UNDERSTANDING AUTISM

gene called neuroligin-3 passes on the condition. In other cases, ASD has been linked to abnormal or extra chunks of DNA on chromosome 15.

The vast majority of autism cases don’t have this clear link. Some suspect it might arise from defects in multiple genes acting together. Over the past two decades researchers have become masterful at fishing them out. One fruitful approach has been to read the DNA sequence of unaffected parents and their autistic child, to see where the child’s DNA differs. This has revealed that often the child has acquired new DNA mutations; in about 10% of cases sizable chunks of their DNA has been either duplicated or deleted – so-called ‘copy number variations’.

While hundreds of genes have been linked to ASD, they haven’t shed much light on the nature of the disorder. The genes tend to participate in global processes: for example, the CHD8 gene regulates how DNA is packaged, while the SCN2A gene codes for a sodium channel that regulates electrical transmission between brain neurons. IN HIS HARVARD LAB, faced with so many potential suspects, Anderson decided to follow a hunch about chromosome 15. He was compelled by evidence of the yin-yang behaviourseffect that followed from having gaining too few or too many bits of this chromosome.

Associated with about 2% of ASD cases, it is one of the few genetic causes of autism that can be detected under the microscope. A band on chromosome 15 is slightly bigger than it ought to be. That’s because this region, known as 15q11-13, has been triplicated. (Curiously, it is only associated with ASD when inherited from the mother.)

On the other hand, children who have lost this same bit of chromosome are the opposite of autistic; they have Angelman’s Syndrome, evidenced by a lower IQ and hyper-social behaviour. They smile and laugh, and make strong eye contact.

One of the genes that stood out in the 15q region was ubiquitin E3-ligase (Ube3a). Other trawls through the DNA of people affected by ASD were also unearthing copy number increases in the same gene. But its known functions – tagging some proteins for destruction, and turning genes on and off – provided few hints as to why it might cause autism.

So the Anderson team created mice with multiple copies of the Ube3a gene. Sure enough, the animals showed symptoms that placed them on the mouse autism spectrum. Males were less interested in new females, vocalised less, and groomed obsessively. The more copies of the gene, the more severe these behaviours became.

Deleting Ube3a had the opposite effect. Mice became more social, spending an average of 15 minutes socialising with newcomers instead of the usual five.

How was Ube3a doing this? The researchers took a look at which genes were affected by the presence of the extra copies. Genes provide recipes for proteins. It seemed Ube3a was acting like the master chef and modifying the menu. The production of about 200 dishes was being dialled down, while 400 others were ramped up.

Dauntingly complex? Not for this team of data detectives. Anderson and his colleagues checked the ASD gene databases to see which proteins on the modified menu also belonged to autism networks. They then checked to see which of the autism- and Ube3a-related proteins bound to each other. Like stepping stones, the idea was to connect the proteins in a pathway that might lead to autism.

One of the proteins dialled down is called cerebellin 1. The more copies of the Ube3a gene, the lower its production. The protein-binding network showed cerebellin 1 snapped together with two interesting autism-related genes: neurexin 1 and glutamate receptor delta subunit 1 (GRID 1). All three act at the synapse, the space between neurons where messages are relayed by neurotransmitters.

The trio form a sandwich across the synapse, with neurexin 1 at the sending side, and GRID 1 at the receiving side and cerebellin 1 in the middle. These proteins help solder particular neurons together to form a circuit.

Now the question became: what types of neurons use this triple solder? There are dozens of types, identified by the transmitters they use. To find out, the team used genetic engineering to insert extra copies of the master chef gene into specific neurons. It was only when excess Ube3a was introduced into glutamatergic or dopaminergic neurons – both associated with reward-dependent behaviours – that mice experienced social problems.

Genes provide recipes for proteins. It seemed the Ube3a gene was acting like the master chef and modifying the menu.