Selective Modulation of α5 GABAA Receptors Exacerbates Aberrant Inhibition at Key Hippocampal Neuronal Circuits in APP Mouse Model of Alzheimer's Disease

Abstract

Selective negative allosteric modulators (NAMs), targeting α5 subunit-containing GABA_A receptors (GABA_ARs) as potential therapeutic targets for disorders associated with cognitive deficits, including Alzheimer's disease (AD), continually fail clinical trials. We investigated whether this was due to the change in the expression of α5 GABA_ARs, consequently altering synaptic function during AD pathogenesis. Using medicinal chemistry and computational modeling, we developed aqueous soluble hybrids of 6,6-dimethyl-3-(2-hydroxyethyl) thio-1-(thiazol-2-yl)-6,7-dihydro-2-benzothiophene-4(5H)-one, that demonstrated selective binding and high negative allosteric modulation, specifically for the α5 GABA_AR subtypes in constructed HEK293 stable cell-lines. Using a knock-in mouse model of AD (APP ^NL-F/NL-F), which expresses a mutant form of human amyloid-β (Aβ), we performed immunofluorescence studies combined with electrophysiological whole-cell recordings to investigate the effects of our key molecule, α5-SOP002 in the hippocampal CA1 region. In aged APP ^NL-F/NL-F mice, selective preservation of α5 GABA_ARs was observed in, calretinin- (CR), cholecystokinin- (CCK), somatostatin- (SST) expressing interneurons, and pyramidal cells. Previously, we reported that CR dis-inhibitory interneurons, specialized in regulating other interneurons displayed abnormally high levels of synaptic inhibition in the APP ^NL-F/NL-F mouse model, here we show that this excessive inhibition was "normalized" to control values with bath-applied α5-SOP002 (1 μM). However, α5-SOP002, further impaired inhibition onto CCK and pyramidal cells that were already largely compromised by exhibiting a deficit of inhibition in the AD model. In summary, using a multi-disciplinary approach, we show that exposure to α5 GABA_AR NAMs may further compromise aberrant synapses in AD. We, therefore, suggest that the α5 GABA_AR is not a suitable therapeutic target for the treatment of AD or other cognitive deficits due to the widespread neuronal-networks that use α5 GABA_ARs.

Keywords: Alzheimer’s disease; GABAA receptors; hippocampus; interneurons; synaptic.

Figure 1

Developing negative allosteric modulator (NAM), α5-SOP002. (A–C) Optimization of α5IA to α5-SOP002. (D) Detailed interactions of α5-SOP002 at the GABA_AR binding site located at the interface between subunit α5 (blue) and γ2 (brown). (E) Surface representation of SH-AI-SOP002 (red) interacting with the α5 GABA_AR at the α5 (blue) and γ2 (brown) subunits’ interface. (F) The upper view of the α5 GABA_A subtype is represented by ribbons. The red arrow points at α5-SOP002. Subunits α5 are shown in blue, β3 in green, and γ2 in brown. (G) Surface and (H) ribbon representation of the α5 GABA_AR.

Figure 2

α5-SOP002 selectively targets α5 subunits of GABA_ARs. Whole-cell recordings in α5β2γ2-, α1β2γ2-, and α2β2γ2-HEK293 cells. HEK293 cells stably expressing α5β2γ2- (A), α1β2γ2- (B), or α2β2γ2-GABA_ARs (C). Immunofluorescent imaging with a 40× oil immersion objective lens shows cell surface expression of α5, α1 or α2- (cyan), β2- (red), and γ2-GABA_AR subunits (green). (A–C) also show all the three channels merged showing α-, β2-, and γ2-GABA_AR subunit co-localization at the cell surface (white) along with the differential interference contrast microscopy (DIC) image of the cells. The scale bar represents 10 μm. All three stable cell lines responded to 10 μM puff-applied GABA (D–F) in control extracellular solution (black traces), an extracellular solution containing 1 μM α5-SOP002 (red traces), and subsequent bath application of diazepam (blue traces) at a holding membrane potential of −60 mV. The corresponding plots for α5β2γ2-HEK293 cells (G–I) show the changes in voltage changes in response to 10 μM GABA puffed locally, in the presence of bath-applied α5-SOP002, and, subsequent addition of diazepam. Only the α5β2γ2-HEK293 cells showed an inverse agonist effect (response to GABA) of α5-SOP002. All three cell lines, however, showed an enhancement of response to GABA in the presence of diazepam. Statistically significant data are shown with *P < 0.05 and **P < 0.01.

Figure 3

α5-SOP002 improved memory in healthy rats performing the eight-arm radial arm maze (RAM) test, consistent with the NAM effect shown by in vitro electrophysiological recordings. (A–C) Analysis of RAM test performed in healthy rats treated with “sham,” α5-SOP002- and L-655, 708. (A) Illustration of the variability of time taken to complete the memory task between all groups during 7 days (training + test phases). (B) Bar graphs compare the time taken to complete the tasks during training day 4 (defined as the information phase; gray) and day 7 (defined as test phase; green) for all groups tested. The time to complete the task during the test phase after administration of either; α5-SOP002 and L-655, 708, was reduced compared to control, suggesting potentiation of spatial memory recall. (C) Bar graphs representing the difference in mean time taken to complete the task between the information and test phase of each group. The a5-SOP002-treated rats had a statistically significant difference while both α5-SOP002- and L-655, 708-treated groups showed a bigger difference compared to the sham group. Results are expressed as mean ± standard error of the mean (SEM; black; sham group, n = 9, orange; L-6, 55 n = 4 and red; a5-SOP002 n = 14; one-way ANOVA, *P < 0.05, **P < 0.01, ****P < 0.0001). (D) Representative images of the anatomy of dendrite targeting, presynaptic CCK schaffer collateral-associated (SCA) cells synaptically connected to a postsynaptic CCK cell (AMCA shows the biocytin labeling during electrophysiological recordings). (E,F) Paired recording, trains of presynaptic action potentials elicited unitary inhibitory postsynaptic potentials (IPSPs), that were reduced by both NAMs selective for a5 GABA_A receptors, L-655, 708 and α5SOP002.

