Cortical GABAergic Dysfunction in Stress and Depression: New Insights for Therapeutic Interventions

1Major depressive disorder (MDD) is a recurring neuropsychiatric illness that is among the leading contributors to social and economic burden, affecting approximately one in five people in the United States. The World Health Organization estimates that MDD will be the second leading cause of disability by 2020. Moreover, MDD induces a high level of personal suffering and suicidality, and also increases the risk of other comorbid medical conditions that can lead to further disability or death.

2The current pharmacological treatment approaches set by national and international guidelines recommend the use of monoaminergic-based drugs, notably, serotonin reuptake inhibitors (SSRI), as first-line medications. Although these drugs provide a significant therapeutic benefit, they still require weeks to months to induce an antidepressant response, and up to 33% of the patients are considered treatment resistant (i.e., fail to respond to two or more antidepressants). Also, the majority of patients experience recurrence after interrupting the treatment, and the adherence of patients to these medicines is relatively low, as they cause undesired side effects, including weight gain, sexual dysfunction, disruption of normal sleep patterns, memory deficits, and others. Moreover, most patients experience a worsening of symptoms in the first weeks of administration, which also contributes to the low treatment adherence.

3Since the late 1980s, when the SSRIs were developed and the monoamine deficiency hypothesis of depression gained more support, there have been no considerable advances in pharmacological treatment for MDD. Considering the relatively low efficacy of monoaminergic drugs, there is an urgent need for development of novel medicaments that address current therapeutic limitations. Indeed, in recent years, a new class of fast-acting, efficacious antidepressants has emerged, showing immense promise in clinical and pre-clinical studies. Among them, ketamine, an NMDA receptor blocker, is the most studied due to its fast (within 2 h of administration) and sustained (up to 7 days) antidepressant effects. These glutamatergic-based drugs have shed light on yet unexplored avenues to explain the pathophysiology of depression, shifting efforts to the discovery and development of new classes of drugs. The accumulating evidence in the literature relating stress, GABA/glutamate deficits in the brain, and MDD, as well as the antidepressant efficacy of drugs that directly interact with these systems, have led to alternative hypotheses to explain the complex neurobiology of affective disorders that overcome inconsistences in monoaminergic theories.

4In this review, we focus on the GABAergic deficit/imbalance hypothesis of MDD, and elaborate this theory in the context of the glutamatergic hypothesis and the monoaminergic and neurotrophic deficits theories. We discuss the dysregulation of GABA neurotransmission and changes in specific GABAergic interneuron subtypes observed in MDD subjects and stress- or genetic-based animal models of depression. Although depression must be seen as a system-wide disorder, a broad range of GABA interneurons that orchestrate excitation–inhibition (E:I) balance in corticolimbic structures are located in the prefrontal cortex (PFC) and several studies point to this region as one of the primary brain regions involved in the pathophysiology of MDD. Indeed, multiple reports show a direct correlation between chronic stress and depression with decreased volume, synaptic atrophy/loss, and altered connectivity in the PFC. For this reason, we discuss in more detail evidence for cortical impairments in depression. Finally, based on recent findings, we will discuss how monoaminergic drugs can also modulate the GABAergic system and will explore novel non-monoaminergic fast-acting pharmacological approaches to treat MDD, including GABAA and GABAB receptor modulators (allosteric modulators, neurosteroids, agonists, and antagonists), NMDA receptors blockers (such as ketamine), and GABAergic interneurons-targeting neuropeptides.

5GABA is the major inhibitory mediator of cortical interneurons in the brain that serves to modulate a wide range of local neurotransmitter systems, most notably, the glutamatergic excitatory counterpart. By targeting specific somatic domains of neighboring glutamatergic principal neurons, GABA interneurons control the E:I balance in the PFC as well as the excitatory output to projecting areas, such as the amygdala, bed nucleus of stria terminalis, and dorsal raphe nucleus. Due to this network orchestration of firing patterns, cortical GABAergic interneurons play an essential role in mediating complex emotional and cognitive processes in the brain. GABA is synthesized from glutamate by glutamate decarboxylase enzymes (GAD65 and GAD67) and stored in vesicles through the vesicular GABA transporter (VGAT1 and 2). The GABAergic signal is terminated by rapid uptake of GABA to glial cells and presynaptic neurons through plasma membrane GABA transporters (GAT1-4).

6One-third of all synapses in the central nervous system (CNS) connects via GABA interneurons, which comprise 20–30% of neocortical neurons and can be classified accordingly to their diverse morphological, electrophysiological, and molecular characteristics. The most common nomenclature segregates interneurons accordingly to their expression profile of neurochemical markers. Three major non-overlapping interneuron groups in the neocortex include those that express the calcium-binding protein parvalbumin (PV), the neuropeptide somatostatin (SST), and the ionotropic serotonin receptor 3. These neurons can further co-express other markers, such as the neuropeptides cholecystokinin (CCK), vasoactive intestinal peptide (VIP), and neuropeptide Y, as well as other calcium-binding proteins, such as calbindin and calretinin. Characterization of distinct subtypes of interneurons helps to identify vulnerable subpopulations that could be relevant to different neuropsychiatric disorders. Specifically, PV and SST interneurons have been extensively studied in stress-related disorders. The most abundant subtype, PV, correspond to 40% of cortical GABA interneurons, and have chandelier or, most commonly, basket cell morphology. PV basket interneurons mainly control firing synchronization and spike timing of neighboring excitatory neurons by providing somatic fast-spiking inhibition to pyramidal cells. On the other hand, 25–30% of cortical interneurons express SST, which consist mainly of Marinotti cells with low-threshold regular spiking properties and an independent high basal firing activity. SST cells make synapses on the dendritic tufts of pyramidal cells but can also inhibit local PV interneurons. These distinct properties and sub-localization confer to both SST and PV interneurons different roles in the cortical microcircuit: while SST cells control the spiking inputs to pyramidal neurons, PV interneurons regulate the spiking outputs from pyramidal neurons to projecting brain areas.

