Reciprocal signaling between translational control pathways and synaptic proteins in autism spectrum disorders

See allHide authors and affiliations

Science Signaling  28 Oct 2014:
Vol. 7, Issue 349, pp. re10
DOI: 10.1126/scisignal.2005832


Autism spectrum disorder (ASD) is a heterogeneous group of heritable neurodevelopmental disorders. Symptoms of ASD, which include deficits in social interaction skills, impaired communication ability, and ritualistic-like repetitive behaviors, appear in early childhood and continue throughout life. Genetic studies have revealed at least two clusters of genes frequently associated with ASD and intellectual disability: those encoding proteins involved in translational control and those encoding proteins involved in synaptic function. We hypothesize that mutations occurring in these two clusters of genes interfere with interconnected downstream signaling pathways in neuronal cells to cause ASD symptomatology. In this review, we discuss the monogenic forms of ASD caused by mutations in genes encoding for proteins that regulate translation and synaptic function. Specifically, we describe the function of these proteins, the intracellular signaling pathways that they regulate, and the current mouse models used to characterize the synaptic and behavioral features associated with their mutation. Finally, we summarize recent studies that have established a connection between mRNA translation and synaptic function in models of ASD and propose that dysregulation of one has a detrimental impact on the other.

Autism spectrum disorder (ASD) is a complex group of heterogeneous neurodevelopmental disorders categorized by three key behavioral abnormalities: restricted interests accompanied by repetitive behavior, deficits in language and communication skills, and inability to engage in reciprocal social interactions (13). These core symptoms often are comorbid with intellectual disability, epilepsy, motor impairment, anxiety, sleep disorder, attention-deficit hyperactivity disorder, and tics (46). The behavioral symptomatology of ASD encompasses a spectrum of wide-ranging phenotypes, which span from mild behavioral and personality traits to severe and debilitating impairments (7).

The remarkable clinical heterogeneity that characterizes ASD is paralleled by an equally multifaceted etiological heterogeneity. ASD is recognized to have a genetic component (twin concordance rate of 73 to 95%) and is extraordinarily heritable (>90%) (8, 9). Recently, linkage and association studies have identified numerous susceptibility genes located on multiple chromosomes, especially 2q, 7q, 15q, and the X chromosome. Thus, ASD is considered a polygenic disorder caused by the additive effect of multiple common genetic variants in combination with unidentified environmental factors (10). These forms of ASD are referred to as nonsyndromic ASD (11).

In contrast to nonsyndromic ASD, in about 10% of cases, ASD appears as part of a syndrome with a known genetic cause (1214). These monogenic forms of ASD can result from genomic DNA mutations, de novo copy number variants (CNVs), and chromosomal rearrangements (such as deletions and translocations) (11). Even when ASD is genetically defined, the clinical symptoms are highly heterogeneous, likely due to differences in the genetic background of the patient and epigenetic effects (11). Nevertheless, monogenic forms of ASD have been paramount for understanding key neurobiological processes and complex physiological pathways that when perturbed increase the risk for ASD.

Recent studies of monogenic forms of ASD have focused on at least two different clusters of genes frequently associated with ASD and intellectual disability: genes encoding for structural synaptic proteins or proteins involved in regulating protein synthesis (reviewed in (11, 15) (Figs. 1 and 2). These studies also suggest that mutations in the same clusters of genes may be pathogenic in nonsyndromic forms of ASD. We hypothesize that mutations in these two clusters of genes interfere with interconnected downstream signaling pathways in neuronal cells resulting in ASD symptomatology.

Fig. 1 Schematic of the hypothetical connection between protein synthesis and synaptic proteins.

