GABA is an abundant inhibitory neurotransmitter in the central nervous system, and along with glutamate, an excitatory neurotransmitter, is responsible for a wide array of brain functions (Reis et al., 2009). It is a neutral amino acid, synthesized from glutamate by a cellular cytoplasmic enzyme called glutamate decarboxylase (Young and Chu, 1990) and plays an important role in maintaining both phasic as well as tonic inhibition in the brain (Sigel and Steinmann, 2012).

GABA, when released from synaptic vesicles, acts postsynaptically via GABA_A receptors, which are ligand-gated anion-selective channels producing hyperpolarization of neuronal membranes and hence brief lasting, but fast and phasic neuronal inhibition. In addition, lower concentrations of ambient extrasynaptic GABA are known to activate different subunits of the GABA_A receptor and hence produce persistent “tonic” neuronal inhibition (Farrant and Nusser, 2005). Any endogenous or exogenous processes that reduce this tonic neuronal inhibition are likely to contribute to heightened anxiety (Maguire et al., 2005). Interestingly, the phasic and tonic GABA inhibitions demonstrate a competitive homeostatic regulation of the total cortical inhibition (Wu et al., 2013). In addition, these inhibitory processes are under the regulatory control of neurosteroids—these are steroids synthesized in the brain (e.g., allopregnanolone) from circulating steroid hormone precursors like progesterone and deoxycorticosterone—via their actions on the GABA_A receptor (Reddy, 2010). Likewise, GABA_A-mediated tonic inhibition also regulates the hypothalamic response to stress by the release of cortisol peripherally (Lee and Maguire, 2014).

In contrast, GABA_B neurotransmission is a slow synaptic inhibition, but more robust and is mediated via metabotropic receptors that produce both, post-synaptic neuronal inhibition as well as pre-synaptic autoreceptor-driven fatigue of synaptic inhibition—these are mediated by different G proteins (Motohashi et al., 1989; Sigel and Steinmann, 2012). The downstream effects of GABA_B receptor activation are complex, partly owing to different isoforms (GB1a and GB1b) of the receptor and site of action (Doly et al., 2016). Nevertheless, it is clear from animal models and human in vivo experiments, that GABA_B-mediated neurotransmission plays a role in reward processing, social motivation, and mood regulation—important ingredients of adequate social functioning (Jacobson et al., 2018). For instance, GABA_B receptor-mediated tonic inhibition regulates the spontaneous firing of locus coeruleus neurons that release noradrenaline, a neurotransmitter critical in maintaining arousal and mood (Wang et al., 2015).

Glutamate, on the other hand, is an excitatory neurotransmitter and the most abundant neurotransmitter in the brain; glutamatergic neurons utilize ~ 60–80% of brain metabolic activity (Rothman et al., 2003). It is synthesized when the enzyme glutaminase facilitates deamidation of glutamine. Glutamate is responsible for a broad range of cortical functions that include but are not restricted to neuronal differentiation, neuronal migration, and synaptic plasticity, which support diverse cognitive abilities (Tapiero et al., 2002). Glutamate also exerts regulatory control over GABA-mediated inhibitory neurotransmission (Bak et al., 2006). Its multitude of pre-synaptic and post-synaptic effects are mediated by both metabotropic and ionotropic receptors. N-methyl-d-aspartate (NMDA) glutamate receptors are a specific class of ionotropic receptors that play a critical role in the development, as well as, the functioning of GABA interneurons (Cohen et al., 2015). A hypofunctioning NMDA-receptor system can potentially result in diminished GABA-interneuron function, an imbalance in the large-scale brain network balance maintained by the excitatory and inhibitory neurotransmitter systems, impaired functional connectivity and impaired synaptic plasticity leading to deficits in learning, memory, and predictive coding (Friston et al., 2016). This model of a GABA/glutamate imbalance resulting in social deficits is posited to underlie major psychiatric disorders including depression (Sanacora et al., 2012) and schizophrenia (Howes et al., 2015).

