GABA_A RECEPTORS AS TARGETS FOR TREATING AFFECTIVE AND COGNTIVE SYMPTOMS OF DEPRESSION

Abstract

In the past 20 years our understanding of the pathophysiology of depression has evolved from a focus on an imbalance of monoaminergic neurotransmitters to a multifactorial picture including an improved understanding of the role of glutamatergic excitatory and GABAergic inhibitory neurotransmission. FDA-approved treatments targeting the glutamatergic (esketamine for major depressive disorder) and GABAergic (brexanolone for peripartum depression) systems have become available. This review focuses on the GABA_A receptor system as a target for novel antidepressants and discusses the mechanisms by which modulation of δ-containing GABA_A receptors with neuroactive steroids or of α5-containing GABA_A receptors result in antidepressant or antidepressant-like actions and clinical data on neuroactive steroids. Moreover, a potential mechanism by which α5-GABA_A receptor positive allosteric modulators may improve cognitive deficits in depression is presented.

Translating knowledge on microcircuitry into novel GABAergic treatments of depression

Major Depressive Disorder (MDD) is the leading cause of ill health and disability worldwide with a 12-month prevalence of approximately 6% and a lifetime risk of 15–18% [1]. MDD is characterized by discrete, typically recurrent episodes of at least two weeks, consisting of lower or depressed mood and/or decreased interest in pleasurable activities, combined with altered cognition and disrupted neurovegetative functions that control energy, sleep and/or appetite [1]. Postpartum Depression (PPD) (recently renamed peripartum depression) refers to a common variant of MDD observed in 10–15% of women during pregnancy or after childbirth. Effective first line treatments for depressive disorders are available, but they lead to remission in less than half of patients, typically act with a therapeutic delay of weeks to months and are often associated with adverse side effects that reduce adherence to therapy. Pharmacologically, these treatments are designed to increase the extracellular concentrations of monoamine neurotransmitters, especially serotonin, and are based on treatments developed more than 60 years ago Due to the lack of reliable biomarkers for MDD, research into the neurobiological basis of MDD has traditionally built on the mechanism of action of monoaminergic antidepressants (see Glossary), which has given rise to the monoaminergic hypothesis of depression [2]. However, it is now widely accepted that reduced monoamine function is insufficient to explain the etiology of depressive disorders.

Research into the etiology of MDD and its treatments has shifted the field’s attention to the major amino acid neurotransmitters, GABA and glutamate, which exert more direct control over neuronal activity than monoamines. This change in focus was noticeably accelerated by insights into the mechanisms of action of ketamine and brexanolone, two recently FDA-approved, rapid-acting antidepressants that directly target glutamatergic and GABAergic neurotransmission. Fast inhibitory and excitatory neurotransmission occur predominantly via neurotransmitter action on ligand-gated ion channels, while G protein-coupled receptors mediate slower effects. GABA acts as the principal inhibitory neurotransmitter in the brain and functions primarily by activating GABA-gated chloride channels known as GABA_A receptors (GABA_ARs)(Text Box 1). Glutamate in turn is the primary excitatory neurotransmitter acting primarily through glutamate-gated cation channels. Together, GABA and glutamate and their receptors exert tight control over neuronal activity and functional plasticity. Even subtle changes in the function of these transmitters or their receptors can profoundly and rapidly affect emotional or cognitive behavior [3] [4]. In this review, we discuss the roles of the GABAergic system and of GABA_A receptors in microcircuits relevant for depression with focus on the mechanisms of and recent clinical data on neuroactive steroids (including a recently FDA-approved compound) modulating the activity of GABA_A receptors. We also present preclinical evidence on α5-GABA_A receptors (α5-GABA_ARs) as potential therapeutic targets for the treatment of depression and provide a model explaining why both positive and negative modulators of α5-GABA_ARs may have antidepressant effects.

Text Box 1: Diversity of GABA_A receptors.

GABA_A receptors (GABA_ARs) are ligand-gated chloride channels and the major inhibitory neurotransmitter receptors in the central nervous system. They are heteropentamers composed of combinations of homologous subunits encoded by 19 different genes. A major subset of these receptors (approximately 80%) composed of α1,2,3,5, β1–3 and γ1–3 subunit are the well-characterized targets for benzodiazepines, a major class of psychoactive therapeutics that are widely used as anxiolytics sedatives, myorelaxants and antiepileptics [3]. These receptors are highly concentrated at synapses, conduct rapidly desensitizing (phasic) currents, and exhibit relatively low affinity for GABA, which further delimits their function to synapses. A smaller subset of GABA_ARs comprised of α, β and δ subunits (overall <5%, but 30% in cerebellum and possibly similarly high levels in dentate gyrus and thalamic nuclei), with the δ subunit conferring increased sensitivity to neuroactive steroids. Moreover, a substantial portion of the α5βγ2 receptors (especially in hippocampus) are localized extrasynaptically and diffusely distributed along neuronal membranes. They exhibit significantly higher affinity for GABA and slower or non-desensitizing currents and mediate a tonic form of neural inhibition. These non-synaptic GABA_ARs by means of their non-desensitizing tonic currents exert profound control over neuronal excitability. While the α and β subunits form the GABA binding site, α and γ subunits form a binding site for benzodiazepines. GABA_ARs contain two neurosteroid binding sites that are formed by the α subunit and the α and β subunit interface, respectively.

