Behavioral phenotypes of genetic mouse models of autism

Abstract

More than a hundred de novo single gene mutations and copy-number variants have been implicated in autism, each occurring in a small subset of cases. Mutant mouse models with syntenic mutations offer research tools to gain an understanding of the role of each gene in modulating biological and behavioral phenotypes relevant to autism. Knockout, knockin and transgenic mice incorporating risk gene mutations detected in autism spectrum disorder and comorbid neurodevelopmental disorders are now widely available. At present, autism spectrum disorder is diagnosed solely by behavioral criteria. We developed a constellation of mouse behavioral assays designed to maximize face validity to the types of social deficits and repetitive behaviors that are central to an autism diagnosis. Mouse behavioral assays for associated symptoms of autism, which include cognitive inflexibility, anxiety, hyperactivity, and unusual reactivity to sensory stimuli, are frequently included in the phenotypic analyses. Over the past 10 years, we and many other laboratories around the world have employed these and additional behavioral tests to phenotype a large number of mutant mouse models of autism. In this review, we highlight mouse models with mutations in genes that have been identified as risk genes for autism, which work through synaptic mechanisms and through the mTOR signaling pathway. Robust, replicated autism-relevant behavioral outcomes in a genetic mouse model lend credence to a causal role for specific gene contributions and downstream biological mechanisms in the etiology of autism.

Autism spectrum disorder (ASD) is a neurodevelopmental syndrome with a prevalence of over 1% of the population (CDC 2014; Elsabbagh et al. 2012; Kim et al. 2011). Diagnosis by the current Diagnostic and Statistical Manual of Mental Disorders (DSM-5) criteria is based on two categories of behavioral symptoms: (1) unusual reciprocal social interactions and impaired social communication; and (2) stereotyped and repetitive patterns of behaviors, with restricted interests and activities (American Psychiatric Association 2013; Lord & Bishop 2015). Associated symptoms, which are present in subsets of individuals with ASD, include intellectual disabilities, anxiety, seizures, hyperactivity, sleep disruption and unusual reactivity to sensory stimuli. Early behavioral interventions are the current standard of care and offer the best long-term outcomes at present (Lord & Jones 2013; Rogers et al. 2012). Intensive behavioral intervention is highly effective in teaching young children to improve their social skills and redirect their repetitive behaviors. However, even the best behavioral therapies do not work for all, are expensive, are labor- and time-intensive and are not available in many geographic regions. Understanding the causes of autism is the first step in the development of effective medical therapeutics to improve symptoms and reverse the disorder’s trajectory.

Hypotheses about the genetic causes of ASD originally arose from observations of a male–female bias, 4:1 or greater, and high concordance between identical twins, 50–90%, as compared to less than 10% for non-identical twins and siblings (Fombonne 2009; Hallmayer et al. 2011; Miles 2011; Nordenbaek et al. 2014; Ritvo et al. 1985; Sandin et al. 2014; Smalley et al. 1988). Intensive searches for the genes causing ASD employed genome-wide association approaches in the early 2000s, progressed to analyses of copy-number variants (CNVs), and are now proceeding with exome and whole genome sequencing in thousands of individuals. Early findings quickly revealed that ASD is not a monogenic disorder. In contrast to disorders such as Huntington’s disease and Fragile X syndrome, there is no one specific gene responsible for all cases of autism. Rather, a growing number of de novo single gene mutations and CNVs have been identified in people with autism (Alarcon et al. 2008; Bucan et al. 2009; Butler et al. 2005; Buxbaum et al. 2007; Cook & Scherer 2008; Crepel et al. 2014; Glessner et al. 2009; Iossifov et al. 2014; Krumm et al. 2015; Kumar et al. 2009; Lawson-Yuen et al. 2008; Leblond et al. 2014; Michaelson et al. 2012; Morrow 2010; Neale et al. 2012; O’Roak et al. 2011; Pinto et al. 2010; Szatmari et al. 2007; Vernes et al. 2008; Wang et al. 2009; Yuen et al. 2015). Mutations in common gene variants and de novo coding mutations may be responsible for up to 50% of ASD cases (Gaugler et al. 2014; Iossifov et al. 2014; Miles 2011). Over 100 risk genes and CNVs for ASD have been published, each one appearing in only a relatively small number of individuals (Butler et al. 2005; Coe et al. 2014; De Rubeis et al. 2014; Gaugler et al. 2014; Iossifov et al. 2014; Li et al. 2014; Parikshak et al. 2013; Pinto et al. 2014; Willsey & State 2015). Epigenetic risk factors have been implicated in ASD, including chromatin remodeling and methylation mechanisms, such as CHD8 (Bernier et al. 2014; Cotney et al. 2015; O’Roak et al. 2012; Wilkinson et al. 2015), HDAC (Foley et al. 2012; Moldrich et al. 2013) and MECP2 (Shibayama et al. 2004; Theoharides et al. 2015), Further, environmental risk factors, such as parental age (Kong et al. 2012) and atypical maternal autoantibodies (Braunschweig et al. 2013; Brimberg et al. 2013; Diamond et al. 2013; Piras et al. 2014), are associated with a higher incidence of ASD.

One of the most intriguing aspects regarding the genetics of ASD is the enigma of how these many risk factors converge to result in the same general cluster of symptoms diagnosed as ASD. One possibility is that there are underlying convergent downstream mechanisms which contribute to ASD symptomotology. No definitive biomarkers have yet been identified across all diagnosed cases. Rather, subsets of biological factors may define subgroups of individuals with ASD. Stratification by subgroup, either by behavioral category or biomarker, may offer the best strategy for focused clinical trials. Intensive searches are underway to define abnormalities in neurophysiology, neuroanatomy, brain chemistry, immune markers and other key biological systems (Ecker et al. 2013; Jeste & Geschwind 2014; Levitt & Veenstra-VanderWeele 2015). High heterogeneity of symptoms across cases suggests that autism is actually multiple disorders, analogous to the plural concept of ‘cancers’, with different genetic etiologies and biological defects, to be treated with different classes of therapeutics. The concept of ‘autisms’ is implicit in the current use of the term ASD, implemented in the 2013 edition of the DSM-5.

