Research Article Low GAS5 Levels As a Predictor of Poor Survival in Patients with Lower-Grade Gliomas

Hindawi Journal of Oncology Volume 2019, Article ID 1785042, 15 pages https://doi.org/10.1155/2019/1785042

Research Article Low GAS5 Levels as a Predictor of Poor Survival in Patients with Lower-Grade Gliomas

Yanfang Wang ,1 Shan Xin,1 Kai Zhang,2 Run Shi ,1 and Xuanwen Bao 3,4

1 Ludwig-Maximilians-Universitat¨ Munchen¨ (LMU), 80539 Munich, Germany 2Department of Cardiology, Sir Run Run Shaw Hospital, Zhejiang University School of Medicine, 310016 Hangzhou, China 3Institute of Radiation Biology, Helmholtz Center Munich, German Research Center for Environmental Health, 85764 Neuherberg, Germany 4Technical University Munich (TUM), 80333 Munich, Germany

Correspondence should be addressed to Run Shi; [email protected] and Xuanwen Bao; [email protected]

Received 1 October 2018; Revised 18 December 2018; Accepted 3 January 2019; Published 3 February 2019

Academic Editor: Srikumar P. Chellappan

Copyright © 2019 Yanfang Wang et al. Tis is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Introduction. Gliomas are infltrative neoplasms of a highly invasive nature. Diferent stages of gliomas feature distinct genomic, genetic, and epigenetic changes. Te long noncoding RNA Growth Arrest Specifc Transcript 5 (GAS5) is an identifed tumour suppressor involved in several cancers. However, the underlying roles of the GAS5 gene in lower-grade glioma (LGG) patients are not clear. Methods. Via bioinformatic analysis based on TCGA-LGG and TCGA-GBM data, we explored the mechanisms of GAS5 expression in LGG (grades II and III) and high-grade glioma (glioblastoma multiforme, grade IV). Te log-rank test and multivariate Cox analysis were performed to fnd the association between GAS5 and overall survival (OS) in LGG patients. Weighted gene coexpression network analysis (WGCNA) and RNA-Seq analysis were applied to fnd the key gene network associated with GAS5. Results. We found that GAS5 expression was downregulated in both LGG and glioblastoma multiforme (GBM) compared with normal brain tissue. Low methylation in the GAS5 promoter region was detected in both LGG and GBM tissues. Te amplifcation type was the predominant type of GAS5 gene alteration in both LGG and GBM. High GAS5 expression was more associated with long overall survival (OS) in LGG patients than in GBM patients. Te multivariate survival analysis of GAS5 and clinical and molecular characteristics in LGG patients further confrmed the association between GAS5 and OS in LGG patients. We then developed a nomogram for clinical use. WGCNA and RNA-Seq analysis indicated that ribosomal biogenesis and translation initiation were the predominant events regulated by GAS5 in LGG patients. Conclusion. Taken together, these results demonstrate that GAS5 expression is associated with OS in LGG patients and that its underlying roles involve the regulation of ribosomal biogenesis and translation initiation, which may aid in identifying a new target for the treatment of LGG.

1. Introduction children with central nervous system (CNS) neoplasms, and no efective therapies currently exist [5]. In most cases, Gliomas are commonly classifed as low-grade glioma (LGG, GBM will recur afer surgical resection, resulting in a poor grades II and III) or glioblastoma multiforme (GBM, grade prognosis. Hundreds of molecular alterations exist in LGG IV). LGGs are infltrative neoplasms that arise most ofen and GBM, making them react diferently to chemotherapy, in the cerebral hemispheres of adults, including oligoden- radiotherapy, and surgical resection. Identifying the key drogliomas, oligoastrocytomas, and astrocytomas [1]. A sub- genes or genome alterations that drive the progression of set of LGGs progress into GBM within a few months, and gliomas will contribute to the understanding of the molecular others may stay stable for several years, causing large varia- mechanism behind gliomas and help to improve the efects of tions in median survival [2, 3]. Hence, individual treatments therapy. should be performed based on the identifcation of histologic Long noncoding RNAs (lncRNAs) are non-protein- class, grade, and reactions to chemotherapy and radiotherapy coding transcripts that play essential roles in cellular regu- [4]. GBM is the most common cause of death among lation at various levels and in diverse biological functions, 2 Journal of Oncology including chromatin modifcation, transcriptional regula- Te salmon and blue modules, which had most signifcant tion, cell diferentiation, immune responses, and epigenetic adjusted p-values, were selected. Genes involved in the regulation [6–8]. GAS5 produces a spliced lncRNA, which salmon and blue modules and their weights as calculated by sensitizes cells to apoptosis by suppressing glucocorticoid- WGCNA were visualized by Cytoscape 3.4.0 [21]. GO analysis mediated induction of several responsive genes [9]. By was performed using the clusterProfler package [22] and binding to the DNA-binding domain of the glucocorticoid Metascape (http://metascape.org/gp/index.html) [23]. Gene receptor, GAS5 acts as a decoy glucocorticoid response ele- set enrichment analysis (GSEA) was applied by the GSEA ment (GRE), thus competing with DNA GREs to bind to the sofware from Broad Institute [24]. glucocorticoid receptor [9]. By analysing GAS5 level and clinical parameters, it has been found that GAS5 correlates with 2.3. Statistical Analysis. Statistical analysis was performed tumour progression and poor prognosis in many tumour using R (R: A language and environment for statistical types [10–13]. Some publications have also indicated that computing. R Foundation for Statistical Computing, Vienna, GAS5 is a predictor of survival in GBM patients [14, 15]. One Austria. URL http://www.R-project.org/). Te association study revealed an association between serum GAS5 level and between GAS5 expression and the clinicopathological fea- the OS of GBM patients [14]. However, the potential roles of tures was evaluated by the diferent mathematical methods GAS5 involvement in gliomas and the regulatory mechanism listed in the table [25]. Te log-rank test was performed to of GAS5 expression in gliomas are not clearly identifed. assess the diference between the Kaplan-Meier curves. Te Besides, the diference of GAS5 expression and the regulatory fold change and Q value for samples with high and low GAS5 mechanism of GAS5 expression in LGG and GBM patients expression were calculated using the limma package [26]. A have not been shown yet. In this study, via bioinformatic Qvalue< 0.05 was considered statistically signifcant. analysis, we explored the regulatory mechanisms of GAS5 expression in gliomas and compared its prognostic value in 3. Results LGG and GBM. In addition, based on TCGA-LGG RNA-Seq datasets, we also applied WGCNA and enrichment analysis 3.1. GAS5 Was Upregulated in Both LGG and GBM Compared to identify the pathways highly associated with GAS5 in LGG with Normal Brain Tissues. GAS5 expression levels were and the network of genes highly associated with GAS5. characterized in several types of solid tumours, including LGG and GBM (Figure 1). Te results indicated that GAS5 2. Materials and Methods was downregulated in both LGG and GBM tissue compared with the normal brain tissue. Ten, we compared the 2.1. Clinical and Omic Data Download. RNA-Seq and DNA expression level of GAS5 between LGG and GBM (Figures methylation datasets from patients with primary LGG or 2(a), 2(b), and S1a). No signifcant diference was identifed GBM were downloaded from the UCSC Xena browser between the LGG and GBM datasets. In addition, grade II, (https://xenabrowser.net/) [16]. GAS5 mRNA expressions grade III, and grade IV gliomas had similar expression levels in diferent types of solid tumours and in corresponding of GAS5, which were all higher than that of normal brain normal tissues were obtained from TCGA datasets and GTEx tissues. Te DNA methylation level of GAS5 was analysed datasets [17]. A box plot was created using ggplot2 [18]. by DNA methylation 450K chips, and the six probes were chosen from the CpG island in the promoter region of GAS5. All probes showed low � values (<0.1), and there 2.2. WGCNA of RNA-Seq Data from LGG Patients and � Network Visualization. Te WGCNA R package was used were no diferences in values from any of the six probes to evaluate the correlation of GAS5 expression and module between LGG and GBM tissues, indicating low methylation membership by the ‘p.weighted’ function [19]. An adjacency of the GAS5 promoter in both LGG and GBM patients matrix was generated to evaluate the weighted coexpression (Figure 2(c)). Te association between GAS5 methylation values between all the subjects in the probe set. Te coex- and GAS5 expression was tested, revealing no signifcant relationship in either GBM or LGG tissues (Figures S1B and pression similarity ��,� wasdefnedastheabsolutevalueofthe correlation coefcient between the profles of nodes � and �: S1C). Ten, to investigate the driver of high GAS5 expression in LGG and GBM tissues, we also examined copy number � � � � alterations (CNAs) in LGG and GBM (Figures 2(d) and 2(e)). ��,� = �� (��,��)� (1) Te results showed that the amplifcation-type alteration where �� and �� were expression value of for genes � and � and contributed to the high expression of GAS5 in both LGG and ��,� represented Pearson’s correlation coefcient of gene � and GBM tissues (P = 0.044 and P = 0.047). gene �. A weighed network adjacency was defned by raising the 3.2.LowGAS5ExpressionWasaPredictorofPoorSurvival coexpression similarity to a power �: in LGG. Te associations between GAS5 expression and the pathological parameters of patients with primary LGG and � ��,� =��,� (2) GBM were summarized in Table 1. In patients with LGG, higher GAS5 expression was signifcantly correlated with with �≥1[20]. We selected the power of � =4and higher tumour purity (Table 1 and Figure S2). Nevertheless, 2 scale free R = 0.95 as the sof-thresholding parameter we could not fnd a similar tendency in patients with GBM. to ensure a signed scale-free coexpression gene network. Furthermore, in LGG patients, higher expression of GAS5 Journal of Oncology 3

