Network analyses of circular RNAs expression profiles and their hub genes in pancreatic ductal adenocarcinoma

DOI: https://doi.org/10.21203/rs.2.15313/v1

Abstract

Pancreatic ductal adenocarcinoma (PDAC) is one of the most common malignant tumor in digestive system. CircRNAs involve in lots of biological processes through interacting with miRNAs and their targeted mRNA. We obtained the circRNA gene expression profiles from Gene Expression Omnibus (GEO) and identified differentially expressed genes (DEGs) between PDAC samples and paracancerous tissues. Bioinformatics analyses, including GO analysis, KEGG pathway analysis and PPI network analysis, were conducted for further investigation. We also constructed circRNA‑microRNA-mRNA co-expression network. A total 291 differentially expressed circRNAs were screened out. The GO enrichment analysis revealed that up-regulated DEGs were mainly involved metabolic process, biological regulation, and gene expression, and down-regulated DEGs were involved in cell communication, single-organism process, and signal transduction. The KEGG pathway analysis, the upregulated circRNAs were enriched cGMP-PKG signaling pathway, and HTLV-I infection, while the downregulated circRNAs were enriched in protein processing in endoplasmic reticulum, insulin signaling pathway, regulation of actin cytoskeleton, etc. Four genes were identified from PPI network as both hub genes and module genes, and their circRNA‑miRNA-mRNA regulatory network also be constructed. Our study indicated possible involvement of dysregulated circRNAs in the development of PDAC and promoted our understanding of the underlying molecular mechanisms.

Introduction

Pancreatic ductal adenocarcinoma (PDAC), which is the most common form of pancreatic carcinoma, remains the fourth highest mortality rate malignant neoplasms in the United States. More specifically, its estimated deaths will be 43,090 in 2017 [1].

The prognosis of PDAC is poor. It is a highly metastatic tumor and therapy resistant. Its survival rate is approximately 25% of 1-year and 5% of 5-year [2]. Beside the dietary habit and environmental factors, PDAC is also related with genetic basis. The family history of pancreatic carcinoma is predisposing to increased risk in direct relatives [3].

The early diagnostic is very hard due to lack of characteristic symptoms and cancer late stage during first oncological visit. It is usually not adequate to treat this disease by surgical resection [4]. There are still need potent specific biomarkers for early diagnosis. System analyses of PDAC may provide better understanding to this disease and contribute to developing more effective approaches.

Circular RNAs (circRNAs) was first declared in 1976 [5]. However, circRNAs were almost completely neglected since then, because it is difficult to be detected using traditional molecular techniques, and hard to be mapped into genomes [6]. With the development of sequencing technologies and bioinformatics methodologies, circRNAs have become the focal research in the biological field [7].

CircRNAs are unusually regarded as stable molecules. Their 3’ and 5’ ends are linked together covalently [8], and thus lack of a free end, which prevents them depredated as the conventional RNA. The special features of circRNA, such as stability, abundance, and tissue-specific expression pattern, make them very attractive for clinical research, as they can be employ as new ways in diagnosis [4].

CircRNAs can repress the function of microRNA (miRNA) by binding to miRNA as sponges [9]. They involve in lots of biological processes by interacting with miRNAs and their target mRNA. CircRNAs have been reported in several types of diseases, including Alzheimer’s disease [10], colorectal cancer [11], atherosclerotic vascular disease [12], and osteoarthritis [13].

To explore the underlying molecular regulation mechanism of circRNAs in PDAC, circRNAs microarray was used to detect the differential expressed circRNAs between PDAC tissues and paracancerous tissues. Our results suggested that the dysregulated expression of circRNAs may play a role in transformation of PDAC.

Materials and Methods

2.1. Microarray data

The circRNA expression profiles were obtained from Gene Expression Omnibus (GEO), where the accession number is GSE79634. It contains 40 samples, including twenty fresh-frozen PDAC samples and twenty paracancerous tissues. The circRNA expression levels of these samples were measured using Agilent–069978 Arraystar Human circRNA microarray V1. Total RNA was extracted from collected samples. After labeling and hybridizing, the acquired array images were analyzed using Agilent Feature Extraction software (version 11.0.1.1). Quantile normalization and subsequent data processing were performed using the R software package.

2.2. Identification of differentially expressed genes

Differentially expressed genes (DEGs), which were significantly differentially expressed among the PDAC samples and the paracancerous tissues, were identified by using the LIMMA package in R language. LIMMA provides several p-value adjustment options to correct for the occurrence of false positive results. It was also known as multiple-testing corrections, where the Benjamini & Hochberg false discovery rate method was selected by default in the package. We defined adjusted p-values ≤0.05 and absolute value of fold change ≥ 2 as the cutoff to determine the statistically significance of gene expression difference.

2.3. GO enrichment analysis

Gene Ontology (GO) is a common useful framework for the model of biology. The GO project provides a computational representation of our evolving knowledge on how genes encode biological functions. It has defined formal ontologies that represent over 40,000 biological concepts at the molecular, cellular and system levels, to describe gene function and relationships between them.

