GABA System as the Cause and Effect in Early Development

Abstract

GABA is the primary inhibitory neurotransmitter in the adult brain and through its actions on GABAARs, it protects against excitotoxicity and seizure activity, ensures temporal fidelity of neurotransmission, and regulates concerted rhythmic activity of neuronal populations. In the developing brain, the development of GABAergic neurons precedes that of glutamatergic neurons and the GABA system serves as a guide and framework for the development of other brain systems. Despite this early start, the maturation of the GABA system also continues well into the early postnatal period. In this review, we organize evidence around two scenarios based on the essential and protracted nature of GABA system development: 1) disruptions in the development of the GABA system can lead to large scale disruptions in other developmental processes (i.e., GABA as the cause), 2) protracted maturation of this system makes it vulnerable to the effects of developmental insults (i.e., GABA as the effect). While ample evidence supports the importance of GABA/GABAAR system in both scenarios, large gaps in existing knowledge prevent strong mechanistic conclusions.

INTRODUCTION

In the adult brain, gamma aminobutyric acid (GABA) is the primary mediator of fast inhibitory neurotransmission through its actions on GABAA receptors (GABAARs). In addition to protecting the brain against excitotoxicity and seizure activity, activation of GABAARs ensures temporal fidelity of excitatory neurotransmission, and regulates concerted rhythmic activity of large populations of neurons known as oscillatory activity. Considering these fundamental roles, it is not surprising that changes in GABA and GABAAR expression are observed in many pathological conditions, such as neurodevelopmental disorders (Deidda et al., 2014), neurodegenerative disorders (Xu et al., 2020) and neuro-psychiatric conditions (Luscher et al., 2011; Schmidt & Mirnics, 2015; Schur et al., 2016)

In the developing nervous system, GABA and GABAARs play integral roles in neuronal proliferation, differentiation, and migration, as well as synaptogenesis and synaptic pruning. GABA activity and GABAAR expression in early development are in turn affected by genetic (Hernandez et al., 2019; Raud et al., 2009; Steudle et al., 2020), maternal (e.g., drug exposure, maternal stress, see below), and environmental (e.g., postnatal exposure to drugs, see below) factors. As such, it is possible to view the GABAergic changes in the above-listed disorders in two ways: Developmental disruptions in GABA and GABAARs can cause significant changes in neurogenesis and synaptic integration, leading to large-scale changes in brain functions (i.e., change in GABAARs as the “cause”). Depending on the magnitude and exact timing of GABA disruptions, these changes can be extensive, causing significant impairment from an early stage, or can be relatively subtle, devoid of functional effects until the system is hit by insults, such as stress or infection. The second possibility is that disruptions in other systems and processes can induce changes in GABA/GABAAR activity, which can then cause problems in the temporal regulation of neuronal activity, concerted actions of neuronal populations, and coordination of neuronal activity across brain regions, leading to impairment in the behavioral outputs of the affected circuitry (i.e., change in GABAARs as the “effect”). Rather than being mutually exclusive, these two processes likely take place simultaneously and exacerbate each other, with GABAAR changes acting both as the cause and the effect of early developmental disruptions.

This review aims to provide an overview of the involvement of GABAARs in the effects of early developmental insults and developmental disorders by focusing on the above-noted “GABAARs as the cause” and “GABAARs as the effect” perspectives. As data on these questions come from different species as well as in vitro models, inter-species translation is a central question. However, while the specific timing of events in the developmental trajectory might differ across species, most of the developmental processes, including those involving the GABA system, are preserved. As such, we have organized the review around developmental processes rather than the exact developmental timing of events. Where data are available, we provide information about the time of processes on developmental trajectory, noting the species. We start the reviewwith a summary of processes involved in the early development of the GABA system (Please refer to Box 1 for an overview of GABAAR structure and function). While we review GABA and GABAARs in normal development, we note the long-term neurodevelopmental and behavioral consequences of disruptions in normal GABAAR development where data are available (i.e., GABAARs as the cause). In the second part of the review, we focus on a number of common developmental insults, such as early developmental drug exposure, maternal immune activation, and early life stress, the effects of these insults on the developing GABA system and the role of GABAergic changes in the overall observed pathology resulting from these insults (i.e., GABAARs as the effect).

BOX 1. Brief Overview of GABAA Receptors.

GABAARs are the primary mediators of fast inhibitory neurotransmission in the mature nervous system, while in the developing nervous system, they play multiple roles in neurogenesis as well as serving a depolarizing function in newborn neurons (see below). GABAARs are Cys-loop ligand-gated ion channels with a heteropentametic structure (Fig. 1). The binding of GABA to the receptor leads to a conformational change resulting in channel opening, allowing the flow of receptor permeable ions (primarily chloride, but also bicarbonate (Farrant & Kaila, 2007; Fritschy & Panzanelli, 2014; Kaila et al., 1989; Sieghart & Sperk, 2002).

The GABAAR pentamer is assembled through combinations of 19 possible subunits: α(1-6), β(1-3), γ(1-3), δ, ε, θ, π, ρ(1-3) (Olsen & Sieghart, 2008; Sieghart, 2006; Sieghart & Sperk, 2002). While thousands of combinations are possible theoretically, only specific subsets of possible configurations are realized with the most common combination being 2α:2β:1γ or δ (Baumann et al., 2002). The subunits are composed of a long extracellular N-terminal domain, four transmembrane domains (TMs) with an intracellular loop between TM3 and TM4, and a short extracellular C-terminal domain (Miller & Aricescu, 2014; Schofield et al., 1987). The five subunits surround the central ion pore with the TM2 of each subunit around the pore. GABA binding between the α and β subunits leads to conformational changes that eventually lead the TM2 domains to open the ion channel (Bera et al., 2002).

Subunit composition determines receptor kinetics, subcellular location, anatomical distribution, and pharmacological properties (Chua & Chebib, 2017; Engin et al., 2018; Rudolph & Knoflach, 2011; Scott & Aricescu, 2019). As the α subunit determines much of the receptor kinetics and pharmacological properties and differentiates anatomical distribution across the brain, it has become customary to refer to “GABAAR subtypes” based on the α subunit expression (Engin et al., 2018; Goldstein et al., 2002; Nathanson et al., 2019; Nusser et al., 1996; Nyiri et al., 2001; Rudolph & Knoflach, 2011). In the adult brain, α1, α2, and α3GABAARs are typically expressed synaptically, where they mediate fast inhibitory currents which can lead to transient hyperpolarization of the postsynaptic neuron (i.e., phasic inhibition), whereas α4, α5, and α6GABAARs are more often expressed extrasynaptically and respond to low concentrations of ambient or spillover GABA and help set an overall threshold for neuronal excitability; i.e., tonic inhibition (Farrant & Nusser, 2005).

In the adult brain, α1GABAARs are the most common subtype with approximately 60% of GABAARs expressing this subunit with the predominant configuration of α1β2γ2 (Sieghart, 2006; Sieghart & Sperk, 2002). α1GABAARs are enriched in the soma and proximal dendrites of principal neurons but are found in all subcellular compartments. They are also expressed preferentially on GABAergic interneurons, particularly on parvalbumin positive (PV+) interneurons (Klausberger et al., 2002; Saunders et al., 2018). 15-20% of GABAARs express α2GABAARs, which are observed in a subset of somatic and dendritic inhibitory synapses (Engin et al., 2012; Nusser et al., 1996). Importantly, α2GABAARs are enriched at the axon initial segment which is contacted specifically by the PV+ chandelier cells (Nusser et al., 1996; Nyiri et al., 2001; Somogyi et al., 1983), where they play essential roles in regulating action potential transmission and network excitability (Hines et al., 2018). α3GABAARs are often listed as synaptic receptors due to their predominantly synaptic localization in brain areas where they are strongly expressed (Baer et al., 1999; Sassoe-Pognetto et al., 2000; Studer et al., 2006; Uygun et al., 2022) . However, both synaptic and extrasynaptic expression of this subtype has been reported in several brain areas (Corteen et al., 2015; Devor et al., 2001; Marowsky et al., 2012), suggesting more heterogenous subcellular localization in a brain-region and cell-type specific manner. Similarly, α5GABAARs, which have long been considered extrasynaptic receptors, can be expressed synaptically as well as in extrasynaptic and perisynaptic locations (Brady & Jacob, 2015; Caraiscos et al., 2004; Christie & de Bias, 2002; Davenport et al., 2021; Hausrat et al., 2015). Moreover, the flexible lateral movement of α5GABAARs in and out of synaptic sites seems to play an essential role in synaptic plasticity and memory formation (Davenport et al., 2021; Ghafari et al., 2017; Hausrat et al., 2015). Unlike α5GABAARs, α4 and α6GABAARs are expressed exclusively extrasynaptically and mediate tonic inhibition (Belelli et al., 2009; Brickley & Mody, 2012; Chandra et al., 2006; Farrant & Nusser, 2005; Nusser et al., 1996). For more detailed information on the expression and functions of GABAARs in the adult nervous system, we refer the readers to existing reviews (Engin et al., 2018; Mody & Pearce, 2004; Rudolph & Knoflach, 2011).