Table 1: Te associations between GAS5 expression and pathological parameters in patients. Attributes GBM LGG GAS expression and ... Statistic P-value FDR (BH) Statistic P-value FDR (BH) Ethnicity (Wilcox test) 0.096 0.620 0.843 0.002 0.487 0.487 Histological type 2.467 0.291 0.843 3.918 0.141 0.197 (Kruskal-Wallis test) Race (Kruskal-Wallis test) 1.110 0.574 0.843 4.180 0.243 0.283 Radiation therapy (Wilcox -0.003 0.843 0.843 -0.025 0.019 0.045 test) Tumour purity (Spearman -0.023 0.785 0.843 0.300 1.775e-11 1.243e-10 correlation) Age (Spearman correlation) -0.064 0.446 0.843 -0.093 0.041 0.073

11 GAS5 log2(RSEM+1)

GBM KIRC KIRP LGG LIHC BLCA BRCA COAD KICH THCA UCEC

normal tumor Figure 1: GAS5 mRNA expression in diferent types of solid tumours and corresponding normal tissues. Te RNA-Seq data of normal brain tissue were obtained from GTEx dataset. was associated with longer OS according to the log-rank GAS5 showed a good predictive efect for survival (Figures test (Figure 3(a)). However, the level of GAS5 expression did 5(a)–5(d)). not show a strong association with the OS of GBM patients (Figure 3(b)). 3.4. Multivariate Survival Analysis of GAS5 with Clinical and Molecular Characteristics in LGG Patients. We applied the 3.3. GAS5 Expression in LGG Patients with Diferent Clin- Cox regression model to conduct a multivariable survival ical and Molecular Characteristics. To further identify the analysis and used Cox regression coefcients to generate relationship between GAS5 expression and overall survival a nomogram (Figure 6(a)). In the multivariable survival of LGG patients, we frst examined GAS5 expression in analysis, we included histologic grade, gender, histological LGG patients with diferent IDH1 status (Figure 4(a)), his- type, IDH mutation status, age, and GAS5 expression level tologic grades (Figure 4(b)), tumour sizes (Figure 4(c)), and as risk factors. Te nomogram predicts LGG patients’ overall histologic types (Figure 4(d)). Te results showed that the 3-year and 5-year survival probability. Te results indicated expression of GAS5 in IDH1 mutant patients was signifcantly that the expression level of GAS5 was highly associated with higher than that in IDH1 wild-type patients. In contrast, overall survival in LGG patients. Calibration plots showed histologic grade, tumour size, and histological type had no that the nomograms performed well compared with an association with GAS5 expression. A Kaplan-Meier survival ideal model (Figure 6(b)). Ten, the risk scores from Cox analysis stratifed by clinicopathological risk factors was regression were assessed with a time-dependent receiver performed to verify the efect of GAS5 on survival. In patients operating characteristic (ROC) curve (Figure 6(c)). Te area with large or small tumours of histologic grade II or III, under the curve (AUC) was 0.897 and 0.825 for 3 years and 5 4 Journal of Oncology

Sample number Primary Copy number GAS5 expression GAS5 DNA methylation N=649 disease change RNAseq (Methylation 450k)

LGG 10 50 samples GAS5 expression

8 GBM GBM LGG tumor type (a) (b)

1.0

0.8

0.6 –value  0.4

0.2

0.0

cg25883402cg17868533cg16905926cg21770828cg09058860cg17170111 GBM LGG (c)

12 12

10 10 log2(RSEM+1) log2(RSEM+1) GAS5 expression GAS5 expression

8 8

p=0.044 p=0.047 6 6 1 copy del no change amplifcation high−amplifcation 1 copy del no change amplifcation Copy number variation Copy number variation (d) (e)

Figure 2: GAS5 expression, DNA methylation, and copy number alteration (CNA) in LGG and in GBM. (a) Heatmap of GAS5 mRNA expression, copy number alteration, and DNA methylation in patients with primary LGG or GBM. Data were obtained from TCGA-LGG and TCGA-GBM. (b) Violin plots of GAS5 expression in LGG and GBM tissues. (c) Bar plots of GAS5 methylation level in LGG and in GBM tissues; six probes were chosen from the CpG island before the transcription start site of GAS5. (d-e) Box plots of GAS5 expression in LGG (d) and GBM (e) tissues with indications of genetic status. Data were obtained from TCGA-LGG and TCGA-GBM. Journal of Oncology 5