1.2 The clinical relevance of GABA and glutamate

The very abundance of these two neurotransmitters (GABA and glutamate), their overarching presence across diverse brain regions and their antagonistic functional properties, provides a mechanistic basis with which the brain executes control over behaviors. The best illustration of this is perhaps from the fact that GABA and glutamate are units of analysis (molecular level) for all the five behavioral domains within the Research Domain Criteria (RDoC) framework (Insel et al., 2010) to study psychiatric disorders. The RDoC is a novel initiative under the National Institute of Mental Health to study five psychological domains relevant to human behavior and functioning (negative and positive valence systems, cognitive systems, social processes systems, and arousal or regulatory systems) in a transdiagnostic manner from genomics to cellular physiology and finally behavior. The ultimate aim of this approach to studying behavioral abnormalities seen across various psychiatric disorders will be to develop precisional and individualized treatments (Insel, 2014). Any disparity in the optimal cortical excitation-inhibition balance can, therefore, express as behavioral abnormalities across the five RDoC domains. A common neurobiological template that illustrates the role of this GABA/glutamate balance is the prefrontal-amygdala fronto-limbic circuit, where GABAergic neurotransmission plays a central regulatory role over processing of emotions in the limbic cortex (Delli Pizzi et al., 2017). Various methods of investigating in vivo GABA functions in humans have indicated GABAergic neurotransmission abnormalities across a broad range of psychiatric disorders—schizophrenia (Benes and Berretta, 2001), depression (Brambilla et al., 2003), anxiety (Long et al., 2013), post-traumatic stress disorder (Rosso et al., 2014), autism (Cochran et al., 2015), and dementia (Bai et al., 2015). Moreover, it is well known that chronic psychological stress can lead to a blunting of the hypothalamo-pituitary-adrenal axis response and hippocampal tissue loss—this effect is also associated with an imbalance in the GABA/glutamate tone (Yang, 2014).

Electro/magnetoencephalography, transcranial magnetic stimulation (TMS), magnetic resonance spectroscopy (MRS), and positron emission tomography (PET) have been used to probe GABA functions at different time-space resolutions. While early and late prepulse inhibitions response to sensory or motor stimuli as assessed by EEG/MEG (Inui et al., 2018) or TMS (McClintock et al., 2011), can reflect GABA_A and GABA_B receptor-mediated neurotransmission with good temporal resolution, MRS measures of GABA concentrations (Egerton et al., 2018) and PET profiling of GABA receptors (Bai et al., 2015; Frankle et al., 2015) provide good spatial resolution of GABAergic functions in the brain. These lines of in vivo measurements have been supplemented by postmortem brain tissue studies, in conjunction with animal model experiments, which consistently report of deficient glutamic acid decarboxylase (GAD-67) expression—a key enzyme in the biosynthesis of GABA—playing a role in a range of psychiatric disorders (Akbarian and Huang, 2006; Akbarian et al., 1995). A note of caution, there are inconsistencies and limited replicability in results from these novel, technologically intensive investigations characterizing GABAergic abnormalities in psychiatric disorders (Taylor and Tso, 2015). A multi-modal combination of different lines of GABA-measurement investigations, leveraging the pragmatic and temporo-spatial advantages of individual investigations, applied in longitudinal, large-scale dense time-series (repeated closely over a period of time) is likely to provide more consistent and reliable results across psychiatric disorders.

Nevertheless, one of the final common behavioral outcomes of these major psychiatric disorders is social withdrawal, often linked to poor quality of life and real-world functional disability. This could be a result of either the cognitive (Volk and Lewis, 2005) or affect processing deficits (Brambilla et al., 2003) triggered by the GABA-glutamate imbalance. It is well established by now that social information processing deficits and residual negative behavioral symptoms like amotivation and anhedonia predict real-world dysfunction in severe psychiatric disorders (Bhagyavathi et al., 2015; Green et al., 2012). Several lines of animal experiments demonstrate the effects of inefficient cortical GABAergic activity and social deficits (Chao et al., 2010). In patients with schizophrenia, where social deficits are a defining feature, GABA_A-mediated intracortical inhibition measured using paired-pulse TMS demonstrated a significant association with social cognition performance (Mehta et al., 2014b). Preclinical pharmacological studies also show a reversal in social deficits by administration of diazepam (Mielnik et al., 2014) and R-baclofen (Silverman et al., 2015)—modulators of GABA_A and GABA_B receptor neurotransmission, respectively. Any interventions targeted toward optimizing this excitation-inhibition balance may potentially be studied as therapies for improving social deficits across a range of psychiatric disorders.