Developmental Processes in the GABA/GABAAR System

Development of GABAergic neurons and GABAARs

GABAergic interneurons develop earlier than glutamatergic neurons (Please refer to Box 2 for an overview of GABAergic interneuron diversity). They are generated from the telencephalon and migrate via the white matter to the cortex. In rodents, GABA staining is seen as early as E10 in the pial surface of the neo-epithelial wall and most GABA+ cells are seen by E12 in the ventricular proliferative zones and decreases significantly by E16 (Wu & Sun, 2015). As development progresses, these GABAergic cells spread through different cortical zones, including the subventricular and ventricular zones by E13, the subplate, marginal zone, and intermediate zones by E14, and the cortical plate by E16 (Wu & Sun, 2015). Whole-cell voltage clamp recording after addition of GABA evokes inward current in E15 cultured neurons (Wu & Sun, 2015).

BOX 2. GABAergic Interneuron Diversity in the Neocortex.

GABAergic interneurons of the adult neocortex display significant diversity based on their molecular identity, physiological and anatomical characteristics, their anatomical location in cortical lamina, output connectivity (i.e., targeting principal neurons versus other interneurons), and subcellular domains targeted by their axons as shown in Fig. 2 (DeFelipe et al., 2013; Tremblay et al., 2016; Wamsley & Fishell, 2017). Moreover, through differences in receptor expression, different GABAergic interneuron subtypes are differentially sensitive to the effects of neuromodulators, such as acetylcholine, serotonin, and dopamine. This immense diversity in interneuron characteristics, activity, and connectivity leads to precise regulation of circuit activity in the neocortex, supporting the range of complex processing accomplished by this structure.

It is possible to classify most neocortical interneurons into three classes based on the expression of molecular markers: Parvalbumin (PV+), somatostatin (SST+), and serotonin receptor 5-HT3a (5-HT3a+) positive interneurons (Rudy et al., 2011).

Two morphologically distinct group of neurons, basket cells and chandelier neurons, comprise the PV+ interneurons (Druga et al., 2023). Basket cells, thus named due to axonal terminals that form a basket-like pattern around the soma and proximal dendrites of target neurons, are fast spiking interneurons with axonal arbors extending through cortical laminae. They are expressed strongly in cortical layers 4 and 5. Chandelier neurons, on the other hand, have more local axonal arbors that terminate at the axon initial segment of the target neurons, exerting strong inhibitory control over signal propagation. Chandelier neurons are primarily observed in layers 2 and 6 and have a fast-spiking activity pattern (Markram et al., 2004).

While PV+ interneurons target somatic and perisomatic compartments of pyramidal neurons, SST+ interneurons typically target distal dendrites (Riedemann, 2019). Martinotti cells, arising primarily from layer 5, are SST+ interneurons with a low-threshold burst firing pattern with axonal arbor terminating mainly in layer 1 and locally in layer 5. The non-Martinotti SST+ interneurons, found most abundantly in layer 4, have high maximum firing frequency, more similar to PV+ fast-spiking interneurons than Martinotti cells. While Martinotti cells target pyramidal neurons, the primary target of non-Martinotti cells is PV+ interneurons.

Vasoactive intestinal peptide positive (VIP+) interneurons comprise a substantial subpopulation of 5-HT3a+ interneurons (Rudy et al., 2011). VIP+ interneurons are found predominantly in layers 2-3 with their axonal arbors spreading across laminae and target primarily SST+ interneurons. Another 5-HT3a subpopulation, reelin+ interneurons, similarly have translaminar axonal arbors and can be further subdivided into neurogliaform cells, which are found in all cortical layers, and small basket cells of layer 1. Reelin+ interneurons are characterized by a late spiking firing pattern and high input resistance (Tremblay et al., 2016; Wamsley & Fishell, 2017).

While the above classification already suggests several subclasses of interneurons with diverse properties, it should be emphasized that this is an extremely simplified overview (Huang & Paul, 2019). Many of these interneurons’ subtypes have further subclasses that express other markers in addition to the primary marker. For instance, subgroups of SST+ interneurons express the calcium-binding protein calretinin, neuropeptide Y (NPY), and reelin, though reelin+/SST+ interneurons have different properties than the reelin+/5-HT3a+ interneurons.

Another possible classification for cortical interneurons is based on developmental origin. GABAergic interneurons, which eventually integrate into cortical laminae and networks, originate from proliferative zones in the medial (MGE), lateral (LGE), and caudal (CGE) ganglionic eminences, and the preoptic area (POA).

MGE is the most significant source of GABAergic interneurons. In mice, more than half of cortical interneurons are thought to originate from this area. PV+ and SST+ interneurons both originate from the MGE, with PV+ originating from the ventral aspect of MGE and SST+ originating from the dorsal. Transcription factor Nkx2.1 plays essential roles in the generation of PV+ and SST+ interneuron progenitors such that the removal of Nkx2.1 during early development results in a molecular switch from these MGE-derived interneurons to CGE-derived interneurons, leading to developmental seizures (Butt et al., 2008). Transcription factors Lhx5 and Sox6 play essential roles in specification and migration of PV+ and SST+ neurons, such that removal can lead to reduction in the expression of PV+ and SST+ neurons in favor of other interneuron subtypes or an imbalance between levels of PV+ and SST+ interneurons (Azim et al., 2009; Batista-Brito et al., 2009).

The second largest source of interneuron progenitors is CGE, where most 5-HT3a+ interneurons originate from. Progenitors cells from CGE express Coup-tf1 and Coup-tf2. Loss of function in Coup-tf1 leads to reduced numbers of VIP+ interneurons in favor of increased PV+ interneuron density, suggesting the importance of this factor in the proliferation of CGE-derived progenitors (Lodato et al., 2011). Coup-tf2 plays essential roles in the caudal migration of CGE cells (Kanatani et al., 2008).

Some NPY+ and reelin+ interneurons originate from the POA. While, as noted above, NPY and reelin expressing interneurons have some overlap with interneurons expressing other markers, POA-derived cortical neurons do not express PV, SST, calretinin, or VIP (Gelman et al., 2009).Finally, there is some evidence that LGE serves as a source of a small proportion of GABAergic interneurons in the cortex (Anderson et al., 2001). However, the contribution of this area seems relatively small.

In humans, during late gestation, GABAergic neurons move from the white matter to the cortex, and migration ceases within six months after birth. The highest concentration of GABAergic neurons in the cortex is observed at term. GABAergic neuron density in the subplate region remains stable from midgestation through infancy, potentially aiding in the formation of connections between the cortex and thalamus. (Xu et al., 2011).

GABAARs are expressed even before the formation of GABAergic synapses and the spatiotemporal expression of GABAARs are thought to facilitate GABA autocrine and paracrine roles in proliferation and migration. In rodents, GABAAR subunit immunoreactivity is present at E14 (Laurie et al., 1992). In rats, during pregnancy and postpartum, GABAAR levels in the hippocampal formation and cerebral cortex change. There is a decrease in GABA subunit levels during pregnancy, followed by an increase after giving birth. Eventually, the levels return to pre-pregnancy levels by postnatal day 7 (P7) (Licheri et al., 2015).

Administration of agonist or antagonists has demonstrated the important role of GABAARs on cell survival, growth, differentiation. Liu et al. (1997) explored the role of GABAAR function and its impact on cell survival and growth using cell cultures from E14 rat pups. Immunohistochemistry and RT-PCR techniques were employed to confirm the presence of GABAAR subunits (α, β, γ, and δ) and to assess functional activity via chloride ion influx measurements. Competitive GABAAR antagonist bicuculline hindered the survival of 5-HT and TH neurons, while negative GABAAR modulator dieldrin specifically reduced the survival of 5-HT neurons (Liu et al., 1997). Furthermore, cell cultures of striatal neurons from E18 Wistar rats subjected to a 3-day treatment regimen involving various GABA treatments demonstrated a remarkable ability to boost cell survival and stimulate growth beyond the levels observed in control cultures. However, when combined with bicuculline or picrotoxin, the effects of GABA were nullified, resulting in reduced cell survival below untreated levels, indicating the necessity of GABA binding to GABAARs for this effect. Similarly, muscimol, a GABAAR agonist, effectively promoted cell survival while including calcium blockers diminished GABA-induced cell survival. To summarize, these findings collectively underline the pivotal role of GABAAR in promoting cell survival and growth in the rat striatal cortex. Calcium (Ca2+) was also identified as a mediator of these effects (Ikeda et al., 1997). When ICR rat embryos were subjected to GABAAR antagonists, agonists, or taurine at varying intervals between E10 and E14, the subsequent histochemical analysis revealed that GABAAR antagonists impeded the transition of cells from neural progenitors (NPs) to radial glia and interfered with their differentiation into "upper-layer neurons." The study emphasized the collaborative role of taurine and GABAARs in neurogenesis mediation (Tochitani et al., 2021). Earlier research findings indicate that the administration of a GABAAR agonist, phenobarbital, on E10 and E11 influenced the regulation of mitotic spindle orientation in mouse neural progenitors on E12 (Tochitani et al., 2010). Basic fibroblast growth factor (bFGF) was found to elevate the expression of the α1 subunit of the GABAA receptor. However, this effect was not observed for other GABAAR subunits (α2, α3, or α5). Interestingly, GABA or muscimol activation of the GABAARs inhibited the proliferative influence of bFGF, resulting in an increased number of differentiated neurons. These findings underscore the stimulating role of bFGF in cell proliferation and GABAAR expression in developing neocortical progenitors. Additionally, GABA was identified as a feedback regulator, curtailing the process of cell division. Muscimol increases intracellular calcium in human dopaminergic neurons derived from induced pluripotent stem cells during days 5 to 21 of differentiation. This indicates that GABAARs, linked to calcium signaling and phenotype determination, impact neuron differentiation, potentially affecting tyrosine hydroxylase expression and differentiation (Antonov et al., 2016). Exposure to GABAAR inhibitors during pregnancy (in utero) reduces parvalbumin+ (PV+) GABAergic interneurons in the striatum due to impaired survival or differentiation. Additionally, exposure to GABAAR agonists in the cortex of newborn rats may cause abnormal migration and heterotopias (Galanopoulou, 2008).

