Skip to content

Advertisement

  • Research
  • Open Access

Biomarker microRNAs for prostate cancer metastasis: screened with a network vulnerability analysis model

Journal of Translational Medicine201816:134

https://doi.org/10.1186/s12967-018-1506-7

  • Received: 1 March 2018
  • Accepted: 5 May 2018
  • Published:

Abstract

Background

Prostate cancer (PCa) is a fatal malignant tumor among males in the world and the metastasis is a leading cause for PCa death. Biomarkers are therefore urgently needed to detect PCa metastatic signature at the early time. MicroRNAs are small non-coding RNAs with the potential to be biomarkers for disease prediction. In addition, computer-aided biomarker discovery is now becoming an attractive paradigm for precision diagnosis and prognosis of complex diseases.

Methods

In this study, we identified key microRNAs as biomarkers for predicting PCa metastasis based on network vulnerability analysis. We first extracted microRNAs and mRNAs that were differentially expressed between primary PCa and metastatic PCa (MPCa) samples. Then we constructed the MPCa-specific microRNA-mRNA network and screened microRNA biomarkers by a novel bioinformatics model. The model emphasized the characterization of systems stability changes and the network vulnerability with three measurements, i.e. the structurally single-line regulation, the functional importance of microRNA targets and the percentage of transcription factor genes in microRNA unique targets.

Results

With this model, we identified five microRNAs as putative biomarkers for PCa metastasis. Among them, miR-101-3p and miR-145-5p have been previously reported as biomarkers for PCa metastasis and the remaining three, i.e. miR-204-5p, miR-198 and miR-152, were screened as novel biomarkers for PCa metastasis. The results were further confirmed by the assessment of their predictive power and biological function analysis.

Conclusions

Five microRNAs were identified as candidate biomarkers for predicting PCa metastasis based on our network vulnerability analysis model. The prediction performance, literature exploration and functional enrichment analysis convinced our findings. This novel bioinformatics model could be applied to biomarker discovery for other complex diseases.

Keywords

  • Bioinformatics model
  • Network vulnerability analysis
  • MicroRNA biomarkers
  • Prostate cancer metastasis

Background

Prostate cancer (PCa) is one of the common malignant tumors worldwide. In western countries, it has become the second major cause of cancer death among men [1]. The incidence and mortality of this disease were also increasing in Asia during the last decade [2]. At present, the degree of PCa is stratified as low- or high-risk based on the Gleason score, prostate-specific antigen level and other clinical indices. However, such classification seems to be insufficient to monitor PCa progression, especially at the time of metastasis initiation [3]. The early detection of PCa metastatic signature is crucial for evaluating PCa outcome, therefore screening key molecules as biomarkers for predicting PCa metastases is of clinical significance.

MicroRNAs are a group of endogenous, small non-coding RNAs with approximately 22–24 nucleotides in length [4]. They regulate gene expression through base-paring with target messenger RNAs (mRNAs) at the post-transcriptional level and thereby play important roles in a number of important cellular processes [5]. Extensive efforts have been made to identify reliable microRNAs as biomarkers because microRNAs are remarkably stable and specific to be detected in tissues, blood as well as other bodily fluids [68]. The expression level of circulating microRNAs also exhibits characteristic alteration in individuals with different pathological settings [9].

Currently, substantial investigations are devoted to discovering microRNA biomarkers for PCa metastasis evaluation, most of which are experimental. Firstly the differentially expressed (DE) or dysregulated microRNAs from large-scale microRNA expression data were extracted as outliers and then the candidates were further validated by low-throughput experiments, such as real-time PCR, etc. [3, 10, 11]. Although experimental methods are powerful enough to detect the abnormal change of microRNA expression between different condition groups, e.g., primary PCa (PPCa) and metastatic PCa (MPCa) [12], it is not easy to identify the driver or key molecules at the systems level. As we known, the mechanism of PCa metastasis is complex, where the dysregulation of microRNAs tends to be highly heterogeneous due to the genetic and environmental factors [13, 14].

Nowadays, computational approaches based on systems biology and network science are well performed on detecting microRNA or gene signatures for the diagnosis and treatment of complex diseases such as cancers [15, 16], diabetes [17] and neurodevelopmental disorders [18]. In particular, Zhang et al. introduced a correlation and clustering based framework to identify microRNA-mRNA network modules for differentiating PPCa and MPCa subtypes [19]. We here aim to develop a novel bioinformatics method to screen single microRNA biomarkers for predicting PCa metastasis, which is propitious to PCa prognosis and therapy.

In our previous studies, we proposed a network-based bioinformatics model called Pipeline of Outlier MicroRNA Analysis (POMA) to detect microRNA biomarkers for cancer diagnosis and prognosis [2022]. The model integrates microRNA/mRNA expression data with the structural information of microRNA-mRNA regulatory network. Two measurements, i.e., number of single-line regulation (NSR, the number of genes that uniquely regulated by a single microRNA) and transcription factor (TF) gene percentage (TFP, the percentage of TF genes targeted by a single microRNA), were defined to quantify the regulatory power of independent microRNAs. These features characterize network systems vulnerability because the abnormal change of unique regulatory relationships cannot be compensated by other interactions. Statistical evidences demonstrated that microRNAs with significantly high NSR and TFP values are more likely to be biomarkers [20, 21]. In this study, we improved and updated the model as MicroRNA Biomarker Discovery (MicroRNA-BD) by analyzing the network vulnerability and considering the functional importance of genes that are uniquely regulated by given microRNAs. We then applied the model to screen key microRNAs as biomarkers for predicting PCa metastasis. The schematic pipeline of the MicroRNA-BD model is shown in Fig. 1.
Fig. 1
Fig. 1

The schematic pipeline for MPCa microRNA biomarker identification. LIMMA linear models for microarray data analysis; eBayes the empirical bayes; adj.p value adjusted p value; FC fold change; DE differentially expressed; MicroRNA-BD microRNA biomarker discovery; NSR number of single-line regulation; TFP transcription factor gene percentage; UTP percentage of transcription factor genes in microRNA unique targets; ROC receiver operating characteristic curve; GO Gene Ontology; MPCa metastatic prostate cancer

Methods

Data collection

The microRNA and mRNA expression dataset for biomarker prediction (GSE21036 [23] and GSE3325 [24]) were downloaded from gene expression omnibus (GEO) [25]. Among them, GSE21036 was generated by Agilent-019118 Human microRNA Microarray 2.0 G4470B, and contained microRNA expression data from 142 prostate tissue samples. Here 113 of the samples including 99 PPCa and 14 MPCa were extracted for further analysis. The mRNA expression dataset GSE3325 was performed on Affymetrix Human Genome U133 Plus 2.0 Array, 5 and 4 individual samples for PPCa and MPCa were selected, respectively. Moreover, another independent dataset GSE26964 [26] with 6 PPCa and 7 MPCa samples from Capitalbio mammal microRNA V3.0 platform was selected for result validation. The details are summarized in Table 1.
Table 1

Summary of the microRNA and mRNA dataset used in this study

Category

RNA type

GEO accession

Platform

Sample source

Number of samples (PPCa/MPCa)

Prediction

microRNA

GSE21036

GPL8227

Prostate tissue

113 (99/14)

mRNA

GSE3325

GPL570

Prostate tissue

9 (5/4)

Validation

microRNA

GSE26964

GPL8469

Prostate tissue

13 (6/7)

PPCa primary prostate cancer; MPCa metastatic prostate cancer

Besides, the previously reported PCa microRNA biomarkers were manually collected and integrated from: (1) the review for PCa microRNAs by Vanacore et al. [27], (2) NCBI PubMed using the retrieval terms as “(prostate cancer[tiab]) AND (microRNA*[tiab] OR miRNA*[tiab]) AND (biomarker*[tiab] OR marker*[tiab] OR indicator*[tiab] OR predict*[tiab])”.

