Study Suggests Autism-Prevention Strategy


by Jim Schnabel

February 6, 2014

French scientists have announced that an FDA-approved drug, given to pregnant rodents just before delivery, prevented the development of autism-like signs in their offspring.

The drug, bumetanide, has already shown some effectiveness in early clinical trials in autistic children. The new study, in the Feb. 7 issue of Science, hints that bumetanide could largely block the development of autism if administered around the time of birth-although, even if that were the case, scientists first would have to find a way to determine which babies should be so treated.

"We know that the earlier you diagnose and the earlier you treat, the more likely success will be," said Yehezkel Ben-Ari, a scientist at Aix-Marseille University who led the research team.

The GABA switch

Aside from the therapeutic implications, the findings suggest that an important cause of autism disorders, and perhaps the major cause, is a delay in a developmental phenomenon known as the "GABA switch."

GABA is a neurotransmitter molecule found virtually everywhere in the brain. In adults it has an essential "inhibitory" function: When it activates GABA receptors on a neuron, the effect is to suppress the spiking activity of that neuron. But in the fetal brain, GABA receptors have the opposite, "excitatory" effect.

Why evolution favored an excitatory role for GABA in early development, and a switchover to  an inhibitory role in maturity, is still debated. But scientists do at least know the immediate cause of this dramatic shift. GABA receptor function in a given neuron depends on the level of chloride ions in that neuron, and in the fetal brain chloride levels are high, making GABA receptors excitatory. In the mature brain, chloride levels are lower, making the receptors inhibitory. The switch seems to occur (in rodents) in the first weeks after birth, as chloride transporter genes change their fetal expression patterns and cause neuronal chloride levels to drop.

In a study reported in Science in 2006, Ben-Ari and his colleagues found evidence for a sudden, temporary reversal in GABA function (from excitatory to inhibitory) in the brains of mice at the time they are born-triggered somehow by a surge in the mother's production of the hormone oxytocin during delivery. The researchers suggested that this transient, oxytocin-driven event helps quiet and protect the brain during birth.

In the new study, they found evidence that it plays a much more important role in brain development-so important that without it, autism may result.

Rescued from autism

First, Ben-Ari's team determined that in hippocampal neurons the normal GABA switch in the first weeks after birth fails to occur-or is at least greatly delayed-in two very different rodent models of autism.

One model is a mouse model of Fragile X syndrome, a relatively common genetic form of autism in humans. (A separate research group, led by Anis Contractor at Northwestern University, reported similar results for Fragile X mice in a paper in January.) The other model is a rat "valproate model" in which autism is induced by prenatal exposure to the epilepsy drug valproate. Rodents in both models usually grow up with striking autism-like signs, ranging from asocial behaviors to brain wave changes and abnormal vocalizations.

Was the lack of a normal GABA shift in these two models merely coincidental, or did it account for their autism-like signs? Ben-Ari's team decided to find out. They forced the GABA switch to occur in these rodents by adding bumetanide to the drinking water of their mothers on the day before delivery. Bumetanide, a drug originally used to treat heart failure, lowers chloride levels in neurons by blocking the activity of a key chloride-importer molecule.

The result: the baby rodents grew up with virtually no autism-like signs.

The researchers suspected that the temporary, oxytocin-triggered reversal of GABA function at birth might be a key initiator of the GABA switch that occurs in the first weeks of rodents' lives. So they took ordinary, non-autism-model rats and mice and blocked this at-birth effect by suppressing oxytocin receptor signaling with an antagonist drug. Just as in the autism-model animals, the GABA switch failed to occur: Neuronal chloride levels remained elevated at birth and for weeks thereafter-and these baby rodents, which otherwise should have developed normally, began exhibiting autism-like signs.

"It's a very nice piece of work. I was very impressed by it when I read it," says Contractor, the author of the Fragile X study.

What comes next

The results make sense of a lot of prior data on autism, including findings of deficiencies in oxytocin signaling and abnormal effects of GABA circuit activity. Above all, the findings point to the possibility that the delay or failure of the GABA switch in a baby's brain may be the causative factor that most or perhaps all autism-spectrum disorders have in common-and thus may be the prime target for therapies.

If further research confirms that hypothesis, then a big question will be whether autistic children can be treated early enough in life to benefit as dramatically as the rodents did in this study. With current diagnostic methods, autism normally isn't diagnosed until the child is a few years old at least. "We are limited by the diagnostic problem-it's not very simple to diagnose autism early in children or babies," says Ben-Ari.

However, he and colleagues recently reported promising results from bumetanide clinical trials in several dozen children aged 3-11, and have founded a company, Neurochlore, to develop the drug for use in autism. "We are now doing our European phase 2 dose-ranging trial which is meant to be finished by the end of the year in five centers and two countries," Ben-Ari says.

The apparent clinical benefit of bumetanide beyond infancy hints that the GABA switch may occur relatively gradually in humans, over years rather than weeks-but that is another big question that remains to be addressed. In any case, it seems likely that researchers eventually will want to test interventions in the perinatal or even prenatal period.

It probably wouldn't be feasible to dose all babies with a drug such as bumetanide-which has side-effects including potassium depletion-to reduce autism risk. "You couldn't do that," says Susan L. Connors, an autism researcher at Harvard Medical School and co-author of a commentary in Science on the new findings. "You'd have to have a biomarker to determine which of these children at birth is likely to develop autism."

One biomarker is the Fragile X gene mutation, which can be found in parents before the baby is born. "We know people whose children are at a high risk of developing Fragile X syndrome, so that might be a good target group," says Contractor.

In principle, excess neuronal chloride levels in newborns would be an even better biomarker, if a delay or disruption of the GABA switch is indeed a major cause of autism. But it seems unlikely that those levels will be easily measurable in live babies any time soon. "What I'm hoping," says Connors, "is that there's a change downstream from intracellular chloride that is evident in accessible tissues, like blood cells."

Another target for interventions might be events that weaken or inhibit the GABA switch. Premature birth and delivery complications, for example, have been linked to higher autism risk. Some researchers, including Ben-Ari, suspect that scheduled Caesarean section births-which probably involve diminished natural oxytocin signaling-may be another risk factor.

Undoubtedly more research will now be done on the GABA switch, in humans and in rodents, to clarify the biological factors involved and the downstream effects, and reveal further possible targets of intervention. Contractor notes, for example, that any link between oxytocin signaling and the expression of chloride transporter genes has yet to be explained. "We still need to work out what the mechanism is," he says. "And even though they've used two separate models here, it would be nice to look at other models of autism too."