Can Autism Speak to Neuroscience?

Autism is a severe neurodevelopmental disorder with early childhood onset. The symptoms, first described by Leo Kanner in 1943, produce significant impairments in social, communicative, cognitive, and behavioral functioning. These symptoms, which persist throughout life, disrupt families and lead to significant disability. Only in the last decade or so have we started to undertake the studies needed to understand etiology. From these, we know that, despite disagreements about prevalence, autism spectrum disorders may affect as many as 73 per 10,000 (Kadesjo et al., 1999; Fombonne, 2006). We also know that genes play a greater role in the risk of autism than in any other common neuropsychiatric disorder, with a reported monozygotic-twin concordance/dyzygotic-twin concordance ratio of ∼25 and heritability ≥90% (Bailey et al., 1996; Risch et al., 1999). Although genes exert a strong influence in aggregate, it is clear that autism is genetically complex, meaning that multiple genes of relatively small effect must interact to produce risk in combination with nongenetic factors.

Despite recent neurobiological findings (as described below) and the clear importance of genes, no defining set of pathophysiological mechanisms have been unambiguously elucidated nor have risk genes been unambiguously identified. This situation is similar to that for other major neuropsychiatric disorders such as schizophrenia, bipolar disorder, and major depression, and reflects several factors. Perhaps most important is the difficulty of studying brain and behavior.

Getting started on the basic neuroscience of autism

A satisfactory understanding of autism will require advanced knowledge of brain structure and function at multiple levels of analysis. Given a lack of convincing animal models, much of what we know about the neurobiological underpinnings of autism have been derived from clinical research on affected children and their immediate families or by postmortem analysis of brains of affected individuals. Clinical investigation has generated a rich tapestry of descriptive information about the autism phenotype and has generated multiple hypotheses about etiology and pathophysiology. For example, postmortem and magnetic resonance imaging (MRI) studies have identified abnor-malities in several major cortical and subcortical brain structures in autism (Courchesne et al., 2005), and neuropsychological studies have implicated specific impairments in executive functioning and in processing social and emotional information (Dawson et al., 2005). This body of research is the first step in identifying what fundamentally goes awry in the developing brain of children with autism. Ultimately, basic neuroscience research may provide the most clues to understanding etiologic and pathophysiologic mechanisms.

The neurobiology of social behavior represents a particularly promising avenue of inquiry. Indeed, the social impairments are what distinguish autism from other developmental disorders. These include difficulties in reciprocal social interactions, abnormalities in nonverbal social communication such as eye-to-eye gaze, and abnormal emotional content in spoken language (Joseph et al., 2002; Baron-Cohen and Belmonte, 2005; Grice et al., 2005; Johnson et al., 2005; Dapretto et al., 2006). Underlying these symptoms and signs are apparent abnormalities in the neural underpinnings of social cognition (Dawson et al., 2005; Bolte et al., 2006). Studies of emotional processing and social cognition are a recent development in neurobiology and have yet to develop a critical mass of investigators.

Several neuropathological features have been reported, e.g., Purkinje cell loss in the cerebellar cortex (Bauman and Kemper, 2005). Head circumference and MRI studies have implicated neonatal brain undergrowth followed by rapid and excessive postnatal brain growth (Courchesne and Pierce, 2005). Given well replicated observations of elevated platelet serotonin in a subset of autistic probands and first-degree relatives, considerable excitement has been generated by recent work (McCauley et al., 2004; Sutcliffe et al., 2005) implicating multiple rare alleles at the serotonin transporter locus (SLC6A4) in the etiology of autism.

