Home About us Editorial board Ahead of print Current issue Search Archives Submit article Instructions Subscribe Contacts Login 


 
 Table of Contents  
ORIGINAL ARTICLE
Year : 2014  |  Volume : 10  |  Issue : 7  |  Page : 114-124

Identification of melanoma biomarkers based on network modules by integrating the human signaling network with microarrays


1 Department of Dermatology, Huashan Hospital, Shanghai Medical College, Fudan University, Shanghai, China
2 Sbarro Institute for Cancer Research and Molecular Medicine, Center of Biotechnology, College of Science and Technology, Temple University, Philadelphia, Pennsylvania, USA

Date of Web Publication29-Nov-2014

Correspondence Address:
Lianjun Chen
Department of Dermatology, Huashan Hospital, Shanghai Medical College, Fudan University, 12 Wulumuqi Zhong Road, Shanghai 200040
China
Login to access the Email id

Source of Support: None, Conflict of Interest: None


DOI: 10.4103/0973-1482.145816

Rights and Permissions
 > Abstract 

Background: Melanoma is a leading cause of cancer death. Thus, accurate prognostic biomarkers that will assist rational treatment planning need to be identified.
Methods: Microarray analysis of melanoma and normal tissue samples was performed to identify differentially expressed modules (DEMs) from the signaling network and ultimately detect molecular markers to support histological examination. Network motifs were extracted from the human signaling network. Then, significant expression-correlation differential modules were identified by comparing the network module expression-correlation differential scores under normal and disease conditions using the gene expression datasets. Finally, we obtained DEMs by the Wilcoxon rank test and considered the average gene expression level in these modules as the classification features for diagnosing melanoma.
Results: In total, 99 functional DEMs were identified from the signaling network and gene expression profiles. The area under the curve scores for cancer module genes, melanoma module genes, and whole network modules are 92.4%, 90.44%, and 88.45%, respectively. The classification efficiency rates for nonmodule features are 71.04% and 79.38%, which correspond to the features of cancer genes and melanoma cancer genes, respectively. Finally, we acquired six significant molecular biomarkers, namely, module 10 (CALM3, Ca 2+ , PKC, PDGFRA, phospholipase-g, PIB5PA, and phosphatidylinositol-3-kinase), module 14 (SRC, Src homology 2 domain-containing [SHC], SAM68, GIT1, transcription factor-4, CBLB, GRB2, VAV2, LCK, YES, PTCH2, downstream of tyrosine kinase [DOK], and KIT), module 16 (ELK3, p85beta, SHC, ZFYVE9, TGFBR1, TGFBR2, CITED1, SH3KBP1, HCK, DOK, and KIT), module 45 (RB, CCND3, CCNA2, CDK4, and CDK6), module 75 (PCNA, CDK4, and CCND1), and module 114 (PSD93, NMDAR, and FYN).
Conclusion: We explored the gene expression profile and signaling network in a global view and identified DEMs that can be used as diagnostic or prognostic markers for melanoma.

Keywords: Human signaling network, melanoma, microarray analysis, module, network motif


How to cite this article:
Huang C, Sheng Y, Jia J, Chen L. Identification of melanoma biomarkers based on network modules by integrating the human signaling network with microarrays. J Can Res Ther 2014;10, Suppl S3:114-24

How to cite this URL:
Huang C, Sheng Y, Jia J, Chen L. Identification of melanoma biomarkers based on network modules by integrating the human signaling network with microarrays. J Can Res Ther [serial online] 2014 [cited 2019 Sep 17];10:114-24. Available from: http://www.cancerjournal.net/text.asp?2014/10/7/114/145816

Chunyun Huang and Youyu Sheng contribute equally to this work



 > Introduction Top


Melanoma is an aggressive and lethal skin cancer. In the United States, it is currently the fifth and seventh most commonly diagnosed cancer in men and women, respectively, with its incidence increasing to 194% from 1975 to 2011. [1],[2] Approximately 76,100 new melanoma cases with 9,710 deaths are expected in the United States by the end of 2014. [1] Melanoma is a leading cause of cancer death because of its metastatic potential. Despite recent advances in therapeutic strategies, melanoma with distant metastasis still portends a poor prognosis with a 5-year survival rate of 16%. [2] Thus, an accurate diagnosis, prognosis, and assessment of progression are critical. Histopathology is usually used to diagnose melanoma; however, this approach cannot differentiate melanoma from certain types of benign nevi. Hence, the development of molecular biomarkers that will assist the diagnosis and prognosis of melanoma is urgently needed.