Differentially expressed microRNAs (DE-microRNAs) and mRNAs (DE-mRNAs) extraction

The DE-microRNAs and DE-mRNAs were extracted based on the comparison of their expressions between PPCa and MPCa samples using the empirical bayes (eBayes) method in linear models for microarray data analysis (LIMMA) R package [28, 29]. The Benjamini–Hochberg false discovery rate method was applied to adjust raw p values [30]. For the gene that is related to multiple probes, we assigned it to the probe that had the most significant variation across its expression profile. The adjusted p value (adj.p value) < 0.05 and |log2fold-change| > 1 were chosen as the cut-off.

MPCa-specific microRNA-mRNA network construction

The MPCa-specific microRNA-mRNA network was constructed in two steps: First, a human microRNA-mRNA network (termed as the reference network) was built based on both experimentally validated and computationally predicted microRNA-mRNA regulatory data. Compared with our previous work [20, 21], the microRNA-mRNA pairs were updated and moreover, the latest nomenclature of microRNAs in miRBase (Release 21) [31] was fully considered before network reconstruction. Here the experimental data were mined from miRTarBase (version 4.5) [32], TarBase (version 6.0) [33], miRecords (version 4.0) [34], and miR2Disease [35] whereas the predicted data included information from HOCTAR (version 2.0) [36], ExprTargetDB [37], and starBase (version 2.0) [38]. To reduce the false positive rate, microRNA-mRNA pairs validated by low-throughput experiments, e.g, real-time PCR etc. were considered in this study while the predicted pairs were selected only when they existed at least in two of the three computational prediction databases. In the second step, the DE-microRNAs and DE-mRNAs were mapped onto the reference to extract the MPCa-specific microRNA-mRNA network.

Biomarker microRNA identification based on network vulnerability analysis

As illustrated in Fig. 2, the microRNA-mRNA relationship can be classified into four types based on their regulatory modes. The POMA model proposed in our previous studies pays more attention to the single-line regulatory power of microRNAs and focuses on the biological functions of their targets [20, 21]. After analyzing the sub-structure of microRNA-mRNA network, we found that microRNAs still held the potential to uniquely regulate genes with crucial functions, e.g., transcription factor (TF) genes. As an example, the TF gene G_10 in Fig. 2 is uniquely regulated by M_4. The alteration of single-line regulation is not compensated, and TFs are often key players in many important biological processes, thus it is reasonable to assume that the dysregulation of such regulatory patterns is more likely to alter the system stability and eventually cause the systematic disorder. To strengthen the importance of microRNA regulation on TF genes, we here applied a novel parameter called the unique-regulated TF gene percentage (UTP). Numerically, it is equivalent to the percentage of TF genes in the microRNA unique targets. Finally, our previous model POMA was improved as MicroRNA-BD to identify microRNA biomarkers with the following three measurements for network vulnerability characterization.
Fig. 2
Fig. 2

Schematic description of microRNA-mRNA regulatory types. Four types were defined here, i.e., TF or non-TF genes regulated by multiple or single microRNAs. For example, G_1 was uniquely regulated by M_1 whereas TF gene G_5 was co-regulated by M_2 and M_3. The co-regulatory sites are robust since one of the regulations altered can be compensated by others. Here the unique regulatory sites, i.e., single-line regulations, are considered as the vulnerable structure in the network. Meanwhile, microRNAs that target more TF genes seem to be functionally important. M microRNA; G gene; TF transcription factor

Step 1:: 

NSR (number of single-line regulation) measurement is the number of genes that are uniquely regulated by a single microRNA. MicroRNAs with significantly high NSR values (p value < 0.05, Wilcoxon signed-rank test) were extracted based on the analysis of MPCa-specific microRNA-mRNA network

Step 2:: 

TFP (transcription factor gene percentage) measurement is the percentage of TF genes targeted by a single microRNA. MicroRNAs with significantly high TFP values (p value < 0.05, Wilcoxon signed-rank test) were selected from those screened in Step 1;

Step 3:: 

UTP (unique-regulated TF gene percentage) measurement is defined as the percentage of TF genes in microRNA unique targets. MicroRNAs with UTP > 0 in Step 2 were identified as candidate biomarkers.

Performance evaluation

We performed the receiver-operating characteristic (ROC) analysis on both prediction and validation microRNA datasets to evaluate the performance of identified microRNA biomarkers for classifying MPCa and PPCa. The ROC curve and the area under curve (AUC) were drawn and calculated for each of the identified microRNAs using the R package ‘epicalc’ [39]. The percentage of the reported MPCa microRNA biomarkers in the predicted set was defined as prediction precision to quantify the performance of the model.

Functional enrichment analyses

To validate the association between the targets of candidate microRNA biomarkers and PCa metastasis, we performed Gene Ontology (GO) annotation, Kyoto Encyclopedia of Genes and Genomes (KEGG) [40] pathway analyses and Ingenuity Pathway Analysis (IPA) [41] using Database for Annotation, Visualization and Integrated Discovery (DAVID, version 6.7) [42] and IPA program [41], respectively. Here the targets of identified microRNA biomarkers were retrieved from human microRNA-mRNA network and Benjamini–Hochberg method was used to adjust raw p values for enrichment analysis. The top ten significantly enriched terms (adj.p value < 0.05) were selected and further studied for their correlations with PCa metastasis through literature validation.

Results

Biomarker microRNAs for predicting PCa metastasis

A total of 67 literature reported PCa microRNA biomarkers were manually collected (see Additional file 1). The human microRNA-mRNA network included 48,868 regulatory pairs among 618 microRNAs and 9526 genes/mRNAs. As shown in Fig. 3, Additional files 1 and 2, respectively, PCa biomarker microRNAs had significant regulatory power in the network, which convinced the priori evidence for biomarker discovery [20, 21]. Moreover, more than 53.7% (36/67) of these microRNAs had UTP > 0, which was approximately twice greater than that in the whole network (27.5%, 170/618).
Fig. 3
Fig. 3

Topological and functional characterization of reported PCa microRNA biomarkers. a NSR distribution of reported PCa microRNA biomarkers and all microRNAs in human microRNA-mRNA network. b TFP distribution of reported PCa microRNA biomarkers and all microRNAs in human microRNA-mRNA network. The statistical significance was calculated using Kolmogorov–Smirnov test. NSR number of single-line regulation; TFP transcription factor gene percentage; PCa prostate cancer

Based on the selected sample data, 91 DE-microRNAs and 970 DE-mRNAs were statistically identified. The MPCa-specific microRNA-mRNA network contained 505 microRNA-mRNA regulatory pairs among 65 microRNAs and 263 mRNAs (see Additional file 3). Seven microRNAs, i.e., miR-204-5p, miR-101-3p, miR-145-5p, miR-198, miR-152, miR-130a-3p and miR-363-3p tended to have significantly high NSR and TFP values in MPCa-specific microRNA-mRNA network (see Additional file 4). The UTP measurements for miR-130a-3p and miR-363-3p are 0, therefore, the remaining five microRNAs, i.e., miR-204-5p, miR-101-3p, miR-145-5p, miR-198, and miR-152 were screened as putative biomarkers for predicting PCa metastasis after MicroRNA-BD filtration.

