Modulation of the mammalian GABA_A receptor by type I and type II positive allosteric modulators of the α7 nicotinic acetylcholine receptor

Abstract

Background and Purpose:

Positive allosteric modulators of the α7 nAChR (α7-PAMs) possess promnesic and procognitive properties, and have potential in the treatment of cognitive and psychiatric disorders including Alzheimer’s disease and schizophrenia. Behavioral studies in rodents have indicated that α7-PAMs can also produce antinociceptive and anxiolytic effects that may be associated with positive modulation of the GABA_AR. The overall goal of this study was to investigate the modulatory actions of selected α7-PAMs on the GABA_A receptor.

Experimental Approach:

We employed a combination of cell fluorescence imaging, electrophysiological, functional competition, and site-directed mutagenesis to investigate the functional and structural mechanisms of modulation of the GABA_A receptor by three representative α7-PAMs.

Key Results:

We show that the α7-PAMs at micromolar concentrations enhance the apparent affinity of the GABA_A receptor to the transmitter and potentiate current responses from the receptor. The compounds were equi-effective at binary αβ and ternary αβγ GABA_A receptors. Functional competition and site-directed mutagenesis indicate that the α7-PAMs bind to the classic anesthetic binding sites in the transmembrane region in the intersubunit interfaces, which results in stabilization of the active state of the receptor.

Conclusion and Implications:

We conclude that the tested α7-PAMs are micromolar-affinity, intermediate- to low-efficacy allosteric potentiators of the mammalian αβγ GABA_AR. Given the similarities in in vitro sensitivities of the nicotinic α7 and α1β2γ2L GABA_A receptors to α7-PAMs, we propose that the doses used to produce nicotinic receptor-mediated behavioral effects in vivo are likely to modulate GABA_AR function.

INTRODUCTION

Nicotinic acetylcholine receptors (nAChRs) and γ-aminobutyric acid type A receptors (GABA_ARs) are members of the pentameric ligand-gated ion channel superfamily (Ortells and Lunt, 1995). nAChRs are permeable to mono- and divalent cations and presynaptically modulate transmitter release or contribute to excitatory neurotransmission, whereas GABA_ARs are permeable to monovalent anions (e.g., Cl⁻ and HCO₃⁻) and mediate fast inhibitory neurotransmission in the brain. These receptors perform many important physiological functions in the central and peripheral nervous systems, and their malfunction by excessive or decreased activity is associated with neuropsychiatric and neurological diseases, including cognitive impairment, depression, anxiety, drug addiction, chronic pain, and epilepsy (Cerne et al., 2022; Dineley et al., 2015; Hernandez and Macdonald, 2019; Papke et al., 2020).

A class of positive allosteric modulators with selectivity for the α7 nAChR (α7-PAMs) has been suggested to have potential in the treatment of cognitive and psychiatric disorders including Alzheimer’s disease and schizophrenia (Hurst et al., 2005; Timmermann et al., 2007). Based on their macroscopic mechanism of action, α7-PAMs are classified as type I (e.g., NS-1738) and type II modulators (e.g., PNU-120596 and PAM-2) that differ in how they modulate the rate of desensitization and reactivate desensitized receptors (Andersen et al., 2016; Arias et al., 2011; Bertrand et al., 2008; Collins et al., 2011; Papke et al., 2018). Preclinical studies have demonstrated that, besides promnesic and procognitive properties likely mediated by potentiation of the α7 nAChR, some α7-PAMs elicit behavioral effects, such as antinociception and anxiolysis, that may be associated with positive modulation of the GABA_AR (Alzarea and Rahman, 2019; Arias et al., 2020; Potasiewicz et al., 2017; Potasiewicz et al., 2015; Quadri et al., 2018; Targowska-Duda et al., 2019a). However, to the best of our knowledge, there are no published reports on the activity of α7-PAMs at GABA_ARs.

The goal of this study was to investigate the modulatory actions of α7-PAMs on the GABA_A receptor. Using a combination of imaging fluorescence and electrophysiology, we show that representative type I (NS-1738) and type II (PAM-2, PNU-120596) α7-PAMs potentiate the mammalian GABA_AR by stabilizing the active state, and that the effect is mediated by interactions with the classic anesthetic binding sites located in intersubunit interfaces in the transmembrane domain.

METHODS

Fluorescence imaging assays using GABA_ARs-expressing HEK293T cells

Fluorescence imaging experiments were carried out on HEK293T-α1β3γ2L cells using gabazine-oregon green (Gzn-OG), a derivative of the GABA_AR competitive antagonist gabazine that becomes fluorescent upon binding to the orthosteric binding site of the GABA_AR (Sakamoto et al., 2019). The α1, β3, and γ2L subunits were transiently expressed in HEK293T cells as previously described (Sakamoto et al., 2019). HEK293T-α1β3γ2L cells were preincubated with Gzn-OG (100 nM) for 5 min at 25 °C, followed by removal of the medium and wash with HBS (2 mL x 2). Cells with α1β3γ2L-bound Gzn-OG were titrated with 0.01 μM - 30 mM GABA in the absence (control) or presence of fixed concentrations of PAM-2, NS-1738, or PNU-120596 dissolved in 0.1% DMSO/HBS. Fluorescence was measured by confocal laser scanning microscopy (CLSM) on single confocal sections. In control experiments, incubation in 0.1% DMSO had no effect on CLSM images of Gzn-OG binding to the GABA_AR (Sakamoto et al., 2019).

Cell fluorescence imaging analysis was performed using a Carl Zeiss CLSM (LSM-800, Germany) equipped with a 63×, NA = 1.40 oil objective, and a GaAsP detector. CLSM images were acquired using 488 nm excitation derived from diode lasers (0.75%, Gain 750 V) and setting emission at 500 nm, with control of the focus using the Definite Focus module included in LSM-800.