An increasing number of diagnostic and prognostic biomarkers for cancers have been identified from genome-wide high-throughput expression profiles. [3],[4],[5],[6] These biomarkers provide a valuable platform for personalized medicine. However, analysis of individual genes is not sufficient to elucidate molecular mechanisms; thus, network-based methods have been proposed. Network motifs are small, repeated, and conserved biological units that serve as "building blocks" within molecular networks (e.g. transcriptional regulatory networks, signaling networks, and metabolic networks). [7] Motifs can respond to external stimuli by modulating gene expression. Therefore, identifying cancer susceptibility biomarkers, combined with network motifs and gene expression profiles, [8] can significantly improve the diagnosis and prognosis of cancer. [9]

Network motifs were first systematically described in Escherichia coli. [10] In recent years, network motifs have attracted increasing attention in molecular network research. [11] A recently developed network motif-based approach integrates biological network topology and high-throughput gene expression data to identify markers not as individual genes, but as network motifs. Shellman et al. [12] compared the motif distributions in the metabolic networks of 21 species across six kingdoms of life. Wu et al. [13] investigated the response of cervical carcinoma to epidermal growth factor (EGF) using network motifs in the regulatory network. Other studies investigated signaling network-based biomarkers in breast cancer, [9],[14],[15] lung cancer, [16],[17] gastric cancer, [18] and so on. [19],[20] Large-scale sequencing investigations have revealed that signaling pathways initiate melanoma formation. Changes in signaling networks such as cell proliferation and apoptosis not only indicate disruptions that lead to carcinogenesis, but also reveal alterations in expression-correlation differential scores between normal and disease samples. The present study identified signaling network-based molecular biomarkers for the diagnosis and prognosis of melanoma.

To classify melanoma samples, we selected differentially expressed modules by integrating the gene expression profiles and the human signaling network through a network module-based method. Compared with individual marker genes selected without network information, the identified modules were more efficient in differentiating diverse melanoma samples.


 > Materials and Methods Top


Microarray data

Gene expression datasets were extracted from the GEO database (http://www.ncbi.nlm.nih.gov/geo/) [Table 1]. To integrate the gene expression and signal network datasets, we mapped the gene expression value of each gene into the network.
Table 1: Gene expression profiles of melanoma samples


Click here to view


Human signaling network

Basing on a previous study, [21] we manually curated the signaling molecules and the interactions between these molecules from a comprehensive signaling pathway database, BioCarta (http://www.biocarta.com/), to establish the human signaling network. We extended this network by extracting and annotating the signaling molecules and their relations from another literature-mined signaling network that contains B500 signaling molecules [22] and from the Cancer Cell Map (http://cancer.cellmap.org/cellmap/), a database that contains 10 manually curated signaling pathways for cancer. Consequently, the signaling network contains 1,634 nodes and 5,089 links that include 2,403 activation links, 741 inhibitory links, 1,915 physical interaction, and 30 unknown links.

Cancer genes and melanoma cancer genes

We downloaded the cancer genes and melanoma cancer genes from F-census online database (http://210.46.85.180:8080/fcensus/). This database is designed for browsing, evaluating, and selecting functionally consistent cancer genes from various data sources, including the CGC, OMIM, AGCOH, CancerGenes, TSGDB, and TGDB databases, as well as two gene lists (H-list and R-list) identified by two types of high-throughput techniques. Finally, we obtained 2104 cancer genes and 56 melanoma cancer genes.

Network analysis

With the Cytoscape software, Cytoscape is open-source software for integration, visualization and analysis of biological networks. It can be extended through Cytoscape plugins, enabling a broad community of scientists to contribute useful features, we constructed the signaling network and mapped the cancer genes and the melanoma cancer genes into the network. We also analyzed the topological information of these genes in the signaling network and applied the plug-in of Cytoscape-MINE to select network motifs.

Differentially expressed modules

We mapped the gene expression profiles with disease and normal samples from the GEO database into the motifs and calculated the significant differential scores for each motif according to the expression profiles.

Given a particular motif M, we calculated the expression-correlation differential score S as



where M i is the score of module i, n is the number of genes in module i, k is the number of genes both in module i and expression data, and S j is the differential score for gene j between normal and disease samples in expression data. The score calculation was based on the Wilcoxon rank test.

Classification

The classification features were selected by screening for significant changes in gene expression between normal and disease samples. We compared the classification accuracy for diverse classification: Modules, melanoma cancer genes, and cancer genes. Classification accuracy was estimated by using tenfold cross-validation, and a support vector machine served as the classifier.


 > Results Top


Signaling network with cancer genes and melanoma cancer genes

After calculating other topological features, we found the average shortest path of cancer genes by comparing the melanoma cancer genes with the whole genes in the signaling network [Figure 1]a. The cancer and melanoma cancer genes shared the similar betweenness centrality [Figure 1]b, but the closeness centrality of melanoma cancer genes was higher than cancer genes [Figure 1]c, which indicated that melanoma cancer genes were likely to act a cluster than the whole cancer genes. The cluster efficiencies of cancer genes and melanoma cancer genes were higher than that of whole genes in the network [Figure 1]d. This result suggests that cancer genes interact with other genes through modules to change the normal biological process. In the human signaling network, the average degrees of whole network genes, cancer genes, and melanoma cancer genes were 5.73, 8.36, and 13.57, respectively [Figure 1]e. Thus, special cancer genes can affect many genes to disrupt the normal biological functions. Basing on the degree distribution, we found that most genes interacted with the genes situated up to the third degree [Figure 2].
Figure 1: Topological features for the whole signaling network, cancer genes and melanoma cancer genes (a) The average shortest path, (b) betweenness centrality, (c) closeness centrality, (d) clustering coefficient, (e) degree, (f) topological coefficient

Click here to view
Figure 2: The degree distribution for the whole signaling network, the degree distribution for the whole signaling network and melanoma cancer genes. (a) The degree distribution for the whole signaling network, (b) the degree distribution for cancer genes, (c) the degree distribution for melanoma cancer genes

