Synapses are specialized and asymmetrical structures formed through the initial formation of dendritic or axonal filopodia to bring nascent presynaptic sites closer to targeted postsynaptic specializations. In the neuron, the actin-capping protein (CP) works in concert with actin-related protein 2/3 (Arp2/3) to prevent further elongation of filopodia structures (Akin and Mullins, 2008; Fan et al., 2011), likely promoting the formation of more stable, putative synaptic structures. Subsequently, the anti-capping protein-enabled/vasodilator-stimulated phosphoprotein (Ena/VASP) regulates presynaptic actin polymerization, which affects synaptic protein anchoring and the eventual bouton size (Lin et al., 2010). The process is followed by the recruitment of cell adhesion molecules (CAMs) to stabilize axo-dendritic contact. The importance of CAMs in synapse formation, specialization, and function is well understood and has been extensively examined in other reviews (Dalva et al., 2007; Kilinc, 2018; Südhof, 2018). Previous reviews on the CAMs’ ASD-associated mutations and their link to autism development have provided many insights into their roles in ASD pathology (Ye et al., 2010; Redies et al., 2012; Gandawijaya et al., 2020; Tromp et al., 2021).

The DS Cell Adhesion Molecule (DSCAM), also known as chromodomain helicase DNA binding protein 2 (CHD2), is a member of the neuronal immunoglobulin superfamily of molecules broadly expressed in the nervous system and implicated in the process of pathfinding (Liu et al., 2009), axon branching (Wang et al., 2002), and dendritic arborization (Maynard and Stein, 2012). The gene coding for DSCAM has been mapped to chromosome 21q22.12 → q22.3, commonly duplicated in Down syndrome patients (Yamakawa, 1998), and its upregulation has been linked to altered neuronal circuits and learning and behavioral abnormalities in early studies on the animal models of Down syndrome (Sago et al., 1998). In addition to its link with Down syndrome, DSCAM is a strong ASD risk gene recurringly identified in various genome association studies (Iossifov et al., 2014; Sanders et al., 2015; Wang et al., 2016). Mutations that result in the premature termination of DSCAM expression have also been identified within the human autistic population, supporting the importance of DSCAM function in general neurological function.

DSCAM exhibits prominent synaptogenic capabilities. In Drosophila and Aplysia, DSCAM is involved in the regulation of synapse targeting (Millard et al., 2010) and the activity-dependent clustering of α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptors (Li et al., 2009), leading to an alteration in presynaptic size (Kim et al., 2013). This observation is likely due to the preferential interaction with, and aggregation of, postsynaptic proteins associated with nanoclusters of protein determinants of synapse maturity and functionality, similar to effects observed with neurexin ligands (Nozawa et al., 2022). Furthermore, DSCAM can negatively regulate synaptic maturation through competitive interaction with neuroligin 1 (NLGN1) and the inhibition of NLGN1-NRXN1β-mediated synaptic specification (Chen et al., 2022) in the developing mouse cortex. Using an induced pluripotent stem cell (iPSC)-derived model of telencephalic neurons, an early truncated variant of DCAM (amino acid 684) resulted in a reduced DSCAM/N-methyl-D-aspartate receptor subunit 1 (NR1) colocalization. As a result, the synaptic N-methyl-D-aspartate (NMDA) receptor response was compromised (Lim et al., 2021), suggesting DSCAM’s role in synaptic maintenance.

CASK is an alternative presynaptic scaffold protein encoded by chromosome Xp11.4. Due to its localization on the X chromosome, pathogenic variants are disproportionately identified in the female-dominant X-linked microcephaly with pontine and cerebellar hypoplasia (OMIM 300749) and FG syndrome 4 (OMIM 300422) (Piluso et al., 2009; Moog et al., 2015), with a comparatively rare occurrence in general autistic patients (Gupta et al., 2014; Seto et al., 2017). Identified mutations are widely distributed along the gene without preferential mutation spots (Moog et al., 2015), while observed phenotypes vary in individuals. Hence, little can be deduced regarding the dominance of mutations and the key impact of CASK on the development of intellectual disorders and ASD. In normal physiology, CASK is a neurexin-binding protein (Hata et al., 1996) with a suspected role as a synaptic organizer that links cell adhesion and transsynaptic signaling with vesicle exocytosis (Butz et al., 1998). Despite an absence of effect on neuron excitability, and microstructure of synapses when CASK is ablated (Atasoy et al., 2007), there is a reduction in spontaneous synaptic events and synaptic vesicle cycling in the Drosophila neuromuscular junction (Chen and Featherstone, 2011) that corroborate the potential function of CASK in presynaptic vesicle exocytosis. Further work is required to pinpoint the precise neurological function of CASK and its correlation with the observed neurodevelopmental conditions.

