Abstract: Autism is a complex genetic disorder, but single-gene disorders with a high prevalence of autism offer insight into its pathogenesis. Recent evidence suggests that some molecular defects in autism may interfere with the mechanisms of synaptic protein synthesis. We propose that aberrant synaptic protein synthesis may represent one possible pathway leading to autistic phenotypes, including cognitive impairment and savant abilities. Autism is a complex genetic disorder, but single-gene disorders with a high prevalence of autism offer insight into its pathogenesis. Recent evidence suggests that some molecular defects in autism may interfere with the mechanisms of synaptic protein synthesis. We propose that aberrant synaptic protein synthesis may represent one possible pathway leading to autistic phenotypes, including cognitive impairment and savant abilities. In 1943, Kanner described 11 children affected by a unique disorder that he designated "early infantile autism" (Kanner, 1943Kanner L. Nervous Child. 1943; 2: 217-250Google Scholar). It has since become clear that autism is comprised of a clinically heterogeneous group of disorders, collectively termed "autism spectrum disorders" (ASDs), that share common features of impaired social relationships, impaired language and communication, and limited range of interests and behavior. Cognitive impairment is common in autism, and ∼70% of autistic individuals suffer from mental retardation (Fombonne, 1999Fombonne E. Psychol. Med. 1999; 29: 769-786Crossref PubMed Scopus (654) Google Scholar). Neuropathological studies have reported minor and inconsistent abnormalities in the autistic brain, but recent morphometric analysis has demonstrated enlargement of the hippocampus and amygdala (Schumann et al., 2004Schumann C.M. Hamstra J. Goodlin-Jones B.L. Lotspeich L.J. Kwon H. Buonocore M.H. Lammers C.R. Reiss A.L. Amaral D.G. J. Neurosci. 2004; 24: 6392-6401Crossref PubMed Scopus (566) Google Scholar). As first noted by Kanner, a high prevalence of macrocephaly is observed among children with ASDs, possibly due to an early period of excessive brain growth (Courchesne et al., 2004Courchesne E. Redcay E. Kennedy D.P. Curr. Opin. Neurol. 2004; 17: 489-496Crossref PubMed Scopus (166) Google Scholar). Remarkably, as many as 10% of autistic individuals paradoxically exhibit restricted domains of normal or even superior skills, termed "savant abilities," on a background of cognitive disability (Heaton and Wallace, 2004Heaton P. Wallace G.L. J. Child Psychol. Psychiatry. 2004; 45: 899-911Crossref PubMed Scopus (90) Google Scholar). Savant syndrome can involve excellence in a variety of cognitive or artistic domains, but declarative memory is most consistently accentuated. Autism is among the most heritable neuropsychiatric disorders, and available evidence points to a complex genetic basis (Persico and Bourgeron, 2006Persico A.M. Bourgeron T. Trends Neurosci. 2006; 29: 349-358Abstract Full Text Full Text PDF PubMed Scopus (410) Google Scholar). However, several disorders caused by single-gene mutations are associated with autism, often accompanied by cognitive impairment (Table 1). In these disorders, a substantial proportion of affected individuals meet diagnostic criteria for ASDs, reflecting a greatly increased risk of autism conferred by the mutation. Conversely, a small but significant percentage of children presenting with ASDs carry mutations in one of these genes. Although such monogenic disorders collectively account for a minority of cases of autism (10%–15%), the molecular alterations in these disorders may identify common pathogenic pathways shared by ASDs.Table 1Single-Gene Disorders with High Rates of AutismGeneDisorderRate of AutismRate in AutismMRGene FunctionFMR1Fragile X syndrome15%–30%2%–5%+Translational repressorTSC1/2Tuberous sclerosis complex25%–60%1%–4%+Inhibitor of mTORPTENPTEN hamartoma syndrome (ASD with macrocephaly)ND1%+Inhibitor of PI3K/mTOR signalingNF1Neurofibromatosis type I4%0%–4%+Ras GAPMECP2Rett's syndrome100%2%+Global transcriptional repressorUBE3AAngelman's syndrome40%1%+E3 ubiquitin ligaseCACNA1CTimothy's syndrome60%<1%+L-type voltage-gated calcium channel (Cav1.2)NLGN3/4Familial ASDND<1%+Synaptic adhesionNRXN1Familial ASDND<1%+Synaptic adhesionSHANK3Familial ASD (22q13 microdeletion syndrome)ND<1%+PSD scaffoldingSeveral monogenic human disorders are characterized by cognitive impairment and autism. The estimated rate of autism spectrum disorders (ASDs) in each disease and the estimated rate of each disease in children with ASDs are indicated (rate of autism and rate in autism, respectively). MR refers to the association of mental retardation with each disorder. ND indicates that the prevalence of ASDs among individuals carrying mutations in the specified gene has not been determined. Open table in a new tab Several monogenic human disorders are characterized by cognitive impairment and autism. The estimated rate of autism spectrum disorders (ASDs) in each disease and the estimated rate of each disease in children with ASDs are indicated (rate of autism and rate in autism, respectively). MR refers to the association of mental retardation with each disorder. ND indicates that the prevalence of ASDs among individuals carrying mutations in the specified gene has not been determined. The identification of mutations in neuroligins as rare genetic causes of autism suggested that defects in synaptic function may be intimately involved in autism pathogenesis (Zoghbi, 2003Zoghbi H.Y. Science. 