Fear factor: The genetics of anxiety 20/4/07. By Jon Turney Tolstoy declared that every unhappy family is unhappy in its own way. After 100 years of genetics, psychologists might disagree.

Amid all the subtle differences that fascinate a novelist, depression does seem to run in families. This is the first line of evidence that the origins of this most common psychiatric disorder are influenced by a person's genetic make-up. Going beyond this to work out just how genes might predispose someone to depression, or other psychiatric disorders, has proved frustratingly difficult, though.

Now, Professor Jonathan Flint and colleagues at the Wellcome Trust Centre for Human Genetics in Oxford are beginning to make real progress in uncovering the details – using a new approach to tracking small genetic effects in mice.

The idea of heritable factors influencing psychiatric conditions goes back a long way, but Professor Flint first got on the trail of genes as a researcher in London in the early 1990s. A series of papers offering evidence of specific genetic associations with manic depression and schizophrenia appeared around that time, but the effects tended to disappear on closer analysis. "So either psychiatrists didn't know what they were talking about when they made a diagnosis, or there were no genetic effects," he says.

Unwilling to accept either conclusion, he wondered whether there was another way into the problem. Together with the experimental psychologist Professor Jeffrey Gray, he decided the behaviour of mice might be it. "It was the one-eyed leading the blind, as neither of us knew anything then about genetics in mice." But Professor Gray knew people who did, notably his then expatriate colleague Professor David Fulker at the Institute for Behavioral Genetics in Boulder, Colorado.

It was a bold move at the time, as it involved extrapolating from animals to humans in an area of hot dispute about whether stable traits exist in people, let alone mice. You have to argue that the two mammals are alike enough to have similar basic processes going on in their brains, and that some of the results are related, too.

On that basis, a collaboration with Boulder ensued, which quickly bore fruit. The then Dr Flint found himself first author of a landmark paper, published in Science a dozen years ago, and boldly entitled 'A simple genetic basis for a complex psychological trait in laboratory mice'. The team had made a series of crosses in a standardised strain of mouse, and looked at the genes of those in each generation that were high or low on 'emotionality', on basic behavioural tests. Although they only used 84 markers to cover the whole mouse genome, they were delighted to find three that seemed to account for all of the variation in this trait.

But what had they actually found? Technically, these were quantitative trait loci (QTLs), regions of a chromosome associated with a characteristic that can be measured on a regular scale. This is a long way from fixing on a particular gene, still less knowing how it might affect the brain in a way that influences behaviour. The stretches of chromosome highlighted by this analysis probably contained hundreds of different genes, maybe thousands.

That was in 1995, before the age of complete genome sequences. So has the constant refinement of detailed genetic information since then – with multiple genomes of mice and men piling up in the databanks – pinpointed the precise changes that make a mouse more or less neurotic? Sadly not.

The whole business of turning information about the rather loosely defined QTLs into tightly identified genes, or even changes within those genes, has proved hugely problematic. Professor Flint recalls that small word 'simple' in his old, much-cited paper with a rueful smile: "That title haunts me to this day."

The QTL problem is not restricted to behavioural traits. It turns out to be a rather common obstacle in the current phase of post-genomic biology. In principle, we can find out pretty well anything about a creature's genome we want, in as much detail as anyone cares to ask for. In practice, the devil in those details still has a few tricks up his sleeve to confound the gene-hunters.

Many researchers have tried to come up with general strategies for hunting down individual genes that affect traits of interest. One approach is to induce new changes in genes and track their effects. Point mutations (a single base pair) induced with chemicals can now be tracked relatively easily, and painstaking phenotypic analysis reveals which ones might be in genes connected with a particular trait.

This method "has produced some interesting biology", according to Professor Flint, but not so far for behaviour. This is probably because the phenotypes are affected by so many complex influences, he suggests.

Professor Flint and his lab took another route, using a new colony of mice with mixed-up genes, instead of the simple pedigrees normally demanded of laboratory suppliers (see box). As the method they wanted to test was generally applicable to complex phenotypes, not just behaviour, they worked up a detailed profile on their mice that went well beyond behaviour. Their battery of tests included measures related to type II diabetes and asthma, as well as anxiety, and a host of basic indicators such as blood pressure, immune function, growth rate and so on. These are well-studied mice.

Establishing this new resource, and demonstrating how it can be used, has taken a good few years, also spanning Professor Flint's move from London to Oxford, first to the Institute of Molecular Medicine (now the Weatherall Institute), then to the Centre for Human Genetics. But it is now ready for wider exploitation. "We have a very general and comprehensive tool for examining what I would call genetic architecture."

That achieved, his own interest remains the problem he began with. So where does this leave the genetics of depression, anxiety or even just neuroticism? In general, the analysis now points to around ten loci that can affect each phenotype. All of them together account for between half and three-quarters of the variation in a trait. The picture for anxiety looks likely to fit this typical pattern.

So, not surprisingly, the ability to detect smaller and smaller effects makes the influence of more and more genes visible. We have moved a long way from any notion of a 'gene for depression'. Soon, there will also be clues from the transcriptome – snapshots of which genes are active based on levels of messenger RNAs read off from the DNA. And this in turn may need to be related to the levels of particular proteins in different parts of the brain.

Where will this lead? Like many others, Professor Flint looks toward the emergence of a new understanding of systems biology. That is, understanding what is really going on will mean finding ways of relating data from all these different levels. It is already clear that there are many complex interactions between the genes that can affect these behaviours, and between the genes and the environment and life history of each individual mouse. Doubtless the same holds true for people.

But genes are still one way of getting a handle on new therapeutic possibilities for psychiatric illness. On one hand, he says: "The problem with all these genetic approaches is that they chase not genes, but genetic effects." On the other: "In the end, all we are interested in is the mechanism. We don't care about the genes as such."

So the real goal is "to use the genetics to give us some information about the physiology we don't know about". One direction he plans to explore there is the production of new neurons, especially in the hippocampus – as the effects of existing antidepressants are thought to be bound up with neurogenesis. But it is one thing to know which process is affected, and another to understand how this comes about. Professor Flint wants to go to the next level. "What molecules are involved? That's the sort of question we could address."

Jon Turney is a science writer based in London.

Image credit: Adrian Cousins

Further reading

Flint J et al. A simple genetic basis for a complex psychological trait in laboratory mice. Science 1995;269(5229):1432–5. Abstract

Willis-Owen SA, Flint J. Identifying the genetic determinants of emotionality in humans; insights from rodents. Neurosci Biobehav Rev 2007;31(1):115–24. Abstract

Valdar W et al. Simulating the collaborative cross: power of quantitative trait loci detection and mapping resolution in large sets of recombinant inbred strains of mice. Genetics 2006;172(3):1783–97. Full text

Mott R. Finding the molecular basis of complex genetic variation in humans and mice. Phil Trans R Soc B 2006;361(1467),393–401. Abstract

Flint J et al. Strategies for mapping and cloning quantitative trait genes in rodents Nat Rev Genet 2005;6(4):271–86. Abstract