Inhibitory control of the excitatory/inhibitory balance in psychiatric disorders

Introduction

Psychiatric disorders, including autism, schizophrenia, bipolar disorder, attention deficit hyperactivity disorder (ADHD) and depression, affect millions of people and are a major socio-economic burden^1–3. The identification of underlying genetic defects and risk factors is becoming increasingly efficient because of genome-wide interrogation methodologies, yet owing to the complex multifactorial origin of most cases, a conclusive molecular diagnosis is made for only a minority of patients. Therefore, the underlying causes for these conditions are poorly understood, and often treatment is still based on symptomology^4–6. In 2003, Rubenstein and Merzenich proposed autism spectrum disorders (ASDs) to be caused by an increase in the ratio between excitation and inhibition, called the E/I balance⁷. Since then, this hypothesis has been substantiated by a vast number of studies and also has been implicated in other psychiatric disorders such as schizophrenia⁸, consistent with their partially overlapping phenotypes⁹. Recently, the focus has shifted to changes to the inhibitory side of the E/I balance^10,11, in particular to one class of inhibitory neurons, parvalbumin (PV)-positive interneurons¹². In this review, we focus on the role of the inhibitory system in psychiatric disorders and explore the changes to the inhibitory systems in different disorders. We then discuss the role and function of PV interneurons and highlight the changes to this specific class of interneurons in the various psychiatric disorders.

Evidence for inhibitory dysfunction in psychiatric disorders

Since Rubenstein and Merzenich postulated their hypothesis of a reduced E/I balance in ASDs, there has been an increasing amount of evidence for disrupted inhibitory control in psychiatric disorders. This evidence comes from post-mortem studies and studies of patient phenotypes.

Firstly, post-mortem studies on patient brains have revealed consistent changes to the inhibitory system in various disorders. Studies of autistic brains revealed reduced expression of the gamma-aminobutyric acid (GABA) synthesizing enzymes GAD65 and GAD67, as well as various GABA receptor subunits, in parietal cortex and cerebellum^13,14. In schizophrenia, reductions of interneuron markers have been found in the prefrontal cortex^15–18, a region strongly implicated in this condition¹⁹. Interestingly, in recent years, this reduction has been shown to be caused by a reduction of the expression of the interneuron markers rather than a reduction of the number of interneurons^20–22, which indicates reduced activity of these neurons^23–25. In addition, both increased and decreased numbers of specific interneuronal subtypes are reported in bipolar disorder^17,26, while a reduced inhibitory function is reported in depression^27,28 and bipolar disorder²⁹.

Secondly, patients with psychiatric disorders display phenotypes that are strongly correlated to impaired inhibition. Epilepsy is a common comorbidity with psychiatric disorders and has consistently been linked to impaired inhibitory function^30–33. In patients with autism, it is estimated that the prevalence of epilepsy comorbidity is around 25%^34,35. However, this is dependent on the type of autism, and the prevalence can be as high as 80% in Rett syndrome³⁶, a monogenic form of autism caused by mutation in the MeCP2 gene³⁷. It is currently unclear whether schizophrenia is a risk factor for epilepsy. A limited number of studies have been dedicated to this question, and contradicting results have been reported^38,39. However, patients with epilepsy show an increased risk of schizophrenia or schizophrenia-like psychosis⁴⁰. Likewise, patients with epilepsy show an increased risk for ADHD^41,42.

Another recurrent phenotypic change is the altered power of gamma oscillations, as measured with electroencephalography or magnetoencephalography in humans, indicating changes in neuronal synchrony⁴³. Gamma oscillations are important for integration of information in neuronal circuits and have been linked to various functions, including attention⁴⁴ and memory⁴⁵. It was shown that PV-positive interneurons⁴⁶, specifically PV-positive basket cells (see below), play an important role in these osciliations^43,47,48. Changes in gamma oscillations are consistently found in patients with schizophrenia⁴⁹, affecting different regions, including the prefrontal cortex^50,51. Interestingly, while a decrease in gamma power is linked to negative symptoms of this disorder, such as psychomotor poverty⁵², increased gamma power has been observed during positive symptoms, such as hallucinations⁵³. In addition, computational studies suggest a central role for inhibitory synaptic scaling in maintaining a stable neuronal network⁵⁴ and found changes in inhibitory transmission to be sufficient to explain the changes in gamma oscillations in schizophrenia⁵⁵. Together, altered inhibitory control is believed to lead to a change in the power of gamma oscillations, which play a central role in schizophrenia⁵⁶.