The fluorescence intensity of a single cell in the absence (F_o) and presence (F) of GABA was determined by enclosing the regions of interest. Non-specific Gzn-OG fluorescence was determined at 30 mM GABA. To estimate the apparent IC₅₀ for GABA, non-linear regression fitting of the F/F_o ratios (mean±SEM) was performed using KaleidaGraph 4.5 (Synergy Software, Reading, PA, USA), according to the following logistic equation:

$$\text{F}/{\text{F}}_{\text{o}}=1–\left[1/(1+\text{log}((\right[\text{GABA}]/{\text{IC}}_{50}{)}^{{\text{n}}_{\text{H}}}))]$$ (1)

where [GABA] is the concentration of GABA, IC₅₀ is the ligand concentration that produces half-maximal inhibition, and n_H is the Hill coefficient. The estimated IC₅₀s were transformed to apparent K_i (equilibrium inhibition constant) using the Cheng-Prusoff equation (Cheng and Prusoff, 1973):

$${\text{K}}_{\text{i}}=\frac{{\text{IC}}_{50}}{1+\frac{[\text{Gzn-OG}]}{{\text{K}}_{\text{d},\text{Gzn-OG}}}}$$ (2)

where [Gzn-OG] is the initial concentration of Gzn-OG (100 nM), and K_d,Gzn-OG is the K_d for Gzn-OG (55 nM; (Sakamoto et al., 2019)).

Electrophysiological recordings of GABA_ARs expressed in Xenopus laevis oocytes

Rat (α1β2γ2L or α1β2) and human (α1β3γ2L, α1β3, or α4β2δ) wild-type and mutant GABA_ARs were expressed in Xenopus laevis oocytes, and two-electrode voltage-clamp recordings were conducted as described in detail previously (Germann et al., 2019a; Shin et al., 2017). In brief, oocytes purchased from Xenopus 1 (Dexter, MI, USA) were injected with a total of 3.5-6 ng of cRNA per oocyte in the ratio of 5:1 (α:β) or 1:1:5 α:β:γ), and incubated in ND96 (96 mM NaCl, 2 mM KCl, 1.8 mM CaCl₂, 1 mM MgCl₂, 5 mM HEPES; pH 7.4) with supplements (2.5 mM Na pyruvate, 100 U/ml penicillin, 100 μg/ml streptomycin and 50 μg/ml gentamycin) for 1-2 days.

For electrophysiological recordings, the oocytes were placed in a recording chamber (RC-1Z, Warner Instruments, Hamden, CT, USA) and clamped at −60 mV. Solutions were gravity-applied from glass syringes and switched manually using 4-port bulkhead switching valves and medium pressure 6-port bulkhead valves (IDEX Health and Science, Rohnert Park, CA, USA). The current responses were amplified with an OC-725C amplifier (Warner Instruments, Hamden, CT, USA), digitized with a Digidata 1200 series digitizer (Molecular Devices), and stored using pClamp (Molecular Devices).

The direct activating effects of α7-PAMs were studied by exposing oocytes expressing α1β2γ2L GABA_ARs to 50 μM NS-1738, PAM-2, or PNU-120596. The duration of drug application was 30-60 s. For normalization purposes, each oocyte was also tested with 1 mM GABA+50 μM propofol that was considered to generate a response with the peak probability of being in the active state (P_A) of ~1 (Shin et al., 2017).

The modulatory effects of α7-PAMs were examined under three experimental protocols. First, modulation was established by coapplying α7-PAM during a steady-state response to a low concentration (P_A<0.10) of GABA. Drug effects were quantified by calculating fold-potentiation from the ratio of the peak response to GABA+α7-PAM to the steady-state response to GABA immediately before application of the modulator. In most cases, the peak and steady-state responses in the presence of GABA before the application of α7-PAM were similar (within 10% of each other), i.e., there was little desensitization during exposure to a low concentration of GABA. Analogous experiments were conducted replacing GABA with the allosteric agonist 3α5βP, R(+)-etomidate, phenobarbital, or propofol. The results are expressed as % of control, to eliminate variation originating from small differences in P_A of control response. Second, the α7-PAM concentration-response relationships were recorded by coapplying various concentrations of the drugs with a low concentration (P_A<0.10) of GABA and measuring concentration-dependent changes in the peak response. The drug applications were 20-40 s in duration, with successive applications separated by 1-3 min washouts. Each cell was exposed to the full range of α7-PAM concentrations. The concentration-response relationships were initially analyzed by fitting the Hill equation to the data to estimate the potentiating EC₅₀, maximal effect, and Hill coefficient. In both protocols, each cell was also tested with 1 mM GABA+50 μM propofol to obtain a reference response for normalization. Lastly, the effects of α7-PAMs were tested by coapplying a single, high concentration (50 μM) of an α7-PAM during a steady-state response to a saturating concentration (1 mM) of GABA.

Mechanistic analysis of GABA_AR potentiation by α7-PAMs

For mechanistic analysis of potentiation concentration-response curves, the raw current amplitudes were converted to P_A units through normalization to the peak response to 1 mM GABA+50 μM propofol tested in the same oocyte (Akk et al., 2018; Eaton et al., 2016). The peak current responses were analyzed in the framework of a cyclic two-state concerted transition model (Fig. S1). The model and its properties have been described in detail previously (Forman, 2012; Steinbach and Akk, 2019). The concentration-response curves for the α7-PAMs were fitted to the state function:

$${\text{P}}_{\text{A}}=\frac{1}{1+{\text{L}}^{*}{\left[\frac{1+[\text{drug}]/{\text{K}}_{\text{R},\text{drug}}}{1+[\text{drug}]/({\text{K}}_{\text{R},\text{drug}}{c}_{\text{drug}})}\right]}^{{\text{N}}_{\text{drug}}}}$$ (3)

where L* expresses the P_A of activity in the presence of low GABA alone and is calculated as (1-P_A,GABA)/P_A,GABA. K_R,drug is the equilibrium dissociation constant of an α7-PAM in the resting receptor, c_drug is the ratio of the equilibrium dissociation constant of a drug in the active receptor to K_R,drug, [drug] is the concentration of the α7-PAM under study, and N_drug is the number of binding sites.

To determine whether two drugs (e.g., a combination of NS-1738 and PAM-2, or a neurosteroid with an α7-PAM) interact with distinct or overlapping binding sites, the observed potentiating effect in the presence of a drug combination was compared to predicted responses calculated using two models. First, a prediction was made assuming pure energetic additivity. In this model, each drug interacts with a distinct set of binding sites (Fig. S2), and one drug (e.g., drug X), alone or in combination with GABA, acts by increasing P_A,background (reducing L*) at which the response to the other drug (e.g., drug Y) is measured. The predicted P_A was calculated as:

$${\text{P}}_{\text{A}}=\frac{1}{1+{\text{L}}^{*}{\left[\frac{1+[\text{Y}]/{\text{K}}_{\text{R},\text{Y}}}{1+[\text{Y}]/({\text{K}}_{\text{R},\text{Y}}{c}_{\text{Y}})}\right]}^{{\text{N}}_{\text{Y}}}}$$ (4)

where L* expresses background activity in the presence of drug X or GABA+drug X, and the other terms relate to the concentration and properties of drug Y as described above for Eq. (3).

