SAPAP Scaffold Proteins: From Synaptic Function to Neuropsychiatric Disorders

Excitatory (glutamatergic) synaptic transmission underlies many aspects of brain activity and the genesis of normal human behavior. The postsynaptic scaffolding proteins SAP90/PSD-95-associated proteins (SAPAPs), which are abundant components of the postsynaptic density (PSD) at excitatory synapses, play critical roles in synaptic structure, formation, development, plasticity, and signaling. The convergence of human genetic data with recent in vitro and in vivo animal model data indicates that mutations in the genes encoding SAPAP1–4 are associated with neurological and psychiatric disorders, and that dysfunction of SAPAP scaffolding proteins may contribute to the pathogenesis of various neuropsychiatric disorders, such as schizophrenia, autism spectrum disorders, obsessive compulsive disorders, Alzheimer’s disease, and bipolar disorder. Here, we review recent major genetic, epigenetic, molecular, behavioral, electrophysiological, and circuitry studies that have advanced our knowledge by clarifying the roles of SAPAP proteins at the synapses, providing new insights into the mechanistic links to neurodevelopmental and neuropsychiatric disorders.


Introduction
Synapses are fundamental elements of neural circuits and networks that convey all aspects of brain function, and pathological alterations in synaptic structure and function are broadly held to underlie many neuropsychiatric disorders, such as autism spectrum disorder (ASD), schizophrenia, obsessive compulsive disorder (OCD), cognitive disorders, and mood disorders [1][2][3][4][5][6][7]. In the mammalian brain, the vast majority of synapses are excitatory synapses (primarily glutamatergic), which occur predominantly at contacts between presynaptic axons and postsynaptic tiny, actin-rich protrusions known as dendritic spines. During synaptic plasticity at excitatory synapses, the molecular composition of postsynaptic membranes and the chemical modification of synaptic proteins are key determinants of the number, strength, morphology, and even connectivity of neuronal synapses [8][9][10]. The postsynaptic density (PSD) at the tip of a dendritic spine's head, which is an electron-dense thickening assembly situated underneath the postsynaptic membrane, is composed of numerous proteins, including adhesion proteins, membranetethered receptor and ion channels, scaffold proteins, signaling molecules, and cytoskeletal proteins [8,9,11]. These components of the PSD assemble into a dynamic macromolecular complex and are crucial for synaptic transmission and plasticity [12]. During development, and in response to stimulation or inhibition, the PSD dynamically undergoes changes in molecular composition and structure in order to tune the strengths and/or efficacy of synaptic signaling.
Despite the potential importance of SAPAP proteins in these neuropsychiatric disorders, the in vivo functional roles of SAPAP proteins are not yet fully understood. Here, we review the most recent findings that have emerged from human genetics studies exploring the links between mutations in SAPAP genes and neuropsychiatric disorders. We further re-view evidence from the recent studies of mutant mice concerning the physiological roles of SAPAP proteins at the excitatory synapses and provide deeper insights into the mechanistic links to neuropsychiatric disorders.

SAPAP-Interacting Proteins
With the N-terminal conserved 14 aa repeat domain, SAPAP proteins directly interact with the GK domain of the PSD-95 family scaffolding proteins-a subfamily of the membrane-associated guanylate kinase (MAGUK) family composed of PSD-95, PSD-93, SAP97, and SAP102 [1,36]. PSD-95 and other members of the MAGUK family not only interact directly with the NR2A and NR2B subunits of NMDARs and Shaker-type K + channels through C-terminal PDZ ligand motifs to promote the clustering of these proteins in the PSD [1,11], the neuronal cell adhesion molecule neuroligin [39,40] via its third PDZ domain, but also interact indirectly with AMPA receptors (AMPARs) through interaction with the transmembrane protein stargazin/TARPs [41][42][43], or other auxiliary proteins [44] to regulate the synaptic trafficking of NMDARs and AMPARs [45,46]. Signaling molecules including SynGAP (a synaptic Ras-GTPase activating protein) and neuronal nitric oxide synthase (nNOS) also interact with the PDZ domains of PSD-95 and present in a large complex with PSD-95 and the NMDARs in the brain [47,48]. Both the SH3 and GK domains in PSD-95 bind to and induce clustering of the kainate subtype ionotropic glutamate receptors [49], which have been linked to schizophrenia and other psychiatric disorders, such as intellectual disability (ID) and ASD [50].
An additional partner to the 14 aa repeat domain of SAPAP proteins is S-SCAM [51], which also interacts with NMDAR subunits, neuroligin 1, and β-catenin at excitatory synapses [51,52], neuroligin 2 and β-dystroglycan at inhibitory synapses [52], and the scaffold protein tamalin [53]. In addition, the N-terminal region of SAPAPs also interacts with neurofilaments (NFLs), but not with actin or tubulin. The NFL-interacting region is different from the regions interacting with PSD-95 and S-SCAM [54].