Click here to view


Melanoma cancer gene-related modules

Through the plugin MINE of Cytoscape, we obtained 135 modules. Combining the gene expression profiles, we calculated the module scores in disease and normal samples. Then, we selected 99 significant modules by using t-test (P < 0.05) for further research [Supplementary Figures S1 [Additional file 1], Figure S2 [Additional file 2], Figure S3 [Additional file 3] and Figure S4 [Additional file 4] and six melanoma cancer gene-related modules. We acquired the following six most significantly different modules that contained melanoma cancer genes [Figure 3]: Modules 10, 14, 16, 45, 75, and 114. Literature search showed that most of the genes in the identified modules were associated with melanoma cancer [Table 2]. This result suggests that the abnormal modules can dysregulate the function of melanoma.
Figure 3: The signaling network and 6 significant differentially expressed modules

Click here to view
Table 2: Six significant modules associated with melanoma


Click here to view


Classification evaluation

To classify the disease and normal samples, we compared two features: Module features (whole network modules, cancer module genes, and melanoma module genes) and nonmodule features (cancer genes and melanoma cancer genes). For the module features, we extracted all the genes in the modules and assessed the gene expression in all samples. The area under the curve scores for the cancer module genes, melanoma module genes, and whole network modules were 92.4%, 90.44%, and 88.45%, respectively [Figure 4]a-c. The classification efficiency rates for the nonmodule features were 71.04% and 79.38%, which corresponded to the features of cancer genes and melanoma cancer genes, respectively [Figure 4]d and e.
Figure 4: ROC curve of diverse classification features. (a) Classification accuracy of differentially expressed modules (DEMs) related to cancer genes, (b) classification accuracy of DEMs related to melanoma cancer genes, (c) classification accuracy of all DEMs, (d) classification accuracy of cancer genes, (e) classification accuracy of melanoma cancer genes

Click here to view



 > Discussion Top


Differences in the average shortest path, clustering coefficient, and degree were determined after comparing the network topology characteristics of the cancer genes, melanoma cancer genes, and whole network genes. The average shortest paths for the cancer genes, melanoma cancer genes, and whole network genes were 2, 3.5, and >4, respectively. These results indicate that the cancer genes interact with the neighbor genes to disrupt the normal signaling pathways and trigger cancer development. The degree of the melanoma cancer genes was higher than those of the cancer genes and whole network genes. This result suggests that the melanoma cancer genes play pivotal roles in the signaling network. The cancer genes and the melanoma cancer genes had higher clustering coefficients than the whole network genes. This finding indicates that these genes are more likely to activate functions as modules instead of as single genes. We considered modules 10 and 14 as examples.

Module 10 comprises Ca 2+ and six genes. Alpha-type platelet-derived growth factor receptor (PDGFRA) is the melanoma cancer gene, and protein kinase C (PKC) is the other cancer gene. PDGFRA is a protein encoded by the PDGFRA gene in humans. This gene encodes a cell surface tyrosine kinase receptor for members of the platelet-derived growth factor family. These growth factors are mitogens for cells of mesenchymal origin. The identity of the growth factor bound to a receptor monomer determines whether the functional receptor is a homodimer or a heterodimer, composed of both PDGFRA alpha and beta polypeptides. Studies in knockout mice, where homozygosity is lethal, indicate that PDGFRA is particularly important for kidney development because mice heterozygous for the receptor exhibit defective kidney phenotypes. [23] PDGFRA is a receptor tyrosine kinase (RTK) that is activated by mutations or small deletions in a subset of gastrointestinal stroma tumors. [24] PDGFRA is overexpressed in some melanomas, [25] suggesting that this protein can also be a target of somatic mutations in some melanomas. Meanwhile, the genetic mutations of PDGFRA are not influenced by the age, thickness, and ulceration status of primary melanomas. PDGFRA mutations have been observed in 4.6% of melanomas, particularly in acral and mucosal melanomas. [24] PKC is a family of kinase enzymes that control the function of other proteins by phosphorylating the hydroxyl groups of serine and threonine amino acid residues on these proteins. Consequently, PKC enzymes are activated by signals, such as increases in the concentration of diacylglycerol or Ca 2+ . Hence, PKC enzymes play important roles in several signal transduction cascades. Soares et al. [26] proposed that the simultaneous activation of the phospholipase C (PLC)-PKC-MEK1/2-ERK1/2 and phosphatidylinositol-3-kinase (PI3K) pathways is the principal mechanism responsible for the proliferative effect elicited by inosine and its significant role in melanoma cancer progression. Moreover, Halder et al. [27] observed a marked variation in the expression of PKCα and PKCd isotypes in B16F10 melanoma tumor cells compared with normal melanocytes and suggested the presence of a reciprocal PKC signaling pathway. They speculated that this pathway regulates ceramide generation and provides important clues to target melanoma cancer by manipulating the PKCd-ceramide axis. Cerne et al. [28] found that PKC inhibitors combined with IR significantly decrease the viability, proliferation, and clonogenic potential of GNAQ (mt) but not GNAQ (wt)/BRAF (mt) cells compared with IR alone. This combined treatment increases the antiproliferative and proapoptotic effects of IR on GNAQ (mt) cells by delaying the resolution of DNA damage and enhancing the induction of proteins involved in cell cycle arrest, cell growth arrest, and apoptosis. These findings suggest that future clinical trials should consider combining PKC inhibitors with RT in GNAQ (mt) UM.