1.00 1.00

0.75 0.75

0.50 0.50 Survival probability Survival probability 0.25 0.25 p < 0.0001 p = 0.042

0.00 0.00 0 500 1000 1500 2000 2500 0 500 1000 1500 2000 2500 Time in days Time in days High GAS5 Low GAS5 Strata expression expression High GAS5 Low GAS5 Strata expression expression (a) (b)

Figure 3: Te associations between GAS5 expression and OS in LGG (a) and GBM (b) patients.

years, respectively. Te patients were separated into 2 groups, from the salmon and blue modules with coefcients higher namely, those with high and low risk scores according to the than 0.5 were selected and subjected to GO enrichment Cox regression model. Te Kaplan–Meier curve, which was analysis (Figures 8(c) and 9). rRNA processing, ribonu- generated by the log-rank test, showed good performance in cleoprotein complex biogenesis, translation factor activity, predicting the OS of LGG patients (Figure 6(d)). ribosomal subunit biogenesis, and other terms were enriched in the analysis, indicating the key roles of GAS5 in the 3.5. WGCNA of RNA-Seq in LGG. To identify the gene regulation of ribosomes in LGG tissues (Figures 9(a), 9(c), network that was highly associated with GAS5 expression, we and 9(e)). Te genes involved in GO analysis and their fold further analysed LGG RNA-Seq data by WGCNA. WGCNA changes between the high GAS5 expression group (the 50 has been widely applied to study coexpressed gene networks samples with the highest GAS5 expression) and the low GAS5 by scale-free weighted network construction and module expression group (the 50 samples with the lowest GAS5 detection. In our analysis, 43 modules were detected; the expression) are shown on the circular plots (Figures 9(b), clustering of genes is shown as a dendrogram (Figures 7(a) 9(d), and 9(f)). RPL18A, RPL23, RPL24, PRL32, PRS12, and and S3). Te module similarity was quantifed and is shown other ribosome proteins were highly expressed along with as a heatmap below (Figure 7(b)). Te weighted network high GAS5 expression, resulting in the upregulation of rRNA of identifed genes from RNA-Seq is shown as a heatmap processing and ribosome biogenesis. Finally, the network below, which depicts the topological overlap matrix among connections in the salmon and blue modules with weight all genes (Figure 7(c)). Ten, 26 modules were identifed to be parameters higher than 0.19 from WGCNA were loaded into related with GAS5 expression by the heatmap of associations Cytoscape and analysed. A variety of ribosome proteins, such between modules and GAS5 expression (Figure 7(d)). as RPL10, RPL12, RPL17, and RPS24, were included in the network (Figures 10(a) and 10(b)). High interconnectivity 3.6. Enrichment Analysis of Genes Highly Associated with within the gene cluster is depicted by thick edges. Te GAS5 Expression in LGG. Based on the results above, the genes most strongly related to GAS5 were analysed in the salmon and blue modules, which had the most signifcant TCGA dataset and validated with a CGGA (Chinese Glioma relationship with GAS5 expression, were selected. Te scat- Genome Atlas) dataset (Figures 10(c) and 10(d)). Te most terplot below illustrates genetic signifcance between GAS5 related genes included RPL6, RPL37A, RPL37, RPS24, and expression (the correlation of genes with GAS5 expression) C20orf199, which were included in the salmon module. Te and module membership (the correlation of genes with results further confrmed the relationship between salmon clusters) for the salmon module (Figure 8(a)) and the blue module and GAS5. module (Figure 8(b)). Te results indicated that genes that had high correlations with the salmon and blue modules were 3.7. RNA-Seq Analysis of the High and Low GAS5 Expression also strongly associated with GAS5 expression. Te genes Groups. To validate the results of WGCNA, we performed 6 Journal of Oncology

13 13 p=0.85 p < 0.0001

12 12

11 11 log2(RSEM+1) GAS5 expression log2(RSEM+1) 10 GAS5 expression 10

9 9

Wild type IDH1 mutation G2 G3 Histologic grade IDH1 mutation status (a) (b)

13 13 p=0.17 p=0.046

12 12

11 11 log2(RSEM+1) GAS5 expression log2(RSEM+1) GAS5 expression 10 10

9 9

Small tumor size Large tumor size Astrocytoma Oligoastrocytoma Oligodendroglioma Tumor size Histologic type (c) (d)

Figure 4: GAS5 expression in LGG patients with diferent IDH1 statuses (a); histologic grade (b); tumour sizes (c); and histological type (d).

GO analysis and GSEA in the high and low GAS5 expression and ribonucleoprotein complex biogenesis were signifcantly groups. Te workfow is shown in Figure 11(a). Te volcano upregulated (Figures 11(e), 11(f), and 11(g)). An MA plot was plot indicates the fold changes and Q values of diferentially generated to show the diferentially expressed genes with high expressed genes (Figure 11(b)). A heatmap of the 200 difer- signifcance levels between the high and low GAS5 expression entially expressed genes with the highest signifcance levels groups (Figure S4). between the high and low GAS5 expression groups is shown in Figure 11(c). Ten, diferentially expressed genes were 4. Discussion analysed by GO analysis (Figure 11(d)). Te results indicated a high degree of association between GAS5 expression and LncRNAs are non-protein-coding transcripts longer than 200 translation initiation. GSEA was performed using all genes nucleotides and have been demonstrated to play critical roles from GAS5 high-expressed group and GAS5 low-expressed in diverse cellular processes and the regulation of biological group, indicating translation initiation, ribosomal biogenesis, functions [7, 27, 28]. LncRNAs regulate gene expression at Journal of Oncology 7

1.00 1.00

0.75 0.75

0.50 0.50 Survival probability Survival Survival probability Survival 0.25 0.25 p < 0.0001 p = 0.0085

0.00 0.00 0 500 1000 1500 2000 2500 0 500 1000 1500 2000 2500 Time in days Time in days

High GAS5 Low GAS5 High GAS5 Low GAS5 Strata expression expression Strata expression expression (a) (b)

1.00 1.00

0.75 0.75

0.50 0.50 Survival probability Survival probability 0.25 0.25 p = 0.00012 p < 0.0001

0.00 0.00 0 500 1000 1500 2000 2500 0 500 1000 1500 2000 2500 Time in days Time in days High GAS5 Low GAS5 High GAS5 Low GAS5 Strata expression expression Strata expression expression (c) (d)

Figure 5: Kaplan-Meier survival analysis for LGG patients according to GAS5 expression levels, stratifed by clinicopathological risk factors. (a, b) Tumour size. (c, d) Tumour grade. P values were calculated using the log-rank test.

