Mechanism of Inhibition of the GluA2 AMPA Receptor Channel Opening by 2,3-Benzodiazepine Derivatives: Functional Consequences of Replacing 7,8-Methylenedioxy with 7,8-Ethylenedioxy Moiety

Abstract

2,3-Benzodiazepine (2,3-BDZ) compounds are a group of AMPA receptor inhibitors and are drug candidates for treating neurological diseases involving excessive AMPA receptor activity. We investigated the mechanism by which GluA2Q_flip receptor channel opening is inhibited by two 2,3-BDZ derivatives, i.e.,1-(4-aminophenyl)-3,5-dihydro-7,8-ethylenedioxy-4H-2,3-benzodiazepin-4-one (2,3-BDZ-11-2) and its 1-(4-amino-3-chlorophenyl) analogue (2,3-BDZ-11-4). Both compounds have a 7,8-ethylenedioxy moiety instead of the 7,8-methylenedioxy feature present in the structure of GYKI 52466, the prototypic 2,3-BDZ compound. Using a laser-pulse photolysis approach with a time resolution of ~60 μs and a rapid solution flow technique, we characterized the effect of the two compounds on the channel-opening process of the homomeric GluA2Q_flip receptor. We found that both 2,3-BDZ-11-2 and 2,3-BDZ-11-4 are noncompetitive inhibitors with specificity for the closed-channel conformation of the GluA2Q_flip receptor. However, 2,3-BDZ-11-4 is ~10-fold stronger, defined by its inhibition constant for the closed-channel conformation (i.e., K_I = 2 μM), than 2,3-BDZ-11-2. From double-inhibitor experiments, we determined that both compounds bind to the same site, but this site is different from two other known, noncompetitive binding sites on the GluA2Q_flip receptor previously reported. Our results provide both mechanistic clues to better understand AMPA receptor regulation and a structure-activity relationship for designing more potent 2,3-BDZ compounds with predictable properties for this new noncompetitive site.

One of the most salient reasons for characterizing the mechanism of action for a series of structurally related regulatory molecules is to find “atomic descriptors” and to use them to design new compounds with predictable properties. In a structural sense, an atomic descriptor can be correlated to a unique position on an existing structure; for example, a single-atom substitution or addition at a particular position leads to a dramatic change in its function, such as a change of binding site on the same molecular target. Consequently, one can develop new analogues to find new binding sites and to achieve more quantitative control of the target activity. As the fourth in a series of mechanistic studies designed to establish a more quantitative structure-activity relationship for 2,3-BDZ compounds, we herein describe the functional consequence of replacing 7,8-methylenedioxy with 7,8-ethylenedioxy moiety on the 2,3-benzodiazepine (2,3-BDZ) structure, represented most prominently by GYKI 52466 [i.e., 1-(4-aminophenyl)-4-methyl-7,8-methylenedioxy-5H-2,3-benzodiazepine] (1). GYKI 52466 is the prototypic 2,3-BDZ compound, based on which hundreds of derivatives have been synthesized (2).

2,3-BDZ compounds are supposedly antagonists of the α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) subtype of glutamate ion channel receptors (2-4). AMPA receptors mediate the majority of fast excitatory synaptic transmission in the central nervous system and are critically involved in neuronal development and brain activities, such as learning and memory (5, 6). Excessive activation of AMPA receptors is implicated in some neurological diseases, such as ischemia, epilepsy, and amyotrophic lateral sclerosis (7). Therefore, antagonists of AMPA receptors are drug candidates to treat such neurological diseases (2, 8). In fact, 2,3-BDZ compounds exhibit desirable anticonvulsant and neuroprotective properties in cellular and animal models (9). To date, however, the mechanism of action of these compounds on AMPA receptors is not well understood, and a quantitative structure-function relationship has not yet been established. This deficiency is mainly attributed to the fact that an AMPA receptor opens its channel in the microsecond (μs) time scale but desensitizes in the millisecond (ms) time domain (10). Previous studies of 2,3-BDZ compounds have not been conducted in the time scale in which the receptors are in the functional state. Consequently, it has not been possible to design 2,3-BDZ derivatives with predictable properties relevant to the time scale of the receptor function.

To characterize the functional consequence of replacing 7,8-methylenedioxy with 7,8-ethylenedioxy moiety on the 2,3-benzodiazepine structure, we focus on two 2,3-BDZ compounds, i.e., 1-(4-aminophenyl)-3,5-dihydro-7,8-ethylenedioxy-4H-2,3-benzodiazepin-4-one (2,3-BDZ-11-2) and its analogue 1-(4-amino-3-chlorophenyl)-3,5-dihydro-7,8-ethylenedioxy-4H-2,3-benzodiazepin-4-one (2,3-BDZ-11-4) (Zappalà et al., 2006; Micale et al., 2008) (Figure 1). 2,3-BDZ-11-2 and 2,3-BDZ-11-4 are more similar to 2,3-BDZ-2 (i.e., 1-(4-aminophenyl)-3,5-dihydro-7,8-methylenedioxy-4H-2,3-benzodiazepin-4-one) (11) because they all have a carbonyl group at C-4 of the diazepine ring, whereas GYKI 52466 has a C-4 methyl group (Figure 1). However, in both 2,3-BDZ-11-2 and 2,3-BDZ-11-4, the 7,8-methylenedioxy feature, as in 2,3-BDZ-2, is replaced with a 7,8-ethylenedioxy moiety. Compared with 2,3-BDZ-11-2, 2,3-BDZ-11-4 contains an additional chlorine atom at C-3 position of the aminophenyl ring (Figure 1).

Figure 1.

Chemical structures of GYKI 52466, 2,3-BDZ-2, 2,3-BDZ-11-2, and 2,3-BDZ-11-4. Their chemical names are given in the Abbreviation list and in the text.

Given these structural differences, we asked the following questions: What is the mechanism by which the GluA2Q_flip channel opening is inhibited by 2,3-BDZ-11-2 and 2,3-BDZ-11-4? Does the ring enlargement in 2,3-BDZ-11-2 at the 7,8 position change its potency and binding site with respect to 2,3-BDZ-2? What is the functional consequence of adding a chlorine atom at the C-3 position of the aminophenyl ring? Answers to these questions will provide us a quantitative understanding of the functional consequences of these structural changes. Experimentally, we used a laser-pulse photolysis technique, together with a photolabile precursor of glutamate or caged glutamate, which provides a time resolution of ~60 μs (10). This technique is suitable for measuring the rate of AMPA receptor channel opening and therefore enables us to elucidate the mechanism of inhibition for these compounds without the complication of channel desensitization that occurs in the ms time scale (11–13). Furthermore, we chose to study these compounds with a homomeric GluA2Q_flip channel, because the unedited isoform of GluA2 or GluA2Q is abnormally expressed in some neurological disorders (14). Among the four AMPA receptor subunits, GluA2 is the one that controls the Ca²⁺ permeability of native AMPA receptor assemblies (15), and an intracellular Ca²⁺ overload leads to neuronal death (16).