Readers of Genes, Brain and Behavior are well aware of methods to interrogate genetic hypotheses of human disorders by targeting the homologous mutation in another species and then explicating the consequent phenotypic outcomes. Knockout (KO) and humanized knockin mice, and more recently KO rats, have been generated for many of the single gene mutations and CNVs that were identified in ASD populations and for comorbid neurodevelopmental disorders such as Fragile X and tuberous sclerosis (TSC) (Baudouin et al. 2012; Ey et al. 2011; Silverman et al. 2010a; Zoghbi & Bear 2012). Some of these mutant mouse models are now being employed in preclinical testing of pharmacological targets to treat the core symptoms of ASD (Silverman & Crawley 2014; Spooren et al. 2012; Vorstman et al. 2014).

As genetic mouse models emerged, our behavioral neuroscience laboratory invested in methods development to design mouse behavioral assays with high relevance to the diagnostic symptoms of autism (Crawley 2004). Because the clinical phenotype of this uniquely human disorder is complex and heterogeneous, we initiated discussions with autism clinical experts, to understand the critical symptoms that could be most meaningfully modeled in mice. Clinical researchers, including colleagues at the University of California Davis MIND Institute, Weill Cornell Medical College, University of North Carolina, University of Washington, University College London and the National Institute of Mental Health Intramural Research Program, kindly allowed us to observe diagnostic interviews and watch videotapes of children with ASD. Knowledge gained through these sessions, and from lectures and conversations with many other generous colleagues working with children, adolescents and adults with ASD, guided our thinking in the development of mouse behavioral assays that dovetail with the natural behavioral repertoire of mice. Considering the types of social approach abnormalities and inappropriate social behaviors that are common in ASD, we developed a mouse 3-chambered social approach assay (Moy et al. 2004; Nadler et al. 2004), refined methods for scoring reciprocal social interactions in juvenile and adult mice (McFarlane et al. 2008), adapted measures for the detection of responses to social olfactory cues (Yang & Crawley 2009), and developed call categories for ultrasonic vocalizations emitted in response to social cues during reciprocal social interactions (Scattoni et al. 2008, 2011). Further, we established observational scoring methods to quantify motor stereotypies and repetitive behaviors, such as self-grooming and digging, along with assembling a set of established behavioral assays relevant to anxiety, intellectual impairment, hyperactivity and sensory reactivity (McFarlane et al. 2008; Moy et al. 2008a,b; Roullet & Crawley 2011; Silverman et al. 2010a,b, 2012, 2013, 2015; Wohr et al. 2011a; Yang et al. 2011, 2012a, 2015), which are now widely used. A small subset of these assays is illustrated in Figure 1. This review presents examples and summaries of ASD-relevant phenotypes discovered by our lab and many other excellent behavioral genetics labs, revealing the phenotypic consequences of targeting mutations in ASD risk genes.

Figure 1. Examples of rodent behavioral assays with face validity to the diagnostic symptoms of autism.

(a) Two mice interacting in a Noldus Phenotyper 3000 reciprocal social interaction chamber equipped with an Avisoft ultrasonic microphone. The inset shows representative ultrasonic vocalizations recorded during adult male–female social interaction. (b) Close-up of two mice displaying ‘crawl over and under’ during the reciprocal social interaction test session. (c) 3-Chambered social approach apparatus offers automated scoring of time spent with a novel social partner vs. time spent with a novel object. (d) BTBR mouse engaged in repetitive self-grooming. Photos by Jane Hayes, Michael Pride, Jill Silverman and Mu Yang, MIND Institute, University of California Davis School of Medicine.

A remarkable number of risk genes for ASD code for synaptic proteins. Cell adhesion proteins, including contactin-associated proteins, neuroligins and neurexins connect dendrites with axons to promote synapse formation. Postsynaptic scaffolding proteins, such as shanks and neuroligins, strengthen synapses and maintain synaptic transmission. Postsynaptic receptors, such as NMDA and metabotropic glutamate receptors, GABA receptors of varying subunit compositions, serotonin transporter and receptor subtypes, and oxytocin receptors, mediate excitatory and inhibitory synaptic signals. Sodium channels, potassium channels and downstream signaling pathways, such as the PTEN/PI3 kinase/Akt/mTOR pathway, mediate postsynaptic events and critical cellular functions. Mutations and common variants of the genes for these proteins have been identified in small numbers of individuals with autism and related disorders (Butler et al. 2005; Cheah et al. 2013; Frazier et al. 2014; Han et al. 2012; Krey et al. 2013; Rosander & Hallbook 2015; Tavassoli et al. 2014; Veenstra-VanderWeele et al. 2012; Weiss et al. 2003).

Mice with targeted mutations in many of these genes were generated by outstanding molecular genetics laboratories and generously donated to public repositories such as The Jackson Laboratory. Behavioral phenotypes have been published for some of these mutant lines. In most cases, one original publication describes the behavioral, electrophysiological, neuroanatomical, and/or biochemical phenotypes of the new mouse model of autism. In some cases, the first findings have been replicated by the same laboratory in additional publications. In a few cases, behavioral phenotypes have been replicated by other laboratories. We summarize some of the strongest findings below. Table 1 provides descriptions of gene mutations associated with human ASD. Table 2 summarizes the behavioral phenotypes in the corresponding mouse models, focusing on a subset of ASD risk genes that are involved in synaptic function and the mTOR signaling pathway. Here, we will refer to mice without any functional alleles (homozygous null) as KO mice, mice with one functional allele as heterozygous (Het) mice, mice with targeted amino acid substitutions as knockin mice, and littermate controls with both functional alleles as wildtype (WT) mice.

Table 1.

