The Evidence for GABA Dysregulation in Autism; Social Processing Deficits and Associated Genetic and Epigenetic Mechanisms

The Evidence for GABA Dysregulation in Autism; Social Processing Deficits and Associated Genetic and Epigenetic Mechanisms

Nina Parrella University of Amsterdam

Author Note

Nina F. Parrella, student of the Masters in Brain and Cognitive Sciences, Institute of Interdisciplinary Science, University of Amsterdam.

Contact: nina.parrella@student.uva.nl

Through the use of both disability-first and person-first terms, this literature review endeavours to acknowledge the deliberations and contention within the scientific community surrounding the conceptualisation of autism. Kenny et al. (2015) demonstrated that there is no unanimously accepted term to describe autism, however disability-first terms are favoured by autistic adults and parents while person-first terms are favoured by professionals.

Abstract

The neurotransmitter γ-aminobutyric acid (GABA), the main inhibitory neurotransmitter, has a prominent role in regulating neural development and functioning. Despite their diverse aetiologies, certain behavioural deficits observed across neuropsychiatric conditions have been hypothesised to arise from dysfunction in particular aspects of GABAergic inhibitory signalling. Such disturbances, particularly during early development, can disrupt the cellular balance of excitation and inhibition in neuronal circuits. This disordered neural circuitry may account for the social processing deficits observed in autism, a highly heterogeneous and heritable lifelong neurodevelopmental condition, among other pathologies, including Schizophrenia Spectrum Disorders (SSD). GABA dysregulation has been implicated as an underlying contributor to a phenotype involving social processing and inter-personal deficits. While this premise has not been subject to abundant direct testing, this review aims to unify diverse streams of pathophysiological and genetic evidence to elucidate the relevance of GABA to this phenotype. Approaching therapeutic targets by focusing on micro-circuitry related to specific symptoms is a promising approach because the aetiology of idiopathic autism involves intricate interactions between heritable and environmental factors. Through the compilation of research from several disciplines including molecular neurobiology, genetics, and systems neuroscience, the role of GABA’s relation to pervasive social processing deficits observed in autism is discussed. Underlying circuitry, risk-genes, and environmental factors that contribute to such symptomatology are found to be

multidimensional and heterogeneous. It is evident that further efforts are required to explore the possible role of environmental and epigenetic elements that may modulate social

processing deficits in neuropathology.

Keywords: GABA dysregulation, inhibitory / excitatory imbalance, autism, social

The Evidence for GABA Dysregulation in Autism; Social Processing Deficits and Associated Genetic and Epigenetic Mechanisms

Clinically and genetically heterogeneous neuropsychiatric conditions such as autism have complex neurophysiological substrates. Independent from identifiable genetic or environmental causes, idiopathic autism is 4-5 times more prevalent in males (Yoo, 2015). Unlike secondary autism, idiopathic forms of autism, also referred to as ‘intrinsic autism’ (Liu et al., 2016), have largely unknown aetiology and account for approximately 85% of identified cases (Coghlan et al., 2012; Liu et al., 2016; Betancur, 2011). The diverse range of symptoms observed across the autism spectrum independently result from region-specific susceptibilities (Etherton et al., 2018). Developmental, environmental, genetic and epigenetic factors are thought to regulate these vulnerabilities (Etherton et al., 2018). It is the interaction of these effects that produces the diversified clinical spectrum of autism and other

neurodevelopmental pathologies, including Schizophrenia Spectrum Disorders (SSD) (Owen, O’Donovan, Thapar and Craddock, 2011). While genetic factors associated with complex behavioural phenotypes, such as social processing deficits, are beginning to materialize, the neurophysiological substrates that underly neuropsychiatric disorders remain predominantly unknown. An emergent hypothesis suggests that disparate genetic abnormalities can bring about similar neuropsychiatric phenotypes (Yizhar et al., 2011; Markram and Markram, 2010). Exploring this relatively recent view requires clarification of how circuit-level mechanisms are altered according to the varied genetic factors which underlie common pathophysiology (Yizhar et al., 2011). This literature review endeavours to provide such clarification.

Functional balance between excitatory and inhibitory networks is required for information transfer in the brain, and necessitates the maintenance of appropriate ratios of excitatory versus inhibitory synaptic inputs (Gao and Penzes, 2015). Glutamate’s activation

of excitatory synapses causes cell depolarisation, which increases the chance of the cell firing an action potential, while GABA’s activation of inhibitory synapse has the opposite effect (Gao and Penzes, 2015). Balanced excitatory / inhibitory synaptic ratio is fundamental in maintaining the cell’s firing patterns within a small range (Gao and Penzes, 2015).

Neurophysiological substrates underlying social processing deficits present in autism could involve hyper-functionality and excessive neuronal processing in local neuronal micro-circuits. Implicated areas are likely to include the posterior superior temporal sulcus (pSTS), amygdala, and prefrontal cortex (PFC), to name a few (Pizzarelli and Cherubini, 2011). Such atypical circuitry allegedly involves malfunctions in certain aspects of inhibitory gamma-aminobutyric acid (GABA) neurotransmission (Coghlan et al., 2012). GABA’s primary role is to regulate excitatory neural processes by reducing the neuronal activity of target cells by binding to GABA receptors that are present on the cell surface (Wu and Sun, 2015). An emergent and promising theory proposes that certain autistic features result from an

imbalance of excitatory / inhibitory neurotransmission (Gogolla et al., 2009; Rubenstein and Merzenich, 2003; Silverman et al., 2015).

Autism and SSD are cognitive disorders with complex genetic architectures that share some behavioural phenotypes (Canitano and Pallagrosi, 2017; Gao and Penzes, 2015).

Excitatory / inhibitory imbalance is a pathophysiological mechanism that has been implicated in both disorders (Gao and Penzes, 2015) and may underlie respective social processing deficits (Pinkham et al., 2008; Canitano and Pallagrosi, 2017). In both disorders, neuronal activation reductions in networks involved in social processing imply that these impairments may involve specific patterns of neural abnormalities that may be deficit-specific rather than disorder-specific (Pinkham et al., 2008). Psychosocial dysfunction has been linked to an increased excitation / inhibition ratio (Yizhar et al., 2011). Investigating the role of right temporal excitatory / inhibitory ratio in autism and schizotypal spectra phenotype, Ford,

Nibbs and Crewther, (2017) revealed that increased excitatory / inhibitory ratio in the right auditory region may lead to behavioural and cognitive characteristics related to social processing and interpersonal deficits (Ford, Nibbs and Crewther, 2017a). It is possible that, regardless of clinical diagnosis, individuals presenting with specific social processing deficits may exhibit the same pattern of neural abnormalities (Gao and Penzes, 2015).

Research in genetics aims to provide a firm causal foundation that elucidates molecular pathways, cellular and circuit functioning, and ultimately relates to human behaviours (De la Torre-Ubieta, 2016). Current genetic evidence suggests that, rather than being singular, Mendelian disorders with one underlying genetic explanation, autism and SSD are polygenic, comprised of multiple distinctive etiologically conditions (Yang et al., 2018). The onset and progression of these disorders are therefore produced by interactions between multiple genes components (Canitano and Pallagrosi, 2017). Complex interactions between several genes, in association with environmental and epigenetic factors, may lead to the aetiology of social processing deficits.

With the ultimate objective of establishing therapeutic supports for neurodivergent individuals in need, it is necessary to establish causal links between genetic architecture, targetable pathophysiological pathways and clinically relevant symptoms (Selimbeyoglu et al., 2017). Since impairment in social cognition is a primary deficit for autistic individuals, examining the brain circuitry relevant to social processes can provide insights into the mechanisms underlying this deficient domain (Pelphrey et al., 2011; Gotts et al., 2012). The mechanisms underlying GABA-related deficits that cause specific symptomatology are not commonly explored. Research findings from molecular neurobiology; involving animal models, human imaging techniques and post-mortem studies are integrated in this review. Even though autism research tends to be central to this topic, individuals presenting with specific social processing deficits may also show similar patterns of neural abnormalities,

regardless of clinical diagnosis. This review highlights evidence of abnormalities in GABAergic neurotransmission in neuropsychiatric conditions, such as autism and SSD, characterised by core symptomatology that result from impairments in social cognition. Initially, neurobiology relevant to social processing deficits is explained, followed by a general overview of the GABA system. GABA dysregulation observed in autism and social processing deficits is then discussed, as well as emerging GABAergic therapies. Genetic and epigenetic mechanisms associated with social processing-related GABA dysregulation are also detailed.