Behavioral assessments on animals exposed to GABAAR agonists or antagonists during embryonic development unveiled alterations in socialization behavior, potentially linked to autism spectrum disorder (ASD) symptoms (Tochitani et al., 2021). Neonatal cholinergic lesions in rats disrupt locomotor activity and object exploration, increasing movement and decreasing object interaction. Muscimol reduces exploratory behaviors in control and cholinergic-lesioned rats, indicating GABAergic involvement, but these lesions don't alter GABAAR agonist reactivity in juvenile rats. In a spatial open field test, cholinergic-lesioned rats exhibit hyperactivity, particularly evident when objects are introduced, with no such effect during the initial test phase (Scattoni et al., 2003).

The Excitatory/Inhibitory Switch

The polarity of the overall effect of GABA binding to its receptors is determined by the chloride differential between the intracellular and extracellular spaces. The Na+−K+−Cl− cotransporter (NKCC1), the main Cl− importer, is expressed at high levels during early neuronal development, leading to higher intracellular concentrations of Cl−. Opening of GABAARs following GABA binding results in the opening of the central Cl− pore, leading to a net efflux of Cl− and membrane depolarization (Ben-Ari et al., 2007; Murata & Colonnese, 2020; Peerboom & Wierenga, 2021). With the upregulation of the K+-Cl− cotransporter (KCC2), a net Cl− exporter, and the gradual downregulation of NKCC1 expression, the electrochemical gradient along the neuronal membranes shift to lower intracellular Cl− concentrations (Rivera et al., 1999). This leads to a shift in the effect of GABAAR activity from depolarizing to hyperpolarizing. In addition to changing the overall effect of GABAAR activation, this shift towards inhibitory GABA also serves as a signal to stop the migration of cortical GABAergic interneurons (Bortone & Polleux, 2009).

The GABA switch is region and cell-type specific (Murata & Colonnese, 2020). The switch occurs earlier in some brain regions (e.g., sensory cortices) compared to others (e.g., hippocampus) in the developing brain. Specific GABAergic inputs (e.g., long-range GABAergic inputs to parvalbumin + (PV+) interneurons of the dentate gyrus (Bao et al., 2017) and GABAergic inputs onto CA3 interneurons (Banke & McBain, 2006)) remain excitatory even in the adult brain. The switch also occurs earlier in females than males, at least in certain brain regions (Banke & McBain, 2006; Murguia-Castillo et al., 2013; Nunez & McCarthy, 2007) and the expression of KCC2 is regulated by sex hormones throughout development (Galanopoulou, 2008b; Galanopoulou & Moshe, 2003; Nunez & McCarthy, 2007).

While the GABA switch is highly evolutionarily conserved, from reptiles to humans (Ben-Ari, 2002; Hyde et al., 2011), the exact developmental timing of the switch varies across species. KCC2 mRNA can be detected as early as E10 in the neural tube of the mouse embryo (Horn et al., 2010)) and by E16 in the cortical interneurons derived form the medial ganglionic eminence (MGE) (Batista-Brito et al., 2008; Bortone & Polleux, 2009) By birth, KCC2 is detectable in deep layers of cortex, in the telencephalon, as well as the brain stem (Shimizu-Okabe et al., 2002; Uvarov et al., 2009; Wang et al., 2002). KCC2 expression increases drastically in the first two postnatal weeks, reaching adult-like levels by P14 (Awad et al., 2018; Rivera et al., 1999; Wang et al., 2002). Based on observations in the hippocampus, it is often stated that the GABA switch occurs in the second week of life in rodents, but as noted above, the switch occurs at different times in different brain regions, and based on changes in KCC2 expression noted above, it is likely that the functional switch may occur much earlier in some brain regions.

The colloquial labelling of the GABA switch as the “excitatory/inhibitory switch” has also brought on the assumption that GABAAR activation can have inhibitory effects only after the GABA switch is complete. However, it is important to note that depolarization does not always mean excitation, as the functional consequence of depolarization can be shunting inhibition in some networks. Indeed, in some brain regions (e.g., visual cortex), GABA is already inhibitory by P3 (Murata & Colonnese, 2020). In a similar vein of generalization, it is often stated that the GABA switch is already complete by birth in humans, based on the observations that the switch can happen prenatally in some brain regions in nonhuman primates (Khazipov et al., 2001) and that in humans, the KCC2 protein can be detected at moderate levels in some brain regions at birth (Vanhatalo et al., 2005). However, KCC2 levels are upregulated within the first year of life, to reach adult-like levels after age 1, and NKCC1 expression continues to be downregulated for at least the first 2.5 postnatal months in humans (Dzhala et al., 2005), suggesting that the functional switch likely occurs within the first year of life for most brain regions.

Early induction of the GABA switch through experimentally overexpressing KCC2 or through the overexpression of the inwardly-rectifying K+ channel K_ir2.1 resulted in immature morphology in cortical neurons, with shorter dendritic length and reduced dendritic branching compared to controls (Cancedda et al., 2007), suggesting that depolarizing GABA plays an important role in the maturation of cortical neurons. Interestingly, this study found no effect of this manipulation on migration. Delayed or incompletely achieved GABA switch during early development has been implicated in several neurodevelopmental disorders. For instance, GABA switch to hyperpolarization was delayed in the cortex of mice in mouse models of Fragile X Syndrome due to delayed downregulation of NKCC1 (He et al., 2014; Tyzio et al., 2014). A similar delay was observed in rats induced with an autism-like phenotype through in utero exposure to valproate (Tyzio et al., 2014). Indeed, bumetanide, an NKCC1 blocker, alleviated symptoms in individuals with autism spectrum disorders (ASD) or Fragile X Syndrome (Lemonnier et al., 2012; Lemonnier et al., 2013), emphasizing the integral role of the delayed and/or incomplete GABA switch in the development of symptoms in these disorders. A study investigating cells derived from mouse models of 22q11.2 deletion syndrome, which confers risk for ASD and schizophrenia, found similar delays in the GABA switch accompanied by disrupted network activity (Amin et al., 2017). Application of bumetanide ameliorated hyperexcitability and restored wild-type-like synchronized network activity in mutant cells. A more immature NKCC1/KCC2 ratio accompanied by depolarizing effects of GABA is also involved in the phenotype of many seizure disorders and attenuating NKCC1 or promoting KCC2 activity have been shown to reduce seizure susceptibility in animal models as well as leading to symptom reduction in human patients (Moore et al., 2017; Talos et al., 2012; Wang et al., 2018).

Developmental Changes in GABAAR expression

As noted above, GABAARs comprise a heterogeneous group of receptors with subunit composition determining several aspects of receptor properties and distribution. Early brain development is characterized by a number of specific shifts in the expression levels of different GABAAR subtypes. In rat embryos, the α3 subunit is the earliest subunit to be detected, shortly after the first GABAergic neurons are observed. This subunit, whose expression is relatively limited in the adult brain, is expressed at high levels during early neuronal development (Laurie et al., 1992; Liu et al., 1997). α3 expression remains high throughout early development in the diencephalon and telencephalon and is downregulated around the time of the completion of the GABA switch. Extensive α3 expression in early neurodevelopmental stages and a gradual downregulation has been observed in species as diverse as zebrafish (Monesson-Olson et al., 2018), rodents (Laurie et al., 1992), pigs (Miller et al., 2017), and humans (Duncan et al., 2010) suggesting an evolutionarily conserved process. Widespread α2 and α5 expression is also observed shortly after the initial detection of α3 and the expression of these receptors also decrease gradually, disappearing completely from some brain regions and being downregulated in many others in both rodents and nonhuman primates (Awad et al., 2018; Fritschy et al., 1994; Hornung & Fritschy, 1996; Laurie et al., 1992). α1 subunit follows the opposite pattern: Work in nonhuman primates show this subunit appearing perinatally and increasing in expression dramatically after birth (Hendrickson et al., 1994; Hornung & Fritschy, 1996; Meinecke & Rakic, 1992). A similar pattern was observed in the rat brain where α1 was undetectable at E14, started appearing at low levels in several brain regions shortly before birth, followed by a steady postnatal increase leading to adult-like high expression starting around P12 (Fritschy et al., 1994; Laurie et al., 1992; Van Eden et al., 1995). Stark postnatal increase in the expression of the α1 subunit, accompanied by gradual decreases in the expression of α3 and α5 subunits is also observed in human cortex, while the reduction in α2 expression seems to be more delayed and smaller in magnitude (Duncan et al., 2010).

While the effects of disrupting a specific developmental upregulation or downregulation trend in different GABAAR subtypes have not been studied directly, some clues can be gained from GABAAR gene-targeted mice. Studies from global knockout mice or conditional knockout mice generated through Cre-mediated recombination involving Cre transgenes that are expressed early during prenatal development provide an opportunity to indirectly observe the effect of disrupting the developmental changes in the targeted GABAAR. Mouse lines matching this description are often characterized by adaptations in GABAAR subtypes other than the targeted GABAAR, suggesting that halting the normal developmental changes in one GABAAR might lead to a disruption of developmentally-mediated changes in other GABAAR subtypes (Engin et al., 2015; Sur et al., 2001; Suryanarayanan et al., 2011; Witschi et al., 2011). A case for comparison is GABAAR conditional knockout mice generated using cre transgenes that start expression later in postnatal development, after the above-mentioned developmental changes in GABAAR subunit expression have taken place. These mice show no adaptations in GABAARs not targeted by the specific manipulation, even in brain regions where the expression of the targeted subunit is reduced, suggesting the presence of a sensitive window for large-scale compensations to take place (Engin et al., 2016; Engin et al., 2015).