The ability of non-neuronal HEK293 cells artificially expressing postsynaptic-specific CAM to induce the formation of presynaptic specializations in cocultured-primary neurons exemplifies the synaptogenic capability of CAMs (Scheiffele et al., 2000). Conversely, the reverse applies to neurexin—its presynaptic expression is sufficient to induce the formation of postsynaptic compartments (Graf et al., 2004). Despite their distinct role in synaptogenesis, the deletion of individual synaptic organizers does not adversely disrupt neurological development. Given the frequent presence of alternative genes, promoters, splice site in CAM coding sequences (Chih et al., 2006; Schor et al., 2013; Treutlein et al., 2014; Li et al., 2020), and preferential binding partners, the possible permutations of the interactions (Fowler et al., 2017; Nozawa et al., 2022), revealing the precise nanoscopic architecture of CAM interactions, is a monumental task. The importance of CAMs in neurological development can only be addressed with the resolution of the relative contribution of each type of CAMs in neurological development and function.

Presynaptic proteins that support the function of CAMs and synaptic organization complicate the process of synaptogenesis and specification. Liprin was identified as an interacting partner and regulator of the localization of the adhesion molecule leukocyte common antigen-related receptor protein tyrosine phosphatase (LAR-RPTP) (Serra-Pagès et al., 1995). As LAR-RPTPs were required for the synaptogenic function of presynaptic neurexin (Han et al., 2020) and postsynaptic Slit- and Trk-like proteins (Slitrks) (Yim et al., 2013), presynaptic Liprin can modulate synapse specification. The vertebrate Liprin family of proteins consists of four different isoforms of Liprin-α (α1–4), two Liprin-β (β1, 2), and a single KazrinE (Sakamoto et al., 2012). In mammals, liprin-α1 is ubiquitously expressed throughout the body, while Liprin-α2 and 3 are predominant in the central nervous system (CNS). Rare de novo missense mutations and chromosomal rearrangement that disrupt the intronic sequence of the Liprin-α1 coding gene PPFIA1 were found in autism proband and patients (Schluth-Bolard et al., 2013; Iossifov et al., 2014).

Liprin-α is also involved in the interaction and recruitment of several components of the presynaptic vesicle release machinery, such as the calcium/calmodulin-dependent protein kinase II (CaMKII), calcium/calmodulin-dependent serine protein kinase (CASK), and regulating synaptic membrane exocytosis protein 1 (RIM1) (Spangler and Hoogenraad, 2007; Spangler et al., 2013), to facilitate synaptic transmission. Interestingly, activity-dependent phosphorylation of neurexin by CASK destabilizes the CASK-Liprin-Neurexin complex, leading to dissociation and an increase in turnover of neurexin (LaConte et al., 2016). Consistent with the observation, mutations in the Caenorhabditis elegans homolog of Liprin-α syd-2 caused the a decrease in active site electron density, mislocation of synaptobrevin-GFP labeling in the active zone, and an increase in the overall size of the presynaptic active site, without affecting synaptic density (Zhen and Jin, 1999). A similar observation was made in the mammalian system—the depletion of Liprin-α2/3 in mice results in the disruption of the active zone ultrastructure, synaptic vesicle tethering, and vesicle exocytosis (Spangler et al., 2013; Wong et al., 2018). Liprin-α is further involved in postsynaptic Slitrk6 and neuroligin 2 (NLGN2)-mediated presynaptic differentiation in cultured rat hippocampal neuron culture (Han et al., 2018), likely via its function as a presynaptic anchor and organizer.

