Fifty-six patients with urothelial carcinoma of the bladder who received radical cystectomy at the Fudan University Shanghai Cancer Center (Shanghai, China) were included in the study. Urothelial carcinoma of the bladder was diagnosed histopathologically. Written informed consent was obtained from all patients and the study was approved by the Institutional Review Board of Fudan University Shanghai Cancer Center (Shanghai, China). Of the 56 pairs of samples, five pairs were used in lncRNA microarray analysis [one pair was excluded from analysis according to three dimension principal component analysis (3D-PCA)] and 51 pairs were analyzed by qPCR. The tumor sample and matched normal bladder tissue from each subject were snap-frozen in RNA ladder immediately after resection and stored in the tissue bank.

If the proportion of cancer cells in a tissue section was >80% then the frozen block was subjected to RNA extraction. Total RNA was extracted from 56 pairs of snap-frozen urothelial carcinoma and matched normal bladder tissues using TRIzol reagent (Invitrogen Life Technologies, Carlsbad, CA, USA) according to the manufacturer’s instructions. The RNA integrity was evaluated by a NanoDrop ND-1000 spectrophotometer (NanoDrop products, Wilmington, DE, USA).

RNA purified from total RNA following the removal of rRNA was amplified and transcribed into fluorescent cRNA along the entire length of the transcripts without 3′ bias utilizing a random priming method and cDNA was labeled and hybridized to the Human LncRNA Array V2.0 (8×60 K, Arraystar Inc., Rockville, MD, USA). A total of 33,045 lncRNAs and 30,215 coding transcripts, which were collected from the most authoritative databases, including RefSeq, UCSC Known Genes, Ensembl and the associated literature, were detected by the microarray.

The microarray work was performed by KangChen Bio-tech (Shanghai, China). In brief, the Arraystar lncRNA array protocol was as follows: i) the RNA sample, kit and reagents were prepared, including TRIzol reagent, Biopulverizer (Biospec, Bartlesville, OK, USA) and Mini-Bead-Beater-16 (Biospec); ii) Total RNA Clean-up and RNA quality control; iii) labeling reaction was prepared; iv) labeled/amplified RNA and labeled cRNA QC were purified; v) hybridization was performed; vi) microarray wash was conducted; vii) scanning was performed; and viii) the data were extracted using Agilent feature extraction software (Agilent Technologies, Santa Clara, CA, USA).

The arrays were scanned by the Agilent Scanner G2505B (Agilent Technologies) and the acquired array images were analyzed by Agilent Feature Extraction software (version 10.7.3.1; Agilent Technologies). Quantile normalization and subsequent data processing were performed using the GeneSpring GX v11.5.1 software package (Agilent Technologies).

To discover the function and associated pathways of differentially expressed mRNAs, GO and pathway analyses were performed. GO annotations of microarray genes were downloaded from NCBI (http://www.ncbi.nlm.nih.gov/), UniProt (http://www.uniprot.org/) and the Gene Ontology (http://www.geneontology.org/). The elim Fisher algorithm was used to perform a GO enrichment test and GO categories with P<0.05 were reported. Pathway annotations of microarray genes were download from KEGG (http://www.genome.jp/kegg/) and a Fisher exact test was performed in order to locate the significant enrichment pathway. The resulting P values were adjusted using the Benjamini Hochberg false discovery rate (BH FDR) algorithm. Pathway categories with a FDR<0.05 were reported.

The network construction procedures included the following: i) preprocessing of data: if one coding gene has different transcripts the median value is taken to represent the gene expression values, without special treatment of lncRNA expression values; ii) data were screened and the subset of data were removed according to the lists of the differential expression of lncRNA and mRNA obtained from the GO and pathway analyses; iii) the Pearson correlation coefficient was calculated and the R value was used to calculate the correlation coefficient between lncRNA and coding genes; and iv) Pearson’s correlation coefficient was used for screening; RNAs with a Pearson’s correlation coefficient of ≥0.99 were considered significant and the lncRNA-mRNA co-expression network was constructed by Cytoscape software (The Cytoscape Consortium, San Diego, CA, USA).

