This issue consists of six articles reporting on various mouse model studies of autism spectrum disorder (ASD). Each proposes to examine and inform on an important aspect of gene–behavior or gene–brain mechanisms relevant to ASD. Each topic is thought provoking for clinicians, and illuminates some of the mechanisms that researchers are considering to explain the etiology and pathophysiology of ASD and, importantly, its characteristic inherent heterogeneity. I believe that these articles will help many clinicians understand the relationship of the numerous reports of genes in autism to causation, the role of variable genetic background in syndrome heterogeneity, and the role of potentially diverse developmental neurobiological mechanisms in pathophysiology. The key issues considered in these articles are: what mechanisms from gene to brain account for the shared features of autism across all cases, and what mechanisms account for the notable heterogeneity in individual features of the syndrome across cases of otherwise comparable severity. These articles also introduce genes related to syndromic autism, cases with an identifiable phenotype that is associated with a gene or chromosomal disruption, and genes associated with “idiopathic” cases of autism. The discovery of related genes across these two categories has narrowed the conceptual gap between such cases.

Studies relying on animal models always raise the issue of how valid the results and conclusions based on animals are for humans. In autism, the issue is magnified because ASD is a disorder of higher cortical functions unique to humans. A key question is whether the more elementary behavior exhibited by animals in these domains is a valid analogue of the human manifestations, since assertions of validity are often but not always supported by such analogies. Some animal models target a neuropathological feature. The issue then becomes whether the histology is equivalent, whether it is produced by the same processes as in the human condition, and whether the functional consequences are analogous.

The issue of validity is a concern that can actually go both ways as scientists searching for candidate drugs well know, i.e., what works in animals may not work in humans but likewise what may work in humans may not work in animals. That is, “false-positive results” and “false-negative results” are both problematic with animal models of human conditions. The approach to such issues involves typical scientific principles. Animal models are not expected to provide a definitive answer, i.e., have appropriate expectations, but rather clues that must be interpreted within a matrix of findings and with reservations, i.e., understand what you are seeing at a mechanistic level with awareness of species differences. Regardless of limitations, animal models are providing the greatly needed opportunity to rapidly investigate issues of importance to translational medicine and that would be impossible or very challenging to address in studies of affected individuals. Animal models are also being used to rapidly discover molecular mechanisms and to assess potential treatments.

With all of that said, what stands out from these studies for clinicians? The study on “Social Peers Rescue Autism-Relevant Sociability Deficits in Adolescent Mice” by Yang et al. involved an inbred strain of mice (BTBR) that exhibits behavior deemed relevant to the three categories of diagnostic symptoms for ASD and an inbred strain of mice that exhibit high sociability and otherwise mouse-typical patterns of communication and activity (B6). Their previous study reported that cross-fostering of BTBR pups with B6 mothers did not improve sociability in the BTBR mice in adulthood, indicating that early maternal care was not a major factor in determining outcome in these mice (Yang et al., 2007). The present strategy involved putting the BTBR mice at weaning with the B6 mice for 20 or 40 days, which resulted in significant improvement in social approach behavior. The 20-day peer intervention was as effective as the 40 days. Repetitive behavior was not changed. So what might these findings mean for ASD? All kinds of interesting inferences could be drawn from these studies ranging from: the behavior of the mother mouse did not cause the social difficulties in her offspring, nor could a social mouse mother change social outcome in a genetically low social line of mouse pups. The cross-strain differences in mouse behavior, the fidelity of behavior from generation to generation of these mice, and its persistence regardless of early maternal behavior speak to the strong influence of genes on behavior. These studies also suggest that there is a period in mouse development after weaning when their brain is susceptible to the beneficial influence of the behavior of peer mice and in particular peer interactions in a naturalistic setting. This phenomenon could represent an example of timing intervention with a critical period in brain development. If the next study introduces the social peer to the nonsocial peer in adulthood, an improvement in BTBR mice might mean a life-long potential for change and the role of learning mechanisms. Adult mouse response to intervention would also be reminiscent of the changing concepts in humans about brain plasticity after brain injury in adults. The prior reports of behavioral rescue of adult mice with syndromic autism have supported the possibility of substantial improvement in adulthood in some cases and situations.

The recent very early intervention approach by Dawson et al. (2010) suggests that intervention beginning at or near 12–14 months may have a great impact on reducing emerging manifestations of ASD. Does this mean that even earlier would be better or does it mean that there is a window of opportunity largely confined to when mechanisms underlying manifestations are active? This will be a very hard question to answer in infants.

