Neurodevelopmental Animal Models of Schizophrenia: Role in Novel Drug Discovery and Development

Abstract

Schizophrenia is a devastating mental illness that is associated with a lifetime of disability. For patients to successfully function in society, the amelioration of disease symptoms is imperative. The recently published results of two large antipsychotic clinical trials (e.g., CATIE, CUtLASS) clearly exemplified the limitations of currently available treatment options for schizophrenia, and further highlighted the critical need for novel drug discovery and development in this field. One of the biggest challenges in schizophrenia-related drug discovery is to find an appropriate animal model of the illness so that novel hypotheses can be tested at the basic science level. A number of pharmacological, genetic, and neurodevelopmental models have been introduced; however, none of these models has been rigorously evaluated for translational relevance or to satisfy requirements of “face,” “construct” and “predictive” validity. Given the apparent polygenic nature of schizophrenia and the limited translational significance of pharmacological models, neurodevelopmental models may offer the best chance of success. The purpose of this review is to provide a general overview of the various neurodevelopmental models of schizophrenia that have been introduced to date, and to summarize their behavioral and neurochemical phenotypes that may be useful from a drug discovery and development standpoint. While it may be that, in the final analysis, no single animal model will satisfy all the requirements necessary for drug discovery purposes, several of the models may be useful for modeling various phenomenological and pathophysiological components of schizophrenia that could be targeted independently with separate molecules or multi-target drugs.

Introduction

Schizophrenia is a severe chronic brain disorder that afflicts approximately 1% of the world’s population. The heterogeneous disorder produces a lifetime of disability, and afflicts all areas of the patient’s life, ranking as one of the leading causes of disability in the United States and other developed countries (1). Symptoms of schizophrenia are commonly divided into three domains: positive (e.g., delusions, hallucinations, agitation); negative (e.g., depression, anhedonia); and, cognitive dysfunction (e.g., poor attention, deficits in executive function, disorders of working and spatial memory). Whereas positive and negative symptoms of schizophrenia tend to be episodic, cognitive deficits often precede the manifestation of psychosis and usually persist throughout the course of the illness (2). Furthermore, cognitive dysfunction is now recognized to be central to the functional disability of the disorder, having the most substantial impact upon the long-term outcome of the illness, yet the focus on developing therapeutic treatments for management of cognitive symptoms has been limited (3, 4).

Current Treatment and Limitations

Treatment options for patients with schizophrenia include typical (first-generation) and atypical (second-generation) antipsychotics. A range of adverse reactions (e.g., extrapyramidal side effects, sedation, anhedonia) of the first-generation antipsychotics led to the development of second-generation antipsychotics with lower D2 receptor affinity and a higher affinity for the 5-HT2A receptor. Results of industry-funded trials suggested that second-generation compounds offered significant advantages over the first-generation drugs, including better efficacy for positive and negative symptoms, enhanced cognitive effects, and improved tolerability (5). However, it is now recognized that these newer atypical agents also have a range of side effects (e.g., weight gain, endocrine disturbances, anticholinergic effects, hypotension, seizures) that can lead to morbidity, impaired quality of life and poor compliance (6, 7).

With rising cost of mental healthcare and lack of evidence of patients with improved outcomes, the NIMH in the U.S. and the NHS Health Technology Assessment R&D Office in the U.K. funded clinical trials to determine clinical superiority of second-generation antipsychotics (5). In terms of effectiveness, results from the U.S. Clinical Antipsychotic Trials of Intervention Effectiveness (CATIE) showed no difference between second-generation antipsychotics (with the exception of olanzapine) and the first-generation antipsychotic perphenazine, the primary outcome being discontinuation of the drug and switching to another antipsychotic (8). Longitudinal assessment of neurocognition and psychosocial functioning indicated no evidence of superiority in the treatment for negative and cognitive symptoms (9). Similarly, the U.K. Cost Utility of the Latest Antipsychotic Drugs in Schizophrenia Study (CUtLASS) showed no advantages of second-generation antipsychotics in terms of symptoms or quality of life, the primary outcome being the total score on the Quality of Life Scale (QLS) and Positive and Negative Syndrome Scale (PANSS) score being a secondary outcome measure (5, 10). These results suggest that no new drugs have achieved superior efficacy for psychosis, nor have they successfully addressed the cognitive and negative symptoms of the disorder (11).

Schizophrenia-Related Animal Models in Drug Discovery and Development

Despite fifty years of drug development research, discovery platforms of schizophrenia have (to date) repeatedly produced compounds with similar mechanisms of action (i.e., primarily dopamine receptor antagonism and, to a secondary extent, serotonin receptor antagonism). This is most likely due to our relatively poor understanding of the etiology and pathophysiology of schizophrenia, as well as the lack of appropriate animal models for screening new compounds. As more knowledge of the pathophysiology of schizophrenia accrues, it is essential that appropriate animal models of the illness be developed that have better translational value. Typically, animal models of human illness are expected to meet the requirements of “face,” “construct” and “predictive” validity (see reviews 12–14).

