Identification of novel long non-coding RNAs in clear cell renal cell carcinoma

Background Long non-coding RNAs (lncRNA) play an important role in carcinogenesis; knowledge on lncRNA expression in renal cell carcinoma is rudimental. As a basis for biomarker development, we aimed to explore the lncRNA expression profile in clear cell renal cell carcinoma (ccRCC) tissue. Results Microarray experiments were performed to determine the expression of 32,183 lncRNA transcripts belonging to 17,512 lncRNAs in 15 corresponding normal and malignant renal tissues. Validation was performed using quantitative real-time PCR in 55 ccRCC and 52 normal renal specimens. Computational analysis was performed to determine lncRNA-microRNA (MiRTarget2) and lncRNA-protein (catRAPID omics) interactions. We identified 1,308 dysregulated transcripts (expression change >2-fold; upregulated: 568, downregulated: 740) in ccRCC tissue. Among these, aberrant expression was validated using PCR: lnc-BMP2-2 (mean expression change: 37-fold), lnc-CPN2-1 (13-fold), lnc-FZD1-2 (9-fold), lnc-ITPR2-3 (15-fold), lnc-SLC30A4-1 (15-fold), and lnc-SPAM1-6 (10-fold) were highly overexpressed in ccRCC, whereas lnc-ACACA-1 (135-fold), lnc-FOXG1-2 (19-fold), lnc-LCP2-2 (2-fold), lnc-RP3-368B9 (19-fold), and lnc-TTC34-3 (314-fold) were downregulated. There was no correlation between lncRNA expression with clinical-pathological parameters. Computational analyses revealed that these lncRNAs are involved in RNA-protein networks related to splicing, binding, transport, localization, and processing of RNA. Small interfering RNA (siRNA)-mediated knockdown of lnc-BMP2-2 and lnc-CPN2-1 did not influence cell proliferation. Conclusions We identified many novel lncRNA transcripts dysregulated in ccRCC which may be useful for novel diagnostic biomarkers. Electronic supplementary material The online version of this article (doi:10.1186/s13148-015-0047-7) contains supplementary material, which is available to authorized users.


Background
Renal cell carcinoma (RCC) is one of the most common malignancies; its incidence is varying substantially worldwide: RCC incidence is high in Europe and North America and low in Asia and South America [1]. Today, many small-sized renal tumors are diagnosed; however, imaging modalities do not allow precise differentiation between renal cell carcinoma and non-malignant renal tumors. Performing tumor biopsy and histopathological classification is sometimes challenging, and definitive exclusion of malignancies still requires surgical exploration. As active surveillance protocols for small renal lesions find the way into daily clinical practice, a better estimation of tumor aggressiveness becomes necessary. Thus, identifying novel biomarkers would be helpful for the management of patients with renal tumors.
Nucleic acids are under discussion as potential biomarkers for patients with RCC [2]. Long non-coding RNAs (lncRNA) are a class of RNA molecules arbitrarily defined as being longer than 200 nucleotides and not translated into a protein. Initially, it was thought that lncRNAs represent transcriptional noise, but it is nowadays recognized that lncRNAs may have biological roles. They are regulating imprinting, dosage compensation, cell cycle, pluripotency, retrotransposon silencing, and meiotic entry, for example [3]. However, knowledge on lncRNA expression and their function is still in its infancy, but lncRNA expression seems to be highly tissue and tumor specific [4]. So far, few are known on lncRNAs expression in RCC [5][6][7][8]. We therefore performed microarray experiments to study the expression of 32,183 lncRNA transcripts in a cohort of patients with clear cell RCC (ccRCC) to determine a comprehensive lncRNA profile.

