This article has Open Peer Review reports available.
Meta-analysis of differentially expressed genes in osteosarcoma based on gene expression data
- Zuozhang Yang†1Email author,
- Yongbin Chen†2,
- Yu Fu†3,
- Yihao Yang†1,
- Ya Zhang†1,
- Yanjin Chen1 and
- Dongqi Li1
© Yang et al.; licensee BioMed Central Ltd. 2014
Received: 27 May 2014
Accepted: 30 June 2014
Published: 14 July 2014
To uncover the genes involved in the development of osteosarcoma (OS), we performed a meta-analysis of OS microarray data to identify differentially expressed genes (DEGs) and biological functions associated with gene expression changes between OS and normal control (NC) tissues.
We used publicly available GEO datasets of OS to perform a meta-analysis. We performed Gene Ontology (GO) enrichment analysis, Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis and Protein-Protein interaction (PPI) networks analysis.
Eight GEO datasets, including 240 samples of OS and 35 samples of controls, were available for the meta-analysis. We identified 979 DEGs across the studies between OS and NC tissues (472 up-regulated and 507 down-regulated). We found GO terms for molecular functions significantly enriched in protein binding (GO: 0005515, P = 3.83E-60) and calcium ion binding (GO: 0005509, P = 3.79E-13), while for biological processes, the enriched GO terms were cell adhesion (GO:0007155, P = 2.26E-19) and negative regulation of apoptotic process (GO: 0043066, P = 3.24E-15), and for cellular component, the enriched GO terms were cytoplasm (GO: 0005737, P = 9.18E-63) and extracellular region (GO: 0005576, P = 2.28E-47). The most significant pathway in our KEGG analysis was Focal adhesion (P = 5.70E-15). Furthermore, ECM-receptor interaction (P = 1.27E-13) and Cell cycle (P = 4.53E-11) are found to be highly enriched. PPI network analysis indicated that the significant hub proteins containing PTBP2 (Degree = 33), RGS4 (Degree = 15) and FXYD6 (Degree = 13).
Our meta-analysis detected DEGs and biological functions associated with gene expression changes between OS and NC tissues, guiding further identification and treatment for OS.
Osteosarcoma (OS), the most common non-haematological primary malignant tumor of bone, occurs most commonly in the metaphyseal regions of long bones mainly in adolescents and young adults, but also in patients over 40 years of age . Though the survival rate has been improved after the introduction of neoadjuvant chemotherapy, the need for advances in treatments is still very urgent [2, 3]. Therefore, an in-depth understanding of the pathobiology of OS is needed to develop rational treatment options for OS. Cytogenetic analyses have revealed that most conventional OShave complex karyotypes with numerous and highly variable genomic aberrations . Many genes become dysregulated due to genomic aberrations, and DNA copy number and DNA methylationand and gene expression data combined to identify oncogenes and tumor suppressor genes in OS [5, 6].
As the high-throughput technologies have been used in many fields, detection of expression level across the whole genome is a useful way to find unusual genomic alteration in OS patients with microarray. Recently, researchers have used this technique to more comprehensively increase knowledge about the cellular and molecular changes in OS [7–13]. Although these studies have shown lists of differently expressed genes (DEGs), there tends to be inconsistencies among studies due to limitations of small sample sizes and varying results obtained by different groups, accomplished by different laboratory protocols, microarray platforms and analysis techniques . Recent studies have shown that the systematic integration of gene expression data from multiple sources, so-called meta-analyses, can increase statistical power for detecting differentially expressed genes while allowing for an assessment of heterogeneity, and may lead to more robust, reproducible and accurate predictions [15, 16]. Similar meta-analysis has never been conducted for OS, and we first perform a meta-analysis of gene expression data sets from various OS studies to overcome the limitations of individual studies, resolve inconsistencies, and reduce the likelihood that random errors are responsible for false-positive or false-negative associations, and lay a foundation for uncovering the pathology of OS and further generating new therapies for OS.