EXPERIMENTAL PROCEDURES

Receptor Expression and Cell Culture

Human embryonic kidney (HEK)-293S cells were cultured in Dulbecco’s modified Eagle’s medium supplemented with 10% fetal bovine serum at 37°C in a 5% CO₂-humidified incubator (17). HEK-293S cells were co-transfected to express GluA2Q_flip, together with the weight ratio of 1:0.2:10 for green fluorescent protein to large T-antigen to plasmid DNA (18). The GluA2Q_flip plasmid used for transfection was ~5–10 μg/35-mm Petri dish (11). The cells were used for recording 48 hours after transfection.

Whole-Cell Current Recording

Glutamate was used as the agonist in this study. Briefly, the resulting whole-cell current response was recorded using an Axopatch-200B amplifier at cutoff frequency of 2–20 kHz by a 4-pole low-pass Bessel filter and acquired at a 5–50 kHz sampling frequency using a Digidata 1322A digitizer (Molecular Devices Corp.) (17). All recordings were collected at −60 mV, pH 7.4, and 25 °C. The electrode used for whole-cell recording was made from glass capillary and then fire polished (17). The electrode was filled with an internal solution containing 110 mM CsF, 30 mM CsCl, 4 mM NaCl, 0.5 mM CaCl₂, 5 mM EGTA, and 10 mM HEPES (pH 7.4 adjusted by CsOH). The extracellular bath buffer contained 150 mM NaCl, 3 mM KCl, 1 mM CaCl₂, 1 mM MgCl₂, and 10 mM HEPES (pH 7.4 adjusted by NaOH). pCLAMP 8 was used for data collection.

Laser-Pulse Photolysis Measurement

A laser-pulse photolysis technique was used to liberate free glutamate surrounding a cell on a μs time scale from photolysis of 4-methoxy-7-nitroindolinyl-caged-L-glutamate (13, 19). The caged glutamate was applied to an HEK-293S cell suspended in the extracellular solution by a flow device (20). The cell was equilibrated with the caged glutamate for at least 250 ms before it was irradiated by a 355-nm, 8-nanosecond laser pulse. A single laser pulse of 200–1000 μJ was generated by a pulsed Q-switched Nd:YAG laser and then coupled into a fiber optic (13). To calibrate the concentration of the photolytically released glutamate, we applied two free glutamate solutions with known concentrations to the same cell before and after a laser flash. The current amplitudes obtained from this calibration were compared with the amplitude from laser photolysis, with reference to the dose-response relation.

Measurement of the Channel-Opening Rate Constant (k_op) and Channel-Closing Rate Constants (k_cl)

Using the laser-pulse photolysis technique combined with whole-cell recording, we determined the observed channel-opening rate constant, k_obs, as a function of glutamate concentration from which k_op and k_cl were calculated. In the laser-pulse photolysis measurement, the channel-opening kinetic process followed a single exponential rate expression, in eq 1, for ~95% of the rise time.

$$A_t=A_{max}(1-e^{-k_{\mathit{obs}}t})$$ (1)

where A_t represents the current amplitude at time t, and A_max represents the maximum current amplitude. From eq 1, k_obs was calculated. Various k_obs values as a function of glutamate concentration can be described using a general mechanism of channel opening (17).

$$\text{R}+\text{nL}\stackrel{K_1}{\rightleftharpoons }{\text{RL}}_{\text{n}}\underset{k_{\text{cl}}}{\overset{k_{\text{op}}}{\rightleftharpoons }}{\overline{\text{RL}}}_{\text{n}}$$

In the scheme described above, R stands for the active, unliganded form of the receptor, L the ligand or glutamate, RL_n the closed-channel forms with n ligand molecules bound, and $\overline{{RL}_n}$ the open-channel state. The number of glutamate molecules to bind to the receptor and to open its channel, n, can be from 1 to 4, assuming that a receptor is a tetrameric complex and each subunit has one glutamate binding site. However, our kinetic results of the channel-opening rate process for AMPA receptors can be fully explained by the assumption that binding of two glutamate molecules per receptor complex is sufficient to open an AMPA receptor channel (21). For simplicity and without contrary evidence, it is further assumed that glutamate binds with equal affinity or K₁, the intrinsic equilibrium dissociation constant, at all binding steps. As such, k_obs can be expressed by eq 2 (17).

$$k_{\mathit{obs}}=k_{cl}+k_{op}{(\frac{L}{L+K_1})}^2$$ (2)

In deriving eq 2, the ligand-binding rate was assumed to be fast relative to the channel-opening rate. This assumption was supported by the observation in which the channel-opening rate process, represented by k_obs, was first-order in the absence (17) and presence of any inhibitors (11–13), including those in this study.

Effect of an Inhibitor on k_op and k_cl

To elucidate the mechanism of inhibition by 2,3-BDZ-11-2 and 2,3-BDZ-11-4, we characterized the effect of each inhibitor on both k_op and k_cl (11–13). If 2,3-BDZ-11-4, for instance, is an uncompetitive inhibitor or an open-channel blocker, it will inhibit only k_cl but not k_op (11–13). If 2,3-BDZ-11-4 is a competitive inhibitor, it only inhibits k_op but not k_cl. Conversely, if 2,3-BDZ-11-4 is a noncompetitive inhibitor, it will inhibit both k_cl and k_op (see eq 3 – this equation was used to analyze the effect of both 2,3-BDZ-11-2 and 2,3-BDZ-11-4 in “Results”, as both turned out to be noncompetitive inhibitors).