Activation of group I mGluRs results in the activation of mTORC1 signaling, which increases protein synthesis. mTORC1 phosphorylates p70 S6 kinase 1 (S6K1) and 4E-BP2; phosphorylation of 4E-BP2 releases eIF4E and results in the association of eIF4E with eIF4G to form the active eIF4F (eIF4E-eIF4G-eIF4A) complex. eIF4F promotes the binding of mRNAs to ribosomes and recruits MNK, which phosphorylates eIF4E, and eIF4B, which is phosphorylated by S6K1. The eIF4F complex and the polyadenylated tail act synergistically with Mnk-dependent phosphorylation of eIF4E to stimulate cap-dependent translation initiation. Cap-dependent protein synthesis translates some mRNAs that encode for synaptic proteins located in the PSD. It is possible that mutations in genes encoding for proteins involved in the regulation of the mTORC1 pathway result in aberrant synthesis of synaptic proteins such as neuroligins, SHANK, SAPAP, NMDA receptors. The altered synthesis of these proteins would generate changes in molecular, structural, and synaptic plasticity, contributing to ASD pathophysiology and ASD-like behaviors. The protein products of genes associated with ASD are circled in bold. Credit: Heather McDonald/Science Signaling

Fig. 2 Schematic of the hypothetical connection between synaptic proteins and protein synthesis.

Intracellular signal transduction is initiated by the activation of neurotransmitter receptors that are organized with scaffolding proteins and adhesion molecules in the PSD. Receptor stimulation triggers the activation of intracellular signaling cascades including the mTORC1 and ERK pathways, which results in increased translation (see also Fig. 1). Given the importance of synaptic proteins in this type of signal transduction, mutations affecting genes encoding for these proteins could result in abnormal signaling that ultimately results in aberrant protein synthesis. The protein products of genes associated with ASD are circled in bold. Credit: Heather McDonald/Science Signaling

Here, we present an overview of the monogenic forms of ASD caused by mutations in these two clusters of genes, the molecular function of their protein products, and the current mouse models used to characterize the neurobiological features of these mutations. Altered dendritic morphology and synaptic pathophysiology in ASD mouse models have been intensely investigated and extensively described. However, for the most part, the relationship of these synaptic phenotypes to ASD-associated behaviors is still correlative, and thus, we do not discuss them in detail. Rather, we speculate and discuss the functional molecular signaling connections between the protein products of these two gene clusters in ASD (Figs. 1 and 2), which almost certainly affects synaptic and circuit pathophysiology.

ASD-Associated Genes Encoding Proteins That Regulate Translation

Fragile X mental retardation protein

Nearly all individuals with fragile X syndrome (FXS) have a trinucleotide (CGG) repeat expansion adjacent to the fragile X mental retardation 1 (FMR1) promoter that leads to the transcriptional silencing and subsequent loss of its protein product, fragile X mental retardation protein (FMRP) (16, 17). Recently, it was discovered that the silencing of FMR1 is mediated by the formation of a DNA-mRNA duplex between the promoter and the trinucleotide repeat region of the mRNA (18). Epidemiological studies show that FXS is the most common disorder that is associated with inherited intellectual disability and ASD (19, 20), occurring in about 1:5000 males and roughly half as many females. Affected males with FXS usually have other neurological and psychiatric conditions in addition to ASD and intellectual disability, including motor abnormalities, speech delay, hyperactivity, and anxiety. Postmortem neuropathological studies have revealed an increase in spine-like protrusions on apical and basal dendrites in the cerebral cortex of individuals with FXS (21, 22).

FMRP is an RNA-binding protein that is involved in many aspects of the posttranscriptional regulation of mRNA, such as stability, dendritic transport, and translational control. In particular, its function as a repressor of protein synthesis has been intensively studied, but the molecular mechanism responsible for this repression remains controversial. Experimental evidence indicates a role for FMRP in the initiation (2328) and elongation (2933) steps of protein synthesis. In the initiation model, FMRP inhibits translation initiation by interacting with the eukaryotic initiation factor 4E (eIF4E)–binding protein (4E-BP)–like protein CYFIP1, which is associated with eIF4E, the cap-binding translation factor for mRNAs (27). On the other hand, there is experimental evidence indicating that the function of FMRP as a translation repressor is at the level of the elongation (2933). Moreover, ribosomal run-off of these mRNAs demonstrated that FMRP is associated with mRNAs bound to stalled ribosomes (34), and ribosome transit assays indicate that elongation is enhanced in mice that lack FMRP (35). It is possible that FMRP acts by inhibiting both the initiation and elongation steps of translation that depends on mRNA identity and/or neuronal stimuli.

