The cloning of GBR cDNAs (Marshall et al., 1999) made it possible to genetically ablate individual receptor subunits in mice, thereby establishing a direct genetic link between GBR dysfunction and disease. Due to the obligate heterodimeric nature of GBRs, knockout of either the GB1 subunit (comprising the GB1a and GB1b isoforms) or the GB2 subunit results in similar synaptic deficits and pathologies (Gassmann and Bettler, 2012; Gassmann et al., 2004; Schuler et al., 2001), including complete loss of both pre- and postsynaptic GBR responses, spontaneous seizures, increased susceptibility to induced seizures, cognitive impairments, hyperactivity, altered circadian activity, and hyperalgesia (Gassmann et al., 2004; Schuler et al., 2001). The occurrence of seizures in GB1 and GB2 knockout mice supports findings from antagonist studies and highlights the key role of GBRs in maintaining the excitation–inhibition balance in the brain through inhibitory signaling. Mice with a heterozygous deletion of the GB1 or GB2 subunits have not been systematically analyzed; however, available data suggest that heterozygous GB1-deficient mice exhibit only mild functional and behavioral deficits (Kaupmann et al., 2003; Schuler et al., 2001). Selective ablation of the GB1a subunit abolishes presynaptic GBR-mediated inhibition of neurotransmitter release, while deletion of the GB1b subunit disrupts postsynaptic inhibition through GIRK channels (Vigot et al., 2006). Notably, only GB1a-deficient but not GB1b-deficient mice exhibit a proconvulsive phenotype (Vigot et al., 2006), highlighting the critical role of presynaptic GBRs in limiting glutamate release and preventing excessive excitation, hypersynchronous network activity, and seizure generation. Similarly, GB1a-deficient mice show pronounced impairments in learning and memory, likely due to disinhibited glutamate release and subsequent saturation of synaptic plasticity mechanisms (Vigot et al., 2006). In comparison, GB1b-deficient mice display milder phenotypes, including hyperactivity, disrupted circadian cycles, spatial memory deficits, and impaired fear conditioning, a form of associative learning (Gassmann and Bettler, 2012). While the therapeutic benefits of baclofen and PAMs largely align with disease phenotypes observed in GBR-deficient mice, the cognition-enhancing effects of the GBR antagonist SGS742 (Froestl et al., 2004) appear at odds with the pronounced learning and memory deficits reported in GBR-deficient mouse models.

Genetic and genomic technologies provide powerful tools for identifying variants in GABBR1 and GABBR2 that may predispose individuals to disease or directly contribute its pathogenesis. Given the broad expression of GBRs throughout the central nervous system, and the diverse pathologies observed in GBR-deficient mice, genetic variants that impair receptor function are likely to contribute to disease (Gassmann and Bettler, 2012; Pin and Bettler, 2016). Genome-wide association studies (GWAS Catalog, https://www.ebi.ac.uk/gw intellectual disability as/) have identified single nucleotide polymorphisms and other genetic variants in GABBR1 and GABBR2 that are associated with schizophrenia, anxiety and depression/mood disorders, autism spectrum disorder (ASD), post-traumatic stress disorder, alcohol use disorder, insomnia, Alzheimer’s disease, and pain (Table 1). Based on statistical significance and replication across independent cohorts, the strongest genetic associations have been identified for depression and schizophrenia. However, since all GWAS-associated variants in GABBR1 and GABBR2 reside in non-coding regions, their impact on GBR function remains unclear. Non-coding variants are thought to influence disease by modulating gene expression or alternative splicing of transcript isoforms. Their regulatory effects are often modest and cell-type specific, which further complicates the functional validation of disease-associated variants (Gallagher and Chen-Plotkin, 2018; Wainberg et al., 2022).