2.4. KEGG pathway analysis

Kyoto encyclopedia of genes and genomes (KEGG) database resource is an encyclopedia of genes and genomes for understanding high-level functions and utilities of the biological system from molecular-level information, especially large-scale molecular datasets generated by genome sequencing and other high-throughput experimental technologies. It assigns functional meanings to genes and genomes both at the molecular and higher levels. Molecular level functions are maintained in the KO (KEGG Orthology) database, while higher-level functions are defined in the forms of KEGG modules, KEGG pathway maps and BRITE hierarchies [14].

2.5. PPI network analysis

It is widely accepted that the interactions and partnerships between proteins are play important roles in biological regulation. The protein–protein interaction networks, which are constructed from either known or predicted interactions, provide us with topological summary of our research subjects.

A larger number of online resources contain this protein–protein interaction information. These interactions were generated either from experimental techniques [15, 16] or from computational prediction [17–19]. A group of databases integrate both known and predicted interactions, to provide much high comprehensiveness and coverage. Typical databases include GeneMANIA [20], FunCoup [17], I2D [21], and STRING [22].

Highly complex biological functions are often undertaken by tightly interacted groups of proteins. PPI network was constructed in Cytoscape, and the significant modules were identified using the plug-in Molecular Complex Detection (MCODE).

2.6. Target prediction and co-expression network

To further investigate the functional roles of circRNA, the miRNAs regulated by these circRNA and corresponding target genes were extracted from CIRCNET database. The circRNA-microRNA-mRNA co-expression networks were constructed based on this information by using Cytoscape.

Results

3.1. Differentially expressed genes

Differentially expressed genes (DEG) were identified from microarray data, using the criteria of adj. P. Value<0.05 and |log2FC|>1. A total of 291 circRNA genes were screened out between the PDAC tissues and paracancerous tissues after preprocessing of microarray data, including 128 up-regulated circRNAs and 163 down-regulated circRNAs. The expression levels of these dysregulated expression circRNAs were distinguishable and variations, which were demonstrated from the volcano plot (Figure 1A) and heatmap (Figure 1B). The top 10 up-regulated and down-regulated circRNAs were listed in Table 1. For example, hsa_circ_0006220 (TADA2A), hsa_circ_0000977 (NOL10) and hsa_circ_0001666 (FAM120B) were up-regulated with top magnitudes. Meanwhile, hsa_circ_0000518 (RPPH1) hsa_circ_0072088 (ZFR) and hsa_circ_0005273 (PTK2) were down-regulated with top magnitudes.

3.2. Enrichment analysis of DEGs

The Gene Ontology (GO) enrichment analyses were performed on these 128 upregulated and 163 downregulated circRNAs respectively. Table 2 was the summary of GO enrichment analysis results, which listed the number of enriched GO terms at different significant level (P<0.10 or P<0.05). Table 3 and Table 4 were top 10 GO terms (P<0.05), respectively. GO enrichment analysis revealed that numerous genes were involved in the biological processes. For example, according to the gene ontology enrichment analysis results, the up-regulated circRNAs were enriched (P <0.05) in 80 biological processes, including biological regulation, metabolic process, and gene expression, etc. Meanwhile, the downregulated circRNAs were enriched (P <0.05) in 151 biological processes, including single-organism process, cell communication, and signal transduction, etc. These processes were associated with PDAC.

3.3. KEGG pathway analysis

KEGG pathway analysis results were listed in Table 5. The cutoff used for this analysis was P<0.05. According to these results, the upregulated circRNAs were enriched in 2 pathways including cGMP-PKG signaling pathway, and HTLV-I infection, while the downregulated circRNAs were enriched in 10 pathways, including Protein processing in endoplasmic reticulum, insulin signaling pathway, regulation of actin cytoskeleton, etc.

3.4. PPI network analysis and hub genes screening

The PPI (protein-protein interaction) network of the dysregulation genes were constructed by Cytoscape software, where the interaction information was extracted from the STRING database. As shown in Figure 2A, the red triangle nodes represent upregulated genes and green circular nodes represent downregulated genes, respectively.

The PPI network was further analyzed by screening its hub genes and identifying significant modules. In the PPI networks, the nodes with high degree are defined as hub genes. The top 10 hub genes were ACACA, ACACB, GSK3B, LRRK2, PRKG1, UBQLN1, FOXO1, AXIN1, MTHFD1L and PTK2. The most significant hub proteins in the PPI network were ACACA (degree = 19) and ACACB (degree = 18). We also analyzed the PPI network using MCODE, which is a Cytoscape plug-in that finds clusters (or modules, highly interconnected regions) in a network. In PPI network, clusters are often protein complexes and parts of pathways. There were 3 modules identified in this PPI network (Figure 2A). The most significant modules contained ACACB, ACACA, AACS, ALDH7A1 and MTHFD1L. The other two modules contained GNAI2, MAP2K2, LRRK2, CAP1, EPS15L1, and EPS15.

3.5. CircRNA-microRNA-mRNA co-expression network

As shown in Table 6, there were four genes identified as both hub genes and module genes in the PPI network analysis, including ACACA, MTHFD1L and LRRK2. The circRNA-microRNA-mRNA co-expression networks were constructed based on these genes.