Identification of eligible OS gene expression datasets
Where xi represents raw intensity data for each gene; represents average gene intensity within a single experiment and δ represents standard deviation (SD) of all measured intensities.
The significance analysis of microarray (SAM) software was then used to identify the DEGs between pathological and control samples. This procedure identifies DEGs by carrying out gene specific t-statistics, with a “relative difference” score for each gene. The D value was defined as the average expression change from different expression states to the standard deviation of measurements for that gene. Genes exhibiting at least two-fold changes corresponding to a false discovery rate (FDR) less than 0.05 were selected as the significantly DEGs .
Functional classification of DEGs
In order to interpret the biological significance of the DEGs, we performed Gene Ontology (GO) enrichment analysis to investigate their functional distribution in OS. The online based software GENECODIS (http://genecodis.cnb.csic.es) was used to perform this analysis . In addition, we also performed the pathway enrichment analysis based on the Kyoto Encyclopedia of Genes and Genomes (KEGG) database.
PPI network construction
The protein-protein interactions (PPIs) research could reveal the functions of proteins at the molecular level and help discover the rules of cellular activities including growth, development, metabolism, differentiation and apoptosis . The identification of protein interact ions in a genome-wide scale is an important step for the interpretation of the cellular control mechanisms . In this analysis, we used Biological General Repository for Interaction Datasets (BioGRID) (http://thebiogrid.org/) to construct PPI network and visualized the distribution characteristics of the top 10 up- and down-regulated DEGs in the network in Cytoscape .
Short overview of the studies included
Characteristics of the individual studies
Sample count (case:control)
GPL96 [HG-U133A] Affymetrix Human Genome U133A Array
GPL96 [HG-U133A] Affymetrix Human Genome U133A Array
GPL6244 [HuGene-1_0-st] Affymetrix Human Gene 1.0 ST Array
GPL6244 [HuGene-1_0-st] Affymetrix Human Gene 1.0 ST Array
GPL10295 Illumina human-6 v2.0 expression beadchip (using nuIDs as identifier)
in vivo/in vitro
GPL6102 Illumina human-6 v2.0 expression beadchip
GPL6947 Illumina HumanHT-12 V3.0 expression beadchip
GPL570 [HG-U133_Plus_2] Affymetrix Human Genome U133 Plus 2.0 Array
Detecting genes associated with OS
The top 10 most significantly up- or down-regulated DEGs
Official full name
Hes-related family bHLH transcription factor with YRPW motif 1
Palmitoyl-protein thioesterase 1
Polypyrimidine tract binding protein 2
Collectin sub-family member 12
FXYD domain containing ion transport regulator 6
Ras and Rab interactor 2
Ribosome production factor 1 homolog (S. cerevisiae)
Natriuretic peptide receptor 3
Leiomodin 1 (smooth muscle)
Growth arrest-specific 6
Retinol dehydrogenase 5 (11-cis/9-cis)
Aldehyde oxidase 1
LIM and senescent cell antigen-like domains 2
Latent transforming growth factor beta binding protein 2
Regulator of G-protein signaling 4
Phosphatidic acid phosphatase type 2B
Cardiotrophin-like cytokine factor 1
The top 15 enriched KEGG pathway of DEGs
Cytokine-cytokine receptor interaction
Complement and coagulation cascades
Pathways in cancer
Regulation of actin cytoskeleton
Protein digestion and absorption
Staphylococcus aureus infection
Systemic lupus erythematosus
Protein-Protein interaction (PPI) network construction
Osteosarcoma (OS) is an aggressive cancer demonstrating both high metastatic rate and chemotherapeutic resistance. A comprehensive analysis of the mechanism underlying OS development is of crucial importance for management policy. In this paper, we chose a meta-analysis approach that combines differently expressed genes (DEGs) from microarray datasets to highlight genes that were consistently expressed differentially with statistical significance, and performed GO enrichment analysis and KEGG pathway analysis, and construct the protein-protein interaction (PPI) networks.