Examples of autism risk genes identified by human genetic studies

Gene	Identification method	Investigated population	Reference for genetic study
CNTNAP2
Association with ASD	Linkage study	Old Order Amish families (4 children and 6 parents) with an autosomal recessive founder (null mutation)	Strauss et al. (2006)
	Linkage, association and gene expression studies	172 parent–child trios from the Autism Genetic Resource Exchange (AGRE) resource	Alarcon et al. (2008)
	Genome-wide linkage study, family based-association mapping	National Institute of Mental Health Autism Genetics Initiative Repository; Stage I: 72 muliplex families (148 affected, 292 controls); Stage II: 1295 parent–child trios (145 mulitplex families with 303 affected children)	Arking et al. (2008)
	Cytogenetics	34 year old female	Rossi et al. (2008)
	Cytogenetics	11 year old male	Poot et al. (2010)
Limited association with ASD	Cytogenetics, linkage and resequencing	635 ASD cases and 942 controls	Bakkaloglu et al. (2008)
	Genetic association analyses; transmission disequilibrium test	186 multiplex (408 trios) and 323 simplex families with ASD from the AGRE resource	Sampath et al. (2013)
No association with ASD	Next generation sequencing	2704 ASD cases, 2747 controls	Murdoch et al. (2015)
NLGN1
Association with ASD	Whole genome copy number variant study	859 ASD cases, 1409 controls; 1336 ASD cases, 1110 controls	Glessner et al. (2009)
Limited association with ASD	Mutation analysis, linkage analysis, association analysis	30 ASD cases; 19 families with 41 ASD cases; 100 families with 122 ASD cases	Ylisaukko-oja et al. (2005)
NLGN3
Association with ASD	Amino acid sequencing, linkage study	36 ASD sibling pairs, 122 ASD trios, 350 unrelated controls	Jamain et al. (2003)
Limited association with ASD	Mutation analysis, linkage analysis, association analysis	30 ASD cases; 19 families with 41 ASD cases; 100 families with 122 ASD cases	Ylisaukko-oja et al. (2005)
NLGN4
Association with ASD	Cytogenetic analysis, fluorescence in situ hybridization (FISH)	8 females, 3 with ASD	Thomas et al. (1999)
	Amino acid sequencing, linkage study	36 ASD sibling pairs, 122 ASD trios, 350 unrelated controls	Jamain et al. (2003)
	Linkage analysis and gene sequencing	10 members of a French family with ASD and intellectual disability, 200 controls	Laumonnier et al. (2004)
	Direct sequencing and mutation analysis	148 unrelated ASD cases, 48 ADHD and bipolar cases, 288 unaffected controls	Yan et al. (2005)
	Chromosome analysis	Family with 1 ASD proband, 96 controls	Lawson-Yuen et al. (2008)
	Single nucleotide polymorphism microarrays and karyotyping	427 unrelated ASD families	Marshall et al. (2008)
Limited association with ASD	Mutation analysis, linkage analysis, association analysis	30 ASD cases; 19 families with 41 ASD cases; 100 families with 122 ASD cases	Ylisaukko-oja et al. (2005)
No association with ASD	PCR genetic screen, denaturing high performance liquid chromatography	196 ASD cases from 183 multiplex and 13 simplex families	Vincent et al. (2004)
NRXN1
	Gene scanning and sequencing	103 Caucasian ASD cases, 61 Afro-American ASD cases, 535 Caucasian controls	Feng et al. (2006)
	Linkage study, using comparative analysis of hybridization intensities	1496 multiplex ASD families (at least 2 affected individuals) (7917 family members)	Szatmari et al. (2007)
	Single nucleotide polymorphism microarrays and karyotyping	427 unrelated ASD families	Marshall et al. (2008)
	Cytogenic analysis	1 ASD case study	Zahir et al. (2008)
	Whole genome CNV study	859 ASD cases, 1409 controls; 1336 ASD cases, 1110 controls	Glessner et al. (2009)
	Comparative genomic hybridization microarrays	3540 cases with developmental disorders, ASD, intellectual disability	Ching et al. (2010)
SHANK1
	Microarray	1158 Canadian and 456 European ASD cases, 15122 controls	Sato et al. (2012)
	Meta-analysis of copy-number and coding-sequence variants	5657 ASD cases, 19163 controls; 76 0 – 214 7 ASD cases, 492–1090 controls depending on the SHANK gene	LeBlond et al. (2014)
SHANK2
	Genome-wide microarray scan; DNA sequencing	396 ASD cases, 184 individuals with intellectual disability, 659 unaffected controls	Berkel et al. (2010)
	Dense genotyping arrays	996 European ASD cases, 1287 matched controls	Pinto et al. (2010)
	Meta-analysis of copy-number and coding-sequence variants	5657 ASD cases, 19, 163 controls; 76 0 – 214 7 ASD cases, 492–1090 controls depending on the SHANK gene	LeBlond et al. (2014)
SHANK3
	FISH analysis; direct sequencing	226 families with at least 1 ASD child, 270 controls	Durand et al. (2007)
	DNA sequencing and microarray-based comparative intensity analysis	400 Canadian ASD cases; HapMap collection was used for comparison	Moessner et al. (2007)
	Single nucleotide polymorphism microarrays and karyotyping	427 unrelated ASD families	Marshall et al. (2008)
	Gene sequencing	427 ASD cases, 190 controls	Gauthier et al. (2009)
	Denaturing high performance liquid chromatography, direct sequencing, multiplex ligation-dependent probe amplification, array comparative genomic hybridization	3 cohorts: 133 American ASD cases; 88 Italian ASD cases; 104 American ASD cases; 560 American controls and 422 Italian controls	Boccuto et al. (2013)
	Meta-analysis of copy-number and coding-sequence variants	5657 ASD cases, 19, 163 controls; 76 0 – 2147 ASD cases, 492–1090 controls depending on the SHANK gene	LeBlond et al. (2014)
PTEN
	Direct DNA sequencing	18 ASD cases with macrocephaly	Butler et al. (2005)
	Direct DNA sequencing, multiplex ligation-dependent probe amplification	88 ASD cases with macrocephaly	Buxbaum et al. (2007)
	Direct DNA sequencing	2 ASD cases with macrocephaly	Herman et al. (2007)
	Direct DNA Sequencing	114 cases of ASD, developmental delay or macrocephaly	Varga et al. (2009)
	Direct DNA sequencing	93 cases of ASD or developmental delays with macrocephaly	McBride et al. (2010)

Table 2.

