Christopher Walsh, MD, discusses the challenges behind identifying the causes of autism.
by Michelle Pflumm, Ph.D.
Donnie T. had no interest in going out for ice cream or visiting the candy store. What he really loved to do was watch tops spin. "He was happiest when he was left alone," his physician Leo Kanner, MD, of Johns Hopkins Hospital wrote in 1938, "almost never cried to go to his mother, [and] never seemed to notice his father's homecomings."
We now understand that the "fascinating peculiarities" Kanner saw in 5-year-old Donnie, which he called "autism," are likely due to miswiring or missing neural connections in the brain. What lies behind these changes, however, continues to fascinate researchers to this day.
A complex genetic disease
In the late 1970s, researchers began to appreciate the strong role of genetics in autism. Studies of twins showed that, if one twin has autism, the second twin is much more likely to be affected if the twins are identical (essentially 100 percent genetically identical) than if they are fraternal (25 percent genetically identical). Autism also is associated with several rare genetic diseases, further suggesting genetic mechanisms.
These studies raised hopes that by understanding the genetic cause of autism, it might be possible to treat the disease. But finding specific genes has proven to be much more complicated. Autism is defined by a set of distinct behavioral symptoms that can be extremely variable, that can be triggered by multiple genetic changes and environmental factors, and that may even represent more than one disease.
"You can have a kid who has an I.Q. of 60 and has seizures and a kid who has an I.Q. of 140 who is at MIT and they both have autism," says Charles Nelson, PhD, Research Director of the Developmental Medicine Center at Boston Children's Hospital. "You've got to deal with that somehow."
The hunt for genes begins
In the early 1960s, when karyotyping became available, researchers inspected the chromosomes of people with autism, seeking deleted or duplicated regions that might contain the critical genes. They largely came up empty-handed.
Some experts, however, suspected that deletions and duplications were there, but were simply too small to be detected by this technique. With the introduction of chromosomal microarray analysis in 2005, however, researchers could detect deleted or duplicated regions at roughly 100-fold greater resolution.
Using chromosomal microarray analysis, the Genetics Diagnostic Laboratory of Boston Children's Hospital, led by director Bai-Lin Wu, PhD, noticed that several patients with a diagnosis of autism spectrum disorders had missing or extra sections of chromosomes 15 and 16. Deletions or duplications of these chromosome regions--15q on chromosome 15 and 16p on chromosome 16--occur in about 1 percent of autistic children.
These regions are now under active investigation to zero in on possible causative genes within them.
"I don't think we're going to find one genetic cause that explains 50 percent of autism," says David T. Miller, MD, PhD of the Division of Genetics and the Department of Laboratory Medicine, a coauthor on these studies. "It's going to be an incremental process. Even if it's 1 percent at a time, that's still progress."
The Genetics Diagnostic Laboratory team. From left: David Miller, MD, PhD; Bai-Lin Wu, PhD; and Yiping Shen, MD, PhD.
The most comprehensive clinical study to date, a collaborative effort between the Genetic Diagnostic Laboratory at Children's and the Boston-based Autism Consortium, detected potentially significant deletions and duplications in 18 percent of the 933 patients tested, with about 7 percent being clearly associated with the disease. This represents a big improvement over traditional karyotyping.
"A lot of genetic studies found chromosomal deletions and duplications mainly for technical reasons," says Christopher Walsh, MD, PhD, chief of the Division of Genetics. "With the new technology we now have, they are easy to see. Point mutations [changes in a single base pair or "letter" of the gene sequence] are probably going to be more common than deletions or duplications, but they are much more difficult to find."
Looking to the Middle East
The big problem is that most of the genetic changes associated with autism are rare. Most of them also occur spontaneously, rather than being inherited, and they are often unique. Only by testing thousands and thousands of people can even deletions be tied to the disease.
To find these rare genetic changes, Walsh's team has been using a classic genetics approach: homozygosity mapping. The idea: Take a look at extremely large families that have a closely shared common ancestry (often because of marriage between cousins) that "enriches" them for rare susceptibility genes. Then, compare their chromosomes side by side to identify small fragments of the genome that are shared between family members with the disease. These areas can then be further explored with DNA sequencing, to find specific point mutations, or simply by chromosomal microarray analysis to find deletions or duplications associated with the disease.
