Better animal models of human psychiatric disorders will be vital to developing new types of treatment. Credit: VICTOR HABBICK VISIONS/SPL

Dangle a mouse by its tail, and it will wriggle and strain to escape before eventually recognizing the hopelessness of its situation. Measure the time it takes to abandon thoughts of helping itself, and you have one of the classic animal tests for depression.

Except it's not, says Laurence Tecott, a research psychiatrist at the University of California, San Francisco. "We can't say that that mouse is depressed, and we can't say you would be if you were strung up by your tail," he says. The reason we have not seen a genuinely new class of drug in psychiatry for 50 years, he asserts, is largely because animal models are woefully inadequate representations of human-specific disorders.

You'll hear the same story from many others. But things are not as hopeless for scientists as they may seem for the dangling mouse; some recent papers offer tantalizing hints of a way forward. "No one is going to create a mouse model of suicidality," says Eric Nestler, a neuroscientist at the University of Texas Southwestern Medical Center at Dallas. "But sensible models of important aspects of the neurobiology underpinning psychiatric disorders are just around the corner."

Classical animal tests for psychiatric disorders are based on responses to clinically proven drugs. What the tests don't necessarily do, however, is reflect the cause or the biological basis of the disorder they are supposed to mimic. Most researchers agree that it's time to apply recent findings about the human brain to creating more useful mouse models — often by deleting, adding or mutating candidate susceptibility genes.

For instance, scientists have already modelled a core symptom — abnormal social relations — of several autism-related disorders1, 2. And schizophrenia, a disorder believed to result in part from faulty neurodevelopment, has also been a focus of recent research, because a particularly strong candidate gene has been identified — that encoding the protein DISC-1.

Head start

Psychiatrist Akira Sawa, from the Johns Hopkins University in Baltimore, Maryland, for example, has created transgenic mice with a disrupted Disc-1 gene and used brain scanning to show that such mice have enlarged brain ventricles — especially on the left side — as is usually seen in humans with schizophrenia3. "This brain anatomy is a very objective characteristic of schizophrenia," says Sawa. "And being able to reproduce it in mice with the candidate association gene further supports the hypothesis that DISC-1 is involved in schizophrenia." His lab is now looking at proteins that interact with DISC-1 in key biochemical signalling pathways, and how environmental factors such as stress may affect the system.

A mouse mutant (left) for Disc-1, a gene implicated in schizophrenia, quickly gives up when forced to swim. Credit: W. LI

Alcino Silva, a research psychiatrist at the University of California, Los Angeles, has gone a step further. His team has created a 'conditional mutant' mouse, in which the Disc-1 gene can be switched on or off at will. He found that activation of the gene on just one specific day — the seventh after birth — was sufficient to cause a range of symptoms analogous to those seen in schizophrenia later in life4. The mice developed specific types of memory deficit and depressive-like behaviours, and were less sociable with other mice. In addition, the hippocampus, a part of the brain that is important in memory, had notably different anatomy at the cellular level. "When we went back to look at natural mutations in the gene in humans, we found that the mutation was correlated with reduced sociability," says Silva. "This may not be an accurate model of schizophrenia, but it is telling us a lot about the disease."

Few, if any, scientists want or expect to recreate an entire psychiatric disorder in a mouse. This would be beside the point, not least because current clinical diagnostic criteria, such as those enshrined in the Diagnostic and Statistical Manual of Mental Disease5 (DSM-IV for the fourth, and current, edition) are also considered inadequate. Most neuroscientists and psychiatrists agree that they add to confusion by giving names to clusters of symptoms that overlap with each other. "Schizophrenia and bipolar disorder, as defined by DSM-IV, share many symptoms, as do anxiety and depression," says neuroscientist Steve Hyman, provost of Harvard University in Cambridge, Massachusetts. "The emerging models focus on biomarkers of specific symptoms, rather than a DSM label."