2003; 302: 826-830Crossref PubMed Scopus (520) Google Scholar, Persico and Bourgeron, 2006Persico A.M. Bourgeron T. Trends Neurosci. 2006; 29: 349-358Abstract Full Text Full Text PDF PubMed Scopus (410) Google Scholar). Consistent with this view, mouse models of mutations that cause ASDs in humans consistently point to disrupted synaptic function: excessive or diminished excitatory synaptic connectivity (Chao et al., 2007Chao H.T. Zoghbi H.Y. Rosenmund C. Neuron. 2007; 56: 58-65Abstract Full Text Full Text PDF PubMed Scopus (346) Google Scholar, Hanson and Madison, 2007Hanson J.E. Madison D.V. J. Neurosci. 2007; 27: 4014-4018Crossref PubMed Scopus (82) Google Scholar) and alterations in the balance of excitation and inhibition (Dani et al., 2005Dani V.S. Chang Q. Maffei A. Turrigiano G.G. Jaenisch R. Nelson S.B. Proc. Natl. Acad. Sci. USA. 2005; 102: 12560-12565Crossref PubMed Scopus (464) Google Scholar, Tabuchi et al., 2007Tabuchi K. Blundell J. Etherton M.R. Hammer R.E. Liu X. Powell C.M. Sudhof T.C. Science. 2007; 318: 71-76Crossref PubMed Scopus (667) Google Scholar). It appears that the "autistic neuron" has too many or too few, too strong or too weak, excitatory synapses relative to the level of inhibition. However, mutations in ASDs that directly affect synaptic structure are rare. In this Essay, we focus on recent insights into the function of gene products mutated in several other autistic disorders. We propose that dysregulation of synaptic protein synthesis in these disorders may lead to altered synaptic development and plasticity and autistic phenotypes. The gene products mutated in several single-gene disorders associated with autism act as negative regulators of protein synthesis (Figure 1). Fragile X syndrome (FXS) is an X-linked form of mental retardation caused by transcriptional silencing of the FMR1 gene (Bagni and Greenough, 2005Bagni C. Greenough W.T. Nat. Rev. Neurosci. 2005; 6: 376-387Crossref PubMed Scopus (375) Google Scholar). The prevalence of ASDs in FXS is reportedly 15%–30%, and conversely, up to 5% of children presenting with ASDs are found to have FXS. The product of the FMR1 gene, the fragile X mental retardation protein (FMRP), binds to specific mRNAs and represses their translation. FMRP is estimated to interact with more than 400 distinct mRNAs (Brown et al., 2001Brown V. Jin P. Ceman S. Darnell J.C. O'Donnell W.T. Tenenbaum S.A. Jin X. Feng Y. Wilkinson K.D. Keene J.D. et al.Cell. 2001; 107: 477-487Abstract Full Text Full Text PDF PubMed Scopus (869) Google Scholar). Loss of FMRP expression in FXS would be expected to cause translational derepression of these target mRNAs. Indeed, there is a substantial (∼20%) and anatomically widespread increase in the rate of cerebral protein synthesis in the Fmr1 knockout mouse, which models the silencing of FMRP expression responsible for the human disorder (Qin et al., 2005Qin M. Kang J. Burlin T.V. Jiang C. Smith C.B. J. Neurosci. 2005; 25: 5087-5095Crossref PubMed Scopus (198) Google Scholar). Tuberous sclerosis complex (TSC) is an autosomal dominant disorder caused by mutations in hamartin (TSC1) or tuberin (TSC2) (Kwiatkowski and Manning, 2005Kwiatkowski D.J. Manning B.D. Hum. Mol. Genet. 2005; 14 Spec No. 2: R251-R258Crossref PubMed Scopus (316) Google Scholar). Central nervous system involvement in TSC is characterized clinically by a high prevalence of ASDs (25%–60%), cognitive impairment, and epilepsy (Wiznitzer, 2004Wiznitzer M. J. Child Neurol. 2004; 19: 675-679Crossref PubMed Scopus (147) Google Scholar). TSC1 and TSC2 form a heterodimeric complex with GAP activity against the small GTP-binding protein Rheb, resulting in inhibition of the mammalian target of rapamycin (mTOR) kinase (Figure 1). More specifically, TSC1/2 inhibit the activity of the rapamycin-sensitive mTOR-raptor complex (mTORC1). mTORC1, which is activated by a sequential kinase cascade downstream of phosphoinositide-3 (PI3) kinase, is a major regulator of cellular growth in mitotic cells (Wullschleger et al., 2006Wullschleger S. Loewith R. Hall M.N. Cell. 2006; 124: 471-484Abstract Full Text Full Text PDF PubMed Scopus (4363) Google Scholar). One of the principal effector mechanisms of mTORC1 is activation of cap-dependent translation. Recognition of the 5′ mRNA cap by eIF4E, which leads to recruitment of eIF4G and the small ribosomal subunit, is the key regulatory step in translation initiation (Richter and Sonenberg, 2005Richter J.D. Sonenberg N. Nature. 