GABA_AR subtypes, frequently defined by their α subunits, have diverse physiological and pharmacological functions. α1-GABA_ARs increase sedation, while α2-GABA_ARs and α5-GABA_ARs reduce anxiety and increase resilience to chronic stress. α3-GABA_ARs and α5-GABA_ARs modulate sensorimotor gating [3]. In young adult animals, α5-GABA_ARs reduce cognitive functions, as α5 NAMs improve memory, while in aged animals they increase cognitive functions, as α5 PAMs improve memory [98]. In young adult animals, α5-GABA_ARs in the dentate gyrus granule cells [75] and in CA1 pyramidal neurons [76] improve performance in high interference tasks. A specific GABA_AR subtype may have different functions based on its location in specific neuronal populations or circuits, which gave rise to the “circuit pharmacology” concept [3]. For example, recently it was shown that the amnestic action of etomidate is mediated by α5-GABA_ARs on interneurons in the hippocampal CA1 region and not by α5-GABA_ARs on CA1 pyramidal neurons [99].

It is interesting to note that while GABA_ARs have been implicated in depression preclinically and clinically, classical benzodiazepines do not have antidepressant actions on their own, although antidepressant plus benzodiazepine therapy is more effective than antidepressants alone in an early phase [100]. This could be due to benzodiazepines targeting a majority of GABA_ARs, including receptors on both excitatory and inhibitory neurons. The recent finding that allopregnanolone acts preferentially on δ-GABA_ARs on PV interneurons in the BLA [38] to modulate network and behavioral states suggests that actions on specific subtypes of neurons may contribute to the antidepressant effects of these compounds.

The role of GABAergic system and of GABA_A receptors in the psychobiology of depressive disorders

The GABAergic deficit hypothesis of MDD posits that reduced GABA function contributes causally to MDD, and that such defects are normalized following successful antidepressant therapy [5] [6] [7] [8] [9] [10]. GABA functional deficits that contribute to MDD may involve genetic, developmental, and environmental aberrations that impact the integrity of GABAergic interneurons and synapses. Reduced brain concentrations of GABA and their normalization by successful treatment are among the most replicated findings in MDD [5]. Consistent with these in vivo findings, analyses of postmortem brain samples of subjects with depression show reduced expression of glutamic acid decarboxylase (GAD), the principal enzyme responsible for the synthesis of GABA [5] as well as reduced expression of peptide markers that identify select subsets of GABAergic interneurons, especially somatostatin (SST, further discussed below). Notably, similar but more pervasive GABA-related gene expression changes have been described in schizophrenia, which shows reduced expression of both GAD orthologs (GAD1 and 2, rather than just GAD1 as in MDD), as well as SST and parvalbumin (PV), a marker of a functionally distinct class of GABAergic interneurons (Fig. 1, see also further below) [11]. MDD is further associated with reduced plasma and cerebral spinal fluid levels of progesterone-derived neurosteroids that act as endogenous positive allosteric modulators (PAM) of GABA_ARs. In particular, allopregnanolone and pregnanolone are reduced in MDD, downregulated by chronic stress in rodents, and normalized by antidepressants, concomitant with improvement of depression and normalization of stress induced behavior (Fig. 2) [6].

Fig. 1: Cortical GABAergic microcircuit controlling stress vulnerability and resilience relevant for depressive disorders and antidepressant drug action.

The large majority of SST cells (i.e. >70%) represent Martinotti cells that preferentially innervate the distal apical tufts of pyramidal neurons (PN) were in cortical Layer 1 [105]. PV-positive interneurons preferentially innervate the somata, proximal dendrites, and axon initial segments of PNs and thereby exert powerful control over action potential firing of these neurons. VIP cells preferentially inhibit SST cells, which causes disinhibition of PNs. PV and SST cells also innervate each other. In particular, inhibition of PV cells by a subset of SST cells in layer 4 that are distinct from Martinotti cells, produces disinhibition of PNs [105]. The distinctive functional properties of SST cells suggest they exert a neuroprotective role. First, the activity of SST neurons, but not PV neurons in the prefrontal cortex scales with the strength of excitatory synaptic inputs from the basolateral amygdala (BLA) (1). Second, GABAergic inputs to PN dendrites buffer the temporal and spatial spread of NMDA receptor-mediated Ca²⁺ entry at adjacent glutamatergic synapses on dendritic spines [106] (2). Third, GABAergic inhibition of dendritic spines by SST neurons is principally mediated by postsynaptic α5-GABA_ARs, presumably composed of 2α5, 2β and 1γ2 subunit [92] (3). Fourth, genetically increasing the excitability of SST cells mimics the behavioral effects of antidepressant drug treatment [7] including stress resilience [8]. Lastly, SST neurons in the mPFC are direct targets also of excitatory inputs from the vHPC [71] (4).

Fig. 2: Pathological and therapeutic roles of neurosteroids in the BLA in depression.

PV interneurons in the BLA are critical for governing the activity of BLA principal neurons and network states driving behavioral states. PV interneurons in the BLA uniquely express the δ-GABA_ARs. Chronic stress has been shown to decrease the expression of δ-GABA_ARs in the BLA, which is the primary site of action for neurosteroids, suggesting that neurosteroid signaling may be impaired in the BLA following chronic stress exposure. Consistent with deficits in neurosteroid signaling, chronic stress is associated with a reduction in the expression of 5α-reductase (Srd5a1/2), a key enzyme in neurosteroidogenesis. Further, the endogenous levels of allopregnanolone are decreased in the BLA following chronic stress. Given that chronic stress is a major risk factor for MDD and the evidence that neurosteroids exert robust anxiolytic and antidepressant effects, impaired endogenous neurosteroid signaling may contribute to deficits in emotional processing and MDD. Impaired endogenous neurosteroidogenesis related to MDD may also help explain the therapeutic efficacy of exogenous NAS treatment for restoring the healthy behavioral state and treating MDD.

