Analysis of competing endogenous RNA network to identify the key RNAs associated with prostate adenocarcinoma

Accepted Manuscript Title: Analysis of competing endogenous RNA network to identify the key RNAs associated with prostate adenocarcinoma Authors: Yehong Li, Zhongwen Yang PII: DOI: Reference:

S0344-0338(18)30854-9 https://doi.org/10.1016/j.prp.2018.08.029 PRP 52172

To appear in: Received date: Revised date: Accepted date:

11-7-2018 15-8-2018 26-8-2018

Please cite this article as: Li Y, Yang Z, Analysis of competing endogenous RNA network to identify the key RNAs associated with prostate adenocarcinoma, Pathology - Research and Practice (2018), https://doi.org/10.1016/j.prp.2018.08.029 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Analysis of competing endogenous RNA network to identify the key RNAs associated with prostate adenocarcinoma Running title: RNAs in prostate adenocarcinoma

1

IP T

Yehong Li1, Zhongwen Yang2* Department of Spinal Surgery, East Section of Jining No. 1 People's Hospital, Jining,

2

SC R

Shandong 272011, P.R. China.

Department of Urology Surgery, Jining No. 1 People's Hospital, Jining, Shandong

N

author: Zhongwen Yang

A

*Corresponding

U

272011, P.R. China.

M

Address: Department of Urology Surgery, Jining No. 1 People's Hospital, No. 6

ED

Jiankang Road, Central District, Jining City, Shandong Province 272011, P.R. China. Tel: +86-18653780252

A

CC E

PT

E-mail: [email protected]

1

Abstract Objectives: Prostate adenocarcinoma (PRAD) is the most common cancer in men. The aim of this study was to reveal the critical long non-coding RNA (lncRNAs),

IP T

microRNA (miRNAs) and mRNAs involved in the pathogenesis of PRAD. Methods: The level 3 mRNA and miRNA sequencing data of PRAD were

SC R

downloaded from The Cancer Genome Atlas database. Using the edgeR package of R, the differentially expressed mRNAs (DEGs), lncRNAs (DE-lncRNAs) and miRNAs (DE-miRNAs) between PRAD and normal tissues were screened. The Cox

N

U

proportional hazards regression method in the survival package was used to select the

A

lncRNAs significantly related to clinical characteristics. After the miRNA-lncRNA

M

and miRNA-mRNA pairs were predicted, a regulatory network was constructed by the Cytoscape software. For the DEGs involved in the network, enrichment analysis

ED

was conducted by the Fisher algorithm.

PT

Results: Compared to the normal samples, 25 DE-lncRNAs, 1421 DEGs and 68 DE-

CC E

miRNAs were identified in the PRAD samples. The down-regulated MESTIT1 had a significantly negative correlation with overall survival. A total of 44 DE-miRNA-DElncRNA pairs were predicted, including the PCA3-miR-96 and UCA1-miR-96.

A

Meanwhile, 33 DEGs targeted by miRNAs (for example, miR-96-CYP19A1) were found to correlate with cancers. Conclusion: Functional enrichment analysis showed that the reproductive development process (which involved TDRD1) was enriched for the DEGs implicated

2

in the lncRNA-miRNA-mRNA regulatory network. The lncRNAs MESTIT1, PCA3, and UCA1; mRNAs CYP19A1 and TDRD1; as well as miR-96 might affect the pathogenesis of PRAD. Keywords: Prostate adenocarcinoma; Long non-coding RNA; MicroRNA;

A

CC E

PT

ED

M

A

N

U

SC R

IP T

Regulatory network; Enrichment analysis

3

I N U SC R

1 Introduction

A

Prostate adenocarcinoma (PRAD) is a cancer of the prostate, a gland of the male reproductive system [21]. PRAD has no obvious symptoms in

M

early stages, while in later stages it can induce the presence of blood in the urine, difficulty in urination, fatigue, and pain in the back and pelvis

ED

[4]. The risk factors of PRAD include a family history of the disease, older age, and ethnicity, and approximately 99% of PRAD cases happen in people older than 50 years of age [44]. Globally, PRAD is the second most frequent cancer and its mortality ranks fifth among cancers in men

PT

[3]. In 2012, PRAD affected 1.1 million men and resulted in 307,000 deaths [3]. As the most common cancer in men, PRAD has been reported

CC E

in 84 countries [3], and its incidence in the developing world is increasing [2]. Non-metastatic PRAD may be cured by surgery, chemotherapy or radiation therapy, while metastatic PRAD may be treated by targeted therapy [4]. Thus, it is important to investigate the molecular mechanisms of PRAD.

A

In recent years, research on understanding the pathogenesis of PRAD has gained importance. As a member of the c-ETS (ETS) transcription factor family, the ETS-related gene 1 (ERG1) is frequently up-regulated in malignant prostate epithelial cells, and its level in PRAD cells can serve as an predictor of disease-free survival following radical prostatectomy [36]. Bonci et al. reported that miRNA-15a (miR-15a) and miR-16 functioned as tumor suppressors in PRAD by regulating cell proliferation, survival, and invasion [5]. MiR-34a suppressed PRAD regeneration

4

I N U SC R

and metastasis through inhibition of the Cluster of Differentiation 44 (CD44), and miR-34a may be a potential therapeutic target against prostate

A

cancer stem cells [31]. Li et al. reported that miR-21 contributes to invasion, motility, and resistance to apoptosis in prostate cancer cells, partly

M

through the regulation of the myristoylated alanine-rich protein kinase c substrate (MARCKS), tropomyosin 1 (TPM1), and programmed cell death 4 (PDCD4) [29]. The long non-coding RNA (lncRNA) H19–miR-675 inhibits prostate cancer metastasis through regulation of the

ED

transforming growth factor β (TGFBI), which may play diagnostic and therapeutic roles in advanced PRAD [45]. The post-transcriptional

PT

regulation of gene expression is a complex process, which can be influenced by various patterns [7]. All types of RNAs can function as regulators of gene expression by acting as competing endogenous RNAs (ceRNAs) [38]. Through competitive binding to miRNAs, ceRNAs may

CC E

release the target genes from miRNAs and thus increase the expression of those target genes [26]. Therefore, it is necessary to understand the ceRNA regulatory mechanisms of PRAD to comprehensively understand its pathogenesis.

