PSD95: a synaptic protein implicated in schizophrenia or autism?

Abstract

The molecular components of the postsynaptic density (PSD) in excitatory synapses of the brain are currently being investigated as one of the major etiologies of neurodevelopmental disorders such as schizophrenia (SCZ) and autism. Postsynaptic density protein-95 (PSD-95) is a major regulator of synaptic maturation by interacting, stabilizing and trafficking N-methyl-D-aspartic acid receptors (NMDARs) and α-amino-3-hydroxy-5- methyl-4-isox-azoleproprionic acid receptors (AMPARs) to the postsynaptic membrane. Recently, there has been overwhelming evidence that associates PSD-95 disruption with cognitive and learning deficits observed in SCZ and autism. For instance, recent genomic and sequencing studies of psychiatric patients highlight the aberrations at the PSD of glutamatergic synapses that include PSD-95 dysfunction. In animal studies, PSD-95 deficiency shows alterations in NMDA and AMPA-receptor composition and function in specific brain regions that may contribute to phenotypes observed in neuropsychiatric pathologies. In this review, we describe the role of PSD-95 as an essential scaffolding protein during synaptogenesis and neurodevelopment. More specifically, we discuss its interactions with NMDA receptor subunits that potentially affect glutamate transmission, and the formation of silent synapses during critical time points of neurodevelopment. Furthermore, we describe how PSD-95 may alter dendritic spine morphologies, thus regulating synaptic function that influences behavioral phenotypes in SCZ versus autism. Understanding the role of PSD-95 in the neuropathologies of SCZ and autism will give an insight of the cellular and molecular attributes in the disorders, thus providing treatment options in patients affected.

Introduction

Synaptic dysregulation of dendritic spines during neurodevelopment is becoming increasingly linked to neurological diseases. Schizophrenia (SCZ) and autism are highly prevalent disorders characterized by synaptic abnormalities that lead to social impairments and cognitive deficits in individuals (Hutsler and Zhang, 2010, Selemon and Goldman-Rakic, 1999). The on-set and symptomology of these maladies are currently being explored at the dendritic spines to gain a better understanding of the effects and potential treatment options. SCZ is a heterogeneous mental health disorder that affects 1.1% of the human population. Typically, the on-set of SCZ occurs during the adolescent age range and consists of positive, negative and cognitive symptoms. Positive and negative symptoms involve hallucinations/delusions and emotional blunting, respectively. The cognitive dysfunctions include impaired working memory, lack of executive functions and attention deficits. In contrast, autism is a neurodevelopmental disorder that affects on average 1 in 68 of the childhood population and is considered within the autism spectrum disorder (ASD) (Toro et al., 2010). Autistic patients experience symptoms that include behavioral abnormalities such as repetition, reduced vocal communication, and aberrant social interactions. The prevailing theories of SCZ and autism etiologies suggest that a disruption of the synapse during neurodevelopment will cause impairments in synaptic plasticity and synaptic processing that are characteristic of these disorders.

The postsynaptic density (PSD) is a dense localized area within dendritic spines of excitatory synapses and is comprised of receptors, kinases, structural proteins and signaling molecules associated with synaptic plasticity. Perhaps the most abundant protein of the PSD is postsynaptic density protein-95 (PSD-95) (Cheng, 2006, Cho et al., 1992), a member of the membrane-associated guanylate kinase family (MAGUK), a scaffolding protein located at excitatory synapses and is involved in the stabilization, recruitment and trafficking of N-methyl-D-aspartic acid receptors (NMDARs) and α-amino-3-hydroxy-5-methyl-4- isox-azoleproprionic acid receptors (AMPARs) to the postsynaptic membrane (Chen et al., 2000, Kornau et al., 1995). PSD-95 is an essential component involved in glutamatergic transmission, synaptic plasticity, and dendritic spine morphogenesis during neurodevelopment (Funke et al., 2005, Gilman et al., 2011, Kim, 2004). Therefore, it is plausible that PSD-95 dysfunction during development may alter synaptic plastic events at the dendritic spines that contribute to the malformations of the synapse-associated with neurological disorders.

