Gene therapy — which typically involves delivering healthy copies of a gene to replace faulty ones — is being tested for any number of brain-related disorders, including Parkinson’s and Alzheimer’s diseases. But despite knowing the genes that cause a number of autism-related disorders, there has been little discussion of applying this technique to neurodevelopmental disorders.

A number of animal studies have shown that replacing faulty genes in animal models of single-gene disorders related to autism — including Rett, fragile X and Angelman syndromes — can reverse behavioral and physiological symptoms. 

But how applicable are these findings to people?

A review published 27 June in Neuropharmacology outlines the challenges of developing gene therapy for these disorders.

Steven James Gray, a researcher at the Gene Therapy Center at the University of North Carolina-Chapel Hill, points out two major hurdles. The first lies in the practical difficulties of delivering the corrective copy of a gene across the large human brain. Second, the dose of the corrective copy has to be just right. Studies in animal models suggest that too much of certain genes can have harmful side effects, and too little does nothing.  

At this point, we can only begin to think about gene therapy for the single-gene disorders linked to autism. The genetic culprits for the majority of cases of autism are unknown, but likely to be many.

Rett syndrome, caused by mutations in a gene called MeCP2, offers the perfect illustration of these challenges.

The syndrome is one of the most common causes of intellectual disability in females. Girls who carry a defective copy of MeCP2 begin showing signs of the disorder, such as speech problems and delayed growth, around 18 months of age. Because the gene is on the X chromosome, boys carry only one copy of the gene, and those who inherit a mutation typically die in utero or soon after birth.

Animal research on mouse models with MeCP2 mutations have shown proof-of-principle that gene therapy can reverse symptoms of Rett.

But the experiments also highlight the delicate balance required for a therapeutic effect: In mice engineered to carry a version of the gene that can be turned on and off at will, reactivating it in 70 percent of cells reverses signs of the disorder, but doing so in only half the cells does not. Expressing the gene in control mice leads to severe motor problems.

Given the differences between the size and complexity of mouse and humans brains, it’s unclear what magic percentage of treated neurons might be needed to treat the disease in people, or how well the available delivery methods can disperse good copies of the gene.

Angelman syndrome, caused by mutations in the maternal copy of the UBE3A gene, is characterized by developmental delay and speech deficits. Duplication of the gene is linked to autism.

As with Rett, animal research suggests that gene therapy might help treat Angelman syndrome. In one experiment, researchers found that injecting good copies of the gene into the hippocampus of mice with a UBE3A mutation improves learning deficits. But here again, too high a dose of the gene proves detrimental, as evidenced by autism-like symptoms in people with the duplication.

These are just the technical hurdles. Clinical trials of gene therapy have had a checkered past, including some serious safety issues in early human tests.

As Gray notes, gene therapy technologies are becoming more sophisticated. Scientists are developing safer and more efficient methods for delivering corrective genes, as well as molecular tricks for targeting specific cell types.

One promising strategy for treating Rett syndrome, for example, might be to limit gene delivery to astrocytes, non-neuronal brain cells that help regulate the brain’s chemical environment. Animal studies show that expressing the gene in just these cells alleviates symptoms of the disorder.

But given the delicate dosing issues and the unpredictable nature of gene therapy, I think drugs being tested for these disorders are a more realistic option.