Neurobiology of social processing deficits associated with autism

The neural substrates of social cognition have amassed attention in recent years, as the neuroscientific community strive to disentangle how the human brain parses the social world (Lahnakoski et al., 2012). Neural processes that underly social behaviours are challenging to identify due to their association with high degrees of ambiguity and

unpredictability. The possibility that underlying incentives for social interactions may differ between species presents further difficulties in revealing such mechanisms. Social

engagement in humans initially presents as joint attention, whereby infants use eye gaze to direct the attention of others and demonstrate engagement. For these skills to develop and eventually fulfil the complex communication expectations faced in adolescence and adulthood, abilities must advance in order to interpret complex social stimuli and continuously make informed choices of actions whilst predicting the behaviours of

communication partners (Markowitz et al., 2018). The fundamental skills required for social interactions have complex underlying aetiology, involving neural micro-circuitries that are yet to be identified. Impairments in social cognition can be explored not only for

neuroscientific advantage, but for the crucial advancement of knowledge surrounding neuro-pathologies and subsequent clinical implications.

The various social processing deficits are potentially unique in their neurological underpinnings, indicating aetiological heterogeneity (Pelphrey et al., 2011) as well as neural complexity. Impaired social processing is a prominent feature of neuropsychiatric conditions. Observation of subsequent symptoms can aid the diagnostic process of autism, as well as other neuropathological conditions. There is emerging evidence for altered anatomical structure, functional activation, and connectivity of the social brain regions in individuals with autism (Gotts et al., 2012; Kennedy and Adolphs, 2012; Pelphrey et al., 2011). Studied social brain regions include the fusiform face area (FFA) (Kennedy and Adolphs, 2012; Spencer et al., 2011), pSTS (Kaiser et al., 2010; Di Martino et al., 2009; Philip et al., 2012), amygdala (Baron-Cohen et al., 2000), striatum (Horder et al., 2018), somatosensory cortex (Etherton et al., 2011) hippocampus (Etherton et al., 2011; Han, et al., 2014; Pizzarelli and Cherubini, 2013), lateral septum (Bredewold et al., 2015) and disrupted connectivity of the theory-of-mind (ToM) network (Kana et al., 2014; see Figure 1, Barak et al., 2016).

Mediating an array of social functions, including joint attention, interpreting intentions, perceiving emotions, and processing faces; distinctive functional activation of these areas can serve as potential neural markers for autism and neuropathological conditions (Tyng et al., 2017; Barak et al., 2016).

Anatomical and functional imaging studies also support the notion that multiple sensory brain regions underlie the atypical social behaviours present in neurodevelopmental and psychological disorders (Baron-Cohen et al., 2000; Mundy, 2018). Communication and socialisation encompass the integration of auditory, visual, and somatosensory information processing (Chattopadhyaya and Cristo, 2012). The amygdala, among other social-cognitive network regions, is central for interpreting socio-emotional cues, storing memories,

establishing fear associations, anxiety, and regulating autonomic and hormonal responses (Tyng et al., 2017). The prefrontal cortex is also implicated as relevant to regulation of

emotional behaviour, social cognition, executive function and language (Rinaldi, 2008). While some non-invasive imaging studies initially suggested that the pre-frontal cortex is not sufficiently activated in autism (Castelli et al., 2002), hyper-activation in this region has also been identified (Supekar et al., 2013). The cerebellum has been recently acknowledged for an unappreciated role in controlling the reward circuitry and social behaviour, interacting with the ventral tegmental area in mice showing social preference (Carta et al., 2019).

Connected with the fusiform gyrus, orbitofrontal cortex, and amygdala (Lahnakoski et al., 2012), the pSTS is a key site of pathophysiology in autism (Alaerts et al., 2015).

Participating in processing social perception and cognition (Lahnakoski et al., 2012), the pSTS has been identified as both structurally (Boddaert et al., 2004) and functionally (Di Martino et al., 2009; Alaerts et al., 2015) divergent in autism. Aberrant functional

connectivity in the pSTS has been linked with social processing deficits and emotion

recognition (Alaerts et al., 2015). These traits have been linked to underconnectivity between key areas in the temporal and fronto-parietal lobes (Alaerts et al., 2014). The pSTS is

considered a major neural hub that connects several social processing networks. Atypical neural connections in the pSTS may lead to mirror neuron system dysfunction as

hypothesized by the broken-mirror theory of autism (Ramachandran and Oberman, 2006; Alaerts et al., 2015).

Despite opposing claims that spatial organisation to the STS does not exist (Hein and Knight, 2008), Deen et al. (2015) valuably identified selective STS subregions that respond to particular types of social input, and are arranged along a posterior-to-anterior axis. Regions responsive to language and theory of mind, faces and voices, and faces and biological motion have been delineated (Deen et al., 2015). Thus, there is argument for the human STS having a rich spatial structure, containing relatively domain-specific areas, and regions that respond to multiple types of social information (Deen et al., 2015).

As mentioned in the previous section, social communication difficulties are phenotypically shared across neuropsychiatric disorders, including autism and SSD (St Pourcain et al., 2017; Stanfield et al., 2017). Stanfield et al. (2017) identified overlaps between autism and SSD in social cognitive difficulties, but their functional magnetic resonance imaging (fMRI) investigation established significant distinguishable differences social brain activity between the conditions (Stanfield et al., 2017). These findings were consistent with the notion that superficially similar conditions are associated with unique underlying mechanisms (Stanfield et al., 2017).

While establishing insights into the pathogenesis of autism and its associated symptoms is in its early stages, the involvement of altered neuronal connectivity is

uncontroversial. Diffusion tensor imaging studies (DTI) can highlight aberrations in anterior-posterior and interhemispheric white matter tracts (Rudie et al., 2013). Disordered functional connectivity is postulated to reflect an increased ratio of excitation / inhibition in sensory, mnemonic, social, and emotional systems (Rubenstein and Merzenich, 2003). Rudie et al. identified altered patterns of cortical activation in autistic individuals while they executed social and cognitive tasks. Abnormal intrinsic functional connectivity (anterior-posterior and/or interhemispheric connections) was also detected in autistic participants involved in this study (Rudie et al., 2012). This hypothesis is derived from clinical studies that indicate an imbalance of excitatory / inhibitory neurotransmission in brain regions of relevance to autism (DeVito et al., 2007), as well as recent research indicating dysfunctional GABAA receptors in autistic patients (Mendez et al., 2012; Stamou et al., 2013).

Through inclusion of associated literature, this review proposes that neurochemical abnormalities may be attributable to specific spectrum phenotypes, such as social processing deficits, rather than to spectrum conditions in their complex and heterogeneous entirety. Therefore, evaluating the glutamate/GABA+ ratio related to this phenotype, often observed in

individuals with diagnoses of autism and SSD, forms the basis of this multi-faceted investigation. Apparent impairment of glutamatergic and GABAergic synaptic function suggests that excitatory / inhibitory imbalance could contribute to symptomatology observed in autism (Pelphrey et al., 2011; Gotts et al., 2012). Consistent with this, mouse models with altered synaptic transmission in exhibit social dysfunction (Etherton et al., 2011) and directly increasing the excitatory / inhibitory ratio in the medial prefrontal cortex (mPFC) of the brain, using optogenetic stimulation, led to impaired social interactions in mice (De la Torre-Ubieta, 2016). This review intends to evaluate how the discussed neurobiology leads to social processing and inter-personal deficits.

Figure 1. Anatomical and functional brain areas relevant to social behaviours. Image taken from Barak et al. 2016.