In the second model, predictions were made assuming that drugs X and Y compete for common or overlapping binding sites (Fig. S3). The predicted peak responses were calculated as:

$${\text{P}}_{\text{A}}=\frac{1}{1+{\text{L}}^{*}{\left[\frac{1+[\text{X}]/{\text{K}}_{\text{R},\text{X}}+[\text{Y}]/{\text{K}}_{\text{R},\text{Y}}}{1+[\text{X}]/({\text{K}}_{\text{R},\text{X}}{c}_{\text{X}})+[\text{Y}]/({\text{K}}_{\text{R},\text{Y}}{c}_{\text{Y}})}\right]}^{\text{N}}}$$ (5)

where L* expresses background activity, and other terms relate to the concentrations and properties of drugs X and Y as described above. N is the number of common binding sites for the combined drugs. This approach has been described in detail previously (Shin et al., 2019).

Modeling results were compared by calculating the difference in second-order Akaike information criterion scores of the two models (Burnham et al., 2011; Wagenmakers and Farrell, 2004):

$$\text{Δ= n ln}\left(\frac{{\text{RSS}}_{\text{Model 1}}}{\text{n}}\right)-\text{n ln}\left(\frac{{\text{RSS}}_{\text{Model 2}}}{\text{n}}\right)$$ (6)

where n is the number of replicates, RSS is the residual sum of squares, and Models 1 and 2 refer to the models considering distinct sites and same sites for the combined drugs, respectively. Akaike weights (w) for each model were calculated as:

$$w_{\text{shared sites}}\text{=}\frac{\text{exp}[-\frac{1}{2}\text{Δ}]}{\text{exp}[-\frac{1}{2}\text{Δ}]+1}$$ (7)

and

$$w_{\text{distinct sites}}\text{=1−}{w}_{\text{shared sites}}$$ (8)

Data and statistical analysis

The data and statistical analysis comply with recommendations and requirements on experimental design and analysis in pharmacology (Curtis et al., 2018). The presented results are from at least five independent replicates per experiment. No statistical methods were used to estimate appropriate sample size. Statistical analyses were conducted on the independent values, and all data are included in data analysis and presentation. Some data are expressed as % of control response to reduce unwanted sources of variation. Data (mean±SEM) analysis was done using Excel 2016 (Microsoft, Redmond, WA, USA) and Stata/IC 12.1 (StataCorp LLC, College Station, TX, USA). Curve fitting was performed using Origin 2020 (OriginLab Corp., Northampton, MA, USA). Dunnett’s test was used to compare the effects of α7-PAMs on receptors activated by different agonists and to determine the effects of amino acid substitutions on drug activity. Student’s t-test was used to compare drug effects on receptor subtypes. P<0.05 was assigned statistical significance. Randomisation or blinding of the operator or data analysis were not undertaken due to the nature of experiments.

Materials

Salts, solvents, and HEPES were purchased from Sigma-Aldrich. R(+)-Etomidate was purchased from Toronto Research Chemicals Inc (North York, Ontario, Canada). GABA, propofol, phenobarbital, DMEM high glucose, fetal bovine serum, and 0.01% poly-L-Lys solution (70-150 kDa) were purchased from Sigma-Aldrich (St. Louis, MO, USA). The steroid pregnanolone (3α5βP) was purchased from BioTechne Corp. (Minneapolis, MN, USA). Lipofectamine 2000, Opti-MEM, and trypsin-EDTA were obtained from Thermo Fischer Scientific (Waltham, MA, USA). NS-1738 and PNU-120596 were obtained from Cayman Chemical (Ann Arbor, MI, USA) and Adooq Bioscience (Irvine, CA, USA). PAM-2 was synthesized as described previously (Arias et al., 2011). The structures of NS-1738, PAM-2, and PNU-120596 are provided in Fig. 1. Key protein targets and ligands in this article are hyperlinked to corresponding entries in http://www.guidetopharmacology.org, and are permanently archived in the Concise Guide to PHARMACOLOGY 2021/22 (Alexander et al., 2021).

Figure 1. Structures of α7-PAMs.

Molecular structures of NS-1738 [N-(5-Cl-2-hydroxyphenyl)-N’-[2-Cl-5-(trifluoromethyl)phenyl]-urea], PAM-2 [(E)-3-(furan-2-yl)-N-(p-tolyl)-acrylamide], and PNU-120596 [N-(5-Cl-2,4-dimethoxyphenyl)-N’-(5-methyl-3-isoxazolyl)-urea].

RESULTS

α7-PAMs increase the apparent affinity of GABA at the α1β3γ2L GABA_AR

To determine the effects of α7-PAMs on the apparent affinity of GABA at the α1β3γ2L GABA_AR, GABA_AR-bound Gzn-OG fluorescence was monitored at varying concentrations of GABA in the absence and presence of 100 μM NS-1738, PAM-2, or PNU-120596 (Fig. 2A). Increasing concentrations of GABA decreased α1β3γ2L-bound Gzn-OG fluorescence, making it nearly undetectable at 1 mM GABA. Further reduction in fluorescence was observed when GABA was combined with an α7-PAM, with NS-1738 being more effective than PAM-2 or PNU-120596 at increasing displacement of Gzn-OG by GABA.

Figure 2. Effects of α7-PAMs on GABA-induced displacement of the fluorescent probe Gzn-OG.

(A) Confocal images of HEK293T-α1β3γ2L live cells upon addition of Gzn-OG (100 nM), in the absence and presence of increasing concentrations (1 μM-10 mM) of GABA administered alone or in the presence of 100 μM NS-1738, PAM-2, or PNU-120596. (B-D) Fluorescence intensity of HEK293T-α1β3γ2L-bound Gzn-OG in the presence over absence (F/F_o) of 0.01 μM-30 mM GABA in the absence or presence of NS-1738 (B), PAM-2 (C), or PNU-120596 (D). The GABA IC₅₀s estimated from non-linear regression analysis of F/F_o vs [GABA] relationships (mean±SEM; n=14-21 cells/titration) were transformed to apparent K_i values using Eq. (2). (E) Effects of α7-PAMs on α1β3γ2L GABA_AR-bound Gzn-OG (n = 14-21 cells/titration). The α7-PAMs at up to 100 μM did not displace Gzn-OG from the orthosteric binding site.