As shown in Table 2, four of the identified microRNAs, i.e., miR-204-5p, miR-101-3p, miR-145-5p, and miR-152, were significantly down-regulated in MPCa group whereas miR-198, was over-expressed in MPCa samples compared with PPCa. The ROC curves for their PCa metastasis prediction performance were shown in Fig. 4. In the prediction set GSE21036 and another independent validation set GSE26964, the AUC ranged from 0.70 to 0.99 and from 0.71 to 0.93, respectively. Overall, miR-145-5p, miR-204-5p, and miR-152 achieved the best performance (AUC > 0.80) on PPCa and MPCa subtyping, and the AUC distribution of the five microRNAs in two datasets was highly consistent, which indicated the predictive power of the identified microRNA biomarkers for discriminating between MPCa and PPCa.
Table 2

Details for the identified microRNA biomarkers

microRNA ID

Adj.p value (PPCa vs MPCa)

log2 (FC)

Target number

NSR

TFP

UTP

miR-204-5p

3.43E−08

− 2.0896

21

13

0.2857

0.3846

miR-101-3p

1.26E−08

− 1.0684

24

3

0.2917

0.3333

miR-145-5p

8.00E−25

− 3.2157

12

3

0.2500

0.3333

miR-198

7.67E−05

1.2564

12

5

0.3333

0.2000

miR-152

3.67E−08

− 1.0146

17

6

0.2941

0.1667

PPCa primary prostate cancer; MPCa metastatic prostate cancer; adj. p value adjusted p value; FC fold change; NSR number of single-line regulation; TFP transcription factor gene percentage; UTP percentage of transcription factor genes in microRNA unique targets

Fig. 4
Fig. 4

ROC analysis for the identified microRNA biomarkers. The AUC distribution in the prediction set GSE21036 and another independent validation set GSE26964 ranged from 0.70 to 0.99 and from 0.71 to 0.93, respectively. Red curve: GSE21036; blue curve: GSE26964. PPCa primary prostate cancer; MPCa metastatic prostate cancer; ROC receiver operating characteristic curve; AUC area under the curve

Literature validation of the identified microRNA biomarkers

All of the five microRNAs were involved in PCa pathogenesis according to PubMed literature reports. In particular, two microRNAs (40%, 2/5), i.e., miR-145-5p and miR-101-3p, have been reported as potential biomarkers for human PCa metastasis [27, 43, 44]. As a tumor suppressor microRNA, the down-regulation of miR-145 (namely miR-145-5p) could cause cell invasion and migration in PCa progression [27]. Huang et al. showed that miR-145 regulated the characteristics of cancer stem cells and played important roles in the progression of PCa bone metastasis [45]. Moreover, miR-145 is a direct target of p53, and the loss of wild-type p53 could promote PCa bone metastasis by partially repressing miR-145 expression [46]. Chakravarthi et al. showed that the loss of miR-101 (namely miR-101-3p) may affect the expression of SUB1 and lead to the activation of known oncogenes driving PCa metastasis [47]. Besides, miR-204-5p and miR-152 were also associated with PCa progression and metastasis. Lin et al. [48] found that miR-204-5p was a tumor suppressor and could promote apoptosis through regulating BCL2 in PCa cells. Todorova et al. studied the effect of miR-204 (namely miR-204-5p) modulation on important TFs for PCa bone marrow metastasis and uncovered that this microRNA was dysregulated in MPCa in vitro [49]. Theodore et al. analyzed the microRNA expression profile data and found the ethnic difference of miR-152 expression between African American (AA) and Caucasian PCa patients. On the other hand, the epigenetic regulation of miR-152 and DNMT1 may contribute to the aggressiveness of PCa tumors, especially to AA PCa patients [50]. Last but not least, miR-198 was found to be up-regulated in high grade (Gleason score ≥ 8) PCa tumors, which would help recognize the aggressive behavior of PCa [51]. However, the filtered two microRNAs, i.e., miR-130a-3p and miR-363-3p, did not show any reliable correlation with PCa metastasis so far. In summary, MicroRNA-BD outperformed previous models and improved the microRNA biomarker prediction precision from 28.6% (2/7) to 40% (2/5) in this case study.

We further evaluated the relationship between the targets of identified microRNA biomarkers and PCa metastasis. As illustrated in Fig. 5, the TF gene STAT1, which was uniquely regulated by miR-145-5p in the MPCa-specific microRNA-mRNA network, could be activated by Endoglin. In the study, Endoglin was shown to suppress the cell invasion of PCa and inhibit PCa metastasis [52]. Meanwhile, several uniquely-regulated non-TF genes, such as DDR2 and MRC2, also participated in the PCa metastatic progression [53, 54]. For instance, DDR2 regulated the promoter activity of parathyroid hormone-related protein and thereby facilitated the bone metastasis of PCa [53]. In addition, EMP1 suppressed PCa cell proliferation and invasion by regulating caspase-9 and VEGFC protein [55], and it was co-regulated by miR-101-3p, miR-204-5p and miR-152 in the network. Some other co-regulated genes, including PTEN and MYC, were also functional during PCa evolution [56, 57].
Fig. 5
Fig. 5

Identified biomarker microRNAs and their targets in MPCa-specific microRNA-mRNA network. Elliptic, triangular and rectangular nodes represent microRNAs, TF genes and non-TF genes, respectively. Nodes in grey represent genes that are uniquely regulated by single microRNAs in the network. MPCa metastatic prostate cancer; TF transcription factor

Gene ontology (GO) annotation

The GO analysis was performed using the online tool DAVID at three levels, i.e., biological process (BP), cellular component (CC), and molecular function (MF). The statistically significant terms (adj.p value < 0.05) of each level were summarized in Additional file 5. Here we mainly focused on the top ten significantly enriched terms for in-depth analyses. As shown in Table 3, at the BP level, the most significant terms were closely relevant to cell cycle, metabolic processes and cell death. Accumulating evidence demonstrated that a number of genes as well as non-coding RNAs played functions in PCa metastasis by dramatically activating or inhibiting the cell cycle process [5860]. Wang et al. analyzed the gene expression data of castration-resistant PCa and found that the identified regulatory modules were also enriched in the phosphorus metabolic process [61]. At the CC level, the enriched terms were mainly concentrated on nuclear lumen, nucleoplasm, and organelle lumen. Nucleophosmin (NPM1) is a nucleoprotein and associated with tumor growth. Destouches et al. showed that the phosphorylated NPM1 may interact with androgen receptor in nucleoplasm, which is biologically important in PCa progression [62]. At the MF level, most enriched terms were linked with molecular activities, including transcription regulator activity, transcription repressor activity, transcription activator activity, etc. Grubb RL et al. found that the transcription regulatory protein STAT3 differed statistically in PCa with high Gleason grade (≥ 8) [63]. Xiao et al. showed that the loss expression of PLZF, a transcription repressor in oncogenesis, correlated with PCa tumor aggressiveness [64], which highlighted the functional importance of transcription repressor activity in PCa metastasis.
Table 3

Top ten significant GO terms enriched by targets of the identified microRNA biomarkers