The GABA System

In the developed human brain, GABA, known as the primary inhibitory neurotransmitter (Coghlan et al., 2012), is synthesised from the excitatory neurotransmitter glutamate, and regulates glutamatergic excitation (see Figure 2; Coghlan et al., 2012). Regulation of

glutamate and GABA levels underlies successful neural plasticity and migration (Stagg et al., 2011). Excessive levels of extra-neuronal glutamate can interfere with synaptic development and neuronal migration, leading to various forms of abnormalities in cellular systems (El-Ansary and Al-Ayadhi, 2014). In relation to neurodevelopmental conditions, including autism, findings suggest that aberrant excitation and inhibition in local neural circuits may significantly influence interregional hyper-connectivity (Supekar et al., 2013).

For the synthesis, release, breakdown, and stabilisation of GABA, numerous proteins are required. Glutaminase synthesises glutamate from glutamine and glutamic acid

decarboxylase (GAD) synthesises GABA from glutamate (Coghlan et al., 2012). GABA synthesis usually occurs in inhibitory interneurons, in addition to occasional synthesis in activated astrocytes (Kim and Yoon, 2017; Coghlan et al., 2012). The release of synaptic GABA involves vesicular inhibitory amino acid transporter (VIAAT), which loads GABA into synaptic vesicles (Coghlan et al., 2012). Reuptake of extracellular GABA into the presynaptic interneurons involves GAT1, GAT2 and GAT3 transporters (Madsen et al., 2009) and glial cells, such as astrocytes, are also involved in the reuptake of GABA that escapes the synapse (Madsen et al., 2009). The breakdown of GABA into glutamate requires GABA transaminase (GABA-T) (Madsen et al., 2009).

GABA function can be influenced by the indirect effects of transcription factors that arise through the regulation of gene expression, receptor trafficking, and downstream

perturbation, even when a particular mutation relates to GABA indirectly (Coghlan et al., 2012).

Synaptic maturation and refinement of neuronal circuitry relies upon balanced

glutamatergic (excitatory) and GABAergic (inhibitory) signalling, permitting the regulation of cognition, emotion, and behaviour (Arjam et al., 2019). Cortical excitatory / inhibitory imbalance can result from disruption to any component of the glutamate and GABA

signalling systems. In neuropathology, this imbalance can lead to hyperexcitability in brain regions involved in language, social interaction, and multisensory perception (Rubenstein, 2010; Harada et al., 2011).

Figure 2. Major synaptic pathways responsible for GABA synthesis, breakdown, release and reuptake. Abbreviations: gaba = gamma-aminobutyric acid; glu = glutamate; gln = glutamine. GLNase = glutaminase. GAD = glutamate decarboxylase. GABA-T = GABA transaminase. VIAAT = vesicular inhibitory amino acid transporter. GAT = GABA transporter. GABA-A = GABAA _{receptor (note: there are}

numerous subtypes of this receptor, but these are not shown, for simplicity.) GS = glutamine synthetase. Not pictured: GABAB _{receptors; GABA}A _{or GABA}B _{autoreceptors. Image taken from Coghlan et al. 2012.}

GABA Receptors

Produced and released by cytoarchitecturally diverse inhibitory interneurons, GABA acts on both ionotropic GABAA _{receptors (ligand-gated chloride channels) and metabotropic} GABAB _{receptors (Coghlan et al., 2012). Under most physiological conditions, GABA} binding to GABAA _{receptors leads to inhibition of cell firing, while activation of GABA}B receptors can influence neurotransmitter release and cause an increase in excitation and neuronal hyperpolarisation (Bowery et al., 2002).

As observed in rodents by Fagiolini et al. (2004) specific subunits of the GABAA receptor system appear to have functional roles in neurodevelopment. In humans, migration and differentiation of GABAergic interneurons occurs during late gestation (Xu et al., 2011). This period is subject to perinatal hypoxia-ischemia, in which development of GABAergic function is vulnerable (Xu et al., 2011). GABA acts an excitatory neurotransmitter for the neurogenesis observed during early stages of neurodevelopment, despite its inhibitory role in the adult brain (Sibilla and Ballerini, 2009).

GABAergic dysfunction associated with autism

The presence of GABA in vivo can be detected and analysed using in situ quantification via microsensor or microelectrode implantation, as well as non-invasive

imaging techniques in vivo like proton magnetic resonance spectroscopy (1H-MRS) (Puts and Edden, 2012). 1H-MRS quantifies low molecular weighted metabolites such as GABA, and separates multiple metabolites along a chemical shift spectrum (Schür et al., 2016; Juchem and Rothman, 2014). GABA levels have been found to differ significantly across brain regions in SSD subjects (Kegeles et al., 2012; Schür et al., 2016). Furthermore, when directly comparing studies that vary methodologically, consistency across findings may be difficult to establish due to regional differences, difficulty in disentangling the GABA signal from the macromolecular signal, and variable tissue composition (Schür et al., 2016). The significance

of these methodological issues cannot be overlooked as they may result in problematic variability in GABA levels across 1H-MRS studies (Schür et al., 2016).

Growing bodies of evidence indicate that identifiable differences in neurotransmitter excitation and inhibition could serve as biomarkers and treatment targets for autism (Arjam et al., 2019). Lower levels of GABA have been identified in specific brain regions in autism (Harada et al., 2011). Reduced GABAergic neurotransmission in the forebrain has been identified in individuals with autism (Han et al., 2012). Rodent models also demonstrate impaired GABAergic and glutamatergic neurotransmission (Gogolla et al., 2009; Baudouin, 2014). Such dysfunction has been elucidated by post-mortem studies (see Table 3). Table 3 details several post-mortem studies implicating GABA function in brain regions relevant to social processing deficits in autism. Some of these specified studies indicate decreased GABAA and GABAB protein expression (Oblak, Gibbs and Blatt, 2011; Blatt and Fatemi, 2011), while others observed decreased expression of GAD, the enzyme that catalyses the conversion of glutamate to GABA (Blatt and Fatemi, 2011) and increased AMPA-type glutamate receptor messenger RNA (Purcell et al., 2001).

Quantified using 1H-MRS, reduced GABA in frontal (Kubas et al., 2012),

motor/sensorimotor (Gaetz et al., 2014), temporal/auditory (Gaetz et al., 2014; Rojas et al., 2014) and occipital (Drenthen et al., 2016) regions has been identified in individuals with autism (Kirkovski et al., 2018). Barnes et al. (2015) proposed that lower GABA levels result from a loss of GABAergic interneurons, while Bruining et al. (2015) suggest such levels are compensatory for the inconsistent excitatory effects of GABA (Schür et al., 2016). A meta-analysis of 1H-MRS literature conducted by Schür et al. (2016) provided evidence for lower brain GABA levels in adults with autism compared with healthy controls (Schür et al., 2016). Through such investigations, it is evident that integration of the evidence for altered in vivo GABA levels across neuropsychiatric disorders is currently scarce (Schür et al., 2016).

It has been proposed that a deficit in the GABA synthesising enzyme, GAD, results in surplus cortical glutamate in autism (Fatemi et al., 2002). Likelihood of GABAA receptor down-regulation (cortical disinhibition resulting from reduced mediation of GABA

inhibition) has also been postulated (Fatemi et al., 2002; Fatemi et al., 2009). An imbalance in the excitation/inhibition ratio has also been widely attributed to N-methyl-D-aspartate receptor (NMDAr) hypo-function, which ultimately reduces GABAergic interneurons

inhibitory output, and increases cortical excitation through disinhibition of excitatory neurons (Lisman et al., 2008). NMDArs have been implicated in auditory processing deficits observed in both autism and SSD (Kantrowitz and Javitt, 2010).

Horder et al. (2013) showed that adult subjects did not demonstrate differences in GABA concentration in the striatum or in the mPFC (Horder et al., 2013), while Kubas et al. (2012) showed that children have reductions in these regions (Kubas, et al., 2012; Gaetz, 2014). Horder et al. (2018) propose that GABA abnormalities vary according to age and have advocated for longitudinal examination of GABA in autism to reveal its maturational

trajectory (Horder et al., 2018).

GABA dysregulation associated with social processing deficits

The process of natural selection is likely to have led to the development of various neural mechanisms responsible for social processing among different species (Sungur et al., 2015). While neural mechanisms involved in socializing are not yet well defined in rodents (De la Torre-Ubieta, 2016), rodent models may still reveal insights relating to relevant ancestral mechanisms (Sungur et al., 2015). Several pertinent investigations using rodents have analysed social behaviour deficits reminiscent of human autism phenotypes and their mechanisms (see Table 1).