Examples of genetic mouse models of autism

Mouse model	Mouse phenotype	Reference
Cntnap2 knockout mouse	Reduced social interactions Increased repetitive behaviors Reduced juvenile ultrasonic vocalizations Hyperactivity Seizures Abnormal cortical migration Asynchronous neurotransmission Reduced number of interneurons	Penagarikano et al. (2011)
Neuroligin-1 knockout mouse	Minimally impaired social approach Repetitive behavior Spatial learning deficits Impaired hippocampal long term potentiation Reduced NMDA/AMPA glutamate receptor ratio at cortico-striatal synapses	Blundell et al. (2010)
Neuroligin-2* knockout mouse	Increased anxiety-like behavior Decreased pain sensitivity Decreased motor coordination Reduced exploratory activity Developmental milestone delays Reduced ultrasonic pup vocalizations Decreased inhibitory synaptic puncta, with no change in overall synapse number Reduced inhibitory synaptic neurotransmission	Chubykin et al. (2007), Blundell et al. (2009) and Wohr et al. (2013)
Neuroligin-3* knockout mouse	Reduced fear conditioning Olfactory impairments Hyperactivity Decreased total brain volume	Radyushkin et al. (2009)
Neuroligin-3* knockin mouse (R451C substitution)	Enhanced water maze spatial learning acquisition Increased protein levels of inhibitory synaptic markers Increased inhibitory neurotransmission in the somatosensory barrel cortex Increase in AMPA-mediated excitatory neurotransmission and enhanced long-term potentiation in the hippocampus Increased dendritic branching in the hippocampus	Tabuchi et al. (2007) and Etherton et al. (2011)
Neuroligin-3* knockin mouse (R451C substitution)	Reduced ultrasonic vocalizations in pups Reduced acoustic startle to high decibel tones Minor developmental differences in growth and righting reflexes	Chadman et al. (2008)
Nlgn4 knockout mouse (gene trap insertion downstream of exon I)	Reduced social interaction time Impaired social approach and social novelty recognition Reduced ultrasonic vocalizations Reduced volume of total brain, cerebellum and brain stem	Jamain et al. (2008) and El-Kordi et al. (2013)
Nlgn4* knockout mouse (gene trap insertion downstream of exon I)	No genotype differences in ultrasonic vocalizations, anxiety-related behaviors or locomotor activity	Ey et al. (2012)
Neurexin 1α* knockout mouse	Increased grooming Decreased prepulse inhibition Reduced excitatory synaptic neurotransmission	Etherton et al. (2009)
Neurexin 1α* knockout mouse	Reduced exploratory activity in novel environments Increased aggressive behaviors Mild increases in anxiety-like behavior	Grayton et al. (2013)
Shank1* knockout mouse (targeted replacement of PDZ domain (exons XIV and XV))	Anxiety-like phenotype Impaired contextual fear memory Enhanced spatial learning Impaired long term memory retention Reduced dendritic spine size and smaller postsynaptic densities Weaker basal synaptic neurotransmission	Hung et al. (2008)
Shank1* knockout mouse (targeted replacement of PDZ domain (exons XIV and XV))	Mild anxiety-like phenotype Reduced exploratory locomotion	Silverman et al. (2011)
Shank1* knockout mouse (targeted replacement of PDZ domain (exons XIV and XV))	Reduced pup ultrasonic vocalizations Reduced male scent marking and abnormal ultrasonic vocalizations in response to female pheromones	Wohr et al. (2011b)
Shank2 knockout mouse (lacking all Shank2 isoforms)	Fewer social contacts during same-sex interactions Reduced social novelty recognition Repetitive grooming Abnormal ultrasonic vocalizations Hyperactivity Increased anxiety-like behavior Reduced dendritic spines Impaired glutamatergic neurotransmission	Schmeisser et al. (2012)
Shank2 knockout mouse (exon VI and VII microdeletion and a frameshift for both Shank splice variants)	Reduced home-cage social interaction Reduced social preference Repetitive jumping Reduced ultrasonic vocalizations Impaired spatial learning Decreased glutamatergic NMDA receptor function	Won et al. (2012)
Shank3A* heterozygous mouse (ankyrin repeat domain deletion (exon IV-IX))	Reduced social sniffing in male–female reciprocal social interactions Reduced ultrasonic vocalizations Reduced long term potentiation Transient dendritic spine expansion during long term potentiation Reduced basal AMPA neurotransmission	Bozdagi et al. (2010)
Shank3A* heterozygous and knockout mice (ankyrin repeat domain deletion (exon IV-IX))	Reduction of select parameters (e.g. sniffing, following, front approach and push-crawl) in juvenile reciprocal social interaction in male heterozygous and knockout mice Increased repetitive grooming in male heterozygous and knockout mice Impaired novel object recognition in knockout mice Mild motor learning deficits in knockout mice Lower pain sensitivity in knockout mice (C57BL/6 background) Impaired synaptic transmission, induction and long-term potenti-ation in knockout mice	Yang et al. (2012b) and Drapeau et al. (2014)
Shank3A knockout mouse (ankyrin repeat domain deletion (exon IV-IX))	Reduced social interaction time Repetitive behaviors Aberrant ultrasonic vocalizations Reduced activity and motor learning in males Impaired acquisition in Morris water maze Reduced postsynaptic protein and glutamate receptor protein levels Longer dendritic spines Reduced long term potentiation	Wang et al. (2011)
Shank3A* knockout mouse (ankyrin repeat domain deletion)	Reduced preference for social novelty	Peca et al. (2011)
Shank3B knockout mouse (PDZ domain deletion)	Reduced social interaction and social novelty recognition Self-injurious repetitive grooming Increased anxiety-like behavior Reduced postsynaptic proteins and glutamate receptor protein levels Neuronal hypertrophy and reduction in dendritic spines Increased caudate volume Reduced neurotransmission in corticostriatal circuits	Peca et al. (2011)
Shank3* knockout mouse (exon 21 deletion, including the Homer binding domain)	Increased repetitive grooming in older knockout mice Impaired spatial learning Avoidance of inanimate objects Impaired motor coordination Increased pain sensitivity Decreased excitatory neurotransmission and reduced long-term potentiation Increased mGluR5 receptor levels in synaptic fractions	Kouser et al. (2013)
Nse-Cre Pten conditional knockout mouse (Pten deletion restricted to subsets of postmitotic cortical and hippocampal neurons)	Decreased social interaction with juvenile mice Reduced sociability and preference for social novelty Spatial learning deficits Reduced exploratory activity in the center of the open field Increased startle responses and reduced sensorimotor gating Spontaneous seizures Macrocephaly, including hippocampal dentate gyrus enlargement Hippocampal granule cell and cortical neuron hypertrophy	Kwon et al. (2006), Ogawa et al. (2007) and Zhou et al. (2009)
Pten heterozygous mouse	Reduced sociability and preference for social novelty Increased repetitive behavior Decreased aggression and reduced social investigation in the resident-intruder test Reduced sensorimotor gating Increased depression-like behavior in male heterozygous mice Macrocephaly and higher brain mass	Page et al. (2009); Clipperton-Allen and Page (2014) and Clipperton-Allen and Page (2015)

The asterisk (*) following the name of a mouse model indicates that its social behaviors were normal, using assays with high face validity to the types of social deficits which characterize autism.