A

We hypothesized that certain ceRNA regulatory relationships might play important roles in the pathogenesis of PRAD. In this study, the differentially expressed mRNAs (DEGs), lncRNAs (DE-lncRNAs) and miRNAs (DE-miRNAs) between PRAD and normal tissues were identified, and the correlations between DE-lncRNAs and clinical characteristics were analyzed. After the miRNA-lncRNA and miRNA-mRNA

5

I N U SC R

pairs were predicted, the lncRNA-miRNA-mRNA regulatory network was visualized. In addition, an enrichment analysis was performed for the

A

DEGs involved in the network.

M

2 Material and methods

ED

2.1 Data source

PT

The level 3 mRNA and miRNA sequencing data of PRAD (up to June 2016) were downloaded from The Cancer Genome Atlas (TCGA, http://cancergenome.nih.gov/) database [41], which included 550 mRNA and 527 miRNA sequencing samples. According to the barcode

CC E

numbers, the mRNA sequencing samples were mapped to the miRNA sequencing samples, and 293 matched samples were obtained (including 241 PRAD tissues and 52 normal tissues). This study used the sequencing data downloaded from the TCGA public database and did not involve

A

patients; thus no ethical review and informed consent were needed. 2.2 Differential expression analysis The mRNA and miRNA sequencing data in level 3 had been preprocessed and normalized, and hence the matched data could be directly used for the following analyses. The HUGO Gene Nomenclature Committee (HGNC, http://www.genenames.org/) database [13] was used to separate

6

I N U SC R

the lncRNAs and mRNAs in the mRNA sequencing data (which included both the mRNA sequences and lncRNA sequences) downloaded from

A

TCGA database. Based on the 2,775 lncRNAs and 19,004 protein coding genes in the HGNC database [13], a total of 673 lncRNAs and 18,137

M

mRNAs were identified from the mRNA sequencing data downloaded from the TCGA database. Using the edgeR package [37] of R, DEGs, DE-lncRNAs, and DE-miRNAs in PRAD and normal tissues were screened. Their p-values were adjusted to the false discovery rate (FDR) by

ED

the multtest package [16]; FDR < 0.05 and |fold change (FC)| > 1.5 were considered as the thresholds. Finally, 25 DE-lncRNAs (10 upregulated

PT

lncRNAs and 15 downregulated lncRNAs), 1421 DEGs (539 upregulated DEGs and 882 downregulated DEGs) and 68 DE-miRNAs (35

CC E

upregulated miRNAs and 33 downregulated miRNAs) were screened from the PRAD samples. 2.3 Associations between the DE-lncRNAs and clinical characteristics The clinical characteristics were obtained from the downloaded data, including age (61 years or < 61 years old), tumor recurrence (yes or no),

A

tumor status (with tumor or tumor-free), new tumor formation (yes or no), AJCC (American Joint Committee on Cancer) TNM (tumor node metastasis) stage (T3 + T4 or T1 + T2), radioactive therapy (yes or no) and overall survival time (days). Based on the expression values of the DE-lncRNAs, the lncRNAs that associated significantly with each clinical characteristic were selected using the Cox proportional hazards regression in the survival package [32]. The lncRNAs that correlated with survival time were divided into up-regulated and down-regulated

7

I N U SC R

lncRNAs according to their expression values, and the Kaplan-Meier (KM) survival analysis [27] was performed for survival time and survival

A

state.

M

2.4 Prediction of the DE-lncRNAs and DEGs targeted by DE-miRNAs

ED

Using the miRcode (a searchable map for predicting miRNA targets, http://www.mircode.org/) [23] and starBase (a database for investigating protein-lncRNA, protein-small non-coding RNA, protein-mRNA and protein-pseudogene interaction maps, http://starbase.sysu.edu.cn/) [42]

PT

databases, the DE-lncRNAs targeted by the DE-miRNAs were predicted. The miRTarBase database, which is often updated via manual

CC E

searching of research papers, provides the most comprehensive and current information on the miRNA-target pairs validated by experiments [8, 20]. Meanwhile, the DEGs targeted by the DE-miRNAs were predicted by the miRTarBase database (http://mirtarbase.mbc.nctu.edu.tw) [8, 20]. Using the allOnco database (http://www.bushmanlab.org/links/genelists), which integrates cancer genes included in several databases [1, 15, 22,

A

39], the cancer-related miRNA targets were further identified. 2.5 Construction of the lncRNA-miRNA-mRNA regulatory network and enrichment analysis

8

I N U SC R

Combined with the predicted miRNA-lncRNA and miRNA-mRNA pairs, the lncRNA-miRNA-mRNA regulatory network was constructed by the

CC E

PT

ED

M

the network. The formula was as follows:

A

Cytoscape software (http://cytoscape.org/) [28]. Using the Fisher algorithm [10], enrichment analysis was conducted for the DEGs implicated in

Among the formula, N, M, and K represent the total gene number in the whole genome, the number of pathway genes, and the number of DEGs, respectively. Meanwhile, p indicates the probability of that no less than x genes in K DEGs are pathway genes.

