Breaking into the autistic brain
A prismatic research approach sheds light on the biology of autism spectrum disorders
by Parizad Bilimoria
In one laboratory, nerve cells in a plastic dish glow green and red under a microscope, revealing their shapes and the locations of specific molecules. In another laboratory, mice that have grown up in dark environments are having electrical activity recorded from their brains’ vision centers. In yet another lab, a toddler looks at pictures of faces while his eye movements are tracked by a camera.
These are just a few of the laboratories at Children's Hospital Boston that are trying to understand autism, one of the biggest mysteries in contemporary neuroscience.
Autism is now recognized as a varied group of developmental disorders, known as autism spectrum disorders or ASDs. It is defined at the behavioral level, and its three hallmark features are known: impaired social interaction, communication difficulties and repetitive behaviors.
What’s not clear, says Charles A. Nelson, PhD, Research Director of the Division of Developmental Medicine and the Richard David Scott Chair in Pediatric Developmental Medicine Research at Children’s, is how autism arises and what the brain is like in someone diagnosed with autism, as compared with the typical brain. He and his colleagues are approaching these questions in a variety of ways.
Seeds for study
Some of the greatest advances in autism biology have come from studying a handful of rare neurologic disorders that are caused by a single gene and sometimes include features of autism. Four of the best examples are Rett syndrome, Fragile X syndrome, tuberous sclerosis complex and Angelman syndrome. “It’s not that we think of these disorders as being part of autism,” Nelson explains. “It’s that a sizeable number of the kids with these syndromes wind up looking like they have autism.”
There’s one main reason why single-gene disorders are such a powerful research tool: mouse models. When the affected gene is known, researchers can mutate or delete that gene in mice. This sometimes allows them to mimic the human disorder and track its root cause.
Although different genes are affected in each disorder, some commonalities are coming into focus. Each gene encodes a control knob of sorts—a molecule whose activity influences the production or survival of other molecules in the brain. Researchers believe that these different knobs, in different ways, regulate the intricate web of connections in the brain, and, in particular, the quantity and quality of synapses, the points of contact between nerve cells and the building blocks of brain circuits.
Dysfunction at the junction
In Rett syndrome, which primarily affects girls and is often associated with autism, the affected gene makes MeCP2, a protein that dims the activity of many genes in nerve cells that influence synapse formation or function. The absence of MeCP2 causes dendrites, tree-like structures on neurons that receive incoming messages, to contain fewer spines, or spots for synapses to grow. The synapses that do develop do not function normally.
“We think that the synapses are overly immature,” says Omar Khwaja, MD, PhD, director of the Rett Syndrome Program. “They make weak connections, but these connections stay in a very immature, plastic state and fall apart easily.”
In Fragile X syndrome—associated with autism in at least a third of cases—the problem is somewhat the opposite. The affected gene makes FMRP, a protein that restricts the manufacture of many proteins at synapses. In its absence, dendrites grow wildly and contain more spines. But these spines are longer and thinner than normal—again, a sign of immaturity.
“To use an analogy, FMRP is the gardener who carefully prunes the tree,” explains Jonathan Picker, MD, PhD, director of the Fragile X Program. “If it’s not pruned, the twigs and branches get out of control, the leaves get all tangled up and the tree becomes unhealthy.”
To a lesser extent, Angelman syndrome too can include some of the behavioral symptoms of autism. In this case, the causative gene makes a protein called Ube3a that marks a “hit list” of other proteins for degradation. The synapses of mice lacking Ube3a have defective plasticity—meaning they have trouble tuning their properties to adapt to the activity patterns of the neurons they join together.
Together with Michael Greenberg, PhD, now chair of Neurobiology at Harvard Medical School, Judith Steen, PhD Assistant Professor of Neurology, has recently found evidence suggesting that in the absence of Ube3a, a protein called Arc builds up at the synapse. This buildup appears to contribute to impaired synapse plasticity in Angelman mice.
Sometimes more than one problem occurs at once: impaired synapse plasticity occurs in the Rett and Fragile X mice too, and Angelman syndrome is also thought to affect dendritic spines.
But is “garden variety” autism, for which there’s no single genetic cause, really a disorder of brain connectivity, as these rare neurogenetic syndromes appear to be? The little we know so far lends itself to this hypothesis. Christopher Walsh, MD, PhD, chief of Genetics at Children’s, points out that a number of genes that don’t single-handedly cause autism, but have been linked to an increased risk, turn out to be active at the synapse. [To learn more about Children’s research into the genetics of autism, click here.]