Kip1 the transcriptional, posttranscriptional, and epigenetic levels. of E2F1 to the P27 promoter in prostate cancer [33]. Te expression levels of lncRNAs are frequently deregulated GAS5 expression is also markedly downregulated in gastric in cancer [29–32]. For instance, GAS5 has been reported as cancer tissues and infuences gastric cancer cell proliferation a crucial tumour suppressor in a variety of human cancers. via regulating the expression of E2F1 and P21 [11]. One GAS5 expression is signifcantly decreased in prostate cancer study revealed that GAS5 was signifcantly downregulated in cells compared with prostate epithelial cells [33]. Ectopic lung adenocarcinoma tissues compared with paired adjacent expression of GAS5 induces cell-cycle arrest in the G0–G1 nontumorous tissue samples. Te overexpression of GAS5 Kip1 phase by increasing the activity of the P27 promoter. In varied inversely with the expression of EGFR pathway and addition, GAS5 interacts with E2F1 and enhances the binding IGF-1R proteins in lung adenocarcinoma tissues [12]. 8 Journal of Oncology

coxph regression Points 0 20 40 60 80 100 MALEFEMALE Gender 3 Oligoastrocytoma and Oligodendroglioma –year Astrocytoma 5–year Histological type

GAS5 expression 13 12.5 11.512 11 10.5 10 9.5 9 8.5 8 IDH1 mutation

Wildtype IDH mutation

G2 G3 Histologic grade 0.4 0.6 0.8 1.0 Observed OS Age 10 15 20 25 30 35 40 4550 55 60 65 70 75

Total-points-to-outcome nomogram:

Total points

150 200 250 300 350 400 0.0 0.2 0.0 0.2 Pr( Time < 1095) 0.4 0.6 0.8 1.0 0.005 0.01 0.015 0 025 0.04 0.06 0.1 0.2 0.3 0.5 0.7 0.85 0.96 0992 0.999 Nomogram–predicted OS Pr( Time < 1825) 0.02 0.04 0.06 0.1 0.2 0.3 0.4 0.6 0.8 0.9 0.97 0994 1 X – resampling optimism added, B=1000 Based on observed–predicted (a) (b) 1095 1825 1.00

1.00

0.75 0.75

0.897 0.825 TP 0.50 0.50

0.25 Survival probability 0.25 p < 0.0001

0.00 0.00 0 1000 2000 3000 4000

0.00 0.25 0.50 0.75 1.00 0.00 0.25 0.50 0.75 1.00 Time in days FP High GAS5 Low GAS5 Strata expression expression (c) (d)

Figure 6: Multivariate survival analysis of GAS5 and clinical and molecular characteristics in LGG patients. Data are AUC (95% CI) or p values. ROC: receiver operating characteristic. AUC: area under the curve. (a) Te nomogram to predict overall survival in LGG patients. (b) Plots depict the calibration of each model in terms of agreement between predicted and observed 3-year and 5-year outcomes. Model performance is shown by the plot, relative to the 45-degree line, which represents perfect prediction. (c) Time-dependent ROC curves indicate the prognostic accuracyof the nomogram over 3 and 5 years. We used AUCs at 3 and 5 years to assess prognostic accuracy.(d) Kaplan-Meier survival analysis was conducted for LGG patients in the high-risk-score group and the low-risk-score group from a Cox regression model, and p values were calculated using the log-rank test. Journal of Oncology 9

Eigengene adjacency heatmap 1

Cluster Dendrogram

1.00.90.80.70.60.50.4 0.8

0.6

Height 0.4

0.2

Module colors 0

GAS expression level GAS expression level

(a) (b) Network heatmap plot Module-trait relationships MEgreenyellow MEwhite MEcyan 1 MEplum1 MEpurple MEtan MEturquoise MEdarkmagenta MEred MEblue MEsalmon MElightsteelblue1 MEdarkgrey 0.5 MEroyalblue MEviolet MEorange MElightcyan MEmidnightblue MEsaddlebrown MEyellowgreen MElightgreen MEsteelblue 0 MEpink MEdarkturquoise MEyellow MEdarkgreen MEgrey60 MEdarkolivegreen MEmagenta MEorangered4 MEskyblue MEpaleturquoise -0.5 MEsienna3 MEskyblue3 MEmediumpurple3 MEivory MEbrown MEdarkorange MEgreen MEblack MElightyellow MEdarkred -1 MElightcyan1 MEgrey (c) (d)

Figure 7: WGCNA of RNA-Seq data from LGG patients and GBM patients. (a) Dendrogram of modules identifed by WGCNA. (b) Heatmap of module-GAS5 expression associations. (c) Topological overlap matrix among detected probes from RNA-Seq. (d) Heatmap for module similarity. (e-f) Scatterplot of gene signifcance for GAS5 expression versus module membership in the salmon (e) and blue modules (f).

Recent studies have implied that lncRNAs participate the association of glucocorticoid receptors with their DNA in the development and malignancy of GBM by regulating recognition sequences by binding to their DNA-binding glioma stem cell self-renewal, proliferation, diferentiation, domain through its double-stranded RNA glucocorticoid therapeutic response, and resistance [34–37]. In this study, receptor element mimic [9]. Tus, GAS5 is upregulated we focused on GAS5, one crucial noncoding gene in glioma in growth-arrested cells and sensitizes mammalian cells to progression [37–39]. GAS5 plays a critical role in the control apoptosis [40]. In this study, we focused on the mRNAs of mammalian apoptosis and acts as a competitive glucocor- coexpressed with GAS5 in gliomas and how GAS5 controlled ticoid response element to mediate cell population growth the survival of glioma patients. By characterizing the lev- by suppressing glucocorticoid receptors [9]. GAS5 inhibits els of GAS5 expression based on data from TCGA-LGG 10 Journal of Oncology

cor=0.94,p=1.1e−119 0.6 cor=0.51,p=2.3e−158 0.8

0.4 0.6

0.2 Gene signifcance for GAS5 for Gene signifcance Gene signifcance for GAS5 for Gene signifcance 0.4

0.0 0.25 0.50 0.75 0.4 0.6 0.8 Module Membership in blue module Module Membership in salmon module

threshold negative positive threshold negative positive (a) (b)

Spliceosomal complex

Mitochondrial protein complex Mitochondrial membrane part SRP-dependent cotranslational protein

Ribosome biogenesis/ subunit biogenesis Cytoplasmic Translation factor translation activity, RNA binding

Maturation of 5.8S rRNA Regulation of translation

Ribosomal large subunit Rough endoplasmic biogenesis reticulum

Maturation of LSU-rRNA 5S RNA binding

Nucleolar part (c)

Figure 8: Te relationship between gene signifcance and module membership for the salmon and blue modules. (a-b) Scatterplot of gene signifcance for GAS5 expression versus module membership for the salmon (a) and blue modules (b). Red indicates a negative correlation with GAS5, and blue indicates a positive correlation with GAS5. (c) Te network of gene ontology analysis for biological processes, cellular components, and molecular functions based on genes in the salmon and blue modules. Journal of Oncology 11