Interestingly, a separate class of endogenous neuroactive steroids (NAS) act as negative allosteric modulators (NAMs) of GABA_ARs. They include steroids that are sulfated at the 3-position of the steroid A-ring (in place of a hydroxyl group in the 3α-position in NAS PAMs) or have a hydroxyl group in the 3-β position, such as pregnenolone sulfate (PREGS) and dehydroepiandrosterone sulfate (DHEAS) or isopregnanolone or epipregnanolone, respectively. In addition to acting as NAMs of GABA_ARs, these compounds also modulate NMDA receptors. In contrast to the reduced levels of allopregnanolone implicated in the pathophysiology of depression, elevated levels of this separate class of NAS have been associated with depression [12]. Consistent with stress as a primary vulnerability factor of depressive disorders, a growing body of recent preclinical studies indicate that SST-expressing GABAergic cells are highly vulnerable to chronic stress (Text Box 2). This is evidenced by stress-induced gene expression changes that map to SST cells [13], as well as a number of activity-induced mechanisms of plasticity that are uniquely observed in these cells. In particular, SST neurons in the medial prefrontal cortex differ from PV cells in that they show activity-induced strengthening (facilitation) of glutamatergic inputs from the amygdala and correspondingly enhanced synaptic inhibition of their neocortical pyramidal target cells [14] (Fig 1). By contrast, corresponding glutamatergic inputs to PV cells show activity-induced adaptation [15]. Similar differences in activity-induced plasticity of glutamatergic inputs to SST and PV cells have been described for the hippocampus [16]. Deliberately increasing the excitability of SST cells in mice results in behavior that mimics that of antidepressant drug treatment, along with resilience to chronic stress exposure, indicating that activity- or stress-induced SST-mediated inhibition of distal dendrites of pyramidal neurons is neuroprotective [7, 8]. Conversely, stress or chemogenetically induced increases in activity of PV cells, which preferentially target the proximal dendrites, somata and axon initial segments of pyramidal cells, lead to anxiety- and anhedonia-like changes in motivated behavior [17, 18]. These preclinical studies predict that selective potentiation of GABAergic inhibition at pyramidal cell dendrites, has therapeutic utility for depressive and other stress-related psychiatric disorders. α5 subunit-containing GABA_ARs (α5-GABA_ARs) seem to uniquely fit that bill, at least in the prefrontal cortex and hippocampus, as further detailed below. In addition, α2-GABA_ARs are similarly (although less selectively) positioned and have been shown to promote resilience to psychosocial stress, at least in male mice [19]. The recent finding that α2-GABA_ARs on D2-receptor-positive but not on D1-receptor positive neurons (presumably on medium spiny neurons in the nucleus accumbens) promote resiliency to chronic stress but do not modulate anxiety-like behavior [19] is consistent with a report that repeated optogenetic activation of D2-positive medium spiny neurons promotes susceptibility to mild social stress [20].

Text Box 2: Depressive disorders show excessive vulnerabilities to stress that are translatable.

For preclinical studies of depression, it is important to use animal models that provide results that are translatable to humans. Depressive disorders are uniquely human conditions whose diagnosis is entirely reliant on questionnaires of patients. This is a major hurdle for basic research approaches that require translating these disorders to rodents. To circumvent this problem, preclinical research has made use of the notion that chronic and excessive stress are primary vulnerability factors for depressive disorders, along with virtually all other psychiatric conditions. In further support of chronic stress-based models of depressive disorders, antidepressant drug treatments of rodents invariably ameliorate and reverse the detrimental behavioral effects of chronic stress exposure. Conceptually, stress-sensitive behavioral parameters of rodents can be subdivided into positively regulated motivated behavior that assess hedonic drive and anhedonia (i. e., chronic stress results in anhedonia-like reductions in pleasurable behavior of rodents such as eating palatable food, social interaction, and sexually motivated behaviors) and negatively motivated behaviors that relate to anxiety (i. e., chronic stress exposure amplifies the anxiety-like behavioral inhibition of rodents to explore naturally aversive situations such as novelty, wide open spaces, and bright light).

In support of a key role of GABAergic transmission in depressive disorders, chronic stress exposure of rodents has been shown to reproduce diverse GABA-related molecular and cellular changes implicated in MDD, including downregulation of GAD and SST [13, 69, 70], along with a chronically elevated glutamatergic tone and functional defects in glutamatergic synaptic transmission [101]. Mice with genetically engineered reductions in GABA_AR function show chronic stress axis activation and they phenocopy chronic stress induced changes in positively and negatively motivated behavior, functional defects in glutamatergic synapses, and deficits in adult hippocampal neurogenesis [102, 103]. Moreover, conventional antidepressants and ketamine can reverse GABA_AR deficit-induced changes in motivated behavior and the defects in GABAergic and glutamatergic synapse function observed in these models [9, 102]. Altogether these preclinical studies indicate that reduced GABA concentrations and other symptoms of GABAergic deficits in patients can contribute causally to depressive disorders and act upstream of changes in glutamate and other neurotransmitter systems implicated in depression [104].