In pursuing an understanding of the mechanisms underlying the therapeutic benefits of yoga, we take the example of depression, a common psychiatric disorder with perhaps the largest global disability burden across medical disorders (Vos et al., 2012). Depressive disorders are characterized by symptoms of intense low mood, anxiety, a lack of interest in pleasurable activities, easy tiredness, thoughts of hopelessness, helplessness, and worthlessness and a wide range of disturbances in sleep, appetite, sexual, and social functioning—lasting for 2 weeks or longer (American Psychiatric Association et al., 2000). Among the various neurobiological models of mood disorders like depression, there is multi-level support for the GABA hypothesis of emotion dysregulation. This is evident from observations of depression-specific reductions in the cerebrospinal fluid GABA levels (Gerner and Hare, 1981), lower MRS-measured GABA concentrations in the occipital and anterior cingulate cortices of depressed patients (Rao et al., 2011), and reduced TMS-measured short interval intracortical inhibition—a measure of GABA_A neurotransmission and cortical silent period—a measure of GABA_A neurotransmission—in the motor cortex of patients with depressive disorders (Radhu et al., 2013). Interestingly, lower GABA levels in the subgenual anterior cingulate cortex of depressed individuals (Gabbay et al., 2012) complement findings of reduced blood flow and glucose metabolism in the same region during depression (Drevets et al., 1997), thus highlighting the role of an inefficient prefrontal default mode control over limbic and executive networks (Northoff and Sibille, 2014). Studies from human social experiments suggest that social stress increases peripheral (serum or salivary) cortisol levels; this is often associated with the symptoms of social isolation or anxiety—this being modulated by sex hormones like testosterone (Lozza et al., 2017) or the bonding hormone, oxytocin (Heinrichs et al., 2003). Interestingly, evidence from preclinical animal experiments, as well as, human experiments suggests a reciprocal regulatory relationship between prefrontal GABA-glutamate equilibrium, and the functioning of the hypothalamo-pituitary-adrenal axis that modulates cortisol release to stress (Mody and Maguire, 2012). Patients with depression are also likely to have increased peripheral cortisol levels as compared to healthy comparison subjects; this elevated cortisol is reduced by antidepressant therapy with medications and yoga (Thirthalli et al., 2013). This study also demonstrated a significant positive association between a drop in depression severity ratings and drop in peripheral serum cortisol levels in patients who received yoga therapy.

4.2 Connecting the dots: The mechanistic basis of yoga—Effects on GABA

Indeed, the most common biological mechanisms of yoga that have been studied over the years are physiological markers of perceived stress (peripheral cortisol levels and sympathetic nervous system activity). A meta-analysis examining the effects of yogasanas with or without mindfulness-based meditation techniques, on such physiological markers of stress, reports diminished peripheral cortisol levels, and regulation of sympathetic activity—diminished blood pressure and heart rate (Pascoe et al., 2017) after yoga.

