If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Department of Neuroscience, Program of Neuroscience, Graduate School of Biomedical Sciences, Tufts University School of Medicine, Boston, Massachusetts
Department of Neuroscience, Program of Neuroscience, Graduate School of Biomedical Sciences, Tufts University School of Medicine, Boston, Massachusetts
Department of Neuroscience, Program of Neuroscience, Graduate School of Biomedical Sciences, Tufts University School of Medicine, Boston, MassachusettsBuilding Diversity in Biomedical Sciences Program, Tufts University School of Medicine, Boston, MassachusettsDepartment of Biology, Hamilton College, Clinton, New York
Department of Neuroscience, Program of Neuroscience, Graduate School of Biomedical Sciences, Tufts University School of Medicine, Boston, Massachusetts
Department of Neuroscience, Program of Neuroscience, Graduate School of Biomedical Sciences, Tufts University School of Medicine, Boston, Massachusetts
Department of Neuroscience, Program of Neuroscience, Graduate School of Biomedical Sciences, Tufts University School of Medicine, Boston, Massachusetts
Department of Neuroscience, Program of Neuroscience, Graduate School of Biomedical Sciences, Tufts University School of Medicine, Boston, MassachusettsBuilding Diversity in Biomedical Sciences Program, Tufts University School of Medicine, Boston, MassachusettsLouis Stokes Alliance for Minority Participation, Tufts University, Medford, Massachusetts
Department of Neuroscience, Program of Neuroscience, Graduate School of Biomedical Sciences, Tufts University School of Medicine, Boston, MassachusettsDivision of Endocrinology, Department of Pediatrics, Boston Children’s Hospital and Harvard Medical School, Boston, MassachusettsF.M. Kirby Neurobiology Center, Boston Children’s Hospital and Harvard Medical School, Boston, Massachusetts
Department of Neuroscience, Program of Neuroscience, Graduate School of Biomedical Sciences, Tufts University School of Medicine, Boston, Massachusetts
Department of Neuroscience, Program of Neuroscience, Graduate School of Biomedical Sciences, Tufts University School of Medicine, Boston, MassachusettsBuilding Diversity in Biomedical Sciences Program, Tufts University School of Medicine, Boston, MassachusettsDepartment of Chemistry, University of Puerto Rico, Cayey, Puerto Rico
Department of Neuroscience, Program of Neuroscience, Graduate School of Biomedical Sciences, Tufts University School of Medicine, Boston, Massachusetts
Department of Neuroscience, Program of Neuroscience, Graduate School of Biomedical Sciences, Tufts University School of Medicine, Boston, MassachusettsDivision of Endocrinology, Department of Pediatrics, Boston Children’s Hospital and Harvard Medical School, Boston, MassachusettsF.M. Kirby Neurobiology Center, Boston Children’s Hospital and Harvard Medical School, Boston, Massachusetts
Department of Neuroscience, Program of Neuroscience, Graduate School of Biomedical Sciences, Tufts University School of Medicine, Boston, Massachusetts
Chronic stress is a major risk factor for psychiatric illnesses, including depression. However, the pathophysiological mechanisms whereby stress leads to mood disorders remain unclear. Allopregnanolone acts as a positive allosteric modulator preferentially on δ subunit–containing GABAA (gamma-aminobutyric acid A) receptors. Accumulating clinical and preclinical evidence supports the antidepressant effects of exogenous administration of allopregnanolone analogs; yet, the role of endogenous allopregnanolone in the pathophysiology of depression remains unknown.
Methods
We utilized a chronic unpredictable stress (CUS) mouse model, followed by behavioral and biochemical assays, to examine whether altered neurosteroid signaling contributes to behavioral outcomes following CUS. We subsequently performed in vivo CRISPR (clustered regularly interspaced short palindromic repeats) knockdown of rate-limiting enzymes involved in allopregnanolone synthesis, 5α-reductase type 1 and 2 (5α1/2), in addition to lentiviral overexpression of 5α1/2 in the basolateral amygdala (BLA) of mice that underwent CUS to assess the impact of 5α1/2 on behavioral outcomes.
Results
The expression of δ subunit–containing GABAA receptors and endogenous levels of allopregnanolone were reduced in the BLA following CUS. Treatment with an exogenous allopregnanolone analog, SGE-516, was sufficient to increase allopregnanolone levels in the BLA following CUS. Knockdown of 5α1/2 in the BLA mimicked the behavioral outcomes associated with CUS. Conversely, overexpression of 5α1/2 in the BLA improved behavioral outcomes following CUS.
Conclusions
Our findings demonstrate that chronic stress impairs endogenous neurosteroid signaling in the BLA, which is sufficient to induce behavioral deficits. Further, these studies suggest that allopregnanolone-based treatments may directly target the underlying pathophysiology of mood disorders suggesting that targeting endogenous neurosteroidogenesis may offer a novel therapeutic strategy.
Neurosteroids, such as allopregnanolone, have been implicated in the pathophysiology of psychiatric illnesses, including anxiety, depression, and posttraumatic stress disorder (
). Few studies have directly explored the pathophysiological link between endogenous neurosteroids and mood disorders. Deficits in neurosteroids and a reduction in the expression of rate-limiting enzymes involved in endogenous neurosteroidogenesis, 5α-reductases, have been demonstrated in preclinical models of chronic stress (
Social isolation-induced increase in alpha and delta subunit gene expression is associated with a greater efficacy of ethanol on steroidogenesis and GABAA receptor function.
