|Home | About | Journals | Submit | Contact Us | Français|
The molecular mechanisms of major mental illnesses, such as schizophrenia and bipolar disorder, are unclear. To address this fundamental question, many groups have studied molecular expression profiles in postmortem brains and other tissues from patients compared with those from normal controls. Development of unbiased high-throughput approaches, such as microarray, RNA-seq, and proteomics, have supported and facilitated this endeavor. In addition to genes directly involved in neuron/glia signaling, especially those encoding for synaptic proteins, genes for metabolic cascades are differentially expressed in the brains of patients with schizophrenia and bipolar disorder, compared with those from normal controls in DNA microarray studies. Here we propose the importance and usefulness of genetic mouse models in which such differentially expressed molecules are modulated. These animal models allow us to dissect the mechanisms of how such molecular changes in patient brains may play a role in neuronal circuitries and overall behavioral phenotypes. We also point out that models in which the metabolic genes are modified are obviously untested from mental illness viewpoints, suggesting the potential to re-address these models with behavioral assays and neurochemical assessments.
Schizophrenia (SCZ) and bipolar disorder (BPD) are major mental illnesses with onset in adulthood. Although the clinical manifestation and disease course are substantially different from each other, these two disorders share genetic risks at several loci and have some similarities in the patterns of cognitive impairment (Berrettini, 2004; Craddock et al., 2006; Hill et al., 2008).
To address molecular mechanisms underlying mental illnesses, many groups have studied molecular expression profiles in postmortem brains and other tissues from patients compared with those from normal controls. Since 2000, in addition to candidate molecule approaches, an unbiased methodology for expression study with microarrays has been frequently employed (Horvath et al., 2011; Mirnics et al., 2006; Mirnics et al., 2000). More recently, RNA-seq has been introduced as a revolutionary tool for transcriptomics (Twine et al., 2011; Wang et al., 2009). At the protein level, several types of proteomic approaches are also utilized (English et al., 2011; Wittmann-Liebold et al., 2006). A major concern in these molecular profile studies is that this approach may not be able to address brain circuitry-dependent mechanisms.
To complement this limitation, rodents (rats and mice) are used to model mental diseases (Chen et al., 2006; Cryan and Holmes, 2005; Nestler and Hyman, 2010). These models have been characterized primarily by behavioral changes: for example, hyperlocomotion, social interaction deficits, impaired prepulse inhibition, and working memory deficits in rodents are used as indicators of behavioral deficits possibly corresponding to positive symptoms, negative symptoms, disturbance in information processing, and cognitive deficits in SCZ patients, respectively (Arguello and Gogos, 2006). In addition, mild but significant abnormalities in anatomy and histology are used as translatable characteristics between mice and humans, including enlarged ventricles, dendritic changes, and interneuron deficits (Jaaro-Peled et al., 2010). Nonetheless, due to the uncertain etiology of mental illnesses, the construct and etiological validity of these models is always debated. The models that modulate genes whose expression is altered in patient tissues may reflect disease pathophysiologies, although there is no guarantee that the model really mimics disease etiologies. Such animal models may complement the difficulty to study brain circuitry-dependent mechanisms at functional levels using expression studies of patient cells and tissues. For example, transient overexpression of dopamine D2 receptor specifically in the striatum can cause persistent functional abnormalities in the prefrontal cortex, affecting some endophenotypes relevant to schizophrenia (Kellendonk et al., 2006), while transient knockdown of DISC1 specifically in the cells of pyramidal neuron lineage in the developing cortex affects postnatal mesocortical dopaminergic maturation and causes functional deficits in mesolimbic dopaminergic neurons, leading to neurochemical and behavioral deficits relevant to SCZ in adulthood (Niwa et al., 2010). These two papers may support the idea of how gene manipulations in mice can provide us with clues to our understanding of circuitry-based mechanisms of mental illnesses. Specific targets found in human gene expression studies can be a source for genetic mouse models that could further enhance such insight into the mechanisms underlying mental illnesses.
In this review, we first summarize DNA microarray studies in postmortem brains from patients with SCZ and BPD. From the many genes in the literature we discuss those that have been found to change in more than one study. In the second part, we review mouse models built by modulating genes that have been identified in human brain gene expression studies. Finally, we discuss a promising research strategy for better understanding major mental illnesses by combining human molecular profiling and studies with mouse models.