Upon targeted interactions between CAMs, there is a rapid and bulk recruitment of preassembled active zone proteins in the form of piccolo-bassoon transport vesicles (PTVs) and synaptic vesicle protein transport vesicles (STVs) (Sabo and McAllister, 2003; Shapira et al., 2003; Sabo et al., 2006; Tao-Cheng, 2007). These are thought to be readily integrated into the nascent presynaptic structure, and form the basis for further defined recruitment and localization of the functional subunits of the presynaptic release machinery. Regulating synaptic membrane exocytosis (RIM), RIM-binding protein (RIM-BP), protein-rich in the amino acids glutamic acid (E), leucine (L), lysine (K), and serine (S) (ELKS), and Liprins interact with each other to form a large protein network that recruits the necessary molecular components for effective vesicle exocytosis (Südhof, 2012). Rare inherited mutations in RIM and de novo RIM-BP variants have been identified in the human ASD population (Bucan et al., 2009; Iossifov et al., 2012; Dong et al., 2014). The localization of voltage-gated calcium channels (VGCC) in the presynaptic compartment is directly dependent on the interaction between their cytoplasmic domain and the PDZ domain of RIM and RIM-BP (Kaeser et al., 2011), while the interaction of RIM with Munc13 places Munc13 near the soluble N-ethylmaleimide sensitive factor attachment protein receptor (SNARE) machinery for the efficient coupling of vesicle priming and docking events (Betz et al., 2001; Dulubova et al., 2005; Ma et al., 2011). Indeed, the deletion of RIM or the RIM-BP results in the loss of presynaptic Ca²⁺ channels, reduced synaptic release probability in the immediate proximity of the presynaptic active zone, and impairment in activity-dependent neurotransmitter release (Kaeser et al., 2011; Liu et al., 2011; Zarebidaki et al., 2020). When combined with mutations in presynaptic factors associated with RIM or RIM-BP function, basic presynaptic homeostatic plasticity fails in Drosophila (Genç et al., 2020), suggesting a potential convergence of seemingly unrelated autism risk genes to the susceptibility of disturbed presynaptic function.

Distinct from the neurotransmitter-containing synaptic vesicles, large dense core vesicle (LDCV) exocytosis is regulated by the Ca²⁺-dependent secretion activator (CAPS), a two-member multidomain protein (Berwin et al., 1998; Renden et al., 2001). However, the enrichment of CAPS proteins in the neuron presynaptic compartments where there is very little LDCV (Sadakata et al., 2006) raised questions surrounding their possible involvement in synaptic vesicle exocytosis. Subsequent studies revealed their fundamental importance in priming synaptic vesicles for fusion and neurotransmitter release. CAPS-1 ablation in mouse hippocampal neurons drastically reduced the readily releasable pool (RRP) size and halved the evoked excitatory postsynaptic current (EPSC) amplitude, despite comparable release probability during RRP release and EPSC kinetics (Jockusch et al., 2007). Furthermore, the double knockout of CAPS renders a significant proportion of the analyzed neurons (39%) incapable of eliciting an EPSC response even with hypertonic buffer treatment. The functional defects observed are due to the loss of docked vesicles with CAPS-1 knockout (Shinoda et al., 2016). CAPS mutations are further linked to the disrupted release of neuropeptides involved in the modulation of social behavior (Fujima et al., 2021). This is in line with the observation that ASD patients have lower blood oxytocin concentration (Zhang et al., 2016), and the oxytocin-treatment-dependent improvement in social skills is only present in individuals with lowered oxytocin levels (Parker et al., 2017). Nonetheless, there have been mixed results in clinical trials involving oxytocin-related treatments, and the efficacy of oxytocin in enhancing the social function of ASD patients remains unclear. Although CAPS mutations are rare in human autism patients (Sadakata et al., 2007; Bonora et al., 2014), likely due to CAPS’ functional importance in the nervous system, gene mutations directly impacting the priming function of CAPS-1/2 would be easily translated into behavioral changes.