The IC₅₀s of fluorescence curves (Fig. 2) were transformed to apparent K_i using Eq. 2. In the presence of increasing concentrations of NS-1738 or PAM-2, the GABA concentration-response curves were shifted to lower transmitter concentrations (Fig. 2B–C). The observed shift in K_i between 0 (control) and 100 μM α7-PAM was more pronounced for NS-1738 (from 16.8±1.5 to 0.67±0.04 μM; a 25-fold decrease) compared to that for PAM-2 (from 18.5±0.8 to 4.9±0.2 μM; a 3.8-fold decrease), indicating that the former is more potent/efficient than the latter. The decrease in the GABA K_i values indicates an increase in the apparent binding affinity of GABA in the presence of α7-PAM. PNU-120596 exhibited complex pharmacological behavior (Fig. 2D). Leftward shift in F/F_o curves was observed at 3-10 μM, while higher concentrations had no effect (18.1±1.7 μM at 30 μM) or resulted in a small rightward shift (23.2±1.2 μM at 100 μM) compared to control values (16.5±1.2 μM), suggesting a decrease in GABA affinity.

We also assessed whether α7-PAMs directly affect the interaction of Gzn-OG with the orthosteric site at GABA_ARs (Fig. 2E). The results indicated that 3-100 μM NS-1738, PAM-2 or PNU-120596 do not modify fluorescence from α1β3γ2L-bound Gzn-OG. Overall, these results suggest that α7-PAMs allosterically and in a concentration-dependent manner increase the apparent affinity of GABA to the GABA_AR.

α7-PAMs activate and potentiate the α1β2γ2L GABA_AR by interacting with an allosteric binding site

The first set of electrophysiological results indicated that α7-PAMs weakly directly activate the α1β2γ2L GABA_AR (Fig. 3A–C). Exposure to 50 μM NS-1738, PAM-2, or PNU-120596 elicited responses that were 0.86±0.51%, 0.70±0.19%, or 0.38±0.09% (mean±SEM; n=5 oocytes/drug) of the peak response to 1 mM GABA+50 μM propofol, used as a comparative standard, in the same oocytes (Fig. 3D).

Figure 3. The α7-PAMs weakly directly activate the α1β2γ2L GABA_AR.

Representative GABA_AR currents induced by 50 μM NS-1738 (A), PAM-2 (B), or PNU-120596 (C), and a reference response to 1 mM GABA+50 μM propofol (Prop) from the same set of oocytes. (D) Summary of direct activation. The plot gives peak responses to each α7-PAM at 50 μM normalized to the response to 1 mM GABA+50 μM propofol. The parameter values can be converted to units of P_A through multiplication by 0.01.

The α7-PAMs potentiated responses of the α1β2γ2L receptor to 2-5 μM (P_A=0.05-0.06; EC_6-7) GABA. Application of 50 μM NS-1738, PAM-2, or PNU-120596 potentiated the steady-state response to GABA to 383±38% (n=12), 300 ± 43% (n=9), or 133±11% (n=5) of control, respectively. Sample current traces and summary are given in Fig. 4. Potentiation of GABA-activated receptors combined with weak direct activation indicates that the α7-PAMs do not interact with the transmitter binding site.

Figure 4. The α7-PAMs potentiate submaximal GABA responses in the α1β2γ2L GABA_AR.

(A) Representative GABA (2 μM)-induced currents in the absence and presence of 50 μM NS-1738, PAM-2, or PNU-120596. (B) Summary of potentiation by 50 μM α7-PAM. Potentiation is expressed in % of the control response, calculated as the ratio of the peak response to GABA + α7-PAM to the steady-state response to GABA before the application of α7-PAM.

The concentration-response relationships for α1β2γ2L GABA_AR potentiation by α7-PAMs were established by activating the receptor with a low concentration (P_A=0.05-0.13) of GABA in the absence or presence of 0.3-50 μM NS-1738, 1-100 μM PAM-2, or 1-50 μM PNU-120596. Sample current traces for GABA coapplied with each drug at 1 or 50 μM are given in Fig. 5A. Fitting the potentiation concentration-response curves with the Hill equation gave an EC₅₀ of 13±2 μM (n=6) for NS-1738, and 26±8 μM (n=5) for PAM-2 (Fig. 5B). PNU-120596 is a weak potentiator. While small concentration-dependent effects were consistently observed at all tested concentrations the fits did not converge.

Figure 5. Concentration-dependence of the potentiating activity of α7-PAMs in the α1β2γ2L GABA_AR.

(A) Representative GABA_AR currents in the presence of 2-4 μM GABA in the absence or presence of 1 or 50 μM α7-PAM. (B) Concentration-response curves for NS-1738 (circles), PAM-2 (squares), and PNU-120596 (triangles). The data points give means±SEM from 5-6 cells. The curves are fits to Eq. 3 with fitting results provided in the text. The fits in the presence of PNU-120596 did not converge.

To gain further insight into the properties of the studied α7-PAMs, we analyzed the concentration-response data in the framework of a two-state concerted transition model (Forman, 2012; Steinbach and Akk, 2019). With the number of α7-PAM binding sites constrained to 2, fitting the P_A data to Eq. 3 yielded a K_R,NS-1738 of 14.4±2.7 μM and a c_NS-1738 of 0.334±0.019, and a K_R,PAM-2 of 24.8±10.5 μM and a c_PAM-2 of 0.524±0.030. Thus, NS-1738 has higher affinity and is a more efficacious potentiator than PAM-2, contributing −1.29 kcal/mol (vs −0.76 kcal/mol for PAM-2) free energy change (ΔG) towards stabilizing the active state of the receptor. We note that the estimated ΔG values are independent of the number of imposed binding sites.

From its effect at the highest concentration employed (50 μM), we could estimate the upper limit for the value of c (i.e., lower limit for free energy change) for PNU-120596. From P_{A,50 μM PNU-120596} = 1/(1 + L×c_PNU-120596^N), we estimate a maximal c of 0.804±0.029 (ΔG=−0.26±0.04 kcal/mol) for PNU-120596.