7GABA interneurons express two subtypes of GABA receptors: GABAA and GABAB. The most prominent receptor, GABAA, is a ligand-gated Cl- ion channel (ionotropic) and has been extensively characterized as the target of many psychotropic agents, including benzodiazepines, ethanol, and barbiturates. These receptors are mostly located post-synaptically and control fast synaptic inhibition. GABAA receptors are tetrameric or pentameric in structure that are made up of multiple subunits (6α, 4β, and 3γ) in distinctive combinations that assemble together around a central chloride pore. GABAB receptors are Gi-coupled receptors (metabotropic) and composed of a heterodimer of two homologous subunits: GABAB1 and GABAB2; they are mainly located at pre-synaptic sites, functioning as autoreceptors and inhibiting GABA release, although they can also be found post-synaptically. Given the broad spectrum of neuronal activity controlled by GABA interneurons, it is increasingly clear that imbalance in the GABAergic system and hence in the E:I balance can contribute to the pathophysiology of several psychiatric disorders, including MDD.

8Although the adaptive, innate stress response is essential for body homeostasis and survival, it is widely recognized that responses to sustained, chronic stress can become dysregulated and result in illness and abnormal behaviors. In the brain, chronic stress can produce changes in neurotransmitter function and appropriate neuroplasticity responses that could precipitate depression in humans and, therefore, has been extensively used as a rodent model for depression.

9GABAergic neurons play an important role in the termination of stress response through regulation of the hypothalamus–pituitary–adrenal (HPA) axis, and disruption of this regulatory response contributes to the abnormal effects of chronic stress exposure. For example, chronic stress causes down-regulation of the transmembrane K-Cl cotransporter (KCC2), rendering GABA inputs ineffective to synaptic inhibition of the HPA axis. Moreover, deletion or mutation of the γ2 subunit of GABAA receptors (heterozygous γ2 knockout: γ2+/-) result in reduced GABAA receptor binding and consequent HPA axis hyperactivity, leading to anxiogenic and pro-depressive behaviors. A similar pro-depressive profile is found in α2 knockout mice. Therefore, genetic modifications in GABAA receptors subunits have been increasingly used as animal models to study the influence of GABAergic system in the pathophysiology of anxiety and depression, as well as pharmacological approaches that have therapeutic potential.

10Considering that GABA receptors are highly expressed and GABAergic interneurons are abundant in the PFC and exert an important GABAergic inhibitory control over HPA axis activity it is conceivable that the PFC GABAergic system plays an essential role in emotional processing that is vulnerable to stress. In this respect, acute psychological stress (threat-of-shock condition) decreased approximately 18% of PFC GABA levels relative to a “safe” condition in healthy subjects. In rodents, repeated immobilization stress increased GAD activity and GABA turnover, and reduced GABA levels in the frontal cortex, an effect that was also reported after a 3-week of chronic mild stress (CMS) exposure. In a learned helplessness paradigm, a model of depression, rats that failed to learn the shuttlebox task showed a 25% reduction of GABAA receptors in cortical synaptoneurosomes. Similar results were observed after other types of chronic stress, such as cold and isolation. A recent study reported that 9-weeks CMS exposure resulted in decreased cortical GABAA receptor function, decreased release probability at peri-somatic GABAergic synapses, and reduced postsynaptic GABAB receptor mediated inhibition in anhedonic rats, leading to higher excitability of pyramidal neurons. Also, chronic unpredictable stress (CUS) or CMS exposure decreased innervation and function of GABAergic axons, and levels of GAD67, VGAT, and GAT3 in the PFC. Besides chronic stress exposure of adult animals, there is also evidence that early life stress exposure impacts the GABAergic system later in life in the adult brain. Maternal separation stress and alteration of maternal care in the early (first weeks) postnatal period decreased expression of GABAA receptors in the frontal cortex and other limbic areas, as well as induced anxiety and depressive-like behaviors in adulthood.

11Collectively, these data provide support for the hypothesis that stress causes major changes in the GABAergic system in the PFC that could result in abnormal behavioral and synaptic responses, including dendritic reorganization of interneurons, as well as alterations of electrophysiological respones, that results in defective output from pyramidal neurons to other brain areas. However, even though numerous reports suggest that chronic stress decreases GABA levels and function, other studies have reported opposite effects. Chronic immobilization stress induced a small increase in GABAA receptor binding in the frontal cortex and chronic social defeat stress increased GABAA-containing α5 subunit in the PFC and hippocampus of susceptible mice. Likewise, chronic restraint stress (21 days) induced an increase in GABAA-α1 subunit mRNA expression in the mPFC but not α2, α3, α4, or γ2. Moreover, found that CUS (14 days) increased the frequency of miniature inhibitory postsynaptic currents in the infralimbic area, as well as inhibitory appositions and terminals onto glutamatergic cells, suggesting a stress-induced enhancement of prefrontal inhibition. While difficult to reconcile, it is important to highlight that the results of stress studies may differ depending on the type and duration of the stressor, the GABA receptor subunit analyzed, and the specific subregions of the PFC studied.