Observations based on the study of infants with an older sibling with autism have supported the principle of neurological windows of developmental opportunity linked to stages of brain development. These studies have reported the absence of manifestations of autism at 6 months of age and only subtle signs in motor, sensory, and visual interests between 9 and 12 months. The manifestations most typically associated with autism, including cognitive decline, temperament and activity emerge between 12 and 24 months. This timing of the emergence of dramatic signs and symptoms of autism is consistent with the period in brain development when the brain evolves the structure and function to support higher order abilities. Would intervention between 0 and 6 months prevent the emergence of autism or improve outcome even more? How early can one provide remediation and expect an impact? Before deficits emerge and thus before the affected skills emerge? These are interesting and important questions to consider.

Researchers involved in studies of infant siblings of children with autism have reported that parents are frequently and quickly implementing the training and skills they acquired for their first child with autism to their second affected child. It will be interesting to see if the change in parenting with a second child alters outcome, though that effect will be very difficult to disentangle from the complexity of other relevant factors. This complexity is exactly why researchers are attempting to isolate factors and to glean evidence about the impact of intervention in animal models; analogous studies in humans would take large numbers of families, decades, and most of the resources for autism.

The study “Modifying Behavioral Phenotypes in Fmr-1 KO Mice: Genetic Background Differences Reveal Autistic-Like Response” by Spencer et al. highlights an extremely important principle to understanding ASD heterogeneity and the broader autism phenotype in family members. This study demonstrates that the manifestations of Fragile X syndrome in a knockout mouse vary depending on the genetic background of the mouse strain. This finding highlights that autism is the result of the interactions of multiple genes and that an individual's genetic background plays a significant role in the presence or absence of manifestations. Many research studies publishing findings implicating genes in autism focus on the altered gene, which is obviously important, but omit the role of familial inheritance and other spontaneous changes in the individual's DNA that also may play a major role in whether or not and to what extent manifestations of autism occur. This is likely what the genome wide scans are telling us—that there is an interplay between rarer, “pathogenic” mutations and an underlying genetic background of relatively common polymorphisms. A broad ASD phenotypic spectrum is the result. Such interaction would explain how members of the same family could display no manifestations, a broad range of severity of ASD manifestations, or intellectual disability without autism while presumably having the same “ASD gene.” The present study provides a well-organized system for modeling the genetic interactions underlying behavioral phenotypes while highlighting the enormous complexity involved in applying such models to genetic syndromes in humans. It is unlikely that anyone reading this article will forget this example of gene×gene interactions in influencing disease expression.

The article “The Autism Risk Genes MET and PLAUR Differentially Impact Cortical Development” by Eagleson and colleagues presents yet another important dimension of thinking about the pathophysiology of autism while demonstrating the utility of mouse models for assessing the biologic plausibility of a growing number of genetic candidates for ASD risk. Specifically, this study is both a detailed analysis of MET expression in the developing and adult mouse brain as well as an examination of interneuron populations in a conditional MET knock-out mouse model. Specifically, the authors find that while PLAUR function is known to be necessary for interneuron differentiation and survival, eliminating MET from the ventral telencephalon (the birthplace of practically all interneurons in the mouse) has no discernable effect on these populations. Furthermore, MET is not expressed in cortical and hippocampal interneurons at any time during development. Instead, MET expression is limited to cortical projection neurons and is particularly focused within the maturing axon.

This study further refines and clarifies seeming inconsistencies in previous research which is reviewed and synthesized as contextual background. Specifically, the argument for the potential role of dysfunction of either of these two genes in disturbing the balance between excitatory and inhibitory influences is strengthened. Such an inhibitory–excitatory imbalance has long been hypothesized as a plausible mechanism of cortical dysfunction in ASD, particularly in cases associated with seizures. The properties of these genes suggest that this could well be a mechanism in cases resulting from genes impacting the development of the circuitry between projection neurons and interneurons. Seizures have also been identified in ASD with intellectual disability and epilepsy in cases associated with migrational disturbances, cortical heterotopias and the Caspr gene, suggesting that there may be several different mechanisms that result in epilepsy in ASD. As the authors point out, differential lineage-, region-, and time-specific alterations in the expression of various candidate genes likely underlies the great clinical heterogeneity observed in ASD.

As a final point, the well-organized review of research pertaining to altered MET and PLAUR function in ASD illustrates the breadth and depth of studies necessary to link genetic candidates with plausible neurobiological developmental mechanisms—a difficult gap to bridge. Modeling molecular dysfunction and examining expression in experimental animals are integral to this process and serve to complement and guide genetic analyses and functional assays.