Mouse models of genetic risk factors for autism

The CNTNAP2 gene, located on chromosome 7, encodes contactin-associated protein-like 2 (CASPR2), a member of the neurexin superfamily of proteins, functioning as a cell adhesion molecule and receptor (Rodenas-Cuadrado et al. 2014). This protein, which contains a putative PDZ binding domain, mediates interactions of neurons and glia during central nervous system development. It also is located in myelinated axons and directs potassium channel localization within differentiating neurons (Poliak et al. 1999, 2003). CASPR2 directly binds to the transcription factor FOXP2 (forkhead box protein P2), which has been implicated in speech and language development (Fischer & Hammerschmidt 2011; Vernes et al. 2008). Several mutations in the CNTNAP2 locus, including rare, common and deletion variants, have been associated with ASD (Alarcon et al. 2008; Arking et al. 2008; Poot et al. 2010; Rossi et al. 2008; Strauss et al. 2006), although more recent studies indicate a limited contribution of CNTNAP2 dysregulation to ASD (Bakkaloglu et al. 2008; Murdoch et al. 2015; Sampath et al. 2013). Mice lacking Cntnap2 exhibited reduced juvenile ultrasonic vocalizations, reduced social interaction time and increased repetitive behaviors (Penagarikano et al. 2011). In addition to these deficits in ASD-related behaviors, Cntnap2 KO mice also displayed abnormal neuronal cortical migration, asynchronous neuronal firing in the cortex, a reduced number of inhibitory interneurons, behavioral perseveration in a cognitive task, hyperactivity and seizures (Penagarikano et al. 2011).

Neuroligins are cell adhesion molecules located at the postsynaptic side of the synapse, interacting with their presynaptic partner proteins, the neurexins (Bang & Owczarek 2013). Neuroligins contribute to synaptic neurotransmission through their influence on synaptic formation and are distributed at excitatory and inhibitory synapses in an isoform-dependent manner (Hu et al. 2015). For example, neuroligin-1 is primarily located at excitatory synapses (Budreck et al. 2013; Chih et al. 2005; Song et al. 1999), while neuroligin-2 is found at inhibitory synapses (Varoqueaux et al. 2004); neuroligin-3 can be found at both these locations (Budreck & Scheiffele 2007). At the synapse, neuroligins bind to PSD-95, a scaffolding protein important for postsynaptic strengthening and synapse organization, particularly with ion channels and receptors, such as the glutamatergic NMDA receptor (Bolliger et al. 2001; Irie et al. 1997; Kim et al. 1995, 2008; Kornau et al. 1995; Niethammer et al. 1996; Shipman & Nicoll 2012). Several studies suggest that neuroligins are involved in synapse modulation and specification rather than synapse formation (Chubykin et al. 2007; Krueger et al. 2012; Varoqueaux et al. 2006). Neuroligin proteins encoded by X-linked genes, such as NLGN3 and NLGN4 which map to Xq13 and Xp22.3, respectively, have been associated with ASD in large genome-wide scans (Auranen et al. 2002; Glessner et al. 2009; Philippe et al. 1999), but strong associations have not been found in all studies (Vincent et al. 2004; Ylisaukko-oja et al. 2005). Using amino acid sequencing in linkage and proband case studies, deletions and frameshifts in NLGN3 and NLGN4 sequences have been identified in individuals with ASD (Jamain et al. 2003; Laumonnier et al. 2004; Lawson-Yuen et al. 2008; Marshall et al. 2008; Thomas et al. 1999; Yan et al. 2005).

Neuroligin-1 KO mice exhibited minimal deficits in social behavior, but displayed increased grooming and spatial learning impairments, along with impaired hippocampal long-term potentiation (Blundell et al. 2010). Neuroligin-2 heterozygous and KO mice showed normal social interactions in social approach, but displayed increased anxiety-like behavior, decreased pain sensitivity and poor motor coordination (Blundell et al. 2009; Wohr et al. 2013). In addition, neuroligin-2 KO mice had decreased inhibitory neurotransmission, as well as decreased immunostaining of inhibitory synapse markers (Blundell et al. 2009; Chubykin et al. 2007). Neuroligin-3 knockin (R451C) mice, with an arginine to cysteine substitution at site 451, did not display robust autism-relevant behaviors, but rather had mild developmental differences, (e.g. slower righting reflexes), enhanced spatial learning acquisition and reduced acoustic startle (Chadman et al. 2008; Etherton et al. 2011; Tabuchi et al. 2007), suggesting that this ASD-related point mutation delayed development, altered learning and reduced sensitivity to stimuli. Neuroligin-3 knockin mice also displayed increased inhibitory neurotransmission in the barrel cortex, increased excitatory neurotransmission and enhanced long-term potentiation in the hippocampus, increased dendritic branching in the hippocampus, as well as increased protein levels of inhibitory synaptic markers, while KO mice did not (Etherton et al. 2011; Tabuchi et al. 2007). Neuroligin-3 KO mice displayed normal sociability, but impairments in fear conditioning and olfaction, as well as hyperactivity and decreased total brain volume (Radyushkin et al. 2009). Characterization of Nlgn4 KO mice revealed that loss of this neuroligin resulted in reduced sociability and ultrasonic vocalizations, as well as a reduction in total brain volume (El-Kordi et al. 2013; Jamain et al. 2008). However, phenotypic analysis of later generations of the same line of Nlgn4 KO did not find any genotype differences in sociability, ultrasonic vocalizations, anxiety-related behaviors or general locomotor activity (Ey et al. 2012).

Neurexins are a class of cell adhesion proteins found on the presynaptic terminal of synapses that bind to neuroligins (Bang & Owczarek 2013). Numerous association studies have identified mutations in the NRXN1 gene, located on chromosome 2, in intellectual disabilities and other syndromes, including several cases of autism (Ching et al. 2010; Feng et al. 2006; Glessner et al. 2009; Gregor et al. 2011; Marshall et al. 2008; Szatmari et al. 2007; Zahir et al. 2008). Neurexin-1α KO mice displayed increased grooming, reduced locomotor activity, reduced sensorimotor gating and increased aggression (Etherton et al. 2009; Grayton et al. 2013). Further studies are necessary to determine the exact contribution of specific neurexin and neuroligin mutations to ASD-relevant behaviors.