We performed a meta-analysis using 8 publicly available GEO data sets to identify common biological mechanisms involved in the pathogenesis of OS. OS is a kind of rare tumor, where material from clinical samples is scarce, therefore data from both bone tissues and cell lines were included in our meta-analysis. In total, 979 genes across the studies were consistently expressed differentially in OS (472 up-regulated and 507 down-regulated). The up-regulated gene with the lowest P-value (P = 5.08E-15) was CPE (carboxypeptidase E), which is a carboxypeptidase that cleaves C-terminal amino acid residues and is involved in the biosynthesis of peptide hormones and neurotransmitters, including insulin , but at the present the role and association with OS have not yet been reported. In line with previous findings, We found that some genes have been closely related to the development of OS among the top ten up-regulated DEGs, such as HEY1, FXYD6, EFNA1. HEY1, one of target genes of NOTCH1, was reported to be up-regulated in OS from p53 mutant mice, suggesting that activation of Notch signaling contributes to the pathogenesis of OS . Another study also found that HEY1 and other downstream target genes of Notch signaling including HES1, NOTCH1 and NOTCH2, were elevated in canine osteosarcoma by gene expression microarray analysis and reverse transcriptase - quantitative PCR (RT-qPCR) . Olstad OK et al. applied directional tag PCR subtractive hybridization to construct a cDNA library generated from three different human osteosarcoma (OS) target cell lines (OHS, SaOS-2 and KPDXM), and identified FXYD6 was enriched in OS cell lines . EFNA1 was significantly elevated in OS samples by using genome-wide microarrays, and in vitro study on the functional role of EphA2 and EFNA1 showed that EFNA1 ligand binding induced increased tyrosine phosphorylation, receptor degradation and downstream mitogen-activated protein kinase (MAPK) activation .
The down-regulated DEGs with the lowest P-value (P = 1.86E-48) was NPR3 (natriuretic peptide receptor 3) that acts as a decoy/clearance receptor functioning to limit the effects of natriuretic peptides. NPR3, which is an important anabolic regulator of endochondral bone growth, is enriched in bone marrow-derived mesenchymal stem cells, and there is no relevant report to OS at present. In our meta-analysis Gas6 was identified to be one of the top ten down-regulated DEGs. In OS cell lines, rhGas6 could activate Axl to protect the tumor cells from apoptosis caused by serum starvation, and promote tumor cells’ migration and invasion in vitro .
In order to uncover the biological roles of the DEGs from OS, we performed a GO categories enrichment analysis. We found GO terms for molecular functions significantly enriched in protein binding and calcium ion binding, while for biological processes, the enriched GO terms were cell adhesion and negative regulation of apoptotic process, and for cellular component, the enriched GO terms were cytoplasm and extracellular region. To further evaluate the biological significance for the DEGs, we also performed the KEGG pathway enrichment analysis. Focal adhesion, ECM-receptor interaction and Cell cycle in our KEGG analysis were found to be highly enriched. Many signal transduction pathways involved in OS development were stimulated by bone morphogenetic proteins (BMPs), transforming growth factors (TGFs), Notch family proteins and Wnt family proteins, and components of each of these pathways have been implicated in OS. Interestingly, we noted that the most significant pathway in our KEGG analysis was Focal adhesion. Focal adhesions are associated with cell migration dynamics. However in the human cells focal adhesion would initially appear to be contradictory to their migratory phenotype. It has been proved previously that knockdown of paxillin in highly metastatic OS sub-lines M112 and 132 would inhibit cell migration .
Furthermore the results from PPI network analysis of the top 10 up-regulated and down-regulated DEGs indicated the significant hub proteins containing PTBP2, RGS4 and FXYD6. PTBP2, a member of PTB (polypyrimidine tract binding protein) family of RNA-binding proteins which plays a critical role in development through the regulation of post-transcriptional events, is expressed in the nervous system including the brain, the neural retina and the spinal cord and the intermediate mesoderm . PTBP2 regulates the generation of neuronal precursors in the embryonic brain by repressing adult-specific splicing , but the function involved in OS development has not been discovered. Our result of PPI suggested that PTBP2 may play an important role in the development.
