January 01, 2018. doi:10.12123/npcd201801003
BIOWED January 01, 2018
Metrics: PDF views | HTML views
Dan Zhang, Jiaru Luo, Yunzhi Ling*
All authors are from Biowed (China) Co., Ltd.
Keywords: Uveal melanoma; GEO database; DEGs; Go analysis; KEGG analysis
Received: September 01,2017 Accept: November 01,2017 published:January 01,2018
Uveal melanoma (UVM) is the most common malignant ocular tumor among adults, with the characteristics of strong infiltration, prone to distant metastasis, and poor prognosis after metastasis. Up until now, many studies suggested the metastasis of uveal melanoma is associated with multiple factors, such as the abnormalities of chromosomes, gene expressions and signaling pathways. In this study, to further understand the underlying mechanism and molecular basis of the UVM metastasis, we selected the datasets of GSE27831 from GEO database, and made a comparative analysis between the 11 cases of UVM metastasis samples and 17 cases of non-metastasis samples using the bioinformatics methods. According to the results of differentially expressed gene screening: compared with the non-metastasis samples, which were considered as control group, 46 up-regulated genes and 323 down-regulated gene existed in the UVM metastasis group. The GO results showed system development (GO:0048731, 1.05E-10), cell periphery (GO:0071944, 9.26E-08), brain-derived neurotrophic factor binding (GO:0048403, 1.12E-09), etc. were significantly enriched, and the hierarchical tree of GO category suggested motor activity (GO: 0003774, 9.14e-4), calmodulin binding (GO: 0005516, 2.070e-4), and cis-trans isomerase activity (GO: 0016859, 8.921e-5), etc., may associated with the UVM metastasis process. Meanwhile, the Kegg analysis showed that 369 DEGs were involved in 19 signaling pathways, including protein digestion and absorption, focal adhesion, ECM-receptor interaction, PI3K-Akt signaling pathway, p53 signaling pathway, etc. In addition, the network of protein- protein interaction indicated 129 nodes existed and multiple hub proteins were identified, including POMC, HIST2H2BE, CABP7, CACNA1A, NGFR, etc. Throughout the study, we were able to preliminarily explore the changes of gene expression profile of metastatic UVM, and hopefully we can generate more accurate UVM transcriptome profiles based on large sample size, and investigates some potential biomarkers and therapy targets for metastatic UVM in the future.
Uveal melanoma (UVM) is the most common primary malignant tumor in adult eyes, predominantly found in Caucasians(1). The UVM morbidity is the second most prevalent ocular malignant tumor in China. UVM mainly originates from the pigment cells and nevi cells of uveal tissues and is associated with a high incidence of local infiltration and distant metastasis, key characteristics of this disease(2). Among that, local infiltration can occur outside the eye, and the distant transfer usually occurs in liver, lungs, kidneys, brain tissue, etc. Statistical data indicates that in clinical, the majority of the UVM patients are at risk of distant metastasis, especially in the common site of liver(3). In addition, for the UVM patients, the prognosis is very poor once metastasis is diagnosed, and patients would die within several months (median survival time varies from <4-12.5 months), and despite the improvements in diagnosis and treatments of the primary tumor, no effective treatment for metastatic UVM has been found(4,5).