Though studied mainly in schizophrenia, changes in gamma oscillations have been observed in other psychiatric disorders, including autism, ADHD and bipolar disorder^57–63. For example, children with autism show a reduced gamma frequency modulation to a visual task⁶⁴, whereas in ADHD, increased power and synchrony were observed^59–62. Together, post-mortem and patient studies point to an important role for altered inhibitory function in various psychiatric disorders and indicate a vital role for inhibition in the maintenance of the E/I balance in the healthy brain.

The central role of parvalbumin-positive interneurons in E/I balance

Cortical and hippocampal synaptic inhibition is mediated by inhibitory interneurons, most of which use GABA as their neurotransmitter. While interneurons make up around 20% of the total neuronal population, they are highly diverse^65,66. For example, different classes of interneurons are specialized to target the dendrites, soma or axon initial segment (AIS) of pyramidal neurons⁶⁵. This large variety of cell types is believed to illustrate the distinct functions that these cells have in regulation of the network⁶⁷. Cortical interneurons can be segregated in three non-overlapping groups by means of specific markers: PV, somatostatin (SOM) and the serotonin receptor 3a (5HT3aR), accounting for 40%, 30% and 30% of the total interneuron population, respectively⁶⁸. 5HT3aR-positive cells mainly originate from the caudal ganglionic eminence and are further divided as vasoactive intestinal peptide (VIP)-positive and VIP-negative interneurons⁶⁸. VIP-positive interneurons mainly inhibit other interneurons and play an important role in disinhibition of the local circuit⁶⁹, where they receive excitatory input from other cortical areas^70,71. VIP cells mainly inhibit SOM cells⁷² but also target PV interneurons²³ and are involved in the regulation of the behavioural state of the network^71,73. Recent studies have suggested a direct inhibition by VIP interneurons of pyramidal cells in cortex^74,75. Despite the prominent, mainly disinhibitory, function of VIP cells in the network, only a limited number of studies have implicated VIP cells to be involved in psychiatric disorders^26,76.

SOM interneurons are a diverse class of interneurons originating from the medial ganglionic eminence (MGE)⁶⁸. These interneurons target non-SOM interneurons⁷² as well as the dendritic domain of pyramidal neurons, including dendritic spines⁷⁷. SOM interneurons regulate the integration of local excitatory input^78,79 and have been shown to regulate synaptic plasticity via the control of dendritic calcium spikes in pyramidal cells, affecting learning tasks⁸⁰. Increasing evidence implicates SOM interneurons in psychiatric disorders. Disinhibition of SOM interneurons leads to an anti-depressive–like phenotype in mice⁸¹, and reduced levels of SOM in cerebral spinal fluid have been linked to major depression and mood disorders⁸². In addition, a recent article shows a role for SOM interneurons in gamma oscillations in the visual cortex⁸³, hinting towards a possible role for SOM interneurons in the changes in gamma oscillations observed in psychiatric disorders.

PV interneurons are MGE-derived and are electrophysiologically identified by their fast-spiking phenotype. Although PV interneurons make up only a small part of the entire neuronal population^84,85, these interneurons are strongly implicated in psychiatric disorders and have been shown to play an important role in the regulations of the E/I balance^8,10,86. PV interneurons are involved in gamma oscillations (see above), and various mutations in disease-linked genes affect PV interneuron function (discussed below) (Table 1). Different subtypes of PV interneurons are distinguished: basket cells, chandelier cells, bistratisfied cells, and, in hippocampus, oriens-alveus-lacunosum-moleculare cells^87,88 form the largest of these groups, the first two of which are most widely studied. Here, we will discuss both types and focus on their respective roles in the network.

Chandelier cells

Chandelier cells, or axo-axonic cells, are a group of interneurons that target the AIS of pyramidal neurons^95,96. These cells form vertically oriented clusters of axon terminals, called cartridges, giving them a chandelier-like appearance. A single pyramidal cell receives contacts from multiple chandelier cells⁹⁷, forming an average of 3 to 5 boutons each⁹⁸ depending on the brain region⁹⁹ and age¹⁰⁰. The synapses are enriched for the GABAAα2 receptor subunit^101,102 and pre-synaptically express the high-affinity GABA transporter 1 (GAT1)¹⁰³. The function of chandelier cells in the network remains largely unknown. However, chandelier cell activity has been shown to increase with increasing network activity¹⁰⁴. In addition, the axon terminals of chandelier cells are specifically absent from the epileptic focus^105,106, and pharmacological induction of seizures leads to a loss of chandelier cells in rats¹⁰⁷, suggesting a role in preventing excessive excitatory activity in the network for these interneurons¹⁰⁴.