Face Validity

The degree of phenomenological similarity between the animal model and the human condition it is meant to simulate is known as “face validity.” In the context of schizophrenia, challenges to face validity immediately arise due to the nature of the symptoms of the illness. For examples, some features of schizophrenia are uniquely human (e.g., language disorders) and others (hallucinations, delusions) are impossible to determine in animals. However, several symptoms known to have an important impact on illness outcome in schizophrenia can be modeled successfully in animals. These include deficits of information processing (e.g., prepulse inhibition) and several domains of cognition (e.g., attention, working memory). Other symptoms that can be modeled successfully in animals include hyperactivity, sensitivity to psychostimulants, and deficits in latent inhibition.

Construct Validity

The level of homology in pathological mechanisms (i.e., between the animal model and the human illness) that underlies disease symptoms is known as “construct validity.” Demonstrating construct validity for animal models of schizophrenia is difficult since neither the underlying neurobiological substrates of the behavioral symptoms nor the cognitive deficits have been clearly established. However, several animal models exhibit some pathophysiological features that have been detected in schizophrenia (e.g., neurotransmitter deficits, enlarged ventricles, decreased hippocampal volume, decreased synaptic connections).

Predictive Validity

The most critical (but most difficult to achieve) feature of an animal model for drug discovery purposes is predictive validity. Predictive validity in drug discovery is primarily defined by the degree to which the model can be used to predict efficacy of a new therapeutic agent in humans (reviewed, 13). Typically, in drug discovery research, assessing the effects of positive-control compounds provides this type of validation. In schizophrenia, positive controls for some of the symptoms of schizophrenia (e.g., efficacy in prepulse inhibition [PPI]) impairment models, amphetamine locomotor assays, etc.) have been identified; however, this has not been the case for negative and cognitive symptoms since drugs that reliably improve these symptoms in human patients are not currently available.

As noted above, schizophrenia is a very complex and heterogeneous disease, and, therefore, efforts to model the illness in its entirety are probably unrealistic. There are, however, a number of distinct behavioral, structural and molecular components of schizophrenia that can be modeled in animals (summarized in Table 1; see also reviews 14–18). Thus, focusing on specific symptoms of the disorder or modeling changes in neurobiological substrates that are known to be involved in schizophrenia may represent the most rational approach to establishing face and construct validity in animals models (14, 19).

Table 1.

Desirable Characteristics of a Schizophrenia-Related Animal Model

Behavioral Features

Elevated Locomotor and Stereotypic Activity -Increased sensitivity to dopaminergic psychostimulants -Increased sensitivity to NMDA antagonists

Abnormal Gating -Decreased Prepulse Inhibition (PPI) -Deficits in P20 (analogous to human P50 component of ERP)

Deficits in Habituation

Deficits in Latent Inhibition

Social Behavior -Social withdrawal -Deficits in social recognition

Cognitive Dysfunction -Deficits in attention -Deficits in speed of processing -Deficits in visual learning and memory -Deficits in working memory -Deficits in reason and problem solving

Structural and Molecular Features

Abnormalities in the Neuronal Organization of Prefrontal Cortex -Reduction in cortical volume -Reduced neuropil -Elevated neuronal density -Altered laminar distribution of neurons

Abnormalities in Cortical Neurotransmitter Function -Alterations in GABA neurons -Decreased tyrosine hydroxylase immunoreactive axons -Changes in gene expression of subunits of NMDA receptor

Other Changes in Neuroanatomy -Enlarged third and lateral ventricles -Decreased gray matter in frontal and temporal cortices -Enlarged caudate nucleus

Elevations in locomotor activity and increased sensitivity to dopaminergic psychostimulants are commonly described characteristics of a useful schizophrenia-related animal model. Amphetamine-induced hyperlocomotion has been especially useful in preclinical research for screening new drugs with potential antipsychotic activity (20). The model is thought to mimic the hyperdopaminergic tone and the related stereotypical behaviors often observed in schizophrenic patients (21). Some have challenged the face and construct validity of this approach, however, and argue that a focus on these characteristics was primarily inspired by antipsychotic (i.e., antidopaminergic) pharmacology and not schizophrenia symptoms per se (22). More recently, an elevation in locomotor activity due to glutamate (NMDA) antagonists (e.g., phencyclidine) is recognized as a desirable feature of a schizophrenia-related model and based on N-methyl-D-aspartate (NMDA) alterations that have been observed in postmortem schizophrenia brains. Abnormal sensory gating, deficits in habituation, impaired social behavior, deficits in latent inhibition and cognitive dysfunction are behavioral features that correspond well with schizophrenic symptomatology. Sensory gating deficits (i.e., the inability to properly filter sensory stimuli) are most commonly assessed via measurements of prepulse inhibition (reduction of startle reaction to a sensory stimulus when preceded by weaker stimulus) and the N40 potential (analogous to human P50 event- related potential). Habituation (often assessed in open field locomotor activity assays in rodents) refers to the reduction of response associated with repeated exposure to unimportant stimuli, necessary for the development of selective attention. Latent inhibition (often assessed in operant tasks, drinking behavior- and shock response-related tasks, etc.), in contrast, refers to the retardation in learning about the significance of stimuli due to nonconsequential repeated pre-exposures. Deficits in social interaction and social recognition reflect abnormalities in social cognition and are indicative of the negative sympotomatology of schizophrenia. Attention, speed of processing, visual learning and memory, working memory, and reasoning and problem solving are cognitive domains of schizophrenia that can be modeled via a variety of methods in animals (e.g., 5-choice serial reaction, attentional set-shifting, novel object recognition, water maze tasks).

