Autism is a syndrome with many causes. David Sulzer and his colleagues at Columbia University examined their new hypothesis that some of these processes converge during a developmental period from early childhood through adolescence when cortical synapses — the connections that provide for communication between neurons — lose approximately half of their overall density, a phenomenon known as ‘synaptic pruning.’

Examining temporal lobe cortical neurons, they found that individuals with autism have severely limited synaptic pruning, resulting in synaptic overconnectivity that reaches levels of nearly twofold more density during their teens than in unaffected people. This overconnectivity may underlie fundamental features of autism, including the high incidence of epileptic seizures and overly sensitive responses to social stimuli. The researchers also found that those with autism possess high levels of activation of a master regulator of protein synthesis and degradation, known as mTOR, in these neurons. This has previously been suggested in diseases with a high incidence of autism, such as tuberous sclerosis (TSC), but was unknown in autism.

Sulzer and his group then found that people with autism have a low level of a cellular degradative process known as autophagy (‘self-eating’) that provides for turnover of the cell’s components and is turned off when mTOR is highly active. They thus hypothesized that overactive mTOR and decreased autophagy may be responsible for the synaptic overconnectivity in individuals with autism.

To test this hypothesis, the researchers developed two mouse lines of TSC mutations with high mTOR activity, another line with deletion of autophagy in cortical neurons, and a fourth mutant line that has both the TSC and autophagy mutations. Each of these mutant mouse models displays autism-like behaviors and does not exhibit normal levels of synaptic pruning. When mTOR is inhibited by the drug rapamycin, both synaptic pruning and behaviors are normalized in the TSC mutants, but not in the autophagy-deficient mice, nor in the double-mutant animals.

These results indicate that the major action of mTOR on synaptic pruning during development is due to inhibition of neuronal macroautophagy, and that neuronal autophagy is required for the majority of synaptic pruning during normal development. Nevertheless, as autophagy-deficient mice still retain approximately 30 percent normal pruning, additional processes must be involved.

As the individuals studied were sporadic cases, meaning that they did not possess a known mutation that caused the disorder, it is possible that many, perhaps a majority, of people with autism have increased mTOR activity in neurons, with decreased synaptic pruning due in large part to loss of normal neuronal autophagy. Because this was a small group of people, Sulzer and his colleagues suspect that not all additional cases will fit this pattern, consistent with autism being a syndrome of multiple diseases.

The ability to normalize synaptic pruning during development in the mouse models suggests that both autism-like behavior and abnormal synaptic development might be rescued by focusing on mTOR/autophagy pathways, for example by the use of drugs that would restore normal levels of neuronal autophagy. This can occur in mice even after the process has been initiated, suggesting that treatment could be helpful even after diagnosis. This is extremely important in a disorder such as autism, as in the vast majority of cases, there is no effective way to predict who will develop it.