Agonist-specific Gating of NMDA Receptors

Abstract

The mechanism by which ligand binding at extracellular receptor domains gates a transmembrane ion-conducting pore is insufficiently understood. Examining a channel’s activation reaction in the presence of agonists with distinct efficacies may inform this issue and may help identify agonist-dependent transitions. We have recently applied this approach to NMDA receptors composed of GluN1 and GluN2A subunits. Based on our results with several subunit-specific partial agonists we concluded that agonist effects were distributed over several of the multiple transitions that make up NMDA receptor gating and that these changes were subunit independent. Here we examine an additional GluN2A partial agonist, 4-fluoro-D, L-glutamic acid, and we summarize the observed kinetic changes of all nine partial agonists investigated. These results support the premise that regardless of the subunit-type to which they bind, agonists influence multiple equilibria within the NMDA receptor reaction and may stabilize a slightly different family of conformers.

Introduction

NMDA receptors are ligand-gated ion-channels that require both glycine (Gly) and glutamate (Glu) binding to extracellular receptor domains before the transmembrane pore opens^{1, 2}. This distinctive feature of NMDA receptors results from specific binding-sites for Gly and Glu on GluN1 and GluN2 subunits, respectively and the incorporation of two GluN1 and two GluN2 subunits in the functional tetrameric receptor^3-5. In addition, the gating reaction is initiated only after all agonist-binding sites are occupied. A number of low efficacy ligands have been identified for both subunits^6-8. The ability to trace the contributions of GluN1- or GluN2-specific ligands to the overall gating mechanism has been exploited to investigate the mechanism of partial agonism at NMDA receptors and to identify possible subunit-specific steps within the gating reaction^9-11.

Recently we examined current traces generated by individual GluN1/GluN2A receptors activated with series of partial agonists in the presence of the full co-agonist and compared these with traces obtained with the endogenous agonists Gly and Glu. Based on these results we concluded that partial agonists produced sweeping changes in the single-channel kinetics of the receptors studied, and that these changes were not subunit-specific¹². However, aside from common mechanisms observed across all the ligands tested our results also revealed agonist-specific kinetic effects. To focus on this latter, insufficiently emphasized aspect of partial agonism at NMDA receptors, we characterized the kinetics of GluN1/GluN2A receptors with an additional Glu-site partial agonist, 4-fluoro-D, L-glutamic acid (FGA), and we summarize here the observed kinetic changes of all nine partial agonists investigated. We interpret these results to propose that regardless of the subunit-type to which they bind, partial agonists influence multiple steps within the NMDA receptor gating reaction. In addition, each ligand can produce distinct activations suggesting that it may stabilize a slightly different spectrum of NMDA receptor conformers.

Results

To investigate the effects of the partial agonist 4–fluoro–D, L glutamate (FGA) (Fig.1A) on the gating kinetics of NMDA receptors, we recorded single-channel current traces from on-cell patches containing only one GluN1/GluN2A receptor after including in the patch pipette saturating concentrations of FGA, 1 mM (EC₅₀ = 0.1 mM)⁷ and the endogenous co-agonist Gly (0.1 mM, EC₅₀= 2.6 μM)¹³. In these experimental conditions, we obtained four one-channel records that were each 10 - 70 min long and contained together >70, 800 events over 137 min of recording.

Figure 1.

