Mental illness would seem to be singularly human, a subjectively experienced distortion of consciousness affecting uniquely human attributes: thought, feeling, and language.
For that reason, it presents a unique challenge to the basic scientist seeking to reproduce its symptoms in experimental animals commonly used in the laboratory. Diabetes, high blood pressure, cancer, or other somatic conditions may be reliably replicated, but how does one reproduce depression, anxiety, or delusional paranoia in a laboratory rat?
"Modeling mental illness in animals still seems to many people to be an outrageous idea," Barbara Lipska, Ph.D., of the clinical brain disorders branch of the National Institute of Mental Health (NIMH), told Psychiatric News. "People cannot believe that psychiatric disorders can be modeled in a rat or a mouse or a primate because these disorders are believed to be inherently human. Delusions, hallucinations—how are we possibly able to reproduce these symptoms in an animal, and even if we do it, how would we know, since the animals cannot communicate verbally?"
But Lipska and other scientists say that in fact they are able to produce with greater and greater reliability certain behaviors in experimental animals—if not the underlying neuroanatomical or biochemical disorder itself—that are analogous to the behaviors in humans with mental illness and that are the phenomenological reflection of that underlying human disorder.
Daniel Weinberger, M.D., chief of the clinical brain disorders branch, said that new genetic technologies and other refinements are expanding the research potential in animal modeling. Today, scientists at NIMH are modeling schizophrenia in rats and disorders of memory and cognition in mice.
Elsewhere, researchers including Rene Hen, Ph.D., of Columbia University College of Physicians and Surgeons, and Irwin Lucki, Ph.D., of the University of Pennsylvania, are modeling depression and anxiety in laboratory animals.
"You can do experiments in animals that you cannot do in humans," Weinberger said. "The reason we do it is to help us understand underlying disease mechanisms, test causal hypotheses, and find new treatments."
Experimentally reproducing the symptoms of schizophrenia—a complex disorder with a spectrum of symptoms and no definitively known cause—epitomizes both the pitfalls and the possibilities of modeling mental illness in animals.
Lipska and colleagues at NIMH have succeeded in producing an array of symptoms in laboratory rats that serve as surrogates of the symptoms of schizophrenia in humans. Moreover, the appearance of the symptoms mimics in important ways the pattern of the human disorder, providing important laboratory corroboration for developmental hypotheses about its cause.
She explained that rats are particularly suited to modeling a complex disease because they are smarter than mice and are known to be particularly social animals. She and her colleagues have employed a "lesion model," inducing an insult to the rat hippocampus—a region of the brain known to be implicated in human schizophrenia—through the injection of a toxin.
They found that rats with a damaged hippocampus displayed a variety of symptoms analogous to the human disease—hyperlocomotion in response to stress (similar to the vulnerability to stress experienced by many patients), deficits in memory testing (much like the deficits in working memory in humans with the disease), and reduced social contacts (analogous to the social withdrawal exhibited by many people with schizophrenia).
"We can reproduce a constellation of behavioral symptoms," Lipska said. "Schizophrenia is not just one symptom; it’s a syndrome of behavioral abnormalities, so we consider it very important to be able to produce an array of behavioral changes. If the model reproduces this constellation of changes, then we can say to a certain extent that it is a good model of the disease."
More important even than the symptoms themselves is the similarity between the pattern of behavioral symptoms in the animals and the pattern of disease as it frequently occurs in humans. It is a remarkable serendipity that would appear to lend support to the "neurodevelopmental" hypothesis of schizophrenia.
The hypothesis holds that schizophrenia begins with a genetic mutation that results, among some who carry the vulnerability, in a defect in development during the third trimester of gestation, when critical brain structures are formed.
This developmental defect may lay more or less dormant until adolescence or early adulthood when the environmental and social stresses of the period begin to overwhelm the individual whose higher brain functions are compromised; it is during this period that many patients experience their first psychotic episode.