The present study has some limitations. First, heterogeneity and confounding factors may have distorted the analysis. Clinical samples might be heterogeneous with respect to clinical activity, severity, or gender. Although we conducted global normalization for different data sets, the heterogeneity of various microarray platforms used in different studies can’t remove. Second, there are differences in gene expression between tissues such as bones, cell lines and lung that were not taken into account. However, our meta-analysis integrated data from different studies which may enable us to detect genes that we would otherwise have missed in an analysis. Despite these limitations, our discover have important implications for the molecular mechanisms of OS,but further experimental research is still need to confirm our study.
In conclusion, by this meta-analysis based on gene expression data of osteosarcoma we have shown the underlying molecular differences between NC tissues and osteosarcoma, including DEGs and their biological function which may contribute to the successful identification of therapeutic targets for osteosarcoma. Further functional studies may provide additional insights into the role of the differentially regulated genes in the pathophysiology of osteosarcoma.
This research was supported in part by grants (no. 81260322/H1606 and no.81372322/H1606) from the National Natural Science Foundation of China, a grant (no. 2012FB163) from the Natural Science Foundation of Yunnan Province, a grant (no. 2011FB201) from the Joint Special Funds for the Department of Science and Technology of Yunnan Province-Kunming Medical University, and a grant (no. D-201242) from the specialty fund of high-level talents medical personnel training of Yunnan province.
- Bielack S, Carrle D, Casali P: Osteosarcoma: ESMO clinical recommendations for diagnosis, treatment and follow-up. Ann Oncol. 2009, 20 (suppl 4): iv137-iv139.Google Scholar
- Grimer RJ: Surgical options for children with osteosarcoma. Lancet Oncol. 2005, 6 (2): 85-92.View ArticlePubMedGoogle Scholar
- Akiyama T, Dass CR, Choong PF: Novel therapeutic strategy for osteosarcoma targeting osteoclast differentiation, bone-resorbing activity, and apoptosis pathway. Mol Cancer Ther. 2008, 7 (11): 3461-3469.View ArticlePubMedGoogle Scholar
- Boehm A, Neff J, Squire J, Bayani J, Nelson M, Bridge J: Cytogenetic findings in 36 osteosarcoma specimens and a review of the literature. Fetal Pediatr Pathol. 2000, 19 (5): 359-376.View ArticleGoogle Scholar
- Yen C-C, Chen W-M, Chen T-H, Chen WY-K, Chen PC-H, Chiou H-J, Hung G-Y, Wu H-TH, Wei C-J, Shiau C-Y: Identification of chromosomal aberrations associated with disease progression and a novel 3q13. 31 deletion involving LSAMP gene in osteosarcoma. Int J Oncol. 2009, 35 (4): 775-788.PubMedGoogle Scholar
- Kresse SH, Ohnstad HO, Paulsen EB, Bjerkehagen B, Szuhai K, Serra M, Schaefer KL, Myklebost O, Meza‒Zepeda LA: LSAMP, a novel candidate tumor suppressor gene in human osteosarcomas, identified by array comparative genomic hybridization. Genes Chromosom Cancer. 2009, 48 (8): 679-693.View ArticlePubMedGoogle Scholar
- Paoloni M, Davis S, Lana S, Withrow S, Sangiorgi L, Picci P, Hewitt S, Triche T, Meltzer P, Khanna C: Canine tumor cross-species genomics uncovers targets linked to osteosarcoma progression. BMC Genomics. 2009, 10 (1): 625-View ArticlePubMedPubMed CentralGoogle Scholar