It is in this context, that it would be critical to examine if the cortisol reducing effects of yoga are actually related to the effects of yoga on the GABA-glutamate equilibrium in the brain. This has been proposed as a putative mechanistic framework for the effects of meditation that can be empirically examined (Austin, 2013; Elias and Wilson, 1995) and extended to the practice of yoga. There is indeed emerging evidence from human studies that have employed magnetic resonance spectroscopy (MRS) to study cortical GABA responses to yoga therapy. A pilot MRS study revealed a 27% increase in whole-slab GABA levels in regular yoga practitioners after 60-min of yoga practice, compared to a similar duration of reading (Streeter et al., 2007). Subsequent analyses revealed that this increase was best observed in the thalamus. In an ensuing study by the same group, healthy individuals were randomized to receive 12 weeks of yoga therapy or metabolically matched intermittent walking. The yoga group demonstrated a significantly greater increase in left thalamus GABA levels on MRS, which correlated with their self-reported mood and anxiety state improvements (Streeter et al., 2010). This finding indeed highlights how yoga therapy is likely to drive improvement in mood and anxiety symptoms in clinical populations by engaging with the cortical GABAergic system. In a clinical proof-of-concept demonstration, the same group was able to replicate these GABA-lowering effects of 12 weeks yoga therapy in two patients with major depression and chronic low backache, in conjunction with clinical improvement (Streeter et al., 2012). Future studies may examine this “engagement” of GABAergic function with yoga in patients with depressive or anxiety disorders. Furthermore, it may be examined if increasing cortical GABA levels are associated with a decrease in peripheral cortisol levels with yoga therapy.

Cross-validating these MRS findings of elevated cortical GABA levels post-yoga, with other in vivo measurements of GABA function can further strengthen this hypothesis of yoga driving the GABA-glutamate equilibrium for optimal social functioning. TMS-derived cortical silent period (CSP) is the GABA_B-mediated isoelectric electromyograph obtained secondary to suppression of motor activity when a suprathreshold stimulus is delivered to the motor cortex hand area when the contralateral hand is voluntarily held in contraction (Kobayashi and Pascual-Leone, 2003). Fig. 1 gives an illustration of this phenomenon. In a pilot observation of healthy volunteers recruited for a TMS experiment (Mehta et al., 2014c), we identified six participants who were experienced-yoga practitioners (average of 10-years yoga practice). We compared their TMS-derived cortical inhibition parameters with 10 age, gender, and education-matched healthy individuals who had not practiced yoga regularly. The two groups (n = 6 experienced-yoga practitioners; n = 10 non-practitioners of yoga) were comparable on TMS-derived resting motor threshold and short and long interval cortical inhibition (Mann-Whitney U > 31; P > 0.3). However, the yoga group had significantly longer CSP compared to the non-yoga group (Mann-Whitney U = 72; P = 0.016; see Fig. 2). We examined these effects in a prospective study, where TMS was used to measure cortical inhibition (short interval intracortical inhibition and CSP) before and after 1 month of yoga training (average of 18 sessions of 60-min duration yoga training) in beginners attending a yoga appreciation course. We found significantly enhanced CSP following the yoga training; this also had a positive correlation with the number of yoga training sessions attended (Govindaraj et al., 2018). Despite the absence of a control arm, this study reaffirmed our initial pilot observations of enhanced GABA_B-mediated cortical inhibition after yoga training in a dose-dependent manner. In an independent study (Guglietti et al., 2013), a similar effect of enhanced GABA_B-driven cortical silent period duration was found in experienced meditators after single session meditation practice compared to television viewing (control task) in a group of individuals who were non-meditators. Interestingly, there was no effect of meditation practice on GABA_A-mediated short interval intracortical inhibition in this study as well. Another study examining brief (5 days) yoga therapy versus intermittent walking in a randomized controlled experiment on 40 patients with depression, demonstrated a significant enhancement of the diminished CSP in the yoga group but not in the walking group (Jakhar, 2017). This rather specific effect of yoga and meditation on GABA_B-mediated neurotransmission may be studied as a neuromarker of treatment response to yoga therapy. However, not all in vivo human studies examining the effect of yoga on GABA have found a GABA enhancing effect. A randomized controlled trial examining the effects of 12 weeks of memory enhancement training (n = 11) or equal duration of yogic meditation (n = 14) in elderly individuals with mild cognitive impairment demonstrated no change in GABA levels in the dorsal anterior cingulate cortex and hippocampus as measured by MRS; there was no correlation between change in GABA levels and memory improvement in either group (Yang et al., 2016). The impact of clinical diagnosis, the age of the individuals, and site of examining GABA using MRS could explain the discrepancies observed in this study as opposed to earlier studies quoted above. Nevertheless, it is important to further contextualize these observed effects of yoga on GABAergic systems, elicited via MRS and TMS studies in keeping with other theories of the mechanism of yoga. An active component of yoga practice—om chanting, which may enable experiencing vibratory sensations were studied using functional imaging. When compared to rest states, om chanting produced deactivations in several limbic structures (cingulate, parahippocampal gyri, and amygdala)—such an effect was not observed with a control condition (Gangadhar et al., 2011). These observations also support the view of an enhanced GABA tone and hence inhibition of the prefrontal cortex regulation exerted on the lower limbic regions, thus resulting in limbic deactivations. This regulation of the excitation-inhibition equilibrium in the frontal cortex can have beneficial effects on social cognition, bonding, connectedness, and behaviors. It is in this context that the GABA regulation via yoga can be understood as a mechanism that (a) reduces perceived stress by regulating the release of cortisol and (b) increases pro-social cognitions and behaviors via its modulation of the neuropeptidergic system in the hypothalamus.