2005; 433: 477-480Crossref PubMed Scopus (719) Google Scholar). A family of eIF4E-binding proteins, 4E-BPs, impedes this process by sequestering eIF4E and blocking its association with eIF4G. Phosphorylation of 4E-BPs by mTORC1 relieves this inhibition, promoting eIF4E release and activation of cap-dependent initiation. Phosphorylation of another mTORC1 substrate, p70 ribosomal protein S6 kinase (S6K), also facilitates translation. Inactivation of TSC1/2 in hippocampal neurons upregulates mTORC1 activity (Tavazoie et al., 2005Tavazoie S.F. Alvarez V.A. Ridenour D.A. Kwiatkowski D.J. Sabatini B.L. Nat. Neurosci. 2005; 8: 1727-1734Crossref PubMed Scopus (364) Google Scholar, Meikle et al., 2007Meikle L. Talos D.M. Onda H. Pollizzi K. Rotenberg A. Sahin M. Jensen F.E. Kwiatkowski D.J. J. Neurosci. 2007; 27: 5546-5558Crossref PubMed Scopus (318) Google Scholar, Ehninger et al., 2008Ehninger D. Han S. Shilyansky C. Zhou Y. Li W. Kwiatkowski D.J. Ramesh V. Silva A.J. Nat. Med. 2008; 14: 843-848Crossref PubMed Scopus (637) Google Scholar), suggesting that loss of TSC1/2 function elicits enhanced translation in neurons. Loss-of-function mutations in another negative regulator of PI3K-mTOR signaling, the PTEN phosphatase, have also been linked to ASD pathogenesis. PTEN antagonizes PI3K-dependent signaling by converting the second messenger PIP3 to PIP2, and loss of PTEN function in neurons leads to heightened mTORC1 activity (Kwon et al., 2006Kwon C.H. Luikart B.W. Powell C.M. Zhou J. Matheny S.A. Zhang W. Li Y. Baker S.J. Parada L.F. Neuron. 2006; 50: 377-388Abstract Full Text Full Text PDF PubMed Scopus (686) Google Scholar). Inactivating mutations in PTEN are responsible for several related familial hamartoma-tumor syndromes, the clinical spectrum of which includes macrocephaly associated with ASDs. Macrocephaly occurs in up to 20% of ASD cases, and PTEN mutations have been identified in ∼5% of ASD patients with macrocephaly (e.g., Butler et al., 2005Butler M.G. Dasouki M.J. Zhou X.P. Talebizadeh Z. Brown M. Takahashi T.N. Miles J.H. Wang C.H. Stratton R. Pilarski R. Eng C. J. Med. Genet. 2005; 42: 318-321Crossref PubMed Scopus (546) Google Scholar). The connection between macrocephaly and PTEN mutations is interesting in view of the role of the PI3K-mTOR pathway in regulation of cell size, an effect thought to be mediated through translational control (Backman et al., 2002Backman S. Stambolic V. Mak T. Curr. Opin. Neurobiol. 2002; 12: 516-522Crossref PubMed Scopus (75) Google Scholar, Fingar et al., 2002Fingar D.C. Salama S. Tsou C. Harlow E. Blenis J. Genes Dev. 2002; 16: 1472-1487Crossref PubMed Scopus (802) Google Scholar). In mitotic cells, activation of this pathway increases cell size, whereas inhibition decreases cell size. It is not clear to what extent macrocephaly in ASD patients may be due to increased neuronal size; however, inactivation of either TSC1/2 or PTEN in mice causes neuronal hypertrophy and macrocephaly (Backman et al., 2002Backman S. Stambolic V. Mak T. Curr. Opin. Neurobiol. 2002; 12: 516-522Crossref PubMed Scopus (75) Google Scholar, Meikle et al., 2007Meikle L. Talos D.M. Onda H. Pollizzi K. Rotenberg A. Sahin M. Jensen F.E. Kwiatkowski D.J. J. Neurosci. 2007; 27: 5546-5558Crossref PubMed Scopus (318) Google Scholar). Taken together, the association of mutations in FMRP, TSC1/2, and PTEN with autism suggests that defects in translational repression may represent one possible mechanism leading to autistic phenotypes. It will therefore be of interest to determine whether defects in additional components of such pathways might contribute to autism susceptibility. As in all cells, elaborate mechanisms regulate protein synthesis in neurons to ensure adaptive responses to a changing environment (Kelleher et al., 2004bKelleher 3rd, R.J. Govindarajan A. Tonegawa S. Neuron. 2004; 44: 59-73Abstract Full Text Full Text PDF PubMed Scopus (459) Google Scholar, Klann and Dever, 2004Klann E. Dever T.E. Nat. Rev. Neurosci. 