Breeding Schizophrenia in the Lab

by Tom Valeo

November 19, 2013

How would a schizophrenic mouse behave?

It would hear voices, and experience paranoid delusions, like many humans with the disorder. It would become apathetic, have trouble concentrating, withdraw from other mice, and perhaps develop alogia, also known as poverty of speech.

Producing a testable mouse that displays the conspicuous symptoms of such a distinctly human disease seems impossible, especially since the disease manifests itself primarily through the uniquely human capacities for language and imagination. Without a reliable animal model, however, the search for drug treatments will lag, and any insights into the brain malfunctions that produce the symptoms of schizophrenia will depend on the study of human patients, whose brains are usually physically and morally off-limits.

But mice don't get Alzheimer's disease or Parkinson's disease either, and useful mouse models of those diseases exist. Or rather, mouse models exist for various aspects of those diseases. Why couldn't schizophrenia be broken into components, with each one modeled by a separate strain of mouse?

That's precisely what researchers at trying to do. A mouse has been bred at Johns Hopkins University in Baltimore, for example, that lacks the enzyme BACE1, which contributes to the accumulation of protein fragments believed to damage neurons in Alzheimer's disease. However, these mice also become hyperactive when placed in an unfamiliar situation-a behavior reminiscent of the agitation of people in the grip of schizophrenia. The mice also continue to startle in response to a loud noise or bright flash even when alerted by a warning signal, which normally quells the startle reflex-except in people with schizophrenia, Alzheimer's disease, and a few other disorders. In addition, the mice appear to lack interest in new mice, and they show deficits in working memory-two more behaviors that resemble symptoms of schizophrenia. The clincher? All of these behaviors improve when the mice are given clozapine, a common antipsychotic drug.

Why would a mouse created to help researchers find treatments for Alzheimer's disease seem to display symptoms of schizophrenia?

Researchers found that BACE1, in addition to producing the protein fragments believed to initiate Alzheimer's, also plays a role in the pathway for neuregulin1, a protein implicated in schizophrenia. (It appears to disrupt the formation of dendritic spines on neurons.)

"When we discovered that BACE1 plays a role in the so-called neuregulin pathway, we started to look at the BACE1 knockout mice as a possible model of schizophrenia," said Alena V. Savonenko, an associate professor at the Johns Hopkins University School of Medicine, and the lead author of a paper on the subject published in PNAS in 2009. (please see full citation below, 1)

Savonenko does recognize the limitations of this model of schizophrenia. "No animal model could encompass the complexity of this human disease, so you must try to interpret what part of the symptoms this model represents," she said. "Only by having a number of models, each providing information about one particular mechanism behind the symptoms, could you ever develop a complete model of the disease."

Since schizophrenia runs in families, researchers hope to create strains of mice engineered to carry human genes suspected of contributing to symptoms. This type of genetic engineering has become commonplace. The only problem is that schizophrenia appears to be caused by contributions from dozens and possibly hundreds of genes.

Joshua Gordon, a psychiatry professor at Columbia University, focuses on one of them-22q11, a gene regarded as one of the largest genetic risk factors for schizophrenia. A microdeletion in the gene increases the risk of schizophrenia 30 times. Two of his collaborators, Maria Karayiorgou, of the New York State Psychiatric Institute, and Joseph A. Gogos, of Columbia University, developed a mouse model of schizophrenia by knocking out 22q11. When they studied the mice they observed deficits in working memory and other cognitive functions; this work was published in Nature in 2010.(2)

"By deleting 22q11, the connections between the hippocampus and the prefrontal cortex of the mouse are disrupted," Gordon said. "With genes, I know I'm studying a causative agent, and I can test my hypothesis that the 22q11 gene causes cognitive symptoms in schizophrenia by interfering with the hippocampal prefrontal system."

Lin Mei, director of the Institute of Molecular Medicine and Genetics at Georgia Regents University in Augusta, has made mouse models lacking genes for ErbB4 and neuregulin-1, which work together to balance inhibition and excitation in pyramidal cells in the prefrontal cortex; some of this work was published in PNAS in 2009. (3)

"Because of the Human Genome Project, some susceptibility genes associated with schizophrenia have been identified, and they gave us a handle on how schizophrenia may work," Mei said. "Neuregulin and ErbB4 are both susceptibility genes, so we generated mouse models of each and found behavioral deficits in them."

Some of the mice appear to be hyperactive, for example, and run around furiously for no apparent reason. Some continue to jump when startled, despite a warning signal that alerts them to the approaching stimulus. Others have problems with working memory.

"We are very excited by this," Mei said, "but we have a long way to go to understand such complex behavior."

Mei suspects that schizophrenia, like autism, exists on a broad spectrum, with symptoms varying widely from person to person. "I predict that in 50 years schizophrenia will be grouped into 20 or even 50 different diseases, each one caused by a different combination of mechanisms," he said. "Neuregulin and ErbB4 may account for no more than 2 percent of cases. There's no way for us to create a model that would duplicate every single symptom in schizophrenic patients, but we hope we can create models piece by piece to tease out the mechanism of schizophrenia."

That means piecing together a multitude of tiny clues until a clear picture of schizophrenia begins to emerge-an arduous and perhaps futile task in the opinion of Peter C. Williamson, chair of neuroscience and mental health at the University of Western Ontario's Schulich School of Medicine.

"People have done some very clever things to mimic aspects of schizophrenia," he said. "I think it's possible to model some aspects of the illness, but not the illness itself."

Williamson is the co-author, with John C. Allman of the California Institute of Technology, of The Human Illnesses: Neuropsychiatric Disorders and the Nature of the Human Brain.  They contend that neuropsychiatric disorders such as schizophrenia, bipolar disorder, autism, and others, result from the malfunction of characteristics unique to the human brain.

"These are human disorders," they state in the preface. "If we could understand more about what makes the human brain unique, then maybe we might understand how it breaks down, resulting in these conditions. The brain breaks down in discrete ways, and how it breaks down tells us something about how it was put together." (4)

Williamson considers some of the mouse models potentially useful, but he still catches himself asking, "Is this really relevant to schizophrenia?"

His answer? "Probably not. I think schizophrenia, and probably bipolar disorder and autism as well, are uniquely human disorders. They affect circuits present only in humans. They depend on higher cortical networks that have developed only in humans."

In their book Williamson and Allman contend that the human illnesses depend heavily on the uniquely human ability to think about the thoughts, feelings, and actions of the self and others.

"Other animals really don't do that," Williamson said. "Our brain is designed to do that, and several regions have evolved to enable us to do that. The disorders we have probably arise from this capacity. That's why when I look at people trying to develop a mouse model of schizophrenia, I think, good luck."



Sources cited:

1. Alteration of BACE1-dependent NRG1/ErbB4 signaling and schizophrenia-like phenotypes in BACE1-null mice. Savonenko AV, Melnikova T, Laird FM, Stewart KA, Price DL, Wong PC. PNAS 2009;105(14):5585-5590.

2. Impaired hippocampal-prefrontal synchrony in a genetic mouse model of schizophrenia. Torfi Sigurdsson, Kimberly L. Stark, Maria Karayiorgou, Joseph A. Gogos, and Joshua A. Gordon. Nature 2010;464(7289):763-767.

3. Neuregulin 1 regulates pyramidal neuron activity via ErbB4 in parvalbumin-positive interneurons. Lei Wen et al. PNAS 2010;107(3):12111-1216.

4. The Human Illnesses: Neuropsychiatric Disorders and the Nature of the Human Brain, by Peter C. Williamson and John M. Allman. Oxford University Press 2011.