The C-terminal region of SAPAP proteins subsequently interacts with the PDZ domain of the SHANK protein family, including SHANK1, SHANK2, and SHANK3 [55,56]. SHANKs then not only bind to the Homer family proteins-thereby linking metabotropic glutamate receptors (mGluR1/5) to SAPAPs [57]-but also provide a link between the SAPAP proteins and the actin cytoskeleton through interactions with the F-actin-binding protein cortactin [55,58]. SAPAPs can bridge the PSD-95 family and SHANK family to form the PSD-95/SAPAPs/SHANKs core complex [36], which is thought to be a major scaffold organizer in orchestrating the synaptic formation and plasticity at glutamatergic synapses. Additionally, SAPAP-SHANK interaction also involves the participation of N-terminal extension sequences of the PDZ domain [59,60]. These extension sequences govern the exquisite and strong interaction between SAPAPs and SHANKs. A recent study also shows that PSD-95, SAPAP1, and SHANK3 can form scaffold complexes with one another at different developmental stages [61].
The proline-rich domain of SAPAPs binds to the third SH3 domain of nArgBP2 [62]. nArgBP2, which is important for neuronal dendritic development and spine synapse formation [63,64], then binds to signaling molecules including the ubiquitin ligase CBL, protein kinases ABL and PYK2, and various proteins involved in the regulation of cell adhesion and the actin cytoskeleton [63][64][65].
The DLC domain of SAPAP proteins binds directly to DLC2 and DLC8 [38,66,67], which colocalize with SAPAP/PSD-95 in the PSD and also tend to distribute into the deep part of the spine [38]. DLC is an accessory subunit shared by dynein and myosin-V, and is highly enriched in dendritic spines [38,67], suggesting that DLC may link SAPAPs to actinand microtubule-based motors and act as a motor-cargo adaptor.
The structure of the GH1 domain of SAPAP proteins is believed to be dynamic [68]. The function and interacting partners of the GH1 domain have not yet been well described [36]; however, one recent study identified a patient with cortical malformations who carried a novel SAPAP4 mutation, affecting the last part of the GH1 domain and leading to loss of a polyproline-rich domain of SAPAP proteins [34].

Expression of SAPAPs in the CNS
During postnatal development, the different SAPAP members are expressed differently. The temporal patterns of Sapap1 and Sapap2 expression in individual brain regions do not change prominently after birth [19]. For example, the expression of Sapap1 mRNAs in the cortex seems to slightly peak around postnatal day 90. However, the spatiotemporal expression pattern of Sapap3 during early postnatal development is noticeable, with peak expression observed around 2-3 weeks after birth in the cortex, striatum, and thalamus [19,71]. Although high expression levels are seen in the cortex within the first 3 weeks after birth, the distribution of Sapap4 mRNA appears to slightly decline afterwards [19]. Additionally, recent studies have indicated that both Sapap1 and Sapap4 are expressed in progenitor cells-the latter notably strongly expressed in the cortical progenitors and migrating neurons-suggesting the important roles of SAPAPs during early corticogenesis [34,72].
In the adult rodent brain, the Sapap mRNAs and proteins display overlapping yet distinct distribution characteristics. All four Sapap mRNAs are expressed abundantly in the cerebral cortex, hippocampus, and olfactory bulb, but at low or undetectable levels in the hypothalamus and substantia nigra [18,19]. Sapap1 mRNA is expressed abundantly in the cortex, hippocampus, amygdala, cerebellum, and olfactory bulb, moderately in the thalamus, and at low levels in the striatum [13,18]. Notably, Sapap1 mRNA is only expressed in a small subset of cells in the striatum and displays heterogeneous expression patterns in different subregions of the thalamus [18], but it is uniformly distributed throughout the whole cerebral cortex.
Distinct from the other SAPAP family members, Sapap2 mRNAs are expressed in a limited pattern-mainly in the forebrain, with relatively low expression levels. Sapap2 displays its highest expression levels in the hippocampus, with moderate levels in the striatum and almost undetectable levels in the cerebellum, thalamus, and amygdala [13,18]. In the cortex, Sapap2 mRNAs show relative strong expression in layers 2 and 3 [18].
The most abundant Sapap detected in the striatum and various thalamic nuclei is Sapap3 [18,19]. Moreover, Sapap3 mRNAs are also highly expressed in layers 1-3 of the cortex, but only moderately expressed in the amygdala [13,18]. An intriguing aspect of the subcellular distribution of Sapap3 is its dendritic localization. It has been found that several mRNAs are localized in the dendrites [73,74], and an extensive body of literature directly linked synaptic plasticity with local protein synthesis and degradation in dendrites [75][76][77] (reviewed in [73,[78][79][80]). In the hippocampus, Sapap3 is the only family member that is highly localized in dendrites [18,19], which possibly allows rapid turnover of proteins and thus provides a mechanism for quick, activity-dependent changes in synaptic protein complex composition [73,80,81], suggesting that SAPAP3 may play a unique role in synaptic plasticity.