12In addition to these preclinical studies, there is accumulating evidence that dysfunction of the GABAergic system is associated with the pathophysiology of MDD and that normalization of GABA is associated with the remission of depressive symptoms. Pioneering studies showed that patients with depression have lower GABA levels in the plasma and the cerebrospinal fluid (CSF). Further studies extended this work through positron emission tomography (PET) imaging methods, which permits a direct and noninvasive quantification of GABA levels in the brain. These studies showed that GABA levels are reduced in unmedicated patients with MDD in several cortical areas, including the prefrontal, occipital, and anterior cingulate (ACC) cortices. Significant reduction in the ratio GABA/creatine + phosphocreatine was found in the ACC of female veterans with suicidal behavior. Likewise, reduced GAD67 protein or gene expression were found in the dorsolateral PFC (dlPFC) and subgenual ACC (SgACC) of depressed patients, although other studies reported no significant effects. Besides GABA levels, several studies reported decreased expression of GABAA receptors subunit genes in MDD cortices, including decreased α1, α3, α4, γ1, β2, and ρ1. However, there were also reports of increased expression of certain subunits, including α5, γ2, β3, and δ in MDD subjects suggesting that different GABAA receptor subunits may play distinct roles in the etiology of MDD.

13Studies regarding the participation of GABAB receptors in the pathophysiology of MDD have received less attention and, therefore, the literature remains unclear. Although GABAB1 and GABAB2 subunits were reported to be decreased in the lateral cerebellum of MDD subjects, no evidence was found for altered GABAB receptor binding in the frontal cortex or hippocampus. However, some variables should be considered as potential confounds in this study, as some of the MDD patients were taking antidepressants at the time of death, and in some cases there was a long post-mortem interval before tissue collection. Despite these limitations, it is notable that preclinical studies report that helpless rats showed decreased expression of GABAB receptors in the frontal cortex, and GABAB1 subunit knockout animals displayed antidepressant-like responses in the forced swim test. Taken together, the results demonstrate that modulation of GABAB receptors induces antidepressant effects (see the section “Conclusion and Future Directions”), and warrant additional studies with more cutting edge tools to further investigate the role of GABAB receptors in depression and treatment response.

14Numerous studies suggest that the reduction in cortical GABA levels observed in MDD subjects and stressed rodents could not only result from decreased levels of the GABA synthetic enzymes GAD65/67, but could also result from a reduction in the density of specific GABA interneuron subpopulations (Table 1). MDD patients show a reduced volume of brain areas such as the PFC and hippocampus. Also, abnormalities in the GABAergic system in cortical areas can also robustly affect other brain regions. For example, low GABA levels in the ACC of MDD patients are associated with reduction in hippocampal volume. Reduced SST gene expression, mRNA, or protein levels were found in the CSF, SgACC, dlPFC, and amygdala of MDD subjects, and in the medial PFC (mPFC) and hippocampus of animals exposed to CUS. Interestingly, female MDD subjects show a more robust reduction in SST expression than males, suggesting that SST could be related to the twofold greater incidence of MDD in females. Although SST expression in MDD subjects and chronically stressed animals is decreased, reported dendritic hypertrophy of Martinotti cells (which includes SST-expressing interneurons) in the mPFC of mice exposed to chronic restraint stress, without changes in spine density.

15The association between SST interneurons and the pathophysiology of MDD has been more directly supported through pharmacological and genetic manipulations in rodents. Mice lacking SST (SST-KO) exhibited increased anxiety- and depressive-like behaviors; elevated basal plasma corticosterone; and reduced BDNF, GAD67, and cortistatin genes expression. Disinhibition of SST interneurons by deletion of GABAA-containing γ2 subunit in SST neurons (SSTCre:γ2f/f) resulted in enhanced inhibitory input to pyramidal cells in the hippocampus and cingulate cortex, and consequently produced anxiolytic- and antidepressant-like phenotypes. Interestingly, acute chemogenetic-induced inhibition of SST interneurons in the mPFC promoted anxiety and depressive-like responses, whereas chronic silencing or chemical ablation had the opposite effect. Furthermore, mice with constitutive, heterozygous deletion of the BDNF gene (BDNF+/-) or with targeted disruption of exon IV (BDNFKIV), causing a reduction or blockade of activity-dependent BDNF expression and depressive-like behaviors, showed reduced SST and NPY gene expression in the cingulate cortex.

16Somatostatin co-localizes with calbindin and NPY, and these neuropeptides have also been implicated in mood disorders. Reductions in calbindin and NPY markers were found in the frontal cortex of MDD patients, as well as in the PFC and hippocampus of rats submitted to different rodent models of depression, including CUS, BDNF mutant mice, or the Flinders sensitive line of rat. However, other studies failed to detect statistical differences in chronic stress models. Additionally, NPY neuronal density was increased in the orbitofrontal cortex and mPFC (IL) of rats that were considered resilient to CUS behavioral effects, but unchanged in anhedonic animals. In depressed patients, the levels of NPY in the CSF or plasma were inversely correlated with anxiety symptoms and with attempted suicide. Other markers, such as CCK and calretinin, seem not to be consistently affected in stress models or MDD, notwithstanding a study reporting decreased immunoreactivity of these peptides in the mPFC of rats submitted to CMS.