Results
Microarray: screening for aberrantly expressed lncRNAs A gene expression microarray was used to identify dysregulated lncRNAs in 15 corresponding tumor and normal renal tissue samples as a discovery cohort. Among the 32,183 analyzed lncRNA transcripts, we observed differential expression (fold change >2) in 1,308 transcripts: 568 lncRNA transcripts were upregulated and 740 were downregulated in ccRCC samples. The 20 most differentially expressed lncRNAs in ccRCC and normal renal tissue are listed in Table 1. A hierarchical cluster analysis based on centered Pearson correlation coefficient was performed to determine different expression profiles in ccRCC and normal tissue. As shown in Figure 1, the lncRNA expression profile allowed distinguishing cancerous and normal tissue samples highly accurately.
We also analyzed whether lncRNA expression levels were correlated with poor prognostic parameters; the Bonferroni method was applied to correct for multiple hypothesis testing (adjusted significance value p < 0.0083). None of the lncRNAs was significantly correlated with staging nor grading. However, there was a tendency towards lower lnc-ERCC5-1 (p = 0.034) and RP3-368B9 (p = 0.016) levels in Fuhrman grade 3/4 tumors, lower lnc-ERCC5-1 (p = 0.026) expression in vascular invasive tumors, and advanced AJCC stage in samples with low lnc-RP3-368B9 (p = 0.011). lncRNAs were not associated with progression-free survival, overall survival or cancer-specific survival (all p > 0.05, data not shown), as determined using Cox regression analysis.
(See figure on previous page.) Figure 2 Validation of lncRNA dysregulation. The expression of 13 target lncRNAs in renal cell carcinoma (red) and normal renal (green) tissue was validated using quantitative real-time PCR; the expression levels were normalized using ACTB and PPIA as reference genes. We confirmed significant (all p < 0.001) overexpression of lnc-BMP2-2 (B), lnc-CPN2-1 (C), lnc-FZD1-2 (F), lnc-ITPR2-3 (G), lnc-SLC30A6-1 (K), and lnc-SPAM1-6 (L) in renal cell carcinoma, whereas lnc-ACACA-1  In addition, lncRNAs may be of prognostic relevance: Fachel et al. [7] reported a 26-gene lncRNA expression profile, but not a single lncRNA, associated with the survival of ccRCC patients. However, the small sample size (n = 16) and the lack of a validation cohort limit the informative value of the study [7]. Malouf et al. [8] identified four lncRNA expression clusters among the study cohort, and the survival was significantly reduced in one subgroup (C2). We did not observe a correlation of lncRNAs (used in the validation study) with any clinicopathological parameter nor survival data. Reasons for the failure to identify prognostic relevant lncRNAs include: (i) a microarray study not powered for the identification of expression differences in localized (n = 5) and advanced RCC (n = 10), (ii) the selection of lncRNAs for validation based on the differential expression of tumor and normal tissue samples, and (iii) the high number of patients with localized ccRCC (56% AJCC stage 1 or 2) in the validation cohort. We expect that future studies designed to identify prognostic relevant lncRNAs in RCC patients will be able to present some candidates, because altered expression of specific lncRNAs was predictive of cancer-specific survival in former studies on other human malignancies; i.e. upregulation of HOTAIR in breast cancer [20] and SChLAP1 in prostate cancer patients [21] was associated with a poor prognosis.
Somewhat surprisingly, neither siRNA-mediated knockdown of lncRNA-CPN2-2 nor lnc-BMP2-2 resulted in a significant reduction of cell proliferation. However, the lncRNA knockdown was of moderate success (relative expression change in siRNA-treated cell lines 2-to 5-fold, despite of the extensive optimization of siRNA treatment in preliminary experiments), and thus the toxicity of the transfection reagent may inhibit determining the functional relevance of both lncRNAs. Many lncRNAs regulate nuclear events and must therefore be localized in the nucleus [26]. As is known, the knockdown of nuclear RNAs is often difficult [27], and therefore our efforts to determine the functional relevance of lncRNAs failed.

Conclusions
In summary, our study reveals that dysregulation of approximately 4% of the lncRNA transcripts occurs in ccRCC, and altered lncRNA expression may modulate fundamental cellular processes. lncRNA profiles allow to accurately distinguish ccRCC and normal renal tissue. Thus, lncRNAs may be used as non-invasive biomarker for RCC patients.