A

3 Results

3.1 Associations between the DE-lncRNAs and clinical characteristics

9

I N U SC R

Based on the Cox proportional hazards regression, the DE-lncRNAs that significantly associated with each clinical characteristic were screened

A

and listed in Table 1. Especially, the down-regulated HLA complex group 22 (HCG22) and the MEST intronic transcript 1 (MESTIT1), as well

M

as the up-regulated chromosome 9 open reading frame 163 (C9orf163) had significant association with overall survivals. The KM survival curves showed that HCG22, MESTIT1, and C9orf163 had significantly negative correlations with overall survival. In other words, a lower

ED

expression corresponded to a higher overall survival (Figure 1).

PT

3.2 Prediction of lncRNAs and mRNAs targeted by miRNAs

CC E

A total of 236 DE-lncRNA-associated miRNA-lncRNA pairs were obtained upon combining the miRcode and starBase databases, among which 44 DE-miRNA-DE-lncRNA pairs (for example, the prostate cancer associated 3 (PCA3)-miR-96 and the urothelial cancer associated 1 (UCA1)miR-96) involving 14 DE-miRNAs and 10 DE-lncRNAs were screened and listed in Table 2. Based on the miRTarBase database, the 14 DE-

A

miRNAs and their target genes were extracted and listed in Table 3. Through further mapping to the allOnco database, 33 DEGs targeted by miRNAs (for example, the miR-96-cytochrome P450, family 19, subfamily A, polypeptide 1 (CYP19A1)) were found to be correlated with cancers (Table 4). Additionally, we found that not all miRNA-targets pairs presented the negative correlation.

10

I N U SC R

3.3 Construction of the lncRNA-miRNA-mRNA regulatory network and pathway enrichment analysis

A

Combined with the predicted DE-miRNA-DE-lncRNA and DE-miRNA-DEG pairs, the lncRNA-miRNA-mRNA regulatory network was

M

constructed (Figure 2). Using the Fisher algorithm, 6 pathways and 12 function terms were enriched for the DEGs implicated in the network,

ED

including a reproductive development process (p-value = 3.18E-03; which involved tudor domain containing 1, TDRD1) (Table 5, Figure 3). 4 Discussion

PT

Compared to the normal samples, 25 DE-lncRNAs (10 upregulated lncRNAs and 15 downregulated lncRNAs), 1421 DEGs (539 upregulated

CC E

DEGs and 882 downregulated DEGs) and 68 DE-miRNAs (35 upregulated miRNAs and 33 downregulated miRNAs) were screened from the PRAD samples. The KM survival curves showed that MESTIT1 had a significantly negative correlation with overall survival. MEST expression increases during senescence of human primary prostate cells and decreases with the development of PRAD, indicating that MEST may be

A

involved in tumor suppression in PRAD [12]. Therefore, MESTIT1 might play an important role in the development of PRAD. Forty-five DE-miRNA-DE-lncRNA pairs (for example, PCA3-miR-96 and UCA1-miR-96) were predicted base on miRcode and starBase databases. PCA3 is overexpressed in most PRAD cases [19], and it produces a prostate-specific noncoding mRNA that can act as a biomarker for

11

I N U SC R

PRAD diagnosis [17]. Urinary PCA3 is a potential marker for detection of PRAD; a PCA3-based nomogram can better recognize men harboring

A

high-risk PRAD and confirm whether further evaluation is needed [9, 11]. UCA1 acts as an oncogenic lncRNA in some malignant tumors such

M

as bladder cancer [40] and renal cell carcinoma [30], and it may be used as a promising biomarker for human cancers. UCA1 plays an essential role in the tumorigenesis of PRAD, and UCA1 loss-of-function inhibits cell proliferation and promotes cell apoptosis partially by inactivating the

ED

Kruppel-like factor 4 (KLF4)-keratin 6/13 (KRT6/13) cascade [34]. MiR-96 functions in regulating PRAD cellular proliferation and promotes

PT

PRAD progression through inversely regulating the expression of Forkhead box O1 (FOXO1), which may provide new insights for developing novel therapeutic schedules for PRAD [14, 18, 43]. In PRAD cells, miR-96 can enhance or repress hypoxia-induced autophagy via primarily

CC E

inhibiting autophagy related 7 (ATG7) or mechanistic targeting of rapamycin (MTOR) depending on its expression level [33]. In this study, PCA3 was upregulated and UCA1 was downregulated while miR-96 was found to be upregulated, which implied that miR-96 could target to

A

various lncRNAs and emphasized complexity of miR-96 regulatory network in the progression of PRAD. Additionally, we found that PCA3 and UCA1 were correlated with various miRNAs and similarly, miR-96 was also associated with different lncRNAs in Table 2. Therefore, the regulatory network of DE-miRNA-DE-lncRNA pairs including PCA3-miR-96 and UCA1-miR-96 needed to be explored in a larger complicated

12

I N U SC R

relationship network. These suggested that targeting of PCA3 and UCA1 by miR-96 might function in the pathogenesis of PRAD and

A

corresponding experimental studies still were needed to be conducted to verify these explorations.

M

Thirty-three DEGs targeted by miRNAs (for example, miR-96-CYP19A1) were found to correlate with cancers. Polymorphic alleles of 5α-

ED

reductase, CYP3A4, CYP17, CYP19, insulin-like growth factor 1 (IGF-1), and insulin-like growth factor binding protein 3 (IGFBP-3) have been reported to be related to PRAD [35]. Kanda et al. demonstrated that CYP19A1 polymorphisms may influence PRAD risk and survival through

PT

regulation of promoter activities, which may impact the sex hormone milieu [25]. Functional enrichment analysis showed that the reproductive

CC E

development process (which involved TDRD1) was enriched for the DEGs implicated in the lncRNA-miRNA-mRNA regulatory network. As a direct target of early growth response (EGR), TDRD1 plays an important role in EGR overexpression in early stages of PRAD [6, 24]. Thus, CYP19A1 targeted by miR-96 and TDRD1 might also be involved in PRAD.