Module 10 showed that PDGFRA indirectly interacts with PKC through Ca 2+ . Calmodulin 3 (CALM3) or PHKD/PHKD3/HEL-S-72 encodes the CALM3 protein. Calmodulin, [29] the principal mediator of the calcium signal, regulates numerous processes pertinent to neural function, and is encoded by three genes (CALM1, CALM2, and CALM3) located on different chromosomes. However, the CALM gene family is differentially active at the transcriptional level. CALM3 is located at chromosome 19 and is at least fivefold more actively transcribed than CALM1 or CALM2. Moreover, the structural alterations in the CALM3 gene are not associated with the altered Ca 2+ homeostasis. [30] Thus, CALM3 can be a potential biomarker with the melanoma cancer genes to dysregulate the biological process. PDGFRA actively indirectly interacts with the cancer gene PKC through phosphatidylinositol 4,5-bisphosphate 5-phosphatase A (PIB5PA)/PLC-g. PIB5PA is an enzyme encoded by the INPP5J gene in humans. [31],[32] Previous studies have shown that PIB5PA is commonly downregulated or lost in melanomas, contributing to the elevated activation of PI3K/Akt in melanoma cells. Ye et al. [33] proved that PIB5PA deficiency renders melanoma cells resistant to RAF/mitogen-activated protein kinase (MEK) inhibitors. This finding suggests that PIB5PA restoration may be a useful strategy to improve the therapeutic efficacy of inhibitors in melanoma treatment. In the subnetwork, PIKs and PLC-g actively interacted with PIB5PA, and PLC-g and PDGFRA activated PI3K. This result indicates that PI3K contributes to melanoma development. The PI3K family of enzymes is involved in cellular functions, such as growth, proliferation, differentiation, motility, survival, and intracellular trafficking, all of which are involved in cancer development. In addition, PI3K is a family of related intracellular signal transducer enzymes that can phosphorylate the hydroxyl group on the third position of the inositol ring of phosphatidylinositol. [34] The pathway with oncogene PIK3CA and tumor suppressor PTEN (gene) is implicated in the insensitivity of cancer tumors to insulin and IGF1, as well as in calorie restriction. [35],[36] Wang et al. [37] found that PI3K/AKT inhibition primarily suppresses proliferation and that MEK/ERK1/2 and PI3K/AKT co-inhibition synergistically induces apoptosis. PLC-g encoded by small wing [31] links growth and patterning/differentiation by modulating some MAPK outputs once activated by the insulin pathway; particularly, sl promotes growth and suppresses ectopic differentiation in the developing eye and wing, allowing cells to attain a normal size and differentiate properly. sl mutants possess a combination of both growth and patterning/differentiation phenotypes. Therefore, PLC-g functions when growth ends and differentiation begins; it also coordinates these two processes. [38] PLC-g was the kernel node in module 10, which physically interacted with PDGFRA and upregulated CALM3, PI3K, PIB5PA, and PKC. Thus, PLC-g is a module biomarker. We enriched the genes of module 10 in GO functions and KEGG pathways with the hypergeometric test. P < 0.05 was considered to indicate significance. In terms of molecular function, the module participates in PDGFRA activity and N-terminal myristoylation domain binding. In terms of biological function, the module genes are involved in the positive regulation of ryanodine-sensitive calcium-release channel activity and in the regulation of sequestered Ca 2+ release into the cytosol by the sarcoplasmic reticulum. We also found that these genes participate in the glioma and calcium signaling pathways.