2004; 5: 931-942Crossref PubMed Scopus (324) Google Scholar, Sutton and Schuman, 2006Sutton M.A. Schuman E.M. Cell. 2006; 127: 49-58Abstract Full Text Full Text PDF PubMed Scopus (575) Google Scholar). There is abundant evidence for transcriptional regulation by neuronal activity and by downstream intracellular signals that are integrated in the neuronal cell body. However, patterns of activity at the thousands of synapses on each neuron dictate where and how this mRNA is used to synthesize proteins to fine-tune neuronal function. Markers of mRNA translation suggest that synaptic activity-induced protein synthesis occurs locally at or near dendritic spines in response to the synaptic release of glutamate. Two types of postsynaptic glutamate receptor have been implicated in translational regulation: the Gq-coupled metabotropic glutamate receptors 1 and 5 (mGluR1 and 5) and the calcium-permeable NMDA receptors (NMDARs). In addition to activating intracellular signaling pathways, there is evidence that NMDAR activation stimulates release of brain-derived neurotrophic factor (BDNF), which can induce neuronal protein synthesis through activation of synaptic TrkB receptors. Excitatory synaptic activity drives translation by activating these receptors, and this process is normally held in check by negative regulators such as FMRP, PTEN, and TSC1/2. Mutations that cause either overactivation of these receptors or decreased negative regulation of protein synthesis knock the system out of balance. We propose that a functional consequence is synaptic dysfunction that may manifest in humans as autism. Protein synthesis in the neuron has been extensively studied in the context of memory formation in adults and, more recently, experience-dependent cortical development. In both contexts, the immediate encoding and storage of information do not require protein synthesis. However, new gene expression is required for these changes to endure for longer than a few hours. Experimental forms of synaptic plasticity display similar temporal and molecular distinctions. Hippocampal long-term potentiation (LTP) and long-term depression (LTD) exhibit persistent late phases (L-LTP and L-LTD, respectively) that require new gene expression and transient early phases (E-LTP and E-LTD, respectively) that are insensitive to inhibitors of transcription and translation. Investigation of the molecular mechanisms mediating this process has highlighted the central importance of translational regulation. The presence of mRNAs and polyribosomes in proximity to synaptic sites suggested that local protein synthesis in dendrites could rapidly supply new gene products to activated synapses. The discoveries that BDNF-induced LTP and mGluR-dependent LTD, both of which require protein synthesis, can be supported by isolated dendrites further suggested a crucial role for translational upregulation in response to synaptic activity (Kang and Schuman, 1996Kang H. Schuman E.M. Science. 1996; 273: 1402-1406Crossref PubMed Scopus (725) Google Scholar, Huber et al., 2000Huber K.M. Kayser M.S. Bear M.F. Science. 2000; 288: 1254-1257Crossref PubMed Scopus (737) Google Scholar). Comparison of the effects of transcriptional and translational inhibitors, combined with metabolic labeling, has shown that establishment of L-LTP is mediated by rapid upregulation of the translation of pre-existing mRNAs (Kelleher et al., 2004aKelleher 3rd, R.J. Govindarajan A. Jung H.-Y. Kang H. Tonegawa S. Cell. 2004; 116: 467-469Abstract Full Text Full Text PDF PubMed Scopus (662) Google Scholar). Recent work from several groups has established that the ERK, MAPK, and mTOR signaling pathways couple synaptic activity to the translational machinery during both protein synthesis-dependent LTP and LTD (Figure 1). The ERK pathway, which is activated downstream of NMDAR, mGluR, and TrkB receptors, plays important roles in synaptic plasticity and memory (Sweatt, 2004Sweatt J.D. Curr. Opin. Neurobiol. 2004; 14: 311-317Crossref PubMed Scopus (785) Google Scholar). ERK activation is required for stimulation of cap-dependent translation and phosphorylation of eIF4E, 4E-BPs, and S6 in hippocampal neurons in response to L-LTP induction, NMDAR activation, and BDNF (Kelleher et al., 2004aKelleher 3rd, R.J. Govindarajan A. Jung H.-Y. Kang H. Tonegawa S. Cell. 2004; 116: 467-469Abstract Full Text Full Text PDF PubMed Scopus (662) Google Scholar). Targeted disruption of ERK signaling in the hippocampus prevents these translational responses and causes selective deficits in L-LTP and long-term memory. Inhibition of mTORC1 activity in hippocampal neurons attenuates synaptic activity-induced translation, 4E-BP phosphorylation, L-LTP, and BDNF-induced LTP (Tang et al., 2002Tang S.J. Reis G. Kang H. Gingras A.C. Sonenberg N. Schuman E.M. Proc. Natl. Acad. Sci. USA. 2002; 99: 467-472Crossref PubMed Scopus (576) Google Scholar, Kelleher et al., 2004aKelleher 3rd, R.J. Govindarajan A. Jung H.-Y. Kang H. Tonegawa S. Cell. 2004; 116: 467-469Abstract Full Text Full Text PDF PubMed Scopus (662) Google Scholar). Similarly, activation of mGluR1/5 leads to ERK- and mTOR-dependent increases in eIF4E, 4E-BP, and S6 phosphorylation, and inhibition of the ERK or mTOR pathways blocks mGluR-dependent LTD (Gallagher et al., 2004Gallagher S.M. Daly C.A. Bear M.F. Huber K.M. J. Neurosci. 2004; 24: 4859-4864Crossref PubMed Scopus (201) Google Scholar, Hou and Klann, 2004Hou L. Klann E. J. Neurosci. 