Bioinformatics Analysis of Potential Therapeutic Targets and Prognostic Biomarkers amid CXC Chemokines in Ovarian Carcinoma Microenvironment

Background Ovarian cancer (OC) is one of the leading lethal gynecologic cancers of women around the world. More than 70% of patients are diagnosed with stage III or IV with poor outcome. This is partly because of lacking early effective screening techniques and potential biomarkers of OC. CXC chemokines in tumor microenvironment (TME) and their interaction with relative receptors can excite the downstream signaling pathways to influence tumor progression. However, the role of CXC chemokines in OC has not been identified. Methods ONCOMINE, GEPIA, Kaplan–Meier plotter, cBioPortal, TIMER, Metascape, and LinkedOmics were applied in our study. Results The transcriptional levels of CXCL1/8/9/10/11/12/13/14/16/17 were significantly elevated while CXCL3 was obviously reduced in OC vs normal ovarian tissue. CXCL8/9/11/13 were correlated with clinic pathological stage. Patients with low expression of CXCL8/9/11/13 were associated with better prognosis. We also found that CXCL3 and CXC12 could be used as potential prognostic markers of OC through Kaplan–Meier plotter. Patients with high expression of CXCL3/12 had a significantly better prognosis. Their functions focus on locomotion, signaling, response to stimulus, undergoing the process of multiorganism, immune system, biological regulation, etc. The differentiated CXC chemokines mainly participate in cytokine-cytokine receptor interaction, chemokine signaling pathway, IL-17 signaling pathway, and toll-like receptor signaling pathway. Our results showed that CXC chemokines were highly correlated with infiltration of immune cells. The kinase targets of differentially expressed CXC chemokines are mainly in ATM, LYN, LCK, PLK1, FYN, CDK2, and ATR. Conclusions Our results may provide a new insight for selecting precision biomarkers of targeted therapy of OC.


Introduction
Ovarian cancer is the most leading lethal gynecological cancer around the world. Patients with epithelial ovarian cancer (EOC), the most common pathological type of OC, are always diagnosed at late stages with poor five-year survival rate (FIGO stage III (51%) or IV (29%)) [1,2]. Actually, cytoreductive/primary debulking surgery (PDS) followed by combined chemotherapy is an effective way to deal with early stage (stages I-IIA) patients, with a 5-year survival rate at around 90%. However, more than 70% of patients are diagnosed with advanced stages III-IV with poor outcome [3]. CA125 is still utilized as a traditional marker to diagnose and prognose this disease. Although targeted therapies, immunotherapies, and other combination therapies have showed their merits, it is far more sufficient to find more biomarkers and therapeutic targets to improve our early diagnostic standard and ameliorate the prognosis of OC patients.
Intricate cross-talk between cancer cells and other cells in TME exerts potential influence in neoplasia and tumor progression. Cytokines secreted by cancer cells and other TME cells exert significant influence on this orchestrated interaction. Chemokines (chemotactic cytokines), comprising pairs of ligands and their associated receptors, are a superfamily of low molecular weight cytokines. On the basis of their structure, chemokines are classified as C, CC, CXC, or CX3C, where X represents a nonconserved amino acid substitution [4]. Nearly 50 cytokines of human are divided into two functional groups: inflammatory chemokines and homeostatic chemokines. e primary function of chemokines is mediating cell trafficking and recruiting proinflammatory cells to the sites of inflammation [5]. Chemokines affect tumor progression via several aspects, including angiogenesis, proliferation, migration, invasion, and metastasis [6][7][8]. While chemokines may play a dual role in tumor development, some may favor tumor growth while some may favor antitumor immunity [5]. As their intriguing roles in tumors, chemokines represent an area of intense interest and study.
Similar to other tumor types, chronic inflammation is an important condition for ovarian cancer progression. e CXC group works as a key mediator in inflammation of OC. ey are responsible for the recruitment and activation of immune cells in the inflammatory milieu; in addition, they mediate pro-and antiangiogenic effects. Chemokines are gaining importance in the field of ovarian cancer for being angiostatic and angiogenic in function [9]. Although several potential roles of chemokines in OC have been investigated, identifying precision therapy targets remains destitute. Recently, bioinformatics analysis of certain gene family in malignant tumors based on public databases has been emerging for us to delve into more practical information which may be applied in clinic therapies in the future [10][11][12]. Herein, in our study, we took advantages of public databases to comprehensively analyze the expression CXC family members in OC and further to evaluate their diagnostic and prognostic value to find more appropriate biomarkers applied in clinic.

