The GRIN gene family encodes the following classes of NMDA receptor (NMDAR) subunits: the glycine-binding GluN1, which is the product of GRIN1, glutamate-binding GluN2, which has 4 paralogs (A-D) encoded by GRIN2A, GRIN2B, GRIN2C, and GRIN2D, respectively, and the glycine-binding GluN3, encoded by GRIN3A and GRIN3B ^[1][45]. GluN2A and GluN2B subunits are highly expressed in the cerebral cortex and hippocampus, and to date they have been the most well-studied ^[2][72]. There are significant physiological differences between NMDAR containing GluN2A and GluN2B, such as different opening probability, which is higher for GluN2A-NMDAR ^[3][73]. Furthermore, since both glutamate and glycine are less potent at GluN2A compared to GluN2B, the deactivation time course following the removal of glutamate is shorter for GluN2A-NMDAR ^[4][5][74,75]. Physiologically, GRIN2 gene expression changes throughout the developmental stages. The GluN2B subunit is strongly expressed during the prenatal phase, and then its expression drops during postnatal stages. When it comes to GluN2A, its expression appears to be low during the prenatal period and rises after delivery ^[6][7][76,77]. Presumably, the main mechanism responsible for the changes seen in protein and gene expression of NMDAR subunits is DNA hypermethylation ^[8][78]. One of the identified factors inducing dysregulation in the GRIN2 gene expression is exposure to methamphetamine. In rodents, repeated administration of methamphetamine led to a decrease in GRIN2A expression in the hippocampus and decreased GRIN2B expression in the striatum ^[9][79]. Rats prenatally exposed to methamphetamine are considered a suitable animal model for ADHD, presenting symptoms like hyperactivity or memory malfunction ^[10][11][80,81]. Another factor influencing the GRIN gene expression is prenatal nicotine exposure (PNE) ^[12][82]. PNE is a well-studied ADHD risk factor, supported both by animal research ^[13][14][83,84] and by cohort studies ^[15][16][85,86]. All things considered, there may be a link between ADHD and altered GRIN gene expression, but further research is needed to confirm the association.

Numerous variants and mutations of the GRIN genes have been found in patients with diverse neuropsychiatric disorders, including ADHD ^[17][18][87,88]. However, ADHD is not the most frequent condition among patients harbouring GRIN mutations. Intellectual disability, epilepsy, and autism spectrum disorder are all much more common in GRIN-mutant patients, but certain GRIN mutations are still strongly linked to ADHD ^[19][89]. One example of this is GRIN2A gene variants that were the first NMDAR-related genes associated with an increased risk of ADHD ^[20][21][90,91]. Research on single nucleotide polymorphisms (SNPs) in 205 families revealed a connection between specific GRIN2B variants and ADHD ^[22][92]. Mutations were found in all domains (ATD, ABD, TM, and CTD), but most frequently in the ABD and TM regions ^[23][46]. Mice with experimentally introduced GRIN2A(N615S) mutation showed hyperactivity and dysregulated attentional levels. Since the asparagine amino acid residue GluN2A(N615) controls the magnesium block, it is suggested that symptoms caused by the GRIN2A(N615S) mutation may result from magnesium block suppression and enhanced calcium permeability ^[24][93]. In the case study of 5 children with de novo GRIN2B mutations, behavioural tests showed prominent hyperactivity, impulsivity, distractibility, and a short attention span. Patients with ADHD-resembling phenotype carried the following GRIN2B mutations: t(9;12)(p23;p13), t(10;12)(p21;p13), (p.R682C) in ABD, (p.A636P) in M3 domain and (p.T268SfsX15) in ATD ^[25][26][94,95]. Other identified polymorphisms in patients with increased inattention measured in the Continuous Performance Test (CPT) are rs2229193 in GRIN2A and rs2284411 in GRIN2B ^[27][96]. rs2284411 could be pharmacologically relevant, since children with that polymorphism showed significantly better treatment responses to methylphenidate ^[28][97].