17Although several studies have failed to detect robust differences in the expression of PV in rodent models and MDD subjects, there are reports that PV interneurons contribute to regulation of E:I within the PFC that influences emotional responses. One study has reported a decrease in PV gene expression in SgACC post-mortem tissues of depressed patients, and a modest reduction in PV immunoreactivity was found in the orbitofrontal cortex (p = 0.05, Brodmann area 47). In rodents, CMS caused a reduction in PV neuron number in the mPFC of anhedonic rats, whereas SST density was unchanged. A decrease in PV cell number was also found in rats submitted to chronic social isolation. By contrast, there was a report of increased PV expression in the PFC of female mice after 2 weeks of CUS. Also, mice that showed helplessness behavior in response to inescapable stress exposure showed a reduction in excitatory synaptic transmission onto PV interneurons in the mPFC, and selective chemogenetic inactivation of PV cells further increased helplessness responses.

18However, other genetic approaches have reported complex behavioral changes. Knockdown of the Gad1 transcript specifically in PV interneurons (Pvalb/Gad1 Tg) produced a decrease in PV-induced GABAergic activity in multiple brain regions, leading to sensoriomotor gating deficits, increased novelty seeking, and decreased fear expression. Besides the PFC, PV cells located in the hippocampus also play an important role in the modulation of affective behaviors as well as memory. Acute activation of PV interneurons in the dentate gyrus of the hippocampus through DREADD-hM3D-Gq virus and CNO administration (30 min) did not affect depressive-like behavior in the tail suspension test, but produced anxiolytic-like responses and increased fear extinction. Also, CMS or chronic social isolation induced a decrease in PV-immunoreactive cells in the hippocampus, whereas CCK and calbindin expression remained unchanged. Thus, divergence in the responses observed among these studies in the literature highly are related to the brain region studied, the experimental protocol (timing, stress paradigms), and sex differences. Also, it is noteworthy that PV interneurons are under inhibitory control of other interneuron populations, such as SST, making the resultant responses even more complex.

19Taken together, the studies mentioned highlight the complexity of segregating the GABAergic system into different subclasses of interneurons to study the pathophysiology of depression, but demonstrate the importance of understanding how these cells locally interact and integrate diverse neurocircuits that control affective behavioral responses.

20Although extensive efforts have been conducted to develop new therapeutic interventions, the current pharmacological treatment approaches still recommend the use of SSRIs as first-line medications for the treatment of MDD. These drugs, along with other classic antidepressants, such as tricyclics and monoamine oxidase inhibitors, primarily facilitate monoaminergic systems, including 5-HT and norepinephrine. However, the emergence of fast-acting antidepressants, notably ketamine, provide evience for other neurotransmitter systems for the treatment, as well as pathophysiology of MDD. Evidence that normalization of GABA-mediated E:I imbalance in the PFC is a shared mechanism of action between different classes of antidepressants, providing further support for the involvement of GABAergic dysfunction in the etiology of MDD. In this section, we will review the literature showing how first-line monoaminergic antidepressants and rapid-acting agents can influence the GABAergic system. We will discuss how fast-acting antidepressants provide a new understanding of the pathophysiology of depression, leading to connections between the glutamatergic, GABAergic, and neurotrophic hypotheses of depression. Finally, we will discuss the antidepressant potential of agonists, antagonists, or allosteric modulators of GABAA and GABAB receptors, as well as neuropeptides that target specific subpopulations of GABA interneurons and cannabinoid agents.

21The correlation between GABA deficits in the brain, stress, and MDD became more evident with investigations showing that SSRIs, electroconvulsive therapy, and transcranial magnetic stimulation normalize the reduction in cortical and plasmatic GABA levels, as well as in GAD67 expression in MDD subjects and rodents subjected to chronic stress. Besides decreased GABA levels, MDD patients and chronically stressed animals have reduced levels of allopregnanolone (brain and plasma), an endogenous neurosteroid that acts as a GABAA receptor positive allosteric modulator (discussed in more detail below). This deficit was reversed by chronic administration of SSRIs such as fluoxetine; interestingly, in vitro evidence suggests that SSRIs can directly interact with the enzymes involved in neurosteroid synthesis. Chronic treatment with the classic monoaminergic antidepressant desipramine, but not fluoxetine, also normalized the elevated serum corticosterone levels and the pro-depressive behaviors of γ2+/- mice. In this same study, subchronic treatment with desipramine had no effect, suggesting that this drug acts over time to balance GABAergic inhibition deficits. In another study, chronic fluoxetine treatment induced pro-depressive and anxiogenic-like effects in γ2+/- mice, pointing to a requirement of GABAA-containing γ2 subunit in the antidepressant effect of SSRIs. Interestingly, studies reported that fluoxetine can act directly as an allosteric modulator of GABAA receptors.

22Indeed, direct interactions between the GABAergic and serotoninergic systems in the raphe nucleus and cortical regions have been reported. In the PFC, both pyramidal glutamatergic neurons and GABAergic interneurons, notably PV positive cells, express serotoninergic receptors (mainly 5HT1A and 5HT2A). 5HT1A receptor agonists, such as 8-OH-DPAT, have a preferential action on GABA interneurons, resulting in pyramidal neuron disinhibition and enhancement of cell firing in PFC and targeted subcortical structures, such as the ventral tegmental area. On the other hand, the excitability of pyramidal neurons in the mPFC can be inhibited by activation of GABA interneurons through 5HT3 receptors. In this regard, multimodal drugs that are high affinity 5HT3 receptor antagonists, such as vortioxetine, show antidepressant efficacy in clinical studies and have been used as atypical antidepressants to treat MDD. Moreover, the majority of serotoninergic cell bodies in the raphe nucleus express GABAB receptors, which control serotoninergic cell firing as well as the release of monoamines in other brain regions. The antidepressant-like effects of the GABAB antagonist CGP56433A were abolished by prior treatment with a tryptophan hydroxylase inhibitor, which depletes serotonin levels. Additionally, several different monoaminergic antidepressants increase GABAB receptor binding and function in the rat frontal cortex.