The process of synaptic vesicle recruitment, priming, fusion, and recycling during chemical neurotransmission has been extensively examined (Okamoto and Südhof, 1997; Rizo and Rosenmund, 2008; Südhof, 2013; Imig et al., 2014; Chanaday et al., 2019). Due to the ubiquitous nature of Ca²⁺ in excitable cell function, function-modifying mutations in the VGCC have been identified in several neurological and neuromuscular conditions, including ASD (Bidaud et al., 2006; Lu et al., 2012). In the neuron, activity-dependent activation of the VGCC and influx of Ca²⁺ in the presynaptic terminal is a critical step in the release of neurotransmitters. The formation and assembly of the SNARE complex are heavily dependent on the presence of Ca²⁺ (Chen et al., 1999) whereby the forebrain-specific ablation of calcium voltage-gated channel subunit α1 A (CACNA1A) results in deficits in a variety of cognitive functions related to learning and memory and circadian rhythms (Mallmann et al., 2013). Depending on the site of impact, mutations in calcium voltage-gated channel subunit α (CACNA) which encodes for the main pore-forming subunit of the VGCC likely have an impact on the activation profile and kinetics of VGCC function.

Synaptic vesicles containing neurotransmitters can translocate between presynaptic terminals (Staras et al., 2010), and the vesicle distribution can be affected by synapsin (SYN) and β-catenin (Bamji et al., 2003; Pechstein et al., 2020), which sequester mobile vesicles. SYN is a family of evolutionarily conserved presynaptic proteins (Candiani et al., 2010) that plays a critical role in the structural and functional organization of the presynaptic terminal. Particularly, it reversibly tethers synaptic vesicles to the actin cytoskeleton which is crucial for the establishment of the vesicle pool for efficient neurotransmission (Baldelli et al., 2007; Cesca et al., 2010). Nonsense (Q555X) and missense (A550T and T567A) mutations, and maternally-inherited frameshift (A94fs199X) and missense (Y236S and G464) mutations have been identified in SYN1 and SYN2, respectively (Fassio et al., 2011; Corradi et al., 2014). Defects in directed targeting of the A550T and T567A SYN1 mutants, the lack of expression of the nonsense A94fs199X variant of SYN2, and the ablation of SYN2 function with Y236S and G464R revealed the importance of the SYN family activity in the regulation of synaptic function. The onset of SYN expression coincides with neuronal differentiation and peaks during synaptogenesis (Haas and DeGennaro, 1988; Melloni and Degennaro, 1994). Beyond this, SYN knockout in mice results in a differential impact on the populations of synaptic vesicles within the excitatory and inhibitor terminals (Gitler, 2004; Chiappalone et al., 2009), alongside epileptic phenotypes and ASD-like behaviors (Li et al., 1995).

It is recognized that spontaneous neurotransmitter release is important in the regulation of presynaptic maturation in the developing nervous system (Choi et al., 2014), as well as the regulation of postsynaptic receptor clustering and synaptic strength (Saitoe et al., 2001). Spontaneous release is dependent on the function of the putative presynaptic calcium sensors synaptotagmin (SYT), double C2-like domain-containing protein alpha (DOC2α), and double C2-like domain-containing protein beta (DOC2β) (Xu et al., 2009; Groffen et al., 2010), which differs in their cell-type specific expression profile (Courtney et al., 2018), sensitivity toward Ca²⁺, kinetics, and preference for phospholipid binding (Kojima et al., 1996; Groffen et al., 2006). Calcium binding changes the conformation of C2 domains, allowing them to insert into, buckle, and bring the presynaptic membrane closer to the synaptic vesicle to promote membrane fusion and neurotransmitter release (Martens et al., 2007). Mutations of SYT1 clustered mainly within the C2B domain which affected exocytosis rates following sustained action potential stimulation (Baker et al., 2015, 2018). SYT with M303K, D304G, and D366E mutations specifically failed to localize or relocalize to the presynaptic terminals following exocytosis. Comparatively, the DOC2A M225I variant likely affects its interaction with Munc13 (Kumar et al., 2009) and consequently SNARE-dependent fusion of synaptic vesicles. Hence, phenotypes associated with autism-associated SYT and DOC2 mutations originate from the disturbed neuronal transmission.