SorCS2 (sortilin-related VPS10 domain-containing receptor 2; chromosome 4) is a large gene composed of 30 exons, belonging to the VSP10p (Vacuolar Protein Sorting 10 protein)-domain receptors gene family, which encodes SorCS2 protein. SNPs in this gene have been associated with a multitude of neuropsychiatric disorders including bipolar disorder ^[29][98], schizophrenia ^[30][99], and ADHD ^[31][100]. Interestingly, in a Genome-Wide Association Study (GWAS) conducted on adult ADHD patients, SNP in the 1 intron of SorCS2 gene rs4689642 has been recognised as the most relevant ADHD-associated polymorphism ^[31][100]. Furthermore, the study conducted on monozygotic twins discordant for ADHD has shown that SorCS2 gene methylation (thus silencing) leads to reduced grey-matter volume in precentral and posterior orbital gyri, which induces symptoms of ADHD in affected children ^[32][101]. Therefore, better understanding of the role of SorCS2 at the molecular level can provide a novel insight into not well-known ADHD aetiology.

VSP10p act as sorting receptors and regulators of neuronal viability and function by controlling the intracellular trafficking of targeted proteins ^[33][102]. Both constituents of VSP10p, Sortilin and SorCS2, can form a complex with neurotrophin receptor p75NTR. The complex is required to control the release and function of pro-neurotrophins, such as pro-BDNF (pro-Brain-Derived Neurotrophic Factor) and pro-NGF (pro-Nerve Growth Factor). The mentioned pro-neurotrophins are precursors for the respective neurotrophins, BDNF and NGF, essential for the promotion of neuronal survival, death and synaptic plasticity ^[33][34][102,103]. The deficiency of SorCS2 disrupts the formation of the SorCS2-p75NTR complex, leading to a decrease in the release of BDNF. Furthermore, the absence of SorCS2 impairs the neuron’s capacity to respond to BDNF through the binding of its receptor—TrkB. Correspondingly, SorCS2-deficient mice presented impaired LTP, LTD, neurite outgrowth, and dendritic spines formation. Lack of SorCS2 abolished NMDAR-dependent synaptic plasticity in the mouse hippocampus, resulting in a phenotype similar to ADHD. SorCS2-deficient mice displayed impairment of long-term memory formation and higher tendency to take a risk and stimulus-seeking behaviour, however, this was accompanied by higher stress vulnerability ^[35][69]. Furthermore, studies have shown that SorCS2 acts as a selective regulator of NMDAR trafficking towards the surface of hippocampal neurons, as well as regulating dendritic spines density (synaptic plasticity) in pyramidal neurons of CA2 region. In the same study, SorCS2-deficient mice exhibited significant social memory deficits, however, without abnormalities in other hippocampal-dependent behaviours ^[36][104]. In addition, the SorCS2-p75NTR complex is also considered an essential regulator of development of dopaminergic projections. The lack of any complex subunits resulted in reduced dopamine levels and metabolism, as well as dopaminergic hyperinnervation of the frontal cortex. Interestingly, as the combined effects of dopaminergic dysregulation are associated with abnormal response to psychostimulants, administration of amphetamine on double knockout models displayed a paradoxical calming response ^[37][105]. Furthermore, SorCS2-deficient mice exhibited an altered dopaminergic firing pattern within the ventral tegmental area (VTA). The dopaminergic transmission was shifted from an irregular to a more regular pattern, along with an associated change in dopaminergic receptor sensitivity (namely, decrease in D1 and increase in D2 sensitivity). Behaviourally, mice presented a general reward deficit, novelty-induced hyperactivity, and yet paradoxical tranquillity in response to amphetamine—a phenotype reminiscent of ADHD ^[38][106].