Analogously, Lipska and colleagues also saw a dormant period among their lesioned rats, a delay between the introduction of the insult to the hippocampus and the appearance of the behavioral changes. When the changes did occur, it was at a period of the rat’s life that correlates with the period of human adolescence, when schizophrenia often appears.
The timing of the insult to the rat hippocampus at seven days after birth—a period of brain development that corresponds to the in-utero period when human brain structures are forming—is likewise crucial. If the toxin is introduced earlier or later, the behavioral changes do not occur or occur differently.
"We have to produce the lesion at a critical time, which shows that—as in the human disease—there is some critical window of vulnerability," Lipska reported.
In similar work, Lalit Srivastava, Ph.D., and colleagues at McGill University School of Medicine in Montreal have experimented with lesion models, but also with modeling changes in neuronal activity in rat brains, reflective of the subtle changes believed to accompany human schizophrenia.
Srivastava told Psychiatric News that the effort is an attempt to reproduce more faithfully the complexity of the disease and to address a difficulty with the "lesion model"—namely, that the brains of humans with schizophrenia do not, in fact, show large holes or lesions.
"What we are doing is based on the belief that schizophrenia is not really a frank lesion, but a lot of subtle changes in neuronal functioning," he said. "To follow up this line of reasoning, what we are doing is not killing cells, but simply blocking certain types of neuronal activity."
In particular, Srivastava and colleagues have targeted the NMDA glutamate receptor in circumscribed regions of the brain, especially the prefrontal cortex, believed to be involved in schizophrenia.
After injecting the rat brains with an NMDA-antagonist, known as AB5, the animals are left to develop in their cages. As with Lipska’s lesioned rats, Srivastava’s rats experienced a delay before the onset of behavioral changes. These changes include diminished social interaction, mirroring the social withdrawal common among humans with the disease.
Srivastava and colleagues have also been able to model some symptoms of schizophrenia using genetically altered mice. (While rats are better animal models because of their superior intelligence and their social nature, the mice genome is more completely mapped than that of the rat.)
He explains that past research published by his laboratory showed that postmortem brain samples of patients with schizophrenia showed decreased levels of a protein known as N-CAM, believed to be involved in neuronal development.
Following up on this finding, Srivastava and colleagues used a genetic "knockout" strategy to delete the gene for N-CAM in experimental mice. The result was that the mice displayed certain behavioral problems analogous to the human disease; in particular, they displayed what scientists call a deficit in "prepulse inhibition of startle."
This refers to a defect, also seen in humans with schizophrenia, in processing of sensory motor information. Healthy people, when tested with a loud or disruptive noise, exhibit a startle reflex; but in test conditions in which the loud noise is preceded by a softer, less-intrusive noise—a warning of sorts—healthy people will be able to modulate their startle reflex to the louder noise.
Not so with some schizophrenia patients; even when they are "warned" by the softer noise, they are unable to inhibit the startle reflex that occurs when they hear the louder noise. So, too, Srivastava’s genetically altered mice are unable to inhibit the startle reflex.
But Srivastava noted that research has failed to find a genetic association between the N-CAM gene and schizophrenia, and the postmortem finding in brain samples does not suggest a causative association.
"Gene models have problems because although schizophrenia is a genetic disorder, it is more likely to be due to abnormal action of multiple genes working in concert to predispose an individual to the disease," Srivastava said. "And this predisposition is acted upon by environmental factors to lead to frank schizophrenia."
In just this way are animal models likely to remain incomplete reflections of complex human mental illness. "In contrast to some somatic disorders, modeling psychiatric disorders is also problematic because we don’t know all of the causative or pathological hallmarks," he told Psychiatric News. "The goodness of a model depends on what we know about the disease. But there is not one schizophrenia, and there may not be one depression. These are a spectrum of disorders, with a range of symptoms and severity.
"What are we modeling?" Srivastava asked. "Are we modeling the total disease? I prefer to think that we are testing hypotheses. Since we don’t know the real causes and pathological hallmarks of complex mental disorders, we use animal models to test hypotheses that have been advanced." ▪