Most DNA microarray analysis studies have used postmortem brains of patients with SCZ, BPD, and related mental illnesses. However, there are three major limitations in studies with postmortem brains, which may possibly generate inconsistency among the results:
Nonetheless, we believe that the molecular information from patient brains is still very important for dissecting SCZ and BPD, as they are “brain” disorders. Furthermore, several efforts have been made in order to overcome these limitations (Lewis, 2002; McCullumsmith and Meador-Woodruff, 2010). Thus, here we searched published literature for DNA microarray studies of postmortem brains from patients with SCZ and/or BPD and summarized the genes that were differentially expressed in more than one study when compared to controls. These genes were then categorized according to their functional properties using DAVID Functional Annotation Bioinformatics Microarray Analysis (Huang da et al., 2009). The genes were clustered into four major groups as follows: synaptic genes (Table 1), genes associated specifically with neuronal and/or glial functions (Table 2), metabolic genes (Table 3), and genes with general functions (signaling, protein degradation, cytoskeletal) (Table 4).
Although multiple brain regions have been examined in gene expression studies of SCZ and BPD, the majority of research has focused on the prefrontal cortex for the microarray studies. Therefore, many genes listed in the tables are data from the prefrontal cortex. Other brain regions analyzed include the anterior cingulate cortex (McCullumsmith et al., 2007), the hippocampus (Altar et al., 2005; Konradi et al., 2004; Mexal et al., 2005), and the temporal cortex (Aston et al., 2004; Hemby et al., 2002). There is one study focusing on the altered thalamic gene expression in both SCZ and BPD (Chu et al., 2009). Of note, although we focus on DNA microarray studies in this review article, altered levels of BDNF and TrkB in SCZ brains have been reported via in situ hybridization and Western blotting (Hashimoto et al., 2005; Weickert et al., 2003).
The number of molecules differentially expressed is the largest for the category of “synaptic genes” (in particular, genes involving GABAergic and glutamatergic synapses) in postmortem SCZ and BPD brains, which is consistent with a proposal that SCZ is a synaptic disorder (Mirnics et al., 2001a).
Most of the dysregulated genes in this synapse group are from studies of the prefrontal cortex areas. Among them, the expression of V-erb-b2 erythroblastic leukemia viral oncogene homolog 3 (avian) (ERBB3) is reduced in postmortem tissues from the prefrontal cortex (Hakak et al., 2001; Tkachev et al., 2003) and the temporal cortex (Aston et al., 2004) in SCZ, as well as the prefrontal cortex (Tkachev et al., 2003) in BPD. Also, the expression of Synapsin II (SYN2) has been shown to be decreased based on two different microarray studies in the prefrontal cortex in SCZ (Arion et al., 2007; Mirnics et al., 2000). Moreover, the expression of Monoamine oxidase A (MAOA) is increased in the prefrontal cortex in both SCZ and BPD (Shao and Vawter, 2008). Those genes whose expression change has not been reported in the prefrontal cortex but other brain regions are Neurexin 3 (NRXN3) (Chu et al., 2009; Mexal et al., 2005), Tyrosine 3-onooxygenase/tryptophan 5-monooxygenase activation protein, zeta polypeptide (YWHAZ) (Chu et al., 2009; Konradi et al., 2004), and Tyrosine 3-monooxygenase/tryptophan 5-monooxygenase activation protein, eta polypeptide (YWHAH) (Chu et al., 2009; Vawter et al., 2001). Expression of these three genes is reduced in the temporal cortex, the hippocampus, or the thalamus in SCZ or both SCZ and BPD. Very interestingly, they are also genetically associated with SCZ or BPD (Grover et al., 2009; Jia et al., 2004; Novak et al., 2009; Toyooka et al., 1999).