Neuroactive steroids (NAS) as modulators of GABA_A receptors

GABA_ARs with unique subunit compositions mediate distinct forms of GABAergic inhibition: synaptic GABA_ARs mediate phasic GABAergic inhibition, whereas extrasynaptic receptors mediate tonic GABAergic inhibition. Synaptic GABAergic inhibition is mediated by postsynaptically concentrated receptors, which contain the γ2 subunit as a distinctive feature, rapidly desensitize and mediate transient GABAergic inhibition. Tonic GABAergic inhibition is a persistent form of inhibition that is mediated by non-desensitizing, extrasynaptic GABA_ARs, which are distinguished from postsynaptic GABAARs by the presence of α4, α5, α6, and/or δ subunits [21]. Synaptic and extrasynaptic GABA_ARs are subject to dynamic exchange between postsynaptic or perisynaptic and extrasynaptic membrane sites, allowing for extrasynaptic receptor-mediated modulation of synaptic (phasic) currents and contribution of synaptic receptors and tonic currents [22] [23] [24]. α5-GABA_ARs not only produce tonic inhibition through extrasynaptic GABA_ARs but can also mediate synaptic currents and regulate dendritic outgrowth and spine maturation in synaptic locations [25]. Nevertheless, synaptic and extrasynaptic receptors can be distinguished pharmacologically, with the γ2 subunit conferring sensitivity to benzodiazepines (see Text Box 1) and the δ subunit conferring increased sensitivity to NAS [26]. Two distinct NAS binding sites have been identified on GABA_ARs that result in indirect (GABA-mediated) and direct receptor activation at low and high NAS concentrations, respectively [27, 28]. Allosteric potentiation of GABA responses involves binding of NAS to the α-subunit transmembrane domain of GABA_ARs, whereas direct activation of GABA_ARs by NAS involves binding at the interface between α and β subunits [27, 28]. In addition to allosteric modulation of GABA_AR channel function, NAS have also been shown to act indirectly through metabotropic receptors and facilitate the cell surface accumulation and functional expression of synaptic GABA_ARs [29] [30] [31] [32]. The known mechanism of action of NAS as positive allosteric modulators of GABA_ARs and the well-established antidepressant effects of these compounds provide further support for the GABAergic deficit hypothesis of depression [6].

Chronic stress is a major risk factor for psychiatric illnesses and has been widely used to model features of mood disorders in animals (Text Box 2). Neurosteroids have been implicated in the pathophysiology of psychiatric illnesses, including anxiety and depression, based on findings from both clinical and preclinical studies (for review see [33] [34]). Deficits in the capacity for endogenous neurosteroidogenesis, evident from a decrease in the levels of allopregnanolone and a reduction in the expression of key neurosteroidogenic enzymes, are implicated in the pathophysiology of depression (Fig. 2) [33]. In addition, the primary site of action of NAS, δ GABAARs, is also reduced in chronic stress models [35]. Collectively, these findings suggest that impaired endogenous neurosteroid synthesis may contribute to behavioral deficits associated with chronic stress. Knockdown of the GABA_AR δ subunit in PV neurons of the basolateral amygdala (BLA) has been shown to induce behavioral impairments. Knockdown of 5α-reductase type 1 and 2, key neurosteroidogenic enzymes, in the BLA has similarly been shown to induce behavioral deficits which are reminiscent of those observed following chronic stress [35]. Conversely, overexpression of 5α-reductase type 1 and 2 in the BLA improves the behavioral outcomes following chronic stress [35]. These data implicate impaired endogenous neurosteroidogenesis in stress-induced behavioral impairments and suggest that harnessing endogenous neurosteroidogenesis may have therapeutic potential for the treatment of psychiatric illnesses (Fig. 2). Interestingly, the antidepressant effects of selective serotonin reuptake inhibitors (SSRIs) may also involve actions on the neurosteroidogenic pathway. SSRI treatments increase the levels of endogenous NAS [36] and enhance the conversion of 5α-dihydroprogesterone to allopregnanolone by 3alpha-HSD type III [37]. The translocator protein 18k Da (TSPO) plays a key role in steroidogenesis and has been proposed to be a therapeutic target for depression [34]. Treatment with SGE-516, a synthetic NAS analog developed by SAGE Therapeutics with molecular pharmacology similar to allopregnanolone, prevents the behavioral deficits associated with chronic stress [38]. NAS have been shown to exert antidepressant-like effects in a wide range of animal models and are thought to contribute to the antidepressant effects of other antidepressants (for review see [39–42]). Thus, there is exhaustive preclinical evidence supporting the anxiolytic and antidepressant effects of NAS.

Neuroactive steroid (NAS) as therapeutics in MDD

Zuranolone (see Fig. 3) is a synthetic GABAergic NAS with molecular pharmacology similar to allopregnanolone. In clinical studies, it has shown promise as a treatment for MDD in adults. Antidepressant effects of zuranolone were demonstrated in several Phase 2 and 3 clinical trials in patients with MDD, as a monotherapy or as an add-on treatment [43]. Zuranolone improved the quality of life of patients with MDD in a randomized, placebo-controlled phase 2 trial [44] and exerted antidepressant effects in a double-blind, phase 2 trial in patients with MDD [43]. Based on these promising results from clinical trials, a New Drug Application (NDA) was submitted to the FDA for the use of zuranolone for the treatment of MDD.

Fig 3: Chemical formulas of neuroactive steroids and modulators of α5-GABA_ARs.

Left: Formulas of neuroactive steroids used in clinical studies of depressive disorders. Right: Formulas of modulators of the α5-GABA_AR used in preclinical studies.

