Introduction

Top Abstract Introduction Methods Results Discussion Appendix References

BIII 277 CL (fig. 1, left) and BIII 281 CL are enantiomeric benzomorphans with neuroprotective and anticonvulsant properties (Carter et al., 1995; Pschorn and Carter, 1996; Yamaguchi S, Kokate T and Rogawski MA, unpublished observations). The novel benzomorphans are structurally related to the prototypic sigma ligand N-allylnormetazocine (SKF 10,047). In addition to its sigma binding properties, SKF 10,047 is well known to interact with mu opiate and NMDA receptors (Zukin, 1982). In contrast to SKF 10,047, which has higher binding affinities for mu and sigma sites than for NMDA receptors, radioligand binding studies have demonstrated that BIII 277 CL and BIII 281 CL are selective for NMDA receptors. Indeed, BIII 277 CL exhibits >200-fold higher binding affinity for NMDA receptors than for mu and sigma sites.

View larger version (9K): [in this window] [in a new window]   Fig. 1.   Structures of BIII 277 CL (left) and (-)-N-allylnormetazocine (right). Both benzomorphans have the absolute configurations 1R. Chiral centers are indicated by asterisks.

BIII 277 CL and BIII 281 CL are unusually high-affinity ligands for the NMDA receptor, as assessed by radioligand binding with the channel blocker [³H]dizocilpine (Huettner and Bean, 1988). As illustrated in figure 1, the (-)-enantiomer of SKF 10,047 has a similar stereochemistry to BIII 277 CL. (-)-SKF 10,047 displaces [³H]dizocilpine with micromolar potency (K_i = 1.5 µM) (Grauert et al., 1997). In contrast, BIII 277 CL displaces [³H]dizocilpine with a K_i value of 4.5 nM, so its binding affinity is comparable to that of dizocilpine itself (K_i = 4.0 nM), one of the most potent NMDA receptor ligands known (Carter, 1995; Rogawski, 1993). BIII 277 CL also potently inhibits NMDA receptor responses in functional biochemical assays (e.g., NMDA-induced [³H]norepinephrine release) and can protect against NMDA-induced lethality in mice (Carter et al., 1995). The distomer BIII 281 CL, although less potent than BIII 277 CL, also possesses relatively high affinity for the [³H]dizocilpine binding site, and with a K_i value of 685 nM, is ~2-fold more potent than (-)-SKF 10,047 (Grauert et al., 1997).

Benzomorphans have a relatively rigid structure. For enantiomers that are stereochemically constrained in this way, structural variations that increase the affinity of the eutomer typically result in a decreased affinity of the distomer. The high binding affinity of BIII 281 CL in comparison with, for example, (+)-SKF 10,047 (K_i = 1.2 µM) is therefore unexpected. To gain insight into the mechanism by which the novel benzomorphans block NMDA receptors and attempt to understand the uncharacteristically high affinity of the distomer, we studied the interaction of BIII 277 CL and BIII 281 CL with NMDA receptors by assessing the effects of the drugs on whole-cell and single-channel NMDA receptor currents in cultured hippocampal neurons. Our findings indicate that BIII 277 CL and BIII 281 CL selectively block NMDA-activated currents in a use- and voltage-dependent manner, probably by binding to a site within the receptor channel pore. Quantitative analysis of the concentration-block relationships and blocking kinetics have allowed us to develop a binding model that may explain the high affinity and unusual binding properties of the enantiomers.

	Methods

Top Abstract Introduction Methods Results Discussion Appendix References

Cell culture. Hippocampal neurons from 19-day-old Sprague-Dawley rat embryos (Harlan, Indianapolis, IN) were grown in monolayer culture on 35-mm polystyrene Petri dishes (Falcon 3001; Becton Dickinson Labware, Oxford, CA) precoated with Matrigel (Collaborative Biomedical Products, Bedford, MA) as described previously (Donevan et al., 1992). The cultures were used for electrophysiological recording after 6 to 12 days in vitro.