Sapap4 is highly expressed in many regions of the brain, such as the hippocampus, cortex, amygdala, and cerebellum [13,18,19]. In the hippocampus, the expression of Sapap4 mRNA and proteins is relatively high in CA1 and CA3, but low in the dentate gyrus. Additionally, Sapap4 mRNA shows high expression in layer 5 of the cortex. In addition to these areas, Sapap4 is also strongly expressed in the adult rodent thalamus [17,18] and striatum [13,82]-second only to Sapap3. Of particular interest, Sapap4 mRNA is strongly expressed in the locus coeruleus [18] as well as the ventricular zone and migrating neurons [34].
In general, the SAPAP proteins are predominantly localized at glutamatergic and cholinergic synapses, and not at GABAergic or glycinergic synapses [18], suggesting that SAPAPs may be a general "core" postsynaptic component of excitatory synapses, but not of inhibitory synapses. Moreover, subsequent studies have found that SAPAP3 and SAPAP4 may tend to present at corticostriatal and thalamostriatal excitatory synapses, respectively [13,82]. These findings of molecular heterogeneity of SAPAPs at different synapses imply that SAPAP family members may be localized at synapses in a circuitselective manner, and that the molecular specificity in SAPAP proteins attributes a unique and specific function to the SAPAP family members. A systematic analysis of the SAPAPs according to the subregions and developmental stages of the brain remains to be determined and would be important for understanding the functions of SAPAPs in the brain.
The SAPAPs themselves also act as substrates for phosphorylation and ubiquitination. Previous in vivo and in vitro studies have indicated that SAPAPs can be phosphorylated by kinases such as PKA, PKC, CDK5, ERK1, P38 MAPK, AKT1, and CaMKII [70,[84][85][86]. SAPAP1 protein complexes show increased binding in specific kinases and phosphatases upon induction of LTP [87]. Interestingly, PSD-95-SAPAP interaction and SAPAP-DLC interaction, which are dependent upon CaMKII phosphorylation of SAPAP, are crucial for proper SAPAP targeting and synaptogenesis [86,88]. Moreover, PKC phosphorylation of SAPAP proteins can also promote PSD-95-SAPAP complex formation and further enhance the clustering and formation of AMPAR nanodomains [84], while CaMKII phosphorylation of SAPAPs can control the turnover of SAPAPs at synapses in a bidirectional manner and phosphorylation of Ser54 in SAPAP promotes its removal from synapse sites [86]. Notably, overexpression of the SAPAP turnover mutants not only results in the loss of polyubiquitination and degradation of SAPAP from synapses but also eliminates activitydependent remodeling of PSD-95, SHANK and regulation of surface AMPAR levels [86], indicating that SAPAP is critical for regulation of postsynaptic protein organization and synaptic structure. Moreover, SAPAPs can also be ubiquitinated by an E3 ligase (TRIM3), leading to its subsequent degradation and loss of SHANK1 from postsynaptic sites, which is involved in regulation of dendritic spine morphology [89]. CDK5 phosphorylation of SAPAP appears to be central in triggering ubiquitination and degradation of SAPAPs, as well as remodeling of the synaptic actin structures [70] (Figure 2). Together, the specialized functional network organized by PSD-95-SAPAP-SHANK scaffold complexes facilitates the activity-dependent remodeling of dendritic spines [86,89]; thus, the disruption of the activity-dependent turnover of PSD scaffold proteins such as SAPAPs can affect synaptic formation and plasticity, leading to abnormal synaptic and behavioral phenotypes. AMPAR, α-amino-3-hydroxy-5-methyl-4-isoxazole propionic acid receptor; NMDAR, N-methyl-Daspartic acid receptor; KCh, Shaker-type K + channels; SAPAP, SAP90/PSD-95-associated protein; mGluR1/5, Group I metabotropic glutamate receptors; CaMKII, Calcium/calmodulin-dependent protein kinase II; PKC, Protein kinase C; SynGAP, Synaptic GTPase Activating Protein; nNOS, Neural Nitric Oxide Synthase; S-SCAM, synaptic scaffolding molecule; nArgBP2, Neural Abelson-related gene-binding protein 2; CBL, E3 ubiquitin ligase casitas B-lineage lymphoma; ABL, tyrosine kinase; PYK2, proline-rich tyrosine kinase 2; APP, amyloid precursor protein; P, phosphorylation; FMRP, fragile X mental retardation protein; TRIM3, Tripartite motif-containing 3; Ub, ubiquitin; DLC, dynein light-chain; CDK5, cyclin-dependent kinase 5; TRAP, transmembrane AMPAR regulatory protein.