Patients
Biomaterials including fresh-frozen tissues are prospectively collected in the Biobank at the CIO Köln Bonn at the Universitätsklinikum Bonn according to standard operating procedures. Tissue from patients undergoing radical nephrectomy or nephron-sparing surgery was snap frozen, and samples from tumor and normal renal tissue were stored at liquid nitrogen. Hematoxylin-and eosin-stained sections were performed to confirm the histology of the samples. The final histological diagnosis was made on the formalin-fixed paraffin-embedded tissue samples. All cases used in this study were reviewed by an experienced uropathologist; the TNM classification (7th edition from 2009) was applied. We used a discovery cohort with 15 patients (ccRCC and adjacent normal renal tissue) for screening 32,183 lncRNA transcripts; the independent validation cohort consisted of 55 ccRCC and 52 normal renal tissue samples. The detailed clinical-pathological parameters are reported in Table 4. All patients gave written informed consent prior collection of biomaterials. The study was approved by the Ethikkommission at the Universitätsklinikum Bonn (number: 280/12).

Cell culture and siRNA experiments
The cell lines Caki-1, Caki-2, and A-498 were obtained from the DSMZ (Braunschweig, Germany). All cell lines were maintained at 37°C, 5% CO 2 in RPMI 1640 culture medium, supplemented with 10% heat-inactivated fetal

Cell proliferation assay
The cell lines were seeded at a concentration of 1.5 × 10 4 cells per well into a 96-well plate. After forward transfection (siRNA concentration 1.2 pmol), the cells were cultured up to 72 h to determine the proliferative activity after siRNA-mediated knockdown of lnc-BMP2-2 and lnc-CPN2-1. The EZ4U assay (Biomedica, Vienna, Austria) was used to determine cell viability using a 340 ATTC Spectra Thermo SLT photometer (Crailsheim, Germany) according to the manufacturer's protocol.

RNA isolation
Fresh-frozen tissue samples (approximately 50 mg) were cut on liquid nitrogen and homogenized with Yttrium

Microarray
The microarray experiments were performed by Biogazelle (Zwijnaarde, Belgium) as a contract service. RNA isolates from 15 corresponding normal renal and ccRCC tissues (100 ng total RNA) were sent on dry ice to Biogazelle. A custom microarray (Agilent SurePrint G3 technology) based on LNCipedia 2.1 [10] was used to study 32,183 lncRNA transcripts belonging to 17,512 lncRNAs. Background substracted and normalized (log2-based) expression data were provided from Biogazelle for further data analysis. Expression data of lncRNAs for ccRCC and normal tissues was groupwise normalized deducting group mean value. Pairwise fold change, for each gene, was calculated by dividing ccRCC group mean of RNA expression level by normal group mean of expression levels. Pearson's correlation coefficient was used for correlation analyses (R-Base Package v3.0.2). A heatmap was created using gplots v2.12.1 and RColorBrewer v1.0-5.

Real-time PCR
Quantitative real-time PCR was performed to validate the microarray experiments using an independent cohort of 55 ccRCC and 52 normal renal tissue samples and to confirm the success of siRNA-mediated lncRNA knockdown. Complementary DNA (cDNA) was synthesized with 1 μg RNA using the PrimeScript RT Reagent Kit with gDNA Eraser. Real-time PCR was performed with 5 ng cDNA template using the 1× SYBR Premix Ex Taq II with ROX Plus and 10 pmol/μl PCR primers (see Additional file 3: Table S1 for primer sequences); all reagents were from Takara Bio, Saint-Germain-en-Laye, France. PCR was performed using an ABIPrism 7900 HT Fast Real-Time PCR System (Applied Biosystems, Foster City, CA, USA). Data analysis was performed using Qbase + (Biogazelle) with PPIA and ACTB as reference genes in the 2-ΔΔCT algorithm. SPSS Statistics v21 (IBM, Ehningen, Germany) was used for statistical analyses (Mann-Whitney-U test, Cox regression analysis).

MicroRNA quantification
Quantitative real-time PCR was used to determine the tissue levels of several FOXG1-2 microRNA targets in each ten normal and malignant renal tissues using the Qiagen miScript SYBR Green PCR technology (Hilden, Germany): 1,000 ng total RNA was reverse transcribed and real-time PCR was performed with 5 ng cDNA template. Pre-designed miScript Primer Assays for miR-519a, miR-519b, miR-519c, miR-548, miR-3120, miR-4284, miR-4658, and SNORD43 were used. PCR was performed using the ABIPrism 7900 HT Fast Real-Time PCR System. Data analysis was performed using Qbase + with SNORD43 as reference genes in the 2-ΔΔCT algorithm. IBM SPSS Statistics v21 was used for statistical analyses (Pearson correlation test).