A

5 Conclusions

In conclusion, 25 DE-lncRNAs, 1,396 DEGs and 68 DE-miRNAs in PRAD samples were identified by bioinformatic analysis. In addition, lncRNAs MESTIT1, PCA3, and UCA1; mRNAs CYP19A1 and TDRD1; as well as miR-96 might be the key RNAs that play important roles in

13

I N U SC R

the pathogenesis of PRAD. However, these findings are derived from bioinformatic analysis and should be validated by further experimental

A

research.

M

Declarations of interest

ED

The authors declare that they have no competing interests.

CC E

None.

PT

Funding

Acknowledgements

A

None.

14

I N U SC R A

References

M

[1] K. Akagi, T. Suzuki, R.M. Stephens, N.A. Jenkins, N.G. Copeland, RTCGD: retroviral tagged cancer gene database. Nucleic acids research

ED

32 (2004) D523-D527.

PT

[2] P.D. Baade, D.R. Youlden, L.J. Krnjacki, International epidemiology of prostate cancer: geographical distribution and secular trends. Molecular nutrition & food research 53 (2009) 171-184.

CC E

[3] W.S. Bernard, P. Christopher, World cancer report 2014. World Health Organization (2014). [4] P.A.T.E. Board, Prostate Cancer Treatment (PDQ®). (2016).

A

[5] D. Bonci, V. Coppola, M. Musumeci, A. Addario, R. Giuffrida, L. Memeo, L. D'Urso, A. Pagliuca, M. Biffoni, C. Labbaye, The miR-15a– miR-16-1 cluster controls prostate cancer by targeting multiple oncogenic activities. Nature medicine 14 (2008) 1271-1277.

15

I N U SC R

[6] J.L. Boormans, H. Korsten, A.J. Ziel‐van der Made, G.J. Leenders, C.V. Vos, G. Jenster, J. Trapman, Identification of TDRD1 as a direct

A

target gene of ERG in primary prostate cancer. International journal of cancer 133 (2013) 335-345.

M

[7] S. Carpenter, E.P. Ricci, B.C. Mercier, M.J. Moore, K.A. Fitzgerald, Post-transcriptional regulation of gene expression in innate immunity.

ED

Nature Reviews Immunology 14 (2014) 361-376. [8] C.-H. Chou, N.-W. Chang, S. Shrestha, S.-D. Hsu, Y.-L. Lin, W.-H. Lee, C.-D. Yang, H.-C. Hong, T.-Y. Wei, S.-J. Tu, miRTarBase 2016:

PT

updates to the experimentally validated miRNA-target interactions database. Nucleic acids research 44 (2016) D239-D247.

CC E

[9] F.K. Chun, A. de la Taille, H. Van Poppel, M. Marberger, A. Stenzl, P.F. Mulders, H. Huland, C.-C. Abbou, A.B. Stillebroer, M.P. van Gils, Prostate cancer gene 3 (PCA3): development and internal validation of a novel biopsy nomogram. European urology 56 (2009) 659-668.

A

[10] İ.H. Damar, F. Altunkaş, A. Çelik, F. Koç, M. Karayakalı, K. Karaman, A. Arısoy, K. Ceyhan, Fragmented QRS frequency in patients with cardiac syndrome X. Anatolian journal of cardiology (2015). [11] I.L. Deras, S.M. Aubin, A. Blase, J.R. Day, S. Koo, A.W. Partin, W.J. Ellis, L.S. Marks, Y. Fradet, H. Rittenhouse, PCA3: a molecular urine assay for predicting prostate biopsy outcome. The Journal of urology 179 (2008) 1587-1592.

16

I N U SC R

[12] J.R. Dobosy, D. Jarrard, M. Onwueme, J. Desotelle, V. Fu, Expression of the imprinted human Peg1/MEST gene is altered in senescent

A

prostate cells and prostate cancer tissues. Cancer research 65 (2005) 655-655.

ED

acids research 34 (2006) D319-D321.

M

[13] T.A. Eyre, F. Ducluzeau, T.P. Sneddon, S. Povey, E.A. Bruford, M.J. Lush, The HUGO gene nomenclature database, 2006 updates. Nucleic

[14] A. Fendler, M. Jung, C. Stephan, A. Erbersdobler, K. Jung, G.M. Yousef, The antiapoptotic function of miR-96 in prostate cancer by

PT

inhibition of FOXO1. PLoS One 8 (2013) e80807.

CC E

[15] P.A. Futreal, L. Coin, M. Marshall, T. Down, T. Hubbard, R. Wooster, N. Rahman, M.R. Stratton, A census of human cancer genes. Nature Reviews Cancer 4 (2004) 177-183.

A

[16] Y. Ge, S. Dudoit, T.P. Speed, Resampling-based multiple testing for microarray data analysis. Test 12 (2003) 1-77. [17] J. Groskopf, S.M. Aubin, I.L. Deras, A. Blase, S. Bodrug, C. Clark, S. Brentano, J. Mathis, J. Pham, T. Meyer, APTIMA PCA3 molecular urine test: development of a method to aid in the diagnosis of prostate cancer. Clinical chemistry 52 (2006) 1089-1095.

17

I N U SC R

[18] B.S. Haflidadóttir, O. Larne, M. Martin, M. Persson, A. Edsjö, A. Bjartell, Y. Ceder, Upregulation of miR-96 enhances cellular proliferation

A

of prostate cancer cells through FOXO1. PLoS One 8 (2013) e72400.