There’s still another way brain connectivity can be disrupted in autism-related disorders. Tuberous sclerosis complex (TSC), a disorder characterized by benign tumors across multiple organ systems, appears to involve abnormal growth of the axon—the long, thin structure that emanates from a neuron and outputs information to other neurons. Axons can travel all the way from one brain region to another in bundles known as tracts. Mustafa Sahin, MD, PhD, director of the Tuberous Sclerosis Program, and Simon Warfield, PhD, director of the Computational Radiology Laboratory, have found that in the brains of TSC patients, about half of whom fall on the autism spectrum, axon tracts are disorganized and structurally abnormal.
“There might be a disconnection in the way the different brain regions are connected—especially brain regions that have to do with language and social cognition,” Sahin suggests. His data substantiate an emerging hypothesis that autism is a “developmental disconnection syndrome” in which long-range connections (carrying information from one brain region to another) are disrupted. [To learn more about Sahin’s work on axon miswiring in TSC and a drug targeting this defect, click here.]
While there’s still much to learn, and while autism isn’t the only disorder that affects brain connectivity, we now have a starting point for understanding what’s different in the brain of a child with autism. It’s not that the neurons are degenerating or dying in large numbers, or that they’re located in entirely the wrong places. It’s that they’re not communicating properly.
But the disconnected brain development may not arise only from anatomic problems of the synapse, axon or dendrites. Autism may also involve a disconnect between the brain and its external environment—an inability of the brain to change properly based on input from the outside world.
Our experiences—what we see or hear, for example—shape our brain connections, most actively during certain windows of development called “critical periods” or “sensitive periods.” Critical periods are timed differently for different brain functions. Takao Hensch, PhD, Michela Fagiolini, PhD, and Chinfei Chen, MD, PhD, of the F.M. Kirby Neurobiology Center, believe these periods are disrupted in autism and other neurodevelopmental disorders.
They and others have long used the visual system as a model to learn how experience molds brain circuits. About a decade ago, Hensch and Fagiolini did experiments in the visual cortex which found that plasticity—the ability to adapt to input from the environment—depends on the balance between two types of neural signals: excitatory and inhibitory. Without proper development of neurons that send inhibitory signals—specifically, a variety called parvalbumin (PV) cells—the critical period for this brain region is not triggered.
Striking a balance
Intriguing new evidence suggests that PV cells are reduced or defective in many autism-related mouse models. Could this mean that the brain’s excitatory/inhibitory balance—and, consequently, critical periods—are altered in neurodevelopmental disorders? That idea would be consistent with what’s been observed in children with ASDs: symptoms are not apparent at birth, but emerge around age two or three, a time when the brain is actively resculpting its connections in response to the child’s experiences, or interactions with his or her environment.
Fagiolini, Hensch and Chen are testing this idea, examining critical periods in the mouse visual system to see how experience shapes circuit development. Fagiolini notes that visual development in Rett mice is initially normal, followed by a sudden regression—mirroring the developmental pattern in children with Rett syndrome.
The mice start out being able to see. But just a couple of weeks after their eyes open, their vision starts to fail. (At around the same time, they start shaking, losing weight and having difficulty coordinating movements, as happens to children with Rett syndrome.) In addition, electrophysiological testing reveals abnormalities in the visual cortex: Neurons are silent when they should be active.
Learning from experience
But, amazingly, this visual regression doesn’t occur if the Rett mice are reared in darkness, depriving them of visual experience, suggesting that experience, and the brain’s ability to process it, could be key to our understanding of Rett syndrome and ASDs in general.
Chen’s lab provides corroborating evidence that Rett syndrome interferes with the brain’s ability to respond normally to experience. Her laboratory studies an earlier point in visual circuitry: the retina’s connection to the thalamus, where sensory information from the eyes is processed before it travels to the cortex. This work has revealed two periods of refinement in this circuit: One occurs independently of visual experience, before the mice even open their eyes, while the other requires light exposure.
It’s this latter phase, requiring visual experience, that seems to be disrupted in Rett mice. Before eye opening, both Rett mice and normal mice and lose excess, unnecessary retina-thalamus connections. After eye opening, the two groups diverge: normal mice stabilize their synaptic refinements, whereas the Rett mice maintain a large number of immature, weak synapses. Chen believes these mice are unable to use sensory experience to stabilize the appropriate synapses.