Since their discovery, there has been a debate about the actions of chandelier cells (reviewed by Wang and colleagues⁹⁸). Some studies using brain slice recordings showed a depolarizing¹⁰⁸ and even excitatory¹⁰⁹ action for these interneurons, whereas others report an inhibitory action¹¹⁰. However, chandelier cell membrane potential fluctuations resembling in vivo patterns appear strongly inhibitory¹¹¹, and recordings in vivo also suggest an inhibitory role for these interneurons^112,113.

Whereas patient studies of schizophrenia patients consistently identify both a reduction in the number or length of chandelier cell cartridges^114,115 as well as the misregulation of proteins associated with chandelier cell synapses^12,116–118, mouse research on this interneuron class is hampered by the absence of a strategy to specifically target these interneurons. As a result, a limited number of studies focus on chandelier cells but instead report on PV interneurons in general, or PV basket cells, which provide a more accessible target of study because of their relative abundance (Figure 1). Nonetheless, chandelier cells are considered to play an important role in psychiatric disorders^11,98,119, and the development of strategies to specifically target this interneuron subtype would be an important step towards understanding the role of these interneurons.

Figure 1. Genes linked to psychiatric disorders affect inhibition on different subcellular aspects.

(Left) Genes affecting the input, intrinsic properties or output of chandelier cells. (Right) Genes affecting the input, intrinsic properties or output of parvalbumin-positive (PV) basket cells. (Middle) Genes affecting input, intrinsic properties or output of interneurons, without the interneuron subtype being identified. Pyramidal cells are shown in red, and interneurons are shown in blue.

Targeting the perisomatic region (basket cells) or AIS (chandelier cells) gives these interneurons strong control over pyramidal cell excitability, and regulation of the synaptic strength of these interneurons is important for normal function of the network. For example, Xue and colleagues have shown that pyramidal neurons receive an amount of synaptic inhibition that is proportional to the amount of synaptic excitation they receive¹³⁸, maintaining the E/I balance on the pyramidal cell. Manipulation of pyramidal cell activity leads to a compensatory change in inhibitory drive onto these cells, specifically from PV interneurons¹³⁸. In addition, PV interneuron activity is reduced during learning and increased during fear conditioning, and an experimental increase of PV interneuron activity leads to impaired learning²³. Apart from synaptic connections, basket cells¹³⁹ and chandelier cells^111,140 connect via electrical connections, called gap junctions. This electrical coupling synchronizes the interneurons¹³⁹, which in turn allows them to synchronize the network¹⁴¹ (for example, in gamma oscillations¹⁴²).

Together, these studies indicate that control of the E/I balance by PV interneurons is important for normal network function and that PV interneuron-mediated inhibition can be regulated upon alterations in the network state. Changes in PV interneuron-mediated inhibition would shift the E/I balance and lead to a disruption in network function. Indeed, various parameters affecting inhibitory function of PV cells are consistently found to be altered in psychiatric disorders. In the next section, we focus on studies on animal models of these conditions and discuss how different changes affecting PV cell activity lead to a shift of the E/I balance (Table 1).

Altered PV interneuron activity is caused by changes to different subcellular aspects

Alterations to the inhibitory drive, affecting the E/I balance, can arise in different ways. Reduced excitatory input onto interneurons, reduced intrinsic excitability of interneurons, and a reduction in inhibitory synapse number or strength onto pyramidal cells all result in a shift of the E/I balance towards excitation. Indeed, patient and animal studies of psychiatric disorders consistently report changes affecting inhibitory function. A complicating factor in the interpretation of these results comes from the dynamic ability of neuronal networks to adapt to changes, known as homeostatic plasticity. Homeostatic plasticity is the ability of neurons to maintain their levels of excitability within a narrow range and is a constantly active feedback process¹⁴³. For example, classic experiments have shown that blocking of action potentials leads to a strengthening of excitatory¹⁴⁴ and a weakening of inhibitory synapses¹⁴⁵. Apart from their synaptic input, neurons can regulate their intrinsic excitability¹⁴⁶, which is observed both in excitatory and inhibitory neurons in culture¹⁴⁷ and in vivo¹⁴⁸. This means that genetic mutations affecting a specific neuronal property in a specific cell type can trigger homeostatic processes affecting other properties or cell types. In this way, mutations affecting both inhibitory or excitatory cells could ultimately affect inhibitory function⁸⁶. It is therefore difficult to distinguish the direct effect of a gene related to a psychiatric disorder from the network adaptation it causes.