Solutions. Before each experiment, the culture medium was replaced with bathing solution containing 145 mM NaCl, 10 mM HEPES, 2.5 mM KCl, 0.1 mM CaCl₂ and 10 mM glucose. The osmolality of the bathing solution was adjusted to 316 to 323 mOsm with sucrose and the pH to 7.4 with NaOH. The bathing solution also contained 1 µM tetrodotoxin to block voltage-dependent Na⁺ channels and 1 µM strychnine to block glycine-activated Cl^- currents. For whole-cell and single-channel recording, patch pipettes were filled with intracellular solution containing 145 mM CsCl, 1.0 mM MgCl₂, 0.1 mM CaCl₂, 10 mM HEPES and 1 mM EGTA. The osmolality of the pipette solution was adjusted to 305 to 310 mOsm with sucrose and the pH to 7.2 with KOH.

Whole-cell and single-channel recording. Patch pipettes (3-8 M) were prepared from filament-containing thin-wall glass capillary tubes (World Precision Instruments, Sarasota, FL) using a four-stage horizontal puller (model P-80/PC Flaming Brown; Sutter Instrument, Novato, CA). Micropipette tips were routinely fire-polished and, for single-channel recordings, were coated with Sylgard (Dow Corning, Midland, MI). Whole-cell and single-channel currents were monitored with Axopatch 1B and 200A patch-clamp amplifiers (Axon Instruments, Burlingame, CA), respectively, and digitally acquired using the Axotape software package (Axon Instruments). Unitary NMDA receptor channel currents were filtered at 1 kHz (-3 dB; four-pole, low-pass Bessel filter) and digitally sampled at 10 kHz. All experiments were performed at room temperature (23-25°C) and, unless otherwise indicated, at a holding potential of -60 mV.

Perfusion. Drug solutions were applied to the cell surface with a rapid perfusion system consisting of a seven- or eight-barrel array of flow tubes in which all barrels emptied via a common glass tip (Tang et al., 1989). One barrel contained bathing solution, and the others contained agonist or agonist plus a blocking drug (BIII 277 CL or BIII 281 CL) except in the experiment of figure 4, in which the blocking drug was at times applied alone. In the whole-cell recordings, test solutions were applied for 10-sec periods separated by wash intervals in which the cell was continuously perfused with bathing solution (except in the experiment of figure 6, in which switching between test solutions occurred immediately). In single-channel recordings, test solutions were applied in 20- to 60 sec-duration epochs, separated by 30- to 90-sec wash periods. The agonist solutions used in the whole-cell recordings were 30 or 300 µM NMDA, 100 µM KA, 100 µM AMPA or 1 µM GABA. In single-channel recordings, a concentration of 2 µM NMDA was used. All NMDA-containing solutions also contained 10 µM glycine. BIII 277 CL and BIII 281 CL were added directly to the agonist-containing solutions from 30 mM stock solutions in H₂O, stored at -20°C. BIII 277 CL and BIII 281 CL were synthesized by Boehringer Ingelheim (Ingelheim am Rhein, Germany). All other drugs and chemicals were obtained from Sigma Chemical (St. Louis, MO) or Aldrich Chemical (Milwaukee, WI).

Data analysis in whole-cell recordings. Percentage block of whole-cell currents was calculated according to the formula B = (1 -I_D/I_o) × 100, where B is the percent block, I_o is the control current evoked by agonist (usually NMDA) at the end of a 10-sec application (before drug application) and I_D is the steady-state current evoked by agonist at the end of a 10-sec application in the presence of the blocking drug. Steady state was determined when there was no further decline of the current in at least three successive drug applications. "Time" in the kinetic analyses represents the total duration of agonist application and assumes that there is negligible recovery from block during the periods between agonist applications. Nonlinear curve fitting was carried out with NFIT (Island Products, Galveston, TX), and statistical comparisons were performed with the PROC NLIN procedure of SAS/STAT (SAS Institute, Cary, NC).

Data analysis in patch recordings. Patch currents were analyzed using the FETCHAN and pSTAT modules of pCLAMP (Axon Instruments). Openings were determined as current level changes exceeding a 50% threshold criterion. Openings briefer than 200 µsec were ignored. For determination of open times and burst durations, only patches demonstrating infrequent multiple openings were used for analysis. Bursts were defined as a series of two or more openings in which closed intervals briefer than 5 msec were ignored. P_o and NP_o values were determined from idealized records; N was taken to be the maximum number of simultaneous openings observed in the experiment. Reported P_o values were determined in patches with infrequent multiple openings and overestimate the true single-channel open probability to the extent that the currents records represent activity from multiple channels.