Activation of GluN1/GluN2A receptors by the glutamate-site partial agonist FGA. (A) Chemical structures of glutamate (Glu) and 4-fluoro-D,L-glutamate (FGA); single-channel open probabilities (P_o) are indicated for each ligand. (B) Continuous current trace (30 s) recorded from a one-channel on-cell patch activated with FGA and Gly (displayed filtered at 1 kHz, open is down). Dashed box isolates the period expanded below (displayed filtered at 5 kHz). (C) Top, closed interval distribution for the record illustrated in (B) (36 min; 123,671 events). Inset, time constants (τ) and relative areas (a) for the E₁ - E₅ closed exponential components (thin lines). Bottom, E₂ and E₃ closed components for FGA (black) and for Glu (blue) are superimposed to illustrate the specific changes observed for these two components. (D) Open interval distributions for FGA (black) and Glu (blue). Inset, time constants (τ) and relative areas (a) for each ligand. (E) The state model illustrated was fitted to the entire sequence of open and closed intervals in each FGA–record and means of the optimized rate constants (n = 4) are given in s^-1 for each transition. Asterisks indicate rates that were significantly different from those optimized with data obtained with Glu (n = 5) (p < 0.04).

The overall appearance of single-channel currents elicited by FGA and Gly were similar with those we obtained previously with Glu and Gly (Fig.1B)¹². The calculated open probability for the observed GluN1/GluN2A channels in the presence of FGA was P_o = 0.33 ± 0.04 (n = 4). This represents 52% of the reported single-channel activity with Glu and Gly (P_o = 0.64 ± 0.06, n = 5)¹² and is in close agreement with the 55% macroscopic efficacy measured for this agonist relative to Glu⁷. This congruence suggested that the ensemble of behaviors we captured in our single-channel recordings represented a fair sample of the activity previously measured in whole-cell recordings.

Inspection of the mean durations of closed and open intervals (Fig. 1C, D) indicated that the lower efficacy of FGA as compared to Glu could be attributed mainly to ~2-fold longer closures: 15 ± 3 ms (FGA) vs. 6.4 ± 1.7 ms (Glu) since the calculated reduction in mean open durations was not statistically significant: 7.4 ± 0.8 ms (FGA, n = 4) vs. 11.2 ± 1.1 ms (Glu, n = 5, p > 0.08).

Because each of the records analyzed reflected the activity of the same channel throughout, it was informative to compare entire closed interval distributions for the two agonists tested. This analysis showed that the increase in the mean closed time could be assigned specifically to an increase in the time constants for the second and third exponential components E₂ and E₃: τ₂, 2.32 ± 0.15 ms vs. 1.74 ± 0.08 ms; and τ₃, 7.14 ± 0.7 ms vs. 4.6 ± 0.19 ms (Fig. 1C). In addition, except for the fastest closed component E₁, all remaining four closed components E₂ – E₅, had larger areas for FGA as compared to Glu. The result that FGA, a glutamate-site agonist prolonged both the τ₂ and τ₃ time constants is in agreement with our previous report that ligands at either GluN1 or GluN2 have similar effects on gating, at least at the level of closed time constants¹². However, these results also expose distinct changes to the structure of open and closed interval distributions elicited by FGA when compared to other partial agonists at the same site, such as HCA, SYM, HQA and QA^{9, 12}

To examine in closer detail the kinetic mechanism of FGA, we fitted a linear state model to the sequences of open and closed intervals recorded (Fig. 1E). This analysis identified the activation rate constant k₂₁, the deactivation rate constant k₂₃, and the rate for entry into a short desensitized state k₃₅, as the only three out of the ten rate constants that were significantly different for FGA as compared to those obtained by fitting the same model to Glu-elicited currents. This result further argues for a lack of correlation between the activated subunit and the observed kinetic changes, and for a delocalized, diffuse effect of partial agonists on the gating reaction. However, the changes we observed with this ligand also brought forth the important point that each of the partial agonists examined so far owed their lower efficacies to subtly distinct mechanisms.

