Allopregnanolone Mediates Affective Switching Through Modulation of Oscillatory States in the Basolateral Amygdala

Abstract

Background: Brexanolone (allopregnanolone) was recently approved by the Food and Drug Administration for the treatment of postpartum depression, demonstrating long-lasting antidepressant effects. Despite our understanding of the mechanism of action of neurosteroids as positive allosteric modulators of GABA_A (gamma-aminobutyric acid A) receptors, we still do not fully understand how allopregnanolone exerts persistent antidepressant effects.

Methods: We used electroencephalogram recordings in rats and humans along with local field potential, functional magnetic resonance imaging, and behavioral tests in mice to assess the impact of neurosteroids on network states in brain regions implicated in mood and used optogenetic manipulations to directly examine their relationship to behavioral states.

Results: We demonstrated that allopregnanolone and synthetic neuroactive steroid analogs with molecular pharmacology similar to allopregnanolone (SGE-516 [tool compound] and zuranolone [SAGE-217, investigational compound]) modulate oscillations across species. We further demonstrated a critical role for interneurons in generating oscillations in the basolateral amygdala (BLA) and a role for δ-containing GABA_A receptors in mediating the ability of neurosteroids to modulate network and behavioral states. Allopregnanolone in the BLA enhances BLA high theta oscillations (6-12 Hz) through δ-containing GABA_A receptors, a mechanism distinct from other GABA_A positive allosteric modulators, such as benzodiazepines, and alters behavioral states. Treatment with the allopregnanolone analog SGE-516 protects mice from chronic stress-induced disruption of network and behavioral states, which is correlated with the modulation of theta oscillations in the BLA. Optogenetic manipulation of the network state influences the behavioral state after chronic unpredictable stress.

Conclusions: Our findings demonstrate a novel molecular and cellular mechanism mediating the well-established anxiolytic and antidepressant effects of neuroactive steroids.

Keywords: Basolateral amygdala; GABA; Interneurons; Neurosteroids; Oscillations; Stress.

Figure 1.. Neuroactive steroids altered brain network dynamics across species. A-C, humans; D-F, rats; G-I, mice.

A, Heatmaps organized by frequency band showing regions where percentage power change in Human EEG during SAGE-217 treatment (55mg) is elevated above vehicle. B, Power spectral mean (± SEM) difference between SAGE-217 and vehicle (averaged from frontal electrodes F3, Fz, and F4). C, Mean (± SEM) change in power across frequency bands (n=7 human participants). D, Schematic for cortical EEG recordings in awake rats. E, Left: Power spectral mean difference between treatment (SAGE-217) and vehicle, Right: Mean (± SEM) change in power across frequency bands (n_SAGE-217=20 rats). F, Same as E, for SGE-516 (n_SGE-516=27 rats). G, Schematic for BLA LFP recordings in awake mice. H, Power spectral mean (± SEM) difference between SGE-516 (5 mg/Kg) and saline IP applications in Wt and Gabrd^−/− mice. I, Mean (± SEM) change in power across frequency bands (n_Wt=6 mice, n_Gabrd−_/−=5 mice). Power for vehicle and treatment groups in power spectral densities and summary plots was calculated as the percentage change from baseline. Shaded regions and error bars represent SEM.

Figure 2.. Acute IP application of allopregnanolone altered BLA network oscillations in freely moving mice partly through GABA_AR δ subunit-containing receptors.

A, Schematic for LFP recordings in awake mice. B, Representative spectrogram of BLA oscillations during IP application of vehicle and allo (10 mg/Kg). Normalized power^6−12Hz is higher during acute application of allo compared to vehicle treatment. C, Power area^6−12Hz normalized to baseline during vehicle and allo acute treatment across time D, Average dot plot; light color lines indicate individual experiments, dark lines the average and error bars the SEM. E, Normalized power area difference between acute allo and vehicle treatment across multiple frequency bands. F, High(green)/Low(pink) theta power ratio in Wt and Gabrd^−/− mice. n =13 mice, n - -=5 mice.