Circuit breaker

Particularly valuable biomarkers, says Hyman, are the neural circuits relevant to disease that have been identified in people using techniques such as functional magnetic resonance imaging (fMRI), which measures changes in oxygenation of blood flowing through the brain. For example, a circuit that runs between the portion of the cortex at the front of the brain, and the striatum and the thalamus deeper in the brain, has been associated with obsessive-compulsive disorder. The symptoms of this disorder can sometimes be alleviated by drugs such as selective serotonin reuptake inhibitors (SSRIs), which enhance the effects of the neurotransmitter serotonin.

Guoping Feng, a neurobiologist at Duke University Medical Center in Durham, North Carolina, hit this target in mice, although not by design. Six years ago, he knocked out a gene in mice called Sapap3, which encodes a key protein involved in regulating receptors for the neurotransmitter glutamate. He noticed that the genetically altered mice groomed themselves until they bled.

At first, he had no idea why. He put things together when he remembered that SAPAP3 is the only member of the SAPAP protein family that is highly expressed in the striatum. And he thus ended up with the first model in which specific disturbances in this neural circuit are associated with a compulsive-like behaviour6. Selectively expressing the missing gene in the striatum alleviated the abnormal behaviour, as did giving the mice SSRIs. "Drug companies have contacted us about testing small molecules they have in their cupboards which regulate glutamate receptors," says Feng.

Current tests continue to pick up drugs that work through the same old mechanisms. Michael Spedding ,

Such research is also stimulating companies to rethink their approach to testing drugs.

During recent decades, the available arsenal of psychiatric drugs has been fine-tuned so that fewer patients have severe side effects. But there has been little improvement in efficacy. SSRIs don't work in one-third of patients with depression, for example, and antipsychotic drugs have little impact on the debilitating cognitive deficits of schizophrenia, even though they control disruptive symptoms such as hallucinations. "Current tests continue to pick up drugs that work through the same old mechanisms," says Michael Spedding, director of research at the pharmaceutical company Servier in Paris, "which is why most schizophrenia drugs target the dopamine system, even though that's not good enough therapeutically." Spedding is now trying to develop surrogates in animals that involve the relevant neural circuits.

Mark Tricklebank, head of science at Eli Lilly's Centre for Cognitive Neuroscience in Windlesham, Surrey, UK, says that many industrial pharmacologists are starting to change their approach in this way. His group, for instance, is developing an oxygen electrode to measure blood-oxygenation levels directly in particular areas of the brain in rodents, as a surrogate of fMRI signals seen in humans.

But it will be some years before the new approach becomes mainstream. "Trying to make drugs is a ferociously complex process," says Paul Chapman, head of GlaxoSmithKline's Centre for Research in Cognitive and Neurodegenerative Disorders in Singapore, "and tolerance to risk varies from company to company". The stakes are huge. A sixth-generation SSRI, Lexapro (escitalopram), has so far made almost US$2 billion for New-York-based Forest Laboratories; by contrast, the last major attempt to launch a genuinely new psychiatric drug, based on the neurotransmitter substance P, failed to show efficacy in several clinical trials for depression.

But it is also early days scientifically. Neuroscientists are quick to point out that their new knowledge about the brain is still preliminary and sketchy. Current candidate association genes are just that — candidates. And the interesting neural circuits are not proven disease mechanisms, just promising correlates.

In the meantime, Tecott is preparing to unveil a new approach to assessing mouse behaviours. It's a holistic approach, whereby the 'lifestyle' of a mouse is assessed in its home cage — a very particular home, where every move is monitored in space and time, along with every lick, every milligram of food eaten, and every snooze. The data can then be mined for information about patterns of behavioural change in the short and long term, for example during the female reproductive cycle, or as a consequence of drug administration.

"Looking at spontaneous behaviour is like looking at brain function on display," says Tecott. "This will help open up ways of finding new psychiatric drugs at last."