There is overwhelming evidence from human and animal studies that suggest PSD-95 disruption is linked to the neuropathologies of SCZ and autism. Specifically, a variety of sequencing techniques and analytics were used to identify PSD-95 mutations in SCZ and autism patients (Table 1). For instance, exome sequencing studies of SCZ patients show disrupted mutations of proteins located in the excitatory synapses of the PSD such as NMDAR and PSD-95 (Fromer, 2014, Purcell and Duncan, 2014). Further evidence reveals a significant decrease in PSD-95 mRNA and protein expression levels in the dorsolateral and dorsomedial prefrontal cortex of schizophrenic postmortem patients (Catts, 2015), suggesting an association between PSD-95 dysfunction and SCZ. PSD-95 has also been shown to be involved in a network of interactions with high-risk ASD genes that include SHANK, HOMER, neuroligin, and FMR1 (De Rubeis et al., 2014, Gilman, Iossifov, 2011, Tsai, 2012). Additionally, in mouse studies, deletion of the DLG4 gene (encodes PSD-95) causes behavioral abnormalities such as an increase in repetitive behavior, decreased vocalization, and irregular social interactions consistent with phenotypes observed in ASD patients (Feyder, 2010). “Furthermore, in a recent genetic study, DLG4 was identified as a candidate gene disrupted in intellectual disability (ID) (Lelieveld et al., 2016) (Table 1), a cognitive and mental disorder characterized by a reduction of dendritic spines. Interestingly, PSD-95 has direct interactions with ID-related proteins within the excitatory PSD that include Arc and Interleukin-1-receptor accessory protein-like 1 (IL1RAPL1), responsible for regulating spine density and function (Fernández et al., 2017, Pavlowsky et al., 2010, Valnegri, 2011). Therefore, PSD-95 deficiencies could attribute to the loss of spines and cognitive impairments associated with the ID. In this review, we provide an in-depth discussion of the role of PSD-95 in glutamatergic transmission and speculate on its implications in SCZ and autism. Moreover, we outline evidence that illustrates the effects of PSD-95 dysfunction and NMDAR regulation that include the formation of silent synapses. Lastly, we propose potential treatment options that restore PSD-95 within dendritic spines as a therapeutic option for SCZ and autism.

Table 1.

Mutations identified in the DLG4 gene in schizophrenia, autism and intellectual disability.

Mutation	Disorder	Methods	Reference
DLG4; Singleton	Schizophrenia	Exome sequence analysis	(Purcell and Duncan, 2014)
DLG4; de novo copy number variation (CNV)	Autism	NETBAG analysis	(Gilman et al., 2011)
DLG4-G241S	Autism	Ion PGM platform sequencing	(Xing et al., 2016)
DLG4; de novo	Autism	smMIP screening	(Stessman et al., 2017)
DLG4; de novo	Intellectual disorder	Meta- analysis	(Lelieveld et al., 2016)

PSD-95: a major component of the neurodevelopment of excitatory synapses

PSD-95 (also known as SAP90, synapse-associated protein 90) is encoded by the DLG4 (discs large homolog 4) gene in humans and is a major member of the MAGUK family. It functions as a scaffolding protein at excitatory synapses of the PSD. PSD-95 consists of three N-terminal PDZ domains (PSD-95, dlg, and zonula occludens-1), src homology domain (SH3), and a catalytically inactive guanylate cyclase domain (GUK) (Cho, Hunt, 1992, Kim, 2004). PSD-95 binds via PDZ domains directly to carboxy-terminal tails of NMDA receptor subunits, NR2A and NR2B, and to the AMPA receptor accessory proteins through stargazin/TARPs (Zhang, 2013) (Figure 1). Moreover, PSD-95 is a major component of a large network of proteins within PSD including ion channels, receptors, adhesion proteins, scaffolding proteins, and signaling molecules that influence glutamatergic transmission. For instance, PSD-95 has direct interactions with K⁺ channels, neuroligin, and nNOS; and indirect interactions with mGluR1/5 via GKAP (Brenman et al., 1996, Irie et al., 1997, Kim et al., 1997, Kim and Niethammer, 1995) (Table 2). Additional members of the MAGUK family include SAP102, SAP97, and PSD-93: and are also responsible for the recruitment and stabilization of NMDA and AMPA receptors.

Figure 1.