Synaptic Formation and Maturation
As an indispensable central member of the axis of PSD-95-SAPAP-SHANK scaffold complexes in the PSD [4,8,9,11], SAPAPs play critical roles in initial synaptic formation and development. SAPAPs are always present before NMDAR and AMPAR clusters become synaptic [90], suggesting a structural role of SAPAPs in synapse formation. Indeed, the SAPAP-SHANK interaction is known to be essential for synaptic targeting of SHANK1 [55,91], whereas the interaction between SAPAPs and PSD-95 is crucial for synaptic localization of NMDARs [90], SHANKs [92], and SAPAPs [88]. Moreover, the SAPAP-DLC2 interaction is primarily located in dendritic spines, activitydependent [93], and also thought to play important roles in proper targeting of SAPAP proteins to synapses [86] and controlling NMDARs' function in synaptic transmission [94] ( Figure 2).
Consistent with an essential function in SAPAP-mediated synaptogenesis and synaptic activity [88], the studies of SAPAPs in vivo have also highlighted a potential role of SAPAPs in synaptic development. For example, the deletion of the Sapap3 gene results in an increase in "juvenile" NMDAR subunit composition [13] and silent synapses (i.e., those containing no AMPARs) [95], suggestive of immature corticostriatal synapses. Intriguingly, the absence of Sapap4 also leads to similar alterations in the hippocampus [16], suggesting critical roles of SAPAP3 and SAPAP4 in promoting the early postnatal switch in NMDAR subunit expression and regulating the maturation of excitatory synapses.
Correspondingly, loss of SAPAPs in mice not only leads to disruption of protein interactions in the PSD [15], but also affects the dendritic arborization, spine numbers, axon caliber, and the structure of excitatory synapses in different brain regions [13,14,16,17,35,96]. In particular, SAPAP2 has been identified as a critical component involved in modulating synaptic function and development. For Sapap4, cell-type-specific differential isoform expression is seen during the transition from immature to mature neurons, probably reflecting the dynamic, diverse, and complex functions of SAPAPs [97]. More recently, SAPAP4 has been found to be required for proper cortical development [34]. These findings indicate an important role of SAPAPs in regulating excitatory synaptic formation and maturation.

Synaptic Transmission and Plasticity
Homeostatic plasticity has essential consequences for maintaining the stable function of neural circuits over a wide range of spatiotemporal scales [98]. There is evidence that SAPAPs can contribute to synaptic scaling by regulating the accumulation of postsynaptic receptors and scaffold proteins at synaptic sites [4,86]. The ubiquitination and degradation of SAPAPs at synapses is thought to be required for the normal activity-dependent remodeling of postsynaptic scaffold proteins such as PSD-95 and SHANKs, as well as excitatory synaptic scaling [86]. Consistent with the role of SAPAPs in synaptic scaling, deletion or perturbation of different SAPAP family members in mice also leads to alterations in the PSD levels of receptors and scaffold proteins, including NMDARs, AMPARs, mGluRs, Homer, SHANK, and αCaMKII [13][14][15][16][17]99]. Strikingly, not only the deletion of the repeat region R1 in SAPAP [86], but also the removal of SHANK [100], leads to the loss of AMPAR-containing synapses, along with impaired synaptic transmission.
The absence of SAPAP2 can lead to postsynaptic and presynaptic deficits, including a decrease in AMPAR-mediated postsynaptic response and an increased paired-pulse ratio (PPR), which indicates a reduction in the probability of presynaptic release in the orbitofrontal cortex (OFC) [14]. Two recent independent studies using different knockout strategies revealed that Sapap4-mutant mice exhibit decreased NR2A-mediated NMDAR currents and increased AMPAR transmission in the hippocampus [16], impaired synaptic transmission, and decreased AMPAR-mediated postsynaptic responses in the nucleus accumbens (NAc) [35], without alterations in the presynaptic neurotransmitter release. Moreover, SAPAP4 deficiency also impairs neuronal network function and affects synaptic plasticity in the hippocampal CA1 region, impairing long-term depression (LTD) but enabling the induction of long-term potentiation (LTP) [16].
Due to its strong association with OCD, the synaptic roles of SAPAP3 have been well studied. Sapap3-mutant mice exhibit significant deficits in corticostriatal synaptic function, including reduced AMPAR-mediated postsynaptic responses and elevated signaling through NMDAR and mGluR5 [13,95,99,101], without significantly affecting thalamostriatal AMPAR synaptic function [82]. Furthermore, there is ample evidence for input-or circuit-specific impairments of corticostriatal synaptic transmission [82,[101][102][103][104][105][106][107]. Beyond participating in glutamatergic system, SAPAPs have been implicated in modulating the monoaminergic system. Sapap3-mutant mice display upregulation of serotonin turnover in cortical and striatal regions and dihydroxyphenylacetic acid/dopamine ratios in the OFC [108], along with alterations in dopamine receptor density within the NAc [109], indicating an important role of SAPAP proteins in glutamate-monoamine interplay, which is relevant in psychiatric disorders such as schizophrenia, mood disorders, and OCD [110,111]. A follow-up study revealed an anomalous excessive form of synaptic plasticity (i.e., endocannabinoid-mediated LTD) expressed at striatal excitatory synapses, requiring mGluR5 signaling in Sapap3-mutant mice [99].