To measure the effect of each inhibitor on k_cl and k_op, we varied the ligand (i.e., glutamate) concentration. This is because k_obs is a function of ligand concentration, and the magnitude of k_obs is contributed by both k_cl and k_op terms (eq 2) (10–13). Therefore, if the ligand concentration is low (i.e., L ≪ K₁), eq 2 is reduced to k_obs ≈ k_cl. Similarly, in the presence of an inhibitor, eq 3 is reduced to k_obs ≈ k_cl^′. In other words, the effect of an inhibitor on k_cl and its inhibition constant ( $\overline{K_I}$) for the open-channel state are determined with the use of eq 4 (here eq 4 is transformed into the linear form to obtain $\overline{K_I}$ by linear regression). At a higher ligand concentration, where k_obs > k_cl, the k_op value can be determined by the difference between k_obs and k_cl or by rearranging (eq 2) such that k_obs − k_cl = k_op [L/(L + K₁)]². Similarly, the effect of an inhibitor on k_op and the inhibition constant (K_I) for the closed-channel state (here the closed-channel state refers to the unliganded, singly and doubly liganded forms, assuming n = 2) can be determined with the use of eq 5.

$$k_{\mathit{obs}}=k_{cl}(\frac{\overline{K_I}}{\overline{K_I}+I})+k_{op}{(\frac{L}{L+K_1})}^2(\frac{K_I}{K_I+I})$$ (3)

$$\frac{1}{k_{\mathit{obs}}}=\frac{1}{k_{cl}}+\frac{1}{k_{cl}}\frac{I}{\overline{K_I}}$$ (4)

$${(k_{\mathit{obs}}-k_{cl}^{\prime })}^{-1}={[\frac{k_{op}{L}^2}{{(L+K_1)}^2}]}^{-1}(1+\frac{I}{K_I})$$ (5)

We previously established the criteria by which k_cl can be determined from the measurement of k_obs (10, 11, 17). For GluA2Q_flip, k_cl is numerically equal to the k_obs value obtained at 100-μM glutamate concentration, which corresponds to ~4% of the fraction of the open-channel form (10, 11, 17). As such, the effect of 2,3-BDZ-11-2 or 2,3-BDZ-11-4 on k_cl was determined at this glutamate concentration (11). The effect on k_op was determined at a glutamate concentration of 300 μM (11). At this concentration the difference between k_op and k_cl could be already detected while the energy used for laser photolysis was still well tolerated by the cell.

Effect of an Inhibitor on Current Amplitude

The effect of 2,3-BDZ-11-2 or 2,3-BDZ-11-4 on the whole-cell current amplitude (A) was measured to calculate independently an inhibition constant. Specifically, we used a low glutamate concentration (i.e., L ≪ K₁), at which most receptors were in the closed-channel state, to determine the inhibition constant for the closed-channel state (eqs 6a and 6b). To determine the inhibition constant for the open-channel state, we used a saturating ligand concentration (L ≫ K₁), at which most receptors were in the open-channel state (11–13). For GluA2Q_flip, we chose 100 μM and 3 mM for the low and the high glutamate concentrations, respectively. These concentrations correspond to ~4% and ~95% of the open-channel form, respectively (11–13).

$$\frac{A}{A_I}=1+I\frac{{(\overline{{RL}_2})}_o}{K_I}$$ (6a)

where in eq 6a, ${(\overline{{RL}_2})}_o$ represents the fraction of the open-channel state and is proportional to the current amplitude. In eq 6b, this fraction is expressed as a function of the fraction of all receptor forms.

$${({\overline{RL}}_2)}_o=\frac{{\overline{RL}}_2}{R+RL+{RL}_2+\overline{{RL}_2}}=\frac{L^2}{L^2(1+\mathrm{\Phi })+2K_{1}L\mathrm{\Phi }+K_1^2\mathrm{\Phi }}$$ (6b)

Experimentally, a flow device was used to apply glutamate of known concentration in the absence and presence of an inhibitor (11). The time resolution of the flow device, determined by the rise time of the whole-cell current response (10–90%) to saturating glutamate concentrations, was 1.0 ± 0.2 ms (11–13). For data analysis, the observed amplitude of the whole-cell current was corrected for receptor desensitization (20). Furthermore, full inhibition by 2,3-BDZ-11-2 or 2,3-BDZ-11-4 was achieved only by preincubating the GluA2Q_flip receptor with either inhibitor for at least 6 seconds, similar to other 2,3-BDZ compounds we studied previously (11–13).

Double-Inhibitor Experiment to Assess Binding Site

To investigate whether two inhibitors bound to the same site or to two different sites on GluA2Q_flip, we examined the effect of two inhibitors on the whole-cell current amplitude (eqs 7 and 8) and compared that with the effect of just one inhibitor (eq 6a) (11–13). Specifically, the amplitude was used, similar to eq 6a, to plot A/A_I,P vs. one inhibitor concentration. Here, one inhibitor is represented as I in molar concentration and the other is P. Based on the assumptions that binding of one inhibitor excludes the binding of the other (i.e., in one-site model, or A·I and A·P complexes are allowed but not A·I·P complex form), the ratio of the current amplitude is given in eq 7.

$$\frac{A}{A_{I,P}}=(1+\frac{P}{K_P})+\frac{I}{K_I}$$ (7)

Conversely, for a two-site model in which there are two sites for I and P separately and binding of one inhibitor is independent of the binding of the other (i.e., A·I, A·P and A·I·P complex formed are all allowed), the ratio of the current amplitude is given in eq 8.

$$\frac{A}{A_{I,P}}=(1+\frac{P}{K_P})+(1+\frac{P}{K_P})\frac{I}{K_I}$$ (8)

In the double-inhibitor experiment, the concentration of one inhibitor was kept constant while the concentration of the other was varied. An apparent inhibition constant obtained from the two-inhibitor experiment (or the slope of the A/A_I,P plot; see eqs 7 and 8) was compared to that obtained from the one-inhibitor experiment (or the slope of the A/A_I plot; see eqs 6a and 6b). All other conditions were the same as those described for measuring the effect of an inhibitor on the current amplitude.

The Origin 7 software was used for both linear and nonlinear regression analysis in this study. Unless otherwise noted, each data point shown in a plot was an average of at least three measurements collected from at least three cells. The error reported refers to the standard deviation of a fit.

RESULTS AND DISCUSSION

2,3-BDZ-11-2 and 2,3-BDZ-11-4 Inhibited the Channel-Opening Rate of GluA2Q_flip

2,3-BDZ-11-2 and 2,3-BDZ-11-4 inhibited the GluA2Q_flip channel activity, shown in a reduction of whole-cell current (Figure 2A). Because neither 2,3-BDZ-11-4 (Figure 2B) nor 2,3-BDZ-11-2 (Figure S1 in Supporting Information) affected the rate of channel desensitization at various concentrations of glutamate and inhibitors, we focused our investigation on the effect of these two compounds on the rate of channel opening of GluA2Q_flip, using the laser-pulse photolysis technique. Figure 3A shows a representative whole-cell current trace generated by photolysis of the caged glutamate. In the presence of 2,3-BDZ-11-4, for example, the rise of the whole-cell current was slowed and the amplitude reduced, indicating that 2,3-BDZ-11-4 inhibited the opening of the GluA2Q_flip channel. In the absence or presence of an inhibitor, the channel-opening rate could be described by a first-order rate process for over 95% of the rising phase (Figure 3A). In fact, a single exponential rate for the current rise was observed, without exception, for all inhibitor and glutamate concentrations in the present and previous studies of GluA2Q_flip (11, 17). These results therefore support the notion that the rate of the current rise in the laser-pulse photolysis measurement was representative of the channel-opening rate rather than the ligand-binding rate process (11, 17). The reduction of the rate of current rise was thus ascribed to the inhibition of channel opening by an inhibitor.

