Topological and functional discovery in a gene coexpression meta-network of gastric cancer

Department of Surgery, Stanford University, Palo Alto, California, United States
Cancer Research (Impact Factor: 9.33). 01/2006; 66(1):232-41. DOI: 10.1158/0008-5472.CAN-05-2232
Source: PubMed


Gastric cancer is a leading cause of global cancer mortality, but comparatively little is known about the cellular pathways regulating different aspects of the gastric cancer phenotype. To achieve a better understanding of gastric cancer at the levels of systems topology, functional modules, and constituent genes, we assembled and systematically analyzed a consensus gene coexpression meta-network of gastric cancer incorporating >300 tissue samples from four independent patient populations (the "gastrome"). We find that the gastrome exhibits a hierarchical scale-free architecture, with an internal structure comprising multiple deeply embedded modules associated with diverse cellular functions. Individual modules display distinct subtopologies, with some (cellular proliferation) being integrated within the primary network, and others (ribosomal biosynthesis) being relatively isolated. One module associated with intestinal differentiation exhibited a remarkably high degree of autonomy, raising the possibility that its specific topological features may contribute towards the frequent occurrence of intestinal metaplasia in gastric cancer. At the single-gene level, we discovered a novel conserved interaction between the PLA2G2A prognostic marker and the EphB2 receptor, and used tissue microarrays to validate the PLA2G2A/EphB2 association. Finally, because EphB2 is a known target of the Wnt signaling pathway, we tested and provide evidence that the Wnt pathway may also similarly regulate PLA2G2A. Many of these findings were not discernible by studying the single patient populations in isolation. Thus, besides enhancing our knowledge of gastric cancer, our results show the broad utility of applying meta-analytic approaches to genome-wide data for the purposes of biological discovery.