SAPAPs' Expression, FMRP, and Neuropsychiatric Disorders
Substantial studies have shown the genetic and epigenetic dysregulation of SAPAP genes in patients afflicted with psychiatric disorders or related animal models [21][22][23][112][113][114][115][116]. Fragile X mental retardation protein (FMRP) is considered to be a translational repressor that plays a key role in regulating local mRNA translation in response to synaptic signaling. In Fmr1knockout (KO) mice-a model for fragile X syndrome-the protein levels of several FMRP targets are increased in PSD fractions either from the neocortex or hippocampus, including Sapap1-3, Shank1, Shank3, and various glutamate receptor subunits [117]. Conversely, Sapap3 is decreased in both the OFC and medial prefrontal cortex (mPFC) of Fmr1-KO mice [118], which may contribute to the deficits in cognitive flexibility found in fragile X syndrome. Additionally, aggressive experience also increases the expression of Sapap3, coupled with a decrease in FMRP phosphorylation that is dependent on the activation of mGluR5 [119]. The postsynaptic targets of FMRP also include Sapap4, mGluR5, and many other scaffold proteins that comprise the glutamate receptor interactomes, such as PSD-95, Homer1, and Neuroligins 1-3 [120][121][122]. The mRNA for Sapap4 is significantly increased by the activation of mGluRs in WT but not Fmr1-KO neurons [123]. Diminished mGluR-induced dendritic localization of Sapap4 mRNA is found in the hippocampal neurons of Fmr1-KO mice [123], indicating a surprising coordinate regulation between Sapap4 and mGluRs that may be altered in fragile X syndrome. In accordance with these findings, increased mGluR5-signaling-dependent AMPAR endocytosis or altered mGluR5-Homer scaffolds are observed in OCD models with Sapap3-mutant mice, fragile X syndrome models with Fmr1-KO mice, and autism models with Shank3-mutant mice, which all include repetitive behaviors, among other symptoms [95,101,124,125], suggesting a potential shared genetic mechanism of different neuropsychiatric disorders (Figure 2).
Altered expression levels of SAPAPs are also observed in different neuropsychiatric disorders (Table 1). Sapap1 expression in the NAc is significantly increased in both animal models of phencyclidine-induced schizophrenia and unmedicated patients [115]. Sapap2 mRNA expression, which is significantly associated with its DNA methylation status, was associated with coping with stress in an animal model of post-traumatic stress disorder (PSTD) [112]. Similarly, the effects of Sapap2 methylation on memory ability in AD patients are also thought to be mediated by changes in SAPAP2 expression in the dorsal lateral prefrontal cortex (dlPFC) [22]. Increases in Sapap3 levels have been observed in both epilepsy patients and murine models [116]. Sapap4 has been indicated to be involved in psychostimulants' actions and bipolar disorder (BD). In taste-aversion-resistant rats after cocaine exposure, the mRNA levels of Sapap4 in the amygdala were strikingly lower than those in taste-aversion-prone rats [126]. Remarkably, sensitivity to cocaine and amphetamine was also changed in Sapap4-deficient mice [17,35]. Interestingly, in human neural progenitor cells derived from BD patients, Sapap4 has been identified as a target gene of miR-1908-5pa BD-associated epigenetic regulator [127]-suggesting that epigenetic dysregulation of SAPAP4 may be involved in the pathogenesis of BD. Taste-aversion-resistant rats after cocaine exposure [126] Epigenetic dysregulation of the expression of SAPAP genes, mainly mediated by changes in DNA methylation and histone acetylation, may confer risks of various neuropsychiatric disorders ( Table 2). Epigenetic modifications of the Sapap2 gene have been implicated in PTSD [112], alcohol and cannabis use [21,114], schizophrenia [113], and AD [22]. One study revealed that histone deacetylases and methyl-CpG-binding protein 2 (MeCP2) inhibit repetitive behaviors through regulation of the Sapap3 gene [128]. Additionally, recent studies have also reported an association between altered methylation of SAPAP1 and stress responses to violence [129] and a link between epigenetic dysregulation of SAPAP4 and early-onset cerebellar ataxia [23,130]. Identified as a target of the BD-associated microRNA miR-1908-5p N/A Neural progenitor cells BD patients [127] Abbreviations: AD, Alzheimer's disease; BD, bipolar disorder; PTSD, post-traumatic stress disorder.