The importance of the SNARE complex in vesicle exocytosis cannot be overstated. Spontaneous membrane fusion in living organisms is energetically unfavorable—a large amount of directed force is required to overcome the repulsive forces between the lipid structures and lateral tension on the membrane surface (Chernomordik et al., 1987; Kozlovsky et al., 2002). Cooperative interactions between the vesicle SNARE (v-SNARE) synaptobrevin, and target-localized SNAREs (t-SNAREs) syntaxin-1 and synaptosomal-Associated Protein, 25kDa (SNAP-25) complex provide sufficient mechanical force for the fusion of opposing bilayers (McNew et al., 2000). Amisyn is a brain-enriched protein with a tomosyn- and synaptobrevin (VAMP)-liked coiled-coil-forming domain that competes with synaptobrevin-2 (VAMP2) for the assembly of the SNARE complex and inhibition of SNARE-dependent vesicle priming (Kondratiuk et al., 2020). Amisyn-containing SNARE complexes are more thermally stable than conventional VAMP2-containing SNARE complexes (Scales et al., 2002), but fusion-incompetent due to the absence of a transmembrane anchor in Amisyn. Interestingly, amisyn or syntaxin binding protein 6 (STXBP6) knockout in mice does not lead to any behavioral abnormalities (Liu et al., 2021), suggesting that Amisyn is not critical for neuronal development but may be involved in the regulation of activity-dependent of synaptic vesicular release. To date, further information regarding the regulatory mechanism of Amisyn function is absent, with only speculations of the full implication of STXBP6 mutation in ASD.

Following activity-dependent exocytosis, synaptic vesicles can be regenerated through the reuse of synaptic vesicles that have transiently docked and fused with the synaptic membrane (kiss-and-run fusion) (Alabi and Tsien, 2013), clathrin-mediated endocytosis of fully collapsed lipid structures from the plasma membrane, activity-dependent bulk endocytosis (Cheung et al., 2010) or fast endophilin-mediated endocytosis (Watanabe and Boucrot, 2017). Dual Specificity Tyrosine Phosphorylation Regulated Kinase 1A (DYRK1A) encodes a chromosome 21-associated proline-directed serine/threonine kinase with dual function in the regulation of gene transcription and clathrin-mediated endocytosis. Recurring DYRK1A haploinsufficiency has been observed in ASD and is associated with microcephaly, intellectual disability, and epileptic seizures (Earl et al., 2017).

At the early stages of neurodevelopment, DYRK1A phosphorylation of Lin-52 dREAM MuvB core complex component (LIN52) is required for the assembly of the dimerization partner (DP), retinoblastoma (RB)-like, E2F, and multi-vulval class B (MuvB) (DREAM) complex which coordinates cell cycle-dependent gene expression (Litovchick et al., 2011; Sadasivam and DeCaprio, 2013). Despite the presence of a nuclear localization sequence, the majority of the DYRK1A remained localized to the cytoplasm of neurons (Hämmerle et al., 2003; Martí et al., 2003). DYRK1A was detected in isolated latherin-coated vesicles and colocalized with latherin in mouse neurons. Mass spectrometry analysis revealed the phosphorylation of adaptins, dymain 1 (DYN1), amphiphysin 1 (AMPH1), and synaptojanin 1 (SYNJ1) by DYRKA1 which inhibits the onset of latherin-mediated endocytosis and promotes the dissociation of clathrin structures on vesicles (Murakami et al., 2012). While the autism-associated R205X and E239X truncations are linked to defects in dendritic growth and spine development in rodents (Dang et al., 2018), the normalization of DYRK1A expression postnatal is sufficient to ameliorate synaptic and the functional changes linked to altered synaptic plasticity (Duchon and Herault, 2016). Similarly, the multimodular intersectin 1 (ITSN1) has been shown to interact with endocytic proteins and control endocytosis in various cell types and organisms, with contradictory effects observed for mammalian synaptic vesicle endocytosis (Gubar et al., 2013).