A number of important structural and molecular abnormalities in the brains of schizophrenic patients has been identified through postmortem studies and, more recently, brain imaging technology. Such features in an animal model help establish construct validity, and include abnormalities in the neuronal organization of the prefrontal cortex (e.g., reduced volume, reduced neuropil, altered laminar distribution of neurons, elevated neuronal density), abnormalities in cortical neurotransmitter function (e.g., alterations in GABA neurons, decreased axons immunoreactive to an essential enzyme for the conversion of dopamine—tyrosine hydroxylase, changes in gene expression of subunits of the glutamate NMDA receptors), as well as enlarged ventricles, decreased gray matter, and enlarged caudate nucleus (16, 17). However, it is important to note that the features listed here are not all inclusive; and, though the behavioral, structural, and molecular abnormalities mentioned are indicative of schizophrenia, they can be observed in patients who do not suffer from schizophrenia.

Neurodevelopmental Hypothesis

The neurodevelopmental hypothesis of schizophrenia posits that the behavioral outcome of the disorder is due to an abnormality in neurodevelopmental processes from early (prenatal or perinatal) insult that begin long before the onset of clinical symptoms (23–25). This hypothesis has been modified more recently to include the role of genetic predisposition and environmental influences (26). Numerous studies have reported developmental delays, social impairments, and general cognitive deficits in preschizophrenic children (27). Studies of brain pathology have revealed abnormalities in neuronal migration, enlargement in cerebroventricular systems, and changes in gray and white matter (18, 28). Several gene candidates (discussed further in the “Genetic Models” section) implicated in schizophrenia are involved in neurodevelopment (e.g., neuronal migration, cell proliferation, axonal outgrowth, and synaptogenesis) and include neuregulin 1 (NRG1), glutamic acid decarboxylase 1 (GAD1), disrupted-in-schizophrenia-1 (DISC1), and dysbindin (DTNBP1) (29–33). The emergence of this considerable amount of evidence has strengthened the link between maldevelopment and schizophrenia.

There has also been a significant amount of research on the role of ecogenetics (i.e., interactions between genetic vulnerability and environmental risk factors) in schizophrenia (34). The incidence of schizophrenia is 2–6% in second degree relatives (e.g., grandparent, aunt, uncle), 6–17% in first degree relatives (e.g., parents, siblings) and nearly 50% in monozygotic twins (35). Although the etiology of schizophrenia is clearly related to genetic factors, the degree of concordance among monozygotic twins and the large percentage of schizophrenic individuals with no affected relative suggest that genetic liability alone is not sufficient for the development of the disorder, and that environ- mental influences also play a critical role. Maternal malnutrition, maternal infection, and labor-delivery complications significantly increase the risk of the offspring developing schizophrenia later in life (36). Studies have also shown that individuals with schizophrenia are more likely to have experienced prenatal adverse events, obstetric complications, or exposure to harmful stressors in comparison to healthy individuals (27). In light of the considerable amount of evidence supporting the neurodevelopment hypothesis, animal models focusing on the role of ecogenetics and mal- development in schizophrenia may be crucial in under- standing the pathophysiology of the disorder, as well as in drug discovery.

Neurodevelopmental Animal Models of Schizophrenia

Disadvantages of using conventional pharmacologic and lesion models of schizophrenia in drug discovery research include a lack of etiological validity, a lack of subtle changes across several neural systems that more accurately reflect the pathophysiology of the disorder, and limitations related to the novel therapeutic agent directly attenuating the effects of the animal manipulation itself (i.e., the drug- or lesion-related effect) instead of what may be disrupted in schizophrenia. Neurodevelopmental animal models, in contrast, have etiological validity, can mimic the delayed onset of symptoms, and produce abnormalities in multiple neural systems associated with schizophrenia; thus, better approximating the pathophysiology of the disorder (3). Several neurodevelopmental animal models have been developed that reflect environmental stressors thought to increase the risk of developing schizophrenia later in life. For the purpose of this review, neurodevelopmental animal models were grouped based on experimental design into prenatal stress models, prenatal/neonatal immune challenge models, perinatal/neonatal stress models, pharmacological/ lesion models, and genetic models.

Prenatal Stress Models

Several studies have shown that maternal stress due to famine, war, and natural disasters during pregnancy increases the risk of a child developing schizophrenia later in life (36–38). Schizophrenia-related rodent models developed to study the effects of prenatal stress include restraint stress, variable stress, maternal protein malnutrition, and vitamin D deficiency (see Table 2).

Table 2.