23In recent years, the mechanisms underlying the actions of ketamine have been extensively studied because of its rapid (within hours), sustained (up to 7 days), and efficacious effects (effective in patients considered treatment resistant). Related agents, including ketamine stereioisomers and metabolites, have also demonsrated rapid effects in rodent models. These drugs share the ability to influence, directly or indirectly the enhancement of glutamatergic signaling in the brain, promoting post-synaptic AMPA-mediated calcium influx that leads to BDNF release by pyramidal neurons. Extracellular BDNF, in turn, activates TrkB receptors in the membrane, resulting in stimulation of intracellular signaling cascades, including Akt, eukaryotic elongation factor 2 kinase (eEF2K), and mTORC1 that results in synaptic actions that contribute to antidepressant behavioral responses.

24The molecular and cellular mechanisms underlying the rapid enhancement of glutamatergic signaling in the PFC by ketamine have been of particular interest. One hypothesis is that ketamine first targets NMDA receptors specifically located in cortical interneurons, notably, SST and PV subtypes. Because these GABA inhibitory neurons are tonic firing they would be more sensitive to antagonist blockade as tonic activity would remove the Mg2+ block of the NMDA receptor allowing ketamine to enter the channel pore and block further activation of Ca2+ entry. Blockade of GABAergic interneuron firing would thereby decrease GABA release, resulting in disinhibition of excitatory pyramidal neurons and subsequently produce a glutamate burst that could drive activity dependent synaptic plasticity. An alternative hypothesis is that ketamine acts directly on pyramidal neurons to block resting state NMDA receptor activity driven by spontaneous glutamate release that produces synaptic changes via deactivation of eEF2K, resulting in increased synthesis of synaptic proteins. These two theories may not be mutually exclusive, although it is difficult to explain how NMDA receptors would be at resting levels in the presence of a known glutamate burst. In either case, there is an increase in synaptic protein synthesis that underlies long-lasting changes (approximately 1 week) that correspond to the time course for the antidepressant behavioral actions of ketamine. It is also possible that more long-lasting ketamine metabolites contribute to the sustained actions of ketamine. The glutamate burst produced by ketamine appears to be contradictory with evidence of elevated glutamate levels in the brains of MDD subjects, although other studies have reported no significant differences. However, it is important to note that the ketamine-induced burst of glutamate is transient, lasting approximately 1 h, and then levels return to control. Although transient, the glutamate burst results in activity-dependent synaptic changes that are long-lasting.

25Although much of the current work has focused on glutamate synaptic changes in the actions of ketamine, there is also increasing evidence that GABA alterations contribute to the ketamine response, by reestablishing E:I balance in the PFC via homeostatic self-tuning adaptations. This local reorganization could influence microcircuits in target regions by reestablishing firing patterns, and thereby promoting antidepressant effects. This idea is supported by recent evidence that the fast antidepressant effects of ketamine are accompanied by a robust increase in GABA levels in the mPFC of MDD patients and in the ACC of rats subjected to CUS, although another study failed to detect differences in the occipital cortex. One possibility for these discrepancies, in addition to the different cortical subregions analyzed, is the timepoint at which MRS data were collected. Whereas one study was conducted during ketamine infusion, the other was carried out after ketamine; by the end of the infusion, it was shown in the former study that the increase in amino acid responses was no longer detectable.

26Also, a SPECT study reports that S-ketamine administration leads to alterations of GABAA receptor binding in the dorsomedial PFC of healthy subjects. Likewise, studies in cultured murine neurons provide evidence that ketamine increases the activity of extrasynaptic GABAA receptors in the cortex and hippocampus. Combined administration of sub-effective doses of muscimol, a potent and selective agonist of GABAA receptors, and ketamine, produced antidepressant-like effects in female mice. In this same study, the antidepressant effects of ketamine were blocked by the GABAB agonist baclofen, suggesting that the antidepressant actions of ketamine could involve activation of GABAA and blockade of GABAB receptors. In support of glutamatergic and GABAergic interactions in ketamine responses is data showing that a single dose of the ketamine induced antidepressant-like effects and normalized the glutamatergic deficits, including reduced cell surface NMDA and AMPA receptor levels and impaired synaptic function in the hippocampus and mPFC of γ2+/- mice. Moreover, ketamine potentiated pre- and post-synaptic GABAergic synapses selectively in the ACC of these animals. In addition, we have found that a single dose of ketamine increases markers of GABA in the PFC, including increased levels of VGAT, GAD, and gephyrin (, SfN abstract). Thus, although more studies are needed to clarify how ketamine modulates the GABAergic system, the current evidence indicates that ketamine enhances GABA levels/function in the brain as well as GABAA receptors activity.