Biological process

ribosome biogenesis

translational initiation size 75 80 nuclear-transcribed mRNA catabolic process 84 89 viral gene expression p.adjust fold change 6.522929e-122 9 2.640266e-99 7 5.280531e-99 5 nuclear-transcribed mRNA catabolic process, nonsense-mediated decay 7.920797e-99 3 1.056106e-98 Count category cotranslational protein targeting to membrane 75 cotranslational protein targeting to membrane 80 establishment of protein localization to endoplasmic reticulum 85 protein targeting to ER 90 SRP-dependent cotranslational protein targeting to membrane SRP-dependent cotranslational protein targeting to membrane 95 translational initiation

protein targeting to ER

establishment of protein localization to endoplasmic reticulum

protein localization to endoplasmic reticulum

0.300 0.325 0.350 0.375 GeneRatio (a) (b)

Cellular component

ribosome

ribosomal subunit fold change 9 cytosolic part 7 5 3 p.adjust cytosolic ribosome 1.331083e-121 category 1.252153e-24 cytosolic large ribosomal subunit 2.504307e-24 cytosolic part 3.756460e-24 cytosolic ribosome large ribosomal subunit 5.008614e-24 ribosomal subunit ribosome Count 40 size cytosolic large ribosomal subunit 50 60 46 7070 62 80 79 focal adhesion 90 95

cell-substrate adherens junction

small ribosomal subunit

cytosolic small ribosomal subunit

0.15 0.20 0.25 0.30 0.35 GeneRatio (c) (d)

Molecular function

structural constituent of ribosome

rRNA binding

category  mRNA 5 -UTR binding mRNA binding rRNA binding structural constituent of ribosome translation factor activity, RNA binding translation initiation factor activity translation factor activity, RNA binding Count fold change 20 9 40 7 translation initiation factor activity 60 5 80 3 p.adjust size ribonucleoprotein complex binding 0.0005 8 0.0010 34 0.0015 60 unfolded protein binding 86

 mRNA 5 -UTR binding

5S rRNA binding

translation elongation factor activity

0.1 0.2 0.3 GeneRatio (e) (f)

Figure 9: Enrichment analysis of the genes in the salmon and blue modules. (a-b) Gene ontology analysis for biological processes and the gene regulation involved in biological processes. (c-d) Gene ontology analysis for cellular components and the gene regulation involved in cellular components. (e-f) Gene ontology analysis for molecular functions and the gene regulation involved in molecular functions. 12 Journal of Oncology

EXOSC4 FAM195BFAM128A DHRS4 GADD45GIP1 DHRS4L2 DDT BGLAP RPL36RPL29RPL7A GSTP1 RPL34 RPL6 CUTA RPL35A RPL23 GSTZ1 C12orf44 C20orf199 COX8A RPL38 IFI27L2 CCDC56 BTF3 CCDC12 RPL41 MDP1 CHMP4A RPL17 C9orf169 FAU MPG CISD3 RPL35 C2orf79 DCXR RPL27 MRPL11 RPL23A C21orf70 ENDOG RPL13 MRPL12 HSPB9 RPL14 C19orf43 LOC442454 MRPL14 IL17RE RPL12 C17orf89 JMJD4 RPL24 MRPL21 RPL10 C12orf57 LSMD1

RPS12 MRPL23 MLST8 NACA C11orf48 MRPL24 RPL27A MRPS7 RPL26 C11orf31 MRPL27 C16orf92 RPL30 RPL10A NDUFS3 RPL31P11 ALKBH3 NDUFS6 RPL19 NDUFA11 PFDN5 C16orf42 PHPT1 C2orf7 RPS3A PIN4 NSMCE1 RPL23P8 PSMC3 CHCHD5 RPL32 MRPS34 R3HCC1 RPSAP58 RPS23 DNLZ RNF181 SNHG5 TSSC4 RPS21 ATP5D SAT2 RPLP2 TIMM13 SCAND1 RPS7 ALKBH7 SCGB3A1 RPL37 SF3B5 RPS20 HINT2 SHBG RPS10 EDF1 RPS17 TADA3 RPS6 LOC100271831 RPS27 SAC3D1 THAP7 MRPL53 RPS3 NDUFB10 TIMM10 RPL37A NUBP2 RPS4X MRPL54 TOMM7 FLJ44635 NDUFB11 RPS27A NDUFA13 TXNL4A RPS14 RPS28 RPL31 NDUFB7 USE1 RPS18 NTHL1 NHP2 VPS28 UBA52 PSMD4 RPS13 OVCA2 RPS29 TIMM17B EEF1B2 NACA2 PSMB6 ZNF358 RPS15ANACAP1RPS24 PSMB7 TCEB2 PSMD13 SURF1 SURF2

(a) (b)

C2 C20 0o 0 0.6 o 4 0 0.8 r 4 3 1.2 rf 24 1.6 f1 S2 2.4 1 S 3.2 9 P 1.8 9 P 2.4 9 9 R 2.4 R 1.8 2.4 3.2 1.6 1.2 3 4 0.8 0.6

0 0 0 0

0.8 3 0.6 4

1.6 2.4 G 1.2

3.2

R 5

P 1.8

4.2 S

2.4 5

L L

1.8

1.2 A

6 6

1.6 G

3.2 2.4 0.6

0.8

4 3 0 0 0

0 4 2.4 0.8 0.6 3.2 1.8 1.6 2.4 R 1.2 1.2 R 2.4 7 P P 1.6 3.2 3 L 0.6 1.8 L 0.8 4 L 37 0 2.4 3 0 R P A 3 37 7A RPL

−1 1 −1 1 (c) (d)

Figure 10: Visualization of genes in the salmon and blue modules with weights higher than the threshold (weight > 0.19) and correlations between GAS5 and the most closely related molecules. Node sizes indicate betweenness centrality, which refects the amount of control that this node exerts over the interactions of other nodes in the network. Edge sizes depict the weight of the connection between genes. (a) Genes in the salmon module with weights higher than 0.19. (b) Genes in the blue module with weights higher than 0.19. (c) Te correlations between GAS5 and the most closely related genes in the TCGA dataset. (d) Te validation of the association in the CGGA dataset.

and TCGA-GBM patients, we demonstrated a signifcant there was no signifcant association between the methylation downregulation of GAS5 expression in LGG and GBM com- level and the expression level of GAS5. Amplifcation-type pared with normal brain tissue. One previous study revealed alteration was associated with increased GAS5 expression that GAS5 expression decreased as the histologic grade of in both LGG and GBM tissues. GAS5 expression played glioma increased [38]. However, by analysing TCGA-LGG diferent roles in afecting the OS of LGG and GBM patients. and GBM RNA-Seq datasets, we arrived at the conclusion GAS5 expression was strongly associated with OS in LGG that histologic grade did not infuence GAS5 expression. patients. We also examined GAS5 expression in LGG patients Te discrepancy may be due to the sample numbers. In with diferent molecular characteristics and histologic grades. this study, we analysed the expression levels of GAS5 in GAS5 expression was higher in IDH1 mutant patients than 702 samples, far more than Zhao’s study used. One previous in patients with wild-type IDH1. To confrm the association report revealed that hypermethylation of CpG islands in between GAS5 and OS in LGG patients, we performed promoter regions contributes to the downregulation of GAS5 Kaplan-Meier survival analysis on LGG patients according expression [10]. Te methylation level of the GAS5 gene to their GAS5 expression levels, stratifed by clinicopatholog- was detected by the six probes targeting the CpG island ical characteristics. Kaplan-Meier survival analysis showed before the transcription start site of GAS5. Both LGG and high association between GAS5 expression level and OS in GBM tissues had low GAS5 methylation levels. However, both grade II and grade III gliomas. We also performed Journal of Oncology 13