Module 14 comprises 13 genes, including six other cancer genes (SRC, LCK, GRB2, CBLB, GIT1, and PTCH2) and one melanoma cancer gene (KIT). In this module, SRC has the highest degree and plays a vital function in the signaling subnetwork. SRC is a non-RTK that is expressed in cancer cells. It is also known as a proto-oncogene that is highly similar to the v-src gene of Rous sarcoma virus. This proto-oncogene regulates embryonic development and cell growth. The protein it encodes is a tyrosine-protein kinase whose activity can be inhibited by c-SRC kinase phosphorylation. Mutations in this gene might be involved in the malignant progression of colon cancer, and two transcript variants encoding the same protein have been found for this gene. [39] Girotti et al. [40] showed that inhibiting the EGF receptor (EGFR)-SRC family kinase (SFK)-STAT3 signaling pathway overcomes BRAF inhibitor resistance in melanoma. They found that the EGFR/SFK pathway mediates resistance to vemurafenib in BRAF-mutant melanoma and that BRAF and EGFR or SFK inhibition blocks the proliferation and invasion of these resistant tumors, thereby providing potentially effective therapeutic options for these patients. In signaling network 14, SRC, Src homology 2 domain-containing (SHC), and downstream of tyrosine kinase (DOK) physically interacted with one another. SHC is a structurally conserved protein domain within the Src oncoprotein [41] and in many other intracellular signal-transducing proteins. [42] Proteins with SH2 domains are allowed to dock to phosphorylated tyrosine residues on other proteins. These domains are commonly found in adapter proteins that aid in the signal transduction of RTK pathways. [43] Metastatic melanoma is an aggressive type of cancer, with a survival expectation of above 6 months only in rare cases. Despite advances in the characterization of underlying molecular pathways and in the development of specific targeted treatments, the available chemo-and immunotherapy cannot significantly prolong the survival of advanced-stage melanoma patients. Pasini et al. [44] found that Rai like protein, a newly identified SHC family member, is selectively expressed during the transition to metastatic melanoma and thus is a potential melanoma-specific drug target. The DOK family of adaptor proteins consists of seven members that share a structural topology characterized by an NH2-terminal pleckstrin homology domain, a central phosphotyrosine-binding domain, and SH2 target motifs in the carboxyl-terminal moiety. [29],[45] These proteins act as common substrates for multiple PTKs, including RTKs and non-RTKs, from which they modulate signaling pathways involved in various natural cellular processes, such as proliferation, apoptosis, growth, and migration. [46],[47] Several members of the DOK protein family are identified as modulators of cell proliferation/growth pathways. Overexpression of DOK proteins in human NK cells reduces cell activation induced by NK cell-activating receptors. Deregulation of specific DOK members has also been associated with specific cancers. [48] SRC actively promotes KIT, transcription factor 4 (TCF-4), GIT1, SAM68, and VAV2 in module 14. TCF-4 directly interacted with KIT, GIT1, and SAM68 while indirectly linked with other nodes of module 14. TCF-4 is a basic helix-loop-helix TCF whose encoded protein recognizes an Ephrussi-box ('E-box') binding site ('CANNTG'), a motif first identified in immunoglobulin enhancers. This gene is broadly expressed and may play an important role in nervous system development. [49] Moreover, Gheidari et al. [50] found that TCF-4 silencing sensitizes the colon cancer cell line to oxaliplatin, a common chemotherapeutic drug. Eichhoff et al. [51] showed that the phenotype switching behavior of melanoma cells is regulated by differential LEF1/TCF-4 activity and that TCF-4 binding sites are the principal regulatory regions that direct versican production. Their results provided new insights into versican promoter regulation during melanoma progression. Some melanomas arising from acral, mucosal, and chronically sun-damaged sites harbor activating mutations and amplification of the type III transmembrane RTK KIT. Among patients with advanced melanoma harboring KIT alterations, treatment with imatinib mesylate results in significant clinical responses in a subset of patients; thus, KIT can be a therapeutic target in metastatic melanoma. [52],[53] Basing on the analysis of KIT, SRC, and TCF-4, we concluded that the genes in a common module shared the same molecular functions and influenced the same biological processes in melanoma. The pathway enrichment suggests that the module genes participate in 10 crucial pathways, such as those in cancer, natural killer cell-mediated cytotoxicity, and acute myeloid leukemia. In terms of biological processes, the module genes are involved in the regulation of signaling pathway and cell communication. The module genes also promote EGF receptor binding, molecular adaptor activity, and so on.

To explore the similar functions of genes in the same module, we analyzed the enrichment of biological processes [Figure 5], cellular components [Figure 6], and molecular functions [Figure 7]. In module 16, 11 genes participate in enzyme-linked receptor protein signaling regulation, myeloid leukocyte differentiation, negative regulation of cell differentiation, negative regulation of programmed cell death, and melanocyte differentiation. In terms of molecular functions, these genes influence receptor signaling protein activity, SMAD binding, protein kinase activity, and so on. All of the genes in module 45 are highly enriched in the biological processes of cell division, regulation of cell cycle, positive regulation of fibroblast proliferation, and mitotic cell cycle. Moreover, these genes are principally located in the intracellular region and nucleus but are more specifically found in the cyclin-dependent protein kinase holoenzyme complex, the male pronucleus, a protein complex, and nucleoplasm. In terms of molecular function, the module genes participate in cyclin-dependent protein kinase activity, kinase binding, and cyclin binding. The genes in module 75 are involved in the transition across the mitotic cell cycle, the interphase of the mitotic cell cycle, the response to the metal ion, translesion synthesis, and so on. The module genes promote DNA polymerase processivity factor activity, purine-specific mismatch base pair DNA N-glycosylase activity, dinucleotide insertion or deletion binding, and DNA insertion or deletion binding. We found that all the genes in the module are located in the intracellular region and nucleus. These genes consist of the cyclin-dependent protein kinase holoenzyme complex, nucleoplasm, PCNA complex, and nuclear lumen. According to functional enrichment, the module genes also participate in DNA polymerase processivity factor activity, purine-specific mismatch base pair DNA N-glycosylase activity, dinucleotide insertion or deletion binding, DNA insertion or deletion binding, and so on. We enriched all the genes in module 114 according to biological process, molecular function, and cellular component. Results showed that the module genes mainly function in the detection of mechanical stimuli involved in pain sensory perception, T-cell proliferation, and so on. In terms of molecular function, the module genes regulate guanylate kinase activity, kinase activity, transferase activity, and transferring phosphorus-containing groups.
Figure 5: The functional enrichment of biological process for module 14, 16, 45, 75, 114

Click here to view
Figure 6: The functional enrichment of cellular component for module 10, 14, 16, 45, 75

Click here to view
Figure 7: The molecular function for module 10, 14, 16, 45, 75