Category

GO terms

Number of enriched genes

Adj.p value

BP

Mitotic cell cycle

60

3.52E−07

Phosphorus metabolic process

112

4.00E−06

Phosphate metabolic process

112

4.00E−06

Regulation of apoptosis

97

3.51E−06

Positive regulation of cell proliferation

61

2.83E−06

Cell cycle process

75

3.16E−06

Regulation of programmed cell death

97

2.92E−06

Regulation of cell death

97

3.02E−06

Cell cycle

93

4.11E−06

Regulation of transcription from RNA polymerase II promoter

88

6.10E−06

CC

Nuclear lumen

162

1.17E−13

Nucleoplasm

110

7.25E−12

Organelle lumen

182

1.05E−11

Membrane-enclosed lumen

183

2.45E−11

Intracellular organelle lumen

176

5.37E−11

Nucleoplasm part

77

1.95E−10

Intracellular non-membrane-bounded organelle

212

5.33E−06

Non-membrane-bounded organelle

212

5.33E−06

Chromatin remodeling complex

18

3.26E−05

Nucleolus

72

1.81E−04

MF

Transcription regulator activity

163

1.16E−09

Transcription repressor activity

50

1.35E−06

Transcription factor binding

67

4.45E−06

Transcription activator activity

56

1.36E−05

Transcription factor activity

100

1.41E−04

Protein kinase activity

68

5.01E−04

Transcription cofactor activity

46

1.07E−03

Protein serine/threonine kinase activity

50

3.67E−03

DNA binding

190

5.00E−03

Phosphoprotein phosphatase activity

25

8.40E−03

GO gene ontology; BP biological process; CC cellular component; MF molecular function; adj.p value: adjusted p value

Pathway enrichment analysis

To investigate the functional mechanisms of the five microRNA candidates, we performed the KEGG and IPA pathway enrichment analysis on their targets using DAVID and IPA program, respectively. The significantly enriched terms (adj.p value < 0.05) were listed in Additional files 6 and 7, respectively. Here the top ten significant terms were mainly selected for further literature exploration. As shown in Fig. 6a, the most meaningful KEGG terms were Axon guidance, Pathways in cancer, Cell cycle, Prostate cancer, and MAPK signaling pathway. Ding et al. found that Semaphorin 4F (S4F), which played important roles in embryologic axon guidance, was a key regulator in the tumor microenvironment and a biomarker of aggressive PCa [65]. McNair et al. uncovered that the cell cycle-coupled expansion of AR activity promoted the progression of PCa and was related to the development of PCa metastases [66]. As shown in Fig. 7, the targets enriched in the prostate cancer pathway had close relations with cell cycle. They potentially mediated the process of cell proliferation and cell survival. Two tumor suppressors, i.e., PTEN and p27 (CDKN1B), which showed the prognostic or therapeutic value in PCa metastasis and recurrence [56, 67], were functionally regulated by the identified microRNAs.
Fig. 6
Fig. 6

Pathway enrichment analysis for targets of the identified microRNA biomarkers. The statistical significance level (adj. p value) was negative 10-based log transformed. a The top ten significant KEGG terms. b The top ten significant IPA terms. adj.p value adjusted p value; KEGG Kyoto Encyclopedia of Genes and Genomes; IPA ingenuity pathway analysis

Fig. 7
Fig. 7

The prostate cancer pathway enriched in KEGG. Objects with pentagrams are acting locus by mapped genes. KEGG Kyoto Encyclopedia of Genes and Genomes

As shown in Fig. 6b, several pathways were also significantly enriched in IPA, such as molecular mechanisms of cancer, axonal guidance signaling, and ERK/MAPK signaling etc. Substantial efforts have convinced that the MAPK signaling was involved in the progression of advanced or metastatic PCa [6870]. As illustrated in Fig. 8, the targets of identified biomarkers were almost enriched in the hubs of ERK/MAPK signaling, which demonstrated the regulatory power of these microRNAs. Chen et al. reviewed the relation between PI3K/AKT signaling and PCa tumorigenesis and pointed out that this pathway regulated tumor cell invasion during the metastasis of PCa cells [71]. Kassi et al. reported that glucocorticoids could mediate the transcriptional regulation of genes which were functional in PCa cell growth, inflammation, differentiation, apoptosis, and metastasis, and the glucocorticoids receptor signaling participated in PCa through cross talking with other signaling cascades [72]. Another well-studied pathway associated with PCa development is TGF-β signaling. It is reported that the TGF-β pathway held the potential to maintain tissue homeostasis and was functional during cancer cell proliferation [73]. Bonci et al. found that the concomitant decrease of miR-15/16 and increase of miR-21 could abnormally activate TGF-β signaling, leading to the invasion, migration and distant bone metastasis of PCa cells [12]. Zhang et al. demonstrated that the TGF-β pathway was significantly enriched by genes in the identified biomarker modules for PCa subtyping [19]. In this study, as illustrated in Additional files 7 and 8, respectively, the TGF-β signaling was also shown statistical significance as the target of the five microRNAs. Moreover, TGFBR1 and TGFBR2, two transforming growth factor beta receptors with the power of transferring TGF-beta signal from the extracellular space to the cytoplasm, were closely regulated by these microRNAs, which could strengthen the importance of our findings.
Fig. 8
Fig. 8

The ERK/MAPK signaling enriched in IPA. Objects with purple circles or triangles are acting locus by mapped genes. ERK extracellular signal-regulated kinases; MAPK mitogen-activated protein kinase; IPA ingenuity pathway analysis

We noticed that the targets of the identified microRNAs were also correlated with bladder cancer, glioma, pancreatic cancer and colorectal cancer, which indicated the similarity of pathogenesis between PCa metastasis and other cancers. For example, TRAP1/Hsp75 has been reported to be a molecular marker in metastatic PCa. Li et al. proved that the expression of TRAP1 was increased in glioma compared with its normal controls, and it could be a useful prognostic factor in glioma management [74]. The RNASEL germline variants were associated with not only familial PCa, but also pancreatic cancer, which indicated the potential mechanisms between pancreatic cancer and PCa development [75].

Discussion

PCa is a commonly diagnosed cancer among males around the world. While the overall survival rate has increased these years, the metastasis is still a leading cause of PCa death [76]. The early detection of metastatic signature is important for monitoring PCa prognosis and helping design personalized therapeutic strategies. It is widely acknowledged that microRNAs are a class of functional regulators in biological processes and have good sensitivity and specificity to be biomarkers for disease initiation and progression [77].

In this study, we integrated microRNA/mRNA expression data with network structural knowledge and improved the bioinformatics model to screen candidate microRNA biomarkers for predicting PCa metastasis. Compared with the previous work, we updated the reference microRNA-mRNA network by carefully integrating recently reported human microRNA-mRNA associations based on microRNA nomenclature. Besides the single-line regulatory power (NSR) and biological roles of microRNAs (TFP), we considered the functional importance of genes that are uniquely regulated by single microRNAs in this model. It is reasonable that the single-line regulatory site in the network is relatively vulnerable and TFs are important regulators in various biological activities. The incorporative analysis of structural and functional characteristics in microRNA-mRNA regulation would strengthen the evalutaion of microRNA regulatory power. Hence we defined a new parameter called UTP to quantify the special regulation of given microRNAs, that is, the percentage of TF genes in microRNA unique targets.

Based on this computational model, a total of five microRNAs, i.e., miR-204-5p, miR-101-3p, miR-145-5p, miR-198, and miR-152, were identified as candidate biomarkers for PCa metastasis prediction. Among them, miR-145-5p and miR-204-5p were validated as tumor suppressor microRNAs, and their down-regulation could disorder cell cycle processes and finally result in PCa invasion [27, 48]. More importantly, miR-145-5p and miR-101-3p have been reported as biomarkers for evaluating PCa metastasis previously [43, 44]. The remaining three were also confirmed to be connected with the development of metastatic or high-grade PCa according to PubMed literature searches. Compared with POMA solely using NSR and TFP as filters, the MicroRNA-BD model improved the prediction precision from 28.6% (2/7) to 40% (2/5) in this case study. Furthermore, the AUC values for the prediction performance of the five microRNAs, respectively, ranged from 0.70 to 0.99 and from 0.71 to 0.99 in two datasets, i.e., the prediction set and another independent validation set, which indicated the power of the identified biomarkers for PCa prognosis and metastasis tracking.