Fmr1 null mice exhibit high amounts of basal protein synthesis in the brain (36), and using high-throughput sequencing of RNA isolated by cross-linking immunoprecipitation (HITS-CLIP), at least 842 FMRP target mRNAs encoding for pre- and postsynaptic proteins have been identified (34). The postsynaptic proteins include Src homology 3 (SH3) and multiple ankyrin repeat domains protein 1 to 3 (SHANK1 to SHANK3), SAP90/PSD-95–associated protein 1 to 4 (SAPAP1 to SAPAP4), synaptic Ras guanosine triphosphatase (GTPase)–activating protein 1 (SynGAP1), and neuroligins; the presynaptic proteins include the neurexins, among others. These findings suggest that synaptic proteins and regulators of protein synthesis may functionally cooperate to generate the FXS phenotype (Figs. 1 and 2). In addition to directly repressing translation, FMRP affects protein synthesis by acting indirectly on signaling pathways involved in translational control. Increased signaling by the mammalian target of rapamycin complex 1 (mTORC1) is seen in Fmr1 null mice (37) that likely is induced by the increased abundance of the GTPase PIKE, which connects the activation of metabotropic glutamate receptor 5 (mGluR5) to the phosphatidylinositol 3-kinase (PI3K)–mTORC1 signaling pathway in the hippocampus (37, 38). Moreover, several mRNA targets of FMRP encode repressors of the mTORC1 signaling pathway, including tuberin (TSC2) and phosphatase and tensin homolog (PTEN) (34). Thus, it is possible that FMRP silencing may have an indirect, secondary effect on protein synthesis by repressing the translation of components of the mTORC1 signaling pathway.

Fmr1 null mice exhibit enhanced mGluR-dependent synaptic plasticity (39), increased density of dendritic spines, and numerous filopodia-like spines in the cortex, recapitulating a pathological feature observed in FXS patients (21, 4042). Moreover, Fmr1 null mice display a range of phenotypes that mimic many of the symptoms observed in individuals with FXS. For example, these mice display hyperactivity, altered sensorimotor gating (filtering out unnecessary stimuli), deficits in learning and memory, increased susceptibility to audiogenic seizures, increased body growth rate, and macroorchidism (43).

Eukaryotic initiation factor 4E

Several studies suggest an association between ASD and mutations in EIF4E. Genetic variants in chromosome 4q, which contains the EIF4E locus, have been described in patients with ASD (44, 45). Notably, in ASD subjects, several of these common genetic variants in the EIF4E gene are associated with a clinical phenotype characterized by repetitive and stereotyped behaviors, but not intellectual disability (46). A de novo chromosomal translocation involving the promoter region of the EIF4E gene in a boy with classic nonsyndromic ASD has been described (47). In addition, a nucleotide insertion in the promoter region of the EIF4E gene that increases promoter activity was discovered in two unrelated families with autistic siblings. These genetic studies link mutations in EIF4E to ASD; however, further investigations are needed to clearly establish a causal connection.

eIF4E binds to the cap structure at the 5′ terminus of mRNA and regulates the initiation step of cap-dependent translation (48, 49). The main role of eIF4E in translation initiation is in the formation of the eIF4F initiation complex, which brings mRNAs to the ribosome. The critical step in the formation of the complex is the direct association of eIF4E with eIF4G (50), an mRNA-ribosome bridging factor, and the indirect association of eIF4E with the RNA helicase eIF4A (51). The interaction of eIF4E with eIF4G is regulated by 4E-BPs, which repress translation by blocking the interaction of eIF4E with eIF4G (52). Upon stimulation, 4E-BP is phosphorylated and inactivated by mTORC1, thereby enabling eIF4E to associate with eIF4G to form the eIF4F complex (53). As well, eIF4E is phosphorylated by mitogen-activated protein kinase (MAPK)–interacting kinases 1 and 2 (Mnk1/2), a substrate of extracellular signal–regulated kinase (ERK). In some experimental conditions, the phosphorylation of eIF4E is correlated with the rate of protein synthesis (54). Thus, eIF4E and cap-dependent protein synthesis can be regulated by both mTORC1 and ERK signaling (55) (Figs. 1 and 2).

