Skip to content

When you choose to publish with PLOS, your research makes an impact. Make your work accessible to all, without restrictions, and accelerate scientific discovery with options like preprints and published peer review that make your work more Open.


Autism, Seizures, and the Amish

(Jonathan Bailey, NHGRI)
(Jonathan Bailey, NHGRI)

On December 4, I visited the Clinic for Special Children in the heart of Pennsylvania Dutch country, where a tiny staff cares for 2000+ patients with a variety of inherited illnesses. Last week’s post described a family in which 5 of 6 children have a seizure disorder that includes autistic features. Investigation of this syndrome over the past 15 years beautifully illustrates the evolution of gene discovery methods before sequencing catapulted us into a new technological age. (Warning: jargon ahead.)

Today, to identify the gene behind an undiagnosed condition in a child, researchers compare the exome (protein-encoding part of a genome) sequences of parents and possibly siblings to identify causative gene variants (alleles). It’s fast.

In the pre-genome era, researchers followed an indirect trajectory to get from phenotype to genotype:

• Noting that symptoms “run in families.”
• Finding the condition to be more common among identical twins than fraternal twins, and among siblings of patients than in the general population.
• Identifying abnormal chromosomes in people with the condition.
• Using genome-wide association studies (GWAS) to identify patterns of genetic variation (at sets of single-base sites called single nucleotide polymorphisms, or SNPs) that might flag a disease-causing variant.
• Identifying the gene and making a mouse model to test treatments.

When the Amish left Switzerland in the early 1700s to escape religious persecution and settled in Pennsylvania, they brought a sampling of the European gene pool. Reproducing among themselves amplified mutations and resulted in “runs of autozygosity” in their genomes – sections of chromosomes that have the same DNA sequences on both copies. Runs of autozygosity indicate that two relatives inherited sets of gene variants from a shared ancestor, such as second cousins from a great-grandparent. These alleles, dubbed “identical by descent” (IDB), make a powerful tool for gene discovery if they show up exclusively in people who have the same disease.

The boy I met at the Clinic and 4 of his 5 siblings have cortical dysplasia-focal epilepsy (CDFE) syndrome. It arises from a single missing DNA base in a gene called CNTNAP2.

Like many tales of gene discovery, the finding that mutations in CNTNAP2 lie behind a variety of brain conditions – autism, seizures, schizophrenia, Tourette syndrome, and language disorders – began with different threads. Let’s focus on the autism connection.

Even the oldest genetic techniques demonstrate an inherited component to autism. Identical twins are much more likely to both have it than fraternal twins; siblings of a child with autism have a risk 75 times that of the general population.

In 1998, the International Molecular Genetic Study of Autism Consortium used GWAS to identify 6 regions in the genome that tracked with people who have autism – the top contender was on the long arm of the seventh largest chromosome, or “7q.”

Disturbed communication between neurexins and neuroligins may underlie autism. (Rachelbash1 from Wikimedia Commons)
Disturbed communication between neurexins and neuroligins may underlie autism. (Rachelbash1 from Wikimedia Commons)

In 1999, researchers implicated CASPR2, a type of protein called a neurexin that, when abnormal, disrupts sending of a nerve impulse. Neurexins align with other proteins called neuroligins to create the synapses that form as a young child begins to explore the world, consolidating memory into learning.

In 2003 came reports of rearranged chromosomes that disrupt the gene that encodes CASPR2, CNTNAP2, in people with Tourette syndrome and in 2007 with people who have intellectual disability, developmental delay, impaired speech, and hyperactivity, but not Tourette’s. These different conditions aren’t surprising – effects vary depending on where in the brain neurexin levels are imbalanced.

In 2006, researchers at the Clinic For Special Children and the Translational Genomics Research Institute matched a mutation in CNTNAP2 to CDFE syndrome in closely-related Amish kids. Seizures begin at about the age that autistic features tend to emerge — 14 to 20 months.

Before seizures begin, symptoms of CDFE are subtle: minor motor delays, poor deep tendon reflexes, and slightly-large heads. Children have difficulty concentrating, imitating people, and planning movements, such as crawling, cruising, and walking. The seizures are frequent and severe, and their onset heralds the ebbing of skills – language, cognitive, and social. After several years the seizures cease, but intellect remains stalls in childhood and the individual requires lifelong care.