- Sadikovic B, Yoshimoto M, Chilton-MacNeill S, Thorner P, Squire JA, Zielenska M: Identification of interactive networks of gene expression associated with osteosarcoma oncogenesis by integrated molecular profiling. Hum Mol Genet. 2009, 18 (11): 1962-1975.View ArticlePubMedGoogle Scholar
- Sadikovic B, Yoshimoto M, Al-Romaih K, Maire G, Zielenska M, Squire JA: In vitro analysis of integrated global high-resolution DNA methylation profiling with genomic imbalance and gene expression in osteosarcoma. PLoS One. 2008, 3 (7): e2834-View ArticlePubMedPubMed CentralGoogle Scholar
- Kuijjer ML, Peterse EF, van den Akker BE, Briaire-de Bruijn IH, Serra M, Meza-Zepeda LA, Myklebost O, Hassan AB, Hogendoorn PC, Cleton-Jansen A-M: IR/IGF1R signaling as potential target for treatment of high-grade osteosarcoma. BMC Cancer. 2013, 13 (1): 245-View ArticlePubMedPubMed CentralGoogle Scholar
- Kresse SH, Rydbeck H, Skårn M, Namløs HM, Barragan-Polania AH, Cleton-Jansen A-M, Serra M, Liestøl K, Hogendoorn PC, Hovig E: Integrative analysis reveals relationships of genetic and epigenetic alterations in osteosarcoma. PLoS One. 2012, 7 (11): e48262-View ArticlePubMedPubMed CentralGoogle Scholar
- Both J, Wu T, Bras J, Schaap GR, Baas F, Hulsebos TJ: Identification of novel candidate oncogenes in chromosome Region 17p11. 2-p12 in human osteosarcoma. PLoS One. 2012, 7 (1): e30907-View ArticlePubMedPubMed CentralGoogle Scholar
- Ying M, Liu G, Shimada H, Ding W, May W, He Q, Adams G, Wu L: Human osteosarcoma CD49f − CD133+ cells: impaired in osteogenic fate while gain of tumorigenicity. Oncogene. 2013, 32 (36): 4252-4263.View ArticlePubMedGoogle Scholar
- Siddiqui AS, Delaney AD, Schnerch A, Griffith OL, Jones SJ, Marra MA: Sequence biases in large scale gene expression profiling data. Nucleic Acids Res. 2006, 34 (12): e83-e83.View ArticlePubMedPubMed CentralGoogle Scholar
- Feichtinger J, Thallinger GG, McFarlane RJ, Larcombe LD: Microarray meta-analysis: From data to expression to biological relationships. Computational Medicine. 2012, Springer, 59-77.View ArticleGoogle Scholar
- Ramasamy A, Mondry A, Holmes CC, Altman DG: Key issues in conducting a meta-analysis of gene expression microarray datasets. PLoS Med. 2008, 5 (9): e184-View ArticlePubMedPubMed CentralGoogle Scholar
- Barrett T, Wilhite SE, Ledoux P, Evangelista C, Kim IF, Tomashevsky M, Marshall KA, Phillippy KH, Sherman PM, Holko M: NCBI GEO: archive for functional genomics data sets—update. Nucleic Acids Res. 2013, 41 (D1): D991-D995.View ArticlePubMedGoogle Scholar
- Tusher VG, Tibshirani R, Chu G: Significance analysis of microarrays applied to the ionizing radiation response. Proc Natl Acad Sci U S A. 2001, 98 (9): 5116-5121.View ArticlePubMedPubMed CentralGoogle Scholar
- Tabas-Madrid D, Nogales-Cadenas R, Pascual-Montano A: GeneCodis3: a non-redundant and modular enrichment analysis tool for functional genomics. Nucleic Acids Res. 2012, 40 (Web Server issue): W478-W483.View ArticlePubMedPubMed CentralGoogle Scholar
- Giot L, Bader JS, Brouwer C, Chaudhuri A, Kuang B, Li Y, Hao Y, Ooi C, Godwin B, Vitols E: A protein interaction map of Drosophila melanogaster. Science. 2003, 302 (5651): 1727-1736.View ArticlePubMedGoogle Scholar