The relationship between eIF4E, cap-dependent translation, and ASD has been recently studied by overexpressing eIF4E in a transgenic mouse (56). These mice show increased protein synthesis in the brain and aberrant behaviors reminiscent of ASD, including impairments in social interactions and repetitive or perseverative behaviors. The ASD-like behaviors were corrected by blocking the interaction between eIF4E and eIF4G with the cap-dependent translation inhibitor 4EGI-1. Notably, mice with genetic deletion of 4E-BP2, the predominant 4E-BP isoform in the brain, exhibit ASD-like behaviors that mimic those displayed by eIF4E transgenic mice (57). Thus, mice with increased eIF4E-dependent translation display ASD-like behaviors, strongly suggesting a link between exaggerated protein synthesis and ASD.

TSC1 and TSC2

Tuberous sclerosis complex (TSC) is a multisystem disorder characterized by the presence of benign tumor-like lesions (hamartomas) in many organs, such as the brain, skin, eyes, kidneys, and heart (58). TSC is an autosomal dominant inherited disorder caused by loss-of-function mutations in either TSC1 (encoding hamartin, also referred to as TSC1) or TSC2 (encoding tuberin, also referred to as TSC2). These mutations include missense or nonsense mutations, insertions, and deletions involving nearly all exons in TSC1 and TSC2 (4, 5, 59). The impact of the different mutations on clinical phenotypes is extremely variable with respect to symptoms and disease severity, and in part is dependent on which TSC gene is affected (60). Seizures are the most common neurological symptom, occurring in up to 90% of the patients, whereas intellectual disability and ASD occur in about 50% of patients (58).

TSC1 and TSC2 form a heterodimeric complex that can regulate protein synthesis by controlling mTORC1 activity. TSC1 and TSC2 are phosphorylated by many kinases and factors, including Akt, ERK, glycogen synthase kinase–3β, adenosine monophosphate–activated kinase, and cyclin-dependent kinase 1 (6164). The active TSC1-TSC2 complex inhibits mTORC1 through activation of the small GTPase Ras homolog enriched in the brain (Rheb). Rheb activates mTORC1 when it is bound to guanosine triphosphate (GTP). The TSC1-TSC2 complex has GTPase activity localized in the GAP domain of TSC2. When phosphorylated by Akt, the GAP activity of TSC1/TSC2 is increased, which in turn hydrolyzes GTP bound to Rheb, thereby inhibiting mTORC1 (6164) (Fig. 1). Therefore, in the absence of either TSC1 or TSC2, high amounts of Rheb-GTP lead to constitutive activation of mTORC1 signaling, thereby resulting in dysregulated protein synthesis and cell growth (64). Several mouse models of TSC have been used to understand the etiology of this disorder. For example, heterozygous genetic deletion of either Tsc1 or Tsc2 in mice results in cognitive and synaptic impairments consistent with ASD (6567). Tsc1 and Tsc2 mutant mice display ASD-like phenotypes in the absence of neuropathological brain tumors and epilepsy, suggesting that cognitive dysfunction in TSC arises independently. However, it should be noted that specific genetic ablation of Tsc1 in either astrocytes (68) or neurons (66, 69) results in epilepsy and lethality. Either genetic reduction or complete deletion of Tsc1 in cerebellar Purkinje cells (PCs) results in ASD-like behaviors, including impaired social interaction, altered ultrasonic vocalizations, and increased repetitive behaviors that are correlated with decreased PC excitability and changes in the number and morphology of PCs (70).

In the aforementioned mouse models, postnatal and postdevelopment treatment with the mTORC1 inhibitor rapamycin ameliorates multiple behavioral and synaptic phenotypes (66, 6972). Thus, inhibition of mTORC1 activity in adult TSC mice is sufficient to correct ASD-like phenotypes, suggesting that these behaviors are caused by persistently increased mTORC1 signaling rather than irreversible pathophysiological changes that occur during brain development.