Social motivation is thought to involve neural circuitry of the amygdala, ventral striatum, orbital, and ventromedial prefrontal cortex, along with mesolimbic reward circuitry

comprised of the nucleus accumbens, ventral tegmental area (VTA), and cerebellum (Rein et al., 2019; De la Torre-Ubieta, 2016). Synaptic dysfunction is likely to compromise social reward processing, and has been reported in the pre-frontal cortex and ventral tegmental area of SHANK3+/ΔC mice (Rein et al., 2019). Inhibitory synaptic malfunctions (involving

GABAA-mediated inhibition) occur in the basolateral nucleus of the amygdala of FMR1 knock-out mice (Olmos-Serrano et al., 2010). As this region is responsible for regulating fear and anxiety, such an abnormality could lead to dysfunctional social behaviour. Largely modulated by GABA and glutamate neurotransmission, the lateral septum (LS) is understood to regulate social play in juvenile rats (Bredewold et al., 2015). Bredewold et al. (2015) suggests a role for GABA neurotransmission in such behavioural regulation. Horder et al. (2018) examined glutamate and GABA levels in autistic individuals and in six rodent models of the disorder. Consistent with their mouse models, and with the concept of an excitatory / inhibitory imbalance in autism, adults with idiopathic autism were found to have decreased glutamate concentration in the striatum compared to controls (Horder et al., 2018). The association between striatal glutamate and the severity of social impairment implies that this abnormality is clinically significant (Horder et al., 2018).

Research in both rodents (see Table 1) and humans (see Tables 2 and 3) continues to develop the available evidence for altered excitation and inhibition of neural processes that may underlie social processing deficits. Reduction in somatosensory cortex GABA has been associated with tactile dysfunction in autism (Puts et al., 2017). Such dysfunction, resulting from loss of presynaptic inhibition of somatosensory neuron transmission, has been shown to result in social interaction deficits in mouse models of autism (Orefice et al., 2016).

Transmagnetic stimulation (TMS) has been used to establish deficits in GABAA receptor in autistic humans (Enticott et al., 2013) and for GABAergic modulation that augmented social capacities in mice with autistic-like behaviours (Tan et al., 2018). Other imaging (fMRI and

1H-MRS) studies of GABA in autism that implicate regions associated with social processing are summarised in Table 2. The table highlights related research that links social processing deficits with increased inhibitory / excitatory ratio (Ford et al., 2017a) and high glutamate and low GABA concentrations (Ford et al., 2019) in the superior temporal cortex (STC), reduced glutamate concentration in the striatum (Horder et al., 2018), and hypo-activity and hypo-connectivity with the fronto-parietal action observation network (AON) (Alaerts et al., 2014). Grouping these preliminary findings is useful in seeking clarification of GABAergic distribution and processes that contribute to social functioning in autism.

Potential GABAergic therapies for social processing deficits

The benefit of GABAergic pharmacologic compounds to target symptomatology in autism remains inconclusive (Brondino et al., 2015) as responses to medications containing conventional GABAergic agents are variable (Bruining et al., 2015). Rodent models with mutations in GABA receptor subunits and / or reduced GABAergic or parvalbumin positive inter-neurons display autistic phenotypes of social processing deficits and repetitive

behaviours (Brielmaier et al., 2014; Silverman et al., 2015). Benzodiazepines modulate GABAA receptors to increase cortical inhibition, and GABA agonists have amended behavioural and electrophysiological abnormalities in FMR1, SCN1A and BTBR mouse models (Han et al., 2014; Han et al., 2012). The sedative properties of benzodiazepines present a challenge in targeting GABAergic transmission in therapeutics, however. Such sedative properties are mediated by action at the α1 GABAAR subunit and may account for the lack of exploration of α1 GABAAR modulators in clinical trials (Löw et al., 2000).

Some pertinent preclinical tests that hypothesise that enhancing inhibitory

neurotransmission reduces specific symptoms are detailed in Table 1. Silverman et al. (2015) discovered that increasing GABAergic signalling via the GABAB _{receptor agonist R-baclofen} reversed lack of sociability of BTBR mouse models (Silverman et al., 2015). It is important

to recognize that the replicated improvements in sociability by R-baclofen in BTBR mice did not generalize to male–female reciprocal social interactions (Silverman et al., 2015).

Furthermore, direct infusion of D-cycloserine (a partial agonist of NMDAr) to the basolateral amygdala restored social processing deficits of TBR1 mice even though TBR1 has an

established role in deep-layer neuron generation and cortical lamination (De la Torre-Ubieta, 2016).

Due to the comorbidity of autism and epilepsy, benzodiazepines and other

antiepileptic drugs are common medications for autistic individuals (Deidda, Bozarth and Cancedda, 2014). Di Martino and Tuchman (2001) showed that antiepileptic drugs improved socialisation and communication skills in autistic individuals. Adolescents and young adults showed improvements in emotion recognition after 10 months of treatment with bumetanide, which is typically used to treat swelling and high blood pressure (Hadjikhani et al., 2013; Deidda, Bozarth and Cancedda, 2014). Reinforcing the hypothesis of the atypical

GABAergic transmission in autism, further correlation between compromised GABAergic transmission and autism was found in a mouse mutant of GABAA receptors where the mutation was restricted to forebrain interneurons (Han et al., 2012). These animals exhibited autistic behaviours, including decreased social interaction, that were rescued by

benzodiazepine treatment (Han et al., 2012).

Use of environmental enrichment (EE) to target sociability symptoms in autism typically involves exposure to enriched sensorimotor environments (Aronoff et al., 2016). EE is considered influential over GABAergic transmission (Deidda, Bozarth and Cancedda, 2014; Cancedda et al., 2004), and has potential to target behavioural, synaptic and

connectivity in models of neurodevelopmental disorders (Hannan, 2014; Deidda, Bozarth and Cancedda, 2014). Behavioural impairments including nociception (the detection of harmful stimuli), pre-pulse inhibition, stereotyped behaviours, grooming, memory deficits, anxiety,

social processing deficits were rescued in several mice models of autism exposed to an EE setting during development (Deidda, Bozarth and Cancedda, 2014). Lonetti et al. (2010) established that EE rescued cortical long-term potentiation (LTP; the persistent strengthening of synapses based on repeated activity), motor skills, spatial learning and reduced anxiety behaviours in MECP2 null mice, indicating that the inhibitory GABAergic system could be accountable (Lonetti et al., 2010).

Genetic components related to autism and SSD

Due to its highly heterogeneous genetic aetiology, there have been countless genes associated with autism (Huguet et al., 2013), suggesting the involvement of diverse

pathophysiological mechanisms. A number of proposed transcriptomic signatures (specific sets of RNA molecules within a cell) and genetic risk variants implicate underlying molecular and cellular processes across the autistic population (De la Torre-Ubieta, 2016); however, it remains unclear how these allegedly distinctive markers can manifest as analogous

phenotypes (De la Torre-Ubieta, 2016). The most frequent chromosomal alterations observed in ~5% of cases of autism are 15q11–q13 duplications, and deletions of 2q37, 22q11.2, and 22q13.3 (Betancur, 2011). Single gene mutations commonly involved in autism are fragile X syndrome (FMR1), tuberous sclerosis (TSC1, TSC2), neurofibromatosis (NF1), Angelman syndrome (UBE3A), Rett syndrome (MECP2), and PTEN (Betancur, 2011). Present in

synaptic genes, NLGN3, NLGN4X (Jamain et al., 2003), SHANK3 (Durand et al., 2007), and SHANK2 (Berkel et al., 2010) are rare mutations that have also been associated with autism (Betancur, 2011).