PSD-95 interactions in the PSD. An illustration describing the molecular organization of the postsynaptic density (PSD) located in the dendritic spine of glutamatergic synapses. Postsynaptic density-95 (PSD-95) contains direct and indirect interactions with many macromolecules at the PSD. PDZ1 domains of PSD-95 bind directly to N-methyl-D-aspartic acid receptors (NMDARs), more specifically, NR2-containing NMDA-receptors. PSD-95 also interacts with ErbB4, neuroligin, and nNos. PSD-95 has indirect interactions with α-amino-3-hydroxy-5-methyl-4- isox-azoleproprionic acid receptors (AMPARs) via stargazin. Other indirect interactions include group 1/5 mGluRs via GKAP, shank, and homer; and actin polymers via GKAP and cortactin. nNos, Neuronal nitric oxide synthase; GKAP, guanylate kinase-associated protein; mGluR, metabotropic glutamate receptors.

Table 2.

PSD-95 interacting protein network.

Interacting protein	Interacting domain	Reference
NMDA receptor subunit NR2	PDZ domains (1-2)	(Kornau et al., 1995)
Stargazin (Interacts with AMPA receptor subunits GluR1, GluR2, and GluR4)	PDZ domain (1)	(Chen et al., 2000)
K+ channel	PDZ domains (1-2)	(Kim and Niethammer, 1995)
Neuroligin	PDZ domain (3)	(Irie et al., 1997)
GKAP (Interacts with mGluR1/5 via SHANK-Homer; and F-actin via SHANK)	GK domain	(Kim et al., 1997)
nNos	PDZ domain (2)	(Brenman et al., 1996)
ErB4	PDZ domains (1-2)	(Garcia et al., 2000)
SPAR	GK	(Pak and Sheng, 2003)
Kalirin	PDZ domains (1-3)	(Penzes et al., 2001)
SynGAP	PDZ domains (1-3)	(Kim et al., 1998)
AKAP	SH3 and GK domains	(Colledge, 2000)
Src family: Src, lyn and Yes	PDZ domains (1-3) and SH3 domain	(Kalia and Salter, 2003)
IRSp53	PDZ domain (2)	(Soltau, 2004)
CRIPT	PDZ domain (3)	(Passafaro et al., 1999)
Arc	PDZ domains (1-2)	(Fernández et al., 2017)
IL1RAPL1	PDZ domains (1-2)	(Pavlowsky et al., 2010)

PSD-95 has long been associated with synaptic plasticity of glutamatergic synapses during neurodevelopment due to its interaction and functional implications of NMDA and AMPA receptors. Synaptic plastic processes such as long-term potentiation (LTP) and long-term depression (LTD) are heavily involved in synaptic maturation of the dendritic spine; therefore, alterations of the PSD may compromise the integral process of spine formation leading to neuropathologies of the synapse. Thus, PSD-95 may act as a key component involved in regulating synaptic strength by controlling spine formation and/or spine elimination/pruning.

PSD-95 mediates NMDA receptor clustering and function

The PSD site for excitatory glutamatergic transmission is mainly composed of glutamatergic receptors, including both NMDA and AMPA receptors. NMDA receptors are essential for synaptic plasticity and cortical development, and functional processes such as learning and working memory (Collingridge, 2013, Dumas, 2005). NMDA receptors consist of a hetero-tetrameric complex that contains an obligatory homodimer of NR1 and homodimers or heterodimers of either NR2A-D or NR3A-D subunits. NR2-containing subunits are involved in mediating calcium (Ca⁺) influx at the postsynaptic membrane; however, the open probability and duration of Ca⁺ flux are subunit specific. For instance, NR2B-containing NMDA receptors have slower kinetics and a slower decay time compared to NR2A-containing NMDA receptors, thus resulting in a larger flow of Ca⁺ within the synapse. PSD-95 influences NMDA receptor transmission via direct interaction and stabilization of specific NR2-containing NMDA receptors to the postsynaptic membrane. PSD-95 RNAi knockdown increases NR2B clustering at the postsynaptic synapse in cultured hippocampal neurons (Bustos, 2014). Additionally, in the hippocampus of a PSD-95 knockout mouse model, a reported increase in NMDAR decay was shown, suggesting a high presence of NR2B-containing NMDA receptors, thus corroborating previous findings (Beique, 2006). NR2B is essential for synaptic maturation during development (Monaco et al., 2015) and cognitive processes within the adult (Wang et al., 2013). However, an overabundance may be hazardous due to the significant increase in Ca⁺ conductance that could lead to excitotoxicity and neuronal death (Hardingham, 2006, Monaco, Gulchina, 2015). Therefore, a downregulation of PSD-95 may subsequently result in a substantial increase in Ca⁺ influx. Ca⁺ excitotoxicity within specific areas may impair tissue function, leading to neuropathologies depending on the brain region.

