ABOVE: © ISTOCK.COM, JAMESBREY

While researchers have found plenty of gene variants that seem to increase the risk of an autism diagnosis, it’s not clear why some people carrying these mutations develop autism spectrum disorders and some do not. In a study published today (September 17) in Science Advances, researchers point to a potential answer: severe infections during early childhood. After an early immune challenge, male mice with a mutated copy of the tuberous sclerosis complex 2 (Tsc2) gene developed deficits in social behavior linked to changes in microglia, the immune cells of the brain. And an analysis of the hospital records of more than 3 million children showed that children, particularly boys, who were hospitalized for infections between ages 18 months and four years were more likely that healthy peers to receive a future autism spectrum disorder (ASD) diagnosis.

“We have genetic models, and we have a lot of in utero exposure models and early life stress models, but it’s pretty rare that people are blending the two to find that gene [and] environment interaction,” says Audrey Brumback, a pediatric neurologist at the University of Texas at Austin Dell Medical School who was not involved in the work. Plus, “we’re so neuron centric in neuroscience, [but] a huge chunk of our brain is non-neuronal,” she adds. “It’s really exciting to see work that’s exploring those non-neuronal cells.”

“We knew that mutations predispose [people] for autism, but if you look in patients with genetic mutations, not everyone with that mutation has autism, and the question is why?” says neuroscientist Alcino Silva of University of California, Los Angeles (UCLA). One such type of mutation, linked to autism in about half of the people who carry the variants, are in the tuberous sclerosis complex 1 or 2 genes and can have a range of symptoms in addition to autism. Mice with a mutation in Tsc2 have some of the same symptoms, but until about a decade ago, the social deficits that can show up in people with the mutations had not been recreated in the mouse model. Then, in 2010, Silva’s group showed that challenging the immune systems of pregnant mice caused ASD-like behavior in their Tsc2 mutant offspring.

In the new study, Silva and colleagues further explore the interactions of genetics and environment, this time at later stages of development. They injected either an immune stimulant known as PolyI:C or saline into wildtype mice and Tsc2 heterozygotes at postnatal days 3, 7, and 14. After the mice reached adulthood, the researchers tested their social behaviors with the three-chamber social interaction test, in which mice are exposed to a chamber that’s empty on one side and contains a new mouse on the other. Twenty-four hours later, the chamber contains the now-familiar mouse on one side and a new mouse on the other side. All of the mice spent more time with the new mouse on the first day than on the empty side of the chamber. But only male Tsc2 heterozygotes who’d received the immune stimulant in early childhood spent equal time with the familiar mouse and the new mouse on the second day—instead of preferring the unfamiliar mouse, as the animals normally do—indicating that their social memory was impaired.

“It was super interesting that these deficits were unique to social memory and did not result in impaired sociability—one of the key hallmark tasks used to assess social interactions in mouse models of ASD,” Annie Ciernia, a neuroscientist at the University of British Columbia who was not involved in the study, writes in an email to The Scientist. “This suggests that postnatal viral infections (which PolyI:C mimic) could be disrupting unique neural circuits important for social memory that are vulnerable during early postnatal development.” 

Mice use ultrasonic vocalizations to communicate, and it’s been shown before that Tsc2 heterozygotes don’t vocalize like their wildtype siblings, instead making more short calls that mother mice may be less responsive to. Silva’s group collaborated with that of Stephanie White, a UCLA biologist and vocal learning expert, to investigate the effect of infections on these vocalizations. The team showed that early immune activation exacerbated the differences in vocalizations between wildtype mice and Tsc2 heterozygotes, and write in the paper that this “may parallel early ASD social communication deficits” seen in humans.

Then, the researchers analyzed gene expression in the brains of the adult mice and found that genes associated with microglia and interferon signaling were more active in male Tsc2 heterozygotes that received the immune stimulant, but not in any of the other mice. Using a drug to deplete the microglia in these mice reversed the defects in social behaviors, even after microglia reappeared months later.

“This is one of the first examples of how repopulation [of microglia] opens a new opportunity to reshape microglia function in the adult and provides the potential for novel therapeutic delivery in adults with ASD,” Ciernia writes.

The team also found that mice without functional interferon signaling—due to either a genetic mutation or injection of the drug rapamycin—don’t develop deficits in either social memory or vocalizations after simulated infections. Taken together, the findings point to a role for interferon signaling by microglia in the development of ASD-like symptoms in mice. The differences in the development of microglia in males and females may help explain the sex differences in the response to immune activation, Silva says, adding that autism is about four times more common in boys than in girls.

Finally, in what he calls “a Hail Mary,” Silva asked a friend, computational biologist Andrey Rzhetsky of the University of Chicago, to look at dataset of more than 3.5 million health insurance claims to see if there was any relationship between severe infections and autism in humans. “He comes back months later and says, ‘That’s the biggest association I’ve ever found in this dataset,’” says Silva. Male children, regardless of genetic status, who were hospitalized with infections between the ages of 18 months and four years were 40 percent more likely to be diagnosed with ASD later than were boys who weren’t hospitalized for infections, while for girls, hospitalization for infection at this age was associated with a 30 percent greater chance of ASD diagnosis. The difference for girls was not statistically significant, however.

“This paper has to be [understood] as proof that you need to vaccinate your kids,” since infectious diseases can not only be fatal, but can also raise the risk of ASD among children who survive, says coauthor Manuel López Aranda, a neuroscientist at UCLA.

The combination of basic science and the clinically relevant data analysis of more than 3 million children is “a slam dunk,” says Tanjala Gipson, a pediatric neurologist at Le Bonheur Children’s Hospital in Memphis, Tennessee, who did not participate in the study. Open questions include: “how do I know my child is at risk? Do I need to be worried about every fever? Do I need to be worried about every infection?” she says. Thus, one next step would be determining whether there are biomarkers that indicate when children are more at risk.

Rapamycin, the drug the authors used to ameliorate the effects of simulated infection on the mice, is already being studied for tuberous sclerosis, the genetic disorder caused by Tsc1 and Tsc2 mutations, she notes. “It’s another reason for hope, and there’s always room for hope.”

Clarification (September 17): The paragraph about the association found between hospitalization for infection and autism diagnosis in children has been amended to state that the association was not statistically significant for girls.