3 December 2007. Models were in vogue at this year’s annual meeting of the Society for Neuroscience in San Diego. Not the air-brushed ones that you might see in glossy magazines, mind you, but models that are far more attractive to researchers. Francine Benes, McLean Hospital, Belmont, Massachusetts, chaired a mini-symposium dedicated to animal models of schizophrenia. Presentations ran the gamut from genetic-based models to chemical and toxin treatments.
Weidong Li from the University of California, Los Angeles, started the symposium by describing the most recent addition to a growing list of DISC1 mouse models. Mutations in DISC1 (disrupted in schizophrenia 1) have been linked to schizophrenia and other major psychiatric illnesses, and the protein is believed to play a major role in neurodevelopment. To test whether DISC1 failure during development translates to deficits later in life, Li and colleagues have engineered a reversible, dominant-negative DISC1 construct that can be turned on and off at will. This reversible, inducible DISC1 mouse was just recently described in PNAS, and SRF has already outlined the major findings (see SRF related news story). Briefly, Li described how turning on the dominant-negative DISC1 during early development (postnatal day 7) causes problems for adult mice, including compromised spatial working memory, depression, and asocial behavior. The mouse supports the neurodevelopmental model of schizophrenia and could prove useful in teasing out exactly what developmental pathways are affected.
Another genetic-based model was described by Tobias Halene from RWTH Aachen University, Germany. This model is based on the glutamate hypothesis of schizophrenia, which suggests that NMDA-type glutamate receptor function is compromised in the disease (see SRF current hypothesis). Halene and colleagues bred transgenic mice that are NMDA receptor hypomorphs, meaning that they only produce about 5-7 percent of the normal levels of the NR1 glutamate receptor subunit. When put through a battery of behavioral and electrophysiological measurements, these animals showed some deficiencies compared to normal mice. The typical P20 and N40 sound-induced event-related potentials were higher and lower, respectively, than in wild-type animals (similar changes in analogous human P50 and N100 ERPs, have been proposed as endophenotypes of schizophrenia). In tests of behavior, the NMDAR hypomorphs spent more time in open sectors on an elevated maze, and they spent less time checking out new mice introduced into their environment. But the animals showed no difference in locomotor activity. Halene observed that previous publications have observed much higher locomotor activity in NMDAR-compromised animals, and he suggested that the environment in which the tests are made is crucial. In the elevated maze and the sociability test, he detected no locomotion difference, as measured by beam breaks, between hypomorphs and normal mice. Halene concluded that NMDAR deficiency results in a lack of behavioral inhibition that might mimic some of the symptoms and behaviors in schizophrenia.
The NMDA theme was also featured in several other presentations. Chalon Majewski-Tiedeken and colleagues the University of Pennsylvania use the NMDA antagonist ketamine to modulate NMDAR hypofunction. Ketamine is a widely used veterinarian anesthetic, but recreational abuse of the drug has been linked to emotional and behavioral disturbances, including anxiety, hallucinations, paranoia, and cognitive disruption, that are also experienced by people with schizophrenia. Ketamine binds noncompetitively to NMDA receptors in GABAergic, serotonergic, and noradrenergic neurons, but whether the drug induces any permanent anatomical damage is unclear.
Majewski-Tiedeken has addressed this by giving frequent, sub-anesthetic doses of ketamine to several strains of inbred mice. She reported that brain sections of treated mice tested positive for both apoptosis (caspase 3 immunolabeling) and neurodegeneration (de Olmos silver staining) in the CA3 layer of the hippocampus. She suggested that excitotoxicity, as a result of compensatory glutamate increase, and cell death may explain some of the long-lasting effects of ketamine exposure. This may sound counterintuitive, given that the NMDA antagonist memantine is currently approved for treating neurodegenerative conditions such as Alzheimer’s disease. Majewski-Tiedeken suggested that the situation may be highly complex and that in Alzheimer’s, stroke, or other conditions where a toxic insult has already occurred, glutamate antagonists may be protective at low doses but toxic at high doses.