Nonetheless, changes to the function of PV interneurons, either direct or indirect, are consistently observed in psychiatric disorders affecting input, output and intrinsic properties, which all lead to an altered inhibitory action of these cells onto their targets (Figure 1).

Changes to synapses formed by PV interneurons

Changes to PV synapses are abundantly studied and identified in various conditions. Post-mortem studies of schizophrenia patients consistently identify changes to the cartridges formed by chandelier cells, showing a decrease in pre-synaptic GAT1 expression^116,117 and an increase in the expression of post-synaptic GABAAα2¹¹⁸. These changes would lead to an increased inhibitory drive from chandelier cells and are believed to be compensatory for a reduced activity of these cells, as indicated by reduced levels of GAD67 in PV cells¹⁹⁰. In addition, a recent study shows a reduction in the density of a specific type a cartridges, calbindin-positive cartridges, in schizophrenia¹¹⁴.

Besides changes in the input to PV interneurons, mutations in Erbb4 also lead to a reduction in synapses formed by PV interneurons on pyramidal neurons, specifically from chandelier cells^120,128. In addition, overexpression of Nrg1, the ligand for ErbB4, in pyramidal neurons increases bouton density on both the AIS and the soma of pyramidal neurons. The increase in bouton density on the AIS originates from chandelier cells since only these cells target the AIS. The origin of the increase in perisomatic bouton density is not clear since a recent report has shown that synapses formed by cholecystokinin (CCK) basket cells require ErbB4 function to form perisomatic synapses¹⁹¹. Future research should unveil whether the increase in perisomatic boutons density arises from PV or CCK basket cells.

Tuberous sclerosis is a disorder whose symptoms include epilepsy and autism¹⁹². Loss-of-function mutations in the mammalian target of rapamycin (mTOR)-negative regulators TSC1 or TSC2 underlie this condition¹⁹³. Slice recordings from TSC1 KO neurons revealed a reduced inhibitory drive onto CA1 pyramidal neurons, while excitatory input was unaltered¹²⁹. TSC1 KO neurons were created by sparsely targeting neurons in a conditional TSC1 KO mouse with a cre-expressing adeno-associated virus¹²⁹. The finding that sparse KO of TSC1 leads to a similar phenotype as in neurons from full KO animals indicates that these results are cell-autonomous rather than compensatory¹²⁹.

Angelman syndrome, caused by loss-of-function mutations or deletion of E3 ubiquitin ligase UBE3A¹⁹⁴, is characterized by epilepsy and autism^195,196. Mouse models for Angelman syndrome recapitulate human phenotypes¹⁹⁷ and initially were found to show a reduced excitatory drive onto pyramidal neurons¹⁹⁸. However, inhibitory input was also found to be decreased, caused by defects in synaptic vesicle cycling¹³⁰. It was hypothesized that the reduced inhibition could outweigh the reduced excitation, leading to the epilepsy and autism phenotypes obeserved^130,199. Consistent with this idea, a recent study using in vivo whole cell recordings shows that pyramidal neurons in Ube3a KO mice display decreased orientation selectivity²⁰⁰, indicative of reduced inhibition^201,202. However, these mice also show increased excitability of pyramidal neurons, which is non–cell-autonomous, suggesting that pyramidal neurons homeostatically increase their excitability because of a relative decrease of excitation²⁰⁰. While it is unclear whether reduced inhibition or reduced excitation has a stronger impact on pyramidal neurons in Angelman syndrome, the change in their relative contribution, resulting in an altered E/I balance, seems to play a pivotal role in this condition.