A qualitative summary of gating changes elicited by a series of partial agonists, including FGA is illustrated Table 1. This survey led us to conclude that when compared to the endogenous agonists Glu and Gly, each of the ligands studied was less efficacious in unique ways. For example, although ALA, ACBC, SYM and QA alike decreased all three forward rates (k₃₂, k₂₁, k_1O), their effect on the magnitude of the reverse rate constants was dissimilar: QA had Glu-like closing rate constants; ALA increased only k₁₂; whereas ACBC and SYM, despite acting at separate subunits, increased the same two closing rate constants k₁₂ and k_O1. Based on this analysis we conclude that at least for the GluN1/GluN2A receptors investigated here, each partial agonist is ‘partial’ for different reasons. Still, in all cases the effects were scattered across all rate constants. ACBC and HCA showed the most pervasive changes, with eight out of ten rate constants being significantly different (p<0.05). In contrast, DCS and FGA showed the most localized effect, changing only three rate constants: k₃₂, k_1O, and k₁₂ for DCS; and k₂₁, k₂₃, and k₃₅ for FGA.

Table 1.

Agonists-specific gating of GluN1/GluN2A receptors

rate	GluN1 agonists				GluN2A agonists					agonist sensitivity
	DCS	ALA	ACPC	ACBC	HCA	SYM	HQA	QA	FGA
activation
k32	↓	↓	↓	↓	↓	↓	—	↓	—	7
k21	—	↓	↓	↓	—	↓	↓	↓	↓	7
k1O	↓	↓	—	↓	↓	↓	—	↓	—	6
deactivation
k23	—	—	—	—	—	—	↑	—	↑	2
k12	↑	↑	—	↑	↑	↑	↑	—	—	6
kO1	—	—	↑	↑	↑	↑	↑	—	—	5
entry into desensitized state
k35	—	↓	↓	↓	↓	↓	↓	↓	↓	8
k24	—	—	—	↑	↓	—	↑	—	—	3
recovery from desensitized state
k53	—	—	—	—	—	—	—	—	—	0
k42	—	—	—	↓	↓	—	—	↓	—	3

Summary of agonists-specific changes, relative to the natural agonist, in the rate constants optimized by fits with the model in Fig. 1E. Arrows indicate significant (p<0.05) increase (upward arrow, red) or decrease (downward arrow, blue) in the value for the respective rate constant. Except for FGA investigated in this study, all other data are compiled from Kussius and Popescu 2009 (ref. 12)

The results summarized in Table 1 also suggest that almost all the rate constants postulated in the reaction mechanism are sensitive to the identity of the agonist activating the channel. Although none of the ligands tested changed all rate constants, each rate constant was sensitive to at least one of the partial agonists examined, except for k₅₃, which represents recovery from the main desensitized state C₅. Only two of the agonists tested, HQA and FGA, both acting at GluN2A subunits, increased the closing rate constant k₂₃. This may indicate that this rate constant and the recovery rate from desensitization, k₅₃, are the least agonist-sensitive transitions. Most ligands decreased the activation rate constants k₃₂ and k₂₁ and also decreased the desensitization rate constant k₃₅. This result may indicate that agonists that are less effective at opening the channel are also less effective at desensitizing the receptor.

Discussion

Fast intercellular communication in the nervous system is largely accomplished by chemical activation of ion-channels^{14, 15}. The mechanism by which the energy resulting from the ligand-binding reaction is allocated within the agonist-receptor complex remains poorly understood¹⁶. A seminal advance in attempting quantitative descriptions of fast synaptic transmission was achieved when the theory put forth for substrate-enzyme pairs in 1913 by Michaelis and Menten¹⁷ was extended in 1957 by del Castillo and Katz¹⁸ to ligand-channel pairs to postulate both quiet (AR) and active (AR*) conformations for agonist-receptor complexes. This formalism divorced binding from gating and defined agonist efficacy as the ratio of opening and closing rate constants. It postulated that the AR ↔ AR* equilibrium is agonist-dependent and the AR and AR* complexes are energetically, structurally and functionally similar regardless of the identity of bound ligand. Modern views of gating mechanisms in neurotransmitter-gated channels attest to a more complex picture^{11, 19-23}.