Figure 2.

(A) A pair of representative whole-cell current traces from GluA2Q_flip channels expressed in HEK-293 cells in the absence (left) and presence (right) of 2,3-BDZ-11-4. The whole-cell current recording was at −60 mV, pH 7.4, and 22°C. The concentration of glutamate and inhibitor was 3 mM and 10 μM, respectively. (B) Effect of 2,3-BDZ-11-4 on the channel desensitization rate constants determined at 100 μM glutamate (●) and at 3 mM glutamate (○) (the same data for 2,3-BDZ-11-2 is shown as Figure S1 in Supporting Information).

Figure 3.

(A) In a laser-pulse photolysis experiment, 2,3-BDZ-11-4 inhibited both the rate and the amplitude of the whole-cell current rise reflecting the opening of the GluA2Q_flip channel. The top trace was the control (k_obs = 2425 s⁻¹; A = 0.52 nA), and the bottom one contained 4 μM 2,3-BDZ-11-4 (k_obs = 2274 s⁻¹; A = 0.24 nA). In both cases, the concentration of the photolytically released glutamate was estimated to be 300 μM. (B) Effect of 2,3-BDZ-11-4 on k_cl obtained at 100 μM glutamate and as a function of 2,3-BDZ-11-4 concentration. From this plot, a ${\overline{K_I}}^{\ast }$ of 24 ± 5.0 μM was obtained using eq 4. (C) Effect of 2,3-BDZ-11-4 on k_op obtained at 300 μM glutamate and as a function of 2,3-BDZ-11-4 concentration. From this plot, a $K_I^{\ast }$ of 9.0 ± 3.0 μM was determined using eq 5. The plots for the effect of 2,3-BDZ-11-2 on the channel-opening rate constants are provided in Supporting Information as Figure S2. (D) Effect of 2,3-BDZ-11-4 on the amplitude of the whole-cell current in the absence and presence of 2,3-BDZ-11-4, determined from laser-pulse photolysis measurements. A K_I of 2.3 ± 0.2 μM was obtained from the plot of A/A_I values versus 2,3-BDZ-11-4 concentration for the closed-channel state (●; 100 μM glutamate). The K_I was determined to be 2.4 ± 1.0 μM at 300 μM of photolytically released glutamate (○). Each data point shown in this plot was an average of at least three measurements collected from at least three cells.

We then characterized exclusively the effect of an inhibitor on k_cl and determined the inhibition constant for the open-channel form ( ${\overline{K_I}}^{\ast }$) at 100-μM glutamate concentration, using eq 4 (10–13). Specifically, a ${\overline{K_I}}^{\ast }$ of 24 ± 5.0 μM for 2,3-BDZ-11-4 was determined for the open-channel state (Figure 3B), and a ${\overline{K_I}}^{\ast }$ of 43 ± 11 μM was determined for 2,3-BDZ-11-2 (Figure S2A in Supporting Information) (Note that for clarity the data presented in the text are mostly for 2,3-BDZ-11-4, while the data for 2,3-BDZ-11-2 are presented in Supporting Information. Furthermore, 2,3-BDZ-11-4 and 2,3-BDZ-11-2 have an identical mechanism of action). At a higher ligand concentration (i.e., 300 μM glutamate; see also Experimental Procedures), the effect of an inhibitor on k_op was further determined. A $K_I^{\ast }$ of 9.0 ± 3.0 μM for the closed-channel state was obtained (eq 5) for 2,3-BDZ-11-4 (Figure 3C). Similarly, a $K_I^{\ast }$ of 39 ± 9.0 μM was obtained for 2,3-BDZ-11-2 (Figure S2B in Supporting Information). These values are also summarized in Table 1.

Table 1.

Summary of the inhibition constants of 2,3-BDZ-11-4, 2,3-BDZ-11-2, 2,3-BDZ-2, and GYKI 52466, obtained from rate and amplitude measurements, for the closed- and open-channel states of GluA2Q_flip

	Rate Measurementa		Amplitude Measurement
Inhibitor	$K_I^{\ast }$ (μM)b (closed channel)	${\overline{K_I}}^{\ast }$ (μM)b (open channel)	KI (μM)b,d	KI (μM)b,e	KI (μM)c,d (closed channel)	$\overline{K_I}$ (μM)c,f (open channel)
2,3-BDZ-11-4	9.0 ± 3.0	24 ± 5.0	2.3 ± 0.2	2.4 ± 1.0	2.0 ± 0.1	14 ± 0.4
2,3-BDZ-11-2	39 ± 9.0	43 ± 11	19 ± 1.0	24 ± 4.0	21 ± 0.1	33 ± 1.0
2,3-BDZ-2g	48 ± 5.0	194 ± 20	25 ± 1.0	23 ± 1.0	25 ± 1.0	7.0 ± 1.0
GYKI 52466h	61 ± 11	128 ± 30	15 ± 1.0	16 ± 1.0	14 ± 1.0	30 ± 2.0

^aThe constants obtained from rate measurements represent those in the first step of inhibition, as in Figure 5, whereas those obtained from the amplitude measurements represent the overall inhibition constants.

^bLaser-pulse photolysis measurement.

^cFlow measurement.

^dMeasurements at 100 μM glutamate for the closed-channel state.

^eMeasurements at ~300 μM glutamate.

^fMeasurements at 3 mM glutamate.

^gRitz (11).

^hRitz (12)

Effect of 2,3-BDZ-11-2 and 2,3-BDZ-11-4 on the Amplitude of Whole-Cell Current Determined by Laser-Pulse Photolysis Measurement

As shown in a laser-pulse photolysis measurement (Figure 3A), 2,3-BDZ-11-4 inhibited the time course of the current rise and concurrently the amplitude of the whole-cell current. From the amplitude of the whole-cell current in the absence and presence of 2,3-BDZ-11-4 (Figure 3D), a K_I of 2.3 ± 0.2 μM was obtained, using eq 6, at 100 μM glutamate for the closed-channel state. Similarly, a $\overline{K_I}$ of 2.4 ± 1.0 μM was estimated at 300 μM glutamate (Figure 3D). These inhibition constants obtained from the amplitude in the laser measurement are also summarized in Table 1.