We further investigated the pathogenic mechanisms of the identified microRNAs in PCa metastasis through GO and pathway enrichment analyses. The cell cycle process, for example, is one of the most significant terms enriched in GO and KEGG pathway, which supported the pivotal view of cell cycle-mediated PCa carcinogenesis [66, 78, 79]. The prostate cancer pathway, ERK/MAPK signaling, and TGF-β signaling are well-studied in PCa metastasis, and we found that most of the targets of identified microRNA biomarkers were the key components of these pathways, meanwhile, some tumor suppressor genes, e.g., PTEN and p27, were regulated by the identified microRNAs. In addition, the remaining pathways, such as axonal guidance [65], PI3K/AKT signaling [71], glucocorticoids receptor signaling [72], pancreatic cancer [75], and molecular mechanisms of cancer [80] etc., were all involved in PCa progression and metastasis according to previous reports.

Compared with the existing approaches for PCa metastasis investigation, our model detects microRNA biomarkers based on statistical evidences from a combination of network sub-structural and functional analyses. In the study of Zhang et al., five microRNA-mRNA modules were identified for PPCa and MPCa respectively based on the correction and clustering analysis on microRNA and mRNA datasets [19]. Similar to this idea, our study also utilized microRNA-mRNA network information. However, the network in our model was concentrated more as a biological system, where special regulatory patterns altering its stability were concluded as the principle for biomarker prediction. Another major difference between the two approaches regards to the methodology for sub-network extraction. In contrast to the two types of network modules, i.e., PPCa- and MPCa-module, built in Zhang et al. using clustering and condensing techniques, [19], our model constructed microRNA-mRNA network specific to PCa metastasis in order to capture the changing signatures during PCa evolution, and only single microRNAs could be screened as candidate biomarkers for PCa subtyping. Considering the results, the miR-145-5p detected by our model was also involved in one of the modules in Zhang et al. [19], which convinced the underlying power of this microRNA for PCa metastasis predicting. From pathway angles, both Zhang et al. [19] and Bonci et al. [12] demonstrated the importance of TGF-β signaling as microRNA targets in PCa progression and metastasis. As described in the section of ‘Pathway enrichment analysis’, we found that genes regulated by the five biomarker microRNAs were similarly enriched in TGF-β signaling, which indicated the pathway-level consistency of results for PCa carcinogenesis decoding across these studies. Due to the complexity and diversity in PCa development, biomarker microRNAs identified by different methods tended to be highly heterogeneous. In systems biology viewpoints, living organisms are often treated as a holistic framework, thus identifying module or network biomarkers catering to the dynamical nature of PCa pathogenesis for personalized prognosis and treatment is our next-step action.

It should not be ignored that some limitations still existed in this study. Firstly, genes in the present model were treated equally. As we known, the importance of genes in different biological activities is not the same, more functional annotations need to be weighted reasonably. Secondly, only TF genes were selected in the study, the specific knowledge to PCa metastasis could be considered in order to provide precise strategies for MPCa early detection and treatment. Thirdly, only 618 microRNAs were recorded in our reconstructed human microRNA-mRNA network, the network scale should keep pace with the development of newly identified microRNA-mRNA associations. Last but the most important, we are trying to collect human PPCa and MPCa samples to further perform wet lab verifications for future carcinogenic exploration and translational application.

Conclusion

In this study, a total of five microRNAs, i.e, miR-204-5p, miR-101-3p, miR-145-5p, miR-198, and miR-152, were identified as candidate biomarkers for predicting PCa metastasis based on a novel bioinformatics model. The prediction performance, literature exploration and functional enrichment analysis convinced our findings. More clinical validations are needed in our future translational application.

Abbreviations

PCa: 

prostate cancer

PPCa: 

primary prostate cancer

MPCa: 

metastatic prostate cancer

LIMMA: 

linear models for microarray data analysis

eBayes: 

the empirical bayes

adj.p value: 

adjusted p value

FC: 

fold change

DE: 

differentially expressed

TF: 

transcription factor

MicroRNA-BD: 

microRNA biomarker discovery

POMA: 

pipeline of outlier microRNA analysis

NSR: 

number of single-line regulation

TFP: 

transcription factor gene percentage

UTP: 

unique-regulated TF gene percentage

ROC: 

receiver operating characteristic curve

AUC: 

area under the curve

GO: 

gene ontology

BP: 

biological process

CC: 

cellular component

MF: 

molecular function

KEGG: 

Kyoto Encyclopedia of Genes and Genomes

DAVID: 

Database for Annotation, Visualization and Integrated Discovery

IPA: 

ingenuity pathway analysis

Declarations

Authors’ contributions

BS and YL conceptualized the research and developed the bioinformatics model. YL, CD, and ZS collected the data. YL, FC, LS, XT and HD performed the computational analysis. YL, JC and BS drafted and revised the manuscript. All authors read and approved the final manuscript.

Acknowledgements

The authors gratefully thank the academic editor and the anonymous reviewers for their insightful comments and suggestions to improve this manuscript. They also thank Prof. Luonan Chen at Key Laboratory of Systems Biology, Shanghai Institute of Biological Sciences, Chinese Academy of Sciences for providing the IPA analysis.

Competing interests

The authors delcare that they have no competing interests.

Availability of data and materials

The data generated or analysed during this study are available from the corresponding author upon reasonable request.

Consent for publication

Not applicable.

Ethics approval and consent to participate

Not applicable.

Funding

This study was supported by the National Natural Science Foundation of China (Grant Nos. 31670851, 31470821,31600670, 91530320, and 31770903), National Key Research and Development Program of China (2016YFC1306605).

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.

Authors’ Affiliations

(1)
Center for Systems Biology, Soochow University, Suzhou, 215006, China
(2)
Department of Genetics & Systems Biology Institute, Yale University School of Medicine, West Haven, CT 06516, USA
(3)
Department of Statistics and Data Science, Yale University, New Haven, CT 06511, USA
(4)
Jiangsu Health Vocational College, Nanjing, 211800, China
(5)
Suzhou Industrial Park Institute of Services Outsourcing, Suzhou, 215123, China
(6)
School of Chemistry, Biology and Material Engineering, Suzhou University of Science and Technology, Suzhou, 215011, China
(7)
Center for Translational Biomedical Informatics, Guizhou University School of Medicine, Guiyang, 550025, China
(8)
Institute for Systems Genetics, West China Hospital, Sichuan University, Chengdu, 610041, China