Materials and Methods
2.1. ONCOMINE Analysis. ONCOMINE (http://www. oncomine.org) dataset is an online web-based cancer microarray database [13]. e mRNA expression of CXC chemokines in ovarian cancer was compared with normal controls. In our study, the cutoff of p value and fold change were set as 0.01 and 2, respectively. Student's t-test was used to analyze the differential expression of CXC in OC.
2.2. GEPIA Dataset. GEPIA (http://gepia.cancer-pku.cn) is an interactive web server applied in analyzing the mRNA sequencing data based on 9763 tumors and 8587 normal samples in the Cancer Genome Atlas (TCGA) and Genotype-Tissue Expression dataset project (GTEx). GEPIA provides customizable functions such as differential expression analysis, profiling according to cancer types and pathological stages, survival analysis, similar gene detection, and dimensionality reduction analysis [14]. In this study, we delved into a differential mRNA expression analysis of tumor and normal tissues, pathological stage analysis of CXC in OC via the module "Single Gene Analysis" of GEPIA.

2.3.
e Kaplan-Meier Plotter Analysis. Kaplan-Meier plotter (http://www.kmplot.com) is an online database, including microarray gene expression data and survival information from Gene Expression Omnibus, TCGA, and the Cancer Biomedical information of 2190 OC patients [15]. OC patients were divided into two groups by median expression (high vs low) to analyze overall survival (OS), progression-free survival (FPS), and postprogression survival (PPS) through Kaplan-Meier plotter, with the hazard ratio (HR) of 95% confidence interval (Cis) and log rank p value.

TCGA and CBioPortal Analysis.
e CBioPortal (http:// www.cbiportal.org/) provides information of complex cancer genomics and clinical profiles from 105 cancers in TCGA database [16]. In this study, the genetic alteration, coexpression, and the network module of CXC chemokines were analyzed from CBioPortal. 606 EOC samples (TCGA, Firehose legacy) were analyzed. mRNA expression z scores (RNA Seq V2 RSEM) were obtained using a z score threshold of ±2.0. Protein expression z scores (mass spectrometry by CPTAC) were obtained using a z score threshold of ±2.0.
2.5. TIMER. TIMER (http://cistrom.shinyapps.io/timer/) is a database which can systematically describe infiltration of different immune cells along with certain genes and their clinic influence [17]. In our study, we used "Gene" module to evaluate correlation of CXC chemokines with infiltration of immune cells, including B cell, CD8+ T cell, CD4+ T cell, macrophage, neutrophil, and dendritic cell. "SCNA" module was used to compare immune infiltration levels in OC with different somatic copy number alterations.
2.6. Metascape. Metascape (http://metascape.org) is a free and reliable tool for gene annotation and enrichment analysis [18]. It is a useful database to analyze common and unique pathways within a group of targeted genes. In our study, Metascape was used to conduct pathway and analyze correlation neighbor genes with CXC chemokines.

2.7.
LinkedOmics. LinkedOmics (http://www.linkedomics. org/) provides comprehensive multiomics data analysis across 32 TCGA cancer types [19]. "LinkInterpreter" module was used to evaluate biological value from kinase target enrichment, miRNA target enrichment, and transcription factor target enrichment of CXC chemokines. Gene Set Enrichment Analysis (GSEA) was performed to analyze a minimum number of genes (size) of 3 and a simulation of 500. Results were analyzed statistically using the Spearman correlation test. e p value cutoff was 0.05.

CXC Chemokines Differentiated Expression in OC.
We first delved into the sixteen CXC chemokines' transcriptional expression levels in OC via ONCOMINE dataset. e results are shown in Figure 1 and Table 1. e transcriptional levels of CXCL1, CXCL8, CXCL10, CXCL11, CXCL12, CXCL13, and CXCL14 were significantly elevated while CXCL3 was obviously reduced in ovarian cancer vs normal ovarian tissue. ese data were consistent with the research dataset of Welsh and Yoshihara who demonstrated that CXCL3 was remarkably decreased in ovarian serous adenocarcinoma or ovarian serous surface papillary adenocarcinoma compared with the normal tissue [20]. Adib et al. also found CXCL1 (p � 0.004) in OC was increased with a fold change of 2.735 [21]. e research of Lu dataset showed that CXCL8 was elevated in ovarian mucinous adenocarcinoma compared with that in normal ovarian tissue [22]. Welsh [23]. e same results of CXCL13 were supported by Bonome dataset [24]. e fold change of CXCL14 was 2.762 (p � 3.21E − 5), 3.372 (p � 0.002), 2.683 (p � 0.035), and 2.149 (p � 6.37E − 6) in the dataset of Hendrix [25], Lu [22], and Bonome [24], respectively.
We then assessed the transcriptional levels of CXC chemokines and their correlation with pathological stage of clinic in the database of GEPIA. Our results showed that CXCL1, CXCL8, CXCL9, CXCL11, CXCL12, CXCL16, and CXCL17 were remarkably elevated in ovarian cancer compared to those in normal tissue ( Figure 2). Also, we found CXCL8 (p � 0.018), CXCL9 (p � 0.020), CXCL11 (p � 0.039), and CXCL13 (p � 0.029) were highly correlated with clinic pathological stages ( Figure 3). e expression of