“A child with autism could be interpreting the world in a completely different way,” Chen suggests. “If you try to break that down at the level of neural circuits or synapses, it could be that you’re emphasizing one input over another, such that your sensory information is incorrect or different. That just veers you off track.”
A therapeutic window
Hensch and his colleagues are now looking at critical periods in other mouse models for ASDs and in other sensory areas of the cortex -- and possibly even in higher-order cognitive functions linked to socialization.
What makes the critical period findings so exciting is that they suggest tangible opportunities for clinical research. There are some commonly prescribed FDA-approved drugs that alter excitatory/inhibitory balance in the brain. These drugs could potentially be used to manipulate critical periods in neurodevelopmental disorders, helping circuits develop more normally without the need to correct the underlying genetic abnormality.
“Gene therapy is probably difficult, if not hopeless, if you have to restore exactly the gene that is defective. In many causes of autism, it is not just one gene,” explains Hensch. “But if it is a common circuit problem, well, there are already a lot of drugs available that adjust excitatory/inhibitory balance—if only we knew when and for how long to give these drugs, and where specifically.”
Nelson’s team, meanwhile, is directly observing brain activity in infants that might be indicative of autism. Using electroencephalograms (EEGs), his team is looking for tell-tale changes in brain activity in infants who have an older sibling with autism. These infants have an approximately 1 in 5 chance of developing autism themselves; if it can be caught early, when their brains are most actively wiring, they may be more responsive to behavioral or other interventions.
While the infant-sibling study is not yet complete, the latest data suggest that certain EEG alterations may provide “functional biomarkers”— non-invasive ways to predict autism risk—at as early as three months of age. One of the clearest differences in infant EEG activity appears to be in the gamma band, a type of brain activity that may reflect how the brain integrates information. Infants at high risk for developing autism have much less gamma activity than those at low risk. “If we think of autism as a disconnection in the brain, gamma may reflect the brain’s ability to put information together,” Nelson says.
This finding dovetails in intriguing ways with the mouse data. Gamma activity is a rhythm thought to be generated by PV cells, the very cells that initiate critical periods. “If you directly manipulate the PV cell—which you can do in mice—then gamma oscillations are strengthened or weakened,” Hensch says.
Since Hensch and his colleagues suspect that PV cells are defective in autism, the next step may be to go back and study gamma oscillations in the mouse models, to see how well Nelson’s observations correlate with the mouse models and to search for molecular manipulations that can help normalize this type of brain activity.
Testing for autism
Since patients on the autism spectrum often have trouble communicating, having a new way to assess their brain function and response to treatments would be invaluable. In addition to EEG signals, Nelson is examining eye-tracking during face processing, speech and language tasks. Both types of functional biomarkers could also help categorize different varieties of autism, and perhaps be used in evaluating the efficacy of new treatments.
Rare neurologic disorders, once again, provide a good starting point. Nelson and Sahin are looking for functional biomarkers of autism in infants with TSC, about half of whom will eventually fall on the autism spectrum. The early risk factors they identify could help in directing future treatments for TSC patients, and possibly also for other autism patients whose neuronal signaling pathways and brain connectivity are similarly disrupted.
Indeed, Sahin will soon launch a clinical trial of a drug that may help correct abnormal axon development and several other features of TSC. “One drug is not going to be useful in all autisms that we identify,” Sahin says. “Having said that, I think that there will be more than the TSC patients that might benefit from this treatment in the future.”
And, if clinical trials in the works for Rett syndrome, Fragile X and Angelman syndrome improve these patients’ cognitive and behavioral outcomes, they might benefit some “garden variety” autism patients as well—perhaps identified by functional biomarkers. [To learn more about these clinical trials, click here. To learn more about Nelson’s infant-sibling study, click here].
Extending the biomarker idea, Nelson is collaborating with Fagiolini, Hensch and Khwaja to examine visual function and brain activity in patients with Rett syndrome. The idea is to adapt the behavioral tests of visual acuity and recordings of activity in the visual cortex that Fagiolini developed to study the mouse visual system, and use them as biomarkers of brain function in patients with Rett or other autistic disorders.
A prismatic approach
Understanding the autism spectrum is going to require a spectrum of research—with molecular and cellular biology, electrophysiology, cognitive neuroscience and clinical work coming together to shed light on the many unknowns of the autistic brain.
“We are trying actively to put together a very strong group on autism that attacks different levels of the same questions,” says Fagiolini. “No one single lab can do everything and there is no one answer. By combining the different approaches I think we will go further.”