2004; 24: 6352-6361Crossref PubMed Scopus (407) Google Scholar, Banko et al., 2006Banko J.L. Hou L. Poulin F. Sonenberg N. Klann E. J. Neurosci. 2006; 26: 2167-2173Crossref PubMed Scopus (194) Google Scholar). Collectively, these findings support a requirement for ERK- and mTOR-regulated translation in protein synthesis-dependent synaptic plasticity and memory. Analysis of synaptic plasticity in a mouse model of FXS illustrates how loss of the normal constraints on neuronal translation may give rise to synaptic and cognitive impairment. Fmr1 knockout mice display a range of phenotypes reminiscent of the human disorder. Interestingly, mGluR-dependent LTD, which requires translation but not transcription for its expression, is significantly enhanced in the hippocampus of Fmr1 knockout mice (Huber et al., 2002Huber K.M. Gallagher S.M. Warren S.T. Bear M.F. Proc. Natl. Acad. Sci. USA. 2002; 99: 7746-7750Crossref PubMed Scopus (1013) Google Scholar). Further supporting the view that increased translation of FMRP target mRNAs underlies this LTD phenotype, hippocampal mGluR-dependent LTD is rendered insensitive to translational inhibitors in Fmr1 knockout mice (Hou et al., 2006Hou L. Antion M.D. Hu D. Spencer C.M. Paylor R. Klann E. Neuron. 2006; 51: 441-454Abstract Full Text Full Text PDF PubMed Scopus (365) Google Scholar, Nosyreva and Huber, 2006Nosyreva E.D. Huber K.M. J. Neurophysiol. 2006; 95: 3291-3295Crossref PubMed Scopus (223) Google Scholar). Presumably, essential LTD proteins are constitutively overexpressed in the absence of FMRP. What might be the consequence of this manifestation of "hyperplasticity" in the hippocampus? Recently it was shown that inhibitory avoidance learning in rats induces LTP in the CA1 region of the hippocampus (Whitlock et al., 2006Whitlock J.R. Heynen A.J. Shuler M.G. Bear M.F. Science. 2006; 313: 1093-1097Crossref PubMed Scopus (1308) Google Scholar). In addition to its role in LTD, mGluR-dependent protein synthesis has also been implicated in the reversal of LTP (Zho et al., 2002Zho W.M. You J.L. Huang C.C. Hsu K.S. J. Neurosci. 2002; 22: 8838-8849Crossref PubMed Google Scholar). Thus, one might anticipate that a manifestation of exaggerated LTD would be impaired retention or increased extinction of inhibitory avoidance memory. This phenotype has been observed in Fmr1 knockout mice (Dolen et al., 2007Dolen G. Osterweil E. Rao B.S. Smith G.B. Auerbach B.D. Chattarji S. Bear M.F. Neuron. 2007; 56: 955-962Abstract Full Text Full Text PDF PubMed Scopus (732) Google Scholar). Thus, exaggerated mGluR-dependent protein synthesis in the absence of FMRP may yield specific patterns of cognitive impairment in adults with FXS. Cognitive impairment and autism may also reflect, at least in part, defects in synaptic development and/or developmental plasticity. Ocular dominance plasticity in the developing visual cortex, a commonly studied model for experience-dependent rearrangement and refinement of synaptic connectivity, requires both ERK activation and new protein synthesis, suggesting that similar translational mechanisms contribute to synaptic plasticity in the developing and adult brain (Maffei and Berardi, 2002Maffei L. Berardi N. Neuron. 2002; 34: 328-331Abstract Full Text Full Text PDF PubMed Scopus (2) Google Scholar, Taha and Stryker, 2002Taha S. Stryker M.P. Neuron. 2002; 34: 425-436Abstract Full Text Full Text PDF PubMed Scopus (81) Google Scholar). Studies of ocular dominance plasticity in the fragile X mouse model again revealed hyperplasticity—an exaggerated response to visual deprivation (Dolen et al., 2007Dolen G. Osterweil E. Rao B.S. Smith G.B. Auerbach B.D. Chattarji S. Bear M.F. Neuron. 2007; 56: 955-962Abstract Full Text Full Text PDF PubMed Scopus (732) Google Scholar). Neuronal activity-induced protein synthesis is crucial for long-lasting modifications of neural circuits. However, these findings in the mouse model of FXS suggest that impairments in synaptic and cognitive function can result from too much of what is normally a good thing. Further evidence that excessive translation can compromise cognitive function comes from mouse models deficient in PTEN and TSC1/2 function, in which elevated mTORC1 activity is associated with impaired hippocampal memory (Kwon et al., 2006Kwon C.H. Luikart B.W. Powell C.M. Zhou J. Matheny S.A. Zhang W. Li Y. Baker S.J. Parada L.F. Neuron. 2006; 50: 377-388Abstract Full Text Full Text PDF PubMed Scopus (686) Google Scholar, Goorden et al., 2007Goorden S.M. van Woerden G.M. van der Weerd L. Cheadle J.P. Elgersma Y. Ann. Neurol. 