27In addition, it was reported that fast-acting agents, such as Ro-25-6981, a GluN2B-selective NMDA receptor antagonist, induce antidepressant effects by promoting GABAB receptor surface expression and increasing postsynaptic GABAB-mediated resting L-type calcium channel activity, resulting in an increased intracellular calcium, recruitment of BDNF/mTORC1 pathways, and protein synthesis. Accordingly, MDD and suicide patients have decreased levels of blood and brain BDNF levels and transcripts; reduced BDNF levels in blood were absent in patients taking antidepressants. However, while monoaminergic antidepressants take weeks to modulate neurotrophic factor expression, the rapid elevation in BDNF “release” and signaling by ketamine is shared by other fast-acting agents, such as the non-selective muscarinic receptor antagonist scopolamine, the NMDA receptor modulator GLYX-13 (rapastinel), the ketamine metabolite (2R,6R)-Hydroxynorketamine [(2R,6R)-HNK], and the mGlu2/3 receptor antagonist LY341495 and may explain, at least in part, the fast versus slow response rates of these agents. Moreover, in vitro and in vivo studies suggest that BDNF induces antidepressant-like effects via increased phosphorylation of γ2 subunit, resulting in an increase of GABAA receptor accumulation and stability in the cell surface, and in an enhancement of synaptic inhibition efficacy in the hippocampus and PFC. Thus, the upregulation of GABAB receptors induced by NMDA receptor blockade and consequent activation of BDNF/mTORC1 signaling, as well as the role of BDNF on GABAA receptors phosphorylation and enhancement of GABAergic mIPSC amplitude and frequency, could be a link associating the GABA/glutamate balance deficits to the neurotrophic theory of depression (Figure 1).

28Deletion studies of α2-containing GABAA subunit demonstrate a role of these receptors in depressive behaviors and suggest that agonists or positive modulators could produce antidepressant effects. Benzodiazepines (BZD), GABAA positive allosteric modulators developed in the 1950s, are one of the most widely used thereapeutic agents for the treatment of psychiatric disorders, due to anxiolytic actions at the GABAA-α2 receptor and hypnotic effects at the GABAA-α1. However, the efficacy of classical BZD monotherapy for the treatment of MDD has not been consistently reported, in part due to methodological confounds (i.e., small sample size, variable duration of treatment, and cotreatment with antidepressant agents), as well as high comorbidity with anxiety disorders. However, the triazolobenzodiazepine alprazolam was shown to induce significant antidepressant effects similar to tricyclic drugs in several meta-analysis studies. This antidepressant potential has been attributed to its differential chemical structure formed by a triazol ring fused to the diazepine ring. Recently, selective agonists or positive modulators of GABAA-containing α2/α3 subunits, such as TPA023 and eszopiclone, have been developed and are proposed as potential antidepressants. When co-administered with SSRIs, eszopiclone, a preferential α2/α3-GABAA positive modulator, induced a faster onset of efficacy and greater treatment response, suggesting a synergistic effect. Although it is still unclear if increasing the activity of GABAA receptor is effective in relieving depression symptoms, given that some BZD seem to have greater antidepressant efficacy (i.e., alprazolam versus diazepam), a more thorough understanding of the role of different subunits/subtypes of GABAA receptors could result in the development of more selective and efficacious antidepressant drugs.

29Recently, a new class of fast-acting antidepressants that specifically target the GABAA-containing α5 subunit has emerged. These receptors were shown to be up-regulated in the cortex and hippocampus of depressed patients and stressed mice. Interestingly, preclinical studies have demonstrated that both positive and negative allosteric modulators of GABAA-α5 receptors produce rapid antidepressant-like effects or prevent the behavioral responses induced by chronic stress. This apparent discrepancy could be due to ketamine-like induction of a glutamate burst for negative modulators and GABAA receptor dependent effects of positive modulators. In one study, MRK-016, a negative allosteric modulator of GABAA-α5 receptors and partial inverse agonist of the BZD-binding site, produced a transient increase in electroencephalogram γ power, similar to ketamine. These effects of MRK-016 were abolished by NBQX, an AMPA-type glutamate receptor antagonist, suggesting a fast recruitment of the glutamatergic system and activity-dependent effects. Importantly, perhaps because of the restricted distribution of GABAA-α5 receptors in the brain (mainly in the cortex and hippocampus), MRK-016 did not induce the typical ketamine-like side effects indicative of psychotomimetic or cognitive impairment. Unfortunately, further development of this compound has been discontinued because of low tolerability in elderly subjects. Another GABAA-α5 negative allosteric modulator, L-655,708, also restored alterations in hedonic behaviors induced by chronic stress and the excitatory synaptic strength in the CA1 region of the hippocampus. A single dose of L-655,708 increased the expression of the GluA1 subunit of the AMPA receptor, suggesting that, similar to MRK-016, it produces an indirect potentiation of excitatory synapses. Other GABAA-α5 negative modulators have been tested as cognitive enhancers in clinical trials, but these agents have not been tested for effectiveness in depression.

30Surprisingly, similar antidepressant-like effects were found after acute and chronic enhancement of α5-GABAA activity by a positive modulator; however, this response was restricted to female mice and absent in males, suggesting sex-effects. Since the behavioral sex differences could not be explained by differential pharmacokinetic effects (i.e., different brain concentrations), it is possible that α5-GABAA positive modulators interact with steroid hormones to produce an antidepressant response. Indeed, GABAA receptor subunits, such as δ, are highly sensitive to neurosteroids (see next section) and are differentially modulated across the estrous cycle.