The biological underpinnings shared by autism and SSD may emerge during early neural development and advance during childhood development (De Lacy and King, 2013; Canitano and Pallagrosi, 2017). The two polygenic disorders manifest through interactions between multiple susceptibility genes, of which some are common (Canitano and Pallagrosi,

2017). Mutant mice with SHANK3 mutations have been linked to both autism and SSD, sharing synaptic and behavioural phenotypes (Zhou et al., 2016), as well as CNTNAP2 (Strauss et al., 2006), and 16p13.11 microdeletion / microduplication (Pinto et al., 2010; Canitano and Pallagrosi, 2017).

Potential genetic mechanisms associated with social processing-related GABA dysregulation

Social processing deficits in both autism and SSD are hypothesised as relating to excitatory / inhibitory imbalance (Canitano and Pallagrosi, 2017). Such processes may result from abnormalities in genes coding for glutamatergic and GABAergic receptors or synaptic proteins (Canitano and Pallagrosi, 2017). Studies using rodent models of autism can be fruitful in addressing genetic links to GABA’s role in social processing deficits (Goodrich et al., 2018; Horder et al., 2018; Pizzarelli and Cherubini, 2013). Rodent models permit the molecular, cellular, circuit, and behavioural analyses of neurodevelopmental conditions (De la Torre-Ubieta, 2016). They also allow for trialling of potential therapeutics that target specific symptoms before humans are tested. There are a range of well-established assays that are now standard assessments for autism-related phenotypes in rodent models, allowing assessment of social interactions and communication. These behavioural phenotypes are believed to be comparable to social processing deficits in humans. One strategy relevant to communication behaviours is the ultrasonic vocalization (USV) test, which measures the frequency and properties of vocal communication in multiple settings (De la Torre-Ubieta, 2016).

Ronald and Hoekstra (2011) state that gene identification can be tackled by studying sub-phenotypes, or endophenotypes (biological markers that are easily detectable) related to autism. For example, Mazefsky et al. (2008) demonstrated that social dysfunction and nonverbal communication symptoms showed common genetic influences in autistic twins.

This poses a considerable challenge, not only due to the need to assess the mechanistic roles of numerous genetic candidates for autism, but also the need to navigate their potential implications on the GABA system whilst acknowledging the multiple brain regions relevant to variable manifestations of social processing deficits, as detailed in the sections above.

Through reviewing several rodent models related to autism (see Table 1), it is evident that different mechanisms can result in similar manifestations of social behaviours (De la Torre-Ubieta, 2016). The amygdala plays a central role in social and emotional processing in autism (Baxter et al., 2012). Goodrich (2018) focused on the PAC1R gene in the developing rodent and human amygdala, which may act as a modifier of social behavioural phenotypes (Goodrich, 2018). PAC1R is expressed in the amygdala from mid-neurogenesis at critical periods to alter brain development trajectories (Goodrich, 2018). After fear conditioning, rodents had increased amygdala PAC1R expression (Dias and Ressler, 2012).

Chromosomal aberrations are estimated for 2-5% of autistic individuals

(Wiśniowiecka-Kowalnik and Nowakowska, 2019). They are most often found in cases of autism with associated dysmorphic features. Structural chromosomal alterations are rare and include deletions, duplications, inversions, and translocations. While their precise role in autism is undetermined, some are recurrent, such as the 15q11q13 duplication, effecting 1-3% of children with autism (Wiśniowiecka-Kowalnik and Nowakowska, 2019). One of the most studied neurotransmitter-related genes linked to autism are those of the GABA receptor cluster on chromosome 15q11q13. This is known as a complex locus containing three

GABAA receptor subunit genes; GABRB3, GABRA5, and GABRG3 (Hogart et al., 2007). A

developmental deficit of GABAA receptors function would affect neurogenesis and

maturation of neuronal network (Hogart et al., 2007). Among the different GABAA receptor genes listed above, the targeted deletion of the GABRB3 gene results in abnormalities reminiscent of the autistic phenotype, including social behaviour deficits, cognitive deficits,

motor stereotypes, and seizures (Pizzarelli and Cherubini, 2011). This could be because the GABRB3 gene encodes for the β3 subunit, which is highly expressed developmental stages (Pizzarelli and Cherubini, 2011). Consortium studies of autism have that revealed copy number variants (CNVs) and single gene mutations in GABA receptor subunit genes are present in a significant percentage of cases of autism (Hogart et al., 2009; Ma et al., 2005; Silverman et al., 2015). GABRB3, GABRA5, and GABRG3 genes have been described as specific risk factors (McCauley et al., 2004), and GABRA4 was found to be involved in autism on its own, and also through interaction with GABRB1 (Ma et al., 2005).

The physiological nature of certain cell types renders them uniquely vulnerable to perturbations of broadly expressed genes (Xu et al., 2014). Several studies have identified mutations in different GABAA receptor subunits as risk factors for the development of autism (Coghlan et al., 2012) and autism-related genes have been found to be mostly expressed in GABAergic interneurons (Xu et al., 2014). Mutations in interneuron-associated genes DLX1 and DLX2 have been described as susceptibility factors for autism (Guptill et al., 2007; Deidda, Bozarth and Cancedda, 2014).

Fast-spiking parvalbumin (PV) GABAergic interneurons are pivotal in regulating the pyramidal neuron activity that ultimately instigates behaviours (Ferguson and Gao, 2018). In both autism and SSD, PV GABAergic interneurons have diminished functioning in the prefrontal cortex (PFC) and have been implicated as responsible for maintaining a suitable balance of excitation and inhibition in cortical circuits throughout the brain (Ferguson and Gao, 2018). Their important role in cortical micro-circuitry influences social interaction and emotional regulation (Ferguson and Gao, 2018). A significant loss of PV interneurons has been detected in the somatosensory cortex of FMRP, MeCP2 and Neuroligin 3 mutants, valproic acid-treated mice (Gogolla et al., 2009), and Engrailed-2 null mutant (En2−/−_{) mice} (Sgadò et al., 2013). These genes (En2, FMRP, MeCP2 and Neuroligin 3) and epigenetic

mechanisms (valproic acid exposure) have also been implicated in autism observed in humans (Ajram et al., 2019; Brielmaier et al., 2012).

There are several ways by which changes to glutamate / GABA metabolism can lead to an excitation / inhibition imbalance (Ajram et al., 2019). Ajram et al. (2019) highlighted many genetic contributions to GABA (and glutamate) signalling pathways, including mutations on Neuroligin 3 and Neurexin 1, polysynaptic density protein 95, SHANK3,

CNTNAP2, and GABAA receptor subtype genes (Ajram et al., 2019). Deletion of MeCP2 in

GABAergic neurons supports an important role for the corresponding protein in the function of inhibitory neurons (Chao et al., 2010). Chao et al. (2010) demonstrate that mice with a loss of MeCP2 in GABAergic neurons develop autistic traits including social processing deficits (Chao et al., 2010).

Some autistic individuals carry mutations in genes encoding postsynaptic cell-adhesion molecules called neuroligins (Nlgs) (Rubenstein and Merzenich, 2003). Nlgs are postsynaptic adhesion molecules, often associated with presynaptic neurexins (Corthals et al., 2017). They are central in the formation of signalling complexes that allow for the

maturation, and functional alterations of synaptic connections between neurons (Corthals et al., 2017). Nlg sub-types are expressed differentially for certain synapses; Nlg1 is

predominantly expressed at excitatory glutamatergic synapses, while Nlg2 is selectively expressed at inhibitory synapses, where it recruits GABA receptors (Varoqueaux, 2004). While Nlg3 and Nlg4 appear at both excitatory and inhibitory synapses, Nlg3 has a preference for GABAergic synapses (Corthals et al., 2017). Alterations in Nlg genes have been identified in autism (Jamain et al., 2008; Ornoy et al., 2016) and mutations of the same genes caused autism-like phenotypes in rodent models (Südhof, 2008). Tabuchi et al. (2007) explored one such mutation in mice (the Arg451→Cys451 [R451C] substitution in Nlg3). Their mutant mice presented with impaired social interactions but advanced spatial learning

capacities. An increase in inhibitory synaptic transmission accompanied these behavioural changes. Such alterations were not observed in mice with Nlg3 deletion, which suggests the R451C mutation results in a gain of function (Tabuchi et al., 2007). These study findings imply that increased inhibitory synaptic transmission may contribute to autism, which were unforeseen as autism thought to be associated with a loss of inhibitory drive (Tabuchi et al., 2007; Rubenstein and Merzenich, 2003; Harada et al., 2011).

