Wu et al. Cancer Cell Int (2018) 18:107 https://doi.org/10.1186/s12935-018-0603-2 Cancer Cell International PRIMARY RESEARCH Open Access Expression profle analysis of antisense long non‑coding RNA identifes WDFY3‑AS2 as a prognostic biomarker in difuse glioma Fan Wu1,2,3,4*†, Zheng Zhao1,2,3†, Ruichao Chai1,2,3, Yuqing Liu1,2,3, Kuanyu Wang1,2,3, Zhiliang Wang1,2,3, Guanzhang Li1,2,3, Ruoyu Huang1,2,3, Haoyu Jiang1,2,3 and Kenan Zhang1,2,3 Abstract Background: Increasing evidence has shown that long non-coding RNAs (lncRNAs) are important prognostic bio- markers and epigenetic regulators with critical roles in cancer initiation and progression. However, the expression and clinical prognostic value of antisense lncRNAs in difuse glioma patients remain unknown. Methods: Here, we profled diferentially expressed antisense lncRNAs in glioma using RNA sequencing data from Chinese Glioma Genome Atlas database. Cox regression was performed to evaluate the prognostic value. Gene oncol- ogy (GO) and gene set enrichment analysis (GSEA) were used for functional analysis of antisense LncRNAs. Results: Expression profling identifed 169 aberrantly expressed antisense lncRNAs between lower grade glioma (LGG) (grade II and III) and glioblastoma multiforme (GBM), 113 antisense lncRNAs between LGG IDH-wt and IDH- mut samples, and 70 antisense lncRNAs between GBM IDH-wt and IDH-mut samples, respectively. Among them, three antisense lncRNAs (WDFY3-AS2, MCM3AP-AS1 and LBX2-AS1) were signifcantly associated with prognosis and malignant progression of patients. WDFY3-AS2, the top one of downregulated antisense lncRNAs in GBM with fold change of 0.441 (P < 0.001), showed specifc decreased expression in classical, mesenchymal, LGG IDH-wt, GBM IDH-wt or MGMT promoter unmethylated stratifed patients. Chi square test found that WDFY3-AS2 was signifcantly associ- ated with the clinical and molecular features of glioma. Univariate and multivariate Cox regression analysis indicated that WDFY3-AS2 was independently correlated with overall survival (OS) of patients. Kaplan–Meier analysis found that patients with high WDFY3-AS2 expression had longer OS than the low expression ones in the stratifed cohorts. Additionally, GO and GSEA showed that gene sets correlated with WDFY3-AS2 expression were involved in regulation of synaptic transmission, glutamate receptor and TNF signaling pathway. Conclusion: Our fndings provided convincing evidence that WDFY3-AS2 is a novel valuable prognostic biomarker for difuse glioma. Keywords: Glioma, LncRNAs, Antisense, Prognosis, Biomarker *Correspondence: [email protected] †Fan Wu and Zheng Zhao contributed equally to this work 4 No. 6, Tiantan Xili, Dongcheng District, Beijing 100050, China Full list of author information is available at the end of the article © The Author(s) 2018. This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creat​iveco​mmons​.org/licen​ses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creat​iveco​mmons​.org/ publi​cdoma​in/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Wu et al. Cancer Cell Int (2018) 18:107 Page 2 of 10 Background which is located in chromosome 4q21.23. WDFY3-AS2 Gliomas, the common primary brain tumor, mostly downregulation was closely correlated with tumor grade remain incurable, despite aggressive therapies with sur- and poor prognosis in patients. Bioinformatic analy- gery, chemo and radiotherapy [1]. Glioblastoma mul- sis predicted that WDFY3-AS2 was involved in synap- tiforme (GBM) (IDH wild type) is the most aggressive tic transmission, glutamate receptor and TNF signaling glioma in adults characterized by glioma cell invasiveness pathway. Tese results indicated that WDFY3-AS2 could and robust neovascularization, with a median survival of be a potential prognostic biomarker for difuse glioma. 15 months and 5-year survival rate less than 3% follow- ing standard treatment [2]. Due to the infltrative growth Methods and inherent resistance to chemo and radiotherapy, it is Patients and tissues urgent to identify new targets to increase glioma patients’ 309 difuse glioma samples from the CGGA database survival. Although recent studies have found several were included in this study. Each sample was diagnosed genetic mutations and deregulated signaling pathways, by two neuropathologists according to the World Health these fndings are still insufcient to uncover the molecu- Organization (WHO) classifcation guidelines of central lar basis of gliomagenesis [3]. Since the coding genes only nervous system tumors [13]. Te study was approved by account for 1–2% of whole human genome [4], noncod- the ethics committee of Tiantan Hospital, and the written ing RNAs are emerging as new therapeutic targets in informed consent was obtained from all patients. glioma. Long non-coding RNAs (lncRNAs) are defned as Datasets transcripts longer than 200 nucleotides without protein- Te RNA sequencing data and corresponding clini- coding potential [5]. Antisense lncRNAs are a class of cal information (age, gender, histology, methylguanine lncRNAs relative to their complementary protein-coding methyltransferase (MGMT) promoter status, isocitrate genes, which are orientated in an antisense direction [6]. dehydrogenase (IDH) mutation status and 1p/19q status) Recent studies have demonstrated that antisense lncR- were downloaded from CGGA database (http://www. NAs are involved in multiple biological processes, includ- cgga.org.cn) [14]. Te characteristics of glioma patients ing cell proliferation, survival, migration, invasion and are summarized in Table 1. IDH mutation and MGMT apoptosis through epigenetic modifcation and chromatin promoter status were determined by DNA pyrosequenc- remodeling [7]. Moreover, accumulating reports found ing as described in previous study [15]. 1p/19q status was that dysregulated expression of antisense lncRNAs plays determined using dual-color fuorescence in situ hybridi- a crucial role in development and progression of vari- zation. According to the standard quality control crite- ous cancers. For example, it was reported that antisense ria, the raw RNA sequencing data was frst processed to lncRNA HAND2-AS1 was signifcantly down-regulated defne improper reads which would be removed. Ten, in endometrioid endometrial carcinoma, and HAND2- the gene expression was estimated using the reads per AS1 overexpression inhibited invasion and metastasis kilobase transcriptome per million reads (RPKM). Te through inactivating neuromedin U [8]. Zhu et al. found calculated gene expression data was further used to com- that upregulation of lncRNA ZEB1-AS1 promoted tumor pare the expression diference among samples [15]. metastasis and predicted poor prognosis in hepatocellu- lar carcinoma [9]. In glioma, Mineo et al. reported that Gene oncology (GO), Kyoto encyclopedia of genes lncRNA HIFA-AS2, upregulated in mesenchymal glioma and genomes (KEGG) and gene set enrichment analysis stem cells, facilitated the maintenance of self-renewal (GSEA) in hypoxia niches [10]. Han et al. found that HOXA11- GO was applied to analyze the main function of difer- AS, a novel cell cycle-associated lncRNA, could serve as entially expression genes. KEGG was performed to ana- a biomarker and therapeutic target for glioma patients lyze the pathway enrichment (http://david .ncifc rf.gov/). [11]. Our previous study reported that upregulation of GSEA was carried out to identify gene sets of statistical lncRNA HOXA-AS3 promoted tumor progression and diference between two groups by using GSEA v3 soft- predicted poor prognosis in glioma [12]. Hence, under- ware (http://www.brodi nstit ute.org/gsea/index .jsp) [16]. standing the role and prognostic value of cancer associ- ated antisense lncRNAs would reveal new perspectives Statistical analysis on the molecular basis of gliomagenesis. Patients were divided into high-expression and low- In this study, we profled diferentially expressed expression groups based on WDFY3-AS2 level using the antisense lncRNAs in difuse glioma. Combined with median value. Kaplan–Meier with 2-sided log-rank test prognosis and tumor progression, we identifed a new was used to evaluate the OS diferences between these two antisense lncRNA WDFY3-AS2 with length of 3383 nt groups. Chi square test was performed to detect diference Wu et al. Cancer Cell Int (2018) 18:107 Page 3 of 10 Table 1 Clinical characteristics of glioma patients clinicopathological features of the selected patients Characteristics n are listed in Table 1. Out of the whole transcriptome mRNAs, 420 antisense lncRNAs were identifed. SAM Age with R package “samr” was performed to detect the dif- 43 166 ≤ ferentially expressed antisense lncRNAs between LGG > 43 143 and GBM. Totally, 70 antisense lncRNAs were down- Gender regulated in GBM samples compared with LGG samples, Male 194 while 99 antisense lncRNAs were upregulated in GBM Female 115 samples. Volcano plot was used for showing the antisense Subtype lncRNAs of diferential expression (Fig. 1a). Hierarchical Classical 69 clustering analysis conducted with R package “pheatmap” Mesenchymal 65 showed systematic variations in expression of these 169 Proneural 99 antisense lncRNAs among samples (Fig. 1d). Considering