References

  1. Siegel RL, Miller KD, Jemal A. Cancer statistics, 2015. CA Cancer J Clin. 2015;65:5–29.View ArticlePubMedGoogle Scholar
  2. Chen J, Shao P, Cao Q, Li P, Li J, Cai H, Zhu J, Wang M, Zhang Z, Qin C, Yin C. Genetic variations in a PTEN/AKT/mTOR axis and prostate cancer risk in a Chinese population. PLoS ONE. 2012;7:e40817.View ArticlePubMedPubMed CentralGoogle Scholar
  3. Fu Q, Liu X, Liu Y, Yang J, Lv G, Dong S. MicroRNA-335 and -543 suppress bone metastasis in prostate cancer via targeting endothelial nitric oxide synthase. Int J Mol Med. 2015;36:1417–25.View ArticlePubMedGoogle Scholar
  4. Bartel DP. MicroRNAs: genomics, biogenesis, mechanism, and function. Cell. 2004;116:281–97.View ArticlePubMedGoogle Scholar
  5. Cui Q, Yu Z, Purisima EO, Wang E. Principles of microRNA regulation of a human cellular signaling network. Mol Syst Biol. 2006;2:46.View ArticlePubMedPubMed CentralGoogle Scholar
  6. Dang TM, Wong WC, Ong SM, Li P, Lum J, Chen J, Poidinger M, Zolezzi F, Wong SC. MicroRNA expression profiling of human blood monocyte subsets highlights functional differences. Immunology. 2015;145:404–16.View ArticlePubMedPubMed CentralGoogle Scholar
  7. Chen X, Ba Y, Ma L, Cai X, Yin Y, Wang K, Guo J, Zhang Y, Chen J, Guo X, et al. Characterization of microRNAs in serum: a novel class of biomarkers for diagnosis of cancer and other diseases. Cell Res. 2008;18:997–1006.View ArticlePubMedGoogle Scholar
  8. Mitchell PS, Parkin RK, Kroh EM, Fritz BR, Wyman SK, Pogosova-Agadjanyan EL, Peterson A, Noteboom J, O’Briant KC, Allen A, et al. Circulating microRNAs as stable blood-based markers for cancer detection. Proc Natl Acad Sci USA. 2008;105:10513–8.View ArticlePubMedPubMed CentralGoogle Scholar
  9. He F, Lv P, Zhao X, Wang X, Ma X, Meng W, Meng X, Dong S. Predictive value of circulating miR-328 and miR-134 for acute myocardial infarction. Mol Cell Biochem. 2014;394:137–44.View ArticlePubMedGoogle Scholar
  10. Wang L, Song G, Tan W, Qi M, Zhang L, Chan J, Yu J, Han J, Han B. MiR-573 inhibits prostate cancer metastasis by regulating epithelial-mesenchymal transition. Oncotarget. 2015;6:35978–90.PubMedPubMed CentralGoogle Scholar
  11. Damodaran C, Das TP, Papu John AM, Suman S, Kolluru V, Morris TJ, Faber EN, Rai SN, Messer JC, Alatassi H, Ankem MK. miR-301a expression: a prognostic marker for prostate cancer. Urol Oncol. 2016;34(336):e313–20.Google Scholar
  12. Bonci D, Coppola V, Patrizii M, Addario A, Cannistraci A, Francescangeli F, Pecci R, Muto G, Collura D, Bedini R, et al. A microRNA code for prostate cancer metastasis. Oncogene. 2016;35:1180–92.View ArticlePubMedGoogle Scholar
  13. Tang Y, Yan W, Chen J, Luo C, Kaipia A, Shen B. Identification of novel microRNA regulatory pathways associated with heterogeneous prostate cancer. BMC Syst Biol. 2013;7(Suppl 3):S6.View ArticlePubMedPubMed CentralGoogle Scholar
  14. Lin Y, Chen J, Shen B. Interactions between genetics, lifestyle, and environmental factors for healthcare. Adv Exp Med Biol. 2017;1005:167–91.View ArticlePubMedGoogle Scholar
  15. Yan W, Wang S, Sun Z, Lin Y, Sun S, Chen J, Chen W. Identification of microRNAs as potential biomarker for gastric cancer by system biological analysis. Biomed Res Int. 2014;2014:901428.PubMedPubMed CentralGoogle Scholar
  16. Zhao XM, Liu KQ, Zhu G, He F, Duval B, Richer JM, Huang DS, Jiang CJ, Hao JK, Chen L. Identifying cancer-related microRNAs based on gene expression data. Bioinformatics. 2015;31:1226–34.View ArticlePubMedGoogle Scholar
  17. Sun SY, Liu ZP, Zeng T, Wang Y, Chen L. Spatio-temporal analysis of type 2 diabetes mellitus based on differential expression networks. Sci Rep. 2013;3:2268.View ArticlePubMedPubMed CentralGoogle Scholar
  18. Cogill S, Wang L. Support vector machine model of developmental brain gene expression data for prioritization of Autism risk gene candidates. Bioinformatics. 2016;32:3611–8.PubMedGoogle Scholar
  19. Zhang W, Edwards A, Fan W, Flemington EK, Zhang K. miRNA-mRNA correlation-network modules in human prostate cancer and the differences between primary and metastatic tumor subtypes. PLoS ONE. 2012;7:e40130.View ArticlePubMedPubMed CentralGoogle Scholar
  20. Zhang W, Zang J, Jing X, Sun Z, Yan W, Yang D, Shen B, Guo F. Identification of candidate miRNA biomarkers from miRNA regulatory network with application to prostate cancer. J Transl Med. 2014;12:66.View ArticlePubMedPubMed CentralGoogle Scholar
  21. Yan W, Xu L, Sun Z, Lin Y, Zhang W, Chen J, Hu S, Shen B. MicroRNA biomarker identification for pediatric acute myeloid leukemia based on a novel bioinformatics model. Oncotarget. 2015;6:26424–36.PubMedPubMed CentralGoogle Scholar
  22. Zhu J, Wang S, Zhang W, Qiu J, Shan Y, Yang D, Shen B. Screening key microRNAs for castration-resistant prostate cancer based on miRNA/mRNA functional synergistic network. Oncotarget. 2015;6:43819–30.PubMedPubMed CentralGoogle Scholar
  23. Taylor BS, Schultz N, Hieronymus H, Gopalan A, Xiao Y, Carver BS, Arora VK, Kaushik P, Cerami E, Reva B, et al. Integrative genomic profiling of human prostate cancer. Cancer Cell. 2010;18:11–22.View ArticlePubMedPubMed CentralGoogle Scholar
  24. Varambally S, Yu J, Laxman B, Rhodes DR, Mehra R, Tomlins SA, Shah RB, Chandran U, Monzon FA, Becich MJ, et al. Integrative genomic and proteomic analysis of prostate cancer reveals signatures of metastatic progression. Cancer Cell. 2005;8:393–406.View ArticlePubMedGoogle Scholar
  25. Edgar R, Domrachev M, Lash AE. Gene expression omnibus: NCBI gene expression and hybridization array data repository. Nucleic Acids Res. 2002;30:207–10.View ArticlePubMedPubMed CentralGoogle Scholar
  26. Peng X, Guo W. Altered expression of microRNAs in bone metastasis compared with human prostate cancer. 2012. https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE26964. Accessed 10 Apr 2018.
  27. Vanacore D, Boccellino M, Rossetti S, Cavaliere C, D’Aniello C, Di Franco R, Romano FJ, Montanari M, La Mantia E, Piscitelli R, et al. Micrornas in prostate cancer: an overview. Oncotarget. 2017;8:50240–51.View ArticlePubMedPubMed CentralGoogle Scholar