However, inspection of the inhibition constants obtained from the rate (Figure 3B and 3C) and amplitude data (Figure 3D) from the same experiment (i.e., the laser-pulse photolysis measurement) for the same inhibitor (e.g., 2,3-BDZ-11-4) showed a clear discrepancy. For instance, a $K_I^{\ast }$ of 9.0 ± 3.0 μM for the closed-channel state of GluA2Q_flip was obtained from the rate data for 2,3-BDZ-11-4 (i.e., the data on column 1, Table 1) but a K_I of 2.3 ± 0.2 μM was obtained from the amplitude data (i.e., the data on column 3, Table 1). For 2,3-BDZ-11-2 we observed a smaller but statistically significant difference in the inhibition constant determined from the amplitude data as compared with the rate data (Figure S3 in Supporting Information; Table 1).

Effect of 2,3-BDZ-11-2 and 2,3-BDZ-11-4 on the Amplitude of Whole-Cell Current from Flow Measurement

As described above, the rate and the amplitude data from the laser-pulse photolysis experiment produced different inhibition constants. To investigate this difference, we used a solution flow technique with known concentrations of glutamate and measured the current amplitude. From the plot of A/A_I as a function of inhibitor concentration, we independently characterized the inhibition constant for each inhibitor. For 2,3-BDZ-11-4, a K_I of 2.0 ± 0.1 μM for the closed-channel state and a $\overline{K_I}$ of 14 ± 0.4 μM for the open-channel state (Figure 4) were determined. For 2,3-BDZ-11-2, a K_I of 21 ± 0.1 μM for the closed-channel state and $\overline{K_I}$ of 33 ± 1.0 μM for the open-channel state were calculated (Figure S4 in Supporting Information, and all these values are also summarized in Table 1). By comparison, inhibition constants obtained from the amplitude data from both the flow and the laser experiments were the same. For instance, at the same glutamate concentration (i.e., either 100 or 300 μM), the K_I value for 2,3-BDZ-11-4 estimated from the amplitude data for the closed-channel state was 2.3 μM by the flow measurement and 2.0 μM by the laser experiment (Table 1). Therefore the apparent discrepancy in inhibition constant between the rate and the amplitude measurements was real (the difference will be discussed below in detail).

Figure 4.

Inhibition constants of 2,3-BDZ-11-4 estimated from the amplitude of the whole-cell current through GluA2Q_flip channels by the solution flow measurement. At 3 mM glutamate (○), a $\overline{K_I}$ of 14 ± 0.4 μM was obtained, corresponding to the inhibition constant for the open-channel state. Similarly, a K_I of 2.0 ± 0.1 μM was obtained for the closed-channel state at a glutamate concentration of 100 μM (●).

It should be pointed out that at 3 mM glutamate, where ~95% of the channels were in the open-channel state (17), an inhibition constant was pertinent to the open-channel state of GluA2Q_flip (see also eqs 6a and 6b). However, reaching 3 mM glutamate by laser-pulse photolysis of caged glutamate was not technically feasible in our experiment. Furthermore, an inhibition constant obtained from the amplitude data collected from the laser experiment at 300 μM glutamate (data in column 4) was closer to an inhibition constant obtained from 100 μM glutamate (data in either column 3 or 5), rather than to 3 mM glutamate (data in column 6). This result was anticipated, because the fraction of the open-channel form that corresponds to 300 μM glutamate is ~10%, which is closer to ~4% at 100 μM glutamate but far from ~93% at 3 mM glutamate or the maximum limit set by the channel-opening probability for GluA2Q_flip (17).

2,3-BDZ-11-2 or 2,3-BDZ-11-4 Inhibits Channel Opening by a Two-Step Process

As described above, the inhibition constants calculated from the amplitude data from both the laser-pulse photolysis and the flow measurements were in good agreement (Table 1). However, those constants were 2–4-fold smaller than the inhibition constants estimated from rate measurement (Table 1). In other words, as compared with the full inhibition or full antagonism observed from the amplitude measurement, the effect of an inhibitor on rate projected only partial inhibition or partial antagonism. This discrepancy can be accounted for by the mechanism of inhibition that we have proposed previously for other 2,3-BDZ inhibitors (Figure 5) (11–13). Specifically, the initial binding of 2,3-BDZ-11-4, for instance, to the receptor forms a loosely bound channel complex (e.g., ${\overline{\mathit{IRL}}}_2^{\ast }$), and such a receptor-inhibitor complex is partially conducting, thereby resulting in only partial inhibition of receptor activity. Additional reduction of the current amplitude comes from the second step in which the receptor-inhibitor intermediate isomerizes rapidly into a more tightly bound complex (e.g., ${\overline{\mathit{IRL}}}_2$); such a complex is no longer capable of conducting ions. This two-step inhibition process involving the formation of loose intermediates applies to both the closed-channel and open-channel states (Figure 5).

Figure 5.

A minimal mechanism of the inhibition of the GluA2Q_flip receptor by 2,3-BDZ-11-2 and 2,3-BDZ-11-4. L represents ligand or glutamate, and the number of ligands that bind to and open the channel is assumed to be two. R represents the active, unliganded form of the receptor, and I represents an inhibitor. For simplicity and without contrary evidence, it is assumed that glutamate binds with equal affinity or K₁, the intrinsic equilibrium dissociation constant, at all binding steps. An asterisk indicates those species in the intermediate state, i.e., loose receptor-inhibitor complexes, whereas those species bound with inhibitor but without asterisk represent those in the final state of the receptor complexes. All species related to R, RL, and RL₂, including those bound with inhibitors, are in closed-channel states, whereas those related to $\overline{{RL}_2}$ refer to the open-channel state. Because channel desensitization rate constant is unaffected by 2,3-BDZ-11-2 (Figure S1) or 2,3-BDZ-11-4 (Figure 2B), the channel desensitization process is not included in this scheme, nor our analysis.