The miRNAs regulated by these circRNA and corresponding target genes were extracted from CIRCNET database. Due to limited space, no more than five top significant regulated miRNAs of each circRNA were reserved for the construction of co-expression network, and so do their target genes.

Figure 3 (A-C) are the circRNA-microRNA-mRNA networks of ACACA, LRRK2 and MTHFD1L, respectively. The purple nodes at the center of each network were the circRNA genes to be analyzed. The red nodes were the corresponding circRNA, the green nodes were miRNAs, and the light blue nodes were regulated gene. The purple edges linked circRNAs and their transcripts. The red edges linked transcripts and their regulated miRNAs. The gray edges linked miRNAs and their target gene.

According to Figure 4A, different transcripts of a specific circRNA could regulated the same miRNA, and different miRNA could regulate the same target gene. The more the transcripts, the more likely the co-regulation exists, and the more complex the network. For example, in the co-expression network of ACACA, both circ-ACACA.171 and circ-ACACA.75 regulated three common miRNAs, i.e. miR–137, miR–338–3p, and miR–376a–5p. Meanwhile, both miR–338–3p and miR–1288–3p regulated SLFN12L.

For more complex regulatory networks with multiple circRNA, different circRNAs could co-regulate the same miRNA. For example, there are two circRNA in Figure 4 miR–508–5p is regulated by two transcripts of ACACB (circ-ACACB.74 and circ-ACACB.25) and one transcript of MTHFD1L (circ-MTHFD1L.11).

In the circRNA-miRNA-mRNA regulatory network, there also existed closed loops. That is, circRNA regulated miRNA through its specific transcript, and miRNA may also regulate circRNA in turn, thus forming a closed loop structure. For example, in Figure 4, ACACB regulated miR–6849–3p through circ-ACACA.15, and miR–6849–3p in turn inhibits the expression of ACACB, thereby forming a closed loop.

Discussion

Pancreatic ductal adenocarcinoma is an aggressive malignant neoplasm with extremely high mortality rate, mainly because of difficult early diagnostic and poor prognosis. Current available therapeutic methods are almost ineffective [23]. The potential improvement of the diagnosis may exist in the strictly molecular biomarkers [24]. CircRNA is a new special kind of RNAs. Recent evidences revealed that circRNAs can function as miRNA sponges and affect disease by regulating gene expression.

In the present study, we employed bioinformatics technology to investigate the underlying molecular mechanism of PDAC. Differentially expressed profiles of circRNAs in PDAC tissues were observed and validated compared with the paracancerous tissues, indicating possible involvement of these dysregulated circRNAs in the development of PDAC.

By using microarray data, we observed a total of 291 circRNAs differentially expressed between PDAC tissues and paracancerous tissues, including 128 up-regulated circRNAs and 163 down-regulated circRNAs. Among these differentially expressed genes, hsa_circ_0006220, hsa_circ_0000977 and hsa_circ_0001666 were upregulated with top magnitudes, while hsa_circ_0000518, hsa_circ_0072088 and hsa_circ_0005273 were downregulated with top magnitudes. The differential expression levels of these genes may be related to the involvement in the transcription level regulation on PDAC.

Cumulative evidence has demonstrated that co-expression genes usually participate in similar biological process. In order to better understand the interactions of these differentially expressed genes, we further applied GO and KEGG pathway analysis on them.

The GO enrichment analysis results revealed that up-regulated DEGs were mainly involved metabolic process, biological regulation, and gene expression, and down-regulated DEGs were involved in cell communication, single-organism process, and signal transduction.

As is well-known, the symptoms of the patients with cancer, such as cachexia, wasting syndrome, is related to metabolic disorders. Recent studies showed that, in the case of caloric deficiency, tumor induced changes in host metabolism, leading to tumor immunosuppression. For example, in pre-cachexia mice with PDA, IL6 reduced liver ketogenic ability by inhibiting PPAR alpha. When these mice challenged with caloric deficiency, it would lead to hypoketonemia, and induce the elevation of glucocorticoid levels. A variety of tumor immune pathways was inhibited by above stress response, and thus promoted the progression of pancreatic cancer [25]. Furthermore, poor survival rates in patients with PDAC were attributed to abnormally high levels of metastasis. Study showed that the metabolic reprogramming of tumor infiltrating macrophages played a key role in PDAC metastasis [26]. In addition, autophagy is required to maintain energy homeostasis by degrading unnecessary cellular components and molecules. In recent years, it has been found that autophagy can regulate the metabolism of tumor. Autophagy plays a crucial role in the maintenance of tumor survival. PDAC cells require autophagy and glutamine transporters to maintain intracellular glutamine levels [27].

Therefore, we hypothesized that the high expressed circRNAs may promote the metastasis of pancreatic cancer through various metabolic process, such as, stress response, metabolic reprogramming and glutamine metabolism.