), these data suggest that deficits in endogenous neurosteroidogenesis may contribute to pathophysiological mechanisms underlying psychiatric illnesses (
). This hypothesis is supported by clinical evidence that finasteride, a 5α-reductase inhibitor, is associated with mood disorders, including anxiety and depression, referred to as post-finasteride syndrome, the pathophysiology of which may involve deficits in endogenous neurosteroid signaling (
Neuroactive steroid levels are modified in cerebrospinal fluid and plasma of post-finasteride patients showing persistent sexual side effects and anxious/depressive symptomatology.
). GABAARs are heteropentameric ligand-gated ion channels that mediate the majority of inhibitory neurotransmission in the brain. Recently, we demonstrated that δ-GABAARs are uniquely expressed on parvalbumin (PV) interneurons in the basolateral amygdala (BLA) and mediate the effects of exogenous allopregnanolone on behavioral states (
). These data suggest that exogenous neurosteroid actions in the BLA can impact mood; however, the impact of endogenous neurosteroid signaling in the BLA on mood remains unexplored. This study attempts to fill this gap in knowledge by investigating whether perturbations in endogenous neurosteroid signaling in the BLA contribute to chronic stress–induced behavioral changes.
Here, we directly examine the contribution of endogenous neurosteroid signaling in the BLA in mediating changes in behavioral states associated with chronic stress. These studies build upon well-established evidence that chronic stress is a risk factor for psychiatric disorders (
). We demonstrate that chronic unpredictable stress (CUS) induces changes in the capacity for endogenous neurosteroid signaling in the BLA, evident from a decrease in allopregnanolone levels measured in the BLA but not circulating plasma, decreased expression of the neurosteroidogenic enzymes, 5α-reductase type 1 and 2 (5α1/2), and their main site of action, δ-GABAARs, in the BLA. CRISPR (clustered regularly interspaced short palindromic repeats)–mediated knockdown of genes encoding for 5α1/2 in the BLA was sufficient to mimic the impact of CUS on behavioral states. Similarly, knocking down the predominant site of action of endogenous 5α-reduced neurosteroids, δ-GABAARs, in the BLA also mimicked behavioral deficits associated with CUS. Importantly, increasing the capacity for endogenous neurosteroidogenesis by overexpressing 5α1/2 in the BLA was sufficient to improve behavioral states following CUS. These findings implicate endogenous neurosteroid signaling in the BLA in the pathophysiology of mood disorders and provide a potential mechanistic underpinning for the antidepressant effects of allopregnanolone.
Methods and Materials
Detailed methods are described in the Supplement. Briefly, CUS mice underwent a 3-week protocol consisting of alternating overnight stressors (Figure S1A). Following CUS, liquid chromatography with tandem mass spectrometry was conducted on plasma, whole-brain, and BLA samples to detect levels of allopregnanolone in vehicle-treated mice or mice treated with SGE-516 (tool compound), a synthetic analog of allopregnanolone developed by SAGE Therapeutics, and quantitative reverse transcription–polymerase chain reaction was performed to measure 5α1/2 expression. Cre-dependent AAV (adeno-associated virus) viral vectors were constructed to facilitate the expression of guide RNAs targeting genes for δ-GABAARs and 5α1/2 (sgGabrd and sg5α1/2, respectively). sgGabrd was stereotaxically administered bilaterally to the BLA of PV-Cas9 mice to specifically ablate δ-GABAARs from PV interneurons in the region, given the unique expression of these receptors in PV interneurons in the BLA (
). The same stereotaxic procedures were performed in constitutive Cas9 mice to ablate 5α1/2 from all cell types in the BLA. To overexpress 5α1/2 in the BLA, 5α1/2 lentiviral constructs were stereotaxically injected bilaterally into the BLA of C57BL/6J mice. All mice were given a 1-week surgical recovery period prior to any behavioral testing to assess for stress-induced helplessness and avoidance behaviors.
Results
Behavioral Aberrations Following CUS
Given that chronic stress is a well-established risk factor for the development of psychiatric disorders, we utilized a CUS paradigm to examine behavioral sequelae of chronic stress. Mice subjected to CUS displayed increased avoidance behaviors, evident from a significant decrease in total time spent in the center of the open field (control: 57.8 ± 6.2 seconds; CUS: 39.7 ± 2.6 seconds) (Figure 1A, B) (control n = 14, CUS n = 15, p = .0107 unpaired t test) without a change in locomotor behavior (control: 2252 ± 135.4 beam breaks; CUS: 2193 ± 75.6 beam breaks) (Figure 1A, B) (p = .7009 unpaired t test). Mice subjected to CUS also exhibited an increase in stress-induced helplessness demonstrated by a shorter latency to immobility (control: 73.5 ± 7.4 seconds; CUS: 35.9 ± 5.9 seconds) (Figure 1C) (control n = 10, CUS n = 11, p = .0008 unpaired t test) and an increased total time spent immobile (control: 180 ± 13.8 seconds; CUS: 233.2 ± 7.7 seconds) in the tail suspension test (Figure 1C) (p = .0027 unpaired t test). Similarly, CUS mice exhibited a decreased latency to immobility (control: 64.6 ± 12.22 seconds; CUS: 6.2 ± 0.6 seconds) (Figure 1D) (control n = 10, CUS n = 15, p = .0010 unpaired t test) and an increased total time immobile in the forced swim test (control: 149.6 ± 17.9 seconds; CUS: 219.8 ± 5.9 seconds) (Figure 1D) (p = .0033 unpaired t test).