  28. Ritchie ME, Phipson B, Wu D, Hu Y, Law CW, Shi W, Smyth GK. limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Res. 2015;43:e47.View ArticlePubMedPubMed CentralGoogle Scholar
  29. Smyth GK. Linear models and empirical bayes methods for assessing differential expression in microarray experiments. Stat Appl Genet Mol Biol. 2004;3:1–25.View ArticleGoogle Scholar
  30. Benjamini Y, Hochberg Y. Controlling the false discovery rate—a practical and powerful approach to multiple testing. J R Stat Soc Ser B Methodol. 1995;57:289–300.Google Scholar
  31. Kozomara A, Griffiths-Jones S. miRBase: annotating high confidence microRNAs using deep sequencing data. Nucleic Acids Res. 2014;42:D68–73.View ArticlePubMedGoogle Scholar
  32. Hsu SD, Tseng YT, Shrestha S, Lin YL, Khaleel A, Chou CH, Chu CF, Huang HY, Lin CM, Ho SY, et al. miRTarBase update 2014: an information resource for experimentally validated miRNA-target interactions. Nucleic Acids Res. 2014;42:D78–85.View ArticlePubMedGoogle Scholar
  33. Vergoulis T, Vlachos IS, Alexiou P, Georgakilas G, Maragkakis M, Reczko M, Gerangelos S, Koziris N, Dalamagas T, Hatzigeorgiou AG. TarBase 6.0: capturing the exponential growth of miRNA targets with experimental support. Nucleic Acids Res. 2012;40:D222–9.View ArticlePubMedGoogle Scholar
  34. Xiao F, Zuo Z, Cai G, Kang S, Gao X, Li T. miRecords: an integrated resource for microRNA-target interactions. Nucleic Acids Res. 2009;37:D105–10.View ArticlePubMedGoogle Scholar
  35. Jiang Q, Wang Y, Hao Y, Juan L, Teng M, Zhang X, Li M, Wang G, Liu Y. miR2Disease: a manually curated database for microRNA deregulation in human disease. Nucleic Acids Res. 2009;37:D98–104.View ArticlePubMedGoogle Scholar
  36. Gennarino VA, Sardiello M, Mutarelli M, Dharmalingam G, Maselli V, Lago G, Banfi S. HOCTAR database: a unique resource for microRNA target prediction. Gene. 2011;480:51–8.View ArticlePubMedPubMed CentralGoogle Scholar
  37. Gamazon ER, Im HK, Duan S, Lussier YA, Cox NJ, Dolan ME, Zhang W. Exprtarget: an integrative approach to predicting human microRNA targets. PLoS ONE. 2010;5:e13534.View ArticlePubMedPubMed CentralGoogle Scholar
  38. Li JH, Liu S, Zhou H, Qu LH, Yang JH. starBase v2.0: decoding miRNA-ceRNA, miRNA-ncRNA and protein-RNA interaction networks from large-scale CLIP-Seq data. Nucleic Acids Res. 2014;42:D92–7.View ArticlePubMedGoogle Scholar
  39. Chongsuvivatwong V. Epicalc: epidemiological calculator, R package version 2.15.1.0. 2012. https://cran.r-project.org/src/contrib/Archive/epicalc/. Accessed 18 Apr 2018.
  40. Kanehisa M, Goto S. KEGG: kyoto encyclopedia of genes and genomes. Nucleic Acids Res. 2000;28:27–30.View ArticlePubMedPubMed CentralGoogle Scholar
  41. Kramer A, Green J, Pollard J Jr, Tugendreich S. Causal analysis approaches in ingenuity pathway analysis. Bioinformatics. 2014;30:523–30.View ArticlePubMedGoogle Scholar
  42. da Huang W, Sherman BT, Lempicki RA. Systematic and integrative analysis of large gene lists using DAVID bioinformatics resources. Nat Protoc. 2009;4:44–57.View ArticleGoogle Scholar
  43. Watahiki A, Wang Y, Morris J, Dennis K, O’Dwyer HM, Gleave M, Gout PW, Wang Y. MicroRNAs associated with metastatic prostate cancer. PLoS ONE. 2011;6:e24950.View ArticlePubMedPubMed CentralGoogle Scholar
  44. Pang Y, Young CY, Yuan H. MicroRNAs and prostate cancer. Acta Biochim Biophys Sin. 2010;42:363–9.View ArticlePubMedGoogle Scholar
  45. Huang S, Guo W, Tang Y, Ren D, Zou X, Peng X. miR-143 and miR-145 inhibit stem cell characteristics of PC-3 prostate cancer cells. Oncol Rep. 2012;28:1831–7.View ArticlePubMedGoogle Scholar
  46. Ren D, Wang M, Guo W, Zhao X, Tu X, Huang S, Zou X, Peng X. Wild-type p53 suppresses the epithelial-mesenchymal transition and stemness in PC-3 prostate cancer cells by modulating miR145. Int J Oncol. 2013;42:1473–81.View ArticlePubMedGoogle Scholar
  47. Chakravarthi BV, Goswami MT, Pathi SS, Robinson AD, Cieslik M, Chandrashekar DS, Agarwal S, Siddiqui J, Daignault S, Carskadon SL, et al. MicroRNA-101 regulated transcriptional modulator SUB1 plays a role in prostate cancer. Oncogene. 2016;35:6330–40.View ArticlePubMedPubMed CentralGoogle Scholar
  48. Lin YC, Lin JF, Tsai TF, Chou KY, Chen HE, Hwang TI. Tumor suppressor miRNA-204-5p promotes apoptosis by targeting BCL2 in prostate cancer cells. Asian J Surg. 2017;40:396–406.View ArticlePubMedGoogle Scholar
  49. Todorova K, Metodiev MV, Metodieva G, Zasheva D, Mincheff M, Hayrabedyan S. miR-204 is dysregulated in metastatic prostate cancer in vitro. Mol Carcinog. 2016;55:131–47.View ArticlePubMedGoogle Scholar
  50. Theodore SC, Davis M, Zhao F, Wang H, Chen D, Rhim J, Dean-Colomb W, Turner T, Ji W, Zeng G, et al. MicroRNA profiling of novel African American and Caucasian Prostate Cancer cell lines reveals a reciprocal regulatory relationship of miR-152 and DNA methyltranferase 1. Oncotarget. 2014;5:3512–25.View ArticlePubMedPubMed CentralGoogle Scholar
  51. Walter BA, Valera VA, Pinto PA, Merino MJ. Comprehensive microRNA profiling of prostate cancer. J Cancer. 2013;4:350–7.View ArticlePubMedPubMed CentralGoogle Scholar
  52. Lakshman M, Huang X, Ananthanarayanan V, Jovanovic B, Liu Y, Craft CS, Romero D, Vary CP, Bergan RC. Endoglin suppresses human prostate cancer metastasis. Clin Exp Metastasis. 2011;28:39–53.View ArticlePubMedGoogle Scholar
  53. Yan Z, Jin S, Wei Z, Huilian H, Zhanhai Y, Yue T, Juan L, Jing L, Libo Y, Xu L. Discoidin domain receptor 2 facilitates prostate cancer bone metastasis via regulating parathyroid hormone-related protein. Biochim Biophys Acta. 2014;1842:1350–63.View ArticlePubMedGoogle Scholar
  54. Caley MP, King H, Shah N, Wang K, Rodriguez-Teja M, Gronau JH, Waxman J, Sturge J. Tumor-associated Endo180 requires stromal-derived LOX to promote metastatic prostate cancer cell migration on human ECM surfaces. Clin Exp Metastasis. 2016;33:151–65.View ArticlePubMedGoogle Scholar