Different from the skin melanoma, due to the lack of lymphatic vessels in eyes, the metastasis of UVM is mainly through bloodstream, which results in the patients not able to spot the early metastasis in time, and among some of the patients, the UVM metastasis had already happened before the diagnose(6). However, the invasion and metastasis of tumor tissue is a multi-cascade reaction process, including the abnormal tumor cells proliferation, morphological changes, invasion of the basement membrane, transferred to the blood vessels, the blood transmission, and the formation of metastases in the target sites(7). Recently, more and more studies show the process of UVM metastasis is closely associated with a variety of factors such as chromosomal abnormalities, gene mutations, cytokines, miRNAs and signal transduction pathway abnormalities, etc.(8). Among these factors, studies of UVM cytogenetics indicate that the abnormalities mainly occur in Chromosome 1, 3, 6, 8, 11, 13(9-12). The loss of chromosome 3 is the most important risk factor for UVM metastasis, and the theory is widely accepted. Meanwhile, the hypothesis, which states that the loss of chromosome 3 increases the incidence of UVM metastasis and mortality rate, has been proven in several studies(13,14). Moreover, metastasis related genes, such as Bcl-2, MDM-2, p53, GNAQ, GNA11, BRAF, BAP1, c-Kit, c-Met, PTP4A3, etc., is also an important study focus of UVM(15,16). For example, in Pópuloħ 's study, the vivo experiments of mouse showed GNAQ mutation could promote the metastasis of UVM(17); Laurent's study found PTP4A3 overexpression in uveal melanoma cell lines significantly increased cell migration and invasiveness in vivo, suggesting a direct role for this protein in metastasis(18). Hence, to further evaluate the roles of metastasis related genes is important to explore the underlying mechanism in UVM.
Microarray, which is a highly efficient technology and large-scale access to biological information, can detect and analyze the differentially expressed genes between normal and tumor tissues. In this study, to reveal the mechanism of pathogenesis and metastasis from the genome level and provide new targets and marker for UVM treatment, we performed the data excavation of UVM metastasis related genes and make a bioinformatics analysis using the Gene Expression Omnibus (GEO) database.
Materials and Methods
In this study, we selected the GSE27831 datasets from NCBI GEO database and screened the differentially expressed genes (DEGs), which are related with the metastasis of UVM. In the GSE27831 datasets, researchers collected 29 UVM samples, and conducted the miroarray experiments by using the Affymetrix Human Genome U133 Plus 2.0 Array.
Evaluation of Data Quality
In the miroarray experiments, not all of the experiments would be successful, and many factors can cause failure. Among the multiple influential factors, the main causes are possibly idue to technology, such as the quality of chip-self, experiments design, or the samples itsself existed degradation, so the quality evaluation is an important operation before the followed up array analysis.
To ensure the reliability of the array data analyzed in the study, we used R software and Bioconductor to process, analyze, annotate, and visualize the CEL raw file data provided in the GEO datasets. For the quality evaluation of Arrays, we used the methods of average and advanced data fit, and combined with affyPLM, simpleaffy R packages to conduct; meanwhile, affy, CLL R packages and AffyRNAdeg function were used to analyzed the sample quality, and the results were visualized via RNA degradation curves.
According to the quality evaluation of the chip data, the unqualified samples were excluded. Then the sample data included in the analysis usually need to subject three steps including background correction, standardization and aggregation to obtain the gene expression matrix for the next differential gene expression analysis. In this study, we used the RMA integration algorithm to preprocess the chip data and obtain a gene expression matrix.
Samples of UVM non-metastasis group and metastasis group were considered as control and experiment group respectively. Using t test, R software and limma package were applied to calculate differently expressed probe sets between control and experiment group. During the screening process, the genes with P<0.05 and |log2(Fold change)|>1 were selected as the significantly differentially expressed genes (DEGs). A heat map analysis was conducted using the "pheatmap" function of R/Bioconductor package "ggplot"(19).
GO and KEGG Analysis
The Gene ontology (GO) was analyzed via the online Gene Ontology Enrichment Analysis Software Toolkit (GOEAST) (http://omicslab.genetics.ac.cn/GOEAST/) to facilitate the interpretation of biological roles of DEGs (20), and the GO functions were performed according to three categories including biological process, molecular function, and cellular components. In addition, the pathway enrichment of DEGs was analyzed using the online tool of KOBAS (http://kobas.cbi.pku.edu.cn/annotate.php).
Protein Protein Interaction (PPI) Network Construction
In order to find candidate genes involved in the process of UVM metastasis, PPI network of DEGs were constructed according to the data from STRING database (https://string-db.org/). Furtherly, the PPI network of DEGs were visualized via Cytoscape(21).