There are five subtypes of dopamine receptors: D1, D2, D3, D4, and D5. The D1 and D5 dopamine receptors belong to the D1-like family, whereas the D2, D3, and D4 receptors belong to the D2-like family ^[39][107]. The D4 receptor is responsible for signalling in the mesolimbic pathway, which takes part in several cognitive processes, such as motivation, desire for rewards, reinforcement learning, and emotional regulation. D4 is encoded by the DRD4 gene located on chromosome 11 at 11p15.5 ^[40][108]. DRD4 exhibits numerous polymorphisms in its coding sequence, and the most common polymorphism occurs in a region encoding the third intracellular loop of the receptor. This polymorphism results in variable number of tandem repeats of a 48-base pair sequence in the third exon ^[41][42][109,110]. The most common DRD4 polymorphism products are D4.2, D4.4 and D4.7, characterized by 2, 4, and 7 repeats of a proline-rich sequence of 16 amino acids, respectively ^[41][109]. D4.7 has been associated with various psychiatric disorders such as ADHD, substance addiction, and personality traits associated with impulsivity ^[43][111]. In vitro studies have implied that the sensitivity of the D4.7 for dopamine is half that of the D4.2 and D4.4 ^[44][112], and this allele was found in 41% of ADHD patients and only 21% of the control group ^[43][111]. ADHD child patients with 7 repeated alleles exhibit more imprecise and impulsive responses on neuropsychological tasks ^[44][112]. Furthermore, mice that expressed the D4.7 variant showed enhanced exploratory and novelty-seeking behaviours, similar to the phenotypic trait of human ADHD. Mechanistically, D4R binds to the SH3 domain of postsynaptic scaffolding protein PSD-95, which is connected to the C-terminus of NMDA receptor subunits (GluN1) by the PDZ domain ^[45][113]. Activation of different variants of D4R, like D4.7, and D4.4 decrease the NMDAR function in the PFC at varying degrees. The activity of interconnected neurons in PFC, which are dependent on NMDAR and responsible for synchronised network activity, is more strongly inhibited by D4.7 compared to D4.4 ^[46][114]. Activation of D4.7 induces greater suppression of both GluN1/PSD-95 binding and NMDAR surface expression in neurons in comparison to D4.4 activation ^[45][113]. Presumably, inhibition of GluN1/PSD-95 binding causes NMDAR hypofunction, which leads to impairment of synchronous network activity and suppressed PFC activity, characteristic for ADHD patients ^[46][47][114,115]. Thus, D4.7R variant might be an attractive target in development of future therapies, as several studies have found a significant association between the various DRD4 polymorphisms and better response to methylphenidate as compared to placebo ^[48][49][116,117]. However, studies on the impact of the D4.7 variant on the response to methylphenidate present conflicting findings. Some studies show no significant association ^[50][118], while others report a reduction in the response ^[51][119]. Factors such as sample size, population differences, and the presence of other genetic and environmental influences could have contributed to these discrepancies. Additionally, administration of D-cycloserine (partial NMDAR agonist) mitigated high novelty-seeking behaviour in D4.7-expressing mice, which emphasizes a link between NMDAR modulation and ADHD pharmacogenetics ^[45][113].

BAIAP2 (also known as IRSp53) is an abundantly expressed, postsynaptic adaptor protein. It is implicated in the regulation of actin filaments assembly during dendritic spines development, as well as the regulation of NMDAR-mediated synaptic transmission and LTP ^[52][120]. BAIAP2 is encoded by the BAIAP2 gene, located on the chromosome 17. SNPs within the mentioned gene have been suggested to be involved in ADHD aetiology, as well as abnormal cerebral lateralisation, which is also associated with ADHD-related symptoms ^[53][54][121,122]. The hippocampal neurons of BAIAP2-KO mice showed a selective increase in NMDAR activity, however, without significant changes in AMPAR-mediated transmission. This was followed by a substantial increase in LTP and, functionally, deficits in learning, memory, and social interactions ^[52][55][120,123]. Interestingly, both direct and indirect inhibition of NMDAR normalised social interactions in BAIAP2-deficient mice ^[55][123]. Furthermore, even a moderate reduction in BAIAP2 level led to a significant increase in hippocampal NMDAR density ^[56][124]. Consistently, re-expression of BAIAP2 in BAIAP2-mutant mice resulted in the restoration of NMDAR-mediated synaptic transmission and proper NMDAR/AMPAR ratio in the medial PFC region (mPFC). However, despite improvement in social interactions, the hyperactivity- and anxiety-like behaviour induced by BAIAP2-knockout were not rescued in such conditions ^[57][125]. Additionally, the deletion of BAIAP2 suppressed neuronal firing variability and dynamics within excitatory mPFC neurons, especially in those encoding social information. Administration of NMDAR antagonist (memantine) restored burst firing in mPFC neurons and rescued social deficits ^[58][126]. Altogether, the mentioned data emphasize the key role of BAIAP2 in regulating NMDAR-dependent signal transduction and, as a result, proper neuropsychological function.