2007; 62: 648-655Crossref PubMed Scopus (199) Google Scholar, Ehninger et al., 2008Ehninger D. Han S. Shilyansky C. Zhou Y. Li W. Kwiatkowski D.J. Ramesh V. Silva A.J. Nat. Med. 2008; 14: 843-848Crossref PubMed Scopus (637) Google Scholar). To facilitate consolidation of synaptic modifications, newly synthesized proteins must be targeted specifically to active synapses. The phenomenon of synaptic tagging and capture suggests that synaptic stimulation creates an immobile "tag" at active synapses that "captures" essential protein products (Frey and Morris, 1997Frey U. Morris R.G. Nature. 1997; 385: 533-536Crossref PubMed Scopus (1191) Google Scholar). Induction of L-LTP in one synaptic pathway can effectively convert E-LTP to L-LTP in a second independent synaptic pathway, suggesting that proteins synthesized in response to stimulation of one group of synapses are available to other stimulated synapses within the same neuron. Synaptic tagging and capture are also observed with LTD, and in fact, crosscapture has been reported between LTP and LTD: induction of L-LTP in one synaptic pathway can convert E-LTD to L-LTD in another pathway, and vice versa (Sajikumar and Frey, 2004Sajikumar S. Frey J.U. Neurobiol. Learn. Mem. 2004; 82: 12-25Crossref PubMed Scopus (251) Google Scholar). Crosscapture suggests that induction of L-LTP and L-LTD may stimulate the synthesis of overlapping sets of proteins capable of enabling either process, consistent with evidence for the use of similar ERK- and mTOR-dependent translational mechanisms outlined above. In essence, synaptic capture implies that the local availability of essential proteins is sufficient to promote the consolidation of LTP and LTD. If the supply of essential proteins is limiting, competition can also be observed among groups of tagged synapses for consolidation of the accompanying synaptic changes (Fonseca et al., 2004Fonseca R. Nagerl U.V. Morris R.G. Bonhoeffer T. Neuron. 2004; 44: 1011-1020Abstract Full Text Full Text PDF PubMed Scopus (132) Google Scholar). From a behavioral standpoint, new proteins synthesized during a learning episode enable associative and competitive interactions among neighboring synapses experiencing LTP and LTD, and the resulting pattern of persistent synaptic weight changes is likely to be crucial for proper long-term memory representation (Govindarajan et al., 2006Govindarajan A. Kelleher R.J. Tonegawa S. Nat. Rev. Neurosci. 2006; 7: 575-583Crossref PubMed Scopus (256) Google Scholar). By altering the levels and/or identities of available proteins, dysregulation of neuronal protein synthesis would interfere with the establishment of appropriate patterns of synaptic modification. In particular, increased availability of plasticity-related proteins would promote consolidation of synaptic changes that would otherwise be lost. Such inappropriate synaptic consolidation would generally be expected to compromise the cognitive performance of neural circuits, reducing the specificity and signal-to-noise ratio of synaptic changes that underlie normal learning. It is interesting to consider the possibility that excessive synaptic capture and consolidation may in certain situations allow for enhanced long-term memory formation. Superior declarative memory, which depends on the hippocampus, is a common feature of savant abilities in autistic individuals (Heaton and Wallace, 2004Heaton P. Wallace G.L. J. Child Psychol. Psychiatry. 2004; 45: 899-911Crossref PubMed Scopus (90) Google Scholar). Whereas effective memory consolidation and retention typically benefit incrementally from repeated exposure to new information, individuals with savant abilities are remarkable in their capacity for rapid or "single trial" learning. Thus, in the case of mnemonic savant abilities, excessive protein synthesis could promote rapid and efficient synaptic capture and consolidation of hippocampal memory traces, regardless of their salience, while at the same time causing a more generalized impairment of cognitive function. In this way, cognitive impairment and savant abilities could be two sides of the same coin. It will be of interest to determine whether mutations causing elevated synthesis or abundance of neuronal proteins are associated with savant abilities in autism. The functions of gene products mutated in other monogenic disorders that overlap with autism lead us to speculate that overexpression of plasticity-related proteins may be one possible molecular mechanism underlying autism (Table 1). Neurofibromatosis type I is caused by inactivating mutations in neurofibromin (NF1), a Ras GAP, resulting in upregulation of Ras-dependent ERK and mTOR activation (Dasgupta and Gutmann, 2003Dasgupta B. Gutmann D.H. Curr. Opin. Genet. Dev. 2003; 13: 20-27Crossref PubMed Scopus (73) Google Scholar). Mutations in the E3 ubiquitin ligase UBE3A have been identified in Angelman's syndrome, suggesting that ubiquitin-dependent protein turnover may be impaired in this disorder, possibly leading to elevated synaptic protein levels (Jiang and Beaudet, 2004Jiang Y.H. Beaudet A.L. Curr. Opin. Pediatr. 