A fourth study by Sakurai et al. on “Haploinsufficiency of GTF2I, a Gene Deleted in Williams Syndrome, Leads to Increase in Social Interactions” focuses on the hypersociability of individuals with Williams syndrome, which early in life appears to be a strength but at older ages emerges as indiscriminate and inappropriate social behavior. These authors and others have dissected out a single behavioral aspect of a syndromic condition (Williams–Beurin syndrome, loss of one copy of a chromosome region containing about 28 genes) and associated it with a single gene loss (loss of a single copy of the gene GTF2I) through genotype–phenotype correlation studies. They then generated a mouse that nicely models the targeted social behavioral abnormality. It is not common to have such a straightforward link between a gene(s) and a behavior, and it will be very interesting to determine the precise mechanisms by which this gene impacts brain development to affect social behavior. The results of the next step in this research—generating a cortical conditional knockout model that avoids the embryonic lethality due to neural tube failure to examine the effects of the complete loss of the protein on cortical anatomy, function, and development—should be helpful in understanding this genotype–phenotype relationship and its implications for the hypersocial behavior in Williams syndrome.

The next study “Absence of Preference for Social Novelty and Increased Grooming in Mice Lacking the Integrin B3 Receptor Subunit” by Carter et al. seeks to explore the potential role of serotonin in ASD by examining the influence of the subunit gene that is a quantitative trait locus for whole blood 5-HT levels. They examined the behavioral differences between mice with the absence or diminished expression of this gene to litter mates and found that these mice had no differences in sociability. In a novel social situation, however, they did not show an increased interest in novel mice and instead exhibited increased grooming. This altered behavior is not typical of “ASD mice” but the question is interesting because of the known functions of beta 3 integrin at the synapse. One of the two popular genetic theories about autism hypothesizes that autism is a disorder of synaptic development/maintenance/function. Hence, it is very interesting to explore the impact of known synaptic genes on phenotype in animal models. To clarify the significance of the findings, it will be important to see the results of a conditional knockout in cortical neurons but not in cortical blood vessels. Although infarct or hemorrhage were not observed in these mice, problems with vasculature may still affect cortical development and swamp any effects of intrinsic defect in the neurons.

The final study “Behavioral Profiles of Mice Carrying Synaptic Gene Mutations Associated with Autism Spectrum Disorders” by Ey and colleagues explores the model of ASD as a heterogeneous group of synapsopathies, a model informed by the identification of potentially synapse altering mutations in genome-wide studies of ASD. Specifically, the authors extensively review the literature pertaining to behaviors in mice carrying mutations of these genes, e.g. neuroligins, neurexins and Shanks, and attempt to delineate meaningful phenotypic clusters. They similarly examine models of syndromic ASD, such as the Fmr1 and Mecp2 transgenic mice.

Primarily, their review illustrates the extensive diversity of behavioral abnormalities observed in synaptic mouse models, likely reflected in the marked phenotypic heterogeneity in ASD patients, at least in part. However, as the authors point out, the variation in testing both within and between studies may be magnifying this to some extent. Despite this, a few interesting trends are noted. For example, mutations in Nlgn and Shank genes tend to produce similar abnormalities of social and vocal behaviors coupled with preserved or even enhanced learning, a pattern reminiscent of high-functioning ASD in humans. Mouse models of syndromic ASD, on the other hand, tend to exhibit differences in body weight, fear conditioning, some social behaviors, seizure propensity, stereotypy, and locomotion. Interestingly, this fundamental divide between syndromic and nonsyndromic ASD mouse models (particularly with respect to increased neurological involvement) is also consistent with patterns observed in humans.

These findings highlight the potential utility of mouse models for examining, in vivo, the behavioral consequences of certain ASD-related mutations over the course of development. While clearly widely divergent from humans, particularly in the areas of higher cognitive functioning most noticeably impacted in ASD, mouse models can clearly be used to investigate equally informative attributes such as abnormal social interaction, deficits in communication, and repetitive behaviors. Additionally, specific mouse models have already, as the authors point out, guided the development of potentially effective pharmacologic, behavioral, and environmental therapies.

In summary, after reading these articles, clinicians will have a much clearer appreciation of current concepts and evidence about the etiology and pathophysiology of autism and its heterogeneity. It is unlikely that they will ever again think that very little is known about the cause of autism. Rather I hope they will come away impressed with the scope of what has been discovered, carrying in mind an integrated model of autism from gene to behavior. They will also be impressed with the complexity of the questions yet to be addressed but also the relentless progress of science in seeking answers using innovative designs and advancing technology.