Effects of Developmental Insults on GABAARs

Exposure to Drugs

Several classes of drugs that human infants can get exposed to pre- or perinatally, such as alcohol, anesthetics, benzodiazepines, z-drugs, antiepileptics such as primidone and cenobamate, and neuroactive stereoid analogs such as brexanolone, either interact with GABAARs directly or have active metabolites that do so (Froestl, 2011; see Fig. 1 for a depiction of binding sites of GABAergic drugs on the GABAAR). Other agents, such as cocaine, amphetamines, and opioids, do not interact directly with GABAARs, but they have indirect effects on GABAAR expression and function (Barker & Hines, 2020). Considering that even a single exposure to some of these agents can have long-lasting effects on the GABAARs of the mature nervous system (Jacob et al., 2012; Werner et al., 2016; Zurek et al., 2014), it is not far-fetched to hypothesize that some GABAAR-related changes induced by these agents on the developing nervous system can be permanent and profound. Moreover, as noted in the above sections, GABAergic neurons and GABAARs predate most other systems in the brain and play integral roles in the development of other systems, suggesting that early manipulation of GABAAR activity by drugs can have diffuse effects that permeate brain morphology and physiology beyond the GABA/GABAAR system itself.

Fig 1. The GABAA receptor.

Side (A), top (B), and zoomed-in (C) illustration of the GABAA receptor. Composed of five subunits, GABAA receptors are heterotetrameric Cys-loop ligand-gated ion channels (A). Each subunit is composed of four transmembrane domains (TM), a long extracellular N-terminal domain, and a short extracellular C-terminal domain (B, C).

Early developmental exposure to alcohol

Ethanol modulates the activity of GABAARs by binding their multiple low and high affinity allosteric sites (Hanchar et al., 2006; McCracken et al., 2010; Mihic et al., 1997; Wallner et al., 2014) and modifying their expression, trafficking, and subcellular localization. Chronic ethanol administration influences both receptors and intracellular second messenger pathways of GABA neurons (Kumar et al., 2009; Maas et al., 2005).

Exposure of the fetus to alcohol during pregnancy leads to fetal alcohol spectrum disorders, resulting in sensory, motor, behavioral, and cognitive abnormalities during postnatal development and later in life (Mattson et al., 2011) In the behavioral domain, prenatal alcohol exposure (PAE) affects memory, learning, attention, executive functions, decision making, social behavior, and stress responses (Cullen et al., 2013; Cuzon Carlson et al., 2020; Middleton et al., 2012; Minetti et al., 1996; Ohta et al., 2012; Olmstead et al., 2009; Richardson et al., 2002; Skorput et al., 2015; Subbanna & Basavarajappa, 2022; Wang et al., 2020). The mechanisms and manifestations of PAE were explored in multiple animal models. These studies suggest that the severity of deficits due to PAE is dependent on the gestational stage and/or the timing and degree of alcohol exposure (O'Leary et al., 2010).

PAE induces alteration in neuronal apoptosis that occurs as a natural process of early development (Creeley & Olney, 2013; Ikonomidou et al., 2000; Ikonomidou et al., 2001; Olney, Wozniak, et al., 2002; Subbanna & Basavarajappa, 2022), changes in expression and distribution of GABA neurons (Marguet et al., 2020; Shenoda, 2017; Smiley et al., 2015), inhibits or delays differentiation of functional activity of GABAARs (Subbanna & Basavarajappa, 2022) and changes expression of their subunits (Hsiao et al., 2002), alters modulation of developmental synapses (Wang et al., 2013), excitatory/inhibitory balance and organization of neural networks (Everett et al., 2012; Galindo et al., 2005; Krawczyk et al., 2016), and produces abnormalities in neural plasticity (Basavarajappa & Subbanna, 2023; Hayward et al., 2004).

It was shown that the brain is particularly sensitive to ethanol during synaptogenesis, which occurs in humans during the last trimester of gestation and in rodents during first two postnatal weeks (Sanderson et al., 2009; Wang et al., 2013). During this time PAE inhibits NMDA glutamate transmission and excessively potentiates GABAARs (Olney, Tenkova, et al., 2002), which are excitatory on early stages of development. This potentiation and excessive GABA release (Sanderson et al., 2009) trigger excessive apoptotic neurodegeneration (Ikonomidou et al., 2000; Ikonomidou et al., 2001; Moulder et al., 2002; Olney, Tenkova, et al., 2002), which in turn plays a critical role in establishing neuronal connections and development of neuronal circuits. Possibly, activation of GABAARs influences apoptotic cascade by calcium influx via L-type voltage-gated calcium channels (Davies, 2003; Moulder et al., 2002).

PAE is a significant cause of interneuronopathy (Granato, 2006; Marguet et al., 2020; Skorput et al., 2015; Smiley et al., 2015) that leads to altered development of neural networks and changes in their activity. Following P7 ethanol exposure, the reduction of more than 30% of cortical GABA neurons containing calretinin and parvalbumin (PV+) was observed in adult mice (Smiley et al., 2015). Significant reduction in PV+ cells in cingulate cortex (Moore et al., 1998) and hippocampus (Madden et al., 2020; Moulder et al., 2002) was shown. Generation of calretinin containing interneurons was delayed and their distribution was altered in hippocampal CA3. However, an increase in some interneuron populations was also reported (e.g., PV+ neurons in PFC) (Skorput et al., 2015), possibly due to PAE-induced altered migration of GABA interneurons during brain maturation (Cuzon et al., 2008; Lee et al., 2022; Leger et al., 2020; Shenoda, 2017; Skorput et al., 2015). This altered distribution of GABA interneurons leads to excitatory/inhibitory imbalance in neuronal networks. There is some evidence that the effect of ethanol on NKCC1 might be an integral player in ethanol-induced interneuronopathy. Studies demonstrate that PAE potentiates the depolarizing action of GABA in GABAergic interneurons by increasing the expression of NKCC1. Notably, the treatment of pregnant dams with NKCC1 antagonist bumetanide prevented interneuronopathy in PFC of ethanol exposed offspring (Skorput et al., 2019).

In addition to influencing the expression and migration of GABAergic neurons, PAE also changes their activity and physiological qualities. In the CA3 region of hippocampus in neonatal rats, PAE increased the frequency of giant depolarizing potentials (Galindo & Valenzuela, 2006; Galindo et al., 2005) by increasing GABA release at interneuron-to-pyramidal neuron synapses and activating the postsynaptic GABAARs. In adult PAE offspring, activity of the hippocampal network was characterized by pathological sharp waves, and excitatory/inhibitory synaptic imbalance was detected both in hippocampus and cortex (Delatour et al., 2020; Krawczyk et al., 2016). PAE delays development of GABAergic miniature postsynaptic currents in hippocampal CA3 (Everett et al., 2012) and in septal neurons (DuBois et al., 2004). Also, it increases post-tetanic potentiation of electrically stimulated GABA release in hippocampus that may contribute to alterations in synaptic plasticity (Hayward et al., 2004). Reduced synaptic plasticity-related protein expression and impaired LTP and LTD in PAE offspring were also observed (Hsiao et al., 2002; Richardson et al., 2002; Subbanna & Basavarajappa, 2022).

The effect of alcohol on GABA signaling during development is not uniform and is area- and cell-type dependent. Thus, a significant decrease of GABA levels was reported in the thalamus, pons, cerebellum and hippocampus, no change in posterior colliculus, occipital cortex, temporal cortex, hypothalamus, septum or striatum and a significant increase in frontal cortex, olfactory bulbs, anterior colliculus and amygdala of 3-week-old rat pups prenatally exposed to alcohol (Ledig et al., 1988). GABAARs in prefrontal cortex, hippocampus, medial septum/diagonal band of Broca, hypothalamic areas, cerebellum were shown to be particularly sensitive to PAE (Everett et al., 2012; Hsiao et al., 1998; Hsiao et al., 1999; Iqbal et al., 2004; Krawczyk et al., 2016; Skorput et al., 2015; Volgin, 2008) and neurons of certain cortical layers were particularly sensitive to PAE during period of synaptogenesis (Sanderson et al., 2009).

PAE-induced alterations in the expression of different GABAAR subunits in different brain areas occur not only at early stages of the development but is protracted to adolescence and adulthood and is area specific. For instance, PAE on gestational day 8 (G8) significantly decreased expression of GABAAR ɑ5 subunit in the developing rat brain, while in adult rats ɑ5 expression increased (Toso et al., 2006). On the other hand, PAE upregulated expression of β2/3-subunit in the cortex and hippocampus of the young adult guinea pigs (Iqbal et al., 2004). Rats with chronic ethanol exposure during P2–12 had increased δ-subunit expression at P28 in cerebellum that correlated with delay in development of motor functions (air righting reflex), while basal ganglia were unaffected (Diaz, Vollmer, et al., 2014).

