Screening Of Differentially Expressed MiRNAs And Integrated Analysis Of MicroRNA Regulatory Network In Nasopharyngeal Carcinoma Tissue

BackgroundNasopharyngeal carcinoma (NPC) is a primary nasopharyngeal mucosa malignant tumor. It has a remarkably unusual ethnic and geographic distribution in Southern China and Southeast Asia, especially those of Cantonese origin, which is known as the "Guangdong tumor". The characteristics are insidious onset, highly malignant local invasion and early distant metastasis. Although the multimodality therapy including intensity-modulated radiotherapy (IMRT) is used to control the tumor, the average 5-year survival rate for NPC patients remains at 70%. Thus it is necessary to develop the individualized cancer therapy. However, the individualized cancer therapy should be based on the study of molecular marker screening and differential expressed genes analysis. Screening out the expression profile which is related to the NPC metastasis and angiogenesis, and analyzing the function of important molecular markers can help to figure out the mechanism of NPC. It is significant to improve the NPC treatment and reduce the patients’burden.MicroRNAs (miRNAs) are a recently discovered class of small non-coding RNAs of 18-24 nucleotides in length that modulate gene expression post-transcriptionally. The mature miRNA targets the 3’untranslated region (3’UTR) of its target mRNA, and induces mRNA degradation. It is showed that over 50% of microRNAs locate in cancer related regions. Their abnormal expression plays a key role as oncogenes or tumor suppressor genes, and be involved in the initiation and progression of cancer cells. As miRNAs involve in biological development, cell proliferation, differentiation and apoptosis. In recent years a growing number of studies have found that the miRNA regulated the tumor through the network, and the network was usually based on expression profile. SETD1A regulated p53 through the miR-30 and miR-590-5p network in breast cancer, prostate cancer, lung cancer and other tumors. In pancreatic cancer, miR-21, miR-23a and miR-27a were screened out and conducted an interaction network with the tumor suppressor genes PDCD4, BTG2 and NEDD4L. They worked together to regulate the proliferation, invasion and metastasis of pancreatic cancer. All these studies suggested that the miRNA regulation network plays an important role in tumor development, and might be used as potential therapeutic targets.In nasopharyngeal carcinoma, a lot of miRNAs were showed to play important roles in the cancer development. The dysregulation of miRNAs had an impact on nasopharyngeal carcinoma cell proliferation, migration and invasion, and that was also related to the prognosis of patients. miR-26a were down-regulated in NPC cell lines and nasopharyngeal tissues and inhibited proliferation and invasion of NPC by targeting EZH2. The same, miR-9 was downregulated in NPC and nasopharyngeal carcinoma cells and regulated cell proliferation, migration and invasion by targeting CXCR4. Up to 2016, there were over 30 miRNAs involved in NPC pathogenesis. According to the literature review of miRNAs function and mechanism in NPC tissue, we found there lacked the network analysis based on miRNA expression profile. So we would combine deep sequencing and bioinformatics analysis to screen out differentially expressed microRNAs, and analyze the microRNA regulatory network in nasopharyngeal carcinoma tissue comprehensively and systematically.To better characterise the specific signature of NPC cells, we expanded our observations by applying laser capture microdissection (LCM) on NPC and chronic nasopharyngitis samples. And deep sequencing was used to get a precise differentially expressed miRNA profile. Then several bioinformatics methods were applied to analyze the microRNA regulatory network. In our study, we used not only the Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis but also the Ingenuity pathway analysis (IPA). The GO and KEGG analysis focus on the target genes’ function distribution, while the IPA is good at constructing the network. Previous researches showed IPA can be used to build miRNA regulation network in breast cancer, liver cancer, pancreatic cancer and so on. The networks could help to reveal tumor molecular mechanism and then improve the treatment and prognosis. However, there is no similar study in NPC tissue yet. The miRNA regulation network in NPC progression should be explored