From the same laboratory at U. Penn, Richard Ehrlichman reported that NMDAR antagonists can elicit in mice some of the same alterations in brain waves seen in schizophrenia patients. In humans these waves, or electrophysiological oscillations, are generally lumped into four main categories: theta (4-7.5 Hz), alpha (8-13 Hz), beta (14-30 Hz) and gamma (30 Hz and above). Disturbances in these bands, particularly the gamma, have been recorded in patients with schizophrenia and have been proposed as endophenotypes of the disease (see SRF related news story and SRF news story).
Ehrlichman described studies to determine if NMDA hypofunction or dopamine hyperfunction in mice can mimic some of the human endophenotypes—hyperactivity of dopaminergic neurons is another well-accepted hypothesis for schizophrenia (see SRF current hypothesis). He administered ketamine and amphetamine to mice to elicit hypo- and hyperfunction, respectively, and then used electroencephalograms (EEGs) to measure various frequency ranges. Ehrlichman recorded baseline EEGs and also evoked EEGs following an audible paired click. The EEG waveforms were band-pass filtered and averaged for each mouse.
He reported that ketamine had no effect on baseline wave powers except in the gamma frequency range where the power was increased. The NMDA antagonist also elicited decreases in evoked theta and alpha waves but increased gamma waves. These responses are consistent with gamma and theta changes seen in people with schizophrenia. He found that amphetamine had no effect on baseline power in any frequency range, but it did significantly decrease the power of the evoked theta wave. The findings suggest that dopaminergic hyperfunction alone is insufficient to recapitulate in mice the changes seen in human patients.
A slightly different angle on the dopamine hypothesis was offered by Alain Louilot, University Louis Pasteur, Strasbourg, France. Louilot and colleagues are interested in why latent inhibition is reduced in some people with schizophrenia. Latent inhibition (LI) is a phenomenon in which a conditioned response is reduced or even eliminated if the stimulus is first given without the condition (think how Pavlov’s dogs would have responded to a bell if it was originally rung in the absence of food). LI is believed to be important for processing sensory information, something that often presents difficulty to people with schizophrenia.
Louilot and colleagues wondered if LI in adults is related to problems in the hippocampus during neurodevelopment. To test this in rats, the researchers introduced a reversible lesion in the subiculum, the output center of the hippocampus, by injecting tetrodotoxin (TTX) at postnatal day 8. The idea is that this lesion would disrupt the normal development of the brain. The researchers then looked for electrophysiological abnormalities in neurons several stages downstream, specifically neurons in the anterior striatum that are innervated by midbrain dopamine neurons. Sixty-two days after TTX injection, the researchers measured in vivo electrical activity while testing latent inhibition. The authors found that developmental inactivation of the subiculum induced abnormalities of the LI-related dopaminergic response in the anterior striatum. The findings suggest that latent inhibition deficits in adults may have their origins in early developmental changes.
Barbara Gisabella and colleagues at McLean Hospital have used a similar strategy to probe the relation between GABAergic currents in pyramidal cells of layers CA2/3 of the hippocampus and input from the basolateral nucleus (BLa) of the amygdala. Previously, the Benes lab showed that there is loss of GABAergic current in the hippocampus following infusion of picrotoxin, a GABA-A antagonist, into the amygdala. Because altered GABAergic tone is thought to be a major facet of schizophrenia, this picrotoxin treatment could serve as a model for the disease.
To gain a better understanding of exactly what happens in the hippocampus following picrotoxin infusion, Gisabella has examined hippocampal slices. Her hypothesis is that increased activity from the BLa in response to GABA antagonism might cause changes in the membrane properties of fast-spiking interneurons in the hippocampus.
Gisabella reported that administration of picrotoxin to postnatal-day-30 rats led to altered electrophysiology in hippocampal slices taken 15 days later. In the CA2/3 layer she found decreased action potential duration, decreased resting membrane potential, and increases in spike frequency. In contrast, she reported that in the CA1 layer, electrical activity was similar in slices taken from saline- and picrotoxin-treated animals.