In contrast to non-coding GWAS variants, missense variants identified through whole-exome sequencing (WES) in affected individuals offer a more direct and potentially causal link to disease. GABBR1 and GABBR2 are classified as haploinsufficient genes, as indicated by their LOF intolerance (pLI) scores of 1 in the gnomAD database (https://gnomad.broadinstitute.org/), indicating strong selective pressure against protein-truncating variants. In contrast, mouse models with heterozygous deletion of Gabbr1 exhibit only mild functional or behavioral deficits (Kaupmann et al., 2003; Schuler et al., 2001), suggesting species-specific differences in dosage sensitivity or compensatory mechanisms. Both genes also exhibit significant constraint against missense variation, with missense Z-scores of 5.54 (GABBR1) and 4.11 (GABBR2) in gnomAD, suggesting that protein-altering mutations are generally not well tolerated and are more likely to be deleterious and potentially disease-causing. ClinVar (https://www.ncbi.nlm.nih.gov/clinvar/), a database documenting human genetic variants and their clinical significance, reports 80 missense variants in GABBR1 and 433 in GABBR2 (Figure 2). Among these, seven monoallelic de novo variants in GABBR1 and fourteen in GABBR2 are classified as pathogenic or likely pathogenic. Additional variants with strong evidence of pathogenicity have been reported in the literature but have not yet been included into ClinVar. These variants are listed in Table 2 (GABBR1) and Table 3 (GABBR2), and have been mapped onto the structural model of GBRs (Figure 3). The missense tolerance ratio (MTR) provides a codon-level measure of selective constraint derived from human population sequencing data (Traynelis et al., 2017). Many, though not all, pathogenic variants in GABBR1 and GABBR2 cluster in regions with low MTR scores, consistent with strong purifying selection against amino acid substitutions in these regions (Figure 2). Due to limited functional validation and incomplete clinical annotation, most missense variants in these genes are currently classified as VUS (Nykamp et al., 2017). Nonetheless, several of these VUS have been identified in individuals with phenotypes consistent with GBR-related disorders (Figure 2). Notably, many of these VUS are located in low-MTR regions, particularly within GABBR2, supporting that they may be pathogenic and warrant further investigation.

Several algorithms have been developed to predict the pathogenicity of single nucleotide variants (MacArthur et al., 2014). These algorithms are used in conjunction with variant frequency data from case cohorts and reference population databases, such as gnomAD, BRAVO, and Regeneron (Karczewski et al., 2020). To further assess the potential pathogenicity of GABBR1 and GABBR2 VUS associated with GBR-related disorders, we used three in silico prediction tools: REVEL (Ioannidis et al., 2016), CADD (Kircher et al., 2014), and AlphaMissense (Cheng et al., 2023). Scores for REVEL, CADD and AlphaMissense were obtained from the dbNSFP database (https://www.dbnsfp.org/) and compared to those of known pathogenic variants (Figure 4). The results show that all three tools reliably classify the majority of known pathogenic variants as deleterious. Notably, many GABBR1 and GABBR2 VUS linked to GBR-related disorders also received high pathogenicity scores, suggesting they may be disease-causing.

Pathogenic variants in GABBR1 are commonly associated with a clinical phenotype that includes neurodevelopmental delay and/or epilepsy (Cediel et al., 2022). Affected individuals typically present with early motor delays, speech and language impairments, ID, learning difficulties, and/or behavioral abnormalities. This phenotypically heterogeneous neurological disorder, caused by monoallelic de novo missense GABBR1, is designated as NEDLC. Pathogenic variants in GABBR2 were initially identified in individuals with DEE59 (Euro et al., 2014) or with clinical features resembling atypical Rett syndrome (Lopes et al., 2016; Vuillaume et al., 2018; Yoo et al., 2017). Individuals with the latter presentation are now more accurately diagnosed with NDPLHS. This condition is characterized by developmental stagnation or regression in early childhood, typically manifesting as loss of purposeful hand movements, impaired or lost language abilities, and ID. Since these initial reports, additional pathogenic de novo mutations in GABBR2 have been discovered in cohorts of individuals with ID (Carneiro et al., 2018; Deciphering Developmental Disorders, 2017), ASD (Al-Sarraj et al., 2021; Takata et al., 2018), and drug-resistant epilepsy (Kim et al., 2021; Rochtus et al., 2020). Notably, the recurrent de novo missense variant GABBR2 p.Ala567Thr, which affects a highly conserved residue within the third transmembrane helix (TM3), has been identified in more than 10 unrelated individuals presenting with NDPLHS.