α7-PAMs weakly desensitize GABA-activated α1β2γ2L GABA_ARs

Exposure to type II α7-PAMs reduces α7 nAChR desensitization, which can manifest as enhancement of steady-state current during a long pulse of saturating ACh or other agonists (Andersen et al., 2016; Collins et al., 2011). To determine whether a similar mechanism underlies potentiation of the α1β2γ2L GABA_AR by α7-PAMs, we coapplied 50 μM NS-1738, PAM-2, or PNU-120596 during the steady-state phase of the response to saturating (1 mM) GABA when ~80% of receptors are in the desensitized state. With any of the three α7-PAMs, we observed a reduction, rather than increase, in steady-state current (Fig. 6A). NS-1738 reduced the steady-state response to 44±8%, PAM-2 to 70±4%, and PNU-120596 to 90±2% of control (Fig. 6B).

Figure 6. The α7-PAMs desensitize responses to saturating GABA in the α1β2γ2L GABA_AR.

(A) The receptors were activated with saturating (1 mM) GABA. Once steady-state current level was reached, the cell was exposed to GABA+50 μM NS-1738 or PAM-2. (B) Summary of modulation. Modulation is expressed in % of the control response, calculated as the ratio of the response to GABA+α7-PAM to the steady-state response to GABA before the application of α7-PAM.

Drug competition-based approach to differentiate between the binding sites for α7-PAMs in the α1β2γ2L GABA_AR

Next, we measured receptor potentiation in the presence of multiple α7-PAMs. The goal of these experiments was to determine whether the type I α7-PAM NS-1738, and the type II α7-PAMs PAM-2 and PNU-120596 act on the α1β2γ2L GABA_AR through distinct or same binding sites. This approach, described in detail previously (Shin et al., 2019), is based on comparing observed potentiation in the presence of drug combinations to calculated responses predicted by models in which the combined drugs interact with distinct (Eq. 4, Fig. S2) vs same or overlapping sites (Eq. 5, Fig. S3). In the distinct site model, the receptor can simultaneously bind both drugs, and the drugs act independently and energetically additively. This situation is analogous to combining an orthosteric agonist (e.g., GABA) and an allosteric agonist (e.g., propofol). In the same/overlapping site model, a site can at any given time be occupied by only one drug. This is analogous to combining two orthosteric agonists (e.g., GABA and β-alanine) or two structurally similar allosteric agonists (e.g., the neurosteroids 3α5βP and 5βTHDOC).

In the first set of experiments, we tested the combination of NS-1738 and PAM-2. Oocytes expressing α1β2γ2L GABA_ARs were exposed to 0.5 μM GABA (P_A=0.039±0.013; n=5), followed by applications of GABA+20 μM NS-1738 (P_A=0.153±0.043) and GABA+NS-1738+25 μM PAM-2 (P_A=0.158±0.044). The calculated P_A for the triple combination of GABA+NS-1738+PAM-2 are 0.146±0.043 using the same site model and 0.267±0.065 using the distinct site model. The Akaike weights (w), that express the likelihood that a particular model provides a better description of observations, were 0.9997 for the same site model and 0.0003 for the distinct site model.

We also tested the combination of NS-1738 and PNU-120596. We posited that if the two drugs bind to overlapping sites, then coapplication of the lower-efficacy potentiator PNU-120596 is expected to reduce receptor potentiation by NS-1738 through competitive inhibition. In contrast, the combination of NS-1738 and PNU-120596 acting through distinct sites is expected to increase potentiation observed in the presence of NS-1738 alone. In 5 oocytes, coapplication of 50 μM PNU-120596 reduced the response to low (0.1-0.5 μM) GABA+10 μM NS-1738 from 0.084±0.028 to 0.074±0.024. Given the lack of exact affinity and efficacy parameters for PNU-120596, we are unable to make precise model-based predictions for receptor behavior in the presence of the combination of NS-1738 and PNU-120596. In the distinct site model, using the upper-limit estimate for c_PNU-120596 of 0.804, we calculate predicted P_A for the combination of GABA+NS-1738+PNU-120596 of 0.114, 0.103, and 0.097 with K_R,PNU-120596 constrained to 10, 50, or 100 μM, respectively. Thus, the distinct site model predicts 15-35% potentiation when PNU-120596 is coapplied with GABA+NS-1738. Employing the same site model, we calculate P_A of 0.047, 0.065, and 0.073 with a c_PNU-120596 of 0.804 and K_R,PNU-120596 constrained to 10, 50, or 100 μM, respectively. Thus, qualitatively we infer that NS-1738 and PNU-120596 act through a common site in the α1β2γ2L GABA_AR to account for reduction in current response when PNU-120596 is coapplied with NS-1738. Furthermore, the results of simulations suggest that the K_R of PNU-120596 is ~100 μM.

Drug competition-based approach to differentiate between the binding sites for α7-PAMs, benzodiazepines, neurosteroids, R(+)-etomidate, propofol, and phenobarbital in the α1β2γ2L GABA_AR

To determine if the sites for α7-PAMs overlap with previously identified allosteric binding sites for benzodiazepines, neurosteroids and anesthetics in the GABA_AR, we compared α7-PAM potentiation of binary αβ and ternary αβγ receptors, or measured α1β2γ2L receptor potentiation by combinations of an α7-PAM and the neurosteroid 3α5βP, R(+)-etomidate, propofol, or phenobarbital. The data indicate that the high-affinity benzodiazepine binding site in the α/γ interface in the ECD and the neurosteroid binding site in the β/α interface in the TMD are not involved in the actions of NS-1738 or PAM-2 on the α1β2γ2L receptor (Figs. 7, S5; Supporting Information).

Figure 7. Effects of α7-PAMs on α1β2γ2L GABA_ARs activated by allosteric agonists.

The panels show representative current responses to the steroid pregnanolone (3α5βP), R(+)-etomidate (Eto), phenobarbital (PhB), or propofol (Prop) in the absence and presence of 50 μM NS-1738 (A-B) or PAM-2 (C-D). The summary plots give normalized potentiation by an α7-PAM of receptors activated by GABA (reproduced from Fig. 4) or an allosteric agonist. Asterisks indicate significant differences (Dunnett’s test) between potentiation observed in receptors activated by allosteric agonist vs. GABA.