Phosphatase and tensin homolog

The gene encoding phosphate and tensin homolog (PTEN), located on chromosome 10q23, is a candidate risk gene for ASD and macrocephaly (7376). Different studies have suggested a causal role for PTEN mutations in a subset of individuals with ASD. Recently, a frameshift variant of PTEN was identified in a patient with extreme macrocephaly, ASD, intellectual disability, and epilepsy, supporting the theory that mutations in this gene are involved in the etiology of ASD and macrocephaly (77). In general, PTEN mutations are more frequent in ASD children that develop macrocephaly than in those that do not (78, 79).

PTEN is a phosphatase that removes the 3′ phosphate from phosphatidylinositol 3,4,5-trisphosphate (80) and thus inhibits PI3K signaling, thereby inactivating Akt and mTORC1. In contrast, deletion of PTEN results in a constitutively active Akt-mTORC1 signaling pathway (81) (Fig. 1). Given the importance of PI3K-Akt-mTORC1 signaling in the control of cell growth, survival, and proliferation, it is not surprising that PTEN inactivation leads to human cancers and neurological disorders (81).

Mice with Pten deletions have been studied mostly to clarify the role of PTEN in neuronal hypertrophy and number because the most obvious phenotype in human patients is macrocephaly. Overall, the effect of genetic deletion of Pten during development is dramatic, resulting in brain enlargement and gross anatomical abnormalities that are often accompanied by the development of seizures and premature death (82, 83). Several studies have directly addressed the role of PTEN mutations in ASD. These studies bypassed the severe developmental phenotype by deleting Pten in mice in either a cell-specific or inducible manner using conditional genetic technology. For example, mice in which Pten is ablated in a subset of postmitotic cortical and hippocampal neurons develop macrocephaly and display ASD-like behaviors, including impaired social interactions, seizures, anxiety, and cognitive deficits (84). Treatment with the mTORC1 inhibitor rapamycin reverses the neuronal hypertrophy and ameliorates the seizures and social impairments (85). Moreover, mice with germline Pten haploinsufficiency (Pten+/−) exhibit increased total brain mass and ASD-like behavioral impairments, such as abnormal social behavior and sensorimotor gating (86), increased repetitive behaviors, and depressive-like behaviors (87). These ASD-like behaviors are exacerbated in Pten+/− mice crossed with serotonin transporter heterozygote mice (Slc6a4+/−). SLC6A4 is also an ASD susceptibility gene (86). These findings demonstrate that deficiencies in PTEN and SLC6A4 can cooperate to contribute to ASD-like behavioral phenotypes.

ASD-Associated Genes Encoding Proteins Involved in Synaptic Function


Phelan-McDermid syndrome (PMS) is a genetic disorder characterized by ASD and intellectual disability. Patients exhibit impairments in communication skills that are often accompanied by reduced socialization and stereotypical movements, as well as aggressive behavior and seizures (8890). The disorder is caused by variable length deletions in the terminal region of the long arm of chromosome 22, which contains the ASD- and PMS-associated gene SHANK3 (9092). Duplications, CNVs, microdeletions, and mutations in SHANK3 are also found in patients with ASD and intellectual disability (9398).

SHANK3 is a member of the SH3 and multiple ankyrin repeat domains protein family, also known as proline-rich synapse-associated proteins (ProSAPs). The SHANK proteins are expressed abundantly in the central nervous system, are enriched in the postsynaptic density (PSD) of excitatory synapses (99101), and interact with cytoskeleton and scaffolding proteins, which in turn bind to receptors to create a matrix for the stabilization and organization of the PSD. Indeed, SHANK proteins bind to SAPAPs, which interact with PSD-95 proteins associated with glutamate receptors (102). Moreover, SHANK proteins bind to the Homer family of scaffolding proteins, which are associated with mGluRs (103) (Fig. 2). Also, SHANK proteins are involved in regulating the cytoskeleton by binding cortactin (104), inositol 1,4,5-trisphosphate (IP3) receptors, and F-actin (105, 106).