Figure 1CUS-induced behavioral deficits. CUS increased avoidance behaviors, with decreased total time spent in the center of the open field without altered overall locomotor behavior (A) control n = 14, CUS n = 15. (B) Representative heat maps of mobility during the open field test of control and CUS mice. CUS also increased stress-induced helplessness, with CUS mice exhibiting a decreased latency to immobility time and an increased total time spent immobile during the tail suspension test (C) control n = 10, CUS n = 11. CUS mice also displayed stress-induced helplessness in the forced swim test, exhibiting a decreased latency to immobility and increased total time immobile compared with controls (D) control n = 10, CUS n = 15. (E) PCA performed on all behavioral outcomes demonstrated different clusters for mice subjected to CUS compared with controls. (F) Summary of behavioral outcomes for mice that underwent CUS for transparency across experiments. ∗p < .05, ∗∗p < .01, ∗∗∗p < .001 using an unpaired t test. CUS, chronic unpredictable stress; EPM, elevated plus maze; L/D, light/dark; PCA, principal component analysis.
CUS did not alter avoidance behaviors in the elevated plus maze as demonstrated by similar outcomes in open arm time (control: 94.9 ± 15.8 seconds; CUS 97.3 ± 16.6 seconds) (Figure S1B) (control n = 9, CUS n = 10, p = .9166 unpaired t test), open arm entries (control: 37.3 ± 6.5 entries; CUS 38.5 ± 5.3 entries) (Figure S1) (p = .8914 unpaired t test), open arm distance (control: 301 ± 82.8 cm; CUS: 274.1 ± 65.1 cm) (Figure S1B) (p = .8020 unpaired t test), and locomotor behavior (control: 1459 ± 464.8 beam breaks; 1232 ± 414.5 beam breaks) (Figure S1B) (p = .7203 unpaired t test). CUS also did not alter behavioral outcomes in the light/dark box test, as demonstrated by similar time spent in the light zone (control: 169.8 ± 15.2 seconds; CUS: 178.0 ± 7.3 seconds) (Figure S1C) (control n = 14, CUS n = 15, p = .6313 unpaired t test) and locomotor behavior (control: 1491 ± 104.7 beam breaks; CUS: 1417 ± 74.7 beam breaks) (Figure S1C) (p = .5744 unpaired t test). To validate our battery of behavior outcomes for chronic stress, we performed a principal component analysis of all behavioral metrics from automated behavior scoring (Figure 1E; Figure S2A). A Bayesian information criteria test indicated a 3-cluster Gaussian mixed model to best fit the first 4 principal components explaining over 60% of variance (Figure S2B, C). Two clusters were predominantly control female or control male mice, 73% and 80%, respectively (Figure 1E). The third cluster was composed entirely of CUS males and females, which appeared as a distinct cluster in the principal component analysis space (Figure 1E). Thus, CUS induced robust changes in behavior, which distinguished between experimental groups, making this model and these outcome measures useful for investigating mechanisms whereby stress alters behavioral outcomes. The summary of behavioral changes following CUS are shown in Figure 1F.
CUS Decreases δ-GABAAR Expression in the BLA
To examine brain regions impacted by CUS, we conducted brain-wide c-Fos immunofluorescence from control and CUS mice. The average number of c-Fos–positive cells within the BLA was increased in mice subjected to CUS (control: 13.1 ± 1.4 cells; CUS: 31.5 ± 4.6 cells) (Figure 2A, B) (control n = 10, CUS n = 6, p = .0091 unpaired t test). Thus, this manuscript focused on potential pathophysiological changes in the BLA contributing to behavioral deficits following CUS.
Figure 2δ-GABAARs are downregulated in the BLA following CUS. (A) Representative images of c-Fos immunofluorescence in the BLA of control and CUS mice. The average number of c-Fos–positive cells in the BLA is reduced in mice subjected to CUS compared with controls (B) control n = 10 mice, CUS n = 6 mice, average 5 sections per mouse. (C) Representative images of immunofluorescence colabeling of PV interneurons and δ-GABAARs in the BLA. The average number of δ-positive cells in the BLA from control and CUS mice (D) control n = 7 mice, CUS n = 5 mice, average 5 sections per mouse. (E) (Top panel) Representative Western blot for δ-GABAAR expression in total protein isolated from the BLA from control and CUS mice. (Bottom panel) The average OD expression of δ-GABAARs and β-tubulin per 25 μg of total protein. Control n = 5 mice, CUS n = 5 mice. ∗∗p < .01, ∗∗∗p < .001 using an unpaired t test. δ-GABAAR, δ subunit–containing gamma-aminobutyric acid A receptor; BLA, basolateral amygdala; CUS, chronic unpredictable stress; OD, optical density; PV, parvalbumin.
Behavioral states, such as fear, avoidance behaviors, and stress-induced helplessness, have been shown to be governed by PV-positive interneurons in the BLA (
), we hypothesized that behavioral abnormalities following chronic stress might involve changes in δ-GABAAR expression on PV interneurons in the BLA. Following CUS, there was a significant reduction in cells expressing δ-GABAARs in the BLA (control: 85.6 ± 6.3 cells per section; CUS: 46.3 ± 2.8 cells per section) (Figure 2C, D) (control n = 7, CUS n = 5, p = .0004 unpaired t test). These findings were further validated using Western blot analysis, demonstrating a reduction in δ-GABAAR expression in the BLA in mice subjected to CUS (control: 66.3 ± 4.0 optical density [OD] units/2 μg total protein; CUS: 37.0 ± 4.4 OD units/25 μg total protein) (Figure 2E) (control n = 5, CUS n = 5, p = .0012) with no change in β-tubulin expression (control: 81.4 ± 8.1 OD units/25 μg total protein; CUS: 105.5 ± 20.01 OD units/25 μg total protein) (Figure 2E) (p = .3124 unpaired t test). Collectively, these data demonstrate that the predominant site of action for neurosteroid signaling is decreased within the BLA following CUS, which may contribute to the associated behavioral deficits.