Amish_-_On_the_way_to_school_by_Gadjoboy-cropKevin Strauss, MD, Erik Puffenberger, PhD, and Holmes Morton, MD, from the Clinic and their colleagues used 100,000-SNP microarray devices  to analyze the DNA of four children with CDFE syndrome from three Amish families. They found an autozygous region 7.1 million bases long on the suspected area on 7q. (Today algorithms quickly spot autozygosity in exome sequences.)

The 7.1 million bases include 83 genes, but only a few made sense. The team first sequenced a gene called CENTG3 known to cause other brain disorders. But the sick kids didn’t have mutations in it.

Then Dr. Puffenberger, the geneticist on the team, found a shortcut: he noticed one SNP in the middle of CENTG3 that was heterozygous in two kids (two different variants), rather than homozygous (the same variant in both chromosome copies), marking an end to the identical-by-descent region. “A recombination event in the middle of the gene allowed Eric to get rid of a lot of it to find the mutation. It’s a perfect example of ‘chance favors the prepared mind’,” Dr. Morton told me. That discovery cut the region of interest on 7q to 3.8 million bases.

The second candidate gene, CNTNAP2, straddles the region of interest. Each child was missing a single base from both copies of chromosome 7 there, and each parent had the same mutation, but in only one copy. They’re carriers. It was Mendel’s first law at work.

The team had found their gene. They then looked further in the community, and among 105 healthy Amish people, four were carriers. Nine of 18 patients who had partial seizures but no specific diagnosis, from 7 families, had CDFE syndrome.

The seizures were puzzling. “One mutation can cause different types of seizures. Four kids in one family respond differently. Some are very disabled, some not very affected,” Dr. Morton said. Three kids had surgery to relieve the seizures, but relief didn’t last. However, the surgeries supplied samples of brain tissue that enabled the researchers to better describe what was going wrong.

Connectivity in the seizure-prone brains is a mess. Gray matter and white matter boundaries blur, and some parts of the cerebral cortex are thickened. The neurons themselves aren’t quite right. They’re too round, too tightly packed, with deranged dendritic trees. Dots on the neurons hint at too many nuclei from glia, the supportive cells that comprise most of the nervous system. The 2006 research report waxed poetic, describing the amygdala, the seat of emotions, in the epileptic brains as “eccentric microscopic islands of partially matured neuronal precursors in tight clusters” cloaked in glia run amok.

The portrait of the Amish epileptic brain made sense in light of the work on the neurexin protein CASPR2 (which stands for contactin-associated protein-like 2). The neurexin forms a scaffolding at the nodes of Ranvier. The nodes are the exposed sites on an axon between pillows of  myelin, the insulating material that is actually the cell membranes of glia wrapped around the neuron like a bandaid around a finger. Nerve impulses hurdle the nodes, sending messages fast enough to sustain life.

The CASPR2 proteins in the Amish kids are stunted. They do not traverse the neuron cell membranes and dip into the cytoplasm as they should, and as a result nearby potassium channels collapse. These channels normally allow potassium ions to rush out of the nerve cells as an impulse passes, resetting it. So without the neurexin scaffolding, the neuron can’t reset itself. Transmission halts. And, somehow, seizures begin. I don’t think it’s known whether the seizures induce the autistic features or they arise directly – further genetic studies should indicate that.

Daniel Geschwind, MD, PhD and professor of neurology at the David Geffen School of Medicine at UCLA, was working on autism genes, and read the 2006 paper in the New England Journal of Medicine. “He called and said, ‘you found my gene!’ A nice collaboration began, and he made a mouse with the Amish mutation,” said Dr. Morton. The mice have the CNTNAP2 gene knocked out, and like people, they have seizures and autistic features.

“A mouse usually runs the around cage, normally social and chattering. These mice were neither,” Dr. Morton explained. The mice also displayed repetitive behaviors and had seizures.