The SHANK family of genes located on chromosome 22q encodes scaffolding proteins that assist in the synaptic organization of excitatory glutamatergic neurons by binding to postsynaptic density proteins, signaling molecules, postsynaptic receptors and cytoskeletal proteins (Boeckers et al. 2002; Grabrucker et al. 2011; Lim et al. 1999; Naisbitt et al. 1999; Sheng & Kim 2000; Tu et al. 1999). SHANK3 can bind to neuroligins, suggesting disrupted cell adhesion may contribute to ASD (Meyer et al. 2004). Genetic studies have identified de novo and inherited mutations in SHANK1 (Sato et al. 2012), SHANK2 (Berkel et al. 2010, 2012; Pinto et al. 2010) and SHANK3 (Boccuto et al. 2013; Durand et al. 2007; Gauthier et al. 2009, 2010; Marshall et al. 2008; Moessner et al. 2007). A recent meta-analysis of SHANK mutations has suggested that ASD severity due to SHANK mutations may be related to which gene is mutated, such that SHANK3 mutations have a higher frequency and penetrance in individuals with ASD, compared to SHANK1 and SHANK2 (Leblond et al. 2014). 22q13 deletion syndrome, also known as Phelan-McDermid syndrome, is caused by a deletion on the distal part of the long arm of chromosome 22 and is associated with ASD-like behaviors, including disrupted social behavior, repetitive behaviors, motor dysfunctions, seizures and moderate to severe intellectual disability (Kolevzon et al. 2014; Phelan & McDermid 2012). SHANK3 is one of the most commonly mutated genes within the Phelan-McDermid critical region and is thought to underlie most of the neural consequences of this deletion (Phelan & McDermid 2012).

While Shank1 KO mice do not display robust autism-relevant social deficits (Silverman et al. 2011), Shank1 KO mice emitted fewer ultrasonic vocalizations as pups, exhibited reduced scent marking and abnormal vocalizations as adults and had motor impairments (Silverman et al. 2011; Wohr et al. 2011b). Shank1 KO mice also displayed dendritic spine abnormalities, including weaker basal synaptic neurotransmission (Hung et al. 2008). Shank2 KO mice had abnormalities in several ASD-relevant behaviors, including reduced sociability, as measured by fewer social contacts during same-sex interactions, and reduced preference for social novelty, as well as higher levels of repetitive behaviors (e.g. grooming and jumping) and abnormal ultrasonic vocalizations (Schmeisser et al. 2012; Won et al. 2012). In addition, Shank2 KO mice had a reduced number of hippocampal dendritic spines and reduced glutamatergic neurotransmission in the hippocampus (Schmeisser et al. 2012; Won et al. 2012).

Mutant mice have been generated for each of the Shank3 isoforms, with deletions in various domains of the Shank3 gene, some of which displayed social deficits while others displayed normal sociability (Jiang & Ehlers 2013). Reduced sociability, reduced ultrasonic vocalizations and high levels of repetitive self-grooming were dependent on which isoform was deleted (Bozdagi et al. 2010; Drapeau et al. 2014; Kouser et al. 2013; Peca et al. 2011; Wang et al. 2011; Yang et al. 2012b). Reduced basal neurotransmission as well as abnormalities in neuronal morphology (e.g. neuronal hypertrophy, dendritic spine deficits) have been identified in most of these models (Bozdagi et al. 2010; Kouser et al. 2013; Peca et al. 2011; Wang et al. 2011; Yang et al. 2012b), underscoring the importance of the Shank proteins in maintaining normal synaptic function and neuronal structure. Therefore, several different ASD-relevant mutations in the SHANK gene family have been modeled in mice, and while the phenotypes differ between the specific models, it appears that complete or partial loss of Shank proteins may be detrimental to normal social behaviors and may induce high levels of repetitive behaviors in mice.

ASD risk genes in the mTOR signaling pathway

While many of the identified risk genes for ASD involve synaptic proteins, mutations in several components of the mTOR pathway are also implicated in ASD (Bourgeron 2009), suggesting that normal function of this intracellular signaling pathway is necessary for proper synaptic transmission and neuronal activity. The mTOR pathway is critical for protein synthesis, cellular proliferation and growth (Hershey et al. 2012; Laplante & Sabatini 2012). Activated tyrosine kinase receptors recruit and activate phosphoinositide 3-kinase (PI3K), which converts phosphatidylinositol (4,5)-biphosphate (PIP₂) to phosphatidylinositol (3, 4, 5)-triphosphate (PIP₃). PIP₃ recruits many proteins to the membrane through pleckstrin homology domains, including PDK1 and the serine/threonine kinase AKT, an important downstream effector of PIP₃. Phosphatase and tensin homolog located on chromosome 10 (PTEN) is a lipid and protein phosphatase that negatively regulates Akt activity by working in opposition to PI3K, converting PIP₃ back to PIP₂ (Maehama & Dixon 1998; Stambolic et al. 1998). Akt is fully activated following phosphorylation by PDK1 and mTOR complex 2 (mTORC2), one of two complexes that involves the evolutionarily conserved serine/threonine kinase, mammalian target of rapamycin (mTOR) (Bayascas & Alessi 2005; Sarbassov et al. 2005). Phosphorylated Akt is involved in a wide range of cellular processes, including cell proliferation, survival and growth, via a myriad of downstream signaling proteins, including TSC1 and TSC2. Activated Akt inhibits a protein complex composed of TSC1 (hamartin) and TSC2 (tuberin) through phosphorylation (Huang & Manning 2009; Tee et al. 2002). This disinhibits the GTPase Rheb and actives mTOR complex 1 (mTORC1) (Sato et al. 2008). Once activated, mTORC1 upregulates protein synthesis by phosphorylating several key proteins, including S6 kinase 1 (S6K1), which subsequently regulates Fragile X mental retardation protein (FMRP) through phosphorylation (Narayanan et al. 2008). Several key proteins, such as PTEN, TSC1, TSC2 and FMRP, are strongly implicated in subsets of ASD cases, and are discussed below. However, additional components of the mTOR pathway are also under investigation for their contribution to ASD-related deficits, such as the mTOR substrate eIF4E (Gkogkas et al. 2013).