Multiple mouse models with deletions of the Shank genes have been intensively studied. Four different lines of Shank3 mice, each with a specific deletion of exons encoding for the functional interaction domains, exhibit behavioral deficits consistent with ASD, including social deficits, communication alterations, repetitive and stereotyped behaviors, and abnormal learning and memory that are accompanied generally by changes in synaptic function and molecular composition of the PSD (107109). Notably, Shank3B mutant mice, which lack the PDZ domain of the protein, exhibit a particularly severe phenotype. Consistent with the marked expression of Shank3 in the striatum, the Shank3B null mice groomed so excessively that they exhibited self-inflicted skin lesions and displayed anxiety-like behaviors and impaired social interactions. Genetic deletion of Shank1 results in abnormal grooming behavior and impairments in ultrasonic vocalization, but normal social interactions (110, 111), contextual fear, and long-term spatial memory (112). Overall, these PMS mouse models suggest that molecular changes perturbing synaptic and structural functions at the PSD of excitatory synapses are likely to generate ASD-like phenotypes.

Neuroligins and neurexins

Several mutations and deletions in genes encoding for neuroligin-3 (NLGN3), neuroligin-4 (NLGN4), and neurexin-1 (NRXN1) are associated with ASD and intellectual disability (44, 113125). A base pair substitution (A335G) in the NLGN4 promoter was observed in a boy with autism and intellectual disability. This mutation causes increased activity of promoter and, subsequently, increased NLGN4 expression (126), curiously suggesting that increased or decreased abundance of neuroligin-4 may be similarly detrimental to neuronal function and result in ASD-like phenotypes.

Neuroligins and neurexins are synaptic cell adhesion molecules that are critical for synaptic efficacy and plasticity (127132). Neurexins are type 1 membrane proteins encoded by three genes (NRXN1, NRXN2, and NRXN3), which generate larger α-neurexins and shorter β-neurexins from independent promoters (133). Furthermore, each gene undergoes extensive alternative splicing that is capable of generating thousands of neurexin isoforms (134). Neuroligins are endogenous ligands for neurexins (127) and are encoded by four genes (NLGN1, NLGN2, NLGN3, and NLGN4) located on the X chromosome (135). Neuroligins are type 1 membrane proteins like neurexins, but have a simpler domain structure and less diversity. All neuroligins are enriched in PSD, but neuroligin-1 and neuroligin-2 are exclusively localized to excitatory and inhibitory synapses, respectively, whereas neuroligin-3 may be present in both (130, 136138) (Fig. 2).

Mouse models recapitulating the genetic mutations or deletions of Nrxns and Nlgns described in ASD patients have been important for understanding their association with the clinical manifestation of the disorder. For example, mice with either a genetic deletion of Nlgn3 (139) or a knock-in allele containing an R451C substitution in Nlgn3 (140) display ASD-like behaviors that were mostly restricted to social and communication domains, such as impairments in ultrasonic vocalization, social interaction, and memory. Similarly, mice with a deletion of the Nlgn4 ortholog exhibited impaired social interactions and ultrasonic vocalization (141). This indicates that neuroligins are important in the generation of normal social skills and vocalization.

The studies performed on mice with genetic ablation of the genes that encode the neurexins are more difficult to interpret, given the high degree of genetic redundancy. Mice with a genetic deletion that results in a lack of all neurexin α-isoforms die 1 day after birth, whereas mice with ablation of a single gene are viable but show compromised respiratory function and die prematurely (128). It will be interesting to study the role of neurexins in a specific neuron type (or postdevelopmental time frame) to avoid the lethal phenotype and establish a link with ASD.


Rare genetic variants in DLGAP2, which encodes SAPAP2, occur in some ASD patients (142). Although not clearly established, a possible involvement of the proteins of the SAPAP family in ASD is intriguing, given their demonstrated interaction with the proteins of the SHANK family, which as discussed above are more clearly described in ASD (9496). However, the involvement of SAPAP3 in obsessive-compulsive spectrum disorders (OCD), trichotillomania, and Tourette syndrome is fairly well established (143146).