	Results

Top Abstract Introduction Methods Results Discussion Appendix References

BIII 277 CL and BIII 281 CL block of whole-cell NMDA receptor currents. In whole-cell recordings, perfusion with 30 and 300 µM NMDA (coapplied with 10 µM glycine) elicited rapidly rising inward currents that desensitized ~10% to 20% during 10-sec duration applications. To minimize run down during the long applications of BIII 277 CL and BIII 281 CL required to demonstrate block, NMDA was applied repetitively in 10-sec duration pulses separated by 10-sec wash periods. Pairs of NMDA/blocking drug and wash solution pulses were successively applied until steady-state block was achieved.

View larger version (21K): [in this window] [in a new window]   Fig. 2.   BIII 277 CL and BIII 281 CL block NMDA-evoked currents in a use-dependent fashion. A, Representative traces demonstrating the progressive inhibition by 0.3 µM BIII 277 CL and 0.3 µM BIII 281 CL of currents induced by 30 µM NMDA (+10 µM glycine). A slight inward drift in holding current as observed here often occurred in prolonged recordings. B, Time course of block in experiments similar to those illustrated in A. To assess the extent of run down, a series of experiments were performed with 30 µM NMDA (+10 µM glycine) alone. Each data point represents mean of data from three or four cells; error bars indicate S.E.M. The data for BIII 277 CL () and BIII 281 CL () block were fit to single exponential functions; the control (NMDA + glycine alone; ) data were fit to an arbitrary curve. Unless otherwise noted, the holding potential in this and subsequent figures is -60 mV.

View larger version (19K): [in this window] [in a new window]   Fig. 3.   Concentration-dependent inhibition of NMDA-evoked currents by BIII 277 CL () and BIII 281 CL (). The ordinate indicates the percent steady-state block of currents evoked 30 µM NMDA (+ 10 µM glycine). Each point represents the mean percent block values of data from three or four cells; error bars indicate S.E.M. and where not shown were smaller than the size of the symbols. The data points were fit to the logistic equation B = Bmax/[1 + (IC50/c)n, where n indicates nH, a parameter indicating the steepness fit; B is the percent block; c is the concentration of BIII 277 CL or BIII 281 CL; and IC50 is the concentration resulting in Bmax/2 block. The parameter values are given in the text.

Concentration dependence of block. Figure 3 presents the concentration dependence of the steady-state block in experiments similar to those illustrated in figure 2. The percent block values were fit according to the logistic equation given in the caption of figure 3. The IC₅₀, B_max and n_H values derived from the fits were 5.2 nM, 99.9% and 0.67 for BIII 277 CL and 58 nM, 95% and 1.2 for BIII 281 CL, respectively. The n_H value for BIII 277 CL was significantly different from 1 (P < .05), whereas the n_H value for BIII 281 CL was not.

Use dependence of block. The progressive inhibition observed in experiments like that of figure 2 suggested that the benzomorphans exerted their block in a use-dependent fashion. To provide further support for a use-dependent blocking model, the experiment of figure 4 was conducted with BIII 277 CL applied for a prolonged period in the absence of NMDA and the degree of block subsequently assessed with a pulse of NMDA in the absence of the drug. As illustrated in figure 4, the current amplitude obtained after a 100-sec application of BIII 277 CL was nearly identical to the control current amplitude elicited before the drug application, indicating that BIII 277 CL does not bind and block closed channels. Note, however, that in the same experiment, coapplication of NMDA and BIII 277 CL results in the progressive development of block. The current elicited by a subsequent application of NMDA in the absence of drug is only minimally larger in amplitude than the final current level achieved in the presence of drug, demonstrating a persistence of the block. A similar experiment with BIII 281 CL gave comparable results (data not shown).

View larger version (17K): [in this window] [in a new window]   Fig. 4.   BIII 277 CL block requires open channels. Perfusion with 0.3 µM BIII 277 CL for 100 sec in the absence of NMDA failed to inhibit the current on a subsequent application of 300 µM NMDA (+10 µM glycine). However, coapplication of BIII 277 CL and NMDA resulted in a progressive inhibition of the current. A final application of NMDA alone demonstrates the persistence of the block.