Results of studies compiled by Wiśniowiecka-Kowalnik and Nowakowska (2019) further indicate that an imbalance in excitation and inhibition synaptic inputs could explain deficits in social and cognitive functions present in autistic individuals. Mutations and CNVs implicate genes that encode proteins involved in chromatin remodelling (CHD8, BAF155), as well as synaptic cell adhesion molecules (Neurexin and Neuroligin families, CNTN4),

neurotransmitters, scaffolding protein in synapse (SHANK2 and SHANK3), and ion channel proteins (CACNA1A, CACNA1H, SCN1A, SCN2A) (Wiśniowiecka-Kowalnik and

Nowakowska, 2019). Implicated proteins are also involved in neuronal networks that

influence synaptic gene transcription and translation pathways (FMR1, TSC1, TSC2, PTEN, NF1, CYF1P1), ubiquitination pathway (UBE3A, PARK2, TRIM33), and also participate in protein synthesis and degradation, in addition to the development, formation, and function of synapses and neurons (Wiśniowiecka-Kowalnik and Nowakowska, 2019).

CNVs are a type of structural variation in the genome, which can be duplication or deletion events (Thapar and Cooper, 2013). They affect multiple base pairs and

are considered a central class of risk factors for several neuropsychiatric conditions (Thapar and Cooper, 2013). Evidence exists for the association between CNVs and autism. Most CNV investigations have focused on the diagnostic criteria for autism, rather than attempting to understand of the impact of CNVs on ‘sub-phenotypes’ such as social processing deficits (Merikangas, 2015). A recent study was conducted in autistic children to investigate whether

the presence of CNVs was associated with particular clinical features (Napoli et al., 2018). The collated data indicated that autistic individuals with causative CNVs exhibit more severe symptoms and experience more severe difficulties in communication skills than individuals with non-causative CNVs or without CNVs (Napoli et al., 2018). Other studies similarly suggest that combination of rare or de novo CNVs and common genetic variants have a significant influence on phenotypic diversity (Bourgeron, 2016; Vicari et al., 2019). Merikangas et al. (2015) analysed 1590 individuals with autism to examine phenotypic manifestations of specific CNVs, including CNVs previously implicated in autism with intellectual disability, and CNVs involving genes specifically expressed in brain

(Merikangas, 2015). The results showed that CNVs previously implicated in autism with intellectual disability were associated with communication and language delay, while those considered differentially expressed genes related to a broad range of presentations of adaptive functioning (Merikangas, 2015; Vicari et al., 2019).

While there is data indicating the presence of genotype-phenotype associations relating to social processing deficits, more targeted studies are required to determine whether risk variants are associated with adaptive abilities or with manifestations of symptomatology in autism. Discovering candidate genes is an essential step in the diagnostic process, and more comprehensive analysis of the underlying cellular pathways and processes will be necessary to understand the pathophysiology of neurodivergent symptomatology. The numerous susceptibility genes potentially contributing to autism likely act in combination with complex gene-environment interactions resulting in alterations in brain development. Other cohorts of genes that exclusively influence symptomatology must be taken into consideration, whether they behave in a homeostatic manner to ameliorate autistic

possible role of environmental and epigenetic elements in phenotype modulation (Vicari et al., 2019).

Potential epigenetic mechanisms associated with social processing deficits

Epigenetic mechanisms can be heritable or environmental changes in gene expression that occur through covalent modifications of chromatin (the DNA–histone protein complex in the cell nucleus) (Roth et al., 2009), without changing the underlying DNA sequence

(Eshraghi et al., 2018). Epigenetic mechanisms result in enduring changes in gene activity states and effect transient variation in gene transcription necessary for activity-dependent modulation in gene expression related to cognitive processes (Roth et al., 2009). Autism, which is considered a multigenic condition, is recognised as dependent upon a myriad of epigenetic factors (Loke et al., 2015). While effects are still largely unknown, many studies have investigated the transgenerational inheritance of autistic symptoms in order to establish epigenetic marks for autistic aetiology (Choi et al., 2016; Pu et al., 2013).

Environmental factors have been shown to exert some control in epigenetic determination. In rodents, valproic acid exposure led to autistic-like phenotypes in male offspring that can be epigenetically transmitted (Choi et al., 2016). Diet variation has also been implicated as instrumental in the development of autism. An enzyme coded by gene MTHFR is essential for generating 5-methyl-tetrahydrofolatein and plays a major role in DNA methylation during neural development, which is associated with a significant increase in risk for developing autism (Pu et al., 2013). A meta-analysis that compared autism

prevalence in children from countries with folate acid fortified diet and those who are without food fortification, established an association between MTHFR C677T polymorphism and children without fortified food (Pu et al., 2013).

Activation of immune responses that determine susceptibility to autism can be influenced by epigenetics. A link between increased risk of autism and maternal immune

activation (MIA) during pregnancy has been suggested (Wiśniowiecka-Kowalnik and Nowakowska, 2019). A self-report population study in Denmark found an increase in prevalence of infantile autism in offspring of mothers that suffered from influenza infection during pregnancy and a further increased prevalence of autism in children of mothers who experienced a prolonged period of fever (Atladóttir et al., 2012). Chronic asthma is one of the most common causes of MIA (Gunawardhana et al., 2014). Epigenetic changes including hypermethylation (of FAM181A, CHFR, and AURKA genes) and hypomethylation (in MAP8KIP3 and NALP1L5 immune genes) in foetal blood samples have been identified in individuals exposed to maternal asthma (Gunawardhana et al., 2014). The results of the study with a mouse model of maternal allergic asthma (MAA) suggested that alterations in

maternal immune responses during pregnancy increase the risk of autism. MAA in mice caused changes in social and repetitive behaviours in the offspring (Vogel Ciernia et al., 2017). Through differential expression analysis, genes involved in controlling microglial sensitivity to the environment and shaping neuronal connections in the developing brain were identified (Vogel Ciernia et al., 2017). Differentially expressed genes significantly

overlapped genes with altered expression in the cortex in autism, supporting a role for microglia in the pathogenesis of autism (Vogel Ciernia et al., 2017).

In SSD, there is growing evidence that DNA methylation and the associated chromatin remodeling contributes to dysfunction of GABAergic neurons, playing a role in the down-regulation of reelin (RELN) and GAD1 (Roth et al., 2009). Tremolizzo et al. (2002), observed that chronic treatment of mice with L-methionine (MET), produces a schizophrenic-like phenotype, replicating some of the molecular aspects of SSD, including increased RELN promoter methylation and down-regulation of RELN and GAD1 in GABAergic neurons (Tremolizzo et al., 2002). Dong et al. (2004) then demonstrated that both RELN and GAD1 promoters show increased recruitment of methyl-CpG binding

proteins, such as MeCP2 (Dong et al., 2005). While there is strong evidence for epigenetic changes in RELN and GAD1 in SSD, it is not suggested that epigenetic dysfunction alone results in susceptibility to the disorder (Roth et al., 2009). As shown by research using genome-wide epigenetic approaches, many genes are likely implicated (Mill et al., 2008).

Epigenetic studies can be formulated through finding indicators of causative genetic and environmental factors (Eshraghi et al., 2018). While the investigations discussed mainly focus on providing greater insight into the physiological mechanisms that predispose

individuals to autism, it would be fair to assume specific phenotypes are also influenced by epigenetic mechanisms. New biomarkers for autism severity will aid medical practitioners to more precisely fine-tune the plans of treatment (Eshraghi et al., 2018). It is evident that epigenetic mechanisms associated with specific symptoms are not widely recognised, however, some gene-targeted medications intend to use epigenetics as a tool for

pharmacological progress. A study by Qin et al (2018) found that treatment with romidepsin alleviated social processing deficits in SHANK-3 deficient mice. In doing so, this study highlights an epigenetic mechanism underlying social processing deficits linked to SHANK3 deficiency, demonstrating the connectedness of deviations in synaptic, transcriptional and epigenetic pathways. The study also illuminates potential therapeutic strategies for

individuals bearing SHANK3 mutations (Qin et al., 2018).