Figure 3.. Allopregnanolone enhanced tonic currents in PV⁺ interneurons but not in principal cells of the BLA.

A, PV⁺ interneurons highly express GABA_AR δ subunit-containing receptors and immunoreactivity for these receptors is absent in PV-Gabrd^−/− mice. B, Representative traces from intracellular recordings in principal and PV⁺ neurons during 100 nM allo application and block by gabazine. C, PV-Gabrd^−/− mice displayed a significant reduction in δ immunoreactive cells in the basolateral amygdala when compared with controls (Wt: n=10 slices (3 mice), PV-Gabrd^−/−: n=11 slices (4 mice)). D, Tonic current in principal cells (pyr) and PV⁺ interneurons (PV) during baseline and 100 nM allo application. E, Percentage potentiation of tonic current between allo and baseline application; n_PV=9 cells, n_Pyr=10 cells.

Figure 4:. Isolated mouse BLA networks ex-vivo generated brain oscillations in the gamma and theta band-range.

A, Schematic illustration of LFP setup for recording of ex-vivo gamma (γ) oscillations in mouse BLA slices; COR: cortex, BLA: basolateral amygdala, CEA: central amygdala. All experiments were performed in the presence of High K⁺/KA solution. B, top: Representative gamma oscillation LFP trace from BLA and bottom: Wavelet transformation. C, Peak frequency box plot of BLA gamma oscillations (n=42 slices). D-E, Gyki 53655 application to BLA gamma oscillations. D, Representative LFP traces during baseline (black) and Gyki 53655 application (blue). E, Normalized gamma peak power; dots represent mean and shaded region represents SEM (n=7 slices; 6 at 20 μM, 1 at 10 μM). F-G, Gabazine application to BLA gamma oscillations. F, Representative LFP traces during baseline (black) and gabazine application (orange). G, Normalized gamma (top) and theta (bottom) peak power change; dots represent mean and shaded region represents SEM (n=9 slices at 10 μM).

Figure 5.. PV⁺ interneurons synchronized and controlled the BLA network.

A, Schematic of viral infusion of AAV-DIO-ChR2-mCherry in mouse BLA. B, Representative fluorescence images of ChR2-mCherry expression in BLA; CEA: Central Amygdala, BLA: basolateral amygdala, PIR: Piriform Cortex; ENT: Entorhinal Cortex. C-D, ex-vivo aCSF (n=8 slices; light intensity=10mW), E-F, ex-vivo High K⁺/KA (n=12 slices; light intensity=10mW), G-H, in-vivo awake (n=5; light intensity=45–55mW). C, E, G: left: schematic illustration of LFP-opto setup for ChR2 experiments; right-top: Representative gamma oscillation LFP trace from BLA where orange squares represent zoomed traces and bottom: Wavelet transformation. D, F, H, LFP power ratio between pulsed stimulation and baseline periods where power was measured at each stimulation frequency.

Figure 6.. Allopregnanolone promoted anxiolysis in control but not in Gabrd^−/− mice.

Schematics for behavioral tests and associated phenotypes in Wt mice infused with vehicle or 5 µg (2.5 µg/µl -inject) Allo: A, open field; D, light dark box; G, elevated plus maze, J, Tail suspension test. Time that mice spend in B, center(Wt: 11; 10, Gabrd^−/−: 9; 11), E, lit area(Wt: 11; 13, Gabrd^−/−: 9; 11), H, open arms(Wt: 11; 12, Gabrd^−/−: 9; 11), or K, immobile (Wt: 12; 11, Gabrd^−/−: 8; 10). Number of Entries in C, center(Wt: 11; 11, Gabrd^−/−: 9; 11); F, lit area(Wt: 10; 13, Gabrd^−/−: 9; 11), I, open arms(Wt: 11; 12, Gabrd^−/−: 9; 11). L, Summary of average absolute time difference in allo vs vehicle infused conditions across behavioral tests. Error bars represent SEM. Brackets and stars represent unpaired t-tests between vehicle vs allo treated groups. The n’s for each behavioral test and mouse strain are shown

Figure 7.. Synthetic Allopregnanolone analog SGE-516 prevents behavioral deficits induced by chronic unpredictable stress (CUS) in mice.