Occlusion of BIII 277 CL and BIII 281 CL block by Mg⁺⁺. The use dependence of the block produced by BIII 277 CL and BIII 281 CL is compatible with but does not prove that block requires open channels. Open-channel blockers typically occlude current flow through the channel by binding within the channel pore. To provide further support for a channel pore blocking model, we investigated the effect of Mg⁺⁺ on the blocking action of BIII 277 CL and BIII 281 CL. Mg⁺⁺ is known to enter the pore of the NMDA receptor channel and bind to a site within the permeation pathway (Ascher and Nowak, 1988). As shown in figure 5, 3 mM Mg⁺⁺, when applied with NMDA, caused an immediate block of the current and partially prevented BIII 281 CL from producing its typical long lasting block. Thus, after coapplication of 3 µM BIII 281 CL with NMDA and Mg⁺⁺, a subsequent application of NMDA alone gave a response amplitude that was reduced by only 27%, not the expected 97%. The modest block produced by BIII 281 CL in the presence of Mg⁺⁺ presumably occurs as a result of displacement of Mg⁺⁺ by the drug. In the experiment of figure 5, subsequent coapplications of NMDA and BIII 281 CL induced a dramatic and persistent decrease in the NMDA-evoked current. An identical experiment performed with BIII 277 CL produced similar results (data not shown). The occlusion of the blocking action of the benzomorphans by Mg⁺⁺ confirms a pore blocking mechanism.

View larger version (18K): [in this window] [in a new window]   Fig. 5.   Mg++ occludes BIII 281 CL block. Currents evoked by 30 µM NMDA (+10 µM glycine) were blocked nearly completely by 3 mM Mg++. Simultaneous application of 0.3 µM BIII 281 CL together with NMDA and Mg++ produced no further block. A subsequent application of NMDA immediately on removal of Mg++ and BIII 281 CL evokes a current that is much larger than expected from the degree of inhibition that occurs when BIII 281 CL is subsequently applied with NMDA alone. In the absence of Mg++, BIII 281 CL causes a dramatic, persistent block.

Effects of BIII 277 CL and BIII 281 CL on KA-, AMPA- and GABA-evoked currents. To assess the specificity of the blocking action of BIII 277 CL and BIII 281 CL, we examined whether the drugs inhibit currents evoked by 100 µM KA, 100 µM AMPA and 1 µM GABA, respectively, at concentrations that are >500-fold higher than their IC₅₀ values for block of NMDA-evoked currents. Currents activated by KA and GABA were nondesensitizing, whereas currents activated by AMPA exhibited rapid desensitization (typically in <100 msec) to a steady-state level. Drug block was assessed on steady-state AMPA-evoked currents. As illustrated in figure 6A (which is representative of three separate experiments), the currents evoked by these three agonists were not affected by 10 µM BIII 277 CL. However, 30 and 300 µM BIII 281 CL produced a modest inhibitory effect (9.3 ± 1.1%; n = 3 and 25.4 ± 1.7%; n = 3, respectively) on KA-evoked current but had no effect on AMPA- and GABA-evoked currents, even at these very high concentrations (fig. 6B).

View larger version (22K): [in this window] [in a new window]   Fig. 6.   Effects of BIII 277 CL, and BIII 281 CL on currents evoked by KA, AMPA and GABA. A, 10 µM BIII 277 CL fails to affect 100 µM KA, 100 µM AMPA or 1 µM GABA currents. B, 30 and 300 µM BIII 281 CL produce minimal inhibition of KA-evoked currents and negligible effect on AMPA and GABA currents.

Voltage dependence of block. As illustrated in figure 7A, the level of block achieved with 0.3 µM BIII 277 CL was substantially greater at -60 mV than at +60 mV, indicating that the block is voltage dependent. The voltage dependence was further analyzed according to the approach of Woodhull (1973) in which the voltage-dependent binding affinity is expressed according to K_D(V) = K_D(0) exp(zFV/RT) where K_D(0) is the dissociation constant at 0 mV transmembrane potential, z is the charge of the blocking particle,  is the fraction of the total electric field sensed by the blocking particle at its binding site and F, R and T are the Faraday constant, the universal gas constant and the absolute temperature, respectively. Assuming that binding of a single BIII 281 CL molecule occludes current flow through the channel, the fractional block can be described by the logistic equation I_D/I_o = [1 + (c/K_D)]^-1, where c is the concentration of the drug, and I_D and I_o are as described in Methods. Incorporation of the Woodhull relationship in this logistic equation allows the linearized form to be derived ln (I_o/I_D - 1) = ln [c/K_D(0)] - zFV/RT in which K_D(0) and z can be determined from a plot of ln (I_o/I_D - 1) against V. The voltage dependence of block is expressed in this fashion in figure 7B. A linear least-squares fit to the data for BIII 281 CL provides an estimate for K_D(0) of 130 nM and of 0.46 for z. BIII 277 CL was not subjected to the same analysis because the flat slope of its concentration-block relationship (fig. 3) suggested that it may bind in two modes with distinct affinities.