4.3 Connecting the dots: The mechanistic basis of yoga—Effects on oxytocin

Oxytocin is a social neuropeptide—secreted by the hypothalamus, that, along with vasopressin, regulates a broad range of social behaviors in humans. These behaviors include coping to stress (Neumann, 2002), empathy (Rodrigues et al., 2009), pair bonding (Carter et al., 2008), social anxiety, and social cognition (Heinrichs and Domes, 2008; Veenema and Neumann, 2008). Recent advances in biochemical estimations of peripheral oxytocin levels (Carter et al., 2007) and genotyping of the oxytocin receptor gene (Tost et al., 2010) have demonstrated important associations between peripheral oxytocin levels, oxytocin receptor gene polymorphisms and psychopathology in depression and schizophrenia. Similarly, in schizophrenia higher peripheral oxytocin is found to be associated with greater pro-social behaviors and less severe positive symptoms of delusions and hallucinations (Rubin et al., 2010). An inverse relationship between peripheral oxytocin and depression severity has been found consistently in pregnant women; however, this relationship varies in non-pregnant women and men, indicating other factors that regulate this relationship (Massey et al., 2016). Recent work has demonstrated how the practice of imitation and being imitated—important ingredients of yogasanas, especially in a group setting or a teacher-student setting—can act as an endogenous oxytocin nebulizer (Aoki et al., 2014; Delaveau et al., 2015). This is supported by observations from a randomized controlled trial that demonstrated significantly greater elevations of serum oxytocin levels after yoga therapy but not after a waitlisted period in patients with schizophrenia (Jayaram et al., 2013). Similarly, a mindfulness-based mind-body intervention demonstrated enhanced salivary cortisol in cancer survivors, compared to sleep hygiene education (Lipschitz et al., 2015). Interestingly, intranasal oxytocin administration is shown to enhance mu suppression—an indirect measurement of mirror neuron system activity (Perry et al., 2010). The mirror neuron system is a promising social brain substrate that is likely to be involved in the reflexive, embodied simulation-driven deciphering of intentions underlying actions, empathy, and related social cognitive processes (Carr et al., 2003; Iacoboni et al., 1999). In clinical populations, a deficient mirror neuron system activity, predominantly in untreated schizophrenia patients, was found to be associated with social cognition deficits in theory of mind and emotion processing (Mehta et al., 2014c). In contrast, a heightened responsivity (possibly secondary to frontal disinhibition) of this neural network is associated with hyper-imitative behaviors (Mehta et al., 2013a), hallucinations (McCormick et al., 2012), and affective instability or mood dysregulations (Mehta et al., 2014a). Given its role in social information processing and social behavior determination, it is intuitive to surmise the role of mirror neuron system activation in yoga interventions, which involve a range of imitative exercises, that are often performed in group-settings. Based on these observations, we have earlier proposed that this property of yogasanas, where imitation and being imitated are integral, can potentially facilitate optimal activations of the mirror neuron system via oxytocin release, and hence result in the improvements in social cognition and behavior performances, especially in patients with schizophrenia (Mehta et al., 2016). Studies examining the impact of group-settings of yoga practice versus individual or solo yoga practice on contemporary measurements of putative mirror neuron system activity in the premotor cortex, inferior frontal gyrus, and inferior parietal lobule, will help in refining our understanding of mechanisms underlying yogasanas.

