From mice to humans: Genetic syndromes may be key to finding autism treatment

Boy and a mouse eye-to-eye
(Aliaksei Lasevich/

A beautiful, happy little girl, Emma is the apple of her parents’ eyes and adored by her older sister. The only aspect of her day that is different from any other 6-month-old’s is the medicine she receives twice a day as part of a clinical trial for tuberous sclerosis complex (TSC).

Emma’s mother was just 20 weeks pregnant when she first heard the words “tuberous sclerosis,” a rare genetic condition that causes tumors to grow in various organs of the body. Prenatal imaging showed multiple benign tumors in Emma’s heart.

Emma displays no symptoms of her disease, except for random “spikes” on her electroencephalogram (EEG) picked up by her doctors at Boston Children’s Hospital. The medication she is receiving is part of the Preventing Epilepsy Using Vigabatrin in Infants with TSC (PREVeNT) trial. Her mother desperately hopes it is the active antiepileptic drug, vigabatrin, rather than placebo.

This trial is crucial to understanding the relationship between tuberous sclerosis and epilepsy — and also, possibly, autism, which affects about half of all children with tuberous sclerosis.

Multiple autisms

TSC is just one of several genetic syndromes that include autism-like behaviors and that may provide clues to treating autism in general.

“In the field of neurodevelopmental disorders and autism, it is pretty clear that there is a lot of heterogeneity,” says Mustafa Sahin, MD, PhD, a neurologist and director of the Translational Neuroscience Center at Boston Children’s.

That heterogeneity means that developing treatments can be a daunting task. “The approach that we have taken is to look at comparisons between different genetic disorders that result in autism,” says Sahin. His lab does this at two different levels — animal and human — seeking commonalities between the diseases.

Pathways to curing autism
Patients with well-defined genetic syndromes that cause ASD can be phenotyped to identify allied biomarkers. Their genetic variants can be modeled in neurons derived from their own cells (via induced pluripotent stem cells or iPSCs) and in mice. Such models can be used to test the effects of drugs and advance preclinical and clinical trials. Insights from syndromic forms of ASD might also help researchers develop treatments for patients with non-syndromic (idiopathic) ASD who have similar pathologies. (Adapted from Science)

A push for autism mouse models

Efforts to develop treatments for autism spectrum disorder (ASD) have faced a significant obstacle: a shortage of validated and robust animal models. In 2013, Autism Speaks launched the Preclinical Autism Consortium for Therapeutics (PACT) to develop and compare seven different mouse models for common genetic causes of ASD.

Tuberous sclerosis already has a mouse model, which has revealed, among other things, differences in myelination of nerve fibers and defective recycling of mitochondria.

Recently, Sahin’s group, with Alex Rotenberg, MD, PhD at Boston Children’s and researchers at the University of California, Davis, validated another ASD mouse model: inactivation of the SHANK3B gene. The same mutation causes Phelan-McDermid syndrome (PMS) in humans (also caused by deletion of the entire 22q13 chromosomal region). Symptoms include low muscle tone, developmental delays, autism-like behaviors and intellectual disability.

“Within PACT, behavioral analysis and EEG studies were performed in a very rigorous way, by replicating each result in two different cohorts of mice,” says Sahin.

Behavioral assays for autism
Mouse behavioral assays in the SHANK3B study (click to expand)

As reported in Molecular Autism, the independently-bred cohorts of mutant mice showed similar autism-like behaviors such as repetitive grooming, deficits in social interactions and differences in sensory responses, anxiety-related behaviors, learning and memory. Both cohorts also had abnormal EEG signatures indicating an inhibited response to stimuli, analogous to those observed in humans, and a lower susceptibility to induced seizures than the control mice.

This EEG pattern, though counter-intuitive, could be a useful biomarker of PMS in clinical trials, serving as a measurable gauge of disease severity or response to a treatment. “Electrophysiology is a readout of brain activity, and it’s relatively translational in that you can do almost the same experiment in the mouse model as in patients,” says Sahin.

Human studies: Comparing different autism syndromes

Sahin’s lab at Boston Children’s Hospital is the lead site for the Developmental Synaptopathies Consortium (DSC), created in 2014. Ten U.S. medical centers have been tasked with studying patients with TSC, Phelan-McDermid syndrome and PTEN Hamartoma Tumor Syndrome (PHTS), another cause of ASD whose manifestations also include skin lesions and macrocephaly.

As with PACT, the goal is to better understand shared mechanisms of autism that might inform potential treatment approaches. “Ideally, the animal effort and the human effort will feed into each other and allow us to test compounds in the animal models and then proceed to clinical trials in patient populations,” says Sahin.

PHTS is a case in point. A change in the PTEN gene disrupts the mTOR pathway that controls cell growth, causing tumors and autism. Inhibitors of the mTOR pathway have been found to be effective in mouse models and have led to a clinical trial that’s now enrolling.

Children 6 to 21 years old will receive the mTOR inhibitor everolimus or placebo for six months while being evaluated for neurocognitive improvements. “We will also have the patients undergo EEG as a biomarker to interrogate the effect of the drug on the brain and its electrophysiological activity,” says Sahin.

Preventing autism?

The PREVeNT clinical trial for tuberous sclerosis, which will enroll 80 infants at seven different centers, is also using EEG as a biomarker. To be enrolled, a baby must have “an abnormal EEG signature,” known to be a predictor of clinical seizures a few months down the line. The babies will undergo repeat EEGs over a period of two years.

Sahin thinks that an earlier TSC trial, which also used everolimus to improve neurocognition, failed because it did not use EEG as a biomarker and because the patients were older (6 to 21 years) with well-established symptoms.

The PREVeNT trial represents a newer idea in child neurology: treating patients before they begin exhibiting symptoms of their genetic disease, based on biomarkers. As in Emma’s case, TSC can be detected even before birth.

“[Preventive treatment] is a transformative idea and can apply not only to seizures, but also to other neurodevelopmental aspects of TSC, such as autism, and to diseases other than TSC,” says Sahin.

Finding “convergence”

What causes the malfunctions in the brain leading to autism? That is the larger question Sahin hopes to answer with his research on TSC, PHTS and PMS. Autism symptoms, like stilted social interactions and repetitive behaviors, potentially arise from disruptions at any one of several junctions within a widespread neuronal network. Mouse models of different forms of autism, along with reliable biomarkers, can help illuminate these neuronal circuitries.

Treating the many forms of autism
Shared pathways for genetic causes of autism might lead to common treatment plans. (Adapted from Science)

With this understanding, researchers could target diseases that affect common pathways (such as the mTOR pathway in TSC and PHTS) with the same or similar drug. Studying different genetic diseases that cause autism will also help scientists like Sahin piece together the puzzle that is autism.

Meanwhile, Emma is so far living a seizure-free life and is developing normally, her mother reports. Whether this is due to vigabatrin remains to be seen.

For information on enrolling in the PREVeNT and PHTS (PTEN) trials, email