Washington DC

Gene therapy has attracted plenty of fanfare but provided very little in terms of positive results. Giving people new genes to remedy defects in their old ones turns out to be a difficult business. But solid if little-noticed progress is being made in an approach that turns the concept on its head: rather than curing conditions, researchers are finding ways to study brain disease by inserting faulty genes into healthy animals.

Several pioneers of the technique presented their latest results at this year's annual meeting of the Society for Neuroscience, held on 12–16 November in Washington DC. They say the animal models they have created are easier to make than those based on alternative methods, mimic human versions of disease better, and can be applied to a wider range of species. “They open up possibilities that other models don't allow,” says Anders Bjorklund, a specialist in neural transplantation at the University of Lund in Sweden.

Some of the most well-developed models are for Huntington's disease, a fatal movement disorder caused by a single faulty gene. The mutant gene expresses altered huntingtin protein, which results in damaged brain cells. One of the first Huntington animal models to be created by gene therapy was achieved in 2002 by Nicole Déglon of the Atomic Energy Commission in Orsay, France (J. Neurosci. 22, 3473–3483; 2002). Déglon injected a virus containing the faulty huntingtin gene into the striatum of rats. The virus inserted itself into neural DNA, causing expression of the huntingtin protein and tissue damage typical of the disease.

Compared with other techniques for creating transgenic animal models, such as adding or deleting genes from embryonic cells and then breeding the animals that develop, the process is quick. “We can create these models in months rather than years,” says Déglon. “And we can replicate the damage seen in late stages of disease, which is not often seen in other transgenic models.”

Déglon is now writing up results from a macaque Huntington model. She says her team has observed motor deficits such as muscle contractions that are typical of the condition, and that post-mortem analysis of brain tissue shows the gene causes damage similar to that seen in humans.

The ability to create primate models using this technique has persuaded others to use viral vectors. Deniz Kirik, a colleague of Bjorklund's at Lund, is finishing a three-year study in which he tracked marmosets infected with a mutant version of the gene for alpha-synuclein — the protein involved in Parkinson's disease. He says the model, the first transgenic primate model of the disease, recreates human symptoms well. The technique also allowed him to use a new control in his experiment: by injecting the faulty gene into just one side of the marmosets' brain, he was able to show that movement problems only developed on the side of the body controlled by that part of the brain.

Other primate models of Parkinson's have used chemicals to kill the neurons that are normally damaged by the disease. “This model is much better — the pathology is driven by mechanisms similar to those in humans,” says Kirik.

Such advantages mean that the use of viral-vector models is likely to grow beyond the ten or so groups that have taken them up over the past few years. But Bjorklund notes that the technique will not replace traditional transgenic models. For example, conventional techniques allow large numbers of animals with identical mutations to be bred, which is useful for drug screening. The surgical skills needed to inject the virus into the correct brain areas are also difficult, adds Bjorklund, and may put off some labs.