While computational predictions are valuable for assessing the potential pathogenicity of variants, functional studies are essential to determine their impact on protein function, including whether they cause GOF or LOF effects and to what extent these alterations influence receptor activity. Furthermore, functional studies help elucidate molecular disease mechanisms—an essential step toward accurate diagnosis and the development of targeted therapies. Cell-based assay systems that enable direct and selective measurement of GBR activity have proven to be both cost-effective and highly informative for functionally characterizing missense variants in GABBR1 and GABBR2 (Bielopolski et al., 2023; Cediel et al., 2022; Vuillaume et al., 2018). These analyses have revealed a number of functional alterations in pathogenic variants, which may also occur in combination: (i) reduced or absent surface expression, leading to decreased or abolished GABA efficacy at the receptor; (ii) a significant reduction in the potency of GABA at the receptor; and (iii) increased constitutive activity. A reduction in surface expression is observed for several pathogenic variants in GABBR1 and GABBR2, which can be located either in the extracellular VFTD or within the TMD. Notably, GABBR1 p.Gly673Asp in TM3 and GABBR2 p.Gln430Pro in the VFTD fail to reach the cell surface, rendering the receptors completely inactive. A decrease in GABA potency is observed at GABBR1 p.Glu368Asp and p.Ser321Leu, both situated in the VFTD near the orthosteric binding site. These variants likely decrease the receptor’s affinity for GABA. Increased constitutive activity is predominantly associated with variants located in the TMDs of GB1 and GB2. Specifically, GABBR1 p.Ile809Ser and p.Ile847Val, along with GABBR2 p.Ala567Thr, p.Ser695Ile, p.Met702Val, p.Ile705Asn, and p.Ala707Thr, all exhibit enhanced constitutive activity to varying degrees. Constitutive activity in these variants is reversed by the competitive GBR antagonist CGP54626, except for p.Ser695Ile, which is fully active in the absence of GABA (Vuillaume et al., 2018). Structural mapping of these variants onto available GBR models reveals their localization along the TMDs (Figure 3). Structural data suggest that amino acid substitutions within the TMDs can stabilize the active state of GB2, thereby enabling G protein activation even in the absence of GABA binding (Liu et al., 2021; Shaye et al., 2021). Increased constitutive activity is also observed with variants in the VFTDs. Molecular dynamics simulations suggest that GABBR1 p.Gly531Ser, located in a hinge region outside the orthosteric binding site, and GABBR2 p.Arg212Gln, situated in the VFTD, both induce local conformational changes that stabilize the active state of the receptor. This aligns with the allosteric activation mechanism of GBRs, in which agonist binding to the VFTD of GB1 induces conformational changes transmitted to the TMD of GB2, ultimately activating the G protein (Shaye et al., 2021). Notably, due to their elevated baseline activity in the absence of GABA, all constitutively active variants exhibit a corresponding reduction in GABA efficacy. The GABBR1 variant p.Gly110Ala, situated in SD2, has not yet been functionally characterized, but may selectively impair the function of presynaptic GBRs. Overall, functional studies have revealed both LOF and GOF variants in GABBR1 and GABBR2.