M

[19] D. Hessels, J.M.K. Gunnewiek, I. van Oort, H.F. Karthaus, G.J. van Leenders, B. van Balken, L.A. Kiemeney, J.A. Witjes, J.A. Schalken,

ED

DD3 PCA3-based molecular urine analysis for the diagnosis of prostate cancer. European urology 44 (2003) 8-16. [20] S.-D. Hsu, F.-M. Lin, W.-Y. Wu, C. Liang, W.-C. Huang, W.-L. Chan, W.-T. Tsai, G.-Z. Chen, C.-J. Lee, C.-M. Chiu, miRTarBase: a

PT

database curates experimentally validated microRNA–target interactions. Nucleic acids research (2010) gkq1107.

CC E

[21] P.A. Humphrey, H. Moch, A.L. Cubilla, T.M. Ulbright, V.E. Reuter, The 2016 WHO Classification of Tumours of the Urinary System and Male Genital Organs—Part B: Prostate and Bladder Tumours. European urology 70 (2016) 106-119.

A

[22] J.-L. Huret, S. Le Minor, F. Dorkeld, P. Dessen, A. Bernheim, Atlas of genetics and cytogenetics in oncology and haematology, an interactive database. Nucleic acids research 28 (2000) 349-351. [23] A. Jeggari, D.S. Marks, E. Larsson, miRcode: a map of putative microRNA target sites in the long non-coding transcriptome. Bioinformatics 28 (2012) 2062-2063.

18

I N U SC R

[24] L.A. Kacprzyk, M. Laible, T. Andrasiuk, J.C. Brase, S.T. Börno, M. Fälth, R. Kuner, H. Lehrach, M.R. Schweiger, H. Sültmann, ERG

A

induces epigenetic activation of Tudor domain-containing protein 1 (TDRD1) in ERG rearrangement-positive prostate cancer. PLoS One 8

M

(2013) e59976.

ED

[25] S. Kanda, N. Tsuchiya, S. Narita, T. Inoue, M. Huang, S. Chiba, S. Akihama, M. Saito, K. Numakura, H. Tsuruta, Effects of functional genetic polymorphisms in the CYP19A1 gene on prostate cancer risk and survival. International journal of cancer 136 (2015) 74-82.

PT

[26] R.V. Kartha, S. Subramanian, Competing endogenous RNAs (ceRNAs): new entrants to the intricacies of gene regulation. Frontiers in

CC E

genetics 5 (2013) 8.

[27] D.G. Kleinbaum, M. Klein, Survival analysis: a self-learning text, Springer Science & Business Media, 2006.

A

[28] M. Kohl, S. Wiese, B. Warscheid, Cytoscape: software for visualization and analysis of biological networks. Data Mining in Proteomics: From Standards to Applications (2011) 291-303. [29] T. Li, D. Li, J. Sha, P. Sun, Y. Huang, MicroRNA-21 directly targets MARCKS and promotes apoptosis resistance and invasion in prostate cancer cells. Biochemical and biophysical research communications 383 (2009) 280-285.

19

I N U SC R

[30] Y. Li, T. Wang, Y. Li, D. Chen, Z. Yu, L. Jin, L. Ni, S. Yang, X. Mao, Y. Gui, Identification of long-non coding RNA UCA1 as an

A

oncogene in renal cell carcinoma. Molecular medicine reports 13 (2016) 3326-3334.

M

[31] C. Liu, K. Kelnar, B. Liu, X. Chen, T. Calhoun-Davis, H. Li, L. Patrawala, H. Yan, C. Jeter, S. Honorio, The microRNA miR-34a inhibits

ED

prostate cancer stem cells and metastasis by directly repressing CD44. Nature medicine 17 (2011) 211-215.

PT

[32] T. Lumley, T. Therneau, The survival package. R News 4 (2004) 26-28. [33] Y. Ma, H.-Z. Yang, B.-J. Dong, H.-B. Zou, Y. Zhou, X.-M. Kong, Y.-R. Huang, Biphasic regulation of autophagy by miR-96 in prostate

CC E

cancer cells under hypoxia. Oncotarget 5 (2014) 9169. [34] X.-Y. Na, Z.-Y. Liu, P.-P. Ren, R. Yu, X.-S. Shang, Long non-coding RNA UCA1 contributes to the progression of prostate cancer and

A

regulates proliferation through KLF4-KRT6/13 signaling pathway. International journal of clinical and experimental medicine 8 (2015) 12609. [35] R.K. Nam, W.W. Zhang, J. Trachtenberg, M.A. Jewett, M. Emami, D. Vesprini, W. Chu, M. Ho, J. Sweet, A. Evans, Comprehensive assessment of candidate genes and serological markers for the detection of prostate cancer. Cancer Epidemiology Biomarkers & Prevention 12 (2003) 1429-1437.

20

I N U SC R

[36] G. Petrovics, A. Liu, S. Shaheduzzaman, B. Furasato, C. Sun, Y. Chen, M. Nau, L. Ravindranath, Y. Chen, A. Dobi, Frequent

A

overexpression of ETS-related gene-1 (ERG1) in prostate cancer transcriptome. Oncogene 24 (2005) 3847-3852.

M

[37] M.D. Robinson, D.J. McCarthy, G.K. Smyth, edgeR: a Bioconductor package for differential expression analysis of digital gene expression

ED

data. Bioinformatics 26 (2010) 139-140.

[38] R. Sen, S. Ghosal, S. Das, S. Balti, J. Chakrabarti, Competing Endogenous RNA: The Key to Posttranscriptional Regulation. Scientific

PT

World Journal 2014 (2014) 896206-896206.

CC E

[39] T. Sjöblom, S. Jones, L.D. Wood, D.W. Parsons, J. Lin, T.D. Barber, D. Mandelker, R.J. Leary, J. Ptak, N. Silliman, The consensus coding sequences of human breast and colorectal cancers. Science 314 (2006) 268-274.