The mode of action of 2,3-BDZ-11-4 and 2,3-BDZ-11-2 can be explained by this mechanism of inhibition (Figure 5). Firstly, in the laser-pulse photolysis measurement, the effect of an inhibitor on both the channel-closing rate and the whole-cell amplitude was associated with a rapid channel-opening process. Therefore, a stronger inhibition or a smaller K_I value calculated from the amplitude, which was an equilibrium measure, suggested that a larger value of the inhibition constant observed from the rate would have to reflect a fraction of the overall inhibition (e.g., the values in column 1 vs. those in column 3 or column 5). Secondly, the partial inhibitory effect on the channel-opening rate would reflect the first step in Figure 5 for the formation of inhibitor-receptor complex; additional inhibition was thus generated from the second step in which the partially conducting inhibitor-receptor complex isomerizes into a totally inhibitory complex. Thirdly, in the presence of 2,3-BDZ-11-4, for example, the rate of channel opening of the GluA2Q_flip receptors was slow compared with the control (Figure 3A). A slower rate should reflect the first step involving the formation of the loose receptor-inhibitor intermediate; the second step, however, should be faster than the first step. If the rate of the second step were slow or comparable to that of the first step, full or virtually full inhibition would be expected in that the inhibition constants determined from the rate would be equal or nearly equal to those obtained from the amplitude data. A significant difference in rate is also consistent with the fact that over the entire concentration range for both glutamate and the inhibitors, only a single exponential rise for the rate of channel opening was observed. Therefore, the effect of an inhibitor on the channel-opening rate was analyzed adequately by use of the one-step process (eqs 4 and 5). That (1/k_obs) increased linearly with increasing inhibitor concentration (Figure 3B and 3C), as predicted (by eqs 4 and 5), for both the closed- and the open-channel states, further supports the notion that the rate of the isomerization reaction would be faster than the first step. Consequently, ${\overline{K_I}}^{\ast }$ and $K_I^{\ast }$ values (Table 1) obtained from the rate measurement reflected partial inhibition of the open- and the closed-channel states, respectively, by 2,3-BDZ-11-4 or 2,3-BDZ-11-2 in the first step (Figure 5).

The finding that 2,3-BDZ-11-4 and 2,3-BDZ-11-2 inhibited both k_op and k_cl, albeit partial, is consistent with a noncompetitive mechanism of inhibition for both inhibitors but inconsistent with either a competitive or an uncompetitive mechanism. By a competitive mode of action, 2,3-BDZ-11-4 would be expected to compete with glutamate for the same binding site; therefore, only the effect on k_op, but not on k_cl, would be expected. Consequently, there would be no [ $\overline{K_I}/(\overline{K_I}+I)$] term associated with k_cl in eq 3, and thus 1/k_obs as in eq 4 would be independent of inhibitor concentration. By an uncompetitive mode of action, commonly known as open-channel blockade, 2,3-BDZ-11-4 would inhibit the open-channel state only. As such, only the effect on k_cl, but not on k_op, would be expected. In this case, the [K_I/(K_I + I)] term associated with k_op in eq 3 does not exist. As a result, the (k_obs − k_cl′) term, as in eq 5, would not depend on inhibitor concentration.

2,3-BDZ-11-2 and 2,3-BDZ-11-4 Inhibited the Closed-Channel State More Strongly than the Open-Channel State of Both the Flip and the Flop Isoforms of GluA2

On the basis of the overall inhibition constants (Table 1), both 2,3-BDZ-11-2 and 2,3-BDZ-11-4 were selective for the closed-channel state of the GluA2Q_flip channel (e.g., K_I of 2.0 ± 0.1 μM versus $\overline{K_I}$ of 14 ± 0.4 μM for 2,3-BDZ-11-4, Figure 4). It is also worth noting that 2,3-BDZ-11-4 showed a comparable potency and selectivity for the flop variant of GluA2Q or GluA2Q_flop. GluA2Q_flop is the alternatively spliced isoform of GluA2Q_flip, and the two isoforms differ by only 8 amino acids (22). However, the homomeric GluA2Q_flip and GluA2Q_flop channels have different kinetic properties such that the flop variant of GluA2 has the same k_op but a larger k_cl than the flip variant (23). For this study, we used the flow technique and determined the overall inhibition constants for the closed-channel and open-channel states of both types of homomeric channels. We found that 2,3-BDZ-11-4 inhibited the GluA2Q_flop receptors expressed in HEK-293 cells with a K_I of 2.4 ± 0.1 μM for the closed-channel and a $\overline{K_I}$ of 19 ± 0.2 μM for the open-channel states (Figure 6)., Furthermore, 2,3-BDZ-11-2 inhibited the GluA2Q_flop receptors with a K_I of 25 ± 0.3 μM for the closed-channel and a $\overline{K_I}$ of 33 ± 1.0 μM for the open-channel states (Figure S5 in Supporting Information). These inhibition constants were comparable to those of GluA2Q_flip (Table 1), suggesting that 2,3-BDZ-11-4 and 2,3-BDZ-11-2 were equally effective on GluA2Q_flop, and both were also selective for the closed-channel state of the GluA2Q_flop channel.

Figure 6.

Inhibition constants of 2,3-BDZ-11-4 for the flop variant of the GluA2Q or GluA2Q_flop. The constants were estimated from the amplitude of the whole-cell current in the absence and presence of 2,3-BDZ-11-4, collected from the flow measurement. A $\overline{K_I}$ of 19.0 ± 0.2 μM was obtained at 3 mM glutamate concentration (○), which corresponds to the inhibition constant for the open-channel state. A K_I of 2.4 ± 0.1 μM was obtained at a glutamate concentration of 100 μM (●), which corresponds to the closed-channel state.

2,3-BDZ-11-2 and 2,3-BDZ-11-4 Bind to the Same Site on the GluA2Q_flip Receptor

Our results show that 2,3-BDZ-11-2 and 2,3-BDZ-11-4 are noncompetitive inhibitors, although 2,3-BDZ-11-4 is stronger than 2,3-BDZ-11-2. Based on the structural (Figure 1) and functional similarities between the two compounds, we asked whether they competed at the same noncompetitive site or bound to two different sites on the same receptor. The outcome would provide important insights into the structure-function relationship of 2,3-BDZ derivatives. To address this question, we carried out a double-inhibitor experiment (see Experimental Procedures) in which 2,3-BDZ-11-2 and 2,3-BDZ-11-4 were used simultaneously to inhibit the GluA2Q_flip receptor. The double-inhibition constant, K_I′, was 1.8 ± 1.0 μM (Figure 7A), which was comparable to the K_I of 2.0 ± 0.1 μM for 2,3-BDZ-11-4 alone (Figure 4 and Table 1). This result suggested that the two inhibitors compete for the same noncompetitive site on GluA2Q_flip receptors. If 2,3-BDZ-11-2 and 2,3-BDZ-11-4 bound to two different noncompetitive sites, a much stronger inhibition produced by the presence of two inhibitors would likely be observed in the same concentration range (Figure 7A, dashed line, simulated by eq 8).