PDAC is a highly metastatic tumor with poor prognosis, and intercellular communication is the key to metastasis [28]. Research showed that PDAC-derived exosomes induced liver pre-metastasis niche formation in nude mice, and increased the load of liver metastasis. Kupffer cells uptaked PDAC-derived exosomes, resulting in transforming growth factor β (TGF-β) and upregulation of fibronectin production by hepatic stellate cells. This fibrotic microenvironment increases the number of bone marrow-derived macrophages. The authors found that macrophage migration inhibitory factor (MIF) was highly expressed in PDAC derived exosomes, and MIF promote metastasis, and liver metastasis before niche formation. It can be seen that the MIF of the exocrine body is initially transferred to the liver, which may be a prognostic indicator of PDAC liver metastasis [28].

The low expression of circRNAs might be a tumor suppressor gene, which inhibit tumor pre-metastasis niche formation and tumor metastasis. Therefore, we can take this as the target for prevention and control, promote the expression of these circRNAs, block the cell mediated intercellular communication, and prevent the metastasis of pancreatic cancer.

According to the KEGG pathway analysis, the upregulated circRNAs were enriched cGMP-PKG signaling pathway, and HTLV-I infection, while the downregulated circRNAs were enriched in 10 pathways, including protein processing in endoplasmic reticulum, Insulin signaling pathway, regulation of actin cytoskeleton, etc. Recent evidence indicated that both cGMP-PKG and Insulin were play important roles in PDAC.

Our KEGG analysis results show that a large number of down-regulated circRNA of pancreatic cancer were enriched in the pathway of Protein processing in endoplasmic reticulum. Similarly, Peng et al. used the KEGG analysis and found that many endoplasmic reticulum (ER)-associated proteins were altered in solid pseudopapillary tumor of the pancreas (SPTP), suggesting that endoplasmic reticulum stress may play an important role in SPTP tumorigenesis. They also verified 7 proteins in the pathway by immunohistochemistry, including ERO1LB, TRIM1, BIP, SEC61B, P4HB, GRP94 and, PDIA4. Six of these proteins (except SEC61B) were consistent with the LC-MS/MS results [29].

Our KEGG analysis results also showed that the down regulated circRNAs of pancreatic cancer were enriched in insulin signaling pathway. It has been reported that hyperinsulinemia was a clear risk factor for pancreatic cancer, and an increase in hyperinsulinemia associated with obesity and type 2 diabetes heralded an increased incidence of cancer. The role of insulin in pancreatic cancer progression was worth to be further studied [30]. They suggested that the overexpression of insulin signaling contributes to the survival and proliferation of human immortalized pancreatic ductal cells and metastatic pancreatic cancer cells, but does not affect normal pancreatic ductal cells. They also suggested that cell survival associated insulin pathway may be involved in the progression of pancreatic cancer [30]. There were also literatures about insulin signaling pathway and chemotherapeutic resistance of pancreatic cancer. Ireland et al. found that tumor associated macrophages (TAM) and myofibroblasts activated insulin /IGF receptor by secreting IGF1–1 and 2, and directly promote chemoresistance of pancreatic cancer cells resistance. Analysis of biopsies from patients with pancreatic cancer showed that 72% of the patients’ tumor cells had expressed activated insulin / IGF receptor. Therefore, inhibition of IGF might contribute to the treatment of PDAC patients with chemotherapy resistance, and activation of insulin /IGF1R might be used as a biomarker for the identification of chemotherapy resistance [31]. In addition, adaptor protein CrkII played an important role in the proliferation and invasion of some malignant tumors. However, the specific mechanism of insulin like growth factor 1 (IGF–1) -CrkII signal induced proliferation of PDAC was not clear. Studies have further demonstrated that the CrkII in insulin signaling pathway mediated IGF–1 signaling and affected PDAC through Erk1/2 and Akt pathway, suggesting that CrkII gene and protein could be used as an effective therapeutic targets for PDAC [32].

We further applied PPI network analysis to these aberrant expression genes. Ten hub genes with the top degree were obtained from the PPI network. We also identified three top significant modules by clustering. A total of eleven module genes were contain in these three module. Four genes were existed both as hub genes and module genes, including ACACA, MTHFD1L and LRRK2.