Next, we examined whether this decrease in δ-containing GABAARs may contribute to behavioral deficits that are observed following CUS. A reduction in δ-GABAARs in the BLA was achieved by injecting sgGabrd (guide RNA encoding for the gene Gabrd specific to the δ-subunit of GABAARs) into the BLA of PV-Cas9 mice to knockdown δ-GABAARs on PV interneurons in the BLA (Figure 3A). This approach was sufficient to decrease the number of δ-GABAAR–positive cells in the BLA (control: 39.3 ± 3.1 cells per section; sgGabrd 7.0 ± 1.0 cells per section) (Figure 3B) (p ≤ .0001 unpaired t test) and induced an increase in stress-induced helplessness and avoidance behaviors. In the elevated plus maze, mice with reduced δ-GABAARs in the BLA exhibited a decrease in open arm entries (control: 27.4 ± 5.0 entries; sgGabrd: 15.5 ± 1.6 entries) (Figure 3C) (p = .0079 unpaired t test) and a decrease in total distance traveled in the open arm (control: 138.8 ± 38.8 cm; sgGabrd: 75.54 ± 7.9 cm) (Figure 3C) (control n = 5, sgGabrd n = 13, p = .0285 unpaired t test), an indication of increased avoidance behaviors. There was no significant difference in total time in the open arm for sgGabrd mice (control: 122.3 ± 22.1 seconds; sgGabrd: 132.9 ± 15.4 seconds) (Figure 3C) (p = .7129 unpaired t test). However, there was also a significant decrease in basic movements in the elevated plus maze for sgGabrd mice that these mice did not display in other behavioral tests (control: 519.8 ± 51.1 movements; sgGabrd: 258.5 ± 15.4 movements) (Figure 3C) (p ≤ .0001 unpaired t test). Mice with reduced δ-GABAARs in the BLA also displayed an increase in stress-induced helplessness demonstrated by an increase in total time immobile in the forced swim test (control: 161.0 ± 13.0 seconds; sgGabrd: 222.5 ± 10.5 seconds) (Figure 3D) (control n = 5, sgGabrd n = 13, p = .0046 unpaired t test), although there was no significant difference in latency to immobility in this test (control: 94.6 ± 19.3 seconds; sgGabrd: 62.8 ± 5.6 seconds) (Figure 3D) (p = .1780 unpaired t test). Collectively, these findings indicate that reduced δ-GABAAR expression on PV interneurons in the BLA is sufficient to induce behavioral deficits, such as those observed following CUS (Figure 1). These data further suggest that the decrease in δ-GABAAR expression observed following CUS may contribute to the behavioral deficits associated with CUS. However, these animals did not differ in the total amount of time spent in the center of the open field (control: 94.7 ± 6.9 seconds; sgGabrd: 74.65 ± 8.2 seconds) (Figure S3A) (control n = 5, sgGabrd n = 13, p = .0811 unpaired t test) or in locomotor behavior (control: 2489 ± 158.6 beam breaks; sgGabrd: 2141 ± 84.4 beam breaks) (Figure S3A) (p = .0976 unpaired t test). Similarly, sgGabrd mice did not display any behavioral changes in the light/dark box test as measured by the total time spent in the light zone (control: 184.4 ± 22.7 seconds; sgGabrd: 189.7 ± 24.2 seconds) (Figure S3B) (control n = 5, sgGabrd n = 13, p = .8762 unpaired t test) and locomotor behavior (control: 1115 ± 26.2 beam breaks; sgGabrd: 1140 ± 43.9 beam breaks) (Figure S3B) (p = .6295 unpaired t test). The summary of behavioral changes in mice with a reduction in δ-GABAARs on PV interneurons in the BLA is shown in Figure 3E.
Figure 3CRISPR knockdown of δ-GABAARs in the BLA induced behavioral deficits. (A) A schematic of the sgRNA construct for knockdown of Gabrd and overview of viral targeting strategy in the BLA of parvalbumin/Cas9 mice. (B) Representative colabeling for δ-GABAARs (diaminobenzidine) and sgRNA expression (GFP) confirmed a loss of δ-GABAARs in GFP-positive cells. The average number of cells expressing δ-GABAARs in the BLA was reduced in mice injected with sgGabrd compared with controls (inset). sgGabrd mice exhibited a reduction in the total distance traveled, number of entries, and a reduction in basic movements with no change in the amount of time spent in the open arm of the EPM compared with controls (C) control n = 5, sgGabrd n = 13. sgGabrd mice exhibited an increase in the total time spent immobile in the forced swim test compared with controls without any change in the latency to immobility (D) control n = 5, sgGabrd n = 13. (E) A summary of behavioral outcomes for knockdown of δ-GABAARs from parvalbumin interneurons in the BLA for transparency across figures and experiments. ∗p < .05; ∗∗p < .01, ∗∗∗p < .001, ∗∗∗∗p < .0001 using an unpaired t test. δ-GABAAR, δ subunit–containing gamma-aminobutyric acid A receptor; BLA, basolateral amygdala; CRISPR, clustered regularly interspaced short palindromic repeats; EPM, elevated plus maze; GFP, green fluorescent protein; L/D, light/dark; PAM, protospacer adjacent motif; sgRNA, single guide RNA.