- Li S, Armstrong CM, Bertin N, Ge H, Milstein S, Boxem M, Vidalain P-O, Han J-DJ, Chesneau A, Hao T: A map of the interactome network of the metazoan C. elegans. Science. 2004, 303 (5657): 540-543.View ArticlePubMedPubMed CentralGoogle Scholar
- Shannon P, Markiel A, Ozier O, Baliga NS, Wang JT, Ramage D, Amin N, Schwikowski B, Ideker T: Cytoscape: a software environment for integrated models of biomolecular interaction networks. Genome Res. 2003, 13 (11): 2498-2504.View ArticlePubMedPubMed CentralGoogle Scholar
- Murthy SR, Pacak K, Loh YP: Carboxypeptidase E: elevated expression correlated with tumor growth and metastasis in pheochromocytomas and other cancers. Cell Mol Neurobiol. 2010, 30 (8): 1377-1381.View ArticlePubMedPubMed CentralGoogle Scholar
- Jeffrey KD, Alejandro EU, Luciani DS, Kalynyak TB, Hu X, Li H, Lin Y, Townsend RR, Polonsky KS, Johnson JD: Carboxypeptidase E mediates palmitate-induced β-cell ER stress and apoptosis. Proc Natl Acad Sci. 2008, 105 (24): 8452-8457.View ArticlePubMedPubMed CentralGoogle Scholar
- Engin F, Bertin T, Ma O, Jiang MM, Wang L, Sutton RE, Donehower LA, Lee B: Notch signaling contributes to the pathogenesis of human osteosarcomas. Hum Mol Genet. 2009, 18 (8): 1464-1470.View ArticlePubMedPubMed CentralGoogle Scholar
- Dailey DD, Anfinsen KP, Pfaff LE, Ehrhart E, Charles JB, Bønsdorff TB, Thamm DH, Powers BE, Jonasdottir TJ, Duval DL: HES1, a target of Notch signaling, is elevated in canine osteosarcoma, but reduced in the most aggressive tumors. BMC Vet Res. 2013, 9 (1): 130-View ArticlePubMedPubMed CentralGoogle Scholar
- Olstad OK, Gautvik VT, Reppe S, Rian E, Jemtland R, Ohlsson C, Bruland O, Gautvik KM: Molecular heterogeneity in human osteosarcoma demonstrated by enriched mRNAs isolated by directional tag PCR subtraction cloning. Anticancer Res. 2002, 23 (3B): 2201-2216.Google Scholar
- Fritsche‒Guenther R, Noske A, Ungethüm U, Kuban RJ, Schlag PM, Tunn PU, Karle J, Krenn V, Dietel M, Sers C: De novo expression of EphA2 in osteosarcoma modulates activation of the mitogenic signalling pathway. Histopathology. 2010, 57 (6): 836-850.View ArticleGoogle Scholar
- Han J, Tian R, Yong B, Luo C, Tan P, Shen J, Peng T: Gas6/Axl mediates tumor cell apoptosis, migration and invasion and predicts the clinical outcome of osteosarcoma patients. Biochem Biophys Res Commun. 2013, 435 (3): 493-500.View ArticlePubMedGoogle Scholar
- Azuma K, Tanaka M, Uekita T, Inoue S, Yokota J, Ouchi Y, Sakai R: Tyrosine phosphorylation of paxillin affects the metastatic potential of human osteosarcoma. Oncogene. 2005, 24 (30): 4754-4764.View ArticlePubMedGoogle Scholar
- Noiret M, Audic Y, Hardy S: Expression analysis of the polypyrimidine tract binding protein (PTBP1) and its paralogs PTBP2 and PTBP3 during Xenopus tropicalis embryogenesis. Int J Dev Biol. 2012, 56: 747-753.View ArticlePubMedGoogle Scholar
- Licatalosi DD, Yano M, Fak JJ, Mele A, Grabinski SE, Zhang C, Darnell RB: Ptbp2 represses adult-specific splicing to regulate the generation of neuronal precursors in the embryonic brain. Genes Dev. 2012, 26 (14): 1626-1642.View ArticlePubMedPubMed CentralGoogle Scholar
- The pre-publication history for this paper can be accessed here:http://www.biomedcentral.com/1471-2350/15/80/prepub
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/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.