Candidate neural systems, circuits, and pathways

We are at a critical period in autism research. The study of emotion and social cognition are beginning to mature along with ever better structural and functional neuroimaging tools (Dapretto et al., 2006). Human genetics is entering an era in which complex phenotypes can be tackled effectively; maps of human variation that will permit whole genome association studies (Risch and Merikangas, 1996; Clark et al., 2005; Palmer and Cardon, 2005) and inexpensive sequencing of human chromosomes (Altshuler et al., 2000) are becoming a reality. Important clues to pathogenesis should eventually come from genetics; particular attention will need to be focused on identifying the neural systems that are affected in this disorder. Sophisticated use of animal models should become ever more informative, especially as associated genetic variants are identified. Models will include rodents, flies, and zebrafish (Shahbazian et al., 2002; Tropepe and Sive, 2003; McBride et al., 2005); these may be applied to study key molecules, pathways, and circuits involved in etiology and pathophysiology. Understanding autism requires approaches that integrate basic studies that extend from genetics, molecular and cellular biology, developmental biology, neurophysiology, and neuroanatomy to systems and cognitive neuroscience.

There are several neural systems that are natural candidates, including systems that regulate social behaviors, higher-order cognition, communication, fear and anxiety, and attention. These systems involve circuits that integrate information processed in the cerebral cortex (including the prefrontal cortex), basal ganglia, thalamus, amygdala, hippocampus, hypothalamus, and cerebellum and are modulated by subcortical monoamingergic systems.

Understanding the mechanisms that regulate patterning and plasticity of the cerebral cortex are likely to have important implications for understanding how cortical areas in the autistic brain form, make connections with other brain regions, develop processing networks, and adapt to changes in input (Hensch, 2005; Sur and Rubenstein, 2005). Furthermore, given that ∼30% of autistic people develop epilepsy, a syndrome characterized by excessive cortical excitation and decreased signal-to-noise (Gillberg and Billstedt, 2000), it is possible that abnormalities in the ratio of excitation to inhibition in key neural circuits, especially in the cortex and hippocampus (Hussman, 2001), may contribute to the pathogenesis of autism (Rubenstein and Merzenich, 2003; Levitt et al., 2004). Although many cases of autism may be characterized by increased excitation, there are also disorders with autistic features, such as Rett syndrome, a disease caused by loss of function mutations in the MeCP2 gene, in which the inhibition may dominate. Mice with a null allele of MeCP2 appear to have increased ratio of inhibition/excitation (Dani et al., 2005).

In addition, these mice show increased expression of the Dlx5 transcription factor (Horike et al., 2005). The Dlx gene family has a central role in regulating the development and function of forebrain inhibitory neurons (Panganiban and Rubenstein, 2002; Cobos et al., 2005). Recently, five nonsynonymous Dlx2 and Dlx5 mutations have been identified in autistic probands (Hamilton et al., 2005). Although it is premature to conclude that there is a causal link that connects MeCP2, the Dlx genes, and autism, this line of investigation illustrates one current approach that is aimed at elucidating the molecular and cellular basis for some forms of autism.

Given that two of the hallmarks of autism are a deficit in social cognition and pronounced fear and anxiety, it is important to scrutinize the function of limbic forebrain structures including the amygdala and hypothalamus. Fear conditioning as a model system for studying emotional learning and memory, and elucidation of the neural circuitry for fear learning, may inform studies of autism (LaBar and LeDoux, 2006). Likewise, it is important to explore the role of specific nuclei and subnuclei in the amygdala on component processes of social behavior (Zirlinger and Anderson, 2003; Schumann et al., 2006).

Cerebellar connections with the limbic system and with the cerebral cortex have been posited as mediating abnormal activity that underlies the motor and sensory abnormalities, language difficulties, socioemotional difficulties, and disordered cognitive processing. Cerebellar pathology could lead to abnormal activity in cerebello-limbic and cerebello-thalamo-cortical pathways, which in turn could be expressed as autistic behavior (Dum and Strick, 2006).

Autism is associated with a particular profile of impaired and spared language abilities, as well as nonverbal social deficits. It has been argued (Walenski et al., 2006) that the status of these abilities can be partly explained in terms of dependence on neural systems that subserve nonlanguage functions, including procedural and declarative memory. Future directions include innovative paradigms to examine integrative explanatory theories that attempt to account for these deficits in the broader context of brain and behavior in autism.