GABAARs altered in PAE offspring cause changes in the synaptic activity of other transmitters, (e.g., increases the level of dopamine and glutamate in the basolateral amygdala) (Baculis et al., 2015; Diaz, Jotty, et al., 2014; Zhou et al., 2010), which further contributes to behavior abnormalities (Baculis et al., 2015; Diaz, Jotty, et al., 2014; Hayward et al., 2004; Iqbal et al., 2004; Toso et al., 2006; Zhou et al., 2010). Besides, altered GABA signaling was shown to be the key mechanism of increased vulnerability to benzodiazepines and drugs of abuse in adolescent and adult PAE offspring (Barbier et al., 2009; Barbier et al., 2008; Osborn et al., 1998).

Early developmental exposure to anesthetics

Most general anesthetics directly bind to GABAARs at allosteric sites and achieve their therapeutic effects through positive modulation of different GABAAR subsets (Bonin & Orser, 2008; Drexler et al., 2011; Garcia et al., 2010; Harrison et al., 1993; Jenkins et al., 2002; Nishikawa & Harrison, 2003; Sonner et al., 2007; Straub et al., 2013; Werner et al., 2011). Each year, over a million infants and toddlers in the US alone receive general anesthetics for surgical procedures. While studies of early anesthetic exposure in experimental animals revealed highly deleterious long-term effects of early anesthetic exposure , studies in humans have been less conclusive, with most studies showing effects only in cases of longer and multiple exposures (Davidson et al., 2016; Feng et al., 2020; Hu et al., 2017; Kalkman et al., 2009; McCann et al., 2019; O'Leary et al., 2016; Sprung et al., 2012; Warner et al., 2018). However, many human studies have focused on motor and cognitive domains, with the emotional and social domains of development still relatively under-studied (Walkden et al., 2019). Moreover, while animal studies report pathology and increased vulnerability to further insults in adult animals that were exposed postnatally, human studies mostly focus on the shorter-term effects in childhood and adolescence due to the difficulties associated with decades-long longitudinal studies. Thus, the possibility of yet uncovered pathology and increased vulnerability in humans remains.

Exposure to anesthetics can lead to immediate (Chang et al., 2023; Li et al., 2015) and long-lasting (Tan et al., 2011; Zurek et al., 2014) changes in GABAAR expression and surface trafficking. Despite this knowledge, the question of whether early developmental exposure to anesthetics might lead to permanent changes in GABAAR expression and/or trafficking has not been directly studied. Still, studies report reduced response to diazepam in adult mice that were exposed to general anesthetics postnatally (Fredriksson et al., 2007), suggesting some lasting changes in the expression, trafficking, and/or function of GABAARs containing the ɑ1, ɑ2, ɑ3, and ɑ5 subunits (see below).

Early developmental exposure to benzodiazepines and related drugs

Benzodiazepines are positive allosteric modulators that bind to a subset of GABAARs, namely, those containing the α1, α2, α3, and α5 subunits. A group of chemically dissimilar drugs that provide allosteric modulation through the benzodiazepine site, colloquially referred to as Z-drugs, are pharmacodynamically similar but often exhibit some selectivity for the α1 subunit. Benzodiazepines and related drugs are used by pregnant and nursing women to control anxiety symptoms and seizures (Hanley & Mintzes, 2014; Yonkers et al., 2017). They are also used as sedatives, anti-seizure drugs, and anxiolytics in all age groups, including neonates and young children (Ng et al., 2002; O'Sullivan et al., 2015; Witek et al., 2005).

Exposure to benzodiazepines or Z-drugs during the early or late gestational period, or during the early postnatal period through mother’s milk leads to significant behavioral changes in rodents, including learning and memory deficits, changes in exploratory and social behavior (Chesley et al., 1991; File, 1986, 1987; Frieder et al., 1984; Lauer et al., 1987; Mikulecka, Subrt, Parizkova, et al., 2014; Mikulecka, Subrt, Stuchlik, et al., 2014; Xu et al., 2015) Similar behavioral abnormalities in the cognitive and emotional domains were reported in animals postnatally exposed to the neurosteroid allopregnanolone, a positive allosteric modulator of GABAARs with a binding site distinct from the benzodiazepine site (Darbra & Pallares, 2009; Modol et al., 2014). Barbiturates, which are nonselective positive allosteric GABAAR modulators (and at high doses, agonists), administered during the second postnatal week also caused long-term impairments in sensory gating, social behavior, and learning and memory (Forcelli et al., 2012).

These changes spanning multiple behavioral domains and lasting into adulthood suggest that early exposure to benzodiazepines and related drugs can lead to widespread alterations in brain function. Indeed, at sedative doses, benzodiazepines diazepam and clonazepam and barbiturates phenobarbital and pentobarbital caused widespread apoptotic neurodegeneration in the brains of P7 rats (Bittigau et al., 2002). Early postnatal exposure to the benzodiazepine flunitrazepam or the neurosteroid allopregnanolone significantly impacted neuronal migration in the cortex, causing an accumulation of parvalbumin positive interneurons in the deep layers (Grobin et al., 2003; Grobin et al., 2004). At the level of GABAARs, early work showed that exposure to benzodiazepines in the third postnatal week resulted in increased sensitivity to convulsants and reduced diazepam stimulation of GABA-mediated chloride uptake in male offspring, suggesting changes in GABAAR expression and composition (Bitran et al., 1991; Kellogg et al., 1991). In rats given clonazepam between P7 and P11, there was an initial increase in α2 expression in the hippocampus and the cortex that persisted for at least a week, accompanied by a decrease in the expression of the δ subunit (Kubova et al., 2020). However, these changes did not persist into adulthood. Still, flunitrazepam binding in parts of the hippocampus, some amygdalar nuclei, and the periaqueductal gray was reduced in adulthood, suggesting that some changes in other benzodiazepine-sensitive GABAAR subunits must have lasted into adulthood.

Early developmental exposure to other drugs

Drugs of abuse (cocaine, amphetamines and opioids) and their metabolites act on GABAARs indirectly and influence GABAergic system via multiple extra- and intracellular signaling pathways.

Influence of psychoactive drugs and their derivatives on GABA system and GABA inhibitory function was discussed in a number of articles and reviews (Barker & Hines, 2020; Hosseinzadeh Sahafi et al., 2023; Jiao et al., 2015; Vaughan et al., 1997; Ye & Ren, 2006), however, the data on their perinatal effect on GABAergic system are scarce. It was shown that maternal exposure to drugs of abuse affects structure, distribution, and functions of the developing GABA neurons and GABAARs, leading to excitatory/inhibitory imbalance and subsequent behavioral deficits later in life.

Prenatal (E8-E15) exposure to cocaine led to altered tangential migration of GABA neurons from the ganglionic eminence to the cerebral cortex of mice through BDNF mechanisms (McCarthy et al., 2011), that produced a deficit of GABA neurons in the embryonic cerebral wall (Crandall et al., 2004). The alterations in distribution were shown to be cell- and region-specific (Crandall et al., 2004; Stanwood, Washington, & Levitt, 2001; Stanwood, Washington, Shumsky, et al., 2001; Wang et al., 1995). For instance, GABA neuron distribution in cocaine-exposed offspring was affected in both medial prefrontal and somatosensory cortices but not in the olfactory bulb (Crandall et al., 2004).

Another sensitive period for prenatal cocaine exposure was reported in rabbits at E16-E25 (Stanwood, Washington, & Levitt, 2001; Stanwood, Washington, Shumsky, et al., 2001), the time of peak corticogenesis and the onset of D1 dopamine receptor expression, which in turn influenced the development of GABA neurons. The authors emphasized that following this exposure, most changes in GABA system of juvenile rabbits (P20) occurred in the cortical regions containing a high density of tyrosine hydroxylase-immunoreactive (TH+) fibers (dopaminergic input), including medial prefrontal, entorhinal, and piriform cortices, while the areas with low TH expression (primary somatosensory, auditory and motor cortices) did not show changes in cortical structure (Stanwood, Washington, Shumsky, et al., 2001). Prenatal cocaine specifically increased the number of PV+ neurons and dendrites in anterior cingulate cortex (Wang et al., 1996) and decreased both the number of PV+ neurons in the mPFC of adult P60 mice (McCarthy & Bhide, 2012) and the length and density of GABA axo-axonic contacts (Morrow et al., 2003).

Altered distribution of cortical GABA neurons was also observed in the offspring of opioid (buprenorphine) exposed mothers both in human fetal tissue culture and in the tissue of adult rats (Nieto-Estevez et al., 2022). Also, region specific re-distribution of GABA, and particularly PV+ neurons was detected in mice prenatally exposed to morphine (Maharajan et al., 2000).

Exposure to certain drugs of abuse elicited changes in selected subunits of GABAARs, GABA synthesizing enzymes, and proteins specific for GABA neurons (Bigl et al., 1982; Gamble et al., 2022; Lum et al., 2021; Shumsky et al., 2002). Noticeably, the changes observed in juvenile (P15) mice can be smoothed or even inverted in adult (P20-P60) animals suggesting developmental compensatory processes (Bigl et al., 1982; Shumsky et al., 2002). For instance, in juvenile (P20) rabbits prenatally exposed to cocaine, a significant increase in the ɑ1 subunit in ACC lamina III and a reduction in β2 subunit in lamina II was observed. However, no differences from control were observed at P60 (Shumsky et al., 2002). Also, exposure to d-amphetamine and amitriptyline from E7 to birth initiated decrease of GABA level in the whole brain on P15 as compared to control and its increase on P21 and in adult rats. The rats had enhanced locomotor activity at P15, which decreased relative to the age-matched controls at P21 (Bigl et al., 1982).