in our further study.Thus, our study aimed to screen out the differentially expressed miRNAs between the NPC and chronic nasopharyngitis. Then the function distribution of target genes was analyzed via bioinformatics methods, and the miRNA regulation network was built. We hope that this study will improve the better understanding of NPC pathogenesis and the development of novel effective therapies for NPC.Materials and Methods1. Clinical date1.1 12 primary NPC biopsy specimens and 8 chronic nasopharyngitis biopsies were obtained for the LCM and deep sequencing. Both tumor and normal tissues were histologically confirmed by H&E (hematoxylin and eosin) staining. No patient received chemotherapy or radiotherapy before diagnosis. The clinical features of nasopharyngeal carcinoma group control of nasopharyngeal carcinoma in cases of epithelial tissue information age distribution and the sex ratio was consistent with the group of NPC.1.2201 primary NPC biopsy specimens and 25 chronic nasopharyngitis biopsies were obtained for the Quantitative reverse transcription-PCR (qPCR). The clinical features were consistent with the above.65 NPC biopsies and 20 chronic nasopharyngitis biopsies were selected for the primary validation. And then the cohort would expand to 201 primary NPC biopsy specimens and 25 chronic nasopharyngitis biopsies to figure out the correlation between miRNA and clinical stages.2 Tissue samples collection and frozen sectionThe specimens were immediately put into liquid nitrogen after biopsy. Frozen section temperature was between -21℃ to -25℃, the maximum transverse of biopsy was chose to section, and the thickness of section was 8 um. Each sample got 18 to 20 sections, and fixed them in anhydrous ethanol for 10 minutes. Then the sections were hematoxylin and eosin (H&E) stained before LCM.3 Laser capture microdissection (LCM)LCM was performed on a MMI Cellcut Microdissection Instrument (Molecular Machines & Industries,Swiss). Phase contrast images were acquired using OlympusIX71 microscope. The microdissected tissue was immersed in 100ul RNAlater. Then the samples were stored in -80℃ refrigerator.4 Total RNA extraction and quality examinationTotal RNA were extracted from NPC and chronic nasopharyngitis tissues according to the protocol of Trizol and RNAiso Plus. Agilent 2100 bioanalyzer and ND-1000 trace nucleic acid quantitative instrument were used to access the quality of total RNA respectively. The results were showed by the date of RIN (RNA Integrity Number),28 s/18s and OD260/OD280. The total RNA that passed the examination were used for deep sequencing and qPCR.5 Library construction and deep sequencingThe library construction was according to BGI TruSeq Small RNA sample preparation process. To achieve optimal tissue miRNA profiles, we carried out high-throughput next-generation sequencing (Illumina, BGI, Shenzhen) of 12 NPC samples and 8 chronic nasopharyngitis samples by following the manufacturer’s recommended protocols. We screened the high quality clean read sequences by the alignment to NCBI GenBank data and miRBase 21.0 for the further analysis. Then we applied transcripts per million (TPM) to normalize the expression of miRNA in two groups (NPC and controls). And we calculated fold change (FC) and P-value via Genespring, and corrected P-value into false discovery rate (FDR) using the Benjamin and Hochberg method. FDR≤0.05 and |log2FC|≥2 were set as the cut-offs to screen out differentially expressed miRNAs.6 Real-time PCRTotal RNA and small RNA were extracted from microdissected tissues and cells. The RNA was reversely transcribed into cDNA. Quantitative real-time PCR (qPCR) was performed using SYBR Green PCR master mix. All samples were normalized to U6 and fold changes were calculated through relative quantification (2-ΔCp and 2-ΔΔ Cp).7 Target genes prediction, GO and KEGG analysisFour softwares (RNAhybrid, TargetScan, Miranda, PITA) were used for target gene prediction and only the genes identified by all four approaches were selected out. And we chose the overlapped genes targeted by oncogenic miRNAs for further study, as well as the tumor suppressor miRNAs. To understand the functions of predicted target genes, Gene ontology (GO) and Kyoto encyclopedia of genes and genomes (KEGG) enrichment analysis were performed.8 IPA work flowThe miRNA regulatory network analysis was performed using the IPA software (http://www.ingenuity.com). The created genetic networks describe functional