A protein interaction based model for schizophrenia study
© Hsu et al. 2008
Published: 12 December 2008
Skip to main content
© Hsu et al. 2008
Published: 12 December 2008
Schizophrenia is a complex disease with multiple factors contributing to its pathogenesis. In addition to environmental factors, genetic factors may also increase susceptibility. In other words, schizophrenia is a highly heritable disease. Some candidate genes have been deduced on the basis of their known function with others found on the basis of chromosomal location. Individuals with multiple candidate genes may have increased risk. However it is not clear what kind of gene combinations may produce the disease phenotype. Their collective effect remains to be studied.
Most pathways except metabolic pathways are rich in protein-protein interactions (PPIs). Thus, the PPI network contains pathway information, even though the upstream-downstream relation of PPI is yet to be explored. Here we have constructed a PPI sub-network by extracting the nearest neighbour of the 36 reported candidate genes described in the literature. Although these candidate genes were discovered by different approaches, most of the proteins formed a cluster. Two major protein interaction modules were identified on the basis of the pairwise distance among the proteins in this sub-network. The large and small clusters might play roles in synaptic transmission and signal transduction, respectively, based on gene ontology annotation. The protein interactions in the synaptic transmission cluster were used to explain the interaction between the NRG1 and CACNG2 genes, which was found by both linkage and association studies. This working hypothesis is supported by the co-expression analysis based on public microarray gene expression.
On the basis of the protein interaction network, it appears that the NRG1-triggered NMDAR protein internalization and the CACNG2 mediated AMPA receptor recruiting may act together in the glutamatergic signalling process. Since both the NMDA and AMPA receptors are calcium channels, this process may regulate the influx of Ca2+. Reducing the cation influx might be one of the disease mechanisms for schizophrenia. This PPI network analysis approach combined with the support from co-expression analysis may provide an efficient way to propose pathogenetic mechanisms for various highly heritable diseases.
Schizophrenia is a severe mental disorder with grave personal and social costs . Approximately 1% of the population develops schizophrenia during their lifetime. Over the years, many genes have been reported to be responsible for the susceptibility to schizophrenia . In general, schizophrenia is considered to be a complex disease with multiple genetic and environment etiological factors. Linkage analysis, association and positional cloning studies and candidate gene approaches  have been successful in identifying risk genes. The way in which multiple genes, each possibly having a small individual contribution, leads to vulnerability and then the pathophysiology, remains to be elucidated. In order to figure out the relationship among those genes, we should investigate not only in gene-gene interaction level but also a whole picture at the protein level. Recent works to map the protein-protein interaction (PPI) in human to curate human metabolism and regulatory networks offer the relationships among different disease genes [4, 5]. The protein clusters in the network may represent the modules with biological functions . It is also reported that if the disease candidate genes are treated as a phenotype, these genes are likely to be function together in the normal cell .
In this study, we provided a novel strategy by taking advantages of PPI to discover the regulatory mechanisms among disease candidate genes. We speculated that disease candidate genes may cluster together in a functional network at a protein level. Protein complexes interact with preferred partners to form a biological module serving a specific collective function . When using a network-clustering method by calculating the pairwise distance in the protein interaction network , two major protein clusters were found which were involved in synaptic transmission and signal transduction protein cluster. We proposed a model to explain the interaction between NRG1 and CACNG2 which not only fell into the synaptic transmission cluster at protein interaction level but also associated at the gene-gene interaction level.
Recent molecular studies implicate neuregulin1 (NRG1) as the most promising risk factor for schizophrenia [9, 10]. Liu and colleagues also found suggestive linkage evidence of schizophrenia to loci near NRG1 on chromosome 8p21 in an ethnically distinct Taiwanese sample . There is also evidence that this genetic risk is elevated when accompanied by genetic changes in the gene for ErbB4, one of neuregulin's binding partners. NRG1-mediated ErbB signalling has important roles in neural development [12–14], as well as in the regulation of neurotransmitter receptors thought to be involved in the pathophysiology of schizophrenia . Hahn and colleagues suggest that enhanced endogenous NRG1-ERBB4 signalling may be responsible for N-methyl-D-aspartate receptors (NMDARs) hypofunction of the disease state . NMDA receptors are a major subtype of glutamate receptors and mediate slow excitatory postsynaptic potentials (EPSPs). Glutamate is the major excitatory neurotransmitter in the brain, and it has been proposed that disruption in glutamate signalling may underlie many of the symptoms of schizophrenia . NRG1 reduces the tyrosine phosphorylation of NMDA receptors, a modification that is triggered by the binding of NMDA or glutamate. NMDAR hypofunction may contribute to the symptomatic features of schizophrenia . ERBB4 associates with NMDAR via DLG4 (also called PSD95), and the binding to DLG4 is probably involved in the enhanced activation of ERBB4. This association provides a physical link between ERBB4 and the NMDAR. These findings add to our basic understanding of glutamatergic transmission, which has been implicated in the pathogenesis of schizophrenia.
CACNG2, also known as stargazin, was found to interact directly with AMPA receptor and allow interaction of the receptor with the scaffold proteins of the postsynaptic density, such as DLG4 [19, 20]. In a previous linkage study of schizophrenia that included Taiwanese samples, CACNG2 was also reported as a vulnerability gene for neuropsychologically defined subgroups of schizophrenic patients [21–23]. Bats and colleagues found that a mutation in PDZ domain of CACNG2 will increase AMPA receptor diffusion. CACNG2 regulates trafficking of AMPA-type glutamate receptors and stabilizes them at the postsynaptic density when neurotransmitters are received .