PSD-95: a regulator of NMDA receptor development

Glutamatergic receptor composition varies at the PSD of excitatory synapses during neurodevelopment. For instance, PSD scaffolding proteins – PSD-95 and SAP102 – are involved in NMDA receptor subunit localization and stabilization at the postsynaptic membrane that indeed facilitates the maturation of the PSD. Within the brain, PSD-95 expression levels show an increase from early life to adulthood and become more stable within the adult animal (Glantz, 2007, Gray, 2006), in contrast with SAP102 expression levels decreasing with age. Concomitantly, there is an NR2B-to NR2A-subunit switch that occurs during early development in most brain regions, facilitating synaptic maturation (Dumas, 2005). NR2B protein expression levels are highly abundant during early development and decline into adulthood; however, NR2A levels begin low but soon rise to peak along the maturation of associate learning. This would suggest that SAP102/NR2B-NMDAR complexes are switched by PSD-95/NR2A-NMDAR complexes during neurodevelopment, further corroborating PSD-95 as a molecular mediator of NMDA receptor development within the PSD. However, the development of prefrontal cortex is known to be protracted until young adult. Physiologically, NR2B-containing NMDA receptors also play a dominant role in synaptic plasticity within the adult prefrontal cortex, responsible for working memory function and cognitive performances (Monaco, Gulchina, 2015, Wang, 2008). NR2B receptors conduct a large amount of Ca⁺ and Na⁺ due to the prolonged open channel state, thereby prolonging depolarization. An overabundance of NR2B at the synapse is thus becoming a risk factor for excitatory synapses during the delayed prefrontal cortical development (Monaco, Gulchina, 2015). Therefore, a PSD-95/SAP102 expression level, as well as an NR2B/NR2A composition at the postsynaptic membrane during development, is extremely critical for normal synaptic maturation, especially for the prefrontal cortex. Thus, a disruption of PSD-95 within the PSD may lead to abnormal glutamatergic transmission due to a shift in NMDA receptor subunits and properties. For instance, a reduction in PSD-95 expression will result in a dramatic increase in NR2B-containing NMDAR presence that may induce hyperexcitability at the synapses, leading to cell death or neuronal damage.

PSD-95 modulates synaptic plasticity via AMPA receptors

AMPA receptors, composed of GluR1-GluR4 subunits, are heavily involved in synaptogenesis and essential for mediating synaptic strength. PSD-95 binds directly to stargazin/TARPs, responsible for interacting with AMPA receptor subunit, GluR1 (Schnell, 2002); and thus is shown to be a key regulator of synaptic plasticity and glutamatergic transmission (Bustos, 2014). Corroborating evidence describes that there is a significant reduction of AMPAR-mediated current and GluR1 protein expression levels in the hippocampus of PSD-95^−/− mice (Beique, 2006), suggesting that a downregulation of PSD-95 leads to synaptic depression. Furthermore, overexpressing PSD-95 causes an increase in AMPA-receptor presence and an increase in dendritic spine number and density, thereby inducing synaptogenesis (El-Husseini et al., 2000). These results suggest that PSD-95 acts as a regulator of synaptic plastic processes such as LTP and/or LTD via AMPA receptor scaffolding; thus determining the dendritic spine size, shape and number for synaptic transmission. Anomalous spine formation and elimination during neurodevelopment are associated with neuropsychiatric disorders; for instance, schizophrenic patients display a reduction of spine numbers, in contrast to autism patients that exhibit an increase in spines (Penzes et al., 2011).