Impaired Neurosteroidogenesis in the BLA Following CUS
To further explore whether there are deficits in endogenous neurosteroid signaling following CUS, we measured allopregnanolone levels in plasma, whole-brain, and BLA samples from control and CUS mice using liquid chromatography with tandem mass spectrometry. Plasma levels of allopregnanolone were not significantly different between control and CUS mice (control: 1.6 ± 0.8 ng/mL; CUS: 0.6 ± 0.1 ng/mL) (Figure 4A; control n = 11, CUS n = 13, p = .087747 unpaired t test), although there was a trend toward a decrease in the plasma of CUS mice. Similarly, there was no significant difference in whole-brain levels of allopregnanolone (control: 3.2 ± 1.1 ng/g; CUS: 1.9 ± 0.8 ng/g) (Figure 4B; control n = 6, CUS n = 4, p = .873461 unpaired t test). We did, however, observe a significant decrease in allopregnanolone levels in the BLA following CUS (control: 3.8 ± 1.1 ng/g; CUS: 1.8 ± 0.3 ng/g) (Figure 4C; control n = 4, CUS n = 5, p = .041426 unpaired t test). Interestingly, treatment with a synthetic allopregnanolone analog, SGE-516, prevented the CUS-induced reduction in allopregnanolone levels in the BLA (CUS+SGE-516 plasma: 1.0 ± 0.2; whole brain: 4.4 ± 1.3; BLA: 4.2 ± 0.4) (Figure 4A; CUS+SGE-516: plasma n = 14; Figure 4B; whole brain n = 8; Figure 4C; BLA n = 8; plasma p = .1964; whole brain p = .3296; BLA p = .0025). These data demonstrate a reduction in endogenous neurosteroid levels in the BLA following CUS which are likely brain region specific.
Figure 4Exogenous neurosteroid analog SGE-516 promoted restoration of endogenous allopregnanolone in vivo and in vitro through 5α-reductases type 1 and 2. Liquid chromatography with tandem mass spectrometry measurement of allopregnanolone levels from control, CUS, and CUS+SGE-516 mice. (A) Plasma, control n = 10, CUS n = 13, CUS+SGE-516 n = 14; (B) whole brain, control n = 6, CUS n = 4, CUS+SGE-516 n = 8; and (C) BLA samples, control n = 4, CUS n = 5, CUS+SGE-516 n = 8. (D) Quantitative reverse transcription–polymerase chain reaction measurement of BLA mRNA levels of Srd5a1 and Srd5a2 normalized to β-actin and control levels, control n = 16; CUS n = 20; CUS+SGE-516 n = 16 pooled BLA samples from 4 mice per experiment. (E) Representative images of Srd5a1 and Srd5a2 expressing cell lines showing cell confluency by DAPI staining (Left panel) and Srd5a1 and Srd5a2 GFP tag (Right panel). Quantification of NADP+ assay readout for luminescence from (F)Srd5a1 and (G)Srd5a2 expressing cell lines following application of control, progesterone, and SGE-516 treatments. ∗p < .05, ∗∗p < .01. BLA, basolateral amygdala; CUS, chronic unpredictable stress; GFP, green fluorescent protein; HEK, human embryonic kidney; mRNA, messenger RNA; NADP+, nicotinamide adenine dinucleotide phosphate; PROG, progesterone; RLU, relative luminescence units.
Given our observation that endogenous allopregnanolone levels are reduced in the BLA, we examined the expression of 5α1/2, the rate-limiting enzymes for allopregnanolone synthesis, in the BLA following CUS. Quantitative reverse transcription–polymerase chain reaction revealed a decrease in transcript levels of Srd5a1 and Srd5a2, which encode for 5α1/2, respectively, in the BLA following CUS (Srd5a1: 0.3 ± 0.1 fold change; Srd5a2: 0.6 ± 0.2 fold change) (Figure 4D; control n = 16, CUS n = 20 pooled BLA samples from 4 mice per experiment, Srd5a1p = .0031, Srd5a2p = .0831 t test). Exogenous SGE-516 treatment prevented the reduction in Srd5a1 and Srd5a2 expression in the BLA following CUS (SGE-516 Srd5a1: 2.0 ± 0.7 fold change; SGE-516 Srd5a2: 1.2 ± 0.4 fold change; Figure 4D) (n = 16 pooled BLA samples from 4 mice per experiment). These data suggest that endogenous neurosteroidogenesis is impaired in the BLA following CUS, which we proposed may contribute to behavioral deficits.
To probe whether SGE-516 may influence 5α1/2 activity, we utilized human embryonic kidney cells which overexpressed either Srd5a1 or Srd5a2 and examined the impact of either progesterone or SGE-516 treatment on the activity of these enzymes by measuring NADP+ (nicotinamide adenine dinucleotide phosphate) production as a proxy. Both progesterone and SGE-516 treatment increased NADP+ synthesis in human embryonic kidney cells expressing Srd5a1 (control: 2642.4 ± 506.1 relative luminescence units [RLU]; progesterone: 4928.3 ± 1820.1 RLU; SGE-516: 4277.1 ± 2656.8 RLU) (Figure 4F; control n = 8–12 replicates per experimental group; progesterone p = .0003, SGE-516 p = .05 unpaired t test) and Srd5a2 (control: 2839.3 ± 953.7 RLU; progesterone: 4484.3 ± 2332.0 RLU; SGE-516: 4591.2 ± 1342.7 RLU) (Figure 4G; control n = 8 replicates per experimental group; progesterone p = .04, SGE-516 p = .005 unpaired t test).