We predicted the target genes of these four circRNA genes, and constructed circRNA-miRNA-Gene regulatory networks for them. On one hand, These genes regulated the expression of miRNAs by transcribing a variety of circRNAs, and further regulated the target genes through miRNAs. According to these regulatory networks, the more circRNAs, the more complex the regulation of miRNA and target genes. There existed multiple miRNAs regulated by the same circRNA. There also existed the same miRNA regulated by several circRNAs of a specific gene. In addition, circRNAs of different genes might regulate the same miRNA. On the other hand, On the other hand, miRNA also could bind to the target gene, and then regulated circRNAs via negative feedback, thus forming a closed loop. For example, circ-ACACB.74, circ-ACACB.25, and circ-MTHFD1L.101 targeted miR–508–5p simultaneously, and miR–508–5p regulated the expression of ACACB and MTHFD1L through the negative feedback. Therefore, bioinformatic analyses showed that circRNA targeting miR–508–5p might involve in the extremely complex regulation mechanism of PDAC. There exist literatures that miR–508–5p involves in the regulation mechanisms of liver gastric cancer and hepatocellular carcinoma. Shang Y. et al. reported that miR–508–5p play an important role in multidrug resistance (MDR) of gastric cancer by targeting ABCB1 and ZNRD1 [33]. Further research suggested that the underlying mechanism was related to the miR–27b/CCNG1/P53/miR–508–5p axis, and it was possible to reverse MDR of gastric cancer by restoring the expression levels miR–27b and miR–508–5p [34]. Yan H. et al. found out that miR–508–5p was aberrant expressed in hepatocellular carcinoma (HCC) [35]. Wu et al. suggested that miR–508–5p might act as a tumor suppressor of HCC progression by targeting MESDC1 [36]. Similar to gastric cancer and hepatocellular carcinoma, pancreatic adenocarcinoma is also a disease belong to the digestive system. We have reason to believe that miR–508–5 p may also be in pancreatic adenocarcinoma plays an important role. It is worth to study the problem of whether miR–508–5p participates in PDAC.

Declarations

Data Availability

The microarray data used to support the finding of this study have been deposited in the GEO database (dataset ID: GSE79634), which have been cited.

Consent for publication

The authors confirm that written consents have been obtained from the patients, to publish this manuscript.

Competing interests

The authors declare that there are no competing interests in this work.

Authors’ contributions

Yanyan Tang wrote the manuscript and performed the analysis work. Ping Zhang designed this study and revised the manuscript. All the authors have read and approved the final manuscript.

Acknowledgements

This study was supported by grants from the Natural Science Foundation of Hunan Province (2018JJ3634), the Humanities and Social Science Youth Fund of Ministry of Education of China (19YJCZH254), and the Scientific Research Project of the Health and Family Planning Commission of Hunan Province (B20180809).

Reference

1.Siegel RL, Miller KD and Jemal A. Cancer Statistics, 2017. CA: a cancer journal for clinicians. 2017; 67(1):7–30.

2.Chand S, O’Hayer K, Blanco FF, Winter JM and Brody JR. The Landscape of Pancreatic Cancer Therapeutic Resistance Mechanisms. International journal of biological sciences. 2016; 12(3):273–282.

3.Sarnecka AK, Zagozda M and Durlik M. An Overview of Genetic Changes and Risk of Pancreatic Ductal Adenocarcinoma. Journal of Cancer. 2016; 7(14):2045–2051.

4.Li C, Heidt DG, Dalerba P, Burant CF, Zhang L, Adsay V, Wicha M, Clarke MF and Simeone DM. Identification of pancreatic cancer stem cells. Cancer research. 2007; 67(3):1030–1037.

5.Sanger HL, Klotz G, Riesner D, Gross HJ and Kleinschmidt AK. Viroids are single-stranded covalently closed circular RNA molecules existing as highly base-paired rod-like structures. Proceedings of the National Academy of Sciences of the United States of America. 1976; 73(11):3852–3856.

6.Jeck WR and Sharpless NE. Detecting and characterizing circular RNAs. Nature biotechnology. 2014; 32(5):453–461.

7.Chen LL and Yang L. Regulation of circRNA biogenesis. RNA biology. 2015; 12(4):381–388.

8.Lasda E and Parker R. Circular RNAs: diversity of form and function. Rna. 2014; 20(12):1829–1842.

9.Memczak S, Jens M, Elefsinioti A, Torti F, Krueger J, Rybak A, Maier L, Mackowiak SD, Gregersen LH, Munschauer M, Loewer A, Ziebold U, Landthaler M, Kocks C, le Noble F and Rajewsky N. Circular RNAs are a large class of animal RNAs with regulatory potency. Nature. 2013; 495(7441):333–338.

10.Lukiw WJ. Circular RNA (circRNA) in Alzheimer’s disease (AD). Frontiers in genetics. 2013; 4:307.

11.Xie H, Ren X, Xin S, Lan X, Lu G, Lin Y, Yang S, Zeng Z, Liao W, Ding YQ and Liang L. Emerging roles of circRNA_001569 targeting miR–145 in the proliferation and invasion of colorectal cancer. Oncotarget. 2016; 7(18):26680–26691.

12.Burd CE, Jeck WR, Liu Y, Sanoff HK, Wang Z and Sharpless NE. Expression of linear and novel circular forms of an INK4/ARF-associated non-coding RNA correlates with atherosclerosis risk. PLoS genetics. 2010; 6(12):e1001233.

13.Liu Q, Zhang X, Hu X, Dai L, Fu X, Zhang J and Ao Y. Circular RNA Related to the Chondrocyte ECM Regulates MMP13 Expression by Functioning as a MiR–136 ‘Sponge’ in Human Cartilage Degradation. Scientific reports. 2016; 6:22572.

14.Kanehisa M, Furumichi M, Tanabe M, Sato Y and Morishima K. KEGG: new perspectives on genomes, pathways, diseases and drugs. Nucleic acids research. 2017; 45(D1):D353-D361.