Click here to view



 > Conclusion Top


We analyzed melanoma cancer biomarkers using the module strategy and found that melanoma cancer occurs in the form of modules and not signal genes. Functional enrichment analysis and interaction network analysis revealed that genes in the same module share similar biological functions. We acquired six significant molecular biomarkers, namely, module 10 (CALM3, Ca 2+ , PKC, PDGFRA, PLC-g, PIB5PA, and PI3K), module 14 (SRC, SHC, SAM68, GIT1, TCF-4, CBLB, GRB2, VAV2, LCK, YES, PTCH2, DOK, and KIT), module 16 (ELK3, p85beta, SHC, ZFYVE9, TGFBR1, TGFBR2, CITED1, SH3KBP1, HCK, DOK, and KIT), module 45 (RB, CCND3, CCNA2, CDK4, CDK6), module 75 (PCNA, CDK4, and CCND1), and module 114 (PSD93, NMDAR, and FYN). However, confirmatory experiments still need to be conducted.[55]


 > Acknowledgments Top


The present work (no. 20134363) was sponsored by Shanghai Municipal Bureau of Health.

 
 > References Top

1.
Siegel R, Ma J, Zou Z, Jemal A. Cancer statistics, 2014. CA Cancer J Clin 2014;64:9-29.  Back to cited text no. 1
    
2.
Weinstein D, Leininger J, Hamby C, Safai B. Diagnostic and prognostic biomarkers in melanoma. J Clin Aesthet Dermatol 2014;7:13-24.  Back to cited text no. 2
    
3.
Alizadeh AA, Staudt LM. Genomic-scale gene expression profiling of normal and malignant immune cells. Curr Opin Immunol 2000;12:219-25.  Back to cited text no. 3
    
4.
Alizadeh AA, Eisen MB, Davis RE, Ma C, Lossos IS, Rosenwald A, et al. Distinct types of diffuse large B-cell lymphoma identified by gene expression profiling. Nature 2000;403:503-11.  Back to cited text no. 4
    
5.
Ben-Dor A, Bruhn L, Friedman N, Nachman I, Schummer M, Yakhini Z. Tissue classification with gene expression profiles. J Comput Biol 2000;7:559-83.  Back to cited text no. 5
    
6.
Ambrogi V, Mineo TC, for the Multidisciplinary Tor Vergata University Study Group for Malignant Pleural M. Clinical and biologic prognostic factors in malignant pleural mesothelioma. Thorac Cancer 2012;3:289-302.  Back to cited text no. 6
    
7.
Milo R, Shen-Orr S, Itzkovitz S, Kashtan N, Chklovskii D, Alon U. Network motifs: Simple building blocks of complex networks. Science 2002;298:824-7.  Back to cited text no. 7
    
8.
Zhang Y, Xuan J, de Los Reyes BG, Clarke R, Ressom HW. Network motif-based identification of breast cancer susceptibility genes. Conf Proc IEEE Eng Med Biol Soc 2008;2008:5696-9.  Back to cited text no. 8
    
9.
Chen L, Qu X, Cao M, Zhou Y, Li W, Liang B, et al. Identification of breast cancer patients based on human signaling network motifs. Sci Rep 2013;3:3368.  Back to cited text no. 9
    
10.
Shen-Orr SS, Milo R, Mangan S, Alon U. Network motifs in the transcriptional regulation network of Escherichia coli. Nat Genet 2002;31:64-8.  Back to cited text no. 10
    
11.
Awan A, Bari H, Yan F, Moksong S, Yang S, Chowdhury S, et al. Regulatory network motifs and hotspots of cancer genes in a mammalian cellular signalling network. IET Syst Biol 2007;1:292-7.  Back to cited text no. 11
    
12.
Shellman ER, Burant CF, Schnell S. Network motifs provide signatures that characterize metabolism. Mol Biosyst 2013;9:352-60.  Back to cited text no. 12
    
13.
Wu SF, Qian WY, Zhang JW, Yang YB, Liu Y, Dong Y, et al. Network motifs in the transcriptional regulation network of cervical carcinoma cells respond to EGF. Arch Gynecol Obstet 2013;287:771-7.  Back to cited text no. 13
    
14.
Chuang HY, Lee E, Liu YT, Lee D, Ideker T. Network-based classification of breast cancer metastasis. Mol Syst Biol 2007;3:140.  Back to cited text no. 14
    
15.
Fröhlich H. Network based consensus gene signatures for biomarker discovery in breast cancer. PLoS One 2011;6:e25364.  Back to cited text no. 15
    
16.
Wang YC, Chen BS. A network-based biomarker approach for molecular investigation and diagnosis of lung cancer. BMC Med Genomics 2011;4:2.  Back to cited text no. 16
    
17.
Li K, Li Z, Zhao N, Xu Y, Liu Y, Zhou Y, et al. Functional analysis of microRNA and transcription factor synergistic regulatory network based on identifying regulatory motifs in non-small cell lung cancer. BMC Syst Biol 2013;7:122.  Back to cited text no. 17
[PUBMED]    
18.
Liu G, Li DZ, Jiang CS, Wang W. Transduction motif analysis of gastric cancer based on a human signaling network. Braz J Med Biol Res 2014;47:369-75.  Back to cited text no. 18
[PUBMED]    
19.
Li W, Chen L, Li X, Jia X, Feng C, Zhang L, et al. Cancer-related marketing centrality motifs acting as pivot units in the human signaling network and mediating cross-talk between biological pathways. Mol Biosyst 2013;9:3026-35.  Back to cited text no. 19
[PUBMED]    
20.
Liang AJ, Hong Y, Sun Y, Gao MZ, Zhao XM. The regulation network and network motif analysis in ovarian cancer. Eur J Gynaecol Oncol 2013;34:170-4.  Back to cited text no. 20
    