The sedative-anesthetic R(+)-etomidate binds in the TMD in the β(+)/α(−) interface (Li et al., 2006; Kim et al., 2020). To determine whether α7-PAMs bind to the site through which the anesthetic R(+)-etomidate potentiates the GABA_AR, we compared α7-PAM-induced potentiation of receptors activated by GABA or R(+)-etomidate. Coapplication of 50 μM NS-1738 with 2-4 μM R(+)-etomidate (P_A=0.053±0.015) potentiated the peak response to 125±8% of control (n=5), which is significantly different from the effect of 50 μM NS-1738 on GABA-activated receptors (383% of control; see above). Similarly, 50 μM PAM-2 coapplied with R(+)-etomidate (P_A=0.041±0.013) potentiated the peak response to 140±17% of control (n=6). This is significantly different from its effect on GABA-activated receptors (300% of control; see above). The data are summarized in Fig. 7.

To gain quantitative insight into α7-PAM-induced potentiation of receptors activated by R(+)-etomidate, we compared observed potentiation of receptors activated by R(+)-etomidate with predicted responses calculated using the same site vs distinct site models. The observed P_A in the presence of R(+)-etomidate+NS-1738 was 0.068±0.022, whereas the same site model predicted a P_A of 0.007±0.002, and the distinct site model predicted a P_A of 0.253±0.055. For R(+)-etomidate+PAM-2, the same site and distinct site models predict P_A of 0.008±0.002 and 0.097±0.030, respectively. The observed P_A in the presence of R(+)-etomidate+PAM-2 was 0.050±0.013. Thus, for both modulators, the observed response is intermediate to predictions made by the two models.

The reason for the inability of either model to accurately predict the responses is not immediately clear from these results. One possibility is that while the sites for R(+)-etomidate and α7-PAMs are not overlapping, which would predict inhibition of R(+)-etomidate-activated GABA_ARs by α7-PAMs, they are also not acting independently, possibly due to physical proximity of the ligands in their respective sites or allosteric crosstalk between the sites. An alternative possibility is that α7-PAMs interact with other sites in the receptor, besides the β/α interface that contains the R(+)-etomidate site. Those sites then mediate the residual potentiation in the presence of R(+)-etomidate and NS-1738 or PAM-2. In such a hybrid model, the α7-PAMs compete with R(+)-etomidate at the anesthetic binding sites at the β/α interfaces but potentiate receptor function through distinct site(s).

To test these hypotheses and to probe the involvement of homologous sites in other intersubunit interfaces, we measured α7-PAM-mediated potentiation of receptors activated by phenobarbital (Fig. 7). Phenobarbital binds at the α/β and γ/β interfaces with ~10-fold selectivity over the β/α interface (Chiara et al., 2013; Kim et al., 2020), and the involvement of the α/β and γ/β interfaces in the actions of α7-PAMs is expected to manifest as reduced potentiation of phenobarbital-elicited currents. The application of 50 μM NS-1738 enhanced the response to 1 mM phenobarbital to 431±52% of control (n=5). This is indistinguishable from the effect NS-1738 has on receptors activated by GABA (383% of control; see above). In contrast, 50 μM PAM-2 was without effect on receptors activated by 1 mM phenobarbital (103±6% of control, n=5). The finding that NS-1738 similarly potentiates GABA- and phenobarbital-activated receptors suggests that the α/β and γ/β interfaces, relative to β/α, make smaller contributions to gating by NS-1738. An alternative explanation is that the α/β and γ/β interfaces can simultaneously bind phenobarbital and NS-1738. The observation that phenobarbital-activated receptors are not potentiated by PAM-2 suggests that its effects are predominantly mediated by the α/β and γ/β interfaces where the sites for phenobarbital and PAM-2 overlap each other.

The anesthetic propofol binds to the β/α interfaces that accommodate R(+)-etomidate and to the α/β and γ/β interfaces that bind phenobarbital (Chiara et al., 2013; Jayakar et al., 2014). Currents elicited by 10 μM propofol (P_A=0.06-0.07) were potentiated to 145±11% (n=5) or 123±9% of control (n=6) in the presence of 50 μM NS-1738 or PAM-2, respectively (Fig. 7). Both are significantly different from the effects on GABA-elicited currents, thereby confirming the involvement of the interfacial anesthetic binding sites in the actions of α7-PAMs.

Mutational approach to identify the binding sites for α7-PAMs in the α1β2γ2L GABA_AR

Next, we tested the effects of mutations to several previously identified allosteric binding sites on the actions of α7-PAMs. The goal was to confirm the observations in functional competition experiments that indicated the involvement of anesthetic binding sites but lack of involvement of the neurosteroid binding site in the actions of NS-1738 and PAM-2. The underlying assumption in these experiments is that a change in the magnitude of potentiation in a mutant receptor indicates involvement of the specific site or interface. First, we examined the effect of the α1(Q241L) mutation to the neurosteroid binding site (Akk et al., 2008; Germann et al., 2021; Hosie et al., 2006) on potentiation by α7-PAMs. The data indicate that the α1(Q241L)β2γ2L receptor is potentiated similarly to the wild-type receptor (Supporting Information), thus supporting the idea that the neurosteroid site is not involved in the actions of α7-PAMs.

The binding site for R(+)-etomidate in the β/α interface is lined by the N265 (TM2) and M286 (TM3) residues from the β(+) side, and the L232 and M236 residues (both in TM1) from the α(−) side (Chiara et al., 2012; Li et al., 2006). Previous mutational studies and substituted cysteine modification experiments have indicated the involvement of these residues in the actions of R(+)-etomidate on the GABA_AR (Nourmahnad et al., 2016; Reynolds et al., 2003; Stewart et al., 2008). Our results show that receptor potentiation by NS-1738 is affected by the β2(M286W) and α1(M236C) mutations, but not by the β2(N265M) or α1(L232C) mutations. Potentiation by PAM-2 was reduced in β2(M286W) and β2(N265M) mutants at the β(+) side, and in α1(L232C) and α1(M236C) mutants at the α(−) side. This confirms the involvement of the anesthetic binding site at the β/α interface in the actions of NS-1738 and PAM-2. The data are summarized in Fig. 8A and Table 1.

Figure 8. Effects of selected mutations on the potentiating activity of α7-PAMs.