Behavioral, Structural, and Neurochemical Phenotypes of Prenatal Stress Models

Prenatal Stress Models	Structural and Neurochemical Phenotype	Behavioral Phenotype	References
Restraint Stress	Changes in dopamine sensitivity in nucleus accumbens; increased HPA axis response; inhibition of neurogenesis in hippocampus	Anxiety- and depressive-like behaviors; learning deficits in Morris water maze; locomotor activity and sensitivity to amphetamine	39, 40, 41, 42, 43, 44, 45
Variable Stress	Dysregulation of genes involved in NMDA, GABAergic, and synaptic function in the frontal pole and hippocampus	Increased response to amphetamine; impaired social interaction; disrupted PPI (sensory gating) and N40	46, 47, 48, 49
Maternal Protein Malnutrition	Altered DA and 5-HT response to stress; increased NMDA and DA receptor binding in striatum and hippocampus of female rats	Increased stereotypic response to apomorphine; increased locomotor response to amphetamine; disrupted PPI; (behaviors only in females)	50, 51, 52
Vitamin D Deficiency	Dysregulation of proteins involved in mitochondrial function and synaptic plasticity; enlarged lateral ventricles; reduced cortical thickness	Impaired habituation; altered PPI with additional chronic postnatal vitamin D deficiency; impairment of latent inhibition	53, 54, 55, 56, 57, 58, 59

Restraint Stress

Offspring of pregnant females exposed to restraint stress during the last week of pregnancy show several behavioral changes and alterations in neurochemistry. For example, several studies have indicated that restraint stress increases sensitivity to amphetamine-induced locomotor activity in offspring. This is often interpreted as a sign of a functional imbalance in dopaminergic transmission, which is thought to underlie several symptoms of schizophrenia (39). One study found that restraint stress resulted in alterations of dopamine receptor densities in the meso-limbic dopaminergic system, possibly the cause of the increased amphetamine sensitivity in the animals (40). Prenatal restraint stress also induced impairments in the Morris water maze (a hippocampal-related spatial learning task) and reduced neurogenesis in the dentate gyrus (41). Restraint stress also increased HPA axis response, altered dopamine and norepinephrine levels, and was associated with deficits in the forced swim test and in home cage emergence, results thought to correspond to depressive- and anxiety-like behaviors, respectively (42–45).

Variable Stress

Variable prenatal stress models were developed in response to evidence that repeated application of the same type of stress allows an animal to adapt over time, resulting in reduced activation of the HPA axis. In contrast, unpredictable stress exposure induces a more robust HPA axis response (46). Similar to restraint stress, variable stress produced increased sensitivity to amphetamine-induced locomotor activity and deficits in prepulse inhibition (PPI), considered a measure of sensorimotor gating which is often deficient in schizophrenia patients (46). Animals exposed to variable prenatal stress also showed a reduction in social interaction and dysregulation of genes involved in NMDA, GABAergic, and synaptic function in the frontal pole and hippocampus (47–49).

Maternal Protein Malnutrition

Protein malnutrition throughout pregnancy was also found to produce PPI deficits in rodents; however, unlike variable prenatal stress, the deficits were age and sex dependent occurring only in females (50). Maternal protein malnutrition induced sex dependent increases in NMDA and DA receptor binding in the striatum and hippocampus, increased stereotypic response to apomorphine, and increased locomotor response to amphetamine in female rats (51). Prenatal malnutrition altered extracellular DA and 5-HT levels in response to stress as well (52).

Vitamin D Deficiency

Vitamin D is known to be involved in normal brain development, and deficiency of this vitamin has been proposed as a risk factor for the development of schizophrenia. Studies show dysregulation of proteins involved in mitochondrial function, synaptic plasticity and neurotransmission in rodents exposed to prenatal vitamin D deficiency (53–55). Vitamin D deficiency produced enlarged lateral ventricles and reduced cortical thickness in rodents as well, resembling structural changes observed in schizophrenic patients (56). PPI deficits were only seen with additional chronic postnatal vitamin D deficiency (57). In addition, increased MK-801 sensitive locomotor activity was observed in prenatal vitamin D deficient animals (58). The model also showed impairments of latent inhibition (often observed in patients with schizophrenia), which reflects the ability to ignore stimuli that historically predicted no significant consequence (59).

Prenatal/Neonatal Immune Challenge Models

The original report of Mednick et al. (1988), indicating that persons in Helsinki exposed to the 1957 A2 influenza epidemic during their second trimester of fetal development were at an elevated risk of schizophrenia, has stimulated a considerable amount of interest in the role of maternal infection (37, 60). Rodent models developed to study the effects of prenatal and neonatal infection (i.e., in the context of schizophrenia-related symptoms or pathophysiology) include maternal exposure to the human influenza virus, neonatal exposure to the Borna disease virus, maternal immune activation by bacterial endotoxin lipopolysaccharide (LPS), and maternal and neonatal immune activation by synthetic double-strand RNA polyriboinosinic-polyribocytidilic acid (Poly I:C) (see Table 3).

Table 3.

Behavioral, Structural, and Neurochemical Phenotypes of Prenatal and Neonatal Immune Challenge Models

Prenatal/Neonatal Immune Challenge Models	Structural and Neurochemical Phenotype	Behavioral Phenotype	References
Human Influenza Virus	Dysregulation of genes involved in cell structure and function; significant brain atrophy	Impaired PPI; deficits in exploratory behavior in both open field and novel object test; impaired social interaction	61, 62, 63, 64
Borna Disease Virus	Impaired hippocampal synaptogenesis; increased cortical and cerebellum levels of NE and 5-HT; loss of Purkinje cells in cerebellum	Impaired social behaviors; increased locomotor activity; impaired habituation and PPI (rat strain dependent)	65, 66, 67, 68, 69
LPS	Age-specific changes in accumbal DA levels and striatal DOPAC; dendritic changes in mPFC and hippocampus	Disrupted PPI; increased amphetamine-induced locomotor activity; deficits in social interaction	70, 71, 72, 73, 74, 75, 76
Poly I:C	DA-related pharmacological abnormalities; increase in GABAA receptor subunit α2 in ventral dentate gyrus and basolateral amygdala	Learning deficits in Morris water maze; postpubertal latent inhibition and PPI deficits; impaired object recognition; impaired social behavior	77, 78, 79, 80, 81, 82