Given that cognition, social interaction, sensory function, emotional behavior, and language are disrupted in autism, and given that these functions depend on integrative mechanisms within the prefrontal cortex, it is logical to postulate that there are abnormalities in the structure and function of this brain region (Price, 2006). An avenue for future inquiry is the investigation of how clinical expression depends on the interaction with neural systems in the prefrontal cortex, even if primary deficits in autism are in other brain regions. Another potentially exciting avenue for research is to investigate possible changes in the neuronal circuitry of the thalamus and neuromodulatory systems arising in the brainstem core, basal forebrain, and posterior hypothalamus, which act on glutamatergic thalamocortical neurons, GABAergic thalamic reticular neurons, and local-circuit GABAergic interneurons, as well as functional dysfunction implicating attention and sensory gating (Steriade, 2006).

Future studies will require the critical integration of findings from basic neuroscience paradigms with those from genetic studies. This may provide insights on the nature of risk genes and the timing of their expression during development, thereby providing useful clues to relevant environmental factors. There is increasing appreciation for the importance of gene/environment interactions that extend from infection susceptibility to the effect of sensory experience on chromatin structure (Meaney and Szyf, 2005).

Shared pathophysiology in autism, fragile X syndrome, and Rett syndrome?

Symptomatic commonalities among autism and other pervasive developmental disorders like Angelman’s and Rett syndromes, tuberous sclerosis, and fragile X syndrome may reflect an overlap in affected neural circuits and pathways and perhaps shared pathophysiologic mechanisms. Fragile X syndrome is caused by reduced expression of the fragile X mental retardation protein (FMRP) RNA binding protein whose function is linked to synaptic function. Based on observations made in fragile X, and several other lines of evidence, researchers are currently exploring the possibility that alterations in synapse development and signaling may underlie some forms of autism (Rubenstein and Merzenich, 2003; Levitt et al., 2004). This perspective has been bolstered by the identification of function-altering mutations in some neuroligin genes in a small subset of autistic people. Neuroligin proteins, together with their binding partners (neurexins), regulate the formation of synapses; different neurexin/neuroligin combinations appear (Jamain et al., 2003) to participate in specifying whether new synapses are assembled into excitatory or inhibitory synapses (Graf et al., 2004; Chih et al., 2005; Chubykin et al., 2005; Cline, 2005).

Studies of synaptic plasticity in the hippocampus of the Fmr1 knock-out mouse suggested a connection between metabotropic glutamate receptor (mGluR) signaling and the fragile X clinical phenotype (Huber et al., 2002). The so-called “mGluR theory of fragile X” proposes that exaggerated signaling via group 1 mGluRs is a consequence of the loss of FMRP from neurons (Bear et al., 2004). Increased translation of group 1 mGluRs in dendrites, which would lead to overactive signaling by these receptors, may contribute to slowed altered synaptic function and other symptoms of fragile X. Drugs that target mammalian mGluR signaling rescue in flies behavioral and other deficits that are analogous to human fragile X symptoms (McBride et al., 2005). Such compounds have therapeutic potential for fragile X children, many of whom have autistic features (Bear, 2005). This raises the intriguing possibility that new therapeutics developed to treat fragile X also may have efficacy in treating aspects of autism (Moldin, 2005).

Increased funding for cancer research resulted in an avalanche of data that greatly accelerated our understanding of its underlying mechanisms and established a foundation for current discoveries in cancer biology. Foundations of knowledge are now in place to warrant an increased investment in autism research. We anticipate that in the next 10 years the result will be a marked improvement in the understanding and treatment of this devastating disorder.

Two reviews in this week’s issue examine the rapidly expanding interest in autism research in the neuroscience community. Moldin et al. provide a brief prospective on the overall state of research in autism. DiCicco-Bloom and colleagues summarize their presentations at the Neurobiology of Disease workshop at the 2005 Annual Meeting of the Society for Neuroscience.