Data Collection GAS5_highest Tissues & GAS5_lowest Tissues (n=50) (n=50) −log10 p−value Criteria RNA-seq, Q value < RNA-seq Data of 2 20 0.05, Abs (Log2(FC)) >1 groups included

Analysis Gene ontology GSEA 0 −5.0 −2.5 0.0 2.5 5.0 log2 fold change (a) (b)

Type 3 3 patients patients Type low expression AK3L1 BHLHE40 KIAA1522 high expression CYB561 STBD1 2 high GAS5 expression EPDR1 UGP2 LGALS8 2 6−Sep SLC9A1 SWAP70 MAML3 low GAS5 expression GCLM 1 PLS3 SPATS2L SIPA1L1 HIVEP3 TSHZ2 1 C2orf50 ANKRD45 DNAH9 GO term KLHL26 0 RAB36 ULBP3 C9orf64 EVC C14orf86 cotranslational protein targeting to membrane TGFB2 0 PLA2G5 C11orf63 -1 TOM1L1 RPL39 PPP2R3B RPAIN establishment of protein localization to endoplasmic reticulum CIRBP ALKBH2 TUT1 PRR14 -2 GPSM1 −1 ANKRD13B nuclear-transcribed mRNA catabolic process, nonsense-mediated decay SF3A2 SOX12 C11orf2 ZNF32 AIP -3 PQBP1 CCDC101 protein localization to endoplasmic reticulum SNHG6 WDR74 −2 FAU UBXN1 TAF1D C6orf134 protein targeting to ER PRR3 TYSND1 NPM3 LOC678655 TATDN1 UQCRB −3 ZNF337 protein targeting to membrane ZNF276 TOP3B CLEC2D TIGD1 LOC100128288 TFAP4 SRP-dependent cotranslational protein targeting to membrane ZNF34 ZNF74 CSNK1E EFS PCBP4 PYCR1 CECR5 translational initiation EIF3D PHB2 NOB1 C10orf2 FAM35B PUS3 viral gene expression EEF1A1 EEF1A1P9 PABPC1 EIF3E MGC21881 PLK1S1 viral transcription HNRNPA1L2 EEF2 NPM1 NAP1L1 IGBP1 CCNB1IP1 ZMYND17 RPL7 RPL3 EIF3L RPL17 RPL10A EEF1B2 RPL31 HNRNPA1 RPL6 RPS23 RPL15 EIF3H NACA BTF3 RPLP0 GNB2L1 RPS2 RPLP2 RPL27A RPL8 EEF1D RPL18A RPL13 RPS3A RPL34 RPS29 RPL27 RPL37 RPL37A RPL32 RPL26 RPS27A RPS10 RPL38 RPS21 RPL36 RPL24 RPL19 RPS20 RPSA RPSAP58 RPL29 RPL14 RPS14 RPL35A RPS7 EIF3F RPL23A RPS12 C6orf48 C20orf199 NCRNA00188 RPS3 EEF1G RPL23 RPL30 RPS15A RPS6 RPL7A RPL12 IMPDH2 RPS18 RPL10 RPS4X RPS25 RPS13 RPS24 RPL4 SNHG7 GLTSCR2 RPL22 RPL5 RPS8 RPS17 RPLP1 RPS9 RPS16 POLR2F GAS5 SNHG1 PHF21B EBF4 PCGF2 FXYD2 GNB3 RRN3P1 LOC338799 C2orf27A UPK2 LOC254559 FAM110B NCRNA00219 TNK2 RPL21 ATOH8 RCOR2 NOG NEU4 C12orf34 SOX8 OLIG2 OLIG1 SHD BMP2 FERMT1 AMH DAPL1 HMX1 GO term

Enrichment plot: GO_TRANSLATIONAL_INITIATION Enrichment plot: GO_RIBOSOME_BIOGENESIS Enrichment plot: GO_RIBONUCLEOPROTEIN_COMPLEX_BIOGENESIS 0.5 0.6 0.5 0.5 0.4 0.4 0.4 0.3 0.3 0.3 0.2 0.2 0.2 Enrichment score (ES) Enrichment score Enrichment score (ES) Enrichment score

0.1 (ES) score Enrichment 0.1 0.1 0.0 0.0 0.0

1.0 1.0 1.0 0.5 0.5 0.5 0.0 0.0 0.0 −0.5 −0.5 −0.5 − − − Ranked list metric (PreRanked) 1.0 1.0 Ranked list metric (PreRanked) 1.0 Ranked list metric (PreRanked) metric list Ranked 0 2,500 5,000 7,500 10,000 12,500 15,000 17,500 20,000 0 2,500 5,000 7,500 10,000 12,500 15,000 17,500 20,000 0 2,500 5,000 7,500 10,000 12,500 15,000 17,500 20,000 Rank in Ordered Dataset Rank in Ordered Dataset Rank in Ordered Dataset

Enrichment profle Hits Ranking metric scores Enrichment profle Hits Ranking metric scores Enrichment profle Hits Ranking metric scores

(e) (f) (g)