relationships among miRNAs and genes based on known associations in the databases. Networks related were ranked according to their biological relevance to the gene list provided.Candidate miRNAs are imported into IPA online database (http://www.ingenuity.com), which is a software for predicting the Target genes differentially expressed micrornas, rating and the connection between the credibility to them, and then reuse miRNA Target Filter function selection and tumor targeting the function of the related relationship, the miRNA regulation network built on the basis of these connections.9 Cell cultureThe human NPC cell line CNE2 were cultured in RPMI-1640 with 10% FBS and 1% Penicillin-Streptomycin antibiotic solution. All cells were cultured in a humidified atmosphere of 95% air and 5% C02 at 37℃.10 MiRNA transfectionThe transfection was performed using Lipofectamine 2000 reagent, and a working concentration of miRNA was 20 nmol in a 6-well plate.11 Western blotTotal protein was isolated and quantitated using BCA assay. The protein lysates were separated by 6% or 10% SDS-PAGE, and put 30 μg protein in every lane, electrophoretically transferred to PVDF membrane. Then, the membrane was incubated with antibodies and detected by chemiluminescence.12 Statistical analysisStatistical analyses were conducted using spss19.0 statistical software. All experiments were performed for three times.2-ΔCp was used for distinguished the NPC tissues and the control group, and also for the analysis of miRNAs in the different stages. The data are shown as the mean±SEM unless otherwise noted. Two-tailed Student’s t test was used for comparison of two independent groups. Comparison of three groups of or more than three sets of data using single factor analysis of variance, the variance homogeneity, using one-way ANOVA, the multiple comparisons using LSD method; Variance not neat when using approximate method of F test Welch, multiple comparison using Dunnett T3. P values of<0.05 were considered statistically significant.Results1. Differentially expressed miRNAs between NPC and controlsIn order to isolate tumor cells from non-tumor cells, LCM was performed on NPC and chronic nasopharyngitis specimens. According to the NCBI GenBank, Rfam (10.1) and mirbase21.0 database, we could match 827 human miRNA from clean date. After normalization and T-test on the clean data,8 differentially expressed miRNAs were screened,4 up-regulated (miR-205-5p, miR-92a-3p, miR-193b-3p and miR-27a-5p) and 4 down-regulated (miR-34c-5p, miR-375, miR-92b-3p and miR-449c-5p). The difference ratio and FDR of these 8 miRNAs were as follow, miR-205-5p (fold change=3.9103,FDR<0.01), miR-92a-3p (foldchange=4.5575,FDR<0.01), miR-193b-3p (fold change=257.719, FDR=0.03), miR-27a-5p (fold change=353.84, FDR=0.017), miR-34c-5p (fold change=-565.0915, FDR<0.01), miR-375 (fold change=-226.8843, FDR<0.01), miR-92b-3p (fold change=-10.7984, FDR<0.01), miR-449c-5p (fold change=-1968.9108, FDR<0.01). From the heat map and scatter plot analysis, we found these miRNAs could well distinguish the NPC samples from controls, out the differences between the miRNA expression spectrum can be NPC well apart from the normal controls. It meant the result had a good concordance and could be repeated in the same kind of tissue.2. Validation of expression of miRNAs and the relationship between the clinical stages and miRNAsTo confirm the deep sequencing results, we used qRT-PCR to assess the expression of 8 miRNAs with an independent cohort (including 65 NPC patients and 20 chronic nasopharyngitis patients). Expression of 7 miRNAs measured by qRT-PCR was dramatically different between NPC and chronic nasopharyngitis tissues and significantly correlated with their sequencing data. They were miR-205-5p (t=-5.209, P<0.001), miR-92a-3p (t=-4.457, P=0.001), miR-193b-3p (t=-3.64, P<0.001), miR-27a-5p (t=-2.977, P=0.005), miR-34c-5p (t=3.776, P=0.001), miR-375 (t=2.179, P=0.004), miR-92b-3p (t=-1.297, P=0.198) and miR-449c-5p (t=1.347, P=0.007). And miR-205-5p, miR-92a-3p, miR-193b-3p and miR-27a-5p are overexpressed, miR-34c-5p, miR-375 and miR-449 c-5p are downregulated. Because miR-92b-3p was not reliably measured by qRT-PCR in the tissue specimens, it was excluded from further analysis.The dynamic expression levels of miRNAs were revealed that the patterns were classified into 2 groups, and the varying tendencies between control and NPC were consistent with the sequencing result. Moreover, the