Although synaptophysin (SYP) is the most abundant synaptic vesicle membrane protein, its function is enigmatic. The existing report suggests the diverse roles of SYP function, including synaptogenesis (Tarsa and Goda, 2002), and the biogenesis of synaptic vesicles (Leube et al., 1989; Cameron et al., 1991; Kwon and Chapman, 2011). Mutant mice lacking SYP are viable and does not exhibit changes in neuronal structure and behavior (McMahon et al., 1996). The lack of phenotypes is accounted for by the redundancy of function between the synaptophysin and synaptogyrin family of proteins (Raja et al., 2019). Interestingly, SYP knockout results in the mislocalization of VAMP2 with no impact on vesicle turnover, likely reflecting a defect in SYP-dependent VAMP2 retrieval during synaptic vesicle endocytosis (Gordon et al., 2011). As such, while SYP variants have been identified in a population of Vietnamese autistic patients (Tran et al., 2020) and subjects recruited by the Autism Sequencing Consortium (Satterstrom et al., 2020), their role in ASD is not known.

Regardless of the mechanism adopted for the replenishment of synaptic vesicles, empty vesicles need to be rapidly refilled in the anticipation of a new round of neuronal activity and neurotransmission. Partially filled synaptic vesicles have a lower release probability (Rost et al., 2015) and can affect signal transmission. Vacuolar H⁺-ATPase (vATPase) is first required to establish a proton gradient across the vesicle membrane before neurotransmitters can be efficiently loaded into the vesicle lumen through the established electrochemical gradient (Farsi et al., 2017). The precise molecular composition and arrangement of synaptic-vesicle-associated vATPase have not been fully resolved.

Unlike conventional ASD risk genes, the correlation of nuclear receptor coactivator (NCOA7) with autism pathogenesis was revealed through the multidimensional examination of shared co-expression relationships of previously identified autism candidate genes with normal neurodevelopmental processes (Mahfouz et al., 2015). Up to date, there is only a single recessive case of autism known to be due to mutation in nuclear receptor coactivator 7 (NCOA7) (Autism Sequencing Consortium et al., 2019), and very little is known about NCOA7 function in neuronal physiology. NCOA7 is widely expressed throughout all developmental time frames of the mouse brain and interacts with various cytosolic V1 subunits of the vATPase in vivo. The loss of NCOA7 results in the increase in the number of proximal neurites of cultured primary neurons and a reduction in inhibitory synapses in layer 2/3 of the somatosensory cortex, which is related to the impaired social behavior observed (Castroflorio et al., 2021). A closer examination of potential ASD risk factors linked to vATPase function could reveal novel mechanisms for the development of neurodevelopmental and neuropsychiatric disorders.

The Na⁺/H⁺ exchanger 6 and 9 (NHE6/9 or SLC9A6/9) has been identified as candidate gene of interest for attention deficit hyperactivity disorder (ADHD) (Lasky-Su et al., 2008) and ASD (Morrow et al., 2008). Nonsense mutations that result in a similar premature termination of the last transmembrane segment of NHE9 have been observed in the closely related NHE1 and NHE6, associated with lower cognitive ability and epilepsy (Cox et al., 1997; Gilfillan et al., 2008). The sensitivity of Na uptake by glutamatergic synaptic vesicles toward a low micromolar amount of 5-(N-ethyl-N-isopropyl)amiloride, an inhibitor of NHEs suggests the involvement of NHEs in the filling of synaptic vesicles (Goh et al., 2011). Furthermore, mutations in the secretory carrier-associated membrane protein 5 (SCAMP5), a synaptic vesicle enriched protein responsible for the trafficking and synaptic localization of NHE6, were reported in idiopathic ASD and autism-like neurodevelopmental disorder with the manifestation of epilepsy (Castermans et al., 2010; Hubert et al., 2020). The knockdown of SCAMP5 in rat hippocampal neurons mislocalized NHE6 and hyperacidified the synaptic vesicles within the neuron (Lee et al., 2021). Given the prominent endosomal function of NHE, a careful evaluation of the contribution of NHEs is warranted before any conclusion can be made regarding its role in ASD development.