Figure 11: RNA-Seq analysis of LGG tissues with the highest and lowest GAS5 expression. (a) Workfow for RNA-Seq and downstream analysis based on diferentially expressed genes and all genes in RNA-Seq. (b) Volcano plot for diferentially expressed genes in the groups with high and low expression of GAS5. (c) Heatmap of the 200 diferentially expressed genes with the highest signifcance levels between the groups with high and low expression of GAS5. (d) Heatmap of gene ontology analysis for biological processes. (e-g) Gene set enrichment analysis for the groups with high and low expression of GAS5. 14 Journal of Oncology multivariate survival analysis of GAS5 with clinical and Authors’ Contributions molecular characteristics in LGG patients. Te Cox HR model indicated that GAS5 was strongly associated with the Xuanwen Bao and Run Shi conceived and designed the exper- overall survival of patients. Based on the results discussed iments. Xuanwen Bao, Shan Xin, and Kai Zhang analysed above, a nomogram was developed to predict the 3-year the data. Xuanwen Bao and Yanfang Wang wrote the paper. and 5-year survival probabilities of LGG patients. Time- All authors read and approved the fnal manuscript. Yanfang dependent ROCs and log-rank tests further confrmed the Wang and Shan Xin contributed equally to this work. results. To understand the molecular mechanism behind the Acknowledgments efect of GAS5 in LGG tissues, we performed WGCNA and downstream enrichment analysis. Te results indicated that We would like to thank Dr. Omid Azimzadeh, Dr. Natasa the genes in the salmon and blue modules played important Anastasov, and Dr. Michael Rosemann for helpful discussions roles in ribosome function, which was consistent with previ- and suggestions. ous reports on the involvement of GAS5 in ribosomal RNA biosynthesis [41]. In addition, the genes highly correlated with GAS5 expression showed intensive connection when Supplementary Materials visualized by Cytoscape. A variety of the hub genes were Figure S1: the GAS5 expression in diferent stages of glioma ribosomalproteins,e.g.,RPL13,RPL37,PRL38,andPRL10A, (a) and the correlation of GAS5 methylation level and GAS5 indicating the strong association between GAS5 and ribo- expression level in GBM (b) and LGG (c) tissues. Figure somal function. Te results were consistent with a previous S2: the associations between GAS5 expression level and study,whichrevealedthatGAS5mayactasa“ribo-repressor” tumour purity in LGG (a) and GBM (b) patients. Figure S3: of glucocorticoid receptors to infuence cell survival and the dendrogram of modules identifed by WGCNA. Figure metabolic activity during starvation by modulating the tran- S4: MA plot of diferentially expressed genes with high scriptional activity of the glucocorticoid receptor [40]. To signifcance levels in the groups with high and low GAS5 further confrm the results from WGCNA, we also performed expression groups in LGG tissues. (Supplementary Materials) RNA-Seq analysis and downstream GO analysis and GSEA. Translation initiation was the most predominant result from GO, implying that GAS5 may be an important regulator of References transcription initiation. One study demonstrated that GAS5 was enriched with eIF4E, which had two RNA binding motifs [1]Q.T.Ostrom,H.Gittleman,P.Farahetal.,“CBTRUSstatistical for GAS5 [42]. Te deletion of either motif inhibited the report: primary brain and central nervous system tumors diagnosed in the United States in 2006–2010,” Neuro-Oncology, binding of GAS5 to eIF4E. In addition, GSEA results also vol. 15, supplement 2, pp. ii1–ii56, 2013. indicated upregulation of translation initiation and ribosomal biogenesis in group with high GAS5 expression, which was [2] Network CGAR, “Comprehensive, Integrative Genomic Analy- consistent with GO analysis and previous research [43]. sis of Difuse Lower-Grade Gliomas,” Te New England Journal Altogether, the data presented here suggest that GAS5 of Medicine,vol.372,no.26,pp.2481–2498,2015. plays essential roles in the physiological process of LGG. Te [3] M. J. Van Den Bent, “Practice changing mature results of mechanism may include ribosome biogenesis and regulation RTOG study 9802: Another positive PCV trial makes adjuvant of translation initiation. In conclusion, based on molecular chemotherapy part of standard of care in low-grade glioma,” and clinical analysis, our study demonstrated that GAS5 Neuro-Oncology,vol.16,no.12,pp.1570–1574,2014. is linked tightly to ribosomal biogenesis and translation [4] A. S. Jakola, K. S. Myrmel, R. Kloster et al., “Comparison of a initiation in LGG tissues and acts as a predictor of survival strategy favoring early surgical resection vs a strategy favoring in patients with early-stage gliomas. watchful waiting in low-grade gliomas,” Journal of the American Medical Association, vol. 308, no. 18, pp. 1881–1888, 2012. Data Availability [5]E.Boufet,V.Larouche,B.B.Campbelletal.,“Immunecheck- point inhibition for hypermutant glioblastoma multiforme Te data used to support the fndings of this study are resulting from germline biallelic mismatch repair defciency,” available from the corresponding author upon request. Journal of Clinical Oncology, vol. 34, no. 19, pp. 2206–2211, 2016. [6] P.Han and C.-P.Chang, “Long non-coding RNA and chromatin Ethical Approval remodeling,” RNA Biology,vol.12,no.10,pp.1094–1098,2015. [7] J. T. Lee, “Epigenetic regulation by long noncoding RNAs,” Tis article does not contain any studies with human partici- Science, vol. 338, no. 6113, pp. 1435–1439, 2012. pants performed by any of the authors. [8] S.Carpenter,D.Aiello,M.K.Atianandetal.,“Alongnoncoding RNA mediates both activation and repression of immune Conflicts of Interest response genes,” Science,vol.341,no.6147,pp.789–792,2013. [9]T.Kino,D.E.Hurt,T.Ichijo,N.Nader,andG.P.Chrousos, We declare that we have no fnancial or personal relationships “Noncoding RNA Gas5 is a growth arrest- and starvation- with other people or organizations that could inappropriately associated repressor of the glucocorticoid receptor,” Science infuence our work. Signaling,vol.3,no.107,p.ra8,2010. Journal of Oncology 15

