Radiotherapy and Oncology 108 (2013) 451–457
Contents lists available at SciVerse ScienceDirect
Radiotherapy and Oncology journal homepage: www.thegreenjournal.com
MicroRNA expression
Identification of a microRNA expression signature for chemoradiosensitivity of colorectal cancer cells, involving miRNAs-320a, -224, -132 and let7g Junius Salendo a, Melanie Spitzner a, Frank Kramer b, Xin Zhang c, Peter Jo a, Hendrik A. Wolff d, Julia Kitz e, Silke Kaulfuß f, Tim Beißbarth b, Matthias Dobbelstein c, Michael Ghadimi a, Marian Grade a, Jochen Gaedcke a,⇑ a e
Department of General and Visceral Surgery; b Department of Medical Statistics; c Department of Molecular Oncology; Department of Pathology; and f Department of Human Genetics, University Medical Center Göttingen, Germany
a r t i c l e
i n f o
Article history: Received 10 April 2013 Received in revised form 26 June 2013 Accepted 28 June 2013 Available online 7 August 2013 Keywords: MiRNA Rectal cancer Preoperative chemoradiotherapy let-7g MiR320
d
Department of Radiotherapy and Radiooncology;
a b s t r a c t Background and purpose: Preoperative chemoradiotherapy (CRT) represents the standard treatment for locally advanced rectal cancer. Tumor response and progression vary considerably. MicroRNAs represent master regulators of gene expression, and may therefore contribute to this diversity. Material and methods: Genome-wide microRNA (miRNA) profiling was performed for 12 colorectal cancer (CRC) cell lines and an individual in vitro signature of chemoradiosensitivity was established. Functional relevance of selected miRNAs was established by transfecting miRNA-mimics into SW480 and SW837 cells. The prognostic value of selected miRNAs was assessed in 128 pretherapeutic patient biopsies. Results: Thirty-six miRNAs were identified to significantly correlate with sensitivity to CRT (Q < 0.05) including miR-320a and other miRNAs involved in the MAPK-, TGF- and Wnt-pathway. Transfection of selected miRNAs (let-7g, miR-132, miR-224, miR-320a) each induced a shift of sensitivity. High expression of let-7g was associated with a good prognosis in rectal cancer patients (P = 0.03). Conclusions: This is the first report of a miRNA expression signature for in vitro chemoradiosensitivity of CRC cell lines. Many of the identified miRNAs have not been linked to the response to CRT and may represent potential molecular targets to sensitize resistant cancers. If further validated, let7g expression may serve as predictive biomarker. Ó 2013 Elsevier Ireland Ltd. All rights reserved. Radiotherapy and Oncology 108 (2013) 451–457
The treatment of malignant diseases is accompanied by the clinical dilemma of varying sensitivity to different treatment modalities. The identification of the underlying mechanisms and predictive markers as well as deciphering methods to overcome resistance is an urgent need for cancer patient therapy. In advanced stages of the disease, rectal cancer patients are treated with a combination of 5-FU-based chemotherapy and radiation followed by standardized surgery [1,2]. Some patients respond very well to this treatment concept, and consequently reveal a better prognosis. In contrast, patients with resistant tumors require alternative therapeutic strategies [3]. Therefore, it remains of utmost clinical importance to analyze the underlying mechanisms of resistance to chemoradiotherapy (CRT), which may enable the identification of predictive markers as well as potential molecular targets to realize an individualized rectal cancer treatment [4,5].