PTEN mutations are most often associated with a variety of hamartoma syndromes, including Cowden syndrome (Eng 2003), which are characterized by benign focal malformations. PTEN was identified as an ASD candidate gene after several case and prospective studies revealed an association in individuals with ASD and macrocephaly (enlarged head size) (Butler et al. 2005; Buxbaum et al. 2007; Herman et al. 2007; McBride et al. 2010; Varga et al. 2009). Modeling PTEN mutations in mice has focused on two distinct strategies: conditional KO mice, in which distinct cell types lack the gene and protein product, and heterozygous null mice that possess constitutive haploinsufficiency. Conditional KO mice with a Pten deletion restricted to a subset of post-mitotic hippocampal and cortical neurons exhibited deficits in social interaction, sociability and preference for social novelty as well as impaired performance in the Morris water maze and macrocephaly (Kwon et al. 2006; Zhou et al. 2009). Similarly, heterozygous KO mice with a constitutive Pten mutation exhibited macrocephaly and deficits in sociability (Clipperton-Allen & Page 2014), but these sociability impairments were limited to female mice in one study (Page et al. 2009). Additionally, Allen, Page and colleagues also showed that Pten heterozygous mice exhibited elevated repetitive behaviors (i.e. digging, self-grooming). To date, the results obtained with Pten mutant mice suggest that Pten dysfunction leads to several ASD-relevant behaviors, although more extensive behavioral characterization of the various Pten conditional mouse models would further elucidate this phosphatase’s contribution to ASD.

Tuberous sclerosis is a genetic condition resulting from mutations in the TSC1 or TSC2 gene, which negatively regulate mTOR activity (Curatolo & Maria 2013). Individuals that carry mutations in either TSC gene have a higher than expected occurrence of ASD-like features (50%) or an ASD diagnosis (29%) (Curatolo et al. 2010). Tsc1 heterozygous mice exhibited deficits in social interactions, hippocampus-dependent contextual fear conditioning, and hidden platform Morris water maze tasks (Goorden et al. 2007). Social deficits were not identified in Tsc2 heterozygous mice, except when the mice were also treated prenatally with Poly I:C, a maternal immune activation model (Ehninger et al. 2012). However, Tsc2 mutations did produce learning and memory deficits in Morris water maze hidden platform performance and contextual fear conditioning (Ehninger et al. 2008). A dominant negative Tsc2 mutation that binds to Tsc1 and inhibits its activity also led to reduced social interactions and a reduced preference for social novelty, although these mice exhibited normal sociability in the 3-chambered social approach task (Chevere-Torres et al. 2012). A specific Tsc1 disruption limited to cerebellar Purkinje cells produced deficits in sociability, repetitive behavior and cognitive flexibility (Tsai et al. 2012). Similarly, Tsc2 disruption limited to cerebellar Purkinje cells replicated deficits in sociability and social preference, which were accompanied by increased repetitive behaviors (Reith et al. 2013). The study of these Tsc1 and Tsc2 mutant mice has provided knowledge as to how these single gene mutations can model some ASD-relevant phenotypes and contribute to the overall ASD phenotype.

Fragile X syndrome (FXS) is caused by a hypermethylated CGG repeat expansion in the FMR1 gene that leads to a drastic reduction in its protein product, FMRP. FMRP is an RNA binding protein that has been implicated in regulating protein expression (Bagni & Greenough 2005; Chen & Joseph 2015; Crawford et al. 2001; Kazdoba et al. 2014). Individuals with FXS are characterized by intellectual disability and a variety of physical abnormalities, and frequently display social dysfunction, anxiety and repetitive behaviors (Berry-Kravis et al. 2002; Lightbody & Reiss 2009). The diagnostic criteria for ASD commonly occur in individuals with FXS, with recent estimates ranging from 18% to 47% (Clifford et al. 2007; Demark et al. 2003; Hatton et al. 2006; Kaufmann et al. 2004; Rogers et al. 2001). Conversely, the proportion of FXS in the ASD population has been estimated at 3–8% (Cohen et al. 1991; Fombonne et al. 1997). The first and most widely tested mouse model of FXS is the Dutch-Belgian Consortium Fmr1 KO mouse, which has been maintained on multiple background strains. Recent work with the Fmr1 KO mouse has evaluated ASD-relevant phenotypes, including social deficits and repetitive behaviors. Fmr1 KO mice on a C57BL/6 (B6) background exhibited decreased sociability in some cases (Dahlhaus & El-Husseini 2010) or decreased sniffing during a social approach test when maintained on a B6/FVB hybrid background (McNaughton et al. 2008). In contrast, other studies revealed normal sociability and reduced preference for social novelty in Fmr1 KO mice maintained on FVB and B6 backgrounds (Liu & Smith 2009; Pietropaolo et al. 2011). In tests of direct social interaction, Fmr1 KO mice on a B6 background showed deficits in social interaction in some studies (Mineur et al. 2006), but as with much of the Fmr1 mouse behavioral literature, there are caveats to these behavioral impairments (Kazdoba et al. 2014). Paylor and colleagues demonstrated increased social interaction in Fmr1 KO mice on a B6 background (Spencer et al. 2005, 2008). Importantly, the Paylor team presented data interpreted as higher social anxiety in these mice. Fmr1 KO mice also exhibited elevated repetitive behaviors, another core symptom of ASD. Specifically, Fmr1 KO mice on B6 or hybrid backgrounds had elevated levels of self-grooming, but not when they were maintained on an FVB background (McNaughton et al. 2008; Pietropaolo et al. 2011). Fmr1 KO mice on B6 and B6 hybrid backgrounds also had higher levels of marble burying (Gholizadeh et al. 2014; Spencer et al. 2011; Veeraragavan et al. 2012). Intellectual disabilities which characterize FXS have been evaluated in a variety of cognitive tasks. In some reports, Fmr1 KO mice demonstrated cognitive deficits consistent with the intellectual impairments that characterize FXS. Deficits have been observed in contextual, cued and trace fear conditioning and/or context discrimination in Fmr1 KO mice on B6 and sighted FVB backgrounds (Auerbach et al. 2011; Ding et al. 2014; Paradee et al. 1999; Zhao et al. 2005). However, many other reports have described normal cognitive abilities in Fmr1 KO mice maintained on B6, FVB/129 hybrid, and albino B6 backgrounds, including fear conditioning (Baker et al. 2010; Dobkin et al. 2000; Peier et al. 2000; Uutela et al. 2012; Van Dam et al. 2000). Similarly, spatial navigation and reversal deficits were observed during Morris water maze testing in Fmr1 KO mice maintained on B6, albino B6 and FVB backgrounds in some labs (Baker et al. 2010; D’Hooge et al. 1997; Kooy et al. 1996; The Dutch-Belgian Fragile X Consortium et al. 1994) but these impairments were not seen in other studies (Paradee et al. 1999; Uutela et al. 2012; Yan et al. 2004). Thus, although the gene and its product are absent in Fmr1 mice, behavioral phenotypes relevant to the human syndrome and to ASD appear to be variable, with phenotypes potentially dependent on a variety of methodological and environmental factors. While genetic background is one potential factor, there does not appear to be a clear segregation of behavioral outcomes on the B6 vs. FVB backgrounds, either in the social and repetitive behavioral domains most relevant to autism, or in the cognitive domains most relevant to FXS. In contrast to the inconsistent behavioral literature on Fmr1 mice, investigations of neuroanatomical, electrophysiological, genetic and biochemical phenotypes in Fmr1 KO mice have allowed researchers to gain considerable insight into biological mechanisms underlying Fragile X syndrome.