21.
Cui Q, Ma Y, Jaramillo M, Bari H, Awan A, Yang S, et al. A map of human cancer signaling. Mol Syst Biol 2007;3:152.  Back to cited text no. 21
    
22.
Ma'ayan A, Jenkins SL, Neves S, Hasseldine A, Grace E, Dubin-Thaler B, et al. Formation of regulatory patterns during signal propagation in a Mammalian cellular network. Science 2005;309:1078-83.  Back to cited text no. 22
    
23.
Entrez Gene: PDGFRA platelet-derived growth factor receptor, alpha polypeptide. Available from: http://wwwncbinlmnihgov/gene/5156. [Last accessed on 2014 Aug 07].  Back to cited text no. 23
    
24.
Heinrich MC, Corless CL, Duensing A, McGreevey L, Chen CJ, Joseph N, et al. PDGFRA activating mutations in gastrointestinal stromal tumors. Science 2003;299:708-10.  Back to cited text no. 24
    
25.
Barnhill RL, Xiao M, Graves D, Antoniades HN. Expression of platelet-derived growth factor (PDGF)-A, PDGF-B and the PDGF-alpha receptor, but not the PDGF-beta receptor, in human malignant melanoma in vivo. Br J Dermatol 1996;135:898-904.  Back to cited text no. 25
    
26.
Soares AS, Costa VM, Diniz C, Fresco P. Inosine strongly enhances proliferation of human C32 melanoma cells through PLC-PKC-MEK1/2-ERK1/2 and PI3K Pathways. Basic Clin Pharmacol Toxicol 2014.  Back to cited text no. 26
    
27.
Halder K, Banerjee S, Bose A, Majumder S, Majumdar S. Overexpressed PKCd downregulates the expression of PKCa in B16F10 melanoma: Induction of apoptosis by PKCd via ceramide generation. PLoS One 2014;9:e91656.  Back to cited text no. 27
[PUBMED]    
28.
Cerne JZ, Hartig SM, Hamilton MP, Chew SA, Mitsiades N, Poulaki V, et al. Protein kinase C inhibitors sensitize GNAQ mutant uveal melanoma cells to ionizing radiation. Invest Ophthalmol Vis Sci 2014;55:2130-9.  Back to cited text no. 28
    
29.
Berger AH, Niki M, Morotti A, Taylor BS, Socci ND, Viale A, et al. Identification of DOK genes as lung tumor suppressors. Nat Genet 2010;42:216-23.  Back to cited text no. 29
    
30.
Ibarreta D, Tao J, Parrilla R, Ayuso MS. Mutation analysis of chromosome 19 calmodulin (CALM3) gene in Alzheimer's disease patients. Neurosci Lett 1997;229:157-60.  Back to cited text no. 30
    
31.
Dunham I, Shimizu N, Roe BA, Chissoe S, Hunt AR, Collins JE, et al. The DNA sequence of human chromosome 22. Nature 1999;402:489-95.  Back to cited text no. 31
    
32.
Entrez Gene: PIB5PA phosphatidylinositol (4,5) bisphosphate 5-phosphatase, A. Available from: http://wwwncbinlmnihgov/gene?Db=gene and Cmd=ShowDetailView and TermToSearch=27124. [Last accessed on 2014 Aug 07].  Back to cited text no. 32
    
33.
Ye Y, Li Q, Hu WL, Tseng HY, Jin L, Zhang XD, et al. Loss of PI (4,5) P2 5-Phosphatase A Contributes to Resistance of Human Melanoma Cells to RAF/MEK Inhibitors. Transl Oncol 2013;6:470-81.  Back to cited text no. 33
    
34.
Temple S, Alvarez-Buylla A. Stem cells in the adult mammalian central nervous system. Curr Opin Neurobiol 1999;9:135-41.  Back to cited text no. 34
    
35.
Entrez Gene: Cell pathway on overdrive prevents cancer response to dietary restriction. Mar 11, 2009. Available from: http://www.PhysOrg.com. [Last retrieved on 2009 Apr 22].  Back to cited text no. 35
    
36.
Kalaany NY, Sabatini DM. Tumours with PI3K activation are resistant to dietary restriction. Nature 2009;458:725-31.  Back to cited text no. 36
    
37.
Wang XY, Cheng Y, Du JP, Chen DG, Yang XS, Yang ZM, et al. Synergistic effect of targeted inhibition of MEK/ERK and PI3K/AKT survival signaling pathways on induction of apoptosis in melanoma. Sichuan Da Xue Xue Bao Yi Xue Ban 2014;45:355-61.  Back to cited text no. 37
[PUBMED]    
38.
Murillo-Maldonado JM, Zeineddine FB, Stock R, Thackeray J, Riesgo-Escovar JR. Insulin receptor-mediated signaling via phospholipase C-gamma regulates growth and differentiation in Drosophila. PLoS One 2011;6:e28067.  Back to cited text no. 38
    