Impaired Neurosteroidogenesis in the BLA Is Sufficient to Induce Behavioral Deficits
Given that CUS induced brain region–specific deficits in endogenous neurosteroidogenesis, we next sought to directly investigate whether behavioral consequences of CUS involve impaired neurosteroidogenesis in the BLA. Utilizing constitutive Cas9-expressing mice, we knocked down 5α1/2 in the BLA using single guide RNAs targeting genes encoding these enzymes, Srd5a1 and Srd5a2 (Figure 5A). This approach reduced transcript levels of 5α1/2 (Srd5a1: 0.5 ± 0.1 fold change; Srd5a2: 0.7 ± 0.1 fold change) in the BLA (Figure 5B) (control n = 15, sg5α1/2 n = 8, Srd5a1p = .0001; Srd5a2p = .0037 t test). sg5α1/2 mice exhibited increased avoidance behaviors, spending less time in the light compartment of the light/dark box (control: 286.7 ± 21.7 seconds; sg5α1/2: 233 ± 13.5 seconds) (Figure 5C; control n = 15, sg5α1/2 n = 8, p = .0480 unpaired t test) with no change in locomotor behavior (control: 2423 ± 54.3 beam breaks; sg5α1/2: 2305 ± 62.3 beam breaks) (Figure 5C; p = .1745 unpaired t test). sg5α1/2 mice also displayed increased stress-induced helplessness behaviors, evident from a shorter latency to immobility (control: 76.07 ± 5.4 seconds; sg5α1/2: 31.7 ± 5.6 seconds) (Figure 5D; p ≤ .0001 unpaired t test) and an increase in total time immobile (control: 173.2 ± 10.3 seconds; sg5α1/2: 237.2 ± 13.0 seconds) (Figure 5D; control n = 14; sg5α1/2 n = 9, p = .0013 unpaired t test) in the tail suspension test. However, sg5α1/2 mice did not exhibit any behavioral changes in the open field test, spending similar time in the center (control: 76.5 ± 8.8 seconds; sg5α1/2: 81.9 ± 12.0 seconds) (Figure S4; control n = 14, sg5α1/2 n = 9, p = .7216 unpaired t test) and exhibited similar locomotor behavior (control: 1878 ± 70.6 beam breaks; sg5α1/2: 1718 ± 124.2 beam breaks) (Figure S4; p = .2814 unpaired t test). Collectively, these results suggest that impaired neurosteroid signaling in the BLA is sufficient to induce behavioral deficits reminiscent of those observed following CUS. The summary of behavioral changes in mice with reduced 5α1/2 expression in the BLA is shown in Figure 5E.
Figure 5Knockdown of Srd5a1 and Srd5a2 in the BLA induced behavioral deficits. (A) (Left panel) A schematic of the sgRNA construct for the knockdown of Srd5a1 and Srd5a2 in the BLA of constitutive Cas9 mice. (Right panel) The experimental timeline. The average expression of Srd5a1 and Srd5a2 mRNA in the BLA of sg5α1/2 mice decreased compared with controls measured using quantitative reverse transcription–polymerase chain reaction normalized to β-actin levels (B) control n = 15, sg5α1/2 n = 8. (C) (Top panel) Representative heat maps of movement of control and sgSrd1/2 mice in the L/D box test. (Bottom panel) Knockdown of 5α1/2 increased avoidance behaviors, evident from a decrease in the amount of time spent in the light chamber of the L/D box with no change in basic movements (C) control n = 15, sg5α1/2 n = 8. Knockdown of 5α1/2 in the BLA increased stress-induced helplessness, evident by a decreased latency to immobility and an increased total time spent immobile during the tail suspension test (D) control n = 14, 5α1/2 n = 9. (E) Summary of behavioral outcomes for transparency across behavioral tests. ∗p < .05, ∗∗p < .01, ∗∗∗p < .001, ∗∗∗∗p < .0001 using an unpaired t test. 5α1/2, 5α-reductase type 1 and 2; BLA, basolateral amygdala; EPM, elevated plus maze; L/D, light/dark; mRNA, messenger RNA; PAM, protospacer adjacent motif; sgRNA, single guide RNA.
Increasing the Capacity for Neurosteroidogenesis in the BLA Is Sufficient to Rescue Behavioral Deficits Following CUS
Given that we observed a decrease in the expression of 5α1/2 in the BLA following CUS (Figure 4), which was sufficient to increase avoidance and stress-induced helplessness behaviors (Figure 5), we next sought to investigate whether increasing 5α1/2 in the BLA may prevent the behavioral consequences of CUS. Utilizing lentiviral (LV) constructs, we overexpressed 5α1/2 in the BLA and subsequently subjected mice to CUS. Overexpression of 5α1/2 increased transcript levels in the BLA (Srd5a1: 1.9 ± 0.3 fold change; Srd5a2: 1.5 ± 0.1 fold change) (Figure 6C; control n = 7, LV 5α1/2 n = 4, Srd5a1 p = .0673; Srd5a2 p = .0360 t test). Following CUS, mice with increased 5α1/2 expression in the BLA (CUS+LV 5α1/2 mice) did not exhibit signs of CUS-induced avoidance behaviors with no difference in the amount of time spent in the center of the open field (control: 73.4 ± 8.3 seconds; CUS+LV 5α1/2: 101.5 ± 11.2 seconds) (Figure 6D; control n = 11, CUS+LV 5α1/2 n = 8, p = .0639 unpaired t test), with no change in locomotor behavior (control: 2613 ± 126.5 beam breaks; CUS+LV 5α1/2: 2738 ± 315.9 beam breaks) (Figure 6D; p = .7236 unpaired t test). CUS+LV 5α1/2 mice did not exhibit CUS-induced avoidance behaviors in the light/dark box test, exhibiting similar amounts of time spent in the light chamber (control: 193.7 ± 17.5 seconds; CUS+LV 5α1/2: 215.9 ± 26.5 seconds) (Figure 6E; control n = 11, CUS+LV 5α1/2 n = 8, p = .4966 unpaired t test) with no difference in locomotor behavior (control: 2234 ± 109.6 beam breaks; CUS+LV 5α1/2: 2242 ± 200.3 beam breaks) (Figure 6E; p = .9729 unpaired t test). These data suggest that overexpression of 5α1/2 is protective against the behavioral impacts of chronic stress. In fact, following CUS, CUS+LV 5α1/2 mice spent more time in the open arm of the elevated plus maze (control: 24.9 ± 2.8 seconds; CUS+LV 5α1/2: 60.6 ± 13.7 seconds) (Figure 6F; control n = 7, CUS+LV 5α1/2 n = 8, p = .0356 unpaired t test), displayed a trend toward traveling further in the open arm (control: 329.3 ± 52 cm; CUS+LV 5α1/2: 749.4 ± 195.9 cm) (Figure 6F; p = .0721 unpaired t test), and a trend to perform more entries into the open arm (control: 31.6 ± 5.6 entries; CUS+LV 5α1/2: 52.4 ± 9.5 entries) (Figure 6F; p = .0840 unpaired t test) with no change in locomotor behavior (control: 2632 ± 153.0 beam breaks; CUS+LV 5α1/2: 2751 ± 199.2 beam breaks) (Figure 6F; p = .6442 unpaired t test). CUS+LV 5α1/2 mice did not exhibit increased stress-induced helplessness as a result of CUS, demonstrating a similar latency to immobility (control: 30.8 ± 5.1 seconds; CUS+LV 5α1/2: 30.7 ± 3.3 seconds) (Figure 6G; control n = 8, CUS+LV 5α1/2 n = 8, p = .9875 unpaired t test) and total time immobile in the tail suspension test (control: 185.9 ± 8.5 seconds; CUS+LV 5α1/2: 181.1 ± 9.1 seconds) (Figure 6G; p = .7052 unpaired t test). These data demonstrate that enhancing the synthesis of 5α-reduced neurosteroids in the BLA is sufficient to overcome the behavioral deficits induced by CUS. The summary of behavioral changes in CUS mice with overexpression of 5α1/2 in the BLA is shown in Figure 6H.