⇑ Corresponding author. Address: Department of General and Visceral Surgery, Robert-Koch-Straße 40, 37075 Göttingen, Germany. E-mail address:
[email protected] (J. Gaedcke). 0167-8140/$ - see front matter Ó 2013 Elsevier Ireland Ltd. All rights reserved. http://dx.doi.org/10.1016/j.radonc.2013.06.032
Very recently, we established a cell line based chemoradiosensitivity model [6]. We identified genes whose expression levels correlated significantly with the sensitivity to CRT. Many of these genes were involved in the mitogen-activated protein kinase (MAPK) and Wnt signaling pathway or cell cycle genes. In the present study, we now aimed to analyze the potential relevance of specific microRNAs (miRNA) on the response of CRC cell lines to RCT. MiRNAs are highly conserved, noncoding RNAs ranging between 21 and 24 nucleotides in size. They play a crucial role in the post-transcriptional regulation of mRNA and consecutively influence physiological and patho-physiological processes [7]. MiRNAs are involved in different stages of colorectal cancer (CRC) pathogenesis by regulating the expression of oncogenes and tumor suppressor genes [8]. Furthermore, they are involved in the regulation of radioresistance [9,10]. However, a comprehensive miRNA expression analysis of CRC cell lines with respect to their intrinsic sensitivity to 5-FU-based CRT remains to be established. We therefore applied comprehensive miRNA profiling to a previously established CRC cell line model [6]. The vast majority of the identified miRNAs have not been identified to play a role
452
MicroRNA chemoradiosensitivity signature
in chemoradioresistance so far. Representative miRNAs were then functionally analyzed for their potential influence on resistance by determining the survival fraction after miRNA-mimic transfection and consecutive CRT. The independent analysis of rectal cancer biopsies identified let7g as promising marker to predict response to CRT and as an interesting therapeutic candidate. Materials and methods Cell lines and cell culture Cell lines, procedures to exclude Mycoplasma contamination and cross-contamination of cell lines, and the establishment of an in vitro protocol for 5-FU-based chemoradiotherapy have been recently described in detail [6]. For functional validation of selected miRNAs, the respective mimics for miR-320a, let-7g, miR132, and miR-224 were purchased from Qiagen (Hilden, Germany) and transfected into SW480 and SW837 cells as recently described [11] using Amaxa Nucleofactor System (Lonza, Cologne, Germany). AllStars negative control siRNA (Qiagen) served as negative control. Successful transfection was confirmed through western blot analysis of putative target gene products of the respective miRNA mimics. Chemo-/radiotherapy and determination of cell survival The establishment of the cellline resistency model has previously been described. To validate identified target miRNAs, cells were exposed to chemotherapy (3 lM of 5-FU) and radiation (doses of 1, 2, 4, 6 and 8 Gy) 96 h after transfection. Chemotherapy and radiation were performed either alone or as combined treatment. Cell survival was determined by a standard colony formation assay as previously described [6,12]. The efficacy of mimic transfection was compared to negative control and Dose Modifying Factor (DMF) was calculated. RNA isolation For each cell line, total RNA was isolated from three different passages (passage four to six, at 60–70% confluence) using the miRNeasy Mini kit (Qiagen), according to the manufacturer’s instructions. RNA quantity, quality and integrity were analyzed using a Nanodrop and 2100 a Bioanalyzer (Agilent Technologies, Inc., Santa Clara, CA, USA). Only samples with a RNA integrity number >9.5 were considered for further experimentation. Patient biopsies were stored in RNAlater, and total RNA was isolated using Trizol as previously described [13]. Expression profiling Each cell line was analyzed as a triplicate using the 60 K Agilent Human microRNA Microarray (Agilent Technologies). In brief, 500 ng of total RNA and Spike-In were labeled by ligation to Cy3conjugated 30 ,50 -cytidine-bisphosphate using T4 RNA ligase. Spin columns were used for desalting (BioRad, Hercules, CA, USA). Subsequently, hybridization was carried out for 20 h. Slides were washed and finally scanned on an Agilent DNA microarray scanner G2505B (Agilent Technologies) at 3 micron resolution. The respective miRNA expression data have been uploaded to GEO (Reviewer Accession Link: http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi? token=hjmzxmywcccsmty&acc=GSE40976). Semi-quantitative real-time PCR The miScript SYBR Green PCR Kit (Qiagen) and the QuantiTect SYBR Green PCR Master Mix (Qiagen) were used to detect miRNA