"When you have a family that is large enough, with three or four affected people," says Walsh, "you can identify one or two places in the genome where the gene has to be."
Walsh's team recruited nearly 100 families with autism in parts of the world where marriage between cousins is common, such as the Middle East. Coupled to chromosomal microarray analysis, the team zeroed in on six likely susceptibility genes in five out of the first 78 families studied. All the genes are close to or located within the chromosome deletions. Many of them appear to be involved in wiring up circuits in the brain.
"The new genes confirm the theory that a lot of autism seems to have to do with connections and with the changes in these connections that enable learning," says Walsh.
Now, the Walsh team is sequencing the chromosome regions associated with autism in other families in hopes of identifying additional susceptibility genes and point mutations. In the future, Walsh hopes to extend this type of analysis to families in the United States with a shared ethnic ancestry, to find additional genes that might contribute to the disease.
Making more connections
With over 100 genes implicated in autism and countless possible combinations of them, it is increasingly clear that identifying the key missing or miswired brain connections in autistic children will be difficult indeed. And add to that the growing consensus that autism isn't one disease, but many, each involving different sets of genes. That's where computational power can come to bear, to crunch the genetic data and look for patterns.
"Autism is a multifaceted, polygenic disease," says Dennis Wall, PhD, director of the Computational Biology Initiative at Harvard Medical School and a researcher with the Children's Hospital Informatics Program (CHIP). "In order to disentangle that forest, we feel it is extremely important to implement network biology approaches to pull together connections we already know about and use them to develop a more comprehensive picture of the disease."
Wall is using what he calls a cross-disease approach, looking at other behavioral disorders that share symptoms with autism, such as ataxia (loss of muscle coordination) epilepsy (seizure disorder) and major depressive disorder, and thereby may share risk-associated genes.
"In many of these other diseases, the genetic causes are better understood," says Wall. "We can use this information about these diseases as a launchpad to make predictions of new genes in autism."
Wall and his team identified 13 diseases that share 66 of the 127 previously identified autism-associated genes. By including additional genes that interact or are tightly co-expressed with these 66 genes, they expanded the list to a network of 334 possible susceptibility genes.
But are any of these 334 genes actually associated with autism? Wall and his team took a look at a list of genes identified by the UC Davis M.I.N.D. Institute to be dysregulated in the bloodstream of people with autism. They found that 289 of these genes (87 percent) showed a significant difference in gene expression in people with autism, suggesting they are indeed promising risk-associated genes.
Now, in collaboration with Children's Director of Informatics Isaac Kohane, MD, PhD, and Director of Genomics Louis Kunkel, PhD, Wall and his team are refining their predictions of autism susceptibility genes. By further fine-tuning the underlying networks of genes disrupted in autism, Wall hopes to better understand the variability of the disease, and ultimately identify key genes to target in the disease.
A promising future
With a growing list of genes identified, autism experts are beginning to define sub-types of the disease. By coupling the power of bioinformatics with next-generation microarray and gene-sequencing technologies, new therapeutic targets may eventually emerge in this puzzling disease.
"Instead of autism looking like a large grey area, we are starting to see its sub-syndromes in sharper relief," say Walsh. "The more we understand each of these conditions, the more we can tailor the treatment of each child with the disease."
- Arresting autism
Charles Nelson MD is developing behavioral tests to identify children as early as 3-6 months of age suspected of having autism.
- A blood test for autism?
Louis Kunkel, PhD, and Isaac Kohane, MD, PhD, introduce a potential autism test that can be performed in a typical clinical diagnostic laboratory.
- Get involved
Logon to Bionotate, a collaborative Web project, to help Wall's team identify potential genes implicated in autism spectrum disorders.
- Breaking into the autistic brain
Children's researchers work across disciplines to study brain development in autism.
- Can drugs improve cognitive outcomes?
A new wave of clinical trials at Boston Children's tests drugs aimed at improving cognitive outcomes in genetically-based developmental disorders.