Ganaxolone, (see Fig. 3) a positive allosteric modulator of GABA_ARs with similar pharmacology to allopregnanolone developed by Marinus Pharmaceuticals, has demonstrated antidepressant effects in an open-label, uncontrolled pilot study in postmenopausal women with depression and insomnia [45]. The mean MADRS score was decreased by the end of the eight-week treatment and persisted over a two-week taper period. Nearly half of the subjects experienced the antidepressant response and remission throughout the study. Ganaxolone treatment also improved symptoms assessed using the Inventory of Depressive Symptomatology-Self-Report, MADRS Reduced Sleep subscale, and Symptoms of Depression Questionnaire (SDQ) [45]. These studies are promising but additional follow-up studies are required to establish the antidepressant efficacy of ganaxolone in MDD.

Eliem Therapeutics has developed ETX-155, a novel NAS that acts as a positive allosteric modulator at GABA_ARs [46] with a planned phase II trial for MDD.

NAS as therapeutics in postpartum depression (PPD)

NAS have been implicated in PPD due to the timing of symptom onset, which coincides with a dramatic decline in endogenous NAS levels. Preclinical studies further support a role for NAS in the pathophysiology of PPD. Fluctuations in NAS levels have been shown to alter GABA_AR subunit expression and function over the ovarian cycle [47] and throughout the peripartum period [48]. Mice engineered to lack the expression of the GABA_AR δ subunit (Gabrd^−/− mice) exhibit behavioral deficits that are restricted to the postpartum period along with deficits in maternal care [48]. Treatment with a selective agonist at these receptors (gaboxadol = THIP) improves postpartum behavior and increases pup survival [48]. The abnormal postpartum behavior of mice lacking the GABAAR δ subunit were associated with altered stress reactivity [49]. Genetically altered mice that fail to suppress the hypothalamic-pituitary-adrenal (HPA) axis during the peripartum period (KCC2/Crh mice) show abnormal postpartum behaviors [50]. Treatment with SGE-516 improved postpartum behavioral outcomes in both of these models [51]. Similarly, allopregnanolone treatment has been shown to improve behavioral outcomes in a mouse model for PPD (involving maternal separation and early weaning) [52]. These preclinical studies implicate deficits in NAS signaling in the pathophysiology of PPD and were key in demonstrating the therapeutic potential of NAS treatment.

Brexanolone (see Fig. 3), a proprietary i.v. formulation of allopregnanolone, has been approved by the US Food and Drug Administration (FDA) as the first treatment specifically for postpartum depression (PPD). Zuranolone, an orally available allopregnanolone analog, has also demonstrated success in clinical trials for the treatment of PPD and an application is currently being reviewed by the FDA. Brexanolone treatment demonstrated rapid and persistent antidepressant effects in women with severe and moderate PPD [53–56]. Brexanolone (i.v., 60 hrs) reduced depression symptoms, decreasing the scores on the Hamilton Rating Scale for Depression [HAM-D] in a double-blind, randomized, placebo-controlled clinical trial across four different sites [56]. The antidepressant effects of brexanolone in women with PPD was reproduced in two multicenter, double-blind, randomized, placebo-controlled, Phase 3 trials, demonstrating a significant reduction in total HAM-D scores [55]. These studies demonstrate both rapid and persistent antidepressant effects, lasting up to 30 days [53, 55, 56]. Similarly, zuranolone (30 mg once daily for 14 days in 153 randomized patients) demonstrated both rapid and sustained antidepressant (as well as anxiety-reducing and insomnia-reducing) effects in a randomized clinical trial in women with PPD, meeting the primary endpoint, significantly decreasing the HAM-D scores at day 15 compared to placebo, with effects lasting for 45 days, i.e., four weeks beyond the end of treatment [57, 58].

Thus, NAS have been established to be effective in the treatment of PPD. The mechanisms underlying the persistent antidepressant effects of NAS remain unclear, but may involve anti-inflammatory effects [34] and the ability of NAS to modulate network states involved in emotional processing [59].

Ganaxolone, has demonstrated antidepressant effects in women with PPD in a randomized clinical trial. Infusion with ganaxolone (i.v., 60 hrs) resulted in a significant reduction in HAM-D scores from baseline at 48 hrs, 60 hrs, and day 34 compared to placebo [60]. Seventy-five percent of women were categorized as responders, defined as a ≥50% reduction in HAM-D scores from baseline and 50% achieved remission, determined by HAM-D scores ≤7 at day 34. Despite these promising initial studies, the antidepressant effects in PPD failed to replicate and ganaxolone has not been moved forward for this indication.

An advantage of this class of compounds is that they are derived from naturally occurring NAS. As a result, very few adverse side effects have been reported with the most common being headache, dizziness, and somnolence [55]. Unlike benzodiazepines, these compounds appear not to lead to tolerance or dependence, nor is there evidence of abuse potential. However, given the short time on the market, it is not clear yet clear how these compounds will be tolerated in the real world, such as how they impact sleep, daily functioning, interact with prescribed drugs, alcohol, and other drugs of abuse. Despite these potential limitations, these compounds have demonstrated remarkable success as antidepressants. Moreover, they make use of a novel mechanism of action and help address an enormous unmet medical need.

α5-GABA_A receptors as a therapeutic target in depression and aging

The connectivity between neurons in the brain is an important requirement for the performance of higher brain functions such as cognition, and restoration of dysfunctional connectivity in brain disorders may thus have broad therapeutic potential. Depression and aging are often associated with reduced markers of cell-to-cell connectivity. For instance, stereological studies showed little-to-no decrease in neuron numbers during aging, and, instead, loss of structures mediating connections between pyramidal cells, namely, dendritic length, synapses, and spines [61]. Similarly, postmortem studies in brain samples of subjects affected with stress-related disorders, such as depression [62] [63], or neurological conditions, such as Alzheimer’s disease, report reduced markers of pyramidal cell connectivity. These findings are reported in frontal cortex [61] and hippocampus, two regions critical for cognition. These structural cellular and spine deficits are widely accepted as contributing to cognitive deficits [61] [64] andevidence suggesting that restoring dendritic GABAergic inhibition has both procognitive and neurotrophic effects in models of chronic stress and aging is emerging.