The in vivo effects of constitutively active variants are likely to be complex and context-dependent. Under conditions of low ambient GABA, constitutive activity and increased GABA potency may enhance GBR signaling, producing a GOF effect. In contrast, during periods of elevated synaptic GABA concentrations, reduced GABA efficacy could lead to a net LOF. While inverse agonists can suppress constitutive receptor activity, they risk further dampening GABA-mediated signaling during synaptic transmission, potentially exacerbating functional deficits. Functional studies in transfected neurons suggest that certain constitutively active GABBR2 variants disrupt receptor trafficking to the neuronal surface, resulting in reduced signaling efficacy and contributing to presynaptic hyperexcitability (Minere et al., 2024). Notably, this synaptic phenotype was reversed by pharmacological enhancement of GBR signaling using a PAM. Moreover, variants such as GABBR2 p.Ser695Ile, which exhibit high constitutive activity, may trigger adaptive cellular mechanisms that ultimately downregulate receptor function. These observations underscore the challenge of selecting an optimal therapeutic strategy based solely on in vitro data, emphasizing the need for a deeper understanding of variant-specific effects in a physiological context. The advent of CRISPR/Cas genome editing has made it relatively rapid and cost-effective to generate mouse models carrying specific variants inserted into the endogenous gene locus. Such models closely replicate the human condition by maintaining physiological expression levels within the native neuronal environment—a critical factor when studying monoallelic variants. These models offer a powerful platform for detailed investigations of synaptic and network function through both in vitro and in vivo electrophysiology. In parallel, they enable comprehensive biochemical profiling of the receptor and its signaling partners, facilitating the identification of adaptive or compensatory mechanisms that may emerge in response to altered receptor function.

Interestingly, both LOF and GOF variants can give rise to overlapping clinical phenotypes (Table 4). As noted above, GOF effects driven by constitutive receptor activity are accompanied by a reduced responsiveness to synaptic GABA, effectively resulting in a concomitant LOF. In addition, both types of variants may disrupt homeostatic mechanisms critical for maintaining neural network stability and the balance between excitation and inhibition (Vertkin et al., 2015). Such disruption likely contributes to the etiology of neurological and psychiatric disorders, including epilepsy, ID, and ASD (Issa et al., 2023; Wondolowski and Dickman, 2013).

Proteomic analyses of brain tissue have identified AJAP1 as a primary interaction partner of GBRs (Schwenk et al., 2016). AJAP1 is a single-pass transmembrane protein broadly expressed in neurons, with its extracellular domain binding to SD1 of GB1a (Figure 1a) (Dinamarca et al., 2019). Through this interaction, AJAP1 trans-synaptically recruits GB1a-containing GBRs to presynaptic sites, thereby influencing their synaptic localization and function (Figure 1b) (Fruh et al., 2024).

Genetic variants in genes encoding GBR-associated proteins, such as AJAP1, may contribute to diseases resulting from GBR dysfunction. GWAS studies have implicated non-coding AJAP1 variants in insomnia and dementia (Table 1). WES and chromosomal microarray analysis have identified individuals carrying either the AJAP1 missense variant p.Trp183Cys, the frameshift variant p.I271Ffs*24, the splice-site variant c.917 + 1G>C, or a complete deletion (Table 5) (Fruh et al., 2024). These individuals predominantly present with global developmental delay, ID, hypotonia, and/or epileptic seizures. These clinical features closely resemble those reported in individuals with LOF variants in GABBR1 (Table 2) (Cediel et al., 2022), indicating that impaired GBR function may contribute to the underlying pathogenesis. The de novo AJAP1 variant p.Trp183Cys replaces a critical tryptophan at position 183 that is essential for SD1 binding. The de novo p.I271Ffs*24 frameshift variant may trigger nonsense-mediated mRNA decay. However, any transcript escaping decay is expected to produce a truncated protein lacking the transmembrane and intracellular domains, while retaining the SD1 binding site. A paternally inherited complete AJAP1 deletion results in LOF. The splice-site variant c.917 + 1G>C is predicted to disrupt normal splicing, potentially leading to nonsense-mediated decay, exon skipping, activation of a cryptic splice site, or intron retention.

