
Abstract
Introduction
-tube and the solutions used for recording. C, Control extracellular solution; G, 1 mM glutamate; M, 20 µM MK-801. Responses illustrated in A and B are from different cells.