A

[40] A. Srivastava, P. Singh, S. Rath, D. Dalela, M. Goel, M. Bhatt, Appraisal of diagnostic ability of UCA1 as a biomarker of carcinoma of the urinary bladder. Tumor Biology 35 (2014) 11435-11442. [41] K. Tomczak, P. Czerwinska, M. Wiznerowicz, The Cancer Genome Atlas (TCGA): an immeasurable source of knowledge. Contemp Oncol (Pozn) 19 (2015) A68-A77.

21

I N U SC R

[42] J.-H. Yang, J.-H. Li, P. Shao, H. Zhou, Y.-Q. Chen, L.-H. Qu, starBase: a database for exploring microRNA–mRNA interaction maps from

A

Argonaute CLIP-Seq and Degradome-Seq data. Nucleic acids research 39 (2011) D202-D209.

M

[43] J.-J. Yu, Y.-X. Wu, F.-J. Zhao, S.-J. Xia, miR-96 promotes cell proliferation and clonogenicity by down-regulating of FOXO1 in prostate

ED

cancer cells. Medical oncology 31 (2014) 1-7.

1894-1903.

PT

[44] M. Zeegers, A. Jellema, H. Ostrer, Empiric risk of prostate carcinoma for relatives of patients with prostate carcinoma. Cancer 97 (2003)

CC E

[45] M. Zhu, Q. Chen, X. Liu, Q. Sun, X. Zhao, R. Deng, Y. Wang, J. Huang, M. Xu, J. Yan, lncRNA H19/miR‐675 axis represses prostate

A

cancer metastasis by targeting TGFBI. FEBS journal 281 (2014) 3766-3775.

22

I N U SC R

Figure captions

A

Figure 1. Kaplan-Meier survival curves for 3 lncRNAs C9orf163 (A), HCG22 (B), and MESTIT1 (C) associated with overall survival.

M

Horizontal axis represents overall survival time. Vertical axis stands for survival ratio. Dotted blue lines and dotted red lines represent samples

ED

with down-regulated and up-regulated lncRNAs, respectively. Figure 2. The lncRNA-miRNA-mRNA regulatory network. Red and blue lines separately represent lncRNA-miRNA and miRNA-mRNA

CC E

respectively.

PT

regulatory relationships. Orange circles, white squares, and white circles stand for differentially expressed lncRNAs, miRNAs and mRNAs,

Figure 3. The Gene Ontology (GO) functions (A) and pathways (B) enriched for the differentially expressed genes implicated in the regulatory

A

network.

23

A ED

PT

CC E A

M

N U SC R

Figr-1

24

I

A ED

PT

CC E A

M

N U SC R

Figr-2

25

I

A ED

PT

CC E A

M

N U SC R

Figr-3

26

I

27

A ED

PT

CC E A

M

N U SC R

I

I N U SC R

Table 1. The differentially expressed lncRNAs (DE-lncRNAs) significantly associated with each clinical characteristic Associated lncRNAs

A

Clinical characteristics

Down-regulated

C2orf48

C7orf65

Biochemical_recurrence (Yes or No)

PCA3,SNHG4

MIMT1,TCL6

Tumor status (With tumor or Tumor free)

C2orf48,SNHG4

EGOT,MIMT1,PGM5P2,PWRN1

New tumor after treatment (Yes or No)

PCA3,SNHG4

-

AJCC TNM staging system (T3 + T4 or T1 + T2)

C2orf48，PCA3

ATXN8OS,TCL6

Radiation therapy (Yes or No)

DGCR10,DGCR9

EGOT,FAM66E

Overall survivals (days)

C9orf163

HCG22,MESTIT1

A

CC E

PT

ED

Age at diagnosis (61 or < 61)

M

Up-regulated

28

29

A ED

PT

CC E A

M

N U SC R

I

30

A ED

PT

CC E A

M

N U SC R

I

I N U SC R

Table 2. The 45 miRNA-lncRNA pairs involving 14 differentially expressed miRNAs (DE-miRNAs) and 10 differentially expressed lncRNAs

A

(DE-lncRNAs)

change of IncRNAs down-

miR-7,miR-449b,miR-122

PT

ATXN8OS

miRNAs

ED

lncRNA

M

Expression

regulation

CC E

down-

EGOT

miR-23c,miR-499-5p

regulation down-

A

EMX2OS

miR-449b

regulation down-

FAM66E

miR-490,miR-23c,miR-187,miR-153,miR-133a,miR-876,miR-7,miR-122 regulation

31

I N U SC R

downHCG22

miR-490,miR-23c,miR-96,miR-876,miR-122,miR-190b regulation

A

downMEG8

miR-206,miR-7,miR-449b

MYCNOS

ED

down-

M

regulation

miR-7,miR-449b,miR-122,miR-551

PT

regulation up-

PCA3

miR-490,miR-23c,miR-96,miR-206,miR-876,miR-7

CC E

regulation down-

PWRN1

miR-490,miR-23c,miR-876,miR-7,miR-122,miR-190b

A

regulation down-

UCA1

miR-23c,miR-96,miR-206,miR-122,miR-190b regulation

32

33

A ED

PT

CC E

IP T

SC R

U

N

A

M

Table 3. The 14 differentially expressed miRNAs (DE-miRNAs) and their target mRNAs miR

Target Targeted mRNAs

NA

count

IP T

miR31 122

ACTA1,ADAM18,ANKRD34C,APOBEC2,ASB10,ATP1A4,BCAN,CD5L,CLD

SC R

N19,CLEC4M,CNGB1,CNTN6,CPEB1,CSTL1,CXCR2,FGB,FOXE3,FOXI1, GABRA3,GALNT8,GCNT4,GRIN3B,GTSE1,HAVCR1,HCG22,HEPHL1,HM GA2,HMX2,HOXC13,IGF2BP1,KBTBD12,HCG22,FAM66E,UCA1,PWRN1,