In this study, before the DEGs screening, to ensure the reliability of the array data, we divided the 29 samples of the GSE27831 datasets into metastasis group and non-metastasis group based on the clinical information of Rosaria's study, and made a quality evaluation of each chip experiment. Among that, there were 11 samples of metastasis and 18 samples of non-metastasis. As shown in Figure. 1, the normalized unscaled standard errors (NUSE) box showed the NUSE values of each sample in metastasis group distributed around 1, which indicated the quality of microarrays were good (Figure.1A); however, in the non-metastasis group, the NUSE value of MU16.CEL sample appeared an offset, which exceeded 1 and suggested a problem may exist in this microarray experiments (Figure.1C). Meanwhile, we also made a fit for the data of microarray CEL files, and we found that there was a serious RNA degradation in the sample of MU16.CEL, while the RNA quality of all other samples were well (Figure1B, 1D). Based the analysis above, we believed a serious problem of RNA quality exists in the MU16.CEL sample, and this sample should be excluded in the subsequent analysis. And the clinical data of the 28 included samples are shown in details in Table.1.
DEGs Screening and Cluster Analysis
Finally, 11 datasets of UVM metastasis samples including GSM685472_MU9, GSM685473_MU10, GSM685523_MU15, GSM685601_MU_3, GSM685602_MU8, GSM685603_MU4, GSM685652_MU_7, GSM686985_MU_31, GSM686988_MU_34, GSM686989_MU_36, GSM687003_MU21 and 17 datasets of UVM non-metastasis samples including GSM685471_MU1, GSM685474_MU11, GSM685475_MU12, GSM685522_MU13, GSM685650_MU5, GSM685651_MU6, GSM686961_MU_17, GSM686962_MU_22, GSM686963_MU_25, GSM686984_MU_30, GSM686986_MU_33, GSM686986_MU_33, GSM686990_MU_40, GSM686991_MU18, GSM687001_MU_2, GSM687002_MU20, GSM687004_MU23 were included in this study. The results of the screening DEGs analysis showed that comparing with the non-metastasis UVM samples, our control group, 369 genes expressions had significant changes in UVM metastasis group, including 46 up-regulated genes and 323 down-regulated genes. Meanwhile, the DEGs was analyzed by the two-way cluster, as shown in Figure.2, there was a difference in the expression of the same gene in cancerous tissues of different samples. The significantly top 10 down-regulated and up-regulated genes are listed in Table.2.
Figure.1 The quality evaluation of GSE27831 datasets (including the metastasis and non-metastasis UVM samples),
which selected from GEO database. (A, C) the affyPLM R package was used to make a fitting and regression for the
Chip raw data, and the results were visualized through normalized unscaled standard errors (NUSE) box to evaluate
the quality of microarray itself; (B, D) the RNA degradation plots of the samples in metastasis group and
non-metastasis group; (E) overview plot of microarrays quality control.
Table.1 Clinicopathologic characteristics of 28 uveal melanoma patients included in this study.
Notes: compared with non-metastasis patients, ***P<0.001, and "n.s" represents P>0.05.
Table.2 Most obviously dysregulated genes sorted by P value in metastatic uveal melanoma compare to non-metastatic tumors.
Figure.2 Heat-map image of the significantly DEGs in UVM samples of metastasis and non-metastasis group. Expression data are described as a data matrix in which each row represents a gene and each column represents a sample. Expression level are described according to the color scale shown at the topright. Red and green indicate high and low expression levels, respectively.