The vast majority of ligand-gated channels appear to assemble as heteromeric complexes of multiple subunits; have multiple, often non-equivalent ligand-binding sites, and in the case of NMDA-sensitive glutamate receptors, have dual-agonist requirement¹⁵. The ideal concept of a two-state gating mechanism is most likely inadequate, a fact plainly demonstrated by the multiplicity of states observed directly in single-channel records for many receptors. Given this molecular complexity, it is not surprising that a mechanistic and quantitative description of agonist efficacy is still lacking. We owe the current view of agonism at glutamate-gated channels to accumulating structural information about various receptor elements.

The observation that glutamate receptor subunits have structurally distinct modules and that the isolated domains preserve their function upon excision from the core protein, initiated a rewarding line of inquiry in the structure and functionality of isolated modules^24-29. In particular, the investigations focused on extracellular agonist-binding domains have shaped the modern view of agonist-mediated gating³⁰. Essentially, these studies have demonstrated that the agonist-binding modules have an overall clamshell shape; ligands bind at the interface of two mobile domains; and the principal difference between the apo and agonist-bound forms rests with a more compact, closed clamshell conformation of the liganded form^{30, 31}. Importantly, for soluble ligand-binding domains partial agonists induced conformations with intermediate degrees of cleft closure as compared to the full agonists on one end and antagonists at the other³². In addition these seminal findings also revealed that regardless of the degree of closure induced, each ligand may stabilize subtly different conformations of the complex³³. It is not known whether the agonist-specific conformations observed in the structures of the isolated ligand-binding domains also translate to distinct families of conformers for full-length functional receptors. Insight into this important issue must come from investigations at the single-molecule level because this is the only known experimental technique that can classify receptor populations based on their active-inactive status and life-time distributions³⁴.

In the experiments summarized in this article, all ligands tested had similar although not identical effects on the gating of GluN1/GluN2A receptors and all gating transitions observed showed agonist-sensitivity. The differences observed in the mechanisms by which each ligand modified receptor activity did not correlate with the subunit-type specifically recognized by the particular agonist suggesting that the free-energy gained from agonist binding at either the GluN1 or the GluN2A subunit makes mechanistically similar contributions to channel activation. However, each agonist produced activations with distinct kinetic patterns perhaps reflecting that entities with unique chemistry and architecture produce specific rearrangements within the binding-pocket. The observations that each agonist initiated different gating reactions and that all gating steps were sensitive to the nature of bound-agonist strongly suggest that the family of conformations elicited by each agonist are not identical. The implication is that agonist-sensitive transitions extend beyond the closing of the clamshell to yet-to-be identified downstream gating elements over which both subunit types have equivalent control.

Despite these recent insights, the issue of partial agonism at ligand-gated channels is far from settled, with the first wave of single-channel investigations leading to somewhat dissimilar conclusions. For example, in the case of the pentameric channels of the nicotinic receptor family, agonist effects appeared localized to only the initial pre-gating step, called priming or flipping^{22, 23}. For tetrameric, AMPA-type glutamate receptors partial agonists shifted the receptor occupancies toward lower conductance states³². Further, within the NMDA receptor family, experiments done in excised patches identified two pre-opening transitions to be agonist- and subunit-specific^{9, 11}, whereas our on-cell experiments summarized here, show distributed ligand-specific but subunit-independent effects of agonists on NMDA receptor gating transitions.

Whether these discrepancies reflect true mechanistic differences or limitations inherent to the preparations or methodologies used remains to be clarified by additional experimentation. To reveal the still hidden mechanism by which ligand binding produces gating it will be necessary to bring to congruence structural data on a variety of functional conformations and kinetic information regarding the route and time course of their interconversion.

Materials and Methods

Cells and transfections

HEK cells (ATCC CRL-1573, gift from Dr. Auerbach, Buffalo, NY) were maintained in Dulbecco’s Modified Eagle Medium (DMEM, Invitrogen, Grand Island, NY) at 37°C in a 5% CO₂ atmosphere and were passed when reaching 80 -90 % confluence. Passages 22 – 35 were used for transfections.