Figure 6Overexpression of Srd5a1 and Srd5a2 improved behavioral outcomes following CUS. (A) (Top panel) Schematic of lentiviral construct and targeting for overexpression of Srd5a1 and Srd5a2 in the BLA and the experimental timeline (Bottom panel). (B) Representative immunofluorescence of GFP-tagged lentiviral targeting in the BLA of LV 5α1/2 mice. (C) The average mRNA expression of Srd5a1 and Srd5a2 increased in LV 5α1/2 mice compared with controls measured using quantitative reverse transcription–polymerase chain reaction in the BLA of LV 5α1/2 mice, which were normalized to β-actin levels, control n = 7, LV 5α1/2 n = 4. (D) Overexpression of 5α1/2 decreased avoidance behaviors in CUS mice, exhibited as no change in the time spent in the center of the open field test compared with controls with no change in the total number of basic movements. Control n = 11, CUS+LV 5α1/2 n = 8. These mice did not differ from controls in the time spent in the light zone or the number of basic movements performed in the L/D box test (E) control n = 11, CUS+LV 5α1/2 n = 8. (F) CUS mice with overexpression of 5α1/2 in the BLA also demonstrated a decrease in stress-induced anxiety, indicated by an increase in the total time spent in the open arm of the elevated plus maze with no change in basic movements compared with controls. LV Srd5a1/2 mice subjected to CUS also exhibited a trend toward traveling further and performing more entries into the open arm of the elevated plus maze. Control n = 7, CUS+LV 5α1/2 n = 8. (G) CUS mice with overexpression of 5α1/2 in the BLA exhibited a decrease in stress-induced helplessness as demonstrated by similar latency to immobility and overall time immobile in the tail suspension test. Control n = 8, CUS+LV 5α1/2 n = 8. (H) Summary of behavioral outcomes for CUS mice with lentiviral overexpression of 5α1/2 in the BLA. p > .05 is not significant; ∗p < .05 using an unpaired t test. 5α1/2, 5α-reductase type 1 and 2; BLA, basolateral amygdala; CUS, chronic unpredictable stress; EPM, elevated plus maze; GFP, green fluorescent protein; L/D, light/dark; LV, lentiviral; mRNA, messenger RNA.
These findings implicate deficits in endogenous neurosteroid signaling in the BLA in the pathophysiological mechanisms contributing to behavioral deficits following chronic stress (Figure 7). Here, we have demonstrated reduced allopregnanolone levels in the BLA following CUS, consistent with a reduction in expression of neurosteroidogenic enzymes, 5α1/2, in this region. Further, the site of action of 5α-reduced neurosteroids, δ-GABAARs, were also reduced in the BLA following CUS. Collectively, these findings suggest that chronic stress impairs endogenous neurosteroid signaling in the BLA, contributing to behavioral deficits following CUS. In fact, the knockdown of enzymes involved in endogenous 5α-reduced neurosteroidogenesis or their predominant site of action induced increased avoidance and stress-induced helplessness behaviors similar to those observed following chronic stress. These novel findings further our understanding of the mechanisms through which chronic stress alters mood, consistent with previous observations of reduced expression of 5α-reductases (
Social isolation-induced increase in alpha and delta subunit gene expression is associated with a greater efficacy of ethanol on steroidogenesis and GABAA receptor function.
) may be due to their ability to overcome pathological reductions in endogenous neurosteroid levels and endogenous neurosteroid signaling.
Figure 7Summary of major findings. CUS increases avoidance behaviors and stress-induced helplessness. Knocking down key enzymes involved in endogenous neurosteroid synthesis, 5α1/2, in the BLA was sufficient to increase avoidance behaviors and stress-induced helplessness. Similarly, knocking down the primary site of action for neurosteroid signaling, δ-GABAARs, also increased avoidance behaviors and stress-induced helplessness. Conversely, overexpression of 5α1/2 prevented the behavioral deficits following CUS, resulting in no change or even a reduction in avoidance behaviors or stress-induced helplessness compared with controls. 5α1/2, 5α-reductase type 1 and 2; BLA, basolateral amygdala; CUS, chronic unpredictable stress; δ-GABAAR, δ subunit–containing gamma-aminobutyric acid A receptor; EPM, elevated plus maze; L/D, light/dark.