Human Influenza Virus

Maternal human influenza infection produced offspring with behavioral deficits in social interaction and exploratory behavior in an open field, as well as deficits in PPI that were reversible by clozapine and chlopromazine (61). Dysregulation of genes involved in cell structure and function suggests that prenatal influenza infection may cause permanent changes in brain morphology and function; indeed, significant brain atrophy has been observed in rodents using this model approach (62, 63). Studies also show a decrease in reelin-positive cell counts in the cortex and hippocampus, thought to play a key role in neuronal migration of the developing brain (64).

Borna Disease Virus

Neonatal Borna disease virus infection caused severe alteration in the cerebellum, including reduced size, substantial loss of Purkinji cells, and increased extracellular levels of norepinephrine and serotonin (65, 66). Neonatal exposure to the Borna disease virus produced alterations in the hippocampus of offspring as well, with impaired synaptogenesis and defects in synaptic organization (67). Behavioral changes in social interaction, locomotor activity, habituation and PPI were observed; however, these changes were found to be dependent on the strain of the rat, possibly demonstrating that the same environmental insult can produce differential neuroanatomical and behavioral abnormalities dependent on genetic vulnerability (68, 69).

Lipopolysaccharide

Prenatal immune activation by bacterial endotoxin lipopolysaccharide (LPS) produces several of the behavioral deficits previously described that are thought to be relevant to schizophrenia. Studies have shown increased amphetamine-induced locomotor activity, deficits in social interaction, increased anxiety and haloperidol-sensitive, age-dependent disruptions in PPI (70–73). Prenatal LPS also produced age-specific changes in accumbal dopamine levels (74, 75). Observations of dendritic changes in the medial prefrontal cortex and hippocampus suggest that LPS prenatal immune activation affects development of pyramidal neurons that may potentially lead to abnormal neuronal connectivity and function (76).

Poly I:C

Immune activation (maternal and neonatal) with the synthetic double-strand RNA polyriboinosinic-polyribocytidilic acid (Poly I:C) has been suggested to be more appropriate in mimicking viral infection than LPS because it is thought that double-strand RNA is important in interferon induction (77). Studies of prenatal and postnatal Poly I:C immune activation produced similar behavioral results as LPS immune activation in rodents, including increased amphetamine-induced locomotor activity, deficits in social interaction, increased anxiety and disrupted PPI (78, 79). Other behavioral changes observed in maternal exposure to Poly I:C included postpubertal emergence of latent inhibition and impaired spatial working memory, while neonatal Poly I:C immune activation impaired object recognition (79–81). Deficits in latent inhibition and novel object recognition were improved by clozapine, but not affected by haloperidol (77). Maternal immune activation during pregnancy by Poly I:C also produced dopamine-related pharmacological abnormalities and increased GABAA receptor subunit α2 in the ventral dentate gyrus and basolateral amygdala (78, 82).

Perinatal/Neonatal Stress Models

Various forms of psychopathology have frequently been associated with obstetric complications or exposure to a traumatic experience during childhood (83, 84). Rodent models developed to study the effects of perinatal and neonatal stress include 24-hour maternal deprivation, isolation rearing, and birth insults (see Table 4).

Table 4.

Behavioral, Structural, and Neurochemical Phenotypes of Perinatal and Neonatal Stress Models

Perinatal/Neonatal Stress Models	Structural and Neurochemical Phenotype	Behavioral Phenotype	References
24-Hour Maternal Deprivation	Cerebellum neuronal degeneration; increased astrocytes in cerebellum and hippocampus (sex dependent)	Impaired PPI; increased susceptibility to apomorphine (rat strain dependent)	85, 86, 87, 89
Isolation Rearing	Decreased cortical and hippocampal synaptic plasticity; increased striatal dopamine receptors in the functional high state	Increased locomotor activity; impaired set-shifting and object recognition; increased social interaction; deficits in PPI (rat strain dependent)	90, 91, 92, 93, 94, 95, 96, 97, 98, 99, 100,101, 102
Birth Insults	Decreased dentate granule cells; increased DA release in nucleus accumbens and striatum; decreased DA release in PFC	Increased amphetamine-induce locomotor activity (age dependent)	103, 104, 105, 106, 107

24-Hour Maternal Deprivation

A single 24-hour episode of maternal deprivation produced postpubertal deficits in PPI that were reversible by haloperidol and risperidone (85). However, it was found that the deficits in PPI, as well as sensitivity to apomorphine, were strain dependent, similar to what was seen in prenatal exposure to the Borna disease virus (86). Maternal deprivation was also found to be associated with neuronal degeneration and an increased number of astrocytes in the cerebellum, which interestingly was attenuated by inhibitors of endocannabinoid inactivation, suggesting a possible role of the endocannabinoid system in neural development (87, 88). Increased numbers of astrocytes in the hippocampus were also observed in animals exposed to maternal deprivation; however, these results were sex dependent (89).