2004; 16: 419-426Crossref PubMed Scopus (69) Google Scholar). Timothy's syndrome is caused by mutations in the L-type voltage-gated calcium channel (VGCC) Cav1.2, which impair channel inactivation and prolong inward calcium ion currents (Splawski et al., 2004Splawski I. Timothy K.W. Sharpe L.M. Decher N. Kumar P. Bloise R. Napolitano C. Schwartz P.J. Joseph R.M. Condouris K. et al.Cell. 2004; 119: 19-31Abstract Full Text Full Text PDF PubMed Scopus (1101) Google Scholar). Although the resulting enhancement of calcium ion influx may have pleiotropic effects, activation of the ERK pathway and CREB-dependent transcription are major effector mechanisms regulated by L-type VGCCs (Dolmetsch et al., 2001Dolmetsch R.E. Pajvani U. Fife K. Spotts J.M. Greenberg M.E. Science. 2001; 294: 333-339Crossref PubMed Scopus (707) Google Scholar). The molecular defects in these disorders suggest that a common consequence may be an overabundance of plasticity-related proteins. Rett's syndrome is caused by loss-of-function mutations in the methyl-CpG binding protein 2 (MeCP2), which can function as both a transcriptional activator and repressor (Chahrour et al., 2008Chahrour M. Jung S.Y. Shaw C. Zhou X. Wong S.T. Qin J. Zoghbi H.Y. Science. 2008; 320: 1224-1229Crossref PubMed Scopus (1247) Google Scholar). In rare cases, a Rett-like syndrome can also be caused by duplication of the MECP2 locus, indicating that decreased and increased MeCP2 dosage produce similar phenotypes. Analysis of mouse models either lacking or overexpressing MeCP2 revealed that neuronal transcription of a large number of target genes is decreased by loss of MeCP2 function, whereas transcription of the same group of target genes is increased by gain of MeCP2 function. Interestingly, the number of excitatory hippocampal synapses is decreased in mice lacking MeCP2 and increased in mice overexpressing MeCP2 (Chao et al., 2007Chao H.T. Zoghbi H.Y. Rosenmund C. Neuron. 2007; 56: 58-65Abstract Full Text Full Text PDF PubMed Scopus (346) Google Scholar), indicating that altered transcription of MeCP2 target genes, which presumably results in altered synthesis of the encoded proteins, produces corresponding changes in synaptic connectivity. As emphasized above for other autistic disorders, these findings suggest that cognitive deficits and autistic features arise in the MeCP2 duplication syndrome as a result of exaggerated protein expression. Conversely, findings from the mouse model of Rett's syndrome suggest that inadequate protein expression can also produce cognitive impairment and autism. Supporting this notion, multidisciplinary studies in mice have demonstrated that defects in synaptic protein synthesis impair hippocampal learning and memory (Kelleher et al., 2004aKelleher 3rd, R.J. Govindarajan A. Jung H.-Y. Kang H. Tonegawa S. Cell. 2004; 116: 467-469Abstract Full Text Full Text PDF PubMed Scopus (662) Google Scholar, Costa-Mattioli et al., 2007Costa-Mattioli M. Gobert D. Stern E. Gamache K. Colina R. Cuello C. Sossin W. Kaufman R. Pelletier J. Rosenblum K. et al.Cell. 2007; 129: 195-206Abstract Full Text Full Text PDF PubMed Scopus (345) Google Scholar). In a recent human genetic study, ASD-linked deletion mutations affected several genes whose expression is stimulated by neuronal activity, further arguing that defects in synaptic activity-induced gene expression contribute to ASD pathogenesis (Morrow et al., 2008Morrow E.M. Yoo S.Y. Flavell S.W. Kim T.K. Lin Y. Hill R.S. Mukaddes N.M. Balkhy S. Gascon G. Hashmi A. et al.Science. 2008; 321: 218-223Crossref PubMed Scopus (533) Google Scholar). Based on these observations, we suggest that the performance of neuronal networks mediating cognition depends on the level of synaptic protein synthesis (Figure 2). Deviations in either direction from the optimal level of synaptic protein synthesis can adversely affect synaptic capture and consolidation, and the resulting perturbations in synaptic connectivity may underlie the development of cognitive impairment and autistic traits. The manner in which synaptic properties are affected by aberrant protein synthesis is likely to depend on the identities of the individual proteins whose expression is altered. A parsimonious view of autism pathogenesis would envision the convergence of diverse molecular triggers on a final common disease-causing pathway. The identification of ASD-linked mutations in the synaptic adhesion molecules neuroligins 3 and 4 has suggested that synaptic abnormalities play a central role in autism (Zoghbi, 2003Zoghbi H.Y. Science. 2003; 302: 826-830Crossref PubMed Scopus (520) Google Scholar, Persico and Bourgeron, 2006Persico A.M. Bourgeron T. Trends Neurosci. 