Abnormalities in glutamatergic neurotransmission and associated signaling (Tamminga, 1998), as well as GABAergic deficits (Benes, 2000) have been consistently reported in SCZ. Several genes listed in Table 1 are glutamate- or GABA-related. Among them, the expression of Gamma-aminobutyric acid A receptor, alpha 5 (GABRA5) and Glutamate receptor, metabotropic 3 (GRM3 i.e. mGluR3) is increased in the prefrontal cortex and anterior cingulate cortex in BPD (Choudary et al., 2005), whereas their expression is downregulated in the prefrontal cortex (Duncan et al., 2010) and the entorhinal cortex in SCZ (Hemby et al., 2002). Several groups have consistently reported that the expression of Glutamate decarboxylase 1 (GAD1) is downregulated in both SCZ and BPD (Duncan et al., 2010; Hashimoto et al., 2008; Konradi et al., 2004; Vawter et al., 2002). Genetic associations of these glutamatergic and GABAergic molecules with SCZ and BPD have also been reported (Cherlyn et al., 2010).
In parallel to the expression data and genetic information described above, synaptic disturbances have been reported in many other studies with human subjects (e.g., decrease in synaptic density in SCZ brains and disturbed cortical synchrony as a sign of GABAergic interneuron deficits) (Lewis et al., 2003; Uhlhaas and Singer, 2010). Thus, the molecules that are altered in expression in the patient brains can be key probes in studying the mechanistic roles of synaptic disturbances in the pathophysiologies of SCZ and BPD.
This category includes genes that are involved not directly in synaptic roles but in functions otherwise associated with neuronal and glial cells. In contrast to the dominance of expression changes in the synaptic genes (Category 1 above) in the prefrontal cortex, the expression changes of this second category are from various brain regions, including the prefrontal cortex, the temporal cortex, and the hippocampus. Although we do not know whether these differences imply the disease pathophysiology or are a mere bias of the brain regions studied, it might be worth mentioning.
In this category, many myelin-related genes are downregulated in SCZ and BPD (Dracheva et al., 2006; Hakak et al., 2001; McCullumsmith et al., 2007; Tkachev et al., 2003). Considering a recent excitement on the detection and characterization of white matter pathology in SCZ and BPD (Connor et al., 2009; Regenold et al., 2007), these expression data may be very important. In SCZ, their downregulation is reported in multiple brain areas (Bowden et al., 2008; Dracheva et al., 2006; Hakak et al., 2001; McCullumsmith et al., 2007; Tkachev et al., 2003). Among them, alteration in the expression of Myelin associated glycoprotein (MAG) is most frequently reported in at least three distinct brain areas in SCZ (Aston et al., 2004; Dracheva et al., 2006; Hakak et al., 2001; McCullumsmith et al., 2007) and the prefrontal cortex in BPD (Tkachev et al., 2003). The expression of 2',3'-cyclic nucleotide 3' phosphodiesterase (CNP) is decreased in the prefrontal cortex (Hakak et al., 2001), the anterior cingulate cortex (Dracheva et al., 2006; McCullumsmith et al., 2007), and the hippocampus (Dracheva et al., 2006) in SCZ. The expression of SRY (Sex determining region Y)-box 10 (SOX10) is reduced in multiple brain regions of SCZ (Dracheva et al., 2006; Iwamoto et al., 2005b) and in the prefrontal cortex of BPD (Tkachev et al., 2003). The expression of Claudin 11 (CLDN11) and Transferrin (TF) is decreased in the prefrontal cortex in both diseases (Arion et al., 2007; Hakak et al., 2001; Tkachev et al., 2003) and in the anterior cingulate cortex in SCZ (Dracheva et al., 2006; McCullumsmith et al., 2007).
In addition to myelin-related genes, the expression of some neuropeptides is also downregulated in SCZ and BPD. The expression of Somatostatin (SST) is reduced in the prefrontal (Hashimoto et al., 2008) and temporal cortices in SCZ (Bowden et al., 2008) as well as in the hippocampus in BPD (Konradi et al., 2004). Neuropeptide Y (NPY) expression is decreased in the prefrontal cortex in both SCZ and BPD (Hashimoto et al., 2008; Kuromitsu et al., 2001).