Isolation Rearing

A large amount of research has been dedicated to the effects of early social isolation in rodents. Behavioral deficits observed in animals exposed to isolation rearing include strain-dependent PPI, increased aggression, increased anxiety, increased locomotor activity, increased food hoarding behavior, and impaired object recognition (90–96). Treatment with methylphenidate was found to improve deficits in anxiety, and treatment with clozapine and fluoxetine was found to improve aggressive behavior in these animals (97). Isolation rearing is also associated with impaired performance in an attentional set-shifting task, a deficit seen in patients with chronic schizophrenia and thought to reflect impaired reasoning and problem solving (98). Exposure to social isolation in rodents produces several neurochemical and neuroanatomical alterations as well, including reduced arrays of chandelier axons in the prelimbic cortex, reduction in the volume of the medial prefrontal cortex, decreased cortical and hippocampus synaptic plasticity, and an increase in the number of striatal dopamine receptors in the functional high affinity state (95, 99–102).

Birth Insults

To determine the possible effects of obstetric complications, a rodent model of anoxia during Caesarean section birth was developed. Exposure to anoxia during birth increased age-dependent amphetamine-induced locomotor activity (103, 104). Altered dendritic morphology in the prefrontal cortex and hippocampus, changes in dopamine turnover in the prefrontal cortex and nucleus accumbens, and a decrease in hippocampal dentate granule cells were observed in animals exposed to anoxia as well (104–107).

Pharmacologic/Lesion Models

Pharmacologic and lesion models have played an important role in biomedical research. An advantage of these models is the ability to provide information about changes in transmitter systems and neural circuits from a specific alteration (3). Rodent models developed to study the effects of specific pharmacological and lesion manipulations on neurodevelopment (and how the effects may relate to schizophrenia) include neonatal PCP exposure, neonatal ventral hippocampus lesion, neonatal exposure to NOS inhibitors, and exposure to antimitotic agents (see Table 5).

Table 5.

Behavioral, Structural, and Neurochemical Phenotypes of Neonatal Pharmacological and Lesion Models

Pharmacologic/Lesion Models	Structural and Neurochemical Phenotype	Behavioral Phenotype	References
Neonatal PCP	Apoptosis of GABAergic interneurons in primary somatosensory, motor, and retrosplenial cortices	Deficits in attention set-shifting	108, 109, 110, 111
Neonatal Ventral Hippocampal Lesion	Decreased dedritic density of pyramidal neurons; alterations in GABAA receptor expression; improper interneuron maturation	Increased amphetamine-induced locomotor activity; PPI deficits; decreased social interaction; hyperresponsive to stress	112, 113, 114, 115, 116, 117, 118, 119, 120
Neonatal NOS Inhibition	Possible changes in NMDA function	Hypersensitivity to amphetamine; deficits in social interaction; PPI deficits (all sex dependent)	121, 122, 123, 124
Antimitotic Agent: MAM or AraC	Decreased tissue weight in hippocampus and PFC; disorganization of pyramidal cell layers in hippocampus	PPI deficits (postpubertal); deficits in radial arm maze; impaired social interaction; increased amphetamine-induced locomotor	125, 126, 127, 128

Neonatal PCP

Neonatal administration of phencyclidine (PCP), an NMDA antagonist, has been shown to cause schizophrenia-like symptoms in humans and to impair attentional set-shifting in adult rats (108). The antipsychotic, sertindole, and ampakine CX516 (an AMPA receptor modulator) were able to attenuate these deficits in attentional set-shifting (109). Neonatal PCP administration was also associated with apoptosis of GABAergic interneurons in the primary somatosensory, motor, and retrosplenial cortices as well as neurodegeneration in the striatum and hippocampus, an effect that could be attenuated by treatment with lithium (110, 111).

Neonatal Ventral Hippocampal Lesion

Perhaps the most thoroughly characterized neuro-developmental model of schizophrenia to date is the neonatal ventral hippocampal lesion (NVHL) model first introduced by Lipska and Weinberger (112). Lesions of the ventral hippocampus in neonatal rats (typically produced by ibotenic acid) produce a complex syndrome that is postadolescent in onset and characterized by frontal cortical-striatal dysfunction. The syndrome in rats is expressed as a variety of symptoms that are observed in schizophrenia, including positive (e.g., deficits in PPI and latent inhibition), negative (e.g., deficits in social behaviors), and cognitive components (e.g., deficits in spatial learning and memory). Other behavioral changes include addiction vulnerability, increased amphetamine-induced locomotor activity, and hyperresponsiveness to stress (reviewed in 113; see also, 114–118). It is interesting to note that the disruption of PPI in this animal model could be attenuated by atypical antipsychotics as well as ORG 24598, a selective glycine transporter 1 inhibitor (118). In addition, several important (schizophrenia-relevant) anatomical and neurophysiological alterations have been observed in the prefrontal cortex of NVHL rats, including decreased dendritic length and spine density of pyramidal neurons, alterations of plasticity, and improper maturation of interneurons during adolescence (118–120). Thus, while on first glance the construct validity of the NVHL model might appear to be questionable, after the animals reach adolescence several of the requirements of face, construct, as well as predictive, validity appear to be met.