39.
SRC SRC proto-oncogene, non-receptor tyrosine kinase [ Homo sapiens (human) ]. Gene Database updated on 9-Nov-2014. Available from:http://www.ncbi.nlm.nih.gov/gene/6714.  Back to cited text no. 39
    
40.
Girotti MR, Pedersen M, Sanchez-Laorden B, Viros A, Turajlic S, Niculescu-Duvaz D, et al. Inhibiting EGF receptor or SRC family kinase signaling overcomes BRAF inhibitor resistance in melanoma. Cancer Discov 2013;3:158-67.  Back to cited text no. 40
    
41.
Tong L, Warren TC, King J, Betageri R, Rose J, Jakes S. Crystal structures of the human p56lck SH2 domain in complex with two short phosphotyrosyl peptides at 1.0 A and 1.8 A resolution. J Mol Biol 1996;256:601-10.  Back to cited text no. 41
    
42.
Russell RB, Breed J, Barton GJ. Conservation analysis and structure prediction of the SH2 family of phosphotyrosine binding domains. FEBS Lett 1992;304:15-20.  Back to cited text no. 42
    
43.
Koytiger G, Kaushansky A, Gordus A, Rush J, Sorger PK, MacBeath G. Phosphotyrosine signaling proteins that drive oncogenesis tend to be highly interconnected. Mol Cell Proteomics 2013;12:1204-13.  Back to cited text no. 43
    
44.
Pasini L, Turco MY, Luzi L, Aladowicz E, Fagiani E, Lanfrancone L. Melanoma: Targeting signaling pathways and RaLP. Expert Opin Ther Targets 2009;13:93-104.  Back to cited text no. 44
    
45.
Mashima R, Hishida Y, Tezuka T, Yamanashi Y. The roles of Dok family adapters in immunoreceptor signaling. Immunol Rev 2009;232:273-85.  Back to cited text no. 45
    
46.
Crowder RJ, Enomoto H, Yang M, Johnson EM Jr, Milbrandt J. Dok-6, a Novel p62 Dok family member, promotes Ret-mediated neurite outgrowth. J Biol Chem 2004;279:42072-81.  Back to cited text no. 46
    
47.
Woodring PJ, Meisenhelder J, Johnson SA, Zhou GL, Field J, Shah K, et al. c-Abl phosphorylates Dok1 to promote filopodia during cell spreading. J Cell Biol 2004;165:493-503.  Back to cited text no. 47
    
48.
Ghanem T, Bracken J, Kasem A, Jiang WG, Mokbel K. mRNA expression of DOK1-6 in human breast cancer. World J Clin Oncol 2014;5:156-63.  Back to cited text no. 48
[PUBMED]    
49.
Kim JT, Liu C, Zaytseva YY, Weiss HL, Townsend CM Jr, Evers BM. Neurotensin, a novel target of Wnt/ß-catenin pathway, promotes growth of neuroendocrine tumor cells. Int J Cancer 2014.  Back to cited text no. 49
    
50.
Gheidari F, Bakhshandeh B, Teimoori-Toolabi L, Mehrtash A, Ghadir M, Zeinali S. TCF4 silencing sensitizes the colon cancer cell line to oxaliplatin as a common chemotherapeutic drug. Anticancer Drugs 2014;25:908-16.  Back to cited text no. 50
    
51.
Eichhoff OM, Weeraratna A, Zipser MC, Denat L, Widmer DS, Xu M, et al. Differential LEF1 and TCF4 expression is involved in melanoma cell phenotype switching. Pigment Cell Melanoma Res 2011;24:631-42.  Back to cited text no. 51
    
52.
Carvajal RD, Antonescu CR, Wolchok JD, Chapman PB, Roman RA, Teitcher J, et al. KIT as a therapeutic target in metastatic melanoma. JAMA 2011 8;305:2327-34.  Back to cited text no. 52
    
53.
Leiter U, Meier F, Garbe C. Targeted therapies for melanoma. Hautarzt 2014;65:600-6.  Back to cited text no. 53
    
54.
Laugesen, I.G, et al., Valrubicin activates PKCa in keratinocytes: A conceivable mode of action in treating hyper-proliferative skin diseases. J Drugs Dermatol, 2013. 12: p. 1156-62.  Back to cited text no. 54
    
55.
Dai J, et al., Large-scale analysis of PDGFRA mutations in melanomas and evaluation of their sensitivity to tyrosine kinase inhibitors imatinib and crenolanib. Clin Cancer Res, 2013. 19: p. 6935-42.  Back to cited text no. 55
    


    Figures

  [Figure 1], [Figure 2], [Figure 3], [Figure 4], [Figure 5], [Figure 6], [Figure 7]
 
 
    Tables

  [Table 1], [Table 2]



 

Top
 
 
  Search
 
Similar in PUBMED
   Search Pubmed for
   Search in Google Scholar for
 Related articles
Access Statistics
Email Alert *
Add to My List *
* Registration required (free)

  >Abstract>Introduction>Materials and Me...>Results>Discussion>Conclusion>Acknowledgments>Article Figures>Article Tables
  In this article
>References

 Article Access Statistics
    Viewed2874    
    Printed84    
    Emailed0    
    PDF Downloaded151    
    Comments [Add]    

Recommend this journal


[TAG2]
[TAG3]
[TAG4]