Epigenetic mechanisms provide molecular explanations for changes in brain function and influence the dynamic nature of the CNS in response to environmental stimuli (Roth et al., 2009). While there is emerging data regarding the role of DNMT and HDAC inhibitors (that induce DNA demethylation and histone acetylation) in alleviating cognitive deficits in rodents (Szyf, 2009; Deutsch et al., 2008), future clinical research should continue to

examine whether epigenetic drugs can alleviate cognitive deficits associated with autism and SSD. During efforts to elucidate the role of epigenetics that underly social processing

deficits, standardisation of symptoms across participant cohorts will assist in managing epigenetic complexities. Larger sample sizes can allow for better confidence in gene-gene and gene-environment interactions observed (Eshraghi et al., 2018). A collective systems-approach has been advised for future studies, necessitating a compilation and integration of findings (Eshraghi et al., 2018).

Future directions and conclusion

As discussed throughout this review, neural substrates underlying autistic phenotypes may be an imbalance of excitation and inhibition in neural circuits (Rubenstein and

Merzenich, 2003). The increased ratio of excitation / inhibition can arise from the

GABAergic dysfunction (Pizzarelli and Cherubini, 2011). Providing functional connectivity level support for excitation / inhibition ratio imbalance, Guo et al. (2019) recently

demonstrated dysfunction of the pSTS, fusiform gyrus, temporal gyrus, and posterior cingulate cortex (PCC) in social cognition in autism. These abnormalities in dynamic functional coordination are related to social impairments in individuals in autism (Lynch et al., 2013). Variability in the dynamic functional connectivity in the pSTS is related to GABA receptor genes expression (Guo et al., 2019). This particular study exemplified an integration of neuroimaging techniques with gene expression decoding analysis to address a common underlying hypothesis.

Dissimilar phenotypic presentations of spectrum conditions indicate aetiological heterogeneity, so determining treatment targets is complex, especially without means to identify biologically responsive individuals (Arjam et al., 2019). Robust evidence indicating GABA modulation as clinically effective for treating the core social interaction deficits observed in autistic humans is still lacking (Selimbeyoglu et al., 2017). However, structural or functional abnormalities in inhibitory neurons in genetic models of cognitive dysfunction continue to be highlighted, and improvements in social behaviours were observed through

systemic pharmacological modulation of GABA transmission (Silverman et al., 2015; Han et al., 2014; Han et al., 2012). At this stage, no pharmacological interventions for core social processing deficits have been approved (Selimbeyoglu et al., 2017). Current therapeutic options for such deficits are limited to behavioural interventions, which often rely upon consistent early intervention.

Continuing to investigate GABA’s role using 1H-MRS holds promise for ascertaining the impact of divergent GABA levels in autism aetiology as well as elucidating phenotypic

ambiguities. To date, 1H-MRS findings concerning GABA levels have shown reduced

concentration in autism (Horder et al., 2018; Gaetz et al., 2014; Harada et al., 2011); however, there are few 1H-MRS studies of GABA, and most studies tend to include small numbers of subjects. Hence, 1HMRS data relating excitation / inhibition imbalance to neuro-psychiatric phenotypes requires extensive maturation. Current findings do still substantiate the importance of GABA in the aetiology of neurodevelopmental disorders (Schür et al., 2016). While evidence suggests that the GABA system remains a promising target for pharmacological interventions for autism, further research could benefit from increased fundamental knowledge of the physiological variation in GABA levels, longitudinal studies, and consensus on the preferred methodology of human 1H-MRS studies (Schür et al., 2016). Far beyond these essential developments lies the potential of diagnostic precision and personalised treatments based upon detailed and accurate brain GABA level measures.

Somewhat counterintuitively, there may also be value in using typically developing individuals to study inhibitory and excitatory signalling related to the social processing deficits observed in neuropsychiatric disorders such as autism and SSD (Baron-Cohen, Wheelwright, Skinner, Martin and Clubley, 2001; Dinsdale, Hurd, Wakabayashi, Elliot and Crespi, 2013; Ford and Crewther, 2016). This is because there are likely to be varying degrees of detectable social processing deficits present within nonclinical population and

dimension-based approaches are necessary for consolidating current understanding of inhibitory and excitatory signalling in clinical phenotypes, particularly in regards to

multidimensional spectrum conditions (Baron-Cohen et al., 2001; Dinsdale et al., 2013; Ford and Crewther, 2016; Ford et al., 2019).

In sum, neurochemical abnormalities may be more specific to spectrum phenotypes rather than spectrum conditions as a whole, with an increased excitatory / inhibitory ratio perhaps more closely related to the social processing deficits. Further research into this area is therefore warranted in order to investigate the translation of these findings to the clinical population. Such investigations may ultimately lead to pharmaceutical interventions that target the glutamatergic and GABAergic systems, systems that may be more effective in reducing the severity of pervasive psychosocial symptoms (Brondino et al., 2015; Bruining et al., 2015). The integration of genetics with clinical phenotypes and other functional genomic techniques such as transcriptomics and epigenomics will lead to a better understanding of the molecular mechanisms involved in autism and ultimately inform clinical care (Loke et al., 2015). Classified genetic variants as etiological factors are identified only in about 25–35% of the cases of autism (Bourgeron, 2015). Some genetic variants associated with autism, such as SHANK3 mutations (Zhou et al., 2016), CNTNAP2 (Strauss et al., 2006), and 16p13.11 microdeletion / microduplication (Pinto et al., 2010), are also relevant to other

neuropsychiatric conditions such as SSD, making the assessment of genetic risk more complex (Wiśniowiecka-Kowalnik and Nowakowska, 2019). Continued investigations with specific objectives are required for better understanding of autism aetiology and distinctive phenotypes among individuals (Wiśniowiecka-Kowalnik and Nowakowska, 2019).

This review has highlighted some evidence that the social behaviour deficits observed in autism and other neuropsychiatric disorders can be characterised by abnormalities in GABA function. The necessity to utilize evidence from several fields, including genetics,

epigenetics, neuroimaging, animal models and post-mortem studies of humans, demonstrates the importance of a multidisciplinary outlook when investigating GABA functioning in brain regions associated with the social processing deficits. While suggesting that more research is needed is somewhat a truism across the discipline of neuroscience, this particular premise shows emerging promise, and the necessity to address social processing deficits in clinical settings is unarguably fundamental for the individuals experiencing such symptoms.

Continuing to directly examine the GABA system in the brains of living humans is urgently warranted and could positively influence pharmacological developments. The

implementation of standardised phenotyping in future investigations would be an appropriate approach in refining evidence and longitudinal developmental studies across both child and adult development would be beneficial in establishing GABA trajectories related to social processing deficits.

Study Rodent phenotype

Molecular, cellular or circuit phenotypes

Measurement / method Target symptoms / behaviours

Results / Molecular function Region of interest Treatment

Bredewold et al. 2015

- Wistar rats - GABA and glutamate modulate lateral septum neuronal activity - Bicuculline is a light-sensitive competitive antagonist of GABAA receptors - Treatment administration - Intracerebral micro-dialysis

- Social play behaviours Bicuculline (dose-dependently) decreased the duration of social play behaviour.

Suggested role for GABA neurotransmission in the lateral septum in the regulation of social play.

- Lateral septum Bicuculline

Brielmaier et al. 2013 - Engrailed-2 null mutants (En2−/−) As above - Neuroanatomical expression profiling - Phenotyping of behaviours - Deficits in reciprocal social interactions as juveniles and adults - Absence of sociability in

adults

En2 is expressed in adult brain structures including the somatosensory cortex, hippocampus, striatum, thalamus, hypothalamus and brainstem.

En2 mutants exhibited: - Fear conditioning impairments - Water maze learning deficits - High immobility

- Reduced prepulse inhibition, - Mild motor coordination impairments - Reduced grip strength

- No differences found on measures of ultrasonic vocalizations in social contexts

- No stereotyped or repetitive behaviours

Disturbances in En2 signalling thought to contribute to social and cognitive deficits.