Single SHANK3 mutations, at the level of point mutations or microdeletions, have been identified in patients with ASD²⁰³. In the human genome, there are three SHANK genes (SHANK1-3), which all code for scaffold proteins located at the postsynaptic density of excitatory synapses, of which SHANK3 is best studied^204–206. Because of this localisation, most studies have focussed on excitatory synapses, where Shank3-deficient mice show reduced cortico-striatal connectivity²⁰⁷, impaired long-term potentiation²⁰⁸ and reduced GluA1 expression²⁰⁹. Recent studies, however, indicate that the inhibitory system is also affected. Shank3 mutant mice lacking exon 9 show reduced inhibitory input onto layer 2/3 pyramidal neurons but an increase of these events in CA1 pyramidal neurons¹³¹. In addition, PV levels are reduced in Shank1 KO and Shank3b KO mice, indicating reduced activity²².

Presynaptic neurexins and their postsynaptic partners neuroligins are a large class of cell-adhesion molecules that have been shown to play important roles in synaptic specificity²¹⁰. Overexpression or knockdown of neuroligins leads to an increase or decrease in synapse number, respectively²¹¹. Neuroligins are expressed at specific synapses: neuroligin 1 (NLGN1) is mainly expressed at excitatory synapses²¹², NLGN2 is expressed at inhibitory synapses²¹³, NLGN3 is expressed at both inhibitory and excitatory synapses²¹⁴ and NLGN4 is expressed at glycinergic synapses²¹⁵. Mutations and deletions affecting human NLGN3, including a gain-of-function mutation, and NLGN4 de novo mutations have been found in Swedish families and have been associated with autism²¹⁶. Nlgn3 deletion in mice leads to increased inhibitory transmission onto pyramidal neurons²¹⁷, specifically from CCK-positive interneurons because of an increased tonic endocannabinoid signalling, whereas PV interneuron connectivity is unaffected¹³⁶. A recent study, however, has shown that conditional deletion of Ngln3 from PV interneurons alters AMPA/NMDA (alpha-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid/N-methyl-d-aspartate) ratio of excitatory input onto these cells and causes reduced gamma oscillations¹²⁴. In addition to the loss of NLGN3, a gain-of-function amino acid substitution (R451C) in NLGN3 is associated with autism²¹⁶. Mice carrying this mutation show a strong reduction in inhibitory drive from PV interneurons while increasing the inhibitory drive from CCK cells¹³⁶.

Nlgn4 KO mice show a reduced number of perisomatic inhibitory synapses in hippocampus and a concomitant reduction in inhibitory input²¹⁸. In addition, a reduced power of evoked gamma oscillations in acute slices was observed²¹⁸. Different mutations in NLGN2 have been linked to autism²¹⁹ and schizophrenia²²⁰. Deletion of Nlgn2 has been shown to specifically reduce the amount of perisomatic synapses on pyramidal neurons in hippocampus¹³⁵ and reduce inhibitory transmission from PV interneurons but not SOM interneurons¹³⁴.

Neurexins are less well studied in the context of psychiatric disorders. There are three neurexin genes, each coding for an α- and β-neurexin. Mutations in neurexin (NRXN1) have been associated with autism²²¹ and schizophrenia, affecting both NRXN1α and NRXN1β^222,223. Nrxn1α KO mice do not show changes in inhibitory drive but do show a reduced excitatory drive onto CA1 pyramidal neurons²²⁴. However, mice carrying mutations in Nrxn1β show a reduced frequency of both inhibitory and excitatory input onto cortical pyramidal neurons²²⁵ suggestive of a reduced number of synaptic contacts. These studies show that neuroligin and neurexin mutations that are linked to autism affect the inhibitory system, including perisomatic-targeting interneurons.

Studies of ADHD rodent models have mainly focussed on the dopamine system and excitatory synapses²²⁶. However, recent studies identify changes to inhibitory connectivity. G protein–coupled receptor kinase interacting protein 1 (GIT1) has been identified by a GWAS as a risk gene for ADHD¹³², but recent studies challenge this claim^227,228. Git1 KO mice show ADHD-like phenotypes, including hyperactivity¹³². While excitatory input to CA1 pyramidal neurons remains unaltered, the frequency of inhibitory inputs is reduced, suggesting a reduction of inhibitory synaptic contacts¹³², consistent with previous studies showing a role for GIT1 in inhibitory synapses²²⁹. In addition, PV expression was reduced while other interneuron markers remained unaltered¹³².