GO enrichment and KEGG enrichment analysis
System development (GO:0048731, 1.05E-10), single-organism developmental process (GO:0044767, 2.05E-10), anatomical structure development (GO:0048856, 2.05E-10) and developmental process (GO:0032502, 4.73E-10) were significantly enriched upon the category of GO biological process; cell periphery (GO:0071944, 9.26E-08), extracellular region (GO:0005576, 9.49E-08), presynaptic active zone (GO:0048786, 1.65E-07) and plasma membrane (GO:0005886, 2.08E-07) were significantly enriched upon the category of GO cellular component; while for the category of GO molecular function, brain-derived neurotrophic factor binding (GO:0048403, 1.12E-09), protein tyrosine kinase activator activity (GO:0030296, 6.10E-08), neurotrophin receptor activity (GO:0005030, 7.41E-07) and neurotrophin binding (GO:0043121, 5.88E-06) were notably enriched. Meanwhile, the hierarchical tree of each GO category (biological process, molecular function and cellular component) shows motor activity (GO: 0003774, 9.14e-4), calmodulin binding (GO: 0005516, 2.070e-4), peptidyl-prolyl cis-trans isomerase activity (GO: 0003755, 8.921e-5) and cis-trans isomerase activity (GO: 0016859, 8.921e-5) may related with the metastasis of UVM (Figure.3).
In addition, the results of KEGG analysis showed: the 369 DEGs were involved in 19 signal pathways, including protein digestion and absorption, focal adhesion, ECM-receptor interaction, Axon guidance, Axon guidance, Cysteine and methionine metabolism, cAMP signaling pathway, PI3K-Akt signaling pathway, p53 signaling pathway, etc. (Table.3)
Figure.3 GO enrichments. The enrichment was conduct based on the P value of each category, and the enrichment score
equates -log10(P-value). Top 10 significantly enriched categories of GO biological process (A), GO cellular component (B),
and GO molecular function (C); (D) the hierarchical tree of each GO category.
PPI Network of DEGs
The PPI network of significantly DEGs consisted 129 nodes. And multiple hub proteins were identified in this network, including POMC, HIST2H2BE, ACTC1, KIT, ADCY8, CABP7, CACNA1A, NGFR, etc. (Figure.4).
Table.3 Kegg enrichment of DEGs: top 15 terms with high enrichment scores.
Figure.4 PPI network of DEGs. The size of the nodes and the width of the edges show their significance (larger indicates lower P value).
For the uveal melanoma, once it is spread to distant organs, the disease will be largely resistant to currently available therapies(22). Nowadays, It is generally accepted that the altered gene expression pattern of a cancer tissue should be associated with the initiation and progress of malignant phenotype. In our study, we selected the GSE27831 datasets from NCBI GEO database for analysis, attempting to reveal which genes expression significantly changed in the process of UVM metastasis from the genome level, and which biological processes and signal pathways these dysregulated genes participated, in order to provided a reliable theoretical basis for the further study of mechanism in UVM metastasis. The present study followed microarray-based 28 uveal melanoma patients, who are not significantly different in ages, gender, primary tumor thickness, tumor largest diameter, but are significantly different in the disease-free survival (DFS). The screening results of differentially expressed genes (DEGs) showed that 369 DEGs exists in the metastatic UVM comparing with non-metastatic UVM, which is involved in 278 categories of GO biological process, 47 categories of GO cellular component, 27 categories of GO molecular functions, 19 signal pathways. At the same time, the analysis of PPI network indicates the existence of 129 nodes and multiple hub proteins, including HIST2H2BE, KIT, NGFR, etc.
In bioinformatics, the gene ontology is widely considered as the primary tool for organization and functional annotation of the molecular aspects of cellular systems(23). In this study, after DEGs screening, we conducted a GO analysis and the hierarchical tree of each GO category (biological process, molecular function and cellular component) showed the motor activity (GO: 0003774, 9.14e-4), calmodulin binding (GO: 0005516, 2.070e-4), peptidyl-prolyl cis-trans isomerase activity (GO: 0003755, 8.921e-5) and cis-trans isomerase activity (GO: 0016859, 8.921e-5) may be related with the metastasis of UVM. Among those, for the GO category of calmodulin binding (GO: 0005516, 2.070e-4), Van et al(24) used Chromatographic procedures and amino acid sequence analysis to identify the series of calcium-binding proteins d in both primary tumors and cell lines of uveal melanoma and found calcium-binding proteins may endow tumor cells with properties related to their malignancy and metastatic phenotype; and Wagner et al(25) suggests both normal and neoplastic uveal melanocytes require an intracellular signal or signals which involves calcium and calmodulin in the few minutes following cell binding to ECM proteins in order for successful cell attachment to occur. For the peptidyl-prolyl cis-trans isomerase activity, studies showed multiple proteins which have peptidyl-prolyl cis-trans isomerase activity play an important role in protein folding, transportation, signal transduction, inflammation, immune regulation, apoptosis and other biological processes. As is well known, the protein phosphorylation at Ser/Thr-Pro site is an important step of intracellular signal transduction. Recent studies found the protein phosphorylation at Ser/Thr-Pro site is just the first step for the changing of protein structure, many proteins must be getting the second change in structure, after which the corresponding protein could exert its function normally, and this structurual change is carried out under the regulation of peptidyl-prolyl cis-trans isomerase (PPIase)(26,27). For example, the protein of Pin, which contains two functional structures including phosphorylation region and peptidyl-prolyl cis-trans isomerase activity region, participates the regulation of multiple signal pathways and could catalyze many protein like p53, β-catenin, cyclinD1, etc., which are closely associated with the occurrence and progression of tumor(28,29).