expression level of miR-34c-5p decreased with the ascent of clinical stage, which implicated that miR-34c-5p might be more important in NPC progression. In the further validation, miR-34c-5p was significantly reduced in NPC patients who was in higher T stage (F= 3.735, P< 0.05). And it was showed the significant correlation with clinical stages respectively (F= 4.985, P< 0.05). But it had no obvious correction with the M and N stages.3. Target genes prediction and bioinformatics analysisFour softwares (RNAhybrid, TargetScan, Miranda, PITA) were used for target gene prediction and only the genes identified by all four approaches were selected out. And we chose the overlapped genes targeted by oncogenic miRNAs for further study, as well as the tumor suppressor miRNAs.666 genes targeted by downregulated miRNAs and 248 genes targeted by upregulated miRNAs were acquired after screening.The selected genes were analyzed by the Gene ontology (GO) enrichment and Kyoto encyclopedia of genes and genomes (KEGG) pathway analysis. The enrichment analysis of GO categories included biological process (BP), molecular function (MF), and cellular component (CC) three parts. The highest enrichment were the terms like "Interleukin-1 secretion", "Mast cell activation involved in immune response", "Fibroblast growth factor binding", "Lipopolysaccharide binding" and "Cortical actin cytoskeleton" (Table 3). These results showed the function of target genes mostly focused on cell growth, secretion, migration and immune system process.Chemokine signaling pathway, Cytokine-cytokine receptor interaction and Cell cycle, Natural killer cell mediated cytotoxicity, etc. were remarkable pathways which could further confirm the target genes’function in tumor cell proliferation, secretion and tumor immunology. So the differentially expressed miRNAs might be involved in the development of NPC by targeting these genes. And the "Natural killer cell mediated cytotoxicity" was the top term in KEGG analysis,34 target genes were working through this pathway,27 of these genes including killer cell immnoglobulin-like receptors (KIRs) could be targeted by miR-34c-5p and miR-449c-5p.4. MiRNA regulatory networks in NPCIPA showed the miR-34 family (including miR-34c-5p, miR-449c-5p) targeted the most genes in the regulation network of miRNA. The further analysis found the relationships between miR-34c and the 5 genes (TP53, CCND1, BCL2, CDK6, MET) in cancer was the key of the network. TP53 and BCL2 could be predicted to be the diagnostic marker of NPC, and CCND1 could be used to assess the effect of treatment of head and neck cancer. All of the above were thought to be involved in PI3K/AKT/mTOR signaling that modulated tumor cells growth and proliferation.To further characterize whether these target genes (MET, CCND1, CDK6, BCL2) respond to miR-34c in NPC cells, CNE-2 cells were transfected with miR-34c mimic or miR-Ctrl. Western blot was performed to confirm these relationships. With the literature review, we analyzed how the miR-34c-5p regulated TP53, CCND1, Bcl2, CDK6 and MET. Then we got a miR-34c-5p centred network, which mediated NPC cellular growth and proliferation, apoptosis and autophagy though the PI3K/AKT/mTOR signaling.Conclusions1. We got 8 diffentially expressed miRNAs in NPC via deep sequencing, including 4 upregulated miRNAs (miR-205-5p, miR-92a-3p, miR-193b-3p, miR-27a-5p), and 4 downregulated miRNAs (miR-34c-5p, miR-375, miR-92b-3p and miR-449c-5p).2. The validation in NPC tissue via qRT-PCR made sure that miR-205-5p, miR-92a-3p, miR-193b-3p and miR-27a-5p were upregulated in NPC, while miR-34c-5p, miR-375, miR-449c-5p were downregulated. And the expression level of miR-34c-5p was related with the clinical stages and T stages.3. There were 248 genes targeted by upregulated miRNAs, and 666 genes targeted by downregulated miRNAs. GO analysis showed the target genes focused on the cells build and motion function, when KEGG pathway analysis indicated that natural killer cell mediated cytotoxicity pathways might be the way of NPC immune regulation by miR-34c-5p and miR-449c-5p.4. We got the miR-34c-5p regulated network through IPA, which functioned in the tumorigenesis and progression of NPC, through regulating the expression level of TP53、MET、CCND1、CDK6 and Bcl2, and their pathway was thought to be the PI3K/AKT/mTOR signaling. The hypothesis provided molecular theory for the development of molecular targeted therapy and might reduce NPC recurrence.