Figure 7.

Double-inhibition experiment to characterize inhibitor binding sites. (A) 2,3-BDZ-11-2 and 2,3-BDZ-11-4 bind to the same site on GluA2Q_flip. In this experiment, the concentrations of 2,3-BDZ-11-2 and glutamate were fixed at 20 μM and 100 μM, respectively. From the amplitude data in the presence of both 2,3-BDZ-11-2 and 2,3-BDZ-11-4 (●), the double-inhibition constant, K_I′, was determined to be 1.8 ± 1.0 μM (using eq 7), as compared with the K_I of 2.0 ± 0.1 μM for 2,3-BDZ-11-4 alone (○; Figure 4 and Table 1). The dashed line is drawn on the basis of the simulated result from eq 8, assuming that the two inhibitors bound to two different sites. (B) 2,3-BDZ-11-2 and 2,3-BDZ-2 do not bind to the same site on GluA2Q_flip. The concentrations of 2,3-BDZ-2 and glutamate were kept at 20 μM and 100 μM, respectively. The double-inhibition constant, ${\overline{K_I}}^{\prime }$, was determined _I to be 10 ± 2.0 μM (●), as compared with the $\overline{K_I}$ of 21 ± 0.1 μM for 2,3-BDZ-11-2 alone (○; Figure S4 in Supporting Information). The dashed line is the simulation based on eq 7, assuming that the two inhibitors bound to the same site. (C) 2,3-BDZ-11-2 and GYKI 52466 do not bind to the same site on GluA2Q_flip. In this experiment, the concentrations of GYKI 52466 and glutamate were fixed at 15 μM and 100 μM, respectively, while the concentration of 2,3-BDZ-11-2 was varied. The double-inhibition constant, ${\overline{K_I}}^{\prime }$, was 14.0 ± 3.0 μM (●), compared with $\overline{K_I}$ of 21 ± 0.1 μM for 2,3-BDZ-11-2 alone (○; Figure S4 in Supporting Information). The dashed line represents the simulated A/A_I values by using eq 7, assuming that the two inhibitors bound to the same site.

We previously reported a unique noncompetitive binding site for 2,3-BDZ-2 (11) and another site for GYKI 52466 (see Figure 1) (12, 13). The molecular determinant of the two sites is thought to be the nature of the C-4 substitution on the diazepine ring. Specifically, the C-4 position in GYKI 52466 is a methyl group whereas in 2,3-BDZ-2 is a carbonyl group. Furthermore, GYKI 52466 is selective for the closed-channel state (12, 13), whereas 2,3-BDZ-2 prefers to inhibit the open-channel state of GluA2Q_flip (11). Because 2,3-BDZ-11-2 and 2,3-BDZ-11-4 share the C-4 carbonyl feature with 2,3-BDZ-2 (Figure 1), it was assumed that they would compete with the 2,3-BDZ-2 binding site. However, double-inhibition experiments revealed that 2,3-BDZ-11-2 and 2,3-BDZ-11-4 did not bind to the 2,3-BDZ-2 site (Figure 7B) nor to the GYKI 52466 site (Figure 7C); in both cases (Figure 7B and 7C), a much stronger inhibition was observed in the presence of either 2,3-BDZ-2 or GYKI 52466, together with 2,3-BDZ-11-2, than inhibition by a predicted single site (the dashed line in both Figure 7B and 7C). These results suggested that the noncompetitive site where 2,3-BDZ-11-2 and 2,3-BDZ-11-4 bound was new or different from either of the two sites we reported previously (11–13).

The Effect of Replacing 7,8-Methylenedioxy with 7,8-Ethylenedioxy Moiety on Binding Sites of 2,3-BDZ Compounds

Structurally, 2,3-BDZ-11-2 is most similar to 2,3-BDZ-2 (Figure 1). The two compounds were synthesized (3, 4) based on the structure of GYKI 52466 (Figure 1) (24, 25). To replace the 4-methyl group of GYKI 52466 with a carbonyl group in order to synthesize 2,3-BDZ-2, the double bond between N-3 and C-4 of the diazepine ring must be saturated (Figure 1). Thus the structural difference between GYKI 52466 and 2,3-BDZ-2 is a replacement of the azomethine moiety with an ε-lactam moiety (alternatively, this structural change can be considered a replacement of the iminohydrazone moiety of GYKI 52466 by an iminohydrazide moiety, resulting in 2,3-BDZ-2). As we previously reported, such a replacement forces 2,3-BDZ-2 to bind to a new noncompetitive site on GluA2Q_flip (11–13). In other words, if a methyl group is at the C-4 position (e.g., GYKI 52466) of the diazepine ring, the compounds bind to one site(12, 13); but if a carbonyl group is at the C-4 position, those compounds bind to another site (11).

When 2,3-BDZ-11-2 and 2,3-BDZ-2 are compared (Figure 1), the only structural difference is that the 7,8-methylenedioxy moiety on 2,3-BDZ-2 is replaced with a 7,8-ethylenedioxy moiety as in 2,3-BDZ-11-2. This structural change also results in the change of the binding site. Furthermore, based on inhibition constants (Table 1), 2,3-BDZ-11-2 is roughly as effective as 2,3-BDZ-2 in inhibiting the closed-channel state of GluA2Q_flip (i.e., K_I is 21 μM for 2,3-BDZ-11-2 and 25 μM for 2,3-BDZ-2; Table 1) but it is almost 5-fold less potent in inhibiting the open-channel state. Thus, the functional consequence of replacing the 7,8-methylenedioxy moiety on 2,3-BDZ-2 with the 7,8-ethylenedioxy moiety, yielding 2,3-BDZ-11-2, is that such a replacement causes a change of binding site for 2,3-BDZ-11-2 but at a cost of the reduction of the potency. However, that reduction in potency for 2,3-BDZ-11-2, which is only for the open-channel state, turns it into a closed-channel-preferring inhibitor, unlike 2,3-BDZ-2 (11).