Neonatal NOS Inhibition

It has been suggested that nitric oxide affects neurodevelopmental processes in the central nervous system. Neonatal exposure to a nitric oxidase inhibitor resulted in sex-dependent behavioral deficits that included hypersensitivity to amphetamines, deficits in PPI, and social interaction (121, 122). Behavioral deficits in latent inhibition that were reversed by glycine 1 inhibitors (GlyT1) and α-7 nicotinic receptor agonists were also observed in animals exposed to prenatal NOS inhibition, suggesting alterations in glutamatergic and cholinergic systems (123, 124).

Antimitotic Agent

Some studies have suggested that low doses of antimitotic agents given prenatally will perturb the developmental progression of neurogenesis (a process thought to be altered in schizophrenia). Prenatal exposure to the antimitotic drug, cytosine arabinoside (Ara-C), produced postpubertal deficits in PPI and disorganization in hippocampal pyramidal cell layers (125). Prenatal exposure to the mitotic inhibitor, methylazoxymethanol (MAM), produced increased amphetamine-induced locomotor activity, deficits in social interaction, impairments in set-shifting tasks, and deficits in radial arm maze performance (126–128). Prenatal MAM exposure also resulted in decreased tissue weight in the hippocampus and prefrontal cortex (127).

Genetic Models

Considerable evidence supports an important role of genetics in schizophrenia and, accordingly, a vast number of risk genes have been studied in mutant mouse models (for review, 29, 129, 130). While the most promising gene candidates for schizophrenia have failed to be replicated in multiple populations, several transgenic mouse models have been useful for studying the behavioral consequences of specific gene/molecular alterations and the mechanisms potentially underlying the pathogenesis of schizophrenia (129). For the purposes of this review, mouse models specifically designed to study the schizophrenia-related gene candidates that are also known to play an important role in neurodevelopment will be discussed. These include neuregulin 1 (NRG1), disrupted-in-schizophrenia-1 (DISC1), and dysbindin (DTNBP1) (see Table 6).

Table 6.

Behavioral, Structural, and Neurochemical Phenotypes of Genetic Models

Genetic Models	Structural and Neurochemical Phenotype	Behavioral Phenotype	References
NRG1 (phenotypes dependent on domain mutations)	Decreased NMDA receptor function; enlarged lateral ventricles; decreased dendritic spine density	Increased locomotor activity; PPI deficits; impaired social interaction; deficits in latent inhibition; impaired delayed alteration memory task performance	131, 134, 135, 136, 137, 138 139, 140
DISC1	Decreased mPFC volume; decreased hippocampal dendritic complexity; enlarged lateral ventricles; reduced parvalbumin GABAergic neurons in the hippocampus and mPFC	Impaired working memory; disrupted PPI; altered latent inhibition; hyperactivity; depressive-like symptoms; abnormal spatial working memory	144, 145, 146, 147, 148, 149
DTNBP1	Decreased DA levels; decreased steady-state snapin; deficiencies in neurosecretion and vesicular morphology	Increased anxiety; decreased activity; impaired social interaction; working memory deficits; recognition memory deficits; emotional learning and memory deficits	159, 160, 161, 162, 163, 164, 165

Neuregulin 1 (NRG1)

Neuregulin 1 (NRG1), part of a family of growth and differentiation factors, was first suggested as a potential candidate gene for schizophrenia in a study of the Icelandic population (131). This association of NRG1 with schizophrenia was later confirmed in Scottish (132) and Irish (133) subjects. NRG1 has been found to be essential for neurodevelopment, with key roles in synapse formation, neuronal migration, synaptic plasticity, and regulation of neurotransmitter systems (134). Homozygous null mice embryos die midgestation; however, heterozygous mutant mice are viable and can reproduce (135). NRG1 hypomorph epidermal growth factor (EFG)-like domain models produce hyperactive mice with impaired PPI only after treatment with MK-801 (135, 136). Most NRG1 proteins are synthesized with a transmembrane (TM) domain, and NRG1 hypomorph TM models also produce hyperactivity that can be reduced by the second-generation antipsychotic, clozapine. These animals also exhibit impaired PPI (that is not reversed by clozapine), altered habituation, increased aggression, and decreased functional NMDA receptors (131, 137, 138). NRG1 immunoglobulin (Ig)-like domain mutant mice, however, do not appear to be hyperactive, although they do exhibit impairments of latent inhibition (139). A more recent model focusing on the deletion of a specific NRG1 isoform (Type III) produced mice with more pronounced PPI deficits and impaired performance on delayed alteration memory tasks, as well as enlarged lateral ventricles and decreased dendritic spine density (140).

Disrupted-in-Schizophrenia-1 (DISC1)