Structural changes initiated by prenatal exposure to drugs may underlie alterations in the inhibitory function of GABAARs. In in vitro studies on adult rabbits prenatally exposed to cocaine, membrane potentials produced by activation of postsynaptic GABAARs in hippocampal CA1 were significantly enhanced (Little & Teyler, 1998). On the contrary, juvenile (Niu et al., 2009) and adult rats(Tan et al., 2015) that were prenatally exposed to morphine had decreased GABAergic inhibition and impaired hippocampal synaptic plasticity. Also, females prenatally exposed to methadone showed decreased excitability and increased inhibition of hippocampal dentate granule cells in patch-clamp studies, which correlated with impairment in spatial working and recognition memory (Gamble et al., 2022).

The excitatory/inhibitory imbalance in offspring of drug-exposed mothers may contribute to deficits in motor, autonomic, cognitive, and behavioral functions: impaired motor skills, SIDS and respiratory complications, impaired attention, cognition, working memory, learning, cognitive flexibility and problem solving, anxiety, depression and less sociability, altered responsiveness to pain and to future drug reward (Banz et al., 2016; Bigl et al., 1982; Chiriboga et al., 1999; Ellis et al., 1993; Gholami et al., 2020; Grecco & Atwood, 2020; Grecco et al., 2022; Jablonski et al., 2016; Kaltenbach et al., 2018; Lum et al., 2021; Macuchova & Slamberova, 2016; Malanga & Kosofsky, 1999, 2003; Riley et al., 2015; Ross et al., 2015; Salas-Ramirez et al., 2010; Slamberova, 2019; Tan et al., 2015).

Maternal Immune Activation

Maternal immune activation (MIA) during pregnancy leads to structural and functional changes in GABAergic system, with these GABAergic changes being one of the mechanisms of the development of neuropsychiatric disorders, including anxiety disorders, schizophrenia and autism later in life. Offspring of mothers with MIA exhibit changes in neuronal excitability along with behavioral abnormalities.

The two most common preclinical models of MIA are the injection of synthetic viral RNA polyinosinic: polycytidylic acid (Poly(I:C)) (Smith et al., 2007; Woods et al., 2021) and bacterial endotoxin lipopolysaccharide (LPS, (Solek et al., 2018; Vojtechova et al., 2021) to evoke antiviral or antibacterial immune responses, respectively, or exposure to individual cytokines (see Fig. 3 for a depiction of the MIA model and a summary of affected GABAergic processes. Poly(I:C) interacts with toll-like receptor (TLR) 3 and initiates the production of antiviral interferons and inflammatory cytokines (Reisinger et al., 2015). LPS binds to TLR4, activating parallel and overlapping inflammatory signaling to that of Poly(I:C) (Reisinger et al., 2015). Brain-region-specific elevation of IL-1ß, IL-6, and TNF-α was shown in the fetal brain following MIA in rodents (Boksa, 2010; Boksa et al., 2016; Patterson, 2009, Garay et al., 2013), and the effects of MIA on GABAAR trafficking via IL-6 (Smith et al., 2007) and TNF (Konefal & Stellwagen, 2017) were demonstrated. Even a single injection of Poly(I:C) to pregnant mice affects GABAergic system in offspring that exhibited multiple behavioral deficits in adulthood. MIA exerts both immediate effects, e.g., on the proliferation of GABA neuron precursors, and long-lasting effects, e.g., on migration of neuroblasts, and on neuronal differentiation and maturation of GABA neurons. MIA during two critical periods of GABA development, early (GD9-11), a period of GABAergic specification and differentiation, and late (GD13-15), a period of GABAergic migration, lead to GABAergic deficits (Nakamura et al., 2022), although the time when MIA occurred determines the specific inflammation-mediated brain and behavioral pathology (Meyer et al., 2006; Meyer et al., 2008). Prenatal PolyI:C was shown to reduce the density of GABA interneuron precursors in cortex and hippocampus with differential developmental impact on their subtypes (Vasistha et al., 2020), e.g. having a layer-specific effect on the distribution of PV+ and SST+ interneurons in the cortex (Vasistha et al., 2020), and on the morphology of PV+ neurons, decreasing the complexity of their dendritic tree.

Fig 3. Maternal immune activation.

Schematics of the maternal immune activation (MIA) model and a summary of the effects of MIA on GABAergic developmental processes.

MIA-initiated elevation of inflammatory signaling and changes in GABAA neurotransmission were shown to involve various mechanisms and are brain region-specific. MIA in mice prevents or delays GABA developmental excitatory/inhibitory shift in hippocampal CA3 (Corradini et al., 2018; Fernandez et al., 2019) via downregulation of KCC2 transcription, which results in higher excitability and susceptibility to seizures into adulthood. Also, MIA at G15 causes altered expression of certain GABAAR subunits, GABA-synthetizing enzymes, and GABA-transporters, leading to behavioral deficits in adolescent and adult offspring P35-60. In the ventral dentate gyrus (DG) and basolateral amygdala, MIA evoked significant increase in GABAAR subunit ɑ2 in adult rats (Nyffeler et al., 2006), although in controls it is gradually reduced during development. Decrease of Gabrb1 in the hippocampus (Su et al., 2022) and decrease of Gabrb3, GAD65 and GAD67, vesicular GABA transporter (VGAT), α2, α3, α4, and α5 GABAAR subunits in PFC of adult animals were shown (Richetto et al., 2014), while in nucleus accumbens increased Gabrb3 was detected (Su et al., 2022). In the VTA, MIA at G15 in rats significantly decreased mRNA expression of β2/3 subunits GABAARs and GAD2, while the atypical antipsychotic risperidone injected to juvenile rats at P35 for 25 days partially reversed the decreased GAD2 expression (Chen et al., 2022).

MIA was shown to reduce PV expression in cortex (Vasistha et al., 2020) and hippocampus (Vojtechova et al., 2021), which decreased GABAergic transmission onto pyramidal neurons. Reduced amount of PV+ interneurons in the somatosensory cortex in parallel with development of autism-like behavior was observed. IL-17a and TNF-α dependent downregulation of microglial Gpr56 expression in fetal brain was proposed as one mechanism that mediates these GABAergic changes: Mice with conditional knockout of microglial Gpr56 had PV+ interneuron deficiency and autism-like behaviors in offspring, while genetically restoring Gpr56 expression in microglia ameliorated these structural and behavioral abnormalities (Yu et al., 2022). However, other work showed that exposure to Poly(I:C) on G9 initiated functional deficits in synaptic release probability by PV+ interneurons in mPFC, but without changes in the total number of PV+ neurons (Canetta et al., 2016). Possibly, the character of changes is dependent on the specific cortical region and the timing of the exposure, however, some change in the morphology, expression, and/or activity of the PV+ interneuron population seems to be a common finding of MIA studies.

MIA changes electrophysiological properties of GABA neurons and alters excitatory-inhibitory network balance. Reduction in GABAergic inhibitory input and hyperexcitation of layer 2/3 pyramidal neurons in anterior cingulate cortex slices was observed in PolyI:C treated offspring (Okamoto et al., 2018). In mature DG neurons decrease in frequency of miniature inhibitory postsynaptic currents (mIPSCs), with no change in mEPSCs (Zhang & van Praag, 2015) was reported. However, in the somatosensory cortex and in CA1 hippocampal neurons, MIA induced decrease in mEPSC frequency in pyramidal neurons (Coiro et al., 2015), with an accompanying increase in their amplitude (Ito et al., 2010). Maximum firing frequency of PV+ interneurons was reduced in MIA animals, while it was described to increase during postnatal development, thus suggesting a maturational delay of PV+ interneurons due to maternal inflammation. Also, offspring of MIA mothers had altered hippocampal ripples (Gao et al., 2019) and other activity changes (Dickerson & Bilkey, 2013).

In adolescent and adult animal behavior studies, MIA produces impaired motor coordination, sensorimotor gating (Kobayashi et al., 2021; Nakagawa et al., 2020; Nyffeler et al., 2006; Su et al., 2022; Zhang & van Praag, 2015), anxiety-like and depression-like behaviors in open field, elevated plus maze, and forced swimming tests (Su et al., 2022) deficit in memory and learning (Richetto et al., 2014; Samuelsson et al., 2006) and social interaction (Machado et al., 2015; Okamoto et al., 2018; Su et al., 2022). Noticeably, local injection of the benzodiazepine clonazepam into the anterior cingulate cortex of PolyI:C mice restored their social behavior (Okamoto et al., 2018) and counteracted other behavioral abnormalities (Yang et al., 2021). MIA-mediated disruption of hypothalamic neurocircuits of maternal care behavior also involves GABA signaling, specifically in the medial preoptic area – VTA pathway (Zambon et al., 2022).

Overall, there is significant evidence of disruptions in the GABA/GABAAR system of offspring from MIA mothers, mostly in the direction of reduced expression of GABAergic neurons and GABAARs, and delayed or stunted maturation of the GABA/GABAAR system. As MIA leads to large-scale behavioral abnormalities and diffuse changes in many brain systems, establishing causality between GABAergic changes and behavioral abnormalities is a missing link in most current studies. However, as noted above, there is some evidence that reversing or reducing GABAergic deficits can lead to the amelioration of specific behavioral deficits. Future studies targeting specific GABAergic deficits in specific brain areas and circuits and linking these to specific behavioral outcomes can help clarify the contribution of GABA/GABAAR system deficits to the behavioral pathology of MIA.