31Endogenous neuroactive ligands synthetized from progesterone, deoxycorticosterone, or testosterone, referred to as neurosteroids interact with a number of targets, most notably GABAA receptors and act as positive or negative allosteric modulators. Numerous preclinical evidence demonstrate that neurosteroids modulate the HPA axis and adaptive responses to stress exposure, and exert anxiolytic or antidepressant effects in rodent models. Specifically, the progesterone-derived neurosteroids, allopregnanolone, a potent positive allosteric modulator of both synaptic and extrasynaptic GABAA receptors, were shown to rapidly modulate BDNF expression in the rat brain, which could explain its fast onset for antidepressant responses. Allopregnanolone has been tested for the treatment of post-partum depression using a formulation developed by SAGE, referred to as brexanolone. The rationale for this study is based on the precipitous drop at the time of delivery of estrogen and progesterone, and consequently a drop in allopregnanolone resulting in a loss of this key positive allosteric modulator of GABAA receptors and a withdrawal like effect. Brexanolone has been delivered intravenously and tested in two Phase II and Phase III trials. Due to its very promising results, it was recently granted a FDA Breakthrough Therapy Designation for the treatment of post-partum depression, and it has also been tested in placebo-controlled Phase III trials for the treatment of MDD. Another compound, SAGE-217, an improved allopregnanolone formula with higher oral bioavailability and longer half-life, which can be used for once daily oral administration, successfully completed a Phase II study for MDD and also received a FDA Breakthrough Therapy Designation. In addition, ganaxolone (Marinus Pharmaceuticals), a neuroactive steroid that acts as a GABAA positive allosteric modulator, was initially developed for the treatment of epilepsy and anxiety, and currently is under Phase II trials for post-partum depression.

32The first prototypical GABAB receptor agonist, bacoflen, was synthetized in 1962 and it was an invaluable pharmacological tool that influenced studies that led to the characterization of GABAB receptors in the 1980s. Years later, with the development of the first GABAB receptor antagonists, phacoflen and saclofen, additional work has lead to the development of compounds that more specifically target GABAB receptors. Given that the GABAB receptor is a heterodimer of two subunits (GABAB1 and GABAB2), that GABAB1 has been reported to have several splice variants (mainly GABAB1A and GABAB1B), and that GABAB receptors are located both pre- and post-synaptically, pharmacological studies targeting these receptors report very challenging and complex results.

33Preclinical studies suggest that GABAB agonists, positive allosteric modulators, and antagonists can produce antidepressant effects; unfortunately, there are very few clinical studies due to the lack of compounds adequate for human testing. In rats, acute administration (i.p.) of baclofen or SKF97541, both GABAB receptors agonists, or CGP7930, a GABAB positive allosteric modulator, induced antidepressant-like effects in the forced swim test, whereas chronic administration increased the escape failures in the learned helplessness test. However, other studies failed to find significant effects for agonists or positive allosteric modulators. In humans, one study reported that bacoflen intensified depressive symptoms in MDD patients. Studies of GABAB antagonists have yielded more consistent results, with a large range of studies showing antidepressant-like effects induced by several different compounds administered either acute- or chronically, such as CGP36742 (also known as SGS742), CGP51176, CGP51176A, CGP56433A, and SCH50911. Interestingly, CGP36742 decreased learned helplessness behavior in rats and increased BDNF and NGF release in the cortex and hippocampus, as well as increased extracellular glutamate and SST in the rat hippocampus. Notably, this is the first GABAB receptor antagonist that underwent clinical trials for cognition-enhancing activity and improved attention in patients with mild cognitive impairment. Thus, considering that: (i) MDD patients in general have an upregulation of GABAB receptors; (ii) genetic deletion of GABAB receptors produce antidepressant-like effects; (iii) GABAB receptors are implicated in the antidepressant actions of fast agents such as ketamine; and (iv) GABAB receptors antagonists offer a promising strategy for the development of novel fast-acting antidepressants, more studies and clinical trials are warranted to identify effective and safe agents.

34Because of the postmortem evidence of selective alterations of GABA interneuron subytpes, it is interesting to speculate on approaches to target the function of specific subpopulations of interneurons based on expression of selective neuropeptides. Preclinical studies demonstrate promising pharmacological evidence for two neuropeptides, NPY and SST, to treat MDD. Intraperitoneal or direct intracerebral (lateral ventricle, hippocampus, amydgala, or septum) infusions of the SST peptide or small molecule SST agonists induce anxiolytic- and antidepressant-like effects in naïve and chronically stressed rodents, as well as exert inhibitory feedback on the HPA axis. There are five (1–5) SST Gi-protein-coupled receptors that are distributed on SST-expressing GABAergic interneurons, and are mainly coupled with induction of K+ conductance leading to neuronal hyperpolarization. The development of selective SST compounds and genetic approaches using specific SST receptor subtypes knockout animals suggest that SST2 receptor, the most abundant subtype in the brain, is a key target receptor for the antidepressant effects of SST. However, the plasma half-life of SST is very short making it unsuitable for clinical trials.

35Two more stable analogs, octreotide and lanreotide, have been tested in clinical studies to treat a wide range of diseases, such as inflammation, tumor growth, and pain; although these drugs show a high affinity to SST2 receptors, they lack selectivity, and induce a broad spectrum of undesired effects in both periphery and CNS. Drugs that act as selective SST2 receptor agonists, such as L-779,976, have never been tested in clinical trials. Thus, given that SST levels were reported to be lower in the brain of MDD patients and stressed rodents, which can be normalized by monoaminergic drugs, and preclinical evidence that SST induces antidepressant-like effects, clinical studies testing the antidepressant potential of selective SST2 receptor analogs with longer half-life merit additional attention.