The brains of the mutant mice showed an abnormal connectivity pattern reminiscent of the earlier histology work. “The front of the brain talks mostly with itself. It doesn’t communicate as much with other parts of the brain and lacks long-range connections to the back of the brain,” Dr. Geschwind said. The group had shown similar abnormalities in the brains of children with autism.

Functional MRI shows distinct and consistent connectivity patterns in the brains of children with autism and the CNTNAP2 risk variant. (Geschwind lab)
Functional MRI shows distinct and consistent connectivity patterns in the brains of children with autism and the CNTNAP2 risk variant. (Geschwind lab)

The striking similarity between the Amish children and the mice provides a testing ground for drugs. Risperidone, prescribed to treat repetitive behaviors in children, had the same effect on the mice, while also improving their nest-building ability. But the drug doesn’t help children socialize.

An obvious candidate drug to improve social skills is the “love hormone” oxytocin. It’s abundant in the same brain neurons that are rich in CASPR2 protein. Could too little oxytocin cause autistic features? Results of supplementing oxytocin are promising, both in mice and children.

Dr. Geschwind and co-workers found that a nasal spray of oxytocin “dramatically improves social deficits” in the mice. Because non-autistic mice didn’t respond, the hormone indeed seems to compensate for a deficit.

Amish farmers already give oxytocin to cows to contract their uterine muscles, and I recall getting it to rev up a stalled labor. But don’t try this at home. Several clinical trials are underway for oxytocin or drugs that boost its activity in the brain to improve socialization in children with autism.

Autism Speaks funded the first clinical trial of oxytocin in 2010, and NIH is sponsoring a larger ongoing trial of oxytocin nasal spray. But as far as I can tell, patients were enrolled based on clinical diagnoses according to the DSM-IV – not the more specific criterion of genotype.

Although the idea to try oxytocin to improve social symptoms in autism didn’t require knowing the underlying mutation, such information can add precision to any conclusions by considering mechanism – which can lead to developing other treatments. In another post I’ll look at how genetic precision enabled Dr. Morton to develop cures for certain inborn errors of metabolism that are much more common among the Amish, but still show up on newborn screens of everyone.

  1. Ricki,

    I use this particular example when I talk about how my research can be used to validate candidate genes identified by GWAS. Because of the founder effect in the Amish, The Clinic for Special Children was able to identify CNTNAP2 by looking at ONLY 7 people (3 parents and 4 children). As you know most GWAS studies require tens of thousands of individuals to identify potential causative SNP’s! I met Dr. Geschwind at the Gordon Conference this summer and was sitting next to Kevin Krauss when Eric Puffenberger mentioned their collaboration at the Clinic meeting that you couldn’t make it to. I immediately showed Kevin my notes from Dan Geschwind’s Gordon talk. It is a small world 🙂


  2. Thanks Jon. I will try to go to this summer’s meeting, which is in Ohio. Are you going? I cannot believe how many story ideas I’ve gotten from a 2 hour visit! If organic chemistry had been taught in the context of inborn errors of metabolism, I probably would’ve made it to medical school and never become a writer. Context is everything!

  3. Hello Ricki,

    I was wondering, have you looked at the following webpage?
    It was generated on 21st October 2013.

    The Ethical Frontier: Recombineering and Its Potential Effect on the Human Genome

    I mentioned the FOXP2 Gene as a possible regulator of the CNTNAP2 Gene in the above thread, and was wondering what you thought about this in regards to individuals expressing ASD who lack the latter gene.

    The reason for my interest is that my son suffers severe autism, but is responding well to the input of others although he continues to be largely non-verbal.

    I have been recently diagnosed as having Asperger’s Disorder, although this name was recently removed from the diagnostic manual.

    Although I had mixed feelings regarding the use of the mice, I am taken aback and very impressed by the speed with which you and your peers managed to make progress and publish.

    Thank God for hardworking, dedicated people like you.

    We at Project Avalon are cheering you on — and in fact some of us are so excited that we are actively contributing to research in our own way, as you can see.

    Thank you again for your hard work and selflessness.


Leave a Reply

Your email address will not be published. Required fields are marked *

Add your ORCID here. (e.g. 0000-0002-7299-680X)

Back to top