e Prognostic Value of CXC Chemokines in OC Patients.
We then analyzed the prognostic value of differentially expressed CXC chemokines in OC through the Kaplan-Meier plotter database. Overall survival (OS) curves are shown in Figure 4. Data suggested that OC patients with high expression of CXCL1 (p � 0.02), CXCL3 (p � 0.0034), CXCL9 (p � 0.0017), CXCL11 (p � 0.004), and CXCL13 (p � 0.0012) were significantly correlated with long OS. However, low expression of CXCL12 (p � 3e − 7) and CXCL14 (p � 0.0052) was associated with long OS. e differential expressions of CXC chemokines of progressionfree survival (PFS) and post-progression survival (PPS) were also assessed ( Figures 5 and 6). We found that high expression of CXCL1, CXCL3, CXCL10, and low expression of CXCL11, CXCL12, CXCL14, CXCL16, and CXCL17 were remarkably correlated with long PFS, while high expression of CXCL3, CXCL9, CXCL10, CXCL12, and CXCL13 was shown with long PPS. Among these, CXCL13 (p � 0.00013) was significantly correlated with long PPS. However, low expression of CXC12 (1.8e − 05) was relevant to long PPS.     Our research suggested that CXCL3 and CXC12 could be used as potential prognostic markers of OC.

Gene Alteration, Coexpression, and Prognostic Value of Alterative CXC Chemokines in Patients with OC.
e TCGA and CBioPortal were used to analyze the gene alteration rate of CXC chemokines in OC. Among those differentially expressed CXC chemokines, our results showed that 5% of CXCL3, CXCL8, and CXCL11 were altered, while 4% of CXCL12 and CXCL14 were mutated. 3% of CXCL16 and 7% of CXCL17 suffered with gene alteration (Figure 7(a)). We then delved into the correlation of potential coexpression of CXC chemokines in OC; there was a moderate-to-high correlation among the expression of CXCL1, CXCL3, and CXCL8 within the differentiated genes, and a high correlation among CXCL9, CXCL10, and CXCL11 (Figure 7(b)), while a low correlation of CXCL12, CXCL13, CXCL14, CXCL16, and CXCL17 was detected (Figure 7(b)). We next explored the prognostic value of altered CXC chemokines in OC; we found there was no statistically significant correlation of overall survival and disease-free survival (Figures 7(c) and 7(d)).

Functional Enrichment Analysis of CXC Chemokines in OC Patients
. GEPIA and Metascape were utilized to analyze the functions, pathway enrichment, and their neighboring genes of differentially expressed CXC chemokines in OC (Figure 8). We delved into the top 10 associated genes of each differentiated CXC chemokine via GEPIA dataset (Table 2). e top 11 GO enrichment items of those differentiated CXC chemokines focused on locomotion, signaling, and response to stimulus. Also, they undergo the process of multiorganism, immune system, biological regulation, etc (Figures 8(a) and 8(b) and Table 3). KEGG pathway enrichment analysis represented that cytokinecytokine receptor interaction, chemokine signaling pathway, IL-17 signaling pathway, and toll-like receptor signaling pathway were significantly involved in the tumorigenesis and pathogenesis of OC (Figures 8(c) and 8(d) and Table 4). Moreover, to better understand the relationship between CXC chemokine family members and OC, we then performed a Metascape protein-protein interaction (PPI) enrichment analysis. e PPI network and MCODE components are shown in Figures 8(e) and 8(f ). Data showed that the biological functions of CXC chemokines are mainly enriched in CXCR chemokine receptor binding, chemokine activity, and chemokine receptor binding in OC.