). Work from our laboratory suggests that the antidepressant mechanism of allopregnanolone and synthetic neuroactive steroid analogs may involve the ability of these compounds to modulate BLA network states, mediating both affective state switching and prolonged antidepressant effects (
). We previously identified PV interneurons in the BLA as critical mediators of behavioral state changes and determined that they contain a high density of δ-GABAARs (
). The ability of neurosteroids to mediate transitions between BLA network and behavioral states likely involves their actions on δ-GABAARs expressed on PV interneurons in the BLA (
). We propose that CUS impairs neurosteroid-mediated enhancement of tonic GABAergic signaling on PV interneurons in the BLA, resulting in the desynchronization of network states and, consequently, behavioral states (Figure 7). Thus, endogenous neurosteroid actions in the BLA likely target δ-GABAARs on PV interneurons to shift behavioral states. This represents a novel, endogenous mechanism mediating affective state switching and may represent a mechanism mediating the episodic nature of psychiatric illnesses.
Most studies investigating neurosteroid influence on mood have relied on exogenous neurosteroid administration. Currently, little is known about the function of endogenous neurosteroidogenesis. We have overcome this limitation by developing CRISPR tools enabling direct manipulation of endogenous neurosteroidogenesis to examine the impact on behavioral states. The current study extends our knowledge by providing insight into the role of endogenous neurosteroid signaling on behavioral states. These data suggest that a physiological function of endogenous 5α-reduced neurosteroid signaling in the BLA involves setting a baseline affective tone. Consistent with this hypothesis, our data demonstrate that knocking down 5α1/2 in the BLA induces behavioral deficits, suggesting that impaired endogenous neurosteroidogenesis in the BLA disrupts affective tone (Figure 5). Conversely, overexpression of 5α1/2 reduces avoidance behaviors (Figure 6), demonstrating that enhanced neurosteroidogenesis can improve behavioral states. These findings suggest that a physiological function of endogenous 5α-reduced neurosteroids is to regulate affective tone.
Previously, our understanding of the physiological function of endogenous neurosteroidogenesis was limited by the inability to manipulate endogenous neurosteroidogenesis and relied on pharmacological tools, such as the 5α-reductase inhibitor finasteride. Clinically, treatment with finasteride has been shown to lead to mood disorders, including anxiety and depression, collectively referred to as post-finasteride syndrome, the pathophysiology of which is thought to involve deficits in endogenous neurosteroid signaling (
Neuroactive steroid levels are modified in cerebrospinal fluid and plasma of post-finasteride patients showing persistent sexual side effects and anxious/depressive symptomatology.
). In animal models, exogenous allopregnanolone exerts anxiolytic and antidepressant effects, while treatment with finasteride blocks the anxiolytic and antidepressant effects of progesterone (
). Collectively, these data suggest that endogenous neurosteroids are capable of impacting mood. This study demonstrates that interfering with endogenous neurosteroid signaling is capable of inducing deficits in behavioral states, suggesting that endogenous neurosteroids have a tonic influence over affective states. Further, we have demonstrated that enhancing endogenous neurosteroidogenesis can improve behavioral states, suggesting that targeting endogenous neurosteroidogenesis may represent a novel therapeutic approach for the treatment of mood disorders (
Neuroactive steroid levels are modified in cerebrospinal fluid and plasma of post-finasteride patients showing persistent sexual side effects and anxious/depressive symptomatology.
). Further studies are required to investigate the therapeutic potential of targeting endogenous neurosteroidogenesis for the treatment of mood disorders.
Acknowledgments and Disclosures
This work was supported by the National Institutes of Health (Grant Nos. R01AA026256, R01NS105628, R01NS102937, R01MH128235, and P50MH122379 [to JLM, NLW, PA, AD, and RP], Grant No. T32DK124170 [to SH], and Grant Nos. R01NS107315, R01DK108797, R21HD098056, and P30DK046200 [to DK]). SGE-516 data presented in this manuscript were part of a sponsored research agreement with SAGE Therapeutics.
We thank Boston Children’s Hospital viral vector core for packaging the single guide RNAs used in this manuscript as well as PharmaCadence for performing LC-MS analytical support.
NLW and JLM contributed to conceptualization of the project. NLW, TD, AE-S, AD, JG, RP, EJS, and JLM contributed to data generation and NLW, PA, LB, GLW, and JLM contributed to data analysis. SH and DK developed the CRISPR constructs used in the study. NLW, PA, GLW, and SH contributed to data visualization. NLW and JLM wrote the original draft of the manuscript and NLW, PA, LB, TD, AD, AE-S, JG, SH, RP, EJS, GLW, DK, and JLM also contributed to manuscript preparation (reviewing and editing).
JLM serves as a member of the Scientific Advisory Board for SAGE Therapeutics, Inc. All other authors report no potential biomedical financial interests or conflicts of interest.
Social isolation-induced increase in alpha and delta subunit gene expression is associated with a greater efficacy of ethanol on steroidogenesis and GABAA receptor function.
Neuroactive steroid levels are modified in cerebrospinal fluid and plasma of post-finasteride patients showing persistent sexual side effects and anxious/depressive symptomatology.
00050-1&id=gr1.jpg){kind=link}
00050-1&id=gr2.jpg){kind=link}
00050-1&id=gr3.jpg){kind=link}
00050-1&id=gr4.jpg){kind=link}
00050-1&id=gr5.jpg){kind=link}
00050-1&id=gr6.jpg){kind=link}
00050-1&id=gr7.jpg){kind=link}