36Early preclinical studies also provided evidence that central administration of NPY induces anxiolytic- and antidepressant-like effects, as well as promotes stress adaptation and resilience. In the brain, at least four subtypes of Gi-coupled receptors for NPY were identified (Y1, Y2, Y4, and Y5) and the antidepressant-like effects of NYP are suggested to be mediated by Y1R. Interestingly, NPY levels were decreased in treatment-resistant MDD patients and increased after treatment with SSRIs, an effect that was inversely correlated to depression severity. NPY administration in humans also represents a challenge due to short half-life, as well as undesired effects. To overcome this problem, clinical studies have focused on the therapeutic potential of intranasal NPY administration. While a recent randomized dose-ranging study found that intranasal NPY is effective for the treatment of posttraumatic stress disorder with reduced side effects, the antidepressant efficacy of intranasal NPY in MDD patients has not been tested. Given that Y1 agonists or Y2 antagonists also show promise in preclinical studies as antidepressants, further studies of selective drugs as well as intranasal administration of NPY in MDD patients are warranted.

37Endocannabinoids, such as anandamide and 2-arachidonoylglycerol, are pivotal endogenous neuromodulators that control GABA and glutamate release in the brain, mainly through actions on cannabinoid type 1 (CB1) and cannabinoid type 2 (CB2) receptors (although some endocannabinoids can also activate transient receptor potential vanilloid type 1 receptors). CB1 and CB2 are Gi/o-coupled receptors mostly located pre-synaptically, and their activation results in hyperpolarization and reduction of neurotransmitter release. In the neocortex, CB1 receptors are expressed by multiple interneuron subpopulations, mostly in CCK-, but are also found in SST-, calbindin-, and VIP-expressing cells, and at lower levels in glutamatergic neurons. Given that (i) CB1 receptors are highly expressed in cortical and limbic regions, (ii) CB1 receptors are expressed in cortical interneurons and glutamatergic pyramidal cells, thereby modulating both GABA and glutamate release, and (iii) endocannabinoids act as retrograde messengers to mediate depolarization-induced suppression of E (DSE) and I (DSI), it is not surprising that the endocannabinoid system plays an important role in orchestrating cortical E:I balance and controlling stress responses. Indeed, in the mPFC, endocannabinoids contribute to the termination of HPA activity during stress responses through inhibition of GABA release, increasing the outflow of principal interneurons to target regions.

38Cannabinoid agents have shown promise for the treatment of anxiety disorders and depression. The most studied compound for therapeutic use is cannabidiol (CBD), the major non-psychotomimetic substance from Cannabis sativa. Although CBD has a low affinity for CB1 and CB2 receptors, it enhances endocannabinoid neurotransmission by interfering with the function of fatty acid amide hydrolase (FAAH), the enzyme responsible for anandamide degradation. Also, CBD acts as an allosteric modulator of 5HT1A receptors and was recently shown to exert direct actions at GABAA receptors. Accumulating clinical and pre-clinical evidence suggests that acute and chronic administration of CBD induces anxiolytic and antidepressant effects, as well as prevents the behavioral consequences of CUS. Interestingly, the rapid molecular changes induced by CBD are similar to several glutamatergic and GABAergic rapid-acting drugs discussed so far, whereas the long-term effects resemble monoaminergic drugs. For example, a single injection of CBD promotes synaptogenesis in the mPFC and induces rapid and sustained antidepressant effects through increased mTORC1/BDNF signaling, and repeated administration of CBD prevents the decrease in neuronal remodeling/function and hippocampal neurogenesis induced by CUS. In spite of these advances in the mechanism of action of CBD and other cannabinoid agents, there are very few studies that have investigated the role of the GABAergic system. Thus, more causal studies should be performed to determine the subtype of interneuron populations that mediate the anxiolytic and antidepressant effects of cannabinoid drugs, as well as other GABA-related cellular and synaptic mechanisms that could be involved.

39For decades, the monoaminergic deficit hypothesis of depression was the prevalent theoretical basis for studies of the mechanisms underlying the pathophysiology and treatment of depression. However, although increased extracellular monoamines underlies the acute actions of monoamingergic agents, altered monoamine levels alone in forebrain areas are insufficient to explain the molecular and cellular changes underlying the antidepressant actions of these agents. Moreover, there is little consensus evidence that depression results from a deficit of monoamines. Thus, research has focused on neurotransmitter systems and microcircuits that can explain both the efficacy of antidepressant drugs and the etiology of MDD. Given growing consensus that MDD patients have a decrease in GABA levels in the brain and the revolutionary discovery that NMDA receptors antagonists, such as ketamine, can produce rapid and sustained antidepressant responses, efforts have been made to link the deficits in amino acid neurotransmitter systems to the pathophysiology of depression. Notably, the GABA deficit and the imbalance of cortical E:I hypothesis of depression provide a broader understanding of depression, as it offers connections with other important conceptual frameworks, such as altered glutamate and neurotrophic factor deficit hypotheses. With recent advances and new approaches, researchers have renewed enthusiasm for the development of fast-acting antidepressants that target the GABAergic and glutamatergic systems and overcome current therapeutic limitations of monoaminergic drugs. Despite recent advances, significant challenges remain, including development of more selective GABA, NMDA, and neuropeptide receptor agonists, antagonists, and modulators, characterization of optimal doses and treatment schedules, and better design of clinical trials. Moreover, genetic, chemogenetic, and optogenetic approaches should be directed to elucidate the role of specific interneuron subtypes and mechanisms underlying the control of behaviors related to mood and emotion, as well as sex-specific differences involved in these processes, with a view to developing more selective and improved antidepressant treatments.

40MF designed and wrote the manuscript, revised the literature, and prepared the figure and table. RD revised, edited, and approved the manuscript, figure, and table, and contributed in writing the manuscript.

41The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.