Download full-text


Available from: Alex Boussioutas, Jun 11, 2014
  • Source
    • "Little is known about the signalling events and transcriptional machinery, leading to increased expression of sPLA 2 s in cancer, yet the published studies point to several growth factor signalling pathways. For example, a series of studies has shown that hGIIA sPLA 2 is overexpressed and secreted by androgen-independent prostate cancer cells due to elevated signalling of the HER/HER2-PI3K-Akt-NF-kB pathway [11] [15] [62], while the β-catenin-dependent Wnt signalling was identified as a major upstream regulator of hGIIA sPLA 2 expression in gastric cancer cells [21] [63]. Collectively, the limited amount of data available so far suggest that the modulated expression of sPLA 2 s in cancer may be a consequence of dysregulated epigenetic, growth factor signalling and/or inflammatory mechanisms, which are often cancer-and cell type-specific. "
    [Show abstract] [Hide abstract]
    ABSTRACT: Secreted phospholipases A2 (sPLA2s) hydrolyse cell and lipoprotein phospholipid membranes to release free fatty acids and lysophospholipids, and can also bind to specific proteins. Several sPLA2s have been associated with various cancers, including prostate, colon, gastric, lung and breast cancers, yet, their role is controversial and seems to be dependent on the cancer type, the local microenvironment and the enzyme studied. There is strong evidence that the expression of some sPLA2s, most notably the group IIA, III and X enzymes, is dysregulated in various malignant tissues, where, as described in a number of in vitro and in vivo studies using mouse models and according to correlations between sPLA2 expression and patient survival, a particular enzyme may exert either a pro- or an anti-tumourigenic role. It is becoming clear that there are multiple, context-dependent mechanisms of action of sPLA2s in different cancers. First, the role of sPLA2s in cancer has traditionally been associated with their enzymatic activity and ability to participate in the release of potent biologically active lipid mediators, in particular arachidonic acid-derived eicosanoids, which promote tumourigenesis by stimulating cell proliferation and cell survival, by abrogating apoptosis and by increasing local inflammation and angiogenesis. Second, several biological effects of sPLA2s were found to be independent of sPLA2 enzymatic activity, arguing for a receptor-mediated mechanism of action. Finally, recent studies have implicated sPLA2s in the regulation of basal lipid metabolism, opening a new window to the understanding of the diverse roles of sPLA2s in cancer. In this short review, we highlight the newest findings on the biological roles of sPLA2s in cancer, with emphasis on their diverse mechanisms of action.
    Biochimie 10/2014; 107. DOI:10.1016/j.biochi.2014.09.023 · 2.96 Impact Factor
  • Source
    • "The GCN also exhibits properties common to most naturally occurring networks such as scale-free, small world and hierarchical topology [3,4]. Due to the availability of large quantities of publically available expression data and the relative ease of construction, GCNs have been constructed for a broad array of organisms including human [2,5,6], yeast [7-9], Arabidopsis [10-13], rice [14,15], maize [16], potato [17] and many more. These networks have elucidated gene sets involved in varied biological systems including cell wall biosynthesis [13], mouse weight [18], and complex trait expression [19-22]. "
    [Show abstract] [Hide abstract]
    ABSTRACT: Background In genomics, highly relevant gene interaction (co-expression) networks have been constructed by finding significant pair-wise correlations between genes in expression datasets. These networks are then mined to elucidate biological function at the polygenic level. In some cases networks may be constructed from input samples that measure gene expression under a variety of different conditions, such as for different genotypes, environments, disease states and tissues. When large sets of samples are obtained from public repositories it is often unmanageable to associate samples into condition-specific groups, and combining samples from various conditions has a negative effect on network size. A fixed significance threshold is often applied also limiting the size of the final network. Therefore, we propose pre-clustering of input expression samples to approximate condition-specific grouping of samples and individual network construction of each group as a means for dynamic significance thresholding. The net effect is increase sensitivity thus maximizing the total co-expression relationships in the final co-expression network compendium. Results A total of 86 Arabidopsis thaliana co-expression networks were constructed after k-means partitioning of 7,105 publicly available ATH1 Affymetrix microarray samples. We term each pre-sorted network a Gene Interaction Layer (GIL). Random Matrix Theory (RMT), an un-supervised thresholding method, was used to threshold each of the 86 networks independently, effectively providing a dynamic (non-global) threshold for the network. The overall gene count across all GILs reached 19,588 genes (94.7% measured gene coverage) and 558,022 unique co-expression relationships. In comparison, network construction without pre-sorting of input samples yielded only 3,297 genes (15.9%) and 129,134 relationships. in the global network. Conclusions Here we show that pre-clustering of microarray samples helps approximate condition-specific networks and allows for dynamic thresholding using un-supervised methods. Because RMT ensures only highly significant interactions are kept, the GIL compendium consists of 558,022 unique high quality A. thaliana co-expression relationships across almost all of the measurable genes on the ATH1 array. For A. thaliana, these networks represent the largest compendium to date of significant gene co-expression relationships, and are a means to explore complex pathway, polygenic, and pleiotropic relationships for this focal model plant. The networks can be explored at Finally, this method is applicable to any large expression profile collection for any organism and is best suited where a knowledge-independent network construction method is desired.
    BMC Systems Biology 06/2013; 7(1):44. DOI:10.1186/1752-0509-7-44 · 2.44 Impact Factor
  • Source
    • "Thus, a deeper understanding of multifactorial mechanisms in HCC progression may be achieved using a systems-wide approach. This shift in paradigm is attracting substantial interest and has been applied successfully to define signaling cascades in cardiac hypertrophy [11], mechanisms of gastric cancer progression [12], and fundamental organizational properties of metabolic networks [13]. More recently, integration of HCC gene expression with topological features of human protein interaction networks resulted in enhanced diagnosis of HCC [14], while network analysis of hepatitis C virus (HCV)-induced HCC elucidated dysfunctional interactions among proteins and aberrant relationships between transcription factors and their target genes [15]. "
    [Show abstract] [Hide abstract]
    ABSTRACT: Hepatocellular carcinoma (HCC) is a leading cause of global cancer mortality. However, little is known about the precise molecular mechanisms involved in tumor formation and pathogenesis. The primary goal of this study was to elucidate genome-wide molecular networks involved in development of HCC with multiple etiologies by exploring high quality microarray data. We undertook a comparative network analysis across 264 human microarray profiles monitoring transcript changes in healthy liver, liver cirrhosis, and HCC with viral and alcoholic etiologies. Gene co-expression profiling was used to derive a consensus gene relevance network of HCC progression that consisted of 798 genes and 2,012 links. The HCC interactome was further confirmed to be phenotype-specific and non-random. Additionally, we confirmed that co-expressed genes are more likely to share biological function, but not sub-cellular localization. Analysis of individual HCC genes revealed that they are topologically central in a human protein-protein interaction network. We used quantitative RT-PCR in a cohort of normal liver tissue (n = 8), hepatitis C virus (HCV)-induced chronic liver disease (n = 9), and HCC (n = 7) to validate co-expressions of several well-connected genes, namely ASPM, CDKN3, NEK2, RACGAP1, and TOP2A. We show that HCC is a heterogeneous disorder, underpinned by complex cross talk between immune response, cell cycle, and mRNA translation pathways. Our work provides a systems-wide resource for deeper understanding of molecular mechanisms in HCC progression and may be used further to define novel targets for efficient treatment or diagnosis of this disease.
    PLoS ONE 04/2012; 7(4):e35510. DOI:10.1371/journal.pone.0035510 · 3.23 Impact Factor
Show more