The members of the SAPAP family, also referred to as guanylate kinase–associated proteins (GKAPs), are postsynaptic scaffold proteins that are localized to the PSD uniquely at excitatory synapses (147). SAPAP proteins are encoded by a family of four genes that are widely, but differentially, expressed throughout the nervous system. SAPAP proteins may link the PSD-95 family proteins with the actin cytoskeleton by interacting with SHANK proteins, which in turn bind the actin-binding protein cortactin (Fig. 2). Therefore, in the current model of PSD organization, PSD-95-SAPAP-SHANK interactions play an important role in the constitution of the large postsynaptic signaling complex at glutamatergic synapses (148).

The most studied member of the SAPAP family is SAPAP3, which is highly expressed in the striatum (147, 149). Genetic ablation of Sapap3 caused behavioral abnormalities consisting of extremely frequent and aggressive self-grooming accompanied by self-inflicted snout lesions and anxiety-like behaviors. Consistently, Sapap3 null mice display synaptic, morphological, and molecular defects at striatal glutamatergic synapses. The behavioral and synaptic phenotypes of Sapap3 null mice are similar to those generated by genetic ablation of Shank3, indicating that genetic changes perturbing these synaptic proteins in the striatum result in specific phenotypes that are consistent with ASD.

Similar to SAPAP3, genetic ablation of SAPAP2 (DLGAP2) in mice increases aggressive behavior and results in impaired social interactions. Moreover, the DLGAP2 null mice exhibit reduced dendritic spine density, changes in receptor composition, and decreased PSD length and thickness (150). Overall, these results suggest that deletion of SAPAP2 may reduce synaptic and postsynaptic responses.


Deletion of SYNGAP1 or a mutation introducing a premature stop codon in SYNGAP1 is found in a few patients with intellectual disability with or without ASD (121, 142). SYNGAP1 encodes a Ras GTPase–activating protein (RasGAP) called synaptic GAP (SynGAP). SYNGAP1 has several alternative start sites, and transcripts can be spliced extensively to generate multiple SynGAP1 isoforms (151, 152). SynGAP1 is a brain-specific protein that is highly enriched at excitatory synapses and colocalizes and interacts with N-methyl-D-aspartate (NMDA) receptors and the PDZ domains of PSD-95 via its C-terminal amino acids (151, 152) (Fig. 2). It inhibits signaling pathways linked to NMDA receptor–mediated synaptic plasticity and AMPA (α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid) receptor membrane insertion (153155). It was shown that SynGAP1 connects Ca2+ influx to activation of the ERK pathway downstream of NMDA receptors (156). Given the multiple isoforms and the possible high degree of redundancy, the impact of deleting SynGAP1 in neurons is not clear. In fact, deletion of SynGAP1 in hippocampal neurons in culture has been reported to both enhance (153, 155) and suppress (154) dendritic spine formation. Syngap1−/− but not Syngap+/− mice die within 2 days after birth (157). Heterozygous mice exhibit hyperactivity, diminished sensorimotor gating, and enhanced startle response. Moreover, they display a reduction in social memory, a tendency toward social isolation, enhanced ERK activation, and impaired hippocampal synaptic plasticity (156). Hippocampal neurons in these mice have an accelerated rate of glutamatergic synapse maturation, consequently disrupting the excitation/inhibition balance (158). These studies indicate that changes in synaptic maturation during the development result in enduring behavioral abnormalities.

Reciprocal Signaling Links Two Clusters of ASD Genes

The studies summarized above are consistent with at least two defined clusters of genes that are involved in ASD and intellectual disability. One cluster encodes for proteins that regulate protein synthesis, a fundamental process for long-lasting changes in synaptic strength and dendritic spine plasticity underlying cognition. The second cluster of genes produces proteins involved in the regulation of synaptic transmission and structure, which are important in the establishment and remodeling of neuronal networks. Currently, there is limited experimental evidence suggesting a direct interaction between the protein products of these two gene clusters. However, their critical and central biological functions strongly suggest that an anomaly in one of these pathways would almost necessarily perturb the other (Figs. 1 and 2).