α5-GABA_ARs are essential key elements of dendritic inhibition (Figs. 1, 4). They are located opposite of SST neurons on pyramidal cell dendrites, at high and intermediate levels in the hippocampus and frontal cortex, respectively [65]. Consistent with loss of cell connection in aging and diseases, markers of GABA function restricted to neuron dendrites are decreased in depression, aging and brain disorders. For instance, SST and GABRA5 (coding for the α subunit of GABA_ARs) gene expression decreases by 50% between ages 20 and 80 [66], and reductions of these markers are reported in brain samples of subjects who suffered from depression, schizophrenia, bipolar disorder and Alzheimer’s disease [66] [67]. In contrast, other GABAergic neurons are less reliably affected in aging and disorders, or display disease-specific pathologies, as for PV-neurons in schizophrenia [68]. The reason for the aging and pan-disorder effects on the SST-neuron/α5-GABA_AR pathway is that SST neurons are selectively vulnerable to develop endoplasmic reticulum stress, leading to protein aggregation and reduced cell function [13, 69]. At the behavioral level, genetic and environmental manipulations in rodents show that reducing SST or SST neuron function induces anxiety-/depressive-like phenotypes and cognitive deficits [13, 70, 71], and that augmenting the function of SST neurons has antidepressant-like activity [7, 8]. Consistent with its role in mediating activity-dependent feedforward tuning of excitatory signal, optogenetic inhibition of SST neurons during active cortical processing enhances dendritic excitability [72]. A role in top-down information integration is supported by computational modeling, which suggests SST neurons contribute to excitatory signal to noise ratio and detection rates of excitatory information [73, 74]. Genetic ablation of α5-GABA_ARs was initially shown to improve memory function, but later studies showed complex anxiety and executive function deficits [65] [75] [76]. Both reduction (using chemogenetics [70], optogenetics [71] or pharmacological tools [77]) and sustained increase (as pursuant to anesthesia [78]) in functional α5-GABA_ARs can induce cognitive deficits (Fig. 4), suggesting that different pharmacological approaches to targeting α5-GABA_ARs may be needed depending on levels of available α5-GABAARs.

Fig. 4: Model of a microcircuit relevant for cognition in depression and restoration of cognitive deficits with an α5-PAM.

A. Procognitive effects of α5-PAMs in depression through improved signal-to-noise ratio at the cellular level. Left panel: Dendrite-targeting somatostatin (SST)-positive interneurons inhibit pyramidal neurons via α5-GABA_ARs, resulting in a sparse baseline activity of the pyramidal neurons. This inhibition is mediated by α5-GABA_AR (short green “B” arrows). Stronger excitatory signals are not inhibited by these SST interneurons (long green “S” arrows). This results in a high signal-to-noise ratio (SNR) which is required for optimal cognitive function. Middle panel: In depression, the excitatory activity of the pyramidal neurons is increased, while the inhibitory activity of the SST interneurons is decreased. The result is that the baseline activity of the pyramidal neurons is not as efficiently inhibited as in the healthy brain so that more baseline signals result in activation of the pyramidal neurons (B) and, while there is no change in the transmission of stronger excitatory signals through the pyramidal neurons (S), this results in a lower signal-to-noise-ratio with poor coding of information and cognitive deficits. Right panel: Administration of an α5-PAM strengthens the inhibition of the pyramidal neurons by the SST interneurons, so that incoming baseline activity is inhibited, as in the healthy condition, and not propagated, while stronger excitatory signals are propagated through the pyramidal neurons, resulting in restoration of a high signal-to-noise-ratio and normalized cognitive function. B. Cognition as a function of the activity of α5-GABA_ARs. There is an optimal level of activity of the α5-GABA_AR system for cognitive performance. Both a lower and a higher activity result in cognitive impairments. In depression, the activity of the α5-GABA_AR system is reduced, resulting in cognitive impairment. Administration of an α5-PAM increases the activity level of the α5-GABA_AR system and thus restores learning and memory (L&M) performance.

Modulators of the α5-GABA_A receptor: α5-PAMs and α5-NAMs

Bidirectional pharmacological modulation of α5-GABAARs can lead to antidepressant-like actions. Clinical and pre-clinical evidence suggest that augmenting dendritic inhibition through α5-GABA_AR potentiation may have therapeutic utility in depression and other conditions characterized by reduced dendritic inhibition, such as aging and age-related disorders [65]. Molecules with preferential activity at α5-GABA_ARs were developed [4, 79]. These α5-PAMs (e.g., GL-II-73, Fig. 3) retain the anxiolytic effects of non-specific GABA_AR PAMs, such as benzodiazepines [3] [4] and show novel procognitive effects on spatial working memory deficits in chronic stress and aging models following acute or chronic treatment, not observed with other GABA_AR PAMs or non-specific benzodiazepines (see also Text Box 1) [4, 80, 81]. Notably the procognitive effects of α5-PAMs were observed only in aged rodents displaying memory impairment consistent with a role for α5-GABA_A-Rs in both pathophysiology and therapeutic remediation of age-related cognitive deficits [77]. Chronic α5-PAM treatment was recently shown to also reverse stress and age-related neuronal pathologies (reduced spine density and dendritic complexity) [82]. Together, this suggests that potentiating α5-GABA_ARs under conditions of reduced GABAergic function could display both symptomatic and disease-modifying therapeutic potential in depression and other brain disorders, although the mechanism by which α5-PAMs exert neurotrophic effects are currently not known. Novel α5-PAMs are in late preclinical development for cognitive deficits in depression (DAMONA Pharmaceuticals) and for the mild cognitive impairment to Alzheimer’s disease transition (Agene Bio). Hippocampal α5-GABA_ARs also modulate dopamine neuron activity in the ventral tegmental area, and α5-PAMs were shown to reduce the hyperdopaminergic activity observed in models of schizophrenia and post-traumatic stress disorder [83], suggesting the therapeutic potential of α5-PAMs may extend to psychotic episodes in depression.