SAPAP1 and Neuropsychiatric Disorders
Several human molecular genetic studies have emerged to show that SAPAP1 is related to several neuropsychiatric disorders, including schizophrenia, OCD, AD, ADHD, ASD, mood disorders, and cortical malformation [24][25][26][27][28][131][132][133][134] (Table 3). The SAPAP1 gene is located on chromosome 18p11, which has been suggested to be a susceptibility region for schizophrenia and/or BD [135,136]. Although several SNPs found in SAPAP1 are not associated with schizophrenia [131,132], a de novo CNV mutation of SAPAP1 has been identified in patients with schizophrenia [24]. The first genome-wide association study (GWAS) with 1465 OCD cases revealed that the two SNPs with the lowest p-values were located within SAPAP1, suggesting a subthreshold statistical significance for SAPAP1 [25]. Furthermore, the fact that two rare CNVs (a 62 kb duplication for SAPAP1 and a 16 kb deletion for SAPAP2) overlapping the FMRP targets SAPAP1 and SAPAP2 were uncovered in another OCD study is of particular interest [26]. Two SAPAP1 SNPs associated with cognitive flexibility have also been identified in ADHD children [27]. In addition, SAPAP1 is also implicated in ASD [133] and recurrent major depressive disorder (MDD) [134]. Interestingly, the expression of SAPAP1, which is significantly downregulated in the hippocampus and cortex of AD patients, is strongly associated with early identified AD variants (i.e., noncoding SNP rs8093731 in Desmoglein-2/DSG2 that acts as an expression quantitative trait loci for SAPAP1) [28] and SAPAP1 is identified as a potential candidate gene involved in AD.

ADHD and MDD
Rare missense mutation c.1922A > G (a benign variant) [132] Associated with the IGAP (International Genomics of Alzheimer's Project) SNP rs8093731 in Desmoglein-2/DSG2; underexpressed in the entorhinal cortex, hippocampus, and frontal and temporal cortex of AD cases [28] SNPs rs1116345 and rs34248 [144] c.1397A > G, p. Asp466Gly, exon 8 in subcortical heterotopias [34] SNPs rs2049161and rs16946051 associated with cognitive flexibility in ADHD [27] SNP rs12455524 in recurrent MDD [134] SAPAP1 De novo CNV deletion [24] Identified as an ASD-associated gene in a genome-wide network analysis [133] Two SNPs located within an intron of SAPAP1 [25] N/A N/A N/A N/A Rare CNVs (62 kb duplication) [26] N/A N/A  Although SAPAP2 has been identified as a promising candidate gene of ASD, the potential role of SAPAP2 in ASD is still unclear. A study using a murine model demonstrated that the deletion of SAPAP2 leads to increased social interaction and aggressive behavior, along with impaired initial reverse learning and synaptic function [14], indicating that SAPAP2 somehow modulates cognition. Correspondingly, a recent cross-species study not only revealed the role of SAPAP2 in age-related working memory decline, but also established an association between SAPAP2 and AD phenotypes at multiple levels of analysis [22], highlighting the possible importance of epigenetic dysregulation of SAPAP2 expression in the pathophysiology of AD. As noted in the above studies, CDK5-one of the key kinases in AD-is involved in triggering amyloid-beta (Aβ)-induced ubiquitination and degradation of SAPAPs [70], while amyloid precursor protein (APP) is one of binding partners of Homer, which is indirectly linked to SAPAPs via SHANKs [7]. Gene overexpression can arise from de novo CNV duplication, and several autism risk genes in duplicated loci-such as SAPAP2, SHANK3, and Neuroligins 1 and 3-are all FMRP targets [29,117,121,138,140], suggesting a promising link between dysfunction of FMRP in regulating the expression of synaptic proteins and the development of ASD.

SAPAP3 and Neuropsychiatric Disorders
After the initial report of the features of Sapap3-mutant mice for OCD-like repetitive self-grooming behaviors [13], a higher frequency of rare nonsynonymous coding variants located in SAPAP3 was found in OCD patients, providing the first clinical evidence and tentatively supporting the link between SAPAP3 and OCD [152]. Several follow-up human genetic studies and other secondary analyses of GWAS have also noted that SAPAP3 is a promising functional candidate gene for OCD and obsessive-compulsive spectrum disorders, such as trichotillomania (TTM) and Tourette syndrome (TS) [31][32][33]152], although some of the mutations identified in the above studies are more commonly found in association with grooming disorders and TTM than with OCD. Convergent evidence from functional imaging and neuropsychological studies has linked OCD symptoms to dysfunction of striatum-based circuitry, especially highlighting cortico-striato-thalamo-cortical circuitry [5,31,154]. Owing to their robust behavioral similarities and sophisticated neural circuitry shared with OCD and related disorders [13,82], Sapap3-mutant mice are now considered to be a well-established animal model for OCD and related disorders. Interestingly, a CNV within the SAPAP3 gene was also identified in a study of about 200 individuals with ASD [30]-a spectrum of disorders that involve prominent compulsive-like repetitive behaviors overlapped with the behavioral phenotypes of OCD.