In addition to genes associated with myelination and neuropeptides, the following molecules may be of scientific interest in this category: expression of Neural precursor cell expressed, developmentally downregulated 8 (NEDD8) is decreased in the temporal cortex (Bowden et al., 2008) and in the hippocampus (Altar et al., 2005) in SCZ. Ubiquitin carboxyl-terminal esterase L1 (Ubiquitin thiolesterase) (UCHL1) is downregulated in several brain regions of SCZ as well (Altar et al., 2005; Vawter et al., 2001). These observations may give us a hint to consider roles for misfolded proteins and endoplasmic reticulum (ER) stress in major mental illnesses (Hayashi et al., 2009; Ottis et al., 2011). The expression of Sema domain, seven thrombospondin repeats, transmembrane domain and short cytoplasmic domain, 5A (SEMA5A) is increased in the temporal cortex (Bowden et al., 2008) while reduced in the hippocampus (Altar et al., 2005) in SCZ.
Recent studies have suggested intrinsic abnormalities in metabolic cascades in SCZ (Holmes et al., 2006; Khaitovich et al., 2008; Kirkpatrick et al., 2009; Prabakaran et al., 2004). Metabolic disturbances are observed in drug-naïve SCZ patients (Thakore, 2004): for example, metabolic profiling of cerebrospinal fluid of drug-naïve patients with first-onset SCZ shows alterations in glucoregulatory processes (Holmes et al., 2006) and individuals with non-affective psychosis display increased prevalence of abnormal glucose tolerance prior to antipsychotic treatment (Fernandez-Egea et al., 2009). Thus, it is now very important to address molecular mechanisms underlying such changes. Interestingly, through this categorization of the past publications of gene expression in SCZ and BPD brains, we came across many genes differentially expressed in these disorders, forming a category of “metabolic genes.”
Many genes in this category are downregulated in SCZ and BPD. Among them, the expression of NADH dehydrogenase (ubiquinone) Fe-S protein 4, 18kDa (NADH-coenzyme Q reductase) (NDUFS4) and Phosphoglycerate mutase 1 (brain) (PGAM1) is reduced in the hippocampus of SCZ (Altar et al., 2005). The expression of ATP synthase, H+ transporting, mitochondrial Fo complex, subunit C3 (subunit 9) (ATP5G3), ATP synthase, H+ transporting, mitochondrial F1 complex, gamma polypeptide 1 (ATP5C1), ATPase, H+ transporting, lysosomal 70kDa, V1 subunit A (ATP6V1A), and Glutamic-oxaloacetic transaminase 1, soluble (aspartate aminotransferase 1) (GOT1) is decreased in the hippocampus of BPD (Konradi et al., 2004). These genes are also downregulated in other brain regions, such as the prefrontal cortex, the superior temporal gyrus, and the thalamus (Arion et al., 2007; Bowden et al., 2008; Chu et al., 2009; Sun et al., 2006). In contrast, the expression of Chitinase 3-like 1 (cartilage glycoprotein-39) (CHI3L1) in SCZ is increased both in the hippocampus (Chung et al., 2003) and the prefrontal cortex (Arion et al., 2007).
The expression of Mitochondrial ribosomal protein L3 (MRPL3) in BPD is increased in the prefrontal cortex (Iwamoto et al., 2005a), whereas it is reduced in the hippocampus (Konradi et al., 2004). Ubiquinol-cytochrome c reductase core protein II (UQCRC2) has been reported with opposite expression changes in the prefrontal cortex of BPD in two different studies (Iwamoto et al., 2005a; Sun et al., 2006), which is speculated to be due to a medication effect (Iwamoto et al., 2005a).
Although the genes categorized in this group have functions not only in the central nervous system but also in many organs, their roles in cellular signaling in neurons and glia are also well known. For example, the sub-category of signal transduction includes Regulator of G-protein signaling 4 (RGS4), and Phosphodiesterase 1A, calmodulin-dependent (PDE1A), which are known as key players in modulating the signals from dopamine and glutamate receptors (Levitt et al., 2006; Menniti et al., 2006). Downregulation of RGS4 in multiple brain regions of SCZ has been consistently reported in several studies (Arion et al., 2007; Bowden et al., 2008; Mirnics et al., 2001b) and expression of PDE1A is increased in the temporal cortex in SCZ (Aston et al., 2004) whereas its expression is increased in the prefrontal cortex in BPD (Iwamoto et al., 2004).
Some genes involved in protein degradation cascade were already classified in the group of “genes associated specifically with neuronal and/or glial functions” (see category 2 above and Table 2), but several of these genes are also listed in the ubiquitin/proteasome/protein degradation subgroup of this category, including Proteasome subunit, alpha type, 1 (PSMA1) (Altar et al., 2005; Vawter et al., 2001).