Plasmids encoding rat GluN1-1a (U08261), rat GluN2A (M91561), and GFP were gifts from Drs. Wenthold (Washington, DC), Auerbach (Buffalo, NY) and Kosman (Buffalo, NY). Open reading frames were sub-cloned into pcDNA3.1(+) under the control of the CMV promoter. Complete sequence of each insert was confirmed periodically by sequencing.

Calcium phosphate-mediated transient transfections were done by incubating the cells for 2 hours with mixtures consisting of (in mM): 140 NaCl, 5 KCl, 0.75 Na₂HPO₄, 6 sucrose, 125 CaCl₂, 25 HEPES/NaOH, pH 7.05, and ~1 μg cDNA (GluN1:GluN2A:GFP = 1:1:1) per 35-mm dish containing cells at ~50% density. After removing the precipitate, the cells were grown 24 – 48 hrs in DMEM supplemented with 2 mM Mg²⁺.

Before each experiment, cells were washed, covered with PBS and placed on the stage of an inverted microscope. Individual cells were selected visually, based on fluorescence intensity and patch clamp recordings were obtained while the cells were still attached to the dish.

Electrophysiology

Steady-state single-channel currents were recorded with the cell-attached patch-clamp technique³⁵. The electrodes were filled with solutions containing (in mM): 150 NaCl, 2.5 KCl, 1 EDTA, 10 HEPBS, 0.1 glycine and 1 FGA (Sigma), adjusted to pH 8 (NaOH). Inward sodium currents were elicited by applying + 100 mV through the recording pipette. Currents were amplified and low pass filtered at 10 kHz (Axopatch200B; 4-pole Bessel), sampled at 20 – 40 kHz (PCI-6229, M Series card, National Instruments, Austin, TX) and written onto computer hard drives with QUB acquisition software (www.qub.buffalo.edu, Buffalo, NY).

Processing and analyses of single-channel current records

Each digital file was displayed on computer monitors with QUB software and inspected visually for simultaneous openings, signal-to-noise ratios, high-frequency artifacts and baseline drift. In the conditions tested, due to low receptor desensitization rate constants (< 10 s^-1) and high open probability within bursts (P_o, 0.3 – 0.6), records consisted of long, almost solid clusters of activity separated by zero-current periods lasting seconds (see Fig. 1B). As a result, records originating from patches containing two or more active channels revealed many superimposed bursts, and were easily detected and excluded from further analyses. Baseline drifts were corrected by resetting the baseline to zero-current levels, as necessary.

Preprocessed data were idealized in QUB with the SKM algorithm after digitally low-pass filtering at 12 kHz. Idealized records were subjected to kinetic analyses and modeling with the MIL algorithm in QUB as described previously^{12, 19}

Significance of differences observed was evaluated with Student’s t-tests assuming equal variance; differences were considered significant for p < 0.05;

Abbreviations and Acronyms

ACBC

1-Aminocyclobutane-1-carboxylic acid

ACPC

1-Aminocyclopropane-1-carboxylic acid

ALA

L-Alanine

DSC

D-Cycloserine

FGA

4-Fluoro-D,L-glutamic acid

GluN1

Glycine-binding NMDA receptor subunit (previously abbreviated NR1)

GluN2

Glutamate-binding NMDA receptor subunit (previously abbreviated NR2)

HCA

L-Homocysteic acid (L-2-amino-4-sulfobutyric acid)

HQA

Homoquinolinic acid (pyridine-2-carboxy, 3-methylcarboxylic acid)

NMDA

N-Methyl-D-aspartic acid

QA

Quinolinic acid (pyridine-2, 3-dicarboxylic acid)

HEPES

N-(2-Hydroxyethyl)piperazine-N’-(2-ethanesulfonic acid)

HEPBS

N-(2-Hydroxyethyl)piperazine-N’-(4-butanesulfonic acid)

SYM

SYM2081 (2S,4R)-4-Methylglutamic acid