What might lead to shorter action potentials and more rapid firing of hippocampal neurons? Gisabella noted that postmortem studies of schizophrenia patients have revealed upregulation of the gene that codes for the hyperpolarization-activated (Ih) potassium channel in layer CA2/3 of the hippocampus. Could these channels contribute to the increased activity? Gisabella found that the Ih current amplitude was increased in hippocampal slices taken from picrotoxin-treated animals, and she proposed a model whereby blockage of GABA in the amygdala leads to an increase in Ih activity and an alteration of the intrinsic membrane properties of CA2/3 interneurons in the hippocampus. These findings could help shed some light on the mechanisms by which interneuron firing is altered in schizophrenia, she said.
Administering other chemicals, not necessarily neurotransmitter agonists, antagonists, or toxins to mice, might also mimic some of the pathology of schizophrenia. Patricia Tueting, University of Illinois, Chicago, described the use of L-methionine to induce DNA methylation and downregulation of key genes that have been implicated in schizophrenia, including those for reelin (see SRF related news story) and glutamic acid deacarboxylase 67 (GAD67), which is necessary for GABA synthesis (see SRF related news story).
Tueting reminded the audience that 30 years ago clinicians had figured that methionine would actually help schizophrenia patients and entered into a small trial to test that hypothesis. But after the amino acid was given to 11 patients, seven of whom actually got worse and four showed no improvement, the idea was dropped. We now know, suggested Tueting, that in schizophrenia DNA methyltransferase activity is elevated in GABAergic neurons, and since methionine is a precursor for S-adenosylmethionine, a methyl donor, the amino acid was probably feeding modification of methyl sensitive genes such as RELN.
Tueting tested this hypothesis by administering methionine to mice and looking at key downstream effects including loss of dendritic spines, which are crucial for the proper function of neural networks. Reelin is, in fact, important for maintenance of spines, and reelin heterozygous mice have both loss of spines and GAD67. Tueting reported that administration of methionine to mice over a 1- to 2-week period resulted in significant loss of spines on the apical dendrites of pyramidal cells in prefrontal cortex, which extend into the GABAergic layer. Tueting demonstrated that it was specifically spines that are located about 3 Sholl units from the cell body that were most affected, the same region that is most affected in reelin heterozygous animals (Sholl units measure the radius out from the cell body). The findings suggest that administration of methionine could be used as a simpler model of reelin deficit. Tueting did not say whether these mice have any behavioral or functional problems, but she did say that the effect was reversible. Withdrawal of methionine was followed about 12 days later by a return to normal spine density, giving the model some valuable flexibility.
Animal models can serve not only to study disease pathology but also to develop and refine treatments, as exemplified by the work of Cara Rabin, University of Pennsylvania. Using animal models, Rabin demonstrated how a long-term delivery system could be used to overcome a major problem in schizophrenia treatment: failure to adhere to medication. Rabin noted that 74 percent of schizophrenia patients are unable to adhere to medication within 2 years of starting on a prescription and that as relapses increase, it becomes more and more difficult for patients to gain stability.
Non-adherence may certainly be linked to adverse side effects, but they can also be attributed to the inconvenience of taking daily medication. Despite this, there has been little study done on using implantable, long-term delivery approaches, said Rabin. One of the reasons for this is that DEPOT-type drug treatments are irreversible, she suggested.
Rabin has developed a removable implant comprising a polymer of polylactic and polyglycolic acids (PLGA) and risperidone, a second-generation antipsychotic drug. She reported that these implants are stable, release drug over a period of at least 56 days, and are easily removed. They also are capable of modifying behavior in animal models, as measured by pre-pulse inhibition to the startle response and event-related potentials. Rabin suggested that this type of long-term delivery could lead to reduced morbidity and mortality from untreated psychosis and perhaps other chronic conditions.—Tom Fagan.