The wild-type and mutant receptors were activated by a low concentration of GABA (P_A=0.04-0.14) and exposed to 50 μM NS-1738 or PAM-2. The drug application protocol is as shown earlier in Figure 4. The plots give normalized potentiation of each α7-PAM at 50 μM. Potentiation is expressed in % of the control response, calculated as the ratio of the peak response to GABA+α7-PAM to the steady-state response to GABA before the application of α7-PAM. The mutations were made to the β/α interface (A), α/β and γ/β interfaces (B), or the α/γ interface (C). The data for α1β2γ2L in (A) are reproduced from Figure 4. The effects of the β3(M227C) and β3(L231C) mutations in (B) were compared to α1β3γ2L data shown in the same panel. All other mutants were compared to the α1β2γ2L data presented in (A). The dotted lines show the levels of responses from α1β2γ2L or α1β3γ2L receptor for comparison. The dashed lines show level of no effect (normalized response=100%). Asterisks indicate significant differences (Dunnett’s test) between potentiation observed in the mutant vs. wild-type receptor.

Table 1.

Summary of potentiation of wild-type and mutant GABA_ARs by NS-1738 and PAM-2.

Receptor	Potentiation by NS-1738 (% of control)	Potentiation by PAM-2 (% of control)
α1β2γ2L	383 ± 38%	300 ± 43%
α1(Q241L)β2γ2L	491 ± 45%	382 ± 29%
α1β2(N265M)γ2L	426 ± 61%	124 ± 6%*
α1β2(M286W)γ2L	613 ±91%*	181 ± 12%*
α1(L232C)β2γ2L	357 ± 55%	197 ± 8%*
α1(M236C)β2γ2L	181 ± 15%*	176 ± 11%*
α1(A291W)β2γ2L	199 ± 22%*	130 ± 7%*
α1β2γ2L(S280C)	344 ± 60%	233 ± 16%
α1β2γ2L(I242C)	376 ±35%	189 ± 11%*
α1β2γ2L(L246C)	262 ± 25%	147 ± 7%*
α1β3γ2L	301 ± 12%	176 ± 12%*
α1β3(M227C)γ2L	322 ± 40%	163 ± 12%
α1β3(L231C)γ2L	217 ± 36%	153 ± 11%

The receptors were activated by low GABA producing a response with P_A < 0.10. The table shows potentiation of the control response by 50 μM NS-1738 or PAM-2. Potentiation was calculated as the ratio of the peak response to a low concentration of GABA + α7-PAM to the response to GABA alone before application of α7-PAM, and is expressed as % of control (100% = no effect). The data are shown as mean ± SEM from at least 5 oocytes per experiment. Potentiation of all β2-containing mutant receptors and the α1β3γ2L receptor was compared to potentiation observed in the α1β2γ2L receptor. Potentiation of β3-containing mutant receptors was compared to potentiation observed in the α1β3γ2L receptor. Dunnett’s test;

^*,P < 0.05.

The involvement of the α/β and γ/β interfaces was investigated by mutations to the α1-A291 and γ2-S280 residues at the α(+) and γ(+) sides, respectively, and mutations to β3-M227 and β3-L231 at the β(−) side. These residues line the binding sites for phenobarbital (Kim et al., 2020). The α1(A291W) mutation reduced potentiation by NS-1738 and PAM-2. While this supports the idea that α7-PAMs act through the α/β interface, the caveat is that the α1(A291W) mutation also modifies the α/γ interface. The γ2(S280C) mutation in the γ/β interface was without effect on potentiation by either drug. At the β(−) side, potentiation by α7-PAMs was not affected by β3(M227C) or β3(L231C). A summary is provided in Fig. 8B and Table 1.

As noted above, reduced potentiation by α7-PAMs in the α1(A291W) mutant may be mediated by one or both of α/β and α/γ interfaces. The α/γ interface is not involved in the actions of GABAergic anesthetics (Nourmahnad et al., 2016). To test whether the α/γ interface contributes to the actions of α7-PAMs, we measured the effects of γ(I242C) and γ(L246C) mutations on potentiation. These mutations, at the γ(−) side, are unique to the α/γ interface. Neither mutation affected receptor sensitivity to NS-1738, however, both mutations reduced potentiation by PAM-2 (Fig. 8C; Table 1).

DISCUSSION AND CONCLUSIONS

Previous behavioral work has revealed that some α7-PAMs have antinociceptive and anxiolytic effects in rodent behavioral experiments. Hypothesizing that these effects involve the GABA_A receptor, we set out to investigate the modulatory activity and location of the putative binding sites of α7-PAMs in the GABA_AR. The main finding of this study is that selected type I (NS-1738) and type II α7-PAMs (PAM-2 and PNU-120596) potentiate the GABA_AR by stabilizing the active state through interactions with anesthetic binding sites in the TMD in the intersubunit interfaces.

Analysis based on a co-agonist concerted transition model indicated that NS-1738 and PAM-2 contribute −1.29 kcal/mol or −0.76 kcal/mol free energy change, respectively, towards stabilization of the active state. This is similar to free energy change contributed by neuroactive steroids or the benzodiazepine diazepam (Cao et al., 2018; Germann et al., 2019b; Shin et al., 2019). PNU-120596 is a weak potentiator (ΔG=−0.26 kcal/mol) of the GABA_A receptor. The observed relative efficacies of α7-PAMs in the GABA_A receptor (NS-1738>PAM-2>>PNU-120596) are different from those in the α7 nAChR where PNU-120596 is more efficacious than PAM-2 (Andersen et al., 2016). The in vitro EC₅₀s for NS-1738 are 3-13 μM in the nicotinic α7 receptor and 13 μM in the α1β2γ2L GABA_AR, and the EC₅₀s for PAM-2 are 5-12 μM in α7 and 26 μM in α1β2γ2L (Arias et al., 2020; Arias et al., 2011; Timmermann et al., 2007; Young et al., 2008). Thus the apparent potencies of NS-1738 and PAM-2 are only marginally lower in the GABA_AR than in the α7 nicotinic receptor indicating that the doses used to produce α7 nicotinic receptor-mediated behavioral effects are likely to modulate α1β2γ2L GABA_A receptor activity. We also measured the effects of α7-PAMs in α1β3γ2L, α1β2, α1β3, and α4β2δ GABA_A receptors (Figs. 4, S4–S5, Supporting Information). The magnitude of potentiation by NS-1738 and PAM-2 is similar in the binary αβ and ternary αβγ receptors (Figs. 4, S5, Supporting Information), but the α4β2δ receptor is inhibited by NS-1738 and PAM-2 during long drug applications (Fig. S4, Supporting Information). We therefore propose that any positive GABAergic effect of α7-PAMs is mediated by binary αβ and ternary αβγ GABA_ARs.