15.Orchard S, Kerrien S, Abbani S, Aranda B, Bhate J, Bidwell S, Bridge A, Briganti L, Brinkman FS, Cesareni G, Chatr-aryamontri A, Chautard E, Chen C, Dumousseau M, Goll J, Hancock RE, et al. Protein interaction data curation: the International Molecular Exchange (IMEx) consortium. Nature methods. 2012; 9(4):345–350.

16.Orchard S, Ammari M, Aranda B, Breuza L, Briganti L, Broackes-Carter F, Campbell NH, Chavali G, Chen C, del-Toro N, Duesbury M, Dumousseau M, Galeota E, Hinz U, Iannuccelli M, Jagannathan S, et al. The MIntAct project—IntAct as a common curation platform for 11 molecular interaction databases. Nucleic acids research. 2014; 42(Database issue):D358–363.

17.Schmitt T, Ogris C and Sonnhammer EL. FunCoup 3.0: database of genome-wide functional coupling networks. Nucleic acids research. 2014; 42(Database issue):D380–388.

18.Zhang QC, Petrey D, Garzon JI, Deng L and Honig B. PrePPI: a structure-informed database of protein-protein interactions. Nucleic acids research. 2013; 41(Database issue):D828–833.

19.Baspinar A, Cukuroglu E, Nussinov R, Keskin O and Gursoy A. PRISM: a web server and repository for prediction of protein-protein interactions and modeling their 3D complexes. Nucleic acids research. 2014; 42(Web Server issue):W285–289.

20.Zuberi K, Franz M, Rodriguez H, Montojo J, Lopes CT, Bader GD and Morris Q. GeneMANIA prediction server 2013 update. Nucleic acids research. 2013; 41(Web Server issue):W115–122.

21.Niu Y, Otasek D and Jurisica I. Evaluation of linguistic features useful in extraction of interactions from PubMed; application to annotating known, high-throughput and predicted interactions in I2D. Bioinformatics. 2010; 26(1):111–119.

22.Szklarczyk D, Franceschini A, Wyder S, Forslund K, Heller D, Huerta-Cepas J, Simonovic M, Roth A, Santos A, Tsafou KP, Kuhn M, Bork P, Jensen LJ and von Mering C. STRING v10: protein-protein interaction networks, integrated over the tree of life. Nucleic acids research. 2015; 43(Database issue):D447–452.

23.Ishiguro S, Yoshimura K, Tsunedomi R, Oka M, Takao S, Inui M, Kawabata A, Wall T, Magafa V, Cordopatis P, Tzakos AG and Tamura M. Involvement of angiotensin II type 2 receptor (AT2R) signaling in human pancreatic ductal adenocarcinoma (PDAC): a novel AT2R agonist effectively attenuates growth of PDAC grafts in mice. Cancer biology & therapy. 2015; 16(2):307–316.

24.Olakowski M, Tyszkiewicz T, Jarzab M, Krol R, Oczko-Wojciechowska M, Kowalska M, Kowal M, Gala GM, Kajor M, Lange D, Chmielik E, Gubala E, Lampe P and Jarzab B. NBL1 and anillin (ANLN) genes over-expression in pancreatic carcinoma. Folia histochemica et cytobiologica. 2009; 47(2):249–255.

25.Flint TR, Janowitz T, Connell CM, Roberts EW, Denton AE, Coll AP, Jodrell DI and Fearon DT. Tumor-Induced IL–6 Reprograms Host Metabolism to Suppress Anti-tumor Immunity. Cell metabolism. 2016; 24(5):672–684.

26..!!! INVALID CITATION!!!

27.Nkembo AT, Salako O, Poku RA, Amissah F, Ntantie E, Flores-Rozas H and Lamango NS. Disruption of actin filaments and suppression of pancreatic cancer cell viability and migration following treatment with polyisoprenylated cysteinyl amides. American journal of cancer research. 2016; 6(11):2532–2546.

28.Costa-Silva B, Aiello NM, Ocean AJ, Singh S, Zhang H, Thakur BK, Becker A, Hoshino A, Mark MT, Molina H, Xiang J, Zhang T, Theilen TM, Garcia-Santos G, Williams C, Ararso Y, et al. Pancreatic cancer exosomes initiate pre-metastatic niche formation in the liver. Nature cell biology. 2015; 17(6):816–826.

29.Zhu Y, Xu H, Chen H, Xie J, Shi M, Shen B, Deng X, Liu C, Zhan X and Peng C. Proteomic analysis of solid pseudopapillary tumor of the pancreas reveals dysfunction of the endoplasmic reticulum protein processing pathway. Molecular & cellular proteomics: MCP. 2014; 13(10):2593–2603.

30.Chan MT, Lim GE, Skovso S, Yang YH, Albrecht T, Alejandro EU, Hoesli CA, Piret JM, Warnock GL and Johnson JD. Effects of insulin on human pancreatic cancer progression modeled in vitro. BMC cancer. 2014; 14:814.