View larger version (15K): [in this window] [in a new window]   Fig. 7.   Voltage dependence of BIII 277 CL and BIII 281 CL block. A, Comparison of the use-dependent inhibition of NMDA-evoked currents by 0.3 µM BIII 277 CL at holding potentials of +60 and 60 mV (different cells). B, Woodhull analysis of the voltage dependence of block by BIII 277 CL () and BIII 281 CL () obtained in experiments similar to those of A. Data points represent mean ± S.E.M. of ln(Io/ID - 1) values from three or four cells.

BIII 277 CL and BIII 281 CL block of single NMDA receptor currents. The blocking effects of BIII 277 CL and BIII 281 CL were further examined in recordings of single NMDA receptor channel currents in outside-out patches. Low concentrations of NMDA (2 µM) were used so channel openings were sufficiently infrequent that unitary channel activity could be discriminated. At a holding potential of -60 mV, perfusion of patches with 2 µM NMDA (+10 µM glycine) evoked inward single-channel currents with a mean conductance of ~54 pS.

View larger version (26K): [in this window] [in a new window]   Fig. 8.   BIII 277 CL block of NMDA-activated single-channel currents in an outside-out membrane patch. A, NPo values for consecutive 500-msec epochs in a patch-perfused externally with 2 µM NMDA (+10 µM glycine) (open bars) and with NMDA plus 30 µM BIII 277 CL (filled bar). Four simultaneous openings occasionally occurred in this recording (not shown). B, Sample current records obtained at the points marked in A. Holding potential, -60 mV. Channel opening is downward.

View larger version (29K): [in this window] [in a new window]   Fig. 9.   Flickery block of NMDA-activated single-channel currents by high concentrations of BIII 281 CL in outside-out membrane patches. Traces show representative currents evoked by control 2 µM NMDA (+10 µM glycine) alone (control) and by 30 and 100 µM BIII 281 CL in the presence of NMDA. Sample burst openings (shown on an expanded time scale in boxes to the right) illustrate the flickery block (particularly evident with 100 µM BIII 281 CL). All traces are from the same patch. Holding potential, -60 mV. Channel opening is downward.

View larger version (22K): [in this window] [in a new window]   Fig. 10.   Effects of BIII 281 CL and BIII 277 CL on open time distributions of NMDA-activated single-channel currents in outside-out membrane patches. Open time distribution histograms were constructed using data pooled from four control patches and three patches exposed to 30 µM BIII 281 CL (B) and three patches exposed to 30 µM BIII 277 CL (C). The histograms were fit to single exponential functions with open values as shown. D, Concentration dependence of 1/open values. Each data point was obtained from fits to histograms similar to those in A-C constructed from data pooled from three or four patch recordings. BIII 277 CL (); BIII 281 CL (). The lines indicate the best fits to the data; the parameters of the fits are given in the text.

	Discussion

Top Abstract Introduction Methods Results Discussion Appendix References

The present study demonstrates that the enantiomeric 6,7-benzomorphans BIII 277 CL and BIII 281 CL inhibit whole-cell and single-channel NMDA receptor currents. The blocking action of the benzomorphans occurred in a slow, use- and voltage-dependent fashion. Once achieved, the block was persistent; only partial recovery was obtained during prolonged wash periods with intermittent agonist applications (up to 40 min). BIII 277 CL had high steady-state blocking potency (IC₅₀ = 5.2 nM at -60 mV) and BIII 281 CL was only modestly less potent (IC₅₀ = 64 nM). Both enantiomers were selective for NMDA receptor currents and, even at high concentrations, did not affect currents evoked by AMPA, or GABA, although there was minimal block of KA currents by BIII 281 CL at 30 and 300 µM.

BIII 277 CL and BIII 281 CL have previously been shown to inhibit binding of [³H]dizocilpine to NMDA receptors in rat cortical synaptosomal membranes under conditions in which the receptors are fully activated by endogenous glutamate and glycine (Carter et al., 1995; Grauert et al., 1997). Moreover, BIII 277 CL inhibits NMDA-induced neurotransmitter release from brain slices and protects against NMDA-induced lethality in mice, suggesting that it is a functional antagonist of NMDA receptors. The present electrophysiological experiments confirm that BIII 277 CL and BIII 281 CL can functionally inhibit NMDA receptors and, for the first time, conclusively demonstrate that they act by a channel-blocking mechanism.