expression. cDNA containing universal tag was generated from total RNA using the miScript Reverse Transcription Kit (Qiagen) according to the manufacturer’s instructions. Forward primers were obtained from Qiagen’s database, and the miScript Universal Primer (Qiagen) was used as reverse primer. To technically validate microarray-based miRNA expression measurements, miR-300, miR-1914⁄ and miR-26b were chosen for normalization, as their expressions showed the least variance within the 12 cell lines as obtained from the microarray analysis (data not shown). To analyze selected miRNAs in patient samples U44, miR-202 and miR874 were used for normalization, Semi-quantitative real-time PCR was performed using the ABI 7900HT system (Applied Biosystem). Pretherapeutic biopsies from patients treated with preoperative CRT To assess the prognostic value of selected miRNAs, their relative expression was assessed in pretherapeutic biopsies from 128 patients. Staging, treatment and follow-up were performed according to the trial guidelines of the German Rectal Cancer Study Group [14,15]. Biopsies were taken in 4 different departments. All patients received a combination of either 5-FU and 50.4 Gy of radiation, or 5-FU, Oxaliplatin and 50.4 Gy of radiation prior to standardized radical surgery and adjuvant chemotherapy (clinical data are summarized in Supplementary Table 2). Expression data were correlated to tumor regression grade (TRG), and DFS. The work on patient biopsies complies with the principles laid down in the Declaration of Helsinki and was approved by the local ethics committee. Western blot analysis Western blot analysis was performed as recently described [12]. Cells were lysed using lysis buffer including a protease and phosphatase inhibitor cocktail. For blocking 5% blotting grade milk was applied and membranes were probed using antibodies against Survivin (1:1000; Cell Signaling, Massachusetts, USA), p-AKT (1:2000; Cell Signaling) and c-MYC (1:2000; Cell Signaling) respectively. As secondary antibody a goat anti-rabbit peroxidase linked antibody (1:30,000; Acris Antibodies, Herford, Germany) was used. Membranes were developed using an enhanced chemiluminescence detection system (ECL Advanced, GE Healthcare, Buckinghamshire, UK), and signals were detected using a CCD-Camera (Image Quant LAS 4000 mini, Uppsala, Sweden). MicroRNA pathway analysis and target prediction To identify affected pathways, first DIANA-miRPath (http://diana.cslab.ece.ntua.gr/pathways), a free web-based tool that performs enrichment analysis of microRNA target genes to all known KEGG pathways was applied [16]. Within this tool, potential microRNA targets were assessed by Diana-microT-4.0 (beta version). Additionally, we applied the commercially available software Ingenuity Pathway Analysis (IPA) v.8.8 (IngenuityÒ Systems, www.ingenuity.com, Ingenuity Systems, Redwood City, CA). IPA is a web-based software to analyze a list of genes, miRNAs or other molecular data that are typically generated by high-throughput analyses. Using a database of biological interactions and functional annotations molecular networks can be created to vizualize functions and relationships of the entered data. Here, we assessed the pathways that are regulated by the mRNA targets of the differentially expressed miRNAs. Only mRNAs were used that were known to be involved in chemoradioresistance as previously established by Spitzner et al. [6]. Computationally predicted targets of each miRNA were identified using microcosm (http://www.ebi.ac.uk/enright-srv/
J. Salendo et al. / Radiotherapy and Oncology 108 (2013) 451–457
microcosm/htdocs/targets/v5/). To validate these targets several filter criteria were applied as we recently described [17]. First, mRNA targets that have not been significantly regulated in the previous mRNA analyses of the CRC cell lines [6] were excluded. Next, only mRNAs were included that were downregulated if the miRNA was upregulated and vice versa. However, this ‘‘overfiltering’’ did not respect miRNA-mRNA interactions with incomplete complementation that only prevents translation. Statistical analysis Expression levels were analyzed using log2 transformation and quantile normalization [18]. Except for control spots, all 1347 features were used without any a priori filtering. Linear models, fitted on a gene-by-gene basis, were used to assess a linear correlation between miRNA expression levels and surviving fractions at 3 lM 5-FU and 2 Gy of X-rays (SF2). We applied an empirical Bayes estimator [19] to compute linear models for hundreds of miRNAs in parallel and to assess their significance. In order to not exceed a false discovery rate (q-value, Q) of 5%, p-values were adjusted for multiple testing using the Benjamini-Hochberg method [20]. All analyses were performed using the free statistical software R (version 2.14.1; http://www.r-project.org). Linear models were computed using the ‘limma’ package. A multiple linear regression model was used to describe the normalized surviving fraction as dependent variable, given the independent variables of irradiation dose, treatment group (control group or miRNA mimic) and replicate. Two different models including either only irradiation dose and replicate or additionally a treatment effect and a treatment irradiation dose interaction term were compared with analysis of variance for each miRNA. A significant outcome with p < 0.05 suggests an influence of the treatment on the dose response. Survival data were visualized using Kaplan–Meier plots and the effect of the individual miRNAs on survival was assessed using Cox proportional hazards regression. A Cox regression model was calculated miRNA-wise, correlating miRNA expression levels and time to event data. Disease free survival was calculated as the time from surgery to local recurrence or distant metastasis. For visualization on Kaplan–Meier plots patients were grouped depending whether they had an expression level above or below median expression level for a particular miRNA. Survival analysis was conducted using the R package survival. P-values