Conversely, clinical and pre-clinical evidence suggest that reducing α5-GABA_AR function may have therapeutic applications in conditions characterized by increased tonic inhibition mediated by elevated α5-GABA_AR function including Down syndrome [84], autism [85, 86]), post-anesthesia, and traumatic brain injury [65] [87]. In depression, interest in α5-GABA_AR NAMs (α5-NAM, e.g., MRK-016, Fig. 3) comes on the tail of studies showing that subanesthetic ketamine induces long-lasting antidepressant effects through a mechanism involving transient and selective antagonism of NMDA receptors on GABAergic interneurons followed by transient disinhibition of pyramidal cells, and a burst of synaptic plasticity that leads to lasting restoration of previously compromised GABAergic and glutamatergic synapses [88] [9] [10]. This form of transient disinhibition of pyramidal cells can be reproduced by much more directly reducing dendritic inhibition of pyramidal cells with an α5-NAM [89, 90]. Similar to ketamine, α5-NAMs are though to lead to a burst of glutamatergic activity that is then followed by NMDA receptor-mediated synaptic plasticity and restoration of previously compromised synaptic transmission [89, 90] [91] [92] (Figs. 1, 5). This suggests a rapid antidepressant effect of α5-NAMs with lesser side-effects than ketamine, due to the regionally and neuroanatomically restricted distribution of α5-GABA_ARs and a mechanism involving more direct disinhibition of glutamatergic neurons compared to ketamine [93], However, clinical development of α5-GABA_AR NAMs is currently lacking.

Fig. 5 (KEY FIGURE): Proposed mechanisms of α5-PAMs and α5-NAMs for normalization of defects in GABAergic and glutamatergic deficits in stress-associated psychiatric conditions.

Behavioral stress involves a rapid increase in glutamatergic synaptic transmission and acts over time to result in defects of GABAergic inhibition as evidenced by reduced expression of glutamic acid decarboxylase (GAD) and of the vesicular GABA transporter [107], reduced SST neuron function [5] [108], and reduced NAS synthesis [109]. Chronic stress and deficits in GABAergic synaptic inhibition are followed by homeostatic-like downregulation of AMPA and NMDA-type glutamate receptors and of the vesicular transporter for glutamate [101] [107]. Potentiation of GABAergic inhibition by PAMs appears to allow for a gradual recovery of glutamate receptor expression and eventually also of GABAergic transmission. Very different from GABA_AR PAMs, acute doses of α5-GABA_AR NAMs lead to a rapid and transient decrease of dendritic GABAergic inhibition, a corresponding increase in glutamatergic transmission and activity-induced and NMDAR-mediated plasticity including especially at α5-GABA_AR expressing dendritic spines [92]. This mechanism of α5-GABA_AR NAMs is reminiscent of the rapid antidepressant mechanism of subanesthetic ketamine, which involves selective and transient inhibition of PV and SST GABAergic interneurons followed by disinhibition of glutamatergic neurons and LTP-like NMDA receptor mediated potentiation of both GABAergic and glutamatergic synapses on pyramidal cells, lasting for days ([9] and reviewed in [10]).

This raises the question of how could both α5-PAM and α5-NAM have antidepressant activities? The role of α5-PAM is consistent with α5-GABAARs mediating silencing of pyramidal cells by SST neurons at baseline, whereas these interneurons are largely silent during internally or externally-generated excitatory stimuli, resulting in large signal-to-noise ratio that is critical for information processing (Figs. 4, 5) [72]. Pathologically reduced SST neuron/α5-GABA_AR function leads to reduced dendritic inhibition, higher baseline pyramidal cell activity and reduced signal-to-noise ratio, which are associated with altered information processing and cognitive symptoms. Experimental and computational modeling data show that α5-PAMs restore those deficits and display procognitive efficacy models of depression (Fig. 4) [70, 71, 74]. On the other hand, α5-NAMs further disinhibit pyramidal cells, leading to increased excitatory signal at baseline and reduced temporal and spatial signal-to-noise-ratio [89, 90, 94]. This non-specific increase in excitatory information may be sufficient to support adaptive behavioral responses under stress conditions, resulting in broad antidepressant activity (Fig. 5), but is predicted to lack procognitive efficacy under conditions of already-reduced filtering capacity mediated by SST neurons/α5-GABA_ARs, such as depression [95]. This model is consistent with the modest-to-no procognitive effects of ketamine in the clinic [96] and of α5-NAMs in preclinical models of reduced GABAergic dendritic inhibition [65, 97].

Concluding Remarks and Future Perspectives.