The mechanisms of potentiation by α7-PAMs differ in the α7 nAChR and the α1β2γ2L GABA_AR. In the nAChR, exposure to type II, and to a smaller degree type I, α7-PAMs recovers receptors driven to the desensitized state during an extended application of agonist (Andersen et al., 2016; Collins et al., 2011). In the GABA_AR, α7-PAMs shift the equilibrium towards the active state in the presence of low concentrations of GABA and slightly reduce GABA_AR steady-state current elicited by saturating GABA. The potentiating and inhibitory effects of α7-PAMs observed at low and high levels of activity, respectively, are likely independent effects. Here, we focused on the potentiating effects of α7-PAMs. We note that several potentiators of the GABA_AR exhibit inhibitory effects at high concentrations (Adodra and Hales, 1995; Akk and Steinbach, 2000; Pistis et al., 1997), and that NS-1738 and PAM-2 inhibit the heteromeric α4β2 and α3β4 nicotinic receptors at high μM concentrations (Arias et al., 2011; Timmermann et al., 2007).

In the α7 nAChR, type I and type II α7-PAMs interact with distinct sites ((Andersen et al., 2016; Bertrand et al., 2008; Targowska-Duda et al., 2019b) but see (Collins et al., 2011)). Our functional analysis indicates that in the α1β2γ2L GABA_AR, NS-1738, PAM-2, and PNU-120596 interact with the same set of sites. Receptor subtype selectivity and drug competition and mutational assays indicated lack of involvement of high-affinity benzodiazepine and neurosteroid binding sites in the actions of α7-PAMs. The involvement of anesthetic binding sites in the intersubunit interfaces in the transmembrane domain was suggested by drug competition and mutational analysis. The β/α interface was implicated by the findings that α7-PAMs only weakly potentiate receptors activated by R(+)-etomidate and that mutations to β2-N265 and β2-M286 at the β(+) side, and α1-L232 and α1-M236 at the α(−) side affect potentiation. Mutations to these residues have been previously shown to modify the actions of R(+)-etomidate, likely through changes in direct interactions with the ligand (Belelli et al., 1997; Li et al., 2006; Siegwart et al., 2002; Stewart et al., 2008). Qualitatively, the extent of loss of potentiation was greater for NS-1738, suggesting that it, compared to PAM-2, makes a relatively stronger contribution through the β/α interface than through other site(s). Not all mutations similarly affected potentiation by NS-1738 and PAM-2. For example, β2(N265M) and α1(L232C) reduced potentiation by PAM-2 but were without effect on potentiation by NS-1738. Interestingly, β2(M286W) enhanced potentiation by NS-1738 but reduced potentiation by PAM-2. We interpret the difference in sensitivity to mutations in the β/α interface as NS-1738 and PAM-2 interacting with different key residues in the β/α interface, as supported by docking data (Fig. S6, Supporting Information). In mutational analysis, we have assumed that the mutations act locally. This may not be so, at least in all cases. For example, the β3(N265M) and β3(M286W) mutations in the β/α interface reduce potentiation by the α/β and γ/β selective R-5-allyl-1-methyl-5-(m-trifluoromethyl-diazirinylphenyl) barbituric acid (R-mTFD-MPAB), suggesting allosteric crosstalk between anesthetic binding sites at different interfaces (Szabo et al., 2019).

We tested the involvement of the α/β and γ/β interfaces by measuring α7-PAM-induced potentiation of receptors activated by phenobarbital. Phenobarbital binds to the α/β and γ/β interfaces with high affinity (Chiara et al., 2013; Kim et al., 2020). NS-1738 potentiated phenobarbital-activated receptors similarly to GABA-activated receptors. The strict interpretation of this finding is that NS-1738 and phenobarbital act independently and energetically additively. We therefore propose that the α/β and γ/β interfaces are minimally involved in the actions of NS-1738. This is supported by the finding that NS-1738 shows strongly reduced potentiation when the receptor is activated by the β/α specific R(+)-etomidate. An alternative explanation is that the α/β and γ/β interfaces can simultaneously accommodate NS-1738 and phenobarbital. In contrast, PAM-2 exhibited no modulatory effect on receptors activated by phenobarbital. This suggests that a major site of action of PAM-2 is at the α/β and γ/β interfaces where it overlaps with the phenobarbital site. The α/γ interface has been termed “orphan” because of lack of protection of cysteine residues in the interface by anesthetics (Nourmahnad et al., 2016). The γ(I242C) and γ(L246C) mutations in the α/γ interface were without effect on potentiation by NS-1738 but reduced potentiation by PAM-2, suggesting that the α/γ interface may be involved in the actions of PAM-2.

In conclusion, we have shown here that the α7-PAMs NS-1738 and PAM-2 potentiate the binary αβ and ternary αβγ GABA_A receptors at micromolar concentrations with gating efficacies comparable to those of neurosteroids or the benzodiazepine diazepam. The data indicate that NS-1738 and PAM-2 act by binding to the classic anesthetic binding sites in the intersubunit interfaces in the TMD. The in vitro sensitivities to NS-1738 and PAM-2 are similar in the nicotinic and GABA_A receptors; accordingly, this implies that the doses used to generate nicotinic receptor-mediated behavioral effects are likely to elicit GABA_AR-mediated effects.

What is already known:

• Positive allosteric modulators of the α7 nAChR (α7-PAMs) produce behavioral effects including antinociception and anxiolysis.

What this study adds:

• α7-PAMs potentiate αβ and αβγ GABA_ARs with apparent affinities similar to those in α7 nAChR.

Clinical significance:

• Doses of α7-PAMs that produce α7 nAChR-mediated behavioral effects in vivo also modulate GABA_AR function.

Abbreviations:

GABA_AR

GABA type A receptor

nAChR

nicotinic acetylcholine receptor

PAM

positive allosteric modulator

PAM-2

(E)-3-(furan-2-yl)-N-(p-tolyl)-acrylamide

NS-1738

N-(5-Cl-2-hydroxyphenyl)-N’-[2-Cl-5-(trifluoromethyl)phenyl]-urea

PNU-120596

N-(5-Cl-2,4-dimethoxyphenyl)-N’-(5-methyl-3-isoxazolyl)-urea

R(+)-etomidate

R-1-(1-ethylphenyl)imidazole-5-ethyl ester

Gzn-OG

gabazine-oregon green

Data availability statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.