Rodent models of schizophrenia -- Annie Bryant

Kellendonk et al. and Moore et al. developed two distinct rodent models of schizophrenia that both incorporate disruption of normal embryonic development. Kellendonk et al. developed a double-transgenic mouse line in which tet-off dopamine-2 receptors (D2Rs) are reversibly overexpressed in the striatum, a region implicated in schizophrenia pathophysiology. They cite studies dating back to Arvid Carlsson's 1963 dopamine hypothesis of schizophrenia, many of which associate aberrant dopamine transmission in the striatum and prefrontal cortex with schizophrenia. Moore et al. delivered the cytostatic agent MAM to timed-pregnant rat dams at embryonic day 17, during which cortical structures are still developing but subcortical and cerebellar structures are all set. They developed this model based on evidence for the role of abnormal development of frontal and limbic cortical circuits in schizophrenic basal ganglia and dopaminergic systems. They suggest that MAM-induced DNA methylation may modulate expression of genes that are essential for cortical neuron development and plasticity, including GAD67, a gene involved in GABA transmission that is downregulated in schizophrenia. 

Both models yield deficits in working memory and behavioral flexibility, two characteristics of schizophrenia symptomology. Kellendonk et al. used the delayed non-match to sample maze tasks, which require an intact PFC to perform successfully. The D2R-transgenic mice exhibit impaired task acquisition in two types of DNMTS tasks, which persisted even when doxycycline was administered. In fact, significantly fewer D2R-transgenic mice reached DNMTS T-maze task criterion after D2R expression was normalized, suggesting that normalization in adulthood after chronic expression has deleterious effects on cognition. Even transgene suppression at P0 wasn't able to prevent the selective cognitive deficits induced by developmental D2R overexpression. However, they report normal locomotion, sensorimotor gating, general cognition, and baseline anxiety in these D2R-transgenic mice, unlike the MAM-E17 model, which yields impaired sensorimotor gating. Moore et al. used pre-pulse inhibition (PPI) tests to show that in MAM-E17 rats, there is a lack of response extinction when a low-decibel sound is played before a high-decibel startle sound. Like Kellendonk et al., Moore et al. show impaired MAM-E17 behavioral flexibility in a reversal learning task, with MAM-E17 rats learning novel discrimination tasks faster but learning reversal tasks much slower than controls. The authors interpret this to mean that while sensory and motor association learning are unaffected, there is a perseverative response in MAM-E17 rats that becomes especially clear upon contingency reversal.

There are also some interesting physiological phenotypes associated with these two models. Moore et al. also examined membrane properties of E-17 neurons in the PFC and ventral striatum (vSTR), which in control mice regularly oscillate between “up” and “down” states of depolarization. The “up” state, from which action potentials are selectively generated, is regulated by convergent excitatory synapses, enabling the neurons to function as “coincidence detectors”. MAM-E17 neurons lost this bistable membrane phenotype, exhibiting a relatively depolarized baseline and higher spike threshold. The authors suggest that this loss of state-dependent spike firing impairs processes like response selection and may underlie problems with perseveration and forming new associations, as striatal output isn’t regulated by thalamic and cortical input. Kellendonk et al. identified what appears to be a fragile counter-balancing mechanism between D1R and D2R levels in the PFC, with even the slightest perturbance to the circuit throwing the balance out of whack. In their D2R-overexpressing model it seems that increased striatal D2R density leads to working memory deficits via excess D1R activation in the PFC, along with elevated overall DA levels and decreased DA turnover. Although D1R antagonists haven’t been clinically successful, Kellendonk et al. note they would like to see if they could mitigate cognitive deficits in the D2R transgenic mice.

To me, the most intriguing points introduced by these two papers are (1) the window of development at which an embryo is susceptible to developing schizophrenia and (2) the role of impaired inhibitory circuitry in schizophrenia pathology. Moore et al. and others found that MAM-E15 induces robust and nonspecific developmental problems, yet that MAM-E17 produces more subtle and schizophrenia-relevant developmental changes. I wish that Kellendonk et al. could have used a model with greater temporal precision, perhaps a Cre-sensitive transgenic system or AAV, to better delineate when during embryonic development the overexpression of D2Rs sets the course for schizophrenia-like symptoms. To be fair, I don’t know if these techniques are possible in utero, which may have restricted the researchers’ options; at least they were able to use ISH to show that the receptors were expressed by E17.5 (mouse). It would be really cool to compare turning the transgenes on and then off during different periods of development to identify when the embryo is susceptible, which could be translatable to human trimesters. While it initially seems disheartening that dox administration at birth to D2R-overexpressing mice couldn’t even prevent symptom onset, this underscores the importance of both identifying early biomarkers and of developing therapies that can effectively counteract what’s already been set in stone developmentally. Additionally, it’s interesting that at least part of these models involve modulating development and gene expression of GABA-ergic cortical neurons, via altered gene expression or altered GABA-ergic neuron development. MAM-E17 neurons were more excitable than their control counterparts with a loss of spike regulation, which could enable erratic firing patterns. This may contribute to learning and cognitive deficits in schizophrenia, since dysregulated neuronal excitability can impede LTP and memory consolidation.