Conclusions

The summary above and in Table 2 provides descriptions of behavioral and biological phenotypes in representative genetic mouse models of ASD. The small subset of genetic mouse models included herein focuses primarily on risk genes that mediate the formation and strengthening of synapses, and postsynaptic downstream signaling through the mTOR pathway. Our selection is presented for its possible usefulness in conceptualizing a cluster of genes with potentially interrelated actions through synaptic and postsynaptic intracellular mechanisms. The appeal of this convergence concept, proposed by many autism researchers (Delorme et al. 2013; Geschwind & State 2015; Silverman & Crawley 2014; Spooren et al. 2012), includes the possibility of developing pharmacological treatments for ASD that act through impaired synaptic mechanisms, perhaps using compounds repurposed from other uses involving synaptic dysfunction. Several intriguing preclinical studies with mouse models of ASD indicate improvements in social behaviors and/or reductions in repetitive behaviors and/or amelioration of cognitive deficits after pharmacological treatments. Promising results from pharmacological interventions in mouse models of ASD and FXS include mGluR5 antagonists (Michalon et al. 2012; Silverman et al. 2012; Tian et al. 2015), GABA agonists (Han et al. 2014; Henderson et al. 2012; Silverman et al. 2015), rapamycin (Burket et al. 2014; Ehninger & Silva 2011; Zhou et al. 2009), d-cycloserine (Burket et al. 2013; Yadav et al. 2012), BDNF and ampakines (Lauterborn et al. 2007; Silverman et al. 2013), IGF-1 (Bozdagi et al. 2013) and oxytocin (Huang et al. 2014; Meziane et al. 2014; Modi & Young 2012; Penagarikano et al. 2015). Descriptions of additional mouse models of ASD are available in several other recent review articles (Ey et al. 2011; Kas et al. 2014; Silverman & Crawley 2014).

Caveats abound for the methods and interpretations of mouse behavioral phenotypes relevant to the core symptoms of autism. The first is genetic background. Just as humans with the same mutation may present with different symptoms, possibly due to protective or susceptibility genes in their genetic backgrounds, mice present with different phenotypes when a mutation is bred into different inbred strains, such as C57BL/6J, FVB/NJ and substrains of 129, each of which has its own idiosyncratic behavioral traits (Crawley et al. 1997). Varying behavioral phenotypes have been reported for mutant mouse models of autism and FXS, as described above, when outcomes of the mutation were placed on divergent backgrounds and directly compared (Moy et al. 2009; Pietropaolo et al. 2011; Spencer et al. 2011). The second is experimental design. Employing large Ns and using WT littermates as controls are essential to avoid over-interpretations of phenotypes which were actually caused by environmental influences that affect mouse behaviors. The third is statistical analysis of behavioral data. Two-way ANOVAs are often required, in which treatment is one factor and genotype is the other factor, rather than simple t-tests. Stringent post-hoc tests, such as Newman-Keuls, Dunnett’s and Tukey’s, avoid the false positives that may be obtained from more forgiving post-hoc tests such as Fisher’s LSD. Statistical comparisons must match the original experimental design. For example, our 3-chambered social approach task is not sensitive enough to compare the absolute number of seconds spent with the novel mouse across genotypes, or across drug doses. Data on social approach are correctly analyzed with a simple paired t-test or equivalent, which compares time with the novel mouse vs. time with the novel object within genotype only, or within a drug dose only, to provide a yes-or-no outcome measure, i.e. sociability or absence of sociability. The fourth is corroboration within the behavioral domain. Conducting two or more assays that interrogate the same behavior (e.g. at least two social tasks, or two repetitive behavior assays, or two anxiety-related tests, or several learning and memory tasks that tap into different cognitive domains) allows for stronger interpretations and generalizability of a finding within a given domain. Including relevant controls, such as open field exploration for tasks that require locomotor activity, and pain or foot shock sensitivity for fear conditioning, will ensure that a motor or sensory artifact is not the cause of a significant effect in the behavioral domain of primary interest. Large numbers of mouse behavioral assays relevant to the diagnostic and associated symptoms of ASD are available to choose from (Crawley 2012; Silverman et al. 2010a), along with a wealth of relevant control tests (Crawley 2007).

One additional caveat, arguably the most important, is replicability. While all scientific discoveries require replication, behavioral findings require extra attention to reproducibility because of the strong influence of the various environmental factors on mouse behaviors. Stressors, such as construction noise or rough handling, can greatly affect scores on sensitive behavioral assays for anxiety-related, social and cognitive phenotypes. These types of variable findings remain primarily anecdotal, because they are difficult to document and publish, although they are common knowledge among behavioral neuroscientists. Loss of behavioral phenotypes across breeding generations may occur due to attritional loss of the mutation or drift in background genes. Basic methodological issues, such as the age of the mice at testing, composition of the chow diet, properties of the testing equipment and testing room, social testing in single vs. mixed genotype dyads, cleaning of the equipment between subject mice, and simple random chance beyond P <0.05, may create conditions that yield a positive finding that cannot be replicated. Following published procedures by expert behavioral neuroscientists can help alleviate some of these concerns. However, the best strategy to ensure high replicability is for the research team to conduct the entire experiment again in a separate cohort of mice prior to publishing. Further, to confirm the ultimate strength of a finding for the scientific community, several different laboratories should conduct essentially the same experiment with the same line of mice. When a social deficit in a genetic mouse model of ASD is replicated across cohorts and by many labs, the strength of a finding is ensured. Highly robust, well-replicated behavioral phenotypes relevant to the symptoms of ASD in a genetic mouse model can then effectively inform our understanding of the role of that gene in the symptomotology of ASD and serve as preclinical outcome measures for therapeutic discovery.