Kegg pathway can find the significant signaling pathways that DEGs participate in, and provide a comprehensive understanding about interactions of genes and relations between up and down stream. In this study, the kegg analysis showed protein digestion and absorption, focal adhesion, ECM-receptor interaction, PI3K-Akt signaling pathway, p53 signaling pathway, etc., were enriched. In previous studies, numerous studies have proven PI3K-Akt signaling pathway participated in several cancers' metastasis(30-32); meanwhile, it was reported that the pathways of focal adhesion, ECM-receptor interaction were also associated with the invasion of cancer cells. In addition, in the PPI network of DEGs, compared with the non-metastatic UVM samples, the expression of hub protein KIT up-regulated 2.82-fold, and the expression of hub proteins NGFR, HIST2H-2BE down-regulated 2.05-fold and 2.11-fold, respectively. Among these proteins, in 2004, All-Ericsson et al(16) reported that c-kit was vastly expressed in uveal melanoma, and the c-kit molecular pathway may be important in uveal melanoma growth, and pointed to its use as a target for therapy with STI571; in Calipel's study(33), in UVM patients, Ninety-five percent of liver metastases expressed KIT at the protein level, which suggested KIT may play an important role in the metastasis process; however, in Lüke's study(34), it was proved that c-Kit expression was not found to be associated with metastasis formation. Hence, it is necessary to further study if the hub protein KIT is really associated with the metastasis of UVM.
All above results suggest that there are differences in gene expression between the metastatic and non-metastatic uveal melanoma. The proteins encoded by these genes involved in multiple GO categories and signal pathways, the dysregulation of which may contribute to the UVM metastasis. Despite a similar study conducted by Zhang et al(35), a quality evaluation was not performed and all samples of GSE27831 datasets were included in their study, which may cause an error. In this study, based on the accurate quality evaluation and analysis, the identified DEGs, the related GO terms and signal pathways that DEGs enriched here provide an important theoretical basis for clinical investigation, meanwhile, all these results also need to be further studied and confirmed in a larger sample size containing more patients by other clinic-related studies.
Bioinformatics can excavate and analyze large amounts of data in microarrays by the methods of rigorous experimental planning, scientific statistical analysis and collection of completed data. In our study, we evaluated the quality of GSE27831 datasets, analyzed the DEGs screened from non-metastatic and metastatic UVM samples using bioinformatics, and preliminary provided partial new targets for diagnosis and theoretical basis for studying the mechanism of uveal melanoma metastasis.
All authors sincerely acknowledge the support given by Biowed (China) Co., Ltd and Guangzhou Algae Technology Information Consuitant Co., Ltd.
1. S. Kaliki, C. L. Shields, and J. A. Shields. Uveal melanoma: estimating prognosis. Indian J. Ophthalmol. (2015) 63(2): 93-102.
2. J. W. Ling, P. R. Lu, Y. B. Zhang, S. Jiang, and Z. C. Zhang. miR-367 promotes uveal melanoma cell proliferation and migration by regulating PTEN. Genet Mol. Res. (2017) 16(3): gmr16039067.