31.Ireland L, Santos A, Ahmed MS, Rainer C, Nielsen SR, Quaranta V, Weyer-Czernilofsky U, Engle DD, Perez-Mancera PA, Coupland SE, Taktak A, Bogenrieder T, Tuveson DA, Campbell F, Schmid MC and Mielgo A. Chemoresistance in Pancreatic Cancer Is Driven by Stroma-Derived Insulin-Like Growth Factors. Cancer research. 2016; 76(23):6851–6863.

32.Liu R, Wang Q, Xu G, Li K, Zhou L and Xu B. The adaptor protein CrkII regulates IGF–1-induced biological behaviors of pancreatic ductal adenocarcinoma. Tumour biology: the journal of the International Society for Oncodevelopmental Biology and Medicine. 2016; 37(1):817–822.

33.Shang Y, Zhang Z, Liu Z, Feng B, Ren G, Li K, Zhou L, Sun Y, Li M, Zhou J, An Y, Wu K, Nie Y and Fan D. miR–508–5p regulates multidrug resistance of gastric cancer by targeting ABCB1 and ZNRD1. Oncogene. 2014; 33(25):3267–3276.

34.Shang Y, Feng B, Zhou L, Ren G, Zhang Z, Fan X, Sun Y, Luo G, Liang J, Wu K, Nie Y and Fan D. The miR27b-CCNG1-P53-miR–508–5p axis regulates multidrug resistance of gastric cancer. Oncotarget. 2016; 7(1):538–549.

35.Yan H, Wang S, Yu H, Zhu J and Chen C. Molecular pathways and functional analysis of miRNA expression associated with paclitaxel-induced apoptosis in hepatocellular carcinoma cells. Pharmacology. 2013; 92(3–4):167–174.

36.Wu SG, Huang YJ, Bao B, Wu LM, Dong J, Liu XH, Li ZH, Wang XY, Wang L, Chen BJ and Chen W. miR–508–5p acts as an anti-oncogene by targeting MESDC1 in hepatocellular carcinoma. Neoplasma. 2016; 64(1).

Notes

Gene Expression Omnibus (GEO, http://www.ncbi.nlm.nih.gov/geo/)

Gene Ontology Consortium, http://www.geneontology.org/

Kyoto encyclopedia of genes and genomes (KEGG) http://www.genome.jp/kegg/

Tables

Table 1. Top 10 upregulated and downregulated circRNAs in the PDAC

ID

adj. P. Val

t

B

logFC

Alias

Gene Symbol

ASCRP002386

1.46E-09

8.96494

15.32025

6.301989

hsa_circ_0006220

TADA2A

ASCRP002932

2.24E-08

7.80154

11.75923

6.156615

hsa_circ_0000977

NOL10

ASCRP004540

3.23E-09

8.60686

14.24016

5.730049

hsa_circ_0001666

FAM120B

ASCRP002384

3.05E-09

8.6488

14.36742

5.702157

hsa_circ_0043278

TADA2A

ASCRP004499

1.60E-06

6.19504

6.66086

3.455038

hsa_circ_0078297

MTHFD1L

ASCRP000816

5.33E-07

6.57821

7.8869

3.135898

hsa_circ_0003600

SOX13

ASCRP000706

1.20E-05

5.45481

4.2992

3.004105

hsa_circ_0013912

POLR3C

ASCRP001007

4.85E-05

4.95945

2.74095

2.978418

hsa_circ_0018909

VDAC2

ASCRP002451

1.50E-06

6.21761

6.73311

2.73586

hsa_circ_0044436

KAT7

ASCRP001589

1.33E-06

6.2597

6.86782

2.427013

hsa_circ_0029634

ZMYM2

ASCRP000018

1.63E-09

-8.92144

15.18985

-3.32345

hsa_circ_0000518

RPPH1

ASCRP004099

4.61E-08

-7.50572

10.83219

-2.93726

hsa_circ_0072088

ZFR

ASCRP004952

2.63E-07

-6.84074

8.72501

-2.89755

hsa_circ_0005273

PTK2

ASCRP000343

7.68E-09

-8.24112

13.12182

-2.65125

hsa_circ_0000520

RPPH1

ASCRP000385

6.05E-09

-8.33767

13.41847

-2.47966

hsa_circ_0000511

RPPH1

ASCRP000315

9.34E-09

-8.15269

12.84926

-2.45408

hsa_circ_0000517

RPPH1

ASCRP004682

3.89E-07

-6.69486

8.25959

-2.42589

hsa_circ_0081188

SLC25A13

ASCRP003040

1.27E-07

-7.14191

9.68275

-2.40068

hsa_circ_0006110

USP34

ASCRP002376

7.35E-07

-6.46544

7.52625

-2.3365

hsa_circ_0003930

GGNBP2

ASCRP003437

1.53E-05

-5.37134

4.03481

-2.25906

hsa_circ_0061749

BRWD1

 

 

Table 2. Summary of GO enrichment analysis

Differentially expressed circRNAs

The number of enriched GO terms

biological process

cellular component

molecular function

p<0.1

p<0.05

p<0.1

p<0.05

p<0.1

p<0.05

up-regulated

135

80

28

17

40

21

down-regulated

233

151

44

27

61

49

 