Another major finding is PSD-95 downregulation leads to “silent synapse” formation (Beique, 2006, Huang, 2015). Silent synapses, also described as immature synapses, can be characterized by a high abundance of NMDA receptors and an absence of AMPA receptors at the postsynaptic membrane, thereby attenuating glutamatergic transmission. The lack of AMPA receptors at the synapse is mainly attributed to a reduction of PSD-95; however, there is also an increase in NR2B clustering. If PSD-95 is involved in recruiting NMDA receptors, then how will a lack of PSD-95 cause an upregulation in NR2B-NMDA containing receptors? Investigators proposed a compensatory increase of SAP-102 and PSD-93 expression levels to explain this phenomenon. Triple knockdown studies of PSD-95/SAP-102/PSD-93 show an attenuation of NMDAR-mediated current, in corroborating with this theory (Chen, 2015); but how exactly each of the three proteins contributes to the alteration of NMDA receptor subunits in the synapses remains to be determined. Nonetheless, PSD-95 deficiency alone leads to silent synapse formation and could be associated with the neuropathologies observed in psychiatric disorders.

PSD-95: genetic implications in SCZ

SCZ is a chronic mental disorder that manifests in adolescence or early adulthood due to changes in molecular and biochemical events that occur during development (Cash-Padgett and Jaaro-Peled, 2013). Patients display an array of clinical symptoms that consist of positive symptoms, negative symptoms, and cognitive impairments that include hallucinations/delusions, emotional blunting, and lack of executive function, respectively. Environmental and genetic factors during development, such as infection, drug use, parental age, prenatal and early postnatal or childhood stress have all been linked to the development of SCZ. Most genes identified as affected by these factors and dysregulated in SCZ are involved in neurodevelopment, neuronal growth and migration. Investigators have identified multiple high-risk genes and protein complexes associated with NMDA dysfunction and SCZ (i.e., neuregulin, dysbindin, and neuroligin) (Schizophrenia Working Group of the Psychiatric Genomics, 2014, Soares et al., 2011). Recently, exome sequencing studies of SCZ patients show disrupted mutations of proteins located at excitatory synapses of the PSD that include PSD-95, SAP-102 and NMDA receptors (Fromer, 2014, Purcell and Duncan, 2014). Therefore, an anomalous interaction between PSD-95 and NMDA receptors could be responsible for the maladaptive changes observed in SCZ during neurodevelopment. For instance, PSD-95 responsible for the stabilization of NMDAR’s at the postsynaptic membrane may be severely compromised, thus, contributing to a malformed PSD and glutamatergic dysfunction. Interestingly, PSD-95 binds directly to high-risk SCZ proteins such as DISC1 and neuroligin, responsible for synapse formation, microtubule network dynamics, axonal elongation, and cell adhesion; respectively (Shinoda, 2007, Taya, 2007). Therefore, disruption of PSD-95 expression/function may influence protein complexes involved in maladaptive synapse formation leading to schizophrenic- like phenotypes.

PSD-95 & NMDAR dysregulation in SCZ

NMDA receptor hypofunction is highly associated with the pathophysiology of SCZ (Lau and Zukin, 2007, Snyder et al., 2013). Studies reveal disruptive mutations of NMDA receptor subunits in the prefrontal cortex, hippocampus, and thalamus in postmortem subjects with SCZ (Clinton et al., 2003, Kristiansen et al., 2006). Additionally, functional imaging studies show reduced activity in the dorsolateral prefrontal cortex (dlPFC) in patients. More specifically, single photon emission computed tomography (SPECT) display “hypofrontality” in SCZ patients (Amen et al., 2011). Furthermore, genetic studies report single-nucleotide polymorphisms (SNPs) and a reduction of NR1 protein and mRNA in the dorsolateral prefrontal cortex in postmortem subjects (Clinton, Haroutunian, 2003, Weickert et al., 2013), suggesting a decrease in translation and lack of NMDA receptor presence at the postsynaptic membrane. Accordingly, schizophrenic patients display a reduction in PSD-95 mRNA and protein expression levels in the dorsolateral and dorsomedial prefrontal cortex (Catts, 2015). This data localized PSD-95 downregulation in specific forebrain tissue that may attribute to cellular phenotypes as well as schizophrenic patient symptomology. Hypothetically, a downregulation of PSD-95 may cause an overall reduction in NMDA receptors presence and function at the postsynaptic membrane. However, evidence within the hippocampus shows that PSD-95 deficiency exhibits no changes in NMDAR receptor amplitude, and is likely attributed to compensation via SAP-102 and PSD-93 (Elias, 2006). Nevertheless, PSD-95 disruption may shift the balance of responsibilities, thus altering tissue function that contributes to abnormal behavior within psychiatric patients.