4.4 The GABA and oxytocin synchrony

This brings the focus to the inter-dependent relationship between GABA and oxytocin neurotransmission. From a neurodevelopmental perspective, it has been demonstrated that maternal oxytocin plays a central role in switching the fetal immature excitatory GABA actions to mature inhibitory functions (Tyzio et al., 2006). Blocking this important neuroprotective process of the oxytocin-induced switch in GABA functions in rodents resulted in behaviors of social isolation, much like the autistic-like phenotype (Tyzio et al., 2014). In addition, we also now understand that oxytocin perhaps acts via GABA receptors (Gough, 2015), in addition to its effects via the oxytocin receptors—both these receptors are found in abundance in the central nucleus of the amygdala, an important center in the social brain. Recent animal experiments have emphasized the inter-dependent role of oxytocin and GABA in the amygdala to regulate social responses. For instance, treatment of prairie voles with oxytocin after a stress-inducing protocol results in recruitment of GABAergic neurons, which in turn inhibit stress-related activation of the hypothalamo-pituitary-adrenal axis (Smith et al., 2016). This cascade of neurochemical events is also likely to confer neuroprotection against inflammation and oxidative stress, common mechanisms of stress-induced neuronal damage (Kaneko et al., 2016). Despite limited evidence for this interactive, inter-dependent relationship between oxytocin, GABA, and cortisol in humans, indirect evidence from human clinical trials suggests that oxytocin administration improves sleep architecture just like benzodiazepines (Braga et al., 2014).

In summary, the practice of yoga is associated with the often self-reported feelings of enhanced social connectedness (Kinser et al., 2013) and the congruent objective improvements in mood (Cramer et al., 2013b), anxiety (Cramer et al., 2018), social behaviors of schizophrenia (Duraiswamy et al., 2007), and autism (Radhakrishna, 2010) and social cognition (Behere et al., 2011) in clinical populations. These “therapeutic” gains are likely to be driven by the GABA and oxytocin enhancing effects secondary to a broad range of components within yoga therapy described above. Oxytocin-dependent or independent GABA enhancement in these clinical conditions can improve social behaviors via its actions on cortisol. Independently, the oxytocin enhancing effects of yoga can modulate the mirror neuron system, thus driving pro-social behaviors. This integrated neurobiological model of therapeutic gains of yoga (Fig. 3) can be studied at several levels of analyses outlined by the RDoC criteria.

However, we concede that not many components of this model have been thoroughly examined with yoga therapy. Although emerging evidence supports the oxytocin and GABA enhancing effects of yoga, these findings need to be replicated, more so in clinical settings. The subsequent chain of physiological effects from oxytocin and GABA elevation to improve social functioning, via regulated cortisol response and mirror neuron system activation is yet to be examined empirically. The influence of independent components of yoga therapy—focused attention, posture-based exercises, chanting, breathing regulation, etc., on individual components of this model are also not teased out; most of the evidence for biological changes with yoga comes from the use of integrated yoga modules. Last, there might be other potential mechanisms (e.g., altering functional connectivity, synaptic plasticity, regulating autonomic brain functions) how yoga might exert its therapeutic gains in clinical conditions, which have not been alluded to in this model. It is likely that the active pathways mentioned in our model are related to some of these processes. These can be therefore added to the model based on the accumulation of stronger and consistent scientific evidence.

Acknowledgments

U.M.M. is supported by the Wellcome Trust/DBT India Alliance Early Career Fellowship, Grant/Award Number: IA/E/12/1/500755.

Funding

The funding agency had no role to play in the content and drafting of this chapter.

Conflict of interest

U.M.M. is one of the Associate Editors at Schizophrenia Research and receives honorarium from Elsevier for this service. None of the other authors report any potential conflict of interest.