Early Life Stress

Early life stress (ELS) refers to experiences of adversity or trauma that occur during critical periods of brain development, particularly during the prenatal period, infancy, and childhood. These experiences can include physical or emotional abuse, neglect, exposure to violence, and other forms of trauma that can disrupt normal brain development and lead to lasting changes in behavior, cognition, and mental health (Hedges & Woon, 2011; Pechtel & Pizzagalli, 2011). Early life stress has been associated with a wide range of negative outcomes, including increased risk for mood disorders (Kaufman & Charney, 2000; Saleh et al., 2017), anxiety and trauma related disorders (Pervanidou et al., 2020), substance abuse (Enoch, 2011), and other negative mental health outcomes (Edwards et al., 2003).

In rodents, ELS is often modeled through maternal separation approaches, that involve separating the dam and the pups, although model parameters, such as the specific timing and duration of the separation procedures in the postnatal period, the length and frequency of the separation vary across laboratories (Nishi et al., 2014; Pryce et al., 2005; Tractenberg et al., 2016; Vetulani, 2013). Another common approach is to limit the availability of bedding and nesting material in the home cage for a limited period (e.g., 1 week) during the early postnatal days, which leads to fragmented maternal care (Walker et al., 2017). Others have used combinations of limited bedding/nesting, separation, and maternal stress to reduce the quality of maternal care and create early developmental stress in the pups (Gapp et al., 2014; Pena et al., 2014).

While findings from all early life stress models are reported indiscriminately here, it should be noted that the nature, severity, and timing of stress varies from one approach to the next and the effects on GABA/GABAAR system are also likely to be nonuniform across models, although no such systematic comparisons have been made to date. Another methodological concern is the difficulty of finding equivalent time periods to represent this early developmental time period across species. For instance, mice are born in a neurodevelopmentally immature state compared to humans and it is customary to assume a P9 mouse as the equivalent of a human newborn. However, as noted in the above sections, the GABA system goes through comparable postnatal developmental changes in humans and model species. For instance, a gradual reduction in the expression of a5GABAARs occurs through the first 5 years of life in humans and nonhuman primates (Datta et al., 2015; Duncan et al., 2010),and in the first 10 postnatal days in rats (Laurie et al., 1992). Gradual increase in a1GABAAR expression and a gradual decrease in a2GABAAR expression is also observed in early postnatal life in rodents, nonhuman primates, and humans alike (Datta et al., 2015; Duncan et al., 2010; Fritschy et al., 1994; Hornung & Fritschy, 1996; Laurie et al., 1992). The GABA switch, while earlier in humans and other primates, also takes place at some point in early postnatal period. As a comparison, a study that evaluated postnatal changes in cortical GABA and glutamate receptor expression in rhesus monkeys found that there were only small changes in glutamate receptors during this time of dynamic changes in GABAARs (Datta et al., 2015).

Based on these findings, it is tempting to hypothesize that the specific protracted maturation of the GABA/GABAAR system could make this system particularly vulnerable to the effects of stress and adversity experienced during this period of dynamic maturation in all above-listed species. Interestingly, very few studies addressed this question directly.

Two studies investigating the effects of early postnatal stress on the timing of the GABA switch yielded conflicting results, with one study reporting accelerated switch to inhibitory GABA (Karst et al., 2023) while the other reported delayed switch (Furukawa et al., 2017). Furukawa et al. used daily maternal separation from P2 to P21 and measured muscimol-induced Ca2+ response in CA3 pyramidal neurons on P2, P8, P11, and P15 and NKCC1 and KCC2 expression in hippocampal subregions on P7, P 14, and P21. They found a deficiency in KCC2 expression, already apparent on P7 and becoming more pronounced with time in maternal separation exposed mice compared to controls. In addition, muscimol failed to induce a Ca2+ response in the CA3 of control animals, suggesting a lack of a depolarizing effect, already on P11, while this was achieved only on P15 in maternal separation animals. Strikingly, the KCC2 deficiency persisted into adolescence, with maternally separated mice showing lower KCC2 compared to controls in CA1 and CA3 on P35. Karst et al., on the other hand, employed the limited bedding/nesting (LBN) approach between P2 and P9 and measured GABA reversal potential, mEPSCs and mIPSCs in the medial prefrontal cortex through ex vivo recordings conducted during and after the LBN period. Based on GABA reversal potential, the authors concluded that the excitatory to inhibitory switch in the medial prefrontal cortex happens between P15 and P21 for control mice, whereas the switch takes place between P6 and P9 in LBN mice. Unlike Furukawa et al., the authors found no difference in medial prefrontal cortex KCC2 expression between control and LBN mice in pooled P9-P15 samples, but found reduced NKCC1 expression in LBN mice, suggesting the accelerated GABA switch might be the result of this early reduction of NKCC1.

The seemingly conflicting findings of the two studies might stem from the different regions investigated. As noted in the above sections, the GABA switch happens at different timepoints in different brain regions and cell types. It is possible that the effect of early life stress on the GABA switch is also region-specific, with accelerated GABA switch in some regions and delayed switch in others. More importantly, the two studies also employed different stress models. While inducing more consistent long-term phenotypes in rats, maternal separation has in general failed to provoke robust long-term effects in C57Bl6 mice, leading some researchers to conclude that this strain is resilient to the effects of maternal separation (Own & Patel, 2013) It is possible that the maternal separation approach used in Furukawa et al. leads to mild stress, leading to a stress-innoculation effect (Meaney, 2001; Parker et al., 2006) rather than a vulnerability effect in the long-run. Some of their behavioral findings at P35, such as increased explorative tendencies in the maternal separation exposed mice, including more exploration of the open arms in elevated plus maze, seem to be in line with this idea. LBN, on the other hand, represents a severe stressor, with most studies noting a long-lasting stress vulnerability phenotype in LBN-exposed offspring (Walker et al., 2017).This hypothesis of stress-innoculation versus vulnerability would also imply that KCC2 might be more sensitive to the effects of mild stress, which leads to delayed GABA switch, whereas NKCC1 might be more sensitive to severe stress, leading to accelerated GABA switch. This hypothesis, along with the hypothesis of the region-specificity of early life stress effects, can be tested through systematic studies that vary the intensity of early life stress and take comparative measures in different brain regions.

In addition to the GABA switch, there is evidence that early experiences can cause changes in GABAAR expression that last into adulthood. For instance, only two days (P9 and P10) of handling combined with maternal separation was found to reduce the expression of α1GABAARs in the dentate gyrus of adult rats, leading to an immature GABAAR subunit expression profile with α2 subunit as the dominant subunit (Hsu et al., 2003). A study that involved daily maternal separation for the whole preweaning period (P2-P21) in rats reported an increase in the expression of the α2 subunit in the mPFC and the central nucleus of the amygdala (Gondre-Lewis et al., 2016) potentially leading to a similar immature GABAAR expression profile with α2 dominance. Unfortunately, this study did not report levels of α1 expression in these brain areas or GABAAR subunit expression in the dentate gyrus, leaving the question of whether an immature α1/α2 ratio might be a general effect of early life stress unanswered. Moreover, the 3-week stress period which covers almost all early developmental changes in the GABA system makes it impossible to pinpoint a sensitive window for the induction of long-lasting changes in GABAAR expression.

While early postnatal insults such as above have the potential to cause changes in GABAAR expression that lasts into adulthood, later stress manipulations do not cause lasting effects by themselves but apparently make the organism more vulnerable to the effects of later life stress on the GABA system. In one study, juvenile rats (P27-P29) were exposed to elevated platform stress, which, by itself had no effect on α1 or α2 subunit expression in the basolateral amygdala (BLA) or the hippocampus on P60 (Jacobson-Pick et al., 2008). However, when a subgroup of these animals were exposed to open field and startle stress as adults, they had changed GABAAR subunit expression in both regions compared to naive controls or controls who were exposed to the adult stressors without the prior juvenile stress. Namely, there was a reduction in α1 and an increase in α2 in the BLA, reminiscent of the above immature GABAAR expression profile. In the hippocampus, on the other hand, both α1 and α2 subunits were increased in comparison to control or adult-stress hippocampi.

Overall, it is clear that ELS can affect the timing of the GABA switch and can lead to lasting effects on the expression of different GABAARs. However, the heterogeneity of the ELS protocols in terms of the specific developmental window they target and in procedural details that affect stress severity makes it impossible to make comparisons between studies and draw conclusions regarding sensitive periods for the observed impacts on the GABA system.

CONCLUSIONS AND FUTURE DIRECTIONS

Development and maturation of GABAergic neurons and GABAARs is a fundamental developmental process that serves as the framework for the development of many other brain systems. As reviewed in the first section of this manuscript, experimental manipulation of GABAergic developmental processes such as the expression and migration of early GABAergic neurons, timing of the inhibitory/excitatory switch, and the expression of GABAARs can lead to large-scale neuronal and behavioral pathology and have been implicated in developmental disorders such as autism spectrum disorders, Fragile X syndrome, and seizure disorders. These findings support the idea that disruptions in the normal development of the GABAergic system can serve as the cause of large scale pathology through cascading effects on other systems whose normal development relies on the GABA system. The specific protracted maturation of the GABA system, with many GABAergic developmental changes spanning the early postnatal period, might potentially make the GABA/GABAAR system particularly vulnerable to the effects of early life insults. The effects of these insults on the GABA system can in turn lead to increased excitability, disrupted network activity, and a range of behavioral alterations which might then be ameliorated through the normalization of GABAergic activity. As reviewed in the second part of this manuscript, there is ample evidence to support the idea that early life insults such as exposure to drugs, maternal immune activation, and early life stress can have a long-lasting impact on the GABA/GABAAR system (See Table 1 for a visual summary of the findings reviewed in this section). Systematic studies linking circuit-specific changes in the GABA/GABAAR system to specific behavioral outcomes are needed to establish causality between disruptions in the GABA/GABAAR system and behavioral outcomes. In addition to providing a mechanistic understanding of developmental processes, such studies can have translational value, as there is a vast arsenal of pharmacological agents targeting the GABA system and the development of more specific GABAAR modulators is an active area of research and development.