Disrupted-in-schizophrenia-1 (DISC1) was suggested as a possible candidate gene when it was found to be disrupted by a balanced (1;11)(q42.1; q14.3) translocation that cosegregates with schizophrenia and related psychopathologies (141, 142). DISC1 has been suggested to play important roles in neurite outgrowth, cell migration and cell signaling (143). Studies of a DISC1 mutant model of a deletion variant in mDisc1 specific to the 129S6/SvEV strain transferred onto a C57BL/6J strain revealed impairments in working memory, decreased mPFC volume, altered synaptic transmission in the hippocampus, and reduced dendritic growth in the dendate gyrus (144, 145). An inducible DISC1 C-terminal fragment (DISC1-cc) transgenic model exhibited abnormal spatial working memory and deficits in social interaction, as well as decreased hippocampal dendritic complexity (146). A model expressing the dominant negative C-terminal truncated DISC1 (DN-DISC1) exhibited enlarged lateral ventricles, hyperactivity, disrupted PPI, and depressive-like symptoms (147). Inducible expression of mutant hDISC1 also produced mice with enlarged lateral ventricles, deficits in spatial working memory, impaired social interaction, and hyperactivity (148). DISC1tr (truncated) transgenic mice are characterized by enlarged lateral ventricles, decreased cortical neurogenesis (and decreases in cerebral cortex), as well as partial agenesis of the corpus callosum. In addition, parvalbumin GABAergic neurons are reduced in the hippocampus and medial prefrontal cortex, and displaced in the dorsolateral frontal cortex. In behavioral studies, DISC1tr mice exhibit increased immobility and reduced vocalization in depression-related tests, as well as impairment in conditioning of latent inhibition (149).

It should be noted, however, that there are some ambiguities in the results of the animal and cell-based studies that need to be addressed before final conclusions can be drawn regarding the role of DISC1 in neurodevelopmental processes that are relevant to schizophrenia. For example, the cellular models based on the knockdown of DISC1 or expression of dominant negative DISC1 indicate clear disruptions in the developmental processes that are critical for normal cortical architecture. However, the aforementioned mutant mice (i.e., with the 29S6/SvEV-derived DISC1 gene transferred onto a C57BL/6J genetic background) showed relatively subtle behavioral abnormalities and no major deficits in cortical architecture. Such a finding leads to confusion, especially regarding a so-called “critical role” of DISC1 in neurodevelopment given that these mutant mice have a 25-bp deletion in a coding exon of the DISC1 gene that would be expected to completely abolish the full-length DISC1 protein. Efforts are currently underway to resolve these questions (see 150).

Dysbindin (DTNBP1)

Several studies have described dysbindin (DTNBP1) as a candidate gene in schizophrenia (151–154). Dysbindin binds to dystrobrevins, components of the dystrophin- associated glycol-protein complex (DGC), and is thought to play a fundamental role in the regulation of synaptic structure and signaling (155, 156). In 1991, Swank and colleagues first described a mouse mutant (sandy) which arose from the DBA/2J strain and was later found to have a deletion of the DTNBP1 gene (157, 158). Behavioral studies on sandy (sdy) mutant mice showed decreased activity, increased anxiety and impaired social interaction, as well as deficits in working memory and recognition memory (159, 160). These mice also displayed an increased freezing response to a conditioned stimulus in a fear-conditioning paradigm suggestive of deficits in emotional and motivated learning and memory (161). Decreased dopamine levels, reduced steady state levels of snapin (a synaptic priming regulator), and deficiencies in neurosecretion were also observed in the sdy mutant mice (162–164). In contrast, a recent study on sdy mutant mice (DTNBP1 KO) from the C57BL/6J strain found no evidence of increased anxiety or increased activity, although the mice were impaired in spatial learning and memory (165).

Conclusions

Novel therapeutic strategies for schizophrenia are critically important in light of the inadequate treatment options currently available. Accordingly, the development of valid and reliable animal models of the illness is essential so that novel treatment-related hypotheses can be tested at the basic science level. While a number of schizophrenia-related animal models have been introduced (see references 166 and 167 for a comprehensive list), few have been rigorously examined for their translational value in the context of novel drug discovery. However, several neurodevelopmental models have met some of the requirements of face validity as evidenced by deficits/alterations in social interaction, PPI, amphetamine sensitivity, latent inhibition, and spatial learning (i.e., features commonly observed in patients with schizophrenia [16]). Some of these behavioral deficits (e.g., PPI, latent inhibition) were reversible by first- and second-generation antipsychotics, thus providing evidence of predictive validity. Furthermore, neuroanatomical alterations (e.g., cortical atrophy, enlarged ventricles, neuronal disorganization) observed in some of the neurodevelopmental models correlate with what is often observed in patients with schizophrenia, thus satisfying some requirements of construct validity (28). Another striking feature of the neurodevelopmental models is the ability to produce age-dependent behavioral and neurochemical deficits as well as strain-specific alterations (i.e., factors that support an important role of ecogenetics in schizophrenia). The age-dependent emergence of schizophrenia-related symptoms in some of the animal models could be very useful for testing hypotheses related to the contemporary (and controversial) topic of therapeutic intervention during the prodromal stages of schizophrenia. There is also an emerging interest in the development of animal models that mimic gene-environment interactions in schizophrenia (i.e., combine mutant mouse transgenics with measurable environmental insults). Such an approach (see 168 for review) may lead to more optimal models for identifying important gene-environment interactions in schizophrenia, as well as for testing novel drug-development strategies.

It is important to note that another potential reason for the disappointing progress in the development of novel therapeutic agents for schizophrenia to date may have been the focus on designing a single compound for treating an illness with multiple dimensions. It has been suggested that a more rational approach would be to deconstruct schizophrenia into its various phenomenological components that could be targeted independently with separate molecules or specifically designed multi-target drugs (169). While it may be that no single animal model would satisfy all the requirements necessary for drug discovery purposes using this approach, several of the neurodevelopmental models discussed in this review could theoretically be used concurrently for such an approach.