  55. Sun GG, Wang YD, Cui DW, Cheng YJ, Hu WN. EMP1 regulates caspase-9 and VEGFC expression and suppresses prostate cancer cell proliferation and invasion. Tumour Biol. 2014;35:3455–62.View ArticlePubMedGoogle Scholar
  56. Noh BJ, Sung JY, Kim YW, Chang SG, Park YK. Prognostic value of ERG, PTEN, CRISP3 and SPINK1 in predicting biochemical recurrence in prostate cancer. Oncol Lett. 2016;11:3621–30.View ArticlePubMedPubMed CentralGoogle Scholar
  57. Nowak DG, Cho H, Herzka T, Watrud K, DeMarco DV, Wang VM, Senturk S, Fellmann C, Ding D, Beinortas T, et al. MYC drives Pten/Trp53-deficient proliferation and metastasis due to IL6 secretion and AKT suppression via PHLPP2. Cancer Discov. 2015;5:636–51.View ArticlePubMedPubMed CentralGoogle Scholar
  58. Xu B, Tao T, Wang Y, Fang F, Huang Y, Chen S, Zhu W, Chen M. hsa-miR-135a-1 inhibits prostate cancer cell growth and migration by targeting EGFR. Tumour Biol. 2016;37:14141–51.View ArticlePubMedGoogle Scholar
  59. Pan XW, Gan SS, Ye JQ, Fan YH, Hong U, Chu CM, Gao Y, Li L, Liu X, Chen L, et al. SMC1A promotes growth and migration of prostate cancer in vitro and in vivo. Int J Oncol. 2016;49:1963–72.View ArticlePubMedGoogle Scholar
  60. Zhang SR, Yang JK, Xie JK, Zhao LC. Long noncoding RNA HOTTIP contributes to the progression of prostate cancer by regulating HOXA13. Cell Mol Biol. 2016;62:84–8.Google Scholar
  61. Wang X, Wen J, Li R, Qiu G, Zhou L, Wen X. Gene expression profiling analysis of castration-resistant prostate cancer. Med Sci Monit. 2015;21:205–12.View ArticlePubMedPubMed CentralGoogle Scholar
  62. Destouches D, Sader M, Terry S, Marchand C, Maille P, Soyeux P, Carpentier G, Semprez F, Ceraline J, Allory Y, et al. Implication of NPM1 phosphorylation and preclinical evaluation of the nucleoprotein antagonist N6L in prostate cancer. Oncotarget. 2016;7:69397.View ArticlePubMedPubMed CentralGoogle Scholar
  63. Grubb RL, Deng J, Pinto PA, Mohler JL, Chinnaiyan A, Rubin M, Linehan WM, Liotta LA, Petricoin EF, Wulfkuhle JD. Pathway biomarker profiling of localized and metastatic human prostate cancer reveal metastatic and prognostic signatures. J Proteome Res. 2009;8:3044–54.View ArticlePubMedPubMed CentralGoogle Scholar
  64. Xiao GQ, Unger P, Yang Q, Kinoshita Y, Singh K, McMahon L, Nastiuk K, Sha K, Krolewski J, Burstein D. Loss of PLZF expression in prostate cancer by immunohistochemistry correlates with tumor aggressiveness and metastasis. PLoS ONE. 2015;10:e0121318.View ArticlePubMedPubMed CentralGoogle Scholar
  65. Ding Y, He D, Florentin D, Frolov A, Hilsenbeck S, Ittmann M, Kadmon D, Miles B, Rowley D, Ayala G. Semaphorin 4F as a critical regulator of neuroepithelial interactions and a biomarker of aggressive prostate cancer. Clin Cancer Res. 2013;19:6101–11.View ArticlePubMedPubMed CentralGoogle Scholar
  66. McNair C, Urbanucci A, Comstock CE, Augello MA, Goodwin JF, Launchbury R, Zhao SG, Schiewer MJ, Ertel A, Karnes J, et al. Cell cycle-coupled expansion of AR activity promotes cancer progression. Oncogene. 2016;36:1655–68.View ArticlePubMedPubMed CentralGoogle Scholar
  67. Zhu W, Xue Y, Liang C, Zhang R, Zhang Z, Li H, Su D, Liang X, Zhang Y, Huang Q, et al. S100A16 promotes cell proliferation and metastasis via AKT and ERK cell signaling pathways in human prostate cancer. Tumour Biol. 2016;37:12241–50.View ArticlePubMedGoogle Scholar
  68. Lim W, Park S, Bazer FW, Song G. Naringenin-induced apoptotic cell death in prostate cancer cells is mediated via the PI3K/AKT and MAPK signaling pathways. J Cell Biochem. 2016;118:1118–31.View ArticleGoogle Scholar
  69. Browne AJ, Gobel A, Thiele S, Hofbauer LC, Rauner M, Rachner TD. p38 MAPK regulates the Wnt inhibitor Dickkopf-1 in osteotropic prostate cancer cells. Cell Death Dis. 2016;7:e2119.View ArticlePubMedPubMed CentralGoogle Scholar
  70. Wang J, Kobayashi T, Floc’h N, Kinkade CW, Aytes A, Dankort D, Lefebvre C, Mitrofanova A, Cardiff RD, McMahon M, et al. B-Raf activation cooperates with PTEN loss to drive c-Myc expression in advanced prostate cancer. Cancer Res. 2012;72:4765–76.View ArticlePubMedPubMed CentralGoogle Scholar
  71. Chen H, Zhou L, Wu X, Li R, Wen J, Sha J, Wen X. The PI3K/AKT pathway in the pathogenesis of prostate cancer. Front Biosci. 2016;21:1084–91.View ArticleGoogle Scholar
  72. Kassi E, Moutsatsou P. Glucocorticoid receptor signaling and prostate cancer. Cancer Lett. 2011;302:1–10.View ArticlePubMedGoogle Scholar
  73. Massague J, Gomis RR. The logic of TGFbeta signaling. FEBS Lett. 2006;580:2811–20.View ArticlePubMedGoogle Scholar
  74. Li S, Lv Q, Sun H, Xue Y, Wang P, Liu L, Li Z, Li Z, Tian X, Liu YH. Expression of TRAP1 predicts poor survival of malignant glioma patients. J Mol Neurosci. 2015;55:62–8.View ArticlePubMedGoogle Scholar
  75. Bartsch DK, Fendrich V, Slater EP, Sina-Frey M, Rieder H, Greenhalf W, Chaloupka B, Hahn SA, Neoptolemos JP, Kress R. RNASEL germline variants are associated with pancreatic cancer. Int J Cancer. 2005;117:718–22.View ArticlePubMedGoogle Scholar
  76. Gupta GP, Massague J. Cancer metastasis: building a framework. Cell. 2006;127:679–95.View ArticlePubMedGoogle Scholar
  77. Lin Y, Qian F, Shen L, Chen F, Chen J, Shen B. Computer-aided biomarker discovery for precision medicine: data resources, models and applications. Brief Bioinform. 2017. https://doi.org/10.1093/bib/bbx158.Google Scholar
  78. Heger Z, Merlos Rodrigo MA, Michalek P, Polanska H, Masarik M, Vit V, Plevova M, Pacik D, Eckschlager T, Stiborova M, Adam V. Sarcosine up-regulates expression of genes involved in cell cycle progression of metastatic models of prostate cancer. PLoS ONE. 2016;11:e0165830.View ArticlePubMedPubMed CentralGoogle Scholar
  79. Erdmann K, Kaulke K, Rieger C, Salomo K, Wirth MP, Fuessel S. MiR-26a and miR-138 block the G1/S transition by targeting the cell cycle regulating network in prostate cancer cells. J Cancer Res Clin Oncol. 2016;142:2249–61.View ArticlePubMedGoogle Scholar
  80. Fu W, Madan E, Yee M, Zhang H. Progress of molecular targeted therapies for prostate cancers. Biochim Biophys Acta. 2012;1825:140–52.PubMedGoogle Scholar

Copyright

Advertisement