Table 1.

Summary of GABAAR subunit expression changes resulting from prenatal and early postnatal developmental insults.

Early developmental insult	Timing of insult	Timing of outcome measure	Affected brain region	Affected GABAAR subtype(s)	Reference
Ethanol exposure	G8 rats	G18 embryo & adulthood	Whole brain	Decreased ɑ5 at G18; increased ɑ5 in adult	Toso et al., 2006
Ethanol exposure	G2-G67 guinea pigs	Near-term fetus & P60	Hippocampus	No change in fetus; increased β2/3 in adult	Iqbal et al., 2004
Ethanol exposure	P2-P12 rats	P28	Cerebellum	Increased δ	Diaz et al., 2014
Benzodiazepine exposure	P7-P11 rats	P12-P18 & adulthood	Hippocampus, Cortex	Increased ɑ2, decreased δ up to P18; no change in adult	Kubova et al., 2020
Cocaine exposure	G8-G29 rabbits	P20 & adulthood	Anterior cingulate cortex	Increased ɑ1 in lamina III, decreased β2 in lamina II at P20; no change in adult	Shumsky et al., 2002
Maternal immune activation (MIA)	G15 mice	Adulthood	Ventral dentate gyrus, basolateral amygdala	Increased ɑ2 in both regions	Nyffeler et al., 2006
MIA	G17 mice	Peripuberty & adulthood	Prefrontal cortex	Increased ɑ2, ɑ4 in peripubertal; decreased ɑ2, ɑ4, ɑ5, increased ɑ3 in adult; no change in ɑ1 at either timepoint	Richetto et al., 2014
MIA	G15 rats	Adulthood	Ventral tegmental area	Decreased β2/3	Chen et al., 2022
MIA	G15 rats	Adulthood	Hippocampus, Prefrontal cortex	Decreased β1 in hippocampus, decreased β3 in cortex	Su et al., 2022
Handling + maternal separation	P9-P10 rats	Adulthood	Dentate gyrus	Decreased ɑ1	Hsu et al., 2003
Maternal separation	P2-P21 rats	Adulthood	Medial prefrontal cortex, central nucleus of the amygdala	Increased ɑ2	Gondre-Lewis et al., 2016

Fig 2. Neocortex GABAergic neurons.

Schematics of some major interneuron types and interneuron connectivity across cortical layers.

Box 3. GABAA receptors in adolescence, puberty, and pregnancy.

GABA system and GABAARs continue to differentiate and mature during adolescence in late-developing brain regions (Chugani et al., 2001; Hashimoto et al., 2009). In humans, GABAARs reach adult levels by age 14 in the basal ganglia, age 17.5 in thalamus, age 18 in frontal cortex, and by age 19.5 in the PFC (Chugani et al., 2001). The developmental shift in expression of GABAAR α subunits in dorso-lateral PFC of primates continues through adolescence (Hashimoto et al., 2009).

GABAARs play an important role in the regulation of gonadotropin-releasing hormone (GnRH) neurons, which are drastically activated during puberty in mammals. Unlike other networks, the switch from GABA depolarization to hyperpolarization in GnRH neurons is delayed until the puberty period (Clarkson & Herbison, 2006). On the other hand, GABAARs in different areas of the brain are also a target for neuroactive steroids (Akk et al., 2007; MacKenzie & Maguire, 2014; Maguire & Mody, 2007, 2009; Smith et al., 2009; Smith et al., 2007), for instance expression of α4βδ GABAARs significantly increases at extrasynaptic sites on CA1 hippocampal pyramidal cells at the onset of puberty in female mice that increases inhibition of these neurons (Smith, 2013; Smith et al., 2009).

At the beginning of puberty ovaries produce increased amounts of estradiol, progesterone, and the neuroactive metabolite of progesterone allopregnanolone (ALP). GABAergic neurons, perineuronal nets (PNN) surrounding PV+ neurons, and GABAARs are influenced by these hormones (Gilfarb & Leuner, 2022), e.g. ALP acts as a positive allosteric modulator of GABAARs, potentiating their inhibitory activity (Concas et al., 1999; MacKenzie & Maguire, 2014).

Changes in GABA system promote increase in inhibition over the course of adolescence. Expression of the GABA synthesizing enzyme GAD, GABA transporters, and the number of inhibitory synapses drastically increases at puberty (Guo et al., 1997; Kilb, 2012; Shen et al., 2020). Increase in PV expression and PV+ neuron activity is observed in PFC and hippocampus (Delevich et al., 2021; Kilb, 2012; Sisk & Zehr, 2005; Smith et al., 2009). Also GABAAR subunit expression changes during adolescence, for instance, GABAR δ subunit expression is increased on hippocampal pyramidal neurons and decreased on PV+ neurons at pubertal onset in mice (Keating et al., 2019; Shen et al., 2020). This increase in the expression of the extrasynaptic α4βδ GABARs on CA1 pyramidal neurons leads to increased tonic inhibition (Smith et al., 2007). Interestingly, ALP, which acts as a positive allosteric modulator of other GABAAR subtypes, has an inhibitory effect on α4βδ GABARs (Smith et al., 2009), at least partially balancing out the increased inhibition from increases in GABAARs and neurosteroids. Moreover, despite generally reported increase of GABA levels during adolescence, in , in some brain regions (e.g., anterior cingulate cortex (ACC)), GABA levels are lower in adolescents compared to young adults. Lower levels of ACC GABA were associated with greater impulsivity in teenagers (Silveri et al., 2013).

The onset of puberty is marked by the beginning of the ovarian cycle comprised of follicular and luteal phases. In humans the level of estradiol increases during the follicular phase till ovulation, after that the increase in progesterone and ALP occurs. At menses estradiol, progesterone, and ALP levels drops and remains at low level for a few days. It was demonstrated that at ovulation GABA concentration reaches the peak in PFC (De Bondt et al., 2015), while at luteal phase, a decrease in GABA concentration is observed (Epperson et al., 2002). Further, increase in gamma oscillation frequency and in neuronal synchrony were detected during the luteal phase relative to the follicular phase (Sumner et al., 2018). Fluctuations in estradiol, progesterone, and ALP affect the GABA-mediated inhibitory tone over the ovarian cycle and cause fluctuations in behavior (Maguire & Mody, 2009).

Rats and mice have 4-6 day estrous cycles, which consist of: metestrus, diestrus, proestrus, and estrus. In mice, estradiol peaks during proestrus and progesterone is maximal during diestrus. In rats, both estradiol and progesterone peak during proestrus. ALP level fluctuates depending on the level of progesterone. Fluctuations of ovarian hormones during estrus cycle can modulate inhibitory tone through GABAARs (Licheri et al., 2015; Lovick, 2012; Lovick & Zangrossi, 2021; Maguire & Mody, 2007; Maguire et al., 2005; Wu et al., 2013), for instance, via cyclic changes in the expression of neurosteroid-sensitive δ subunit. Increase in δ subunit in GABAARs of hippocampus and decrease in hippocampal gamma oscillations were observed during diestrus in mice (Barth et al., 2014). Also, inhibitory tone is modified by PV expression and activity of PV+ neurons. Decrease in amygdala PV expression during proestrus was observed (Zimmermann et al., 2019). Additionally, in midbrain periaqueductal gray (PAG) expression of α4β1δ GABAARs fluctuates over the estrous cycle in relation to changes in the level of ALP, which changes GABAergic tone and anxiety-like behavior (Lovick, 2012; Lovick & Zangrossi, 2021).

As during puberty and the ovarian cycle, significant hormonal changes occur during pregnancy and postpartum. In humans, estradiol and progesterone during pregnancy increases 50- and 10-fold, respectively, as compared to the peak concentrations during the menstrual cycle. In rodents, levels of estradiol and progesterone increase during late pregnancy. ALP rises in humans and rodents reaching the maximum during late pregnancy. The hormonal increase drops just prior to delivery in rats or following delivery in humans. Changes in hormonal level in turn influence different compartments of the GABA system and, in turn, excitability and inhibitory tone(Licheri et al., 2015; MacKenzie & Maguire, 2014; Maguire, 2019; Maguire & Mody, 2009).For instance, increase in δ subunit in hippocampus occurs during pregnancy and decrease and in α4 and γ2 after delivery in rats which alter synaptic transmission, leading to changes in excitatory/inhibitory balance (Biggio et al., 2009; Licheri et al., 2015; Maguire & Mody, 2009). Altered GABAergic signaling was shown to be related to changes in mood, increased anxiety and depressive behavior during pregnancy and postpartum and deficit in maternal care during postpartum period (Maguire, 2019; Maguire & Mody, 2008; Mostallino et al., 2009).

HIGHLIGHTS.

• GABA neurons/receptors are expressed before excitatory counterparts in the brain.

• Development of the GABA system guides the development of other brain systems

• Development of the GABA system continues postnatally.

• Early developmental insults cause long-lasting disruptions in the GABA system.