Activity-dependent changes in PSD composition and/or structure represent molecular mechanisms that drive complex brain functions, including learning and memory. These long-term synaptic and structural changes are critically dependent on dendritic protein synthesis (Fig. 1). Indeed, it recently was shown that aberrant protein synthesis driven by overexpression of the cap-binding translation factor eIF4E causes synaptic impairments and ASD-like behaviors (56), indicating that exaggerated translation directly influences synaptic and structural plasticity. Consistent with this idea, a related study revealed that overexpression of neuroligins is likely responsible for the generation of certain ASD-like phenotypes in eIF4E transgenic mice (57). Moreover, the synaptic, structural, and behavioral abnormalities in mice with exaggerated eIF4E-dependent translation were corrected by reducing protein synthesis and/or diminishing the expression of neuroligins with small interfering RNAs (siRNAs) (56, 57).

The examination of FMRP-regulated target mRNAs—encoding both synaptic proteins and regulators of translation—also supports this idea and demonstrates that both pre- and postsynaptic proteins are part of the transcripts dysregulated in FXS (34), which include SHANK3, SynGAP1, neuroligin-3, and neurexin-1 (34) and SHANK1, SAPAP1, and SAPAP3 (159). This suggests that changes in synaptic and PSD proteins driven by dysregulated protein synthesis may contribute to enduring changes in synaptic plasticity, dendritic morphology, and ASD-like behavioral abnormalities. Another set of mRNA targets of FMRP are proteins directly involved in the regulation of translation, such as TSC2 and PTEN (34, 160), suggesting that synaptic proteins and regulators of mTORC1 activity may interact to give rise to the FXS phenotype.

Conversely, it is possible that the ASD-associated mutations that result in changes in the level and function of synaptic and PSD proteins alter protein synthesis and contribute to the generation of ASD (Fig. 2). Unfortunately, there is limited information regarding the activity of translational control pathways in human patients and mouse models of ASD caused by mutations in genes encoding for synaptic proteins, as discussed above. However, a recent study investigating mGluR signaling in mice with a genetic deletion of Fmr1 reveals a fundamental role of the PSD scaffolding protein Homer1a (36, 161, 162). Altered mGluR5-Homer interactions contribute to abnormal mGluR signaling, altered protein synthesis, and other ASD-like phenotypes in FXS model mice. Genetic deletion of Homer1a restores the normal mGluR5-Homer association and corrects several phenotypes in Fmr1 null mice, including enhanced global protein synthesis (161, 162). Although the effect of mGluR5-Homer interactions on protein synthesis is secondary to the direct role of FMRP in translation, this study indicates the possibility that alterations in synaptic proteins result in aberrant translational control. It is tempting to speculate that ASD linked to SHANK mutations is also associated with alterations in protein synthesis because SHANK directly interacts with Homer (103). Therefore, defects in synaptic protein function could result in aberrant protein synthesis, resulting in abnormal synaptic plasticity and ASD-like behaviors. Future studies are necessary to conclusively address this hypothesis.

Several lines of evidence indicate that loss-of-function mutations, deletions, and overexpression of synaptic and PSD proteins are detrimental and result in ASD-like behavioral phenotypes, as discussed above. This is in agreement with our hypothesis that increased expression of synaptic proteins generated by alteration in protein synthesis could trigger synaptic abnormalities and behaviors associated with ASD. In contrast, investigations concerning the proteins that regulate translation hint at a connection between exaggerated protein synthesis and ASD phenotypes in humans and animal models (56, 57, 163). However, inhibition of de novo protein synthesis impairs long-lasting plasticity and cognition [reviewed in (48, 164)] and likely contributes to cognitive deficits in TSC model mice (165). An intriguing possibility is that excessive translation contributes to aberrant behaviors associated with ASD, whereas insufficient translation contributes to impaired cognition associated with the intellectual disability that often accompanies ASD.

Thus, recent data support the hypothesis that proteins involved in the regulation of translation and synaptic function may be interconnected and act in concert to give rise to synaptic and behavioral aberrations associated with ASD. Future genetic studies are necessary to reveal the molecular players that link these two pathways and to understand whether it is possible to intervene therapeutically at the level of these molecular crossroads.


Funding: Research in the Klann laboratory is supported with funds from NIH grants NS034007 (E.K.), NS047384 (E.K.), and NS087112 (E.S.) and CDRMP-DoD award AR100216. Competing interests: The authors declare that they have no competing interests.
View Abstract

Navigate This Article