Synaptosomal-associated protein, 25 kDa (SNAP-25) is a part of the SNARE complex, which is involved in the exocytotic release of neurotransmitters during synaptic transmission. Furthermore, SNAP-25 plays an important role in modifying NMDAR and kainate receptor density in the postsynaptic membrane ^[59][127]. Protein kinase C (PKC)-mediated phosphorylation of SNAP-25 facilitates the transport of postsynaptic vesicles and their subsequent fusion with the plasma membrane, resulting in the insertion of NMDA channels onto the cell surface ^[60][128]. Since SNAP-25 affects NMDAR density in the postsynaptic membrane, it is not surprising that downregulation of SNAP-25 impairs LTP ^[61][129], which as a result affects synaptic plasticity and memory function. Although the actual mechanism by which SNAP-25 affects psychiatric disorders are not well known, numerous studies have shown a connection between alterations in SNAP-25 levels and symptoms of multiple mental disorders, including ADHD ^[62][130]. Symptoms such as spontaneous hyperactivity are seen in animals with SNAP-25 deletion, known as coloboma mice ^[63][64][131,132]. Mice with a knockout of one of the SNAP-25 genes (complete knockout is lethal) present mild hyperactivity ^[65][133]. The hyperactive phenotype observed in mentioned animals has driven the search for SNAP-25 mutations associated with ADHD among humans. Barr et al. genotyped DNA from 122 patients diagnosed with ADHD and found two significant mutations, the MnlI polymorphism and DdeI polymorphism ^[66][134]. Both haplotypes showed biased paternal transmission to affected probands ^[67][68][135,136]. More recently, some studies have suggested an association between microsatellite repeats within the SNAP-25 and ADHD prevalence ^[69][70][137,138]. In one meta-analysis, four SNAP-25 gene variants were confirmed as ADHD risk genes. These included: rs362987 on intron 4, rs363006 on intron 6, and aforementioned MnlI (3′UTR rs3746544) and DdeI (3′UTR rs1051312) ^[71][139]. A subsequent study found another polymorphism associated with ADHD, the rs362549 ^[72][140]. SNAP-25 polymorphisms influence ADHD severity ^[73][74][141,142]. The impact of the MnlI on symptom intensity in ADHD is the most well-known. Children with the MnlI gene showed significantly decreased local functional connectivity density (lFCD) in the ACC, as well as decreased lFCD in the dorsal lateral PFC ^[75][143]. Another study has found a correlation between altered working memory and carrying the MnlI gene ^[76][144].

SNAP-25 gene variations might be an important predictor for methylphenidate response. The strongest association with pharmacotherapy was seen in children with MnlI polymorphism ^[77][78][145,146]. The mechanisms underlying different responses to treatment among patients with MnlI variants remain unclear, However, some studies suggest that changes in brain metabolite levels and in haemodynamics might play a key role in this phenomenon ^[79][80][81][147,148,149].