2006; 29: 349-358Abstract Full Text Full Text PDF PubMed Scopus (410) Google Scholar). Our proposed model provides an additional molecular mechanism leading to synaptic dysfunction in autism. How might defective synaptic adhesion and aberrant synaptic protein synthesis produce a common synaptic phenotype? Recent evidence indicates that neuroligins stabilize new synapses and specify their functional properties, thereby regulating the balance of excitatory and inhibitory transmission (Chubykin et al., 2007Chubykin A.A. Atasoy D. Etherton M.R. Brose N. Kavalali E.T. Gibson J.R. Sudhof T.C. Neuron. 2007; 54: 919-931Abstract Full Text Full Text PDF PubMed Scopus (402) Google Scholar) (see Essay by Walsh et al., page 396 of this issue). Interestingly, an ASD-linked missense mutation in neuroligin 3 shifts this balance of excitation and inhibition in favor of increased inhibitory drive (Tabuchi et al., 2007Tabuchi K. Blundell J. Etherton M.R. Hammer R.E. Liu X. Powell C.M. Sudhof T.C. Science. 2007; 318: 71-76Crossref PubMed Scopus (667) Google Scholar). Dysregulated protein synthesis could similarly alter the balance of excitation and inhibition by promoting the net strengthening or weakening of excitatory relative to inhibitory synapses. For example, excessive protein synthesis-dependent LTD, as observed in the hippocampus of Fmr1 knockout mice, could promote a net weakening of excitatory relative to inhibitory synapses. Reduced excitatory and/or increased inhibitory synaptic activity has been observed in the hippocampus and neocortex of several other mouse models of ASDs, suggesting that an imbalance of excitation and inhibition may be a common synaptic phenotype underlying autism (Cline, 2005Cline H. Curr. Biol. 2005; 15: R203-R205Abstract Full Text Full Text PDF PubMed Scopus (98) Google Scholar, Dani et al., 2005Dani V.S. Chang Q. Maffei A. Turrigiano G.G. Jaenisch R. Nelson S.B. Proc. Natl. Acad. Sci. USA. 2005; 102: 12560-12565Crossref PubMed Scopus (464) Google Scholar, Chao et al., 2007Chao H.T. Zoghbi H.Y. Rosenmund C. Neuron. 2007; 56: 58-65Abstract Full Text Full Text PDF PubMed Scopus (346) Google Scholar, Hanson and Madison, 2007Hanson J.E. Madison D.V. J. Neurosci. 2007; 27: 4014-4018Crossref PubMed Scopus (82) Google Scholar). It remains to be seen whether altered protein synthesis-dependent plasticity at excitatory and/or inhibitory synapses contributes to imbalances of excitation and inhibition in mouse models of autism. Single-gene disorders that increase autism risk in humans offer an exciting window into autism because they can be modeled in mice. Mouse models for several of these disorders display behavioral phenotypes resembling human autism, including impairments in cognition and social interaction (Moy et al., 2006Moy S.S. Nadler J.J. Magnuson T.R. Crawley J.N. Am. J. Med. Genet. C. Semin. Med. Genet. 2006; 142: 40-51Crossref Scopus (108) Google Scholar). However, it should be emphasized that the synaptic pathology that yields autistic behavior in humans might have distinct effects in other species due to differences in brain circuitry. Thus, the failure of a single-gene mutation in mice to phenocopy the full spectrum of clinical features does not diminish the importance of the mouse models to reveal pathophysiology and, potentially, new treatments. Based on the findings in mice discussed above, we hypothesize that mutations that lead to excessive or dysregulated synaptic protein synthesis could be one mechanism contributing to autism in humans. One prediction of our hypothesis that can be tested in mice is that there should be some observable commonalities in the synaptic dysfunction wrought by the loss of FMRP, TSC1/2, and PTEN. If our hypothesis proves to be correct, approaches to restoring normal levels of protein expression may provide a therapeutic strategy for treating these disorders, and perhaps mental retardation and autism. In the case of FXS, the discovery of enhanced mGluR-dependent LTD suggests that inhibition of mGluR activity may be one way to limit the excessive translational response to mGluR activation (Bear et al., 2004Bear M.F. Huber K.M. Warren S.T. Trends Neurosci. 2004; 27: 370-377Abstract Full Text Full Text PDF PubMed Scopus (1197) Google Scholar). Indeed, a range of phenotypes displayed by mice lacking Fmr1 can be rescued by genetic reduction of mGluR5 activity (Dolen et al., 2007Dolen G. Osterweil E. Rao B.S. Smith G.B. Auerbach B.D. Chattarji S. Bear M.F. Neuron. 2007; 56: 955-962Abstract Full Text Full Text PDF PubMed Scopus (732) Google Scholar). Similarly, the mTOR inhibitor rapamycin rescues abnormal hippocampal LTP and memory in a mouse model of TSC (Ehninger et al., 2008Ehninger D. Han S. Shilyansky C. Zhou Y. Li W. Kwiatkowski D.J. Ramesh V. Silva A.J. Nat. Med. 2008; 14: 843-848Crossref PubMed Scopus (637) Google Scholar). Other approaches that restrain the activity of the molecular machinery regulating synaptic protein synthesis may have therapeutic potential in autism.