[10] X. Shi, M. Sun, H. Liu et al., “A critical role for the long non- [27] A. Fatica and I. Bozzoni, “Long non-coding RNAs: new players coding RNA GAS5 in proliferation and apoptosis in non-small- in cell diferentiation and development,” Nature Reviews Genet- cell lung cancer,” Molecular Carcinogenesis,vol.54,pp.E1–E12, ics,vol.15,no.1,pp.7–21,2014. 2015. [28] P. J. Batista and H. Y. Chang, “Long noncoding RNAs: cellular [11] M. Sun, F.-Y. Jin, R. Xia et al., “Decreased expression of long address codes in development and disease,” Cell,vol.152,no.6, noncoding RNA GAS5 indicates a poor prognosis and promotes pp. 1298–1307, 2013. cell proliferation in gastric cancer,” BMC Cancer,vol.14,no.1, [29] A. M. Schmitt and H. Y. Chang, “Long noncoding RNAs in article 319, 2014. cancer pathways,” Cancer Cell,vol.29,no.4,pp.452–463,2016. [12] S. Dong, X. Qu, W. Li et al., “Te long non-coding RNA, GAS5, [30] A. Lin, C. Li, Z. Xing et al., “Te LINK-A lncRNA activates enhances geftinib-induced cell death in innate EGFR tyrosine normoxic HIF1� signalling in triple-negative breast cancer,” kinase inhibitor-resistant lung adenocarcinoma cells with wide- Nature Cell Biology,vol.18,no.2,pp.213–224,2016. type EGFR via downregulation of the IGF-1R expression,” [31] W. Nie, H.-J. Ge, X.-Q. Yang et al., “LncRNA-UCA1 exerts Journal of Hematology & Oncology,vol.8,no.1,ArticleID43, oncogenic functions in non-small cell lung cancer by targeting 2015. miR-193a-3p,” Cancer Letters,vol.371,no.1,pp.99–106,2016. [13]Z.-Q.Tu,R.-J.Li,J.-Z.Mei,andX.-H.Li,“Down-regulationof [32] A. Bhan and S. S. Mandal, “LncRNA HOTAIR: A master long non-coding RNA GAS5 is associated with the prognosis of regulator of chromatin dynamics and cancer,” Biochimica et hepatocellular carcinoma,” International Journal of Clinical and Experimental Pathology,vol.7,no.7,pp.4303–4309,2014. Biophysica Acta (BBA) - Reviews on Cancer,vol.1856,no.1,pp. 151–164, 2015. [14] J. Shen, T. R. Hodges, R. Song et al., “Serum HOTAIR and GAS5 levels as predictors of survival in patients with glioblastoma,” [33] G. Luo, D. Liu, C. Huang et al., “LncRNA GAS5 inhibits cellular Molecular Carcinogenesis,vol.57,no.1,pp.137–141,2018. proliferation by targeting P27Kip1,” Molecular Cancer Research, vol.15,no.7,pp.789–799,2017. [15]X.-Q.Zhang,S.Sun,K.-F.Lametal.,“Alongnon-coding RNA signature in glioblastoma multiforme predicts survival,” [34]E.-B.Bian,J.Li,Y.-S.Xie,G.Zong,J.Li,andB.Zhao, Neurobiology of Disease,vol.58,pp.123–131,2013. “LncRNAs: New players in gliomas, with special emphasis on the interaction of lncRNAs with EZH2,” Journal of Cellular [16] M. S. Cline, B. Craf, T. Swatloski et al., “Exploring TCGA pan- Physiology,vol.230,no.3,pp.496–503,2015. cancer data at the UCSC cancer genomics browser,” Scientifc Reports,vol.3,ArticleID2652,2013. [35] P. Wang, Z. Ren, and P. Sun, “Overexpression of the long non- [17] G. Consortium, “Te Genotype-Tissue Expression (GTEx) pilot coding RNA MEG3 impairs in vitro glioma cell proliferation,” analysis: multitissue gene regulation in humans,” Science,vol. Journal of Cellular Biochemistry, vol. 113, no. 6, pp. 1868–1874, 348,pp.648–660,2015. 2012. [18] L. Wilkinson, ggplot2: Elegant Graphics for Data Analysis, [36] L. Han, K. Zhang, Z. Shi et al., “LncRNA profle of glioblastoma Springer, 2016. reveals the potential role of lncRNAs in contributing to glioblastoma pathogenesis,” International Journal of Oncology,vol.40, [19] P. Langfelder and S. Horvath, “WGCNA: an R package for no. 6, pp. 2004–2012, 2012. weighted correlation network analysis,” BMC Bioinformatics, vol. 9, article 559, 2008. [37] X. Zhao, Y. Liu, J. Zheng et al., “GAS5 suppresses malignancy of [20] S. Horvath, B. Zhang, M. Carlson et al., “Analysis of oncogenic human glioma stem cells via a miR-196a-5p/FOXO1 feedback signaling networks in glioblastoma identifes ASPM as a molec- loop,” Biochimica et Biophysica Acta (BBA) - Molecular Cell ular target,” Proceedings of the National Acadamy of Sciences of Research,vol.1864,no.10,pp.1605–1617,2017. the United States of America,vol.103,no.46,pp.17402–17407, [38] X. Zhao, P. Wang, J. Liu et al., “Gas5 Exerts Tumor-suppressive 2006. Functions in Human Glioma Cells by Targeting miR-222,” [21] P. Shannon, A. Markiel, O. Ozier et al., “Cytoscape: a sofware Molecular Terapy, vol. 23, no. 12, pp. 1899–1911, 2015. Environment for integrated models of biomolecular interaction [39] Q. Liu, S. Sun, W. Yu et al., “Altered expression of long non- networks,” Genome Research,vol.13,no.11,pp.2498–2504, coding RNAs during genotoxic stress-induced cell death in 2003. human glioma cells,” Journal of Neuro-Oncology,vol.122,no.2, [22]G.Yu,L.-G.Wang,Y.Han,andQ.-Y.He,“clusterProfler:anR pp. 283–292, 2015. package for comparing biological themes among gene clusters,” [40] M. Mourtada-Maarabouni, M. R. Pickard, V. L. Hedge, F. OMICS: A Journal of Integrative Biology,vol.16,no.5,pp.284– Farzaneh, and G. T. Williams, “GAS5, a non-protein-coding 287, 2012. RNA, controls apoptosis and is downregulated in breast cancer,” [23] S. Tripathi, M. O. Pohl, Y. Zhou et al., “Meta-and orthogonal Oncogene,vol.28,no.2,pp.195–208,2009. integration of infuenza, OMICs , data defnes a role for UBR4 in [41] C. M. Smith and J. A. Steitz, “Classifcation of gas5 as a multi- virus budding,” Cell Host & Microbe,vol.18,pp.723–735,2015. small-nucleolar-RNA (snoRNA) host gene and a member of [24] A. Subramanian, P. Tamayo, V. K. Mootha et al., “Gene set the 5’-terminal oligopyrimidine gene family reveals common enrichment analysis: a knowledge-based approach for inter- features of snoRNA host genes,” Molecular and Cellular Biology, preting genome-wide expression profles,” Proceedings of the vol. 18, no. 12, pp. 6897–6909, 1998. National Acadamy of Sciences of the United States of America, [42] G. Hu, Z. Lou, and M. Gupta, “Te long non-coding RNA GAS5 vol. 102, no. 43, pp. 15545–15550, 2005. cooperates with the eukaryotic translation initiation factor 4E [25] S. V. Vasaikar, P. Straub, J. Wang, and B. Zhang, “LinkedOmics: to regulate c-Myc translation,” PLoS ONE,vol.9,no.9,Article Analyzing multi-omics data within and across 32 cancer types,” ID e107016, 2014. Nucleic Acids Research,vol.46,no.1,pp.D956–D963,2018. [43] W. Sun, Y. Yang, C. Xu, and J. Guo, “Regulatory mechanisms of [26] G. K. Smyth, Limma: Linear Models for Microarray Data. long noncoding RNAs on gene expression in cancers,” Cancer Bioinformatics and Computational Biology Solutions Using R and Genetics, vol. 216-217, pp. 105–110, 2017. Bioconductor,Springer,2005. M EDIATORSof INFLAMMATION

The Scientifc Gastroenterology Journal of Research and Practice Diabetes Research Disease Markers World Journal Hindawi Hindawi Publishing Corporation Hindawi www.hindawi.com Volume 2018 Hindawi Hindawi http://www.hindawi.comwww.hindawi.com Volume 20182013 www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 www.hindawi.com Volume 2018

Journal of International Journal of Immunology Research Endocrinology Hindawi Hindawi www.hindawi.com Volume 2018 www.hindawi.com Volume 2018

Submit your manuscripts at www.hindawi.com

BioMed PPAR Research Research International Hindawi Hindawi www.hindawi.com Volume 2018 www.hindawi.com Volume 2018

Journal of Obesity

Evidence-Based Journal of Stem Cells Complementary and Journal of Ophthalmology International Alternative Medicine Oncology Hindawi Hindawi Hindawi Hindawi Hindawi www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 www.hindawi.com Volume 2013

Parkinson’s Disease

Computational and Mathematical Methods Behavioural AIDS Oxidative Medicine and in Medicine Neurology Research and Treatment Cellular Longevity Hindawi Hindawi Hindawi Hindawi Hindawi www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 www.hindawi.com Volume 2018