Latrophilin-3 protein (LPHN3p) is a brain-specific member of a small subfamily of adhesion G protein-coupled receptors. It is encoded by the ADGRL3 gene (also known as LPHN3 gene), located on chromosome 4. Functional studies have demonstrated that LPHN3 variants were expressed mainly within brain regions associated with attention and activity (such as PFC, caudate, hippocampus, amygdala, and cerebellum) and were implicated in both ADHD development and pharmacogenetics ^[82][83][150,151]. Indeed, a multitude of studies has confirmed a crucial role of LPHN3 SNPs in susceptibility to ADHD, as well as its predictive role in ADHD severity, associated comorbidities, and drug responsiveness ^{[84][85][86][87]}[152,153,154,155]. However, the direct molecular mechanism underlying LPHN3p contribution to ADHD development have not yet been fully elucidated. LPHN3p, when combined with its endogenous ligands, acts as a crucial regulator of proper excitatory pyramidal neurons functioning, both in the neocortex and hippocampus. Mechanistically, LPHN3p regulates cortical and hippocampal glutamatergic synaptic formation and density ^[88][89][90][156,157,158]. Deficiency of one endogenous ligand of LPHN3p, namely Leucine-rich repeats transmembrane protein (FLRT3), resulted in significantly reduced NMDAR-mediated excitatory postsynaptic currents (EPSCs) within the hippocampus ^[63][131]. Furthermore, LPHN3-knockout mice demonstrated impairment of early LTP in the CA1 region of the hippocampus with concomitant reduction in NMDAR-GluN1 expression. This resulted in hyperactivity and hippocampal-mediated learning and memory deficits, characteristic of ADHD phenotype ^[91][92][159,160].

PCDH7 is a gene that belongs to the protocadherin gene family and encodes an extracellular protein Protocadherin 7 (PCDH7p). PCDH7p is an integral protein of plasma membrane, which plays role in cell–cell recognition and adhesion. Few variants in PCDH7 have been identified as rare, but at the same time significant risk loci for ADHD development ^[93][94][161,162]. Interestingly, PCDH7p has been found to interact with the N-terminal domain of the GluN1 subunit of NMDAR. Consequently, PCDH7p overexpression resulted in a reduction in synaptic NMDAR current and impairment of dendritic spines morphology within the hippocampus, observed as collapse of spines and abnormal dendritic swelling ^[69][137]. On the other hand, knockout of PCDH7 resulted in elongation of dendritic protrusions beyond typical spine size ^[95][163]. Altogether, PCDH7 constitutes another potential factor affecting synaptic NMDAR function and ADHD risk; however, further research is required to validate this hypothesis.

Recent advances in high-throughput technologies have enabled mapping and holistic analysis of numerous genetic variants, leading to a more comprehensive understanding of molecular changes in normal development and disease. The utility of integrative approaches is particularly important for diseases such as ADHD, where genetic and environmental factors interact with each other ^[96][164]. Multi-omics analysis examines interconnections across genomics, epigenomics, transcriptomics, and metabolomics, aiming to elucidate the biological mechanisms behind ADHD and identify potential biomarkers. This approach allows reusesarchers to determine how environmental factors (e.g., parental smoking, glucocorticoid exposure) influence the child’s genome, ultimately contributing to the manifestation of genetically correlated traits of ADHD phenotype (e.g., childhood aggression, insomnia, tendency to addiction) ^[97][98][165,166]. Furthermore, this approach can unveil new candidate genes implicated in the pathogenesis of ADHD. A multi-omics study by Cabana-Dominguez et al. has identified seven modules of co-expressed genes associated with ADHD. These modules consist of genes that are pivotal for the genetic and epigenetic control of neurodevelopment and immune response ^[99][167]. Among them a number of genes (e.g., SP3, CUX1) are involved in neuronal differentiation, synaptogenesis and synaptic plasticity. IQSEC1 is essential for the maintenance of glutamatergic synapses, while CNTNAP2 for neurocognitive development and neuron-glia interactions ^[99][100][167,168]. While certain research suggests the involvement of these genes in regulating NMDAR expression and glutamatergic signalling ^{[101][102][103][104][105]}[169,170,171,172,173], there is a scarcity of studies focused on investigating these interactions comprehensively. Nevertheless, combination of multi-omics and mechanistic approaches might constitute a promising direction for future research.