Kinase Targets and mRNA Targets of CXC Chemokines in
Patients of OC. We then analyzed the kinase targets and mRNA targets of differentially expressed CXC chemokines from LinkedOmics database ( Table 5). PLK1 and ATM were the top two targets in the CXCL1 kinase target network. e targets of CXCL3 were ATM and CDK2. LCK, ATM and ATR, ATM were the top two kinase targets in the CXCL8 and CXCL14, respectively. LCK and FYN were the targets of CXCL9 and CXCL13. LCK and LYN were the top two targets of CXCL10, CXCL11, CXCL12, CXCL16, and CXCL17 kinase target networks. Similarly, we explored the enriched mRNA targets from LinkedOmics database (results presented in Table 6).  Table 6.
As CXC chemokines and their receptors may be potentially used in molecular targeting of cancers, accumulating researches have focused on CXC chemokine family and their receptors in ovarian cancer. For instance, researchers found that tumor-suppressor miRNA-27b-5p regulated the growth and metastatic behaviors of ovarian carcinoma cells by targeting CXCL1 [27]. CXCL4 insufficiency was involved in specific inflammatory microenvironment of ovarian cancers arising in endometriosis [28]. Recombinant CXCL8 (rIL-8) attenuated si-JMJD2A-suppressed malignancy of OC cells(OCC) and CXCL8 induced proliferation of OCC in 3D Spheroids [29,30]. In a mouse model of high-grade serous ovarian cancer, CXCL10 altered the tumor immune microenvironment and facilitated disease progression [31]. Researchers also found that upregulated CXCL14 and CXCL16 were associated with poor survival outcomes and promoted ovarian cancer cells proliferation [32,33]. In the progression of OC, higher CXCL17 correlated with higher expression of B7-H4 [34]. All these researches showed the importance of CXC chemokines in the progression of OC. However, the patterns of expression and exact roles of CXC chemokine family in OC are obscure. In our study, we systematically explored the expression patterns, prognostic values, gene alteration, coexpressions, correlation, infiltration of immune cells, potential functions, pathways, and kinase/mRNA targets of differentially expressed CXC chemokines in OC.
ese results were in accordance with previous researches; CXCL14 and CXCLL16 were highly expressed in OC and they were associated with poor survival outcomes and promoted OCC proliferation [32,33]. Also, we found * * * * * OV  that CXCL8/9/11 and CXCL13 were highly correlated with clinic pathological stage. Patients with low expression of CXCL8/9/11/13 were associated with better prognosis. Moreover, we utilized the Kaplan-Meier plotter to find the potential prognostic markers of OC. We found patients with high expression of CXCL3 and CXCL12 had a significantly better prognosis. Herein, CXCL3 and CXCL12 may be used as novel prognostic markers of OC in the future. en, we analyzed the genetic alterations and coexpression of differentially expressed CXC chemokines via TCGA and CBioPortal database. e differentially expressed CXC chemokines all contained a certain proportion of   16 Journal of Oncology genetic alteration while the altered genes did not affect the prognosis of OC. We found a low-to-high correlation among these differentially expressed genes; it suggested that these CXC chemokines may play a synergistic role in OC progression. Furthermore, we used GO and KEGG pathway enrichment analysis to identify the main functions and pathways of differentially expressed CXC chemokines in OC. eir functions focused on locomotion, signaling, response to stimulus, undergoing the process of multiorganism, immune system, biological regulation, etc. e cytokine-cytokine receptor interaction, chemokine signaling pathway, IL-17 signaling pathway, and toll-like receptor signaling pathway were the vital pathways that participated in OC. For instance, researchers found that cancer-associated fibroblasts (CAFs) in TME induced epithelial-mesenchymal transition (EMT) and cisplatin resistance in OC via CXCL12/CXCR4 axis [35]. In breast cancer, IL-17-CXCR2 axis facilitated the recruitment of neutrophils to the tumor sites, thus allowing them to play a cancer-promoting role in cancer progression. [36]. We then investigated the kinase targets of differentially expressed CXC chemokines; the kinase targets were mainly in ATM, FYN, LYN, LCK, PLK1, CDK2, and ATR. ese results may provide us with potential therapeutic targets in OC.
Recently emerging data suggested that immune cell infiltration plays a key role in tumor progression [37]. Our research delved into infiltration of six immune cells (B cells, CD8 + T cells, CD4 + T cells, macrophages, neutrophils, and dendritic cells) that were correlated with differentiated CXC chemokines.
e results indicated that CXC chemokines could be immune regulators in OC.
All in all, our research took advantages of public databases to systematically delve into data of CXC chemokines in OC. However, our results may need to be verified by indepth experiments in vivo and in vitro. Multiple clinical trials are needed to validate potential biomarkers of CXC chemokines. We hope our results can provide novel insights for our researchers and these potential targets could be applied in clinic someday.

Conflicts of Interest
e authors declare no conflicts of interest.

Authors' Contributions
Hua Jiang and Qinghua Xi designed the study and supervised the writing of the manuscript. Yunfeng Jin and Qiwang Lin performed data analysis work and wrote the manuscript. He Fei, Lujie Xue, and Li Li helped edit the manuscript. Yunfeng Jin and Qiwang Lin contributed equally to this work.