Another ADHD-linked gene identified by GWASs, cadherin 13 (CDH13)^230–232, is exclusively expressed in inhibitory neurons¹³³. Cdh13 KO mice show an increase in inhibitory synaptic contacts onto CA1 pyramidal neurons, while excitatory inputs remained unaltered¹³³. This increase in inhibitory synaptic contacts could underlie the increase in gamma oscillations observed in ADHD patients discussed before^59–62. Cdh13 KO mice show a reduced number of interneurons at embryonic day 18.5²³³ but not at P21¹³³. Experiments are needed to test whether the changes in interneuron number have a role in the aetiology of ADHD.

In addition to changes in the number and strength of inhibitory synapses, changes are found in the modulation of inhibitory synaptic transmission. Metabotropic GABA receptors, GABA_B receptors, are expressed both pre- and postsynaptically in GABAergic synapses²³⁴. Postsynaptic GABA_B receptors activate potassium channels that hyperpolarize the postsynaptic cell²³⁵. Presynaptic GABA_B receptors, in addition, can reduce calcium inflow and reduce neurotransmitter release^236,237. A reduction of GABA_B subunit expression has been observed in post-mortem studies of patients with schizophrenia^238,239, patients with bipolar disorder²⁴⁰ and patients with major depression²⁴¹ as well as animal models for schizophrenia^242–244. However, more research is required to identify the exact role for GABA_B receptors in psychiatric disorders.

In summary, changes to the inhibitory drive of PV interneurons can be caused by changes affecting the input, output or intrinsic properties of PV interneurons, and animal models of various psychiatric disorders all show alterations to one or more of these aspects, tilting the E/I balance.

Conclusions

Psychiatric disorders are a diverse group of disorders, but changes to the inhibitory system seem to be a point of convergence. Impairment of normal inhibitory function can arise from input to, output from, or intrinsic properties of inhibitory neurons. Altered inhibitory activity or drive leads to changes in signal processing, which in turn is believed to underlie the phenotypic changes observed in the various psychiatric disorders. PV-positive interneurons play a pivotal role in these conditions, possibly through their strong inhibitory effect on pyramidal cell activity due to the axonal or perisomatic targeting of their axons in combination with the nature of their functions in the network.

Changes to either excitation or inhibition will change the ratio between these two types of input, leading to a change in the E/I balance. The examples described above indicate that various psychiatric disorders occur following changes to the input, output or intrinsic properties of specific interneurons, PV interneurons, leading to an altered activity of these neurons. It seems from the studies discussed that specific changes in the E/I balance lead to a disruption of specific function(s) in the network that affect signal processing in a specific way to result in a specific psychiatric phenotype. This might explain why different disorders present different phenotypes, in both patients and animal studies. For example, changes in chandelier cell cartridges seem to be more prominent in schizophrenia. However, other factors are likely to contribute to the development of a specific disorder. Apart from the affected PV cell type, the direction of the altered activity (increased/decreased E/I balance) might be an important factor. Another interesting aspect might be the changes in interneuron-interneuron connectivity, which could lead to altered signal integration and network activity. Some genes have been associated with multiple psychiatric disorders, indicating that a mutation does not in all cases lead to a specific condition. Individual difference in compensatory plasticity could subtly affect network development, steering the developing network towards a specific disorder. It should be noted that while homeostatic changes might compensate a specific alteration, this compensation might disrupt other pathways. While the above-mentioned factors potentially all play a role in the development of specific psychiatric disorders, more research is needed to identify how specific alterations to PV interneurons affect network processing and behaviour.

The notion that psychiatric disorders are caused by changes to the inhibitory drive from PV interneurons means that a restoration of this drive could improve patient symptoms. An interesting possibility of counteracting the changes to the inhibitory system would be to make use of the neuron’s ability for homeostatic plasticity¹⁷⁰. Homeostatic mechanisms function to keep neurons in an optimal range of activity²⁴⁵. Understanding which genetic and molecular mechanisms underlie these homeostatic processes opens the possibility to ‘hijack’ these pathways and manipulate the neuron’s activity in a way to compensate for the altered inhibitory drive. In this way, it might be possible to selectively up- or downregulate the activity levels of specific interneuron populations and thereby ameliorate the symptoms observed in patients with the various disorders.

While psychiatric disorders are still far from being fully understood, the study of the role of the inhibitory system in these conditions might further increase our understanding of both the diseased and the healthy brain and hopefully could lead to treatment or alleviation of the symptoms for those suffering from these conditions.