A-D, SGE-516 application during CUS. A, Timeline of CUS paradigm with LFP and behavioral sessions. B, Left: Time spent immobile in TST (n_control=10, n_CUS=11, n_CUS+SGE-516=12) and Right: % sucrose preference in control/no CUS (white), CUS (black), and CUS+SGE-516 (orange) mice (n_control=15, n_CUS=10, n_CUS+SGE-516=8). C, Time spent immobile in TST plotted against high theta (6–12Hz) power change pre- and post-CUS, in CUS (black, r=0.53, n=10, p=0.11, Pearson correlation) and CUS+SGE-516 (orange, r=−0.62, n=11, p=0.04, Pearson correlation). D, Time spent immobile in TST plotted against theta ratio change (6–12/2–5Hz) in both CUS (black, n=6) and CUS+SGE-516 (orange, n=6) from mice with the most extreme changes (6 with largest and 6 with smallest theta ratio change) (r=−0.84, n=12, p=0.0006, Pearson correlation). E-H, SGE-516 application after CUS. E, Timeline of CUS paradigm with LFP and behavioral sessions. F, Time spent immobile in TST before and after SGE-516 application (n=7). G, Time spent immobile in TST plotted against theta ratio (6–12/2–5HzHz) change pre- and post-CUS, in CUS (black, r=−0.42, n=6, p=0.40, Pearson correlation) and CUS+SGE-516 (orange, r=−0.74, n=8 mice, p=0.04, Pearson correlation) mice. H, TST response difference prior to and following SGE-516 application plotted against theta ratio (6–12/2–5Hz) change pre- and post-CUS (r=−0.86, n=8, p=0.01, Pearson correlation). Power changes in correlation plots were calculated as the percent difference between 4 weeks post-CUS and 24 hours prior to CUS; r-values represent the Pearson correlation coefficient. I-J, BLA entrainment through PV-ChR2 photo-stimulation at 4 and 8 Hz after CUS. I, Timeline of CUS paradigm with opto-behavioral sessions. J, Time spent immobile in TST across different stimulation patterns (n=8).

Figure 8.. Synthetic Allopregnanolone analog SGE-516 prevents altered functional connectivity induced by chronic unpredictable stress (CUS) in C57BL/6J mice.

A, Timeline of CUS paradigm with fMRI sessions. B, Changes of baseline connections (present in baseline) during CUS and CUS+SGE-516 treatment; number of connections is proportional to line width; total of 299 connections present at baseline. C-D, Resting state functional connectivity; type-of-connection legend: black: downregulated, blue: upregulated, green: reversed by SGE-516, pink: potentiated by SGE-516. C, Representative brain volumes of resting state functional connectivity in CUS and CUS+SGE-516 mice. Arrows indicate directionality of connections from the indicated seed regions; arrow color reflects type of connection change based on type-of-connection legend; AC: Auditory Cortex, BLA: Basal Lateral Amygdaloid Area, CA1: Cornu Ammonis 1, CA3: Cornu Ammonis 3, CEA: Central Nucleus of the Amygdala, COA: Cortical Amygdaloid Area, LA: Lateral Amygdaloid Area, LC: Locus Coeruleus, MEA: Medial Amygdaloid Area. D, Connectome map indicating significant changes in BOLD signal connectivity, Outer circle lines represent brain areas of the respective color as indicated in the bottom legend. Mice: ncontrol=5, nCUS=5, nCUS+SGE-516=5.