3. A. M. Abbott, M. P. Doepker, Y. Kim, M. C. Perez, C. Gandle, K. L. Thomas et al. Hepatic Progression-free and Overall Survival After Regional Therapy to the Liver for Metastatic Melanoma. Am. J. Clin. Oncol. (2017).
4. A. D. Singh, L. Bergman, and S. Seregard. Uveal melanoma: epidemiologic aspects. Ophthalmol. Clin. North Am. (2005) 18(1): 75-84, viii.
5. T. Aoyama, M. J. Mastrangelo, D. Berd, F. E. Nathan, C. L. Shields, J. A. Shields et al. Protracted survival after resection of metastatic uveal melanoma. Cancer (2000) 89(7): 1561-1568.
6. A. D. Singh, I. G. Rennie, T. Kivela, S. Seregard, and H. Grossniklaus. The Zimmerman-McLean-Foster hypothesis: 25 years later. Br. J. Ophthalmol. (2004) 88(7): 962-967.
7. G. P. Gupta and J. Massague. Cancer metastasis: building a framework. Cell (2006) 127(4): 679-695.
8. M. Ye, D. Hu, L. Tu, X. Zhou, F. Lu, B. Wen et al. Involvement of PI3K/Akt signaling pathway in hepatocyte growth factor-induced migration of uveal melanoma cells. Invest. Ophthalmol. Vis. Sci. (2008) 49(2): 497-504.
9. K. Sisley, I. G. Rennie, D. W. Cottam, A. M. Potter, C. W. Potter, and R. C. Rees. Cytogenetic findings in six posterior uveal melanomas: involvement of chromosomes 3, 6, and 8. Genes Chromosomes Cancer (1990) 2(3): 205-209.
10. K. Sisley, I. G. Rennie, M. A. Parsons, R. Jacques, D. W. Hammond, S. M. Bell et al. Abnormalities of chromosomes 3 and 8 in posterior uveal melanoma correlate with prognosis. Genes Chromosomes Cancer (1997) 19(1): 22-28.
11. K. Sisley, M. A. Parsons, J. Garnham, A. M. Potter, D. Curtis, R. C. Rees et al. Association of specific chromosome alterations with tumour phenotype in posterior uveal melanoma. Br. J. Cancer (2000) 82(2): 330-338.
12. M. D. Onken, L. A. Worley, J. P. Ehlers, and J. W. Harbour. Gene expression profiling in uveal melanoma reveals two molecular classes and predicts metastatic death. Cancer Res. (2004) 64(20): 7205-7209.
13. B. Damato, J. A. Dopierala, and S. E. Coupland. Genotypic profiling of 452 choroidal melanomas with multiplex ligation-dependent probe amplification. Clin. Cancer. Res. (2010) 16(24): 6083-6092.
14. C. L. Shields, A. Ganguly, C. G. Bianciotto, K. Turaka, A. Tavallali, and J. A. Shields. Prognosis of uveal melanoma in 500 cases using genetic testing of fine-needle aspiration biopsy specimens. Ophthalmology (2011) 118(2): 396-401.
15. S. L. Scholz, I. Moller, H. Reis, D. Susskind, J. A. P. van de Nes, S. Leonardelli et al. Frequent GNAQ, GNA11, and EIF1AX Mutations in Iris Melanoma. Invest. Ophthalmol. Vis. Sci. (2017) 58(9): 3464-3470.
16. C. All-Ericsson, L. Girnita, A. Muller-Brunotte, B. Brodin, S. Seregard, A. Ostman et al. c-Kit-dependent growth of uveal melanoma cells: a potential therapeutic target? Invest. Ophthalmol. Vis. Sci. (2004) 45(7): 2075-2082.
17. H. Populo, J. Vinagre, J. M. Lopes, and P. Soares. Analysis of GNAQ mutations, proliferation and MAPK pathway activation in uveal melanomas. Br. J. Ophthalmol. (2011) 95(5): 715-719.