PSD-95 regulates spine density involved in SCZ

Normal synaptic formation during neurodevelopment is critical for proper synaptic function; therefore, key components within the PSD, such as PSD-95, and SHANK help regulate synaptic growth and sustain molecular organization at the synapse. More specifically, within the prefrontal cortex, PSD-95 plays an essential role in postsynaptic development and function. During development, PSD-95 protein expression levels increase until late adolescence and early adulthood, consistent with synaptic density growth in the human prefrontal cortex (Glantz, 2007). However, PSD-95 is highly susceptible in schizophrenic patients; and thus, may alter normal spine growth and synaptic physiology. Neuroanatomical evidence of SCZ patients shows structural aberrations in specific tissues; for instance, an overall decrease in brain volume and reduced cortical gray matter in forebrain tissue, such as the dorsolateral prefrontal cortex, superior temporal gyrus, and limbic areas (i.e., hippocampal formation, anterior cingulate cortex) were reported in SCZ patients. At the cellular level, reports show a reduction in neuronal number and dendritic spine densities in the hippocampus and dorsolateral prefrontal cortex (Glantz and Lewis, 2000, Kolluri et al., 2005), although this observation appears to be controversial. However, many studies corroborate a significant decrease in pyramidal dendritic spines within superficial layers of the prefrontal cortex (Glantz and Lewis, 2000, Selemon and Goldman-Rakic, 1999), which is due to a spine-turnover rate during development. For example, in normal human subjects, spine numbers increase prior to birth and into childhood, spines are then selectively eliminated during adolescence. Therefore, we presume that within SCZ patients, there is an increase in spine number elimination (due to exaggerated spine pruning) that occurs during adolescence, contributing to the emergence of schizophrenic- like symptoms (Figure 2A).

Figure 2.

Dendritic spine changes in schizophrenia. (A) Graph showing dendritic spine development in patients with schizophrenia (yellow) versus normal brains (black). (B) Hypothetical models for the decrease in number of dendritic spines in SCZ patients. PSD-95 mutations disrupt ErbB4-neuregulin signaling causing AMPAR’s to destabilize from the postsynaptic membrane and a loss of NMDAR’s. However, enhanced neuregulin-Erb4 signaling leads to NMDAR hypofunction and loss of spines. A disrupted or deficient amount of PSD-95 results in a loss of NMDAR’s (NR2A-containing) and AMPAR receptors present at the membrane.

In fact, the molecular mechanisms involved that promote synaptic elimination in SCZ patients remains elusive. We propose that a reduction in dendritic spine numbers within SCZ patients is attributed to a deficient and/or dysfunctional PSD-95 scaffolding protein at the synapse (Figure 2B). Specifically, PSD-95 regulates dendritic spine numbers via synaptic plastic process through its interactions with NMDA- and AMPA-receptors. We suggest that when PSD-95 is disrupted, it will cause a decrease in the stabilization of NMDAR receptors — more specifically, NR2A-containing receptors — at the postsynaptic membrane, causing an attenuation of NMDAR signaling and a loss of spines. Moreover, fewer AMPA receptors will be recruited to the PSD during spine formation, thus severely inhibiting spinogenesis. Nevertheless, PSD-95 deficiency leads to silent synapse formation due to an increase in the NMDAR/AMPAR ratio at the synapse (Beique, 2006, Huang, 2015), and thus has been previously theorized to lead to synaptic elimination (Hanse, 2013). In addition, it is reported that a disruption in ErbB4 signaling, which is highly associated to SCZ, causes a destabilization of AMPA-receptors and loss of NMDA-receptors at the postsynaptic membrane (Li, 2007). Since PSD-95-ErB4 interactions are critical for ErB4 signaling (Barros CS, 2009), we suggest that a PSD-95 mutation would severely diminish the signaling pathway responsible for synaptic growth, consequently causing spine loss. Indeed, an altered neuregulin-1-ErB4 signaling contributes to NMDAR hypofunction observed in postmortem brains of SCZ patients (Hahn, 2006). We presume it is plausible that a reduction in PSD-95 within the frontal cortices of patients with schizophrenia will lead to a dramatic loss in dendritic spines, thus inducing schizophrenic- like symptoms that include cognitive deficits and working memory impairments. Consistently, a reduction in spine numbers will lead to glutamatergic attenuation and hypofrontality within the brain; thus, the overall reduction in neuronal activity in the dlPFC could explain the cognitive deficits and negative symptoms exhibited in patients with SCZ.