The role of GABA_A receptor subtypes as targets for the treatment of depression is increasingly being recognized, with a NAS being FDA-approved for the treatment of postpartum depression and clinical studies with NAS going on for the treatment of MDD. The mechanisms mediating the persistent antidepressant effects of NAS remains unclear and an avenue of active research with the potential to provide insight into the underlying neurobiology of disease as well as a novel antidepressant mechanism of action. Moreover, our increasing knowledge on the function of α5-GABA_ARs, which are primarily expressed in hippocampus and in cortex, two regions which are linked to the pathophysiology of depression, are providing hope that allosteric modulators selective for this particular receptor subtype will have actions that are distinct from those of clinically used benzodiazepines, which rather non-selectively bind to α1-, α2, α3- and α5-GABA_ARs. The limited spatial distribution of α5-GABA_ARs will target α5-PAMs and α5-NAMs to specific neuronal circuits and thus has the potential to minimize adverse side effect when used as antidepressants and/or for improvement of cognition in depression. An eagerly awaited milestone will be the availability of such compounds for studies in humans to confirm that the actions seen in rodent studies translate to humans (See Outstanding questions).

OUTSTANDING QUESTIONS.

1. Will GABA_AR subtype-selective compounds (e.g., α5-PAMs and α5-NAMs) exert antidepressant-like actions in humans ? Will α5-PAMs exert procgnitive effects in depressed subjects?

2. Do α5-NAMs act on interneurons as well as on principal neurons ?

3. Is there a subset of PV interneurons that expresses δ-GABA_ARs ?

4. Why do classical benzodiazepines do not have antidepressant effects, in contrast to neurosteroids ? Absence of modulation of δ-GABA_ARs on PV neurons ? Modulation of multiple α1-GABA_ARs ?

5. Does the antidepressant efficacy, unlike rapid anxiolytic effects of classical/nonspecific benzodiazepines, critically depend on prolonged treatment which leads to lysosomal degradation of α1-GABA_ARs and dendritic spine engulfment?

HIGHLIGHTS.

A neuroactive steroid targeting δ-GABA_A receptors has been approved for the treatment of peripartum depression; clinical trials of neuroactive steroids for the treatment of major depressive disorder are ongoing.

Preclinical studies indicate that both positive and negative modulators of α5-GABA_ARs have antidepressant-like actions; we discuss a model that explains how bidirectional modulation of the same GABA_A receptor population can result in antidepressant-like action.

A model for how positive allosteric modulation of α5-GABA_ARs may lead to an improved signal to noise ratio in cortical microcircuits and thus restored information processing and cognitive function is discussed.

GLOSSARY

Allosteric modulation

Receptor activity change via a modulatory site distinct from the orthosteric site at which the neurotransmitter binds. Can be positive modulation increasing the action of the primary neurotransmitter, or be negative modulation decreasing the action of the primary neurotransmitter. In the case of the GABA_A receptor, the primary neurotransmitter is GABA, and the modulator can be, e.g., a benzodiazepine or a neuroactive steroid. Positive allosteric modulation typically increases the inhibitory function of GABA on the GABA_A receptor, while negative allosteric modulation typically disinhibits the function of GABA on the GABA_A receptor and may thus lead to a relative activation of the respective postsynaptic neuron

Antidepressants

Compounds that decrease the symptoms of depression in humans and of depression-related behavior in animals, e.g., in rodents, increasing resilience to chronic stress. Many clinically used antidepressants are inhibitors of the reuptake of the neurotransmitter serotonin and/or noradrenaline, which typically take weeks to develop their antidepressant effects, while inhibition of the neurotransmitter reuptake is immediate, indicating that processes secondary to the reuptake inhibition are required for their antidepressant effects. There is a medical need for fast-acting antidepressants. Esketamine, a NMDA receptor antagonist which was FDA-approved for the treatment of depression only in 2019, is a fast-acting antidepressant, and experience with long-term treatment is thus limited

Anxiolytics

Compounds that decrease anxiety in humans and anxiety-like behavior in animals. Benzodiazepines, i.e., compounds binding to their own modulatory site on the GABA_A receptor resulting on positive allosteric modulation of GABA-induced chloride currents, have historically been among the most widely used anxiolytic agents. The anxiolytic-like action of the prototypical benzodiazepine diazepam was shown to be mediated by α2-GABA_AR

Axon initial segment

Initial segment of the axon at the base of the neuron that is important for action potential generation

Extrasynaptic

Non-synaptic sites of neuronal surfaces. Extrasynaptic δ-GABA_ARs and α5-GABA_ARs respond to low ambient extracellular GABA concentrations. Note that unlike in hippocampus, α5-GABA_ARs expressed by pyramidal cells of the frontal cortex are localized predominantly postsynaptically on dendritic spines

GABA_A receptor

Ionotropic neurotransmitter receptor built from five subunits with a central pore with selectivity for chloride ions. This receptor is mediating fast synaptic inhibition in the central nervous system. Upon binding of GABA, chloride typically enters the neuron from the outside leading to hyperpolarization, which makes it more difficult to excite the neuron

HAM-D score

Abbreviation for Hamilton Depression Rating Scale. A questionnaire designed to rate severity of depression including questions assessing mood, insomnia, weight loss, anxiety, feelings of guilt, suicidal ideation, agitation, retardation, and somatic symptoms

Medium spiny neurons

GABAergic inhibitory cells in the dorsal and ventral striatum, where they make up approximately 95% of neurons. There are two subpopulations of medium spiny neurons: cells that express D1 dopamine receptors and cell that express D2 dopamine receptors

Phasic inhibition

Transient response by rapidly desensitizing GABA_ARs

Tonic inhibition

Sustained response of peri- and extrasynaptic, non-desensitizing GABA_ARs to ambient GABA