18. C. Laurent, F. Valet, N. Planque, L. Silveri, S. Maacha, O. Anezo et al. High PTP4A3 phosphatase expression correlates with metastatic risk in uveal melanoma patients. Cancer Res. (2011) 71(3): 666-674.
19. M. Reimers and V. J. Carey. Bioconductor: an open source framework for bioinformatics and computational biology. Methods Enzymol. (2006) 411: 119-134.
20. Q. Zheng and X. J. Wang. GOEAST: a web-based software toolkit for Gene Ontology enrichment analysis. Nucleic Acids Res. (2008) 36(Web Server issue): W358-363.
21. P. Shannon, A. Markiel, O. Ozier, N. S. Baliga, J. T. Wang, D. Ramage et al. Cytoscape: a software environment for integrated models of biomolecular interaction networks. Genome Res. (2003) 13(11): 2498-2504.
22. J. J. Augsburger, Z. M. Correa, and A. H. Shaikh. Effectiveness of treatments for metastatic uveal melanoma. Am. J. Ophthalmol. (2009) 148(1): 119-127.
23. R. C. Lovering, E. B. Camon, J. A. Blake, and A. D. Diehl. Access to immunology through the Gene Ontology. Immunology (2008) 125(2): 154-160.
24. P. R. Van Ginkel, R. L. Gee, T. M. Walker, D. N. Hu, C. W. Heizmann, and A. S. Polans. The identification and differential expression of calcium-binding proteins associated with ocular melanoma. Biochim. Biophys. Acta (1998) 1448(2): 290-297.
25. M. Wagner, S. Bielby, I. G. Rennie, and S. Mac Neil. Attachment of human uveal melanocytes and melanoma cells to extracellular matrix proteins involves intracellular calcium and calmodulin. Melanoma Res. (1997) 7(6): 439-448.
26. P. Blume-Jensen and T. Hunter. Oncogenic kinase signalling. Nature (2001) 411(6835): 355-365.
27. T. Hunter. Prolyl isomerases and nuclear function. Cell (1998) 92(2): 141-143.
28. Y. C. Liou, A. Ryo, H. K. Huang, P. J. Lu, R. Bronson, F. Fujimori et al. Loss of Pin1 function in the mouse causes phenotypes resembling cyclin D1-null phenotypes. Proc. Natl. Acad. Sci. U. S. A. (2002) 99(3): 1335-1340.
29. A. Ryo, M. Nakamura, G. Wulf, Y. C. Liou, and K. P. Lu. Pin1 regulates turnover and subcellular localization of beta-catenin by inhibiting its interaction with APC. Nat. Cell Biol. (2001) 3(9): 793-801.
30. F. D. Lublin. MS as a gateway disease. J. Neurol. Sci. (2013) 333(1-2): 73-75.
31. R. A. Grana. Electronic cigarettes: a new nicotine gateway? J. Adolesc. Health (2013) 52(2): 135-136.
32. S. Shukla, S. Chandran, and R. Gadagkar. Ovarian developmental variation in the primitively eusocial wasp Ropalidia marginata suggests a gateway to worker ontogeny and the evolution of sociality. J. Exp. Biol. (2013) 216(Pt 2): 181-187.
33. A. Calipel, S. Landreville, A. De La Fouchardiere, F. Mascarelli, M. Rivoire, N. Penel et al. Mechanisms of resistance to imatinib mesylate in KIT-positive metastatic uveal melanoma. Clin. Exp. Metastasis (2014) 31(5): 553-564.
34. J. Luke, J. Wegner, R. Wegner, K. Nassar, O. Tatar, J. M. Rohrbach et al. Expression of c-Kit and its ligand SCF in primary uveal melanoma. Eur. J. Ophthalmol. (2011) 21(5): 615-624.
35. Y. Zhang, Y. Yang, L. Chen, and J. Zhang. Expression analysis of genes and pathways associated with liver metastases of the uveal melanoma. BMC Med. Genet. (2014) 15: 29.