SAPAP4 and Neuropsychiatric Disorders
In addition to the aforementioned possible roles in the actions of psychostimulants, BD, and fragile X syndrome, SAPAP4 is also associated with several other neuropsychiatric disorders, including ASD, early-onset cerebellar ataxia, and subcortical heterotopias [23,34,153]. An SNP located at the 20q11.21-q13.12 locus that encompasses the SAPAP4 gene has also been linked with a potential role in the development of ASD [153]. SAPAP4 is also implicated in early-onset cerebellar ataxia, according to an epigenetic study in which disruption of the SAPAP4 gene by the chromosome translocation t (8;20) caused monoallelic DNA hypermethylation of the truncated SAPAP4 promoter CpG island 820 bp downstream of the first untranslated exon of SAPAP4 isoform a and perturbed expression of the SAPAP4 gene [23]. A recent study has identified a de novo frameshift variation and a missense variant in SAPAP4 from two families with different types of subcortical heterotopias-which are cortical malformations associated with ID and epilepsy [34]-pinpointing a novel role of SAPAP4 in early cortical development. Although there is only an indirect link between SAPAP4 and mood disorders such as BD [127], 22q13 duplications spanning SHANK3 (a known interaction scaffold protein of SAPAP4) have been found in patients diagnosed with ADHD and BD, respectively [155], suggesting a possible role of SAPAP4 in hyperkinetic disorders and mood disorders.

Mutational Studies in Murine SAPAP Models
Because increasing evidence has pointed to a link between SAPAP family members and some aspects of the pathophysiology of neuropsychiatric disorders, characterizing how the perturbed synaptic functions of SAPAPs are involved informs behavioral phenotypes related to neuropsychiatric disorders in mutant mice, revealing part of what makes neuropsychiatric disorders' neurobiology.
To study the in vivo function of SAPAPs, mutant mice carrying Sapap1-4 deletions have been generated (Table 4). Sapap1-mutant mice exhibit normal spontaneous locomotion and normal reverse learning ability, but subtle decreases in sociability and impaired association of PSD-95 with SHANK3 complexes [15,156], while Sapap2-mutant mice display enhanced social interaction ability, excessive aggressive behaviors, and abnormal synaptic morphology and transmission in the OFC [14]. In contrast, for reverse learning, two independent mutant lines show opposing phenotypes [14,156]. A recent study also found that deletion of SAPAP2 in mice leads to reduced alcohol consumption [21]. Sapap1and -2-mutant mice all exhibit several forms of abnormal behavior relevant to schizophrenia, ASD, and cognitive disorders, such as deficits in learning, memory, and social behaviors.
Owing to the promising association between SAPAP3 and OCD, much more attention has been given to Sapap3-mutant mice-a well-established OCD-relevant murine model displaying repetitive self-grooming behavior, augmented anxiety, cognitive inflexibility, imbalances between goal-directed and habitual behavior, selective deficits in behavioral response inhibition, insensitivity to reward devaluation, altered valence processing, hypolocomotion, disrupted sleep patterns, normal preference motivation for sucrose, and Pavlovian learning [13,102,104,106,[157][158][159][160][161][162][163][164]. Convergent evidence from structural, biochemical, electrophysiological, and neural circuitry studies of Sapap3-mutant mice demonstrates and emphasizes the crucial role of SAPAP3 in corticostriatal synapses and striatum-based circuitry in OCD-like phenotypes. Surprisingly, neither cognitive inflexibility, augmented anxiety, nor aberrant habit formation are correlated with compulsive, repetitive behavior, implying that these different OCD-like behaviors in Sapap3-mutant mice probably involve complex or independent etiologies [157,159,160,162,163]. For example, the compromised PFC-striatal synaptic and circuitry function observed in Sapap3-mutant mice have been thought to be involved in their excessive compulsive and repetitive behavior [13,82,102].
Increased neural activity in prelimbic (PrL) and infralimbic (IL) regions of the medial PFC is associated with impaired reverse learning-usually representing cognitive inflexibility-in Sapap3-mutant mice [158]. Of particular interest, one study revealed an imbalanced cortical input to the central striatum in Sapap3-mutant mice, with more inputs from the secondary motor area (M2) and fewer inputs from the lateral OFC (LOFC) [103]. Notably, the role of SAPAP3 in the OFC appears to be highly heterogeneous.   [17,35] • Impulsive behaviors [17]  • Hyperactivity, reduced rearing behaviors [16] • ↑ Locomotion during the dark phase in home cages [16] • ↓ Self-grooming [16] • ↓ Working memory and spatial learning and memory [16] • ↓ Anxiety [16] • = General circadian activity. [16] Dlgap4   The LOFC is involved in both repetitive behaviors and reverse-learning deficits, but with disparate patterns of abnormal activity. Optogenetic stimulation of the LOFC's pyramidal neurons or LOFC-striatum circuitry reduces compulsive grooming in Sapap3mutant mice [102]. However, inhibition of LOFC GABAergic interneurons results in increased activity in the LOFC's pyramidal neurons and constitutes a direct pathway that leads to the impairment of reverse learning in Sapap3-mutant mice [106].
