Annual Report 2019

SFARI Research Roundup

Simons Foundation Autism Research Initiative

Since its launch in 2003, the Simons Foundation Autism Research Initiative (SFARI) has supported the work of more than 550 investigators in the United States and abroad. In 2019, nearly 300 SFARI Investigators studied a broad array of questions about autism, from its genetic basis to new ways to support people with autism and their families. Below are some highlights of the past year’s research.

This wristwatch-style device uses a slew of biosensors to recognize when a child with autism is on the verge of an aggressive outburst. Credit: © Empatica

Sensing Aggression

Some children with severe autism are prone to aggression, which can include hitting, biting and throwing objects. Aggressive outbursts, which are highly stressful for the children and their caregivers, are often hard to predict. A new biosensor, however, offers a warning when a child with autism is likely to erupt within the next minute, giving caregivers a head start on redirecting the child and making the environment safer.

The biosensor, attached to a wristband, measures heart rate, sweat levels, movement and temperature. To teach the device how to recognize when the child is about to become aggressive, SFARI Investigators Matthew Goodwin of Northeastern University in Boston and Matthew Siegel of Maine Medical Center Research Institute in Portland and colleagues collected sensor data from 20 children and teenagers with severe autism who had been admitted to an inpatient psychiatric unit. The researchers then used a machine learning algorithm to figure out which physiological signs indicated an impending outburst.

The sensor’s warnings predicted aggressive outbursts with 71 percent accuracy, provided the sensor had been recording data for at least three minutes before the warning. The accuracy level rose to 84 percent when the model was personalized for a specific child, the researchers reported in the August 2019 issue of Autism Research. As more data become available, the researchers say, the model should be able to predict outbursts earlier.

Beyond the Coding Regions of the Genome

Over the past 15 years, gene sequencing studies have implicated more than 100 genes in autism. But the protein-coding portions of genes represent less than 2 percent of the human genome, and the remaining 98 percent — the ‘noncoding’ genome — may also play a significant role in autism. Teasing out this role is challenging, since the noncoding genome is enormous and nearly everyone has some mutations there. But two recent complementary studies, in the December 14, 2018, issue of Science and the May 27, 2019, issue of Nature Genetics, have pointed the way forward.

The Science paper was the result of work by SFARI Investigators Stephan Sanders of the University of California, San Francisco; Michael Talkowski of Harvard University; Bernie Devlin of the University of Pittsburgh; and Kathryn Roeder of Carnegie Mellon University, and their colleagues. The researchers examined whole-genome sequencing data from nearly 2,000 families in the Simons Simplex Collection, a database of genetic and phenotypic information from people with autism and their unaffected family members. Overall, the study found, children with autism have approximately the same number of spontaneous mutations as their siblings without autism do. But they have significantly more mutations than their siblings in ‘promoter’ regions of the genome, which appear just before the start of a gene. Some of these promoter mutations, the team found, are associated with genes involved in developmental delay or neuronal differentiation.

In the Nature Genetics paper, Olga Troyanskaya and colleagues at the Simons Foundation’s Flatiron Institute; Robert Darnell of Rockefeller University and colleagues; and Alan Packer of the SFARI science team used a machine learning algorithm that pinpoints how individual noncoding mutations disrupt the way genes turn on and off throughout the body. The algorithm assigns a disease impact score to every nucleotide in the human genome. Among families in the Simons Simplex Collection, the team found, the noncoding mutations in children with autism had significantly higher disease impact scores than the mutations in their siblings without autism.

Combined, these two studies suggest that mutations in noncoding regions of the genome that control gene expression and protein translation significantly contribute to autism.

Skin-Deep Treatment

Many people with autism are highly reactive to touch and other sensory stimuli. A mouse study in the August 8 issue of Cell suggests that an experimental drug called isoguvacine that dampens the activity of touch neurons beneath the skin can reduce touch hypersensitivity and also alleviate some behavioral traits associated with autism. The finding indicates that sensory reactiveness may play a more central, causative role in autism than many researchers had believed.

The researchers, led by SFARI Investigator David Ginty of Harvard Medical School, injected six-week-old mice with a single dose of isoguvacine. They found that the drug moderated reactions to touch in six different mouse models of autism, each of which models a different genetic or environmental cause of autism. The team also gave daily doses of the drug for six weeks to newborn mice missing a copy of either Shank3, an autism risk gene, or Mecp2, a risk gene for the related neurodevelopmental disorder Rett syndrome. This treatment prevented the development of touch hypersensitivity, anxiety and some social impairments.

The team also created mice that lack Shank3 every-where except in the peripheral neural system. These mice react normally to touch and don’t have anxiety or social difficulties. Their behavior suggests that establishing normal functioning in peripheral neurons may prevent certain autism traits if it is done early in development. The researchers concluded that drugs such as isoguvacine that affect peripheral neurons but cannot enter the brain may reduce autism symptoms, while avoiding the undesirable side effects of drugs that act via the brain.

A Window on Individual Neurons

Until recently, researchers studying postmortem brain tissue could analyze alterations in gene expression only at the level of clumps of tissue, not single neurons. Now SFARI Investigator Arnold Kriegstein of the University of California, San Francisco and colleagues have harnessed a new technique to sequence RNA from individual cells in the brains of 15 people with autism and 16 controls. The analysis has identified several types of neurons as playing a crucial role in the condition.

In people with autism, the team reported in the May 17 issue of Science, neurons in layers two and three of the cerebral cortex had significantly more altered genes than neurons in the cerebral cortex’s other four layers. The new work bolsters earlier studies implicating these two layers in autism. The researchers also saw altered gene expression in other cell types, especially microglia; the brain’s immune cells; and astrocytes, star-shaped brain cells that perform numerous tasks, including helping neurons communicate with each other.

Many high-confidence autism risk genes showed up among those that were expressed differently in the neurons of people with autism. Also, differences in gene expression were most pronounced in people whose autism is relatively severe. The team plans to analyze additional brains and regions outside the cerebral cortex to gather further information about which types of neurons are most important in autism.

A Protective Imbalance

Many studies indicate that the brains of people with autism have too much excitatory activity relative to inhibitory activity. This imbalance, a popular theory proposes, causes neurons to spike too often, leading in turn to problems like sensory hypersensitivity. A new study, however, calls this theory into question, suggesting that the imbalance between excitation and inhibition in autism may in fact compensate for other differences rather than cause them.

The study, led by SFARI Investigator Daniel Feldman of the University of California, Berkeley, examined the somatosensory cortex of mouse models for four different autism-linked mutations. The researchers analyzed both in vitro brain slices and recordings of neuronal activity in live mice. In each of the four models, the researchers did find a higher ratio of excitation to inhibition than in control mice. But to their surprise, the researchers also found that the neurons receiving these signals fired at the same rates as those in the control mice.

This finding, the team wrote in the February 20 issue of Neuron, suggests that the skewed ratio of excitation to inhibition may serve as a protective mechanism, helping neurons to spike normally. Some scientists are testing drugs to normalize the signaling imbalance in people with autism, but the new study suggests that such drugs might do more harm than good.