A

N

U

MYCNOS,ATXN8OS

miR-

M

38 133a

ABHD12B,AQP10,ARL14,ATP10B,ATP2B3,CBWD5,CCR3,CDH8,CIDEA,C

ED

MTM5,CPB2,CRP,CSRNP3,CSTL1,CTXN2,DBX2,DCAF12L1,DCC,DPP6,D

PT

RD5,DUOXA2,EBF2,FCAR,FOXA2,FOXG1,FOXI2,FOXL2,FSHB,GALR1, GLDN,GLRA2,GLYATL3,GRIA4,HABP2,HAPLN1,HMGCLL1,HORMAD2,I

CC E

GSF1,FAM66E

A

miR153

40

ABCG4,ACTN3,ADH1A,ANGPTL5,ANXA13,APOBEC2,ATP13A4,ATP1B4, CASQ1,CDC25C,CLEC2A,CNTNAP4,COL10A1,COL19A1,CTNNA3,DBX2, DCAF12L1,DCC,DLX1,DMRT2,DMRTC2,DNAJC5B,DUOXA1,DUSP13,EB F2,EDDM3A,EDDM3B,EGFL6,ELAVL2,EPGN,EVPLL,FABP2,FGF23,FGL 1,GDF10,GIF,GJB4,GLDN,HCG22,HMGCLL1,FAM66E

34

miR30 A1CF,ABO,ADAM2,ANKRD7,ATP1A4,CLEC4M,COL10A1,DIO1,DPP6,EB

IP T

187

F2,FCAR,FUT3,GBX2,GRIA4,GTSE1,GTSF1L,HOXC9,IQCJ,LHX8,LRIT1,L

SC R

RRC36,MCF2,MTMR8,MYOZ2,NEUROG2,PPP1R1C,PRDM13,PRSS35,PT

U

CHD3,RBM46,FAM66E

miR30

N

ACTN3,ADAM18,ADH1A,ADH4,ANGPTL5,ANKRD30B,ANKRD34B,APOB

190b

A

EC4,ASPA,ASTN1,CALCR,CDH18,CDH9,CLCA2,CNTN6,CPNE9,CRB1,TF

M

F2,TNMD,TRIM55,TRIM61,TRIM9,UGT2B28,UGT3A1,ULBP1,VSNL1,VW

A

CC E

206

54

PT

miR-

ED

C2,WDR72,ZIC3,ZIC5,PWRN1,HCG22,UCA1

ABHD12B,ADAM2,ADH1A,ANKRD34B,MCHR2,MGAM,MIOX,MOG,MOV 10L1,MRO,MYL1,MYOZ2,MYPN,NETO1,NGB,NLRP14,NR1H4,PANX3,PA X3,PAX7,PCDH11Y,PIK3C2G,PITX2,PKLR,POU4F2,RDM1,RGS7,RIT2,RP TN,S100A7A,SCEL,SFTPB,SFTPC,SGCZ,SH2D1B,SH3GL3,SIM1,SNTG1,S ORCS3,SPRR4,STOX2,SYT14,TBC1D28,TFAP2B,TGM3,TRPC5,TRPM3,VA T1L,VSNL1,WDR72,WFDC5,XG,ZIC1,ZNF488,UCA1,MEG8,PCA3

35

miR35 449b

ABCC12,ACRV1,ACTN3,APOB,ASTN1,P2RX6,PAPPA2,PPP1R1C,PPP3R2,

IP T

PRG4,PRKAG3,SCGB1D4,SCN1A,SCRT1,SORCS3,SOSTDC1,SPSB4,SULT 4A1,SVOP,TBC1D26,TBC1D28,TDRD1,THPO,THRSP,TRIM67,TRIM9,TRP

SC R

C3,TRPC5,UGT3A1,UNC13C,UNC45B,UTS2,VAT1L,WFDC5,ZIC5,MEG8,

U

EMX2OS,ATXN8OS,MYCNOS

N

miR41

ABCG4,ACRV1,ADAM18,ADAM21,ADAMTS18,ALB,ANGPTL5,ANKRD34C

A

7

M

,APOA2,APOBEC2,APOBEC4,COL2A1,CRP,CSRNP3,CTNNA3,CYP11A1, CYP1A2,CYP3A5,DAPL1,DEFB127,DMBX1,DMC1,DMRT2,DMRTC2,SPIN

ED

K8,SPRR1A,STK32A,STOX2,SULT1E1,TBC1D26,TCP10,TDRD12,TPTE2,T RDN,TRIM9,TROAP,TRPA1,TRPC3,TRPC5,UGT3A1,UGT8,FAM66E,PWR

CC E

PT

N1,MEG8,ATXN8OS,PCA3,MYCNOS

miR-

ADAM18,ADAMTS18,AKR1D1,AMHR2,ANKRD7,AQP2,AQP5,ASTN1,CAL

56

CR,CBLN4,CCKBR,CCL18,CHRM2,CLCN1,CLDN22,CLEC4M,CPEB1,CP

A

96

NE9,CTXN2,CYP19A1,DEPDC1B,DMRT2,DMRTB1,DUSP13,EDAR,EGFL 6,EPYC,ERAS,FBXL21,FCAR,FOXI1,GAD2,GAL,NHLH2,NKAPL,NRG1,OT OR,PCDH11X,PCDH11Y,PF4V1,PKD1L2,PNPLA1,PPP3R2,PRLHR,PROK 2,PRRG3,RFX4,RGS20,RNASE7,RSPH10B2,RTP1,SCARA5,SCN11A,SERPI NA11,SERPINA5,SH3GL3,UCA1,HCG22,PCA3