The results from this study show the presence of the 7,8-ethylenedioxy moiety on the diazepine ring (Figure 1) renders 2,3-BDZ-11-2 (and 2,3-BDZ-11-4) incapable of binding to either the 2,3-BDZ-2 site (Figure 7B) or the GYKI 52466 site (Figure 7C). In other words, if there is an additional change at the 7,8-position in the diazepine ring, the nature of the C-4 substitution is no longer a dominant determinant of both the binding site and the conformational selectivity of the resulting compounds (11–13). Our results suggests that the size of the ring at the 7,8-position seems to be a more critical “atomic descriptor” in that the increase in ring size from a 5-membered to a 6-membered ring (i.e., from dioxole to dioxane moiety) overrides the nature of the C-4 substitution. The impact of the increase in ring size from a dioxole to dioxane moiety on the binding site for the resulting compounds might reflect the intimate molecular contact between the 7,8-position and the receptor site. Two interesting questions now arise. First, can the correlation between a size change of the dioxole ring and the binding site be extended beyond the ring enlargement from the dioxole to the dioxane moiety? Second, can modulation of the dioxole ring, i.e., the methylene group of the methylenedioxy function, be used as a novel method to explore undiscovered, noncompetitive binding sites? For example, the 7,8-methylenedioxy moiety can be replaced with 7,8-ethylidenedioxy or 7,8-isopropylidenedioxy moieties. Alternatively, a hydrogen atom on the 7,8-methylenedioxy moiety can be substituted with a chlorine atom to generate a different analogue. Such changes could produce new 2,3-BDZ analogues with different potency, and these new analogues could also bind to new regulatory site(s) on the same receptor. In other words, these structural changes may allow us to find new sites for developing more potent and more selective 2,3-BDZ compounds.

Comparison of Potency of 2,3-BDZ-11-2 with 2,3-BDZ-11-4

2,3-BDZ-11-4 and 2,3-BDZ-11-2 bind to the same site because both have the 7,8-ethylenedioxy moiety on the diazepine ring. However 2,3-BDZ-11-4 is ~10-fold and ~3-fold stronger in inhibiting the closed- and open-channel states, respectively, than 2,3-BDZ-11-2. These results suggest that substitution of a hydrogen atom by a chlorine atom at C-3 on the aminophenyl ring increases the potency of the resulting compound, i.e., 2,3-BDZ-11-4, without changing the binding site nor the mechanism of action or the conformational selectivity for the GluA2Q_flip receptor. The higher potency of 2,3-BDZ-11-4 seems to be realized even at the first step involving the formation of the initial receptor-inhibitor intermediate, as judged by the inhibition constants for both the open- and closed-channel states between 2,3-BDZ-11-2 and 2,3-BDZ-11-4 (Figure 5 and Table 1). Specifically, the receptor–2,3-BDZ-11-4 intermediate formed in the closed-channel state is ~4-fold more inhibitory than the receptor–2,3-BDZ-11-2 counterpart (i.e., $K_I^{\ast }=9\mu \text{M}$ for 2,3-BDZ-11-4 and 39 μM for 2,3-BDZ-11-2; Table 1). Furthermore, the receptor–2,3-BDZ-11-4 intermediate formed in the open-channel state is 2-fold more inhibitory than the receptor–2,3-BDZ-11-2 (i.e., ${\overline{K_I}}^{\ast }$ is 24 μM for 2,3-BDZ-11-4 but 43 μM for 2,3-BDZ-11-2; Table 1). This analysis suggests that new compounds can be made that bind to this new site but with a higher potency than 2,3-BDZ-11-2. Previously we reported that the N-3 position on the benzodiazepine ring is one place where acetylation can improve the potency of resulting compounds (11–13). The aminophenyl ring is now the second place where chemical modifications can also improve the potency of the resulting compounds without either changing the mechanism of action or the site of binding.

We do not know, however, whether the improved potency of the resulting compound, i.e., 2,3-BDZ-11-4, compared with 2,3-BDZ-11-2, is due to the addition of an electron-withdrawing group (i.e., chlorine atom) or to other factors. If it is the former, the presence of an electron-withdrawing group on the aminophenyl ring could stabilize the interaction between the aminophenyl moiety of the 2,3-BDZ compound with the receptor site (or most likely a pocket that accommodates the aminophenyl ring). If this is the case, substitution of more hydrogen atoms on the aminophenyl ring by additional chlorine atoms or even other electron-withdrawing groups/moieties might yield even more potent 2,3-BDZ derivatives. Other factors, however, could also be involved in defining the potency of the resulting compounds and should also be explored. These factors include shape and stereochemical arrangement of substitutions at the ortho position. It is important to note, however, that in order to construct better inhibitors, any new possibilities for generating chemical modifications at the ortho position must adhere to the 7,8-ethylenedioxy moiety on the diazepine ring in their structures, like 2,3-BDZ-11-4 or 2,3-BDZ-11-2. If new compounds were to bind to a different site, the predictions based on this unique feature of the structure-activity relationship will not be applicable.

In conclusion, we found that substituting the methylenedioxy with ethylenedioxy feature at the 7,8-position in the benzodiazepine ring is more critical than the C-4 substitution in determining the site of binding for the resulting compounds. As a result, the nature of the C-4 substitution as we reported earlier is only a dominant determinant of both the binding site and the conformational selectivity of 2,3-BDZ compounds with the 7,8-methylenedioxy moiety. A ring enlargement, at least from 7,8-methylenedioxy to 7,8-ethylenedioxy ring size, then makes the resulting 2,3-BDZ compounds bind to an unique noncompetitive site, the third one we have reported thus far. That a simple change of structure leads to a dramatic change of site may well reflect an intimate molecular contact between the 7,8-methylenedioxy ring and presumably a tight receptor pocket. We therefore speculate that chemical modifications of the 7,8-methylenedioxy ring, in addition to the ring enlargement, lead to new compounds that may likely bind to new sites. Furthermore, for the 2,3-BDZ compounds with the 7,8-ethylenedioxy feature, chemical modifications of the 1-(4-aminophenyl) ring can improve the potency of the resulting compounds.

ABBREVIATIONS

AMPA

α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid

2,3-BDZ

2,3-benzodiazepine

GYKI 52466

1-(4-aminophenyl)-4-methyl-7,8-methylenedioxy-5H-2,3-benzodiazepine

2,3-BDZ-2

1-(4-aminophenyl)-3,5-dihydro-7,8-methylenedioxy-2,3-benzodiazepine-4-one

2,3-BDZ-11-2

1-(4-aminophenyl)-3,5-dihydro-7,8-ethylenedioxy-4H-2,3-benzodiazepin-4-one

2,3-BDZ-11-4

1-(4-amino-3-chlorophenyl)-3,5-dihydro-7,8-ethylenedioxy-4H-2,3-benzodiazepin-4-one

HEK

human embryonic kidney