Finally, genetic susceptibility factors for SCZ and its interacting proteins have indicated the involvement of cytoskeletal disturbances in major mental illnesses (Kamiya et al., 2008). Consistent with this notion, some genes in the cytoskeleton subgroup are differentially expressed between patients and controls, and most of the reports are from the studies of the prefrontal cortex (Arion et al., 2007; Iwamoto et al., 2004).
Here we summarize mouse models in which the differentially expressed genes detected in patients are targeted. For the sake of simplicity we looked for publications of knockout (KO) models for genes that are downregulated in patients and those of transgenic overexpression models for genes that are upregulated in patients, respectively. Of note, in addition to mimicking the condition of a loss of function, KO mice are useful to study physiological role for the target gene. The information for unpublished KO models has been found at the website of International Knockout Mice Consortium (http://www.knockoutmouse.org/). Compared with abundant KO models, we came across only a few transgenic models in which genes upregulated in patients are overexpressed. Among all the publications on mouse models with disease relevance in human expression studies, none of them referred to the relevance for BPD and only three mouse models did for SCZ. This gap may imply that many mouse geneticists at present generate models that are potentially useful in studies of mental disorders, without acknowledging such merits and potentials towards translation (Figure 1).
Although KO mice have been generated for almost all neuroscience-related genes that are differentially expressed in patients with SCZ and/or BPD, only a few have been characterized from the viewpoint of mental illnesses. Basic neuroscience research has focused mainly on electrophysiology of the cerebellum and the hippocampus, because they are structurally relatively simple and anatomically well defined. Although the hippocampus is relevant to mental illnesses, attention to the circuitry between the hippocampus and the prefrontal cortex or the striatum may be expected for better understanding of major mental illnesses. Furthermore, behavioral changes in the animals have not been well studied yet under the light of endophenotypes for mental illnesses. This is partly because many of these homozygotes become embryonic lethal. Here we introduce three KO mice as representative models, in which genes included in these categories are genetically depleted and characterized from mental illness viewpoints. These include KO mice for Grm3 (mGluR3) (Table 1), syn2 (Table 1), and rgs4 (Table 4).
Although many genes associated with metabolic cascades are differentially expressed in SCZ and BPD brains compared with those from normal controls (Table 3), psychiatry-oriented studies with KO mice deficient in such metabolic genes have rarely been performed. However, considering the high prevalence of metabolic disturbances in SCZ (McEvoy et al., 2005) and BPD (Taylor and MacQueen, 2006), these models will be very useful and provide us with valuable information especially when their behavioral and neurochemical changes are examined in the light of endophenotypes relevant to major mental illnesses. The high incidence of metabolic abnormalities is accounted at least in part by poor life style and side effects of second generation antipsychotics (Correll et al., 2008); however, recent studies have supported a notion that mental illness-associated intrinsic abnormalities include metabolic disturbances, which may be further augmented by these exogenous factors. Mediators of metabolic signaling, such as insulin and leptin, are known to modulate cognitive functions (Benedict et al., 2011; Gerozissis, 2003; Harvey, 2007; Paz-Filho et al., 2008; Shanley et al., 2001). Animal models will be useful to test mechanistic details in the crosstalk of intrinsic susceptibility and exogenous factors.
Here we propose a research strategy in which molecular profiling in patient tissues is combined with characterization of genetic animal models (Figure 1). As etiological and construct validity is still debated in modeling major mental illnesses, this approach to focus on the molecules associated with pathophysiology may be a promising alternative.
Through the overview of the past publications, we now realize that there are several technical barriers and missing points to enhance this research strategy:
We thank Yukiko L. Lema for organizing the manuscript and preparing the figure. We also thank Dr. Minae Niwa and Ms. Ashley Wilson for critical reading. Grant supports: from MH-084018 Silvo O. Conte center (A.S.), MH-069853 (A.S.), MH-085226 (A.S.), MH-088753 (A.S.), Stanley (A.S.), RUSK (A.S.), S-R foundations (A.S.), NARSAD (H.J-P., A.S.) and Maryland Stem Cell Research Fund (A.S.).
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.