36

37

A ED

PT

CC E

IP T

SC R

U

N

A

M

38

A ED

PT

CC E

IP T

SC R

U

N

A

M

Table 4. Cancer-related differentially expressed genes (DEGs) targeted by miRNAs. miRNA

Cancer related target genes

APOBEC2 (down-regulation),CNTN6 (down-regulation),HMGA2 (down miR-122 (up-regulation) (up-regulation)

DCC (down-regulation),FOXG1 (up-regulation),FOXL2 (down-regulatio

IP T

miR-133a (down-regulation)

APOBEC2 (down-regulation),CDC25C (up-regulation),COL19A1 (down miR-153 (up-regulation)

SC R

(down-regulation),GDF10 (down-regulation)

ABO (down-regulation),MCF2 (down-regulation)

miR-190b (up-regulation)

CALCR (down-regulation),CNTN6 (down-regulation),TFF2 (down-regu

miR-206 (down-regulation)

PAX3 (down-regulation),PAX7 (down-regulation)

miR-449b (up-regulation)

PAPPA2 (down-regulation),PRG4 (up-regulation),THPO (up-regulation

miR-7 (up-regulation)

ADAMTS18 (down-regulation),APOBEC2 (down-regulation)

ED

M

A

N

U

miR-187 (down-regulation)

ADAMTS18 (down-regulation),CALCR (down-regulation),CYP19A1 (do (up-regulation)

A

CC E

PT

miR-96 (up-regulation)

39

40

A ED

PT

CC E

IP T

SC R

U

N

A

M

Table 5. The Gene Ontology (GO) functions (A) and pathways (B) enriched for the differentially expressed genes (DEGs) implicated in the regulatory network Term

Count

P-value

Genes HAPLN1, CLCA2, CLDN19, CNTNAP4,

IP T

EGFL6, PCDH11Y, CNTN6, PCDH11X, GO:0007155~cell

ASTN1, BCAN, CLDN22, COL2A1, ACTN3, 24

7.68E-04

OTOR, CTNNA3, CLEC4M, CDH8, CDH9,

SC R

adhesion

COL19A1, CCR3, CDH18, ADAM2, TROAP,

U

HABP2

N

HAPLN1, CLCA2, CLDN19, CNTNAP4,

A

EGFL6, PCDH11Y, CNTN6, PCDH11X,

GO:0022610~biol

ASTN1, BCAN, CLDN22, COL2A1, ACTN3,

7.83E-04

PT

ED

M

24 ogical adhesion

SULT4A1, CYP3A5, APOA2, APOB, CYP11A1, 1.47E-03

FGF23, SULT1E1, FSHB, AKR1D1, NR1H4,

CC E

11

COL19A1, CCR3, CDH18, ADAM2, TROAP, HABP2

GO:0008202~ster oid metabolic

OTOR, CTNNA3, CLEC4M, CDH8, CDH9,

process

CYP19A1 CDH8, CDH9, CLDN19, COL19A1, PCDH11Y,

A

GO:0016337~cell 13

1.56E-03

CDH18, PCDH11X, ASTN1, CLDN22,

-cell adhesion COL2A1, OTOR, CTNNA3, CLEC4M

41

GO:0003006~rep AMHR2, EDDM3B, FOXL2, DMRTC2, DMRT2, roductive 12

3.18E-03

MOV10L1, DMC1, FSHB, LHX8, DMRTB1,

developmental ANKRD7, TDRD1 process

GO:0007610~beh

IP T

DMBX1, SCN1A, PRLHR, CYP11A1, CCKBR, DRD5, ASTN1, CXCR2, GAL, ZIC1, CCL18, 17

3.41E-03

PROK2, CCR3, ADAM2, NHLH2, LHX8, CMTM5

GO:0007423~sen

FOXL2, HMX2, CRB1, HOXC13, FOXG1, 11

3.66E-03

GBX2, POU4F2, COL2A1, ZIC1, FOXI1,

U

sory organ

SC R

avior

FOXE3

N

development

GO:0006811~ion 23

5.39E-03

ED

transport

M

A

CLCN1, TRPM3, CLCA2, SCN1A, SVOP, TRPC3, TRPC5, GABRA3, ATP1B4, TRPA1, GLRA2, HEPHL1, ATP1A4, GIF, GRIN3B, CNGB1, GRIA4, PKD1L2, ATP13A4, ATP2B3,

PT

P2RX6, SCN11A, SCARA5

CC E

GO:0042445~hor mone metabolic

7

PRLHR, CYP11A1, ADH4, DIO1, SULT1E1, 6.87E-03 FSHB, AKR1D1

process

A

GO:0007389~patt ern specification

DLX1, HOXC9, FOXA2, HOXC13, PAX7, 11

1.04E-02 SOSTDC1, FOXG1, GBX2, ZIC1, ZIC3, PITX2

process

42

GO:0048609~rep

FOXL2, EDDM3B, EDDM3A, DRD5,

roductive process

MOV10L1, PROK2, APOB, DUSP13, 15

2.37E-02 SERPINA5, NHLH2, NLRP14, ADAM18,

organism

DMC1, FSHB, TDRD1

GO:0055085~tran

CLCN1, TRPM3, SVOP, SCN1A, TRPC3,

smembrane

16

IP T

in a multicellular

3.81E-02

TRPC5, TRPA1, AQP5, ATP1A4, GIF, CNGB1, ABCC12, AQP10, AQP2, SCN11A, SCARA5

SC R

transport

ED

M

A

N

U

(A)

Count

P-value

PT

Pathway

CALCR, MCHR2, P2RX6, PRLHR,

hsa04080~Neuroactive

CC E

Genes

1.38E-

GALR1, CCKBR, GABRA3, CHRM2,

04

DRD5, GLRA2, GRIN3B, GRIA4,

13

ligand-receptor interaction

A

FSHB

43