Figure 4

Expression of α5 subunit-containing GABA_ARs in CA1. (A–D) Confocal microscopy Z-stacks at 63× magnification showing α5 subunit-containing GABA_AR expression (red, Alexa 488 or Alexa 568 label) on pyramidal neurons (CaMKII- α, green, FITC), CR interneurons (green, Alexa 488), SST interneurons (green, Texas Red), and CCK interneurons (green, Texas Red) in wild-type (i) and APP^{NL−F/NL−F} animals (ii). Panels show individual channels and merged image with the nuclear stain DAPI (blue). Representative cell somata are outlined with white circles. White arrows indicate dendritic co-localization of α5. (E) Representative image labeled with α5 (red) and DAPI (blue) taken at 20× magnification in the CA1 of APP^{NL−F/NL−F} to exemplify the region of data acquisition, arrows indicate the location of the sub-types of cells imaged and analyzed. Layers are labeled: alveus (A), stratum oriens (SO), stratum pyramidale (SP), stratum radiatum (SR), stratum lacunosum moleculare (SLM). (F) Analysis of α5 subunit-containing GABA_AR expression on the soma of the four sub-types of neurons investigated. Each data point represents an average value (from five cells) analyzed form individual animals at 12–18 months of age (n = 5–7 mice studies per cohort). (G) Analysis of α5 subunit-containing GABA_AR expression on the dendrites of CR cells, SST cells, and pyramidal neurons (n = 3 mice per genotype with visible proximal dendrites analyzed for five cells per animal). (F,G) Results are expressed as a scatter plot ± (SEM; results not significant, P > 0.05), of Pearson correlation coefficient as a measure of co-localization, after application of Fisher’s transformation. Data analyzed with a one-way ANOVA and post hoc Tukey test.

Figure 5

Calretinin (CR)-expressing interneurons are functionally restored by NAM of α5 subunit-containing GABA_ARs in APP^{NL−F/NL−F} mice. (A,B) Whole-cell current-clamp recordings of spontaneous inhibitory/excitatory postsynaptic potentials (sIPSPs and sEPSPs) recorded in CR cells in CA1 of 12-month-old wild-type and APP^{NL−F/NL−F} mice, at membrane potentials of −60 mV in control conditions, and after bath-application of α5-SOP002 (red traces). The squares indicate where synaptic events have been enlarged and shown in the inserts. *Indicate, an usually high sIPSPs recorded in the AD model. (C,D) Bar graphs show the average sIPSP and sEPSP amplitude and frequency at −60 mV in CR cells recorded in wild-type mice and the APP^{NL−F/NL−F} mouse model. These data suggest a significantly enhanced amplitude and frequency of inhibition in the AD model, which was “normalized” to control values after bath- application of α5-SOP002. **P < 0.01, Data analyzed with a two-way ANOVA and post hoc Tukey’s test. (E) Paired recording obtained between two putative CR cells recorded in SR of CA1 in the AD model. The unitary IPSPs were not sensitive to zolpidem, reduced by α5-SOP002, and then enhanced by subsequent addition of diazepam, indicating α5 pharmacology. (F) Line graphs show the average unitary IPSP amplitude and width at half amplitude change for each paired recording between two CR cells, in control, and after bath-application of zolpidem, α5-SOP002 and diazepam, recorded at −55 mV in APP^{NL−F/NL−F} mouse model. *P < 0.05, **P < 0.01. Data analyzed with a one-way ANOVA and post hoc Tukey test. Blue (*) are representative traces that have been enlarged in the inserts.

Figure 6

CCK interneurons and pyramidal cells are further compromised by NAM of α5 subunit-containing GABA_ARs in APP^{NL−F/NL−F} mice. (A–D) Whole-cell current-clamp recordings illustrating sIPSPs and sEPSPs recorded in CCK-SCA cells (A,B) and pyramidal cells (C,D) in CA1 of 12-month-old wild-type and APP^{NL−F/NL−F} mice, recorded at a membrane potential of −60 mV in control conditions and after bath-application of α5-SOP002. Bath-application of the α5-SOP002 resulted in a reduction in sIPSP amplitude and frequency, but also increased membrane excitation in both cell types, thus further increasing the aberrant hyperexcitability in the AD model. (E–H) Bar graphs show the overall pharmacological change after applying α5-SOP002 in CCK-SCA and pyramidal cells recorded from wild-type and APP^{NL−F/NL−F} mice at 10–12 months. **P < 0.01. Data analyzed with a two-way ANOVA and post hoc Tukey’s test, see Table 1 for details.