To strengthen a causal link between the p.Trp183Cys variant and GBR dysfunction, mice carrying the orthologous Ajap1 p.Trp183Cys variant were generated. Heterozygous Ajap1 ^Trp183Cys/+ mice mimic the monoallelic p.Trp183Cys genotype observed in patients, enabling the investigation of GBR dysfunctions in the brain. Ultrastructural analysis revealed a significant reduction in presynaptic GBR levels in Ajap1 ^Trp183Cys/+ mice, demonstrating that replacement of tryptophan 183 impairs AJAP1’s ability to recruit GBRs to synaptic terminals (Figure 5). As a consequence, Ajap1 ^Trp183Cys/+ mice exhibited reduced GBR-mediated presynaptic inhibition at both excitatory and inhibitory synapses, along with impaired synaptic plasticity. Similar synaptic deficits were observed in Ajap1 ^−/+ mice, which model the heterozygous deletion of AJAP1 seen in patients. Both Ajap1 ^Trp183Cys/+ and Ajap1 ^−/+ mice thus phenocopy the synaptic impairments reported in GB1a ^−/− mice, which lack presynaptic GBRs (Vigot et al., 2006). Individuals with heterozygous LOF alleles in AJAP1 therefore represent the first clinical cases of presynaptic GBR dysfunction.

PIANP is a single-pass transmembrane protein with sequence homology to AJAP1 (Dinamarca et al., 2019). Like AJAP1, it is a primary interaction partner of GBRs and binds to the SD1 domain of the GB1a subunit, albeit with a tenfold higher binding affinity (Dinamarca et al., 2019). The SD1-binding sites in PIANP and AJAP1 share only weak sequence homology, but PIANP also contains the conserved tryptophan residue essential for binding. Unlike AJAP1, PIANP lacks dendritic sorting motifs in its C-terminal intracellular domain and is expressed in both axons and dendrites. PIANP could therefore interact with GB1a in cis within axons and in trans across synapses in dendrites (Figure 1b). Interestingly, however, PIANP cannot compensate for the loss of AJAP1 at mossy fiber synapses (Fruh et al., 2024).

A case study described a boy with a homozygous nonsense variant in PIANP, who presented with global developmental delay (Winkler et al., 2020; Anazi et al., 2017) (Table 5). An additional individual with a heterozygous variant in PIANP was reported to exhibit musculoskeletal and nervous system abnormalities (Table 5). The synaptic effects of PIANP loss were investigated in Pianp knockout mice, which model the homozygous human condition. Electrophysiological recordings revealed a loss of presynaptic GBR-mediated inhibition at hippocampal synapses, suggesting that PIANP is required for stabilizing presynaptic GBRs, either in cis or through a trans-synaptic mechanism, similar to AJAP1 (Winkler et al., 2020). Behavioral phenotyping in mice demonstrated that Pianp deficiency leads to context-dependent increases in anxiety, spatial learning deficits, an altered stress response, severely impaired social interactions, and enhanced repetitive behaviors—all characteristic features of an autism spectrum disorder-like phenotype.

KCTD8, KCTD12, and KCTD16 interact with most GBRs in the brain and are considered auxiliary subunits of the receptor (Schwenk et al., 2010). They bind to both the receptor and the G protein, regulate the kinetics of the receptor response (Fritzius et al., 2024; Turecek et al., 2014), and also function as scaffolding proteins for effector channels such as VGCCs and HCN channels (Perez-Garci et al., 2025; Trovo et al., 2024). These roles suggest that dysfunctional KCTD proteins could contribute to pathologies associated with GBR dysfunction. GWAS have linked KCTD proteins to a range of neuropsychiatric and neurological conditions, including ASD, bipolar disorder, major depression, alcohol and opioid use disorders, insomnia, dementia, and brain development (Table 1). However, to date, no missense variants in KCTD genes have provided causal links to GBR-related pathologies.