Several lines of evidence provide support for such a channel-blocking mechanism, including the observations that BIII 277 CL and BIII 281 CL only bind to their blocking sites when the channel is in the agonist-gated open state and that Mg⁺⁺ occludes binding of the blocking drugs. The voltage-dependent relief of block accompanying depolarization is also compatible with a channel-blocking mechanism in which the drug binding site is within the channel pore at a site that senses a fraction of the membrane electric field. At physiological pH, the nitrogen of the benzomorphans is protonated (pK_a = 9.6), so the molecules have a charge of +1. Increasing the positivity of the transmembrane potential would then tend to neutralize the interaction between this charge and an electronegative site on the channel protein, thus reducing binding affinity. Analysis of the steady-state fractional block values for BIII 281 CL according to the method of Woodhull provided an estimate of the electrical depth of the binding site of 0.46, which is similar to that of other monovalent cationic channel-blocking ligands (Subramaniam et al., 1994), although there can be considerable variability in this value (see Frankiewicz et al., 1996). The binding affinity derived in this analysis [K_D(0) = 130 nM] corrected to -60 mV is 43 nM, which compares favorably with the IC₅₀ value determined in the steady-state block experiment (64 nM). However, the affinity obtained in the present electrophysiological analyses is substantially higher than the affinity previously obtained with radioligand binding (Grauert et al., 1997). The reason for this discrepancy is not apparent.

In view of the similar use dependence and electrical depth, the channel-blocking mechanism for the benzomorphans is likely to be comparable to that of various structurally diverse channel-blocking NMDA receptor antagonists (see Subramaniam et al., 1994). However, the benzomorphans, particularly BIII 277 CL, have markedly higher affinity for the channel-blocking site than many other NMDA receptor channel blockers (except for dizocilpine; Huettner and Bean, 1988; Halliwell et al., 1989). We developed a model to explain the high affinity of this binding interaction. The model also accounts for the unexpectedly high binding affinity of BIII 281 CL in comparison with SKF 10,047.

The model posits that BIII 277 CL can bind to the NMDA receptor at three interaction points and that binding occurs in several modes characterized by docking at either one, two or all three of the interaction points. Binding with alignment of interaction point 1 (mode 1) occurs rapidly but is of low affinity unless stabilized by binding at additional interactions points. Binding with simultaneous alignment of interaction points 1 and 2 (mode 2) is of moderate affinity, comparable to that of older benzomorphans such as SKF 10,047. Finally, when there is alignment of all three interaction sites (mode 3), binding affinity is very high. The model also requires that transitions between binding modes are energetically unfavorable (i.e., once a drug molecule has approached the binding domain in the orientation required for docking in one mode, it cannot readily assume another binding mode without leaving the vicinity of the binding domain and reentering).

Support for this model comes from the concentration-block isotherms of figure 3 in conjunction with the results of the single-channel recordings. The isotherm for BIII 281 CL is readily fit with n_H value near 1, indicating a one-to-one binding interaction between the drug and its channel acceptor. However, the data for BIII 277 CL are not well fit by such a simple interaction model, and indeed statistical analysis shows that n_H is significantly different from 1. As described in the Appendix, a relationship can be developed based on the three-mode model that does provide an adequate fit to the fractional block data (fig. 11A). Using this relationship, we obtain binding affinities in mode 2 and mode 3 of 58 and 2 nM, respectively. Assuming that BIII 277 CL binds at the same electrical depth as BIII 281 CL, these values corrected to 0 mV are 173 and 6 nM. This latter value compares favorably with the equilibrium [³H]dizoclilpine binding affinity of BIII 277 CL (4.5 nM; Carter et al., 1995). Extensive structure-activity studies with benzomorphans related to BIII 277 CL have indicated that the protonated nitrogen, the methoxypropyl side chain and the tetralin ring system are important structural features for high affinity binding (Grauert et al., 1997). It is tempting to propose that BIII 277 CL docks with the channel acceptor via these three moieties, and moreover that the nitrogen and the methoxypropyl side chain represent the groups relevant to mode 1 binding and mode 2 binding. In figure 11A, the fractional block data for BIII 281 CL are well fit by a logistic equation with K_D value equal to the derived mode 2 binding affinity for BIII 277 CL, consistent with the idea that BIII 281 CL can bind in modes 1 and 2 but not 3. As illustrated in figure 11B, superimposition of the enantiomers to bring the nitrogen and methoxy functionalities into congruence results in the aromatic rings (and the conformationally restrained coplanar tetralin moieties) being out of alignment (nearly orthogonal). Therefore, assuming that the nitrogen and the methoxypropyl side chain represent the groups relevant to mode 1 and mode 2 binding, it is appealing to speculate that the aromatic ring (or tetralin moiety) docks at the third interaction site required for mode 3 binding.