PSD-95: genetic implications in autism

Autism is a neurodevelopmental disorder characterized by behavioral abnormalities such as repetition, reduced vocal communication, and aberrant social interactions; therefore, can be described as a disorder of the synapse. Genomic screenings of autism patients display genetic mutations of synaptic proteins that include PSD-95 (Mariner et al., 1986, Risch et al., 1999). More recently, exome sequencing studies reveal PSD-95 as a high-risk gene involved in autism-spectrum disorders (De Rubeis, He, 2014). Furthermore, network-based analysis of genetic associations (NETBAG) identifies gene clusters such as dlg4, affected by de novo CNV events in autistic individuals (Gilman, Iossifov, 2011). In behavioral studies, PSD-95 knock-out mice exhibit similar phenotypes, consistent with autism spectrum disorders such as repetitive behavior (increased grooming) and decreased vocalization (Feyder, 2010). These studies suggest PSD-95 gene disruption may be associated with autism-related phenotypes observed in autistic patients. However, how a reduced PSD-95 expression in the brain results in an opposite change in dendritic spine, i.e., decrease in SCZ and increase in autism, remains an unsolved dilemma.

PSD-95: interaction with high-risk autism-related proteins

A large network of macromolecules present at the PSD has been implicated in autism, including SHANK, homer, FMRP (fragile x mental retardation protein), and neuroligin; interestingly, these molecules are described to have either direct or indirect interaction with PSD-95 that may alter synaptic structure and activity (Fujita-Jimbo et al., 2015, State, 2010, Tsai, 2012, Xing, Kimura, 2016). For example, PSD-95 is heavily involved in recruiting the cell adhesive molecule neuroligin to excitatory synapses during neurodevelopment. A PSD-95-neuroligin complex is critical for promoting neuronal connections between cells and inducing synaptic maturation (Prange et al., 2004); therefore, an overabundance of this complex could disrupt normal synaptic formation. PSD-95 is also associated with fragile X syndrome via FMRP mutations, consequently leading to a reduction in PSD-95 proteosomal degradation and a decrease in synaptic elimination (Todd et al., 2003, Zalfa et al., 2007). These findings illustrate that high-risk autism-related proteins may alter synaptic formation by regulating PSD-95 ubiquitination, synthesis, and location within the PSD.

PSD-95: alters spine density involved in autism

A major cellular phenotype identified in autism patients is dendritic spine aberrations; for instance, in humans, there is an increase in spine density on apical dendrites of pyramidal neurons in layer 2/3 of frontal, parietal, and temporal lobes. During development in normal subjects, spine pruning occurs as a part of synaptic maturation. However, in the autistic brain, dendritic spines undergo less pruning/elimination, and exhibit rapid growth and formation that occur during childhood (Figure 3A). For example, in the fragile X autism brain, there are increases in spine density that are associated with aberrant synaptic connectivity and plasticity (Irwin et al., 2001). Furthermore, autism is characterized by hyperconnectivity in local circuitry, although observations of hypoconnectivity are seen between brain regions (Penzes, Cahill, 2011). The hyperconnectivity is mainly attributed to an increase in spine density at the dendritic spines; thus, altering synaptic transmission leading to social deficits and cognitive impairments seen in patients.

Figure 3.