Table 3. Top 10 GO enrichment terms of upregulated circRNAs

Ontology

Accession

Name

Gene count

P value

biological process

GO:0065007

biological regulation

73

0.0465

GO:0008152

metabolic process

72

0.0196

GO:0071704

organic substance metabolic process

71

0.0118

GO:0044237

cellular metabolic process

68

0.0160

GO:0044238

primary metabolic process

67

0.0290

GO:0043170

macromolecule metabolic process

64

0.0058

GO:0044260

cellular macromolecule metabolic process

61

0.0034

GO:0071840

cellular component organization or biogenesis

49

0.0048

GO:0019222

regulation of metabolic process

48

0.0050

GO:0016043

cellular component organization

48

0.0051

cellular component

GO:0005622

intracellular

88

0.0192

GO:0044424

intracellular part

87

0.0123

GO:0043226

organelle

82

0.0299

GO:0043229

intracellular organelle

76

0.0438

GO:0005737

cytoplasm

69

0.0496

GO:0044446

intracellular organelle part

58

0.0119

GO:0044422

organelle part

58

0.0198

GO:0032991

macromolecular complex

36

0.0421

GO:0070013

intracellular organelle lumen

34

0.0225

GO:0043233

organelle lumen

34

0.0283

molecular function

GO:0005515

protein binding

65

0.0069

GO:0003824

catalytic activity

44

0.0464

GO:0016740

transferase activity

23

0.0210

GO:0098772

molecular function regulator

17

0.0059

GO:0005524

ATP binding

15

0.0472

GO:0044877

macromolecular complex binding

14

0.0349

GO:0030234

enzyme regulator activity

12

0.0279

GO:0016772

transferase activity, transferring phosphorus-containing groups

12

0.0468

GO:0003712

transcription cofactor activity

11

0.0013

GO:0000989

transcription factor activity, transcription factor binding

11

0.0026

 

 

Table 4. Top 10 GO enrichment terms of downregulated circRNAs

Ontology

Accession

Name

Gene count

P value

biological process

GO:0044699

single-organism process

103

0.0401

GO:0044763

single-organism cellular process

98

0.0139

GO:0051716

cellular response to stimulus

60

0.0460

GO:0007154

cell communication

57

0.0158

GO:0023052

signaling

56

0.0233

GO:0071840

cellular component organization or biogenesis

56

0.0327

GO:0044700

single organism signaling

55

0.0303

GO:0016043

cellular component organization

55

0.0308

GO:0007165

signal transduction

52

0.0290

GO:0044767

single-organism developmental process

50

0.0346

cellular component

GO:0044464

cell part

121

0.0003

GO:0005623

cell

121

0.0004

GO:0005622

intracellular

116

1.15E-06

GO:0044424

intracellular part

115

5.14E-07

GO:0043226

organelle

107

8.95E-05

GO:0005737

cytoplasm

104

2.06E-09

GO:0043229

intracellular organelle

102

4.72E-05

GO:0043227

membrane-bounded organelle

102

7.86E-05

GO:0043231

intracellular membrane-bounded organelle

93

4.52E-04

GO:0044444

cytoplasmic part

88

2.29E-09

molecular function

GO:0005488

binding

99

0.0428

GO:0005515

protein binding

74

0.0105

GO:0097159

organic cyclic compound binding

54

0.0110

GO:1901363

heterocyclic compound binding

53

0.0136

GO:0003824

catalytic activity

53

0.0154

GO:0043167

ion binding

42

0.0121

GO:0046872

metal ion binding

41

0.0083

GO:0043169

cation binding

41

0.0102

GO:0036094

small molecule binding

32

6.87E-04

GO:0000166

nucleotide binding

30

9.57E-04

 

 

Table 5. KEGG analysis of differential expressed circRNAs

 

Entry

Name

Gene count

P value

Up

regulated

hsa04022

cGMP-PKG signaling pathway

5

0.0085

hsa05166

HTLV-I infection

5

0.0356

Down

regulated

hsa04141

Protein processing in endoplasmic reticulum

8

0.0004

hsa04910

Insulin signaling pathway

7

0.0009

hsa04810

Regulation of actin cytoskeleton

7

0.0075

hsa04510

Focal adhesion

6

0.0268

hsa00620

Pyruvate metabolism

4

0.0042

hsa00280

Valine, leucine and isoleucine degradation

4

0.0067

hsa05213

Endometrial cancer

4

0.0088

hsa04012

ErbB signaling pathway

4

0.0346

hsa05215

Prostate cancer

4

0.0357

hsa00640

Propanoate metabolism

3

0.0220

 

Table 6. Hub and Module genes screened from PPI network

Type

Gene Symbol

Intersection

Hub Genes

ACACA, GSK3B, LRRK2, PRKG1, UBQLN1, FOXO1, AXIN1, MTHFD1L, PTK2, UBA2, MAP3K5.

ACACA

LRRK2

MTHFD1L

Module

Genes

ACACA, ACACB, AACS, ALDH7A1, MTHFD1L, GNAI2, MAP2K2, LRRK2, CAP1, EPS15L1, EPS15.