View larger version (13K): [in this window] [in a new window]   Fig. 11.   A, Fitting of the theoretical binding isotherm for the three-interaction-mode model given in Appendix to the fractional block (FB) data for BIII 277 CL of figure 3. K3 and K2 are taken to be 2.0 and 58 nM, respectively. Also shown are Hill equation fits (nH = 1) in which KD is set to K3 (dashed curve) and K2 (dotted curve). Note that the curve with KD = K2 adequately fits the fractional block data for BIII 281 CL. B, Superimposition of the three-dimensional structures of BIII 277 CL (thick lines) and BIII 281 CL (thin lines). When the nitrogen and the methoxypropyl side chain are aligned, the aromatic rings are nearly orthogonal.

In the three-mode model, a small number of channels N₁(c) will have drug bound in mode 1. For simplicity, this population of channels was ignored in the analysis of the Appendix. We proposed that mode 1 binding is represented by the flickery block observed in the single-channel recordings at high drug concentrations. The binding constant for this low-affinity interaction can be estimated from the ratio K₁ = k_-1/k₁, where k₁ is taken from the fits to the data presented in figure 11D, and k_-1 is estimated as 1/_closed (see Results). The estimates of K₁ for BIII 277 CL and BIII 281 CL are 301 and 206 µM, respectively. These concentrations are within the range of concentrations at which flickery block was observed but >2 orders of magnitude greater than the concentrations at which block was nearly saturated in the binding isotherms of figure 3. With these estimates of K₁, the maximum percent block attributable to mode 1 binding at equilibrium is <0.3%. Consequently, it is appropriate to ignore the contribution of block by this low affinity interaction in the determination of K₂ and K₃ (see Appendix). (Under nonequilibrium conditions such as in the sample records of figures 8 and 9 obtained before the full establishment of block in modes 2 and 3, flicker is more apparent.)

Based on the three-mode model, we assume that the major determinant of the time course for the development of block by BIII 281 CL in figure 2B is binding in mode 2 and for BIII 277 CL is binding in mode 3. Because the rate of unbinding at -60 mV is negligibly slow, the apparent forward rates can be derived directly from the fits in figure 2B; these values are 0.7 and 2.8 × 10⁵ M^-1 sec^-1 for BIII 277 CL and BIII 281 CL, respectively. Huettner and Bean (1988) noted the substantially faster forward rate for flickery block in single-channel recordings compared with the much slower rate for block in whole-cell experiments and concluded that P_o under the whole-cell conditions is very low (~0.002). However, if the benzomorphans bind in distinct modes at different rates, this method for the estimation of P_o cannot be applied. Indeed, based on the slow development of single-channel block observed in experiments of the type illustrated in figure 8A, the rate constant for entry into mode 3 block for BIII 277 BI can be estimated as ~10⁴ to 10⁵ M^-1 sec^-1, which is comparable to the forward rate determined in the whole-cell recordings. Thus, P_o is likely to be substantially greater than the estimate of Huettner and Bean (1988) and may approach 1. This seems reasonable given the fact that P_o in the single-channel experiments with a substantially lower NMDA concentration (2 vs. 30 µM) was on average ~0.1 (table 1).

In conclusion, the present study demonstrated that BIII 277 CL and BIII 281 CL are potent and selective channel-blocking NMDA receptor antagonists. The substantially higher affinity of BIII 277 CL in comparison with the distomer and other benzomorphans is compatible with a model in which binding is stabilized at three sites within the channel. The surprising observation that the binding affinity of the distomer is not markedly reduced compared with other benzomorphans is further explained by the model if the binding of the distomer is stabilized at two sites in a similar fashion as conventional benzomorphans.