- Somatosensory cortex - Hippocampus - Striatum - Thalamus - Hypothalamus - Brainstem N / A Etherton et al. 2011. - Neuroligin 3 R451C (NL3R451C) mutation - Increase in AMPA receptor-mediated synaptic transmission - Enhancement of NMDA receptor-mediated transmission - Altered subunit composition of postsynaptic NMDA receptors - Behavioural assays (examination of sociability phenotype) - Electrophysiological recordings - Social interaction deficits

Mutation alters synaptic function in a circuit-dependent manner (effects dependent on synaptic context). - Hippocampal CA1 region: enhances excitatory synaptic

transmission, increases dendritic branching and changes glutamate receptor composition of synapses

- Somatosensory cortex: enhances inhibitory synaptic transmission. - Hippocampal CA1 region - Somatosensory cortex N / A

2018 (standard control strain) used to investigate Pac1r

neural signalling processes essential for social function and emotional processing

(ISH)

- Interactome and gene ontology analysis

deficits - Altered brain

connectivity

zones for generating neurons of the amygdala. High levels of Pac1r expression in emerging mouse amygdala nuclei found at early postnatal stages.

Rodents show increased Pac1r expression in the amygdala after fear conditioning.

Han, et al. 2012 _{- SCN1A+/−} haploinsufficie ncy

Impaired GABAergic neurotransmission due to decreased NaV1.1 activity in GABAergic interneurons - Treatment administration - Behavioural analysis - Hyperactivity - Stereotyped behaviours - Social interaction deficits - Impaired spatial memory - Aversion to social odours

Medication rescued abnormal social behaviours and deficits in fear memory. - Hippocampal CA1 region - Prefrontal cortex Low-dose clonazepam Han, et al. 2014 - BTBR (idiopathic autism) *C57BL/6J and 129SvJ wild-type mice Reduced spontaneous GABAergic neurotransmission - Treatment administration - Behavioural analysis

- Core social interaction deficits

- Repetitive behaviours

Medication improved deficits in social interaction, repetitive behaviour, and spatial learning (positive allosteric

modulation of postsynaptic GABAA _{receptors increased}

inhibitory neurotransmission).

*Negative allosteric modulation of GABAA receptors impaired social behaviour in C57BL/6J and 129SvJ wild-type mice - Hippocampal CA1 region Low non- sedating/non-anxiolytic doses of benzodiazepines

Horder et al. 2018 - Neuroligin-3R451C knock-in (KI) - Neuroligin-3 knock-out (KO) - SHANK3 KO - Mice prenatally exposed to VPA (valproate)

SHANK3 knockout (KO) mice, Nlgn3 R451C knock-in

(KI) mice and Nlgn3 KO rats: Monogenic models carrying mutations in synapse-related genes (linked with autism in humans) Prenatal exposure to valproate (valproic acid, VPA), (an environmental risk factor for ASD)

- Proton magnetic resonance spectroscopy (1H MRS)

- Striatum implicated in social behaviours

Striatal glutamate deficit in VPA-exposed mice, Nlgn3R451C

KI mice, and Nlgn3 KO rats. reduction in striatal glutamate and glutamine levels in the SHANK3 KO mice.

Nlgn3 mutant mice and rats also showed reduced glutamate in the mPFC.

Results are suggestive of a region-specific imbalance due to reduced excitation. - Right striatum - Medial prefrontal cortex N / A Pizzarelli and Cherubini, 2013 - Neuroligin 3 R451C (NL3R451C) knock-in mice

NLs ensure correct cross talk between post and presynaptic specializations

- Electrophysiological recordings

N / A Enhanced frequency of Giant Depolarizing Potentials (GDPs), as compared to controls.

- Hippocampal CA3 region

transmission early in postnatal life may change the excitatory / inhibitory balance.

Rein et al. 2019 _{- SHANK3} deficient (SHANK3+/ΔC ) - BTBR (idiopathic autism) SHANK3+/ΔC mice: Synaptic dysfunction in the PFC and VTA

BTBR mice: changes in the shape and localization of hippocampus and amygdala

- Simplified behavioural approach

- Diminished intent to interact with social stimuli

Both SHANK3+/ΔC and BTBR: decreased open-arm engagement with social stimulus. Deficits not expressed under less pressured experimental conditions.

Compromised areas in mice models: - Pre-frontal cortex (SHANK3+/ΔC) - Ventral tegmental area (SHANK3+/ΔC) - Hippocampus (BTBR) - Amygdala (BTBR) N / A Selimbeyoglu et

al. 2017 - CNTNAP2-deficient mice Reduction in excitatory / inhibitory balance in the medial prefrontal cortex (mPFC) - Optogenetic modulations (increasing the excitability of inhibitory parvalbumin (PV) neurons and decreasing the excitability of excitatory pyramidal neurons)

- Novel interaction time Both (real-time and reversible) modulations acutely rescued deficits in social behaviour and hyperactivity in adult mice lacking CNTNAP2.

- Medial

Pre-frontal cortex N / A

Sgado et al. 2013 - Engrailed-2 null mutants (En2−/−)

Immature cortical connectivity

(En2 regulates neurogenesis and development of monoaminergic pathways)

- Immunohistochemistry - Reduced social interactions

- Locomotor impairment - Defective spatial

learning and memory

Reduced expression of GABAergic marker mRNAs Link between altered function of En2, anatomical deficits of GABAergic forebrain neurons and the pathogenesis of ASD.

- Hippocampal CA1 region - Prefrontal cortex N / A Silverman et al. 2015 - Breeding pairs of C57BL/6J (B6), BTBR T+ Itpr3tf/J (BTBR), C58/J (C58), and 129/SvImJ

Low sociability, high repetitive or stereotyped behaviours. - Treatment administration - Behavioural analysis - Impaired social interactions - Minimal vocalizations in social settings - High levels of repetitive

self-grooming and digging

- Cognitive deficits

Treatmentreversed social approach deficits in BTBRT+Itpr3tf/J (BTBR) and reduced repetitive self-grooming and high marble burying scores in BTBR. Stereotyped jumping in C58/J (C58), at non-sedating doses. S-baclofen produced minimal effects at the same doses.

N / A - R-baclofen enantiomer (GABAB receptor agonist) - S-baclofen enantiomer (less potent) Sungur et al. 2016 _{- SHANK1-/-}

null mutant - SHANK1+/- heterozygous - SHANK1+/+ wildtype controls

SHANKs interact directly or indirectly with all major types of glutamate receptors – NMDA receptors, AMPA receptors, and mGluRs

- Isolation-induced ultrasonic vocalizations (USV) analysis

- Communication/interact

ion deficits SHANK1-/-mice vocalized less and displayed a delay in _{typical use of USV.} Testing under social conditions exposed genotype-dependent deficits regardless of the familiarity of the social context.

communication/interaction deficits, confirming SHANK1’s involvement in acoustic communication across species. Tabuchi et al. 2007 - Neuroligin-3 R451C knock-in (KI) - Neuroligin-3 knock-out (KO) Neuroligin-3 R451C KI but not neuroligin-3 KO mice exhibit increased spontaneous inhibitory synaptic

transmission

- Whole cell recordings - Behavioural analysis

- Impaired social behaviours - Enhanced spatial

learning abilities

Selective increase in inhibitory synaptic strength in neuroligin-3 R451C KI but not in neuroligin-3 KO mice. Impaired social interaction behaviours in neuroligin-3 R451C KI mice.

R451C substitution increases inhibitory synaptic transmission without affecting excitatory synaptic transmission, simultaneously impairing and selectively enhancing specific behaviours (findings are surprising because autistic behaviours have been associated with loss of inhibitory drive). - Hippocampal CA1 region - Somatosensory cortex N / A

Yizhar et al. 2011 - ChR2 mutants (C128S, D156A and C128S/D156A)

Engineered microbial opsin tools (stabilized step-function opsin (SSFO)) used for interrogation of neuronal circuits during social behaviours

- Combinatorial optogenetics in vivo (novel optogenetic tools)

- Social functioning SSFO activation in prefrontal cortical excitatory neurons led to social impairments.

Causal support for the elevated cellular E/I balance hypothesis: identified circuit-physiology manifestations of the resulting social dysfunction.