Dendritic spine changes in autism. (A) Graph showing dendritic spine development in autism patients (orange) versus normal brains (black). (B) Hypothetical models for the increase in number of dendritic spines within autism patients. The absence of FMRP causes a reduction in PSD-95 degradation and subsequent increase in AMPAR recruitment to the postsynaptic membrane. FMPR dephosphorylation causes an increase in PSD-95 translation via group 1 mGluR activation, thus increasing AMPAR recruitment to the postsynaptic membrane.

Scaffold proteins, such as SHANK and homer are heavily involved in anomalous maturation of the synapse; however, PSD-95 may also play a role. Indeed, previous studies show that PSD-95 overexpression causes an increase in AMPA-receptor presence and activity, and is accompanied with an increase in spine size and density in cultured hippocampal neurons (El-Husseini, Schnell, 2000). More specifically, PSD-95 overexpression induces LTP via recruitment of GluR1-containing AMPA receptors to synaptic membranes (Ehrlich, 2004). This finding suggests that an aberrant increase in spine density is consequently due to PSD-95 overabundance.

We propose hypothetical models that highlight the involvement of PSD-95 and dendritic spine growth (Figure 3B). Previous reports show FMRP dysregulation strongly influences the presence and function of PSD-95. For instance, in FMR1 knock-outs, a model for fragile x syndrome, there is an increase in total protein levels of PSD-95 that is consistent with increases in GluR1 protein levels at the synapse (Muddashetty et al., 2007, Todd, Mack, 2003). Indeed, FMRP was eloquently described as a major mediator in synaptic translation of receptors, kinases, scaffolding proteins via eukaryotic elongation factor 4F (eIF4F) (Richter, 2015). For example, FMRP dephosphorylation causes an increase in PSD-95 translation via mGluR1 dependent mediated pathways (Muddashetty, 2011, Nalavadi, 2012). It was further illustrated that PSD-95 ubiquitination is suppressed in the absence of FMRP, thus contributing to a reduction in synaptic elimination and subsequent enhanced synaptic formation (Tsai, 2012). These studies reveal that deactivation or loss of FMRP causes an increase in PSD-95 protein expression at the synapse and a dramatic increase in the recruitment of AMPA-receptors to the postsynaptic membrane. Therefore, we suggest that a disruption of FMRP utilizes PSD-95 scaffolding machinery to induce abnormal dendritic spine formation within the cortical tissue of autism patients.

Conclusions and future perspectives

The dendritic spine is the site of the PSD of excitatory synapses and is tightly regulated by synaptic proteins that influence glutamatergic transmission. However, within neurological disorders, there is a disruption of proteins within a PSD network that leads to the pathogenesis of SCZ and autism. Among these proteins, PSD-95 is involved in the recruitment and stabilization of glutamate receptors and is a major regulator of the maturation of glutamatergic synapses. An imbalance of PSD-95 expression will alter AMPA receptor composition and function, as well as NMDAR stabilization and kinetics, thus modifying glutamatergic transmission. PSD-95 also influences the size and density of dendritic spines during neurodevelopment that has overwhelming effects on synaptic connectivity and activity.

A major question remains, how is PSD-95 implicated in both SCZ and autism? For instance, sequencing studies show that PSD-95 gene mutations are associated with both SCZ and autism (Xing, Kimura, 2016). A plausible hypothesis would be that PSD-95 deficiency is associated with SCZ, whereas an overabundance of PSD-95 is linked to autism. However, this intriguing hypothesis remains to be tested. One way to investigate this theory is to establish a PSD-95 mutant animal model to assess the behavioral correlations with patients that exhibit psychiatric symptoms. Determining the behavioral effects of PSD-95 deficiency or overexpression and its association with SCZ and/or autism symptomology will provide greater insights from clinical aspects. Further, understanding the neuroanatomical and physiological effects of PSD-95 mutations can be used to compare to the established schizophrenic and autistic animal models such as neuregulin and SHANK3, respectively. These studies are important in understanding PSD-95 as a major regulator during neurodevelopment and synaptic transmission and associate these findings to neuropsychiatric disorders.

Highlights.

• PSD-95 mediates NMDA and AMPA receptor clustering and function

• Genetic implications of PSD-95 deficiency in schizophrenia

• PSD-95 & NMDAR dysregulation in schizophrenia

• PSD-95 regulates spine density involved in schizophrenia

• PSD-95 is associated with NMDAR dysregulation and spine change in autism