To validate the microarray results, qPCR was performed. Total RNA was extracted from frozen tumor specimens using TRIzol reagent and then reverse transcribed using the Maxima Probe qPCR Master mix (Fermentas, Waltham, MA, USA) according to the manufacturer’s instructions. lncRNA expression levels in bladder cancer tissues were measured by qPCR using a GeneAmp PCR System 9700 (Applied Biosystems, Waltham, MA, USA). The primers used in this study are shown in Table I. Four lncRNAs that were significantly deregulated (TNXA, CTA-134P22.2, CTC-276P9.1 and KRT19P3) were evaluated in the patients included in this study.

Total RNA (2 μg) was converted to cDNA according to the manufacturer’s instructions. PCR was performed in a total reaction volume of 20 μl, including 10 μl Master Premix (2X, with ROX Reference Dye II), 1 μl of PCR Forward Primer (10 μM), 1 μl of PCR Reverse Primer (10 μM), 0.2 μl Roche probe (100X), 2 μl of cDNA and 5.8 μl of double-distilled water. The qPCR reaction was set at an initial denaturation step of 10 min at 95°C; and 95°C (15 sec), 57°C (30 sec), 72°C (30 sec) for a total 40 cycles. The experiments were performed in triplicate. The samples were normalized to β-actin. The median in each triplicate was used to calculate relative lncRNA concentrations (ΔCt = Ct median lncRNAs - Ct median β-actin). Expression fold changes were calculated using the 2^−ΔΔCt method.

Statistical analysis was performed using Student’s t-test for the comparison of the two groups in microarray, and analysis of variance for multiple comparisons. P<0.05 was considered to indicate a statistically significant result. The statistical significance of the microarray result was analyzed by fold change and Student’s t-test. The FDR was calculated to correct the P-value. The threshold value used to screen differentially expressed mRNAs was a fold change of ≥2.0, and a fold change of ≥8.0 for differentially expressed lncRNAs.

The expression profiles of lncRNAs in paired samples were shown by calculating the log₂ fold-change tumor/normal (T/N). The agreement was formulated as follows: Fold change (FC) cut-off: 2.0. Positive fold change values indicated upregulation and negative values indicated downregulation. Log fold change means log₂ of absolute fold-change (log₂FC). The fold change and P-value were calculated from the normalized expression. One pair of samples was excluded from analysis according to 3D-PCA. We finally identified 3,324 differentially expressed human lncRNAs in four bladder cancer patients (≥2-fold).

A total of 110 lncRNAs were significantly differentially expressed between the tumor and control groups (≥8-fold). Twenty-two lncRNAs were upregulated and 88 lncRNAs were downregulated in the tumor group compared with the controls (Table II). Log₂FC of upregulated lncRNAs in the tumor group ranged between 3.017291 to 4.581319, and −3.00191 to −6.10723 of downregulated lncRNAs. RP11-58A12.3 (log₂FC=−6.10723) was the most significantly downregulated lncRNA and RNU12 (log₂FC=4.581319) was the most significantly upregulated lncRNA. We observed that downregulated lncRNAs were more common than upregulated lncRNAs.

Up to 17,069 coding transcripts were detected in four pairs of samples using 30,215 coding transcripts probes. Among the four pairs of samples, 1,269 mRNAs were upregulated in tumor samples compared with the matched normal tissues, while 851 mRNAs were downregulated. Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis showed that the differentially expressed mRNAs may be associated with p53, bladder cancer, cell cycle and propanoate metabolism pathways (Fig. 1). These results suggest that bladder cancer has a variety of genetic and phenotypic characteristics.

Four lncRNAs were selected for further confirmation of microarray results using qPCR. These lncRNAs were among the most downregulated or upregulated lncRNAs. Data analysis showed that KRT19P3 was upregulated and TNXA, CTA-134P22.2 and CTC-276P9.1 were downregulated in bladder cancer samples compared with matched normal tissues (Fig. 2). These data support a strong correlation between the qPCR result and the microarray data (Fig. 3).

We constructed a coding-non-coding gene co-expression network based on the correlation analysis between the differentially expressed lncRNA and mRNA. We used lncRNAs and mRNAs with Pearson’s correlation coefficients of no less than 0.99 to construct the network. In total, 79 lncRNAs and 103 mRNAs were included in the co-expression network. The co-expression network indicated that one mRNA or lncRNA may correlate with one to tens of lncRNAs. The co-expression network may suggest that the inter-regulation of lncRNAs and mRNAs is involved in bladder cancer.