Three independent Sapap4-mutant mouse lines generated using different deletion strategies have been used to investigate the in vivo functional roles of SAPAP4 [16,17,34,35]. Of these, one mouse line has not been tested for all behaviors [34]; the other two show both shared and distinct defects at the synaptic and behavioral levels [16,17,35]. These two lines both show robust hyperactivity and memory deficits [16,17]. Whereas Dlgap4 geo/geo mice with an exon trap vector integrated in intron 7 of the Sapap4 gene show synaptic defects in the hippocampus and display impaired social interaction, diminished anxiety, and normal grooming behavior [16]-in which the social defects phenotype resembles the ASD-like behaviors observed in various Shank3-mutant mice [57]-mutant mice carrying Sapap4 deletions of exons 3-6 have profound synaptic defects in the NAc and exhibit robust hyperactivity, reduced depression-like behavior, altered sensitivity to amphetamine and cocaine, and reduced sensory responses, without deficits in anxiety and social interaction [17,35], and the behavioral phenotype of this line replicates the BD manic-like hyperkinetic behaviors seen in Shank3 transgenic mice [155]. Consistently, SHANK3 expression in the PSD is significantly upregulated in mice carrying Sapap4 deletions of exons 3-6 [17]. In addition, a recent study has shown that heterozygous mice with Sapap4 deletions (i.e., removal of exon 8) exhibit decreased size in several brain structures, with the hippocampus, corpus callosum, anterior commissure, thalamus, and cortex affected [34], suggesting the potential role of SAPAP4 during brain development. Given that the disruption of SAPAP4 in Dlgap4 geo/geo mice mainly affects the C-terminal region of SAPAP4 (including the binding site for SHANKs) and results in almost total loss of interaction with SHANKs, and that the ablation of exons 3-6 leads to the loss of the vast majority of the SAPAP4 protein (especially the N-terminal region), these two studies provide insights into how different domains or isoforms of the Sapap4 gene may elicit domain-or isoform-specific synaptic and behavioral defects and be involved in different neuropsychiatric disorders.

Conclusions
Emerging evidence indicates that SAPAPs are multifaceted postsynaptic scaffold proteins at excitatory synapses; they are believed to control excitatory synaptic formation and maturation, and play crucial roles in homeostatic plasticity by regulating the accumulation and turnover of glutamatergic receptors and scaffold proteins at synaptic sites [4,86], which may involve FMRP, ubiquitination, and actin cytoskeleton dynamics (Figure 2). Although evidence from direct association studies between neuropsychiatric disorders and SAPAPs is limited, there is ample convergent evidence linking dysfunction of SAPAPs to various cognitive deficits and strongly supportive of the important roles of SAPAPs in neuropsychiatric disorders. Findings from human genetic studies and mutant murine models are coalescing into a picture of the molecular networks that, when dysregulated or disrupted, may lead to synaptic dysfunction and, ultimately, be responsible for various neuropsychiatric disorders, such as OCD, ASD, ADHD, BD, schizophrenia, addiction, AD, and other cognitive disorders.
Undoubtedly, Sapap-mutant mice could be useful tools with which to dissect the effects of SAPAP disturbances on neuronal and circuitry function, and to decipher the neurobiological mechanisms underlying behavioral abnormalities seen in neuropsychiatric disorders, including genetically overlapped psychiatric disorders. Given that different SAPAP members have shared and distinct expression patterns in different developmental stages, brain regions, cell types, and circuitry, with differential regulation of gene expression at the transcript, translational, and epigenetic levels, as well as different interacting partners and signaling molecules and possible compensatory expression of other SAPAP members, it will be very important to clarify the physiological functions of the individual SAPAPs, along with how disrupted SAPAPs may result in molecular, synaptic, and circuitry defects and related abnormal behaviors with regard to neuropsychiatric disorders.
Therefore, future studies on Sapap-mutant mice should take into consideration the alternative splicing variants of the gene, the expression timing, the site, and the cell type where the gene mutation exerts its primary effects, eventually compromising the specific neural circuit. Importantly, the significance of the specific neural circuits to individual SAPAP proteins and behavioral abnormalities relevant to neuropsychiatric disorders still needs to be further explored by combining the manipulation of specific circuitry with conventional and conditional gene manipulation strategies.