Since functionally related genes are usually co-regulated, we went further to check whether these genes were co-regulated in the brain tissue. If disease is considered as a perturbation to the normal state, different brain tumors may perturb a given gene to different extent. The GSE4271 microarray data set deposited in Gene Expression Omnibus at National Center for Biotechnology Information contains 100 samples from 15 assigned subsets . These gene expression data have been used to compute the correlation of the gene expression for each pair of genes and to establish a relevance network . Since disease is treated as a perturbation, this disease sample-derived network actually represents the co-expression in relation to normal cells.
As described previously, DLG4 appears to be a hub, which receives the NRG1-ERBB4 signal and then relays the signal to the NMDA receptor and the CACNG2.
On the other hand, the interaction between CACNG2 and DLG4 may anchor CACNG2 on the cytoskeleton. The anchored CACNG2 may recruit the AMPA receptor to the synaptic region  and increase the cation influx. Phosphorylation of DLG4 will release it from the cytoskeleton  and fail to recruit the AMPA receptors efficiently (Figure 5B). Although the calcium-dependent tyrosine kinase PTK2B is interacting with the DLG4 protein (see Figure 3), it is not clear whether PTK2B is catalyzing this reaction. This enzyme has been shown to regulate the activation of calcium channels . Because there are less AMPA receptors in the synaptic area, the capacity for cation influx (mainly calcium influx) will also decrease. Taken together, DLG4 links two mechanisms to decrease the cation influx at the synaptic area. The presence of gene variations in both NRG1 and CACNG2 may thus create a synergistic effect to affect the influx of Ca2+(Figure 5C). It has been shown that the schizophrenia patient has more NRG1-ERBB4 complex in the synaptic area than controls. This model specifically predicts that schizophrenia patients will have less cation influx in the synaptic area. Because both NMDA and AMPA receptors are triggered by glutamate, this model also predicts that glutamate may play an important role in pathogenesis. Many of the evidence for the glutamate hypothesis of schizophrenia implicate the NMDA-type glutamate receptor. But the glutamate role may be more complex because there are hints that AMPA receptors also contribute to schizophrenia symptoms, both independently or via effects on NMDA receptors .
DLG4, which receives the NRG1-ERBB4 signal and then relays the signal to the NMDA receptor and the CACNG2, links two mechanisms to decrease the cation influx at the synaptic area. On the basis of the protein interaction network, the NRG1-triggered NMDAR protein internalization and the CACNG2 mediated AMPA receptor recruiting may act together in the glutamatergic signalling process. Since both the NMDA and AMPA receptors are calcium channels, this process may regulate the influx of Ca2+. Ca2+ is necessary for transmission at the neuromuscular junction and other synapses. Reducing the synaptic calcium influx due to variants of NRG1 and CACNG2 might explain the basis of schizophrenia. This PPI network analysis approach combined with the support from co-expression analysis may provide an efficient way to propose disease mechanisms for various highly heritable diseases.
Protein-protein interaction data were obtained from Integrated Protein Interaction Resource (IPIR, http://ymbc.ym.edu.tw/ipir). IPIR has integrated protein-protein interaction information from BIND, DIP, HPRD, MINT, MIPS and IntAct databases. In this case, we chose brain, cerebellum, cerebrum and nervous as tissue filter. By using 36 candidate proteins as a data set to look for its primary protein neighbours and secondary protein neighbours, there were 831 proteins retrieved from databases. This network is displayed by Cytoscape which provides basic functionality for a visual representation of the graph and integrated data .
For each biological network investigated, relevant proteins (nodes) and the interaction among them (edges) were assembled as follow. Each edge in the network was assigned a length of one. A pairwise distance matrix contains the length of shortest path between every pair of proteins in the network. Each distance in the matrix was shown as an "association", defined as 1/d 2, where d is the shortest path distance. Generalized Association Plots (GAP) , which is a graphical environment for matrix visualization and information mining, were used to view this results. We have used the Gene Ontology annotations to assign functional category labels to the proteins of the PPI network . GoMiner, a tool for biological interpretation of gene sets, was used to annotate the enriched gene ontology terms of these protein clusters . GoMiner used the Gene Ontology (GO) to identify the biological processes, functions and components represented in gene lists.
The microarray data was obtained from Gene Expression Omnibus (GEO) , and the accession number for the data series is GSE4271 . Robust Multichip Average (RMA) normalization was performed to compute gene expression values for Affymetrix data and to carry out quality assessment using probe-level metrics . After normalizing the microarray data, we used the Pearson's correlation, performed by a Perl module called "Statistics::RankCorrelation", to represent the correlation coefficient of each pair of probe sets.
We thank to Dr. Hsin-Chou Yang in the Institute of Statistical Science, Academia Sinica for helpful discussion and Dr. Chia-Huei Lee in National Health Research Institute for providing array-CGH data. This work was supported by grant NSC-96-3112-B-010-015 from National Research Program for Genomic Medicine, National Science Council (Taiwan).
This article has been published as part of BMC Bioinformatics Volume 9 Supplement 12, 2008: Asia Pacific Bioinformatics Network (APBioNet) Seventh International Conference on Bioinformatics (InCoB2008). The full contents of the supplement are available online at http://www.biomedcentral.com/1471-2105/9?issue=S12.
This article is published under license to BioMed Central Ltd. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.